Systems and Methods for Identifying Cosmetic Agents for Skin Care Compositions

Abstract

Provided are methods and systems for determining functional relationships between a cosmetic agent and a skin tissue condition of interest. Also provided are methods and systems for identifying cosmetic agents that affect a skin aging tissue condition, as well as the use of agents identified by such methods and systems for the preparation of cosmetic compositions, personal care products, or both.

Description

This application claims priority under 35 U.S.C. §119(e) to U.S. Provisional Application 61/445,315 filed on Feb. 22, 2011.

BACKGROUND

Connectivity mapping is a well-known hypothesis generating and testing tool having successful application in the fields of operations research, computer networking and telecommunications. The undertaking and completion of the Human Genome Project, and the parallel development of very high throughput high-density DNA microarray technologies enabling rapid and simultaneous quantization of cellular mRNA expression levels, resulted in the generation of an enormous genetic database. At the same time, the search for new pharmaceutical actives via in silico methods such as molecular modeling and docking studies stimulated the generation of vast libraries of potential small molecule actives. The amount of information linking disease to genetic profile, genetic profile to drugs, and disease to drugs grew exponentially, and application of connectivity mapping as a hypothesis testing tool in the medicinal sciences ripened.

The general notion that functionality could be accurately determined for previously uncharacterized genes, and that potential targets of drug agents could be identified by mapping connections in a data base of gene expression profiles for drug-treated cells, was spearheaded in 2000 with publication of a seminal paper by T. R. Hughes et al. [“Functional discovery via a compendium of expression profiles” Cell 102, 109-126 (2000)], followed shortly thereafter with the launch of The Connectivity Map (-map Project by Justin Lamb and researchers at MIT (“Connectivity Map: Gene Expression Signatures to Connect Small Molecules, Genes, and Disease”, Science, Vol 313, 2006.) In 2006, Lamb's group began publishing a detailed synopsis of the mechanics of C-map construction and installments of the reference collection of gene expression profiles used to create the first generation C-map and the initiation of an on-going large scale community C-map project, which is available under the “supporting materials” hyperlink at http://www.sciencemag.org/content/313/5795/1929/suppl/DC1.

The basic paradigm of predicting novel relationships between disease, disease phenotype, and drugs employed to modify the disease phenotype, by comparison to known relationships has been practiced for centuries as an intuitive science by medical clinicians. Modern connectivity mapping, with its rigorous mathematical underpinnings and aided by modern computational power, has resulted in confirmed medical successes with identification of new agents for the treatment of various diseases including cancer. Nonetheless, certain limiting presumptions challenge application of C-map with respect to diseases of polygenic origin or syndromic conditions characterized by diverse and often apparently unrelated cellular phenotypic manifestations. According to Lamb, the challenge to constructing a useful C-map is in the selection of input reference data which permit generation of clinically salient and useful output upon query. For the drug-related C-map of Lamb, strong associations comprise the reference associations, and strong associations are the desired output identified as hits.

Noting the benefit of high-throughput, high density profiling platforms which permit automated amplification, labeling hybridization and scanning of 96 samples in parallel a day, Lamb nonetheless cautioned: “[e]ven this much firepower is insufficient to enable the analysis of every one of the estimated 200 different cell types exposed to every known perturbagen at every possible concentration for every possible duration . . . compromises are therefore required” (page 54, column 3, last paragraph). Hence, Lamb confined his C-map to data from a very small number of established cell lines. This leads to heightened potential for in vitro to in vivo mismatch, and limits information to the context of a particular cell line. Selection of cell line, therefore, may be critical to the utility of a resulting C-map.

Lamb stresses that particular difficulty is encountered if reference connections are extremely sensitive and at the same time difficult to detect (weak), and Lamb adopted compromises aimed at minimizing numerous, diffuse associations. Since the regulatory scheme for drug products requires high degrees of specificity between a purported drug agent and disease state, and modulation of disease by impacting a single protein with a minimum of tangential associations is desired in development of pharmaceutical actives, the Lamb C-map is well-suited for screening for potential pharmaceutical agents despite the Lamb compromises.

The connectivity mapping protocols of Lamb would not be predicted, therefore, to have utility for hypothesis testing/generating in the field of cosmetics. Cosmetic formulators seek agents or compositions of agents capable of modulating multiple targets and having effects across complex phenotypes and conditions. Further, the phenotypic impact of a cosmetic agent must be relatively low by definition, so that the agent avoids being subject to the regulatory scheme for pharmaceutical actives. Nonetheless, the impact must be perceptible to the consumer and preferably empirically confirmable by scientific methods. Gene transcription/expression profiles for cosmetic conditions are generally diffuse, comprising many genes with low to moderate fold differentials. Cosmetic agents, therefore, provide more diverse and less acute effects on cellular phenotype and generate the sort of associations expressly taught by Lamb as unsuitable for generating connectivity maps useful for confident hypothesis testing.

Contrary to the teachings of Lamb and the prior art in general, the present inventors surprisingly discovered that useful connectivity maps could be developed from cosmetic active—cellular phenotype—gene expression data associations in particular with respect to skin care cosmetics. Specifically, certain aspects of the present invention are based on the surprising discovery that selection of human dermal fibroblast cells as the relevant cell line resulted in construction of connectivity maps useful for hypothesis generating and testing relating to cosmetic agents in treatment of photo-damaged/photo-aged skin, while a combination of fibroblast and keratinocyte cells appeared most suitable for intrinsically aged skin. Skin is a highly complex system, and the effects of aging conditions, whether intrinsic or photo-induced, on skin are diffuse and not fully understood. Therefore, it could not be predicted that a fibroblast cell or a keratinocyte cell, or any combination thereof, could be used to construct a connectivity map effective for generating and testing hypotheses relating to cosmetic actives and genes associated with skin aging.

Skin is a complex, multi-layered and dynamic system that provides a protective covering defining the interactive boundary between an organism and the environment. It is the largest organ of the body and is vitally important to both our health and our self image. As shown in FIG. 1, skin comprises three principal layers, the epidermis, the dermis, and a layer of subcutaneous fat. The majority of cells in the epidermis are keratinocytes that produce a family of proteins called keratins. Keratins contribute to the strength of the epidermis. The epidermis itself may be divided into multiple layers with the outermost layer referred to as the stratum corneum, and the innermost layer referred to as the basal layer. All epidermal cells originate from the basal layer and undergo a process known as differentiation as they gradually displace outward to the stratum corneum, where they fuse into squamous sheets and are eventually shed. In healthy, normal skin, the rate of production equals the rate of shedding (desquamation).

The differentiating epidermal cells form distinct though naturally blended layers. As the cells displace outward, they flatten and become joined by spiny processes forming the stratum spinosum, or spinous layer. The cells manufacture specialized fats called sphingolipids, and begin to express keratins associated with terminal differentiation. As keratin is produced, it is incorporated into the cellular matrix, strengthening the skin and providing structural support to the outer layers. As the cells migrate further outward and develop characteristic granules that contain proteins which contribute to the aggregation of keratins; they now form part of the granular layer. Cells lose their nuclei in the outer part of this layer, and the granules release their contents contributing to cornification. Vesicles containing lipids discharge into the spaces between the cells, creating a barrier structure that has been suggested to function like bricks (cells) and mortar (lipids). As the cells rise into the outermost layer of the epidermis—the stratum corneum, sometimes called the horny layer or the cornified layer—they take the form of flattened discs, tightly packed together. These flattened cells, called corneocytes, are effectively dead. The lipids of the epidermis play an important role in maintaining skin health, as they help the stratum corneum to regulate water loss while providing a virtually impermeable hydrophobic barrier to the environment. Fully mature keratinocytes function to protect the skin from UV light damage, and help effectuate immune response to environmental stimuli.

The dermis, which lies just beneath the epidermis, is composed largely of the protein collagen. Most of the collagen is organized in bundles which run horizontally through the dermis and which are buried in a jelly-like material called the ground substance. Collagen accounts for up to 75% of the weight of the dermis, and is responsible for the resilience and elasticity of skin. The collagen bundles are held together by elastic fibers running through the dermis. The fibers are comprised of a protein called elastin, and make up less than 5% of the weight of the dermis. Fibroblasts function to synthesize collagen and the dermis ground substance, including components glycoproteins and glycosaminoglycans such as hyaluronic acid (which is able to bind water). The junction between the epidermis and the dermis is not straight but undulates—more markedly so in some areas of the body than others. A series of finger-like structures called rete pegs project up from the dermis, and similar structures project down from the epidermis. These projections increase the area of contact between the layers of skin, and help to prevent the epidermis from being sheared off. As skin ages, the projections get smaller and flatter. Networks of tiny blood vessels run through the rete pegs, bringing nutrients, vitamins and oxygen to the epidermis, although the epidermis itself is avascular and nourished by diffusion from the rete pegs. The dermis also contains the pilobaceous units comprising hair follicles and sebaceous glands, apocrine and eccrine sweat glands, lymphatic vessels, nerves, and various sensory structures, including the mechano-sensing Pacinian and Meissner's corpuscles.

Beneath the dermis lies the hypodermis, which comprises subcutaneous fat that cushions the dermis from underlying tissues such as muscle and bones. The fat is contained in adipose cells embedded in a connective tissue matrix. This layer may also house the hair follicles when they are in the growing phase.

Thus, skin is a multilayered complex organ comprising a wide variety of cellular types and structures, including epidermal and dermal connective tissue with blood and lymphatic vessels, the pilosebaceous units, glands, nerves, various sensory structures, the hypodermal adipose tissue, and the elastic fascia beneath the hypodermis. In turn, these structures are composed of a number of different cell types including keratinocytes, melanocytes, neuroendocrine Merkel cells, sebocytes, fibroblasts, endothelial cells, pericytes, neurons, adipocytes, myocytes and resident immunocytes including Langerhans cells, other dendritic cells, T cells and mast cells. Two of the main cell lineages in the skin are epithelial cells, which in general form the linings of the body and the parenchyma of many organs and glands, and mesenchymal cells, which form connective tissue, blood vessels and muscle. Dermal fibroblasts are mesenchymal cells, and keratinocytes are epithelial cells, which comprise most of the structure of the epidermis.

Skin aging is likewise a complex multi-factorial process that results from unrepaired cellular and tissue damage leading to impaired functional capacity. The aging process in skin is the result of both intrinsic and extrinsic factors occurring over decades. Skin is subject to many of the same intrinsic aging processes as other organs, but is also exposed to solar radiation, an important extrinsic factor that contributes to premature skin aging or photo-aging. Another important extrinsic factor potentially contributing to skin aging is smoking. There have been major advances in the understanding of the aging process with the identification of cellular pathways and genes associated with longevity and aging. Several theories have been proposed to explain intrinsic aging, including cellular senescence resulting from telomere shortening, mutations in nuclear and mitochondrial DNA, hormonal insufficiency and oxidative stress. Reactive active oxygen species and the direct effects of ultraviolet radiation (UVR) appear to play major roles in photo-aging. As is the case for aging in general, an integrated understanding of skin aging has not been developed.

Skin researchers have categorized age-inducing factors as either intrinsic or extrinsic, although these are interdependent, reflected for example by the fact that extrinsic factors may accelerate intrinsic aging. One example of the complex interplay of factors involves free radicals, which are both generated internally through normal metabolic processes and produced as a consequence of external factors, including UVR exposure. As a result of the age-associated decline in protective internal antioxidant mechanisms, free radicals can reach higher and sustained levels in cells and alter both proteins and DNA in skin. Levels of altered protein and DNA may accumulate causing damage. In addition, ongoing accumulation of damage secondary to internally-generated free radicals combined with those generated from UVR and other external assaults (surfactants, allergens, and other irritants) can promote a chronic inflammatory state. This chronic inflammation compromises skin health and may accelerate the aging process; for example, proteolytic enzymes are produced, resulting in collagen degradation. Activated inflammatory cells resulting from elevations in circulating pro-inflammatory mediators (e.g., prostaglandins, cytokines, histamines) produce reactive oxygen species that can cause oxidative damage to nucleic acids, cellular proteins and lipids. Accumulated damage caused by reactive oxygen species may stimulate a host of cytokine cascades that results in photo-aging and photo-carcinogenesis. These changes may be tied to the appearance of aging skin.

Other changes resulting from the complex interplay between intrinsic and extrinsic factors that may impact the appearance of fine lines, wrinkles and texture include the following:

• • Epidermal thickness and cellular turnover rate of both the epidermis and the stratum corneum declines and epidermal differentiation is altered.
  • The dermis becomes thinner as major structural molecules including collagen, elastin and glycosaminoglycans decrease in amount. The elastic network in photo-damaged skin becomes disorganized and aggregated and the various structural proteins may be modified by glycation. Metalloproteinase activity increases in photo-damaged skin, contributing to the degradation of collagen and elastin.
  • Convolution of dermal-epidermal junction (rete ridges) flattens with age, resulting in a loss of mechanical strength. This also leads to decreased microcirculation to the upper dermis and, thus, decreased nourishment to the epidermis.
  • Age-related changes in inter- and intra-cellular signaling lead to decreases in collagen synthesis.
  • Changes in hyaluronic acid content within the skin occur with age. Hyaluronic acid is a natural moisturizer within the skin, binding up to 1000 times its weight in water. Age-related declines result in compromised moisturization and firmness.
  • Decreased intracellular energy sources including ATP and NADH lead to an inability of skin cells to sustain youthful skin biochemistry, thereby reducing the skin's ability to maintain and restore youthful skin structure.
  • In the epidermis, a lack of estrogen slows the activity of the basal keratinocytes, and consequently leads to epidermal atrophy. This atrophic, fragile skin is less well protected by the normal surface film of lipids, because of the slow decline in sebum secretion experienced by everyone as they age. The stratum corneum barrier is less effective, and the skin may develop reactions to irritants, particularly if skin care has been inadequate or too aggressive.
  • The time necessary to repair the stratum corneum barrier increases considerably with age: the replacement of skin cells takes about twice as long for people over 75 as for those around 30.

These changes may compromise skin's elasticity, firmness and structure—contributing to areas of collapse and irregularity and ultimately manifesting as fine lines, wrinkles and texture problems.

There are many skin care products available to consumers which are directed to improving the health and/or physical appearance of skin tissue, such as keratinous tissue. Many of such products are directed to delaying, minimizing or even eliminating changes typically associated with one or both of aging and environmental damage to skin. Such products may include one or more of the numerous cosmetic agents known to be useful in improving the health and/or appearance of keratinous tissue. Although many such agents are known, an ongoing need exists to identify cosmetic agents that can provide new or improved benefits to skin tissue. There is also a need to identify additional cosmetic agents that provide similar or improved benefits as compared to existing products but which are easier to formulate, produce, and/or market.

Successful identification of anti-aging cosmetic agents has proven to be difficult due to the multi-cellular, multi-factorial processes that occur in skin over the course of decades. In addition, many desirable cosmetic agents may comprise a mixture of compounds with effects and interactions that may not be fully understood. This is often the case with a botanical or other natural extract that may affect many cellular/pathways. An additional challenge for cosmetic formulators is that cosmetics must be very safe and adverse effects generally are not acceptable. Further, while much is known about skin aging, there is much that is still poorly understood or unknown. Conventional in vitro studies of biological responses to potential cosmetic agents involve testing hundreds or thousands of potential agents in various cell types before an agent that gives the desired result can be identified and moved into a next stage of testing. However, such studies can be hindered by the complex or weakly detectable responses typically induced and/or caused by cosmetic agents. Such weak responses arise, in part, due to the great number of genes and gene products involved, and cosmetic agents may affect multiple genes in multiple ways. Moreover, the degree of bioactivity of cosmetic agents may differ for each gene and be difficult to quantify.

For example, niacinamide is a well-known cosmetic agent producing skin benefits such as improved barrier function and anti-inflammatory activity. Niacinamide is a precursor of NADH, which is involved in more than 100 reactions in cellular metabolism. In contrast to drug agents, which are selected for specificity and which are intended to have measurable effects on structure and function of the body, cosmetic agents are selected for effect on appearance and may not effect structure and function of the body to a measurable degree. Cosmetic agents tend to be non-specific with respect to effect on cellular phenotype, and administration to the body is generally limited to application on or close to the body surface.

The value of a connectivity map approach to discover functional connections among cosmetic phenotypes such as aged skin, gene expression perturbation, and cosmetic agent action is counter-indicated by the progenors of the drug-based C-map. The relevant phenotypes are very complex, the genetic perturbations are numerous and weak, and cosmetic agent action is likewise diffuse and by definition, relatively weak. It is unclear whether statistically valid data may be generated from cosmetic C-maps and it is further unclear whether a cell line exists which may provide salient or detectable cosmetic data.

Surprisingly, the present inventors have provided a C-map approach that is generalizable and biologically relevant for identification of potential cosmetic actives, and demonstrate that the C-map concept is viable by use of benchmark cosmetic actives to query the reference data and by identification of new cosmetic actives.

SUMMARY

Accordingly, the present invention provides novel methods and systems useful for generating potential new actives for the treatment of aged and photo-damaged skin. In particular, by careful selection of cell type, and by generation of a reference collection of gene-expression profiles for known cosmetic actives, the present inventors were surprisingly able to create a connectivity map useful for testing and generating hypotheses about cosmetic actives and cosmetic conditions. The present investigators confirmed the validity of connectivity mapping as a tool for identifying cosmetic agents efficacious in the treatment of aged and photo-damaged skin. Potentially efficacious cosmetic agents were identified using gene expression signatures derived from multi-cellular, full thickness human skin biopsies that were compared to short term in vitro experiments of simple cell culture systems. Based on the counter-intuitive and surprising discovery that a fibroblast cell line appears particularly predicative for use with some skin aging gene expression signatures (e.g., a photo-aging gene expression signature), while a combination of fibroblast and keratinocyte cell lines appears more predictive for other skin aging conditions (e.g., intrinsically aged skin), the invention provides methods and systems uniquely suited for desired treatment targets.

The present inventors discovered that it is possible to derive unique photo-aging and intrinsic aging gene expression signatures for use in a connectivity map, particularly where the photo-aging gene expression signature is derived from a full thickness biopsy of skin that was chronologically or intrinsically aged. The present inventors have also surprisingly discovered methods that utilize a plurality of unique skin aging gene expression signatures in a connectivity map to identify useful cosmetic skin care agents.

The present invention provides embodiments which broadly include methods and systems for determining relationships between a skin aging tissue condition of interest and one or more cosmetic agents, one or more genes associated with the skin aging tissue condition, and one or more cells associated with the skin aging tissue condition. Such methods may be used to identify cosmetic agents without detailed knowledge of the mechanisms of biological processes associated with a skin aging condition of interest, all of the genes associated with such a condition, or the cell types associated with such a condition.

According to one embodiment of the invention, a method for constructing a data architecture for use in identifying connections between perturbagens and genes associated with one or more skin aging conditions is provided. The method comprises: (a) providing a gene expression profile for a control human dermal fibroblast cell; (b) generating a gene expression profile for a human dermal fibroblast cell exposed to at least one perturbagen; (c) identifying genes differentially expressed in response to the at least one perturbagen by comparing the gene expression profiles of (a) and (b); (d) creating an ordered list comprising identifiers representing the differentially expressed genes, wherein the identifiers are ordered according to the differential expression of the genes; (e) storing the ordered list as a fibroblast instance on at least one computer readable medium; and (f) constructing a data architecture of stored fibroblast instances by repeating (a) through (e), wherein the at least one perturbagen of step (a) is different for each fibroblast instance. In another embodiment, the method further requires (g) providing a gene expression profile for a control human keratinocyte cell; (h) generating a gene expression profile for a human keratinocyte cell exposed to at least one perturbagen; (i) identifying genes differentially expressed in response to the at least one perturbagen by comparing the gene expression profiles of (g) and (h); (j) creating an ordered list comprising identifiers representing the differentially expressed genes, wherein the identifiers are ordered according to the differential expression of the genes identified in (i); (k) storing the ordered list created in step (j) as a keratinocyte instance on the at least one computer readable medium; and (l) constructing a data architecture of stored keratinocyte instances by repeating (g) through (k), wherein the at least one perturbagen of step (h) is different for each keratinocyte instance.

Other embodiments provide that the inventive methods are implemented to generate connections useful for identifying cosmetic agents effective for treating aged skin. According to one embodiment, the method comprises querying the data architecture with at least one skin aging gene expression signature, wherein querying comprises comparison of at least one skin aging gene expression signature to each instance in a data architecture of stored instances, wherein the skin aging gene expression signature represents genes differentially expressed in association with at least one skin aging condition. Examples of skin aging conditions include intrinsic skin aging conditions, which are generally age-dependent, and photo-aged skin conditions, which are a form of extrinsic skin aging where the skin is exposed to UV radiation.

Additional embodiments provide methods for formulating a skin care composition by identifying connections between perturbagens and genes associated with one or more skin aging conditions. The methods comprise: (a) accessing a plurality of instances stored in a data architecture, wherein each instance is associated with a perturbagen and a skin cell type and wherein each instance comprises an ordered list comprising a plurality of identifiers representing a plurality of up-regulated and a plurality of down-regulated genes; (b) accessing at least one skin aging gene expression signature stored in a data architecture, wherein the at least one skin aging gene expression signature comprises one or more lists comprising a plurality of identifiers representing a plurality of up-regulated genes and a plurality of down-regulated genes associated with a skin aging condition. According to one embodiment, a skin aging gene expression signature is selected from up-regulated genes set forth in Tables A and C, and down-regulated genes set forth inn Tables B and D. In more specific embodiments, about 25-100 percent of the up-regulated genes are selected from the genes set forth in Tables A or C, and about 25-100 percent of the down-regulated genes are selected from the genes set forth in Tables B and D; (c) comparing the at least one skin aging gene expression signature to the plurality of the instances, wherein the comparison comprises comparing each identifier in the gene expression signature list(s) with the position of the same identifier in the ordered lists for each of the plurality of instances; (d) assigning a connectivity score to each of the plurality of instances, wherein the connectivity score of each instance represents how strongly the gene expression profile represented by the ordered list of an instance correlates to the skin aging gene expression signature; and (e) formulating a skin care composition comprising a dermatologically acceptable carrier and at least one perturbagen, wherein the connectivity score of the instance associated with the at least one perturbagen has a negative correlation or connectivity score.

In other embodiments, methods for generating a gene expression signature for use in identifying connections between perturbagens and genes associated with one or more skin aging conditions are provided. In a specific embodiment, the method comprises: (a) providing a gene expression profile for a reference sample of human skin cells; (b) generating a gene expression profile for at least one sample of human skin cells from a subject exhibiting at least one skin aging condition, (c) comparing the expression profiles of (a) and (b) to determine a gene expression signature comprising a set of genes differentially expressed in (a) and (b); (d) assigning an identifier to each gene constituting the gene expression signature and ordering the identifiers according to magnitude of differential expression to create one or more gene expression signature lists; and (e) storing the gene expression signature list(s) in a data architecture.

Systems useful for identifying connections between perturbagens and genes associated with one or more skin aging conditions are also provided. According to one embodiment, the system comprises (a) a data architecture having stored therein a plurality of instances, and at least one skin aging gene expression signature, wherein the instances and the gene expression signature are derived from a human dermal fibroblast cell, wherein each instance comprises an instance list of rank-ordered identifiers of differentially expressed genes, and wherein the at least one skin aging gene expression signature comprises one or more gene expression signature lists of identifiers representing differentially expressed genes associated with a skin aging condition; (b) a programmable computer comprising computer-readable instructions that cause the programmable computer to execute one or more of the following: (i) accessing the plurality of instances and the at least one skin aging gene expression signature stored in the data architecture; (ii) comparing the at least one skin aging gene expression signature to the plurality of the instances, wherein the comparison comprises comparing the each identifier in the gene expression signature list(s) with the position of the same identifier in the instance list for each of the plurality of instances; and (iii) assigning a connectivity score to each of the plurality of instances, wherein the connectivity score of each instance represents how strongly the gene expression profile represented by the ordered list of an instance correlates to the skin aging gene expression signature. In other embodiments, the system may comprise a microarray scanner for receiving a sample comprising human dermal fibroblast cells and/or human keratinocyte cells; and a second programmable computer for transmitting gene expression data from the scanner to the first programmable computer.

In other aspects, the invention provides inventive gene expression signatures which may exist tangibly in various forms known in the art. For example, a gene expression signature may exist as a set of immobilized oligonucleotides wherein each oligonucleotide uniquely hybridizes to a nucleotide sequence identifying a region of a gene in the signature. In a specific embodiment, a gene expression signature for intrinsically aged skin consisting of genes selected from the genes set forth in Table E is provided. In another specific embodiment, a gene expression signature for photoaged skin consisting of genes selected from the genes set forth in Table F is provided. It is understood that the “genes set forth” in a table refers to gene identifiers designating the genes, and that a genetic signature as set forth herein is set forth according to a gene identifier.

According to yet another embodiment, a computer readable medium comprising the inventive data architecture is provided. The computer readable medium comprises a first digital file stored in a spreadsheet file format, a word processing file format, or a database file format suitable to be read by a respective spreadsheet, word processing, or database computer program, the first digital file comprising data arranged to provide one or more gene expression signature lists comprising a plurality of identifiers when read by the respective spreadsheet, word processing, or database computer program; and wherein each identifier is selected from the group consisting of a microarray probe set ID, a human gene name, a human gene symbol, and combinations thereof representing a gene set forth in Table E and wherein the gene expression signature list comprises between about 50 and about 600 identifiers.

These and additional objects, embodiments, and aspects of the invention will become apparent by reference to the Figures and Detailed Description below.

BRIEF DESCRIPTION OF THE FIGURES

FIG. 1 is a schematic illustration of the epidermal, dermal, and subcutaneous skin layers;

FIG. 2 is a schematic illustration of a computer system suitable for use with the present invention;

FIG. 3 is a schematic illustration of an instance associated with a computer readable medium of the computer system of FIG. 2;

FIG. 4 is a schematic illustration of a programmable computer suitable for use with the present invention;

FIG. 5 is a schematic illustration of an exemplary system for generating an instance;

FIG. 6 is a schematic illustration of a comparison between a gene expression signature and an instance, wherein there is a positive correlation between the lists;

FIG. 7 is a schematic illustration of a comparison between a gene expression signature and an instance, wherein there is a negative correlation between the lists; and

FIG. 8 is a schematic illustration of a comparison between a gene expression signature and an instance, wherein there is a neutral correlation between the lists.

FIG. 9 illustrates morphology of A: telomerized-keratinocytes treated with 0.1% DMSO for 6 hr (vehicle control); B: telomerized-keratinocytes treated with 10 μM all-trans-retinoic acid (tRA) in DMSO for 6 hr; C: BJ fibroblasts treated with 0.1% DMSO for 6 hr (vehicle control); and D: BJ fibroblasts treated with 10 μM tRA in DMSO for 6 hr; wherein the control keratinocytes are observed to have a cobblestone-like morphology, the control fibroblasts are observed to have a spindly morphology; and wherein treatment with 10 μM tRA caused dramatic morphologic changes in the keratinocytes (B compared to A) but had no apparent morphologic effect on the fibroblasts (D compared to C).

DETAILED DESCRIPTION

The present invention will now be described with occasional reference to the specific embodiments of the invention. This invention may, however, be embodied in different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and to fully convey the scope of the invention to those skilled in the art.

Tables A through J are submitted herewith as .txt files in accordance with large tabled data submission guidelines. Tables A through J in the .txt file format are incorporated herein, in their entirety, by this reference. Table A is a table of 300 identifiers for genes up-regulated in response to skin intrinsic aging conditions. Table B is a table of 300 identifiers for genes down-regulated in response to skin intrinsic aging conditions. Table C is a table of 300 identifiers for genes up-regulated in response to skin photo-aging conditions. Table D is a table of 300 identifiers for genes down-regulated in response to skin photo-aging conditions. Table E reflects an exemplary instance according to certain embodiments of the invention and includes a rank-ordered list of identifiers (chip probe set ID) for an illustrative number of high, middle and low ranked genes profiled by the 22,214 probe sets on Affymetrix chip model HG-U133A2.0 in response to exposure of human dermal fibroblast cells to the perturbagen Cynara scolymus (artichoke) leaf extract (instance CMP_—62_—13). Table F reflects an exemplary instance according to certain embodiments of the invention and includes a rank-ordered list of the identifiers (chip probe set ID) for an illustrative number of high, middle and low ranked genes profiled by the 22,214 probe sets on Affymetrix chip model HG-U133A2.0 in response to exposure of human keratinocyte cells to the perturbagen artichoke leaf extract (instance CMP_—62_—13). Table G reflects an exemplary instance according to certain embodiments of the invention and includes a rank-ordered list of the identifiers (chip probe et ID) for an illustrative number of high, middle and low ranked genes profiled by the 22,214 probe sets on Affymetrix chip model HG-U133A2.0 in response to exposure of human dermal fibroblast cells to the perturbagen carob leaf extract (instance CMP_—62_—24). Table H reflects an exemplary instance according to certain embodiments of the invention and includes a rank-ordered list of the identifiers (chip probe set ID) for an illustrative number of high, middle and low ranked genes profiled by the 22,214 probe sets on Affymetrix chip model HG-U133A2.0 in response to exposure of human keratinocyte cells to the perturbagen carob leaf extract (instance CMP_—61_—23). Table I is, a list of identifiers for up-regulated Signature 3, a set of genes up-regulated under photo-aging conditions, optimized for fibroblasts. Table J is a list of identifiers for down-regulated Signature 3, a set of genes down-regulated under photo-aging conditions, optimized for fibroblasts.

Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in the description of the invention herein is for describing particular embodiments only and is not intended to be limiting of the invention. As used in the description of the invention and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise.

As used interchangeably herein, the terms “connectivity map” and “C-map” refer broadly to devices, systems, articles of manufacture, and methodologies for identifying relationships between cellular phenotypes or cosmetic conditions, gene expression, and perturbagens, such as cosmetic actives.

As used herein, the term “cosmetic agent” means any substance, as well any component thereof, intended to be rubbed, poured, sprinkled, sprayed, introduced into, or otherwise applied to a mammalian body or any part thereof. Cosmetic agents may include substances that are Generally Recognized as Safe (GRAS) by the US Food and Drug Administration, food additives, and materials used in non-cosmetic consumer products including over-the-counter medications. In some embodiments, cosmetic agents may be incorporated in a cosmetic composition comprising a dermatologically acceptable carrier suitable for topical application to skin. A cosmetic agent includes, but is not limited to, (i) chemicals, compounds, small or large molecules, extracts, formulations, or combinations thereof that are known to induce or cause at least one effect (positive or negative) on skin tissue; (ii) chemicals, compounds, small molecules, extracts, formulations, or combinations thereof that are known to induce or cause at least one effect (positive or negative) on skin tissue and are discovered, using the provided methods and systems, to induce or cause at least one previously unknown effect (positive or negative) on the skin tissue; and (iii) chemicals, compounds, small molecules, extracts, formulations, or combinations thereof that are not known have an effect on skin tissue and are discovered, using the provided methods and systems, to induce or cause an effect on skin tissue.

Some examples of cosmetic agents or cosmetically actionable materials can be found in: the PubChem database associated with the National Institutes of Health, USA (http://pubchem.ncbi.nlm.nih.gov); the Ingredient Database of the Personal Care Products Council (http://online.personalcarecouncil.org/jsp/Home.jsp); and the 2010 International Cosmetic Ingredient Dictionary and Handbook, 13^th Edition, published by The Personal Care Products Council; the EU Cosmetic Ingredients and Substances list; the Japan Cosmetic Ingredients List; the Personal Care Products Council, the SkinDeep database (URL: http://www.cosmeticsdatabase.com); the FDA Approved Excipients List; the FDA OTC List; the Japan Quasi Drug List; the US FDA Everything Added to Food database; EU Food Additive list; Japan Existing Food Additives, Flavor GRAS list; US FDA Select Committee on GRAS Substances; US Household Products Database; the Global New Products Database (GNPD) Personal Care, Health Care, Food/Drink/Pet and Household database (URL: http://www.gnpd.com); and from suppliers of cosmetic ingredients and botanicals.

Other non-limiting examples of cosmetic agents include botanicals (which may be derived from one or more of a root, stem bark, leaf, seed or fruit of a plant). Some botanicals may be extracted from a plant biomass (e.g., root, stem, bark, leaf, etc.) using one more solvents. Botanicals may comprise a complex mixture of compounds and lack a distinct active ingredient. Another category of cosmetic agents are vitamin compounds and derivatives and combinations thereof, such as a vitamin B3 compound, a vitamin B5 compound, a vitamin B6 compound, a vitamin B9 compound, a vitamin A compound, a vitamin C compound, a vitamin E compound, and derivatives and combinations thereof (e.g., retinol, retinyl esters, niacinamide, folic acid, panthenol, ascorbic acid, tocopherol, and tocopherol acetate). Other non-limiting examples of cosmetic agents include sugar amines, phytosterols, hexamidine, hydroxy acids, ceramides, amino acids, and polyols.

The terms “gene expression signature,” and “gene-expression signature” refer to a rationally derived list, or plurality of lists, of genes representative of a skin tissue condition or a skin agent. In specific contexts, the skin agent may be a benchmark skin agent or a potential skin agent. Thus, the gene expression signature may serve as a proxy for a phenotype of interest for skin tissue. A gene expression signature may comprise genes whose expression, relative to a normal or control state, is increased (up-regulated), whose expression is decreased (down-regulated), and combinations thereof. Generally, a gene expression signature for a modified cellular phenotype may be described as a set of genes differentially expressed in the modified cellular phenotype over the cellular phenotype. A gene expression signature can be derived from various sources of data, including but not limited to, from in vitro testing, in vivo testing and combinations thereof. In some embodiments, a gene expression signature may comprise a first list representative of a plurality of up-regulated genes of the condition of interest and a second list representative of a plurality of down-regulated genes of the condition of interest.

As used herein, the term “benchmark skin agent” refers to any chemical, compound, small or large molecule, extract, formulation, or combinations thereof that is known to induce or cause a superior effect (positive or negative) on skin tissue. Non-limiting examples of benchmark skin agents include niacinamide, trans-retinoic acid, and hexamidine.

As used herein, the term “query” refers to data that is used as an input to a Connectivity Map and against which a plurality of instances are compared. A query may include a gene expression signature associated with one or both of a skin aging condition and a benchmark skin agent.

The term “instance,” as used herein, refers to data from a gene expression profiling experiment in which skin cells are dosed with a perturbagen. In some embodiments, the data comprises a list of identifiers representing the genes that are part of the gene expression profiling experiment. The identifiers may include gene names, gene symbols, microarray probe set IDs, or any other identifier. In some embodiments, an instance may comprise data from a microarray experiment and comprises a list of probe set IDs of the microarray ordered by their extent of differential expression relative to a control. The data may also comprise metadata, including but not limited to data relating to one or more of the perturbagen, the gene expression profiling test conditions, the skin cells, and the microarray.

The term “keratinous tissue,” as used herein, refers to keratin-containing layers disposed as the outermost protective covering of mammals which includes, but is not limited to, skin, hair, nails, cuticles, horns, claws, beaks, and hooves. With respect to skin, the term refers to one or all of the dermal, hypodermal, and epidermal layers, which includes, in part, keratinous tissue.

As used herein, the term “skin aging” refers to a human skin tissue condition resulting from the expression or repression of genes, environmental factors (e.g., sun exposure, UVA and/or UVB exposure, smoking), intrinsic factors (e.g. endogenous free radical production or cellular senescence) or interactions there between that produces one or more of fine lines and/or wrinkles, dry skin, inflamed skin, rough skin, sallow skin, telangectasia, sagging skin, enlarged pores, and combinations thereof.

The term “perturbagen,” as used herein, means anything used as a challenge in a gene expression profiling experiment to generate gene expression data for use in the present invention. In some embodiments, the perturbagen is applied to fibroblast and/or keratinocyte cells and the gene expression data derived from the gene expression profiling experiment may be stored as an instance in a data architecture. Any substance, chemical, compound, active, natural product, extract, drug [e.g. Sigma-Aldrich LOPAC (Library of Pharmacologically Active Compounds) collection], small molecule, and combinations thereof used as to generate gene expression data can be a perturbagen. A perturbagen can also be any other stimulus used to generate differential gene expression data. For example, a perturbagen may also be UV radiation, heat, osmotic stress, pH, a microbe, a virus, and small interfering RNA. A perturbagen may be, but is not required to be, any cosmetic agent.

The term “dermatologically acceptable,” as used herein, means that the compositions or components described are suitable for use in contact with human skin tissue.

As used herein, the term “computer readable medium” refers to any electronic storage medium and includes but is not limited to any volatile, nonvolatile, removable, and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data and data structures, digital files, software programs and applications, or other digital information. Computer readable media includes, but are not limited to, application-specific integrated circuit (ASIC), a compact disk (CD), a digital versatile disk (DVD), a random access memory (RAM), a synchronous RAM (SRAM), a dynamic RAM (DRAM), a synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), a direct RAM bus RAM (DRRAM), a read only memory (ROM), a programmable read only memory (PROM), an electronically erasable programmable read only memory (EEPROM), a disk, a carrier wave, and a memory stick. Examples of volatile memory include, but are not limited to, random access memory (RAM), synchronous RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), and direct RAM bus RAM (DRRAM). Examples of non-volatile memory include, but are not limited to, read only memory (ROM), programmable read only memory (PROM), erasable programmable read only memory (EPROM), and electrically erasable programmable read only memory (EEPROM). A memory can store processes and/or data. Still other computer readable media include any suitable disk media, including but not limited to, magnetic disk drives, floppy disk drives, tape drives, Zip drives, flash memory cards, memory sticks, compact disk ROM (CD-ROM), CD recordable drive (CD-R drive), CD rewriteable drive (CD-RW drive), and digital versatile ROM drive (DVD ROM).

As used herein, the terms “software” and “software application” refer to one or more computer readable and/or executable instructions that cause a computing device or other electronic device to perform functions, actions, and/or behave in a desired manner. The instructions may be embodied in one or more various forms like routines, algorithms, modules, libraries, methods, and/or programs. Software may be implemented in a variety of executable and/or loadable forms and can be located in one computer component and/or distributed between two or more communicating, co-operating, and/or parallel processing computer components and thus can be loaded and/or executed in serial, parallel, and other manners. Software can be stored on one or more computer readable medium and may implement, in whole or part, the methods and functionalities of the present invention.

As used herein, the term “intrinsic aging gene expression signature” refers to a gene expression signature derived from gene expression profiling of an intrinsic aging skin condition.

As used herein, the term “intrinsic aging skin condition” refers to a skin aging condition that derives, in whole or part, from chronological aging of the skin.

As used herein, the term “photo-aging skin condition” refers to a skin aging condition that derives, in whole or part, from exposure to sunlight and/or ultraviolet light (e.g., UVR, UVA, UVB, and/or UVC).

As used herein, the term “photo-aging gene expression signature” refers to a gene expression signature derived from gene expression profiling of a photo-aging skin condition.

As used herein, the term “connectivity score” refers to a derived value representing the degree to which an instance correlates to a query.

As used herein, the term “data architecture” refers generally to one or more digital data structures comprising an organized collection of data. In some embodiments, the digital data structures can be stored as a digital file (e.g., a spreadsheet file, a text file, a word processing file, a database file, etc.) on a computer readable medium. In some embodiments, the data architecture is provided in the form of a database that may be managed by a database management system (DBMS) that is be used to access, organize, and select data (e.g., instances and gene expression signatures) stored in a database.

As used herein, the terms “gene expression profiling” and “gene expression profiling experiment” refer to the measurement of the expression of multiple genes in a biological sample using any suitable profiling technology. For example, the mRNA expression of thousands of genes may be determined using microarray techniques. Other emerging technologies that may be used include RNA-Seq or whole transcriptome sequencing using NextGen sequencing techniques.

As used herein, the term “microarray” refers broadly to any ordered array of nucleic acids, oligonucleotides, proteins, small molecules, large molecules, and/or combinations thereof on a substrate that enables gene expression profiling of a biological sample. Non-limiting examples of microarrays are available from Affymetrix, Inc.; Agilent Technologies, Inc.; Ilumina (two “1”'s?), Inc.; GE Healthcare, Inc.; Applied Biosystems, Inc.; Beckman Coulter, Inc.; etc.

Unless otherwise indicated, all numbers expressing quantities of ingredients, properties such as molecular weight, reaction conditions, and so forth as used in the specification and claims are to be understood as being modified in all instances by the term “about”. Additionally, the disclosure of any ranges in the specification and claims are to be understood as including the range itself and also anything subsumed therein, as well as endpoints. All numeric ranges are inclusive of narrower ranges; delineated upper and lower range limits are interchangeable to create further ranges not explicitly delineated. Unless otherwise indicated, the numerical properties set forth in the specification and claims are approximations that may vary depending on the desired properties sought to be obtained in embodiments of the present invention, Notwithstanding that numerical ranges and parameters setting forth the broad scope of the invention are approximations, the numerical values set forth in the specific examples are reported as precisely as possible. Any numerical values, however, inherently contain certain errors necessarily resulting from error found in their respective measurements.

In accordance with one aspect of the present invention, provided are devices, systems and methods for implementing a connectivity map utilizing one or more query signatures associated with a skin aging condition. The query signatures may be derived in variety of ways. In some embodiments, the query signatures may be gene expression signatures derived from gene expression profiling of full thickness skin biopsies of skin exhibiting a skin aging condition of interest compared to a control (e.g., photo-damaged skin compared to sun protected skin). The gene expression profiling can be carried out using any suitable technology, including but not limited to microarray analysis or NextGen sequencing. Some examples of gene expression signatures include a photo-aging gene expression signature and an intrinsic aging gene expression signature, examples of which are described more fully hereafter. In other embodiments, the query signature may be derived from benchmark agents known to positively affect, treat, or reverse a skin aging condition, wherein the query signature is, derived from a fibroblast and/or keratinocyte cell line exposed to the benchmark skin agent. These query signatures may be used singularly or in combination.

In accordance with another aspect of the present invention, provided are devices, systems, and methods for implementing a connectivity map utilizing one or more instances derived from a perturbagen, such as a cosmetic agent, exposed to a fibroblast (e.g., BJ fibroblasts) and/or keratinocyte (e.g., telomerized human keratinocytes) skin cell line. Instances from more complex cell culture systems may also be used, such as skin organotypic cultures containing both keratinocytes and fibroblasts or ex vivo human skin. Instances from a plurality of cell lines may be used with the present invention.

In accordance with yet another aspect of the present invention, provided are devices, systems and methods for identification of relationships between a skin aging query signature and a plurality of instances. For example, it may be possible to ascertain perturbagens that give rise to a statistically significant activity on a statistically significant number of genes associated with a skin aging tissue condition of interest, leading to the identification of new cosmetic agents for treating a skin condition or new uses of known cosmetic agents.

I. Systems and Devices

Referring to FIGS. 2, 4 and 5, some examples of systems and devices in accordance with the present invention for use in identifying relationships between perturbagens, skin aging tissue conditions, and genes associated with the skin aging tissue condition will now be described. System 10 comprises one or more of computing devices 12, 14, a computer readable medium 16 associated with the computing device 12, and communication network 18.

The computer readable medium 16, which may be provided as a hard disk drive, comprises a digital file 20, such as a database file, comprising a plurality of instances 22, 24, and 26 stored in a data structure associated with the digital file 20. The plurality of instances may be stored in relational tables and indexes or in other types of computer readable media. The instances 22, 24, and 26 may also be distributed across a plurality of digital files, a single digital file 20 being described herein however for simplicity.

The digital file 20 can be provided in wide variety of formats, including but not limited to a word processing file format (e.g., Microsoft Word), a spreadsheet file format (e.g., Microsoft Excel), and a database file format. Some common examples of suitable file formats include, but are not limited to, those associated with file extensions such as *.xls, *.xld, *.xlk, *.Xll, *.xlt, *.xlxs, *.dif, *.db, *.dbf, *.accdb, *.mdb, *.mdf, *.cdb, *.fdb, *.csv, *sql, *.xml, *.doc, *.txt, *.rtf, *.log, *.docx, *.ans, *.pages, *.wps, etc.

Referring to FIG. 3, in some embodiments the instance 22 may comprise an ordered listing of microarray probe set IDs, wherein the value of N is equal to the total number of probes on the microarray used in analysis. Common microarrays include Affymetrix GeneChips and Illumina BeadChips, both of which comprise probe sets and custom probe sets. To generate the reference gene profiles according to the invention, preferred chips are those designed for profiling the human genome. Examples of Affymetrix chips with utility in the instant invention include model Human Genome (HG)-U133 Plus 2.0. A specific Affymetrix chip employed by the instant investigators is HG-U133A2.0, however it will be understood by a person or ordinary skill in the art that any chip or microarray, regardless of proprietary origin, is suitable so long as the probe sets of the chips used to construct a data architecture according to the invention are substantially similar.

Instances derived from microarray analyses utilizing Affymetrix GeneChips may comprise an ordered listing of gene probe set IDs where the list comprises 22,000+ IDs. The ordered listing may be stored in a data structure of the digital file 20 and the data arranged so that, when the digital file is read by the software application 28, a plurality of character strings are reproduced representing the ordered listing of probe set IDs. An example of an ordered listing of probe set IDs for a perturbagen having the INCI name Cynara scolymus (artichoke leaf extract) is set forth in Table E, which specifically includes an ordered list derived from a microarray analysis of fibroblast cells dosed with Cynara scolymus. Table F is a table of rank-ordered identifiers in association with the rank as derived from a microarray analysis of keratinocyte cells dosed with Cynara scolymus. While it is preferred that each instance comprise a full list of the probe set IDs, it is contemplated that one or more of the instances may comprise less than all of the probe set IDs of a microarray. It is also contemplated that the instances may include other data in addition to or in place of the ordered listing of probe set IDs. For example, an ordered listing of equivalent gene names and/or gene symbols may be substituted for the ordered listing of probe set IDs. Additional data may be stored with an instance and/or the digital file 20. In some embodiments, the additional data is referred to as metadata and can include one or more of cell line identification, batch number, exposure duration, and other empirical data, as well as any other descriptive material associated with an instance ID. The ordered list may also comprise a numeric value associated with each identifier that represents the ranked position of that identifier in the ordered list.

Referring again to FIGS. 2, 3 and 4, the computer readable medium 16 may also have a second digital file 30 stored thereon. The second digital file 30 comprises one or more lists 32 of microarray probe set IDs associated with one or more skin aging gene expression signatures. The listing 32 of microarray probe set IDs typically comprises a much smaller list of probe set IDs than the instances of the first digital file 20. In some embodiments, the list comprises between 2 and 1000 probe set IDs. In other embodiments the list comprises greater than 10, 50, 100, 200, or 300 and/or less than about 800, 600, or about 400 probe set IDs. The listing 32 of probe set IDs of the second digital file 30 comprises a list of probe set IDs representing up, and/or down-regulated genes selected to represent a skin aging condition of interest. In some embodiments, a first list may represent the up-regulated genes and a second list may represent the down-regulated genes of the gene expression signature. The listing(s) may be stored in a data structure of the digital file 30 and the data arranged so that, when the digital file is read by the software application 28, a plurality of character strings are reproduced representing the list of probe set IDs. Instead of probe set IDs, equivalent gene names and/or gene symbols (or another nomenclature) may be substituted for a list of probe set IDs. Additional data may be stored with the gene expression signature and/or the digital file 30 and this is commonly referred to as metadata, which may include any associated information, for example, cell line or sample source, and microarray identification. Examples of listings of probe set IDs for intrinsic aging gene expression signatures are set forth in Tables A (up-regulated) and B (down-regulated), and an example of listings of probe set IDs for photo-aging gene expression signatures are set forth in Tables C (up-regulated) and D (down-regulated). In some embodiments, one or more skin aging gene expression signatures may be stored in a plurality of digital files and/or stored on a plurality of computer readable media. In other embodiments, a plurality of gene expression signatures (e.g., 32, 34) may be stored in the same digital file (e.g., 30) or stored in the same digital file or database that comprises the instances 22, 24, and 26.

As previously described, the data stored in the first and second digital files may be stored in a wide variety of data structures and/or formats. In some embodiments, the data is stored in one or more searchable databases, such as free databases, commercial databases, or a company's internal proprietary database. The database may be provided or structured according to any model known in the art, such as for example and without limitation, a flat model, a hierarchical model, a network model, a relational model, a dimensional model, or an object-oriented model. In some embodiments, at least one searchable database is a company's internal proprietary database. A user of the system 10 may use a graphical user interface associated with a database management system to access and retrieve data from the one or more databases or other data sources to which the system is operably connected. In some embodiments, the first digital file 20 is provided in the form of a first database and the second digital file 30 is provided in the form of a second database. In other embodiments, the first and second digital files may be combined and provided in the form of a single file.

In some embodiments, the first digital file 20 may include data that is transmitted across the communication network 18 from a digital file 36 stored on the computer readable medium 38. In one embodiment, the first digital file 20 may comprise gene expression data obtained from a cell line (e.g., a fibroblast cell line and/or a keratinocyte cell line) as well as data from the digital file 36, such as gene expression data from other cell lines or cell types, gene expression signatures, perturbagen information, clinical trial data, scientific literature, chemical databases, pharmaceutical databases, and other such data and metadata. The digital file 36 may be provided in the form of a database, including but not limited to Sigma-Aldrich LOPAC collection, Broad Institute C-MAP collection, GEO collection, and Chemical Abstracts Service (CAS) databases.

The computer readable medium 16 (or another computer readable media, such as 16) may also have stored thereon one or more digital files 28 comprising computer readable instructions or software for reading, writing to, or otherwise managing and/or accessing the digital files 20, 30. The computer readable medium 16 may also comprise software or computer readable and/or executable instructions that cause the computing device 12 to perform one or more steps of the methods of the present invention, including for example and without limitation, the step(s) associated with comparing a gene expression signature stored in digital file 30 to instances 22, 24, and 26 stored in digital file 20. In some embodiments, the one or more digital files 28 may form part of a database management system for managing the digital files 20, 28. Non-limiting examples of database management systems are described in U.S. Pat. Nos. 4,967,341 and 5,297,279.

The computer readable medium 16 may form part of or otherwise be connected to the computing device 12. The computing device 12 can be provided in a wide variety of forms, including but not limited to any general or special purpose computer such as a server, a desktop computer, a laptop computer, a tower computer, a microcomputer, a mini computer, and a mainframe computer. While various computing devices may be suitable for use with the present invention, a generic computing device 12 is illustrated in FIG. 4. The computing device 12 may comprise one or more components selected from a processor 40, system memory 42, and a system bus 44. The system bus 44 provides an interface for system components including but not limited to the system memory 42 and processor 40. The system bus 36 can be any of several types of bus structures that may further interconnect to a memory bus (with or without a memory controller), a peripheral bus, and a local bus using any of a variety of commercially available bus architectures. Examples of a local bus include an industrial standard architecture (USA) bus, a microchannel architecture (MSA) bus, an extended ISA (EISA) bus, a peripheral component interconnect (PCI) bus, a universal serial (USB) bus, and a small computer systems interface (SCSI) bus. The processor 40 may be selected from any suitable processor, including but not limited to, dual microprocessor and other multi-processor architectures. The processor executes a set of stored instructions associated with one or more program applications or software.

The system memory 42 can include non-volatile memory 46 (e.g., read only memory (ROM), erasable programmable read only memory (EPROM), electrically erasable programmable read only memory (EEPROM), etc.) and/or volatile memory 48 (e.g., random access memory (RAM)). A basic input/output system (BIOS) can be stored in the non-volatile memory 38, and can include the basic routines that help to transfer information between elements within the computing device 12. The volatile memory 48 can also include a high-speed RAM such as static RAM for caching data.

The computing device 12 may further include a storage 44, which may comprise, for example, an internal hard disk drive [HDD, e.g., enhanced integrated drive electronics (EIDE) or serial advanced technology attachment (SATA)] for storage. The computing device 12 may further include an optical disk drive 46 (e.g., for reading a CD-ROM or DVD-ROM 48). The drives and associated computer-readable media provide non-volatile storage of data, data structures and the data architecture of the present invention, computer-executable instructions, and so forth. For the computing device 12, the drives and media accommodate the storage of any data in a suitable digital format. Although the description of computer-readable media above refers to an HDD and optical media such as a CD-ROM or DVD-ROM, it should be appreciated by those skilled in the art that other types of media which are readable by a computer, such as Zip disks, magnetic cassettes, flash memory cards, cartridges, and the like may also be used, and further, that any such media may contain computer-executable instructions for performing the methods of the present invention.

A number of software applications can be stored on the drives 44 and volatile memory 48, including an operating system and one or more software applications, which implement, in whole or part, the functionality and/or methods described herein. It is to be appreciated that the embodiments can be implemented with various commercially available operating systems or combinations of operating systems. The central processing unit 40, in conjunction with the software applications in the volatile memory 48, may serve as a control system for the computing device 12 that is configured to, or adapted to, implement the functionality described herein.

A user may be able to enter commands and information into the computing device 12 through one or more wired or wireless input devices 50, for example, a keyboard, a pointing device, such as a mouse (not illustrated), or a touch screen. These and other input devices are often connected to the central processing unit 40 through an input device interface 52 that is coupled to the system bus 44 but can be connected by other interfaces, such as a parallel port, an IEEE 1394 serial port, a game port, a universal serial bus (USB) port, an IR interface, etc. The computing device 12 may drive a separate or integral display device 54, which may also be connected to the system bus 44 via an interface, such as a video port 56.

The computing devices 12, 14 may operate in a networked environment across network 18 using a wired and/or wireless network communications interface 58. The network interface port 58 can facilitate wired and/or wireless communications. The network interface port can be part of a network interface card, network interface controller (NIC), network adapter, or LAN adapter. The communication network 18 can be a wide area network (WAN) such as the Internet, or a local area network (LAN). The communication network 18 can comprise a fiber optic network, a twisted-pair network, a T1/E1 line-based network or other links of the T-carrier/E carrier protocol, or a wireless local area or wide area network (operating through multiple protocols such as ultra-mobile band (UMB), long term evolution (LTE), etc.). Additionally, communication network 18 can comprise base stations for wireless communications, which include transceivers, associated electronic devices for modulation/demodulation, and switches and ports to connect to a backbone network for backhaul communication such as in the case of packet-switched communications.

II. Methods for Creating a Plurality of Instances

In some embodiments, the methods of the present invention may comprise populating at least the first digital file 20 with a plurality of instances (e.g., 22, 24, 26) comprising data derived from a plurality of gene expression profiling experiments, wherein one or more of the experiments comprise exposing dermal fibroblast cells and/or keratinocyte cells (or other skin cells such as human skin equivalent cultures or ex vivo cultured human skin) to at least one perturbagen. For simplicity of discussion, the gene expression profiling discussed hereafter will be in the context of a microarray experiment.

Referring to FIG. 5, one embodiment of a method of the present invention is illustrated. The method 58 comprises exposing a fibroblast cells 60 and/or keratinocyte cells 62 to a perturbagen 64. The perturbagen may be dissolved in a carrier, such as dimethyl sulfoxide (DMSO). After exposure, mRNA is extracted from the cells exposed to the perturbagen and reference cells 66 (e.g., fibroblast or keratinocyte cells) which are exposed to only the carrier. The mRNA 68, 70, 72 may be reverse transcribed to cDNA 64, 76, 78 and marked with different fluorescent dyes (e.g., red and green) if a two color microarray analysis is to be performed. Alternatively, the samples may be prepped for a one color microarray analysis as described in Example 1, and further a plurality of replicates may be processed if desired. The cDNA samples may be co-hybridized to the microarray 80 comprising a plurality of probes 82. The microarray may comprise thousands of probes 82. In some embodiments, there are between 10,000 and 50,000 gene probes 82 present on the microarray 80. The microarray is scanned by a scanner 84, which excites the dyes and measures the amount fluorescence. A computing device 86 may be used to analyze the raw images to determine the expression levels of a gene in the cells 60, 62 relative to the reference cells 66. The scanner 84 may incorporate the functionality of the computing device 86. The expression levels include: i) up-regulation [e.g., greater binding of the test material (e.g., cDNA 74, 76) to the probe than the reference material (e.g., cDNA 78)], or ii) down-regulation [e.g., greater binding of the reference material (e.g., cDNA 78) to the probe than the test material (e.g., cDNA 74, 76)], iii) expressed but not differentially [e.g., similar binding of the reference material (e.g., cDNA 78) to the probe than the test material (e.g., cDNA 74. 76)], and iv) no detectable signal or noise. The up- and down-regulated genes are referred to as differentially expressed. Microarrays and microarray analysis techniques are well known in the art, and it is contemplated that other microarray techniques may be used with the methods, devices and systems of the present invention. For example, any suitable commercial or non-commercial microarray technology and associated techniques may used. Good results have been obtained with Affymetrix GeneChip® technology and Illumina BeadChip™ technology. One illustrative technique is described in Example 1. However, one of skill in the art will appreciate that the present invention is not limited to the methodology of the example and that other methods and techniques are also contemplated to be within its scope.

In a very specific embodiment, an instance consists of the rank ordered data for all of the probe sets on the Affymetrix HG-U133A2.0 GeneChip wherein each probe on the chip has a unique probe set IDentifier. The probe sets are rank ordered by the fold change relative to the controls in the same C-map batch (single instance/average of controls). The probe set IDentifiers are rank-ordered to reflect the most up-regulated to the most down-regulated.

Notably, even for the non-differentially regulated genes the signal values for a particular probe set are unlikely to be identical for the instance and control so a fold change different from 1 will be calculated that can be used for comprehensive rank ordering. In accordance with methods disclosed by Lamb et al. (2006), data are adjusted using 2 thresholds to minimize the effects of genes that may have very low noisy signal values, which can lead to spurious large fold changes. The thresholding is preferably done before the rank ordering. An example for illustrative purposes includes a process wherein a first threshold is set at 20. If the signal for a probe set is below 20, it is adjusted to 20. Ties for ranking are broken with a second threshold wherein the fold changes are recalculated and any values less than 2 are set to 2. For any remaining ties the order depends on the specific sorting algorithm used but is essentially random. The probe sets in the middle of the list do not meaningfully contribute to an actual connectivity score.

The rank ordered data are stored as an instance. The probes may be sorted into a list according to the level of gene expression regulation detected, wherein the list progresses from up-regulated to marginal or no regulation to down-regulated, and this rank ordered listing of probe IDs is stored as an instance (e.g., 22) in the first digital file 20. Referring to FIG. 3, the data associated with an instance comprises the probe ID 80 and a value 82 representing its ranking in the list (e.g., 1, 2, 3, 4 . . . N, where N represents the total number of probes on the microarray). The ordered list 84 may generally comprise approximately three groupings of probe IDs: a first grouping 86 of probe IDs associated with up-regulated genes, a second group 88 of probe IDs associated with genes with marginal regulation or no detectable signal or noise, and a third group 90 of probe IDs associated with down-regulated genes. The most up regulated genes are at or near the top of the list 84 and the most down-regulated genes are at or near the bottom of the list 84. The groupings are shown for illustration, but the lists for each instance may be continuous and the number of regulated genes will depend on the strength of the effect of the perturbagen associated with the instance. Other arrangements within the list 84 may be provided. For example, the probe IDs associated with the down-regulated genes may be arranged at the top of the list 84. This instance data may also further comprise metadata such as perturbagen identification, perturbagen concentration, cell line or sample source, and microarray identification.

In some embodiments, one or more instances comprise at least about 1,000, 2,500, 5,000, 10,000, or 20,000 identifiers and/or less than about 30,000, 25,000, or 20,000 identifiers. In some embodiments, the database comprises at least about 50, 100, 250, 500, or 1,000 instances and/or less than about 50,000, 20,000, 15,000, 10,000, 7,500, 5,000, or 2,500 instances. Replicates of an instance may created, and the same perturbagen may be used to derive a first instance from fibroblast cells and a second instance from keratinocyte cells and a third instance from another skin cell type, such as melanocytes or complex tissue, for example ex vivo human skin.

The present inventors have surprisingly discovered that instances derived from fibroblast cells appear to be more predictive than other cell types when used in combination with a photo-aging gene expression signature. As described more fully hereafter in Example 6, the present inventors compared instances derived from BJ fibroblast cells and keratinocyte cells with a photo-aging gene expression signature and found that instances derived from the fibroblast cells were dramatically over represented in the highest ranking results (the higher the ranking, the more likely the perturbagen is to have a beneficial affect upon the photo aging condition) compared to keratinocyte cells. The inventors also discovered that the up-regulated genes of a photo-aging gene expression signature most closely correlated with instances derived from fibroblast cells. In comparison, the present inventors discovered that a combination of instances derived from fibroblast cells and keratinocyte cells using the same set of perturbagens correlated more closely with an intrinsic aging gene expression signature, with the instances derived from the fibroblast cells being slightly more preferential. The inventors were surprised to find that instances derived from keratinocyte cells were under represented in the top ranking results of the photo-aging gene expression signature, while more equal representation was observed with respect to an intrinsic aging gene expression signature. Still further, both results are surprising considering that the gene expression signatures were derived from full thickness biopsies representing complex, multi-factorial skin aging conditions and further that the photo-aging gene expression signature was derived from a full thickness skin biopsy that was also intrinsically aged. In other words, a photo-aging gene expression signature could be differentiated from biopsy samples containing both intrinsic aging and photo aging phenotypes.

III. Methods for Deriving Skin Aging Gene Expression Signatures

Some methods of the present invention comprise identifying a gene expression signature that represents the up-regulated and down-regulated genes associated with a skin aging condition of interest. A skin aging condition typically involves complex processes involving numerous known and unknown extrinsic and intrinsic factors, as well as responses to such factors that are subtle over a relatively short period of time but non-subtle over a longer period of time. This is in contrast to what is typically observed in drug screening methods, wherein a specific target, gene, or mechanism of action is of interest. Due to the unique screening challenges associated with a skin aging condition, the quality of the gene expression signature representing the condition of interest can be important for distinguishing between the gene expression data actually associated with a response to a perturbagen from the background expression data. One challenge in developing skin aging gene expression signatures is that the number of genes selected needs to be adequate to reflect the dominant and key biology but not so large as to include many genes that have achieved a level of statistical significance by random chance and are non-informative. Thus, query signatures should be carefully derived since the predictive value may be dependent upon the quality of the gene expression signature.

One factor that can impact the quality of the query signature is the number of genes included in the signature. The present inventors have found that, with respect to a cosmetic data architecture and connectivity map, too few genes can result in a signature that is unstable with regard to the highest scoring instances. In other words, small changes to the gene expression signature can result significant differences in the highest scoring instance. Conversely, too many genes may tend to partially mask the dominant biological responses and will include a higher fraction of genes meeting statistical cutoffs by random chance—thereby adding undesirable noise to the signature. The inventors have found that the number of genes desirable in a gene expression signature is also a function of the strength of the biological response associated with the condition and the number of genes needed to meet minimal values (e.g., a p-value less than about 0.05) for statistical significance. When the biology is weaker, such as is the case typically with cosmetic condition phenotypes, fewer genes than those which may meet the statistical requisite for inclusion in the prior art, may be used to avoid adding noisy genes. For example, the gene expression profiling analysis of a photo aging skin condition yielded approximately 4000 genes having a statistical p-value of less than 0.05 and approximately 1000 genes having a p-value of less than 0.001, which could be considered a very strong biological response. The gene expression profiling analysis of an intrinsic aging skin condition yielded approximately 1000 genes have a statistical p-value of less than 0.05 and approximately 400 genes have a p-value of less than 0.001, which could be considered a moderately strong biological response. In these cases, a gene expression signature comprises between about 100 and about 600 genes. Weaker biology may be better represented by a gene expression signature comprising fewer genes.

While a gene expression signature may represent all significantly regulated genes associated with skin aging condition of interest; typically it represents a subset of such genes. The present inventors have discovered that skin aging gene expression signatures comprising between about 200 and about 800 genes of approximately equal numbers of up-regulated and/or down-regulated genes are stable, reliable, and can provide predictive results. For example, a suitable gene expression signature may have from about 200-250 genes, 250-300 genes, 300-350 genes, 350-400 genes, 400-450 genes, 450-500 genes, 500-550 genes, 550-600 genes, 600-650 genes, 650-700 genes, 700-750 genes, and 750-800 genes. However, one of skill in the art will appreciate that gene expression signatures comprising fewer or more genes are also within the scope of the various embodiments of the invention. For purposes of depicting a gene expression signature, the probe set IDs associated with the genes are preferably separated into a first list comprising the most up-regulated genes and a second list comprising the most down-regulated.

Gene expression signatures may be generated from full thickness skin biopsies from skin having the skin aging condition of interest compared to a control. Two gene expression signature types for skin aging can include an intrinsic aging gene expression signature and a photo-aging gene expression signature, which may be derived by comparing gene expression data from a full thickness skin biopsy from skin having the condition of interest and a control. Examples 2 and 3 below describe in greater detail non-limiting methods for deriving these gene expression signatures. Generally, for a photo-aging gene expression signature, biopsies may be taken from sun exposed skin (e.g., extensor forearm) and sun protected skin (e.g., buttocks) of a plurality of older subjects. The subjects may vary in age, but one age range is between about 45 years of age and 70 years of age. A gene expression profiling analysis of the biopsy samples may be performed and one or more photo-aging gene expression signatures derived from a statistical analysis of the results. In some embodiments, the photo-aging gene expression signature comprises about equal numbers of up-regulated and down-regulated genes. In an alternate embodiment, a photo-aging gene expression signature may be derived by comparing a sun exposed site of an older individual (e.g., 45 to 80 y.o.) to a sun exposed site of a younger individual (e.g., 18 to 25 y.o.)

Generally, for an intrinsic aging gene expression signature, biopsies may be taken from sun protected sites (e.g., buttocks) of a plurality of older and younger subjects. The subjects may vary in age, but one age range is between about 45 years of age and 80 years of age for the older subjects and 18 years of age and 25 years of age for the younger subjects. A gene expression profiling analysis of the biopsy samples may be performed and one or more intrinsic aging gene expression signatures derived from a statistical analysis of the microarray results. In some embodiments, the intrinsic aging gene expression signature comprises about equal numbers of up-regulated and down-regulated genes.

In some embodiments, the photo-aging and intrinsic aging gene expression signatures have fewer than 20% of their genes in common, and they reflect different aspects of the biology of aging skin as described. The intrinsic aging gene expression signatures may capture decreased expression of epidermal differentiation markers and down-regulation of pathways involved in the synthesis of lipids important in epidermal barrier function as well as changes related to the dermis, while the photo-aging signatures may be much more reflective of modified biology of the dermis and fibroblasts. Therefore, both photo-aging and intrinsic aging gene expression signatures are useful for identification of cosmetic agents to improve the appearance of aging skin and allow identification of potential perturbagens with differing biological activities.

In other embodiments of the present invention, a gene expression signature may be derived from a gene expression profiling analysis of fibroblast and/or keratinocyte cells treated with a benchmark skin agent to represent cellular perturbations leading to improvement in the skin tissue condition treated with that benchmark skin agent, said signature comprising a plurality of genes up-regulated and down-regulated by the benchmark skin agent in cells in vitro. As one illustrative example, microarray gene expression profile data where the perturbagen is the known skin anti-aging agent all trans-retinoic acid (tRA) may be analyzed using the present invention to determine from the rank-ordered instances in the query results, the genes associated with the highest scoring instances. Thus, a list of genes strongly up-regulated and strongly down-regulated in response to challenge with tRA can be derived, and said list of genes (a proxy for skin anti-aging) can be used as a query signature to screen for skin anti-aging agents. In another embodiment, a signature may be derived to represent more than one aspect of the condition of interest.

IV. Methods for Comparing a Plurality of Instances to One or More Skin Aging Gene Expression Signatures

Referring to FIG. 6 and FIG. 7, a method for querying a plurality of instances with one or more skin aging gene signatures will now be described. Broadly, the method comprises querying a plurality of instances with one or more skin aging gene signatures and applying a statistical method to determine how strongly the signature genes match the regulated genes in an instance. Positive connectivity occurs when the genes in the up-regulated signature list are enriched among the up-regulated genes in an instance and the genes in the down-regulated signature list are enriched among the down-regulated genes in an instance. On the other hand, if the up-regulated genes of the signature are predominantly found among the down-regulated genes of the instance, and vice versa, this is scored as negative connectivity. FIG. 6 schematically illustrates an extreme example of a positive connectivity between signature 90 and the instance 104 comprising the probe IDs 102, wherein the probe IDs of the instance are ordered from most up-regulated to most down-regulated. In this example, the probe IDs 100 (e.g., X₁, X₂ X₃, X₄, X₅, X₆, X₇, X₈) of the gene signature 90, comprising an up list 97 and a down list 99, have a one to one positive correspondence with the most up-regulated and down-regulated probe IDs 102 of the instance 104, respectively. Similarly, FIG. 7 schematically illustrates an extreme example of a negative connectivity between signature 94 and the instance 88 comprising the probe IDs 90, wherein the probe IDs of the instance are ordered from most up-regulated to most down-regulated. In this example, the probe IDs of the up list 93 (e.g., X₁, X₂ X₃, X₄) correspond exactly with the most down-regulated genes of the instance 88, and the probe IDs of the down list 95 (e.g., X₅, X₆, X₇, X₈) correspond exactly to the most up-regulated probe IDs of the instance 88. FIG. 8 schematically illustrates an extreme example of neutral connectivity, wherein there is no consistent enrichment of the up- and down-regulated genes of the signature among the up- and down-regulated genes of the instance, either positive or negative. Hence the probe IDs 106 (e.g., X₁, X₂ X₃, X₄, X₅, X₆, X₇, X₈) of a gene signature 108 (comprising an up list 107 and a down list 109) are scattered with respect to rank with the probe IDs 110 of the instance 112, wherein the probe IDs of the instance are ordered from most up-regulated to most down-regulated. While the above embodiments illustrate process where the gene signature comprises a both an up list and a down list representative of the most significantly up- and down-regulated genes of a skin condition, it is contemplated that the gene signature may comprise only an up list or a down list when the dominant biology associated with a condition of interest shows gene regulation in predominantly one direction.

In some embodiments, the connectivity score can be a combination of an up-score and a down score, wherein the up-score represents the correlation between the up-regulated genes of a gene signature and an instance and the down-score represents the correlation between the down-regulated genes of a gene signature and an instance. The up score and down score may have values between +1 and −1. For an up score (and down score) a high positive value indicates that the corresponding perturbagen of an instance induced the expression of the microarray probes of the up-regulated (or down-regulated) genes of the gene signature, and a high negative value indicates that the corresponding perturbagen associated with the instance repressed the expression of the microarray probes of the up-regulated (or down-regulated) genes of the gene signature. The up-score can be calculated by comparing each identifier of an up list of a gene signature comprising the up-regulated genes (e.g., Tables A, C, I and lists 93, 97, and 107) to an ordered instance list (e.g., Tables E, F, G, H) while the down-score can be calculated by comparing each identifier of a down list of a gene signature comprising the down-regulated genes (see, e.g., Tables B, D, J and down lists 95, 99, and 109) to an ordered instance list (e.g., Tables E, F, G, H). In these embodiments, the gene signature comprises the combination of the up list and the down list.

In some embodiments, the connectivity score value may range from +2 (greatest positive connectivity) to −2 (greatest negative connectivity), wherein the connectivity score (e.g., 101, 103, and 105) is the combination of the up score (e.g., 111, 113, 115) and the down score (e.g., 117, 119, 121) derived by comparing each identifier of a gene signature to the identifiers of an ordered instance list. In other embodiments the connectivity range may be between +1 and −1. Examples of the scores are illustrated in FIGS. 6, 7 and 8 as reference numerals 101, 103, 105, 111, 113, 115, 117, 119, and 121. The strength of matching between a signature and an instance represented by the up scores and down scores and/or the connectivity score may be derived by one or more approaches known in the art and include, but are not limited to, parametric and non-parametric approaches. Examples of parametric approaches include Pearson correlation (or Pearson r) and cosine correlation. Examples of non-parametric approaches include Spearman's Rank (or rank-order) correlation, Kendall's Tau correlation, and the Gamma statistic. Generally, in order to eliminate a requirement that all profiles be generated on the same microarray platform, a non-parametric, rank-based pattern matching strategy based on the Kolmogorov-Smirnov statistic (see M. Hollander et al. “Nonparametric Statistical Methods”; Wiley, New York, ed. 2, 1999)(see, e.g., pp. 178-185). It is noted, however, that where all expression profiles are derived from a single technology platform, similar results may be obtained using conventional measures of correlation, for example, the Pearson correlation coefficient.

In specific embodiments, the methods and systems of the present invention employ the nonparametric, rank-based pattern-matching strategy based on the Kolmogorov-Smirnov statistic, which has been refined for gene profiling data by Lamb's group, commonly known in the art as Gene Set Enrichment Analysis (GSEA) (see, e.g., Lamb et al. 2006 and Subramanian, A. et al. (2005) Proc. Natl. Acad Sci U.S.A, 102, 15545-15550). For each instance, a down score is calculated to reflect the match between the down-regulated genes of the query and the instance, and an up score is calculated to reflect the correlation between the up-regulated genes of the query and the instance. In certain embodiments the down score and up score each may range between −1 and +1. The combination represents the strength of the overall match between the query signature and the instance.

The combination of the up score and down score is used to calculate an overall connectivity score for each instance, and in embodiments where up and down score ranges are set between −1 and +1, the connectivity score ranges from −2 to +2, and represents the strength of match between a query signature and the instance. The sign of the overall score is determined by whether the instance links positivity or negatively to the signature. Positive connectivity occurs when the perturbagen associated with an instance tends to up-regulate the genes in the up list of the signature and down-regulate the genes in the down list. Conversely, negative connectivity occurs when the perturbagen tends to reverse the up and down signature gene expression changes, The magnitude of the connectivity score is the sum of the absolute values of the up and down scores when the up and down scores have different signs. A high positive connectivity score predicts that the perturbagen will tend to induce the condition that was used to generate the query signature, and a high negative connectivity score predicts that the perturbagen will tend to reverse the condition associated with the query signature. A zero score is assigned where the up and down scores have the same sign, indicating that a perturbagen did not have a consistent impact the condition signature (e.g., up-regulating both the up and down lists).

According to Lamb et al. (2006), there is no standard for estimating statistical significance of connections observed. Lamb teaches that the power to detect connections may be greater for compounds with many replicates. Replicating in this context means that the same perturbagen is profiled multiple times. Where batch to batch variation must be avoided, a perturbagen should be profiled multiple times in each batch. However, since microarray experiments tend to have strong batch effects it is desirable to replicate instances in different batches (i.e., experiments) to have the highest confidence that connectivity scores are meaningful and reproducible.

Each instance may be rank ordered according to its connectivity score to the query signature and the resulting rank ordered list displayed to a user using any suitable software and computer hardware allowing for visualization of data.

In some embodiments, the methods may comprise identifying from the displayed rank-ordered list of instances (i) the one or more perturbagens associated with the instances of interest (thereby correlating activation or inhibition of a plurality of genes listed in the query signature to the one or more perturbagens); (ii) the differentially expressed genes associated with any instances of interest (thereby correlating such genes with the one or more perturbagens, the skin tissue condition of interest, or both); (iii) the cells associated with any instance of interest (thereby correlating such cells with one or more of the differentially expressed genes, the one or more perturbagens, and the skin tissue condition of interest); or (iv) combinations thereof. The one or more perturbagens associated with an instance may be identified from the metadata stored in the database for that instance. However, one of skill in the art will appreciate that perturbagen data for an instance may be retrievably stored in and by other means. Because the identified perturbagens statistically correlate to activation or inhibition of genes listed in the query signature, and because the query signature is a proxy for a skin tissue condition of interest, the identified perturbagens may be candidates for new cosmetic agents, new uses of known cosmetic agents, or to validate known agents for known uses.

In some embodiments, the methods of the present invention may further comprise testing the selected candidate cosmetic agent, using in vitro assays and/or in vivo testing, to validate the activity of the agent and usefulness as a cosmetic agent. Any suitable in vitro test method can be used, including those known in the art, and most preferably in vitro models having an established nexus to the desired in vivo result. For example, MatTek human skin equivalent cultures and skin biopsy assays may be used to evaluate candidate cosmetic agents. In some embodiments, evaluation of selected agents using in vitro assays may reveal, confirm, or both, that one or more new candidate cosmetic agents may be used in conjunction with a known cosmetic agent (or a combination of known cosmetic agents) to regulate a skin aging tissue condition of interest.

V. Cosmetic Compositions and Personal Care Products

Because of the desirability of providing various cosmetic skin anti-aging benefits to a consumer, it may be beneficial to incorporate test agents or compounds identified by one or more of the screening methods described herein into a cosmetic composition suitable for topical application to skin. That is, it may be desirable to include the test agent as an ingredient in the cosmetic composition. In certain embodiments, the cosmetic composition may include a dermatological acceptable carrier, the test agent, and one or more optional ingredients of the kind commonly included in the particular cosmetic compositing being provided.

Dermatologically acceptable carriers should be safe for use in contact with human skin tissue. Suitable carriers may include water and/or water miscible solvents. The cosmetic skin care composition may comprise from about 1% to about 95% by weight of water and/or water miscible solvent. The composition may comprise from about 1%, 3%, 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, or 85% to about 90%, 85%, 80%, 75%, 70%, 65%, 60%, 55%, 50%, 45%, 40%, 35%, 30%, 25%, 20%, 15%, 10%, or 5% water and/or water miscible solvents. Suitable water miscible solvents include monohydric alcohols, dihydric alcohols, polyhydric alcohols, glycerol, glycols, polyalkylene glycols such as polyethylene glycol, and mixtures thereof. When the skin care composition is in the form of an emulsion, water and/or water miscible solvents are carriers typically associated with the aqueous phase.

Suitable carriers also include oils. The skin care composition may comprise from about 1% to about 95% by weight of one or more oils. The composition may comprise from about 0.1%, 0.5%, 1%, 2%, 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, or 90% to about 90%, 85%, 80%, 75%, 70%, 65%, 60%, 55%, 50%, 45%, 40%, 35%, 30%, 25%, 20%, 15%, 10%, 5%, or 3% of one or more oils. Oils may be used to solubilize, disperse, or carry materials that are not suitable for water or water soluble solvents. Suitable oils include silicones, hydrocarbons, esters, amides, ethers, and mixtures thereof. The oils may be volatile or nonvolatile.

Suitable silicone oils include polysiloxanes. Commercially available polysiloxanes include the polydimethylsiloxanes, which are also known as dimethicones, examples of which include the DM-Fluid series from Shin-Etsu, the Vicasil® series sold by Momentive Performance Materials Inc., and the Dow Corning® 200 series sold by Dow Corning Corporation. Specific examples of suitable polydimethylsiloxanes include Dow Corning® 200 fluids (also sold as Xiameter® PMX-200 Silicone Fluids) having viscosities of 0.65, 1.5, 50, 100, 350, 10,000, 12,500 100,000, and 300,000 centistokes.

Suitable hydrocarbon oils include straight, branched, or cyclic alkanes and alkenes. The chain length may be selected based on desired functional characteristics such as volatility. Suitable volatile hydrocarbons may have between 5-20 carbon atoms or, alternately, between 8-16 carbon atoms.

Other suitable oils include esters. The suitable esters typically contained at least 10 carbon atoms. These esters include esters with hydrocarbyl chains derived from fatty acids or alcohols (e.g., mono-esters, polyhydric alcohol esters, and di- and tri-carboxylic acid esters). The hydrocarbyl radicals of the esters hereof may include or have covalently bonded thereto other compatible functionalities, such as amides and alkoxy moieties (e.g., ethoxy or ether linkages, etc.).

Other suitable oils include amides. Amides include compounds having an amide functional group while being liquid at 25° C. and insoluble in water. Suitable amides include N-acetyl-N-butylaminopropionate, isopropyl N-lauroylsarcosinate, and N,N,-diethyltoluamide. Other suitable amides are disclosed in U.S. Pat. No. 6,872,401.

Other suitable oils include ethers. Suitable ethers include saturated and unsaturated fatty ethers of a polyhydric alcohol, and alkoxylated derivatives thereof. Exemplary ethers include C_4-20 alkyl ethers of polypropylene glycols, and di-C_8-30 alkyl ethers. Suitable examples of these materials include PPG-14 butyl ether, PPG-15 stearyl ether, dioctyl ether, dodecyl octyl ether, and mixtures thereof.

The skin care composition may comprise an emulsifier. An emulsifier is particularly suitable when the composition is in the form of an emulsion or if immiscible materials are being combined. The skin care composition may comprise from about 0.05%, 0.1%, 0.2%, 0.3%, 0.5%, or 1% to about 20%, 10%, 5%, 3%, 2%, or 1% emulsifier. Emulsifiers may be nonionic, anionic or cationic. Non-limiting examples of emulsifiers are disclosed in U.S. Pat. No. 3,755,560, U.S. Pat. No. 4,421,769, and McCutcheon's, Emulsifiers and Detergents, 2010 Annual Ed., published by M. C. Publishing Co. Other suitable emulsifiers are further described in the Personal Care Product Council's International Cosmetic Ingredient Dictionary and Handbook, Thirteenth Edition, 2006, under the functional category of “Surfactants—Emulsifying Agents.”

Linear or branched type silicone emulsifiers may also be used. Particularly useful polyether modified silicones include KF-6011, KF-6012, KF-6013, KF-6015, KF-6015, KF-6017, KF-6043, KF-6028, and KF-6038 from Shin Etsu. Also particularly useful are the polyglycerolated linear or branched siloxane emulsifiers including KF-6100, KF-6104, and KF-6105 from Shin Etsu. Emulsifiers also include emulsifying silicone elastomers. Suitable silicone elastomers may be in the powder form, or dispersed or solubilized in solvents such as volatile or nonvolatile silicones, or silicone compatible vehicles such as paraffinic hydrocarbons or esters. Suitable emulsifying silicone elastomers may include at least one polyalkyl ether or polyglycerolated unit.

Structuring agents may be used to increase viscosity, thicken, solidify, or provide solid or crystalline structure to the skin care composition. Structuring agents are typically grouped based on solubility, dispersibility, or phase compatibility. Examples of aqueous or water structuring agents include polymeric agents, natural or synthetic gums, polysaccharides, and the like. In one embodiment, the composition may comprises from about 0.0001%, 0.001%, 0.01%, 0.05%, 0.1%, 0.5%, 1%, 2%, 3%, 5% to about 25%, 20%, 10%, 7%, 5%, 4%, or 2%, by weight of the composition, of one or more structuring agents.

Polysaccharides and gums may be suitable aqueous phase thickening agents. Suitable classes of polymeric structuring agents include but are not limited to carboxylic acid polymers, polyacrylamide polymers, sulfonated polymers, high molecular weight polyalkylglycols or polyglycerins, copolymers thereof, hydrophobically modified derivatives thereof, and mixtures thereof. Silicone gums are another oil phase structuring agent. Another type of oily phase structuring agent includes silicone waxes. Silicone waxes may be referred to as alkyl silicone waxes which and are semi-solids or solids at room temperature. Other oil phase structuring agents may be one or more natural or synthetic waxes such as animal, vegetable, or mineral waxes.

The skin care compositions may be generally prepared by conventional methods such as known in the art of making compositions and topical compositions. Such methods typically involve mixing of ingredients in or more steps to a relatively uniform state, with or without heating, cooling, application of vacuum, and the like. Typically, emulsions are prepared by first mixing the aqueous phase materials separately from the fatty phase materials and then combining the two phases as appropriate to yield the desired continuous phase. The compositions are preferably prepared such as to optimize stability (physical stability, chemical stability, photostability, etc.) and/or delivery of active materials. The composition may be provided in a package sized to store a sufficient amount of the composition for a treatment period. The size, shape, and design of the package may vary widely. Certain package examples are described in USPNs D570,707; D391,162; D516,436; D535,191; D542,660; D547,193; D547,661; D558,591; D563,221; 2009/0017080; 2007/0205226; and 2007/0040306.

EXAMPLES

The present invention will be better understood by reference to the following examples which are offered by way of illustration not limitation.

Example 1

Generating Instances

Individual experiments (referred to as batches) generally comprise 30 to 96 samples analyzed using Affymetrix GeneChip® technology platforms, containing 6 replicates of the vehicle control (e.g., DSMO), 2 replicate samples of a positive control that gives a strong reproducible effect in the cell type used (e.g., all trans-retinoic acid for fibroblast cells), and samples of the test material/perturbagen. Replication of the test material is done in separate batches due to batch effects. In vitro testing was performed in 6-well plates to provide sufficient RNA for GeneChip® analysis (2-4 μg total RNA yield/well).

Human telomerized keratinocytes (tKC) were obtained from the University of Texas, Southwestern Medical Center, Dallas, Tex. tKC cells were grown in EpiLife® media with 1× Human Keratinocyte Growth Supplement (Invitrogen, Carlsbad, Calif.) on collagen I coated cell culture flasks and plates (Becton Dickinson, Franklin Lakes, N.J.). Keratinocytes were seeded into 6-well plates at 20,000 cells/cm² 24 hours before chemical exposure. Human skin fibroblasts (BJ cell line from ATCC, Manassas, Va.) were grown in Eagle's Minimal Essential Medium (ATCC) supplemented with 10% fetal bovine serum (HyClone, Logan, Utah) in normal cell culture flasks and plates (Corning, Lowell, Mass.). BJ fibroblasts were seeded into 6-well plates at 12,000 cells/cm² 24 hours before chemical exposure.

All cells were incubated at 37° C. in a humidified incubator with 5% CO₂. At t=−24 hours cells were trypsinized from T-75 flasks and plated into 6-well plates in basal growth medium. At t=0 media was removed and replaced with the appropriate dosing solution as per the experimental design. Dosing solutions were prepared the previous day in sterile 4 ml Falcon snap cap tubes. Pure test materials may be prepared at a concentration of 1-200 μM, and botanical extracts may be prepared at a concentration of 0.001 to 1% by weight of the dosing solution. After 6 to 24 hours of chemical exposure, cells were viewed and imaged. The wells were examined with a microscope before cell lysis and RNA isolation to evaluate for morphologic evidence of toxicity. If morphological changes were sufficient to suggest cytotoxicity, a lower concentration of the perturbagen was tested. Cells were then lysed with 350 ul/well of RLT buffer containing β-mercaptoethanol (Qiagen, Valencia, Calif.), transferred to a 96-well plate, and stored at −20° C.

RNA from cell culture batches was isolated from the RLT buffer using Agencourt® RNAdvance Tissue-Bind magnetic beads (Beckman Coulter) according to manufacturer's instructions. 1 μg of total RNA per sample was labeled using Ambion Message Amp™ II Biotin Enhanced kit (Applied Biosystems Incorporated) according to manufacturer's instructions. The resultant biotin labeled and fragmented cRNA was hybridized to an Affymetrix HG-U133A 2.0 GeneChip®, which was then washed, stained and scanned using the protocol provided by Affymetrix.

In one embodiment, an exemplary test perturbagen was Cynara scolymus (artichoke) leaf extract obtained from Ichimaru Pharcos, JP. Following generally the procedure described above, a resultant rank-ordered listing of 22,214 probe set IDs (e.g., an instance) for the artichoke leaf perturbagen applied to fibroblast cells was generated and are set forth in Table E. Following generally the procedure described above, a resultant rank-ordered listing of 22,214 probe set IDs (e.g., an instance) for the artichoke leaf extract perturbagen applied to keratinocyte cells was generated and illustrative portions are set forth in Table F. In another aspect, an exemplary test perturbagen was Hydrolyzed Ceratonia siliqua (carob) seed extract. Following generally the procedure described above, a resultant rank-ordered listing of 22,214 probe set IDs (e.g., an instance) for the carob seed extract perturbagen applied to fibroblast cells was generated and illustrative portions are set forth in Table G. Following generally the procedure described above, a resultant rank-ordered listing of 22,214 probe set IDs (e.g., an instance) for the carob seed extract perturbagen applied to keratinocyte cells was generated and illustrative portions are set forth in Table H.

Example 2

Deriving a Photo-Aging Gene Expression Signature

A clinical survey study to obtain biopsy specimens for use in the investigation of gene expression patterns associated with sun light-mediated skin aging (photo-aging) was performed. Baseline gene expression patterns were examined in sun-protected and sun-exposed skin from young and aged women to examine gene expression profiles associated with photo-aging. A total of 3 full thickness skin biopsies (−4 mm) were taken from sun-protected (buttocks) and sun-exposed (extensor forearm) body sites from each of 10 young women (aged 18 to 20 years) and 10 older women (aged 60 to 67 years). The older women were selected to have moderate to severe forearm photo-damage. Biopsies were flash frozen in liquid nitrogen and stored at −80° C. until RNA isolation.

Frozen skin biopsies were homogenized in Trizol (Invitrogen) and RNA extracted using the protocol provided by Invitrogen. Since the tissue samples were from full thickness biopsies, RNA was extracted from a variety of cell types within the full-thickness skin sample, including keratinocytes, fibroblasts, melanocytes, endothelial cells, pericytes, nerves, smooth muscle, sebocytes, adipocytes, and immunocytes). RNA was further purified using RNEasy spin columns (Qiagen). Total RNA was quantified using a NanoDrop spectrophotometer (Thermo Scientific, Waltham, Mass.) and quality was confirmed using an Agilent (Santa Clara, Calif.) 2100 BioAnalyzer. Total RNA (5 μg) was converted to GeneChip targets using the Enzo BioArray labeling procedure (Enzo Life Sciences, Farmingdale, N.Y.) and protocol provided. All biotin-labeled GeneChip targets were hybridized to Affymetrix Human Genome HG-U133 Plus 2.0 GeneChips overnight, which were then washed, stained and scanned using the protocol provided by Affymetrix. Forearm and buttock samples were processed on the same day using the same manufacturing lot of GeneChips.

The samples were analyzed on the Affymetrix HG-U133 Plus 2.0 GeneChips, which contain 54,613 probe sets complementary to the transcripts of more than 20,000 genes. However, instances in the provided database used were derived from gene expression profiling experiments using Affymetrix HG-U133A 2.0 GeneChips, containing 22,214 probe sets, which are a subset of those present on the Plus 2.0 GeneChip. Therefore, in developing gene expression signatures from the clinical data, the probe sets were filtered for those included in the HG-U133A 2.0 gene chips.

Using, generally the following selection process, a statistical analysis of the microarray data was performed to derive a plurality of photo-aging gene expression signatures comprising between about 200 and about 600 up-regulated and down-regulated genes.

• • a. Filtering based on Absent/Margin/Present Calls. This filter creates a list of potential genes for inclusion in the gene expression signature. For example, a suitable filter may be that at least 50% of the samples in one treatment group must have a Present call for each probe set. Present calls are derived from processing the raw GeneChip data and provide evidence that the gene transcript complementary to a probe set that is actually expressed in the biological sample. The probes that are absent from all samples are likely to be just noisy measurements. This step is important to filter out probe sets that do not contribute meaningful data to the signature. For both photo-aging and intrinsic aging gene expression signatures, the data was filtered for probe sets with at least 10% Present calls provided by the Affymetrix MAS 5 software.
  • b. Filtering According to a Statistical Measure. For example, a suitable statistical measure may be p-values from a t-test, ANOVA, correlation coefficient, or other model-based analysis. As one example, p-values may be chosen as the statistical measure and a cutoff value of p=0.05 may be chosen. Limiting the signature list to genes that meet some reasonable cutoff for statistical significance compared to an appropriate control is important to allow selection of genes that are characteristic of the biological state of interest. This is preferable to using a fold change value, which does not take into account the noise around the measurements. The t-statistic was used to select the probe sets in the signatures because it is signed and provides an indication of the directionality of the gene expression changes (i.e. up- or down-regulated) as well as statistical significance.
  • c. Sorting the Probe Sets. All the probe sets are sorted into sets of up-regulated and down-regulated sets using the statistical measure. For example, if a t-test was used to compute p-values, the values (positive and negative) of the t-statistic are used to sort the list since p-values are always positive. The sorted t-statistics will place the sets with the most significant p-values at the top and bottom of the list with the non-significant ones near the middle.
  • d. Creation of the Gene expression signature. Using the filtered and sorted list created, a suitable number of probe sets from the top and bottom are selected to create a gene expression signature that preferably has approximately the same number of sets chosen from the top as chosen from the bottom. For example, the gene expression signature created may have at least about 10, 50, 100, 200, or 300 and/or less than about 800, 600, or about 400 genes corresponding to a probe set on the chip. The number of probe sets approximately corresponds to the number of genes, but a single gene may be represented by more than one probe set. It is understood that the phrase “number of genes” as used herein, corresponds generally with the phrase “number of probe sets.” The number of genes included in the signature was based upon the observations in preliminary studies that indicated signatures with from 200 to 800 probe sets equally divided between up- and down-regulated genes provide stable results with regard to the top scoring chemical instances when using the signature to query the provided database.

For photo-aging, three gene expression signatures were selected as follows:

• • (i) “Photo-aging 200” signature (P200), which comprises 100 most significant up- and 100 most significant down-regulated probe sets comparing older arm to younger arm as set forth in Tables C and D, respectively;
  • (ii) “Photo-aging 400” signature (P400), which comprises 200 most significant up- and 200 most significant down-regulated probe sets comparing older arm to younger arm as set forth in Tables C and D, respectively; and
  • (iii) “Photo-aging 600” signature (P600), which comprises 300 most significant up- and 300 most significant down-regulated probe sets comparing older arm to younger arm as set forth in Tables C and D, respectively.

Example 3

Deriving an Intrinsic Aging Gene Expression Signature

A clinical survey study to obtain biopsy specimens for use in the investigation of gene expression patterns associated with chronological (intrinsic) skin aging was performed. Baseline gene expression patterns were examined in sun-protected skin from young and aged women to examine gene expression profiles associated with intrinsic aging. A total of 3 full thickness skin biopsies (−4 mm) were taken from sun-protected (buttocks) body sites from each of 10 young women (aged 18 to 20 years) and 10 older women (aged 60 to 67 years). Biopsies were flash frozen in liquid nitrogen and stored at −80° C. until RNA isolation.

Frozen skin biopsies were homogenized in Trizol (Invitrogen) and RNA extracted using the protocol provided by Invitrogen. Since the tissue samples were from full thickness biopsies, RNA was extracted from a variety of cell types within the full-thickness skin sample, including keratinocytes, fibroblasts, melanocytes, endothelial cells, pericytes, nerves, smooth muscle, sebocytes, adipocytes, and immunocytes). RNA was further purified using RNEasy spin columns (Qiagen). Total RNA was quantified using a NanoDrop spectrophotometer (Thermo Scientific, Waltham, Mass.) and quality was confirmed using an Agilent (Santa Clara, Calif.) 2100 BioAnalyzer. Total RNA (5 μg) was converted to GeneChip targets using the Enzo BioArray labeling procedure (Enzo Life Sciences, Farmingdale, N.Y.) and protocol provided. All biotin-labeled GeneChip targets were hybridized to Affymetrix Human Genome HG-U133 Plus 2.0 gene chips overnight, which were then washed, stained and scanned using the protocol provided by Affymetrix. Samples were processed on the same day using the same manufacturing lot of gene chips.

Using, generally the same sorting process as set forth in Example 2, a statistical analysis of the microarray data was performed to derive intrinsic aging gene expression signatures comprising between about 200 genes and about 600 genes.

• • (i) “Intrinsic Aging 200” signature (1200), which comprises 100 most significant up- and 100 most significant down-regulated probe sets comparing older buttock to younger buttock as set forth in Table A and B, respectively.
  • (ii) “Intrinsic Aging 400” signature (1400), which comprises 200 most significant up- and 200 most significant down-regulated probe sets older buttock to younger buttock as set forth in Table A and B, respectively.
  • (iii) “Intrinsic Aging 600” signature (1600), which comprises 300 most significant up- and 300 most significant down-regulated probe sets older buttock to younger buttock as set forth in Table A and B, respectively.

The photo-aging gene expression signatures and the intrinsic aging gene expression signatures may be used in whole or part, or they may be used in combination as a query in the present invention. In some embodiments, the gene expression signature comprises the set of most up-regulated or most down-regulated genes. In some embodiments, the photo-aging gene expression signature and intrinsic aging gene expression signature may have between about 10% and 50% of their probe set IDs in common. In other embodiments, the photo-aging gene expression signature and the intrinsic aging gene expression signature may have between about 20% and 30% of their probe set IDs in common. Some materials tested in C-map link to only to gene expression signatures developed for one of the types of skin aging because these agents may be affecting different biology. For example, the inventors have found that the photo-aging gene expression signatures show strong preferential linkage to perturbagen instances tested in fibroblasts and the gene expression changes are more reflective of changes in the dermal connective tissue. In contrast, the intrinsic aging gene expression signatures reflect changes in both the epidermis and dermis that occur with aging. Therefore, it is useful to query the connectivity map with separate signatures derived for both intrinsic aging and photo-aging. Also, it is desirable to use signatures with different numbers of genes. Predictions of biological activity based on multiple signatures are more likely to be correct.

Example 4

Photo-Aging Gene Expression Signature and Intrinsic Aging Gene Expression Signature Comparison to Fibroblast and Keratinocyte Instances

The inventors observed that photo-aging gene signatures (e.g., Example 2) have a strong preferential connectivity toperturbagen instances tested in BJ fibroblasts. In other words, among the top ranked instances (i.e., with high connectivity scores) resulting from queries with photo-aging signatures BJ fibroblast instances were strongly over-represented. This effect was most dramatic when the genes up-regulated in photoaged skin were used as the query signatures. A similar result, but less strong, was obtained with the intrinsic aging signatures (e.g., Example 3). A summary of the results of comparing a plurality of instances with the Photo-aging 600 signature is shown in Table 1. These results are completely unpredicted and surprising considering that: i) the gene signatures derived from full thickness biopsies comprising a variety of cell types expected to contribute to the skin aging phenotype, and ii) skin aging is a complex multi-factorial condition involving changes in most of the tissues and structures of skin,

TABLE 1
Distribution of fibroblast (BJ and keratinocyte (tKC)
instances in the top ranked instances from a C-map query
with the Photo-aging 600 signature 1
	Chemical instances tested in each cell type
	Complete
	signature		Up list 3		Down list 3
Rank 2	BJ	tKC	BJ	tKC	BJ	tKC
1-100	79	21	100	0	36	64
101-200	60	40	100	0	36	64
201-300	57	43	100	0	41	59
301-400	51	49	96	4	34	66
401-500	38	62	85	15	15	72
1 This signature comprises the 300 most statistically significant up-regulated and the 300 most significant down-regulated Affymetrix probe sets in the older arm to younger arm comparison described in Example 2, above.
2 Instances were ranked from the most negatively scoring to most positive (i.e., 1 was the most negatively scoring instance). Chemicals scoring negatively are predicted to have beneficial effects on photo-aging.
3 The up list results are from querying a database of instances with only up-regulated probe sets in the gene expression signature. Similarly, the down list results are from querying with a gene expression signature comprising only probe sets that are down-regulated.

Among the top 300 ranked instances linking to the complete photo-aging signature, 196 (65.3%) were with respect to BJ fibroblasts and only 34.7% were with respect to keratinocytes. This was more than a 2-fold over-representation of fibroblasts, since they comprised only 31% of the instances in the database. The preferential connectivity of fibroblasts to the photo-aging signature was much more dramatic when only the up list was used for the query. Remarkably, in that case all of the top 300 instances were in fibroblasts. The up signature contribution accounted for the preferential connectivity of fibroblasts to the complete photo-aging signature, because querying with the down signature alone gave a distribution of top scoring instances more proportional to the percentages of the two cell lines instances in the database. Similar results were obtained with the Photo-aging P200 and P400 signatures described in Example 2. Overall, the strong preferential connectivity of fibroblast instances to the photo-aging signatures is surprising because keratinocytes are expected to contribute to the photo-aging phenotype and all of the materials tested in BJ fibroblasts were also tested in keratinocytes (i.e., the effect is not due to any chemical specificity).

A similar but subtly different pattern was observed when the provided database was queried with intrinsic aging gene expression signatures. Table 2 shows results obtained with the Intrinsic Aging 600 signature (Example 3). Again there was an over-representation of fibroblasts among the top ranked instances with the complete signature, and this effect was due to the up-regulated probe sets in the signature, but the effect was not as strong as with the photo-aging signatures. Furthermore, the top ranking instances connecting to the down list showed some over-representation of keratinocytes relative to their proportion of the samples tested (Table 2).

TABLE 2
Distribution of fibroblast (BJ) and keratinocyte (tKC)instances
in the top ranked instances from a C-map query with the
Intrinsic Aging 600 signature. 1
	Chemical instances tested in each cell type
	Complete
	signature		Up list 3		Down list 3
Rank 2	BJ	tKC	BJ	tKC	BJ	tKC
1-100	68	32	95	5	13	87
101-200	55	45	84	16	17	83
201-300	38	62	72	28	17	83
301-400	37	63	67	33	21	79
401-500	31	69	57	43	26	74
1 This signature comprises the 300 most statistically significant up- and 300 most significant down-regulated Affymetrix probe sets in the older buttock to younger buttock comparison in the clinical study described in Example 3.
2 Instances were ranked as in Table 1.
3 The up list results are from querying the C-Map database with only the up-regulated probe sets in the signature. Similarly, the down list results are from querying only with the down list.

Unexpectedly, the gene expression signatures generated from clinical samples of photo-aged skin show very strong preferential connectivity to chemical instances tested in dermal fibroblasts compared to epidermal keratinocytes. Furthermore, this preferential connectivity was observed with signatures generated from gene expression profiling done on RNA extracted from full thickness biopsies of skin, which comprise a composite sample of all of the cell types within the complex structure of skin. Analysis of the biological processes associated with the genes differentially expressed in photo-aged skin indicated that many of the processes were related to the dermis and fibroblasts including wound healing (see Theme Analysis, Example 5). These results support the hypothesis that the dominant gene expression changes in photo-aged skin occur in dermal fibroblasts. Furthermore, in clinical testing statistically significant anti-wrinkle results were obtained with two botanical extracts predicted to have skin anti-aging activity based on C-Map queries of fibroblast instances (e.g. Examples 6 and 7). In contrast, the keratinocytes instances of these materials did not predict their anti-aging activities.

These differences between the photo-aging and intrinsic aging signatures may be accounted for by the fact that they represent overlapping but different biological changes in aging skin (see Theme Analysis, Example 5). Photo-aging is due to a combination of the effects of UV radiation exposure and intrinsic factors, while intrinsic aging lacks the solar/UV component. Fewer than 20% of the genes in the photo-aging and intrinsic aging signatures were observed to be in common. These results point to the value of using both photo-aging and intrinsic aging signatures when applying C-map to identify cosmetic agents to improve aging skin, because the signatures will allow detection of materials that affect different aspects of the complex aging process.

The under-representation of telomerized keratinocytes in the top ranked C-map instances with queries using the complete photo-aging and intrinsic aging gene expression signatures was not due to a lack of responsiveness of the cells under the conditions of testing. Table 3 sets forth gene expression outlier numbers for 493 chemicals that were tested in both BJ fibroblasts and telomerized keratinocytes using the same stock solutions. Outlier numbers were calculated for each singleton C-map instance and are a measure of the number of probe sets regulated by treatment. In this probe set-by-probe set analysis, the 6 vehicle control samples in a C-map batch are used to form a prediction interval for each probe set, and then for each chemical instance in the batch the probe sets falling outside the interval are counted. This statistic is used because chemical replication is usually done in different C-map batches due to batch effects on the scoring.

TABLE 3
Gene expression outlier numbers for 493 materials tested
in both BJ fibroblasts and telomerized keratinocytes.1
	Number of C-MAP instances
Range of	BJ	Telomerized-
outliers	fibroblasts	keratinocytes
≧2000	123	245
1000-1999	711	725
<1000	292	234
Totals:	1126	1204
1Materials from the same stock tubes were tested in both types of cells. Outlier numbers were calculated for each singleton C-map instance and are a measure of the number of probe sets regulated by treatment. In this probe set-by-probe set analysis, the 6 vehicle control samples in a C-map batch are used to form a prediction interval for each probe set, and then for each chemical instance in the batch the probe sets falling outside the interval are counted. This statistic is used because chemical replication is usually done in different C-map batches due to batch effects on the scoring.

Based on outlier numbers, the keratinocytes tended to be more responsive to treatment than fibroblasts and there were about twice as many keratinocyte instances with high outlier numbers (≧2,000) compared to those with fibroblasts. Additional indications of cellular responsiveness to treatment are morphologic changes. Altered morphology in telomerized keratinocytes treated with various chemicals has been observed. In contrast, morphologic changes in BJ fibroblasts have rarely been observed under the same conditions. A fairly dramatic comparative example is shown in FIG. 9, which documents the morphology of cells treated with 10 μM all-trans-retinoic acid (tRA). This treatment caused marked morphologic changes in the keratinocytes, but had no apparent effect on the morphology of the fibroblasts. Therefore it is surprising that the more responsive keratinocytes were less predictive than fibroblasts with respect to photo-aging.

Example 5

Theme Analysis of Age-Related Gene Expression Signatures

Theme analysis was used as a tool to understand better the results obtained using BJ fibroblasts instances. Theme analysis is a statistical analysis-based method for detecting biological patterns in gene expression profiling data. The method uses an ontology of controlled vocabulary terms developed by the Gene Ontology (GO) Consortium [Ashburner, M. et al. (2000) Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat. Genet, 25, 25-29] that describes the biological processes, molecular functions and cellular components associated with gene products. Analysis involves statistical comparison of a regulated list of genes and a larger reference list of all the expressed genes, to determine if genes annotated to specific GO terms are significantly enriched in the regulated list. This analysis reveals biological patterns when multiple genes associated with a given GO term occur on the regulated list at a frequency greater than expected by chance. Such analysis was performed using Theme Extractor proprietary software and an algorithm that calculates the p value for each ontology term. Data were analyzed for statistical significance by the Fisher's exact test. The approach used here and statistical methods are very similar to Gene Set Enrichment Analysis, which has been described in the literature [Subramanian, A. et al. (2005) Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad Sci U.S.A, 102, 15545-15550].

Table 4 shows GO Biological Process terms that are significantly enriched in the comparison of the older arm to younger arm (photo-aging) results from the clinical gene expression study described in Example 3. This involved the analysis of approximately 12,500 expressed genes. Only the most highly significant themes are shown (p≦1×10⁻⁴), and the theme analysis was done separately for the up- and down-regulated genes. The level of indentation in the terms column generally indicates the level in the GO hierarchy and parent/child relationships between terms. In other words, indented terms often are related to the term above them and describe a more specific process. The numbers of asterisks indicate the p value ranges for each term.

TABLE 4
Theme analysis of genes regulated in photo-aging.
	Old Arm to Young	Terms
	Arm Comparison 1	related to
Gene Ontology Biological Process Terms	Up	Down	dermis
. . . GO:0006956 complement activation	*
. GO:0032501 multicellular organismal process	****
. . GO:0007275 multicellular organismal development	***
. . . GO:0035295 tube development	*
. . . . GO:0030323 respiratory tube development	*
. . . . . GO:0048754 branching morphogenesis of a tube	*
. . . GO:0048731 system development	***
. . . . GO:0048513 organ development	*
. . . . . GO:0048568 embryonic organ development	*
. . . . . GO:0009887 organ morphogenesis	*
. . . . . GO.0009888 tissue development	*
. . . . GO:0007399 nervous system development	**
. . . . GO:0001501 skeletal system development	**		X
. . . . . GO:0048706 embryonic skeletal system development	*		X
. . . GO:0048736 appendage development	**		X
. . . . GO:0035107 appendage morphogenesis	*		X
. . . . GO:0060173 limb development	*		X
. . . GO:0007389 pattern specification process	*
. GO:0032502 developmental process	***
. . GO:0048856 anatomical structure development	***
. . . GO:0009653 anatomical structure morphogenesis	***
. . . GO:0016477 cell migration	*
. . . . . . GO:0002541 activation of plasma proteins involved in acute inflammat...	*
. . . GO:0009611 response to wounding	****		X
. . . . GO:0042060 wound healing	****		X
. . GO:0009605 response to external stimulus	****
. . . GO:0019222 regulation of metabolic process		*
. . . . . . . GO:0045941 positive regulation of transcription		*
. . . . . . . . GO:0045893 positive regulation of transcription, DNA-dependent		*
. . . . . . . . . GO:0045944 positive regulation of transcription from RNA polym...		*
. . . . . . . GO:0006355 regulation of transcription, DNA-dependent		*
. . . . . . . . GO:0045892 negative regulation of transcription, DNA-dependent		*
. . . . . . . . . GO:0000122 negative regulation of transcription from RNA polym...		**
. . . . . . . . GO:0006357 regulation of transcription from RNA polymerase II pr...		**
. . . . . . . GO:0016481 negative regulation of transcription		*
. . . . . . . GO:0051253 negative regulation of RNA metabolic process		*
. . . . . . GO:0051252 regulation of RNA metabolic process		*
. . . . . . . GO:0051254 positive regulation of RNA metabolic process		*
. . . . . . GO:0045935 positive regulation of nucleobase, nucleoside, nucleotide...		**
. . . . . GO:0031324 negative regulation of cellular metabolic process		*
. . . . . . GO:0031327 negative regulation of cellular biosynthetic process		**
. . . . . GO:0031325 positive regulation of cellular metabolic process		*
. . . . . . GO:0031328 positive regulation of cellular biosynthetic process		*
. . . . . GO:0051173 positive regulation of nitrogen compound metabolic process		**
. . . . GO:0060255 regulation of macromolecule metabolic process		*
. . . . . . GO:0010628 positive regulation of gene expression		*
. . . . . . GO:0010629 negative regulation of gene expression		*
. . . . . . GO:0010557 positive regulation of macromolecule biosynthetic process		*
. . . . . . GO:0010558 negative regulation of macromolecule biosynthetic process		*
. . . . . GO:0010604 positive regulation of macromolecule metabolic process		*
. . . . . GO:0010605 negative regulation of macromolecule metabolic process		*
. . . . . GO:0009890 negative regulation of biosynthetic process		**
. . . . . GO:0009891 positive regulation of biosynthetic process		*
. . . . GO:0009892 negative regulation of metabolic process		*
. . . . GO:0048522 positive regulation of cellular process		*
. . . . . . GO:0002053 positive regulation of mesenchymal cell proliferation	*		X
. . . . GO:0048634 regulation of muscle organ development	*		X
. . . . . GO:0016202 regulation of striated muscle tissue development	*		X
. . . . . GO:0010464 regulation of mesenchymal cell proliferation	**		X
. . . . GO:0007165 signal transduction	*	*
. . . . . GO:0009966 regulation of signal transduction		*
. . GO:0043062 extracellular structure organization	*
. . . GO:0030198 extracellular matrix organization	**		X
. . . . . GO:0006350 transcription		*
. . GO:0006928 cellular component movement	*
. . . . GO:0006325 chromatin organization		*
. . . . . GO:0016568 chromatin modification		**
. . GO:0007154 cell communication	*
. . . . . GO:0016485 protein processing	*
. . . . . . GO:0051605 protein maturation by peptide bond cleavage	*
1 Statistical significance is as follows.
P-Value Legend
1.00E−4<	p
1.00E−5<	p	<=1.00E−4	*
1.00E−6<	p	<=1.00E−5	**
1.00E−7<	p	<=1.00E−6	***
	p	<=1.00E−7	****

This analysis revealed that up- and down-regulated genes in photo-aging were associated with quite distinct biological processes. A number of biological processes closely associated with fibroblasts and connective tissue were up-regulated in photo-aging. It is notable that the Gene Ontology terms “response to wounding” and “wound healing” were among the most statistically significant terms (p≦1×10⁻⁷) for the up-regulated gene list. Other terms clearly associated with fibroblasts or mesenchymal cells are marked in Table 4 and these all were significant for the up list. Most of these other terms associated with the up-regulated list relate to developmental processes in which mesenchymal cells play important roles. The genes down-regulated in photo-aging were associated with more general processes such as gene transcription and various aspects of metabolism and signal transduction that relate to various cell types.

The Photo-aging 600 signature gene lists (see Example 2) were also subject to theme analysis (not shown), which revealed a pattern of significant terms quite similar to that obtained with the entire photo-aging data set. The up-regulated gene list contains genes involved in biological processes closely associated with fibroblasts or mesenchymal cells. Again among the most significant terms for the up list included “response to wounding” and “wound healing” (P≦1×10⁻⁵). The results of these theme analyses provide a clear rationale for the specific linkage of fibroblasts to the photo-aging up-regulated gene expression signature and the strong linkage to the overall signature.photo-aging It appears that the dominant processes that are up-regulated in photo-aging involve the dermis and fibroblasts. Overall, the photo-aging gene signatures reflect this dominant biology observed in the gene expression profiling of the photo-aging skin condition.

Theme analysis was also performed on the genes regulated in intrinsic aging, which involved approximately 12,500 expressed genes. Table 5 shows GO Biological Process terms that are significantly enriched in the comparison of the older buttock to younger buttock (intrinsic aging) from the clinical gene expression study described in Example 3. Similarly with respect to Table 4, only the most highly significant themes are shown (p≦1×10⁴) to keep the size of this table manageable.

TABLE 5
Theme analysis of genes regulated in intrinsic aging
	Old Buttock to
	Young Buttock	Terms
	Comparison 1	related to
Gene Ontology Biological Process Terms	Up	Down	epidermis
. GO:0032501 multicellular organismal process	*
. . . GO:0022600 digestive system process		*
. . . GO:0003012 muscle system process	*
. . GO:0007275 multicellular organismal development	*
. . . GO:0048731 system development	***
. . . . GO:0048513 organ development	***
. . . . . . GO:0007398 ectoderm development		*	X
. . . . . . . GO:0021846 cell proliferation in forebrain		*
. . . . GO:0007399 nervous system development		*
. . . . . GO:0007417 central nervous system development		**
. . GO:0007586 digestion		*
. . GO:0001503 ossification	*
. GO:0032502 developmental process	*
. . GO:0048856 anatomical structure development	**
. . . GO:0060021 palate development	*
. . . GO:0061061 muscle structure development	*
. . . . . GO:0043434 response to peptide hormone		**
stimulus
. . . . GO:0042127 regulation of cell proliferation	*
. . . GO:0042180 cellular ketone metabolic process		*
. . . . GO:0043436 oxoacid metabolic process		*	X
. . . . . GO:0019752 carboxylic acid metabolic process		*	X
. . . . . . . GO:0006631 fatty acid metabolic process		*	X
. . . . . . . . GO:0006633 fatty acid biosynthetic process		*	X
. . . . . . GO:0046394 carboxylic acid biosynthetic process		*	X
. . . . GO:0042438 melanin biosynthetic process		*
. . . . GO:0051188 cofactor biosynthetic process		*
. . . . GO:0016053 organic acid biosynthetic process		*	X
. . . GO:0044255 cellular lipid metabolic process		***	X
. . . GO:0051186 cofactor metabolic process		**
. . . . GO:0006732 coenzyme metabolic process		*
. . . GO:0006082 organic acid metabolic process		*	X
. . GO:0019748 secondary metabolic process		*
. . . GO:0006582 melanin metabolic process		*	X
. . . GO:0006629 lipid metabolic process		****	X
. . . . GO:0008202 steroid metabolic process		*	X
. . . . . GO:0006694 steroid biosynthetic process		**	X
. . . . . . GO:0016126 sterol biosynthetic process		**	X
. . . . . . . GO:0006695 cholesterol biosynthetic process		**	X
. . . . . GO:0016125 sterol metabolic process		*	X
. . . . GO:0008610 lipid biosynthetic process		****	X
. . GO:0044281 small molecule metabolic process		****
. . . GO:0044283 small molecule biosynthetic process		*
. . . GO:0006066 alcohol metabolic process		*
. . GO:0055114 oxidation reduction		**
1 Statistical significance is as described in Table 4.

The theme analysis of the genes regulated in intrinsic aging, like that for photo-aging, revealed that the up- and down-regulated genes were associated with very different processes. The down-regulated genes in intrinsic skin aging were associated with a number of processes related to the epidermis and keratinocytes. Most dominant were terms related to lipid biosynthetic pathways that are involved in the synthesis of lipids important in epidermal barrier function. A more specific term related to epithelial biology, “ectoderm development,” was associated with the down-regulated genes. The related term “epidermal development” (p<0.001) was not included in the table because it did not make the p value cutoff, but it was also associated with the down-regulated genes. Terms related to regulation of pigmentation (melanogenesis), which is an epidermal function, were also significant for the down list analysis. As seen in the analysis of photo-aging, up-regulated themes included various developmental processes related to mesenchymal cells. The appearance of terms such as “ossification,” which may seem irrelevant to skin, was due to the fact that various regulatory pathways and related genes are involved in a number of processes associated with cell lineages. Bone matrix cells are of mesenchymal origin. The term “regulation of mesenchymal cell proliferation” (highly significant in the photo-aging analysis) was not included in the table because it did not make the p value cutoff, but was significant at p<0.001 for the up list in intrinsic aging.

The Intrinsic Aging 600 signature gene lists (described in Example 3 and defined as the top 300 up- and down-regulated genes set forth in Tables A and B) were also subjected to theme analysis (not shown), which revealed a pattern of significant terms similar to that obtained with the entire intrinsic aging data set. The up-regulated gene list contains genes involved in biological processes associated with fibroblasts or mesenchymal cells, and the down-regulated list contained genes more closely associated with epidermal processes and keratinocytes. The results of these theme analyses suggest a rationale for the strong connectivity of fibroblasts to the intrinsic aging up-regulated gene expression signature and their preferential linkage to the overall signature, as well as the preferential connectivity of keratinocyte instances to the down-regulated gene expression signature. Overall, the intrinsic aging signatures reflect the biology in intrinsic aging.

The biological theme analyses performed on the genes regulated in photo-aging and intrinsic aging demonstrate differences in the dominant biology in these two aspects of skin aging. The major differences related to an apparently greater down-regulation of processes associated with keratinocyte differentiation in intrinsic aging. These analyses further support the value of using signatures for both photo-aging and intrinsic aging in querying the C-map database to identify cosmetic materials to treat aging skin. These analyses also further support the non-intuitive finding that fibroblasts cell lines are more useful than keratinocyte cell lines for identifying cosmetic agents to improve photo-aged skin. The studies described herein also demonstrate that cell type-specific signals can be detected from full thickness biopsies using the C-map method and system.

Example 6

Connectivity Map Results for Artichoke Leaf Extract and Carob Seed Extract

In this example, 20 materials that were candidates for an anti-aging facial benefit study were tested in fibroblasts and keratinocytes. A microarray analysis was conducted for each sample compound and ordered lists of the probe set IDs were stored as instances in a database. Photo-aging and intrinsic aging gene expression signatures were compared against the instances and connectivity scores were derived.

Of the 20 candidates screened, 2 showed consistent connections to the photo-aging gene expression signatures when tested in BJ fibroblasts. The detailed results for these materials, artichoke leaf extract and carob seed extract, are shown in Table 6. Since both intrinsic aging and photo-aging are considered adverse conditions, it is preferential to find materials that tend to reverse one or both of the gene expression signatures. Such materials will have negative connectivity score, and the more negative the score the stronger the negative connection. It can be seen in Table 6 that all fibroblast instances of artichoke leaf extract consistently linked negatively to the 3 photo-aging gene expression signatures. Artichoke leaf extract also consistently scored negatively with the intrinsic aging signatures when tested in fibroblasts, but the scores were weaker and there were some null or zero scores. Null scores are not considered inconsistent with other scores all showing the same directionality. When tested in keratinocytes, artichoke leaf extract showed generally weak positive connections to both the photo-aging and intrinsic aging gene expression signatures. Therefore, the testing in fibroblasts and keratinocytes gave inconsistent results. Similarly, carob seed extract fibroblast instances showed a pattern of negative connections to the photo-aging gene expression signatures and inconsistent results were obtained with keratiocyte instances with both sets of gene signatures.

TABLE 6
C-MAP scores for two cosmetic supplier materials, Biobenefity ™ and
Glyco-repair ™, tested in fibroblasts and keratinocytes.
	Cell Line	Intrinsic Aging Clinical Signatures	Photo-aging Clinical Signatures
		Utilized for	Intrinsic	Intrinsic	Intrinsic	Photo-aging	Photo-aging	Photo-aging
GeneChip ID	Material1	Instance Data	200	400	600	200	400	600
CMP_62_13	Artichoke	BJ fibroblasts	−0.403	0	−0.333	−0.551	−0.553	−0.547
CMP_62_14	Leaf		−0.211	−0.198	−0.235	−0.251	−0.298	−0.299
CMP_65_51	Extract2		0	−0.222	0	−0.31	−0.264	−0.295
CMP_61_13		Telomerized-	0	0	0	0	0	0.19
CMP_61_14		keratinocytes	0.401	0.346	0.316	0.293	0.268	0.284
CMP_68A_27			−0.251	0	0.137	0.209	0.156	0.149
CMP_62_23	Carob	BJ fibroblasts	0	0	0	−0.342	−0.288	−0.325
CMP_62_24	Seed		0	0	0	−0.41	−0.327	−0.325
CMP_65_55	Extract3		0	0	0.151	0	0	0
CMP_61_23		Telomerized-	−0.321	−0.286	−0.248	−0.44	−0.438	−0.404
CMP_61_24		keratinocytes	0.27	0.25	0.226	−0.374	−0.349	0
CMP_68A_31			0.287	0.263	0.268	0.347	0.348	0.346
1Test compositions were prepared using a concentration of 0.01% of one of artichoke leaf extract or carob seed extract.
2One example of Artichoke leaf extract is available from Ichimaru Pharcos, Japan under the tradename Biobenifity ™
3One example of carob seed extract is available from Silab, France under the tradename Glyco-Repair ™

In view of the discovery that there is a strong preferential linkage of the skin aging gene expression signatures to instances derived from fibroblasts, additional gene expression signatures optimized for the fibroblast linkage were also generated to screen the cosmetic materials. Only the photo-aging data from the aging skin genomics study in Example 2 were used because the fibroblast linkage to the photo-aging gene expression signature was stronger than that to intrinsic aging gene expression signature. The modified gene expression signatures were generated similarly to the photo-aging signatures in Example 3, except that the microarray probe sets were also filtered for genes selected from a microarray analysis of BJ fibroblast cells that had ≧10% Present calls, wherein the microarray analysis of the BJ fibroblast cells was derived from data from 30 control and chemically-treated cultures. This filter was added so that the modified gene expression signature would only contain probes complementary to genes actually expressed in the fibroblasts. Additionally, the probes used in the modified gene expression signature were identified using the following filters.

• • (i) “Signature 1” Filtered on log t-test rank; best ranked 100 up-regulated and 100 down-regulated probe sets in the arm old to young comparison.
  • (ii) “Signature 2” Filtered on log t-test rank; best ranked 150 up-regulated and 150 down-regulated probe sets in the arm old to young comparison.
  • (iii) “Signature 3” (identifiers set forth in Table I) Filtered on log t-test rank; best ranked 200 up-regulated and 200 down-regulated probe sets in the arm old to young comparison.

TABLE 7
Connectivity scores derived from a comparison of artichoke leaf
extract and carob seed extract instances tested in fibroblast cells
compared to the fibroblast-optimized photo-aging signatures.
	Photo-aging signatures
	optimized for fibroblasts
GeneChip			Signa-	Signa-	Signa-
ID	Material 1	Cell Line	ture 1	ture 2	ture 3
CMP_62_13	Artichoke	BJ	−0.721	−0.707	−0.684
CMP_62_14	leaf	fibroblasts	−0.341	−0.366	−0.375
CMP_65_51	extract		−0.417	−0.38	−0.365
CMP_62_23	Carob	BJ	−0.482	−0.43	−0.415
CMP_62_24	seed	fibroblasts	−0.48	−0.41	−0.416
CMP_65_55	extract		0	0	0
1 The C-map instances are the same as in Table 6. Only results with fibroblasts are shown.

The results set forth in Table 7 are similar to those for the photo-aging signatures shown in Table 6 except that the scores are stronger.

An overall weighting of the C-map scores was used to determine whether the materials tested were considered hits against the gene expression signatures. For each signature in a category (e.g. photo-aging or intrinsic aging), the hits were identified and given a weight using the criteria in Table 8. The desired directionality of the scores depends on whether the analysis is intended to identify materials that will ameliorate a condition like photo-aging (negative scores) or that are similar to a known benefit agent (positive scores). This heuristic approach was applied because of the limited number of replicates available for the materials in the C-map database.

TABLE 8
Weight Criteria
3      At least 2 instances in the top 5% of all C-map
       instances1 based on C-map score,
       all instances with correct directionality or null.
       Non-null instances >50% .
2      At least 1 instance in the top 5% of all C-map
       instances based on C-map score, all instances with
       correct directionality or null. Non-null instances >50%.
2      At least 2 instances in the top 10% of all C-map
       instances based on C-map score, all instances with
       correct directionality or null. Non-null instances >50%.
1      1 instance in the top 5% of all C-map instances,
       all instances with correct directionality or null.
       Non-null instances ≧40% AND ≦50%..
1There were more than 4000 chemical instances in the C-map database when these analyses were done.

The average weight was calculated for each set of signatures (e.g. the 3 photo-aging signatures described in Example 2 and derived from Tables C and D). Materials were considered a hit if their average weights across signatures were as described in Table 9.

TABLE 9
Weighting of hits Average hit weig
Strong hit        value = 3
Hit               2 ≦ value < 3
Weak hit          1 < value < 2
Weak linkage      0 < value ≦ 1
indicates data missing or illegible when filed

The results of the weighted analysis for artichoke leaf extract and carob seed extract are shown in Table 10. Based on the average hit weights artichoke leaf extract was considered an overall negative hit against the photo-aging signatures and a weak negative hit against the intrinsic aging signatures when tested in fibroblasts. Carob seed extract was a weak negative hit against the photo-aging signatures optimized for fibroblasts. Considering the strong preferential linkage of the photo-aging signatures to instances derived from fibroblasts (described above), the keratinocyte data were discounted and only the fibroblast data were used as part of the weight of evidence to determine whether artichoke leaf extract and carob seed extract should be advanced to in vivo testing.

TABLE 10
Weighting of the C-map scores from Table
6 and Table 7 to determine hits.
	Average hit weights
		Intrinsic	Photo-
		Aging	aging	Fibroblast
		Signatures	Signatures	Optimized
		(I200,	(P200,	Photo-aging
		I 400,	P400,	Signatures
Material	Cell Line	I600)	P600)	(Signatures 1-3)
Artichoke	BJ fibroblasts	1.33	2	2
leaf	Telomerized-	0	0	0
extract	keratinocytes
Carob seed	BJ fibroblasts	0	0	1.33
extract	Telomerized-	0	0	0
	keratinocytes

Example 7

In Vitro and In Vivo Results for Artichoke Leaf Extract and Carob Seed Extract

C-map is a hypothesis-generating tool whose predictiveness for identifying agents that can improve aging skin should be validated through clinical testing. Artichoke leaf extract and carob seed extract were subjected to in vitro testing to establish a weight of evidence before submitting them for clinical tests. Skin equivalent cultures were treated with the materials as described below and assayed for endpoints related to cosmetic benefits including; procollagen, hyaluronic acid, fibronectin, and inhibition of matrix metalloprotease 1 (MMP1) activity.

In Vitro Testing Procedures:

Human skin equivalent cultures (EFT-400 Full-Thickness Skin Model, MatTek Corporation, Ashland Mass.) were equilibrated overnight at 37° C. and 5% CO₂. Cultures were treated topically with 40 μl of test material for 24 hours. The test materials evaluated were aqueous solutions of 3.0% of a commercial preparation of hydrolyzed carob seed extract or 3.5% of a commercial preparation of artichoke leaf extract. Control cultures were treated with water vehicle alone. After treatment with test material, the cultures were rinsed with PBS. A 5-mm punch biopsy sample was taken for procollagen and hyaluronic acid measurements and the remaining quantities of each culture were used for cell viability (MTT) analysis.

Cultures to be tested for cell viability were transferred to new six-well plates containing 2 ml of MTT solution (MTT kit, MatTek Corporation) in each well and incubated for 3 hours. Cultures were then removed from wells, blotted dry, and transferred to new six-well plates containing 3 ml of extraction solution in each well. One ml of extraction solution was added topically to each culture, and the plates were placed on a shaker for 2 hours at room temperature. An aliquot (200 μl) of extraction solution was removed from each culture well, transferred to a 96-well flat bottom plate, and read at A₅₇₀.

Culture samples for procollagen and hyaluronic acid analysis were incubated with 1 ml of mild protein extraction reagent (T-per, Pierce Protein Research Products, Thermo Fisher Scientific Inc., Rockford, Ill.), then homogenized with a mixer mill (Model MM 300, Qiagen Inc., Valencia, Calif.) for 6 minutes. Samples were then centrifuged at 10,000 rpm for 15 minutes at 4° C. and the soluble fractions collected. Supernatants were analyzed with a Micro BCA Protein Assay Kit (Pierce Protein Research Products, Thermo Fisher Scientific Inc., Rockford, Ill.) to quantify protein concentrations. For analysis of procollagen 1, supernatant samples were diluted 1:5 in assay buffer and analyzed with a commercially available ELISA kit (Takara Bio Inc., Shiga, Japan). For hyaluronic acid analysis, supernatant samples were diluted 1:300 in assay buffer and analyzed with a commercially available ELISA kit (Corgenix, Broomfield Colo.). Procollagen and hyaluronic acid were normalized to protein levels for each culture, and expressed as a % of the vehicle control. Statistical comparisons to vehicle control were made using the Students t-test.

In Vitro Testing Results:

TABLE 11
In vitro Skin Biomarker Responses
	Level	Hyaluronic Acid	Procollagen	Other
Compound	(%)	(% control)	(% control)	(% control)
Carob Seed	3.0	139*	354*	—
Extract
Artichoke Leaf	3.5	160*	270*	MMP-1: 64*
Extract
*indicates p < 0.05; nc = no change vs Control

These results indicate that human skin cultures treated with 3.0% of a commercial preparation of hydrolyzed carob seed extract significantly (p<0.05) increased expression of procollagen (the precursor of dermal matrix collagens) and hyaluronic acid (a matrix component that binds water and hydrates the skin). Cultures treated with 3.5% of a commercial preparation of artichoke leaf extract produced significantly increased expression of procollagen and hyaluronic acid, and also significantly reduced MMP-1 expression (an enzyme increased in skin aging and inflammation that damages dermal matrix).

In Vivo Testing Procedures

The study design was a 13-week, randomized, double-blinded, vehicle controlled, split-face study to evaluate fine lines and wrinkles in 40 to 65 year old women, Fitzpatrick skin type I to III, with moderate to moderately-severe photo-aged facial skin. The duration of the study included 1-week preconditioning with vehicle product, followed by 12 weeks of test product application to the facial skin twice each day. High density digital images of subject facial skin were captured with a Fuji S2 Pro digital SLR camera with a 60 mm Nikon lens. Images were taken at baseline, 4, 8 and 12 weeks. Coded images were evaluated on a 0 to 8 grading scale by expert graders to determine the degree of change in eye area fine lines and wrinkles at 4, 8 or 12 weeks as compared to the matching baseline image for each subject.

The test products used by the panelists contained a commercial preparation of 3.0% hydrolyzed carob seed extract or 3.5% artichoke leaf extract, each prepared in an oil-in-water formulation. The control was the oil-in-water formulation alone.

In vivo Test Results:

TABLE 12
Eye Area Fine Lines & Wrinkle Grades at 4, 8 and 12 Weeks of Test Product Use
	8 Weeks	12 Weeks
	Level	FLW	Statistical	p value vs	FLW	Statistical	p value vs
Facial Product	(%)	Grade	Grouping	Vehicle	Grade	Grouping	Vehicle
Vehicle	—	0.137	a	—	0.230	a	—
Carob Seed extract	3.0%	0.741	bc	0.0302	0.799	bc	0.0213
Artichoke Leaf	3.5%	0.503	ab	0.1207	0.924	c	0.0059
extract

These results indicate that test product containing 3.0% of a commercial preparation of carob seed extract produced a significant (p<0.05) improvement in eye area fine lines and wrinkles as compared to the concurrent vehicle control product at 4, 8 and 12 weeks. Test product containing 3.5% of a commercial preparation of artichoke leaf extract produced a significant improvement in eye area fine lines and wrinkles as compared to the concurrent vehicle control product at 12 weeks. Overall, the in vivo test results demonstrate the predictiveness of the methods described here to identify skin anti-aging agents.

Every document cited herein is hereby incorporated herein by reference in its entirety unless expressly excluded or otherwise limited. The citation of any document is not an admission that it is prior art with respect to any invention disclosed or claimed herein or that it alone, or in any combination with any other reference or references, teaches, suggests or discloses any such invention. Further, to the extent any meaning or definition of a term in this document conflicts with any meaning or definition of the same term in a document incorporated by reference, the meaning or definition assigned to that term in this document shall govern.

The values disclosed herein are not to be understood as being strictly limited to the exact numerical values recited. Instead, unless otherwise specified, each such value is intended to mean both the recited value and a functionally equivalent range surrounding that value.

The present invention should not be considered limited to the specific examples described herein, but rather should be understood to cover all aspects of the invention. Various modifications, equivalent processes, as well as numerous structures and devices to which the present invention may be applicable will be readily apparent to those of skill in the art. Those skilled in the art will understand that various changes may be made without departing from the scope of the invention, which is not to be considered limited to what is described in the specification.

Claims

1. A method for constructing a data architecture for use in identifying connections between perturbagens and genes associated with one or more skin aging conditions, comprising:

(a) providing a gene expression profile for a control human dermal fibroblast cell;

(b) generating a gene expression profile for a human dermal fibroblast cell exposed to at least one perturbagen, wherein at least one perturbagen comprises a cosmetic agent;

(c) identifying genes differentially expressed in response to the at least one perturbagen by comparing the gene expression profiles of (a) and (b);

(d) creating an ordered list comprising identifiers representing the differentially expressed genes, wherein the identifiers are selected from the group consisting of gene names, gene symbols, microarray probe set ID values, and combinations thereof, and the identifiers are ordered according to the differential expression of the genes, and wherein each identifier in the ordered list is in association with a numerical ranking corresponding to its rank in the ordered list;

(e) storing the ordered list as a fibroblast instance on at least one computer readable medium; and

(f) constructing a data architecture of stored fibroblast instances by repeating steps (a) through (e) for between about 50 and about 50,000 instances, wherein the at least one perturbagen of step (a) is different for each fibroblast instance;

wherein a programmable computer performs one or more of steps (c), (d), (e) and (f).

2. A method according to claim 1, wherein the step of generating is performed by extracting a biological sample comprising mRNA from the treated cell and subjecting the biological sample to microarray analysis.

3. A method according to claim 1, wherein the ordered list is arranged so that an identifier associated with a most up-regulated gene is positioned at the top of the ordered list and an identifier associated with a most down-regulated gene is positioned at the bottom of the ordered list.

4. A method according to claim 1, further comprising;

(g) providing a gene expression profile for a control human keratinocyte cell;

(h) generating a gene expression profile for a human keratinocyte cell exposed to at least one perturbagen;

(i) identifying genes differentially expressed in response to the at least one perturbagen by comparing the gene expression profiles of (g) and (h);

(j) creating an ordered list comprising identifiers representing the differentially expressed genes, wherein the identifiers are ordered according to the differential expression of the genes identified in (i);

(k) storing the ordered list created in step (j) as a keratinocyte instance on the at least one computer readable medium; and

(l) constructing a data base of stored keratinocyte instances by repeating (g) through (k), wherein the at least one perturbagen of step (h) is different for each keratinocyte instance.

5. A method according to claim 4, wherein the at least one perturbagen of step (a) is the same as the at least one perturbagen of step (g).

6. A method according to claim 1, wherein the at least one perturbagen is a botanical derived from one or more of a root, stem, bark, leaf, seed, or fruit of a plant.

7. A method according to claim 1, wherein the at least one perturbagen is selected from the group consisting of a vitamin compound, a sugar amine, a phytosterol, hexamidine, a hydroxy acid, a ceramide, an amino acid, and a polyol.

8. A method according to claim 7, wherein the vitamin compound is selected from the group consisting of a vitamin B3 compound, a vitamin B5 compound, a vitamin B6 compound, a vitamin B9 compound, a vitamin A compound, a vitamin C compound, a vitamin E compound, and derivatives and combinations thereof.

9. A method according to claim 7, wherein the vitamin compound is selected from the group consisting of retinol, retinyl esters, niacinamide, folic acid, panthenol, ascorbic acid, tocopherol, and tocopherol acetate.

10. A method for implementing the data architecture according to claim 1 to generate connections useful for identifying cosmetic agents effective for treating aged skin, the method comprising querying the data architecture with at least one skin aging gene expression signature, wherein querying comprises comparing the at least one skin aging gene expression signature to each stored fibroblast instance, wherein the skin aging gene expression signature represents genes differentially expressed in association with at least one skin aging condition.

11. A method according to claim 10, wherein the at least one skin aging condition is selected from intrinsic skin aging conditions; photo-aging skin conditions; and combinations thereof.

12. A method according to claim 10, wherein the at least one skin aging gene expression signature is constructed by a method comprising (i) identifying genes having up-regulated expression in the at least one skin aging condition when compared to a control; (ii) identifying genes having down-regulated expression in the at least one skin aging condition when compared to a control; (iii) creating one or more gene expression signature lists associated with the at least one skin aging gene expression signature comprising identifiers corresponding to a plurality of the genes identified in (i) and (ii); and storing the one or more gene expression signature lists on the at least one computer readable medium.

13. A method according to claim 12, wherein the number of genes having up-regulated expression in the at least one skin aging condition is between about 10 and about 400, and the number of genes down-regulated in the at least one skin aging condition is between about 10 and about 400.

14. A method according to claim 13 wherein the identifiers for from between about 80% and about 100% of the up-regulated genes are set forth as in Table C and wherein identifiers for from between about 80% and about 100% of the down-regulated genes are set forth in Table D.

15. A method according to claim 12, wherein the one or more gene expression signature lists comprises a first list representing a plurality of the up-regulated genes identified in (i) and a second list representing a plurality of down-regulated genes identified in (ii).

16. A method according to claim 12, wherein at least one skin sample is taken from a human subject exhibiting the at least one skin condition, a biological sample is extracted from the skin sample, and a gene expression profile of the at least one skin sample is generated prior to at least one of the steps (i) and (ii).

17. A method according to claim 16, wherein the skin sample comprises cells from epidermal and dermal layers of the human subject.

18. A method according to claim 10, wherein the comparison further comprises assigning a connectivity score to each of plurality of instances, wherein a plurality of connectivity scores represents a positive correlation and a plurality of the connectivity scores represents a negative correlation, and wherein the connectivity score has a value between +2 and −2.

19. A system for identifying connections between perturbagens and genes associated with one or more skin aging conditions, comprising:

(a) at least one computer readable medium having stored thereon a plurality of instances, and at least one skin aging gene expression signature, wherein the instances and the at least one gene expression signature are derived from either a human dermal fibroblast cell or a human keratinocyte cell or both, wherein each instance comprises an instance list of rank-ordered identifiers of differentially expressed genes, and wherein the at least one skin aging gene expression signature comprises one or more gene expression signature lists of identifiers representing differentially expressed genes associated with a skin aging condition;

(b) a programmable computer comprising computer-readable instructions that cause the programmable computer to execute one or more of the following: (i) accessing the plurality of instances and the at least one skin aging gene expression signature stored on the computer readable medium; (ii) comparing the at least one skin aging gene expression signature to the plurality of the instances, wherein the comparison comprises comparing each identifier in the gene expression signature list with the position of the same identifier in the instance list for each of the plurality of instances; and (iii) assigning a connectivity score to each of the plurality of instances.

20. A system according to claim 19, further comprising: a microarray scanner for receiving a sample comprising human dermal fibroblast cells and/or human keratinocyte cells; and a second programmable computer for transmitting gene expression data from the scanner to the first programmable computer, and an array of perturbagens for application to the human dermal fibroblast cells and/or the keratinocyte cells.