Mechanisms of Photoaging and Chronological Skin Aging

Ultraviolet irradiation from the sun damages human skin, causing it to age prematurely. This premature aging process (photoaging) is cumulative with sun exposure, and preferentially affects individuals with lighter skin color. During the last 10 years, substantial progress has been made in understanding the molecular mechanisms responsible for photoaging in human skin. One major conceptual advance elaborated by this work is that UV irradiation invokes a complex sequence of specific molecular responses that damage skin connective tissue. These molecular processes derive from the ability of UV irradiation to exploit highly evolved cellular machinery that regulates responses of cells to physiological and environmental extracellular stimuli. The cellular machinery that mediates UV damage to human skin connective tissue includes cell surface receptors, protein kinase signal transduction pathways, transcription factors, and enzymes that synthesize and degrade structural proteins in the dermis that confer strength and resiliency to skin. Recruitment by UV irradiation of the cellular machinery that damages skin connective tissue is initiated by photochemical generation of reactive oxygen species (ROS). Ultraviolet-induced ROS also cause direct deleterious chemical modifications to cellular components (ie, DNA, proteins, and lipids). Chemical oxidation of cellular components and activation of cellular machinery, both brought about by UV-induced oxidative stress, act in concert to cause photoaging. Insights gained into the molecular basis of photoaging provide new opportunities for therapeutic intervention aimed at prevention.

Signal transduction pathways that are activated by UV irradiation in human skin are depicted in Figure 1. Ultraviolet irradiation causes activation of cell surface cytokines and growth factor receptors. Epidermal growth factor (EGF), interleukin (IL) 1, and tumor necrosis factor α (TNF-α) receptors are activated within 15 minutes following UV exposure (twice the minimal erythema dose) in human skin in vivo. Functional activation of these receptors requires stimulation of distinct tyrosine kinase activities.¹ This addition of phosphate groups to tyrosine residues on the receptors and their associated adaptor proteins is the initial biochemical event in receptor activation and provides specific docking sites for molecules involved in signal propagation within the cell.²

The primary mechanism by which UV irradiation initiates molecular responses in human skin is by photochronical generation of ROS. These ROS include superoxide anion, peroxide, and singlet oxygen. The mechanism(s) of receptor activation by UV irradiation are not well understood. One possibility, which is supported by indirect experimental evidence, is that photochemical generation of ROS oxidize, and thereby inhibit, specific protein-tyrosine phosphatases, which function in opposition to receptor-activated protein-tyrosine kinases by removing phosphate group from receptors or their associated adaptor proteins.³ The result would be a net increase in receptor phosphorylation (activation).⁴ This mechanism of UV activation of cell surface receptors is supported by several publications,^{5 - 6} although direct evidence that protein-tyrosine phosphatases regulate the activation state of cell surface receptors is lacking.

Irrespective of mechanism, UV irradiation activates cell surface receptors, as does ligand binding, and triggers downstream signal transduction pathways. Several studies have demonstrated that UV irradiation stimulates production of hydrogen peroxide, which induces multiple signaling pathways, although the mechanism(s) by which hydrogen peroxide acts is not clear. In human skin in vivo and human keratinocytes, hydrogen peroxide levels are increased within 15 minutes following UV irradiation and continue to accumulate for approximately 60 minutes after UV exposure (Figure 2A). It is important to appreciate that this generation of hydrogen peroxide after UV exposure is distinct from photochemical generation of ROS described above, which occurs only during UV exposure and abates following UV exposure.

Phagocytes contain a multi-subunit complex, nicotinamide adenine dinucleotide phosphate (NADPH) oxidase, which catalyzes the reduction of molecular oxygen to superoxide anion.⁷ The superoxide produced is quantitatively converted to hydrogen peroxide, which is less damaging to cells and serves as a cosubstrate for peroxidases. In turn, hydrogen peroxide can be converted to other ROS, including hydroxyl radical and singlet oxygen. Human skin and human keratinocytes also express NADPH oxidase subunits, and contain NADPH oxidase activity that is induced following UV exposure. In keratinocytes, NADPH oxidase activity is induced 2-fold within 20 minutes following UV exposure (Figure 2B). Interestingly, pharmacological inhibition of UV-induced NADPH oxidase completely abrogates UV-induced hydrogen peroxide generation. Thus, NADPH oxidase is a major enzymatic source of hydrogen peroxide production following UV irradiation in human keratinocytes. The functional role of hydrogen peroxide in mediating UV responses in human skin remains to be determined.

Ultraviolet irradiation activates protein kinase–mediated signaling pathways within 1 hour. These signaling pathways are maximal within 4 hours after UV exposure.^{8 - 10} At this time, immunohistologic examination reveals activation (phosphorylation) of several of signaling kinases in cells throughout the epidermis (Figure 3). The activated kinases up-regulate expression and functional activation of the nuclear transcription factor, AP-1 (composed of Jun and Fos proteins), which then stimulates transcription of genes for matrix-degrading enzymes such as metalloproteinase (MMP) 1 (collagenase), MMP-3 (stromelysin 1), and MMP-9 (92-kd gelatinase).^{11 - 12} Transcription factor AP-1 also interferes with collagen gene expression in human dermal fibroblasts (see the following section). Ultraviolet irradiation also activates the transcription factor NF-κB¹³ that stimulates transcription of proinflammatory cytokine genes, including IL-1β, TNF-α, IL-6, and IL-8, and adhesion molecules including intercellular adhesion molecule-1.¹⁴ Ultraviolet-induced cytokine gene products then act through their cell surface receptors to activate AP-1 and NF-κB and thereby amplify the UV response.

Ultraviolet-induced MMP-1 initiates cleavage of fibrillar collagen (type I and III in skin) at a single site within its central triple helix. Once cleaved by MMP-1, collagen can be further degraded by elevated levels of MMP-3 and MMP-9.¹⁵ Metalloproteinase 1, MMP-3, and MMP-9 activities have been shown to colocalize with collagen in the dermis, following UV irradiation of human skin in vivo.^{10 ,16} Type I collagen molecules are stabilized by intermolecular covalent cross-links. Depending on the extent of degradation, partially degraded collagen can remain cross-linked within the insoluble collagen matrix. These insoluble collagen fragments are susceptible to proteolytic cleavage, in vitro, by proteases with broad specificity such as chymotrypsin. When sun-protected human skin was irradiated and biopsy performed, results showed the level of partially degraded collagen 24 hours following UV irradiation increased 3-fold (Figure 4). Thus, UV-induced MMPs degrade skin collagen and thereby impair the structural integrity of the dermis. In the absence of perfect repair, MMP-mediated collagen damage is expected to accumulate with each successive UV exposure. Such cumulative collagen damage is likely a major contributor to the phenotype of photoaged human skin.

In addition to degrading mature dermal collagen, UV irradiation impairs ongoing collagen synthesis, primarily through down-regulation of type I and type III procollagen gene expression.¹⁷ Two mechanisms contribute to reduced procollagen gene expression. As described earlier, UV irradiation induces the transcription factor AP-1. By binding and sequestering factors that are part of a transcriptional complex required for procollagen transcription, AP-1 interferes with collagen production.^{11 ,18} Transcription factor AP-1 has also been shown to decrease collagen synthesis by blocking the effects of transforming growth factor β (TGF-β), a major profibrotic cytokine, and sequestering one of the signaling proteins it activates both directly and indirectly.^{18 - 22}

Ultraviolet irradiation also interferes with TGF-β–dependent type I procollagen gene expression by down-regulating type II TGF-β receptor, within 8 hours of irradiation, rendering the cells unresponsive to TGF-β effects.²³ In cultured human fibroblasts, UV-induced down-regulation of type II TGF-β receptor and subsequent loss of TGF-β responsiveness results in substantial reduction of type I procollagen gene expression. These data suggest that down-regulation of type II TGF-β receptors, in addition to AP-1–mediated transcriptional repression, contribute to reduced procollagen gene expression observed in human skin in vivo, following UV irradiation.

It is well established that skin pigment provides a significant degree of protection against actinic damage. For a given exposure to UV irradiation, persons with less pigment (lighter skin color) will exhibit greater erythema and less tanning than persons with more pigment (darker skin pigment). This inverse relationship between skin color and UV-induced skin reddening (sunburn) forms the basis of the skin phototype classification system. This classification system identifies 6 skin phototypes, with light-skinned persons who sunburn easily and do not tan (classified as skin phototype I) and dark-skinned persons who do not sunburn and tan readily as phototype VI. However, quantitative assessment of the effects of these pigment-associated phototypes on molecular responses of skin to UV irradiation is lacking. Therefore, the effects of skin color on UV-induced MMP-1 gene expression and formation of thymine dimers were investigated in human skin in vivo.

Subjects were placed in one of 2 groups based on quantitative measure of their surface skin color, as determined by a color meter (Chroma Meter; Minolta Camera Co Ltd, Osaka, Japan). Persons with Chroma Meter values greater than 65 were placed in one group (lightly pigmented), and persons with Chroma Meter values less than 55 were placed in a separate group (darkly pigmented). These 2 groups approximately correspond to skin phototypes I and II and phototypes V and VI, respectively. Subjects were exposed to either a UV-B/UV-A2 source (λmax = 311 nm, filtered to remove UV-C) or a UV-A1 source (λmax = 359 nm, 366 nm, and 374 nm) (Sellas 2000 Watt High Pressure Unit; Sellas GmbH, Gevelsberg, Germany). Lightly pigmented subjects were exposed to twice their minimal erythema dose from the UV-B/UV-A2 source (average minimal erythema dose was 640 mJ/cm² total UV irradiation). Darkly pigmented subjects were exposed to 2 and 4 times the average minimal erythema dose of the lightly pigmented group (640 mJ/ cm² to 1280 mJ/cm²). Skin samples were obtained 24 hours after UV irradiation for determination of MMP-1 messenger RNA (mRNA) levels and DNA photoproducts. Ultraviolet-B/UV-A2 exposure resulted in substantial induction of MMP-1 mRNA (Figure 5) and formation of thymine dimers (Figure 5) in lightly pigmented subjects. In contrast, twice the average exposure of the lightly pigmented group produced only modest MMP-1 mRNA induction (Figure 5) or DNA damage (Figure 5) in the darkly pigmented group.

The effect of skin pigment on induction of MMP-1 mRNA and DNA photoproducts by UV-A1 exposure was investigated next. In the lightly pigmented group, 110 mJ/cm² UV-A1 caused substantial induction of MMP-1 mRNA expression and DNA photoproducts (Figure 6). The response to UV-A1 was similar to that observed in response to UV-B/UV-A2 described above (Figure 5). In contrast, 110 mJ/cm² UV-A1 had little effect on the level of MMP-1 gene expression or DNA photoproducts (Figure 6) in the darkly pigmented group. These data demonstrate that pigmentation provides substantial protection against deleterious effects of both UV-B/UV-A2 and UV-A1 irradiation or is at least correlated with enhanced protection. These findings provide quantitative support for the clinical observation that photoaging is less severe in the darkly pigmented population compared with the lightly pigmented population. These data are also consistent with the low incidence of skin cancer in the darkly pigmented population.

It is worth noting the difference between lightly and darkly pigmented populations in depth from the surface where DNA photoproducts were observed. In the lightly pigmented group, DNA photoproducts were observed in cells throughout all layers of the epidermis and in the upper dermis (Figure 5 and Figure 6). In the darkly pigmented group, the modest level of photoproducts was restricted to post–mitotic cells in the upper epidermis. Because UV-B–induced DNA photoproducts are produced by direct absorption of UV-B irradiation by DNA, these data indicate that skin pigment substantially attenuates penetration of UV irradiation into skin, which was expected owing to its broad UV absorption spectrum^{24 - 25} and preferential distribution in supranuclear "caps" in the path of incident UV irradiation.²⁶

Based on optical properties of the skin, it is often argued that penetration of UV-B is restricted to the epidermis, while penetration of longer wavelength UV-A1 irradiation extends deep into the dermis. While this distinction of skin penetration between UV-B and UV-A irradiation may be generally true, our data on the localization of photoproduct formation indicate that, in lightly pigmented individuals, penetration of UV-B/UV-A2 irradiation extends definitely into the upper dermis, a site of major changes in photoaging.

Many cosmetic and cosmeceutical products claim to improve the appearance of photoaged skin. In general, however, objective evidence that would allow evaluation of these claims is not available. In contrast, a large number of controlled clinical studies have been published demonstrating that topical application of 0.025% to 0.1% all-trans retinoic acid (tRA) improves the appearance of photoaged skin.^{27 - 28} In spite of extensive knowledge regarding the molecular mechanisms by which tRA acts to regulate expression of certain genes, details regarding the molecular basis of tRA action in photoaging remain largely unknown. Type I collagen is the most abundant protein in skin, and type I and type III collagen fibrils provide strength and resiliency to skin. Photoaged skin contains an abundance of degraded, disorganized collagen fibrils and has a reduced production of type I and type III procollagen. All-trans retinoic acid has been shown to induce type I and type III procollagen gene expression in photoaged human skin.²⁹ Because procollagen is the precursor to collagen, it is likely that increased production of procollagen results in increased deposition of collagen fibrils. This laying down of newly synthesized collagen fibrils may lie at the heart of the ability of tRA to improve the appearance of photoaged skin.

If so, then how does tRA induce new collagen synthesis? One clue is the observation that topical tRA induces TGF-β in human and mouse skin.^{30 - 31} As described earlier, TGF-β induces production of type I and type III procollagen and other components of the dermal extracellular matrix. Thus, it is reasonable to envision that tRA induction of TGF-β plays a primary role in tRA's mechanism of action in photoaging. However, this possibility remains to be investigated.

A major obstacle to understanding the actions of tRA in photoaging is lack of suitable in vitro models. In monolayer cultures, human skin fibroblasts continuously produce high levels of type I procollagen and TGF-β1. These cultures are largely insensitive to exogenous TGF-β and tRA. In 3-dimensional collagen gel cultures, human skin fibroblasts express less procollagen than in monolayer culture, but they respond similarly poorly to exogenous TGF-β and tRA. It has recently been demonstrated that fibroblasts cultured from photoaged skin produce amounts of type I procollagen equivalent to fibroblasts from sun-protected skin.³² These data indicate that reduced type I procollagen expression observed in photoaged skin in vivo stems from contextual influences, which are alleviated in culture in vitro. This observation suggests that a suitable in vitro model to investigate the mechanisms by which tRA regulates type I procollagen synthesis in photoaged skin may be difficult to achieve without additional knowledge regarding how alterations in the epidermis and/or extracellular matrix in photodamaged skin affect fibroblast function.

Photoaging involves complex alterations to many important structural components of the dermal extracellular matrix besides collagen. The elastin fiber network is grossly disrupted, and modifications in the structure and composition of anchoring fibrils, proteoglycans, and glycoaminoglycans have been reported.^{33 - 35} It is likely that the mechanisms by which tRA improves photoaged skin involve actions on other dermal components as well as collagen. In addition to its ability to improve existing photoaging, tRA has also been shown to interfere with responses of skin to acute UV irradiation that bring about collagen degradation.¹⁶ Treatment of human or mouse skin with tRA prior to UV exposure inhibits induction of transcription factor c-Jun.¹⁰ c-Jun is necessary to form an active AP-1 complex, which, as described above, both stimulates transcription of MMP-1, MMP-3, and MMP-9, and inhibits expression of type I procollagen and type III procollagen. All-trans retinoic acid does not inhibit UV induction of c-Jun mRNA, but rather blocks accumulation of c-Jun protein.¹⁰ Whether tRA inhibits c-Jun protein synthesis and/or stimulates c-Jun protein degradation is not known.

Regardless of mechanism, suppression by tRA of UV induction of c-Jun protein reduces MMP activities that degrade skin connective tissue and prevents loss of procollagen expression. In essence, tRA is predicted to impede photoaging by mitigating disruption of collagen homeostasis caused by UV irradiation. The metabolic precursor of tRA is vitamin A (all-trans retinol [tROL]). Skin keratinocytes and fibroblasts, as well as most other cell types, possess the ability to convert vitamin A (which is taken up from the circulation) into tRA. However, cellular tRA levels are constrained by catabolic inactivation mediated by specific members of the cytochrome P450 family. Human keratinocytes also possess the ability to esterify tROL. All-trans retinol esters reside in cellular membranes and are considered a stored form of tROL. Under appropriate conditions, tROL esters can be enzymatically hydrolyzed to provide substrate for conversion to tRA, which binds nuclear retinoic acid receptors that tROL does not bind.

Interestingly, human fibroblasts possess little if any capacity to esterify tROL and therefore must depend on uptake of tROL from the circulation for tRA biosynthesis. In human keratinocytes, the rate and capacity of tROL esterification is hundreds of times greater than that of oxidation to tRA.³⁶ Thus, esterification functions to sequester tROL from conversion to tRA. The lack of esterification activity in fibroblasts should permit higher rates of tRA synthesis than that in keratinocytes. Given the capacity of human skin to metabolically convert topical tROL to tRA, a sufficient concentration of tROL (approximately 0.25%-0.5%) should provide therapeutic and preventative benefits equivalent to those of tRA. Comparisons of similar biologically active concentrations of topical tROL and tRA demonstrated less erythema and scaling with tROL than with tRA. Because the severity of tRA adverse effects is concentration dependent, the stringent control of tROL conversion to tRA, as described earlier, likely accounts for the observed difference

Skin, like all organs, ages with the passage of time. Compared with photodamaged skin, sun-protected aged skin appears thinner, more evenly pigmented, laxer, and more finely lined. Often, photoaging and chronological aging have been considered separate entities. However, recent evidence indicates that they share some important molecular features. Photoaging is the superposition of UV irradiation from the sun on intrinsic aging. To exert its biological effects, UV irradiation must be absorbed by molecules (chromophores) in the skin, and the absorbed energy must be converted into chemical reactions. Depending on the chromophore, absorbed energy may cause direct chemical modification of the chromophore itself, or the energy may be transferred from the chromophore to another molecule, which undergoes chemical modification. For example, UV-B absorption by DNA causes cross-linking of adjacent pyrimidines, whereas UV-A–absorbing skin chromophores transfer energy to oxygen to generate ROS, which oxidize cellular constituents including proteins, lipids, and DNA. Photoaging is mediated by direct UV absorption and ROS-mediated photochemical reactions. The cause of chronological aging is far less clear than that of photoaging. Many theories have been advanced to explain chronological aging. One, the free radical theory, states that aging results from accumulation of cellular damage that results from excess ROS that are generated as a consequence of oxidative metabolism.^{37 - 38} Age-associated cellular damage includes oxidation of DNA resulting in mutations, oxidation of proteins resulting in reduced function, and oxidation of membrane lipids resulting in reduced transport efficiency and possibly altered transmembrane signaling. The main source of excess ROS implicated in aging is mitochondrial oxidative energy generation. As a result of accumulated damage, the aged cell has reduced antioxidant capacity,^{39 - 40} further exacerbating ROS-mediated damage and the aged phenotype.⁴¹

Given the central role of ROS in both photoaging and chronological aging, it is possible that the 2 processes have common molecular mediators. As described earlier, transcription factor AP-1 is a critical mediator of acute photodamage that is involved in both overexpression of MMPs and reduction of type I procollagen. In human skin, AP-1 activity is limited by c-Jun expression, since c-Fos is continuously expressed.¹⁰ Interestingly, while c-Fos expression in young (18-28 years old) and aged (>80 years old) skin does not differ, c-Jun expression is elevated in aged compared with young skin.⁴² Both c-Jun mRNA and protein levels are elevated in aged skin, a finding similar to that in UV-irradiated skin. In addition, the activity of the upstream activator of c-Jun, c-Jun N-terminal kinase, is also elevated in aged compared with young skin.⁴² These data suggest that AP-1 activity is increased in aged skin. In support of this possibility, AP-1–regulated MMP-1 and MMP-9 activities are also increased in aged human skin in vivo.⁴³

Increased MMP activity would be expected, over time, to degrade dermal connective tissue. Consistent with this view, insoluble partially degraded collagen, as a percentage of total collagen, is increased 4-fold in aged compared with young human skin (data not shown; unpublished data, 2002). This increased percentage of partially degraded collagen in aged skin is similar to that observed in photoaged skin compared with sun-protected skin. These data indicate that imperfect repair of elevated MMP-mediated collagen breakdown in aged skin, as in UV-irradiated skin, results in accumulation of collagen fragments. Transmission electron microscopy reveals the fragmented nature of collagen fibrils in photoaged and aged human skin in vivo (Figure 7).

In addition to reduction of type I procollagen synthesis following immediate UV irradiation as described earlier, ongoing procollagen synthesis is significantly reduced in severely photodamaged and chronologically aged skin, independent of recent UV exposure.^{29 ,43} Mechanisms underlying the sustained reduction in procollagen synthesis are not fully understood. In full-thickness photoaged skin, total collagen content is only modestly reduced.³⁴ However, procollagen gene and protein expression are substantially reduced in approximately the upper one third of the dermis.⁴⁴ This restricted localization of procollagen reduction likely reflects the depth of penetration of UV irradiation into the skin, which mitigates more extensive reduction of total collagen. In photodamaged skin, the number of fibroblasts in vitro is similar to that in sun-protected skin of the same subjects, and the capacity to synthesize type I procollagen is the same for fibroblasts cultured from both areas.⁴³ Thus, the reduced procollagen observed in photodamaged skin likely results from down-regulation of fibroblast procollagen synthesis by factors within the dermal milieu rather than from inherent alterations in the fibroblasts.

Interestingly, exposure of cultured fibroblasts from either photodamaged or sun-protected skin to partially degraded type I collagen, produced by in vitro treatment of collagen with a mixture of MMPs from human skin, inhibited procollagen synthesis.³² These results suggest that elevated levels of degraded collagen observed in photodamaged skin act to down-regulate type I procollagen synthesis. Of the different MMPs, MMP-1 was the most effective collagenase, followed by MMP-8 and MMP-13.³² As expected, gelatinolytic enzymes (MMP-2 and MMP-9) did not degrade intact collagen and did not inhibit procollagen synthesis, but the combination of MMP-1 and MMP-9 broke down collagen to small peptides.³² Interestingly, these small fragments did not inhibit procollagen synthesis, but rather the larger breakdown fragments of type I collagen negatively regulated its synthesis. Taken together, these data suggest that the high-molecular-weight fragments of type I collagen serve as negative regulators of type I collagen synthesis and that further breakdown of MMP-1–cleaved collagen by MMP-9 can alleviate this inhibition. Thus, UV-induced MMPs damage the dermis by 2 related mechanisms: direct degradation of collagen and indirect inhibition of collagen synthesis by MMP-generated collagen degradation products.

In contrast to photodamaged skin, in aged sun-protected skin, both the number of fibroblasts and their capacity to synthesize type I procollagen are reduced compared with young skin.⁴³ In addition, aged skin, similar to photoaged skin, contains elevated levels of partially degraded collagen. Therefore, it is likely that the inhibitory effects of collagen fragments observed in photoaged skin are also operative in aged skin and are superimposed on an intrinsic decline in collagen synthetic activity.

Reduced mechanical tension may also contribute to diminished number of fibroblasts in aged skin because the loss of mechanical tension results in increased apoptosis in model cell systems.^{45 - 46} How MMP-damaged collagen functions to inhibit new collagen synthesis in vivo is not known, but collagen synthetic activity is regulated by mechanical tension on fibroblasts resulting from attachment to a firm substratum.⁴⁷ Fibroblasts exert contractile forces on the collagen extracellular matrix, and the physical resistance of the matrix to this contraction generates mechanical tension on the fibroblast. The rate of collagen synthesis is proportional to the level of mechanical tension.^{48 - 49} We envision that damaged collagen fibrils are more pliable than native fibrils. As fibroblasts interact with damaged collagen fibrils, the cells experience less resistance and therefore less mechanical tension, resulting in reduced procollagen synthesis. How mechanical tension regulates procollagen synthesis is poorly understood, although many cellular functions are affected by mechanical tension, including cell surface receptor activation, signal transduction, gene expression, and cell growth.^{46 - 47 ,50} The observed disruption in the structure of the collagenous extracellular matrix and attendant reduction of mechanical tension on fibroblasts offer an attractive explanation, at least in part, for reduced procollagen expression in aged and photoaged human skin.

The pathophysiological process of photoaging and skin aging derives largely from aberrant regulation of a multitude of finely tuned molecular mechanisms, which have evolved to maintain the structural integrity of skin connective tissue. These molecular mechanisms enable skin cells to communicate with each other and their environment. Knowledge regarding relevant signal transduction pathways and their effectors provides exciting opportunities for therapeutic intervention to prevent and repair age-related skin damage. Two critical mediators of skin aging are the transcription factor AP-1 and AP-1–regulated MMPs, both of which play critical roles in tumor formation, inflammation, and fibrosis, which are all subjects of intensive basic and clinical research. The promise of this research is the development of new drugs that target these critical mediators of disease. The challenge lies in the application of these new therapies to skin aging.

Accepted for publication July 23, 2002.

This work was supported in part by grant AG139364-01-GJF from the National Institutes of Health, Bethesda, Md, and grants from Johnson & Johnson, New Brunswick, NJ, and the Mary Reynolds Babcock Foundation, Winston-Salem, NC.

All procedures involving human subjects were approved by the University of Michigan institutional review board, and all subjects provided written informed consent.

Corresponding author and reprints: Gary J. Fisher, PhD, Medical Science I, R6447, Department of Dermatology, University of Michigan, Ann Arbor, MI 48109 (e-mail: dianemch@umich.edu).