F. Patrick Ross1 and Angela M. Christiano2
1Department of Pathology and Immunology, Washington University School of Medicine, St. Louis, Missouri, USA.
2Departments of Dermatology and Genetics and Development, Columbia University College of Physicians and Surgeons, New York, New York, USA.
Address correspondence to: F. Patrick Ross, Department of Pathology and Immunology, Washington University School of Medicine, Campus Box 8118, 660 South Euclid Avenue, St. Louis, Missouri 63110, USA. Phone: (314) 454-8079; Fax: (314) 454-5505; E-mail: [email protected] . Or to: Angela M. Christiano, Departments of Dermatology and Genetics and Development, Columbia University, 630 West 168th Street, VC-1526, New York, New York 10032, USA. Phone: (212) 305-9565; Fax: (212) 305-7391; E-mail: [email protected] .
Skin and bone — what comes to mind at hearing this phrase? While certainly a metaphor for disease, it also defines two very different tissues, one a flexible and contiguous outer covering, the other a morphologically diverse hard tissue distributed at over 200 sites in the body. As the accompanying series of Reviews highlights, these tissues are indeed diverse, but there are also surprising similarities. Skin is the interface between the internal organs and the environment, and as such plays a crucial role in the body’s defense mechanism. The skin and its many appendages are responsible for functions as diverse as epidermal barrier and defense, immune surveillance, UV protection, thermoregulation, sweating, lubrication, pigmentation, the sensations of pain and touch, and, importantly, the protection of various stem cell niches in the skin. Bone serves a number of purposes: it provides protection for vital organs, a lever for locomotion, a reservoir for calcium, and the site of adult hematopoiesis. The tissue is composed of osteoblasts, osteoclasts, and their individual precursors plus a complex mixture of mesenchymal, myeloid, and lymphoid cells in the marrow space. Finally, the endothelial microenvironment provides nutrition and is a conduit for the influx and emigration of cells that impact bone biology in several important ways. This Review series guides the reader through these various facets of 2 diverse, yet interdependent, tissues.
While in the adult vertebrate organism, bone and skin spend much of their time as separate entities with vastly different agendas, the skin dermis and the bone originate from a common primordial mesenchyme, and at some points in development the overlying epidermis — in the form of the apical ectodermal ridge — and the outgrowth of the limb are intimately interdependent. In the earliest days of development, and again in times of need such as limb regeneration in tetrapod vertebrates, the 2 tissues must come together and function very much as one. The study of developmental pathways and epithelial-mesenchymal interactions in the skin and bone have revealed some striking parallels, which are reprised in both organs in the adult.
Source: J. Clin. Invest. 116:1140-1149 (2006)
Despite its aesthetically pleasing appendages such as the hairand nails, its pliable nature, its flexibility, and its responsiveness,the skin is a master in the art of self-defense — provingthat a tissue need not be hard in order to be tough. In additionto serving as the body’s outermost protective covering,the skin barrier integrates the body’s physiology withthe terrestrial environment. The epidermal barrier works in2 ways: as an inside-out barrier, to minimize transepidermalwater loss, and as an outside-in barrier, as a sentry to preventinvasion by infectious agents and noxious substances (1-5).
In addition to the 2 major structural layers of the skin, theepidermis and the dermis, the skin is also home to a numberof other cell types and structures that each play a unique rolein its function (6). There are resident dendritic cells, knownas Langerhans cells, which are the antigen-presenting cellsof the skin interspersed among the keratinocytes that providethe first line of defense against invasion. Mast cells residein the dermis, and their degranulation releases vasoactive aminesand other proinflammatory mediators that induce immediate hypersensitivityresponses such as urticaria (7).
The skin is also populated by eccrine (sweat) glands, whichallow for temperature regulation via sympathetic nervous systemcontrol of an intricate network of lymphatic and blood capillariesthat reside in the dermis, as well as the apocrine (scent) glands,which are believed to emit secretions and pheromones for sexualcommunication (8). The sebaceous gland is attached to the hairfollicle and secretes sebum to the skin surface that keeps itsupple and waterproof (9). The hair follicle, in addition togenerating the hair shaft, also provides a protective nicheto several stem cell populations in the skin, including thekeratinocyte stem cell, the melanocyte stem cell, a populationof epidermal neural crest stem cells, and the dermal stem cellcompartment, known as the dermal papilla (10-13). There areseveral contractile cell populations, such as the myoepithelialcells lining the sweat gland, that contract to extrude liquidonto the skin surface during thermoregulation, as well as thearrector pili muscle, which attaches to the hair follicle andis responsible for creating goose bumps when the body’score temperature falls (14).
Adult skin is also home to at least 2 different neural crestcell populations: the melanocyte, which provides pigmentationand UV protection to human epidermis and color to the hair shaft(15), and the Merkel cell, a neuroendocrine cell responsiblefor transmission of touch sensation through the cutaneous nerves,among other functions (Figure 1) (16). There are also significantanatomical variations in skin structure on different parts ofthe body, such as the thick, protective epidermis on the palmsand soles compared with the thin skin on the eyelid, and specializedregions without hair follicles, such as the glabrous (lip) epithelium,or with a high density of hair follicles, such as the scalp.
Thus the skin you’re in is a highly specialized and meticulouslyregulated organ system populated by numerous different celltypes that each contribute uniquely to its multitude of functions(6). Likewise, the spectrum of skin disorders that arise whenthese functions go awry is virtually limitless. The skin-relatedReviews in this series focus on 4 main topics: the epidermalbarrier, autoimmune skin disease, epithelial viral infection,and the relationship between the skin and the central nervoussystem.
The series begins with a Review of epidermal barrier formationby Julia Segre (17). In addition to serving as the body’soutermost protective covering, the skin is a barrier betweenthe body and the terrestrial environment. Barrier function iscritical in newborn animals, as shown by transgenic animal modelswith barrier defects that die shortly after birth from transepidermalwater loss.
Early studies of the skin barrier focused mainly on its remarkablephysiochemical properties and on determining the unique proteincomposition of lipids and the cornified cell envelope (1, 2,4, 5). Genetic approaches and efforts in genomic cloning haverevealed an unusually complex cluster of genes in human chromosome1q21, termed the epidermal differentiation complex, which containsmore than 30 genes involved in terminal differentiation of theskin (18).
Recent studies of the molecular mechanisms governing barrierformation, particularly transcriptional events, have begun toshed light on the molecular underpinnings of this highly regulatedprocess. In her Review (17), Segre draws some intriguing parallelsbetween the formation of the epidermal barrier and the reprisalof these processes in the setting of common skin disorders suchas psoriasis or atopic dermatitis, which display a perturbationin barrier function that may exacerbate these conditions. Takingclues from the transcriptional and molecular events involvedin forming the barrier in utero, this Review proposes that themanagement of inflammatory skin disease may benefit from therapiesthat enhance restoration of the epidermal barrier.
The autoantigens in pemphigus are epidermal desmosomal cadherinsthat mediate the intercellular adhesion of keratinocytes, specificallydesmoglein 1 in pemphigus foliaceous and desmoglein 3 in pemphigusvulgaris (PV). Likewise, the autoantibodies in bullous pemphigoids(BPs) are 2 hemidesmosomal proteins that are involved in cellbasement membrane attachment of basal keratinocytes, specificallythe 180-kDa BP antigen (BP180, also known as type XVII collagen)and the 230-kDa BP antigen (BP230).
The pathognomonic blisters of pemphigus arise from the bindingof circulating autoantigenic IgG molecules to these keratinocyteproteins and the subsequent disruption of intercellular adhesioneither through a direct effect on desmosomal or hemidesmosomalarchitecture, by triggering cell signaling processes that resultin loss of cell adhesion (acantholysis), or both. Thus the criticalrole of the B cell in generating the secreted autoantibodiesfound in pemphigus patients has long been known. Hertl and colleaguesfocus instead on the emerging importance of the T cell in pemphigus(19). These authors describe a "multiple hit" model of lossof immune tolerance in pemphigus. The authors review severallines of evidence that collectively show that neither circulatingIgG nor autoreactive T cells alone are sufficient to elicitPV and suggest that a third population of cells, Tregs, maybe critical in its pathogenesis.
The role of Tregs in suppressing immune responses, controllingautoimmunity, and maintaining tolerance has not been widelystudied in the setting of pemphigus; however, some recent worksuggests a potential imbalance of these cells in PV patients.One study found that Tregs were found in the autoreactive Tcells of a large number of healthy individuals but in only 20%of PV patients, raising the possibility that the Tregs wererequired to keep the autoreactive T cells from evoking disease.A multiple hit model in pemphigus would therefore require anautoreactive B cell, an autoreactive T cell, and finally a lossof Treg suppressor activity in order to set the stage for PV.The requirement of this rare constellation of events might explainthe relatively low population incidence of PV, at only 1–2cases per 100,000 individuals. Hertl and colleagues suggestthat focusing on Tregs may represent a new therapeutic opportunityin the restoration of immune tolerance in the pemphigus familyof autoimmune bullous diseases.
The next Review, by Douglas Lowy and John Schiller (21), summarizesrecent developments in understanding human papillomavirus (HPV)and the development of successful HPV vaccines. HPVs are a largefamily of double-stranded DNA viruses that are virtually ubiquitousand infect epithelial tissues including the skin, cervix, andother mucosae. Over 75 variants have been identified from humantissues, about one-third of which are sexually transmitted andgive rise to a spectrum of lesions within the genital tract.Some HPVs, such as HPV6 or HPV11, cause only benign epitheliallesions such as genital warts, while others, such as HPV16 andHPV18, result in lesions that can progress to invasive cancersof the cervix (22). HPVs that infect the cutaneous epitheliumhave also recently been linked as a cofactor along with UV exposurein the development of nonmelanoma cancer of the skin (23).
Primary infection of HPV in skin or cervical epithelia usuallyoccurs within the long-lived basal stem cells, wherein the virusreplicates and immediate early proteins are expressed. As theepithelial cells begin to differentiate, the viral proteinsE6 and E7 are expressed. The L1 and L2 proteins are not expresseduntil the virus reaches the upper spinous layers, where assemblyoccurs, and finally the production of mature virions takes placein the stratum corneum, from whence intact virions are shed(24).
Initial HPV exposure occurs during sexual activity, where theinfection of the cervical epithelium leads to a squamous intraepitheliallesion caused by either a low-risk HPV (e.g., HPV6 and HPV11)or a high-risk HPV (e.g., HPV16, -18, -31, -33, -45). Increasedsexual activity and the total number of partners results ina higher lifetime risk of HPV exposure. Similarly, in oral mucosa,high-grade HPV lesions can be promoted by smoking and otherfactors and result in invasive squamous cell carcinomas overlong periods of time. Although such occurrences are rare, low-gradesquamous intraepithelial lesions may evolve into invasive squamouscarcinomas over time.
Cervical cancer comprises 10% of all cancers in women and isthe second leading cause of death from cancer among women worldwide.HPV DNA is found in over 90% of cervical cancer lesions. Roughly80% of all cervical cancers occur in less-developed countries,largely due to insufficient resources for cervical cancer screening(e.g., Pap smear). This striking association suggests the possibilityof developing either prophylaxis or new therapies for cervicalcancer based on the manipulation of human immune responses againstHPVs.
Recently, Merck and GlaxoSmithKline announced stunning successesin the development of vaccines against HPVs. These vaccineswere developed against the L1 capsid protein, which induceshigh levels of antibodies. Both vaccines target HPV16 and HPV18,which make up more than 70% of cancers, and the Merck vaccinealso targets HPV6 and HPV11, which account for about 90% ofgenital warts. These vaccines have shown nearly 100% efficacyin the prevention of HPV infection in women. In their Review,Lowy and Schiller (21) address many important considerationsin the design and testing of these vaccines, including publichealth and ethical issues. These vaccines have great potentialto prevent several hundreds of thousands of cancers each year,most of which occur in young, sexually active women.
The development of HPV vaccines represents a powerful illustrationof translational research whereby advances in epithelial virologycan move into the commercial sector for the development of eventuallarge-scale vaccination programs. Perhaps successes such asthe one described by Lowy and Schiller for HPV will carry overinto similar programs for herpesvirus infections, a most importantcause of morbidity in human populations.
To broaden our thinking about the intimate connection betweenthe skin and the nervous system, the final skin-related Reviewin this series, by Ralf Paus and colleagues (25), raises someintriguing new possibilities in managing a symptom unique todermatology: pruritus, or itching. Despite many years of research,the exact cause of itching is unknown and represents a complexphysiological phenomenon. An itch can be defined as an unpleasantsensation that provokes an often uncontrollable desire to scratch,which ironically can be inexplicably pleasurable. Scratchingis believed to relieve the itch by inducing mild pain that causesa temporary distraction from the itch. Frequently, one worsensthe other, since damage to the outer layers of the skin by scratchingcan release additional proinflammatory agents that further exacerbateitching. This phenomenon is known as the itch-scratch cycle,which lies at the center of current neurophysiological researchin the skin (26).
Pruritus can be a symptom of innumerable skin diseases; however,it can also occur when there is no visible evidence of a skinlesion and instead results from a disruption in processing ofthe itch sensation in the brain or the circuitry connectingit with the skin. There are many systemic diseases that cancause itch, including kidney failure, hepatitis C infection,multiple myeloma, and liver disease, among others, suggestingthat stimuli from outside the skin-brain circuit can enter thepathway and cause itching.
Pruritus has been a challenge from the research perspectivedue to the subjective nature of itching itself, the absenceof a precise physiological definition or quantitative measures,and the lack of suitable in vitro or in vivo models to mimicthe symptoms. Despite these challenges, recent advances in understandingthe neurophysiological basis of itch have revealed some surprisingnew insights. The skin is home to a complex network of nervesthat transmit different sensations, including itch, pain, touch,cold, and heat. In the simplest of terms, itching is a responseto a chemical stimulus that transmits these signals back tothe brain (27). Teasing out which nerve fibers are responsiblefor itch has been a formidable task, and for many years it wasbelieved that itch simply hijacked nerve circuits from the painpathways and was merely a modified form of pain. However, thediscovery of itch-specific neurons revealed that there are,in fact, distinct sensory systems for itch and pain (nociception).In their Review (25), Paus and colleagues discuss some potentiallynovel therapeutic opportunities that have arisen from a betterunderstanding of itch from the point of view of the skin aswell as the brain.
The skin-related Reviews within this series focus mainly on4 variations on the theme of the epithelial cutaneous disease,since the epidermal barrier, the destruction of keratinocytesin pemphigus, viral infection of epithelial cells by HPV, andthe origins of itching all relate in some way to the epidermisor its resident cells. There is far more to the skin than theepidermis, and to consider it in isolation would be just scratchingthe surface. The dermis, home of mesenchymal cells in the skin,was not touched upon extensively in these Reviews, nor was thevast subject of developmental pathways such as Wnt signaling(reviewed in refs. 28, 29). Interestingly, in early development,undifferentiated mesenchymal cells can give rise to the fibroblastsof skin dermis, to adipocytes, to cartilage, to muscle, or tobone, suggesting that there are common themes at play in differentiationof these tissues (30). Since mesenchymal cells and signalingpathways are the primary focus of the second half of this Reviewseries, we shall now cut straight to the bone and examine whatlies beneath.
Physiological bone turnover can be divided into 2 temporal phases: modeling, which occurs during development (a topic not addressed in this series; for recent reviews see refs. 31, 32), and remodeling, a lifelong process involving tissue renewal. Remodeling starts with removal by osteoclasts of matrix, a mixture of insoluble proteins in which type I collagen is predominant (>90%) and a poorly crystalline, chemically modified hydroxyapatite. Following resorption, osteoblasts are recruited to the site, where they secrete and mineralize new matrix. Until about age 30–35 bone replacement exceeds or equals removal, thus increasing or maintaining bone mass; thereafter, bone mass decreases, reflecting the predominance of osteoclast activity. The major thrust of the bone-related Review articles contained within this series is to outline selected new and important aspects of osteoblast and osteoclast biology (for an excellent review of the pathophysiology of osteoporosis see ref. 33).
Osteoblasts are specialized fibroblasts that secrete and calcifya specific matrix. Their lineage specification from mesenchymalstems cells is regulated by a plethora of signals. A range ofcytokines modulate osteoblast differentiation, including bonematrix–derived TGF-ß, bone morphogenic protein2 (BMP-2), BMP-4, and BMP-7, and their inhibitors noggin, chordin,gremlin, and sclerostin, the last identified by positional cloningof families with increased bone mass. Similarly, numerous hormonesimpact osteoblast function positively including IGF-1, parathyroidhormone (PTH), PTH-related protein (PTHrP), 1,25(OH)2D3, leptin,glucocorticoids, the Notch pathway, and members of the leukemiainhibitory factor/IL-6 family.
Transcription factors that regulate the osteoblast include arange of homeodomain proteins: the activator protein (AP) familymembers Jun, Fos, and Fra, Smads, CCAAT/enhancer binding proteinß (C/EBPß) and C/EBP, lymphoid-enhancingfactor (a Wnt effector), activating transcription factor 4,Runt-related transcription factor 2 (Runx2), and osterix, thelast 3 of which are considered master genes for osteoblast differentiation.In contrast to the osteoclast (see below), the osteoblast’sbest-characterized intracellular signaling pathway is the p42/44MAPK system.
Osteoblasts ligate existing matrix via ß1 integrins,forming a monolayer that is linked by cadherins. Once active,the cells secrete a matrix containing type I collagen and smallerbut significant amounts of osteocalcin, matrix gla protein,osteopontin, bone sialoprotein, many minor components, and,importantly, growth factors such as BMPs and TGF-ß.Key ectoproteins, including progressive ankylosis gene (ANK)and tissue nonspecific alkaline phosphatase (TNAP), export pyrophosphategenerated intracellularly and cleave this small-molecule inhibitorof calcification, respectively (34). In contrast to their proapoptoticrole in osteoclasts, bisphosphonates increase osteoblast lifespanand perhaps function (35).
The text below focuses on Wnt signaling in osteoblasts and therole of these cells as supporters of the HSC niche and targetsfor osteoclastogenic hormones. Detailed reviews of osteoblastbiology are available (36-40).
In this series, the Review article by Ormond MacDougald andcolleagues (41) focuses on the role of Wnt signaling in osteoblastformation and function. Wnts, a family of secreted glycoproteinswith multiple inhibitors, are ligands for the family of 7-membrane–spanningfrizzled receptors and play a prominent role in both the earlyand later stages of osteoblast differentiation. Wnt antagonistsinclude Dickkopfs (Dkks) and secreted frizzled-related proteins(sFRPs). While their signaling pathways were characterized originallyin terms of development, Wnts regulate numerous cellular functionsand have been linked recently to cancer and stem cell biology.It is also clear that individual Wnts utilize canonical andnoncanonical pathways (42); the endogenous Wnt ligands in boneremain unidentified, although Wnt10b was recently implicatedby MacDougald et al. (43). The canonical pathway is well establishedand involves Wnt-dependent inhibition of proteasome-mediateddegradation of ß-catenin, which forms complexes withmembers of the T cell factor/lymphoid enhancer binding factor(TCF/LEF) family that regulate transcription when it accumulatesin the nucleus. The noncanonical pathway, also frizzled dependent,activates different intracellular signals including the calcium-calmodulin-PKCaxis and the Rho family of small GTPases.
MacDougald and colleagues review the current information onWnts and bone biology. The story began with the identificationof loss- and gain-of-function mutations in LDL receptor–relatedprotein 5 (LRP5), a frizzled coreceptor, that correlate withchanges in human bone mass. More recent studies showed thatß-catenin, which is downstream of the Wnt–LRP5/6–frizzledaxis, is indispensable for osteoblast differentiation in themouse (44). Several papers cited by MacDougald et al. are worthyof comment here. Whereas mice lacking frizzled-related protein1 have increased bone mass arising from markedly decreased osteoblastapoptosis, and Lrp5–/– mice exhibit a predominantdeficit in osteoblast number and function, Col1-Cre; ß-cateninc/canimals show mainly a secondary defect in osteoclasts. Theseresults may reflect different modes of Wnt signaling: canonicalversus noncanonical pathways, Wnt-dependent versus -independentsignaling, or unique roles for different Wnt family members.Identification of downstream targets of Wnt signaling is anunderexplored subject: osteoprotegerin (OPG) was identifiedas a direct target in osteoblasts, and the anabolic genes subjectto Wnt regulation are not known but may include BMPs. Despitethe implied importance of canonical Wnt signaling in osteoblastbiology, the role of the TCF/LEF family of transcription factorsis unclear. Since ß-catenin participates in transcriptionalcomplexes with molecules other than TCF/LEFs, some target genesmay not be regulated via TCF/LEF binding sites. Finally, MacDougaldet al. may understate the problem of targeting glycogen synthasekinase 3 (GSK3), a downstream effector of Wnt/ß-cateninsignaling and a molecule for which a number of potent inhibitorshave been developed (45). As pointed out, long-term treatmentwith GSK3 inhibitors may predispose cells to an oncogenic mutation.On the other hand, lithium ions, a nonspecific inhibitor ofGSK3, have been used for decades without apparent problems inthis regard (45).
In the Review by Tong Yin and Linheng Li (46), the authors summarizeanother new and exciting arena in bone biology, the HSC niche.The model, proposed simultaneously by Li and coworkers and Calviet al. (47, 48) suggests that mesenchymal cells, in the formof bone-lining osteoblasts, are central to the maintenance ofHSCs and their proliferation and/or differentiation. The systemis complex since (a) many ligands and receptors that modulateinteraction between myeloid and mesenchymal cells and henceexpression of paracrine factors are present on both cell typesand (b) many of the ligands and/or receptors are themselvesregulated by cytokines. The authors also summarize data suggestingthat bone marrow endothelium acts as a niche for a functionallydistinct cell population. In related findings, the work of Brandiand Collin-Osdoby indicates that vascular endothelial cellsrespond to inflammatory cytokines by secreting receptor activatorof NF-κB ligand (RANKL) and chemokines that are chemoattractivefor osteoclast precursors and augment the osteoclastogenic capacityof RANKL (49).
Bone marrow is also the environment in which specific metastasesmanifest their osteolytic and/or osteogenic phenotypes. Cancercells facilitate their infiltration into the marrow cavity bystimulating osteoclast formation and function. An initial stimulusis PTHrP generation by lung and breast cancer cells (50-52),thus enhancing mesenchymal production of RANKL and M-CSF whiledecreasing that of OPG (see Osteoclast biology). The resultingincrease in matrix dissolution releases bone-residing cytokinesand growth factors that, by feedback mechanisms, increase growthand/or survival of cancer cells. This loop has been termed "thevicious cycle" (Figure 2) (51). Multiple myeloma uses a differentbut related strategy, namely secretion of macrophage inhibitoryprotein 1α (MIP-1α) and monocyte chemoattractant protein-1 (MCP-1),both of which are chemotactic and proliferative for osteoclastprecursors (53, 54). Metastasis of prostate cancer to bone hasboth lytic and blastic components, with the net result beingdeposition of excess woven bone (55). A recent study suggeststhat BMP-6, produced by cancer cells, stimulates their invasiveand proliferative capacity (56). Similarly, inhibition of VEGFalso blunts prostate metastasis (57), suggesting that multiplecytokines play a role in proliferation and invasion of thiscancer. Future studies will undoubtedly reveal additional moleculesmediating bone loss in metastatic disease.
The osteoclast is generated by differentiation and fusion ofprecursors of the monocyte/macrophage lineage, giving rise toa polykaryon with unique cellular and molecular properties.Two cytokines mediate basal osteoclastogenesis, RANKL (58) andM-CSF, also called CSF-1 (59). Both proteins, produced by marrowstromal cells and their derivative osteoblasts, are membranebound, and thus differentiation of osteoclasts requires directinteraction of these nonhematopoietic, bone-residing cells withosteoclast precursors (60). The discovery of RANKL was precededby identification of its physiological inhibitor OPG, to whichit binds with high affinity (61). M-CSF regulates many aspectsof myeloid precursors and mature osteoclasts, including proliferationand/or survival, differentiation, and the cytoskeletal rearrangementsrequired for efficient bone resorption (59).
A combination of biochemistry and genetics helps to explainhow osteoclasts resorb bone (58, 62). The capacity of osteoclaststo isolate the enclosed area between themselves and the underlyingbone is at the heart of their function. The acidic pH (4.5)decalcifies the tissue, exposing the organic matrix to degradationby lysosomal-derived proteases, particularly cathepsin K (Figure3). The fact that dysfunction of the proton pump, Cl–channel, or cathepsin K results in human diseases of excessbone mass, namely osteopetrosis or pyknodysostosis (58, 62),attests to their critical role in osteoclast function.
This model of bone degradation requires close apposition betweenthe osteoclast and bone, a role provided by integrins in theform of αß heterodimers (63). Consistent with the factthat αvß3 is the principal integrin in osteoclasts,ß3–/– mice are hypocalcemic and generatefewer and shallower resorptive lacunae on dentin slices thando their wild-type counterparts (62). Based on these and manyin vitro observations, small-molecule inhibitors of osteoclastfunction that target αvß3 are in development (64).
Integrin activation mediates both cellular adhesion and transmembranesignaling (65). Important downstream transducers include theproto-oncogene c-src, important for membrane ruffling and osteoclastmigration (66), and Rac and Rho, members of a small subfamilyof the small GTPase superfamily that are central to remodelingof the actin cytoskeleton in many cell types (67) and play asimilar role in osteoclastic bone resorption (68). It is nowclear that bisphosphonates block bone resorption by inhibitingmembrane targeting of a number of small GTPases (69).
The steroid hormone 1,25(OH)2D3 plays a major role in regulatingcalcium and phosphate homeostasis. Deficiency of the hormoneincreases bone loss by altering the RANKL/OPG ratio secondaryto hypocalcemia and resulting in hyperparathyroidism (61). Incontrast, high levels of the steroid directly stimulate mesenchymalcell expression of RANKL and suppresses that of OPG (61) aswell as suppress the proosteoclastogenic hormone PTH (70). Arecent intriguing study identifies a vitamin D analog that preventsbone loss in oophorectomized mice by inhibiting RANKL-inducedexpression of c-Fos, a transcription factor required for osteoclastogenesis(71).
Both endogenous glucocorticoids and their synthetic analogs,which continue to be a mainstay of immunosuppressive therapy,have major impacts on bone biology (38) because of severe osteoporosisarising from decreased bone formation and resorption (low-turnoverosteoporosis). The majority of the evidence focuses on the osteoblastas the prime target, with glucocorticoids increasing apoptosisof these bone-forming cells (72). However, human studies documenta rapid initial decrease in bone resorption, suggesting thatthe osteoclast and/or its precursors may also be impacted bythe steroid via an ill-defined mechanism. One possibility isthat the apoptotic impact on osteoblasts decreases local levelsof RANKL and M-CSF. Alternatively, glucocorticoids have beenshown to decrease osteoclast apoptosis (73).
A wide range of clinical information demonstrates that excessprostaglandins stimulate bone loss by targeting stromal andosteoblastic cells, thus stimulating expression of RANKL andsuppressing that of OPG (33). This increase in the RANKL/OPGratio is sufficient in itself to explain the clinical findingsof increased osteoclastic activity. However, highlighting againthe dilemma of interpreting in vitro studies, prostaglandinsregulate osteoclastogenesis per se in murine cell cultures buthave anabolic consequences in vivo (74).
In addition to M-CSF and RANKL, several proteins play importantroles in osteoclast biology (Figure 4). OPG, an endogenous RANKLinhibitor, is secreted by mesenchymal cells both basally andin response to other regulatory signals, including cytokinesand bone-targeting steroids (61). Genetic deletion of OPG inmice and humans leads to profound osteopetrosis (58, 75), whileglobal overexpression of the molecule in mice results in severeosteoporosis (58). Importantly, postmenopausal women exhibithigher levels of RANKL on their bone marrow stroma, and treatmentwith estrogen reverses this outcome (76). Finally, a recentstudy demonstrates that a humanized, monospecific antibody toRANKL increases bone loss in osteoporosis patients (77). Together,these observations indicate that circulating OPG modulates thebone-resorptive activity of RANKL and suggest that the RANKL/OPGratio in serum will become a clinically important index.
Proinflammatory cytokines suppress OPG expression while enhancingthat of RANKL (61). Thus many patients with osteolytic diseasehave altered serum levels of RANKL and/or OPG as a result ofhigh levels of TNF-, IL-1, PTH, or PTHrP. Serum PTH levels areincreased in hyperparathyroidism of whatever etiology, whereasPTHrP is secreted by bone-targeting lung and breast carcinomas(50, 51). Attesting to the importance of TNF-, antibodies againstthe cytokine or a soluble TNF receptor–IgG fusion proteinpotently suppress bone loss in disorders of inflammatory osteolysissuch as rheumatoid arthritis (78). In humans, an IgG fusionprotein containing the active component of the IL-1 receptorantagonist enhances the ability of anti–TNF- antibodiesto decrease bone loss in rheumatoid arthritis (79).
It has become clear over the past few years that the immunesystem plays an important role in bone biology; perhaps nowhereis this more evident than in the work of the Pacifici group.In this series, the Review by M. Neale Weitzmann and RobertoPacifici explores how estrogen deprivation leads to bone loss(80). In short, they report that estrogen, the main sex steroidregulating bone mass in both men and women (81), suppressesthe capacity of antigen-presenting cells to stimulate T cellactivation, an event that normally results in secretion of arange of proresorptive cytokines including TNF-, IL-6, and RANKL.Notably, in this circumstance RANKL is cleaved from the cellsurface. TNF- and IL-6 act on stromal cells and osteoclast precursorsto enhance bone resorption by regulating expression of pro-(i.e., RANKL and M-CSF) and antiosteoclastogenic (i.e., OPG)cytokines in the case of mesenchymal cells and by synergizingwith RANKL itself in the case of myeloid osteoclast precursors.
While these studies underscore the interface between bone andthe immune system, they represent only a fraction of the workin the "hot" new area of osteoimmunology. The elegant work ofTakayanagi and colleagues suggests that IFN- is an importantsuppressor of osteoclast formation and function (82). Nevertheless,these findings are in conflict with in vivo observations inhumans, including the report that IFN- treatment of childrenwith osteopetrosis ameliorates the disease (83) and the factthat a number of in vivo studies indicate that IFN- stimulatesbone resorption (ref. 84 and references therein). This conundrumhighlights again the need to discriminate between in vitro cultureexperiments using single cytokines and results in vivo. Manyadditional reports have implicated other cytokines in regulationof the osteoclast, including numerous ILs, GM-CSF, IFN-, stromalcell–derived factor 1, MIP-1, and MCP (85-88), but atthis time the results are either contradictory, as for GM-CSFin the murine versus human systems, or lack direct proof inhumans. Future studies are likely to clarify the currently confusingdata set. Finally, several groups have provided evidence thatmolecules traditionally considered as solely immune receptors,such as DNAX activating protein of 12 kDa (DAP12), FcR and triggeringreceptors expressed in myeloid cells, and their ligands on cellsof the stromal and myeloid/lymphoid lineages, are downstreamof RANKL and M-CSF signaling (89, 90). Mutations in DAP12 leadto a rare bone disease, Nasu-Hakola disease, in which patientsexhibit bone cysts and concomitant demyelination (91).
Finally, we return to the question posed in the introduction.What are the common links between skin and bone? One recurringtheme is that Wnts are important and exhibit diversity in functionin both tissues. In the skin, Wnts, BMPs, and other signalingpathways are required for early patterning and for morphogenesisof the skin appendages including hair, teeth, nails, and mammaryglands (28, 29). Likewise, in bone, Wnts, BMPs, and FGFs arerequired, among other factors, for critical aspects of bonedevelopment. In early development, the overlying ectoderm playsan instructive role and secretes morphogens to the underlyingbone in the form of the apical ectodermal ridge during limboutgrowth (92, 93). In the adult, Wnt signaling is reprisedagain during tissue homeostasis and is a key regulator of thehair cycle as well as osteoblast and osteoclast regulation.Both skin and bone are home to specialized stem cell nichesthat provide a safe haven for their relevant progenitor cellpopulations. There are common transcriptional regulators ofbone and skin, such as the vitamin D receptor, whose cascadeof downstream effectors are clearly crucial in both tissues,since mutations in this gene cause profound effects in the formof irreversible alopecia as well as rickets (94). It has alreadybeen shown that hair follicle dermal papilla cells can differentiateinto adipocyte and osteogenic cells in culture (95), and further,that they can repopulate the entire hematopoietic system (96).The dermal papilla is easily identified within the skin by itsstrong expression of alkaline phosphatase, generally considereda marker of bone formation (97). Given the anatomical proximityof the two tissues, it is not beyond the realm of possibilitythat a common adult mesenchymal progenitor cell population supportsboth the skin dermis and underlying bone. Despite their phenotypicextremes of softness and hardness, future studies in skin andbone biology may reveal that they have far more in common thanmeets the eye.
Nonstandard abbreviations used: BMP, bone morphogenic protein; BP, bullous pemphigoid; Dkk, Dickkopf; HPV, human papillomavirus; LRP, LDL receptor–related protein; MIP-1, macrophage inhibitory protein 1; OPG, osteoprotegerin; PTH, parathyroid hormone; PTHrP, PTH-related protein; PV, pemphigus vulgaris; RANKL, receptor activator of NF-B ligand; TCF/LEF, T cell factor/lymphoid enhancer binding factor.
Conflict of interest: The authors have declared that no conflict of interest exists.
Figure 1 . Schematic illustration of several key elements of the skin. As indicated, the skin is divided into 2 main compartments, epidermal (tan) and dermal (pink), separated by a basement membrane (blue). The epidermis serves as the protective barrier, due to the differentiation of proliferative epithelial cells in the basal layer to the terminally differentiated cells in the stratum corneum. The sites and development of HPV infection are indicated at far left: the primary infection occurs in the basal layer, and mature virion sheds in the stratum corneum. Merkel cells are located within the basal layer of the epidermis (purple) and are associated with sensory nerve endings (brown). Two types of cell junctions critical for the integrity of the skin are highlighted here. Desmosomes (upper right inset) form cell-cell junctions, in which cadherins (pink and green), such as desmogleins 1 and 3, are the extracellular bridges and the autoantigens in different forms of pemphigus. Hemidesmosomes (lower right inset) tether the cells to the basement membrane and are composed of a number of proteins, 2 of which — BP230 and BP180 (also known as type XVII collagen) — are autoantigens in BP. Disruption of these interactions results in loss of adhesion of the cells to one another or to the underlying basement membrane.
Figure 2 . Cell-cell interactions in bone marrow. HSCs, the precursors of osteoclasts, reside in a stem cell niche provided by osteoblasts, which, together with stromal cells, derive from mesenchymal stem cells. Bone degradation (arrows) results in release of matrix-associated growth factors, which stimulate mesenchymal cells and thus bone formation. This "coupling" is an essential consequence of osteoclast activity (98). Additionally, matrix-derived factors stimulate cancer cell proliferation in the so-called "vicious cycle." Finally, cancer cells release cytokines that target mesenchymal cells and thus activate bone resorption.
Figure 3 . Mechanism of osteoclastic bone resorption. The osteoclast adheres to bone via binding of RGD-containing proteins (green triangle) to the integrin αvß3, initiating signals that lead to insertion into the plasma membrane of lysosomal vesicles that contain cathepsin K (Ctsk). Consequently, the cells generate a ruffled border above the resorption lacuna, into which is secreted hydrochloric acid and acidic proteases such as cathepsin K. The acid is generated by the combined actions of a vacuolar H+ ATPase (red arrow), its coupled Cl– channel (pink box), and a basolateral chloride–bicarbonate exchanger. Carbonic anhydrase converts CO2 and H2O into H+ and HCO3–. Solubilized mineral components are released when the cell migrates; organic degradation products are partially released similarly and partially transcytosed to the basolateral surface for release.
Figure 4 . Role of cytokines, hormones, steroids, and prostaglandins in osteoclast formation. Under the influence of other cytokines (not shown), an HSC commits to the myeloid lineage, expresses the M-CSF receptor colony-stimulating factor receptor 1 (c-Fms) and then, driven by M-CSF/c-Fms signaling and the RANKL receptor RANK, differentiates into an osteoclast. Mesenchymal cells in the marrow respond to a range of hormonal and cytokine stimuli, secreting a mixture of pro- and antiosteoclastogenic proteins, the latter primarily being OPG. Glucocorticoids (GCs) suppress bone resorption indirectly (by inducing death of osteoblasts) but possibly also target osteoclasts and/or their precursors. Estrogen (E2) inhibits secretion of RANKL and TNF-α by T cells via a complex mechanism (not shown); the sex steroid also inhibits osteoclast function.