Distinct mechanisms underlie pattern formation in the skin and skin ...

Distinct Mechanisms Underlie Pattern Formationin the Skin and Skin Appendages

Randall B. Widelitz,1 Ruth E. Baker,2 Maksim Plikus,1 Chih-Min Lin,1

Philip K. Maini,2 Ralf Paus3 and Cheng Ming Chuong1*

Patterns form with the break of homogeneity and lead to the emergence ofnew structure or arrangement. There are different physiological and patho-logical mechanisms that lead to the formation of patterns. Here, we firstintroduce the basics of pattern formation and their possible biological ba-sis. We then discuss different categories of skin patterns and their potentialunderlying molecular mechanisms. Some patterns, such as the lines ofBlaschko and Naevus, are based on cell lineage and genetic mosaicism.Other patterns, such as regionally specific skin appendages, can be set bydistinct combinatorial molecular codes, which in turn may be set by mor-phogenetic gradients. There are also some patterns, such as the arrange-ment of hair follicles (hair whorls) and fingerprints, which involve geneticsas well as stochastic epigenetic events based on physiochemical principles.Many appendage primordia are laid out in developmental waves. In theadult, some patterns, such as those involving cycling hair follicles, mayappear as traveling waves in mice. Since skin appendages can renewthemselves in regeneration, their size and shape can still change in theadult via regulation by hormones and the environment. Some lesion pat-terns are based on pathological changes involving the above processes andcan be used as diagnostic criteria in medicine. Understanding the differentmechanisms that lead to patterns in the skin will help us appreciate theirfull significance in morphogenesis and medical research. Much remains tobe learned about complex pattern formation, if we are to bridge the gapbetween molecular biology and organism phenotypes. Birth DefectsResearch (Part C) 78:280–291, 2006. VC 2006 Wiley-Liss, Inc.

INTRODUCTION TOPATTERN FORMATION

What is a pattern? Patterning canbe considered as the loss of homo-geneity, when small, random per-turbations to a system are amplifiedthrough a number of local processesand iterations to form recognizablestructure or order (Meinhardt, 1982;Ermentrout and Edelstein-Keshet,1993; Murray, 2003; Chuong et al.,2006a). For example, one of the sim-plest forms of patterning is the

asymmetric conversion of part of ahomogenous field (Fig. 1A; gray) toa different state (Fig. 1B; black). Thenew pattern can be generated asdots, stripes, patches, segments,branches, etc. (Fig. 1C–E), and can bearranged randomly or periodically.What are the mechanisms of bio-

logical pattern formation? In somecases, they may be based on thedistribution of cell lineage so thatcells strictly follow their fates genet-ically (Fig. 1F). In other cases, it

may be based on combinatorial mo-lecular coding which can be inter-preted at the enhancer/transcrip-tion factor level (Small and Levine,1991) or at the cell adhesion level(Steinberg, 1996) (Fig. 1G and H).These molecular changes usuallyappear before the real morphologi-cal changes and are referred to asprepatterns (Nagorcka and Mooney,1992; Forgacs and Newman, 2005).They can explain many downstreamphenomena that follow the gener-ated prepattern, but do not explainthe upstream issue—we do notknow how these molecular codesare set up. For example, morphoge-netic gradient models have beenproposed in which cells can interprettheir position within a morphogengradient, as Wolpert (1969) hasproposed in the French flag model(Fig. 1C). Cells can enter a newstate in a concentration dependentmanner (Fig. 1I and J) (Ashe andBriscoe, 2006). This can explainmany examples of how molecularcodes are set, but still does notresolve the issue as to the origin ofthe pattern—we still do not knowwhat sets up the molecular gradi-ent, for example, how the exactmorphogen and its point of secre-tion are determined. This is whereself-organization comes into play:stochastic events combinedwith physi-

REVIE

W

VC 2006 Wiley-Liss, Inc.

Birth Defects Research (Part C) 78:280–291 (2006)

Randall B. Widelitz, Maksim Plikus, Chihmin Lin, and Cheng Ming Chuong are from the Department of Pathology, Keck School ofMedicine, University of Southern California, Los Angeles, California.Ruth E. Baker and Philip K. Maini are from the Centre for Mathematical Biology, Mathematical Institute, Oxford, UK.Ralf Paus is from the Department of Dermatology, University Hospital Schleswig-Holstein, University of Lubeck, Lubeck, Germany.

Grant sponsor: NIAMS; Grant numbers: AR052397 (to R.W.), AR2177 (to C.M.C.), AR47364 (to C.M.C.), Grant sponsor: Astor TravelFund (to visit the Laboratory of Tissue Engineering; to R.E.B.); Grant sponsor: Lloyds Tercentenary Foundation (to R.E.B.).

*Correspondence to: Cheng-Ming Chuong, MD, PhD, Department of Pathology, University of Southern California, HMR 315B, 2011 ZonalAve, Los Angeles, CA 90033. E-mail: [email protected]

Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/bdrc.20075

Figure 1. Basics of pattern formation. A–E: Schematic drawings showing basic patterns. F–T: Possible mechanisms that can lead topattern changes. U–Z: Additional factors that can influence biological pattern formatting due to growth. E: From Yue et al. (2005). M:From Jiang et al. (2004). U–X: Represent the trunk with the midline on the top. There are three possible ways new cells can be added,which are indicated by the green color. Y,Z: Indicate the changes of field shape which can represent the growth of limb bud, tail bud, orfeather bud. See text for further explanation.

cochemical principles can increasethe order and/or structure of a sys-tem, perhaps resulting in emergentevents, without being guided by anexternal source (Newman and Com-per, 1990; Camazine et al., 2003;Newman et al., 2006). In otherwords, patterns at the global levelmay solely arise as a result of inter-actions between lower level compo-nents.Reaction diffusion models follow-

ing the method first outlined byTuring (1952) have been applied toexplain many biological periodicpatterning processes (Fig. 1K andL) (Gierer and Meinhardt, 1972). Inthis model, the morphogenetic fieldstarts with a homogenous distribu-tion of cells, activators, and inhibi-tors, and random fluctuations initi-ate the periodic patterning process.The activators and inhibitors un-dergo a series of interactions, whichcan include self- and cross-activa-tion and inhibition; both can diffuse,with the inhibitor diffusing furtherthan the activator. With time, pat-terns in the form of dots or stripes inactivator and inhibitor concentrationgradually emerge, with the patterndepending on the ratio of activatorsto inhibitors and the size and shapeof the pattern field. This mechanismhas been proposed to explain theformation of hair follicles (Nagorckaand Mooney, 1992) and feather pat-terning (Jung et al., 1998).Cellular automata models have

also been proposed to explain manybiological phenomena (Fig. 1M)(Wolfram, 1992; Deutsch et al.,2004). In a general cellular autom-ata model, the field is divided into anumber of discrete ‘‘cells,’’ whichevolve through a number of timesteps, according to a set of rulesbased on the states of neighboring‘‘cells.’’ Each ‘‘cell’’ of the model cor-responds to an area of the patternfield and information on this area isstored as the ‘‘state’’ of the cell.Along this line, a digital hormonemodel has been developed to ex-plain the formation of dermal con-densations by feather mesenchymalcells (Shen et al., 2004). The rolethat stochastic interactions mayplay in hair cycling have beenexplored using a cellular automationmodel (Halloy et al., 2002).

Oscillation is another importantproperty that may be used in pat-terning. The oscillation can occur atthe level of a single cell, or at thelevel of an organ (e.g., hair andfeather follicle) (Fig. 1P, S, and T). Aclock and wavefront mechanism in-volving cellular oscillation has beenused to explain the formation ofsomites (Fig. 1N–P) (Pourquie, 2003).Oscillation of hair follicles in hair cy-cling becomes very visible in darklypigmented normal mice, such asC57BL6/J (e.g. chnemus et al.,2005) and in mutant mice, such asnude mice and Msx2 null mice (Mil-itzer, 2001; Ma et al., 2003; Suzukiet al., 2003; Mecklenburg et al.,2005). A model based on the Belou-sov-Zhabotinski reaction wasrecently suggested to explain thisphenomena of wave formation, al-though no underlying molecular ba-sis was identified (Fig. 1R; Fig. 5)(Suzuki et al., 2003).Of course, the physicochemical

events are still genetically basedsince the biological patterns arespecies-specific. A way to conceivethis is that DNA gives rise to RNAand proteins that build cells withunique physicochemical properties.At this level, groups of cells interactwith outcomes based on these prop-erties and the surrounding en-vironment—not just on the mole-cule itself. Therefore, the patternformation process is best appreci-ated as a combination of geneticand epigenetic events, and the re-sults are both deterministic andstochastic, as seen in the finger-prints of homozygotic twins: similarbut nonidentical (reviewed in Jianget al., 2004).Organs can also grow and change

their shape, size, and organizationduring development (Fig. 1S and T).Another level of variation is that themorphogenetic field (in this case,the surface of the animal body)changes in size and shape duringdevelopment (Fig. 1U–X). The wayin which these changes take placecan influence patterns which are ata formative stage. For example, du-ring the expansion of the skin, newdermal and/or epidermal cells maybe added to specific regions such asthose receiving cells from the der-matome or the advancing ventral

body folds. In other cases, new cellsmay be inserted randomly all overthe developing skin (Fig. 1U–X).These growth modes can have dif-ferent consequences for patternsforming on the skin. Similarly, theshape of the morphogenetic fieldmay change, even sometimes re-ducing in size, and this also maylead to variation in patterning; forexample, the formation of stripesrather than spots (Fig. 1Y and Z)(Murray, 2003).Our aim is to try to analyze these

pattern formation processes andidentify the biological bases under-lying them, but much remains to belearned. We have described aboveonly some examples, and they cer-tainly do not exhaust all models thathave been proposed for patterning.We can also contemplate that com-plex pattern formation is generatedthrough a combination of the aboveprocesses, which perhaps results inpatterns that are more robust togenetic and environmental pertur-bations. We will point out these pat-terns and the processes that may beinvolved in the following sections.

DEVELOPMENT-BASEDAND UNALTERABLEPATTERNS

Although the appearance of integu-ments of mammals, birds, and rep-tiles can be very different (Fig. 2A),the development of their skin andskin appendages share similar hier-archical morphogenetic processes(Fig. 2B). In some cases, differenttypes of skin appendages appear,while in others, the patterns of simi-lar types of appendages are ar-ranged differently (Ball, 1999). Wethink that this is controlled throughgenetic and epigenetic controls thatoperate at different levels (Jianget al., 2004). Here we take a closerlook at these regulatory processes.

Regional Specificity

Regional specificity implies that dif-ferent skin regions such as the scalp,beard, eyebrows, face, lips, palms,nails, mammary glands, sweat glands,etc., have different characteristics.Epidermal precursors (or stem cells)

282 WIDELITZ ET AL.

Birth Defects Research (Part C) 78:280–291, (2006)

are initially multipotent and compe-tent to form all these different struc-tures. During development, specialdomains of the dermis begin sending

specific messages to the epidermis.Through a series of epithelial–mesen-chymal interactions, these differentskin domains with special structures

and functions gradually emerge. Theintegument diversifies to endow dif-ferent functions to different parts ofthe human skin. An example of re-gional specificity can be seen bycomparing the different types offeathers present on the breast,wing, tail, etc., of birds. This is mostevident in pheasants, as shown inFig. 2A.How dermal specificity and epider-

mal competence are set up is mostlyunknown. A model based on a skinHox code was proposed, suggestingthat different combinations of Hoxgene expression may be the basis ofskin regional specificity, setting upthe subsequent differences in diffus-ible morphogens and adhesion mol-ecules (Chuong, 1993). DifferentHox expression patterns are shownin different regions of chicken skin(Kanzler et al., 1997; Duboule,1998; Reid and Gaunt, 2002). In-deed there are spatiotemporally de-fined, specific HOX expression pat-terns in human skin (Stelnicki et al.,1998), and the Hox expression pat-terns of dermal cells derived fromdifferent topological skin regions inhumans are different (Chang et al.,2002).Most interestingly, the characteris-

tics of these different skin regionscan be trans-determined. For ex-ample, the engrailed pathway wasshown to be involved in definingthe mesenchymal characteristics ofthe ventral versus the dorsal paw(Loomis et al., 1996). Tbx4 and Tbx5are shown to be involved in definingthe identity of the chicken leg versuswing, and hence scale or featherforming dermis (Rodriguez-Estebanet al., 1999). Epidermal cells cantrans-differentiate and converthairs into glands or scales intofeathers under the influence of ret-inoic acid, or by ectopic expressionof specific molecules such as b-cat-enin (Dhouailly et al., 1980; Robin-son et al., 1990; Widelitz et al.,2000). A recently engineered K14-noggin transgenic mouse showsthat sweat glands are transformedto hairs (Plikus et al., 2004), whilenoggin overexpression under theneuron-specific enolase promotercan convert outer root sheath ke-ratinocytes into sebocytes (Guhaet al., 2004). An adult cornea can

Figure 2. Patterns on avian skin and skin appendages and hierarchical morphogenesis.A: Male and female pheasants show regionally specific skin appendages and sexualdimorphism. Also note the thick pigment stripes and dots in the tail feather. Prum andWilliamson (2002) proposed a reaction diffusion model for feather pigment patterning.B: Different developmental stages of skin appendage morphogenesis. Note the differenttypes of skin appendages, including the schematic radially and bilaterally symmetricfeathers. Modified from Wu et al. (2004).

PATTERN FORMATION IN SKIN AND SKIN APPENDAGES 283


also be diverted to form piloseba-ceous units when they are con-fronted with embryonic hair formingdermis (Pearton et al., 2004). Theseobservations imply that the specificcombinatorial molecular codes mayspecify phenotypes of skin and skinappendages (Chuong, 1993; Fli-niaux et al., 2004; Prin andDhouailly, 2004). Altering thesecodes may lead to a resetting of thephenotypes. The upstream questionconcerning how the molecular codesare set remains unanswered.

Developmental Wave

During skin development, hair orfeather primordia are laid out in atemporal order as they graduallyacquire competence (reviewed inDhouailly et al., 2004). Their arrange-ment and orientation are reflectedas a propagating wave of skin ap-

pendage formation. This process isclearly shown in the chicken skin inFigure 3A. In the spinal tract of thechicken, the formation starts at themidline and spreads bilaterally. Atthe lateral edge, feather primordiaare at the induction stage. Towardthe midline, they progress to formshort feather buds, long feather buds,and feather filaments with branchingmorphogenesis, etc. A morphoge-netic wave sweeping from themidlineto the lateral has been inferred(Sengel, 1990). However, althoughthe lateral row appears after themore medial row, the formation ofthe lateral row does not really have todepend on the medial row (Jianget al., 1999). While this sequentialappearance may be perceived as a‘‘gradient,’’ it is actually a temporalwave since the lateral feather budswill eventually also go through featherbud and filament stages. For a tract,

there has to be a primary row beforethis sequential appearance takesplace. The gradual emergence of budsalong the primary row in the midlinecan be readily visualized by in situhybridization staining for b-catenin(Fig. 3B). Themolecular basis of theseprocess remains unknown.In humans, this is manifested as

hair whorl patterns in the occipitalregion (Fig. 1Q) (reviewed in Plikusand Chuong, 2004). In humanfetuses, lanugo hairs form whorlpatterns both on the scalp and trunkskin (Gworys and Domagala, 2003).On the thoracic wall there arelanugo whorls that begin bilaterallyover the nipples. The whorls col-lide and merge along the midlines(Domagala, personal communica-tion). In adults, whorl patterns aredistinct only on the parietal scalp. Isthe whorl pattern genetically con-trolled? A pair of homozygotic twins

Figure 3. Temporal wave. A: In situ hybridization of Shh in embryonic chicken skin. Midline is indicated by an arrow. Feather bud for-mation starts from the midline, and then lateral buds appear sequentially. From the lateral edge to the midline are regions of no featherprimordia, feather placodes, short buds, long buds, and feather filaments with branch formation. B: In situ hybridization of b-catenin.The feather field first homogenously expresses b-catenin at a moderate level in the morphogenetic zone. Then the periodicallyarranged buds emerge gradually expressing high levels of b-catenin, while the lateral inhibitory zone does not express beta catenin.From Widelitz et al. (2000).

284 WIDELITZ ET AL.


were shown to have one and twowhorls, respectively (Paine et al.,2004). Therefore, there must be anepigenetic component in hair whorldetermination. While conserved mo-lecular pathways underlie all hair fol-licle formation, local environmentaland fortuitous factors can influencethe final hair pattern.In the mouse, transgenic mice

that lose frizzled 6 show the forma-tion of multiple whorls (Guo et al.,2004), suggesting the involvementof the Wnt pathway in this process.Interestingly, some strains of guineapigs also show multiple whorls ontheir skin.The formation of fingerprints is

another example, as discussed inthe next section.

Periodic Patterning

This mechanism is most obviousin the formation of skin append-ages and pigment patterns. Duringthe formation of feather primordia,the epithelium has to become com-petent to respond to induction sig-nals (forming a field). A reactiondiffusion mechanism, involving ac-tivators and inhibitors, is proposedto operate in the dermal mesen-chyme (Jung et al., 1998; Jianget al., 1999). Through this mecha-nism, cells are set to become the pri-mordia of skin appendages, stochas-tically. This then leads to the forma-tion of the feather or hair primordia,evenly spaced with interfollicular skin.Similar processes were proposedfor hair/wool formation (Nagorckaand Mooney, 1985; Moore et al.,1998; Meinhardt and Gierer, 2000).It should be emphasized that theprocess of periodic patterning canbe uncoupled from the developmen-tal wave process discussed in theprevious paragraph.The sequential appearance of

feather buds is so exquisite (Fig.3A) that it led scientists to proposemodels that are based on the useof previous buds as templates(Murray and Oster, 1984). Theexperiments by Jiang et al. (1999)showed that in a reconstitution sit-uation, in which mesenchymal cellsare dissociated into single cells, theperiodic patterns will reform simul-

taneously. So the sequential appear-ance results from a global compe-tence wave imposed on the local per-iodic patterning process.Another dramatic example often

referred to is the dissolution of pig-ment cells that leads to the forma-tion of stripes on zebra fish orzebras and the formation of pig-ment ducts on fish or leopards. Thiswas addressed earlier by Murray(1993). Recently, Prum and Wil-liamson (2002) also applied a reac-tion diffusion mechanism to makea theoretical model of feather pig-ment patterns. However, some pig-ment patterns are controlled genet-ically by enhancer regions, asshown by differences in the DroopyEar mouse mutant. Here, the nor-mal, sharp delineation betweendorsal and ventral pigmentationpatterns is disrupted. This is pro-duced by a loss of function muta-tion in TBox 15, which then allowsagouti to be expressed further dor-sally (Candille et al., 2004).Do these patterns result from

genetic coding or stochastic events?In fact, it is likely to be both. For ex-ample, consider patterns such asfingerprints (Kucken and Newell,2005). They are used for individualidentification because the ridgewidth and possibility for branchingnodal points provide ample possibil-ity for endless variation. Finger-prints among monozygotic twinshave more similar attributes (similarwidth, organization plan) than withunrelated individuals, but they arenonidentical and are sufficiently dif-ferent to be used as individual iden-tifiers (Jain et al., 2002). Thus thereis a nongenetic component at thislevel of tissue morphogenesis, wheremolecular codes become indirectand cells interact based on physical-chemical rules.

Morphogenetic Gradient

We acknowledge the importance ofmolecular codes, and have pro-posed the skin Hox code hypothesisfor regional specificity of skin andskin appendages (Chuong, 1993).Yet, how are these molecular codesset up?A morphogenetic gradient, such

as an Shh gradient, has been used

to explain dorsal–ventral spinalcord determination (Monsoro-Burqand Le Douarin, 1999) and ante-rior–posterior (A–P) limb bud pat-terning (McGlinn and Tabin, 2006).Here we will use a recent exampledescribing how a Wnt 3a gradientin adult feather follicles patternsepithelial stem cells to form eitherradial or bilateral, symmetric feath-ers.In the adult bird, there are radially

symmetric body feathers and bilat-erally symmetric flight feathers.Feathers do not contain the bulgestructure found in hairs that housestem cells. We recently identifiedfeather stem cells located as a con-centric ring sitting at the bottom ofthe feather follicle (Yue et al., 2005).Interestingly, in radially symmetricfeathers, the stem cell ring is placedhorizontally. In bilaterally symmetricfeathers, the stem cell ring tilts to-ward the anterior rachis position.This topological difference led us topropose that there would be a breakof symmetry in the ramogenic planewhere feather branches start to form(Fig. 4). Indeed, in bilaterally sym-metric feathers, we found a Wnt 3agradient from anterior to posteriorpositions that does not exist in radi-ally symmetric feathers. Flatteningthe Wnt 3a gradient using RCAS ret-rovirus-mediated gene misexpres-sion converted bilaterally symmetricfeathers to radially symmetric feath-ers (Yue et al., 2006). A local Wnt 3agradient released from a bead causesthe forming barb ridge branches toswirl toward the Wnt 3a source.Swapping dermal papillae betweenradial and bilaterally symmetricfeathers shows that the dermal pa-pilla determines the gradient config-uration and feather symmetry, whilestem cells can respond to the mor-phogenetic gradient to make differ-ent forms of feathers. This is anexcellent example, showing how amicrogradient within a feather fol-licle can set the organ shape.

PATTERNS THAT CAN BECHANGED IN THE ADULT

Traveling Wave

Hair follicles go through regenerativecycles: they cycle through growth



(anagen), regression (catagen), hairshaft shedding (exogen), and rest-ing phases (telogen) (Paus and

Cotsarelis, 1999; Stenn and Paus,2001; Paus and Foitzik, 2004). Visu-alizing hair waves is facilitated by

hair cycle–dependent changes in in-tegument pigmentation and timedhair loss in nude mice (Militzer,2001; Suzuki et al., 2003), Msx2null mice (Ma et al., 2003; Mecklen-burg et al., 2005), calcineurin B1(Mammucari et al., 2005), etc. As aresult, we observe skin regions cy-cling through discrete stages, eachrepresenting a different ‘‘status’’ ofhair follicle (Fig. 1S and T). Theregions can appear as waves of dra-matic pattern changes across theadult mouse skin.Militzer (2001) analyzedmore than

400 nude mice on albino (NMRI,foxn1nu) and pigmented (C57BL/6,foxn1nu) backgrounds for more thanone year. Pink skin turns dark whenhairs enter anagen III and returns topink when anagen is completed. Skinpigmentation changes travel in awave-like fashion on the skin surfaceof these mice. When mice are young,all hair cycles initially synchronize,but with increasing age the haircycles over different regions de-synchronize. Thus, the skin pigmentpattern breaks into distinct stripesand patches. As mice age, thestripes and patches become nar-rower/smaller and eventually appearrandom.Dramatic traveling hair waves

occur in the Foxn1nu strain of nudemice on the C57BL/6 background(Suzuki et al., 2003). These mutantmice have a distinct defect in theFoxn1 gene that results in fasterhair cycling. Thus the dynamic pig-mentation pattern changes de-scribed above progress faster thanthose observed in classical nudemice. In young mice, the pigmenta-tion oscillation takes place synchro-nously throughout the skin. The widerpigmented stripes progress tobecome narrower bands as a miceage. Some mice (usually more thanseven months) show narrow,roughly evenly-spaced pigmentedstripes that travel along the trunk;however, many mice show irregular,fragmented, or very wide stripes(Fig. 5).The pattern can become more

complex. Ma et al. (2003) reported‘‘cyclic alopecia’’ in Msx2 knockoutmice. This phenotype is due to thefact that hair fibers are defectiveand are dislodged specifically during

Figure 4. Morphogenetic gradient. Left column shows an idealized radially symmetricfeather. Right column shows a bilaterally symmetric feather. A,E: The proximal follicleshows ordered compartments of stem cells (orange color), TA cells, and differentiating cells(ramogenic zone) (Yue et al., 2005). In radially symmetric feathers, the ring is horizontal.In the bilaterally symmetric feathers, the ring is tilted from zero to about 458. Themoleculargradient in the ramogenic zone is shown in shades of blue. B,C: In an open follicle prepara-tion, the feather filament cylinder is opened to form a plane. In radially symmetric feathers,all new barb ridges form at the same time and in parallel. In bilaterally symmetric feathers,the tilting of the stem cell ring results in a discrepancy of maturation due to the fact that theTA cells have to travel (or are displaced) different distances before they reach the ramo-genic zone (m1 and m2). On the anterior side, cells are more mature. The shift in cell posi-tion is represented by vectors AB, AC, and AD. D: According to this model, there should bea molecular gradient along the A–P axis. Indeed, we found a Wnt 3a gradient. Flatteningthe gradient converted feathers from bilateral to radial symmetry (Yue et al., 2006).

286 WIDELITZ ET AL.


the catagen phase. The skin of thesemice during anagen is black andhairy, but during telogen is bald andnonpigmented. As the hairs cycle,the alopecic regions reenter anagenand regain pigmentation in a pro-gressive order. Long-term observa-tion of hairy and bald skin regionsrevealed a ‘‘cyclic alopecia’’ phe-nomenon. Hairs within one skin do-main cycle in waves but not withhairs in neighboring domains (whichalso cycle in waves, but with an in-dependent rhythm).In essence, the ‘‘traveling stripes’’

of the Foxn1nu mice are a manifesta-tion of the same phenomenon. Notch1 activation in keratinocytes can gothrough the RBP or calcineurin B1pathways. Recently, mice with a cal-cineurin B1 deletion also showed acyclic alopecia phenotype (Mammu-cari et al., 2005). In adult humans,most hair follicle cycles independ-ently, so there are no wave-like pat-terns.

Hormone-Based Changesof Appendage Pattern

Since hair or feather follicles can gothrough regenerative cycles, a com-pletely new type of skin appendagecan reform with a new shape andsize through a regenerative cyclingmechanism. This is most obvious insex hormone–dependent changes

(Mayer et al., 2004). Upon puberty,skin appendages in specific regionsare transformed when sex hor-mone (estrogen and androgen)pathways are activated. Sex ste-roids can also affect the melano-genic activity of epidermal melano-cytes, giving rise to hormonally-based changes of skin and skin ap-pendage patterns as evident inbirds (Fig. 2A).This is most apparent in tail

feathers of hens/roosters and pea-hens/peacocks. Sexual dimorphismcharacteristics are also observed inmammals, including humans(Wheeler, 1991). In the humanbeard, axilla, and genital regions,hair follicles are transformed fromthe vellus to the terminal state.With increasing age, the reversetends to occur, leading to andro-genic alopecia. Vellus hairs also canbe transformed into unwanted ter-minal hairs (e.g., on the upper lipand lower legs) when properlystimulated by androgens, leadingto hirsutism. Here, terminal hairs inthe frontal and parietal scalp areaffected, but not those in the occi-pital region. As a result, the type ofhairs that form and the region ofhair growth (hairline) change at dif-ferent ages. The mechanism con-trolling how scalp and occipital hairsrespond to sex hormones is notknown, but appears to be mediated

by differences in dermal papillae,which exhibit varying response tostimulation with androgens orestrogens (Randall et al., 2001;Inui et al., 2002; Conrad et al.,2005). In fact, estrogens and estro-gen receptor-mediated signaling arepowerful mediators or even inducersof wave pattern formation, namelyof hair waves in mice (Ohnemuset al., 2006). Thus, hormonally-based skin lesion patterns are alsothe consequence of region-specificdevelopmental programming.

Environmentally-Based PatternChanges

Change in light/dark cycles producedby the seasonal lengthening andshortening of days or changes intemperature can alter the type or col-oring of skin appendages. This canbe seen in the seasonal (summer/winter) hair coat variation of horses,snow shoe rabbits, etc. In nature,changes in the length of the light pe-riod are translated into changes inthe plasma melatonin and prolactinlevels, which can trigger animals toproduce a longer/shorter or whiter/darker coat to improve their chan-ces for survival during a given sea-son (Rose et al., 1987). Now that weknow that human and rodent skinand hair follicles are extrapituitarysites of melatonin synthesis (Slo-minski et al., 2002, 2003; Kobaya-shi et al., 2005), one wonders towhat extent environmental cues(such as the length of the light pe-riod) can also affect seasonal changesin skin and skin appendage patterns.

Pattern Formation of/in SkinLesions

Since the skin covers the surface ofan individual, patterns on the skinare the most recognizable. Theyhave been used as diagnostic cluesto the dermatologist (Ackerman,1997; Bolognia et al., 2003; Sterryet al., 2006). In addition to develop-mental and physiological causes,patterns on the skin that developcan be due to pathological or artifac-tual causes. A multiauthored reviewfeaturing several view points focus-ing on skin lesion patterns was re-cently edited by Dr. Ralf Paus for

Figure 5. Traveling stripes. A: In the adult mouse, hair follicles go through regenerativecycling. They appear as black in the anagen period. In this mutant nude mouse, hair fila-ments are lost in the telogen period and appear white. This helps us visualize the chang-ing states of hair follicles, which appear as traveling waves (after Suzuki et al., 2003; Pli-kus and Chuong, 2004). Arrows describe the direction of wave propagation. B: Msx2 nullmice show cyclic alopecia in which hair shafts are dislodged at a specific time of haircycle but can also regenerate. As a result, patches of hairy domains (black) and balddomains (white) are formed. These domains alternate between growth and restingphases, and give the impression of traveling stripes. The shape and size of thesedomains, their relative configuration changes over time, and situation in (A) is a specialexample of this phenomenon. Based on Ma et al. (2003).



Experimental Dermatology (Chuonget al., 2006b). Here, we will brieflysummarize those discussions.

Lineage, Genetically-Based

Molecular expression within cells ischanged genetically or epigeneticallyduring development. The changescan be transmitted to daughter cellsbecause they involve somatic muta-tions in DNA or are mediated by epi-genetic mechanisms such as X-chro-mosome inactivation, DNA methyla-tion, etc. This collection of differentpatterns has mainly been studied inhuman diseases. The offspring of themutated cells share a similar abnor-mality. The distinct phenotypes ofthese cells thenmanifest themselves

in the skin. These ectopic changesare named Naevus (Happle, 1995)(see Glossary). There are severalstriking examples in which lesionsare limited to the left or right side ofthe body, regional segments, check-erboard patterns, or linear distribu-tions (Fig. 6A) (Happle, 1993, 1995,2004). The most striking example inthe epidermal lineage is the Blaschkolines (Jackson, 1976). A recent caseof linearly distributed acne turnedout to be due to a somatic mutationin the FGF receptor in one epider-mal cell lineage (Munro and Wilkie,1998). The mechanism leading tothe Blaschko lines is fundamentaland not limited to humans. Whenearly chicken embryo epidermal cells(embryonic day 2 [ED 2]) were la-

beled along the dorsal midline withreplication-defective virus express-ing b-galactosidase, their cellulardescendents showed multiple paral-lel blue lines, resembling Christmastree branches, radiating from themidline across the dorsal skin of latechicken embryos (Fig. 6B and C)(Chuong et al., 1998). However,analyses of these patterns show thatformation of feather primordia orfeather filaments are not based onlineage, but on the local environmentat the time of formation (Fig. 6D).There are many different types of

ectodermal organs on the integu-ment. Many of them share morpho-genetic signaling pathways. Pertur-bation of one pathway can lead tochanges in multiple organs (Plikus

Figure 6. Genetic mosaicism on the skin. A: Lines of Blaschko. Through X chromosome inactivation, the lineage of epithelia cells canbe seen to be distributed in lines horizontal to the A–P axis. Several examples of checkerboard or patch patterns on human skin areseen in several human diseases (Happle, 1995, 2004). After Happle’s viewpoint 2 in Chuong et al. (2006). B: Equivalent lines ofBlaschko in embryonic chicken. Embryos are injected with nonreplicative virus carrying b-galactosidase. C: Line drawing of (B). D: Dif-ferent cell lineages are represented by different colors. Analyses show that individual feather buds or individual barb ridges are madeof cells from different lineages, not from a single lineage. Therefore, the local environment at the time of feather morphogenesis ismore important than lineage. B–D: From Chuong et al. (1998).

288 WIDELITZ ET AL.


et al., 2004). In humans, when amolecule, such as EDA, that is fun-damental to these processes is mu-tated, it can lead to ectodermal dys-plasia that affect multiple epithelialorgans (Bolognia et al., 2003).

Anatomically- orPhysiologically-Based

Distinct regions such as skin ap-pendages, skin ridges, cutaneousnerves, blood vessels, etc., can con-tribute to patterns of skin lesions.When skin lesions develop, theymay follow these obvious anatomi-cal borders or follow ‘‘hidden’’ latentpatterns based on physiological dif-ferences. Through various pathoge-netic mechanisms, these differentskin regions may result in differentsusceptibility to diseases. It is uponthis dynamic landscape that skinlesions develop, and become dis-tributed and shaped.

Artificial

Human behaviors can also cause pat-terned lesions. For example, chronicsun exposure can lead to the charac-teristic ultraviolet (UV)-light-inducedpatterns corresponding to unclothedskin regions. Tattoos, skin paintings,hair dyes, cosmetic surgery, etc.,can lead to further visible patternedchanges on the skin.

CONCLUSION

The skin is an excitable medium. Indevelopment, it conducts reactionsamong signaling molecules thatdetermine the formation of skinappendages or the distribution ofactive melanocytes. In the adult, re-generative hair cycling provides arich opportunity for the skin torenew itself based on hormonal andenvironmental changes. Patterns ofskin lesions provide diagnostic cluesto skin or systematic diseases. Theconvergence of genetic, epigenetic,and regenerative events to generatecomplex patterns on this very visibleorgan also provides a great experi-mental opportunity to study themany unknown mechanisms of bio-logical pattern formation.

ACKNOWLEDGMENTS

This work is based on two of ourrecent multiauthor reviews (ChuongCM, Dhouailly D, Gilmore S, ForestL, Shelley WB, Stenn KS, Maini P,Michon F, Parimoo S, Cadau S,Demongeot J, Zheng Y, Paus R, Hap-ple R. What is the biological basis ofpattern formation of skin lesions?Experimental Dermatology 2006;15:547–564; Plikus M, Chuong C-M.Making waves with hairs. Journal ofInvestigative Dermatology 2004;122:vii–ix) and on further development.We appreciate the contributions ofthe previous coauthors.

REFERENCES

Ackerman AB. 1997. Histologic diagno-sis of inflammatory skin diseases: analgorithmic method based on patternanalysis. 2nd ed. Baltimore: Wilkins &Wilkins.

Ashe HL, Briscoe J. 2006. The interpre-tation of morphogen gradients. Devel-opment 133:385–394.

Ball P. 1999. The self-made tapestry:pattern formation in nature. Oxford:Oxford University Press.

Bolognia JL, Jorizzo JL, Rapini RP, editors.2003. Dermatology. Mosby: London.

Camazine S, Deneubourg JL, Franks NR,et al. 2003. Self-organization in biolog-ical systems. Princeton: Princeton Uni-versity Press.

Candille SI, Van Raamsdonk CD, Chen C,et al. 2004. Dorsoventral patterning ofthe mouse coat by Tbx15. PLoS Biol2:E3.

Chang HY, Chi JT, Dudoit S, et al. 2002.Diversity, topographic differentiation,and positional memory in human fibro-blasts. Proc Natl Acad Sci USA 99:12877–12882.

Chuong CM. 1993. The making of afeather: homeoproteins, retinoids andadhesion molecules. Bioessays 15:513–521.

Chuong CM, Jung HS, Noden D, et al.1998. Lineage and pluripotentiality ofepithelial precursor cells in developingchicken skin. Cell Biol Biochem 76:1069–1077.

Chuong CM, Wu P, Plikus MV, et al.2006a. Engineering stem cells intoorgans: topobiological transformationsdemonstrated by beak, feather andother ectodermal organ morphogene-sis. Curr Top Dev Biol 72:237–274.

Chuong CM, Dhouailly D, Gilmore S,et al. 2006b. What is the biological ba-sis of pattern formation of skin lesions?Exp Dermatol 15:547–564.

Conrad F, Ohnemus U, Bodo E, et al.2005. Substantial sex-dependent dif-ferences in the response of humanscalp hair follicles to estrogen stimula-tion in vitro advocate gender-tailored

management of female versus malepattern balding. J Investig DermatolSymp Proc 10:243–246.

Deutsch A, Dormann S, Maini PK.2004. Cellular automaton modeling ofbiological pattern formation. Boston:Birkhauser.

Dhouailly D, Hardy MH, Sengel P. 1980.Formation of feathers on chick footscales: a stage-dependent morphoge-netic response to retinoic acid. J EmbryolExp Morphol 58:63–78.

Dhouailly D, Olivera-Martinez I, Fliniaux I,et al. 2004. Skin field formation: mor-phogenetic events. Int J Dev Biol 48:85–91.

Duboule D. 1998. Hox is in the hair: abreak in colinearity? Genes Dev 12:1–4.

Ermentrout Gb, Edelstein-Keshet. 1993.Cellular automata approaches to bio-logical modeling. J Theoret Biol160:97–133.

Fliniaux I, Viallet JP, Dhouailly D. 2004.Ventral vs. dorsal chick dermal progen-itor specification. Int J Dev Biol 48:103–106.

Forgacs G, Newman SA. 2005. Biologicalphysics of the developing embryo.New York: Cambridge University Press.

Gierer A, Meinhardt H. 1972. A theoryof biological pattern formation. Kyber-netik 12:30–39.

Guha U, Mecklenburg L, Cowin P, et al.2004. BMP signaling regulates post-natal hair follicle differentiation andcycling. Am J Pathol 165:729–740.

Guo N, Hawkins C, Nathans J. 2004.Frizzled6 controls hair patterning inmice. Proc Natl Acad Sci USA 101:9277–9281.

Gworys B, Domagala Z. 2003. Thetypology of the human fetal lanugo onthe thorax. Ann Anat 185:383–386.

Halloy J, Bernard BA, Loussouam G,Goldbeter A. 2002. The follicular au-tomaton model: effect of stochasticityand of synchronization of hair cycles.J Theor Biol 214:469–479.

Happle R. 1993. Mosaicism in human skin.Understanding the patterns and me-chanisms. Arch Dermatol 120:1460–1470.

Happle R. 1995. Epidermal nevus syn-dromes. Semin Dermatol 14:111–121.

Happle R. 2004. Patterns on the skin.New aspects of their embryologic andgenetic causes. Hautarzt 55:960–961;964–968. [German]

Inui S, Fukuzato Y, Nakajima T, et al.2002. Androgen-inducible TGF-beta1from balding dermal papilla cells in-hibits epithelial cell growth: a clue tounderstand paradoxical effects of an-drogen on human hair growth. FASEBJ 16:1967–1969

Jackson R. 1976. The line of Blaschko: areview and reconsideration: observa-tions of the cause of certain unusuallinear conditions of the skin. Br J Der-matol 95:349–360.

Jain AK, Prabhakar S, Pankanti S. 2002.On the similarity of identical twin fin-gerprints. Pattern Recognit 35:2653–2663.



Jiang TX, Jung HS, Widelitz RB, et al.1999. Self-organization of periodicpatterns by dissociated feather mes-enchymal cells and the regulationof size, number and spacing of pri-mordia. Development 126:4997–5009.

Jiang TX, Wideltz RB, Shen WM, et al.2004. Integument pattern formationinvolves genetic and epigenetic con-trols operated at different levels:Feather arrays simulated by a digitalhormone model. Int J Dev Biol 48:117–136.

Jung HS, Francis-West PH, Widelitz RB,et al. 1998. Local inhibitory action ofBMPs and their relationships with acti-vators in feather formation: implica-tions for periodic patterning. Dev Biol196:11–23.

Kanzler B, Prin F, Thelu J, et al. 1997.CHOXC-8 and CHOXD-13 expressionin embryonic chick skin and cutane-ous appendage specification. Dev Dyn210:274–287.

Kobayashi H, Kromminga A, Dunlop TW,et al. 2005. A role of melatonin in neuro-ectodermal-mesodermal interactions:the hair follicle synthesizes melatoninand expresses functional melatonin re-ceptors. FASEB J 19:1710–1712.

Loomis CA, Harris E, Michaud J, et al.1996. The mouse Engrailed-1 gene andventral limb patterning. Nature 382:360–363.

Ma L, Liu J, Wu T, et al. 2003. Cyclic alo-pecia in Msx2 mutants: defects in haircycling and hair shaft differentiation.Development 130:379–389.

Mammucari C, Tommasi di Vignano A,Sharov AA, et al. 2005. Integration ofNotch 1 and calcineurin/NFAT signalingpathways in keratinocyte growth anddifferentiation control. Dev Cell 8:665–676.

Mayer JA, Chuong CM, Widelitz R. 2004.Rooster feathering, androgenic alope-cia, and hormone dependent tumorgrowth: what is in common? Differen-tiation 72:474–488.

McGlinn E, Tabin CJ. 2006. Mechanisticinsight into how Shh patterns the ver-tebrate limb. Curr Opin Genet Dev16:426–432.

Mecklenburg L, Tychsen B, Paus R.2005. Learning from nudity: lessonsfrom the nude phenotype. Exp Der-matol 14:797–810.

Meinhardt H. 1982. Models of biologicalpattern formation. New York: AcademicPress.

Meinhardt H, Gierer A. 2000. Pattern for-mation by local self-activation andlateral inhibition. Bioessays 22:753–760.

Militzer K. 2001. Hair growth pattern innude mice. Cells Tissues Organs 168:285–294.

Monsoro-Burq AH, Le Douarin N. 1999,Molecular aspects of vertebral chon-drogenesis. J Soc Biol 193:263–268.

Moore GP, Jackson N, Isaacs K, et al.1998. Pattern and morphogenesis inskin. J Theor Biol 191:87–94.

Munro CS, Wilkie AO. 1998. Epidermalmosaicism producing localised acne: so-matic mutation in FGFR2. Lancet 352:704–705.

Murray JD, Oster GF. 1984. Generationof biological pattern and form. IMAJ Math Appl Med Biol 1:51–75.

Murray JD. 1993. Mathematical biology.New York: Springer-Verlag.

Murray JD. 2003. Mathematical biologyII: spatial models and biomedical ap-plications. New York: Springer-Verlag.

Nagorcka BN, Mooney JR. 1985. Therole of a reaction-diffusion system inthe initiation of primary hair follicles.J Theor Biol 114:243–272.

Nagorcka BN, Mooney JR. 1992. Fromstripes to spots: prepatterns which canbe produced in the skin by a reaction-diffusion system. Math Med Biol 9:249–267.

Newman SA, Comper WD. 1990. Genericphysical mechanisms of morphogene-sis and pattern formation. Develop-ment 110:1–18.

Newman SA, Forgacs G, Muller GB.2006. Before programs: the physicalorigination of multicellular forms. IntJ Dev Biol 50:289–299.

Ohnemus U, Uenalan M, Conrad F,Handjiski B, Mecklenburg L, Naka-mura M, Inzunza J, Gustafsson JA,Paus R. 2005. Hair cycle control byestrogens: catagen induction viaestrogen receptor (ER)-alpha ischecked by ER beta signaling. Endo-crinology. 146:1214–1225.

Ohnemus U, Uenalan M, Inzunza J, Gus-tafsson JA, Paus R. 2006. The HairFollicle as an Estrogen Target andSource. Endocr Rev. In Press.

Paine ML, Paine CT, Machin GA. 2004.Hair whorls and monozygosity. J InvestDermatol 122:1057–1058.

Paus R, Cotsarelis G. 1999. The biologyof hair follicles. N Engl J Med 341:491–497.

Paus R, Foitzik K. 2004. In search of the‘‘hair cycle clock’’: a guided tour. Dif-frentiation 72:489–511.

Pearton DJ, Ferraris C, Dhouailly D.2004. Transdifferentiation of cornealepithelium: evidence for a linkage be-tween the segregation of epidermalstem cells and the induction of hair fol-licles during embryogenesis. Int J DevBiol 48:197–201.

Plikus M, Chuong CM. 2004. Makingwaves with hairs. J Invest Dermatol122:vii–ix.

Plikus M, Wang WP, Liu J, et al. 2004.Morpho-regulation of ectodermalorgans: integument pathology andphenotypic variations in K14-nogginengineered mice through modulationof BMP pathway. Am J Pathology164:1099–1114.

Pourquie O. 2003. The segmentationclock: converting embryonic time intospatial pattern. Science 301:328–330.

Prin F, Dhouailly D. 2004. How and whenthe regional competence of chick epi-dermis is established: feathers vs. scu-

tate and reticulate scales, a problemen route to a solution. Int J Dev Biol48:137–148.

Prum RO, Williamson S. 2002. Reaction-diffusion models of within-feather pig-mentation patterning. Proc Biol Sci 269:781–792.

Randall VA, Hibberts NA, Thornton MJ,et al. 2001. Do androgens influencehair growth by altering the paracrinefactors secreted by dermal papillacells? Eur J Dermatol 11:315–320.

Reid AI, Gaunt SJ. 2002. Colinearity andnon-colinearity in the expression ofHox genes in developing chick skin.Int J Dev Biol 46:209–215.

Robinson ME, Gibbins AM, Hardy MH.1990. Persistence of added retinoids inorgan culture media during induction ofmucous metaplasia and glandular mor-phogenesis in hamster cheek pouches.Experientia 46:513–517.

Rodriguez-Esteban C, Tsukui T, Yonei S,et al. 1999. The T-box genes Tbx4 andTbx5 regulate limb outgrowth andidentity. Nature 398:814–818.

Rose J, Oldfield J, Stormshak F. 1987.Apparent role of melatonin and prolactinin initiating winter fur growth in mink.Gen Comp Endocrinol 65:212–215.

Sengel P. 1990. Pattern formation inskin development. Int J Dev Biol 34:33–50.

Shen WM, Will P, Galstyan A, Chuong CM.2004. Hormone-inspired self-organiza-tion and distributed control of roboticswarms. Auton Robots 17:93–105.

Slominski A, Pisarchik A, Semak I, et al.2002. Serotoninergic and melatoni-nergic systems are fully expressed inhuman skin. FASEB J 16:896–898.

Slominski A, Pisarchik A, Zbytek B, et al.2003. Functional activity of serotoni-nergic and melatoninergic systems ex-pressed in the skin. J Cell Physiol 196:144–153.

Small S, Levine M. 1991. The initiation ofpair-rule stripes in the Drosophila blas-toderm. Curr Opin Genet Dev 1:255–260.

Steinberg MS. 1996. Adhesion in devel-opment: an historical overview. DevBiol 180:377–388.

Stelnicki EJ, Komuves LG, Kwong AO,et al. 1998. HOX homeobox genes ex-hibit spatial and temporal changes inexpression during human skin devel-opment. J Invest Dermatol 110:110–115.

Stenn KS, Paus R. 2001. Controls of hairfollicle cycling. Physiol Rev 81:449–494.

Sterry W, Paus R, Burgdorf W. 2006.Dermatology. Stuttgart/New York:Thieme. pp. 16–38.

Suzuki N, Hirata M, Kondo S. 2003.Traveling stripes on the skin of a mu-tant mouse. Proc Natl Acad Sci USA100:9680–9685.

Turing A. 1952. The chemical basis ofmorphogenesis. Philos Trans R SocLond B Biol Sci 237:37–72.

Wheeler MD. 1991. Physical changes ofpuberty. Endocrinol Metab Clin NorthAm 20:1–14.

290 WIDELITZ ET AL.


Widelitz RB, Jiang TX, Lu J, et al. 2000.Beta catenin in epithelial morphogen-esis: conversion of part of avian footscales into feather buds with a mu-tated beta catenin. Dev Biol 219:98–114.

Wolfram S. 2002. A new kind of science.Champaign, IL: Wolfram Media.

Wolpert LJ. 1969. Positional informationand the spatial pattern of cellular dif-ferentiation. Theor Biol 25:1–47.

Wu P, Hou L, Plikus M, Hughes M, Sceh-net J, Susaweang S, Widelitz RB,Jiang TX, Chuong CM. 2004. Evo-devoof amnion integuments and appen-dages. Int J Dev Biol 48:249–270.

Yue Z, Jiang TX, Widelitz RB, et al. 2005.Mapping stem cell activities in thefeather follicle. Nature 438:1026–1029.

Yue Z, Jiang TX, Widelitz RB, et al.2006. Wnt 3a gradient converts radialto bilateral feather symmetry via top-ological arrangement of epithelia. ProcNatl Acad Sci USA 103:951–955.



Distinct mechanisms underlie pattern formation in the skin and skin ...

Documents

Transcript of Distinct mechanisms underlie pattern formation in the skin and skin ...