Wnt3a Gradient Converts Radial to Bilateral Feather Symmetry Via Topological Arrangement of Epithelia

Wnt3a gradient converts radial to bilateral feather symmetry via topological arrangement of epithelia

Zhicao Yue, Ting-Xin Jiang, Randall Bruce Widelitz, and Cheng-Ming Chuong*

Department of Pathology, Keck School of Medicine, University of Southern California, Los Angeles, CA 90033

Edited by Jeremy Nathans, The Johns Hopkins University School of Medicine, Baltimore, MD, and approved November 28, 2005 (received for review August 9, 2005) The evolution of bilaterally symmetric feathers is a fundamental (where barb ridges start to form), and maturing feather branches process leading toward flight. One major unsolved mystery is how (where barb ridges form the ramus and barbule plates that the feathers of a single bird can form radially symmetric downy keratinize to become barbs). In downy feathers (radially sym- feathers and bilaterally symmetric flight feathers. In developing metric), all barb ridges are parallel to the long follicular axis. In downy feather follicles, barb ridges are organized parallel to the flight feathers (remiges), bilaterally symmetric barb ridges con- long axis of the feather follicle. In developing flight-feather folli- verge obliquely toward the anterior follicle, leading to fusion and cles, the barb ridges are organized helically toward the anterior the creation of the rachis (Figs. 1c and 5). Although remiges region, leading to the fusion and creation of a rachis. Here we located in the more distal wing become bilaterally (left–right) discover an anterior–posterior molecular gradient of wingless int asymmetric (3), this will not be studied here. (Wnt3)a in flight but not downy feathers. Global inhibition of the Wnt gradient transforms bilaterally symmetric feathers into radi- Results ally symmetric feathers. Production of an ectopic local Wnt3a Molecular and Cellular Differences of Downy and Flight Feathers. We gradient reoriented barb ridges toward the source and created an first analyzed the difference among embryonic downy (radial), ectopic rachis. We further show that the orientation of the Wnt3a adult downy (more radial), and adult flight (bilateral) feathers gradient is dictated by the dermal papilla (DP). Swapping DPs (Fig. 1a). Sonic hedgehog (Shh) expressing marginal plates were between wing covert and breast downy feathers demonstrates used to delineate the orientation of barb ridges in developing that both feather symmetry and molecular gradients are in accord embryonic feather buds (20, 21). Here we use opened adult with the origin of the DP. Thus the fates of feather epidermal cells feather follicle preparations (Fig. 1aЈ) to further analyze barb- are not predetermined through some molecular codes but can be ridge organization in mature follicles (Fig. 1c). In radially modulated. Together, our data suggest feathers are shaped by a symmetric embryonic downy feathers, barb ridges originate DP3Wnt gradient3helical barb ridge organization3creation of simultaneously early in development or at random positions rachis3bilateral symmetry sequence. We speculate diverse feather around the feather germ (22). In flight feathers, the rachis forms forms can be achieved by adjusting the orientation and slope of

in the anterior, whereas new barbs are continuously generated BIOLOGY molecular gradients, which then shape the topological arrange- from the posterior barb generative zone (9, 19). Through helical DEVELOPMENTAL ments of feather epithelia, thus linking molecular activities to barb-ridge organization, barbs reach the rachis with an angle ␪ organ forms and novel functions. (ref. 23; we use ‘‘helical organization’’ instead of ‘‘helical growth,’’ because this event involves only cell rearrangement, but ͉ ͉ axis determination dermal papilla evo-devo (evolution and ‘‘growth’’ usually implies the involvement of cell proliferation in ͉ ͉ development) morphogenesis skin appendages the context of cell biology). The angle of helical organization is bigger in flight-feather follicles, smaller in adult downy feather s the genomics of different organisms are gradually re- follicles, and 0° in embryonic downy feathers (Fig. 1c). The final Avealed, we need to learn more about how these 1D molec- barb-to-rachis angle in mature feathers results from this angle of ular codes are transformed to form a variety of biological forms, helical barb-ridge organization ␪ and the expansion angle of the as described by D’Arcy Thompson (1). Although the basic barb at the emergence of mature feathers (23). information is genetically determined, the organization of cells How are newly generated epithelial cells guided to form their and their collectives appears to operate at a different level and specific barb patterns? It was proposed that new cells may have follow rules we do not fully understand. Here we use feather a tendency to move toward the rachis, and there may be a one follicles, one of the most complex epithelial organs, to decipher barb–one clone lineage relationship (ref. 17; Fig. 1d, i), or that the architectural principles of how cells are arranged in place and cells may be deposited along the vertical axis of feather follicles time to build the functional forms in the context of ‘‘topobiol- independent of barb organization (ref. 23; Fig. 1d, ii). To ogy’’ (2). differentiate these possibilities, we injected the follicle base with The origin and evolution of feathers have been of great 1,1Ј-dioctadecyl-3,3,3Ј,3Ј-tetramethylindocarbocyanine (DiI). interest (3, 4), particularly with the recent discoveries of feath- Forty-eight hours later, the follicles were opened and photo- ered dinosaurs in Northern China (5, 6). From the many graphed, then superimposed with the image of Shh in situ intermediate feather forms, we learned that feathers evolved hybridization (Fig. 1e). The labeled cells left a straight track. through stepwise evolutionary novelties to produce diverse Therefore, the experiments favored the second possibility, un- morphology with new functions (7, 8). Today’s birds have coupling cell lineage from cell arrangement events during barb- evolved region-specific feather forms, ranging from radially symmetric downy feathers to bilaterally symmetric flight feathers (Fig. 1a). We must look further into the molecular and devel- Conflict of interest statement: No conflicts declared. opmental mechanisms that make these processes possible. This paper was submitted directly (Track II) to the PNAS office. The feather has a follicular structure with the dermal papilla Abbreviations: DP, dermal papilla; Dkk1, Dickkopf1; RCAS, replication-competent avian (DP) at its base (9–11). The feather filament is a cylindrical sarcoma virus; BMP, bone morphogenetic protein; DiI, 1,1Ј-dioctadecyl-3,3,3Ј,3Ј-tetra- structure with mesenchymal pulp inside (Fig. 1b). Moving up the methylindocarbocyanine; PE, papillar ectoderm; Wnt, Wingless int; Shh, Sonic hedgehog. feather shaft from the proximal to the distal end, there are the *To whom correspondence should be addressed. E-mail: chuong@pathfinder.usc.edu. collar (where cells actively proliferate), the ramogenic zone © 2006 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0506894103 PNAS ͉ January 24, 2006 ͉ vol. 103 ͉ no. 4 ͉ 951–955 Downloaded by guest on September 30, 2021 Fig. 1. Cells arrange into diverse feather forms with bilateral or radial symmetry. (a) Gross morphology of feather from a single chicken. On the left, more bilateral symmetry; on the right, more radial symmetry. (aЈ) Open follicle Fig. 2. Perturbation of the Wnt gradient changes feather forms. (a) Con- preparation. (b) Schematic feather follicle drawing. (c) Barb ridge orientation version of feather vanes from bilateral (b) to radial symmetry (r). Some in radially and bilaterally symmetric feather follicles. Whole-mount Shh in situ feathers are chimeric, with bilateral symmetry in the distal vane and radial hybridizations reveal that barb ridges insert into the rachidial ridge (rr) with symmetry in the proximal vane. RCAS genes used (Dkk1 and Wnt3a) are shown the helical insertion angle, ␪.(d and e) Uncoupling cell lineage from barb ridge in yellow. Gradual alteration of barb-to-rachis angle (blue blank arrow and organization. (d) Open follicle preparations stained by Shh in situ hybridiza- arrow) is shown in Dkk1 specimens. (aЈ) Line tracings. (b) Cross sections at the tion. rr (yellow arrow), barb generative zone (bg, green arrowhead), and the ramogenic zone are stained with antibody to virus GAG protein. The nearly possibilities of barb-ridge organization (i and ii). The green arrow represents homogeneous staining suggests a high and even expression level, flattening each possibility. A, anterior; P, posterior. (e) DiI (red)-labeled cells were the endogenous gradient. The result is a radial symmetric feather. Some displaced along a straight track. The photograph was superimposed with the disorganized barb ridges can also be seen. This affects barb-ridge differenti- Shh-stained follicle. Arrowhead, initial injection site. (f and g) A–P molecular ation but not their orientation and will not be pursued here. (c) Localized gradient. (f) Longitudinal (A–P) feather sections with Wnt3a in situ hybrid- RCAS-Wnt3a-transduced follicles. Serial sections are reconstructed in 3D to ization. Similar distributions and levels were observed in downy feathers, but help visualize the spatial configuration. A new rachis (yellow) is located at the an A–P gradient exists in flight feathers. Red blocks, schematic representation side of high viral transduction (purple). Blue, the original rachis. [Scale bars: (a) of Wnt3a gradient. (g) Feather-follicle dissection diagram. Semiquantitative 1 cm, (b) 1 mm.] RT-PCR from anterior (A), middle (M), and posterior (P) regions of a bilaterally symmetric feather shows an A–P gradient. [Scale bars: (a) 1 cm, (b) 100 ␮m, (c–f) 0.5 mm.] titative RT-PCR (Fig. 1g). The expression of Wnt 3a showed graded levels along the A–P axis. Although in the rest of this work, we refer to this graded distribution as a Wnt gradient to ridge organization. The end result is helical organization of barb facilitate discussions, we appreciate that these are transcript ridges toward the rachis. gradients, and it would be ideal to determine a Wnt protein or What molecular mechanism, then, causes barb ridges to activity gradient in the future. organize with a slanted angle? We analyzed cellular and molecular asymmetries at the ramogenic zone. The liver cell adhesion Global Perturbation of Wnt Gradient. To determine the role of the molecule was homogenously expressed throughout the feather endogenous Wnt gradient, we perturbed the gradient using follicle epithelium (ref. 19; not shown). Interestingly, several plucking͞regeneration͞transgenic misexpression techniques Wingless int (Wnt) family members (Wnt3a, Wnt 5a, Wnt8c, and (15). The Wnt antagonist Dickkopf1 (Dkk1) is a secreted factor Wnt11) showed higher expression in the anterior͞rachis side but that inhibits Wnt signaling (24, 25). We used retrovirus repli- none or lower expression in the posterior barb generative zone cation-competent avian sarcoma virus (RCAS) to overexpress (Wnt3a is shown as an example in Fig. 1f; Wnt 5a is shown in Fig. Dkk1 and observed a chimeric feather with gradual conversion 6, which is published as supporting information on the PNAS from the bilateral symmetric feather vane to the more radially web site). To show that these asymmetric expressions constitute symmetrically arranged barbs (Fig. 2 a and aЈ; 75%, n ϭ 25). An a gradient distribution, the follicle was dissected, and the ramo- example (green box) illustrates the gradual alteration of the genic segment was removed as a horizontal disk. The disk was barb-to-rachis angle from Ϸ40° to a much sharper 15° (Fig. 2a, further divided into three portions and analyzed with semiquan- arrows). For the mature feather, because cells are dead, we

952 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0506894103 Yue et al. Downloaded by guest on September 30, 2021 Fig. 4. The DP can alter the Wnt3a gradient and determine feather symme- Fig. 3. Local perturbation of the Wnt3a gradient reorients barb ridges. (a try. (a) Wing covert (Wi) and breast downy (Br) feathers were used to represent Ј ␪ and a ) Control beads (BSA) have no effect on barb ridge organization. , angle bilaterally and radially symmetric feathers. DPs were swapped between these Ј of helical organization. (b and b ) Wnt3a beads (blue) placed near the original two follicle types. Regenerated feather forms are in accordance with the rachis induced an ectopic rachidial ridge (err, small yellow arrow). The original origin of the DPs. (b) Longitudinal sections (along the A–P plane) of control rachidial ridge is indicated by the large yellow arrow. The direction of barb and chimeric follicles were processed for Wnt3a in situ hybridization. A–P ridges, particularly those between the original rachis and ectopic rachis, are gradients were observed in follicles with wing covert DP. Red blocks, sche- Ј reoriented. (c and c ) Wnt3a beads (blue) placed away from the original rachis matic representation of Wnt3a gradient. (large yellow arrow) induced an ectopic rachidial ridge (small yellow arrow) and a new ectopic barb generating zone (ebg, green arrowhead). Because the Ј barb ridges were reorientated, duplicated A–P axes were created. (d and d ) a triangle and is designated as the posterior side (refs. 9, 21; Fig. TGF␤1 beads led to a patch of inhibited barb ridge formation but did not change the orientation of barb ridges or the symmetric form of the feather. 3a, green arrowhead). When a Wnt3a-coated bead was placed The effect was similar to that induced by BMPs. (Insets) Individual expected near the original rachis, it induced a new rachis and redirected gradients (solid red) and their sum (broken red line). (Scale bars: 1 mm.) barb ridges to curve toward the bead. This is particularly obvious for barb ridges located between the original and ectopic rachis (Fig. 3 b and bЈ). When the bead was placed sufficiently away cannot detect their viral expression directly. The transition of from the original rachis, a new ‘‘posterior triangle’’ (i.e., barb morphology in the middle of the vane is most likely, because viral generative zone) was induced together with a new rachis. The transduction became widespread halfway through the growth of barb ridges are now reoriented to form mirror image-duplicated

this feather. axes, A-P-P-A, compared with the original A–P organization BIOLOGY Ј We also tested the effect of ectopic Wnt gene expression. (Fig. 3 c and c ). Control beads with BSA have no effect. Thus, DEVELOPMENTAL Depending on the distribution, two categories of phenotypes the Wnt pathway exerts a direct effect on barb-ridge organiza- were observed. When high levels of RCAS-Wnt3a were ex- tion, whereas their involvement in the formation of the expan- pressed in the whole follicle, it flattened the endogenous Wnt sion angle remains to be determined. The Wnt pathway is gradient, and the feathers became more radially symmetric (Fig. versatile (26), and the downstream cellular mechanism of Wnt3a 2a). The nearly homogenous expression of viral GAG protein is may be determined in future studies. In comparison, TGF␤1 shown in Fig. 2b, and with this protocol, most expression is in the inhibits barb ridge formation but has no effect on the orientation epidermis. Depending on the time and level of viral expression, of adjacent barb ridges (Fig. 3 d and dЈ). EGF or FGF10 also have the feathers may show a reduced rachis size or, in some cases, no no such effects (not shown). rachis at all, producing fewer A–P-polarized feathers (Fig. 2 a and aЈ). Controls using RCAS–LacZ show typical bilateral Epithelial–Mesenchymal Recombination Between Downy and Flight symmetric vanes. These phenotypes are specific to the Wnt Feathers. How was the Wnt3a gradient established? Classical pathway, because RCAS–bone morphogenetic protein (BMP), experiments showed that the DP controls the morphology of –noggin,–Shh, and several other genes do not produce these regenerated feathers, although in some earlier studies, the types of changes (ref. 15 and not shown). papillar ectoderm (PE) was not removed from the DP (17, 18, When localized viral genes were expressed (judged by a 27). Here we developed a way to remove the PE from the DP (see localized region of viral misexpression), we observed an ectopic Materials and Methods). We swapped DPs between wing covert rachis was induced (Fig. 2c). The relationship between the new and breast downy feathers. Wing covert feathers are used here rachis position and the site of high viral transduction is best to represent bilaterally symmetric feathers, because DP swap- appreciated in 3D reconstruction (Fig. 2c). Interestingly, the site ping must be done by using similarly sized follicles; flight-feather of the new rachis was at the site of highest RCAS–Wnt3a virus DPs are too big to be accommodated by breast downy feather levels. All together, in RCAS–Wnt3a feathers (n ϭ 37), 41% follicles. The chimeric feathers showed their symmetric forms showed regions of radial symmetry, and 54% showed new rachis are in accordance with the origin of the DPs (Fig. 4a; 100%, n ϭ positions. 25). The presence and absence of a Wnt3a gradient were also in accord with the origin of the DPs (Figs. 1f and 4b). Mock DP Local Perturbation of the Wnt Gradient. To further analyze the transplant controls did not show changes (not shown). These effect of the Wnt3a gradient in this process, we implanted beads results are consistent with the embryonic recombination study soaked with Wnt3a in the developing feather follicle in vivo and showing that downy͞bilateral feather symmetry is based on allowed it to grow for another 48 h. Open-feather follicles were epithelial–mesenchymal interactions (dictated by the DP), prepared (Fig. 1aЈ), and Shh was used to reveal the orientation whereas barb morphology is determined by an intraepithelial of barb ridges. In flight feathers, barb ridges meet the rachidial event (28). Together, these studies suggest that barb-ridge ridge (Fig. 3a, yellow arrow) with the helical angle ␪ (Fig. 3a). organization is via cell rearrangement, not proliferation, which Opposite to the rachis is the barb generative zone, which forms is in response to local molecular gradients dictated by the DP

Yue et al. PNAS ͉ January 24, 2006 ͉ vol. 103 ͉ no. 4 ͉ 953 Downloaded by guest on September 30, 2021 flexible; they can be organized into different forms during different molting cycles (9) or through regeneration after plucking. Here we show an ectopic Wnt gradient can reorient the topological arrangements of barb-ridge keratinocytes. Wnt3a attracts barb ridges to bend toward sources of higher Wnt3a concentrations, leading to the formation of the rachis. Tuning each process at each step can lead to different 3D configurations of feather branches, thus giving rise to the large diversity of feather forms (23). As demonstrated in the example here, the identification of the molecular bases for these processes is now possible. In our previous work, we showed that BMP caused the formation of an enlarged rachis, whereas noggin led to the production of multiple small rachides in the original rachis location (15). The fusion and creation of new barb ridges involve Shh͞Bmp2 signaling in the marginal plate epithelium (20, 21). This differs from the results here, in that a localized peak of Wnt signaling can specify the location of the rachis, and a homogenous level of Wnt eliminates A–P polarity and rachis formation. On the other hand, ectopic expression of BMP or noggin does not alter the rachis position or A–P axis. In addition to helical organization, bilaterally symmetrical feathers are characterized by the localization of new barb-ridge creation posteriorly and fusion of barb ridges anteriorly to create the rachis (21). Our results indicate that the A–P Wnt gradient plays a role in the polarized localization of these processes to either the anterior or posterior end of the feather follicle. How does the event downstream of the Wnt gradient influence the helical organization of barb ridges in the feather-filament epithelial cylinder? It does not appear to be mediated by ␤-catenin, because ␤-catenin nuclear staining does not appear until keratinocytes are in the differentiating barbule plates above the ramogenic zone (not shown). The Wnt noncanonical pathway has been shown to modulate cell shape and movement within an epithelial sheet and will have to be evaluated in the future (26, 31). Fig. 5. A model linking feather-symmetric forms and molecular gradients. After this work was submitted, an activator–inhibitor model of The basic design of feathers allows variations of feather shape and symmetry to occur by varying just a few parameters (23). Barb ridges are oriented in embryonic feather branching was proposed (32). In this model, parallel to the feather follicles in radially symmetric feathers (␪ ϭ 0) but form activators and inhibitors can transform the smooth circumfer- an angle of helical organization ␪ with the rachidial ridge (rr). The barb ridges ence of feather cylindrical epithelia into discrete numbers of slant obliquely, because they are composed of a vertical component and a barb ridges by the formation of periodically arranged Shh͞ horizontal component. The vertical displacement is caused by feather growth, BMP2-positive marginal plates; thus, a radial symmetric downy shown as vector AB. The force to have barb ridges oriented horizontally to the feather will form. The downy feather also shows occasional anterior side is shown as vector AC, to which the Wnt3a gradient has contrib- random bifurcation, fusion, and initiation of stripes due to uted. There are likely to be other molecules involved, but Wnt3a is illustrated instability in this space-filling patterning process (21, 32). When here as an example, and this model provides a conceptual framework. The sum an additional dorsal͞ventral polarity (or anterior͞posterior in of vectors AB and AC is vector AD, leading to the helical organization of barb ridges toward the rachis. In radially symmetric feathers, there is only a vertical the terminology here) is imposed on top of this local activator– component AB. A gradual increase of the component AC may lead to feather inhibitor mechanism, there is less inhibitor activity in the ante- morphologies ranging from more radial to more bilateral symmetry (Fig. 1A). rior rachis region but more activator activity in the posterior region. Thus, there is extinction of activator stripes as barb ridges approach the anterior midline but emergence of additional (Fig. 5). The symmetry is without regard to the epidermal cell activator stripes in the posterior end. Through this mechanism, origin, and their fates are not predetermined. the radial symmetric feather is transformed into a bilaterally symmetric feather. The Wnt3a gradient we observed here fit the Discussion criteria of this global anterior–posterior polarity well. Indeed, The feather is unique in its complex architecture and seemingly elimination of this gradient leads to the transformation of endless variation of forms (7, 9, 29). We propose the following bilateral symmetric feathers to become radially symmetric (Fig. events during the morphogenesis of feather follicles (Fig. 5). (i) 2). In open adult feather-follicle preparation, the Shh-positive Although embryonic feathers are known to form A–P asymme- stripes vividly illustrate the helical organization, emergence, and try (21, 30), molecules do not form gradients in radially sym- extinction of these barb ridges (Fig. 3; compare with figure 2 of metric feathers, but they form gradient peaks toward the anterior ref. 32). How the global Wnt gradient and local Shh͞BMP2 in flight feathers [A–P gradient, e.g., Wnt3a, Wnt 5a, and Keratin signaling or other unidentified factors are coupled remains to be A (data not shown)]. These molecular gradients help establish determined. asymmetric arrangements of cells in the collar region and What set up the global anterior–posterior Wnt gradient? ramogenic zone. (ii) In bilaterally symmetric feathers, barb Here, we show that swapping DP and epithelial follicles leads to ridges are aligned toward the anterior end and eventually merge newly regenerated feathers with symmetry in accordance with into the rachis with an angle of helical organization ␪. In radially that of the origin of the DPs (Fig. 4). Recently, we showed that symmetric feathers, barb ridges form parallel to the long axis of feather epidermal stem cells are in a ring-configured niche, feather follicles, and ␪ ϭ 0. Feather keratinocyte precursors are horizontally placed in the radial symmetric feathers but tilted

954 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0506894103 Yue et al. Downloaded by guest on September 30, 2021 anteriorly– posteriorly in bilateral symmetric feathers (11). Thus, Retrovirus Production and Misexpression. RCAS viruses were cul- these epidermal stem cells are true stem cells that can be tured and harvested as described (13). RCAS-LacZ, RCAS- modulated into distinct symmetric forms by the different micro- Wnt3a, and RCAS-Dkk1 were used in this study. Dkk1 is a gift environments created by the DP. Furthermore, an A–P Wnt from S. Millar (University of Pennsylvania, Philadelphia) and A. gradient is involved in the property of this microenvironmental Lassar (Harvard University, Boston), which we subcloned into niche (Fig. 5). RCAS-Bryant polymerase envelope protein A (14). Remiges of In summary, we report a mechanism nature uses to convert Ϸ1-mo-old chicks were plucked, transduced with retrovirus, and organ radial symmetry to bilateral symmetry, a simple molecular allowed to regenerate for up to 2 mo (15). Different levels of viral gradient. Distinct feather forms provide a unique opportunity to transduction can be adjusted by using different viral titers or visualize how dialogues between cells and molecules are cast into changing the number of injection sites. The extent of viral feather morphologies, linking molecular activities to organ infection was visualized with GAG immunostaining (Hybridoma forms, and allowing the evolution of novel feather functions. Bank, University of Iowa, Iowa City).

Materials and Methods 3D Reconstructions of Serial Sections. Eight-micrometer feather DiI Labeling and Bead Implantation. DiI (Molecular Probes) and follicle sections were digitized by using IGL TRACE software beads (Wnt3a, TGF␤1:R&DSystems) were prepared as (Boston University, Boston). These images were aligned and described (12). Three-month-old chickens were anesthetized rendered as a 3D view of the feather follicle by using RHINOC- with ketamine and xylocaine (2:1, 10 mg͞kg). DiI and beads were EROS software (16). injected through the follicle wall. Follicles were collected at desired times and photographed or processed for whole-mount Surgery and DP Operations. Three-month-old chickens were anes- in situ hybridization. thetized with ketamine and xylocaine (2:1, 10 mg͞kg). DP͞PE operations were done after Lillie and Wang (17, 18). The DP͞PE RT-PCR Analysis. Total RNA was isolated according to the man- complex is immersed in trypsin͞EDTA in vitro for 10 min at ufacturer’s guide (Qiagen, Valencia, CA, RNAeasy kit). PCR 37°C. PE are then stripped from the DP. DP was then washed in was carried out by using the annealing temperature 60°C. DMEM containing 10% serum and transplanted back into an Primers used were: Wnt3a (GAAGCTGGAAGGACCTCTAT empty follicle. The complete removal of epithelium from the DP and GGTCACAACCGTCAATCCC) (35 cycles), GAPDH was verified with antibody to LCAM (19). All animal care was (GGCGAGATGGTGAAAGTCG and CAGTTGGTGGTG- in accordance with institutional guidelines. CACGATG) (28 cycles). We thank Drs. Richard Prum (Yale University, New Haven, CT), Gerald Immunostaining and in Situ Hybridization. Immunostaining and in Edelman (The Scripps Research Institute, La Jolla, CA), and George situ hybridization were processed as described with an automated Cotsarelis (University of Pennsylvania) for helpful input. We are grateful Discovery system (Ventana, Tucson, AZ) (12). RNA probes to Drs. C. Tabin (Harvard University) (for Shh and RCAS-Wnt3a), A. Lassar (Harvard University) (for Dkk1), S. Millar (University of Penn- included in this study involve Shh and Wnt3a (from A. McMa- sylvania) (for Dkk1), and A. McMahon (Harvard University) (for hon, Harvard University, Boston). Open-feather follicles were Wnt3a) for providing reagents. We thank all Chuong lab members for BIOLOGY prepared by making a cut between the rachis and the barb discussion. This work is supported by National Institute of Arthritis and DEVELOPMENTAL generation zone. The follicles were opened and laid flat on a Musculoskeletal and Skin Diseases Grants AR 42177 and AR47364 (to dish. C.-M.C.) and National Cancer Institute Grant 83716 (to R.B.W.).

1. Thompson, D. (1917) On Growth and Form, reprinted (1992) (Dover, New 15. Yu, M., Wu, P., Widelitz, R. B. & Chuong, C. M. (2002) Nature 420, 308–312. York). 16. Fiala, J. C. & Harris, K. M. (2002) Microsc. Anal. 52, 5–7. 2. Edelman, G. M. (1988) Topobiology: An Introduction to Molecular Biology 17. Lillie, F. R. & Wang, H. (1941) Physiol. Zool. XIV, 103–133. (Basic Books, New York). 18. Lillie F. R. & Wang, H. (1944) Physiol. Zool. 17, 1–32. 3. Feduccia, A. (1999) The Origin and Evolution of Birds (Yale Univ. Press, New 19. Chuong, C. M. & Edelman, G. M. (1985) J. Cell Biol. 101, 1027–1043. Haven, CT), 2nd Ed. 20. Ting-Berreth, S. A. & Chuong, C. M. (1996) Dev. Dyn. 207, 157–170. 4. Prum, R. O. & Brush, A. H. (2002) Q. Rev. Biol. 77, 261–295. 21. Harris, M. P., Fallon, J. F. & Prum, R. O. (2002) J. Exp. Zool. 294, 160–176. 5. Zhou, Z., Barrett, P. M. & Hilton, J. (2003) Nature 421, 807–814. 22. Chuong, C. M. & Edelman, G. M. (1985) J. Cell Biol. 101, 1009–1026. 6. Xu, X., Norell, M. A., Kuang, X., Wang, X., Zhao, Q. & Jia, C. (2004) Nature 23. Prum, R. O. & Williamson, S. (2001) J. Exp. Zool. 291, 30–57. 431, 680–684. 24. Glinka, A., Wu, W., Delius, H., Monaghan, A. P., Blumenstock, C. & Niehrs, 7. Prum, R. O. (1999) J. Exp. Zool. 285, 291–306. C. (1998) Nature 391, 357–362. 8. Chuong, C.-M., Wu, P., Zhang, F. C., Xu, X., Yu, M., Widelitz, R. B., Jiang, 25. Andl, T., Reddy, S. T., Gaddapara, T. & Millar S. E. (2002) Dev. Cell 2, T. X. & Hou, L. (2003) J. Exp. Zool. B. 298, 42–56. 643–653. 9. Lucas, A. M. & Stettenheim, P. R., eds. (1972) Avian Anatomy: Integument. 26. Veeman, M. T., Axelrod, J. D. & Moon, R. T. (2003) Dev. Cell 5, 367–377. Agriculture Handbook 362 (U.S. Dept. of Agriculture, Washington, DC). 27. Cohen, J. & Espinasse, P. G. (1961) J. Embryol. Exp. Morphol. 9, 223–251. 10. Yu, M., Yue, Z., Wu, P., Wu, D.-Y., Mayer, J. A., Medina, M., Widelitz, R. B., 28. Sengel, P. (1976) in Morphogenesis of Skin (Cambridge Univ. Press, London). Jiang, T.-X. & Chuong, C.-M. (2004) Int. J. Dev. Biol. 48, 181–191. p. 188. 11. Yue, Z. C., Jiang, T.-X., Widelitz, R. B. & Chuong, C. M. (2005) Nature 438, 29. Chuong, C.-M., Chodankar, R., Widelitz, R. B. & Jiang, T. X. (2000) Curr. 1026–1029. Opin. Genet. Dev. 10, 449–456. 12. Jiang, T.-X., Jung, H.-S., Widelitz, R. B. & Chuong, C.-M. (1999) Development 30. Chen, C. W., Jung, H. S., Jiang, T. X. & Chuong, C. M. (1997) Dev. Biol. 188, (Cambridge, U.K.) 126, 4997–5009. 181–187. 13. Widelitz, R. B., Jiang, T.-X., Lu, J.-F. & Chuong, C. M. (2000) Dev. Biol. 219, 31. Moon, R. T., Bowerman, B., Boutros, M. & Perrimon, N. (2002) Science 296, 98–114. 1644–1646. 14. Suksaweang S., Lin, C. M., Jiang, T.-X., Widelitz, R. B. & Chuong, C. M. (2004) 32. Harris, M. P., Williamson, S., Fallon, J. F., Meinhardt, H. & Prum, R. O. (2005) Dev. Biol. 266, 109–122. Proc. Natl. Acad. Sci. USA 102, 11734–11739.

Yue et al. PNAS ͉ January 24, 2006 ͉ vol. 103 ͉ no. 4 ͉ 955 Downloaded by guest on September 30, 2021