Comprehensive Analysis of FGF and FGFR Expression in Skin: FGF18 Is Highly Expressed in Hair Follicles and Capable of Inducing A

Home , FGF11, FGF12, FGF14, FGF18, FGF21, FGF22

Comprehensive Analysis of FGF and FGFR Expression in Skin: FGF18 Is Highly Expressed in Hair Follicles and Capable of Inducing Anagen from Telogen Stage Hair Follicles

Mitsuko Kawano,Ãw Akiko Komi-Kuramochi,Ã Masahiro Asada,Ã Masashi Suzuki,Ã Junko Oki,Ã Ju Jiang,z and Toru ImamuraÃ ÃNational Institute of Advanced Industrial Science and Technology (AIST), Higashi, Tsukuba, Ibaraki, Japan; wUniversity of Tsukuba, Tennodai, Tsukuba, Ibaraki, Japan; zDepartment of Dermatology, Juntendo University School of Medicine, Bunkyo, Tokyo, Japan

We quantified the mRNA expression of all 22 fibroblast growth factor family members (FGF) and their four receptors (FGFR) in adult mouse full-thickness skin at various stages of the hair growth cycle. We found that in addition to mRNA encoding FGF previously identified in skin (FGF1, 2, 5, 7, 10, 13, and 22), FGF18 mRNA was also strongly expressed. Expression of these FGF varied throughout hair growth cycle: mRNA expression of FGF18 and 13 peaked at telogen; FGF7 and 10 at anagen V; and FGF5 and 22 at anagen VI. In situ hybridization revealed that FGF18 mRNA is mainly expressed in the anagen inner root sheath and telogen bulge of hair follicles. In culture, FGF18 stimulated DNA synthesis in human dermal fibroblasts, dermal papilla cells, epidermal keratinocytes and vascular endothelial cells. When FGF18 was administered subcutaneously to mice in a uniform telogen state, anagen hair growth was observed. Our findings suggest that FGF18 is important for the regulation of hair growth and the maintenance of skin in adult mice. Key words: dermis/fibroblast growth factor/hair growth/receptor J Invest Dermatol 124:877 –885, 2005

A variety of polypeptide growth factors, including various via FGFR2 IIIb to stimulate keratinocyte proliferation in nor- members of the fibroblast growth factor (FGF) family, are mal and wounded skin (Werner et al, 1994; Marchese et al, expressed in skin. In both mice and humans, FGF are en- 1995; Igarashi et al, 1998; Ohuchi et al, 2000; Kawano et al, coded by twenty-two distinct genes (Ornitz and Itoh, 2001), 2004). Although these studies suggest the importance of whereas four genes encode transmembrane FGF receptor the concerted activities of FGF in skin, the expression of all (FGFR) tyrosine kinases (Jaye et al, 1992). Among them, FGF family members in skin have never been systematically FGF1, 2, 5, 7, 10, 13, and 22, are known to be expressed in investigated in a single study. We therefore established an dermal and hair follicular cells and to regulate hair growth experimental system with which to quantify the expression and skin regeneration (du Cros, 1993; du Cros et al, 1993; of FGF and FGFR mRNA in skin, and comprehensively pro- Hebert et al, 1994; Danilenko et al, 1995; Marchese et al, filed the dynamic changes in their expression over the 1995; Guo et al, 1996; Petho-Schramm et al, 1996; Rose- course of the hair growth cycle. nquist and Martin, 1996; Mitsui et al, 1997; Ortega et al, 1998; Suzuki et al, 1998, 2000; Nakatake et al, 2001; Stenn and Paus, 2001; Beyer et al, 2003; Kawano et al, 2004). For Results instance, earlier experiments revealed the crucial role Expression profiles of FGF and FGFR family members played by FGF5 in regulating the onset of catagen during in skin at various stages of the hair cycle: abundant the mouse hair cycle; that Fgf5 translates into two proteins expression of FGF18 mRNA at telogen phase We initially (FGF5 and FGF5S) through generation of alternatively established an experimental system to quantify the expres- spliced mRNA (Ozawa et al, 1998); and that FGF5S antag- sion of all Fgf and Fgfr family members (described in Ma- onizes the catagen-inducing activity of FGF5 (Suzuki et al, terials and Methods). The level of expression of each FGF 2000). In addition, FGF1, 2, and 7 reportedly promote hair and FGFR mRNA at the indicated stage in the hair cycle is growth, FGF13 (FHF2) is highly expressed in the bulge re- shown in Figs 1 and 2, respectively. In addition, the highest gion of hair follicles, and FGF7 and 10 function in concert expression levels of all the mRNA and the phases of the hair cycle at which they occurred are summarized in Table I. The most strongly expressed genes were Fgf1, 5, 7, 10, 13, Abbreviations: FGF, fibroblast growth factor; FGFR, FGF receptor; 18, and 22 (Fig 1A and Table I). Within this group, the levels HDF, human dermal fibroblasts; HDPC, human hair follicle dermal of FGF7 and FGF10 mRNA (or keratinocyte growth factor papilla cells; HEK, human epidermal keratinocytes; HORSC, human outer root sheath cells; HUVEC, human umbilical vein end- (KGF) 1 and KGF 2, respectively) exhibited similar variation othelial cells; PBS, phosphate-buffered saline throughout the hair cycle, with the highest expression levels

Copyright r 2005 The Society for Investigative Dermatology, Inc. 877 878 KAWANO ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

occurring at anagen V (day 8; Fig 1A). The highest levels of A 8000 Fgf1 Fgf5 Fgf 7 FGF5 and 22 mRNA occurred at anagen VI (day 18: Fig 1A). 6000 Expression of FGF1 mRNA (or acidic FGF) was compara- tively constant, with maximum levels attained at telogen 4000 (day 0; Fig 1A). At their highest levels, the expression of 4 2000 some FGF (e.g., FGF7 and 18) was approximately 1 10

copies/ng mRNA copies per ng mRNA. This level of expression was about 0 60-fold lower than that of b-actin (5.6 105 copies per ng 8000 mRNA at anagen V; Table I), one of the abundant structural Fgf10 Fgf11 Fgf13 proteins in skin, and was similar to that observed for cyto- 6000 kine tumor necrosis factor-alpha in some cell lines (5 103– 4 4000 1 10 copies per ng mRNA in U937 and HL60 cells; Takahashi et al, 2001). 2000

copies/ng mRNA It is noteworthy that even compared to those genes al- 0 ready known to be expressed in skin (i.e., FGF1, 2, 5, 7, 10, 13, and 22), the expression of FGF18 mRNA, which 12000 8000 Fgf18 Fgf22 peaked during telogen (days 0 and 22), was quite strong 10000 6000 (Fig 1A). Indeed, at most stages, the level of FGF18 ex- 8000 pression was much higher than that of the well-studied 6000 4000 FGF2 (or basic FGF; Fig 1B), and the other FGF as well 4000 2000 (Fig 1A). Moderate levels of FGF3, 6, 9, 11, 20, and 21 copies/ng mRNA copies/ng mRNA 2000 mRNA were also expressed (Fig 1A and B), but little or no 0 0 expression of FGF4, 8, 14, or 15 mRNA was detected at any Phase of Hair Cycle (days after depilation) stage of the hair cycle (o100 copies per ng mRNA; B Fig 1B and Table I). Fgf2 Fgf3 Fgf4 900 With respect to FGFR, the mRNA encoding both FGFR1 and FGFR3 were abundantly expressed (a maximum of 600 47,000 copies per ng mRNA for FGFR1 or 28,000 300 copies per ng mRNA for FGFR3; Fig 2). The strongest ex-

copies/ng mRNA 0 pression of the former occurred at anagen VI-catagen (days 18 and 20), whereas that of the latter peaked at anagen VI Fgf6 Fgf8 Fgf9 (day 18; Fig 2). The level of expression of FGFR2 mRNA, 900 including its IIIb subclass, which is specific for epithelial 600 cells, was lower than those of either FGFR1 or -3 mRNA 300 (Fig 2). The level of FGFR4 mRNA was the lowest among the

copies/ng mRNA four receptors (Fig 2). 0

Fgf12 Fgf14 Fgf15 900 FGF18 mRNA is expressed in hair follicles This study addresses the expression and function of FGF18 in skin, so 600 we initially determined the distribution of its mRNA using 300 in situ hybridization. We found that at anagen V (day 8), copies/ng mRNA 0 anagen VI (day 18), and catagen (day 20), FGF18 mRNA

Fgf16 Fgf17 Fgf20 Figure 1 900 Expression of ﬁbroblast growth factor (FGF) family genes in skin 600 throughout the hair cycle. (A and B) A portion of the dorsal skin of male 8-wk-old C3H/HeN mice was depilated during telogen [except 300 T(0)], and then harvested at various stages of the hair cycle (i.e., at the

copies/ng mRNA indicated times thereafter). Total RNA was extracted and further pu- 0 rified to mRNA. After reverse transcription using (dT)12–18 as a primer, Fgf21 Fgf23 the cDNA was amplified with specific primers for the indicated FGF. 900 With the aid of serial dilutions of standard plasmids, the absolute copy number for each FGF mRNA was calculated. To verify the specific am- 600 plification, not only the direct sequencing of the cloned DNA fragment but also the melting temperature and the size of the amplified fragment 300 were routinely monitored at the end of each amplification reaction.

copies/ng mRNA Symbols represent means SE obtained with cDNA from five mice for 0 each point. Note that the full Y-axis scales differ in A and B. The b-actin Phase of Hair Cycle (days after depilation) mRNA copy number per ng mRNA in each skin sample was separately analyzed, and was confirmed to be within the expected range (see Table I). All quantification experiments were carried out at least twice in each set of animal experiment to confirm the results. Moreover, the entire set of animal experiments were performed three times, and the results obtained were very similar. T, telogen; A, anagen; C, catagen. The numbers in the parentheses indicate the days after depilation. 124 : 5 MAY 2005 FGF AND FGFR EXPRESSION IN MOUSE SKIN 879

Table I. Expression of FGF and FGFR mRNA in adult mouse Fgfr1 Fgfr2 skin 50000

40000 Highest mRNA expression

30000 Phase of the hair cycle at which the 20000 highest mRNA expression

copies/ng mRNA Copy number (copies was attained 10000 Gene per ng mRNA) (day after depilation)

0 Fgf1 2900 152 Telogen (0) T(0) A(8) A(18) C(20) T(22) T(0) A(8) A(18) C(20) T(22) Fgf2 751 51 Telogen (0) Fgf3 667 87 Anagen VI (18) Fgfr3 Fgfr4 50000 Fgf4 ()()

40000 Fgf5 4307 578 Anagen VI (18)

30000 Fgf6 542 94 Telogen (0)

20000 Fgf7 7467 475 Anagen V (8)

copies/ng mRNA 10000 Fgf8 ()()

0 Fgf9 306 34 Telogen (22) T(0) A(8) A(18) C(20) T(22) T(0) A(8) A(18) C(20) T(22) Fgf10 4389 341 Anagen V (8) Phase of Hair Cycle (days after depilation) Fgf11 1422 84 Telogen (0) Figure 2 Expression of fibroblast growth factor receptor (FGFR) family Fgf12 188 24 Anagen V (8) genes in skin throughout the hair cycle. A skin cDNA was prepared, Fgf13 4945 87 Telogen (0) amplified with specific primers for the indicated FGFR, and then analyzed as described in the legend to Fig 1. Symbols represent mean- Fgf14 ()() s SE obtained with cDNA from five mice for each stage. All quantification experiments were carried out at least twice in each set Fgf15 ()() of animal experiment to confirm the results. Moreover, the entire set of Fgf16 171 34 Anagen VI (18) animal experiments were performed three times, and the results obtained were very similar. T, telogen; A, anagen; C, catagen. The num- Fgf17 179 7 Anagen VI (18) bers in the parentheses indicate the days after depilation. Fgf18 9822 1,875 Telogen (0) Fgf20 465 50 Telogen (0) was expressed virtually exclusively in the transient portion Fgf21 898 113 Anagen VI (18) of the inner root sheath cells of the hair follicles, close to the Fgf22 2685 307 Anagen VI (18) hair bulb (Fig 3A, C, D; purple signals). At telogen, FGF18 mRNA signals were detected in the bulge region (Fig 3E). Fgf23 ()() The sense probe produced no signal (Fig 3B). Although Fgfr1 47,108 3088 Catagen (20) FGF18 mRNA was most abundant at telogen (Fig 1A), we Fgfr2 10,414 1179 Anagen VI (18) discovered that mouse telogen skin contains high levels of a putative, non-specific inhibitor that strongly suppressed Fgfr3 28,540 2253 Anagen VI (18) in situ detection of even abundantly expressed mRNA such Fgfr4 1028 214 Catagen (20) as b-actin.1 Consequently, an extended period of time was b-actin 562,240 40,910 Anagen V (8) required to detect the rather weak FGF18 mRNA signals. The absolute mRNA copy number for each FGF (Fgf) and FGF receptor (Fgfr) gene was determined as described in the text; each mRNA copy FGF18 induces DNA synthesis in human hair follicle number is indicated with the hair cycle phase at which mRNA expression dermal papilla cells, dermal fibroblasts, epidermal ker- was maximal. The values are means SE of five replicate samples. (), genes with low copy numbers (o100 copies per ng mRNA); FGF, atinocytes, and vascular endothelial cells To examine fibroblast growth factor; FGFR, FGF receptor. the target cell specificity of FGF18 in skin, we examined its mitogenic effect on human hair follicle dermal papilla cells (HDPC), human outer root sheath cells (HORSC), human HORSC and HEK (Fig 4B and D, respectively), but very dermal fibroblasts (HDF), human epidermal keratinocytes weakly in HDPC (Fig 4A), HDF (Fig 4C), and HUVEC (Fig 4E). (HEK), and human umbilical vein endothelial cells (HUVEC). We found that at relatively high concentrations (100 and FGF18 induces anagen in telogen phase hair follicles 1000 ng per mL) FGF18 induced DNA synthesis in HDPC, in vivo The in vivo effect of FGF18 on telogen hair follicles HDF, HEK, and HUVEC (Fig 4A, C, D, and E, respectively), was investigated in C3H/HeN mice. In one set of experi- which is in contrast to FGF7 (KGF), the well-characterized ments (Fig 5) carried out, 21 d after subcutaneous admin- keratinocyte mitogen that stimulates DNA synthesis in istration of FGF18 protein adsorbed onto Sepharose beads, four of five mice exhibited hair follicle growth. The exterior 1Kawano and Imamura, manuscript in preparation. surface of the dorsal skin of one mouse exhibited vigorous 880 KAWANO ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 3 Localization of ﬁbroblast growth factor (FGF)18 mRNA to hair follicles. A portion of the dorsal skin of 8-wk-old C3H/ HeN mice was isolated (E; telogen) or depilated and isolated after 8 d (A and B; anagen V), 18 d (C; anagen VI) or 20 d (D; catagen). The harvested skin samples were immediately frozen in OCT compound and sliced into 10-mm-thick sections. In situ hybridization was then carried out using digoxygenin-labeled antisense riboprobes specific for FGF18 (A, C, D, and E). As a control, the sense probe was also prepared and used for hybridization (B). The bound riboprobe was detected using anti-digoxygenin antibody conjugated with alkaline-phosphatase; the signals were visualized using BM Purple (arrows; purple signals). Some sections were counterstained with Kern- echtrot solution (pink color; right two panels in A and B and the right panel in C and D). Scale bars ¼ 100 mm.

hair growth throughout the entire test area, whereas three The skin from one of the five mice that received FGF18 others exhibited hair growth and/or pigmentation at various showed no apparent changes, whereas none of the skin parts of the test area (photographs not shown). Examination samples from the five control mice exhibited hair growth or of the reverse side of the skin revealed extensive growth of strong pigmentation. Examination of the reverse side of anagen hair follicles in the mice with pigmentation and/or control skin revealed little or no anagen hair follicles, as is hair growth (Fig 5A and B, the reverse side of the pigmented indicated by their white color (Fig 5D). Finally, these results area is shown; black spots indicate anagen hair follicles). were further confirmed by histological examination of skin We also noted that the skin was thicker in these areas than sections (Fig 5C and E; representative results are shown). was the skin from control mice (Fig 5C and E), just as nor- Anagen follicles were often observed in, but not limited mal skin with physiological anagen hair follicles is thicker. to, the approximate area of the implanted beads, suggest- 124 : 5 MAY 2005 FGF AND FGFR EXPRESSION IN MOUSE SKIN 881

A HDPC ing that FGF18 protein was not tightly bound to the Seph- 5 arose beads. When a similar set of experiments in which FGF18 FGF 7 mice were kept for 24 d was carried out, ﬁve of eight mice **

) 4

3 that received FGF18 showed apparent induction of anagen − 3 hair follicles, as judged by pigmentation on the reverse side of the resected skin. By contrast, only two of seven mice 2 that received phosphate-buffered saline (PBS) did so. (dpm x 10 1 Finally, we carried out another in vivo experiment, en- H]-TdR incorporated

3 tailing examination of larger numbers of mice for a longer [ 0 0 10 100 1000 0 10 100 1000 period of time (57 d) after injection of FGF18. In this exper- FGF (ng/ml) iment, all of the FGF18-injected mice showed vigorous hair growth by the end of the observation period, as did some of B HORSC 6 the PBS-injected mice (Fig 6). For comparison, the ratio of FGF18 FGF 7 the hairy area to the total test area in each mouse was 5 ** ** compared between the FGF18-and PBS-administrated ) ** 5

− 4 mice. Statistical analysis showed this ratio to be signiﬁ- ** 3 cantly larger in FGF18-injected than PBS-injected mice [0.424 0.260 (n ¼ 15) vs 0.174 0.124 (n ¼ 11); p 0.01 2 o

(dpm x 10 (ANOVA)]. 1 H]-TdR incorporated 3 [ 0 0 10 100 1000 0 10 100 1000 Discussion FGF (ng/ml) Among the many polypeptide growth factor systems in- C HDF volved in the dynamics of dermal function, the FGF system 8 FGF18 FGF 7 appears to play multiple roles in such physiological proc- ** esses as wound healing and hair growth. No earlier studies, ) 6 4

− however, have presented a comprehensive or quantitative ** 4 analysis of Fgf or Fgfr expression in skin and its append- * ** ages. We have now carried out such a study, and our results clearly show that a number of Fgf genes are active in mouse (dpm x 10 2 skin and that their expression changes dynamically and in H]-TdR incorporated 3 [ 0 differing patterns during the hair growth cycle. This sug- 0 10 100 1000 0 10 100 1000 gests that they play distinct roles in the regulation of hair FGF (ng/ml) growth and related events. D HEK FGFR2 IIIb and its ligands, FGF7, 10 and, presumably, 8 22, comprise an important cell-type-specific system for FGF18 ** FGF 7 *

) * 4 6 * Figure 4 − Target cell speciﬁcity of ﬁbroblast growth factor (FGF)18 protein. 4 (A) Effects of FGF18 and FGF7 on DNA synthesis in human hair follicle dermal papilla cells (HDPC). Serum-starved HDPC were stimulated with the indicated FGF at the indicated concentrations and 5 mg per mL (dpm x 10 2 heparin. After 8 h, [3H]-thymidine was applied to the cells for 18 h, and H]-TdR incorporated

3 the incorporated radioactivity was counted as described in the text. (B) [ 0 Effects of FGF18 and FGF7 on DNA synthesis in human hair follicle 0 10 30 100 0 10 30 100 outer root sheath cells (HORSC). HORSC were starved for 24 h and FGF (ng/ml) stimulated with the indicated FGF at the indicated concentrations in the keratinocyte basal medium containing insulin and 5 mg per mL heparin. E HUVEC After 5 h, [3H]-thymidine was applied to the cells for 21 h, and the incorporated radioactivity was counted. (C) Effects of FGF18 and FGF7 FGF18 FGF 7 6 ** on DNA synthesis in human dermal ﬁbroblasts (HDF). Serum-starved )

4 HDF were stimulated with the indicated FGF at the indicated concen-

− 3 * ** trations without heparin. After 18 h, [ H]-thymidine was applied to the 4 cells for 8 h, and the incorporated radioactivity was counted. (D) Effects of FGF18 and FGF7 on DNA synthesis in human epidermal keratinocytes (HEK). HEK were stimulated with the indicated FGF at the in-

(dpm x 10 2 dicated concentrations in keratinocyte basal medium containing insulin 3

H]-TdR incorporated and 5 mg per mL heparin. After 20 h, [ H]-thymidine was applied to the 3 [ cells for 8 h, and the incorporated radioactivity was counted. (E) Effects 0 0 10 100 1000 0 10 100 1000 of FGF18 and FGF7 on DNA synthesis in human umbilical vein end- FGF (ng/ml) othelial cells (HUVEC). HUVEC were stimulated with the indicated FGF at the indicated concentrations in M199 medium containing 15% FBS and 5 mg per mL heparin. After 19 h, [3H]-thymidine was applied to the cells for 6 h, and the incorporated radioactivity was counted. Columns and bars represent means SE of triplicate samples. Asterisks (Ã and þ ) indicate statistical significance (po0.05 and po0.01, respectively) between FGF-treated and control (0 ng per mL FGF) cells. 882 KAWANO ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 6 Induction of hair growth by ﬁbroblast growth factor (FGF)18 protein in mice with telogen stage hair follicles. Fifteen 56-d-old male C3H/HeN mice were anesthetized, their dorsal hair was gently cut short, and 1 mg of recombinant FGF18 protein in 150 mL of Sepharose bead suspension was injected subcutaneously into the dorsal test area (n ¼ 15) (A). As a control, another group of mice received phosphate- buffered saline in bead suspension (n ¼ 11) (B). After 57 d, all of the mice receiving FGF18 showed vigorous hair growth. The mice were then anesthetized, sacriﬁced, and photographed; three representative mice for each treatment are shown. The ratio of the hairy area to the total test area in each mouse was measured and analyzed statistically (results described in the text).

ously characterized keratinocyte mitogens are indeed ex- Figure 5 pressed at high levels in skin. Furthermore, we noted Anagen-inducing effect of fibroblast growth factor (FGF)18 protein remarkable variation in their mRNA expression throughout in mice with telogen stage hair follicles. (A) Reverse side of resected the hair cycle. It is interesting that both FGF7 and FGF10 are skin from mice receiving FGF18. Five 50-d-old male C3H/HeN mice were anesthetized, their dorsal hair was gently cut short, and 1 mgof maximally expressed at anagen V (day 8), when hair grows recombinant FGF18 protein in 150 mL of Sepharose bead suspension vigorously, whereas expression of FGF22, a presumed au- was injected subcutaneously into the dorsal test area. The mice were tocrine keratinocyte growth factor, is strongly expressed at then maintained on standard laboratory diet and water ad libitum. After 21 d, the exterior appearance of the skin of four of five mice showed anagen VI (day 18), when hair follicle length reaches its pigmentation; among them three also showed hair growth over all or maximum. FGF1, a wide-specificity prototype FGF ligand part of the depilated area (not shown). One mouse apparently did not with the capacity to stimulate keratinocytes, is expressed in respond to FGF18 during that period. The mice were anesthetized and inner root sheath cells and other follicular cells (Kawano sacrificed, the dorsal skin was excised at full-thickness, and its reverse side was photographed. A representative result is shown. (B) Enlarged et al, 2004) and appears to induce dermal papilla cells to photograph of the skin in panel A (squared area). (C) Histology of the secrete growth promoting factors affecting the outer root hair follicles from mice administered FGF18. The skin in A was em- sheath cells (Ota et al, 2002). Thus, although FGF1, 7, 10, bedded in paraffin, sectioned (4 mm thickness), stained with hem- and 22 are all capable of inducing cellular proliferation of the atoxylin (blue) and photographed. A representative result is shown. (D) Reverse side of resected skin from mice receiving phosphate-buffered keratinocyte lineage, they differ in their expressing cells, saline (PBS). The same procedure described in A was conducted ex- mode of action, and target populations. cept that PBS was used instead of FGF18. A representative result is A gene knockout study revealed that FGF5 is responsible shown. (E) Histology of the hair follicles from mice administrated PBS. The skin in D was processed as in C. A representative result is shown. for inducing catagen during the hair cycle, presumably by Scale bars ¼ 100 mm. inhibiting activation of dermal papilla cells (Ota et al, 2002). We found that there was little expression of FGF5 mRNA at telogen (Fig 1A; days 0 and 22), but that the level was 20- regulating keratinocyte proliferation. Although FGF7 and fold higher at anagen VI (day 18), which is consistent with its FGF10 are expressed in dermal fibroblasts and papilla cells ability to induce catagen. and stimulate proliferation of keratinocytes, FGF22 is ex- Other FGF also showed variation in their expression that pressed in epidermal and hair follicle keratinocytes and also was tied to hair cycle progression. Among strongly ex- stimulates keratinocyte proliferation (Nakatake et al, 2001; pressed FGF, FGF13 is expressed in hair follicle bulge cells Beyer et al, 2003). Thus, this FGF system stimulates kera- and some basal layer keratinocytes (Kawano et al, 2004). tinocyte proliferation via both an epithelial–mesenchymal Although its activity related to stem cell function is not yet interaction and epithelial autocrine/paracrine stimulation. understood, the pattern of its mRNA expression observed in The present results confirmed that all three of these previ- this study suggests a role during telogen. 124 : 5 MAY 2005 FGF AND FGFR EXPRESSION IN MOUSE SKIN 883

We showed here for the first time that FGF18 mRNA is guidelines of National Institute of Advanced Industrial Science and strongly expressed in hair follicles. Our in situ hybridization Technology (AIST). All protocols were approved by the animal ex- showed that FGF18 mRNA was expressed in the bulge re- periment committee of AIST Central 6. For hair cycle experiments, 25 seven-wk-old male C3H/HeN mice (Tlr4 þ / þ ) were obtained gion during telogen, and that the site of expression moves from Japan SLC (Hamamatsu, Japan) and maintained on a stand- to the transient portion of the inner root sheath with pro- ard laboratory diet and water ad libitum. After conditioning for gression of hair follicle growth. We also found that FGF18 1 wk, five mice were sacrificed, the dorsal hair was cut short, and stimulates DNA synthesis in dermal fibroblasts and papilla skin samples were isolated and examined as the ‘‘day 0 (telogen)’’ cells, vascular endothelial cells, and also epidermal kera- sample. In the remaining 20 mice, the anagen phase of the hair tinocytes (Fig 4). This was somewhat unexpected, as growth cycle was induced by gently depilating the dorsal hair FGF18 is known to activate FGFR2 IIIc and FGFR3 IIIc, shafts with gloved fingers, without using wax, tape, or depilatory chemicals. Then, at selected stages of the hair follicle cycle (Paus, but activity via FGFR2 IIIb, the keratinocyte-specific recep- 1999)—i.e., on days 8 (anagen V), 18 (anagen VI), 20 (catagen IV), tor, has never been reported. Still, one earlier study showed and 22 (telogen) after depilation—the mice (n ¼ 5 in each group) that administration of recombinant FGF18 to mice induces were sacrificed, the hair was cut short for days 18, 20, and 22 proliferation of hepatic parenchymal cells and pancreatic mice, and full-thickness skin samples (containing epidermis, hair ductal epithelium (Hu et al, 1998), illustrating FGF18’s ca- shafts, hair follicles, sebaceous glands, subcutaneous adipose pacity to exert effects on a wide range of cell types. tissue, dermal muscles, and blood vessels) were excised from the depilated area. The harvested skin samples were then processed FGF18 is also reportedly expressed in the developing for mRNA isolation, in situ hybridization or immunohistochemical brain (Hoshikawa et al, 2002), pancreas (Dichmann et al, staining. The experiment was carried out three times using essen- 2003), and adipose tissue (Gabrielsson et al, 2002). It also tially the same design to confirm the reproducibility of the results. reportedly stimulates proliferation of osteoblasts and For in situ hybridization, the skin samples were immediately chondrocytes (Shimoaka et al, 2002), influences lung frozen in optimum cutting temperature compound using liquid ni- morphogenesis (Whitsett et al, 2002) and improves learn- trogen. For immunohistochemical staining, the samples were first ing and memory (Cavallaro et al, 2002). FGFR-2 and FGFR- fixed for 16–24 h at room temperature in 10% formaldehyde and then embedded in paraffin using standard procedures (Suzuki et al, 3 are expressed in hair follicle matrix cells (Widelitz et al, 1998). 1997), suggesting that these cells may be the targets of FGF 18 activity. Interestingly, it was recently reported that Fgf18 mRNA preparation and reverse transcription Total RNA was is a target gene of Wnt/b-catenin (Shimokawa et al, 2003), prepared from skin samples using Isogen (Nippon Gene, Tokyo, which is an essential signaling pathway for skin and hair Japan) according to the manufacturer’s instructions, after which formation (Van Mater et al, 2003). FGF18 may thus be an the mRNA was purified from the total RNA using an Oligotex-dT30 Super mRNA purification kit (Takara Bio, Osaka, Japan). Samples important modulator of skin and hair formation. (100 ng) of the purified mRNA were then reverse transcribed in a These results prompted us to investigate the activity of total volume of 20 mL using SuperScript II with Oligo (dT)12–18 as FGF18 on hair growth in vivo. Administration to mice with a primer (Gibco BRL, Tokyo, Japan). dorsal telogen phase hair follicles revealed that FGF18 is capable of inducing the anagen growth state, which is con- Design of PCR primers and the cloning of FGF cDNA Specific sistent with its involvement in modulating hair growth. It is primer sets were designed so that the levels of FGF mRNA could be quantitatively analyzed using real-time PCR. The specificity of presently unclear whether FGF18 directly promotes the the primers was mainly ensured by using the whole mouse cDNA proliferation of hair follicle keratinocytes or does so indi- sequence as the background sequence (a new algorithm devel- rectly. Given that, in vitro, FGF18 induced DNA synthesis in oped by Dr Akira Suyama at Tokyo University). The same sets of HDPC and HUVEC but not HORSC, even at a high con- primers were also used to clone the respective cDNA fragments centration, it is unlikely that its ability to potently promote that served as copy number references for quantification. The primers used for FGF18 amplification were 50-AGACGCGGGCTC- hair growth in vivo involves direct mitogenic activity. In- 0 0 stead, the observation that it takes 3 wk or more to induce GAGATGATGTGAGTCGGA-3 (sense) and 5 -GCCCTTGATCCG- GACTTGACTCCCGAAGGT-30 (antisense). The RT mixture was hair growth suggests that FGF18 acts indirectly to modulate diluted ten times with dH2O, after which PCR amplification was the hair growth cycle. FGF18 may induce dermal papilla carried out using Pfu polymerase according to the manufacturer’s cells to secrete growth factors affecting hair follicle cells, or (Stratagene, La Jolla, California) instructions. Thereafter, aliquots of it may affect blood circulation by acting on vascular end- the reaction mixture were run on 2.0% agarose gels to confirm othelial cells. On the other hand, FGF18 exerted a mitogenic the size of the amplified products. To clone the respective cDNA, effect on dermal fibroblasts (Fig 4) and may be involved in the PCR products were ligated into pCR-Blunt II-TOPO vector (In- vitrogen, Carlsbad, California), which was then used to transform the growth of adipocytes (Gabrielsson et al, 2002), sug- Escherichia coli. The nucleotide sequences of the recombinant gesting that the increased skin mass seen in FGF18-treated plasmids were verified using a BigDye terminator cycle sequencing mice may be attributable to a direct effect on those cells. kit and an ABI PRISM 310 Genetic Analyzer (Applied Biosystems, In sum, the results suggest that by acting on various cell Foster City, California). types under specific sets of conditions, FGF dynamically regulate dermal system function. Direct administration of Quantification of mRNA copy number by real-time PCR Real- time PCR amplification was carried out in a Light Cycler (Roche FGF18 and/or modulation of its expression may thus be an Diagnostics, Basel, Switzerland) using 2 mL of RT product diluted effective approach to stimulating hair growth. 1:10 as a template. The absolute copy numbers of each FGF was determined using each respective plasmid DNA as a copy number reference.2 Cyber Green was used for quantification of the ampli- Materials and Methods

Mice and skin sample preparation Throughout this study, all an- imals received humane care that was in accordance with the 2Suzuki M, et al, manuscript in preparation. 884 KAWANO ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

fied DNA. Highly specific primers, as well as optimized annealing absorbed calf serum (CH-CS) (HDF) or Dulbecco’s Modified Ea- temperatures, elongation time and acquisition temperatures were gle’s Medium (DMEM) containing 0.5% CH-CS (HDPC), after which employed in the Light Cycler reaction. To further ensure the the cells were incubated for 48 h (HDF) or 96 h (HDPC) before specificity of the reaction conditions, at the end of the individual stimulation with the indicated concentration of FGF. The recom- runs, the melting temperature (Tm) of the amplified products was binant FGF18 and FGF7 (KGF) proteins were purchased from Wako measured to confirm its homogeneity. After the quantification pro- Pure Chemical Industries (Osaka, Japan). After 18 h (HDF) or 8h cedure, the products were resolved by 2.0% agarose gel elect- (HDPC) in the presence of FGF with 5 mg per mL heparin (HDPC) or rophoresis to confirm that the reaction had amplified the correct without heparin (HDF), [3H]-thymidine (2 mCi per mL; Moravek Bi- DNA fragments of known size. ochemicals; California) was added to the medium, and the cells For the evaluation of the mRNA, we included serially diluted were incubated for an additional 8 h (HDF) or 18 h (HDPC). The DNA samples with the target cDNA sequence in the same Light cells were then harvested, and the incorporated radioactivity was Cycler run; from those measurements, the absolute copy number counted. of the mRNA of interest was calculated for the individual samples. Cultured primary HEK (from adult facial skin; Toyobo) and pri- We also evaluated the copy number of the housekeeping gene b- mary HORSC (from adult male scalp) were subcultured in keratin- actin using the same procedures and confirmed that its abundance ocyte growth medium [keratinocyte basal medium (KBM)-2 was within the expected range. supplemented with bovine pituitary extract, insulin, epidermal growth factor, hydrocortisone, transferrin, epinephrine, and antibi- In situ hybridization The in situ hybridization experiments were otics; Toyobo] and then used within two passages. The cells were carried out using OCT-embedded frozen tissue sections and dig- trypsinized and plated to a density of 1 104 cells per well per 0.5 oxygenin-labeled riboprobes essentially as described previously mL in 48-well (HEK) or 1 104 cells per well per 1 mL in 24-well (Komminoth, 1992). Sense and antisense riboprobes for FGF18 (HORSC) collagen-coated plates and then maintained in the ker- were synthesized using SP6 and T7 RNA polymerases, respec- atinocyte growth medium at 371C. The next day (HEK) or 4 d later tively, in the presence of digoxygenin-dUTP (Roche Applied Sci- (HORSC), the growth medium was replaced with KBM-2 medium ence, Mannheim, Germany). All possible effort was made to avoid containing 8.6 107 M insulin, hydrocortisone, transferring, RNA degradation during the hybridization procedure. The frozen epinephrine (HEK) or with KBM-2 medium containing 8.6 107 tissues were cut into 10 mm sections using a Cryostat (Microm, M insulin (HORC). The HEK were stimulated with the indicated model HM500OM, Walldorf, Germany), immediately placed on concentration of FGF with 5 mg per mL heparin. The HORSC were MAS-coated slide glasses (Matsunami Glass, Tokyo, Japan), dried cultured for 24 h and then stimulated with FGF with 5 mg per mL at 501C for 1 h, and then stored at 801C with a desiccant. We pre- heparin. After 20 h (HEK) or 5 h (HORSC) in the presence of FGF, viously determined that fixing the skin tissue before freezing was [3H]-thymidine (2 mCi per mL; Moravek) was added to the medium, disadvantageous, as it impaired signal detection and reduced the the cells were incubated for an additional 8 h (HEK) or 21 h flexibility of the tissue sections, leading to their easy detachment (HORSC) and assayed as described above. from slide glasses. Before hybridization, the sections were fixed in Cultured HUVEC were subcultured in M199 medium supple- 4% paraformaldehyde in PBS for 10 min at room temperature and mented with 15% FBS, 10 ng per mL FGF1, and 5 mg per mL then washed with PBS for 5 min. Thereafter, the sections were heparin (HUVEC growth medium) and were used for 15 passages. permeabilized with proteinase K (0.01 mg per mL in 2 10 mM Tris- For experimentation, the cells were first trypsinized and plated Hcl, 1 mM EDTA, pH) for 5 min at 371C, rinsed with PBS, again fixed to a density of 2 104 cells per well per 1 mL in 24-well collagen- with 4% paraformaldehyde in PBS for 5 min, and washed with PBS coated plates and maintained for 1 d in M199 medium for 5 min. The fixed sections were soaked twice in 0.1% diethyl- supplemented with 15% FBS at 371C. Thereafter, the cells were pyro carbonate in PBS for 15 min, rinsed with PBS, then soaked in stimulated with the indicated concentration of FGF with 5 mg per 5 saline-sodium citrate for 15 min. Finally, hybridization was mL heparin. After 19 h in the presence of FGF, [3H]-thymidine (2 mCi carried out for 16–18 h at 451C using 500 ng per mL of heat- per mL; Moravek) was added to the medium, and the cells denatured (701C for 10 min) probe in 100 mL of hybridization buffer were incubated for an additional 6 h and assayed as described consisting of 50% (vol/vol) formamide, 5 SSC, 500 mg per mL above. salmon testes DNA, 250 mg per mL tRNA, 5 Denhardt’s solution, and 1 mM dithiothreitol. After hybridization, the slides were rinsed In vivo analysis of FGF18 activity To characterize the effect of with 2 SSC at 451C and then washed twice with 2 SSC for 30 FGF18 on hair follicles in vivo, we essentially followed the protocol min at 451C and twice with 0.2 SSC for 20 min at 451C. The developed by McElwee et al (2004) to demonstrate the activity of bound riboprobe was detected using an anti-digoxygenin antibody macrophage stimulating protein. Briefly, 10 mg of recombinant conjugated with alkaline-phosphatase (1:500, Roche Applied FGF18 protein in 0.75 mL of PBS were mixed with 0.75 mL (packed Science). For visualization, a colorimetric reaction with nitro-blue volume) of Sepharose 4B beads and 7.5 mL of C3H/HeN serum and tetrazolium and 5-bromo-4-chloro-3-indolyl-1-phosphate (NBT/ incubated at 371C for 2 h. The FGF18-treated bead suspension BCIP) was carried out using BM Purple (Roche Applied Science) was used for injection without washing. Control beads were pre- at room temperature in the dark until the purple signals developed pared using the same procedure without FGF18. Fifty-d-old male [24 h for day 8 (anagen V) sections, 48 h for days 18 (anagen VI) C3H/HeN mice were then anesthetized, their dorsal hair was gently and 20 (catagen) sections, and 96 h for day 0 (telogen) sections]. cut short with a trimmer, and the FGF18-treated bead suspension Some sections were further counterstained (pink) with Kernechtrot was injected subcutaneously into the test area (150 mL per mouse; Stain solution (Muto Pure Chemicals, Tokyo, Japan). i.e., 1 mg FGF18 per mouse). In some experiments 56-day-old mice were used. The mice were then maintained for selected numbers of DNA synthesis assays Cultured primary HDF (from adult fore- days on a standard laboratory diet and water ad libitum, after skin; Toyobo, Tokyo, Japan) subcultured in HDF growth medium which they were anesthetized, sacrificed, and photographed using [fibroblast basal medium (FBM) with growth supplement; Toyobo] a digital camera. The full-thickness of the dorsal skin in the test and cultured primary HDPC (from adult scalp, Toyobo) subcultured area was excised and, after its reverse side was photographed, it in HDPC growth medium (Toyobo) were used within two passages was processed for paraffin embedding. The embedded skin sam- for [3H]-thymidine incorporation assays. The cells were trypsinized ples were cut into 4 mm sections using a microtome, stained with and plated to a density of 1 104 cells per well per 0.5 mL in 48- hematoxylin (blue) and observed under a microscope. In some well (HDF) or 1 104 cells per well per 1 mL in 24-well (HDPC) experiments, the ratio of the hairy area versus the total test area collagen-coated plates (Sumilon, Tokyo, Japan) and then main- was measured in the images of each mouse using ImageJ image tained in the respective growth medium at 371C. The next day, the analysis software (http://rsb.info.nih.gov/ij/) and was processed for growth medium was replaced with FBM containing 0.5% charcoal- statistical evaluation. 124 : 5 MAY 2005 FGF AND FGFR EXPRESSION IN MOUSE SKIN 885

Marchese C, Chedid M, Dirsch OR, et al: Modulation of keratinocyte growth We thank Prof. Tadashi Baba at the University of Tsukuba for his sup- factor and its receptor in reepithelializing human skin. J Exp Med port of the graduate study of Mitsuko Kawano, and Prof. Akira Suyama 182:1369–1376, 1995 at Tokyo University and Olympus Optical Co. Ltd for primer design McElwee KJ, Huth A, Kissling S, Hoffman R: Macrophage-stimulating protein software using a novel algorithm. We also thank Dr Kevin J. McElwee promotes hair growth ex vivo and induces anagen from telogen stage hair at the University of British Columbia for his helpful suggestion regard- follicles in vivo. J Invest Dermatol 123:34–40, 2004 ing our in vivo hair growth assay. Akiko Komi-Kuramochi was sup- Mitsui S, Ohuchi A, Hotta M, Tsuboi R, Ogawa H: Genes for a range of growth ported in part by a grant-in-aid from the Ministry of Education, Culture, factors and cyclin-dependent kinase inhibitors are expressed by isolated Sports, Science and Technology, Japan. human hair follicles. Br J Dermatol 137:693–698, 1997 DOI: 10.1111/j.0022-202X.2005.23693.x Nakatake Y, Hoshikawa M, Asaki T, Kassai Y, Itoh N: Identification of a novel fibroblast growth factor, FGF-22, preferentially expressed in Manuscript received May 30, 2004; revised November 25, 2004; the inner root sheath of the hair follicle. Biochem Biophys Acta 1517: accepted for publication January 6, 2005 460–463, 2001 Ohuchi H, Hori Y, Yamasaki M, Harada H, Sekine K, Kato S, Itoh N: FGF10 acts as Address correspondence to: Toru Imamura, PhD, National Institute of a major ligand for FGF receptor 2 IIIb in mouse multi-organ development. Advanced Industrial Science and Technology (AIST), Central 6, 1-1-1 Biochem Biophys Res Commun 277:643–649, 2000 Higashi, Tsukuba, Ibaraki 305-8566, Japan. Email: imamura-toru@aist. Ornitz DM, Itoh N: Fibroblast growth factors. Genome Biol 2REVIEWS3005, 1–12, go.jp 2001 Ortega S, Ittmann M, Tsang SH, Ehrlich M, Basilico C: Neuronal defects and delayed wound healing in mice lacking fibroblast growth factor-2. Proc Natl Acad Sci USA 95:5672–5677, 1998 References Ota Y, Saitoh Y, Suzuki S, Ozawa K, Kawano M, Imamura T: Fibroblast growth factor 5 inhibits hair growth by blocking dermal papilla cell activation. Beyer TA, Werner S, Dickson C, Grose R: Fibroblast growth factor 22 and its Biochem Biophys Res Commun 290:169–176, 2002 potential role during skin development and repair. Exp Cell Res 287: Ozawa K, Suzuki S, Asada M, et al: An alternatively spliced fibroblast growth 228–236, 2003 factor (FGF)-5 mRNA is abundant in brain and translates into a partial Cavallaro S, D’Agata V, Manickam P, Dufour F, Alkon DL: Memory-specific tem- agonist/antagonist for FGF-5 neurotrophic activity. J Biol Chem 273: poral profiles of gene expression in the hippocampus. Proc Natl Acad Sci 29262–29271, 1998 USA 99:16279–16284, 2002 Petho-Schramm A, Muller HJ, Paus R: FGF5 and the murine hair cycle. Arch Danilenko DM, Ring BD, Yanagihara D, Benson W, Wiemann B, Starnes CO, Dermatol Res 288:264–266, 1996 Pierce GF: Keratinocyte growth factor is an important endogenous Rosenquist TA, Martin GR: Fibroblast growth factor signalling in the hair growth mediator of hair follicle growth, development, and differentiation. Am J cycle: Expression of the fibroblast growth factor receptor and ligand Pathol 147:145–154, 1995 genes in the murine hair follicle. Dev Dyn 205:379–386, 1996 Dichmann DS, Miller CP, Jensen J, Scott Heller R, Serup P: Expression and Shimoaka T, Ogasawara T, Yonamine A, et al: Regulation of osteoblast, misexpression of members of the FGF and TGFbeta families of growth chondrocyte, and osteoclast functions by fibroblast growth factor (FGF)- factors in the developing mouse pancreas. Dev Dyn 226:663–674, 2003 18 in comparison with FGF-2 and FGF-10. J Biol Chem 277:7493–7500, du Cros DL: Fibroblast growth factor and epidermal growth factor in hair devel- 2002 opment. J Invest Dermatol 101:106S–113S, 1993 Shimokawa T, Furukawa Y, Sakai M, Li M, Miwa N, Lin YM, Nakamura Y: In- du Cros DL, Isaacs K, Moore GP: Distribution of acidic and basic fibroblast volvement of the FGF18 gene in colorectal carcinogenesis, as a novel growth factorsin ovine skin during follicle morphogenesis. J Cell Sci downstream target of the beta-catenin/T-cell factor complex. Cancer Res 105:667–674, 1993 63:6116–6120, 2003 Gabrielsson BG, Johansson JM, Jennische E, et al: Depot-specific expression of Stenn KS, Paus R: Controls of hair follicle cycling. Physiol Rev 81:449–494, fibroblast growth factors in human adipose tissue. Obes Res 10:608–616, 2001 2002 Suzuki S, Kato T, Takimoto H, et al: Localization of rat FGF-5 protein in skin Guo L, Degenstein L, Fuchs E: Keratinocyte growth factor is required for hair macrophage-like cells and FGF-5S protein in hair follicle: Possible in- development but not for wound healing. Genes Dev 10:165–175, 1996 volvement of two Fgf-5 gene products in hair growth cycle regulation. J Hebert JM, Rosenquist T, Gotz J, Martin GR: FGF5 as a regulator of the hair Invest Dermatol 111:963–972, 1998 growth cycle: Evidence from targeted and spontaneous mutations. Cell Suzuki S, Ota Y, Ozawa K, Imamura T: Dual-mode regulation of hair growth 78:1017–1025, 1994 cycle by two Fgf-5 gene products. J Invest Dermatol 114:456–463, Hoshikawa M, Yonamine A, Konishi M, Itoh N: FGF-18 is a neuron-derived glial 2000 cell growth factor expressed in the rat brain during early postnatal de- Takahashi M, Funato T, Kumura-Ishii K, Kaku M, Sasaki T: Measurement of tumor velopment. Brain Res Mol Brain Res 105:60–66, 2002 necrosis factor-alpha messenger RNA in synovial fibroblasts by real-time Hu MC, Qiu WR, Wang YP, et al: FGF-18, a novel member of the fibroblast growth quantitative reverse transcriptase-polymerase chain reaction. J Lab Clin factor family, stimulates hepatic and intestinal proliferation. Mol Cell Biol Med 137:101–106, 2001 18:6063–6074, 1998 Van Mater D, Kolligs FT, Dlugosz AA, Fearon ER: Transient activation of beta- Igarashi M, Finch PW, Aaronson SA: Characterization of recombinant catenin signaling in cutaneous keratinocytes is sufficient to trigger the human fibroblast growth factor (FGF)-10 reveals functional similarities active growth phase of the hair cycle in mice. Genes Dev 17:1219–1224, with keratinocyte growth factor (FGF-7). J Biol Chem 273:13230–13235, 2003 1998 Werner S, Smola H, Liao X, Longaker MT, Krieg T, Hofschneider PH, Williams LT: Jaye M, Schlessinger J, Dionne CA: Fibroblast growth factor receptor tyrosine The function of KGF in epithelial morphogenesis and wound re-epithe- kinases: Molecular analysis and signal transduction. Biochim Biophys lialisation. Science 266:819–822, 1994 Acta 1135:185–199, 1992 Whitsett JA, Clark JC, Picard L, et al: Fibroblast growth factor 18 influences Kawano M, Suzuki S, Suzuki M, Oki J, Imamura T: Bulge-and basal layer-specific proximal programming during lung morphogenesis. J Biol Chem 277: expression of fibroblast growth factor 13 (FHF-2) in mouse skin. J Invest 22743–22749, 2002 Dermatol 122:1084–1090, 2004 Widelitz RB, Jiang TX, Noveen A, Ting-Berreth SA, Yin E, Jung HS, Chuong CM: Komminoth P: Digoxygenin as an alternative probe labeling for in situ hybridi- Molecular histology in skin appendage morphogenesis. Microsc Res Tech zation. Diag Mol Pathol 1:142–150, 1992 38:452–465, 1997