Runx3 is involved in hair shape determination


  • Eli Raveh,

    1. Department of Cell and Animal Biology, Silberman Life Sciences Institute, Edmond Safra Campus at Givat-Ram, The Hebrew University, Jerusalem, Israel
    Search for more papers by this author
  • Shulamit Cohen,

    1. Department of Cell and Animal Biology, Silberman Life Sciences Institute, Edmond Safra Campus at Givat-Ram, The Hebrew University, Jerusalem, Israel
    Search for more papers by this author
  • Ditsa Levanon,

    1. Department of Molecular Genetics, The Weizmann Institute of Science, Rehovot, Israel
    Search for more papers by this author
  • Yoram Groner,

    1. Department of Molecular Genetics, The Weizmann Institute of Science, Rehovot, Israel
    Search for more papers by this author
  • Uri Gat

    Corresponding author
    1. Department of Cell and Animal Biology, Silberman Life Sciences Institute, Edmond Safra Campus at Givat-Ram, The Hebrew University, Jerusalem, Israel
    • Department of Cell and Animal Biology, Silberman Life Sciences Institute, Edmond Safra Campus at Givat-Ram, The Hebrew University, Jerusalem 91904, Israel
    Search for more papers by this author


Transcriptional regulators of the Runx family play critical roles in normal organ development and, when mutated, lead to genetic diseases and cancer. Runx3 functions during cell lineage decisions in thymopoiesis and neurogenesis and mediates transforming growth factor-β signaling in dendritic cells. Here, we study the function of Runx3 in the skin and its appendages, primarily the hair follicle, during mouse development. Runx3 is expressed predominantly in the dermal compartment of the hair follicles as they form and during the hair cycle, as well as in the nail and sweat gland skin appendages. Distinct expression is also detected periodically in isolated cells of the epidermis and in melanocytes, populating the hair bulb. Runx3-deficient mice display a perturbation of the normal hair coat, which we show to be due to hair type and hair shape changes. Thus, one of the functions of Runx3 in skin may be to regulate the formation of the epithelial derived structural hair by affecting dermal to epidermal interactions. Developmental Dynamics 233:1478–1487, 2005. © 2005 Wiley-Liss, Inc.


Skin appendages are an excellent model for morphogenetic events, as they give rise to different complex structures sculpted from the same basic template—the embryonic ectoderm (Thesleff et al., 1995; Chuong et al., 2000). Like many other organs, these appendages form as a result of reciprocal interactions between the embryonic mesenchymal and epithelial tissues, as was demonstrated in the most well-studied case of hair follicle development (Sengel, 1986).

Hair follicles are initiated when dermal aggregates instruct the overlying ectoderm to form epithelial placodes. In turn, the placode signals to the dermal cells to condensate and give rise to precursors of the dermal counterpart of future hair follicle. This process is followed by another signal from the dermal cells to the overlying epidermal cells to encase the dermal cells and to form a hair follicle. The hair follicle then elongates and grows down into the dermis and starts a complex differentiation plan to form the final output of this organ, the structural hair (Hardy, 1992; Chuong, 1998).

The pit of dermal cells enveloped by the epithelial cells is termed the dermal papilla (DP) and is important for the regulation of hair follicle function. Other mesenchymal cells are not enclosed by the epidermal part and form the dermal sheath (DS), which envelops the hair follicle. The epithelial cells encasing the DP are termed the hair follicle matrix and rapidly proliferate to provide cell progeny, which undergo differentiation into six concentric layers. The three outermost layers form the inner root sheath (IRS), rendering support to the three innermost layers designated, from outside to inside: cuticle, cortex, and medulla, which constitute the hair shaft (Stenn and Paus, 2001). The cells of the hair shaft layers synthesize high levels of hair keratins, and eventually as they are pushed up by younger cells, fill up with hair keratin filaments and undergo cell death to form the structural inert hair rod, which protrudes out of the body surface. The adult hair follicle undergoes cycles consisting of (1) a growth phase (anagen), in which the hair follicle grows and extends into the skin and is active in structural hair synthesis; (2) a regression/retraction phase (catagen), in which the lower segment of the epithelial compartment disintegrates and the DP is pulled up next to the permanent epithelial base, the bulge; and (3) a resting stage (telogen) and again a growth stage, which resembles in many aspects the embryonic morphogenesis (Ebling, 1987; Hardy, 1992; Stenn and Paus, 2001). This process is believed to be triggered by a signal from the DP to the epithelial stem cells in the bulge region (Cotsarelis et al., 1990; Taylor et al., 2000), which reforms the matrix, giving rise to a new lower portion of the hair follicle and new hair production.

The DP, thus, provides the necessary signals governing the proliferation and differentiation of the hair follicle epidermal compartment at all its developmental stages. There is a good correlation between the size of the DP, determined by the number of its mesenchymal cells, and the diameter of the hair fiber produced by the particular hair follicle (Ebling, 1987). An open question remains as to the identity of the molecular signals involved in DP function, especially in the communication between the DP and the epithelial compartments during embryogenesis and in the cycling stages.

The Wnt/wingless pathway, which is involved in multiple developmental processes (Wodarz and Nusse, 1998; Huelsken and Birchmeier, 2001; Logan and Nusse, 2004), also plays an essential role in hair and skin development (Gat et al., 1998; DasGupta and Fuchs, 1999; Fuchs et al., 2001; Andl et al., 2002). A connection between the Wnt pathway and the DP was missing until recently, when the control of the Versican DP marker was studied. Versican is a large chondroitin sulfate proteoglycan, which is highly expressed in the DP of hair follicles (du Cros et al., 1995) and in several other mesenchymal tissues. Study of isolated DP cells expressing green fluorescent protein driven by Versican promoter (Naso et al., 1994; Kishimoto et al., 1999), indicated that components of the Wnt pathway (Wnt5a, Lef-1, β-catenin) are expressed in the DP and that Wnt activation is necessary for maintaining Versican expression as well as for the hair-inducing abilities of DP cells (Kishimoto et al., 2000). Sequence analysis of the promoter revealed a putative binding site for Lef-1/Tcf, which further indicates that Versican is regulated by the Wnt/wingless pathway (Kishimoto et al., 2000).

We searched for other genes that may interact with members of the canonical Wnt pathway and focused on the Runx family, the binding sites of which have often been found close to those of TCF/Lef-1 in enhancers of several genes such as the T cell receptor (Travis et al., 1991; Love et al., 1995). Additionally, both Runx and Lef-1 interact with the transcriptional corepressor Gro/TLE (Levanon et al., 1998), and Lef-1 interacts with Runx2 during transcriptional regulation of the Osteocalcin promoter (Kahler and Westendorf, 2003).

The Runx is a small family of three transcriptional regulators containing a highly conserved DNA binding domain similar to that of the Drosophila pair-rule Runt gene product. The RUNX proteins play roles in proliferation vs. differentiation decisions in various tissues and function as activators or repressors of target genes (Levanon et al., 1994; Speck and Terryl, 1995; Westendorf and Hiebert, 1999; Levanon and Groner, 2004). Studies of human genetic diseases and analysis of knockout mice show that Runx1 is necessary for definitive hematopoiesis and is the most common target of chromosomal abnormalities in leukemias (Nucifora and Rowley, 1995; Wang et al., 1996). Runx2 was found to be essential for osteoblast differentiation and skeletal morphogenesis/bone formation (Ducy et al., 1997; Otto et al., 1997). Runx3 is the smallest member of the family and thought to be the primordial prototype in this paralogous group (Levanon and Groner, 2004). Knockout analysis showed its importance for neurogenesis in the dorsal root ganglia (Inoue et al., 2002; Levanon et al., 2002), for thymopoiesis (Taniuchi et al., 2002; Woolf et al., 2003), and for dendritic cell function (Fainaru et al., 2004).

Expression of Runx3 in the mesenchymal compartment of several organs, such as the whiskers that develop through mesenchymal–epithelial interactions, has been reported previously (Levanon et al., 2001). Here, we study the expression of Runx3 during skin and hair follicle development and demonstrate its role in hair structure determination, presumably by affecting mesenchymal/epithelial cross-talk.


To determine the sites of Runx3 expression in skin and hair during morphogenesis and in the postnatal hair follicle cycle, we have used Runx3-specific antibodies (Levanon et al., 2001). At embryonic day (E) 16.5, before hair placode formation, Runx3 was expressed in a continuous region of dermal cells underlying the primitive epidermis (Fig. 1A). When hair germs are formed (E18.5), a cluster of positively stained cells forms the primordia of the DP adjacent to the epidermal hair germs (Fig. 1B, pre-DP). At birth, as the hair follicle elongates, the Runx3-expressing DP was engulfed by the epithelial compartment and Runx3 could also be detected in the cells of the DS compartment, which is in contact with the base of the DP (Fig. 1C).

Figure 1.

Expression of Runx3 during hair follicle morphogenesis. A: Runx3 antibody staining showing Runx3-positive dermal cells underlying the back skin (BS) epidermis at embryonic day (E) 16.5 (arrows). Arrowheads mark the border between dermis and epidermis. B: At E18.5, Runx3-expressing cells are clustered adjacent to hair germs (HG), forming the primordial dermal papilla (pre). C: At birth, Runx3 is clearly expressed in the mature DP (arrows), and in the primordial DP of secondary follicles, as well as in the dermal sheath cells (arrowheads). D: First anagen at postnatal (P) day 3. The three distinct waves of hair growth (I, first; II, second; III, third) are clearly seen at this stage. A dotted line depicts the border between dermal and epidermal layers. Scale bars = 50 μm.

The pelage of mice contains four different hair types, which differ in their length, width, and shape (Dry, 1926). The first type to form, starting at around E14, is called the guard hair (Primary or tylotrich), followed by the awl type at E17. The last two types to form close to birth, are the auchene and zigzag (underfur) hair, both of which contain bends in their structure (Mann, 1962). To determine whether Runx3 expression differs among these hair types, we analyzed it at day 3 after birth, when the three waves of the hair follicles could be discerned by the extent of their growth into the dermis. As shown in Figure 1D, all the hair types displayed similar Runx3 DP staining during their initial establishment. The DP continued to express Runx3 in most, if not all of its cells, during the entire anagen (growth) stage.

The question then arose whether Runx3 is expressed only during the formation and growth (anagen) phase of the hair follicle, or whether it is found also in other stages of the hair follicle cycle. To address this issue, we immunostained skin sections representing all stages of the hair cycle (Muller-Rover et al., 2001) with Runx3 antibodies and showed persistence of Runx3 expression in catagen (Fig. 2A), telogen (Fig. 2B), and on to the next postnatal anagen growth cycle (Fig. 2C).

Figure 2.

Expression of Runx3 in the dermal papilla (DP) during the hair follicle developmental cycle. A: Mid-catagen stage (V–VI stage) as in (Muller-Rover et al., 2001). The club hair (CH) is indicated. B: Telogen phase. Compact ball-shaped DP cells are attached to the overlying germ cell cap. C: Second anagen phase. SG, sebaceous gland. Scale bars = 50 μm.

The immunohistochemical analysis was confirmed by X-gal staining of skin sections from heterozygous Runx3+/lz mice, which contain a lacZ targeted Runx3 allele (Levanon et al., 2001, 2002). The blue X-gal staining in the DP of the Runx3+/lz hair follicle fully corresponded to the antibody staining in these heterozygous mice, which was similar to that of the wild-type (WT) mice. An even more-intense X-gal staining was also observed in sections from the null mouse (Runx3lz/lz, Fig. 3A), which as expected were negative for the Runx3 antibody staining in the DP (not shown).

Figure 3.

A: β-Galactosidase staining of Runx3 null mouse back skin (BS), at postnatal day (P) 3. Dermal papillae are exclusively stained. B: Epidermal expression of Runx3. Note the periodical arrangement of isolated Runx3-positive interfollicular basal cells (arrows). Inset: Runx3-positive cells also appear in the suprabasal layers (arrowhead) and in the upper outer root sheath (ORS, arrows). Scale bars = 50 μm.

In addition to the major DP dermal expression, we also observed an interesting expression pattern in the epidermis of postnatal mice. Intense Runx3 staining was observed in isolated, evenly distributed cells of the interfollicular epidermis, as well as in the upper outer root sheath. These Runx3-positive cells were mostly located in the basal layer but sometimes also in the supra-basal layers (Fig. 3B). Positive cells were also found in the thickened epidermis of the toes, a region not containing hair follicles, indicating that these cells are not only restricted to the haired skin. Of note, there was no X-gal staining in the epidermis of the heterozygous mice, which have a normal Runx3 antibody-staining pattern in the epidermis. Thus, the lack of lacZ expression in the Runx3-positive epidermal cells may be either due to a different Runx3 splicing form, which does not contain exon 2 (the site of lacZ insertion) or may indicate poor expression of β-galactosidase, which is under the translational control of an internal ribosome entry site element (Levanon et al., 2001).

Another phenomenon that became apparent with the more-advanced stages of anagen (from day 3 onward) was the appearance of Runx3-positive cells in the matrix of the bulb, mostly in close proximity to the DP (Fig. 4A). To distinguish between Runx3-expressing cells of DP from those of the matrix, we used double immunofluorescence staining of Runx3 and β4-integrin (Fig. 4). The later is a marker of the basement membrane that separates the dermal from the epidermal compartments. This double staining clearly demonstrated that the new group of Runx3-expressing cells was indeed located in the epidermal part of the bulb outside the DP. Because these cells were mostly found over the distal part of the DP, where melanocytes are known to reside during anagen (Stenn and Paus, 2001), we performed double stainings using antibodies against Runx3 and against MITF, an early marker of melanocytes (Tachibana, 2000). As shown in Figure 4C, the MITF/Runx3 double-positive cells are distinct from the main Runx3 DP cells, which suggests that these cells are indeed hair follicle-specific melanocytes.

Figure 4.

Epidermal expression of Runx3 in the hair bulb. A: Runx3-positive cells in the hair bulb (arrowheads). B: Images of Runx3 costained with β4-integrin, which marks the basement membrane (BM) at the dermal–epidermal border. Runx3-positive cells appear around the dermal papilla (DP, bottom) in a cross-section of the hair bulb and at the upper matrix cells (arrowhead, top) in a sagittal section. C: Colocalization of MITF and Runx3 (in yellow) in the epidermal part of the bulb, outside of the DP. Scale bars = 50 μm.

Next, we asked whether Runx3 was also expressed in other skin appendages. Sagittal toe sections revealed intense Runx3 staining in the mesenchymal compartment of the nail (Fig. 5A), which corresponds to the DP of the hair follicle. When we examined the distal ventral (pad) part of the toe, expression was also found in the eccrine sweat glands (Fig. 5B). The sweat gland is composed of a tube, which forms loops in the dermis, spirals in the epidermis, and eventually connects in a straight segment into the skin surface. Curiously, intensely stained cells were found in the innermost (luminal) cells of the outflow tube (Fig. 5C), whereas cells of the distal loops were weakly stained.

Figure 5.

Expression of Runx3 in other skin appendages. A: Runx3-positive cells in the nail mesenchyme (arrow). Also stained is the growth plate (GP) of the distal phalanx bone. B: In eccrine sweat glands of the toe pad, Runx3 is highly expressed in the inner cells of the outflow tubes (OT) but weakly expressed in the distal loops (DL) region of the gland. the ventral side of the toe is at the bottom. C: Confocal image of Runx3 co-stained with β4-integrin, demonstrating the epidermal location of the sweat gland-positive cells (arrowheads). The arrow points to one of the Runx3-positive cells in the skin epidermis. Scale bars = 100 μm in A, 50 μm in B,C.

To gain insight as to possible functions of Runx3 in skin, Runx3 null mice were compared with wild-type (WT) mice using histology, analysis of hair follicle markers, and epidermal differentiation markers. No major change was detected in the overall morphology of the skin and its appendages, which may be due to compensation by Runx2 that is also expressed in the DP (Raveh and Gat, unpublished results). We did notice, however, that the pelage of the Runx3 null mice looked different from that of their littermates in that it appeared less dense and the outer guard hair seemed more prominent. This finding prompted us to characterize the hair type profile of the mutant mice.

Hair plucked from WT and Runx3 null mice were compared and a change was found in the relative abundance of the different types: the zigzag type, which are the most abundant hair and which form the pelage under fur, constituted around 70% of the hair counted in WT mice, whereas in the Runx3 null mice their proportion was reduced to 55% (Fig. 6A), and the auchene hair type was seldom found at all. Microscopic examination revealed changes in the shape of the bent hair types, in that the mutant mice manifested much less-prominent bends in the auchenes and zigzags (Fig. 6B). Importantly, the mutant zigzags had only two bends in contrast to the three to five bends in the normal fur (Fig. 6B). Hair length analysis (Fig. 6C) showed that, whereas straight hair types in the mutant mice were not significantly shorter, the length of the zigzag hairs was significantly reduced (by approximately 25%), demonstrating again that the later hair types are the most affected in the Runx3 knockout (KO) pelage. To examine if changes in the durations of the hair cycle stages may be responsible for the changes observed in the hair length of the mutant mice, we have analyzed skin sections from mutant and WT littermates at different ages representing the first two hair cycles and have found no obvious differences (not shown). Thus, we conclude that the reason for the difference in the hair length is probably in the intrinsic hair rod growth rate per time period, which is slower in the mutant mice.

Figure 6.

Analysis of hair type and structure. A: Reduction in the proportion of the zigzag hair type in the Runx3 null mice compared with the ratio in wild-type hair (Student's paired t-test probability < 0.01, n = 3). B: Changes in the shape of the bent hair types in the Runx3 null mouse. Note the lower number of bends and also the less-prominent constrictions in auchene (au) and zigzag (zz) hair types of the null mouse. C: A representative comparison of awl and zigzag hair length of both wild-type (WT) and Runx3 null mice shows a significant reduction in zigzag hair length in the Runx3 null mice. D: Presentation of the ratio between minimal diameters (“waist” regions of hair) mean to the maximal diameters (hair “hip” regions) mean in wild-type vs. Runx3 null mice. High values, as in the null mice, indicate a less-flexible structure due to less-pronounced constrictions. C,D: In both images, values are indicated as mean ± standard deviation. E: Microscopic images of wild-type and null zigzag hairs. Note the bright segments in hairs of the knockout (KO) mice. Scale bar = 50 μm.

Comparison of the WT vs. mutant zigzag hair demonstrated that the WT bend regions appear significantly thinner than those of the mutant regions (Fig. 6D). This change in morphology resulted in a less “flexible” movement of the mutant hair when manipulated under the dissecting microscope. A prominent hallmark of the structural hair is the medulla-originated hair cells, which occupy most of the hair rod volume. The mutant zigzag hair exhibited large regions in which the air cells seemed more refractive to light, resulting in the abundance of bright segments compared with wild-type hair (Fig. 6E). As both pigmented and albino strains of mutant mice showed these changes, they are unlikely to be caused by changes in pigmentation, but rather by an alteration in medulla air cell differentiation and septation. Thus, a change in the mesenchymal compartment of the hair follicle resulted in alteration of the structural output, the hair rod, produced by the adjacent epithelial compartment.


The DP of the hair follicle remains the more enigmatic part of the now well-characterized complex mini-organ. Using immunohistochemistry (IHC) and X-gal staining, we have found previously that Runx3 and Runx1 are expressed in the DP and epidermal compartments of the whiskers, respectively, as in several other organs arising from mesenchymal–epithelial interactions (Levanon et al., 2001). The current study describes the expression of Runx3 in skin and skin appendages and delineates the consequences of its ablation in mice.

In the developing mouse embryo, Runx3 is uniformly expressed in the dermis below the primitive epidermis before initiation of hair morphogenesis. This finding is reminiscent of several other key factors regulating hair follicle development, such as Lef-1 and EDAR (Zhou et al., 1995; Laurikkala et al., 2002), which are uniformly expressed throughout the early skin or in one of its compartments. Subsequently, when hair germs are formed, Runx3-expressing cells are found below the epidermal placode, where they form the pre-DP. The change from uniform to localized expression may be due either to the loss of Runx3 expression in the cells that are not proximal to the placodes or to convergence of the interplacode-positive cells to the placodes and their replacement by nonexpressing cells. Of interest, asymmetrical location of these Runx3-expressing cells is observed at this stage, which could predict the future angling of the hair follicle relative to the skin surface. At later stages of anagen, Runx3 assumes a constant DP expression pattern, as well as DS expression, which is also typical of other DP markers such as Noggin (Botchkarev et al., 1999). The DP expression persists also in the regression (catagen) and rest (telogen) stages of the hair cycle, similar to alkaline phosphatase (Handjiski et al., 1994).

The toes of the mouse contain regions of specialized skin, which harbor the unique nail and eccrine sweat glands appendages (Byrne et al., 2003). As Runx expression in these structures had not been explored previously, we performed an immunohistochemical analysis of this area. We were able to show that, in the nail, Runx3 is expressed in the mesenchymal portion, in parallel to its expression in the DP of the hair follicle. On the other hand, in the eccrine sweat glands, Runx3 is expressed in the inner part of the straight segment of the tube in its more surface-proximal part. Our preliminary results show that Runx1 is also expressed in a complimentary manner to Runx3, i.e., in the epithelial matrix cells of the nail and in the distal loops of the eccrine glands (data not shown). This finding is in agreement with our previous observations of distinct compartmental expression of these two members in the same tissue (Levanon et al., 2001).

Immunohistochemical analysis also detected isolated Runx3-positive cells in the epidermis. These cells first appear at birth, after which their density increases. Because our previous work has shown that Runx3 is mostly confined to the mesenchymal compartment of organs (Levanon et al., 2001), and because most of these epidermal cells did not express cytokeratins in addition to Runx3 (data not shown), we concluded that they are not epidermal but must have migrated into this compartment during development. T-cells bearing the γδ receptor type are known to colonize the epidermis from birth and on, whereas Langerhans dendritic cells also appear from day 3 and on (Paus et al., 1998). Thus, either cell type may represent the Runx3-positive population, as both of these immune cells have been found to express Runx3 (Fainaru et al., 2004). In addition, supporting evidence for the identity of these epidermal cells came from studies, in which epidermal sheets prepared by limited trypsinization of ear skin from mice, were subjected to staining with MHC class II–specific antibodies. This staining revealed a distribution pattern of MHC II–bearing dendritic cells in WT but not in Runx3 KO mice (Fainaru et al., 2004), which is similar to that we observe for Runx3-positive cells in epidermis.

As it became evident that, in advanced stages of anagen, Runx3 expression was also found in cells populating the matrix/pre-cortex compartment, we were interested to explore the identity of this population of cells. The time of appearance and location of these cells suggested that they might represent melanocytes. This assumption was verified by the Runx/MITF colocalization experiments. MITF is expressed in several cell types, many of which are pigmented cells, whereas after birth in the mouse, it is found only in the melanocytes of the hair follicles (see Tachibana, 2000, and references therein), and thus can then serve as a marker for melanocytes. These data are the first demonstration of a Runx family member, which is expressed in melanocytes, and although we have not yet observed a melanocytes associated phenotype in these mice, it is possible that Runx3 together with another Runx may play a role in melanocytes differentiation and/or migration, with obvious potential medical importance.

Unlike Runx1 null mice, which do not survive past day 12 of embryonic development, and Runx2 KO mice, which die at birth, Runx3 null mice bred on heterogeneous genetic background (ICR, MF1) reach adult life and even produce progeny (Levanon et al., 2002). To assess phenotypic effects caused by loss of Runx3, we conducted a thorough comparison between WT and Runx3 KO skin sections using histology, staining with antibodies against hair follicle and against epidermis differentiation markers. As expected, Runx3 null skin did not show any Runx3 staining in all the above cells, and displayed an overall normal morphology (not shown). In light of the major phenotypes observed in mice null for Runx1 and Runx2, this lack of an overt phenotype in Runx3 KO could be explained by compensation for Runx3 function by Runx2, which is also expressed in the DP. Inspection of the fur coat of Runx3 mutant mice, however, did reveal that, despite the possible redundancy of Runx function, a change in hair type composition, as well as in the intrinsic shape of the structural hair, did occur. Curiously, the change mostly affected the auchene and the zigzag hair, which are the latest hair types to develop (Mann, 1962).

A possible explanation of the Runx3 hair coat phenotype is that the Runx family members are differentially regulated in the DP of the unaffected tylotrich and awl hair follicles, compared with the later affected auchene and zigzag hair follicles. Thus, if hypothetically speaking, only the late hair follicles were found to express Runx3, it would have been evident that only they were affected in the mutant. We have performed antibody stainings for Runx3 (Fig. 1D) and Runx2 (not shown) in early postnatal skin, and we can detect their expression in the primary as well as in the latest hair follicles to form in the third wave of morphogenesis. As a result, we cannot link a certain Runx member to a specific hair type. Alternatively, the differential effect may be due to different partners/effectors of Runx3 in the different hair types, or due to alternative splicing of Runx transcripts (Levanon and Groner, 2004), resulting in differential activity in the hair types.

To find candidate genes that may be interacting with Runx3 or belong to the same pathway, we have searched available reviews and annotations of mutations that affect hair follicle structure and development (Trigg, 1972; Sundberg, 1994; Nakamura et al., 2001). There are numerous natural mutations, which affect hair shaft shape and structure, but none result in the same phenotype that we have observed in the Runx3 KO, and none map to the same chromosomal loci of Runx3 or any other Runx (Nakamura et al., 2001).

We did find, however, a possibly relevant mutation, Ragged (Ra), which affects hair type distribution and hair shaft structure/shape (Carter and Phillips, 1954), and has several features in common with the Runx3 mutation that we have presented here. The Ragged heterozygous phenotype includes a reduction in coat density, a lack of the auchene and zigzag hair, and irregularities in air spaces and in septation of the zigzag medulla, whereas the homozygous mice hardly harbor any hair or whiskers (Carter and Phillips, 1954; Slee, 1957). The gene affected in this mutation was found to be Sox18 (Pennisi et al., 2000b), belonging to a family of HMG box containing genes coding for transcription factors taking part in multiple developmental processes in different organs (Wilson and Koopman, 2002). The Sox18 knockout mouse displays a much milder phenotype than the natural mutations, which are trans dominant in nature, but still has a lower ratio of zigzags in its coat (Pennisi et al., 2000a). In addition to the similarity in phenotype between the Runx3 mutant mice and the Ragged/Sox18 mutant mice, like Runx3, Sox18 is expressed in the embryonic mesenchymal part of the hair follicle. However, unlike Runx3, the expression of Sox18 is transient, lasting only until formation of the DP. Considering the above-described similarities between the phenotypes, and because Sox9 was found to be induced by the Wnt/β-catenin pathway (Blache et al., 2004), whereas Sox 17 was shown to directly bind β-catenin and form together a bipartite transcriptional activator (Sinner et al., 2004), it would be interesting to: (1) investigate whether Sox18 and Runx3 interact by determining if Runx3 expression is affected in the Ragged/Sox18 mutants and vice versa, and (2) explore the possible role these factors play in the function of the Wnt pathway in the DP.

The Runx family has been found to be induced by the transforming growth factor-β/bone morphogenetic protein (BMP) pathway (Fainaru et al., 2004; Levanon and Groner, 2004; Miyazono et al., 2004) and to physically interact with the receptor type Smad transcription factors that transduce these pathways signals (Hanai et al., 1999). Like the aforementioned Wnt pathway, the BMP pathway is critical for hair follicle morphogenesis, growth, and differentiation (Blessing et al., 1993; Botchkarev, 2003; Botchkarev and Sharov, 2004). It was found that, when an inhibitor of BMPs, Noggin, is ablated in mice, secondary hair follicles do not form in skin, whereas the primary follicles develop but arrest at a specific stage (Botchkarev et al., 1999, 2002). At later stages of follicle development, the activity of the BMP pathway was shown to be necessary for maturation of the matrix progenitors and their proper differentiation into the hair shaft and IRS cell lineages (Kulessa et al., 2000; Kobielak et al., 2003; Andl et al., 2004). In addition, the Homeobox Msx2 gene, which is likely induced by BMPs of the epithelial compartment, was found to be responsible for hair shaft differentiation by activation of the Foxn1 (nude), Hoxc13, and perhaps also Lef-1, all of which are hair keratin activators (Ma et al., 2003). Most interestingly, the Msx2 knockout mice (Ma et al., 2003) were shown to have hair shaft structure disorders, including medullae that have irregular and disorganized septations and air bubbles, which are reminiscent to those we show in the Runx3 KO zigzag hairs. It is possible, therefore, that DP signaling by Runx3 may feed into the function of the BMP/Msx2 differentiation plan, an avenue that we plan to explore.

In conclusion, we have shown that a gene expressed in the DP can influence the output of the neighboring epithelial compartment—the matrix and hair shaft that produce the structural hair. It is plausible, thus, that Runx3 together with yet undefined regulators elicit mesenchymal–epithelial interactions, which modify the essential developmental pathways collaborating to form the hair follicle and to direct hair synthesis in it.


Mouse Strains

The Hsd:ICR (CD-1) and MF1 mouse strains were used for most studies. Runx3-null mice were generated as described (Levanon et al., 2002) and bred on ICR and MF1 background.

Immunological Procedures

For immunostaining, 8-μm cryostat sections were fixed in paraformaldehyde (4% in phosphate buffered saline [PBS], 10 min), treated with 1.5% hydrogen peroxide for blocking endogenous peroxidase activity where needed, and incubated for 30 min with blocking solution containing 0.05% Triton X-100, 2.5 % normal goat serum [NGS], 1% bovine serum albumin, and 0.25% glycine in PBS before incubation with primary antibodies overnight for IHC or 1 hr for immunofluorescence (IF). When staining with mouse monoclonal antibodies, sections were incubated for 30 min with mouse IgG blocking reagent from the MOM kit (Vector Labs). The following primary antibodies were used at the indicated concentrations: anti-Runx3 (rabbit, 1:2,000 IHC, 1:1,000 IF; Le et al., 1999), anti-MITF (mouse, Zymed; ready-to-use solution), and anti-β4 integrin (rat, 1:4,000; BD Bioscience), diluted in 50% blocking solution. Biotinylated anti-rabbit secondary antibodies were applied diluted in 1.5% NGS PBS for 1 hr and detected by the avidin–biotin peroxidase technique (ABC, Vectastain, Vector Laboratories, Burlingame, CA), using diaminobenzidine reagent (soluble tablets, Sigma). Alternatively, the relevant fluorescein isothiocyanate, red-rhodamine, or Texas Red-conjugated donkey or goat antibodies (1:200–1:300; Jackson Laboratories) were used for detection of primary antibodies. All staining procedures were performed at room temperature. Antibody-stained sections were observed and photographed using an Olympus BX51 microscope and a Magnafire SP digital camera. Confocal images were taken on a Bio-Rad MRC-1024 confocal microscope.

X-gal Staining

Sections were fixed for 30 sec in 0.1% glutaraldehyde in PBS, washed five times in PBS, and incubated in X-gal staining solution at room temperature overnight.

Hair Analysis

Hair from three pairs of 9- to 12-week-old control ICR or MF1 and Runx3-null mice were plucked from the dorsal part of skin and examined under a binocular (Olympus, SZXILLB200). All pelage hair types from back skin were visualized using a microscope (Olympus, BX51) under ×200 magnification, photographed, and measured. Hair type counts were performed twice for each pair, 250–300 hairs counted in each of the repeats, and the mean and standard deviation were calculated. For hair length analysis, samples of awl and zigzag hairs obtained from pairs of WT and Runx3-null mice dorsum region were counted and the mean lengths and standard deviation were calculated. For hair shape analysis, samples of Zigzag hairs from pairs of WT and Runx3-null mice back skin were taken and diameters of wide and narrow parts of 25 hair rods were measured. Mean diameters of the two different hair rod regions were calculated, as well as the proportion between the two means. This procedure was repeated for three mice pairs.


We thank all members of the Gat and Groner laboratories for their help and support. Thanks also to Naomi Melamed-Book for her help with confocal microscopy and to Dr. Allan Bar-Sinai for critical reading of the manuscript.