SEARCH

SEARCH BY CITATION

Keywords:

  • homeodomain;
  • homeobox;
  • HOX;
  • stem cell;
  • keratinocyte;
  • proliferation;
  • skin;
  • epidermis

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

The HOX homeodomain proteins are fundamental regulators of organ and tissue development, where they are thought to function as transcription factors, and HOX gene expression has been associated with numerous types of cancers. Previous studies have demonstrated that enforced expression of the HOXB4 protein transforms cultured fibroblasts and leads to a selective expansion of the hematopoietic stem cell pool, suggesting that this protein might play a role in cellular proliferation. In support of this concept, we now show that enforced expression of HOXB4 in human neonatal keratinocytes results in increased cellular proliferation and colony formation as well as decreased expression of the alpha-2-integrin and CD44 cell surface adhesion molecules. We previously have reported HOXB4 gene expression in the basal and suprabasal layers of developing human skin and now show extensive HOXB4 mRNA in psoriatic skin and basal cell carcinoma. In fetal human skin HOXB4 protein expression was both nuclear and cytoplasmic within epidermal basal cells and in hair follicle inner and outer root sheath cells, whereas strong nuclear signals were observed in the bulge region. In adult skin, HOXB4 protein expression was both nuclear and cytoplasmic, but was predominantly localized to the intermediate and differentiated cell layers. In contrast to the striking gradient patterns of HOX gene and protein expression previously described in developing spinal cord and limb, HOXB4 protein was uniformly detected in all regions of the fetal and adult skin. Although little HOXB4 signal localized to proliferative cell layers, as marked by proliferating cell nuclear antigen (PCNA) staining, in normal adult epidermis, nuclear HOXB4 protein expression substantially overlapped with PCNA-positive cell in a series of samples of hyperproliferative skin. Taken together, these data suggest that nuclear HOXB4 protein may play a role in the regulation of cellular proliferation/adhesion in developing fetal human epidermis and in hyperproliferation conditions, including cancers, in adult epidermis. Published 2002 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

There is growing evidence that the developmentally important HOX homeodomain proteins influence cell proliferation (Zakany and Duboule, 1999; Krosl and Sauvageau, 2000). Much of this evidence comes from surveys that demonstrate increased expression of HOX genes and/or their regulators in a broad range of solid tumors (Celetti et al., 1993; Cillo, 1994) and leukemias (Look, 1997). In addition, overexpression of several HOX genes leads to either expansion of particular hematopoietic compartments (Sauvageau et al., 1997; Crooks et al., 1999) or to overt cellular transformation (Perkins et al., 1990; Thorsteinsdottir et al., 1997; Kroon et al., 1998). Several previous studies showed that enforced expression of the HOXB4 protein resulted in hyperproliferation. HOXB4 overexpression in bone marrow cells led to proliferation of pluripotent progenitors with properties ascribed to hematopoietic stem cells (Sauvageau et al., 1995), whereas overexpression in embryonic stem cells resulted in hyperproliferation of the red cell progenitor pool (Helgason et al., 1996). In addition, enforced expression of HOXB4 led to hyperproliferation and was transforming in fibroblasts, and this phenotype was dependent on coexpression of the PBX homeodomain protein (Krosl et al., 1998). Overexpression of the HOXA7 gene was reported to inhibit keratinocyte differentiation (LaCelle and Polakowska, 2001). It should be noted that HOXB4 gene expression was detected in both papillomas and carcinomas during an reverse transcription-polymerase chain reaction survey (Chang et al., 1998).

Although the mechanism of action of HOX proteins remains unknown, there is substantial circumstantial evidence that these proteins function as transcription factors. HOXB4 and other HOX proteins alone exhibit weak, relatively nonspecific DNA binding (Pellerin et al., 1994; Shen et al., 1996). They gain both binding avidity and specificity by forming heterodimeric DNA binding complexes with PBX proteins (Chan et al., 1994; Chang et al., 1995; Phelan et al., 1995; Lu and Kamps, 1996). The PBX1 gene was first identified as an oncogenic fusion between the PBX homeodomain and the activation domain of the E2A gene product (Kamps et al., 1990; Nourse et al., 1990). Currently however, the regulatory regions of the few downstream targets for HOX proteins, including targets such as p21 (Bromleigh and Freedman, 2000), p53 (Raman et al., 2000a), the progesterone receptor (Raman et al., 2000b), or FGF (Care et al., 1996), do not appear to contain PBX-HOX binding sites. Another level of regulation appears to be the finding that localization of the putative cofactor PBX/EXD proteins to the nucleus is dependent on complex formation with MEIS/HTH homeodomain proteins (Abu-Shaar et al., 1999; Berthelsen et al., 1999). Although few studies have examined HOX protein subcellular localization, we have reported previously that both the HOXB6 protein and MSX2, a non-HOX homeodomain protein, undergo a cytoplasmic to nuclear transition during skin development (Stelnicki et al., 1997; Kömüves et al., 2001). Despite the advances in defining binding partners and some data on subcellular localization, the mechanism of action by which HOX proteins exert their powerful developmental and/or hyperproliferative effects remain largely unknown. Although most studies suggest that HOX proteins directly regulate gene transcription (reviewed in Pradel and White, 1998), we have recently presented an alternative model in which HOX proteins repress transcription by inhibiting CBP histone acetyltransferase activity (Shen et al., 2001).

To gain understanding as to how HOX proteins function, we and others have attempted to characterize their tissue expression patterns and the effects of overexpression and/or homologous deletion phenotypes. We initially reported the detection of Hoxb4 during a survey of Hox gene expression in developing murine skin by using RNase protection analysis (Detmer et al., 1993). In situ hybridization was used to more precisely localize HOXB4 gene expression within developing human skin (Stelnicki et al., 1998). In these studies, the HOXB4 gene appeared to be expressed within the entire developing epidermis, including the hair follicles, as well as transiently in dermal fibroblasts in mid-gestation embryonic skin. The current study was designed to determine both the localization of HOXB4 protein within the layers of the skin, as well as the subcellular localization of the protein. Nuclear HOXB4 protein expression was detected within the basal cell layer of developing epidermis and within the bulge region of the hair follicle. Enforced HOXB4 expression in cultured keratinocytes led to cellular changes previously associated with the transit amplifying keratinocyte pool (Jones and Watt, 1993). HOXB4 mRNA and protein are strongly expressed in the proliferating cell nuclear antigen (PCNA) -positive cell layers of a range of tissue samples from hyperproliferative skin condition, including squamous and basal cell cancers. These data suggest that the HOXB4 protein is a candidate for regulating keratinocyte hyperproliferative in skin diseases.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

HOXB4 Protein Expression Is Prevalent in the Basal Cell Layer in Developing Epidermis

Relatively weak HOXB4 signal was observed throughout the developing epidermis in 10-week gestation fetal scalp (Fig. 1A). At this time, the epidermis is only two to three cells thick and does not contain visible hair follicles. By 16–17 weeks gestation, when follicles are first developing, HOXB4 expression was much stronger and appeared relatively localized to the basal cell layer (Fig. 1B,G,H). There was also strong expression within the inner root sheath of some, but not all, of the developing hair follicles. No signal was detected in the dermal papillae. By 21 weeks gestation, strong HOXB4 expression is detected throughout the developing epidermis as well as in dermal fibroblasts (Fig. 1C), with expression in the majority of developing hair follicles (Fig. 1D, and E). Within the hair follicle, weak signal was detected in the precortex area. Cells in the outer as well as the inner root sheaths stained for HOXB4. Expression was especially prominent within the bulge region. Staining in the hair bulb was absent.

thumbnail image

Figure 1. HOXB4 protein is expressed in the basal cell layer and the hair follicle of the developing epidermis, and in suprabasal layers of adult epidermis. Affinity purified antibodies were used to localize HOXB4 protein in samples of developing fetal or adult human skin. Images in A–F were obtained by using an amplified peroxidase detection and counterstained with Methyl Green to visualize skin morphology, whereas those in G–M were obtained by using an amplified alkaline phosphatase protocol and counterstained with eosin. A,10-week; B, 17-week; C–E, 21-week fetal scalp skin, treated with HOXB4 antibodies. Control panel F (21-week scalp) was treated identically without primary antibody. G, 16-week arm; H, 16-week back; I, adult abdomen. J, adult leg; K, adult shoulder; and L, adult scalp were treated with HOXB4 antibodies, whereas M (adult scalp) was a control tissue that was treated identically without primary antibody. With the exception of L, the inserts are 3× digitally magnified images of the respective boxed areas. Strong HOXB4 signal (brown in A–E and blue in G and H) was detected in the basal cell layer of 16, 17, and 21 week skin (B, C, G, and H). HOXB4 signal was detected in the outer and inner root sheath and was markedly nuclear in the hair follicle bulge region (C–E, the brackets denote the bulge regions). Note the melanin granules in the bulb region represent endogenous pigmentation. N–P: Confocal microscopy images used to confirm nuclear localization of the HOXB4 signals in the hair follicle and bulge region of a 21-week fetal skin sample costained with antibodies to HOXB4 (red signals in N), Sytox Green to detect nuclei (green signals in O), and the electronically merged images (P). Note strong nuclear signals throughout the inner root sheath of the developing hair follicle, which are prevalent in the bulge region of the follicle (P, arrows). The HOXB4 expression pattern in adult skin is complex, with intense, predominantly cytoplasmic expression in the most suprabasal layers (I–M), with weaker, predominantly nuclear signals in middle layers of the epidermis, and essentially no signal detected in the basal cell layer. Essentially identical patterns of HOXB4 expression were detected in samples obtained from different individuals representing diverse regions of the body surface. To test whether HOXB4 expression colocalizes with proliferating cells, proliferating cell nuclear antigen (PCNA) -positive cells were identified by immunostaining. The insert in L shows a tissue section cut in parallel to that shown for HOXB4 expression in L, stained for PCNA, and shown at 2× reduced magnification. The majority of the HOXB4 signal is not colocalized with PCNA-positive proliferating cells, but residual nuclear HOXB4 signal overlaps the upper PCNA-positive cell layers. Scale bars = 120 μm in D (applies to D,E), 240 μm in M (applies to A–C,F–M), 62.5 μm in P (applies to N–P).

Download figure to PowerPoint

HOXB4 Protein Expression Is Predominantly Suprabasal in Adult Epidermis

There are substantial differences in the distribution of HOXB4 protein across the layers of the epidermis, in normal adult skin compared with developing epidermis. In contrast to fetal epidermis, a gradient of increased expression from an essentially negative basal cell layer to intense staining in the suprabasal layers was observed in adult epidermis (Fig. 1I–L; Table 1). Tissues were stained for PCNA to mark the proliferating cell layers (Fig. 1L, insert). In normal adult skin, the majority of the HOXB4 signal is outside of the PCNA positive cell layers, with only occasional HOXB4-positive cells noted in within the upper layers of PCNA-positive cells.

Table 1. HOXB4 Protein Expression Patterns in Normal and Hyperproliferative Epidermisa
SampleNBasal cell layerIntermediate cell layersUpper cell layers
NuclearCytoplasmicNuclearCytoplasmicNuclearCytoplasmic
  • a

    Tissue sections were all stained under identical conditions and were arbitrarily scored on a scale of −/+ < +/− < +1 through ++++, − indicates no signal was detected. BCC, basal cell carcinoma; SCC, squamous cell carcinoma; DF, dermatofibroma; Cav. Hem., cavernous hemangioma; KS, Karposi's sarcoma; SK, sebaceous keratoma.

  • b

    Samples from multiple sites and individuals were pooled for tabulation.

10-week fetal1+−/+
16-week fetal4b+++−/+−/+−/+
18-week fetal1++++−/++
21-week fetal4b++++++−/+++
Normal 4 mo1−/++++/+++/++
Normal adult10b−/++++++++/+++
BCC2++++−/++/−++
SCC2−/+++++++++
DF1+/−++++++++
Eczema2++/−+++
Psoriasis1+++/−++
Cav. Hem.1+/−++
SK.1++−/++/++−/+
KS.1−/++/+−/+++

HOXB4 Protein Is Both Nuclear and Cytoplasmic in Developing Epidermis and Hair Follicles and in Adult Epidermis

A complex pattern of subcellular localization of HOXB4 protein was observed in developing epidermis (summarized in Table 1). In very immature fetal scalp epidermis, essentially all the HOXB4 signal was cytoplasmic (Fig. 1A). However, by 16 weeks gestation, the majority of HOXB4 protein in fetal arm and back was nuclear (Fig. 1G,H), but the expression in 17-week scalp remained predominantly cytoplasmic (Fig. 1B). Within 21-week fetal scalp, there was a both cytoplasmic and nuclear HOXB4 expression within the epidermis (Fig. 1C) and hair follicle shaft (Fig. 1D,E), but light microscopic analysis suggested that expression in the hair follicle bulge region was mainly nuclear. Fluorescence colocalization of HOXB4 and Sytox Green, a nuclear stain, confirmed predominant nuclear HOXB4 protein expression within the developing hair follicle bulge region (Fig. 1N–P). Compared with 16- to 17-week skin, 21-week fetal skin exhibited more cytoplasmic HOXB4 signal in the upper epidermis (compare Fig. 1B,G, or H with 1C). A larger fraction of the HOXB4 signal appears to be cytoplasmic in adult skin, with particularly strong cytoplasmic expression in the most suprabasal epidermal layers (Fig. 1I–L). Although substantial nuclear HOXB4 expression was detected in the intermediate cell layers, only trace nuclear signals were detected in the basal cell layer (Fig. 1I–K, inserts) (summarized in Table 1), which exhibited predominant nuclear HOXB4 expression in early developing fetal skin. Nuclear HOXB4 expression was also detected in the inner root sheath cells of the adult hair follicle (data not shown).

HOXB4 Protein Expression Is Uniform Across the Body

During fetal skin development, HOXB4 protein expression levels appeared to be relatively constant between scalp, back, and arm samples from approximately similar gestation tissues (Fig. 1B,G,H). Examination of HOXB4 protein expression in adult skin from different body regions from different individuals revealed essentially similar expression profiles. Thus HOXB4 signal in the abdomen (Fig. 1I), leg (Fig. 1J), shoulder (Fig. 1K), scalp (Fig. 1L), and arm and face (data not shown) were essentially identical both in location within the outer epidermis and distribution between the cytoplasm and nucleus (compare the lack of signal in the control Fig. 1M) (summarized in Table 1).

Enforced Expression of HOXB4 Is Associated With Increased Proliferation and Decreased Cell Adhesion

Since the basal cell layer, which exhibits strong HOXB4 protein expression, has been shown to contain the highly proliferative transit amplifying cells (Jones and Watt, 1993), we wished to explore the effects of enforced expression of HOXB4 in keratinocytes. A retroviral vector was used to stably transfect cultured human neonatal keratinocytes with a cDNA encoding the HOXB4 protein. Preliminary experiments demonstrated that G418-selected HOXB4-transfected cell pools were synthesizing high levels of HOXB4 mRNA (data not shown). The HOXB4 transfected cells exhibited a 3-fold increase in proliferation rate and an associated 10-fold increase in colony formation (Fig. 2A,B). In previous studies, the rapidly proliferating transit amplifying cell population was identified on the basis of decreased adhesion to collagen matrices, as reflected by decreased cell surface α2-integrin expression (Jones and Watt, 1993). When transfected keratinocytes were tested for adherence, there was a small but statistically significant decrease in the binding of the HOXB4-transfected cells to either plastic (Fig. 2C) or collagen coated plates (not shown). To explore the possible effects of HOXB4 expression on cellular adhesion proteins, specific antisera were used for analysis of cell surface adhesion markers. Fluorescence activated cell sorting (FACS) analysis showed that both α2-integrin and CD44 were decreased 2-fold in the HOXB4-transfected keratinocytes (Fig. 2D,E). However, there was no change in the capacity of the HOXB4 transfected cells to differentiate fully in response to high calcium, as measured by involucrin and transglutaminase expression detected by Western blotting (not shown). Taken together with the expression data for developing skin, these data suggest that expression and nuclear localization of the HOXB4 protein is associated with the regulation of fetal keratinocyte proliferation, self-renewal, and/or adhesion.

thumbnail image

Figure 2. Enforced HOXB4 expression increases keratinocyte proliferation and decreases adhesion and cell surface adhesion protein expression. Cultured keratinocytes were transfected with retrovirus vectors containing either a HOXB4 expression cassette or a control vector containing only the neomycin selection gene. Stable clones were selected and then assayed for proliferation rate (A), colony formation on plastic (B), adhesion to plastic (C), and surface expression of adhesion molecules alpha2-integrin (D) or CD44 (E). Single asterisks indicate P < 0.001; Double asterisks indicate P < 0.01.

Download figure to PowerPoint

HOXB4 Gene Is Expressed in Hyperproliferative Epidermis

Because our data suggested that the overexpression of the HOXB4 gene might be correlated with cellular proliferation and/or transformation, we examined HOXB4 mRNA levels in human hyperproliferative skin specimens by in situ hybridization. The intensity of HOXB4 gene expression did not appear to be up-regulated in hyperproliferative skin compared with normal skin, but the presumptive proliferating cells within psoriatic tissue or the tumor mass strongly expressed the HOXB4 message (Fig. 3).

thumbnail image

Figure 3. HOXB4 mRNA is expressed throughout the epidermis in normal and hyperproliferative skin. In situ hybridization analysis of HOXB4 gene expression (B,D,F) in normal adult human arm skin (top), psoriatic skin (middle), and a basal cell carcinoma (bottom). A,C,E: Bright field exposures of the same samples showing hematoxylin/eosin staining to reveal tissue morphology. An antisense probe to a portion of the HOXB4 coding region, flanking but not including the homeobox, was used to detect mRNA signals as described in Experimental Procedures section. A sense control probe showed no signal (not shown)._art;;1>

Download figure to PowerPoint

Nuclear HOXB4 Protein Expression Overlaps With PCNA-Positive Keratinocytes in Hyperproliferative Epidermis

To further explore the possible relationship between HOXB4 expression and keratinocyte proliferation, we next examined HOXB4 protein expression in a series of human samples representing a range of hyperproliferative skin conditions. Representative samples are presented in Figure 4, and data for all samples are summarized in Table 1. In nontransformed tissues, HOXB4 protein signals mirrored those detected in normal adult skin. Thus there was predominant cytoplasmic staining in the most suprabasal cell layers, as well as substantial nuclear signal in the intermediate epidermal cell layers of psoriatic and eczema samples (Fig. 4A,D). There was also pronounced nuclear HOXB4 protein in the hyperproliferative epidermis associated with dermatofibroma (Fig. 4M). Nuclear HOXB4 protein expression was extensive in the proliferative cell layers of squamous cell carcinoma (SCC) (Fig. 4G), whereas nuclear HOXB4 signal was localized to the outermost cell layer, with a diffuse cytoplasmic signal throughout the tumor mass of a basal cell carcinoma (BCC) (Fig. 4J). The HOXB4 signal in these two tumor types also appeared to mirror the general pattern observed for normal adult skin. HOXB4 protein expression was strongest in the SCC with a more differentiated phenotype, whereas lower intensity signal was detectable in the less differentiated BCC. In each of the tissues examined, strong nuclear HOXB4 protein expression (Fig. 4A,D,G,J,M) substantially overlapped the regions of cellular hyperproliferation, as reflected by positive PCNA staining (insert panels in Fig. 4B,E,H,K,N). However, for some samples (especially notable in the dermatofibroma sample shown in Fig. 4M), cytoplasmic HOXB4 expression was more extensive than PCNA staining.

thumbnail image

Figure 4. Nuclear HOXB4 protein expression overlaps the proliferating cell nuclear antigen (PCNA)-positive regions in hyperproliferative skin disorders. HOXB4 protein (A,D,G,J,M,, blue signals) and PCNA in parallel sections (B,E,H,K,N,, blue signals) were detected by immunohistochemical staining and compared with identically treated control parallel sections in which the primary antisera was omitted (C,F,I,L,O). A series of hyperproliferative skin disorders were examined, including psoriasis (A–C), eczema (D–F), squamous cell carcinoma (G–I), basal cell carcinoma (J–L), and the hyperproliferative epidermis associated with a dermatofibroma (M–O). For each set of three images, adjacent tissue sections were cut and processed, and the same region was photographed. The inserts represent 3× digital amplifications of the small boxed regions. Among the various conditions examined, a similar pattern of HOXB4 protein was detected in similar cell layers but, in the eczema tissue, the PCNA-positive cells included a basal cell population that did not express high levels of HOXB4 (D vs. E). With the exception of the basal cell carcinoma sample, moderate to intense cytoplasmic HOXB4 staining was observed in the most differentiated cells, with strong nuclear staining in the intermediate cell layers, and reduced or negligible staining in the basal cell layer. Most of the cells that showed nuclear HOXB4 staining were in layers containing PCNA-positive cells, especially for the basal cell carcinoma tissue (J and K). Scale bar = 240 microns in O (applies to A–O).

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

The HOX homeodomain proteins are widely ascribed to be transcription factors that function in the nucleus to regulate aspects of tissue-specific gene expression (Zappavigna et al., 1991; Mann, 1995). However, we have reported previously that, in the epidermis, both the HOXB6 (Kömüves et al., 2001) and the non-HOX homeodomain proteins MSX1 and MSX2 (Stelnicki et al., 1997) were either predominantly cytoplasmic in distribution or partitioned between the cytoplasm and nucleus. In contrast, we observed that the PBX proteins, which are putative DNA binding partners for HOXB4, are predominantly nuclear throughout fetal epidermal development (Kömüves et al., 2001). In the current study, much of the HOXB4 is nuclear in both the developing and adult epidermis, particularly in hyperproliferative conditions. Our finding of substantial nuclear HOXB4 expression is consistent with both biochemical data (Chang et al., 1995; Shen et al., 1996), gene transfection studies (Krosl et al., 1998), and transgenic animal studies (Gould et al., 1997; Popperl et al., 2000) that indicate that HOXB4 forms biologically active DNA binding complexes with PBX proteins.

Our data showing substantial cytoplasmic expression, in both fetal and adult epidermis, present several intriguing questions: (1) Are there biological roles for cytoplasmic homeodomain proteins? (2) What are the regulatory mechanisms that control the movement of HOX proteins from the cytoplasm to the nucleus? and (3) Do alternatively spliced forms of the HOX homeodomain proteins account for the observed differences in subcellular distribution? For HOXB4, several different sized mRNA transcripts have been reported but only a single protein coding sequence has been described (Peverali et al., 1990) and its relationship to the various transcripts is unknown. Questions concerning the possible roles of cytoplasmic HOX proteins or possible regulation of the movement of HOX proteins from the cytoplasm to the nucleus remain almost completely unanswered. A mechanism regulating the movement of the PBX homeodomain protein from the cytoplasm to the nucleus has been described (Abu-Shaar et al., 1999; Berthelsen et al., 1999). The nuclear localization of other transcription factors depends on phosphorylation events (Gauthier-Rouviere et al., 1995), and several HOX proteins have been reported to be phosphorylated (Jaffe et al., 1997; Berry and Gehring, 2000; LaCelle and Polakowska, 2001). However, no data exist regarding the mechanisms that regulate the movement of HOX proteins within the cell. Alternatively, we have demonstrated recently that HOXB4 and other HOX proteins bind to CBP/p300 and inhibit histone acetyltransferase activity (Shen et al., 2001). Thus it is possible that cytoplasmic HOX proteins may act to sequester CBP/p300 outside the nucleus, where they are thought to function to regulate gene expression.

Numerous HOX gene expression studies together with data from a broad array of transgenic and gene deletion models have led to the characterization of HOX proteins as “patterning” molecules (reviewed in Krumlauf, 1994). Previous studies suggested that HOX genes and their protein products may be expressed in the skin across the entire body surface (Godwin and Capecchi, 1998; Stelnicki et al., 1998), whereas others propose that HOX genes function in spatially restricted patterns to specify regional identity of the skin (Kanzler et al., 1994, 1997). In the current study epidermal HOXB4 protein expression was detected in similar patterns from almost all of the fetal and adult body regions examined. Choung et al. have proposed that Hox genes play multiple roles in feather and hair development within the skin (Chuong et al., 1993). These authors propose both a “long-range patterning” based on differential Hox gene expression across the body and a “short-range patterning” based on differential Hox gene expression within individual feather buds. Our data support a “short range” model for HOX protein function as patterning molecules within the developing skin, rather than as regulators of large-scale regional epidermal variation across the body. In this regard, it is interesting that HOXB4 protein expression was detected in many but not all developing hair follicles. The paralogous gene Hoxa4 was shown to be expressed in a partially overlapping but different pattern to Hoxb4 in murine hair follicles (Packer et al., 2000). Perhaps there is a “Hox code” for hair development. It should be noted that animals carrying homozygous deletions of either the Hoxb4 or Hoxa4 genes do not exhibit discernible hair or skin defects (Q-C. Yu, and C. Largman), but Hoxc13-deficient animals exhibit major defects in hair shaft formation (Godwin and Capecchi, 1998), and HOXC13 recently has been proposed to be a direct regulator of keratin gene expression (Jave-Suarez et al., 2001).

Our initial impetus for investigating the expression and role of HOXB4 in keratinocyte function was based on earlier studies (Sauvageau et al., 1995; Helgason et al., 1996; Krosl et al., 1998) that suggested that the HOXB4 protein might be a causative agent for cellular hyperproliferation in either developing fetal epidermis and/or adult hyperproliferative skin disorders. Adult epidermal keratinocytes are thought to undergo a series of changes from slowly cycling stem cells to proliferating transit amplifying cells within the lower epidermis to differentiated nonproliferative cells of the upper epidermis during the continuous process of maintenance of the epidermis (Watt, 1998). Our data support a possible role for HOXB4 in the regulation of fetal epidermal development. In support of this possible role, studies on mice in which a beta-galactosidase gene replaced the Hoxb4 gene showed β-gal enzymatic activity in the developing placodes in fetal mouse skin (Whiting et al., 1991). Other genes must regulate the differentiation of the epidermis in the adult skin, because HOXB4 does not appear to be expressed in the requisite basal cell patterns to fulfill this role. However, the expression of nuclear HOXB4 protein within the proliferating cell layers of both overt tumors as well as other hyperproliferative skin samples suggests that the HOXB4 protein might play a regulatory role in epidermal hyperproliferation in adult skin diseases.

EXPERIMENAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Tissue Samples

Tissue samples were obtained from 10-, 17-, and 21-week-old human fetuses and from adult skin samples or archival anatomic pathology specimens following Human Use Protocols approved by the University of California, San Francisco. Tissues were fixed in 4% formaldehyde in phosphate buffered saline (PBS) before paraffin embedding and sectioning.

Antibodies

Polyclonal anti-HOXB4 antisera were raised in rabbits by immunization with a 15 amino acid peptide (GGAAGSAGGPPGRPN) derived from the C-terminal region of the human HOXB4 protein sequence that showed minimal homology to the other HOX-4 paralog proteins, under a collaborative research program with the Berkeley Antibodies, Inc. (Richmond, CA) funded by NIH grant N44DK-3-2219. After affinity purification on immobilized GST-HOXB4, the antibody was shown to recognize bacterially expressed HOXB4 protein on Western blots (not shown). Biotinylated mouse monoclonal anti-PCNA antibody was from Caltag Laboratories (Burlingame, CA). Purified antisera against human CD44 and alpha2-integrin were kindly provided by Dr. Caroline Damsky.

Immunohistochemical Detection of HOXB4 and PCNA Proteins

Endogenous peroxidase or alkaline phosphatase activity was destroyed by treating tissue sections with 3% H2O2 or 0.05% levamisole, respectively (5 min). Tissues were blocked with 4% bovine serum albumin, 0.5% fish gelatin, 0.1% Tween 20 in TBS. Sections were incubated with anti-HOXB4 antibody (1/100), and developed by using donkey anti-rabbit immunoglobulin G (IgG) (Fab fragment) conjugated to peroxidase (Jackson Laboratories, West Grove, PA). Signals were amplified with biotinylated tyramide (DAKO, Carpinteria, CA) for 10 min, and detected with avidin-biotin complex (ABC)–horseradish peroxidase or ABC-alkaline phosphatase, by using diaminobenzidine (Vector Laboratories, Burlingame, CA) or BCIP/NBT (Roche Biochemicals, Indianapolis, IN) substrates, respectively. For immunofluorescent localization of HOXB4, similar procedures were used with treatment with 5% Na borohydride to quench nonspecific autofluorescence. Primary antibodies were detected with peroxidase-labeled donkey anti-rabbit IgG (Fab fragment) by using tyramide-Cy3 reagent (NEN, Boston, MA) for 10 min, for signal amplification. The sections were washed and counterstained with Sytox Green (Molecular Probes, Eugene, OR). As specificity controls for both the colorimetric and fluorescence immunodetection, preimmune serum in lieu of the first antibody or omission of the first antibody resulted in no signal. Biotinylated PCNA antibody staining was detected with ABC-AP followed by BCIP. Sections were counterstained with eosin or Methyl Green for morphologic evaluation.

RNA In Situ Hybridization

DNA templates for generating RNA in situ hybridization probes for human HOXB4 were constructed by subcloning a cDNA fragment, lacking the homeobox and designed to minimize cross-reaction with the other Hox-4 paralog genes, into Bluescript SK+ plasmid in the sense and antisense orientations (Stratagene, La Jolla, CA). Plasmids were linearized to produce sense and anti-sense templates, which were used to synthesize 35-S–labeled probes that were used for RNA in situ hybridization as previously described (Arbeit et al., 1994; Stelnicki et al., 1998).

Retroviral Infection of Keratinocytes

A HOXB4 cDNA encoding a full-length, homeodomain containing protein (Peverali et al., 1990) was subcloned into the MSCVneoEB retroviral vector such that expression of the cDNA is under control of the retroviral 5′ long terminal repeat (Sauvageau et al., 1995). Plasmid DNA was transfected by calcium phosphate precipitation into G+PE cells, and virus containing supernatants were collected 24 and 48 hours after transfection. For infection, 1 × 105 cultured second passage keratinocytes (K2P) were plated in 10-cm dishes and cultured for 1 day with virus-containing supernatants in the presence of 6 μg/ml of Polybrene (Sigma, St. Louis, MO). Cells were then washed with PBS, incubated for 1 day with Dulbecco's modified Eagle's medium supplemented 10% fetal calf serum, and selected for 7–21 days in KGM-SFM medium containing 0.07 mM calcium and 0.5 mg/ml of G-418 (GIBCO Life Technologies). Control infections used the parent retroviral vector containing the neomycin selection gene.

Proliferation Assay

Exponentially growing keratinocytes transfected with HOXB4 or NEO control plasmids were trypsinized and replated in KGM-SFM with 0.07 mM calcium. After a 3- to 5-hr attachment period, cells were washed three times with PBS, and maintained in KGM-SFM with 0.07 mM calcium for 24–72 hr. Cells were trypsinized, treated with trypan blue, and live cells counted by using a hemacytometer. All proliferation assays were performed in duplicate and were repeated by using independently infected K2P populations.

Colony Forming Assays

Stably selected K2P infected with either HOXB4 or NEO control plasmids were plated at a density of approximately 100 cells/100-mm dish and grown in KGM-SFM for 2 weeks, fixed in 5% glutaraldehyde in 1× Hanks' basic saline solution (HBSS), stained with 1% toluidine, and counted. Colonies greater than 36 cells were scored positive.

Cellular Adhesion

Transfected cells were prelabeled for 48 hr by growth in KGM-SFM containing [3H]thymidine. Trypsinized cells were plated onto either plastic- or collagen-coated flasks, and nonadherent cells were removed at each time point by washing with 1× HBSS. Adherent cells were removed by trypsinization, and [3H]thymidine counts were measured in a Beckman LS 6000SE liquid scintillation counter.

Cell Surface Protein and Cellular Differentiation Analysis

Cells were trypsinized, washed with PBS, incubated with specific fluorescein isothiocyanate–conjugated antisera to human alpha2-integrin or CD44 at 4°C, and subjected to FACS analysis by using a Bectin-Dickinson cell sorter. Assays were performed in duplicate from two independent infections. Data are expressed as means and standard deviations. Statistical significance was determined by using the Students t-test. Keratinocytes were grown in KGM-SFM medium containing 1.2 mM calcium to induce cellular differentiation (Pillai et al., 1991). Differentiation was measured by subjecting total cellular proteins to Western blotting analysis by using specific antibodies for human involucrin and transglutaminase (Pillai et al., 1991).

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We thank Robert G. Hawley for the MSCV vector and Caroline Damsky for antibodies to alpha2-integrin and CD44. C.L. received funding from the Department of Veterans Affairs and by the NIH. M. Morimune (A. Lincoln High School) was supported by UCSF Science and Health Education Partnership Summer Internship Program. The Microscopy and Molecular Histology Laboratory at the University of California VA Medical Center is supported by the NIH.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES