Abstract
Genetic alterations found in carcinomas can alter specific regulatory pathways and provide a selective growth advantage by activation of transforming oncogenes. A subset of these genes, including wild-type alleles of GLI or c-MYC, and activated alleles of RAS or β-catenin, exhibit transforming activity when expressed in diploid epithelial RK3E cells in vitro. By in vitro transformation of these cells, the zinc finger protein GKLF/KLF-4 was recently identified as a novel oncogene. Although GKLF is normally expressed in superficial, differentiating epithelial cells of the skin, oral mucosa, and gut, expression is consistently up-regulated in dysplastic epithelium and in squamous cell carcinoma of the oral cavity. In the current study, we used in situ hybridization, Northern blot analysis, and immunohistochemistry to detect GKLF at various stages of tumor progression in the breast, prostate, and colon. Overall, expression of GKLF mRNA was detected by in situ hybridization in 21 of 31 cases (68%) of carcinoma of the breast. Low-level expression of GKLF mRNA was observed in morphologically normal (uninvolved) breast epithelium adjacent to tumor cells. Increased expression was observed in neoplastic cells compared with adjacent uninvolved epithelium for 14 of 19 cases examined (74%). Ductal carcinoma in situ exhibited similar expression as invasive carcinoma, suggesting that GKLF is activated prior to invasion through the basement membrane. Expression as determined by Northern blot was increased in most breast tumor cell lines and in immortalized human mammary epithelial cells when these were compared with finite-life span human mammary epithelial cells. Alteration of GKLF expression was confirmed by the use of a novel monoclonal antibody that detected the protein in normal and neoplastic tissues in a distribution consistent with localization of the mRNA. In contrast to most breast tumors, expression of GKLF in tumor cells of colorectal or prostatic carcinomas was reduced or unaltered compared with normal epithelium. The results demonstrate that GKLF expression in epithelial compartments is altered in a tissue-type specific fashion during tumor progression, and suggest that increased expression of GKLF mRNA and protein may contribute to the malignant phenotype of breast tumors.
INTRODUCTION
Multiple physiological changes lead to acquisition of the malignant phenotype. These include self-sufficiency in growth signaling, insensitivity to growth-inhibitory signals, inhibition of apoptosis, immortalization, induction of angiogenesis, and the ability to invade and metastasize (1 , 2) . Genetic analyses of inherited predispositions to develop specific carcinomas enabled isolation of tumor suppressor genes important in both inherited and sporadic disease (3) , and subsequent functional studies identified certain of these genes as regulators of classical transforming oncogenes. Thus, alterations in the tumor suppressor patched (PTC1), or in other molecules that transduce the hedgehog signal, result in activation of GLI mRNA expression in virtually all basal cell carcinomas of the skin (4, 5, 6, 7) . In contrast, alterations in the adenomatous polyposis coli (APC) pathway activate the β-catenin/TCF-4 complex and transcription of c-MYC and cyclin D1 during colorectal tumor progression (8 , 9) . Modulation of the PTC1 and APC pathways by alteration of the mouse genome provides additional support of a role for these gene products in specific tumor types (10, 11, 12, 13, 14) . Therefore, these pathways exhibit properties of a gatekeeper, indicating that alteration of the pathway in a specific tissue is rate-limiting for tumor progression, and that alterations are found in a large proportion of inherited as well as sporadic tumors (15 , 16) .
By expression cloning we recently identified the zinc finger protein GKLF 3 as a novel transforming oncogene when expressed in RK3E cells, a diploid epithelial cell line derived from primary rat kidney cells and immortalized with adenovirus E1A (17) . cDNA libraries were prepared using mRNA from human oral squamous cell or breast carcinoma cell lines, tumor-types not reported to exhibit frequent genetic alterations that activate well-characterized oncogenes such as RAS, GLI, or β-catenin (18) . Retroviral transduction of these libraries into RK3E cells induced morphologically transformed foci, 11 of which were subsequently attributed to enforced expression of wild-type human c-MYC. Two other transformed foci contained independently derived, wild-type alleles of GKLF. No other genes were identified in the screen, suggesting that only a select subset of all oncogenes are able to transform these cells.
Whereas enforced expression of a human wild-type GKLF transgene in RK3E cells induces morphological transformation in vitro and tumorigenicity in athymic mice, the doubling time of GKLF-expressing cells was considerably longer than for RK3E cells (27 h versus 12 h, respectively; Refs. 17 ,, 19 ). Similar results were obtained for other oncogenes, including GLI and c-MYC, as cells expressing these genes exhibited doubling times of 18 and 19 h, respectively. These oncogenes may therefore function in epithelial cells by interfering specifically with contact inhibition rather than by inducing a more general increase in the rate of cell division.
In support of a role for GKLF as an oncogene, we observed increased expression of GKLF mRNA during progression of squamous cell carcinomas of the oropharynx. As demonstrated by mRNA ISH analysis of surgical specimens, expression in normal epithelial cells is limited to the differentiating compartment. Expression in dysplastic oral epithelium is increased overall and is found in all cell layers, and GKLF is expressed at similar levels in dysplasia and in invasive carcinoma. These results identified loss of the compartment-specific pattern of GKLF mRNA expression in epithelium as a candidate mechanism of tumor progression in oral cancer (17) .
GKLF encodes a DNA-binding transcription factor with functional domains that mediate activation or repression of transcription (20, 21, 22) . GKLF is essential for the barrier function of skin, because homozygous knockout mice exhibit morphologically normal skin but die postpartum due to dehydration (23) . Transcriptional targets of GKLF that may be relevant to epithelial differentiation have been preliminarily identified (23, 24, 25) . As shown by analysis of normal mouse or human tissues, GKLF is preferentially expressed in differentiating epithelial cells of the skin, gut, oral cavity, and thymus (17 , 20 , 23 , 26 , 27) . In contrast to human oral squamous cell carcinoma, expression of GKLF mRNA was found to be reduced in mouse models of intestinal tumorigenesis or hyperproliferation (28 , 29) . Independently, analysis of human colorectal mucosa and tumors by SAGE confirmed the earlier studies in mice (30 , 31) . Specifically, mRNA from specimens of normal colonic mucosa generated GKLF tags at frequencies of 138 or 99 per million SAGE tags, whereas mRNA from a microdissected tumor generated only 20 tags per million. These results suggested that GKLF expression is regulated during neoplastic progression in a tumor type-specific fashion.
Increased expression of specific oncogenes in tumors can result from genetic alterations and play a causal role in tumor progression. Alternatively, expression of some oncogenes is cell cycle dependent, and increased expression can occur as a consequence of increased proliferation or altered cell cycle occupancy of tumor cells. In multiple normal tissues as well as in certain cell lines, GKLF expression is reduced in actively cycling cells compared with terminally differentiated or growth-arrested cells, and enforced expression of GKLF in cultured cell lines can retard cell cycle progression (26 , 32) . These properties predicted that GKLF expression might be reduced in tumors, as observed in colorectal carcinoma. Increased expression in other tumor-types is therefore somewhat unexpected and may result from specific alterations in the pathways that regulate GKLF transcription in normal cells.
To better understand the spectrum of tumor-types that exhibit GKLF activation, we obtained samples of breast carcinoma, colorectal carcinoma, and prostatic carcinoma and analyzed expression of GKLF in malignant cells and in adjacent normal-appearing epithelium (i.e., uninvolved epithelium). The results show that levels of GKLF mRNA and protein are each up-regulated before invasion in a majority of cases of breast cancer, but not in tumors of the colorectum or prostate. In neoplastic lesions of the breast as well as in cultured mammary epithelial cells in vitro, increased GKLF expression appears to precede overtly malignant behavior. The potent transforming activity of GKLF in vitro, the tumor type-specific activation of expression in vivo, and activation early during tumor progression identify this oncogene as a potential effector of tumor progression in the breast.
MATERIALS AND METHODS
Tissue Procurement.
Fresh-frozen and paraffin-embedded samples were obtained through the Tissue Procurement Core Facility of the University of Alabama at Birmingham Comprehensive Cancer Center and through the Southern Division of the Cooperative Human Tissue Network.
mRNA Expression.
ISH was conducted as described (17) using sense and antisense [35S]-labeled riboprobes prepared by in vitro transcription of a cDNA fragment corresponding to the 3′ untranslated region of human GKLF. A GAPDH antisense probe corresponding to bases 366–680 (GenBank accession no. M33197) was synthesized using a commercially available template (Ambion, Inc., Austin, TX). High stringency washes were in 0.1 × SSC and 0.1% (v/v) 2-mercaptoethanol at 58°C for GKLF or 68°C for GAPDH. Slides were coated with emulsion and exposed for 14 days. The number of silver grains/nucleus were counted within representative areas by two individuals, and a score from 0.0 to 4.0 was recorded. A score of 0.0 indicated only nonspecific background, as determined using the sense control, and 1.0 corresponded to an average of four grains/nucleus.
Breast adenocarcinoma cell lines were obtained from the American Type Culture Collection (Manassus, MD). HMECs were described previously and were cultured in mammary epithelial basal media (Clonetics Corp., Walkersville, MD; Ref. 33 ). Extracts were prepared from exponentially growing cells at 70% confluence, and total RNA isolation and Northern blot analysis were performed as described (17) .
Isolation of an Anti-GKLF Monoclonal Antibody.
The region of the human GKLF cDNA encoding bases 479-1197 (GenBank accession no. AF105036) was cloned into plasmid pET-32a and expressed in Escherichia coli BL21(DE3) bacteria as a histidine-tagged protein. Protein was purified from the bacteria after induction with isopropyl-1-thio-β-d-galactopyranoside using a His-Trap Ni-agarose column (Amersham Pharmacia Biotech, Piscataway, NJ) and eluted with 500 mm imidazole. Purified protein was used to immunize two mice, and lymphocytes were fused with murine myeloma cells (PX63-Ag8.653) as described previously (34) . Hybridomas that were immunoreactive in an ELISA assay for the purified antigen were cloned and recloned by limiting dilution. Positive clones were identified by ELISA, and an IgG1 antibody (αGKLF) was purified from ascites on a protein A affinity column.
Immunohistochemistry.
Tissues were fixed in neutral buffered formalin and embedded in paraffin. Deparaffinized tissue sections were incubated with αGKLF at a concentration of 1.0 μg/ml for 1 h at room temperature, and processed as described (35) . Immunodetection was performed using a biotinylated secondary antibody, streptavidin-horseradish peroxidase detection system (Signet Laboratories, Dedham, MA), and the chromogenic substrate diaminobenzidine (Biogenex, San Ramon, CA). Sections were counterstained with hematoxylin. Results were scored by using a 0.0 to 4.0 scoring system where 4.0 corresponds to a saturated signal (36) .
Statistical Analyses.
Paired t tests were used to compare the differences in expression in breast epithelial cells at various stages of tumor progression (37) . Pearson correlation coefficients were used to compare results obtained by ISH to those obtained for the same cases using immunohistochemistry.
RESULTS
GKLF mRNA Expression Is Up-Regulated during Breast Tumor Progression.
Previously, SAGE analysis of purified normal breast epithelial cells detected GKLF transcripts at an abundance of 40 tags per million (31 , 38) , and Northern blot analysis of breast tumor cell lines revealed the presence of GKLF transcripts (17) . Using sense and antisense [35S]-labeled riboprobes, we examined the expression of GKLF mRNA in 31 cases of carcinoma of the breast. Specificity of hybridization was determined by using the sense probe as a negative control or by hybridization of the antisense probe to human foreskin, in which GKLF was specifically detected in suprabasal epithelial cells (not shown).
Expression of GKLF was detected in malignant cells in 21 of 31 cases of ductal adenocarcinoma (68%, Fig. 1 ⇓ , Table 1 ⇓ ). For several cases that exhibited no detectable expression of GKLF, prominent expression of the housekeeping gene GAPDH was observed, indicating that overall mRNA integrity was maintained and that failure to identify GKLF transcripts may reflect reduced levels of expression. GKLF expression was increased in malignant cells of 14 of 19 cases that contained adjacent uninvolved epithelium (Fig. 1A) ⇓ . For 7 of these 14 cases, no specific signal was detected in adjacent uninvolved epithelium. In the other seven cases, expression was detected in both uninvolved and malignant cells, with expression of GKLF in malignant cells increased by 3- to 5-fold compared with uninvolved epithelium. Within tumors, expression of GKLF was specific to malignant cells, with little or no expression detected in stromal components (Fig. 1B) ⇓ .
ISH analysis of GKLF mRNA in carcinoma of the breast. Two distinct cases were analyzed by applying an antisense (GKLF-AS)[ 35S]-labeled RNA probe to sections of paraffin-embedded (A) or fresh-frozen (B) surgical material. Brightfield (left) and darkfield (right) views are shown. Sections were stained with H&E. Hybridization to a sense control probe resulted in an average of 0.4 grains/nucleus (not shown). A, two areas of the same slide are shown, with uninvolved (i.e., morphologically normal) breast epithelium (upper plate) adjacent to an area (lower plate) containing DCIS (arrowheads) and additional uninvolved tissue (arrows). B, IDC admixed with cords of stroma. Scale bars = 160 μm.
mRNA ISH analysis of GKLF in tumorsa
GKLF expression in DCIS was not significantly different from invasive carcinoma, but expression in both lesions was higher than for uninvolved breast epithelium (Table 1 ⇓ , Fig. 2 ⇓ ). In contrast to results obtained in breast tumors, examination of several cases of prostatic carcinoma revealed equal or reduced expression in tumor cells compared with adjacent uninvolved glandular epithelial cells (Table 1) ⇓ . In summary, the results suggest that GKLF mRNA expression is activated in approximately two-thirds of breast carcinomas, and that expression in positive cases is consistently induced in DCIS before invasion.
GKLF mRNA expression in normal and neoplastic breast tissue. The data in Table 1 ⇓ was analyzed using a paired t test. Sample size (N), statistical significance (p), and SE are indicated for each comparison. Uninv., uninvolved ducts.
Characterization of a GKLF-specific Monoclonal Antibody.
An IgG1 isotype antibody raised against bacterially expressed GKLF was subsequently referred to as αGKLF. Immunoblot analysis of GKLF-transformed RK3E cells and control cell lines detected a single protein species of 55 kDa consistent with the predicted size of the full-length polypeptide (data not shown). Compared with RK3E cells or control cell lines transformed by other oncogenes, apparent GKLF abundance was increased by several-fold in each of two cell lines transformed by the human expression vector. The epitope recognized by the antibody may be denaturation-sensitive, as a signal was obtained only after overnight exposure of autoradiographic film using a standard chemiluminescence protocol. The antibody was not sufficiently sensitive to detect GKLF by immunoblot analysis of extracts of human tumor cell lines that express the endogenous GKLF mRNA.
The cell type- and tumor type-specific patterns of GKLF mRNA expression were used to examine the specificity of αGKLF in immunohistochemical assays. These patterns can be summarized as follows. Human GKLF mRNA is detected by ISH in differentiating cells of oral epithelium, and is markedly elevated in oral tumors (17) . The mRNA is not detected in morphologically normal basal or parabasal cells, particularly within epidermal pegs that extend further into the submucosa. Mouse GKLF mRNA is similarly found to be more highly expressed in superficial, differentiating cells of the skin and gut, and is reduced or absent in basal epithelial cells in both tissues (20 , 23 , 26) . In contrast to human oral and breast cancer, GKLF mRNA expression is reduced in mouse colorectal tumors compared with normal epithelium (29) , and is similarly reduced in human colorectal cancer as indicated by SAGE (31) .
The staining pattern of αGKLF exhibited a strict concordance with detection of GKLF mRNA (Figs. 3 ⇓ and 4 ⇓ , Table 2 ⇓ ). In positive tissues, αGKLF exhibited a mixed nuclear and cytoplasmic staining pattern. For uninvolved epithelium, DCIS, and invasive breast carcinoma alike, the average cytoplasmic staining was 1.8- to 2.5-fold greater than nuclear staining, suggesting that subcellular localization was not altered during breast tumor progression in any consistent fashion. Cytoplasmic staining was subsequently used as a more sensitive indicator of overall expression.
Immunostaining of human tissues with αGKLF monoclonal antibody. Each panel (A-C) illustrates adjacent areas of a tissue section. A, uninvolved oral epithelium (left) and invasive oral squamous cell carcinoma (right). Arrowheads indicate the basal cell layer, whereas arrows indicate invasive carcinoma. Staining of tumor cells and of superficial epithelial cells is indicated by a brown precipitate. B, a section of small bowel illustrating increased staining of superficial epithelium (left) compared with cells deeper within crypts (right). C, a case of colorectal carcinoma, with increased staining of uninvolved superficial mucosa (left) compared with adjacent tumor cells (right). Scale bar for C (left panel) = 45μ m; other scale bars = 140 μm.
Immunostaining of breast tissue with αGKLF. A, a tissue section containing uninvolved epithelium (left, arrowheads) adjacent to invasive carcinoma (right). B, a different case showing invasive carcinoma cells with a mixed nuclear and cytoplasmic staining pattern. C, a tissue section containing an uninvolved duct (left panel) adjacent to both DCIS (right panel, arrows) and invasive carcinoma (right panel, arrowheads). Scale bars: A = 120 μm; B = 30 μm; C = 60 μm.
Immunohistochemical analysis of GKLF in tumorsa
In several samples of skin or oral squamous epithelium, αGKLF bound specifically to differentiating suprabasal epithelial cells (Fig. 3A) ⇓ . Compared with adjacent uninvolved epithelium, staining was markedly increased in malignant cells for each of several cases of squamous cell carcinoma with little or no staining of stromal components of the tumor, as shown previously for the mRNA (17) . Likewise, staining was increased in superficial cells compared with cells deeper within epithelial crypts of the small bowel (Fig. 3B) ⇓ or the large bowel (Table 2 ⇓ ; P = 0.043). In contrast to oral and breast tumors, staining was reduced in tumor cells compared with adjacent superficial epithelial cells for each of four cases of human colorectal adenoma or carcinoma examined (Fig. 3C ⇓ , Table 2 ⇓ ; P = 0.027).
Expression of GKLF Protein Is Increased during Neoplastic Progression in the Breast.
Eighteen cases were tested for GKLF expression by immunohistochemistry (Table 2 ⇓ , Fig. 4 ⇓ ). Nuclear and cytoplasmic staining of normal breast epithelium, DCIS, and invasive carcinoma were semiquantitatively assessed. Low-level staining of tumor cells was observed for 6 cases (e.g., cytoplasmic staining ranging from 0.20 to 0.85), with 11 cases exhibiting higher-level staining (e.g., cytoplasmic staining ranging from 1.00 to 1.75). These results are consistent with detection of the mRNA in approximately two-thirds of tumors by ISH. For cases 23–31, which were analyzed by both ISH and immunohistochemical staining, results of the two methods exhibited a close correlation that reached statistical significance for invasive carcinoma cells (N = 8; coefficient = 0.77; P = 0.024). In DCIS, the correlation was moderate, although the sample number was small (N = 7; coefficient = 0.43). Perhaps because of the overall lower level of expression in uninvolved tissue, the correlation was weakest in uninvolved ducts. Minor differences observed for the two methods may be attributed to differences in sensitivity and specificity, to false negative results attributable to partial degradation of mRNA in some surgical samples, or to analysis of nonserial sections of the same tissue block. As observed in uninvolved tissue adjacent to tumors, staining was low or undetectable for each of five cases of reduction mammoplasty (data not shown).
Apparent GKLF expression as determined by nuclear or cytoplasmic immunostaining was increased in both DCIS and invasive carcinoma compared with uninvolved ducts (Table 2 ⇓ , Fig. 5 ⇓ ). For morphologically normal ducts, staining of myoepithelial cells was not significantly different from that of luminal epithelial cells (P = 0.303, data not shown). However, staining of neoplastic cells in DCIS was significantly increased compared with myoepithelial cells within the same ducts (P = 0.0001), which was consistent with other studies indicating similarities between tumor cells and luminal epithelial cells (39) .
Staining of uninvolved and neoplastic breast tissue byα GKLF. The data in Table 2 ⇓ were analyzed using a paired t test. Sample size (N), statistical significance (p), and SE are indicated for each comparison. Uninv., uninvolved ducts.
Analysis of GKLF in Cultured Breast Epithelial Cells.
Northern blot analysis of breast tumor cell lines revealed variable levels of GKLF expression relative to a tubulin control (Fig. 6) ⇓ . GKLF expression was high in MCF7 and ZR75-1; intermediate in BT474, BT20, MDAMB361, and SKBR3; and reduced in MDAMB453 and MDAMB231. Thus, expression in six of eight breast tumor-derived cell lines was increased relative to 184 cells, an HMEC population of finite life span derived from normal breast tissue after reduction mammoplasty (Lane 1). Expression was similarly increased in 184A1 cells (33) . These immortalized cells were derived from 184 cells by treatment with benzo(a)pyrene. They are wild-type for p53 and p105Rb and are anchorage-dependent and nontumorigenic in animals. The results obtained for breast tumor cell lines support the conclusion that GKLF expression is up-regulated at the mRNA level in most breast tumors, whereas activation in 184A1 cells is consistent with identification of GKLF induction as an early event.
Northern blot analysis of GKLF expression in human breast tumor cell lines. Total RNA from the indicated cell lines was analyzed. Lane 1, finite-life span HMECs. Lane 2, benzo(a)pyrene-treated, immortalized HMECs. Lanes 3-10, breast carcinoma-derived cell lines. Lane 11, SCC15, a human oral squamous cell carcinoma-derived cell line. Lane 12, a RAS-transformed rat cell line. The filter was stripped and hybridized to a β-tubulin probe.
DISCUSSION
Oncogenes such as c-MYC, GLI, and GKLF function in a regulated fashion in normal epithelium to control cellular proliferation and differentiation (5 , 6 , 8 , 23 , 40 , 41) . Analysis of well-characterized tumor types such as colorectal carcinoma and basal cell carcinoma of the skin suggests that genetic alterations cluster within specific pathways, rather than within any specific gene, and that these pathways can function as regulators of oncogene transcription (42 , 43) . An activity common to several oncogenes implicated in carcinoma is the ability to induce transformed foci in the RK3E assay (17 , 44 , 45) . This assay is highly specific, as foci result from expression of tumor-derived mutant (but not wild-type) alleles of RAS or β-catenin (Ref. 45 and data not shown), and only GKLF and c-MYC were identified in a large screen (17) . The assay also detects a distinct subset of oncogenes compared with other host cell lines. With the exception of RAS, the oncogenes that transform RK3E cells do not induce foci in NIH3T3 cells (17 , 44 , 45) .
GKLF encodes a zinc finger transcription factor of the GLI-Krüppel family (46) and is distinct from many other oncogenes in that expression in normal tissue is observed in terminally differentiating epithelial cells. In addition, expression is induced in association with cell growth-arrest in vitro (26) . As predicted by these observations, expression in certain tumor-types is reduced compared with the relevant normal epithelia. Thus, GKLF expression is reduced in colorectal tumors, a result supported by multiple approaches including analysis of RNA extracted from tissues (29) , SAGE (31) , and immunohistochemical analysis of human tissues (this work). ISH analysis of several prostatic tumors likewise indicates that GKLF is expressed in normal prostatic epithelium, and that expression can be lost during tumor progression.
In contrast to colorectal and prostatic carcinoma, GKLF expression is activated in both invasive carcinoma and preinvasive neoplastic lesions during progression of most breast carcinomas and virtually all oropharyngeal squamous cell carcinomas. Breast and oral cancers share a number of additional molecular alterations. Loss-of-function mutations frequently affect p53 and p16/CDKN2, whereas a smaller proportion of tumors (5–20%) exhibit gene amplification of c-MYC, cyclin D1, erbB-family members including the EGF receptor and erbB-2/HER-2/neu, or others (47, 48, 49, 50, 51) . Unlike carcinomas of the GI tract or skin, neither breast nor oral carcinoma is reported to exhibit frequent genetic alterations that activate known transforming oncogenes such as RAS, β-catenin, c-MYC, or GLI. By analogy with oncogenes in other tumor types, disruption of the pathways that control GKLF mRNA expression in breast epithelial cells and in oral mucosa represents a potential mechanism of tumor initiation or progression in vivo.
The pattern of GKLF expression in normal epithelia may provide clues as to how GKLF functions in tumor progression. Stratified squamous epithelium contains at least four functionally distinct compartments (52 , 53) . The stem cell compartment is composed of cells within the basal cell layer that exhibit a capacity for self-renewal, but which rarely divide. The transit-amplifying compartment is composed of cells within the basal or parabasal cell layers that exhibit rapid cell division but a reduced capacity for self-renewal. Differentiation occurs within the prickle cell layer that contains identifiable desmosomes, leading to the outermost, keratinized superficial layer. Whereas mechanisms regulating transitions from one compartment to the next remain poorly understood, c-MYC activation can induce stem cells to enter the highly proliferative transit-amplifying compartment (40) . Because self-renewal and rapid cell division occur in distinct cell types, the organization of compartments enables the rapid turnover of epithelial cells while minimizing the possibility of sustaining permanent genetic damage in stem cells.
The observation that GKLF functions normally in the prickle cell layer suggests that each of the three compartments (stem cell, transit-amplifying, and prickle layer) expresses a transforming activity or a critical function (e.g., self-renewal or proliferation) that may contribute to the progression of carcinoma. These compartments appear to be intermingled in dysplastic stratified squamous epithelium, with prickle layer markers including GKLF misexpressed in the basal layers, whereas other basal or parabasal markers are misexpressed in superficial layers. Loss of compartment-specific patterns of gene expression may result in coexpression of the properties of several compartments in a single cell. For example, specific properties of the prickle cell layer, such as reduced cellular adhesion to basement membranes, altered adhesion to other cells, and/or loss of the cellular mechanisms that mediate contact inhibition could confer invasive or metastatic properties to oral carcinomas.
To better understand the mechanism of transformation, we are characterizing transcriptional alterations induced by GKLF when expressed in epithelial cells in vitro. In the future, identification of upstream regulators of GKLF transcription in epithelial cells may elucidate the pathways that regulate GKLF and the mechanism of deregulation of GKLF in specific tumor-types.
Acknowledgments
We gratefully acknowledge Tom Broker and Louise Chow for assistance with ISH, Martha Stampfer for the gift of mammary epithelial cells, and Iuri D. Louro and Pintusorn Hansakul for critical reading of the manuscript.
Footnotes
-
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
-
↵1 Supported by USPHS Grant R29CA65686-05 (to J. M. R.), the University of Alabama at Birmingham Oral Cancer Research Center P50 DE11910-04 (to J. M. R., W. E. G., and J. A. E.), Grants P50 DE08228-10S3 (to J. A. E.) and NIGMS 5T32 GM08361-08 (to K. W. F.), and a gift to the Comprehensive Cancer Center from the Avon Products Foundation Breast Cancer Research and Care Program.
-
↵2 To whom requests for reprints should be addressed, at Department of Medicine, Room 570 Wallace Tumor Institute, University of Alabama at Birmingham School of Medicine, Birmingham, AL 35294-3300. Phone: (205) 975-0556; Fax: (205) 934-9573; E-mail: mruppert{at}uab.edu
-
↵3 The abbreviations used are: GKLF, gut-enriched Krüppel-like factor; DCIS, ductal carcinoma in situ; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HMEC, human mammary epithelial cell; IDC, internal ductal carcinoma; ISH, mRNA in situ hybridization; SAGE, serial analysis of gene expression.
- Received April 12, 2000.
- Accepted September 12, 2000.
- ©2000 American Association for Cancer Research.
References
Volume 60, Issue 22
