KLF6, a Candidate Tumor Suppressor Gene Mutated in Prostate Cancer

See allHide authors and affiliations

Science  21 Dec 2001:
Vol. 294, Issue 5551, pp. 2563-2566
DOI: 10.1126/science.1066326


Kruppel-like factor 6 (KLF6) is a zinc finger transcription factor of unknown function. Here, we show that theKLF6 gene is mutated in a subset of human prostate cancer. Loss-of-heterozygosity analysis revealed that one KLF6allele is deleted in 77% (17 of 22) of primary prostate tumors. Sequence analysis of the retained KLF6 allele revealed mutations in 71% of these tumors. Functional studies confirm that whereas wild-type KLF6 up-regulates p21 (WAF1/CIP1) in a p53-independent manner and significantly reduces cell proliferation, tumor-derived KLF6 mutants do not. Our data suggest thatKLF6 is a tumor suppressor gene involved in human prostate cancer.

Prostate cancer is a leading cause of cancer death in men, with more than 198,000 new cases and 32,000 deaths annually in the United States alone. Loss of heterozygosity (LOH) analyses of sporadic prostate cancers and linkage studies of familial prostate cancer have provided strong evidence for the existence of prostate cancer–susceptibility genes (1). Although a number of tumor suppressor genes, including the retinoblastoma susceptibility gene (RB1), the putative protein tyrosine phosphatase gene (PTEN), andp53, have been implicated in prostate cancer, no single gene has yet been identified which is responsible for the majority of cases (2).

KLF6 (Zf9/CPBP) (GenBank accession number AF001461) is a ubiquitously expressed Kruppel-like transcription factor whose in vivo role has not been fully clarified (3–5). KLF6 contains a proline- and serine-rich NH2-terminal activation domain, and like other Kruppel-like factors, three COOH-terminal C2H2 zinc fingers.KLF6 directly interacts with DNA through a GC box promoter element (3). Putative transcriptional targets ofKLF6 include the genes encoding a placental glycoprotein (4), collagen α1(I) (3), transforming growth factor β1 (TGFβ1), types I and II TGFβreceptors (6), urokinase type plasminogen activator (uPA) (7), and the human immunodeficiency virus long terminal repeat (HIV-1 LTR) (5).

The KLF6 gene maps to human chromosome 10p, a region deleted in ∼55% of sporadic prostate adenocarcinomas (8, 9). Given the role of Kruppel-like factors in the regulation of many cellular processes that include differentiation and development (10), we examined primary prostate tumor samples for specific LOH of the KLF6 gene. Microsatellite markers flanking KLF6 were analyzed in paired normal prostate tissue and in well to poorly differentiated prostate tumor specimens from 22 patients (11, 12). Of the 22 samples analyzed, 17 (77%) displayed LOH of the KLF6 locus (Fig. 1A). To define the minimal region of loss, we designed two microsatellite markers, KLF6M1 andKLF6M2, which flank the KLF6 gene by ∼42 kb and ∼12 kb, respectively (12). Tumor DNA from patients 9 and 10 showed loss of only the tightly flanking KLF6M1 andKLF6M2 microsatellite markers, whereas that from patients 14 and 21 demonstrated loss of only KLF6M1 (Fig. 1A). Representative fluorescent electropherograms for microdissected tumor samples with loss of KLF6M1 are shown in Fig. 1B.

Figure 1

LOH at the KLF6 locus in human prostate tumors. (A) Summary of LOH patterns of 22 prostate tumors. Retained microsatellite markers are indicated in white, markers demonstrating allelic loss in black, and noninformative markers in gray. A hatched circle indicates DNA that could not be amplified. Patient data was grouped according to degree of LOH. Genetic map is not drawn to scale. (B) Representative fluorescent electropherograms for microsatellite marker KLF6M1 for patients with LOH. A XLOH score of less than 0.7 was used for determination of LOH (12).

We then sequenced all four coding exons and intron-exon boundaries of the retained KLF6 allele using genomic DNA from these 22 tumors (13). Twelve of 17 (71%) tumor samples showing LOH of KLF6 had mutations in the retained KLF6 allele (Fig. 1A and Fig. 2, A and B), suggesting that two inactivating events had occurred; thus, as defined by Knudson's “two-hit model” (14), KLF6appears to be a tumor suppressor gene. No mutations were present in the paired normal prostate tissue genomic DNA from these patients, confirming that the tumor-derived mutations were somatic. Analysis of 11 additional prostate cancer patients for which there was insufficient normal DNA available to complete LOH studies also revealedKLF6 mutations in five of the tumors (12).

Figure 2

Sequence analysis reveals KLF6 mutations in human prostate tumors. (A) Tumor samples from patients 1 through 5 showed LOH of the KLF6 locus. Sequence analysis ofKLF6 was performed on PCR-amplified and subcloned tumor-derived genomic DNA, which was then used in functional studies. The mutations are underlined. (B) LCM was used to isolate normal and malignant cells from distinct foci within the same tumor. DNA sequence analysis was performed using genomic DNA extracted from each malignant focus and from surrounding normal prostate tissue. Separate malignant foci (Focus 1 and Focus 2) from the same patient contained different KLF6 mutations (Patient 13 and Patient 17). Patient 18 had a KLF6 mutation in Focus 1 but not in Focus 2. Mutations are indicated by an “N” in the sequencing chromatogram. No mutations were detected in the adjacent normal prostate tissue, confirming that the mutations were somatic.

Interestingly, a number of tumor samples showed compound mutations in KLF6 (Fig. 2B). The presence of multiple single-gene mutations within the same tumor is unusual, but genetic heterogeneity in prostate cancer has been reported previously for the tumor suppressor p53 (15, 16). Here, we show that separate malignant foci within the same tumor can harbor distinct KLF6 mutations. We used laser capture microdissection (LCM) to isolate malignant tissue (2,500 to 10,000 cells) from separate tumor foci (12) within the same tumor in six patients. Five of these patients had KLF6 mutations. In three, the mutation was seen in DNA derived from the cells of one tumor focus and not the other. Furthermore, two patients had distinctKLF6 mutations in different tumor foci. No mutations were detected in the adjacent normal prostate tissue.

In total, 18 of 33 prostate tumors (55%) had KLF6mutations. Of the 26 mutations identified, 23 were within theKLF6 transactivation domain (12). These mutations resulted in 25 nonconservative amino acid changes and the introduction of a premature stop codon. None of these mutations was present in the patient's normal prostate tissue DNA or in germline DNA from 100 chromosomes of 50 unaffected, unrelated individuals. The majority of the mutations affected highly conserved amino acids, suggesting that these residues are functionally important.

To explore the biological activity of KLF6, we generated an NIH 3T3 cell line in which KLF6 expression was regulated by a tetracycline (tet)-responsive promoter (12). Up-regulation of wild-type (wt) KLF6 significantly (P < 0.001) reduced cell proliferation (Fig. 3A). Induction of KLF6 resulted in a fivefold increase in the expression of p21 (WAF1/CIP1), an inhibitor of several cyclin-dependent kinases and a key regulator of the G1/S transition (17), and a reduction in proliferating cell nuclear antigen (PCNA) expression (Fig. 3B) (12). Transient cotransfection assays with wt KLF6 and ap21 promoter luciferase reporter construct revealed thatp21 is a direct transcriptional target of KLF6(Fig. 3C) (12). Transient cotransfection performed with wt KLF6 and p21 promoter deletion constructs (18) revealed that upstream flanking sequences including two GC boxes present in only the deletion construct pW-225 were necessary for high-level transactivation of the p21 promoter by KLF6 (Fig. 3D). Gel shift using oligonucleotides corresponding to these wt and mutated “GC box” motifs further indicated that transactivation of the p21 promoter by KLF6 is mediated through binding to these two GC boxes (12). Additional studies have confirmed that KLF6 up-regulates p21 and suppresses growth in a p53-independent manner (12).

Figure 3

p21 (WAF1/CIP1) is a direct transcriptional target of KLF6. (A) DNA synthesis was assayed in NIH 3T3 fibroblasts expressing KLF6 under the regulation of a tetracycline (tet) responsive promoter. Induction of KLF6 by 24-hour withdrawal of tet reduced cell proliferation by 40% compared to control cells expressing an empty vector (P < 0.001) (B) Immunoblot showing that induction of KLF6 results in a fivefold up-regulation of p21 and a decrease in proliferating cell nuclear antigen (PCNA) expression. (C) Luciferase activity assayed 24 hours after transient cotransfection of a KLF6 cDNA expression vector and a wtp21 promoter reporter construct. An eightfold increase in promoter activity was detected, similar to the activity of a p53 expression vector (*** P < 0.0001 relative to empty vector as assessed by two-way ANOVA) (20) (D) Luciferase activity was assayed 24 hours after transient cotransfection of a KLF6 cDNA expression vector andp21 promoter deletion constructs (pW-225, pW-78, pW-53, pW-35) (18). Transactivation of the p21promoter by KLF6 was seen only with the pW-225 construct. The GC boxes that appear to mediate high-level transactivation of thep21 promoter present in the pW-225 construct are indicated (these GC boxes are absent in the pW-78, pW-53, and pW-35 constructs).

To determine the effect of tumor-derived mutations on KLF6function in this cell culture assay system, we generated cDNAs encoding four KLF6 protein mutants (12) and performed transient cotransfection assays with a p21 promoter reporter (Fig. 4A). Wild-type KLF6 transactivated thep21 promoter 10-fold, whereas none of the four tumor-derived mutants were active to a similar level (P < 0.0001). Unlike wt KLF6, none of these mutants significantly up-regulated the endogenous level of p21 (Fig. 4B) or significantly suppressed the growth of prostate cancer 3 (PC3) cells (Fig. 4B).

Figure 4

Prostate cancer-derived KLF6 mutants show loss of growth suppressive activity. (A) Luciferase activity was assayed 24 hours after cotransfection of an immortalized human embryonal fibroblast cell line (293T) with ap21 promoter reporter construct containing mutated p53 binding sites and the indicated mutant or wt KLF6. A 10-fold increase in p21- promoter activity was detected following expression of wt KLF6 (*** P < 0.0001). The X137 truncation mutant had no transactivating activity and the remaining three tumor-derived mutants transactivated the p21 promoter (**P < 0.05) to a lesser extent than wt KLF6 (ψ indicates a P < 0.0001 relative to KLF6 mutants by two-way ANOVA) (20). (B) PC3 cells were transfected with the R64, D123, X137, P169 tumor-derived mutant proteins or wt human KLF6. Cells were harvested 24 hours later and KLF6 and p21 expression levels were determined by Western blot. All four mutant proteins were expressed. Numbers on the left indicate size in kilodaltons. A threefold up-regulation of endogenous p21 was detected with the wt KLF6 determined by band densitometry (n = 4, P < 0.001 relative to empty vector). In contrast, there was no significant up-regulation of p21 by any of the tumor-derived mutants. DNA synthesis was assayed 40 hours after PC3 cells were transfected with wt or mutant KLF6 protein. DNA synthesis in cells transfected with wt KLF6 was suppressed by 40% compared to empty vector transfected cells (*** P < 0.0001,n = 6, two-way ANOVA) (20). None of the tumor-derived mutants significantly suppressed DNA synthesis (n = 4).

Of the four missense mutations in the DNA binding domain (amino acids 201 to 283) (3, 12), two are predicted to disrupt the critical zinc finger motifs in the protein and thus could conceivably alter protein function. The Cys265 → Tyr265(C265Y) mutation, which occurs in the last zinc finger, is predicted to prevent zinc binding (19) and hence might affectKLF6-DNA interactions. The Leu217 → Ser217 (L217S) mutation affects a residue conserved across 20 zinc finger–containing domains, and this leucine aligns perfectly with the initial cysteine of the three zinc fingers and would therefore be predicted to affect secondary structure. In addition, examination of the primary sequence revealed mutations involving known phosphorylation motifs.

Our data identify KLF6 as a candidate tumor suppressor gene in prostate cancer. Given its ubiquitous expression and its ability to suppress growth, KLF6 may have a general role in the development or progression of other human cancers, particularly those associated with LOH at chromosome 10p15.

  • * These authors contributed equally to this work.

  • Present address: Structural Neurobiology and Proteomics Laboratory, Department of Biochemistry and Molecular Biology, FUHS/Chicago Medical School, 3333 Green Bay Road, North Chicago, IL 60064, USA.

  • To whom correspondence should be addressed. E-mail: frieds02{at}


View Abstract

Navigate This Article