Putative Tumor-Suppressive Function of Krüppel-Like Factor 4 in Primary Lung Carcinoma

The human Krüppel-like factor (KLF) family of transcriptional factors has at least 25 members, all of which contain three domains of Krüppel-like zinc fingers. They regulate a variety of target genes that are involved in differentiation, proliferation, and apoptosis (1).Their roles in oncogenesis and tumor progression have also recently been recognized. KLF4 (formerly GKLF) is a KLF protein that has been reported to activate or repress genes that are involved in cell cycle regulation and differentiation (2, 3). KLF4 expression is frequently lost in various human cancer types, such as colorectal cancer (4), gastric cancer (5), esophageal squamous cell carcinoma (6), intestinal cancer (7), prostate cancer (8), and bladder cancer (9). Recently, KLF4 was shown to undergo promoter methylation and loss of heterozygosity in gastrointestinal cancer (4, 5). Consistent with its tumor-suppressive function, the overexpression of KLF4 reduces the tumorigenicity of colonic and gastric cancer cells in vivo (5, 10). These observations indicate that KLF4 acts as a tumor suppressor. However, high KLF4 expression has been found in primary breast ductal carcinoma (11) and oral squamous cell carcinoma (12). It was also reported that ectopic KLF4 expression in mice induced squamous epithelial dysplasia (13). Thus, KLF4 may function as either a tumor suppressor or an oncogene in tissue type–dependent or cell context–dependent manners.

More recently, KLF4 was reported to play important roles in stem cells (14, 15). Nevertheless, it remains unknown how KLF4 functions in the development and progression of lung cancer. The important roles of KLF4 in stem cells and various cancers prompted us to investigate the level of KLF4 expression in human lung cancer tissues and the effects of its alteration. We found that KLF4 expression was substantially decreased in human lung cancer. Consistently, restoration of or an increase in KLF4 expression significantly inhibited lung cancer cell growth in vitro and suppressed the growth of tumors derived from lung cancer cell lines in vivo. KLF4-mediated antitumor activity was associated with G₁-phase arrest. Together, our data suggest that KLF4 functions as a tumor suppressor in lung cancer and that its down-regulation may contribute to the development and/or progression of lung cancer.

Normal and lung tumor tissue specimens were collected from surgically removed specimens with informed patient consent. The case-matched normal tissues were at least 5 cm away from the edge of corresponding tumors. Both normal and tumor tissues were collected from operation room immediately after tissues were removed from patients and preserved in liquid nitrogen before analysis. Thus, pairs of matched samples were harvested, processed, and analyzed at the same time using the same protocols.

The lung cancer cell lines H460, A549, H322, H1299, H226B, H358, and H2122 and normal human fibroblasts are maintained in our laboratory. Normal bronchial epithelial (HBE) cells and HBE cells transformed by Kras (HBE/Kras) were kindly provided by Dr. John Minna (Southwest Medical School). HBE cells were cultured in serum-free keratinocyte medium (Invitrogen). All other cell lines were routinely cultured in DMEM supplemented with 10% heat-inactivated FCS, 100 units/mL penicillin, and 100 μg/mL streptomycin and maintained in the presence of 5% CO₂ at 37°C.

Adenovirus-mediated KLF4 (Ad-KLF4) was described previously (5). The expansion, purification, titration, and quality analysis of adenovirus were done at the Vector Core Facility of M. D. Anderson Cancer Center as described previously (16).

Western blotting was done as described previously (14) with antibodies for KLF4, p21, and cyclin D1 (Santa Cruz Biotechnology) and β-actin (Sigma).

Total RNA was extracted from frozen lung cancer tissues and lung normal tissues by Trizol reagent (Invitrogen). A 1 μg aliquot of each RNA sample was reverse transcribed in a 20 μL reaction volume using the reverse transcription kit (Clontech). The 10-fold dilutions of the cDNA product were used in real-time PCR. The probe and primer sets for each gene were designed around the junction region of two exons, using the Primer Express software (version 1.0; Perkin-Elmer), so that they are mRNA-specific. Real-time PCR was done using the ABI Prism 7700 Sequence Detection System (Applied Biosystems) according to the protocol of the manufacturer. Human glyceraldehyde-3-phosphate dehydrogenase gene was used as an internal control for relative mRNA amount. The relative RNA level was calculated automatically by the installed software of the instrument with the ΔΔCt method. The relative KLF4 copy number was expressed by log (KLF4/glyceraldehyde-3-phosphate dehydrogenase ratio) value.

Methylation-specific PCR was done using genomic DNA, which was modified with bisulfite according to the manufacturer's instructions (EZ DNA methylation kit). For detecting unmethylated DNA, the forward primer was 5′-TAGTTAAATTAATAAATTTGGTGTATGT-3′ and the reverse primer was 5′-AATAACAATAAATACAACCCTAATCAC-3′; these were designed to amplify a 230-bp sequence between nucleotides −653 and −424 relative to the translation initiation site of the human KLF4 exon 1 region. For detecting methylated DNA, the forward primer was 5′-GGTAGTTAAATTAATAAATTCGGCGTAC-3′ and the reverse primer was 5′-AATAACGATAAATACGACCCTAATCG-3′; these were designed to amplify a 232-bp sequence between nucleotides −655 and −423 relative to the translation initiation site of the human KLF4 exon 1 region. Each PCR of 50 μL consisted of 40 ng DNA and 200 nmol/L each of the forward and reverse primers and 0.5 μL HotStart Taq enzyme (Taqman). Each PCR was hot-started at 95°C for 10 min and then amplified for 35 cycles (94°C for 30 s, 54°C for 30 s, and 72°C for 30 s). PCR products were visualized on 2% agarose gel stained with ethidium bromide.

Genomic DNA from lung cancers and matched normal tissues were treated with bisulfite and amplified as described above. The PCR products were cloned with the Topo TA cloning kit (Invitrogen). The forward primer was 5′-GTGATTTATTTAGTTTTTTATTTTTTTT-3′ and the reverse primer was 5′-TACACAATACTAAACACTACCTCCTCTC-3′; these primers were designed to amplify a 173-bp sequence between nucleotides −1284 and −1112 relative to the translation initiation site of the human KLF4 exon 1 region. The clones were sequenced to determine the CpG methylation rate by the number of methylated CpG pairs to the total number CpG pairs in this region.

Cells (1 × 10³) were plated in a six-well plate. After 9 days of culture, surviving colonies (>50 cells per colony) were counted with crystal violet staining. Triplicate independent experiments were done.

Plasmids encoding KLF4 short hairpin RNA (shRNA) expression were obtained from OriGene Technologies. HBE and HBE/Kras KLF4 knockdown pools were established according to the manufacturer's instructions for the pRS vector kit to produce retrovirus expressing KLF4 shRNA and retrovirus expressing green fluorescent protein shRNA as a control.

The viability of the cell lines was determined using the sulforhodamine B assay as described previously (17). In brief, after adherent cells had been fixed with trichloroacetic acid in a 96-well microplate, the protein was stained with sulforhodamine B, and the absorbance at 570 nm was determined. Relative cell viability was determined by setting the viability of the control cells at 100% and comparing the viability of the treated cells with that of the controls. These experiments were done at least three times.

Cells (2 × 10⁵) were plated in a six-well plate and treated with Ad-KLF4 and Ad-LacZ. After 72 h, cells were fixed in 70% ethanol and stained with propidium iodide. The DNA content was analyzed with an Epics Profile II flow cytometer (Beckman Coulter) with Multicycle software (Phoenix Flow Systems). All experiments were repeated at least twice.

Female nu/nu mice were purchased from Charles River Laboratory. The mice were housed in laminar flow cabinets under specific pathogen-free conditions. All animal experiments were carried out in accordance with the Guidelines for the Care and Use of Laboratory Animals (NIH publication number 85-23) and the institutional guidelines of M. D. Anderson Cancer Center. The ex vivo tumor-suppressive ability of KLF4 was investigated using a tumor xenograft experiment with 5 × 10⁶ parental, empty vector–transformed, and KLF4-expressing H322 cells injected subcutaneously into the right and left hind legs of 6- to 8-week-old nude mice (at least 6 mice per group), respectively. Tumor volumes were monitored by measuring the longest (L) and shortest (W) diameters. The volume was calculated by the formula: V = 0.5 × L × W² (18). The effect of Ad-KLF4 was evaluated in tumors derived from the H322 lung cancer cell line in vivo. A H322 subcutaneous model was established in nude mice. These mice were divided into three groups, which were treated with PBS, Ad-LacZ, and Ad-KLF4, respectively, by intratumoral injection. Tumor volumes were monitored over time.

Differences between the treatment groups were assessed by ANOVA using statistical software (StatSoft). Differences between the results of the in vivo tumor growth experiment were assessed using ANOVA with a repeated-measurement module. P values < 0.05 were regarded as significant.

The critical role of KLF4 in gastrointestinal cancers and stem cells prompted us to determine whether this gene has a role in lung carcinoma. The gene expression of KLF4 in primary lung carcinoma was evaluated in 25 primary tumor tissues by Western blot analysis randomly according to time taken from operation room. The case-matched normal lung tissues were taken from same patients and were at least 5 cm away from the corresponding tumor samples. The matched samples were collected, processed, and analyzed at the same time. The result showed that KLF4 protein was dramatically down-regulated in 21 of 25 (84%) primary lung tumors compared with matched normal lung tissues (P < 0.01; Fig. 1). The apparent down-regulation of KLF4 in tumor samples was also observed when α-tubulin was used as a control (data not shown). Of the 25 primary tumor tissues tested, 13 were squamous carcinoma, 8 adenocarcinoma, and 4 non–small cell carcinoma. The patients had mean age of 65 years (range, 42-82) at the time when tissue samples were collected. Clinical stages varied from stage IA to stage IIIB. The down-regulation of KLF4 was observed in all types of tumors tested. There is no apparent association between KLF4 expression and clinical stages, tumor types, or patients' ages.

To determine whether this down-regulation occurred at the transcriptional level, we tested KLF4 mRNA levels in 10 primary lung tumor tissues and their matched normal lung tissues by quantitative real-time PCR analysis. We found dramatically lower KLF4 copy numbers in 9 of 10 (90%) lung tumors than in the normal tissues (P < 0.01; Fig. 2). These results show that KLF4 expression was down-regulated in primary lung tumors not only at the protein level but also at the RNA level.

To determine whether down-regulation of KLF4 in primary lung tumors is caused by promoter hypermethylation, we determined the methylation status of the KLF4 promoter region. Methylation-specific PCR analysis was done using methylation- or unmethylation-specific primers of the KLF4 promoter. The methylated allele of KLF4 was observed in five primary tumor tissues with low KLF4 expression but not in matched normal tissues (Fig. 3A). We also analyzed the methylated CpG rate in the region of −1284 to −1112 of the KLF4 promoter by bisulfite treatment and then by PCR, cloning of PCR products, and DNA sequencing. CpG island methylation of 3.25% to 5.5% was detected in three primary tumors with low KLF4 expression and 0% to 0.75% in matched normal tissues (Fig. 3B). These results show that hypermethylation of the KLF4 promoter was associated with its transcriptional repression.

We then determined the expression of KLF4 gene in seven cultured lung cancer cell lines. KLF4 expression was dramatically reduced in three cell lines (H226B, H322, and H1299) compared with normal human fibroblasts (Fig. 4A). To determine whether restoring KFL4 expression would affect cell proliferation, we treated these cell lines with an adenovector expressing KLF4 (Ad-KLF4) at a multiplicity of infection of 2,000 viral particles and determined cell viability 72 h after infection. Cells treated with Ad-LacZ were used as controls. Cell viability was dramatically decreased in H322, H1299, and H226B cell lines after Ad-KLF4 treatment compared with the control vector. However, in lung cancer cell lines that had higher levels of KLF4, Ad-KLF4 treatment had no obvious effects on cell viability (Fig. 4B), although equivalent amount of exogenous KLF4 was expressed among those cell lines after adenovector-mediated gene transfer (data not shown).

To identify the mechanism underlying the inhibitory effect of Ad-KLF4, we treated cells with Ad-KLF4 or Ad-LacZ at a multiplicity of infection of 2,000 viral particles for 72 h and performed a cell cycle analysis by flow cytometry. In two KLF4-sensitive cell lines, H322 and H1299, Ad-KLF4 treatment significantly increased the percentage of cells in G₁ phase when compared with PBS- or Ad-LacZ–treated cells (P < 0.01; Fig. 4C); however, Ad-KLF4 had no obvious effect on the cell cycle profiles of H2122, which did not respond to Ad-KLF4 in the cell viability analysis. Ad-KLF4 treatment also did not induce an obvious increase in apoptotic cells (sub-G₁ phase) in H322 and H1299 cells (Fig. 4C) nor cleavage of caspase-3 or caspase-8 (data not shown), indicating that the suppressed cell growth in these cells was caused by G₁-phase arrest, not by apoptosis induction.

It has been reported by others that p21 and cyclin D1, both of them involved in cell cycle regulation, were regulated by KLF4 (19, 20). To test whether KLF4 plays role in regulating p21 and cyclin D1 expression in lung cancers, we first determined the expression of KLF4, p21, and cyclin D1 in 6 pairs of normal lung and lung cancer tissues by Western blot analysis. The result showed that down-regulation of KLF4 in primary tumor tissues was associated with either up-regulation of cyclin D1, down-regulating of p21, or both (Fig. 4D and E). We then determined whether enforced change of KLF4 expression would affect the expression of p21 and cyclin D1 in cultured cells. For this purpose, H322 cells were infected with 2,000 multiplicities of infection of Ad-KLF4 or control vector Ad-LacZ for 48 h. Expressions of KLF4, p21, and cyclin D1 were then determined by Western blot analysis. The result showed that enforced expression of KLF4 led to down-regulation of cyclin D1 and up-regulation of p21, suggesting that KLF4 regulate these two genes in opposite directions. This observation was further supported by knockdown of KLF4 in HBE/Kras cells (Fig. 4F). Knockdown of KLF4 resulted in a dramatic down-regulation of p21 and up-regulation of cyclin D1. For example, knockdown of KLF4 resulted in 40% to 60% reduction of p21 levels when compared with mock-treated or control small interfering RNA–treated cells. Together, these results indicate that KLF4 plays important roles in regulating p21 and cyclin D1 expression in lung cancer cells, which may contribute to its role in regulating cell cycle progression.

To further determine whether the KLF4 gene has an antiproliferation effect in lung cancer cells, we transfected it into H322 cells and selected two stable KLF4-expressing clones (KLF4-c2 and KLF4-c5). H322 cells transfected with empty vector were used as the control. KLF4 protein expression in KLF4-c2 and KLF4-c5 was confirmed by Western blot analysis (Fig. 5A). The anti–cell proliferation effect of KLF4 was assessed by a clonogenic formation assay. The efficiency of clonogenic formation was significantly inhibited (P < 0.01) in KLF4-c2 and KLF4-c5 cells compared with empty vector–transformed controls (Fig. 5A).

We then tested the effect of KLF4 knockdown in normal HBE and HBE/Kras cells. HBE/Kras cells were infected with retrovirus expressing shRNA specific for KLF4. After a brief selection with puromycin, KLF4-knockdown cells were pooled together for further analysis. Parental and green fluorescent protein encoding shRNA-transfected HBE and HBE/Kras cells were used as controls. KLF4 knockdown in HBE and HBE/Kras cells was verified by Western blot analysis (Fig. 5B). Knockdown of KLF4 enhanced the cell growth rate in HBE and HBE/Kras cells (data not shown). A clonogenic formation analysis showed that KLF4 knockdown resulted in increased clonal formation in both HBE and HBE/Kras cells (P < 0.01; Fig. 5B).

The effects of KLF4 gene on in vivo tumor growth was analyzed with KLF4 stably transfected H322 cells. We inoculated 5 × 10⁶ of parental, vector-transfected, and KLF4-transfected (clones KLF4-c2 and KLF4-c5) H322 cells subcutaneously into the rear flanks of nude mice (n = 7-8 per group). Tumor growth was monitored every 2 to 3 days. As shown in Fig. 6A, the tumor growth of KLF4-c2 or KLF4-c5 cells was dramatically retarded compared with that of parental H322 and empty vector–transformed cells. By the end of the experiment, the mean size of in KLF4-expressing tumors was significantly smaller than that of parental H322 and empty vector–transformed tumors (P < 0.01; Fig. 6A).

We also evaluated the in vivo tumor growth of parental H322 cells after Ad-KLF4 treatment by subcutaneously injecting 5 × 10⁶ H322 cells into nude mice. When tumors reached 3 to 5 mm in diameter, treatment was started with PBS or 2 × 10¹⁰ viral particles/injection of Ad-KLF4 or Ad-LacZ. The treatment was repeated every 3 days for a total of three times. As shown in Fig. 6B, tumor growth in the Ad-KLF4–treated group was dramatically suppressed compared with that in the PBS and Ad-LacZ groups (P < 0.01; Fig. 6B). Together, these results show that restoring KLF4 gene expression can suppress tumor growth in vivo.

KLF4 was found to promote cell growth in breast cancer (11) but was down-regulated in other cancers, including colorectal cancer, gastric cancer, esophageal squamous cell carcinoma, intestinal cancer, prostate cancer, and bladder cancer (4–9). In the present study, we found that KLF4 expression was dramatically down-regulated in most primary lung cancers compared with matched normal lung tissues at both RNA and protein levels. However, because of lack of reliable antibodies for immunohistochemical analysis, we are not able to determine what type of cells had reduced expression of KLF4. On the other hand, our results revealed that reduced KLF4 expression was associated with hypermethylation in its promoter region in primary lung cancer tissues, which is consistent with the findings of previous reports of hypermethylation in gastric and colorectal cancers (4, 5). The causal relationship between promoter hypermethylation and reduced KLF4 expression needs to be further investigated. It is also possible that other causes, such a gene deletion or a gene mutation, may also contribute to the reduced expression. Nevertheless, we found that KLF4 expression can be induced in some lung cancer cell lines by treating cells with demethylation agents, such as 5-aza-2′-deoxycytidine (data not shown). It remains to be determined, however, whether demethylation agents can be used to restore KFL4 function in vivo for cancer therapy.

At least 25 Sp1-like/KLF genes have been identified in humans; these are classified into Sp1-like and KLF subfamilies (21, 22). All KLF genes contain three Cys²His² zinc-finger motifs and can recognize and specifically bind to GC-rich sequences (1). These factors are present in species ranging from the nematode Caenorhabditis elegans to humans and appear to have evolved through multiple gene-duplication events (23, 24). The defining feature of KLF proteins is a highly conserved DNA-binding domain (>65% sequence identity among family members) at the carboxyl terminus that has three tandem Cys²His² zinc-finger motifs (1). In addition to DNA binding, these zinc-finger motifs may also function in protein-protein interactions that modulate DNA-binding specificity (25, 26). The amino-terminal regions of KLF proteins are more variable and contain transcriptional activation or repression domains. In addition, KLF proteins have nuclear localization sequences, which can be found immediately adjacent to or within zinc-finger motifs (27, 28).

Because many KLF proteins regulate cell growth in a variety of cell types, it is not surprising that some members of the family also appear to participate in mechanisms leading to carcinogenesis. Sp1 expression and activity have been found to be increased in epithelial carcinomas compared with in benign tumors, such as papillomas, suggesting that Sp1 is involved in tumor progression (29, 30). KLF6 was also recently reported to be a candidate tumor suppressor gene that is mutated in prostate cancer (31), and the ability of KLF6 to inhibit cell growth was reduced by mutations within its transcriptional regulatory domain. Fetal and adult lung tissues have been reported to express high levels of KLF2, which is also named lung KLF (32, 33) and is required for normal embryogenesis and late-stage lung development (34, 35). In this study, we tried to determine whether KLF2 was differentially expressed in normal or malignant lung tissues but obtained no informative data because of the low quality of commercially available antibodies.

The mechanisms by which KLF4 influences cancer development and progression are not yet clear. Evidence indicates that altered KLF4 expression affects the cell cycle (36, 37). In the present study, we found that restoration of KLF4 expression dramatically induced G₁-phase arrest in lung cancer cells, with minimal effect on apoptosis induction, suggesting that KLF4 suppresses tumor cell growth by cell cycle arrest. KLF4 could exert its tumor-suppressive features, at least in part, through activation of p21 (20, 38) and inhibition of cyclin D1 (19). KLF4 can suppress cyclin D1 expression through competition with Sp1 for the same binding site of cyclin D1 promoter (19). Both p21 and cyclin D1 have been reported to play critical roles in cell cycle progression. For example, p21 is a critical negative regulator of G₁-S-phase transition, a major checkpoint for cell cycle progression. The cyclin kinase inhibitor p21 can induce G₁ arrest and block entry into S phase by inactivating cyclin-dependent kinases or by inhibiting activity of proliferating cell nuclear antigen (39). We found that ectopic expression of KLF4 promoted p21 expression and suppressed cyclin D1 expression, suggesting that restoration of KLF4 expression can indeed dramatically change the expression of molecules that are critical in controlling cell cycle progression. p21 is known to be regulated by p53. Interestingly, KLF4 is reported to have opposing effects on p21 and p53 in mouse embryonic fibroblasts, suppressing p53 but inducing p21 (40). Whether KLF4 also plays role in p53 regulation or whether p53 contributes to KLF4-mediated regulation of p21 and cyclin D1 in lung cancers is not yet clear. Nevertheless, our results showed that restoration of KLF4 expression significantly inhibited lung cancer cell growth in vitro and suppressed tumor growth in vivo. Although results from stably transfected cells may have limitation such as potential selection bias or mutagenic effect derived from integration of exogenous DNA, we found the similar tumor-suppressive effect when established tumors were treated with an adenovector expressing KLF4 gene, suggesting that the observed phenotype is likely associated with KLF4 functions. Together, these results suggest that altered KLF4 expression has a role in lung cancer development and that KLF4 may serve as a biomarker or treatment target of lung cancer.

No potential conflicts of interest were disclosed.

We thank Ann Sutton for editorial review and Karen M. Ramirez for technical assistance with the flow cytometric analysis.