Krüppel-like factors 4 and 5: the yin and yang regulators of cellular proliferation

Identification and Initial Characterization of KLF4 and KLF5

A full-length mouse cDNA clone encoding KLF4 (also called gut-enriched Krüppel-like factor or GKLF) was initially isolated from a NIH3T3 cDNA library by reduced-stringency screening with a DNA probe containing the zinc finger region of an immediate early gene product, Zif268 or Egr1 [5]. Mouse KLF4 contains 483 amino acids, has a predicted molecular weight of 53 kD, and is 90% identical to human KLF4. The carboxyl terminus of KLF4 contains three C2H2-zinc fingers that are most closely related to another member of the family, KLF2 [4]. KLF4 is a nuclear protein whose cellular address depends on two nuclear localization signals [6]. A survey of the tissue distribution in adult mice revealed that KLF4 is highly expressed in terminally differentiated, post-mitotic epithelial cells of the intestinal tract [5], a finding consistent with the anti-proliferative effect of KLF4 (see below).

KLF5 (also called intestinal-enriched Krüppel-like factor or IKLF) was initially isolated based on close homology to KLF2 [7]. The coding region of mouse KLF5 contains 446 amino acids and is 88% identical to the human sequence, previously identified as basic transcription element binding protein 2 or BTEB2 [8]. Like KLF4, KLF5 is expressed at a relatively high level in the intestinal tract, its expression is concentrated at the base of the crypt epithelium where active cell division occurs [7].

Expression of both KLF4 and KLF5 genes are developmentally regulated, with a higher level of expression occurring toward the later stage of fetal development [9, 10]. Depending on the tissues, both overlapping and mutually exclusive patterns of expression have been observed for the two genes. The respective expression of KLF4 and KLF5 in differentiated, post-mitotic villus and proliferative crypt epithelial cells of the adult intestinal tract also occurs during embryonic development [10]. The cellular distribution of KLF4 and KLF5 transcripts in the epidermis of the skin in the late stage embryo mirrors that of intestinal epithelium, wherein KLF4 is more highly expressed in terminally differentiated, quiescent suprabasal layer and KLF5 in the proliferative basal layer of the epidermis [10].

KLF4 and KLF5 bind to similar DNA sequences but exert opposing transcriptional regulatory activities

Like many other zinc finger-containing transcription factors, KLF4 and KLF5 bind to cis-DNA elements that are GC-rich. A consensus DNA binding sequence was empirically determined for KLF4 [11]. This sequence is present in many gene promoters and includes the CACCC element and the basic transcription element (BTE). Indeed, KLF4 was subsequently shown to inhibit the promoter of the cytochrome P-450IA1 (CYP1A1) gene in a BTE-dependent manner [12]. In contrast, BTEB2, the human homolog of KLF5, activates transcription through the BTE element [8]. KLF4 and 5 have also been shown to antagonize each other in controlling expression of other genes, despite the two sharing very similar, if not identical, cis-DNA sequences [13]. An example is the gene encoding KLF4 itself. Here KLF4 is an activator of the KLF4 promoter, while KLF5 is an inhibitor, even though both proteins interact with the same cis-element [13]. Moreover, KLF5 abrogates the activating effect of KLF4 on the KLF4 promoter and KLF4 abrogates the inhibitory effect of KLF5 on the same promoter. A potential reason for this competing effect is due to physical competition of the two proteins in binding to a cognate sequence in the KLF4 promoter [13]. A similar competing effect between KLF4 and KLF5 in the transcription of several other genes has also been described, including those encoding smooth muscle α-actin [14] and laminin-1 [15].

KLF4 and 5 exhibit contrasting effects on cellular proliferation

The specificity of KLF4 expression for terminally differentiated, post-mitotic intestinal epithelial cells, rather than proliferating crypt cells, prompted Shields et al to examine the behavior of KLF4 in cultured cell systems [5]. In vitro, the level of KLF4 mRNA correlates with the proliferative state of cells in a manner similar to that seen in vivo; serum-deprived quiescent NIN3T3 cells contain a significant amount of KLF4 mRNA, while actively proliferating cells express little, if any KLF4. When serum-deprived cells are stimulated into proliferation by the addition of fresh serum, the level of KLF4 mRNA decreases. Conversely, KLF4 mRNA levels are increased as proliferating NIH3T3 cells become growth-arrested by either serum starvation or contact inhibition. Finally, forced expression of KLF4 by transfection of proliferating cells results in an inhibition of DNA synthesis [5]. From these observations, it was concluded that KLF4 is a growth arrest-associated gene.

A telling mechanism by which KLF4 inhibits DNA synthesis came from studies that examined the response of KLF4 expression to DNA damage-induced growth arrest. DNA damage caused by methyl methanesulfonate (MMS) [16] or γ irradiation [17] leads to the induction of KLF4 expression in a p53-dependent manner. The increase in KLF4 mRNA level parallels the increase in the level of p21^WAF1/Cip1 mRNA, a major cyclin-dependent kinase inhibitor [16]. Importantly, KLF4 was shown to transactivate the p21^WAF1/Cip1 promoter by binding to a specific Sp1-like cis-element in the proximal p21^WAF1/Cip1 promoter. This same element is also needed for p53 to activate p21^WAF1/Cip1 transcription, although p53 does not directly bind to it. Instead, p53 and KLF4 physically interact with each other, thus allowing p53 to gain access to the p21^WAF1/Cip1 promoter and activate p21^WAF1/Cip1 transcription [16]. A consequence of the p53-dependent activation of p21^WAF1/Cip1 expression following DNA damage is an arrest in the cell cycle at both the G₁/S and G₂/M transition points. KLF4 was shown to be necessary and sufficient in mediating the checkpoint function of p53 at both of these transition points [17, 18]. KLF4 accomplishes this task both through its transcriptional activation of p21^WAF1/Cip1 and through direct suppression of cyclin D1 [19] and cyclin B1 expression [18], which are required for the G₁/S and G₂/M transitions, respectively. Expression profiling of KLF4 using cDNA microarray analysis confirmed that KLF4 activates transcription of many additional genes encoding inhibitors of the cell cycle and suppresses those encoding promoters of the cell cycle [20]. Recent studies from our laboratory indicate that KLF4 is also involved in preventing centrosome amplification after γ irradiation [21] and possibly in preventing aneuploidy after spindle damage (Dalton and Yang, unpublished observations). Taken together, these studies point to a wide-ranging and crucial effect of KLF4 in maintaining the integrity of the cell cycle. The essential role of KLF4 in mediating the DNA damage response of the cell cycle is summarized in Fig. 1.

The observation that KLF4 exerts an important checkpoint function suggests that KLF4 may have a tumor suppressor effect. Indeed, the levels of KLF4 mRNA are significantly reduced in intestinal adenomas in the Apc^MIN/+ mice and familial adenomatous polyposis (FAP) patients when compared to matched normal tissues [22]. A similar reduction in KLF4 mRNA levels is seen in colorectal cancer specimens when compared to matched normal colonic tissues [23]. Moreover, there is evidence for loss of heterozygosity (LOH) in the KLF4 locus in a subset of colorectal cancer specimens [23]. Conversely, re-expression of KLF4 in a colorectal cancer cell line results in diminished tumorigenecity [24]. Collectively, these findings indicate that KLF4 is a potential tumor suppressor in colorectal cancer. A similar down-regulation and growth suppressive effect of KLF4 has also been described in bladder cancer [25].

In contrast to the growth inhibitory effect of KLF4, KLF5 has a growth promoting effect in cultured cells based on a number of studies. The level of KLF5 mRNA is low in serum-deprived, quiescent NIH3T3 cells, but is acutely and significantly increased upon addition of serum or phorbol ester [26]. Constitutive expression of KLF5 in NIH3T3 [26] and intestinal epithelial cells, IEC6 [27], results in an increased rate of proliferation and, eventually, a transformed phenotype as characterized by anchorage-independent growth [26] (Chanchevalap and Yang, unpublished observations). The level of KLF5 transcript is also elevated in NIH3T3 cells transformed by oncogenic H-Ras, and this increase mediates the transforming effect of oncogenic H-Ras [28]. Fig. 2 illustrates the mechanism by which KLF5 mediates this growth-promoting function of oncogenic H-Ras. The increase in KLF5 expression is a direct result of Egr1, which is activated by mitogen-activated protein kinase (MAPK), which is itself stimulated by oncogenic H-Ras. The increase in KLF5 then stimulates expression of the genes encoding cyclin D1 [28], cyclin B1, and Cdc2 (Nandan and Yang, unpublished observations), a change which culminates in accelerated cell proliferation and subsequent transformation. The importance of KLF5 in mediating cell proliferation is further demonstrated by the recent finding that the inhibitory effect of all-trans retinoic acid (ATRA) on proliferation of the intestinal epithelial cells, IEC6, is mediated by suppression of KLF5 expression [27].

KLF4 and KLF5 exert distinct physiologic functions in vivo as revealed by gene knockout experiments

Transgenic experiments involving targeted deletion of Klf4 and Klf5 genes in mice have demonstrated that the two genes exhibit distinct physiologic functions in vivo. Klf4 is normally expressed in the differentiating layers of the epidermis. Klf4^−/− mice die shortly after birth and exhibit a defect in skin barrier function [29]. Conversely, ectopic overexpression of Klf4 in the epidermis results in accelerated formation of the skin permeability barrier [30]. Additional studies have revealed that the critical role of KLF4 in maintaining barrier function is probably based on its ability to coordinately regulate gene clusters that are involved in such function including the Sprr [31] and keratin families [20]. KLF4 is also needed for terminal differentiation of goblet cells in the colon. Klf4^−/− mice demonstrate a 90% reduction in the number of colonic goblet cells, as well as abnormal expression of goblet cell-specific markers [32].

Klf5^−/− mice are embryonic lethal, but experiments using Klf5^+/− mice showed that KLF5 is an important mediator of cardiovascular remodeling upon external stress. KLF5 is normally markedly induced in activated vascular smooth muscle cells and fibroblasts. In response to external stress such as injury induced by a vascular cuff, Klf5^+/− mice showed diminished levels of arterial wall thickening, angiogenesis, cardiac hypertrophy and interstitial fibrosis [33]. This physiologic effect of KLF5 is in part mediated by its ability to activate expression of genes encoding platelet-derived growth factor-A (PDGF-A) and transforming growth factor-β (TGF-β) [33].

Can KLF4 and KLF5 reverse their biological behavior in certain tumors?

The studies above demonstrate that KLF4 and KLF5 exhibit tumor suppressor and oncogenic activities, respectively, in a number of experimental systems. Can their effects be reversed in different settings? Indeed, several studies have indicated that they can. For example, expression of KLF4 is increased in dysplastic oral squamous epithelium [34]. KLF4 mRNA and protein levels are also increased during progression of breast cancer [35], and nuclear localization of KLF4 is associated with an aggressive phenotype in early stage breast cancer [36]. In contrast to the pro-proliferative effect of KLF5 shown above, loss of KLF5 expression has been observed in prostate [37] and breast cancer [38]. Moreover, KLF5 has been shown to reduce colony formation in transformed intestinal epithelial cells [39]. Therefore it appears that the biological behavior of KLF4 and 5 may change in the context of different tumor models, although the reasons for this are not well understood.