Krüppel-like Factor 5 is Important for Maintenance of Crypt Architecture and Barrier Function in Mouse Intestine

Abstract

Background & Aims

Krüppel-like factor 5 (KLF5) is transcription factor that is expressed by dividing epithelial cells of the intestinal epithelium. KLF5 promotes proliferation in vitro and in vivo and is induced by mitogens and various stress stimuli. To study the role of KLF5 in intestinal epithelial homeostasis, we examined the phenotype of mice with conditional deletion of Klf5 in the gut.

Methods

Mice were generated with intestinal-specific deletion of Klf5 (Vil-Cre;Klf5fl/fl).

Morphological changes in the small intestine and colon were examined by immunohistochemistry, immunoblotting, and real-time PCR.

Results

Klf5 mutant mice were born at a normal Mendelian ratio but had high mortality compared to controls. Complete deletion of Klf5 from the intestinal mucosa resulted in neonatal lethality that corresponded with an absence of epithelial proliferation. Variegated intestinal-specific deletion of Klf5 in adult mice resulted in morphological changes that included a regenerative phenotype, impaired barrier function, and inflammation. Adult mutant mice exhibited defects in epithelial differentiation and migration. These changes were associated with reduced expression of Cdx 1, Cdx2, and Eph and ephrin signaling proteins. Concomitantly, Wnt signaling to β-catenin was reduced. Proliferation in regenerative crypts was associated with increased expression of the progenitor cell marker Sox9.

Conclusions

Deletion of Klf5 in the gut epithelium of mice demonstrated that KLF5 maintains epithelial proliferation, differentiation, and cell positioning along the crypt radial axis. Morphological changes that occur with deletion of Klf5 are associated with disruption of canonical Wnt signaling and increased expression of Sox9.

The mammalian intestinal epithelium is a continuously renewing system in which several biological processes including stem cell self-renewal, proliferation, terminal differentiation, migration, and apoptosis are carefully choreographed in order to maintain tissue homeostasis. The epithelium of the small intestine is composed of a crypt and a villus compartment. Similarly, the colon consists of a crypt and a surface epithelium compartment. The bulk of the villus and surface epithelium of the small and large intestine, respectively, is composed of differentiated columnar epithelial cells that are divided into two main classes: absorptive and secretory. These cells are descendants of the multipotent stem cells residing in the base of the crypt and are no longer dividing. In contrast, crypt stem cells are capable of self-renewal and asymmetric division. They give rise to the proliferative progenitor cells, also residing in the crypt compartment, which then become fully differentiated daughter cells as they migrate along the crypt-villus axis of the intestine ^{1, 2}.

Recent studies have illustrated the importance of several signaling pathways including the Wnt, Notch, Hedgehog, and bone morphogenetic protein pathways in controlling proliferation of intestinal stem cells and progenitor cells and their subsequent differentiation into different lineages of mature, differentiated epithelial cells ^3-5. Among these, the Wnt pathway is the best characterized. Wnt genes, of which that are 19 in humans, encode secreted glycoproteins ⁶. Upon binding to cell surface receptors such as Frizzled (Fz) and low-density lipoprotein receptor-related protein (LRP), Wnt elicits a signaling cascade that cumulates in the formation of nuclear β-catenin/T cell factor (TCF) complexes, which then drive cell proliferation ^7-9. It is well established that there is a Wnt gradient along the crypt-to-villus or crypt-to-surface epithelium axis of the small and large intestine, respectively, which provides important signals for proliferation of stem cells and progenitor cells in the crypt region ^{2, 3}. Additionally, Wnt signaling participates in intestinal epithelial cell maturation through regulating expression of the caudal related homologue Cdx1, which, along with Cdx2, promotes development of a mature intestinal phenotype ^{10, 11}. Wnt signaling also controls proper positioning of differentiated Paneth cells and enterocytes along the crypt-villus axis by reciprocally regulating expression of the EphB2/EphB3 receptors and their ligand ephrin B1 ^12-14.

The Krüppel-like factor 5 (KLF5) gene is one of over 20 mammalian genes which exhibit homology to the Drosophila Krüppel gene that is essential for embryogenesis ^15-18. In the intestine, expression of KLF5 is enriched in the proliferating crypt epithelial cells ^{19, 20}. Consistent with its cellular localization, KLF5 has been associated with, and in several instances, regulates, cellular proliferation in both physiological and pathophysiological conditions ^21-25. Until recently, a definitive assessment of KLF5 function in the intestinal epithelium has not been possible, as homozygous deletion of the Klf5 gene is embryonic lethal ²⁶. Here, we use the Cre-LoxP system to generate mice with conditional deletion of the Klf5 gene to investigate the physiological function of KLF5 in maintaining intestinal epithelial homeostasis.

Materials and Methods

Generation of mice with Intestinal-specific deletion of Klf5

C57BL/6 mice carrying Klf5 alleles flanked by loxP sites were described previously ²⁷. C57BL/6 mice carrying the Cre recombinase gene under regulation of the villin promoter (Vil-Cre mice) ²⁸ were obtained from Jackson Laboratories. Klf5^fl/fl mice were crossed with Vil-Cre mice to generate Vil-Cre;Klf5^fl/+ progeny. These mice were backcrossed to Klf5^fl/fl mice to produce Vil-Cre;Klf5^fl/fl offspring (designated Klf5^Δ/S). The Animal Care Committee of Emory University approved all procedures performed, and handling of the mice was in accordance with the Guide for the Care and Use of Laboratory Animals, published by the United States Public Health Service.

In vivo Permeability Assay

Gut permeability assays were performed to determine barrier function in control and mutant mice at 6 weeks of age. Mice were administered FITC dextran by gavage (60 mg/100 g body weight, FD-4, M 4000, Sigma Aldrich, St. Louis, MO) to determine the amount of dextran crossing the gut epithelial barrier and entering the bloodstream. Blood serum was collected 4 hours post-treatment. Levels of FITC dextran in the blood serum were measured by fluorescence intensity of the samples (excitation, 492, nm; emission, 525 nm; Cytofluor 2300). Results were normalized to the amount of serum protein assayed.

Myeloperoxidase (MPO) Activity

Neutrophil infiltration into tissue was quantified by measuring MPO enzyme activity (a marker for neutrophils). Briefly, colon tissues were washed to remove luminal contents and homogenized with a Tissuemizer (Fisher Scientific, Hampton, NH) in 0.5% hexadecyltrimethyl ammonium bromide (Sigma Aldrich) in 50 mM PBS, pH 6.0. Samples were subjected to three rounds of freeze/thaw and homogenized in the Bullet Blender with .1 mm zirconium silicate beads (Next Advance, Averill Park, NY). MPO activity was assayed from cleared supernatants by adding 1 mg/ml of dianisidine dihydrochloride (Sigma Aldrich) and 5 × 10⁻⁴% H₂O₂ (Sigma Aldrich) and measuring the change in optical density at 450 nm. Activity was normalized to total protein assayed.

Immunohistochemistry

Tissue isolated from the colon and small intestine (duodenum, jejunum, ileum) of 8-week-old mice were extracted, cut longitudinally and formed into Swiss rolls. Samples were prepared and subjected to immunohistochemical staining as described previously ²⁹. Immunofluorescence staining was performed by blocking with 3% BSA/PBS/Tween20 solution, incubating with primary antibody overnight at 4°C and incubation with Alexa fluor 488 or Alexa fluor 568 secondary antibodies (Invitrogen, Carlsbad, CA). Images were taken using Zeiss Axioskop2 plus microscope (Carl Zeiss Microimaging, Inc.). Areas of Klf5 deletion were verified in stained sections by co-staining with Klf5 antibody or by comparing with serial sections stained for Klf5.

Alcian blue/Periodic acid Schiff (PAS) staining

Alcian blue/PAS staining of paraffin-embedded sections to identify goblet cells was carried out using the Alcian Blue and PAS Staining Kit (Biocare Medical, Concord, CA) per manufacturer's instructions.

Western Blot Analysis

Protein lysates were prepared from distal ileum or proximal colon where deletion of Klf5 was most complete. Epithelial tissues were scraped into cell extraction buffer (Invitrogen) supplemented with Complete Protease Inhibitor Cocktail tablets (Roche, Indianapolis, IN). Samples were homogenized briefly with a Tissuemizer (Fisher Scientific) and further homogenized with 0.1 mm zirconium silicate beads in the Bullet Blender (Next Advance). Lysates were prepared in a 1:1 ratio with 2X Laemmli Buffer, subjected to polyacrylamide gel electrophoresis and transferred to nitrocellulose membrane. Membranes were blocked for 30 minutes in 5% non-fat dried milk in Tween/Tris-buffered salt solution (20mmol/L Tris-HCl, 3mmol/L KCl, 137 mmol/L NaCl, 0.1% Tween20, pH 7.5) and hybridized overnight at 4°C with primary antibodies.

Quantitative Real-time Polymerase Chain Reaction (qRT-PCR) Analysis

Total RNA from scraped colonic musosa of 3 control and 3 Klf5 mutant mice was prepared using Trizol reagent (Invitrogen) followed by DNase treatment and RNA cleanup with RNeasy columns (Qiagen, Valencia, CA). First strand synthesis of cDNA was performed with the RT² First Strand Kit per manufacturers' instructions (SA Biosciences, Frederick, MD). For analysis of Wnt pathway genes, PCR reactions were set up using the Wnt Signaling Pathway PCR Array plates (SA Biosciences) and analyzed with the manufacturer's software. Reactions were run on the Eppendorf Realplex PCR machine. Information on primers is included in Supplementary Methods.

Antibodies

See Supplementary Methods for a list of antibodies and dilutions used for Western blotting and immunohistochemistry.

Statistics

Numerical values in the figures are expressed as the mean ± S.E. Sample sizes are indicated in the figure legends. P values comparing the data for two groups of mice were calculated using a 2-tailed t test. Differences were considered to be significant if the P value was less than 0.05.

Results

To examine the consequences of intestinal-specific deletion of the Klf5 gene, we generated Vil-Cre;Klf5^fl/fl mice (designated Klf5^Δ/S for “intestinal-specific deletion”) by crossing mice carrying floxed alleles of the Klf5 gene (Klf5^fl/fl) with mice expressing the Cre recombinase under the control of a 12.4-kb regulatory region of the murine villin gene, followed by backcrossing to obtain the mutant mice. Klf5^Δ/S mice were born at a normal Mendelian ratio, but approximately two-thirds of these mice died within two days after birth. The remainder survived up to 8 weeks before dying or being sacrificed due to significant weight loss and rectal prolapse. Deletion of Klf5 in intestinal tissues was confirmed by immunohistochemical staining of Klf5. In mutant mice that died shortly after birth, Klf5 was deleted completely in the small intestine and colon (Figure 1A, panels a-d). Intestinal villi were sparse in newborn mutant mice compared to control mice, with the villus tips having a bulbous rather than tapered shape. In the colon, tissue architecture was distorted, with no identifiable crypt structures. To examine the effect of intestinal-specific deletion of Klf5 on epithelial proliferation in the neonatal gut, tissues from control and Klf5^Δ/S mice were stained for expression of the proliferation marker, Ki67 (Figure 1A, panels e-h). Whereas proliferating cells were abundant in the crypts of control mice, they were virtually absent in corresponding areas in the mutant mice in both the small and large intestine.

Figure 1.

Deletion of Klf5 from the intestinal epithelium of moribund newborn mice. Immunohistochemical staining of Klf5 (A) and Ki67 (B) was performed on tissues from the small intestine and colon of control (Klf5^fl/fl) and Klf5^Δ/S (Vil-Cre;Klf5^fl/fl) newborn mice one day after birth.

In examining Klf5 expression in surviving mice at 8 weeks of age, small and large intestines from Klf5^fl/fl control mice and Klf5 mutant mice were prepared as Swiss rolls and stained for Klf5 (Figure2A). Adult control mice showed Klf5 localized to epithelial cells in proliferating regions of intestinal crypts. In contrast, adult Klf5^Δ/S mice exhibited patchy expression of Klf5 in segments of the small and large intestine, indicating incomplete deletion of the Klf5 gene from intestinal stem cells. Klf5 was deleted in approximately 50% of small intestinal tissues and in 80% of colon tissues (Figure 2B).

Figure 2.

Variegated deletion of Klf5 and evidence of a regenerative phenotype in Klf5 mutant mice. (A) Immunohistochemical staining of Klf5 in Swiss rolls from the small intestine and colon of mice at 8 weeks of age. (Red arrows: Klf5-positive crypts; black arrows: Klf5-negative crypts). (B) Average percent deletion of Klf5 in Swiss rolls of duodenum, jejunum, ileum and colon stained for Klf5 protein, n=5. (C) Histolology in areas of Klf5 deletion. Serial sections from the small intestine and colon of adult control and Klf5^Δ/S mice were stained with Klf5 antibody or H&E.

We next examined morphology of the small and large intestines in control and surviving Klf5 mutant mice. Klf5 expression in areas of representative histology was confirmed by staining serial sections with Klf5 antibody (Figure 2C). Hematoxylin and eosin (H&E) staining indicated that normal crypt architecture was disrupted throughout the gut of the mutant mice. In the small bowel, villi were shortened and crypts were elongated, with some areas exhibiting erosion at the villus tips. In the colon, crypts were severely distorted, with dilated glands and reduced numbers of goblet cells. Somewhat surprisingly, mutant mice displayed active proliferation, with increased crypt lengths and crypt densities in the small intestine and increased crypt fission and proliferation in the colon (Supplementary Figure 1). These results were in contrast to those in newborn mice which showed absence of epithelial proliferation. Altered morphologies in the adult Klf5^Δ/S mice are consistent with a regenerative phenotype.

The average lifespan of Klf5^Δ/S mice with variegated Klf5 expression was approximately 8 weeks, with the mice succumbing to severe wasting or being sacrificed due to rectal prolapse (Figure 3 A and B). Histological analysis of mutant mice with rectal prolapse revealed cryptitis, with infiltration of neutrophils into the epithelium and lamina propria of the small and large intestines, disruption of crypt architecture in the small intestine, and the presence of neutrophil exudate in the glands of the colon (Figure 3C). Neutrophil infiltration was also examined in Klf5 mutant mice by myeloperoxidase (MPO) activity, a lysosomal enzyme abundant in neutrophils (Figure 3D). MPO activity was significantly higher in Klf5 mutant mice than in control mice at 6 weeks of age, prior to the development of rectal prolapse. To determine whether inflammation was associated with epithelial barrier defects, barrier function was measured in control and mutant mice at 6 weeks of age by the FITC dextran method (Figure 3E). Results indicated a significantly increased permeability across the intestinal epithelium of Klf5^Δ/S mice, corresponding to increased MPO activity.

Figure 3.

Wasting and enterocolitis in Klf5^Δ/S mice. (A) Comparison of size difference of control and Klf5^Δ/S littermates at 8 weeks of age. (B) Rectal prolapse is frequently observed in Klf5^Δ/S mice by approximately 8 weeks of age. (C) H&E staining indicates infiltration of immune cells and disruption of normal crypt architecture in the small intestine and colon of Klf5^Δ/S mice with rectal prolapse. (D) Measurement of neutrophil infiltration by myeloperoxidase assay using colon tissue from control and Klf5^Δ/S mice at 6 weeks of age (*P < 0.001, n=7). (E) Measurement of gut permeability of mice at 6 weeks of age with the FITC dextran permeability assay (**P = 0.05, n=7).

Accompanying the perturbed crypt architecture and barrier defects in Klf5 mutant mice were alterations in intestinal epithelial differentiation, with aberrant localization of differentiated cells. In control mice, immunohistochemical staining for carbonic anhydrase-I (CA-I), a colonic marker for epithelial differentiation ³⁰, showed CA-1 restricted to epithelial cells in the upper third of the colonic crypt (Figure 4A). However, in Klf5 mutant mice, CA-I expression was seen throughout the epithelial mucosa. Disruption of normal patterns of terminal differentiation was associated with reduced mRNA expression of key regulators of differentiation in the small and large intestine, Cdx1 and Cdx2 ¹⁰ (Figure 4B).

Figure 4.

Loss of a restricted differentiation zone and changes in epithelial cell lineage with deletion of Klf5. Paraffin-embedded colon sections from adult control and Klf5 mutant mice were stained with various epithelial markers to identify changes in differentiation. (A) Immunohistochemical staining with the colonic differentiation marker, carbonic anhydrase-I. (B) Quantitative real time polymerase chain reaction (qRT-PCR) to measure relative mRNA expression of Cdx1 and Cdx2 in colon tissues from control and mutant mice (*P < 0.001, **P = 0.004, n=3). (C) Alcian blue/Periodic acid-Schiff (PAS) staining to identify goblet cells and immunohistochemical staining of chromogranin A as a marker for enteroendocrine cells.

Epithelial cell lineages also appeared to be altered in the Klf5^Δ/S mice. Numbers of goblet cells were reduced in the colon of mutant mice as determined by Alcian blue/Periodic acid-Schiff staining (Figure 4C). A reduction in goblet cell number was accompanied by a reciprocal increase in the number of enteroendocrine cells, as indicated by chromogranin A staining (Figure 4D). In the small intestine, no significant changes in epithelial cell lineage decisions were noted in mutant mice compared to control mice. Numbers of goblet cells and enteroendocrine cells as a percentage of total epithelial cells were comparable between the two groups (data not shown).

Paneth cell numbers and distribution in the small intestine of mutant and control mice were examined by staining for the Paneth cell marker, lysozyme ³¹. While numbers of Paneth cells per crypt were similar in control and mutant mice, the localization of Paneth cells within the crypts was altered in Klf5 mutant mice. In control mice, Paneth cells were anchored at the very base of the crypts (Figure 5A, panel a). However, in mutant mice, some of the lysozyme-positive cells were no longer retained at the crypt base and could be found at higher positions in the crypts (Figure 5A, panels b,c). Mispositioning of cells was also observed for actively proliferating epithelial cells in Klf5 mutant mice. Whereas cells stained positive for the proliferation marker, Ki67, were localized to the base of the crypts in the small and large intestines of control mice (Figure 5B, panels d,g), Ki67-positive proliferating cells were intermingled with non-proliferating cells throughout the intestinal crypts of mutant mice (Figure 5B, panels e, f, h and i). These results suggest that one of the physiological functions of KLF5 is to control cell migration and position along the crypt/villus axis.

Figure 5.

Impaired migration of epithelial cells in Klf5 mutant mice. Tissues from adult control and Klf5^Δ/S mice were examined for alterations in epithelial cell migration. (A) Immunohistochemical staining of the small intestine with lysozyme as a Paneth cell-specific marker. Right panel shows a higher magnification of Klf5^Δ/S staining. The red arrowhead indicates a Paneth cell that has migrated from the crypt base. (B) Immunohistochemical staining with the proliferation marker, Ki67, in the small intestine and colon. Red dotted lines indicate the normal boundaries between proliferative and differentiated compartments. Right panels are higher magnification images of adjacent panels. Green arrowheads indicate Ki67-positive proliferating cells. Red arrowheads indicate Ki67-negative non-cycling cells.

The scattered distribution of Paneth cells and Ki67-positive cells in the intestines of Klf5 mutant mice is reminiscent of the phenotype observed in mice deficient for the EphB2/EphB3 receptors ^12,13. These receptors are normally expressed in the intestinal crypts. Together with their ligand, ephrin B1 which is expressed in the villi, EphB2/EphB3 regulate positioning of cells in the intestinal epithelium ^{12, 14}. We therefore performed immunohistochemical staining and Western blot analysis for EphB2, EpB3 and ephrin B1 in intestinal tissues from Klf5 mutant and control mice. Deletion of Klf5 in regions of immunohistochemical staining was verified by co-staining for Klf5 or staining serial sections for Klf5 (Supplementary Figure 2). In examining expression of EphB2 and EphB3 in the colon, both receptors were confined to cells at the base of the crypts in control mice (Figure 6A). In the mutant mice, however, EphB2 exhibited a more diffuse staining throughout the crypt, with expression levels only slightly reduced compared to those in control mice (Figure 6A and B). In contrast, EphB3 expression was not detectable in the colon of mutant mice by Western blot or immunohistochemistry. Likewise, ephrin B1, which is normally expressed in cells in the upper region of the colonic epithelium, was undetectable in mutant mice. Similar although less dramatic changes in EphB3 and ephrin B1 expression were seen in the small intestine (Supplementary Figure 3). Thus, KLF5 regulates cell positioning in the gut epithelium by controlling expression of EphB2/EphB3 receptors and the ephrin B1 ligand.

Figure 6.

Disrupted expression of Eph/ephrin proteins in Klf5^ΔIS mice. (A) Immunofluorescence staining of EphB2, EphB3 and ephrin B1 in colon tissues from control and Klf5 mutant mice. White arrows indicate expected localization of the Eph/ephrin proteins. (B) Western blot analysis of Eph/ephrin levels in colon lysates from control and Klf5 mutant mice, n=3.

KLF5 has been reported to promote β-catenin activity through its effects on β-catenin nuclear localization ²⁹. Since differentiation and cell positioning are regulated by Wnt/ β-catenin signaling in the intestine, we investigated whether deletion of Klf5 affects Wnt pathway activation. Immunohistochemical staining of β-catenin revealed reduced nuclear localization at the base of colonic crypts in Klf5 mutant mice compared to controls (Figure 7A). Klf5 expression in regions stained for β-catenin was verified by immunohistochemical staining of Klf5 in serial sections (Supplementary Figure 4.) Levels of β-catenin in nuclear lysates from mucosal scrapings were also reduced in Klf5^Δ/S mice (Figure 7B). Using a quantitative PCR array for analysis of Wnt pathway genes, numerous Wnt-related genes were found to be downregulated in colonic tissues of Klf5^Δ/S mice (Figure7C). Genes with reduced expression included direct targets of Wnt signaling (Ccnd1, Fgf4, Jun, Sox17) as well as genes encoding receptors and ligands for Wnt signaling (Dvl2, Fzd4, Fzd7, Lrp5, Wnt2, Wnt4) and proteins involved in Wnt activation (Lef1 and Pygo1). Thus, deletion of Klf5 results in reduced activation of β-catenin and inhibition of Wnt signaling in the intestine.

Figure 7.

Impaired Wnt signaling and induction of Sox9 in colonic tissues of Klf5^ΔIS mice. (A) Colonic tissues of control and mutant mice were stained by immunohistochemistry to examine localization of β-catenin. Red arrows indicate the presence or absence of nuclear localization of β-catenin in control and mutant mice, respectively. (B) Nuclear lysates prepared from colons of 3 control and 3 mutant mice were pooled and subjected to Western blotting to determine levels of nuclear-localized β-catenin, n=3. Tata-binding protein (TBP) is shown as a nuclear protein loading control. (C) Wnt pathway PCR array analysis was conducted on RNA samples from control and Klf5^Δ/S mice to identify differentially expressed mRNAs. RNA samples were pooled for analysis, n=3. Fold changes in parentheses indicate a fold decrease.

To identify mechanisms promoting colonic epithelial proliferation in regions with no Klf5 expression and reduced Wnt signaling, several candidate pathways and molecules were examined. Canonical TCF/β-catenin targets, cyclin D1 and c-Myc, were reduced or unchanged, respectively, in areas of Klf5 deletion (Supplementary Figure 5). Notch signaling was likewise modestly reduced in colonic tissues of Klf5 mutant mice (Supplementary Figure 6). However, expression of Sry (sex determining region of Y)-box 9 (Sox9), a transcription factor that participates in decisions of self-renewal and differentiation ^{32, 33}, was increased approximately four-fold in colonic tissues of Klf5 mutant mice (Figure 7D and E). Staining of Sox9 in Klf5 mutant mice revealed high levels throughout the regenerative glands. Thus, Sox9 may mediate aspects of the regenerative phenotype in intestinal tissues of Klf5^Δ/S mice.

Discussion

The results presented here demonstrate that Klf5 plays a critical role in homeostasis of the intestinal epithelium, with loss of Klf5 affecting mucosal crypt architecture and barrier function. Deletion of Klf5 from the gut epithelium culminates in a regenerative phenotype that is characteristic of wound repair. This phenotype is likely a compensatory response to changes in the mucosal epithelium that are associated with reduced Wnt signaling. Interestingly, changes in tissue architecture with deletion of Klf5 are more severe in the colon than in the small intestine. This may be due to a more complete deletion of Klf5 in colonic tissues (Figure 2B) or to Klf5 playing a more significant role in maintenance of epithelial homeostasis in the colon.

Our results indicate that Klf5 is not essential to maintain intestinal epithelial proliferation in adult mice, in contrast to its requirement for proliferation in newborn mice. This finding may be attributable to regenerative processes in adult intestinal tissues that are independent of canonical Wnt signaling. In adult Klf5 mutant mice, epithelial cell proliferation was observed in regenerative crypts where KLF5 is deleted and Wnt signaling is reduced. Concomitant with these changes was increased expression of the progenitor cell marker, Sox9. Sox9 is linked to progenitor status in adult liver, exocrine pancreas and intestine, and is proposed to maintain the precursor cell population during physiological cell replacement and/or regeneration after injury ³⁴. Thus, Sox9 may promote proliferation in Klf5 mutant mice by maintaining epithelial cells in a progenitor state.

In regard to changes in tissue morphology, we show a direct effect of KLF5 deletion on epithelial differentiation within the gut epithelium. Defects in terminal differentiation along the crypt-luminal axis are apparent in the colon of Klf5 mutant mice stained with the differentiation marker, carbonic anhydrase-I. Normal expression of CA-I is limited to colonocytes in the upper one third of the crypt ³⁰; however, in KLF5^Δ/S mice, CA-I-expressing cells are scattered throughout the crypt, indicating that terminal differentiation is no longer restricted to the luminal surface. A similar effect is seen with the differentiation marker alkaline phosphatase in mice with transgenic intestinal-specific expression of the Wnt inhibitor Dkk1 ³⁵. Perturbations in terminal differentiation in Klf5 mutant mice were associated with reduced expression of the homeobox transcription factors, Cdx1 and Cdx 2. These factors are reported to stimulate differentiation and inhibit proliferation of intestinal epithelial cells ^{10, 11}. Cdx1 is a transcriptional target of TCF4/β-catenin ³⁶, while Cdx2 is repressed by Sox9 ³⁷. In accordance with reports on regulation of Cdx1 and Cdx2, reduced expression of these factors in Klf5 mutant mice correlates with impaired Wnt signaling and increased Sox9 expression. In examining epithelial lineage decisions, Klf5^Δ/S mice exhibit changes in the prevalence of secretory cells, with fewer goblet cells and increased numbers of enteroendocrine cells in the colon compared to controls. These effects are similar to those seen with deletion of the transcriptional repressor, Gfi1, which drives the selection of goblet and Paneth cells over enteroendorcrine progenitors ³⁸.

Defects in epithelial migration in Klf5 mutant mice are reminiscent of mice with disrupted Eph/ephrin signaling, correlating with altered expression of Eph/ephrin proteins in the mutant mice. In the intestine of EphB3-null mice, localization of Paneth cells is no longer restricted to the crypt base; rather, Paneth cells migrate upwards and are found dispersed throughout the epithelium ¹². Mislocalization of Paneth cells is also observed in mice with intestinal specific deletion of ephrin B1 ³⁹. Other architectural changes in Klf5^Δ/S mice are similar to those reported in EphB2^-/-;EphB3^-/- double knockout mice including intermingling of proliferative and differentiated cells ¹². Future studies will examine whether genes encoding EphB3 or ephrinB1 are direct transcriptional targets of KLF5.

Reduced expression of Eph/ephrin proteins and Cdx1 in Klf5^Δ/S mice indicates that Wnt signaling is inhibited in the mutant mice. Absence of epithelial proliferation in the intestines of Klf5^Δ/S newborn mice (Figure 1B) phenocopies genetic models for disruption of Wnt signaling that include deletion of Tcf4 and transgenic expression of the Wnt inhibitor Dkk-1 ^{35, 40-42}. In these models, the neonatal epithelium was composed entirely of differentiated, non-dividing villus cells, resulting in loss of intestinal stem cell compartments and neonatal lethality. In surviving Klf5^Δ/S mice, we observe downregulation of numerous Wnt target genes, corresponding to reduced expression of active β-catenin at the base of the crypts (Figure 7). Previous evidence from our laboratory demonstrates that changes in expression of Klf5 have a significant impact on β-catenin localization and activity ²⁹. Haploinsufficiency of Klf5 in the Apc^Min/+ mouse rescues the tumor-initiating effects of the Apc^Min mutation by inhibiting β-catenin nuclear accumulation. Furthermore, downregulation of KLF5 in colon cancer cell lines inhibits nuclear localization of β-catenin and reduces expression of TCF/ β-catenin target genes ²⁹. Inhibition of tumor formation by heterozygous deletion of Klf5 is also reported in mice with combined KRas and Apc^Min mutations, indicating a critical role for KLF5 in proliferative and transforming events associated with intestinal tumorigenesis ⁴³.

Homeostasis of the intestinal epithelium is dependent on tight regulation of pathways that regulate cell proliferation, differentiation and migration. An unexpected finding of our work is the necessity of intestinal-specific expression of KLF5 for viability. By targeting expression of Klf5, we have identified a novel effector of Wnt signaling that is critical for maintenance of proper crypt organization and structure.

Abbreviations

Fz

Frizzled

LRP

lipoprotein receptor-related protein

KLF5

Krüppel-like factor 5

MPO

myeloperoxidase

PAS

Periodic acid-Schiff

Sox9

Sry (sex determining region of Y)-box 9

TCF

T cell factor