Dysregulation of Kruppel-like factor 4 during brain development leads to hydrocephalus in mice

Abstract

Kruppel-like factor 4 (KLF4) is involved in self-renewal of embryonic stem cells and reprogramming of somatic cells to pluripotency. However, its role in lineage-committed stem cells remains largely unknown. Here, we show that KLF4 is expressed in neural stem cells (NSCs) and is down-regulated during neuronal differentiation. Unexpectedly, enhanced expression of KLF4 reduces self-renewal of cultured NSCs and inhibits proliferation of subventricular neural precursors in transgenic mice. Mice with increased KLF4 in NSCs and NSCs-derived ependymal cells developed hydrocephalus-like characteristics, including enlarged ventricles, thinned cortex, agenesis of the corpus callosum, and significantly reduced subcommissural organ. These characteristics were accompanied by elevation of GFAP expression and astrocyte hypertrophy. The ventricular cilia, vital for cerebrospinal fluid flow, are also disrupted in the mutant mice. These results indicate that down-regulation of KLF4 is critical for neural development and its dysregulation may lead to hydrocephalus.

Kruppel-like factor 4 (KLF4) is a member of the KLF family of transcription factors (1, 2). It was first identified in the epithelia of the gut and skin, where it is involved in their growth and differentiation (3, 4). Mice with a Klf4 deletion die within 15 h of birth, presumably due to defects in terminal differentiation of epithelial tissues, such as the epidermis and colon (1, 5). Although KLF factors either activate or repress gene expression depending on the cellular context (2), transcriptional profiling reveals a global inhibitory function of KLF4 on macromolecular biosynthesis and the cell cycle in a human colon cancer cell line (6, 7). Recent studies further demonstrate that KLF4 positively regulates self-renewal of embryonic stem cells (8–10). Very importantly, it is one of the original four factors sufficient to reprogram somatic cells into induced pluripotent stem cells (iPS) (11) through direct interaction with OCT4 and SOX2 (12).

Besides its role in embryonic stem cells and iPS, KLF4 was recently demonstrated to function as a repressor of axonal regeneration in retinal ganglion cells or other neurons in the central nervous system (13). However, the function of KLF4 in lineage-committed neural stem cells (NSCs) remains unknown. Our previous microarray analysis showed that Klf4 is significantly up-regulated in NSCs upon deletion of nuclear receptor TLX, which plays an important role in controlling adult NSCs and neurogenesis (14). Such up-regulation of Klf4 in TLX-deleted NSCs was confirmed by RT-PCR (Fig. S1). To address the role of such dysregulated expression in NSCs, we made a mouse line overexpressing KLF4 during brain development. Contrary to its positive role in embryonic stem cells, overexpression of KLF4 reduces self-renewal of neural progenitors and causes hydrocephalus in these transgenic mice.

Hydrocephalus is one of the most common anomalies affecting the nervous system, occurring with an estimated incidence of 1 in 1,000 live births (15). Homeostasis of the cerebrospinal fluid (CSF) in the ventricles is critical for brain development. Accumulation of CSF, caused by either impaired flow, excess production by the choroid plexus or a lack of reabsorption, can result in hydrocephalus (16). Although the precise molecular mechanism for this brain defect is not yet clear, multiple transcriptional regulators have been implicated. These factors include Engrailed 1, Msx1, E2F5, RFX3, RFX4, Foxj1/Hfh-4, and polymerase λ (17). Based on our current data, KLF4 can now be added to this growing list of factors whose dysregulation may lead to hydrocephalus.

Results

Inhibition of Self-Renewal and Differentiation of NSCs by Ectopic KLF4.

RNA in situ hybridization showed Klf4 is expressed in neuroepithelium of the developing mouse forebrain (Fig. S2 A–C). Its expression was further examined by Western blotting analysis. Antibody specificity was verified by antigen absorption using purified KLF4 protein (Fig. S2 D and E). KLF4 has a higher level of expression in embryonic brain, but is gradually down-regulated postnatally (Fig. 1A). Its cell type specificity was examined by using purified cells, including NSCs from embryonic day 14.5 (E14.5) forebrains, primary neurons from E15.5 cortices, and astrocytes from cortices at birth (P0). These cell types were confirmed by morphology and gene expression, such as Nestin for stem cells, Tuj1 for neurons, and GFAP for astrocytes (Fig. 1B and Fig. S2F). We found that KLF4 is expressed in NSCs, but is down-regulated by 90% in neurons and by 40% in astrocytes (Fig. 1B).

Fig. 1.

Enhanced expression of KLF4 inhibits self-renewal and neuronal differentiation of NSCs. (A) Expression of KLF4 in mouse brains by Western blotting analysis. β-actin was used as a loading control. (B) Down-regulation of KLF4 in neurons. Protein lysates from cultured NSCs, neurons, and astrocytes were used for Western blotting. Tuj1, GFAP, Nestin, and β-actin were used as controls. (C) Reduced self-renewal of NSCs by ectopic KLF4. GFP was used a control. The diameter and the number of neurospheres were quantified (n = 3; *P < 0.001). (D) Inhibition of neuronal differentiation of cultured NSCs. The percentage of neurons (Tuj1⁺) and the length of processes of transfected cells were quantified (n = 3; *P < 0.0001).

Previous reports showed that KLF4 contributes to the maintenance of embryonic stem cell self-renewal and pluripotency (8–10). Expression of KLF4 in NSCs raised the possibility that KLF4 may positively regulate growth and self-renewal of NSCs. To test this hypothesis, we infected cultured neurospheres from E13.5 forebrains with lentiviruses expressing either GFP or KLF4-ires-GFP under the hGfap promoter, which drives gene expression in NSCs and astrocytes (18). Contrary to our hypothesis, exogenous KLF4 inhibited both the growth and self-renewal of cultured NSCs, as seen by a 51% and 39% reduction in the diameter and the number of neurospheres, respectively (Fig. 1C). Cell cycle analysis by DNA content showed that ectopic KLF4 resulted in a 39% reduction of cells in the S-phase (18.5% and 11.3% for GFP- and KLF4-ires-GFP-expressing cells, respectively), with a compensatory 18.6% increase of cells in the G0/G1 phase (52.7% and 62.5% for GFP- and KLF4-ires-GFP-expressing cells, respectively). Staining for SOX2, a marker for NSCs, did not detect a significant difference between these two groups, suggesting that KLF4 overexpression did not alter stem cell identity (Fig. S3). To examine the role of KLF4 in neural differentiation, the Klf4-ires-Gfp plasmid under the control of a constitutive CAG promoter was electroporated into E13.5 NSCs. After 3 d in differentiation conditions (by withdrawing growth factors), about 25% of control GFP-transfected NSCs differentiated into Tuj1⁺ neurons with long neurites. However, KLF4-overexpressing NSCs resisted neuronal differentiation, showing rounded morphology with very short processes (Fig. 1D). Apoptosis of these cells was not detected because they were negative for expression of activated caspase-3 (Fig. S3C). These findings suggest that precise regulation of the level of KLF4 is critical to both proliferation of NSCs and their neuronal differentiation.

Impaired Proliferation and Neurogenesis in KLF4 Transgenic Mice.

To examine the role of dysregulated KLF4 in vivo, we generated transgenic mice by crossing TRE-myc-Klf4 (19) with hGfap-tTA (20). In the absence of tetracycline, myc-tagged KLF4 is expressed under the control of the hGfap promoter, which drives gene expression in NSCs and astrocytes during embryogenesis (21). We confirmed gene expression by immunohistochemistry and confocal microscopy using an antibody against the myc epitope. Myc-KLF4 started with restricted expression in the primordial hippocampus at E13.5 and spread to the medial and dorsal regions of the lateral ventricular zones at later developmental stages (Fig. 2A). These cells coexpressed the stem cell marker Nestin but rarely expressed the neuronal marker Tuj1 (Fig. 2A and Fig. S4A). Expression of myc-KLF4 persisted into postnatal stages but was dramatically down-regulated in forebrain (Fig. S4B). Although we found that a subset of myc-KLF4-labeled cells also expressed GFAP in the retrosplenial cortex–hippocampus transition at P1, the majority of myc-positive cells also expressed Nestin (Fig. S4C), suggesting that ectopic KLF4 was mainly restricted to neural progenitors. It should be noted that myc-KLF4 was rarely expressed in the lateral ganglionic eminence (LGE) during all of the developmental stages examined (Fig. 2A and Fig. S4B).

Fig. 2.

Ectopic KLF4 reduces proliferation and neurogenesis of neural progenitors in vivo. (A) Expression of myc-tagged KLF4 is largely in Nestin⁺-progenitor cells in transgenic mice. Higher magnification views were taken from the boxed regions (Tg, transgenic mice; LV, lateral ventricle; LGE, lateral ganglionic eminence; Nes, Nestin). (B) Decreased proliferation of progenitors in transgenic mice examined at E17.5. Proliferating cells were pulse-labeled with BrdU (DV, dorsal ventricular cortex; n = 5; *P < 0.001). (C) Reduction of neurogenesis examined at E17.5 by staining for TBR2, a marker for intermediate neuronal progenitors. TBR2⁺ cells in transgenic mice are largely restricted to the subventricular zone (SVZ), whereas these cells are located in a broader region in the control mice (IZ, intermediate zone).

To detect proliferation, cells were pulse-labeled with bromodeoxyuridine (BrdU) at E16.5 and examined 12 h later. No difference was seen between control and Tg mice in BrdU labeling of the LGE, where ectopic myc-KLF4 was not observed (Fig. 2B). However, a significant reduction of BrdU⁺ cells (42% decrease, n = 5, P < 0.001) was observed in regions where myc-KLF4 was highly expressed, such as the dorsal ventricle (Fig. 2B). Staining for Ki67, an endogenous marker for proliferating cells, also showed a 48% reduction of labeled cells (n = 5, P < 0.0001) in the region with higher myc-KLF4 expression (Fig. S5 A and B). Intermediate neuronal progenitors were examined at E17.5 by staining for TBR2, a T-domain transcription factor that is enriched in these cells (22). In control brains, TBR2⁺ cells were located in both the subventricular zone (SVZ) and the intermediate zone (IZ) of the dorsal cortex (Fig. 2C and Fig. S5C). However, these cells were rarely detected in the IZ in the same region of KLF4-overexpressing mice, with a 31% reduction of Tbr2⁺ cell density (2,351 and 1,614 TBR2⁺ cells per mm² for control and Tg mice, respectively; n = 5, P < 0.0001). Because of lacking compatible antibodies, immunohistochemistry was combined with RNA in situ hybridization to examine the colocalization of TBR2 and Klf4. This analysis showed that cells with higher levels of Klf4 expression did not express TBR2, suggesting a cell autonomous effect of KLF4 in inhibition of neurogenesis, which is consistent with cell culture studies (Fig. 1D and Fig. S5D). On the other hand, staining for activated caspase-3 in the SVZ, IZ, or other cortical regions found no difference (Fig. S6 A and B). These data show that enhanced expression of KLF4 in NSCs inhibited proliferation and neurogenic ability of NSCs.

Dysregulation of KLF4 Leads to Hydrocephalus and Astrocytosis.

Transgenic mice overexpressing KLF4 were born with the expected Mendelian ratio. However, all KLF4-overexpressing mice were growth retarded and had a hairless phenotype (Fig. 3A). By weaning age, most of these mutant mice had enlarged heads (Fig. 3B). Gross analysis of each brain showed compression of the cerebrum at P20, suggestive of increased intracranial pressure. Upon dissection, excess CSF was observed in enlarged ventricles, indicating a hydrocephalus phenotype. This phenotype occurred at varying degrees to all transgenic mice overexpressing myc-KLF4.

Fig. 3.

Mice with enhanced expression of KLF4 develop hydrocephalus. (A) Hairless phenotype of KLF4-transgenic mice examined at P20 (Tg, transgenic mice). (B) Cerebral compression indicated by arrows at P20. (C) Histology showing variable degrees of the hydrocephalus phenotype observed in transgenic mice (Tg) at P14. (D) Nissl-stain of P1 brain showing cell density. Cortical layers (I-VI) are labeled. (E) Astrocytes were examined by staining for GFAP in the cortex (boxed regions) of postnatal day (P) 7 and P14 mice. Lateral ventricle (LV) is outlined.

To characterize the hydrocephalus phenotype, serial brain sections from P14 mice (n = 4 for each genotype) were stained with hematoxylin and eosin (H&E). KLF4-transgenic mice displayed enlarged lateral ventricles, agenesis of the corpus callosum, rudimentary hippocampus, and thinner cerebral cortex compared with control mice (Fig. 3C). In some severe cases, total loss of the septum and the hippocampal formation was seen in transgenic mice. This hydrocephalus phenotype can be detected as early as P1. Serial coronal sections of P1 brains (n = 3 for each genotype) showed that KLF4 overexpression caused dilatation of the lateral ventricles and disruption of the hippocampal formation. Nissl-staining showed that neuronal density in the cerebral cortices of the mutant mice were also reduced compared with their wild-type littermates (Fig. 3D). Accompanying neuronal loss, massive GFAP-positive astrocytes with hypertrophic morphology were detected in the cortical regions at later stages (Fig. 3E, P7 and P14). Some of these cells coexpressed S100β (a marker for mature astrocytes) and SOX2 (a marker for stem cells and activated astrocytes; ref. 23; Fig. S6C). On the other hand, immunohistochemistry analysis did not show myc-KLF4 expression in these GFAP⁺ cells in the cortex, indicating a reactive gliosis response to hydrocephalus (24).

Because the transgene is under the control of the hGfap promoter, a potential role of KLF4 in differentiated astrocytes may lead to the hydrocephalus. Overexpression of KLF4 in cultured astrocytes, however, did not exert any observable effect on either cellular morphology or proliferation (Fig. S7). Gliogenesis mainly occurs during late embryonic and early postnatal development (25). The transgenic expression of KLF4 was turned off by feeding mice with doxycycline (Dox) in drinking water from E17.5 until analysis (P7). Immunostaining confirmed that KLF4 was indeed not expressed in these Dox-treated mice. However, these transgenic mice still developed hydrocephalus (Fig. S8A), suggesting that ectopic KLF4 in early development is sufficient to induce hydrocephalus. To test this hypothesis, we treated pregnant mice with Dox from E8.5 to E17.5 to turn off myc-KLF4 expression, which was confirmed by immunohistochemistry analysis at E17.5. As expected, such treatment completely prevented hydrocephalus (Fig. S8B). These data suggest that dysregulated KLF4 in NSCs but not in astrocytes results in malformation of the brain.

Reduced Subcommissural Organ (SCO) and Disrupted Ependymal Cilia in KLF4-Overexpressing Mice.

Although the third ventricle and the aqueduct appeared normal, the SCO, which may participate in the circulation and reabsorption of CSF (26), was reduced in size by P1 (Fig. 4A). The SCO is composed of ependymal cells, which secrete glycoproteins to form a threadlike structure called Reissner's fiber (RF). The SCO-RF structure in control littermates was clearly identified by staining for CD133, a transmembrane glycoprotein in ependymal cells (Fig. 4B, WT). Even though the ventricular wall was still positive for CD133-staining in KLF4-transgenic mice, the SCO-RF structure was almost nonvisible (Fig. 4B, Tg). Staining for exogenous KLF4 at an earlier stage (P0) showed that it was expressed in the presumed SCO region, suggesting that reduction of the SCO-RF structure may be a cell-autonomous effect of KLF4 overexpression (Fig. 4B, myc-KLF4).

Fig. 4.

Defective SCO and ependymal cilia by overexpression of KLF4. (A) Nissl-stain showing a greatly reduced size of the SCO in transgenic mice (Tg) at P1. Third ventricle (III) is labeled. (B) CD133-staining reveals the SCO-RF complex at P1. The expression of ectopic KLF4 in Tg is shown by staining for myc-tag at P0. (C) Ependymal cilia examined at P1 and P7 by staining for acetylated α-tubulin (Ac-Tub). Note the filamentous protrusion in the control but not in the Tg mice. (D) Quantification of cells with obvious cilia by confocal microscopy (n = 5; *P < 0.0001). (E) Scanning electron microscopy showing the fine morphology of multicilia at P7. (F) Normal cilia on the epithelia of the choroid plexus. (G and H) Multicilia examined at P7 by Ac-Tub staining after turning off the transgene expression at either E8.5-E17.5 or E17.5-P7 (n = 5; *P < 0.0001).

Cilia are flagella-like projections from the cell that are made up of microtubules. Coordinated beating of ependymal cilia is thought to facilitate CSF flow. Defects in cilia structure and activity are associated with hydrocephalus in humans (27, 28) and mouse models (29, 30). Cilia can be identified by staining with an antibody against acetylated α-tubulin. Serial brain sections from P1 and P7 brains were examined with immunohistochemistry and confocal microscopy analysis. Bundles of motile cilia were observed to protrude into the lateral ventricle (LV) lumen in control littermates at both stages examined (Fig. 4C, WT). Although the microtubules in KLF4-transgenic mice could still be identified by staining for acetylated α-tubulin, there were few cilia on ependymal cells lining the LV. If any, they were much shorter and disorganized (Fig. 4 C and D, Tg). This was confirmed by scanning electron microscopy, showing a few stubs of fused cilia on the ependymal cells of transgenic mice (Fig. 4E). In contrast, no apparent difference was observed regarding cilia on epithelia of the choroid plexus (Fig. 4F).

Ependymal cells are derived from NSCs during later embryogenesis and become mature and develop cilia in the first postnatal week (31). To examine the time course of exogenous KLF4-induced dysfunction of ependymal cells, we turned off the transgene expression by Dox treatment for different periods and analyzed cilia formation at P7. No difference on cilia morphology or number was observed when transgene was turned off from E8.5 to E17.5 (Fig. 4 G and H). In contrast, cilia were severely deformed when exogenous KLF4 was shut down from E17.5 to P7, indicating that early expression of KLF4 in NSCs is sufficient to result in malformed ependymal cells and cilia.

Discussion

KLF4 is emerging as an important player in multiple physiological and pathological processes, ranging from stem cell maintenance, cellular reprogramming, skin development, to axonal regeneration (1, 11, 13, 32). Our current study shows that KLF4 is down-regulated during brain development and in differentiated neurons. In contrast to its positive role in promoting self-renewal of embryonic stem cells, overexpression of KLF4 in NSCs inhibits their proliferation and differentiation. Importantly, dysregulated KLF4 in NSCs and NSCs-derived ependymal cells during development leads to hydrocephalus in transgenic mice by disrupting the SCO-RF structure and the ependymal cilia. These data suggest that precise regulation of KLF4 levels is critical to brain development and the maintenance of CSF homeostasis.

A previous study demonstrated that ectopic expression of Ro1, a G(i)-coupled receptor activated solely by a synthetic ligand, in astrocytes leads to hydrocephalus by P15 (33). Using the same “tet-off” system, our detailed analysis showed that exogenous KLF4 is mainly expressed in Nestin-positive NSCs within the ventricular, subventricular, and intermediate zones of the dorsal telencephalon during embryogenesis, which occurs before the onset of astrogenesis around E18 in the developing mouse cortex (34). Postnatally, exogenous KLF4 was significantly down-regulated and was occasionally observed in some GFAP-expressing cells; however, most of these cells also coexpressed Nestin. In addition to NSCs, transgenic KLF4 was also detectable on perinatal ependymal cells, which are derived from embryonic NSCs (31). These data suggest that the hydrocephalic phenotype in KLF4-overexpressing mice may be attributed to its expression in Nestin-positive NSCs and ependymal cells during development. Consistent with this hypothesis, inhibition of exogenous expression of KLF4 by Dox treatment during early embryogenesis but not perinatal stages completely suppressed the malformation of multicilia and the hydrocephalus phenotype.

Hydrocephalus is a direct result of abnormal accumulation of CSF, which is due to imbalanced secretion and reabsorption. CSF is produced by choroid plexus, a structure aligning the brain ventricles. No abnormalities were observed in either this structure or the cilia on epithelia of KLF4 transgenic mice. The aqueduct also appears normal. In contrast, the SCO and the ependymal cilia are severely malformed in mutant mice. The SCO is a highly specialized ependymal organ that secretes high molecular weight glycoproteins. These glycoproteins aggregate and form the threadlike RF, which extends along the length of the CSF tract to the ampulla caudalis. The SCO-RF structure plays a critical role in CSF flow and reabsorption through the Sylvian aqueduct (17). Malformation of the SCO has been implicated in the etiology of hydrocephalus in multiple transgenic mouse strains (30, 35–37). Consistent with these observations, we found that the SCO-RF structure is dramatically reduced in size in KLF4-transgenic mice. Such reduction may be a direct result of the inhibitory effect of KLF4 on ependymal cells.

In addition to the SCO, the cilia on the ventricular ependyma help maintain the laminar flow of CSF. Clinically, abnormal cilia are associated with hydrocephalus in human patients (27, 38). Mutations in mice that lead to dysfunctional ependymal cilia often result in hydrocephalus. These mutations include transcriptional regulators (Foxj1/Hfh-4 and polymerase λ), G protein-coupled receptors (PAC1 and Ro1), and components in the intraflagellar transport complex (Polaris/Ift188, Mdnah5, Hydin, and Spag6) (17). Consistent with their known role in the development of hydrocephalus, the cilia on the ventricular ependyma, but not on the choroidal epithelia, were specifically malformed in KLF4-transgenic mice. Together, these results suggest that CSF flow and reabsorption rather than production may be the cause of the hydrocephalus phenotype in these mice.

It should be noted that multiple members of the KLF family of transcription factors are frequently coexpressed and may have redundant functions. For example, although KLF4 is dispensable for embryonic stem cells, its down-regulation in combination with depletion of both KLF2 and KLF5 leads to differentiation of these stem cells (10). Such functional redundancy is reflected in their shared downstream targets. Similarly, multiple KLF factors can either enhance or suppress axonal growth or regeneration (13). Our preliminary results also showed that down-regulation of KLF4 by shRNAs in NSCs did not significantly alter their proliferation or differentiation (Fig. S9). In addition to Klf4, our global gene expression analysis detected several other Klf genes (Klf2, Klf3, Klf6, Klf7, Klf9, Klf11, Klf12, and Klf13) are expressed in NSCs. Although challenging due to potential redundancy, it will be interesting in the future to tease out the unique or overlapping functions of these Klfs in self-renewal of NSCs and during neural development.

Materials and Methods

For a detailed description of materials and methods, see SI Materials and Methods.

Animals.

The generation of hGfap-tTA and TRE-myc-Klf4 mice has been described in detail (19, 20). No significant phenotypic differences were detected between male and female mice; thus, both genders were included in the analysis. Experimental protocols were approved by the Institutional Animal Care and Use Committee at the University of Texas Southwestern Medical Center.

Immunocytochemistry, Immunohistochemistry, Histology, and Electron Microscopy (EM).

For BrdU experiments, timed-pregnant mice were injected intraperitoneally with BrdU (100 mg/kg) and analyzed after 12 h. For timed induction of transgene, mice were fed with drinking water containing doxycycline (5 μg/mL) for the indicated duration. For scanning EM, brains were perfused, fixed, and processed for viewing with a Philips XL 30 ESEM microscope. All other procedures of histology were done essentially as described (39).

Image Processing and Statistical Analysis.

All of the images within the same panel were identically processed with Adobe Photoshop. Values are presented as mean ± SE. Statistical significance was determined using an unpaired Student t test. A P value of < 0.05 was considered significant.