Dclk1 Deletion in Tuft Cells Results in Impaired Epithelial Repair After Radiation Injury

Abstract

The role of Dclk1⁺ tuft cells in the replacement of intestinal epithelia and reestablishing the epithelial barrier after severe genotoxic insult is completely unknown. Successful restoration requires precise coordination between the cells within each crypt subunit. While the mechanisms that control this response remain largely uncertain, the radiation model remains an exceptional surrogate for stem cell-associated crypt loss. Following the creation of Dclk1-intestinal-epithelial-deficient Villin-Cre;Dclk1^flox/flox mice, widespread gene expression changes were detected in isolated intestinal epithelia during homeostasis. While the number of surviving crypts were unaffected, Villin-Cre;Dclk1^flox/flox mice failed to maintain tight junctions and died at ~5d, where Dclk1^flox/flox mice lived until day 10 following radiation injury. These findings suggest that Dclk1 plays a functional role critical in the epithelial restorative response.

Introduction

Doublecortin-like kinase 1 (Dclk1) has recently been the subject of lineage-tracing experiments whose progeny produced polyps in Apc^min/+ mice, thus identifying it as a tumor stem cell marker ^[1]. Little is known about its role at homeostasis and what if any influence it may exert upon the rest of the intestinal epithelia following injury. Recently it has been reported that Dclk1 marks an overlooked fifth lineage of intestinal epithelial cell subsequently identified as tuft cells utilizing Sox9 and Cox1 ^[2]. While their precise role is still mysterious, recent publications have implicated them in chemosensing through taste transduction-related proteins ^{[2, 3]}. As the mammalian small intestine grows into adulthood, the number of crypts of Lieberkühn steadily increases by a process termed “crypt fission,” in which new crypts are formed by branching off from existing crypts ^[4]. Newly produced progeny comprise the epithelial lineage diversity, migrating onto the villus epithelium ^[5]. Following severe damage from a variety of insults there is an observable loss of these stem cell-containing crypt subunits and a restorative process is initiated. Previous studies have shown that Dclk1 expression drops dramatically ~24h following significant radiation injury and that Dclk1⁺ cells are essentially undetectable in the epithelia by 84h ^[6]. Subsequent label retention experiments showed that following the loss of Dclk1 expression; re-expression could be seen in rare label-retaining quiescent cells located within hyperplastic crypts at 7 and 10 days following severe radiation injury^[7]. A recent report by Cheasley et al demonstrated the downregulation of both Dclk1 and Lgr5 in isolated small intestinal crypts of control mice at 5 days following exposure to 13 Gy radiation ^[8].

Results

To evaluate the role of Dclk1, we created the novel Dclk1-intestinal-epithelial-deficient mice Villin-Cre;Dclk1^flox/flox (fig. S1 and S2A–C). We found no obvious morphologic changes between Villin-Cre;Dclk1^flox/flox mice and their Dclk1^flox/flox littermates (fig. S3A). Furthermore the presence of Cox1 expressing cells identified by immunofluorescence staining established that tuft cells, which normally express Dclk1, were still present in both groups (Fig. 1A). To catalogue the changes within the intestinal epithelia, deep sequencing of isolated epithelial cells was conducted at homeostasis. An examination of stemness-related genes (Fig. 1B), and the Notch (Fig. 1C) and Wnt (Fig. 1D) molecular pathways demonstrated noteworthy alterations. Genes with proposed communication function in tuft cells through chemosensing were also significantly altered (Fig. 1E) ^[9]. Uroguanylin peptide, which is encoded by Guca2b and associated with chemosensing ^[3], was the most differentially expressed gene among the ~20,000 analyzed, which was confirmed by RT-PCR (fig. S3). This data implicating intercellular communication through alteration of chemosensory genes could provide a foundation for how Dclk1, which is expressed in a minority of epithelial cells, exerts its broad influence over intestinal epithelial gene expression. We also found a modest but significant modification of tight and adherens junction-related genes in the Villin-Cre;Dclk1^flox/flox mice (Fig. 1F). A baseline assessment of intestinal epithelial barrier function in uninjured mice from both groups was obtained following oral gavage with the permeability tracer FITC-labeled dextran (80mg/100g of body weight) 4h prior to being killed. Fluorometric analysis of blood via cardiac puncture at the time of death revealed a statistically significant increase in FITC-dextran uptake in the Villin-Cre;Dclk1^flox/flox when compared to Dclk1^flox/flox mice (Fig. 2A). This result appears to confirm a homeostatic dysfunction in junctional adhesion, indicated in the deep sequencing analysis. Indeed a decrease in Claudin-1 protein was seen in Villin-Cre;Dclk1^flox/flox mice by Western blot analysis (Fig. 2B). Further examination of uninjured mice intestine via Western blot and RT-PCR revealed an upregulation of Lgr5 (RNA and protein) and downregulation of Notch1 (protein), but no significant change in Bmi1, Msi1, or Cox1 when compared to Dclk1^flox/flox controls (Fig. 2C and D).

Fig. 1. Dclk1 epithelial deletion and its effect on gene expression.

(A) Intestinal tissue sections from both Dclk1^flox/flox and Villin-Cre;Dclk1^flox/flox mice were stained for anti-Dclk1 (green) and Cox1 (red). Immuohistochemistry confirmed the deletion of Dclk1 from the epithelia of Villin-Cre;Dclk1^flox/flox mice. The presence of tuft cells was identified in both Villin-Cre;Dclk1^flox/flox and Dclk1^flox/flox by the presence of Cox1 (200x). Intestinal epithelial cells were isolated and pooled by group (n=3), RNA was isolated from each group and sent in for deep sequencing by Otogenics Corp. From the library of 20 million reads, sequencing maps were generated in interest groups for stemmness (B), Notch signaling pathway (C), Wnt signaling pathway (D), chemosensing (E), and junctional complexes (F).

Fig 2. Disruption of barrier proteins and alteration of Lgr5 and Notch mRNAs in Villin-Cre;Dclk1^flox/flox mice under basal conditions.

(A) To assess intestinal epithelial barrier function, uninjured Dclk1^flox/flox and Villin-Cre;Dclk1^flox/flox mice were gavaged with FITC-labeled dextran (dose=80mg/100g of body weight) 4h prior to being killed. Blood collected via cardiac puncture at the time of death revealed a modest, but significant increase (p <0.016) in FITC-dextran uptake in the Villin-Cre;Dclk1^flox/flox mice. (B) Western blot analysis of tight junction and adherens junction proteins demonstrates a decrease of Claudin-1 protein in the Villin-Cre;Dclk1^flox/flox mice (lane 2) compared to Dclk1^flox/flox (lane 1). (C) RT-PCR analysis reveals a 4-fold basal upregulation of Lgr5 mRNA (p <0.032), while Notch 1, Bmi1, and Msi1 levels are similar in both Dclk1^flox/flox and Villin-Cre;Dclk1^flox/flox mice. (D) Western blot examination of the same tissues shows basal upregulation of Lgr5 protein, downregulation of Notch1 protein, while Cox1, Bmi1, and Msi1 are similar in both Dclk1^flox/flox (lane 1) and Villin-Cre;Dclk1^flox/flox mice (lane 2). (E) Real-time RT-PCR results demonstrating coregulation of Lgr5 and CBC markers Ascl2 and Smoc2 following TBI with marked suppression seen in all genes throughout the time course in Villin-Cre;Dclk1^flox/flox mice when compared to Dclk1^flox/flox controls.

Next we sought to determine what effect radiation would have on the Dclk1-related deficiencies discovered at homeostasis. Subsequent to severe genotoxic insult, the process of replacing intestinal epithelia and reestablishing the epithelial barrier is initiated. This requires the precise coordination of stem cells and their progeny. Following exposure to a dose of radiation sufficient to cause the loss of crypt subunits (> 8Gy), mice undergo a predictable pattern of replenishment ^{[6, 10, 11]} (Fig. 3A). The use of the well-established radiation injury model as a surrogate for stem cell-associated crypt loss and epithelial replacement mechanics remains exceptional in its reproducibility ^{[6, 10, 12, 13]}. Mice exposed to 12 Gy total body irradiation (TBI), with no differences in the gut histology noted 6h and 24h post-injury (data not shown). At 84h post-injury, regenerative crypts were clearly identifiable in both groups without any statistically significant differences identified by Bromodeoxyuridine (BrdU) immunostaining (Fig. 3B) or H&E (fig. S3B). However molecular examination of intestinal tissues from this time point showed a progression of the deficiencies previously noted in Villin-Cre;Dclk1^flox/flox mice at homeostasis. Western blot analysis of junctional proteins demonstrated decreased levels of Claudin-1, Claudin-7 and E-cadherin when compared to Dclk1^flox/flox (Fig. 3C). RT-PCR and Western blot analysis of key putative stem cell makers showed the downregulation of Bmi1, Msi1 and even more dramatically, Lgr5 (Fig. 3D and E). Of particular note is the near abrogation of Notch1 protein, as it provides confirmation to reports indicating Notch1 regulation via a Dclk1-dependent mechanism ^{[14, 15]} (Fig. 3D).

Fig. 3. Overall stem cell response and junctional complexes appear to be suppressed 84h post-TBI in Villin-Cre;Dclk1^flox/flox mice.

(A) Following high-dose TBI, mouse small intestine undergoes a predictable pattern of replenishment. The initial wave of apoptosis can be observed 6h post-TBI, followed by a second wave of apoptosis and release from mitotic arrest 24h post-TBI. BrdU staining 84h post-TBI can quantify the emergence of surviving crypts. The restoration of a mitotically active crypt-villus axis found 168h post-TBI. (B) The regenerative crypts were identified following Bromodeoxyuridine (BrdU) staining 84h post-TBI in the small intestines from either group, the number of crypts was quantified and no statistical differences were found (200x). (C) Western blot analysis of junctional proteins demonstrated decreased levels of E-cadherin, Claudin-1, and Claudin-7, but not Cox1 in Villin-Cre;Dclk1^flox/flox mice (lane 2) compared to Dclk1^flox/flox controls (lane 1). (D) Western blot and (E) RT-PCR analysis showed the downregulation of Notch1 (p <0.001) and key putative stem cell makers Bmi1 (p <0.085), Lgr5 (p <0.034), and Msi1 (p <0.037) in Villin-Cre;Dclk1^flox/flox (lane 2) mice compared to Dclk1^flox/flox.(lane 1)

By 5d post-TBI, the Villin-Cre;Dclk1^flox/flox mice group began to show signs of physical decompensation. Consequently we chose this time point to perform an additional permeability test as described above. The fluorometric analysis revealed an ~4-fold increase in FITC-dextran uptake in the Villin-Cre;Dclk1^flox/flox mice when compared to Dclk1^flox/flox mice (Fig. 4A). Western blot demonstrated decreases in Claudin-1 and Claudin-7, as well as E-cadherin in Villin-Cre;Dclk1^flox/flox mice (Fig. 4B). Staining for E-cadherin demonstrated strong membranous localization in Dclk1^flox/flox mice and diffuse cytosolic localization in the Villin-Cre;Dclk1^flox/flox mice (Fig. 4C). H&E confirmed the degraded quality of regenerated epithelia in Villin-Cre;Dclk1^flox/flox mice (fig. S3C). A survival curve reveals the dramatic consequence of cellular integrity breakdown, as Villin-Cre;Dclk1^flox/flox mice did not survive beyond 5d, whereas Dclk1^flox/flox were outwardly healthy when killed at 10d (Fig. 4D). Western blot and RT-PCR results from 5d show a deepened suppression of stem cell markers Bmi1, Lgr5, Msi1, and the absence of Notch1 and initial decrease of Cox-1 (protein) (Fig. 4E and F).

Fig 4. Suppression of the restorative response results in shortened survival for Villin-Cre;Dclk1^flox/flox mice following TBI.

(A) Fluorometric analysis of cardiac blood collected 4h after gavage revealed a ~4-fold increase in FITC-dextran uptake in the Villin-Cre;Dclk1^flox/flox mice compared to Dclk1^flox/flox 5 days post-TBI (p <0.0001). (B) Western blot analysis of junctional/adheren proteins in small intestinal tissues show dramatic decreases in Claudin-1, Claudin-7, E-cadherin in Villin-Cre;Dclk1^flox/flox (lane 2) mice compared to Dclk1^flox/flox.(lane 1) (C) Immunohistochemistry for E-cadherin demonstrates strong membranous staining in the Dclk1^flox/flox mice but diffuse cytosolic staining in Villin-Cre;Dclk1^flox/flox mice indicative of protein degradation (200X – insets ~600X). (D) A survival curve reveals the dramatic consequence of such significant loss of cellular integrity, as Villin-Cre;Dclk1^flox/flox mice spontaneously died by day 5, whereas Dclk1^flox/flox controls lived until the termination of the experiment at 10d post-TBI. (E) RT-PCR small intestinal tissues 5d post-TBI demonstrates decreased mRNA expression of Notch1 (p <0.003) and stemmness markers Bmi1 (p <0.010), Lgr5 (p <0.009), Msi1 (p <0.002) and (F) corresponding Western blot analysis. Of particular note is the complete absence of the Notch1 in Villin-Cre;Dclk1^flox/flox mice (lane 2) compared to Dclk1^flox/flox (lane 1). Immunofluorescence staining for both (G) Claudin-1 and (H) Claudin-7 confirm the impaired quality of the tight junctions and loss of cellular integrity following radiation injury in Villin-Cre;Dclk1^flox/flox mice compared to Dclk1^flox/flox controls.

Discussion

Following homeostatic epithelial-Dclk1 deletion, marked gene expression changes in pathways critical to epithelial growth, stemness, barrier function, and taste reception signaling are seen. While the mechanism by which this occurs is not yet known, previous studies have shown guanylin activation to be a mediator of intestinal fluid homeostasis, intestinal cell proliferation/apoptosis, and tumorigenesis ^[16]. In this report investigators propose that guanylyl cyclase C signaling via uroguanylin, mediates regulation of intestinal barrier function. The decrease of this ligand seen in Villin-Cre;Dclk1^flox/flox mice is likely a key factor for the dysregulation seen, particularly in tight junctions. Immunofluorescence staining for both Claudin-1 (Fig. 4G) and Claudin-7 (Fig. 4H) confirms the impaired quality of the tight junctions and loss of cellular integrity following radiation injury in Villin-Cre;Dclk1^flox/flox mice. Although primarily identified as a tumor stem cell by Y. Nakanishi et al., they also demonstrated evidence of stemness via the appearance of rare “blue stripes” in Dclk1CreERT2/+; Rosa26R mice during homeostasis, which increased following exposure to 8 Gy radiation [1]. Additionally Dclk1 does mark rare cells in the intestine that would not readily be considered mature tuft cells and are known to coexpress Lgr5 (fig. S5) ^{[6, 17, 18]}. . An examination of the mRNA levels throughout the radiation time course reveals an unequivocal suppression of not only Lgr5, but companion CBC genes Ascl2 and Smoc2 as well (Fig. 2E). These data taken together establish the basis for a co-regulatory connection between Dclk1 and Lgr5. Yet the full identity, function and lineage of Dclk1-exprssing cells remain unclear and the subject of intense debate ^[17–22]. Therefore it is not improbable that Dclk1-expressing cells in the crypt base may have functions that are particularly relevant to epithelial restoration. Continuous renewal of intestinal epithelium must occur without disrupting the epithelial layer or altering cellular adhesion ^[23–25] and the dramatic phenotypic changes reported here in the face of severe injury are quite compelling. While there is clear evidence that an attempt to reconstitute the gut occurs, the surviving stem cells are unable to participate fully in the epithelial restoration and barrier repair needed for survival in the absence of Dclk1.

Conclusion

Epithelial Dclk1 plays a functional role critical to the epithelial restorative response and is required for satisfactory restitution and survival following severe insult.

Supplementary Material

Supp Fig S1-S5

fig. S1. The excision of Dclk1 gene was confirmed by PCR. Genomic DNA was isolated from the intestinal epithelium of Dclk1^flox/flox and Villin-Cre;Dclk1^flox/flox mice. A ~2.3 kb band corresponding to the floxed allele was detected in the Dclk1^flox/flox mouse. Using the same primers a ~800 bp band was detected in the Villin-Cre;Dclk1^flox/flox mouse, a decrease in molecular weight corresponding to the deletion of the floxed allele.

fig. S2. Dclk1 mRNA and protein levels in mouse small intestinal tissue throughout the experimental time course. (A) Relative Dclk1 mRNA levels increase ~96h post-TBI in Dclk1^flox/flox controls. (B) Dclk1 protein is downregulated 5d post-TBI in Villin-Cre;Dclk1^flox/flox (epithelial-deletion) mice compared to Dclk1^flox/flox control. (C) Dclk1 mRNA expression is downregualted in Villin-Cre;Dclk1^flox/flox throughout the experimental time course when compared to Dclk1^flox/flox controls.

fig. S3. No histologic changes were noted in Villin-Cre;Dclk1^flox/flox mice at homeostasis. Uninjured mouse small intestine stained with H&E and examined for any unique changes to the overall epithelial morphology (A). However at 84hrs (B) and 5d (C) post-TBI incomplete restitution of intestinal epithelium in seen in Villin-Cre;Dclk1^flox/flox mice. (200X).

fig. S4. Guca2b mRNA in Villin-Cre;Dclk1^flox/flox mice is ~80% decreased relative to that of Dclk1^flox/flox mice at homeostasis. N=5 per group, p <0.008.

fig. S5. Lgr5 (β-gal) and Dclk1 (IHC) colocalize in rare CBCs of mouse small intestine. Following 1 day of tamoxifen induction (5mg) histologic examination of small intestine tissue sections (4μm) stained with β-gal and anti-Dclk1 antibody from Lgr5-ires-CreERT2/Rosa26-lacZ mice reveals colocalization in rare CBCs.