Altered Mucus Barrier Integrity and Increased Susceptibility to Colitis in Mice upon Loss of Telocyte Bone Morphogenetic Protein Signalling

FoxL1+-Telocytes (TCFoxL1+) are subepithelial cells that form a network underneath the epithelium. We have shown that without inflammatory stress, mice with loss of function in the BMP signalling pathway in TCFoxL1+ (BmpR1aΔFoxL1+) initiated colonic neoplasia. Although TCFoxL1+ are modulated in IBD patients, their specific role in this pathogenesis remains unclear. Thus, we investigated how the loss of BMP signalling in TCFoxL1+ influences the severity of inflammation and fosters epithelial recovery after inflammatory stress. BmpR1a was genetically ablated in mouse colonic TCFoxL1+. Experimental colitis was performed using a DSS challenge followed by recovery steps to assess wound healing. Physical barrier properties, including mucus composition and glycosylation, were assessed by alcian blue staining, immunofluorescences and RT-qPCR. We found that BmpR1aΔFoxL1+ mice had impaired mucus quality, and upon exposure to inflammatory challenges, they had increased susceptibility to experimental colitis and delayed healing. In addition, defective BMP signalling in TCFoxL1+ altered the functionality of goblet cells, thereby affecting mucosal structure and promoting bacterial invasion. Following inflammatory stress, TCFoxL1+ with impaired BMP signalling lose their homing signal for optimal distribution along the epithelium, which is critical in tissue regeneration after injury. Overall, our findings revealed key roles of BMP signalling in TCFoxL1+ in IBD pathogenesis.


Introduction
Intestinal inflammation has long been considered as a process in which effector immune cells destroy the mucosa and subsequently result in chronic inflammation when left unresolved [1]. Therefore, the mucosa is a target instead of a possible trigger for inflammatory bowel disease (IBD). Studies investigating the deregulation of non-hematopoietic mucosa cells in various phases of IBD have broadened the understanding of their involvement in IBD pathogenesis [2][3][4]. Thus, epithelial and mesenchymal cells also participate in IBD, from pathogen or damage recognition to recruitment of immune cells to the injury site, pathogen or damage elimination and, ultimately, resolution of inflammation [3,5]. Particularly, mesenchymal cells are strategically positioned between the epithelial and immune cell compartments and can consequently regulate epithelial functions and influence mucosal immune cells [5,6].

Electron Microscopy
Colons from post-natal 90-day-old BmpR1a ∆FoxL1+ and control mice were fixed and sectioned as previously described [2]. Transmission electron microscopy images were digitally coloured in blue using Adobe Photoshop CC 2017.

Immunofluorescence and Fluorescence In Situ Hybridisation
Immunofluorescence staining for TC FoxL1+ population was performed as follows; slides were immersed in 0.01 M citric acid buffer (pH 6.0) and microwaved to boil for 6 min for antigen retrieval, then cooled, washed in PBS and incubated for 40 min at room temperature in blocking solution (1% Gelatin from cold water fish skin, 2% BSA, 0.2% Triton X-100 in PBS). Two first primary antibodies anti-Gli1 (1:500, NB600-600, Novus Biological, CO, USA) and anti-PDGFRα (1:150, AF1062, R&D System, MN, USA) were simultaneously diluted in the blocking solution and incubated overnight at 4 • C in a moist chamber. After washing with PBS, Alexa 594 (anti-goat) and Alexa 647 (anti-rabbit) secondary antibodies were diluted in blocking solution, applied to the slides and incubated for 1 h at RT. A second blocking step was performed incubating the slides for 40 min at room temperature in the blocking solution described above. Then, slides were incubated 1 h at RT with anti-CD34 (1:100, ab81289, Abcam, Cambridge, UK) diluted in the blocking solution, followed by 1 h with Alexa 488 (anti-rabbit) at RT. Nuclei were stained with DAPI. Images were captured on a confocal Microscope Zeiss LSM 880 2 photons. Telocytes population was analysed using Fiji ImageJ v 2.1.0, from high-powered fields in a blinded manner on an average of 10 independent fields of the whole colon per animal (N = 6 per group). Staining on frozen sections was performed as followed: tissues were fixed in 100% Ethanol for 15 min at −20 • C [2,8,10,15]. For all other antibodies, immunofluorescence was assessed as previously described [2,8,10] . Images were captured on a Microscope Zeiss Axioscope 5. Carnoy-fixed colon sections were hybridised with a Cy3coupled bacterial 16S rRNA probe (EUB338). A Cy3-coupled nonsense probe (NS_EUB338) was used as the control for non-specific binding. Probe sequences for fluorescence in situ hybridisation were CY3_EUB338_SENSE 5 -/5Cy3/GCTGCCTCCCGTAGGAGT-3 and CY3_NS_EUB338 5 -/5Cy3/CGACGGAGGGCATCCTCA-3 . Nuclei were stained with DAPI. Images were captured on a Leica DM2500 Optigrid.

Quantification of Cell Number, Cell Distribution, Vesicle and Goblet Cell Post-Translational Modifications
The presence of goblet cells was analysed using alcian blue-stained sections from lowpowered fields of well-oriented colonic cross-sections in a blinded manner on an average of 10 independent fields of the proximal, middle and distal colon per animal (N = 4 per group). Goblet cell vesicles were analysed from 10 pictures from transmission electron microscopy images. Using Fiji ImageJ v 2.1.0 they were divided into clear, light grey and dark grey and then quantified (N = 3 per group). For lectins and MUC2 immunofluorescence, we measured the corrected total cell fluorescence (CTCF) by applying the following formula CTCF = Integrated Density-(Area of selected cell X Mean fluorescence of background readings), using Fiji ImageJ v 2.1.0. TC FoxL1+ were labelled according to the most recent consensus in the literature as described in Section 2.6 [7]. Triple immunostaining for PDGFRα, Gli-1 and CD34 was performed, and TC FoxL1+ were identified as PDGFRα + /Gli-1 + /CD34 − . TC FoxL1+ number was averaged from ten different image fields per mouse in each condition (naive, acute, recovery). TC FoxL1+ distribution was also evaluated from these images, where the colon mucosa was separated in 3 different zones, based on epithelial cell identity: stem cells and progenitor cells, transit-amplifying and differentiated zones. TC FoxL1+ were assigned to either one of the zones in all 3 conditions (naive, acute, recovery). Multiple images for each mouse were evaluated and the relative frequency of distribution in percentage was then calculated (zone SC-P + zone TA + zone D = 100%). For myofibroblasts, total number of double-positive cells for Vimentin/αSMA was counted using Fiji ImageJ v 2.1.0, from high-powered fields in a blinded manner on an average of 10 independent fields of the distal colon per animal (N = 4-6 per group).

Statistical Analysis
Statistical significance was calculated using the Mann-Whitney test in GraphPad Prism v8. * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001 were considered significant. Data are presented as the mean ± SEM. For the goblet cell vesicles counting and TC FoxL1+ distribution along the crypt axis, 2-way ANOVA was used as a statistical test. Data are presented as the mean ± SD.

BmpR1a ∆FoxL1+ Mice Have Increased Susceptibility to Experimental Colitis and Delayed Wound Healing
Upon the induction of acute colitis with DSS (herein, acute phase), a reduction in body weight was observed in both male and female BmpR1a ∆FoxL1+ mice compared with the controls at day 5 ( Figure 1A). After recovery from acute colitis (herein, recovery phase), there was no significant difference in weight loss in BmpR1a ∆FoxL1+ mice between the acute and recovery phases compared with the control ( Figure 1B). Noticeably, a 60% death rate was observed in BmpR1a ∆FoxL1+ mice in the first two days of recovery. The body weights of the surviving mice were similar between both groups by day 11 until day 19 (end of the experimental period). We observed a significant 1.6-fold modulation in disease activity index (DAI) between the control and BmpR1a ∆FoxL1+ mice in the acute phase ( Figure 1C). After completing the recovery phase, there was no significant difference in the DAI of surviving animals in both groups. Histological analysis results revealed normal colonic mucosa in naive control and BmpR1a ∆FoxL1+ mice ( Figure 1D). Following the acute phase, crypt erosions, minor immune infiltration and a preserved epithelial layer were observed in the control, whereas complete loss of crypt architecture, strong influx of immune cells and total loss of the epithelial lining were observed in BmpR1a ∆FoxL1+ mice ( Figure 1D). In the recovery phase, the controls showed nearly restored colonic mucosa compared with the surviving BmpR1a ∆FoxL1+ mice that still presented significant immune cell infiltration with some areas having crypt abscess or denuded epithelium ( Figure 1D). H&E staining demonstrated significant differences between the control and BmpR1a ∆FoxL1+ mice in both acute and recovery phases. In both treatments, surviving BmpR1a ∆FoxL1+ mice had a 1.4- fold increase in their histological score after the acute (7 days) and recovery treatments (19 days) compared with the control mice ( Figure 1E). crypt erosions, minor immune infiltration and a preserved epithelial layer were observed in the control, whereas complete loss of crypt architecture, strong influx of immune cells and total loss of the epithelial lining were observed in BmpR1a ΔFoxL1+ mice ( Figure 1D). In the recovery phase, the controls showed nearly restored colonic mucosa compared with the surviving BmpR1a ΔFoxL1+ mice that still presented significant immune cell infiltration with some areas having crypt abscess or denuded epithelium ( Figure 1D). H&E staining demonstrated significant differences between the control and BmpR1a ΔFoxL1+ mice in both acute and recovery phases. In both treatments, surviving BmpR1a ΔFoxL1+ mice had a 1.4-fold increase in their histological score after the acute (7 days) and recovery treatments (19 days) compared with the control mice ( Figure 1E).

Defective Telocytes Lead to the Development of Compromised Mucus Layers and Abnormal Bacterial Infiltration
The loss of the epithelial barrier function has been reported as a cause of IBD [2,23]. However, impaired epithelial permeability was not observed in both groups ( Figure 2A). The mucus layers serve as another protective mechanism in the epithelium against commensal bacteria as they form part of the physical barrier. Goblet cells are responsible for the production of mucus, thereby protecting the epithelial gut lining by forming the mucus barrier [24]. Under naive conditions, we observed no differences in goblet cell number along the colonic crypts in both groups, although we noticed a thinner and discontinuous barrier layer in BmpR1a ∆FoxL1+ mice ( Figure 2B,C). Under acute phase, reduced goblet cells in the colonic crypt were observed in both groups; notably, in BmpR1a ∆FoxL1+ mice, an almost complete depletion of these cells was observed in the distal colon ( Figure 2B,C). The partial erosion of the barrier layer occurred in all mice but was greater in BmpR1a ∆FoxL1+ mice ( Figure 2C). In the recovery phase, we observed a partial restoration in goblet cell count and mucus layers in the control group, whereas delayed mucus layer restoration and low goblet cell density were observed in BmpR1a ∆FoxL1+ mice ( Figure 2B,C).
The loss of the epithelial barrier function has been reported as a cause of IBD [2,23]. However, impaired epithelial permeability was not observed in both groups ( Figure 2A). The mucus layers serve as another protective mechanism in the epithelium against commensal bacteria as they form part of the physical barrier. Goblet cells are responsible for the production of mucus, thereby protecting the epithelial gut lining by forming the mucus barrier [24]. Under naive conditions, we observed no differences in goblet cell number along the colonic crypts in both groups, although we noticed a thinner and discontinuous barrier layer in BmpR1a ΔFoxL1+ mice ( Figure 2B,C). Under acute phase, reduced goblet cells in the colonic crypt were observed in both groups; notably, in BmpR1a ΔFoxL1+ mice, an almost complete depletion of these cells was observed in the distal colon ( Figure 2B,C). The partial erosion of the barrier layer occurred in all mice but was greater in BmpR1a ΔFoxL1+ mice ( Figure 2C). In the recovery phase, we observed a partial restoration in goblet cell count and mucus layers in the control group, whereas delayed mucus layer restoration and low goblet cell density were observed in BmpR1a ΔFoxL1+ mice ( Figure 2B,C). During the acute phase, all mice showed partial erosion of the barrier layer with BmpR1a ∆FoxL1+ mice presenting more advanced erosion. After recovery, BmpR1a ∆FoxL1+ mice still presented a thinner barrier layer while control mice improved. (D) Before treatment, bacteria (red) were localised only in the outer mucus layer and not found in the inner layer (double white arrows) in control mice; during the acute phase, bacteria were in contact with the epithelium but not detected in the inner mucosa. In BmpR1a ∆FoxL1+ mice, during the acute phase, large amounts of bacteria (white asterisks) invaded the colonic mucosa. Following recovery, the control mice restored the sterility of the barrier layer (double white arrow), whereas mutant mice still presented bacteria in contact with the epithelium. Nuclei were counterstained with DAPI (blue). Epithelium is delimited by the white discontinuous line. E: Epithelium. Scale bars: 250 µm (C); 105 µm (D). * p < 0.05; *** p < 0.001 analysed by Mann-Whitney test.
The primary role of healthy mucus layers is to physically protect the epithelial surface against bacteria in the lumen. Given that BmpR1a ∆FoxL1+ mice showed discontinuous mucus layers, we hypothesized that commensal bacteria could directly interact with the epithelium. We found that under naive conditions, bacteria were only detected in the outer mucus layer in the control but within the barrier layer in close contact with the non-inflamed epithelium in BmpR1a ∆FoxL1+ mice ( Figure 2D). Following the acute phase, the control mice presented with typical erosion of the barrier layer and increased bacteria near the epithelia. In contrast, a significant number of bacteria invading the inflamed colonic mucosa in BmpR1a ∆FoxL1+ mice were observed ( Figure 2D). After recovery, the control mice had restored mucus layers and no bacteria were found in the sterile inner layer. Conversely, bacteria were still detected near the epithelial layer in BmpR1a ∆FoxL1+ mice ( Figure 2D).

BMP-Associated Signalling in Telocytes Supports the Maturation of Colonic Goblet Cells
Lineage commitment of gut epithelial cells, along with goblet cell fate specification, involve the expression of transcription factors from the Notch signalling pathway and Klf4 [25,26]. No modulation was found in the commitment genes Hes-1 and Atoh1, or in specification genes Spdef and Gfi1, although a significant reduction was observed in the maturation gene Klf4 (1.4-fold) ( Figure 3A). Ultrastructural examination revealed normal morphology in secretory goblet cells, i.e., a basal nucleus localisation and apical mucin vesicle accumulation, in both groups ( Figure 3B). Noticeably, the controls presented more heterogeneity in vesicles compare to BmpR1a ∆FoxL1+ . Vesicles in BmpR1a ∆FoxL1+ mice were frequently fused together and less organised. We therefore quantified the number of clear, light grey and dark vesicles found in goblet cells in both BmpR1a ∆FoxL1+ and control mice ( Figure 3C). We found no statistical differences for the clear vesicles content and a tendency to fewer dark vesicles in the goblet cells of BmpR1a ∆FoxL1+ mice. However, we observed a significant increase in light grey vesicles (1.7-fold) in mutant goblet cells when compared to controls ( Figure 3C). The decrease in vesicle heterogeneity in BmpR1a ∆FoxL1+ mice led us to investigate mucin diversity and their structural components using RT-qPCR analysis. The mRNA levels of Muc2 (3.01-fold), Fut2 (1.4-fold), Muc4 (2.4-fold) and Argr2 (1.7-fold) were significantly increased in BmpR1a ∆FoxL1+ mice when compared to controls ( Figure 3D). Electron microscopy analysis showed that in control mice, apical microvilli were coated with thick glycocalyx while BmpR1a ∆FoxL1+ mice showed very little coating ( Figure 3E). Typical ultrastructural apical junctional complexes were observed in both groups ( Figure 3E). Immunofluorescence against proteins involved in the tight junction complex, such as ZO-1, claudin-1, claudin-2 and JAM-A revealed no apparent change between both groups, suggesting the presence of a normal junctional complex in BmpR1a ∆FoxL1+ mice ( Figure 3F).

BMP-Associated Signalling in Telocytes Affects Post-Translational Modifications of Mucins
We first investigated the distribution of MUC2 using immunofluorescence and observed its peculiar accumulation at the periphery of goblet cells in BmpR1a ΔFoxL1+ mice compared with its normal distribution in the control ( Figure 4A-C).

BMP-Associated Signalling in Telocytes Affects Post-Translational Modifications of Mucins
We first investigated the distribution of MUC2 using immunofluorescence and observed its peculiar accumulation at the periphery of goblet cells in BmpR1a ∆FoxL1+ mice compared with its normal distribution in the control ( Figure 4A-C). UEA-I lectin recognises the Fucα1,2Galβ1,4 motif, and its lectin reactivity is distributed uniformly along the different regions of the colon. We observed that control mice possessed well-filled vesicles ( Figure 4D); whereas BmpR1a ΔFoxL1+ mice had mucin vesicles with reduced fucosylated residues ( Figure 4E,F). Sialic acid modifications in mucins were also investigated. SNA-lectin in the distal colon was only detected in the upper part of the crypts and exhibited a pattern similar to the one observed in MUC2, i.e., accumulation of goblet cells in the periphery in BmpR1a ΔFoxL1+ mice compared with the control (Figure 4G-I). MAL-II lectin staining in BmpR1a ΔFoxL1+ mice revealed the scattered expression in goblet cells compared with an orderly staining pattern along the distal crypts in control mice UEA-I lectin recognises the Fucα1,2Galβ1,4 motif, and its lectin reactivity is distributed uniformly along the different regions of the colon. We observed that control mice possessed well-filled vesicles ( Figure 4D); whereas BmpR1a ∆FoxL1+ mice had mucin vesicles with reduced fucosylated residues ( Figure 4E,F). Sialic acid modifications in mucins were also investigated. SNA-lectin in the distal colon was only detected in the upper part of the crypts and exhibited a pattern similar to the one observed in MUC2, i.e., accumulation of goblet cells in the periphery in BmpR1a ∆FoxL1+ mice compared with the control (Figure 4G-I). MAL-II lectin staining in BmpR1a ∆FoxL1+ mice revealed the scattered expression in goblet cells compared with an orderly staining pattern along the distal crypts in control mice ( Figure 4J-L). PNA-lectin staining recognises terminal galactose β1,3 GalNAc residues. Without chemical treatment or enzymatic digestion, PNA-lectin was only detected in the Golgi complex. After desulphation-KOH-sialidase digestion, an equal number of PNA-lectin + goblet cells were found in the control and mutant mice ( Figure 4M-O); the peripheral accumulation of lectin in BmpR1a ∆FoxL1+ mice was similar to that of MUC2.

Impaired BMP Signalling Does Not Affect Telocytes' Number but Their Localisation toward the crypt axis following Inflammatory Stress
Ultrastructural analysis revealed the classical features of telocytes, i.e., small bodies and long telopodes, in both groups, with BmpR1a ∆FoxL1+ mice exhibiting more irregularly shaped telopodes than the control ( Figure 5A). At higher magnification, TC FoxL1+ with loss of BMP-associated signalling were surrounded by densely packed collagen fibrils, suggesting the promotion of secretory activities; their telopodes had more pinched off or shed vesicles than that of the control mice. Without chemical treatment or enzymatic digestion, PNA-lectin was only detected in the Golgi complex. After desulphation-KOH-sialidase digestion, an equal number of PNAlectin + goblet cells were found in the control and mutant mice ( Figure 4M-O); the peripheral accumulation of lectin in BmpR1a ΔFoxL1+ mice was similar to that of MUC2.

Impaired BMP Signalling Does Not Affect Telocytes' Number but Their Localisation toward the crypt axis following Inflammatory Stress
Ultrastructural analysis revealed the classical features of telocytes, i.e., small bodies and long telopodes, in both groups, with BmpR1a ΔFoxL1+ mice exhibiting more irregularly shaped telopodes than the control ( Figure 5A). At higher magnification, TC FoxL1+ with loss of BMP-associated signalling were surrounded by densely packed collagen fibrils, suggesting the promotion of secretory activities; their telopodes had more pinched off or shed vesicles than that of the control mice. We next investigated the TC FoxL1+ distribution along the crypts before, during and after an inflammatory flare. Telocytes were defined as Gli1 + /PDGFRα + /CD34 − cells. Before treatment, no difference was detected in the TC FoxL1+ population, distribution and localisation between both groups ( Figure 5B,C). After the acute phase, both control and BmpR1a ∆FoxL1+ mice presented a decrease in TC FoxL1+ population and we found mostly single-labelled (Gli1 + or PDGFRα + ) cells in their stroma ( Figure 5B,C). After recovery, both control and BmpR1a ∆FoxL1+ mice restored their preinjury TC FoxL1+ population ( Figure 5B,C), but only in control mice did they re-establish their localisation along the colonic epithelial axis. We found that the majority of the TC FoxL1+ population was shifted to the upper part of the colonic crypt in BmpR1a ∆FoxL1+ mice ( Figure 5B,D). Following this observation, we investigated the status of myofibroblasts in our model.
Under naive conditions, myofibroblasts were distributed along the crypt vertical axis in both mutant and control mice (Figure 6A,B,G). Upon acute inflammatory stress, myofibroblasts' presence was exacerbated in both control and BmpR1a ∆FoxL1+ mice (Figure 6C,D,G). After recovery, we observed an important enduring population of myofibroblasts scattered in the stroma with a strong presence at the bottom of the crypts near the stem cell region in BmpR1a ∆FoxL1+ mice, whereas the control myofibroblasts population returned to their basal level ( Figure 6E-G). Myofibroblasts counts along the crypt vertical axis in the various conditions confirmed a significant enduring myofibroblasts population following recovery in BmpR1a ∆FoxL1+ mice when compared to controls ( Figure 6G).

Discussion
BMP signalling is involved in the development and homeostasis of the gastrointestinal tract [2,8,10,12]. BMP ligands are widely produced by both the epithelial and mesenchymal compartments [7], and in conjunction with other cascades, BMP maintain the critical balance between cell proliferation and differentiation, thus maintaining gut homeostasis [7,12]. Defective BMP signalling is frequently observed during IBD pathogenesis [2,27]. Recently, we have demonstrated that the targeted disturbance of BMP signalling in mouse TC FoxL1+ led to the reprogramming of the stroma, thus initiating neoplasia in the gastric and colonic epithelia [8,10]. However, the precise role of signalling-impaired TC FoxL1+ in IBD susceptibility and wound healing remains elusive. Without inflammatory stress, BmpR1a ΔFoxL1+ mice possessed dysplastic areas with an enlarged mesenchymal compartment and prominent immune cell infiltration [8]. These findings suggest that impaired BMP signalling in TC FoxL1+ plays an active role during and after an inflammatory flare; however, they must first be triggered to activate an inflammation cascade. In this study, using different DSS challenges, we revealed a novel role for TC FoxL1+ in colon inflammation, resolution and repair. During the acute phase, BmpR1a ΔFoxL1+ mice demonstrated increased susceptibility to experimental colitis due to disrupted mucus layer integrity and functionality. However, mucosal damage in BmpR1a ΔFoxL1+ mice resulted in slow mucosal recovery that proved lethal. After recovery, the signalling-impaired TC FoxL1+ cells showed abnormal shape and their population were disproportionally found at the

Discussion
BMP signalling is involved in the development and homeostasis of the gastrointestinal tract [2,8,10,12]. BMP ligands are widely produced by both the epithelial and mesenchymal compartments [7], and in conjunction with other cascades, BMP maintain the critical balance between cell proliferation and differentiation, thus maintaining gut homeostasis [7,12]. Defective BMP signalling is frequently observed during IBD pathogenesis [2,27]. Recently, we have demonstrated that the targeted disturbance of BMP signalling in mouse TC FoxL1+ led to the reprogramming of the stroma, thus initiating neoplasia in the gastric and colonic epithelia [8,10]. However, the precise role of signalling-impaired TC FoxL1+ in IBD susceptibility and wound healing remains elusive. Without inflammatory stress, BmpR1a ∆FoxL1+ mice possessed dysplastic areas with an enlarged mesenchymal compartment and prominent immune cell infiltration [8]. These findings suggest that impaired BMP signalling in TC FoxL1+ plays an active role during and after an inflammatory flare; however, they must first be triggered to activate an inflammation cascade. In this study, using different DSS challenges, we revealed a novel role for TC FoxL1+ in colon inflammation, resolution and repair. During the acute phase, BmpR1a ∆FoxL1+ mice demonstrated increased susceptibility to experimental colitis due to disrupted mucus layer integrity and functionality. However, mucosal damage in BmpR1a ∆FoxL1+ mice resulted in slow mucosal recovery that proved lethal. After recovery, the signalling-impaired TC FoxL1+ cells showed abnormal shape and their population were disproportionally found at the top of the crypt with a scarce presence at the bottom near the stem cells and progenitors region (Figure 7). This result indicates that although the number of TC FoxL1+ could be important for epithelial recovery, the difference observed regarding the delays in healing in our model does not come from a decrease in TC FoxL1+ in the mucosa. top of the crypt with a scarce presence at the bottom near the stem cells and progenitors region (Figure 7). This result indicates that although the number of TC FoxL1+ could be important for epithelial recovery, the difference observed regarding the delays in healing in our model does not come from a decrease in TC FoxL1+ in the mucosa. A substantial presence of myofibroblasts were found in BmpR1a ∆FoxL1+ mice compared to controls.
(C) After the recovery, mucosa in control mice recovered from the insult while in BmpR1a ∆FoxL1+ mice it presented delayed wound healing. In BmpR1a ∆FoxL1+ mice, TC FoxL1+ population was found to be displaced along the colonic crypt with more TC FoxL1+ found at the top and less at the bottom. Myofibroblast-like cells (vimentin + ; αSMA + ) were still strongly present in the stroma. Myofibroblastlike cells were found to be scattered in the stroma with a robust presence at the base of the crypt.
The participation of telocytes in tissue repair has been reported in several organs [28,29] and two primary roles have been postulated: acting as progenitor cells or modulating stem cell activity [30]. Recent studies have shown that mice with PDGFRα + cells lacking R-spondin 3 exhibit increased sensitivity to DSS-mediated inflammation that affects stem cells [11] and that CD34 + cells are located in regions of active regeneration, thus influencing the progenitors [4].
In this study, recovery experiments revealed that while TC FoxL1+ in control mice resume their natural position along the colonic crypt, myofibroblasts are now more noticeable than TC FoxL1+ near the stem cells in BmpR1a ∆FoxL1+ mice. In recent years, the importance of TC FoxL1+ in stem cell niche regulation, via the secretion of WNT [6,11] and BMP factors [7], has been demonstrated. As morphogens, even a small dysregulation in the concentration or gradient of WNT and BMP ligands could affect stem cell survival and cellular fate [7]. Our results suggest that the delocalisation of TC FoxL1+ population along the crypt in the mutant following recovery could affect tissue homeostasis, leading to more severe inflammation and delayed healing. Currently, some of the secreted factors of TC FoxL1+ have been identified; however, the conducive conditions for TC FoxL1+ survival, expansion and homeostasis remain unclear. BMP is fundamental for epithelial cell differentiation [12] and our results suggest that it might not be different for TC FoxL1+ . In other words, the BMP signalling is most likely required by TC FoxL1+ for its optimal functionality in an autocrine/paracrine feedback manner.
Ultrastructural analysis results demonstrated that TC FoxL1+ in BmpR1a ∆FoxL1+ mice were surrounded by a more complex microenvironment than the control mice. Telocytes are involved in mechanical sensing, serve as scaffold platform for stroma elements, and organise the ECM [31]. Thus, their behaviour is altered depending on the nature of the ECM as shown in other tissues [8,[31][32][33]. Niculite et al. have demonstrated that in the presence of different enriched matrices, telocytes change their adherence, morphology and telopodes [33]. In addition, extracellular vesicles produced by telocytes have been observed in several tissues [31,32]. Upon the release of these extracellular vesicles and other soluble factors, TC FoxL1+ exchange information with the surrounding microenvironment to dictate stem-cell behaviour [6], tissue regeneration [34] and immune cell monitoring [6,35]. We found no significant difference in TC FoxL1+ marker expression and localisation along the colon crypts under naive conditions; however, upon acute inflammatory stress, the TC FoxL1+ population decreased, whereas that of myofibroblasts increased in both control and mutant mice. This reduction in TC FoxL1+ has been previously described in ulcerative colitis [36] and has been hypothesised to facilitate uncontrolled myofibroblast proliferation, thus increasing ECM protein deposition. Excessive ECM production contributes to the disruption of tissue homeostasis and the development of fibrosis in inflammatory diseases [36]. The increase in myofibroblasts seen in both groups during the acute phase was maintained only in the stroma of BmpR1a ∆FoxL1+ mice following recovery, suggesting that apoptosis or transdifferentiation from myofibroblast to fibroblast, normally expected during restitution, does not occur [37]. Hence, this indicates that TC FoxL1+ with impaired BMP signalling influence the homeostasis of the surrounding stromal cells, possibly through secretion of soluble factors or extracellular vesicles [6,34,35]. This paracrine signalling with the surrounding stroma cells will be further developed in upcoming studies.
The dominant phenotype of BmpR1a ∆FoxL1+ mice was the presence of an irregular colonic mucus, more specifically in the barrier layer. Upon an inflammatory insult, bacteria infiltrated the epithelial barrier of mutant mice as shown in the MUC2-deficient mouse model [38]. Recovery experiments demonstrated that in BmpR1a ∆FoxL1+ mice, there was delayed colonic mucosa healing, along with a decreased number of goblet cells, denuded epithelial regions and structural problems in the mucus layers, long after the control group had recovered from the insult. We observed no difference in the expression of genes related to cell-fate decisions in both the secretory and absorbent lineages between control and mutant mice, indicating that TC FoxL1+ with impaired BMP signalling do not influence progenitor cells that directly affect cell determination. However, reduction in the maturation gene Klf4 in BmpR1a ∆FoxL1+ mice suggested a possible defect in goblet cell maturation and functionality following loss of BMP signalling in TC FoxL1+ .
We focused on the analysis of goblet cell functionality, mucus synthesis and production in our mouse model. Our ultrastructural analysis results provide new insights on the vesicular composition of goblet cells and the expression of apical glycocalyx. BmpR1a ∆FoxL1+ mice exhibited alterations in the epithelial surface glycocalyx and goblet cell vesicles without heterogeneity. Furthermore, Muc2 and Muc4 mRNA expression significantly increased, suggesting a shift in mucin ratios, which could support the deregulation observed in vesicle diversity in mutant mice. In addition, BmpR1a ∆FoxL1+ mice had a significant increase in the expression of genes such as Arg2 and Fut2, affecting biosynthesis, final structure and functionality of colonic mucins [14,39]. Anterior Gradient 2 (AGR2) has been associated with MUC2 biosynthesis, particularly in its folding, trafficking and assembly [39]. Meanwhile, fucosyltransferase 2 (FUT2) is involved in mucin maturation. Finally, the abnormal pattern of the MUC2 protein within the goblet cell vesicles in BmpR1a ∆FoxL1+ mice suggests an aberrant glycosylation pattern [40]. Mucins have post-translational modifications that affect their functionality [41]. Altogether, these findings led us to analyse components related to mucus structure, such as glycosylation patterns, that could yield a low mucus quality and support the increased susceptibility to experimental colitis observed in BmpR1a ∆FoxL1+ mice.
O-glycosylation of mucins is initiated by the addition of GalNAc to the hydroxyl groups of serine or threonine to form the Tn antigen [42]. Further steps lead to different core structures containing galactose and terminal residues, such as fucose or sialic acid [43,44]. These terminal residues are known to participate in gut microbiota homeostasis [45,46]. Here, in BmpR1a ∆FoxL1+ mice, UEA-1 and SNA lectin had reduced fucose and α2,6-linked sialic acid residues. Similar to previous studies [21,47], we only detected galactose residues after desulphation and enzymatic digestion, which can be attributed to further glycosylation or sulphation of the T antigen [21,47]. These results suggest that BmpR1a ∆FoxL1+ mice exhibit a defective glycosylation pathway, leading to the production of immature mucin. Hence, the reduction in these residues could not only impair mucus quality and promote bacterial accessibility to the epithelium [48], but also shape the gut microbial community and its relationship with the host [49]. Indeed, because glycans serve as carbon energy sources [46] and attachment sites for the resident microbiota, future studies will be required to analyse the microbiota composition in both groups.
In summary, we have clearly demonstrated the involvement of TC FoxL1+ in the regulation of the healing functions of the colonic mucosa during and after inflammation. BmpR1a ∆FoxL1+ mice showed abnormal mucus quality, and following inflammation, they had increased susceptibility to experimental colitis and delayed healing. Our results suggest that cell-cell paracrine communication or direct interactions between goblet cells and TC FoxL1+ are essential for maintaining the optimal functionality of the secretory cells. In addition, these results revealed that BMP signalling in TC FoxL1+ is key for the regulation of its own homeostasis and communication functionality. Our findings provide new insights into the roles of TC FoxL1+ in the regulation of epithelial homeostasis beyond the stem cell niche.