Sphk2 deletion is involved in structural abnormalities and Th17 response but does not aggravate colon inflammation induced by sub-chronic stress

The chronic inflammatory process that characterizes inflammatory bowel diseases (IBD) is mainly driven by T-cell response to microbial and environmental antigens. Psychological stress is a potential trigger of clinical flares of IBD, and sphingosine-1-phosphate (S1P) is involved in T-cell recruitment. Hence, stress impact and the absence of sphingosine kinase 2 (Sphk2), an enzyme of S1P metabolism, were evaluated in the colon of mice after sub-chronic stress exposure. Here, we show that sub-chronic stress increased S1P in the mouse colon, possibly due to a decrease in its degradation enzymes and Sphk2. S1P accumulation could lead to inflammation and immune dysregulation reflected by upregulation of toll-like receptor 4 (TLR4) pathway, inhibition of anti-inflammatory mechanisms, cytokine-expression profile towards a T-helper lymphocyte 17 (Th17) polarization, plasmacytosis, decrease in IgA+ lymphoid lineage cells (CD45+)/B cells/plasmablasts, and increase in IgM+ B cells. Stress also enhanced intestinal permeability. Sphk2 knockout mice presented a cytokine-expression profile towards a boosted Th17 response, lower expression of claudin 3,4,7,8, and structural abnormalities in the colon. Intestinal pathophysiology should consider stress and S1P as modulators of the immune response. S1P-based drugs, including Sphk2 potentiation, represent a promising approach to treat IBD.

At the receptor level, stress exposure did not affect S1PR1 ( Fig. 1j) but downregulated S1PR2 mRNA expression (Fig. 1k). A stress effect was detected on S1PR3 mRNA levels (see Supplementary Fig. S1 online), and the protein expression revealed a stress-induced increment (Fig. 1l). No genotype differences were detected for S1P receptors.
Therefore, stress increased S1P and downregulated Sphk2 and its degradation enzymes. Moreover, it upregulated S1PR2 and S1PR3. The only Sphk2 deletion effect was a decrease in SGPP2.

Sub-chronic stress led to an immune dysregulation towards an inflammatory phenotype at multiple levels, and Sphk2 deletion could boost the T-helper lymphocyte 17 (Th17) response.
Hematoxylin & Eosin (H&E) evaluation showed inflammatory features in the colon after stress exposure (Fig. 2a-c). The absence of the Sphk2 gene is associated with architectural distortion of the crypts, which is exacerbated in stressed mice (Fig. 2a). However, it did not have an impact on the overall inflammatory score (Fig. 2b).
Toll-like receptor 4 (TLR4) is a crucial receptor for intestinal immune homeostasis. Stress increased TLR4 protein levels (Fig. 2d) and some of its downstream pro-inflammatory proteins, such as phospho mitogen-activated protein kinase (MAPK) p38 (p-p38) and inducible nitric oxide synthase (iNOS) (Fig. 2e,f). Sphk2 deletion did not modulate the TLR4 pathway. Transcript levels of anti-inflammatory mechanisms, such as 15-lipoxygenase  and liver X receptor β (LXRβ) (Fig. 2g,h) decreased after stress exposure, but no genotype effects were detected. Hence, sub-chronic stress caused inflammation in the colon and Sphk2 deletion did not aggravate it, but Sphk2−/− mice presented architectural abnormalities.

Scientific Reports
Sub-chronic stress enhanced intestinal permeability and downregulated some structural proteins, which were differentially expressed in Sphk2−/− mice. Transmission electron microscopy, plasma presence of fluorescein isothiocyanate-dextran average MW 4000 (FITC-D4000), and structural proteins in the colon were assayed to assess intestinal dysfunction. Stress enhanced intestinal permeability, caused bigger tight junction openings (Fig. 6a,b), and increased FITC-D4000 plasma levels in the stressed groups (Fig. 6c). Sphk2 deletion had no effects on intestinal permeability.
Hence, sub-chronic stress increased intestinal permeability but Sphk2 deletion did not affect it. Stress also decreased the expression of several structural proteins in the colon (ZO-1, Cldn5, 8) and Stress Sphk2−/− mice had lower levels of Cldn3, 4, 7, 8.

Discussion
Our data indicate that sub-chronic stress causes weight loss and a plasma corticosterone increase, two effects typically associated with stress axis activation 15 . The Sphk2 deletion does not affect the general stress response, but these mice have a smaller spleen. Spleen size is controversial since it is not associated with a hyper or hypoactivity. A previous study showed that Sphk2 deletion did not modify spleen lymphocyte number but could decrease CD4+ T-cells after certain immunological challenges 16 . Still, no effects compromising the normal immune system activity were recorded in Sphk2−/− mice beyond mild anaphylaxis with a fast recovery that, a priori, does not impact IBD 17 .
By and large, stress as a trigger of intestinal effects is well-documented 3 , and S1P pathways have been intensely studied in IBD 8,9 . However, to our knowledge, this is the first time showing that stress increases S1P in the colon possibly (and supported by the correlation analyses showed in Supplementary Information online) due to a downregulation of Sphk2 and its degradation enzymes, as it was also described in patients 18 . Our results further support that Sphk2 absence may cause an increase in S1P. Moreover, the SGPP2 downregulation detected in Sphk2−/− mice compared to WT would also contribute to the trend to higher S1P levels under control situations. Other studies 13,19 pointed to a S1P increase in both colon and plasma of Sphk2−/− animals, possibly due to an Sphk1 potentiation 20 . However, this mechanism did not seem to happen in our experimental setting, at least at a mRNA expression level. Undeniably, further experiments are required to give a full mechanistic explanation of the S1P pathway functioning in the colon after stress exposure. Figure 2. Inflammatory evaluation, TLR4 pathway expression, and anti-inflammatory mechanisms in the colon of WT and Sphk2−/− mice after sub-chronic stress exposure. H&E staining of 10 µm descending colon sections: individual qualitative evaluation (1-4) of parameters related to inflammation: severity, extent, hyperplasia, goblet cell loss, cryptitis, crypt abscesses, erosion, ulceration, granulation tissue, irregular crypts, crypt loss, and villous blunting (a), overall inflammatory score (1-4) (b), and representative images of each group (c), TLR4 protein expression (d), p-p38 protein expression (e), iNOS protein expression (f), LOX-15 mRNA levels (g), and LXRβ mRNA levels (h). For a and b, data are means of two sections per mice and 5-8 mice per group. For d-h, data are means ± SEM of 5-8 mice per group. The densitometric data of the respective bands of interest are normalized by β-actin (lower band). Blots were cropped (black lines) to improve the clarity and conciseness of the presentation. Two-way ANOVA considering stress (S) and Sphk2−/− as independent variables followed by Tukey's post hoc; **S p < 0.01, ***S p < 0.001; @ (interaction) p < 0.05; + p < 0.05, ++ p < 0.01 vs Control WT;   . Blots were cropped (black lines) to improve the clarity and conciseness of the presentation. Two-way ANOVA considering stress (S) and Sphk2−/− as independent variables followed by Tukey's post hoc; *S p < 0.05, **S p < 0.01, ***S p < 0.001; #Sphk2−/− p < 0.05, ##Sphk2−/− p < 0.01; @ (interaction) p < 0.05, @@ p < 0.01; www.nature.com/scientificreports/ Sub-chronic stress does not change S1PR1, downregulates S1PR2, and induces S1PR3. Although relevant functions in IBD, only chronic but not acute inflammation affect S1PR1 expression 21 . S1PR2 activity has been associated with Na + /K + ATPase inhibition responsible for diarrhea 22 , healthy epithelial barrier function 23 , and suppression of Th17 response 24 . Inflammation 25 and bacteria 26 signal through S1PR3 as has been proved in several tissues; however, there is little information about its role in IBD. Considering all this evidence, our results showing a decrease in S1PR2 and an increase in S1PR3 could indicate a deleterious effect in the colon. S1P accumulation could lead to inflammation, and our results and correlation analyses back up this assertion. Sub-chronic stress causes colon tissue inflammation, activating TLR4 pathways and weakening the expression of some anti-inflammatory molecules. The signature of bacterial TLR4 response appeared in patients with IBD and some polymorphisms were connected to IBD risk 27 . Previous results in our model showed an increase in iNOS expression 15 , and p38 MAPK activity played a role in inflammation-induced colon cancer 28 . Regarding pro-resolving mediators, other studies pointed to a similar decrease in 15-LOX 29 and LXRβ 30 , in congruence with our results.
The combination of sub-chronic stress and Sphk2 deletion does not lead to a synergistic effect on S1P pathways and inflammation. Most research studies in colon use more aggressive models, such as dextran sodium sulfate (DSS), to evaluate the immune response and the S1P role. Particularly, a study found an increase in colon S1P and inflammatory markers; both exacerbated in Sphk2−/− mice 13 . Our stress protocol is sub-chronic and milder than DSS, being chronic exposure the most connected to pathology development 31 . We hypothesize that the subtle changes caused by a sub-chronic stress exposure could shed light on the stress-induced mechanisms contributing to the worsening of IBD clinical activity but are insufficient to impact the processes controlled by Sphk2 at this inflammatory stage. Undoubtedly, future investigation considering these variables will need to be undertaken. Nevertheless, general H&E evaluation showed changes in some parameters in the Control Sphk2−/− mice compared to Control WT (severity, goblet cell loss, irregular crypts, and crypt loss), which were exacerbated in the Stress Sphk2−/− group. These results and the following related to Th17 response, claudin expression, and crypt architecture, suggest the consequences of Sphk2 absence and S1P involvement in intestinal pathophysiology.
A cytokine panel was performed as an indirect measure of immune signaling towards different immune populations. Our sub-chronic stress protocol induces an increase in CX3CL1, which participates in the recruitment and activation of immune cells, agreeing with previous studies showing an IBD improvement after blocking CX3CL1 signaling 32 . OSM, a pleiotropic cytokine of the IL-6 family recently identified as a biomarker of IBD 33 and a promoter of intestinal inflammation related to treatment failure 34 , is upregulated after stress exposure. A similar effect is observed for IL-1β, a pro-inflammatory cytokine involved in Th17 response by promoting an IL-17a production, also shown in our experimental setting 35 . Even more, Th17 cells require the IL-6 receptor activity, which can also bind OSM 36 . Up-regulation of IL-12, IL-22, and IL-23 correlate with other IBD models and current therapies against IBD, evincing the importance of Th17 in bowel inflammation 37 . Interestingly, a dual role has been described for IL-33, acting as an alarmin in acute phases but protecting through Foxp3+ Treg cells in chronic stages of the disease. Considering the sub-chronic nature of our stress protocol, the IL-33 increase is consistent with its involvement in the early stages of IBD 38 . This timeframe mimicked by our stress model could explain the IL-10 induction detected. Mice with a defective IL-10 response spontaneously develop IBD over time. However, the anti-inflammatory properties of IL-10 are crucial in the first days after an immunological insult, trying to restore homeostasis. For this reason, other studies also failed to detect acute colon IL-10 variations but showed a decrease at longer times 39 .
Here, the Sphk2 deletion upregulates OSM, IL-17, IL-22, IL-23, and IL-33, with the highest values expressed in the stressed-knockout mice. These results indicate that the Sphk2 absence could interfere with the adaptive immune response in the colon, making mice prone to a more robust Th17 response. Moreover, the positive correlations among the analyzed ILs further point to this Th17 polarization. Some studies connected the S1P signaling with T-cell differentiation 40 , but controversial results about the Sphk2 role in Th17 response have been reported. The use of Sphk2 inhibitors in a psoriasis model blocked the Th17 differentiation 41 , and both Sphk1 and Sphk1/Sphk2 inhibitors attenuated it 42 . However, these studies showed a decrease in S1P levels contrary to the increase detected in the Stress Sphk2−/− group in our model. Indeed, a recent study has described a strong T cell immunopathology in Sphk2−/− mice after challenged with a virus 43 . Considering the relevance of Th17 in IBD, our results suggest further mechanisms that should be considered when studying S1P pathways and possible actions of S1P-based drugs.
IBD research is revisiting B cells due to their critical contribution to immune homeostasis in the colon. Our experimental setting explores B cell differentiation to produce IgA and IgM. Sub-chronic stress causes a drop in plasmablasts and an increase in plasma cells. Focal or diffuse basal plasmacytosis has been recognized as the earliest feature with the highest predictive value for UC diagnosis 44 . Regarding IgA and IgM, stress decreases IgA+CD45+ cells, B cells, and plasmablasts (which could explain the general drop in total plasmablasts) but increases IgM+ B cells. However, the DSS model caused an increment in IgA+ cells and a drop in IgM+ 45 , contrary to our results. A feasible explanation relies on the disease stage simulated by our sub-chronic stress model. IgA represents a first defense line against pathogens in the gut, and this stress protocol has reported a decline in its levels 15 . IgM+ cells are crucial in acute immune challenges, and a weakened IgA response together with intestinal permeability could be rooted in the loss of immune tolerance against commensal bacteria. As IBD progresses, uncontrolled inflammation exhausts IgM and induces IgA in response to multiple interactions with bacteria due to tissue damage. Interestingly, CD patients in remission showed elevated IgM+, supporting its role in inflammatory control 46 . Undeniably, more exhaustive studies are mandatory to clarify the role of B cells. Regarding Sphk2 involvement, only an interaction effect between stress and genotype variables but with no significant differences within groups was detected in IgM+ plasma cells. A study described that the pharmacological inhibition of Sphk2 had minimal or no impact on basal levels of circulating human B cells from healthy Other studies pointed to S1P as a regulator of IgA-producing cells 48 , agreeing with the described stress effects. However, Sphk2 deletion did not modify S1P levels in our experimental setting and, therefore, seems not to modify B cell differentiation. Inflammation and intestinal barrier dysfunction are two phenomena reciprocally regulated. Stress increases intestinal permeability 15 , regardless of the genotype. Nevertheless, another study found an increase in FITC-D4000 permeability in Sphk2−/− mice, suggesting interference with normal epithelial cell growth 49 . Epithelial integrity especially relies on the intercellular junctions between adjacent cells. Considering this and the irregular crypts detected in Sphk2−/− mice, ZO-1 and claudins were analyzed. On the one hand, stress decreases ZO-1 protein expression and Cldn5, 8; thus, impairing colon epithelial function. Up-regulation of ZO-1 mRNA is consistent with our hypothesis of the homeostatic system trying to respond at this sub-chronic stage, but Sphk2−/− mice do not show this recovery effect. On the other hand, Sphk2 deletion is also responsible for lower levels of Cldn3, 4, 8. While Cldn8 expression is restricted to tight junctions, Cldn3, 4, 5, 7 are also located in the lateral plasma membrane 50 . This particular position has been associated with an active role in dynamic remodeling that could be related to the irregular crypts present in Sphk2−/− mice. Interestingly, the significant negative correlations obtained between crypt architecture defects and ZO-1, Cldn4, 8 have also been suggested in the literature 51,52 . Moreover, downregulation of Cldn7 leads to a defective mucosal architecture 53 and is decreased in Stress Sphk2−/− mice in comparison to Stress WT. Structural abnormalities in the blood-brain barrier of Sphk2 knockout animals further strengthen the idea of the claudin importance for a healthy barrier function 54 . Thus, both sub-chronic stress and Sphk2 deletion modify the expression of structural proteins and claudin differences could be related to the structural abnormalities detected in Sphk2−/− mice.
In a nutshell, our novel findings are summarized and illustrated in Fig. 7. Taken together, data presented herein reinforce the idea that sub-chronic stress induces colonic inflammation by modulating S1P pathways, dysregulating innate and adaptive immune response, and enhancing intestinal permeability. Sphk2 deletion did not significantly contribute to the stress effects in this model, but the Th17 polarization and crypt architecture abnormalities detected in these animals lay the scene for further investigations. Hence, both stress and S1P can modulate the immune system response in the colon and should be considered in intestinal pathophysiology. A critical limitation of this study hinges on the usage of the Sphk2−/− model. Hematopoietic and extra-hematopoietic sources of S1P seem to have different roles in inflammation, and our model is a germinal knockout unable to analyze these differences that could impair the immune response 55 . Moreover, the size of our control groups could not have enough statistical power to detect relevant differences between WT and KO animals under control situations. Despite these limitations, our research supports that pharmacological modulation of S1P-druggable candidates, including Sphk2 potentiation, represents a promising alternative for future studies.  Experimental design. Mice from both genotypes were randomly assigned to stress or control groups, and four experimental groups were formed: Control WT (n = 5), Control Sphk2−/− (n = 5), Stress WT (n = 8), and Stress Sphk2−/− (n = 8) (see Supplementary Fig. S2 online). Sphk2 deletion was confirmed by genotyping (see Supplementary Fig. S3 online).

Stress protocol.
A sub-chronic stress mixed model based on immobilization and ultrasound exposure for 2 h/day during four consecutive days was employed 15 (see Supplementary Fig. S2 online). All experimental procedures started at 11 am to avoid circadian changes affecting the stress response. All the animals were sacrificed at the end of stress exposure using sodium pentobarbital 320 mg/kg i.p. (Vetoquinol, Magny-Vernois, France).

Preparation of biological samples. Blood samples were obtained by cardiac puncture, anti-coagulated
in the presence of ethylenediaminetetraacetic acid (EDTA) 1% w:v, (1 vol EDTA per 50 vol blood), and centrifuged at 1 500 rpm for 15 min to obtain plasma. An aliquot was immediately protected from light and assayed for FITC-D4000 determination. The rest of the plasma was frozen at -80 °C until assayed. After blood collection, mice were perfused via ascending aorta with sterile saline solution 0.9% to remove blood from tissues. Descending colon was excised and cut into pieces for the different analyses (see Supplementary Fig. S2 online). For biochemical determinations, samples were frozen at − 80 °C until assayed.
Intestinal permeability assay. Mice were fasting for 6 h before the sacrifice, but they had ad libitum access to water. One hour before sacrifice, FITC-D4000 dissolved into drinking water (600 mg/kg) was administered by oral gavage. After obtaining plasma, fluorescence was read (excitation = 485 nm; emission = 528 nm) using a Synergy 2 (BioTek Instruments, Inc., Winooski, VT, USA) spectrophotometer. FITC-D4000 plasma concentration was calculated using a standard curve (100-1.5625 µg/ml). H&E staining. Two descending colon fragments of around 0.5 cm per animal were washed gently with phosphate-buffered saline (PBS) 1× pH = 6 and prepared to obtain transverse and longitudinal sections. 10 µm slices were collected using a Leica CM1950 cryostat (Leica, Wetzlar, Germany) and placed on Polysine Microscope Slides (Thermo Fisher Scientific, Waltham, MA, USA). H&E staining procedure consisted of successive incubations following this protocol: tap water 1 min; Mayer's hematoxylin solution (MilliporeSigma, Burlington, MA, USA) 3 min; tap water 1 min; eosin Y 1% alcoholic (MilliporeSigma) 2 min; Milli-Q water 1 min; ethyl alcohol 50º-70º-96º-100º 5 min each; ethyl alcohol 100º 1 min; Xylol 1 min; Xylol 5 min. DPX Mountant for Histology (MilliporeSigma) was employed for observation in ZEISS Axioplan-2 (Zeiss, Oberkochen, Germany). Three members of the team individually evaluated intestinal inflammation following a guide 57 . The different parameters for each section are the average of the three evaluations, and the final parameter score of each animal is the average of its two different colon sections. The overall inflammatory score of each animal relies on a general evaluation detailed in the guide and based on the individual parameter assessment.    www.nature.com/scientificreports/ per section. When possible, we measured the separation between the cell membranes of the adjacent epithelial cells along the length of the TJ complex (0.33 μm approximately) from the apical membrane and reported the maximum distance found between cell membranes for each TJ measured. Quantification was performed using the Image J software (NIH ImageJ, National Biosciences). Primer oligonucleotides for PCR were designed with the Primer3 tool. Target specificity was checked by in silico PCR using the USCS GenomeBrowser and Blast (NCBI) for cDNA and genomic DNA; only primer pairs with no unintended targets were selected (see Supplementary Table S1 online). Relative mRNA concentrations were calculated from the take-off point of reactions using the included software, and Glyceraldehyde-3-Phosphate Dehydrogenase (GAPDH) levels were used as a housekeeper.

Statistical analyses.
Data in text and figures are expressed as mean ± standard error of the mean (SEM).
The Grubbs' test was performed to identify outliers. Considering stress and Sphk2−/− as independent variables, a two-way analysis of variance (ANOVA) was used, followed by Tukey's post hoc test for multiple comparisons. All the results of the ANOVA analyses (F values and dfs) are included in Table 1. For the particular case of Sphk2 mRNA analysis, a two-tailed Student t-test between Control WT and Stress WT was performed. A p-value < 0.05 was considered statistically significant. Data were analyzed using GraphPad Prism 8 (GraphPad Software, San Diego, CA, USA). To further support manuscript discussion, Pearson's correlations between several parameters were analyzed (see Supplementary Figs. S9, S10, S11 and Supplementary Tables S2, S3, S4 online).
Ethics declarations. All experimental procedures adhered to the guidelines of the Animal Welfare Committee of the Universidad Complutense (PROEX052/17) following European legislation (2010/63/EU), and they were carried out in the Animal facility CAI-UCM. Animal studies are reported in compliance with the ARRIVE guidelines, and all efforts were made to minimize animal suffering and reduce the number of animals.

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.