Downregulation of Dickkopf-1 Augments the Therapeutic Effects of Endometrial Regenerative Cells on Experimental Colitis

Background We have demonstrated that endometrial regenerative cells (ERCs) are mesenchymal-like stromal cells and can attenuate experimental colitis, however, its underlying mechanism needs further investigation. Dickkopf-1 (DKK1), a glucoprotein secreted by mesenchymal stromal cells (MSCs), is a classical inhibitor of Wnt/β-catenin pathway which is closely associated with the development of colitis. Therefore, the objective of this study was to investigate whether ERCs could also secret DKK1, and whether the downregulation of DKK1 (DKK1 low -ERCs) would enhance the therapeutic effects of ERCs in attenuation of experimental colitis. Methods BALB/c mice were given 3% dextran sodium sulfate (DSS) for 7 consecutive days and free tap water for 3 days sequentially to induce experimental colitis. Unmodied ERCs, IL-1β-treated ERCs (DKK1 low -ERCs) and glucocorticoid-treated ERCs (DKK1 high ERCs) were injected (1 million/mouse/day, i.v. ) on day 2, 5 and 8 respectively. Colonic and splenic samples were harvested on day 10 after DSS-induction. Results It was found that DKK1 low -ERC treatment markedly attenuated colonic damage, body weight loss and colon-length shortening in colitis mice. Compared with other treatments, cell populations of CD4 + IL-4 + Th2, CD4 + CD25 + FOXP3 + Treg, and CD68 + CD206 + macrophages in spleens were also signicantly upregulated in DKK1 low -ERC group ( p < 0.05). In addition, lower expression of pro-inammatory (TNF-α and IFN-γ), but higher levels of anti-inammatory cytokines (IL-4 and IL-10) and β-catenin were detected in colons in DKK1 low -ERC group ( p < 0.01 vs. other groups). Conclusions DKK1 low -ERCs display augmented immunoregulatory ability and group vs. unmodied ERC group, p < 0.05), while reduced in DKK1 high -ERC group (DKK1 high ERC group vs. unmodied ERC group, p < 0.05). These results suggest that DKK1 low -ERCs mediate the colitis therapy in regulating macrophage phenotypes and promoting immunosupressive M2 phenotype cell increase, which would help to diminish the injury from acute immune response in colons.

In present study, our results demonstrated that ERCs, as mesenchymal-like stromal cells, could secrete an considerable amount of DKK1, and ERCs with downregulated DKK1 secretion (DKK1 low -ERCs) exhibited more powerful therapeutic effects in attenuation of experimental colitis.

ERC preparation
Primary human ERCs were isolated from menstrual blood by a density gradient centrifugation method in accord with previous study [40]. In brief, the mononuclear cells were rstly separated from menstrual blood and then suspended in the Dulbecco's modi ed Eagle's medium (DMEM) high glucose which was supplemented with 1% penicillin/streptomycin and 10% fetal bovine serum. Then cells were seeded in 10 cm dishes and cultured at the 37℃ 5% CO2 condition incubator. ERCs would adhere to the bottom of dishes after overnight incubation and the culture medium was changed every two days to wash away the non-adhered cells. Two weeks later, when cells expanded to 80-90% area of dishes and displayed a spindle-shaped morphology, we split and passaged down the ERCs as the rate of 1:3. Typical cell surface markers of ERCs were detected by a ow cytometry as previous study described [17].
In vitro, we harvested the 3 rd to 7 th generation of ERCs, divided each generation of cells into 3 groups (unmodi ed ERCs, GC-treated ERCs, and IL-1β-treated ERCs), and inoculated them at a concentration of 3.5×10 5 /ml with 2.5ml culture media. In an attempt to alter DKK1 secretion, we added glucocorticoid (GC, 100 nM, as the stimulator) or IL-1β (10ng/ml, as the inhibitor) [26] into respective groups. After cultured for 48 hours, supernatants in each group were collected to implement ELISA analysis. Fifth generation of ERCs, the most suitable candidate for treatments, were prepared for RT-PCR test to further analysis the DKK1 secretion changes.
For in vivo use, we selected the 5 th generation of ERCs, and pre-treated these candidate cells with GC (100nM) or IL-1β (10ng/ml) for 48 hours respectively. Then ERCs were harvested, washed and suspended at a concentration of 1×10 6 /ml for the following therapy.

Experimental groups
The experimental colitis was induced by supplying the mice with 3% (wt/vol) DSS (MP Biochemicals) dissolved tap-water as previous studies described [16]. In this current study, 24 BALB/c mice were randomly allocated into 4 groups: untreated group, unmodi ed ERC group, DKK1 high -ERC group (GCtreated ERCs), DKK1 low -ERC group (IL-1β-treated ERCs) (n=6). All experimental groups were rstly supplied with 3% (wt/vol) DSS (MP Biochemicals) soluted water for seven days, and then replaced with the non-DSS tap water. ERCs or pre-treated ERCs were suspended in phosphate buffered saline (PBS) and injected into experimental mice (1×10 6 cells/ml/mouse, i.v.) on day 2, 5 and 8, respectively. Untreated group was also given the equal volume of PBS as control. Mice body-weights, general conditions and fecal characters were monitored and kept into records daily, convenient for the Disease Activity Index (DAI) assessment and other statistical calculations. DAI is an indicator for disease activity which can comprehensively re ect the severity of in ammation in mice. Its score was calculated by assessing weight loss, fecal character and stool blood, accord to the scoring system (Min = 0, Max = 4) directed by Murthy et al [41].
On day 10, all mice were sacri ced after being fasted for 8 hours. Colons were dissected carefully from ileocecal junction verge to anus, and their lengths were measured. Then, samples were washed with PBS to clean away the contents and longitudinally severed into two parts. One part was xed in 10% formalin buffer preparing for pathology analysis, and the other was reserved at -80 °C for other experiments. Spleen samples were also harvested and split into two parts. One was immediately ground in PBS for FACS; the other was stored at -80°C for ELISA test.
Enzyme-linked immunosorbent assay (ELISA) ELISA was carried out according to the manufacturer's instructions (Boster, Wuhan, China). Supernatants in culture media of ERCs were collected and prepared for measuring the DKK1 secretion level. Equal quality (30mg) of same area intestinal tissues in each group were gathered and grounded with high e ciency tissue lysate buffer (RIPA) and phenylmethyl sulfonyl uoride (PMSF) (Solarbio, Beijing, China) for testing the level of IFN-γ, IL-4, IL-10 and β-catenin. Splenic tissues frozen in -80°C were also ground for detecting the β-catenin expression level. The reaction absorbance was determined at 450 nm with the Microplate Reader (Tecan, Mannedorf, Switzerland) and each sample was performed in duplicates to lessen the error.

Statistical analysis
Experimental data was presented as mean ± standard deviation (SD) and analyzed by SPSS 19.0. Data variance was evaluated by using one-way analysis of variance (ANOVA) (groups≧3) or unpaired twotailed student's t test (groups=2) after normality test. The differences between groups were considered signi cant with p values ≤ 0.05 in statistics.

Results
Pretreatment with Glucocorticoid or IL-1β Changed the DKK1 Secretion by ERCs To determine whether ERCs would secret DKK1 and whether this secretion would be affected by glucocorticoid (GC) or IL-1β, the level of DKK1 in the supernatant of different pre-treated ERCs was measured by ELISA and its concentration was shown signi cantly different in Figure 1. Among which, the 5 th generation of ERCs were recorded with the lowest DKK1 expression when compared with other generation cells ( Figure 1A; *p < 0.05, **p < 0.01, ***p < 0.001). Furthermore, in each generation, the DKK1 secretory amount in IL-1β-treated ERC group was shown lower than that in unmodi ed ERC group and GC-treated ERC group respectively. Speci cally, to further clarify whether there exists statistical difference between different pre-treated ERCs, culture supernatant and homogenate of 5 th generation of ERCs were collected and analyzed for the DKK1 expression at the protein and mRNA level separately. As shown in Figure 1B and 1C, the DKK1 expression was strikingly decreased in IL-1βtreated ERC group (unmodi ed ERCs vs. IL-1β-treated ERCs: p < 0.05, Figure 1B; p < 0.05, Figure 1C), but when ERCs were pre-treated with glucocorticoid (GC), DKK1 expression boosted up (unmodi ed ERCs vs. GC-treated ERCs: p < 0.01, Figure 1B; p < 0.001, Figure 1C).

DKK1 low -ERCs Markedly Ameliorated the Symptoms of DSS-induced Colitis
In present study, we used the 5 th generation of ERCs for the following in vivo experiments due to its lowest DKK1 expression. After the fth day of DSS induction, mice begin to exhibit signi cant bloody stool, weight loss and lethargy. But, in the unmodi ed ERC group, following the treatment of ERCs, bloody stool ( Figure 2A) and body weight loss were found with moderate relieve and improvement ( Figure 2B, unmodi ed ERC group vs. untreated group, p < 0.001). Moreover, the therapeutic effects were further improved when the colitis mice were treated with DKK1 low -ERCs (IL-1β-treated ERCs) (Figure 2A; Figure   2B, vs. unmodi ed ERCs, p < 0.01), while diminished when treated with DKK1 high -ERCs (GC-treated ERCs) ( Figure 2A; Figure 2B, vs. unmodi ed ERCs, p < 0.01). In addition, we have also analyzed the Disease Activity Index (DAI) score in each group. As shown in Figure 2C, the DAI score was apparently decreased in unmodi ed ERC group, when compared with that in untreated group (unmodi ed ERC group vs. untreated group, p < 0.001). While in DKK1 high -ERC group, the DAI score raised up (p < 0.05, DKK1 high -ERC group vs. unmodi ed ERC group), indicating the severe colon injury, but signi cantly decreased in DKK1 low -ERC group (DKK1 low -ERC group vs. unmodi ed ERC group, p < 0.05). Given the above results, it suggests that DKK1 is closely associated with the therapeutic effect of ERCs on the development of DSSinduced colitis.

DKK1 low -ERCs Relieved the Histopathological Damage of DSS-induced Colitis
To observe the changes of colonic morphology, colon samples were collected and their lengths were measured as shown in Figures 2D and 2E. We found that the average length of colons in untreated group is 5.3cm (n=6), indicating a signi cant reduction in colon length due to the severe intestinal in ammation. While, the average length in DKK1 low -ERC group is 7.0 cm, higher than unmodi ed ERC group (6.3cm; Figure 2E, DKK1 low -ERC group vs. unmodi ed ERC group, p < 0.01) and the length in DKK1 high -ERC group is 5.5cm, lower than the unmodi ed ERC group ( Figure 2E, DKK1 high -ERC group vs. unmodi ed ERC group, p < 0.001), which suggested that tissue injury and structural damage were obviously alleviated in DKK1 low -ERC group. Pathological examination also con rmed the above ndings.
DSS intake caused severe injury, while in DKK1 low -ERC group ( Figure 2F a-d), the pathological condition of colon was strikingly better, showing slight damages to crypt structure, glands and epithelium cells, mild in ammatory cell in ltration and less goblet cells loss. Also, the histopathological score of DKK1 low -ERC group was lower than that of the unmodi ed ERC group ( Figure 2G, p < 0.001), and unmodi ed ERC group was lower than the DKK1 high -ERC group ( Figure 2G, p < 0.001). Given together, these results indicate that DKK1 low -ERCs could exhibit optimized therapeutic effect in relieving histopathological damages in DSS-induced colitis, and this optimized effect of ERCs is, at least in part, mediated by DKK1 low expression.

DKK1 low -ERCs Increased the Population of M2 Phenotype Macrophages in Colitis Mice
M2 phenotype macrophages is one of the main subtypes of macrophages which plays the antiin ammatory role in the pathogenesis of experimental colitis. Our previous study has revealed that ERCs could promote the differentation of macrophages to M2 subtype. To determine whether DKK1 has an in uence in regulating M2 in colitis mice, anti-CD68 and anti-CD206 antibodies were used to measure the proportion changes in spleens. Compared with that in untreated group, M2 population was obviously raised in unmodi ed ERC group ( Figure 4B and 4E: untreated group vs. unmodi ed ERC group, p < 0.001).
Moreover, the population of M2 cells was further increased to a higher level in DKK1 low -ERC group (DKK1 low -ERC group vs. unmodi ed ERC group, p < 0.05), while reduced in DKK1 high -ERC group (DKK1 high -ERC group vs. unmodi ed ERC group, p < 0.05). These results suggest that DKK1 low -ERCs mediate the colitis therapy in regulating macrophage phenotypes and promoting immunosupressive M2 phenotype cell increase, which would help to diminish the injury from acute immune response in colons.
Taking together, these data indicate that down-regulating the DKK1 secretion of ERCs, could strikingly polished up the therapeutic effects on colitis and these effects might rely in part on promoting the expression of anti-in ammatory mediators and inhibiting the production of pro-in ammatory mediators.

DKK1 low -ERCs Increased β-catenin Expression in Colon and Spleen
To determine whether DKK1 low -ERCs have an in uence on the activation of Wnt / β-catenin signaling, we measured the β-catenin level in colons and spleens, which is essential for this signaling transduction. As shown in Figure 6, the β-catenin protein level in colons and spleens were increased in unmodi ed ERC group when compared with that in untreated group (p < 0.001, Figure 6A; p < 0.001, Figure 6B). However, in DKK1 low -ERC group, the β-catenin level was further signi cantly raised (p < 0.001, Figure 6A; p < 0.001, Figure 6B; vs. unmodi ed ERC group), but in DKK1 high -ERC group, it declined apparently (p < 0.05, Figure  6A; p < 0.01, Figure 6B; vs. unmodi ed ERC group). In addition, when we analyzed the β-catenin mRNA expression changes in colons, it showed that the trend ( Figure 6C) was consistent with the changes in protein level ( Figure 6A). Taken together, it suggested that DKK1 low -ERCs, signi cantly increased the βcatenin expression, which exhibited anti-in ammatory effects in colitis, and thus eventually enhanced the therapeutic effects of ERCs in DSS-induced colitis.

Discussion
Wnt signaling antagonist DKK1, a secreted glycoprotein, expresses in many species, including human derived stromal cells and mouse colonic tissues [25,45]. DKK1 possesses a conservative gene sequence and its expression in stromal cells can be down-regulated by IL-1β or up-regulated by glucocorticoid stimulation [26]. Relying on competing with Wnt-ligands for LRP5/6 receptor, DKK1 can uniquely inhibit the Wnt/β-catenin signaling transduction, which exhibits anti-in ammatory effects in the development of ulcer colitis [38]. Therefore, in present study, we hypothesized that ERCs, like other stromal cells, could also secrete DKK1, which blocks Wnt mediated anti-in ammatory signal transduction, and thus weaken the immunoregulation of ERCs. Furthermore, we could optimize the therapeutic effect of ERCs on colitis by downregulating the expression of DKK1.
To carry out the current study, we rstly determined DKK1 levels in different groups, including unmodi ed ERC group, GC-treated ERC group and IL-1β-treated ERC group. Our results showed that ERCs indeed secret a considerable amount of DKK1, just like MSCs [46], and the expression of DKK1 could be markedly changed after ERCs stimulated by IL-1β or glucocorticoid (GC), not only at the protein level but also at the level of RNA.
Next, we carried out the following experiments in vivo and found that in DKK1 low -ERC (ERC pretreated with IL-1β) group, clinical characters and pathological manifestations of colitis were strikingly ameliorated, in which the immune-regulatory effects of ERCs may be enhanced by low DKK1 production, as a result of Wnt/β-catenin pathway being activated [34,47]. Comparing with that in DKK1 high -ERC (ERC pretreated with GC) group, we con rmed that DKK1 plays a vital role in affecting therapeutic function of ERCs on experimental colitis.
As we know, the balance between CD4 + T cells (Th1/Th2, Th17 and Tregs) is essential for sustaining the intestinal homeostasis and closely involved in the development of UC [48]. In present study, we analyzed the CD4 + T cell population changes in splenocytes by FACS, and found that, Th1 and Th17 populations in DKK1 low -ERC group, were the lowest among experimental groups, while the populations of Th2 and Tregs in DKK1 low -ERC group presented the highest. These data indicated that in the immune-regulatory function of ERCs, DKK1 participates in the immunoregulation effects on CD4 + T cells.
Among immunocytes activated, DCs and macrophages, are also acting a role in the development of DSSinduced colitis [49]. DCs, the major part of antigen-presenting cells (APCs), exert its effects by MHC class II and CD86 expressed on their cytomembranes [50], promote the migration of immune cells into damaged colons, and then result in inducing T cell-mediated immune response. Macrophages, are illustrated with a diverse plasticity, and its differentiation can be driven by the surrounding settings [51]. Macrophages generally differentiated into two phenotypic subsets: M1 pro-in ammatory sub-type and M2 selectively activated anti-in ammatory sub-type [52]. M1 phenotype cells, which are known as pro-in ammatory macrophages, have the ability to generate tumor necrosis factor alpha (TNF-α), and nitric oxide (NO) to exacerbate the in ammation [53]. While, M2 phenotype cells exhibit immune-regulatory and antiin ammatory effects by producing IL-10, which could promote tissue repairs by its negative immune regulation ability [54].
Thus, in the present study, we evaluated the proportion changes of macrophages and DCs in spleens. As the results shown, the proportion of mature DCs (CD11c + MHCII + / CD11c + CD86 + ) in DKK1 low -ERC group was lower than that of other groups. Meanwhile, the proportion of M2 phenotype macrophages (CD68 + CD206 + ) rised up to the highest. Taking together, these observations suggested that DKK1, secreted by ERCs, negatively affect the immunomodulatory effects of ERCs on macrophages and DCs. In addition, when we downregulate the DKK1 expression of ERCs, this immunoregulatory mechanism tend to be active further and exert effective therapeutics in colitis.
Accumulating documents illustrated that in ammatory mediators in colons (IL-4, IL-10, SOD, IFN-γ, TNF-α, COX-2, MPO and iNOs) orchestrated the pathogenesis of ulcer colitis temporally and specially [55,56]. IL-4, IL-10 and SOD are anti-in ammatory mediators with plenty of protective effects in colitis. IL-4 could assist in inducing T helper 2 cell (Th2) responses, inhibiting Th17 cell development and polarizing macrophages toward M2 phenotype [57][58][59]. IL-10 partcipates in suppressing the antigen presentations and the synthesis of pro-in ammatory cytokines in colitis [60]. SOD, is a critical antioxidant enzyme, which could catalyze the dismutation of toxic superoxides and scavenge superoxide radicals in colons [44]. On the contrary, IFN-γ, TNF-α, MPO, COX-2 and iNOs are pro-in ammatory factors which often strongly exacerbate the in ammations cascade in colitis [61][62][63]. ERCs, used for the immune-related diseases, were reported to have an ability to ameliorate the production and accumulation of in ammatory mediators, and thus to relieve the local tissue injury [19,64]. But, whether down-regulating DKK1 expression in ERCs, could also modulate the imbalance of in ammatory mediator pro les, and exit a more provoking effect has not been illustrated till now. In our present study, as shown in results ( Figure 5), we found a signi cant higher expression of anti-in ammatory mediators (IL-4, IL-10 and SOD) but lower of pro-in ammatory factors in DKK1 low ERC group, when compared with that in unmodi ed ERC group.
Correspondingly, a lower expression of IL-4, IL-10 and SOD but high expression of pro-in ammatory mediators were witnessed in the DKK1 high -ERC group. Taking together, by analyzing in ammatory mediator changes, we believed that downregulating the DKK1 secretion of ERCs indeed improved in ammatory mediator pro les in the pathogenesis of UC.
Wnt/β-catenin signaling was recorded with anti-in ammatory effects in the development of ulcer colitis [35]. As previous studies described, with the activating of canonical Wnt signaling, β-catenin would accumulate in cytoplasm and regulates the expressions of Wnt-related genes [65]. But, in absence of Wntligand stimulus, cytosolic β-catenin will be degraded after rounds of ubiquitination and phosphorylation.
Thus, we measured the β-catenin expressions, both at the protein level and the RNA level, to investigate the activity of canonical Wnt pathway. As shown in gure 6, β-catenin production was signi cantly increased in DKK1 low -ERC group, not only in colons but also in spleens. Intriguingly, high expression level of β-catenin was in accordance with the powerful immune regulation ability of DKK1 low -ERCs exhibiting in colitis. Thus, we concluded that reducing DKK1 expression in ERCs could weaken the antagonistic effect of DKK1 on Wnt signaling, and then the β-catenin accumulation could be promoted and Wnt βcatenin pathway would eventually be activated, so as to improve the immunoregulatory effect of ERCs and optimize their therapeutic effects in the process of ulcer colitis.

Conclusions
In summary, our present study demonstrated that DKK1 low -ERCs, obviously exert a more effective immune regulatory ability and better therapeutic effects in DSS-induced colitis. We con rmed DKK1 low -ERCs signi cantly ameliorated the symptoms and pathological damages of DSS-induced colitis, and modulated the balance of immunocytes in spleens toward a tolerant status. DKK1 low -ERCs also apparently altered the in ammatory mediator pro les in colons, and thus, effectively suppressed the mucosal immune reaction and the detriments in colons. Taking together, these ndings impressed us with an encouraging and viable method to signi cantly enhance the therapeutic effect of ERCs on ulcer colitis by downregulating the DKK1 secretion. DKK1low-ERCs increased β-catenin expression in colonic and splenic tissues. The protein level of βcatenin in colons (A) and spleens (B) were respectively detected. Furthermore, the mRNA expression (C) level of β-catenin in colon was also demonstrated. Data were presented as mean ± SD (*p < 0.05, **p < 0.01, ***p < 0.001). Statistical analysis was performed by one-way ANOVA followed by the LSD test, n = 6. presented as mean ± standard deviation, and P values were calculated by using one-way ANOVA followed by the least signi cant difference (LSD) test. (n=6, *p < 0.05, **p < 0.01, ***p < 0.001). CD11c+CD86+DCs and CD68+CD206+Macrophages respectively. Data were mean ± SD (n=6, *p < 0.05, **p < 0.01, ***p < 0.001). P values were analyzed by one-way ANOVA followed by the LSD test. (E) Percentage of CD4+IL-17+Th17. (F) Percentage of CD4+CD25+Foxp3+Tregs. Data were mean ± SD (n = 6, *p < 0 .05, **p < 0 .01, ***p < 0.001). P values were calculated by one-way ANOVA followed by the LSD test. were calculated according to the scoring system directed by Singh [42] to assess the colonic injury quantitatively. Data were presented as mean ± standard deviation (SD) (*p < 0.05, **p < 0.01, ***p < 0.001). Statistical analysis was calculated by using one-way analysis of variance (ANOVA) followed by the least signi cant difference (LSD) test. DKK1 was secreted by ERCs and can be decreased by IL-1β or increased by GC. (A) The DKK1 level in supernatants of 3rd-7th generation were detected by ELISA kit. Each generation of cells was respectively treated with GC (100nM), IL-1β (10ng/ml) or nothing for 48 hours (n=3). The 5th generation of ERCs were recorded with the lowest DKK1 expression when compared with other generations (vs. the 5th generation of ERCs, *p < 0.05, **p < 0.01, ***p < 0.001). (B) DKK1 level in 5th generation supernatants (n=3). (C) DKK1 mRNA expression in 5th generation of ERCs (n=3). Data were presented as mean ± standard deviation (SD). Statistical analysis was calculated by using one-way analysis of variance (ANOVA) followed by the least signi cant difference (LSD) test.