Human Bone Marrow-Derived Mesenchymal Stromal Cells Reduce the Severity of Experimental Necrotizing Enterocolitis in a Concentration-Dependent Manner

Necrotizing enterocolitis (NEC) is a devastating gut disease in preterm neonates. In NEC animal models, mesenchymal stromal cells (MSCs) administration has reduced the incidence and severity of NEC. We developed and characterized a novel mouse model of NEC to evaluate the effect of human bone marrow-derived MSCs (hBM-MSCs) in tissue regeneration and epithelial gut repair. NEC was induced in C57BL/6 mouse pups at postnatal days (PND) 3–6 by (A) gavage feeding term infant formula, (B) hypoxia/hypothermia, and (C) lipopolysaccharide. Intraperitoneal injections of PBS or two hBM-MSCs doses (0.5 × 106 or 1 × 106) were given on PND2. At PND 6, we harvested intestine samples from all groups. The NEC group showed an incidence of NEC of 50% compared with controls (p < 0.001). Severity of bowel damage was reduced by hBM-MSCs compared to the PBS-treated NEC group in a concentration-dependent manner, with hBM-MSCs (1 × 106) inducing a NEC incidence reduction of up to 0% (p < 0.001). We showed that hBM-MSCs enhanced intestinal cell survival, preserving intestinal barrier integrity and decreasing mucosal inflammation and apoptosis. In conclusion, we established a novel NEC animal model and demonstrated that hBM-MSCs administration reduced the NEC incidence and severity in a concentration-dependent manner, enhancing intestinal barrier integrity.


Mouse Intestine Immunohistochemistry (IHC)
Intestinal samples were routinely formalin fixed-paraffin embedded according to a standardized protocol and cut into 5 µm-thick slices. Intestine sections were stained with the following antibodies: active mouse caspase-3 (R&D System, Minneapolis, MN, USA) and Ki67 antibody (R&D System, Minneapolis, MN, USA). In addition, quantifying Ki67 positive nuclei was performed on 3 sections/samples using the Image J software (n = 5).
Pups exposed to the experimental NEC protocol were dam-separated on PND 3 (approximately 72 h after birth) and placed in a neonatal incubator (Dräger, Lübeck, Germany) at 33 • C, with 60% of humidity, on a 12-h-dark/light cycle.
Pups from the groups subjected to experimental NEC were gavage fed every 3 h with the standard formula for term babies using a 1.9 F silastic catheter (Vygon, Aachen, Germany) starting 1 h after dam separation. Daily feeding volume increased as tolerated (30,40, and 50µL on PND 3, 4, and 5, respectively) ( Figure 1). Orogastric gavage was performed by introducing the silastic catheter in the esophagus of the pup by direct vision, connected to a 1 mL syringe filled with the exact volume of milk to administer. Before every administration, the catheter was primed due to the small volumes of milk used [66][67][68][69][70].
Pups from the groups subjected to experimental NEC were gavage fed every 3 h with the standard formula for term babies using a 1.9 F silastic catheter (Vygon, Aachen, Germany) starting 1 h after dam separation. Daily feeding volume increased as tolerated (30,40, and 50μL on PND 3, 4, and 5, respectively) ( Figure 1). Orogastric gavage was performed by introducing the silastic catheter in the esophagus of the pup by direct vision, connected to a 1 mL syringe filled with the exact volume of milk to administer. Before every administration, the catheter was primed due to the small volumes of milk used [66][67][68][69][70].
Animals were euthanized 72 h after dam separation (PND6) or earlier in case of distress or early clinical signs of NEC. Mice subjected to NEC and injected with PBS were used as a positive control. Breastfed mice, not subjected to NEC, were used as a negative control.

Bodyweight and Survival Rate
Bodyweight was assessed daily, starting on PND 2. In addition, the number and the approximate time of deaths were recorded daily for each experimental group.

Histological Injury Score of NEC in Neonatal Mice
The intestine was harvested, fixed in 10% buffered formalin, embedded in paraffin, sectioned (4 µm), and stained with hematoxylin-eosin using an internal protocol. In addition, small and large intestinal tissues were investigated histologically for intestinal damage by a histopathologist blinded to the experimental conditions.
The severity of the experimental NEC was classified from 0 to 4 using a previously validated scoring system [66].
The definition for each histological grade is as follows: grade 0: intact villi; grade 1: superficial epithelial cell sloughing; grade 2: high number of karyorrhectic nuclei, partially compromised villi structure, basal cells still present; grade 3: complete villous necrosis, but basal cells still appreciable; grade 4: transmural necrosis, basal membrane destroyed.
Samples with histological scores of 2 or higher were considered positive for NEC. Severe NEC was defined as an injury score of grade 3 or 4 [71].

Statistical Analysis
Bodyweight changes were presented as means ± SEM, and two-way ANOVA was used to study differences in growth between experimental groups over time. Survival data were represented by Kaplan-Meier curves and compared by log-rank (Mantel-Cox) test. NEC incidence was compared using Fisher's Exact test and expressed as a percentage. Gene expression analysis and evaluation of Ki67 positive nuclei are represented as means ± SEM. NEC mice treated with hBM-MSCs were compared to NEC mice injected with PBS as controls. Mann-Whitney test was used for statistical evaluation (*** p ≤ 0.001; ** p ≤ 0.01; * p ≤ 0.05). p-values less than 0.05 were considered significant. Graphs and p-values were obtained with GraphPad Prism 9 (GraphPad, La Jolla, CA, USA).

The Present NEC Model Based on Term Infant Formula Feedings Can Induce the Pathological Hallmarks of the Disease
In this study, we set up a new neonatal mouse model of experimental NEC based on previously published methods with slight modifications [43,66]. Representative histologic NEC injury grades are shown in Figure 2.
Mouse pups who underwent the NEC protocol (NEC group) showed a significantly lower survival rate compared with breastfed pups (p ≤ 0.0001; Log-rank Mantel-Cox test) ( Figure S1A,B (Supplementary Materials)).
Mouse pups who underwent the NEC protocol (NEC group) showed a significantly lower survival rate compared with breastfed pups (p ≤ 0.0001; Log-rank Mantel-Cox test) ( Figure S1A,B (Supplementary Materials)).

Body Weight
The starting body weight of mouse pups was comparable across experimental groups before the study on PND 2 and PND 3 (control vs. NEC, not statistically significant). However, control mice showed a remarkable increase in body weight from PND 2 to PND 6. The body weights in the NEC group were dramatically lower (day 4, p < 0.0001; day 5, p < 0.0001; day 6, p < 0.0001; two-way ANOVA with Sidak multiple comparisons post-hoc test) than those in the control group (Table 2).   Controls showed normal development and healthy intestine compared with pups that underwent experimental NEC ( Figure 3B). Controls showed normal development and healthy intestine compared with pups that underwent experimental NEC ( Figure 3B).

hBM-MSCs Reduced Apoptosis and Stimulated Tissue Regeneration in a Neonatal Mouse NEC Model
We further investigated the protective function of hBM-MSCs. Considering that hBM-MSCs could prevent apoptosis in damaged cells exposed to inflammatory insults, we evaluated the expression of the pro-apoptotic marker Caspase 3 on intestine sections [72].
Since the higher administered dose of hBM-MSCs completely abolished NEC incidence, we focused our analysis on the effects obtained by treatment with hBM-MSCs (1 × 10 6 ). While we found several areas positive for the pro-apoptotic marker Caspase 3, we observed a reduced expression of Caspase 3 upon treatment with hBM-MSCs (1 × 10 6 ). ( Figure 5A). Accordingly, when we measured the expression level of the anti-apoptotic gene Bcl2, we observed a significant overexpression in the RNA extract from the intestines of mice treated with hBM-MSCs (p = 0.008) ( Figure 5D, left panel).
In these mice, we also found a reduced expression of the pro-inflammatory cytokine gene IL1b (p = 0.028) ( Figure 5D, right panel) in accordance with the anti-inflammatory function exerted by MSCs [72].
We next evaluated whether hBM-MSCs accelerated the regenerative process upon NEC injury. We stained intestine sections from the different experimental groups with ki67, a marker of active proliferation. The intestines of mice exposed to NEC + PBS showed a higher number of epithelial cells positive for Ki67. On the contrary, in the intestines of pups treated with hBM-MSCs, Ki67+ cells were limited in the intestinal crypts where highly proliferating intestinal stem cells reside (p = 0.048) ( Figure 5B,C).
Moreover, we determined the expression level of ZO-1, one of the critical components of tight junctions (TJs), which is fundamental for mucosal repair [73]. Compared to control pups, pups exposed to NEC + PBS showed low expression levels of ZO-1. However, we found a robust increase in ZO-1 expression in mice exposed to NEC and treated with both hBM-MSCs concentrations, reaching levels similar to those observed in control pups (Control vs. NEC + PBS, p = 0.032; NEC + PBS vs. NEC + hBM-MSCs (0.5 × 10 6 ), p = 0.017; NEC + PBS vs. NEC + hBM-MSCs (1 × 10 6 ), p = 0.032) ( Figure 5E).  In these mice, we also found a reduced expression of the pro-inflammatory cytokine gene IL1b (p = 0.028) ( Figure 5D, right panel) in accordance with the anti-inflammatory function exerted by MSCs [72].
We next evaluated whether hBM-MSCs accelerated the regenerative process upon NEC injury. We stained intestine sections from the different experimental groups with ki67, a marker of active proliferation. The intestines of mice exposed to NEC + PBS showed a higher number of epithelial cells positive for Ki67. On the contrary, in the intestines of

Discussion
NEC still represents a devastating disease for preterm neonates. However, currently, preventive and therapeutic strategies are still scanty.
The study demonstrated that our new mouse model of NEC, based on administering full-term infants' formula that was not hyperosmolar, can successfully induce the disease. We detected an incidence of NEC ≥grade 2 of 50%, divided according to histological score, into 23.1% grade 2 and 26.9% grade 3 NEC. As expected, the pathology development was associated with a lower survival rate of around 40% and a lower body weight in mice that underwent NEC protocol compared to the control group.
Using this novel protocol for NEC induction, we demonstrated that hBM-MSCs treatment reduced NEC incidence in a concentration-dependent manner, with the highest dose of cells preventing the development of the disease. Furthermore, no grade 2 or higher was reported in the analyzed intestines concerning the NEC + PBS group. Interestingly, although not statistically significant, we already observed a global reduction in NEC incidence with the lower concentration of MSCs (0.5 × 10 6 ) compared to NEC + PBS mice, thus suggesting that the administration of hBM-MSCs was able to interfere with the pathology onset.
Recently, stem cell-based therapeutic approaches have been shown to reduce the incidence of NEC, representing a promising treatment [74]. Notwithstanding, their exact mechanism of action is still poorly understood [75,76].
Human BM-MSCs express high growth factor levels that play a key role in tissue regeneration [77]. Several works demonstrated that MSCs release growth factors to main-tain damaged cells and induce a regenerative program in tissue-resident stem cells. For example, MSC-derived hepatocyte growth factor (HGF) was established to protect the infarcted heart [78].
In our model of NEC, hBM-MSCs' administration was associated with the stimulation of tissue regeneration. Indeed, when we analyzed the expression of the proliferative marker Ki67, we found several areas of proliferation in the intestine of NEC + PBS mice, indicating that intestinal cells massively proliferate to repair the damaged tissue. In contrast, in mice treated with hBM-MSCs, we observed a reduced expression of Ki67 in the intestinal cells, with Ki67-positive cells located only at the basis of intestinal crypts, where intestinal stem cells reside and continuously proliferate to guarantee tissue cell turnover, suggesting that the injection of hBM-MSCs were able to accelerate the regenerative process.
Moreover, we also found a reduced expression of the pro-apoptotic factor Caspase 3 in NEC + hBM-MSCs mice. Intestines from mice injected with MSCs showed a regenerated tissue architecture with a barely detectable Caspase 3 signal. In contrast, intestines harvested from NEC + PBS mice were characterized not only by necrosis and dysmorphic tissue organization but also by the presence of several Caspase 3 positive areas, indicating a strong apoptosis induction in the damaged gut. Furthermore, the reduced expression of Caspase-3 was associated with a higher level of Bcl2 expression in hBM-MSC treated mice compared to controls (NEC + PBS), according to previous work showing a primary role of the Bcl2 pathway in the antiapoptotic function of MSCs [81]. We concluded that hBM-MSCs favored resident cell survival by preventing apoptosis, as previously reported in vitro and in vivo, mainly through direct paracrine mechanisms on primarily damaged cells or tissue-resident cells favoring regeneration [82][83][84][85]. hBM-MSCs may prevent excessive cell death by blocking the activation of a detrimental inflammatory program which could interfere with the regenerative process. In line with this, we also found a significant reduction in IL1b expression upon hBM-MSC administration compared to control mice (NEC + PBS).
Finally, we also demonstrated that hBM-MSCs preserved the integrity of the intestinal barrier in our model of NEC, which is fundamental for NEC resolution, considering that the intestinal epithelium offers a physical barrier that protects the host against microbial invasion [86]. Conversely, LPS administration leads to the failure of the intestinal barrier function, thus resulting in an increased mucosal permeability, which plays a key role in promoting bacterial translocation and inflammatory mediators' release [87,88].
Epithelial tight junctions (TJs) are involved in intestinal barrier function and permeability [89]. TJs are localized on the apical membrane of epithelial cells and create a shield to the movement of the solute between cells [90]. They are dynamic structures that can be detached and reassembled in response to stimuli such as LPS and inflammatory mediators [91]. Evidence has demonstrated that TJs disruption and loss of barrier function are often associated with several diseases [86,[92][93][94]. Previous studies have shown that interferon (IFN)-γ-induced damage is related to intestinal barrier disruption and the downregulation of the TJ protein ZO-1, whose decreased expression is enough to determine barrier dysfunction [94].
We hypothesized that the protective role exerted by hBM-MSCs on NEC onset and development could be related to intestinal barrier regeneration. Therefore, we analyzed ZO-1 expression in NEC mice treated with different doses of hBM-MSCs compared to vehicle and breastfed groups.
Our results showed that hBM-MSCs could act on intestinal barrier function by promoting ZO-1 expression in a concentration-dependent manner. In fact, we revealed similar ZO-1 mRNA expression levels between the highest stem cell dose and the breastfed group, suggesting that the most elevated hBM-MSCs' dosing enabled the counteracting of the disease-induced ZO-1 downregulation.
These findings suggest that hBM-MSCs, through their paracrine action, could protect from NEC onset by promoting intestinal barrier recovery by stimulating ZO-1 expression.
Altogether, our results indicated that hBM-MSCs reduced the level of apoptosis, prevented the strong induction of IL1b, and accelerated the regenerative process in our novel experimental model of human NEC.
There are some limitations to our study. First, since NEC is a complex multifactorial disease, none of the many different animal models developed so far can fully recapitulate the underlying pathophysiology of the disease [95]. Consistently, the present model could fail to reproduce all the pathological hallmarks of the disease. In addition, the decrease in body weight observed in all the pups subjected to experimental NEC, regardless of the treatment, is likely due to the technical difficulties inherent in artificial feeding, which were extensively described by many authors [18,20,96]. Moreover, we should underline that in our model, we used a full-term infant formula, which is not hyperosmolar and utterly different in terms of nutrient composition from murine milk, as reported in Table S3.
This could explain why, even if pups were fed every three hours with increasing daily doses for three days, they were always smaller than their breastfed counterparts.
Another limitation is that our experiments were designed to assess the changes occurring in mouse pups at the end of the NEC induction protocol. Therefore, it could not evaluate all the intestinal injuries and their recovery after different treatments.
One additional limitation is the lack of examination into the specifics of mucosal immunity and its significance in developing NEC. In particular, we did not address the role of secretory immunoglobulin A (sIgA), and given its recognized benefits in breast milk, further investigation is required to better understand its contribution to the development of NEC and its potential as a preventive approach [97]. On the other hand, we did not assess the immunological status of the intestinal milieu (in terms of immune cell subpopulations and proliferation) and how it can be affected by BM-MSCs.
Further experiments are needed to better characterize the pathways involved in the immunomodulatory activity of BM-MSCs and their mechanism of action in intestinal restitution in the context of NEC.

Conclusions
Currently, several strategies are used to prevent NEC, such as minimal enteral feeding, prebiotics, probiotics, postbiotics, and breastfeeding [7]. However, morbidity and mortality remain significantly high [98].
Our data showed that hBM-MSCs could reduce NEC incidence in a concentrationdependent manner. Thus, these findings are translationally relevant as they pave the way for new potential MSC-based preventive and therapeutic strategies to be used in NEC clinical management.  Table S1: Term formula milk, human, and mouse breast milk composition.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.