Human umbilical mesenchymal stem cells ameliorate atrophic gastritis in aging mice by participating in mitochondrial autophagy through Ndufs8 signaling.

Introduction
Chronic atrophic gastritis (CAG) refers to a chronic gastric condition characterized by repetitive damage to the gastric mucosal epithelium, resulting in a decrease in intrinsic glandular tissue, with or without fibrosis, intestinal metaplasia, and/or pseudopyloric metaplasia [1]. CAG is a crucial condition that leads to the development of gastric cancer and often evolves from mild superficial gastritis [2].
Serological studies conducted in different regions across the globe report a CAG incidence as high as 27% [3]. Age is a significant risk factor for CAG, and its severity correlates positively with age. In fact, serological studies have shown that the prevalence of CAG increases with age [4, 5].
Changes in mitochondrial function are associated with various degenerative diseases.As cellular mitochondria age, their inefficiency increases, also making them potentially toxic [6]. Aging can lead to increased permeability of the mitochondrial membrane, triggering apoptosis or necrosis [7]. Mitochondria also play a crucial role in inflammatory signal transduction [8]. Autophagy can eliminate dysfunctional or damaged mitochondria, consequently counteracting degeneration, suppressing inflammation, and preventing unnecessary apoptosis [9].Thus, alterations in mitophagy can lead to mitochondrial dysfunction, potentially resulting in a variety of aging-related diseases.
Mesenchymal stem cells (MSCs) are a type of adult stem cells derived from mesoderm. MSCs are used as therapeutic interventions for various inflammatory diseases and are gaining widespread attention. MSCs have shown beneficial effects in several gastric injury models [10]. The synergistic role of placental-derived MSCs can promote the regeneration of gastric mucosal cells in atrophic gastritis related to Helicobacter pylori, thus preventing gastric cancer [11].
NDUFS8, a nuclear-encoded subunit of the human mitochondrial complex I, is one of the core subunits essential for the electron transfer process catalyzed by complex I [12]. Complex I is also considered a primary source of reactive oxygen species (ROS), which are linked to cancer cell survival, proliferation, transformation, and the progression of malignant tumors [13]. However, the role of NDUFS8 in precancerous lesions has not been extensively investigated.
MSC administration in models of H. pylori-induced and high-salt diet-promoted gastric cancer induces autophagy, retains tumor-suppressive 15-hydroxyprostaglandin dehydrogenase, and attenuates apoptosis [14]. MSCs possess strong tissue repair and immune-regulatory capabilities. Considering a regenerative medicine perspective, MSCs-secreted growth factors with anti-apoptotic effects are crucial for tissue regeneration [15]. Growth factors secreted by MSCs can also reduce tissue fibrosis during regeneration [16]. MSCs can selectively migrate to inflammatory or damaged sites, where they survive and proliferate, and thus participate in inhibiting inflammatory responses and repairing damaged areas [17]. MSCs also demonstrate therapeutic benefits in various aspects, with one of the potential mechanisms being improved mitochondrial functions. MSCs can maintain cellular mitochondrial quality by promoting autophagy of the mitochondria. Therefore, we hypothesize that restoring mitochondrial function may offer a promising approach to improving CAG.
Based on these results, we investigated the potential role of human umbilical MSCs (hUMSCs) in ameliorating gastric mucosal epithelium damage by maintaining mitochondrial homeostasis. We explored the potential of Ndufs8 in alleviating mitochondrial damage in aging-induced gastric mucosal epithelial cells (GMECs) by activating mitochondrial autophagy and evaluated the molecular mechanisms of Ndufs8 in atrophic gastritis-associated gastric mucosal injury.

Materials and methods

Experimental design
All experiments were conducted in accordance with the ARRIVE guidelines 2.0. Mice were randomly allocated to cages through a-generated randomization process. Standardized procedures were employed for all measurements to ensure consistency and accuracy.

Cell culture
The hUMSCs from healthy newborns were procured from CellRapy Company (Nanjing,China). The cells were cultured to the 4th generation in MesenCult™ MSC Basal Medium (Stemcell, 05401) at 37 ℃ containing 5% CO2. Before transplantation, cells were subjected to quality control to ensure compliance with specifications related to MSC-specific surface marker expression, cell viability, and the absence of mycoplasma or bacterial contamination. MSC-specific surface markers were detected by flow cytometry, and the following monoclonal phycoerythrin-conjugated antibodies were used: CD45 (BD, 555483), CD34 (BD, 555822), HLA-DP/DQ/DR (BD, 562008), CD19 (BD, 555413), CD11b (BD, 555388), CD44(BD,553133),CD73(BD,561254),CD90(BD,554895),CD105 (BD, 555479). Cells were identified by flow cytometry ((Beckmen, NAVIOS), and FlowJo software was used to analyze the collected data. The differentiation abilities of hUMSCs for adipogenesis(Stemcell,05507), osteogenesis(Stemcell,05504), and chondrogenesis(Stemcell,05455) were, respectively, assessed through oil red staining, alizarin red staining, and alcian blue (AB) staining.

Establishment of the chronic atrophic gastritis mouse model
All animal experiments were approved after a rigorous review by the Institutional Review Board of the Medical School, Nanjing Medical University (IACUC-2008011). The study adhered to current animal research guidelines. C57BL/6 male mice (n = 200, 28-week-old) were purchased from Vital River (Beijing, China) (average weight 20 ± 3g), and were housed in specific-pathogen-free-grade animal facilities. Mice were maintained at a temperature of 25 ± 3 ℃, humidity of 50 ± 10%, and 12/12h light/dark cycle.
During the modeling process, mice were observed weekly for changes in external appearance and weight. When the mice were 20 months old (equivalent to 60 years in humans), enzyme-linked immunosorbent assay (ELISA) was performed to detect mouse serum gastrin-17 (G-17) and pepsinogen I (PG-I). Mice having decreased levels of G-17 and PG-I were considered to have age-related CAG (AR-CAG).

Treatment of chronic atrophic gastritis mice with human umbilical mesenchymal stem cells
Selected CAG mice (n = 50) were randomly assigned to the CAG group (n = 25) and the MSC injection group (n = 25). The effect of hUMSCs on gastric function in AG-CAG mice was investigated after injecting into the tail vein of AG-CAG mice on days 0, 15, and 30. The cells were suspended in 300 μL human serum albumin (HSA), with a dose of 1.0 × 107 cells/kg body weight. This dose of HSA injection was defined according to our previous studies and showed no impact on mice [18]. The mice in the control group were injected with 300 μL HSA. After 30 days of hUMSC injection,mice were humanely euthanized according to IACUC-approved guidelines using isoflurane sedation followed by cervical disarticulation.All mice were healthy until the end of the experiment.

Histopathology and immunofluorescence staining
At the end of the experiment, tissue samples were obtained from mice. Before sampling, the mice in each group had fasted for 24 h with no access to water and were given 3.5 mL/kg of 10% chloral hydrate injected intraperitoneally. To collect the sample, the entire stomach of the mouse was removed and dissected quickly along the greater curvature. The gastric mucosal tissues from the entire lesser curvature and closer to the greater curvature, extending from the esophagus to the duodenal end, were washed in physiological saline and collected and fixed in 10% neutral formalin solution. Routine paraffin embedding, sectioning, hematoxylin and eosin (HE) staining, and AB/periodic acid–Schiff (AB-PAS) staining of the gastric mucosa of mice in each group were performed to observe changes.
For immunofluorescence assays, slices were blocked for 45 min with 5% bovine serum albumin (BSA) at 37 °C, and then incubated overnight with primary antibodies CK-7(Abcam,ab 181598.) Ndufs8(NOVUS,NBP1-85620) at 4 °C. The procedure for fluorescent staining of cells followed the same procedure as tissue staining. The nuclei were stained with DAPI (Beyotime, C1006) and observed under a fluorescence microscope (Nikon, ECLIPSE Ti).

Enzyme-linked immunosorbent assay
Approximately 1 mL of blood was extracted from the retro-orbital venous plexus of each anesthetized mouse. After allowing it to stand at room temperature for 30 min, the blood was centrifuged at 3000 rpm for 15 min and the serum was obtained and stored at −80 °C for hormone analysis. The levels of G-17 (CUSABIO,CSB-E12924m), PG-I (LIANBOKEBIO,LBK-M00955), IL-6 (mlbio, ml098430), IL-8 (mlbio, ml063162)and TNF-α (mlbio, mlC50536-1) were determined following the instructions provided with the respective kits.

Transmission electron microscopy (TEM) results
Tissues were washed using PBS (pH 7.4) and fixed in glutaraldehyde (2.5%) for 24 h. The samples were fixed in 1% osmium tetroxide at room temperature for 1 h, gradually dehydrated in graded ethanol/acetone solutions, and embedded in epoxy resin. Then, ultra-thin Sects. (70 nm) were prepared, stained with 2% uranyl acetate and lead citrate, and visualized by TEM (JEOL, Tokyo, Japan) to capture the cellular ultrastructural images. The experiment was repeated thrice for biological replication (Table 1).

Quantitative Real-Time polymerase chain reaction (qPCR)
Total DNA was extracted using the TIANamp Genomic DNA Kit (TIANGEN Biotech). TBGreen™ pre-mixed ExTaq™II (TaKaRa) was utilized for detecting gene expression. The ®tm7Flex real-time PCR system was applied. The reaction system (total volume 20 μL) was prepared following the TB GreenTM pre-mixed ExTaq kit protocol, employing a two-step PCR amplification program. After the reaction, cycle threshold (Ct) values were obtained employing the @3159ABI Step One Plus sequence detection and software analysis system. Data were analyzed using the 2-ΔΔCT method. The GAPDH gene was used as the internal control and the samples were repeated at least three times. Results were expressed as mean ± multiplication factor, and p < 0.05 indicated statistically significant differences. For the assay, primers were synthesized by Realgene Company (Nanjing, China). Detailed information on the primers used herein is provided in Supplementary Table S1.

Terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) cell apoptosis in situ detection
Processed slices were rinsed twice with PBS, and to each sample, 50 μL TdT enzyme reaction solution was added. The cover glass was placed, and the sample was allowed to react in the dark at 37 °C for 60 min. The treated sample was rinsed thrice with PBS following which 50 μL-100 μL anti-fluorescein antibody working solution was added, and the cover glass was placed and allowed to react in the dark at 37 °C for 30 min. Then, 50 μL-100 μL DAB working solution was added to the sample and incubated at room temperature for 3 min. After washing with PBS, the sample was observed directly under a microscope and photographed.

Western blotting
Cells were lysed using IP lysis buffer, and a mixture of protease and phosphatase inhibitors was introduced and incubated on ice. Then, total protein was extracted and representative samples were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (8%−15%), and transferred to polyvinylidene difluoride membranes. The membrane was blocked with 5% BSA at room temperature, and incubated overnight with specific antibodies at 4 °C. After washing thrice with TBST, the membrane was incubated with respective secondary antibodies at room temperature for 2 h. Finally, the protein bands were detected by the chemiluminescence method and analyzed.

Protein extraction and liquid chromatography-mass spectrometry (LC–MS) analysis
In our study, protemics quantification was based on a 3 vs 3 experimental design, and each group of samples was treated with an appropriate amount of lysis buffer containing 8 M urea. After thorough sonication and subsequent 1 h incubation on ice, the samples were centrifuged at 4 °C and 12,000 g for 30 min. The supernatant was transferred to a fresh centrifuge tube, and protein concentration was detected using the Bradford kit.
An equal volume of protein solution was mixed with dithiothreitol to achieve a final concentration of 5 mM and incubated at 56 °C for 25 min. To this, iodoacetamide was then added to obtain a final concentration of 14 mM, followed by incubation in the dark at room temperature for 30 min. The final urea concentration of the samples was diluted to less than 2 M using 25 mM Tris–HCl. Then, trypsin and Lysis C were added at a mass ratio of 1:100 (enzyme: protein), and the samples were enzymatically digested at 37 °C for 16–18 h. The resulting peptide fragments from trypsin digestion were vacuum freeze-dried after desalting.
The obtained peptide powder was redissolved in formic acid (0.1% v/v) and subjected to elution analysis employing the EASY-nLC 1200 ultra-high-performance liquid chromatography system.
The eluted peptides were ionized and loaded onto the Thermo Scientific Orbitrap Fusion™ Lumos™ Tribrid™ mass spectrometer for analysis. The voltage for the ion source was set at 2.2 kV, and the precursor ions as well as their secondary fragments were detected and analyzed employing the high-resolution Orbitrap. The first MS scan range was set at 350–1500 m/z, with a scan resolution of 60,000, while the second scan resolution was set at 15,000. Data were collected using data-dependent acquisition, with automatic gain control (AGC) set to 8E4 and a dynamic exclusion time for tandem MS scans set at 30 s.

Database search
The secondary MS data were analyzed using Proteome Discoverer (v2.4). The search parameters used the uniprot_mouse_proteome_UP000000589 database (2021.04), trypsin/P as the enzyme cleavage method, allowing for two missed cleavage sites, and a minimum peptide length of six amino acid residues. For fixed modification, cysteine alkylation was set while methionine oxidation and protein N-terminal acetylation were deemed variable modifications.

Isolation of primary gastric mucosal epithelial cells and co-culture with human umbilical mesenchymal stem cells
Mice were euthanized, and gastric mucosal tissues were isolated using established methods [19] and allowed to grow in a 5% CO2 incubator at 37 °C. To establish an in vitro treatment model, a Transwell chamber co-culture system (Corning, 3412) with six-well plates was used with a 0.4 μm membrane. GMECs (1.0 × 105 cells) were seeded on the upper surface of the membrane, while 3.0 × 105 of hUMSCs from the 4th to 6th passage were seeded in the base of the wells. After 48 h of co-culturing, cells were collected for further analysis.

Lentivirus infection of cells
Cells were seeded in a 24-well culture plate, and viral infection was initiated upon reaching approximately 70% confluency. Lentivirus dilution was prepared according to the multiplicity of infection (MOI), and an appropriate concentration of polybrene was added. The culture medium in the original culture dish was replaced with the prepared virus dilution and mixed. Then, the plate was incubated in a CO2 incubator (37 °C, 5% CO2). After 24 h, the lentivirus-containing culture medium was replaced with a normal culture medium. Infection efficiency was detected by qPCR.

Reactive oxygen species (ROS) detection
The cells were collected and the probe was loaded. Diacetyldichlorofluorescein (DCFH-DA)(Beyotime,S0033) was diluted with serum-free culture medium in a ratio of 1:1000, to obtain a final concentration of 10 mmol/L. The cells were collected and suspended in the diluted DCFH-DA, with a cell concentration of 1 × 106–2 × 107 cells/mL. Next, the cells were incubated at 37 ºC in a cell culture incubator for 20 min and then washed thrice with serum-free culture medium. The cells were directly stimulated with either the ROS positive control or the desired drug, or the cells were stimulated separately after dividing into several portions. Usually, the ROS level can increase significantly 20–30 min after cell stimulation.

MitoTracker staining
Mitochondria were labeled by MitoTracker® Mitochondrion-Selective Probes(invitrogen,CM-H2DCFDA). The MitoTracker® stock solution (1 mM) was diluted to a final working concentration of 200 nM in the growth medium. Upon achieving the desired cell confluency, the culture medium was removed, and preheated staining solution with MitoTracker® probes was added and incubated at 37 °C for 30 min. The staining solution was then replaced with fresh preheated medium, and the cells were visualized under a fluorescence microscope or fluorescence microplate.

mtSOX deep red—mitochondrial superoxide detection
The cells were seeded (1 × 105 cells/mL per well) in a 96-well black plate or 6-well plate and cultured overnight at 37 °C with 5% CO2. The culture medium was removed, and the cells were washed twice with Hank's buffered saline solution. After removing the supernatant, mtSOX Deep Red (Invitrogen, M36008) working solution containing 10 μmol/L antimycin was added to the cells and cultured at 37 °C with 5% CO2 for 30 min. The fluorescence intensity was detected by flow cytometry.

Enhanced mitochondrial membrane potential detection kit (JC-1)
About 5 × 105 cells were taken and JC-1 staining working solution was added, mixed by inverting several times, followed by incubation at 37 ºC for 20 min in a cell culture incubator. Next, the cells were centrifuged at 600 g for 3 min at 4 ºC, and the precipitated cells were collected and washed twice with JC-1 staining buffer(Beyotime,C2003S). The cells were resuspended in an appropriate amount of JC-1 staining buffer and observed under a fluorescence microscope or analyzed using a fluorescence spectrophotometer or by flow cytometry.

Annexin V-fluorescein 5-isothiocyanate (FITC)/ propidium iodide (PI) apoptosis detection
The apoptosis level of cells was determined using the Annexin V-FITC/PI Apoptosis Detection Kit (Vazyme,A211-01). The cells were digested with trypsin with no EDTA, centrifuged at 300 g and 4 °C for 5 min, and then 5 × 105 cells were collected. The cells were washed twice with pre-cooled PBS, each time followed by centrifugation at 300 g, 4 °C for 5 min. After adding 100 μL 1 × binding buffer, the cells were blown gently with a pipette to form a single-cell suspension. Then, to the cell suspension, 5 μL Annexin V-FITC and 5 μL PI Staining Solution were added, with gentle blowing. After incubating in the dark at room temperature for 10 min, 400 μL 1 × binding buffer was added to the system and gently mixed. The untreated cell samples were used as negative controls. After staining, the samples were detected by flow cytometry within 1 h.

Statistical analysis
Data were analyzed using GraphPad Prism software (version 8.0). All values were presented as the mean ± standard error of the mean. Unpaired Student's t-test was used to compare two groups, while one-way analysis of variance followed by Bonferroni test was used for multiple comparisons. A P-value < 0.05 was deemed statistically significant.

Results

Gastric mucosal epithelial cell injury in age-related chronic atrophic gastritis mice
The age-related damage to the gastric mucosa in mice was investigated by selecting naturally aging mice with CAG as the AG-CAG mouse model. As the animals aged, AG-CAG mice exhibited sparse and withered fur, reduced activity, poor mental status and food intake, and weight loss.In the non-CAG group mice, the gastric tissues appear normal, with uniform thickness, a transparent and smooth surface, and a pale yellow-white color. In contrast, the gastric walls of AG-CAG group mice are thicker and commonly exhibit white irregular nodules on the surface (Fig. 1A).
HE staining revealed that compared to the control group, mice in the CAG model group showed an irregular arrangement of glandular epithelial cells, reduced gland thickness, enhanced mucosal muscle layer thickness, and glandular atrophy, while a more regular arrangement of glandular cells with less obvious atrophy was noted in the young control group (Fig. 1D).
Intestinal epithelial metaplasia may occur in cases of CAG. Through AB-PAS tissue staining, we found that the normal gastric mucosa epithelium secreted neutral mucus, that appeared red after AB-PAS staining, while cells with intestinal metaplasia secreted mucus containing sulfuric acid and sialic acid, respectively, staining brown-black and blue (Fig. 1E).
Examination of control mice and aged CAG mice by electron microscopy revealed mitochondrial structural changes in gastric tissue. In the control group, the structure of the mitochondria was acceptable with intact membrane, uniform matrix, and parallelly arranged cristae. In the AG-CAG group, cells were elliptical and showed significant damage. In this group, most of the mitochondria were noticeably swollen and enlarged; the matrix was sparse and dissolved, the cristae were reduced and disappeared; some severely injured cells showed damaged membrane, with the occasional presence of autolysosomes (Fig. 1F).
Mouse serum G-17 and PG-I were measured by ELISA. The results indicated a decrease in G-17 (Fig. 1B) and PG-I (Fig. 1C) levels in the CAG model group. Additionally, inflammatory cytokines such as INF-γ (Fig. 1G), IL-6(Fig. 1I) and IL-8, (Fig. 1H) were expressed more in the AG-CAG animal model, with their levels being significantly correlated with gastric function and mucosal epithelial atrophy.

Mesenchymal stem cell transplantation can reduce gastric mucosal damage, positively alter mouse gastric secretion levels, and reduce apoptosis in mice
MSCs derived from umbilical cord tissue were evaluated for their characteristics and identified according to the International Society for Cellular Therapy. Flow cytometry showed high expression of hUMSCs markers; CD105 (97.81%), CD90 (98.33%), CD44 (99.2%), and CD73 (97.72%) were positive, and CD11b (2.29%), CD19 (2.55%), CD34 (2.98%), CD45 (3.86%), and HLA-DR (4.46%) were negative (Supplement Fig. 1A). Assessment of in vitro germ layer differentiation revealed that these MSCs could differentiate into adipocytes, osteoblasts, and chondrocytes (Supplement Fig. 1B).The tests for bacteria, mycoplasma, and endotoxin yielded negative results (Supplement Fig. 1C).
AR-CAG mice were categorized into treatment and solvent injection groups, and were injected with preparations as described earlier (Fig. 2A). The expression of the human-specific gene STEM121 in gastric tissue was confirmed through immunofluorescence, proving the grafting of hUMSCs onto the gastric mucosal epithelial tissue (Fig. 2B).
Changes in gastric function in CAG mice after hUMSC transplantation were assessed by ELISA. Following MSC injection in the CAG model group, G-17 (Fig. 2D) and PG-I (Fig. 2C) levels increased, but no improvement was noted in the solvent control group. MSC injection improved the levels of inflammatory cytokines, such as INF-γ (Fig. 2G),IL-6 (Fig. 2E) and IL-8, (Fig. 2F) in AG-CAG mice.
HE staining revealed that after hUMSC injection, the arrangement of glandular epithelial cells with inflammatory lesions tended to be more regular in the CAG group, with reduced glandular atrophy (Fig. 2H). PAS staining revealed intestinal metaplasia in the gastric mucosal epithelium, which was alleviated by administering hUMSCs(Fig. 2I).
TUNEL analysis indicated a significant decrease in the number of stained cells in the hUMSC group on day 30 after injection, suggesting a decrease in gastric tissue apoptosis (Fig. 2K). Consistent with TUNEL results, western blot showed reduced expression of apoptosis-related proteins Caspase 3,Caspase 9 and BAX and enhanced expression of Bcl-2 in gastric tissue, and the differences were statistically significant (Fig. 2J).

Proteomic signatures of hUMSC injected AG-CAG mice by proteomic and bioinformatic analysis
Proteomic analysis was performed to comprehensively understand the mechanism of MSCs-mediated regulation of gastric function in AG-CAG mice. Based on non-labeled quantitative proteomics, 2605 proteins were identified, and 126 differentially expressed proteins were filtered through Student’s-t test (p < 0.05, fold-change > 1.5). The volcano plot showed a higher fold-change of heregulin (Fig. 3A, B). Actually, we did performed FDR calculation for the raw p values (supplementary table). However, none of proteins meet the requirements of FDR < 0.05.The minimal FDR value is 0.59. In this case, it is more resonable to consider false negative more. Thus, we only use the combination of raw p-value and fold change strategy to screen differentially expressed proteins.
The differentially expressed proteins were subjected to overall functional enrichment analysis. According to the Gene Ontology (GO) biological process annotation, these differentially expressed proteins were mainly enriched in events such as apoptosis, oxidative stress, energy metabolism, and protein translation. Considering cellular localization (cellular component), these proteins were mainly enriched in subcellular locations, such as the mitochondrial respiratory chain, ribosomes, and cell junctions. Correspondingly, in terms of basic molecular functions (molecular function), they were enriched in respiratory chain enzyme-related activity, oxidation–reduction, and antioxidant activities (Fig. 3C). Based on KEGG pathway analysis, these proteins are primarily involved in the tricarboxylic acid cycle, oxidative phosphorylation, and the mTOR signaling pathway (Fig. 3D).
Preliminary insights garnered through functional enrichment suggest that the differentially expressed proteins are predominantly involved in mitochondrial-related functional regulation, encompassing 27 proteins. Analysis from the STING database has elucidated the inter-regulatory relationships among these mitochondrial proteins. A core network of mutual regulation is formed by the proteins Atp5c1, Atp5b, Ndufv2, Ndufs8, Ndufb7, Ndufs6, Ndufv3, Hsd17b10, Sod1, Prdx5, Tsfm, Clybl, Hnrnpk, and Mt3 (Fig. 3E).
The quantitative results of proteomics were validated by RT-PCR analysis of five randomly selected proteins. Ndufs8 were low expressed in the gastric tissue of aged CAG mice, however, their expression increased after hUMSC injection (Fig. 3F), consistent with the reported low expression of Ndufs8 in gastric cancer tissue. The results of western blot analysis were consistent with the quantitative results of proteomics, indicating the reliability of proteomic quantitative analysis.
Concurrently, immunofluorescence of mouse gastric tissue sections revealed that Ndufs8 is primarily localized on the surface of the gastric mucosa (Fig. 3G). The results of Western Blot analysis on proteins extracted from gastric tissues before and after injection of hUMSCs are intriguing. Treatment with hUMSCs appears to elevate the expression of Ndufs8, while significantly reducing the levels of the key mitochondrial autophagy proteins PINK1 and Parkin (Fig. 3H). This suggests that hUMSCs may enhance mitochondrial function in AG-CAG mice through Ndufs8, thereby influencing mitochondrial autophagy.

hUMSCs ameliorate mitochondrial dysfunction and alleviates gastric mucosal epithelial cell damage
To investigate the mechanisms underlying hUMSCs alleviates senescent atrophic gastritis, we isolated primary gastric mucosal epithelial cells (GMECs) from normal mice and employed lentivirus to infect the GMECs, thereby eliminating Ndufs8 expression within the cells. The efficacy of the lentiviral infection, marked by fluorescent labeling, was assessed through fluorescence staining (Fig. 4A). Electron microscopy results of GMECs following the knockout of Ndufs8 reveal that the majority of mitochondria (M) are markedly swollen and enlarged, with a sparse and dissolving matrix, reduced or absent cristae, and in some severely damaged cases, membrane rupture and disintegration. Autophagic lysosomes (ASS) are sporadically present (Fig. 4B).Subsequently, we co-cultured these cells with hUMSCs for 48 h to study the potential effects of hUMSCs on them.
Initially, it was observed that the transcription levels of mitochondrial genes COXIV, Pgc1α, Tfam, and Nrf1 were diminished in the GMECs with a knockout of Ndufs8. However, following co-culturing with hUMSCs, the transcription levels of these genes exhibited an increase (Fig. 4C).
Functionally, using the fluorescent probe DCFH-DA to detect reactive oxygen species (ROS), hUMSCs were observed to reduce the production of ROS in GMECs treated with Ndufs8 siRNA (Fig. 4E, F). Similarly, ATP production was diminished in the Ndufs8 siRNA group; however, co-culturing with MSCs significantly enhanced ATP output (Fig. 4D).
Using MitoTracker® mitochondrial-selective probes to label mitochondria it was evident that the mitochondrial quantity was significantly reduced in Ndufs8 knockout GMECs. However, following co-culture with hUMSCs, the mitochondrial count increased in the GMECs (Fig. 4G, H).
Results from SA-β-gal staining also demonstrate that hUMSCs can mitigate the senescence induced in GMECs by the knockout of Ndufs8 (Fig. 4I, J).
Changes in mitochondrial quantity and quality of GMECs accompanied the overall elimination of ROS. We speculate that hUMSCs may stimulate mitochondrial autophagy in GMECs.

Human umbilical mesenchymal stem cells activate cellular mitochondrial autophagy through the Ndufs8 pathway
To delve deeper into whether hUMSCs modulate changes in mitochondrial autophagy in gastric mucosal epithelial cells through the Ndufs8 pathway, we conducted the following experiment. Consistent with in vivo studies, primary GMECs with Ndufs8 deficiency co-cultured with hUMSCs exhibited reduced mitochondrial ROS levels and enhanced mitochondrial autophagy, while the control group showed no significant changes.
Determination of mitochondrial ROS levels using MitoSOX Red indicated reduced generation of mitochondrial ROS in GMECs upon administration of hUMSCs (Fig. 5A, B).
Next, JC-1 staining was employed to determine mitochondrial membrane potential. Compared to the control group, treatment with hUMSCs led to a decrease in damaged low-potential mitochondria. Further, the ratio of monomers to aggregates of JC-1 was quantitatively assessed by flow cytometry (Fig. 5C, D).
Cell proliferation and apoptosis are correlated with gastric mucosal epithelial damage and the progression of gastric diseases. We examined cell proliferation and cycle dynamics. In GMECs with Ndufs8 deficiency, we observed a reduced number of Ki67-positive cells, indicating lower levels of GMEC proliferation. Co-culturing Ndufs8-deficient GMECs with hUMSCs resulted in an increase in the percentage of ki67-positive cells (Fig. 5G, H).
Utilizing flow cytometry to assess the cell cycle, we noted that Ndufs8-deficient GMECs exhibited a notably shortened S phase and an extended G2/M phase. However, when co-cultured with hUMSCs, the Ndufs8-deficient GMECs displayed a lengthened S phase and a shortened G2/M phase (Fig. 5E, F).
In summary, our research demonstrates that hUMSCs maintain mitochondrial quality and promote mitochondrial autophagy, thereby ameliorating senescent chronic atrophic gastritis (CAG). This effect is likely mediated through the Ndufs8 pathway.

Discussion
Injury to gastric mucosa, inflammation, and apoptosis can be caused due to various reasons that can induce atrophic gastritis [20]. The incidence and detection rates of CAG increase with age, rendering it a serious threat to the health of the elderly [21]. With age, there is a decrease in the repair and regeneration functions of the gastric mucosa and inflammation becomes chronic, leading to abnormal epithelial proliferation, gastric gland atrophy and cancer [22]. However, modern medicine lacks effective treatments, and there is currently no recognized medication for treating CAG.
MSCs, as the most effective therapy for autoimmune diseases, have gained increasing attention in the fields of tissue damage and regenerative medicine attributed to their biological characteristics, including low immunogenicity, chemotaxis toward inflammatory factors, immunosuppression, the potential of multi-lineage differentiation, and nutritional support [23]. MSCs are found in high amounts of umbilical tissue, connective tissue, umbilical cord, placenta, and amniotic membrane [24]. Among various sources of MSCs studied until now, hUMSCs are the most convenient to procure, with no ethical controversies or harm to donors during collection [25]. Compared to other MSCs from adults, MSCs obtained from the human umbilical cord are intermediate between embryonic stem cells and adult stem cells, being more primitive [26]. Their differentiation and proliferation capacities are lower compared to those of embryonic stem cells but significantly higher compared to those of adult stem cells, making them easily accessible and amenable to large scale applications [27]. Numerous studies have already reported the therapeutic application of hUMSCs in gastritis [28], Crohn’s disease [29], and chronic Helicobacter pylori infection [14].
Our research team established a CAG model in aged mice and observed increased GMECs apoptosis, activated inflammatory response, and reduced mitochondrial autophagy in these mice. Intravenous injection of hUMSCs in elderly CAG mice led to significant changes in the pathological features, including reduced intestinal metaplasia and a increased levels of serum G-17 and PG-I. In this study, to explore the mechanisms by which human umbilical cord-derived mesenchymal stem cells (hUMSCs) ameliorate senescent chronic atrophic gastritis (CAG) in aged mice, we employed high-throughput proteomics to analyze gastric mucosal tissues before and after hUMSC treatment. We successfully identified 2,605 proteins, and through Student's t-test, we isolated 126 differentially expressed proteins (p < 0.05, fold change > 1.5).Among these, a key protein identified was NADH dehydrogenase (ubiquinone) Fe-S protein 8 (Ndufs8), which is a nuclear-encoded accessory subunit of mitochondrial complex I (CI).Complex I has also been recognized as one of the main sources of ROS, which is linked to cancer cell survival, proliferation, transformation, and malignancy progression [30].
The results of the KEGG enrichment analysis indicate that Splicing, the TCA cycle, oxidative phosphorylation are the enriched biological pathways. It is well-known that RNA splicing dysregulation is a molecular characteristic of nearly all tumor types. Cancer-related splicing dysregulation can promote tumorigenesis through various mechanisms, leading to increased cell proliferation, reduced apoptosis, enhanced migration and metastatic potential, resistance to chemotherapy, and evasion of immune surveillance. The relationship between RNA splicing mutations and gastric cancer has been confirmed by numerous studies [31–33]. Therefore, we hypothesize that the RNA splicing pathway may also be altered in atrophic gastritis, a precancerous lesion of gastric cancer.
Similarly, the tricarboxylic acid (TCA) cycle is a ubiquitous metabolic pathway in aerobic organisms and is located in the mitochondria of eukaryotes. Metabolism and aging are closely linked [34]. In preclinical models of type 2 diabetes (T2D) and liver disease, high levels of ROS production, low ATP levels, and abnormal mitochondrial morphology have been observed [35]. In the context of human metabolic disorders, alterations in mitochondrial cycling are receiving increasing attention. However, the relationship between mitophagy defects and metabolic diseases is quite complex, potentially involving bidirectional influences. On one hand, mitophagy defects may lead to metabolic abnormalities due to reduced energy supply and excessive oxidative stress. On the other hand, metabolic diseases may cause energy shortages, thereby reducing the homeostatic activity of mitophagy.
Under stressors such as cellular aging, ROS, and nutrient deficiency, mitochondria in cells undergo depolarization damage [36].Cells selectively encapsulate and degrade damaged or dysfunctional mitochondria through autophagy to maintain the stability of the mitochondrial network and the cellular environment [37]. With aging, cells accumulate errors in their nuclear and mitochondrial genomes, leading to damaged organelles and large molecules [38]. According to recent evidence, the interaction between mitochondria and autophagy links aging to health or disease. Studies have also demonstrated the association between mitochondrial autophagy and gastric diseases [39, 40].
Existing research confirms that NDUFS8, in addition to playing a crucial role in neurological disorders [41], also participates in the regulation of various cancers. In non-small cell lung cancer (NSCLC), high expression of NDUFS8 and low expression of NDUFS1 are associated with poor prognosis in patients, suggesting that their expression levels might predict lung cancer outcomes [42]. Elevated levels of NDUFS8 have also been observed in hepatocellular carcinoma (HCC) [43]. Joint studies on lung adenocarcinoma and liver cancer have shown that overexpression of NDUFS8 correlates positively with the long non-coding RNA PPP1R14B-AS1, promoting tumor cell proliferation and migration by enhancing mitochondrial respiration [44]. Recently, a study collected 12 pairs of invasive gastric cancer (IGC) and adjacent normal tissues, employing high-performance liquid chromatography tandem mass spectrometry for proteomic analysis. It revealed proteomic characteristics of invasive gastric cancer, notably a significant downregulation of Ndufs8 protein expression [45]. Consequently, we hypothesize that, as a precancerous manifestation of gastric cancer, Ndufs8 plays a crucial role in the onset and progression of chronic atrophic gastritis (CAG).
In summary, our study represents the inaugural demonstration, both bioinformatically and experimentally, of the role of human umbilical cord-derived mesenchymal stem cells (hUMSCs) in age-related atrophic gastritis. To provide more robust evidence, we are currently developing Ndufs8 knockout mice, which will be used to further validate and explore the effects of hUMSCs on atrophic gastritis.
While our research is not the first to identify the role of mitochondrial autophagy in chronic atrophic gastritis (CAG), to our knowledge, it is the first to use a specific animal model to study age-related atrophic gastritis and to establish the significant role of Ndufs8 in regulating CAG. This will provide new evidence supporting the therapeutic potential of hUMSCs in treating aged CAG.

Conclusion
In summary, we have confirmed that human umbilical cord-derived mesenchymal stem cells (hUMSCs) can colonize the gastric mucosa of age-related chronic atrophic gastritis (AR-CAG) mice and alleviate mucosal damage by enhancing mitochondrial function and promoting mitochondrial autophagy. Moreover, these findings underscore the critical role of the regulatory factor Ndufs8 in age-related atrophic gastritis. These insights could pave the way for the advancement of mesenchymal stem cell therapies in the treatment of elderly patients with atrophic gastritis.

Supplementary Information