1. Introduction
Aging is a universal and inherent biological phenomenon that transpires throughout the course of development, in which immunosenescence can impact the immune regulatory network, leading to compromised immune functionality and diminished resistance against diseases [
1], [
2], [
3]. In mammals, immunosenescence manifests as a state of chronic inflammation, reduced immune reactions toward internal and external antigens, impaired responsiveness to novel antigens, and sluggish activation of preexisting immune memory, culminating in a diminished capacity to combat infectious diseases, hinder tumor growth, and eliminate senescent cells [
2], [
4], [
5]. The decline in immune function observed in aging individuals is associated with heightened vulnerability to infections and the development of more severe diseases. Notably, the elderly population tends to experience greater incidences of respiratory, gastrointestinal, and mucocutaneous infections, underscoring the impairments in mucosal and cutaneous immunity that occur during the aging process [
6].
As the process of aging advances, the bone marrow undergoes a range of transformations, such as modifications in its supporting structure, accumulation of adipocytes, and alterations in the biomolecular microenvironment [
7], [
8], [
9], [
10], [
11]. Within the bone marrow, there is an accumulation of senescent cells that exhibit senescent secretory phenotypes, resulting in the secretion of active components that impact the microenvironment [
9], [
12]. Hematopoietic stem cells (HSCs), which are positioned at the apex of the hematopoietic hierarchy, undergo a notable decline in their ability to self-renew and undergo full-range lineage differentiation [
13]. It has been reported that aged HSCs exhibit an augmented capacity for differentiating into myeloid cells, whereas their potential to differentiate into the lymphoid lineage decreases [
8]. Myeloid skewing reduces the production of B and T cells and increases the contents of myeloid cells, particularly those with immunosuppressive properties. Consequently, the elderly population may experience immune dysfunction and weakened defense against infectious diseases. Thus, exploring significant alterations in the bone marrow microenvironment and immune cell replenishment may aid in comprehending the inefficiency of the pathogen response of the aging immune system.
Clostridium difficile (
C. difficile) is a prevalent etiological agent of acute bacterial diarrhea among individuals aged higher or equal to 65 years, as well as a frequently encountered nosocomial infection [
14], [
15]. The mortality rate associated with
C. difficile infection escalates with advancing age, with a notable increase from 5% in those aged 61-70 years to more than 10% in those aged more than 80 years [
16]. In the United States,
C. difficile was responsible for an estimated 500 000 infections and 29 000 fatalities in 2012 [
17]. Non-invasive
C. difficile strains cause varying degrees of diarrhea and colonitis, worsening mortality rates in the elderly population; during infection, the secreted enterotoxin A (TcdA), cytotoxin B (TcdB), and binary toxin mediate epithelial destruction and proinflammatory functions [
18]. In the context of the host, innate immunity plays an important role in determining the resistance of aged individuals to
C. difficile. A previous study demonstrated that aging hinders the activation of various innate cytokines, resulting in inadequate recruitment of myeloid cells to the large intestine [
19]. This leads to increased bacterial load and toxin production in extraintestinal tissues, ultimately causing mortality. Nevertheless, the specific alterations of myeloid differentiation in the bone marrow that contribute to immune deficiency in mucosal defense remain unclear.
In this study, we investigated the immune cell profiles of bone marrow in young and aged mice using mass cytometry (CyTOF) technology. Single-cell transcriptome data from a public database were reanalyzed to further verify the results from CyTOF. Transcriptomes and immune gene expression in the bone marrow of young and aged mice were measured to confirm the effects of aging on fluctuations in immune pathways. This study lays a theoretical foundation for future immune intervention to prevent and cure C. difficile infection in the elderly.
2. Materials and methods
2.1. Animals and treatments
This protocol was approved by the Animal Ethics and Animal Welfare Committee of China Agricultural University (AW12902202-5-7). Male specific-pathogen-free mice aged 1 (1M), 3 (3M), 6 (6M), 8 (8M), 18 (18M), 20 (20M), 24 (24M), and 28 (28M) months were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. (China) and housed at (24 ± 2) °C with a 12 h dark/light cycle. The relative humidity was 50%-60%, and food and water were provided ad libitum. In this manner, the mice were raised naturally until old age.
In the infection trial, the 24M male mice received the vehicle or 200 μL of
C. difficile VPI 10463 with approximately 1.0 × 10
9 colony-forming units (CFU)·mL
−1 by gavage after treatment with the antibiotic cocktail in drinking water for three days, followed by one day of recovery [
20]. The toxigenic
C. difficile VPI 10463 strain was kindly donated by Professor Yibing Peng from Ruijin Hospital of Shanghai Jiao Tong University.
C. difficile was cultured in a brain heart infusion medium, and the concentration was determined via gradient dilution and culture plate counting. The antibiotic cocktail in drinking water included 0.4 g·L
−1 kanamycin, 0.215 g·L
−1 metronidazole, 0.057 g·L
−1 colistin, 0.045 g·L
−1 vancomycin, and 0.035 g·L
−1 gentamycin. After seven days, the mice were collected to analyze the difference in their susceptibility to
C. difficile. Polymorphonuclear (PMN)-myeloid-derived suppressor cells (MDSCs) were cleared with inVivoMab anti-Gr-1 (BE0075, BioXcell, USA) by intraperitoneal injection. After harvesting, blood was collected in anticoagulant tubes by tearing out the eyeballs and centrifuging at 4 °C and 300
g (
g is gravitational acceleration) for 15 min to obtain plasma. The liver and spleen indices for each mouse were calculated by the organ weight/body weight × 100%. All intestinal tissue samples and chyme stored for determination were kept at −80 °C until use.
2.2. CyTOF for bone marrow cells
The use of the CyTOF method to determine the profiles of immune cells and stem cells in bone marrow has been described in previously published articles [
21], [
22]. After cutting the epiphysis open, the tibia and femur were centrifuged using modified centrifugal tubes to collect bone marrow cells. These were resuspended in phosphate-buffered saline (PBS) to obtain single-cell suspensions, which were then filtered with 70 μm filter membranes. Cells were stained with 194Pt at a final concentration of 0.25 µmol·L
−1 on ice for 5 min to identify the live cells, before blocking with 50 µL of an Fc receptor blocker mix on ice for 20 min. Subsequently, a cocktail containing 41 metal-tagged surface antibodies was used to stain the cells for 30 min on ice. After incubation with Fix and Perm buffers (Thermo Fisher Scientific, USA) at room temperature for 1 h, 191/193Ir was used to stain the DNA overnight at 4 °C, followed by incubation with the intracellular metal-tagged antibody cocktail for 30 min at room temperature after washing with fluorescence-activated cell sorting (FACS) buffer (BD Bioscience, USA). After another wash, the cells were analyzed using a PLT-MC601 instrument (Zhejiang PuLuoTing Health Technology Co., Ltd., China). Data were normalized and manually gated using FlowJo software (BD Bioscience), and the beads, carrier cells, and dead cells circling the single live cells were removed. Then the CD45
+ immune cells were subjected to subgroup clustering, annotation, t-distributed stochastic neighbor embedding (t‐SNE) dimensionality reduction, and data analysis by means of the
x‐shift algorithm.
2.3. Flow cytometry
Isolated bone marrow cells were lysed with red blood cell lysis buffer to eliminate mature erythrocytes. After washing and resuspending with the staining buffer, 100 μL of suspension for each sample aspired into a new dark tube. Three antibodies, including fluorescein isothiocyanate-labeled CD11b antibody (127608, anti-mouse, Biolegend, USA), Alexa Fluor 647-labeled Ly6C antibody (127612, anti-mouse, Biolegend), and phycoerythrin-labeled Ly-6G antibody (127608, anti-mouse, Biolegend), were separately added into dark tubes, according to the instructions from the manufacturer, to stain the cells for 1 h at 4 °C. After washing and resuspending with the staining buffer, the abundances of CD11b+ cells, and PMN-MDSCs (CD11b+, Ly6Clow/neg, and Ly6G+) were detected using the FACSAria III system (BD Difco, Becton, Dickinson and Company, USA). Following similar procedures, the cells in the colonic mucosa were also stained and checked. However, the difference was that the colon tissues were digested with 5 mmol·L−1 ethylenediaminetetraacetic acid (EDTA) to dissociate the epithelia by shaking in Roswell Park Memorial Institute (RPMI) 1640 medium supplemented with 10% fetal bovine serum (FBS) at 37 °C, before acquiring the cells in the mucosa through violent shaking in 10% FBS RPMI 1640 medium containing 300 IU·mL−1 collagenase. In contrast, the staining procedure for the red blood cells was simpler. In brief, 100 μL of blood with EDTA anticoagulant was first stained with three antibodies for 1 h before lysing the erythrocytes in the blood. After collecting the primary results, all data were normalized and manually gated with FlowJo 10.
2.4. Single-cell transcriptome analysis of bone marrow
Single-cell transcriptome data were obtained from the Gene Expression Omnibus with the ID of GSE169608, which contained the data of bone marrow from mice at one month (GSM5210632, 1M), six months (GSM5210633, 6M), and 20 months (GSM5210634, 20M) of age. Single-cell clustering and annotation were performed using the R package “Seurat” (version 4.0). Genes with at least one feature count in more than three cells were kept for further analysis. To remove the cells with low-quality data, cells with more than 5000 or fewer than 500 genes were filtered, in combination with cells with a mitochondria ratio greater than 0.15. Data were normalized using the function “NormalizeData” with the scale.factor = 10 000, and the ten most highly variable genes were identified using the function “FindVariableFeatures” with the nfeatures of 2000. Then, principal component analysis was performed after data integration and scaling, followed by clustering analysis using the “FindNeighbors” and “FindClusters” functions with the top 25 dimensions and a resolution of 0.5. Markers for each cluster were identified using the “FindAllMarkers” function with a value of min.pct of 0.25 and a logfc.threshold of 0.25. Cell types were identified according to the expression of the classical marker genes of each cluster by comparison with the CellMarker 2.0 database. Dimensionality reduction was performed by running the “RunUMAP” function, and the results were visualized using the “DimPlot” function. The uniform manifold approximation and projection (UMAP) plots of genes of interest, such as Cd11b (Itgam), Ly6C2, and Ly6g, were marked with the R packages “ggplot” and “dplyr.” The percentage of cells of interest, including MDSC, PMN-MDSC, PMN-MDSC(C05), and PMN-MDSC(C07), was calculated. We then re-clustered neutrophils in the three stages after merging the three samples and performed gene set enrichment analysis (GSEA) between the 6M and 20M groups.
2.5. PMN-MDSC sorting and co-culture with CD8+ T cells
For the co-culture experiments, 24-well plates were precoated with anti-mouse CD3e IgG1 antibody (60015, STEMCELL, Canada) at a dose of 0.75 microgram per well. CD8+ T cells in the spleen were screened with commercial mouse CD8a positive selection kits (18753, STEMCELL) using magnetic particles, according to the protocol. The screened CD8+ T cells were cultivated in 24-well plates in RPMI1640 with 10% FBS supplemented with 10 mmol·L−1 2-hydroxyethyl buffer, 2 mmol·L−1 L-glutamine, 1 mmol·L−1 sodium pyruvate, 100 μmol·L−1 minimum essential medium non-essential amino acid, and 50 μmol·L−1 β-mercaptoethanol. 30-100 IU·mL−1 of mouse IL-2 (78081.1, STEMCELL) and 0.5 μg·mL−1 anti-mouse CD28 antibody (102102, Biolegend, USA) were added to the medium to activate the cells. CD8+ T cells were seeded in transwell chambers (3470, 0.4 μm, Corning) and co-cultured with PMN-MDSCs in the upper chambers. Next, we circled and sorted out the target cells in line with the flow cytometry method for PMN-MDSCs. After culturing the CD8+ T cells for 24 h, PMN-MDSCs isolated from the bone marrow or the blood of 24M mice were cultivated in the upper chambers with a medium similar to that used for the CD8+ T cells but without the IL-2 and CD28 antibodies. After 48 h of co-culture, the supernatant of the CD8+ T cells was collected for determination.
2.6. Cell viability
The viability of the primary cells in the bone marrow from 3M and 28M mice was determined using 0.4% (m:v; g·mL−1) trypan blue dye. After one night at 4 °C, 0.4% trypan blue solution was added to the cell suspension, which was stained for 2 min. Under the microscope, the dead cells were pale blue, enlarged, and dull, while the living cells remained in a spherical form and were transparent and shiny. Cell viability was calculated as the division of the number of live cells to the number of live and dead cells. In the co-culture system, a Cell Counting Kit-8 cell viability assay kit (G021-1; Nanjing Jiancheng Biological Engineering Institute, China) was used to check the viability of the CD8+ T cells, based on the manufacturer’s instructions with some modifications. Subsequently, the optical density values of the cells were obtained using a microplate reader (BioTek, USA).
2.7. Transcriptome analysis
RNA from bone marrow and colon tissue was extracted with TRIzol (mf034-01, Mei5 Biotechnology Co., Ltd., China). After checking the concentration and quality using a Nanodrop 2000 (Thermo Fisher Scientific) and Agilent Analyzer 5300 (Agilent, USA), followed by 1% agarose gel electrophoresis, the RNA was incubated with Oligo (dT) magnetic beads to isolate messenger RNA (mRNA) for the transcriptome analysis. The results for the RNA quality are provided in Fig. S1 in Appendix A.
The mRNA was treated with a fragmentation buffer to randomly break it down into small fragments of approximately 300 bp; it was then screened by means of magnetic beads. The enriched short mRNA was reverse transcribed into double-stranded complementary DNA (cDNA) and random primers. End Repair Mix was added to repair the cDNA to the flat end, and the cDNA products were amplified via polymerase chain reaction (PCR) for 15 cycles. Targeted PCR products were recycled using 2% agarose gel electrophoresis. cDNA was quantified with TBS380 (Promega, USA) and mixed based on a uniform standard. The mixtures were amplified by bridge PCR to generate clusters, which were then read using the Illumina platform. The generated raw data were processed and analyzed on the Majorbio Cloud Platform‡. Differences in gene expression were counted using the built-in RSEM software. Functional enrichment of differential genes was performed by means of the annotations of the Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases.
2.8. Transcription quantitative PCR for gene expression
The method of reverse transcription quantitative PCR (RT-qPCR) used for gene expression followed a previous article [
23]. In brief, the RNA was extracted from tissues using TRIzol (Thermo Fisher Scientific). After separation with chloroform and precipitation with isopropanol, the RNA was acquired and then cleared with 75% ethanol. The RNA was dissolved in diethylpyrocarbonate-treated water following desiccation. Based on the uniform 1000 ng RNA in a 20 μL volume, all samples were inversely transcribed into cDNA using commercial SuperFast qPCR RT kits (MF012, Mei5 Biotechnology Co., Ltd.). The mRNA expression of the target genes in the samples was checked using the commercial Realtime PCR Super Mix with Low Rox kits (MF797-01, Mei5 Biotechnology Co., Ltd.) and corresponding primers, which was performed on the ABI QuantStudio5 system (Thermo Fisher Scientific). The primers were designed using DNAman5.0 software (Lynnon Biosoft, USA); all the primer sequences are listed in Table S1 in Appendix A. Primers were synthesized by Sangon Biotech Co., Ltd. (China). GAPDH was used as the reference, and the relative gene expression was calculated using the 2
−ΔΔCT method, where CT is cycle threshold.
2.9. Absolute abundance of C. difficile in colonic chyme
The contents of the toxigenic C. difficile in the colonic chyme were checked. Quantitative chyme was extracted from the microbial DNA using commercial Stool DNA Kits (D4015-01*, OMEGA, USA). Then, the RT-qPCR method was used to acquire the CT values for quantitative chyme with the specific primers for TcdB, which are also listed in Table S1. To calculate the C. difficile numbers, a standard curve between the bacterial counts and the CT values in RT-qPCR was established. In brief, the PCR product was cloned into a pESI-T vector with the ampicillin resistance gene and then transformed into competent Escherichia coli (E. coli) DH5α. E. coli with ampicillin resistance was selected from the Luria-Bertani (LB) agar medium and cultured in an LB liquid medium with ampicillin. Specific volumes of E. coli solution were used to extract the DNA and perform RT-qPCR following the same method outlined above. Next, the concentrations of E. coli were quantified via gradient dilutions and culture on LB agar medium. The absolute numbers of C. difficile in the chyme were calculated based on the standard curve.
2.10. Determination of lipopolysaccharide and IL-17A levels
The lipopolysaccharide contents in the plasma were detected using a commercial Mouse Lipopolysaccharides ELISA kit (CSB-E13066m, CUSABIO, China). The IL-17A contents in the plasma and colon tissue were checked using Mouse Interleukin 17 ELISA Kits (CSB-E04608m, CUSABIO). Before the determination of colon tissue, the tissues were homogenized in PBS at a dilution of 1:1000 (g:μL) with the tissue homogenizer, according to the manufacturer’s instructions. The protein concentrations in the homogenized samples were detected using BCA Protein Assay Kits (P0012, Beyotime, China) and were used for the IL-17A contents determination.
2.11. Hematoxylin and eosin and immunohistochemical staining
The tissues acquired for sections were fixed in 4% paraformaldehyde. For the bone marrow, the bone tissues were softened with an EDTA decalcifier (GP1013, Servicebio, China). The subsequent preparation method for tissue sections and hematoxylin and eosin (H&E) staining followed previously published methods [
24], [
25]. For the immunohistochemistry (IHC) staining, after blocking with goat serum for 30 min, the rehydrated sections were incubated with anti-IL-17A rabbit pAb (26163-1-AP, Proteinteck, USA) or anti-myeloperoxidase rabbit pAb (GB11224-100, Servicebio) overnight at 4 °C. The tissues were then incubated with horseradish peroxidase (HRP)-labeled goat anti-rabbit IgG (SE134, Servicebio) at room temperature for 2 h, followed by staining with 3, 3’-diaminobenzidine working solution (DA1010, Servicebio) for 2 min. After another wash with PBS, the sections were sealed. The H&E and IHC sections were viewed using an Olympus BX53 microscope (Olympus Corporation, Japan).
2.12. Statistical analysis
All data are shown as the mean ± standard error of the mean (SEM). Data and statistical analyses were performed using GraphPad (version 9.0.1). Student’s t-test was used for the comparison between two groups, and one-way analysis of variance (ANOVA) and post hoc Tukey’s multiple comparisons were applied to check the discrepancy among three or more groups. P < 0.05 represents a significant difference.
3. Results
3.1. Aging of the bone marrow is characterized by increased PMN-MDSCs and decreased bone trabecular
To investigate the aging of bone marrow, we first analyzed the histopathology of the proximal metaphysis in femur bone marrow from mice that were 3, 12, 18, 24, and 28 months (M) old by H&E staining (
Fig. 1(a)). The most obvious structural change was the abundant bone trabeculae in the young- (3M) and middle-aged (12M) mice, which was obviously reduced in the old (24M) and very old (28M) mice compared with the young mice (
P < 0.01;
Fig. 1(a)). Aside from the structural changes, the primary bone marrow cells isolated from very old mice (28M) exhibited the same numbers (
P > 0.05; Fig. S1(a) in Appendix A) but showed a reduced trend in viability compared with those in young mice (3M; 0.05 <
P < 0.1; Fig. S1(b) in Appendix A).
To further analyze the changes in immune cell subtypes in bone marrow during aging, we designed a panel containing antibodies for 41 markers of immune cells and HSCs (Fig. S1(b)) and measured by means of CyTOF to cluster and annotate 34 cell subgroups (
Fig. 1(b)). These 34 cell subgroups were annotated as MDSCs, monocytes/macrophages, B cells, CD8 T cells, CD4 T cells, natural killer (NK) cells, dendritic cells (DCs), and HSCs (
Fig. 1(c)); the signal intensities of the markers in each subgroup are presented in the heatmaps (Fig. S1(b)). Notably, the MDSCs could be further divided into 13 PMN-MDSC subgroups and two monocytic (M)-MDSC subgroups (
Fig. 1(c) and Fig. S1(c) in Appendix A), among which the relative abundances of the C05, C07, and C08 subgroups of PMN-MDSCs were significantly higher (
P < 0.05;
Figs. 1(d)-(f)) during aging, whereas the relative abundances of the C12 and C15 subgroups decreased significantly (
P < 0.05;
Figs. 1(g) and
(h)). In contrast, there were no significant differences in the percentages of the other nine subgroups of PMN-MDSCs (C06, C09, C10, C11 C13, C14, C16, and C17), in the two subgroups of M-MDSC (C21 and C22), or in the HSCs (C25) and other cell types (
P > 0.05;
Fig. 1(i) and Fig. S2(a) in Appendix A), except for the circulating monocytes (C29), which increased significantly in the bone marrow from the 28M mice (
P < 0.05;
Fig. 1(j)).
To verify the increased PMN-MDSCs in the bone marrow, we evaluated the gene expression of the PMN-MDSC markers
CD11b (
Fig. 1(k)) and
Wfdc17 (
Figs. 1(k) and
(l)). The relative mRNA expression of
CD11b increased significantly in the bone marrow from mice in the 28M group (
P < 0.05;
Fig. 1(k)), and the expression of
Wfdc17 increased significantly in the very old mice (28M) (
P < 0.05;
Fig. 1(l)). Together, these results indicated that bone marrow aging was characterized by a significant increase in the PMN-MDSC population.
3.2. Increased PMN-MDSCs were closely associated with bone marrow aging
To evaluate the influences of the increased PMN-MDSCs during aging on the immune function of bone marrow, we compared the transcript profiles of bone marrow from young (3M) and very old (28M) mice using bulk sequencing (
Fig. 2(a)). The RNA integrity of the bone marrow confirmed the reliability of the transcriptome data (Fig. S2(c) in Appendix A). As shown in the volcano plot (
Fig. 2(b)), 564 genes were upregulated and 268 genes were downregulated.
To verify these transcriptome results, we analyzed some genes with differences in expressions via RT-qPCR, including
Ugt1a5,
Reg3g,
Cdh20, and
Exosc6, which demonstrated results consistent with those of the transcriptome analysis (Fig. S3(a) in Appendix A). According to the enriched KEGG (
Fig. 2(c)) and GO (
Fig. 2(d)) pathway analysis of genes with differential expression, compared with the young mice (3M), several immune-response-related pathways were enriched in the very old mice (28M), including “primary immunodeficiency,” “viral protein interaction with cytokine and cytokine receptor,” “cytokine-cytokine receptor interaction,” “T cell receptor signaling pathway,” “Th17 cell differentiation,” “cell adhesion molecules,” “chemokine signaling pathway,” and “complement and coagulation cascades,” in the KEGG analysis (
Fig. 2(c)), and “immunoglobulin production,” “production of molecular mediator of immune response,” and “immunoglobulin complex, circulating” in the GO analysis (
Fig. 2(d)).
To verify the disorders in immune functions, we measured the mRNA expressions of some differential genes in bone marrow from young (3M), old (24M), and very old (28M) mice, including
Il21,
CD33,
CD3g,
Bcl2a1b, and
HDAC5 (
Fig. 2(e)). Expression of
Il21,
Bcl2a1b, and
CD33 increased significantly in old and very old mice (
P < 0.05;
Fig. 2(e)), whereas the expressions of
CD3g only increased non-significantly compared with young mice (
P > 0.05;
Fig. 2(e)). Moreover,
HDAC5 in young mice was significantly down-regulated in comparison with that in old and very old mice (
P < 0.05;
Fig. 2(e)).
The effects of PMN-MDSCs on the immune functions of bone marrow were also evaluated by means of co-culture with CD8
+ T lymphocytes. The viability of the CD8
+ T cells was significantly decreased by co-culture with PMN-MDSCs, irrespective of whether they were isolated from bone marrow or blood (
P < 0.05;
Fig. 2(f)). In addition, the secretion of perforin from the CD8
+ T cells was inhibited by the PMN-MDSCs, especially those isolated from the original bone marrow (
P < 0.05;
Fig. 2(g)). These findings further confirmed the identity of the PMN-MDSCs and indicated that the increased PMN-MDSCs in bone marrow during aging contributed to the immunosuppression of aged bone marrow.
3.3. Single-cell sequencing analysis revealed increased PMN-MDSC abundance and related changes in immune function in aged bone marrow
To explain the different changes in the PMN-MDSCs, we predicted the differentiation locus of the HSCs using pseudotime analysis. As shown in Fig. S2(b), in very old mice (28M), HSCs (C25) were prone to differentiate into the C07, C08, and C29 subgroups through the C21 and C05 subgroups compared with the HSCs in young mice (3M). The bulk sequencing results also showed that the differential genes regulated hematopoietic and cell development; as shown in
Fig. 2(c), the KEGG enrichment analysis indicated that the differential expressed genes were involved in the hematopoietic cell lineage and MAPK signaling pathways. To confirm the changes in the MDSC and PMN-MDSC communities, we reanalyzed these cell subtypes in the bone marrow using published single-cell sequencing data from a public database. As shown in
Fig. 3(a), UMAP plots of bone marrow from 1M, 6M, and 20M mice revealed 19, 21, and 17 cell populations, respectively. Based on the same markers used in the CyTOF analysis, their expression was shown in the UMAP plots (Figs. S3(c)-(e) in Appendix A). The MDSC and PMN-MDSC subgroups were identified, and were found to increase in the old 20M mice (
Fig. 3(b)). Among the PMN-MDSCs, the C07 subpopulation showed the highest abundance, followed by C05 (
Fig. 3(b)). In accordance with the CyTOF results, the abundance of circulating monocytes (C29) also increased in the 20M group (
Fig. 3(b)). When we visualized the distributions of the total MDSCs and PMN-MDSCs in the UMAP plots (
Fig. 3(c)), we found that the majority of the MDSCs and PMN-MDSCs were distributed in the innate immune cells, including neutrophils, eosinophils, granulocytes, macrophages, and monocytes, among which the neutrophils were the main community in these three age stages. The distribution of the total MDSCs, the C05, C07, C08, C12, and C15 PMN-MDSCs, and the C29 monocytes in the UMAP plots at 1M, 6M, and 20M are shown in Figs. S3(b), (f)-(k) in Appendix A.
For the neutrophils, we performed GSEA for the genes in the 20M mice compared with the 6M mice. According to the results, the genes involved in complement (
Fig. 3(d)), and inflammation response (
Fig. 3(e)), IL6-JAK-STAT3 signaling (
Fig. 3(f)), IL2-STAT5 signaling (
Fig. 3(g)), interferon-alpha signaling (
Fig. 3(h)), interferon-gamma response (
Fig. 3(i)), TGF-beta signaling (
Fig. 3(j)), TNFα signaling via NF-κB (
Fig. 3(k)), and mTORC1 signaling (
Fig. 3(l)) were shown to be enriched and these pathways were upregulated, all of which were involved in regulating immune responses.
3.4. Increased PMN-MDSCs contributed to the susceptibility to C. difficile infection in old mice
Based on the immunosuppressive activity of the PMN-MDSCs, we further investigated whether increased PMN-MDSCs in bone marrow changed the immune response and extra-marrow tissues during aging, based on the
C. difficile VPI 10463 infection experiments (
Fig. 4(a)). After infection, the body weights of the young mice in the 3M group were not affected by
C. difficile infection (
P > 0.05), whereas those of the old mice in the 24M group showed a reduced trend (0.05 <
P < 0.1,
Fig. 4(b)). In addition, the body weights of mice in the 24M +
C. difficile group were significantly lower than those in the 3M +
C. difficile group (
P < 0.05;
Fig. 4(b)). The colonic morphology shown in
Fig. 4(c) also indicated susceptibility to
C. difficile in old mice (24M), characterized by apparent inflammatory foci. Furthermore, the plasma lipopolysaccharide (LPS) levels were significantly increased in 24M mice after
C. difficile infection (
P < 0.01;
Fig. 4(d)), whereas the corresponding increase did not appear in the young mice (3M group) (
P > 0.05;
Fig. 4(d)).
To evaluate the involvement of the PMN-MDSCs in the infection, we detected the proportion of CD11b
+ cells and PMN-MDSCs in the bone marrow, blood, and large intestine using flow cytometry. In the bone marrow (Figs. S4(e) and (f) in Appendix A), the CD11b
+ cells increased in the mice of the 3M group, but not in the old mice (24M), after
C. difficile infection, whereas they remained unchanged in the blood (Figs. S4(e) and (f)). Furthermore, the proportion of PMN-MDSCs in the total CD11b
+ cells in the bone marrow (
Figs. 4(e) and
(g)) and blood (Figs. S4(f) and (h) in Appendix A) increased significantly with aging (
P < 0.05) compared with that in the young 3M mice; they increased with
C. difficile infection as well, especially in the bone marrow (
P < 0.05;
Fig. 4(g)).
Regarding the two cell populations in the large intestine, it was interesting to note that the proportion of CD11b
+ cells decreased after infection with
C. difficile, especially in the mice from the 3M group (
Fig. 4(i) and Fig. S4(h)). Similar to the results in bone marrow and blood, the PMN-MDSCs increased significantly in the large intestine of the 24M group before showing a decreased trend after
C. difficile infection (0.05 <
P < 0.1;
Fig. 4(j) and Fig. S4(h)). The involvement of PMN-MDSCs in
C. difficile infection was also evaluated by the expression of markers for MDSCs and PMN-MDSCs in the colon and bone marrow (
Figs. 5(a) and
(b)). As shown in
Fig. 5(b), expression of
Wfdc17 increased significantly in old infected mice in comparison with young infected mice (24M +
C. difficile group vs 3M +
C. difficile group,
P < 0.01). Taken together, these results suggest that bone-marrow-derived PMN-MDSCs may be delivered to the colon through blood to alleviate the inflammation induced by
C. difficile infection in young animals, whereas the susceptibility to infection during aging may be due to the suppression of mucosal immunity induced by increased PMN-MDSC input from bone marrow to the colon and the reduced delivery of PMN-MDSCs after infection, with accumulation in the bone marrow.
To further verify the above hypothesis, we next investigated the expressions of genes involved in immune response in the colon (
Fig. 5(a)) and bone marrow (
Fig. 5(b)). In the bone marrow, the 24M group showed a non-significant increase in the expression of
CCL5 compared with the 3M group (
Fig. 5(b)). The expression of
CCL8 and
CXCL5 increased significantly in mice from the 24M +
C. difficile group compared with the 3M +
C. difficile group (
P < 0.05;
Fig. 5(b) in Appendix A). Similar results were found in the colon (
Fig. 5(a)); changes in the expression of the genes involved in immunity induced by aging were also only found after
C. difficile infection. The contents of
TNF-α,
IL-10,
IL-12a,
IL-22, and
IFN-γ were all upregulated in the 24 M +
C. difficile group compared with the 3M +
C. difficile group (
Fig. 5(a)). Although the difference in the mRNA expression of
IL-17A was not significant, the contents of IL-17A measured by ELISA and the Th17 cell number evaluated by IHC increased significantly along with age, with or without
C. difficile infection (
P < 0.05;
Figs. 5(c)-(e)). Together with the same load of
C. difficile in the feces of young and old mice after infection (
P < 0.05;
Fig. 5(g)), these findings indicated that the inhibited mucosal defense function, decreased delivery of PMN-MDSC from bone marrow to colon, and resulting inflammation in the colon after infection may explain the susceptibility to
C. difficile in old subjects.
3.5. Clearance of PMN-MDSCs in aged bone marrow significantly decreased the load of C. difficile in the feces and the susceptibility of aged mice
To confirm the role of increased PMN-MDSCs in bone marrow in the susceptibility to
C. difficile in aged mice, we cleared the PMN-MDSCs with anti-Gr-1 antibody by means of intraperitoneal injection (
Fig. 6(a)). The proportion of CD11b
+ cells in the aged mice in the 24M group was increased significantly by antiGr-1 treatment, both in the bone marrow (
P < 0.05; Figs. S5(d) and (e) in Appendix A) and in the blood (
P < 0.05; Figs. S5(f) and (g) in Appendix A), whereas its proportion in the large intestine was not affected (Figs. S5(h) and (i) in Appendix A). In addition, the PMN-MDSC abundances in the CD11b
+ cells in the bone marrow (
Figs. 6(b) and
(c)) and large intestine (
Fig. 6(e); Fig. S5(h)) decreased significantly in the mice from the antiGr-1 group (
P < 0.05) but remained unchanged in the blood (
P > 0.05;
Fig. 6(d) and Fig. S5(f)). After infecting these mice with
C. difficile, the PMN-MDSC abundances in the CD11b
+ cells of the bone marrow and large intestine were still decreased (
P < 0.05; Figs. S6(c)-(h) in Appendix A), while the loads of
C. difficile in the feces of the old mice were significantly decreased in the antiGr-1-treated group (
P < 0.05;
Fig. 6(f)). Furthermore, the plasma levels of LPS decreased significantly after
C. difficile infection in mice pretreated with antiGr-1 to eliminate PMN-MDSC in aged bone marrow (
P < 0.05;
Fig. 6(g)).
To fully evaluate the changes in the colon after pre-eliminating PMN-MDSCs in the old mice, we performed bulk sequencing after infection. The RNA integrity of the colon tissue confirmed the reliability of the transcriptome data (Fig. S6(i)). As shown in
Fig. 6(h), there were 351 upregulated genes and 732 downregulated genes when comparing the antiGr-1 +
C. difficile group with the 24M +
C. difficile group. GO (
Fig. 6(i)) and KEGG (
Fig. 6(j)) enrichment analyses suggested that the differential genes mainly regulated pathways related to “immunoglobulin complex, circulating,” “phagocytosis, engulfment,” “complement activation, classical pathway,” “cell recognition,” and immune response, including “antigen processing and presentation,” as well as the “IL-17 signaling pathway”. In the detection of genes in the “IL-17 signaling pathway,” the expression of proinflammatory factors, including
IL-17A,
IL-17D,
IL-17re,
IL-1β, and
IL-6, were all reduced in the antiGr-1 +
C. difficile group (
P < 0.05;
Fig. 6(k)). These results indicated that pretreating the old mice with antiGr-1 to eliminate PMN-MDSCs in the bone marrow could significantly alleviate the infection of
C. difficile and the increased inflammation in the colon.
4. Discussion
Aging is accompanied by a decline in immune defense function, which means that older people often exhibit life-threatening pathological reactions to infection. This was especially evident during the corona virus disease 2019 (COVID-19) pandemic [
26], [
27]. To achieve healthy aging, it is necessary to enhance resistance to pathogens and reduce abnormal inflammatory responses by intervening in immunosenescence. As the central immune organ, bone marrow regulates the immune homeostasis of the body by exporting immune cells and related factors [
28], [
29]. However, there have been limited studies on the changes in the generation of immune cells in bone marrow and the effects of these changes on the immune function of peripheral organs during aging.
4.1. Aging contributes to the generation of PMN-MDSCs in bone marrow
As an important research tool, CyTOF technology enables the precise identification of immune cell lineages and the analysis of functional associations among subpopulations [
30], [
31]. Using this technology, we found that the abundance of the major subgroups in the PMN-MDSC community increased dramatically in aged bone marrow (
Figs. 1(d)-(f)). A reanalysis of the single-cell transcriptome data of bone marrow from youth to old age also favored this result (
Fig. 3(b)). The flow cytometry results further confirmed the higher abundance of PMN-MDSCs in aged bone marrow (
Figs. 4(e)-(g)). Taken together, these findings suggest that the generation of PMN-MDSCs increases in the bone marrow during aging.
MDSCs include M-MDSCs and PMN-MDSCs, which participate in immune suppression [
32]. Previous research [
33] has focused on their immunosuppression function and their contribution to the carcinogenesis process under pathological circumstances, while only a few study [
34] has reported changes in MDSC abundance in healthy, old individuals. In our study, we first proved that the output of PMN-MDSCs in bone marrow was aggravated with aging, using multiple methods. In tumor-free mice, PMN-MDSCs have a much shorter lifespan than their counterpart neutrophils because of enhanced apoptosis signaling via TRAIL-Rs [
35], [
36]. Although not at the level of significance due to the small sample size, the reduced survival of bone marrow cells in this study (Fig. S1(b)) also confirmed the characteristic of increased PMN-MDSCs. The pseudotime analysis suggested that the differentiation of HSCs in bone marrow was biased toward PMN-MDSCs (Fig. S2(c)).
The driving force for this change in the differentiation direction of HSCs may be related to the changes in bone trabeculae, which provide the major microenvironment for HSCs and are gradually reduced during aging (
Fig. 1(a)) [
37], [
38]. Blocking the IL-1 pathway in the bone marrow can restore HSC function, which also reconstructs the bone trabecula structure [
39]. Therefore, a decrease in the number of bone trabeculae may damage the niche of HSCs and then change the subsequent differentiation fate to PMN-MDSCs. Moreover, chronic inflammation accompanied by aging may affect the microenvironment and change the differentiation fate of PMN-MDSCs. Age-associated inflammation can impair myeloid progenitor differentiation and alter the responses to endogenous and exogenous threats [
40]. In naturally aged colons, the increased IL-17A content and Th17 cells also illustrate the inflammatory state. Therefore, increased generation of PMN-MDSCs in the bone marrow may be the reason for the response to peripheral inflammatory signals and the inhibition of potential excessive inflammation [
41].
Blood circulation is the main pathway for the transport of bone-marrow-generated cells to peripheral organs and tissues. In blood, even though there are fewer PMN-MDSCs, their abundance is increased in aged mice (
Fig. 4(h) and Fig. S4(f)). Similar results were observed in the large intestine (
Fig. 4(j) and Fig. S4(h)). Therefore, it can be speculated that the export of PMN-MDSCs from bone marrow to the large intestine is enhanced during aging. In humans, the abundances of PMN-MDSCs and total MDSCs are also increased in the blood of older people aged 60-100 years [
42], [
43]. In idiopathic pulmonary fibrosis, elevated MDSC levels in the blood were found to be inversely correlated with lung function and positively related to the severity of sepsis [
44], [
45], indicating that MDSC levels may be promising biomarkers for disease progression. Therefore, increased export of PMN-MDSCs may be a reason for immune defection in the elderly population.
In addition, it is widely accepted that PMN-MDSCs are immature and pathologically activated neutrophils [
46]. However, according to the single-cell transcriptome data, the immune cells with the surface markers of PMN-MDSCs only found in young mice were mainly found in neutrophils (
Fig. 3(c)). In the adult and elderly, these markers were found not only in neutrophils but also in other immune cells, including eosinophils, granulocytes, macrophages, and monocytes (
Fig. 3(c)). This finding suggests that PMN-MDSCs are a group of heterogeneous innate immune cells that play different biological functions in anti-infection and defense [
47], [
48], [
49].
4.2. Accumulated PMN-MDSCs disturb the immune homeostasis of the bone marrow and promote the susceptibility of the large intestine to C. difficile
PMN-MDSCs act as a brake in the immune system [
50], [
51], [
52]. The PMN-MDSCs isolated from the bone marrow and blood of aged mice inhibited CD8
+ T cell proliferation and perforin secretion during co-culture. Study [
36] on PMN-MDSCs and M-MDSCs has found that MDSCs promote tumor development and metastasis through local immunosuppression. Therefore, we inferred that the accumulated PMN-MDSCs may inhibit the normal immune response in the bone marrow and peripheral tissues during aging. Bulk sequencing results showed that multiple immune-related pathways—including “Th17 cell differentiation,” “primary immunodeficiency,” and “multiple inflammatory pathways”—were enriched in aged bone marrow (
Fig. 2(c)), along with the expression of some immune-regulated genes in bone marrow, such as
IL21 and
CD33 (
Fig. 2(e)). Moreover, the expression of
HDAC5 was reduced in aged bone marrow (
Fig. 2(e)), which limits histone acetylation and then regulates gene expression via epigenetic regulation to affect mitotic fidelity and cell function [
53], [
54]. In addition, in the single-cell transcriptome analysis, the neutrophil community, including PMN-MDSCs, in aged bone marrow showed abnormal activation of inflammatory and infection-related pathways (
Figs. 3(d)-(i)). As a central regulator of longevity and aging, the mTORC1 signaling pathway was also upregulated. Overactive mTOR pathways lead to cell dysfunction and immunosuppression, and mTOR inhibitors alleviate immunosenescence in mice and humans [
55]. Therefore, accompanied by the increased PMN-MDSCs, the immune functions of the bone marrow are disordered. Together, these findings suggest that aging disturbs the immune regulation function of bone marrow with an increase in PMN-MDSCs, which may increase the susceptibility to pathogens in aged populations.
C. difficile infection is common and lethal in elders, especially those who are hospitalized [
56]. In this study, we found that aged mice also showed increased susceptibility to
C. difficile infection (
Figs. 4(b)-(d)), along with increased PMN-MDSCs in the large intestine (
Fig. 4(j)). In addition, the expression of proinflammatory cytokines and the proportion of Th17 cells and neutrophils increased in the colon of aged mice (
Figs. 5(a)-(g)).
Abernathy-Close et al. [
57] also found that aging dampened the intestinal innate immune response and increased IL-17A expression during the defense against
C. difficile infection. However, in line with this study, the increased susceptibility of aged mice was not associated with the absolute abundance of
C. difficile (
Fig. 5(h)), indicating that the inhibited bone marrow immune response and mucosal immunity of the large intestine may be the key reason for the infectibility [
58]. Some scientists [
58] have highlighted that PMN-MDSCs may play a positive effect in alleviating excessive inflammation to protect tissues, due to the existence of PMN-MDSCs in the bone marrow and tissues of young animals. As described earlier in this study [
59], the increase in PMN-MDSCs in bone marrow may be a result of the response to peripheral inflammation. Under these circumstances, it was reported that PMN-MDSCs induced by the endothelin-a receptor antagonist BQ123 play a key anti-inflammatory role, and exosomes released from PMN-MDSCs can reduce DSS-induced colitis and contribute to wound healing in young mice [
60], [
61].
Therefore, to clarify the role of PMN-MDSC in
C. difficile infection, we eliminated PMN-MDSCs using the anti Gr-1 antibody, which was applied in previous studies [
62], [
63]. As shown in
Fig. 6(f), PMN-MDSC elimination reduced the absolute abundance of
C. difficile and intestinal permeability and inhibited the IL-17 signaling pathway (
Figs. 6(k) and
(l)). The susceptibility and defense capacity of the aged colon improved. Therefore, PMN-MDSCs have a negative effect on increased sensibility to
C. difficile in the aged population, and the elimination of PMN-MDSCs improves the immune response in the aged colon mucosa. As for why PMN-MDSCs are no longer exported from the bone marrow to the blood and then the large intestine to control colonitis induced by
C. difficile infection, we speculated that the abnormal elevations of chemokine
CCL8 and
CXCL5 in aged bone marrow may be one of the reasons. However, it is noteworthy that the anti Gr-1 antibody used to clear the PMN-MDSCs in this study also cleared the M-MDSCs [
64] because of the lack of a specific antibody for PMN-MDSCs, representing a shortcoming of this study. In further studies, different subpopulations of PMN-MDSCs should be isolated to evaluate the effects of subpopulations—rather than the total PMN-MDSC population—on immunosenescence. In particular, M-MDSCs inhibit the immune defense function of the colon mucosa and thus leads to increased susceptibility to pathogens.
5. Conclusions and implications
In brief, cell differentiation of bone marrow in aged mice is biased toward PMN-MDSCs, and the resulting increased PMN-MDSC output from the bone marrow renders the mice more susceptible to C. difficile infection. Basic research on delaying immune senescence should be conducted to further explore the effects of structural degradation and microenvironment changes in bone marrow on the fate and functions of HSCs, clarify the molecular mechanism, and then identify potential pathways for intervention. PMN-MDSCs may be a potential clinical target of immunosenescence, with important research value for the treatment of aging diseases such as antibody therapy and stem cell therapy. This study not only provides a theoretical basis for the increased susceptibility to pathogens caused by aging but also gives a direction for the future development of targeted nutritional interventions and prevention measures.
Acknowledgments
Our profound admiration and respect go to the researchers in this field and in our laboratories for their dedication and hard work. We apologize to scientists whose work is in this field but whose papers are not cited in this study owing to space limitations. This work was supported by the National Key Research and Development Program of China (2022YFF1100504), the 111 project from the Education Ministry of China (B18053), the National Natural Science Foundation of China (32101938 and 32302758), the China Postdoctoral Science Foundation (2022M723422), and the Postdoctoral Fellowship Program of CPSF (GZB20230848).
Compliance with ethics guidelines
Jianmin Wu, Ming Zhang, Hao Zhang, Mingxuan Sheng, Jiazeng Sun, Fang Wu, Haina Gao, Lishui Chen, Zhili Li, Qiyu Tian, Longjiao Zhu, and Bing Fang declare that they have no conflict of interest or financial conflicts to disclose.
Data availability statement
We permit unrestricted use, distribution, and reproduction in any medium provided the original work is properly cited. The RNA-seq data files generated in this study have been deposited in Sequence Read Archive (SRA) of The National Center for Biotechnology Information (NCBI) by the accession number PRJNA1062588 and PRJNA1062443. The raw data supporting the conclusions of this article will be made available by the authors without undue reservation.
Appendix A. Supplementary material
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.eng.2024.06.014.
PDF
(8732KB)
