Peptidoglycan-Mediated Bone Marrow Autonomic Neuropathy Impairs Hematopoietic Stem/Progenitor Cells via a NOD1-Dependent Pathway in db/db Mice

Impairment of bone marrow-derived hematopoietic stem/progenitor cells (HSPCs) contributes to the progression of vascular complications in subjects with diabetes. Very small amounts of bacterial-derived pathogen-associated molecular patterns (PAMPs) establish the bone marrow cell pool. We hypothesize that alteration of the PAMP peptidoglycan (PGN) exacerbates HSPC dysfunction in diabetes. We observed increased PGN infiltration in the bone marrow of diabetic mice. Exogenous administration of PGN selectively reduced the number of long-term repopulating hematopoietic stem cells (LT-HSCs), accompanied by impaired vasoreparative functions in db/db mouse bone marrow. We further revealed that bone marrow denervation contributed to PGN-associated HSPC dysfunction. Inhibition of NOD1 ameliorated PGN-induced bone marrow autonomic neuropathy, which significantly rejuvenated the HSPC pools and functions in vivo. These data reveal for the first time that PGN, as a critical factor on the gut-bone marrow axis, promotes bone marrow denervation and HSPC modulation in the context of diabetes.


Introduction
Diabetes disrupts all bodily organ systems, including the bone marrow. Diabetes is associated with impairment of the compartmentalization and function of bone marrowderived hematopoietic stem/progenitor cells (HSPCs) [1,2]. HSPCs play a crucial role in vascular repair through paracrine mechanisms [3], and impairment of these cells leads to the progression of microvascular complications in subjects with diabetes [1]. Since dysfunction of HSPC populations strongly contributes to diabetes complications involving vessel degeneration, understanding the underlying mechanisms is critical for revascularization purposes.
The roles of the gut microbiome and pathogenassociated molecular patterns (PAMPs) in the pathogenesis of diabetes microvascular complications have garnered interest [4]. Bacteria-derived PAMPs translocate from the gut into the circulation through disrupted gut barriers in subjects with diabetes, thereby causing systemic effects, including chronic low-grade systemic inflammation, dyslipidemia, and insulin resistance [5,6]. We previously revealed the bacterial composition and functional profile of the diabetes microbiome. We determined that the levels of peptidoglycan (PGN), a type of PAMP, were substantially elevated in plasma samples from both humans and rodents with type 1 and type 2 diabetes and that increased PGN levels potentially promote diabetes microvascular complications [5,7]. However, how increased circulating PGN levels affect distant organs, especially the bone marrow, that strongly contribute to microangiopathy in diabetes remains largely unknown.
PGN is a major component of the bacterial cell wall that consists of amino acids and glycan chains of β- (1,4)-linked N-acetylglucosamine (NAG) and N-acetylmuramic acid (NAM) repeats [8]. Accumulating evidence suggests that very low concentrations of bacterial antigens control the size of the bone marrow cell pool [9]. A recent study showed that PGN promotes granulopoiesis and induces neutrophil production via a Toll-like receptor 2-dependent pathway [10], suggesting its involvement in gut-bone marrow communication. However, little is known about the impact of PGN on bone marrow HSPC populations. Thus, we herein tested the hypothesis that the bacterial PAMP PGN participates in the regulation of the bone marrow-derived HPSC pool and functions. PGN levels were markedly increased in the bone marrow extracellular fluid harvested from diabetic db/db mice, which was associated with imbalance of the bone marrow hematopoietic stem cell pool and HSPC dysfunction. Our data further showed that autonomic neuropathy underlies the PGN-mediated impairment of HSPC quantity and function and that bone marrow denervation relies on the nucleotide-binding oligomerization domaincontaining protein 1-(NOD1-) dependent pathway.

Materials and Methods
2.1. Animals. Male diabetic db/db mice are homozygous for the Lepr db mutation, and db/m mice that were heterozygous for Lepr db were used as controls in this study. Four-week-old male db/db mice, their age-matched db/m controls, and B6.SJL (CD45.1) mice were obtained from the Model Animal Research Center of Nanjing University and housed at the Animal Center of Chongqing Medical University (Chongqing, China). The experiments were conducted according to standard animal care protocols that were approved by the Committee on Animal Research of Chongqing Medical University. Experimental mice were randomly divided into different subgroups, and their blood glucose levels and body weights were measured every two weeks. Six-week-old db/db mice were used if their blood glucose levels exceeded 250 mg/dL in at least two measurements. Glycated hemoglobin was assessed on the day prior to euthanasia by the A1CNow+ kit (Bayer HealthCare, Sunnyvale, CA, US). For PGN treatment, 100 μg of E. coli PGN (cat#: tlrl-ksspgn, InvivoGen, San Diego, CA, US) was administered four months after the establishment of diabetes in db/db mice and to their age-matched db/m controls via the tail vein every other day for one week (on the 1 st , 3 rd , 5 th , and 7 th days). An equal amount of saline was injected into the mice as a control. In a separate experiment, the NOD1 receptor-specific inhibitor ML-130 (50 μg) (cat#: ab142177, Abcam, Cambridge, MA, US) was administered together with the purified PGN to establish the db/db+PGN +ML130 group. An equal amount of vehicle (DMSO/saline) was administered to age-matched control mice. For sympathetic nervous system disruption, the mice were administered the neuron toxin 6-hydroxydopamine 3 times via intraperitoneal injection according to a previously published protocol [11]. The animals were anesthetized and euthanized by isoflurane inhalation, followed by cervical dislocation at the study end point.

Bone Marrow Extracellular Fluid
Collection. The left femoral bone was flushed with ice cold phosphate-buffered saline (PBS) (cat#: 10010023, Thermo Fisher Scientific, Waltham, MA, US). Extracellular fluid was collected by centrifugation at 500 g for 10 minutes at 4°C and stored in liquid nitrogen until analysis.

Measurement of Peptidoglycan Levels.
Mouse plasma and bone marrow extracellular fluid were collected for PGN measurement. The experiments were performed using a mouse PGN ELISA kit (cat#: MBS263268 MyBoSource Inc., San Diego, CA, US) according to the manufacturer's instructions as previously described [5]. The absorbance at 450 nm was detected by an ELISA microplate reader. 2.9. CFU Assay. Cells from bone marrow or blood were treated with an ammonium chloride solution to lyse the red blood cells and then plated in MethoCult GF M3434 medium (STEMCELL Technologies) according to the manufacturer's instructions as previously described [13].

Bone
2.10. Long-Term Culture Initiating Cell (LTC-IC) Assay. Primary bone marrow feeder layers were cultured and established using MyeloCult M5300 and then irradiated according to the protocol from StemCell Technologies. 1 × 10 3 sorted LSK cells were cultured on feeder layers at 33°i n 5% CO 2 for 4 weeks with a weekly half media change. Cells were then harvested and plated into MethoCult GF M3434 in triplicate. Colonies were determined after 12 days and expressed as number of the CFUs per 1 × 10 3 sorted LSK cells as previously reported [14].
2.11. Immunohistochemistry. Paraffin-embedded femoral bones were sectioned and deparaffinized, and the slides were processed as previously described [15]. Briefly, after antigen retrieval and nonspecific protein blockage, the slides were stained with either polyclonal rabbit anti-NF200 (Sigma-Aldrich) or polyclonal rabbit anti-tyrosine hydroxylase (Millipore) at 1 : 100 dilutions. After antibody incubation, the slides were treated with R.T.U. Elite ABC reagent and NovaRED (Vector Laboratories), followed by counterstaining with Gill 2 hematoxylin (Thermo Fisher Scientific).
2.12. Data Analysis. All data were evaluated for normal distribution by the JMP 9 software. For multiple comparisons, one-way or two-way ANOVA was used, followed by a post hoc test. The nonparametric Kruskal-Wallis and Mann-Whitney tests were used if the data were not normally distributed. P < 0:05 was considered significant. The results are expressed as the mean ± SD.

Increased Peptidoglycan Levels in the Circulation and
Bone Marrow Extracellular Fluid Are Associated with Depletion of the HSPC Pool and Disruption of HSPC Function in db/db Mice. Bacterial PAMPs can translocate from the intestine into the circulation and even to distant organs through impaired gut barriers in subjects with diabetes. We examined whether type 2 diabetes impacts the levels of a bacterial PAMP, PGN. We found that the PGN levels were markedly increased in db/db mice 4 months after the establishment of diabetes compared to their age-matched db/m controls (Figure 1(a)). We also observed increased PGN levels in extracellular fluid collected from db/db mouse bone marrow (Figure 1(b)), suggesting translocation of the bacterial antigen PGN to distant tissues. Bacterial antigens control the bone marrow stem cell pool. To assess whether increased PGN levels in the bone marrow of diabetic mice are associated with HSPC dysfunction, flow cytometry was used to identify the percentages of various HSPC subpopulations in the mouse cohort ( Figure 1(c)). Even though the percentages of bone marrow Lin -/CD127 -/Sca-1 + c-Kit + (-LSK) cells did not differ between the diabetic and control groups (Figure 1(d)), LK cells isolated from the diabetic bone marrow showed impaired migration ability in response to the chemoattractant CXCL12 and reduced proliferation compared to that of LK cells from db/m control mice (Suppl. Figure 1A and B). These parameters are key indicators of vasoreparative ability in vivo, as cells must self-expand and translocate from the bone marrow to the vessel injury site in response to chemoattractants. We next assessed the colony formation abilities of general HSPCs in both peripheral blood and bone marrow by the colony formation assay. After 10-day culture of ACS-lysed blood cells harvested from diabetic mice, those cells showed a marked decrease in total blood colony-forming units (CFUs). In the CFU assays of bone marrow cells, the total CFUs did not differ between db/m and db/db mice; however, the number of CFU-G/M/ GM (CFU-granulocyte/monocyte/granulocyte, monocyte) was increased in diabetic mice compared to db/m mice, suggesting a shift in hematopoiesis towards proinflammatory cell types (Figure 1(g)). LSK cells were then subdivided into long-term (LT) and short-term (ST) repopulating HSCs; compared to those in the healthy controls, the number of LT-HSCs (Lin -/CD127 -/Sca-1 + c-Kit + /CD34 -CD135 -) was significantly reduced in the diabetic mice, while no change was observed in ST-HSCs (Lin -/CD127 -/Sca-1 + c-Kit + / CD34 + CD135 -) (Figures 1(e) and 1(f)). Because LT-HSCs represent the most primitive hematopoietic stem cells in the bone marrow, their reduced percentage supports that diabetes induces substantial hematopoietic deficiency in the bone marrow. We then performed the long-term culture initiating cells (LTC-IC) assay which is considered a surrogate in vitro assay for testing the function of the most primitive HSC population. db/db mouse LSK cells formed much less colonies than did db/m mouse LSK cells after the long-term culture ( Figure 1(h)). Similar finding was seen in in vivo BrdU incorporation assay, as there was a 36% reduction of BrdU + LT-HSCs in db/db mouse bone marrow (Figure 1(i)), suggesting diabetic LT-HSCs have impaired self-renewal capability. These data suggested that type 2 diabetes resulted in alteration of the bacterial antigen PGN in the circulation and bone marrow, which was accompanied by HSPC depletion and functional impairment.    Figure 2E-F). PGN injection worsened the diabetes-induced reduction in total blood CFUs, which also verified the effect of PGN on the differentiation ability of HSPCs. In addition, the number of CFU-G/M/GM in the bone marrow of db/db mice at 4 months of diabetes was increased as determined by the CFU assay. The number of CFU-G/M/GM of db/db mice treated with PGN was increased to a greater extent that than in mice treated with saline, suggesting that PGN worsened the diabetes-mediated hematopoietic shift towards a more proinflammatory cell type ( Figure 2(e)). PGN treatment caused dramatic reduction in colonies forming by LTC-IC assay using LSK cells and in LT-HSC proliferation by in vivo BrdU corporation assay under diabetic condition, while no effect of PGN was observed in the control ones (Figures 2(f) and 2(g)). We further performed the competitive bone marrow transplantation assay to compare the bone marrow HSPC reconstitution potential among the 4 cohorts. Figures Figure 3(a)). Moreover, the proliferative abilities of LK cells harvested from diabetic mice treated with PGN and saline were similar (Figure 3(b)). We verified that the intravenous injection of PGN disrupted the hematopoietic balance in mice with diabetes by performing CFU assays; however, the colony-forming ability of LK cells from diabetic mice was unaffected by the ex vivo administration of PGN ( Figure 3(c)). Collectively, these data indicate that PGN may aggravate diabetes-induced HSPC impairment via an indirect mechanism.       Figure 5(a), ML130 ameliorated the reduced number of Try-OH + nerve fibers induced by PGN in mice with diabetes. The NOD1 inhibitor also protected against the deleterious effect of PGN on bone marrow autonomic neurons, as shown by NF200-positive staining ( Figure 5(b)). In addition, blockade of NOD1 prevented the reduced percentages of LT-HSCs in db/db mice (Figures 5(c)-5(f)). ML130 restored PGN-induced migration and proliferation dysfunctions of general HSPC populations from diabetic bone marrow (Suppl. Figure 3A-B). ML130 also reestablished the hematopoietic balance that was disturbed by PGN in db/db mice as determined by CFU assays (Figure 5(g)). In the LTC-IC study, diabetic LSK cells from PGN treatment group showed very little colony forming after a long-term culture, while the number of colonies was markedly increased in those pretreated with ML130 suggesting ML130 is protective for maintaining the fitness of more primitive HSPCs (Figure 5(h)). ML130 also restored PGNmediated reduction of BrdU + LT-HSCs in db/db mouse bone marrow ( Figure 5(i)).  Figure 3C-D). Chemical sympathectomy also disturbed the hematopoietic balance maintained by the NOD1 antagonist as determined by CFU assays (Figure 6(h)). Moreover, the LTC-IC study of LSK cells treated with the NOD1 inhibitor no longer showed improved numbers of colonies after chemical sympathectomy ( Figure 6(i)). 6-OHDA also prevented the recovery of proliferation function of LT-HSCs mediated by ML130 treatment (Figure 6(j)). These data suggest that the PGN receptor affects HSPC subpopulations and functions via bone marrow autonomic neurons.

Discussion
This study reported for the first time that PGN is a key regulator of maintaining HSPC homeostasis in diabetes. Such gut-bone marrow cross talk adds a new dimension to the current understanding of the complex pathogenesis of microvascular complications in diabetes. This study further investigated the possible mechanisms by which PGN impairs HSPC function and uncovered the disturbance of sympathetic nervous system innervation in the bone marrow as a potent driver of imbalance of hematopoietic system. A selective inhibitor of NOD1 protected against PGN-induced autonomic neuropathy and therefore promotes bone marrow HSPC rejuvenation in diabetes.
Extensive evidence suggests that both the composition and function of the gut microbiome are altered in type 1 and type 2 diabetes [16][17][18][19]. Increased gut permeability in subjects with diabetes allows the translocation of bacteria, their metabolites, and PAMPs from the gut to distant organs, thereby causing systemic effects [20]. Diabetic nephropathy (DN) and diabetic retinopathy (DR) are two common microvascular complications of diabetes, and intestinal dysbiosis contributes to the development of DN and chronic kidney disease (CKD) [21]. A recent study showed that alteration of the microbiome resulted in elevated plasma acetate levels, thereby inducing the activation of the local renal renin-angiotensin system and the progression of early kidney injury in a diabetic rat model [22]. Our previous studies also suggest an association between the gut microbiome and retinal disorders in subjects with diabetes [5,7]. The intermittent fasting prevented retinopathy by restructuring the gut microbiome and modulating microbederived metabolites in a diabetic db/db mouse model [7]. In another study [5], we compared the microbiome compositions and functions of diabetic Akita mice and their agematched littermates using 16S rRNA sequencing and metatranscriptomic analysis and revealed that PGN biosynthesis pathways were remarkably activated in diabetic mice, thereby increasing the circulating PGN levels. We further determined that PGN directly acts on human retinal endothelial cells and impairs their adherent junctions in vitro [5].
PGN is a very large polymer consisting of glycan chains and amino acids. As a PAMP, PGN is an essential unit of the bacterial cell wall found in most Gram-positive and Gramnegative bacteria [23]. PGN has been reported to trigger chronic inflammation, thereby promoting diet-induced insulin resistance in adipocytes and hepatocytes [24,25].
Here, we showed that increased circulating PGN lead to an elevated level of this type of PAMP in the local bone marrow of db/db mice, which selectively reduced the LT-HSCs and impaired reconstitution and proliferation abilities of primitive HSCs. We previously showed that PGN has a direct    [5].
Since bone marrow-derived HSPCs are important for vascular health maintenance, impairment of their function has long been suggested to remarkably contribute to microvasculature injury in subjects with diabetes. Therefore, the novel finding of this study provides a possibility of PGN indirectly affecting diabetes microvascular complications by disturbing the bone marrow HSPC balance. Interestingly, the deleterious effects of PGN on bone marrow innervation and LT-HSCs were only seen in diabetic mice, but not control ones, which suggests that there are some additional factors in diabetes that interact with the action of PGN. Our previous studies and others have shown that there were various protective mechanisms in healthy bone marrow which were compromised in diabetes, such as the activation of protective renin-angiotensin system (ACE2-Ang1-7-MAS axis), sufficient local levels of neuro-protective, and anti-inflammatory factors, as well as normal peripheral circadian rhythm [5,[26][27][28][29]. Therefore, it is possible that when those beneficial factors were perturbed by diabetes, they were unable to counteract the deleterious effects of PGN. Extensive investigation to understand the precise regulatory complex network of the bone marrow homeostasis is required.
The concept of the gut-bone marrow axis is supported by evidence that the gut microbiome is an extrinsic modulator of the hematopoietic system. In the clinic, bone marrow suppression is often observed in patients administered antibiotics for prolonged periods [30]. In animal studies, mice receiving antibiotics also exhibited impaired hematopoiesis with marked suppression of HSPC populations due to the depletion of intestinal flora [31]. Alteration of the microbiome composition in a Rag -/mouse model led to marked reductions in the LSK, LT-HSC, and ST-HSC populations, and the transplantation of feces from wild-type mice

10
Stem Cells International reversed such changes in the bone marrow of knockout mice [32]. The complexity of the gut microbiome is also strongly correlated with HSPC functions. Cytokine-induced HSPC mobilization was reduced in antibiotic-treated mice, suggesting that endotoxins, as potent cofactors, are involved in HSPC trafficking in the bone marrow [33]. Bioactive secondary metabolites of bacteria and PAMPs are considered major messengers in cross talk between the microbiome and bone marrow. Lipopolysaccharide (LPS) is one of the most well-studied PAMPs in diabetes and has been found to disturb the hematopoietic balance [34,35]. LPS administration drives quiescent LT-HSCs into the cell cycle and promotes the differentiation of HSPCs towards a more proinflammatory cell type, possibly via the direct activation of Toll-like receptor 4 on the cell surface [36,37]. On the other hand, LPS acts on nonhematopoietic cells, such as endothelial cells and mesenchymal stromal cells, and promotes the secretion of proinflammatory cytokines in the hematopoietic microenvironment, which indirectly affects HSPCs by engineering the bone marrow niche [38][39][40][41]. However, the effects of other PAMPs on HSPC populations, including PGN, have remained elusive, especially in the context of diabetes.
We previously performed genomic and metatranscriptomic analysis of the microbiome and revealed that PGN biosynthesis pathways were remarkably enriched, which resulted in increased circulatory levels of the PAMP in subjects with both type 1 and type 2 diabetes. However, whether increased circulating PGN levels modulate bone marrow cells in diabetes remains largely unknown. PGN reportedly translocates from the gut to the bone marrow and promotes neutrophil functions via a NOD1-dependent pathway [42]. Another study showed that PGN accelerates granulopoiesis by TLR2-dependent paracrine mechanisms and participates in innate immunity and host defense [10]. In our current study, we report for the first time that PGN administered in vivo not only modulated the HSPC pool size but also regulated their migration, proliferation, and reconstitution in subjects with diabetes. PGN receptors are widely expressed on hematopoietic cells, including HSPCs [43]. Interestingly, unlike PGN administration via the tail vein, the ex vivo administration of PGN did not worsen HSPC migration and proliferation, suggesting that the effects of PGN on HSPCs in subjects with diabetes are not attributable to the direct binding of cell surface receptors.
The bone marrow in subjects with diabetes is characterized by neuropathy, microangiopathy, and inflammation [2]. As a result, alteration of the bone marrow microenvironment disrupts the hematopoietic system, causing imbalance and functional impairment of HSPCs, which finely regulates the process of myelopoiesis and the pool size of progenitor cells with vasoreparative potential [2,11]. From a clinical perspective, the depletion of HSPCs is associated with the progression of microvascular complications in patients with diabetes and is a predictor of adverse outcomes [28,44]. PGN receptors were reported to be highly expressed in neurons, and a transgenic mouse model of PGN-sensing molecules showed altered expression levels of synaptic-related genes and behavioral disorders, suggesting a possible association between PGN and the central neuron system [45,46]. Direct PGN injection into the mouse brain parenchyma induced microglial cell activation and neurotoxicity via the PI3K-dependent signaling pathway [47]. We investigated whether the in vivo injection of PGN could impair HSPC function by compromising bone marrow autonomic neuropathy, as we did not observe a direct effect of PGN on HSPCs ex vivo. In this study, we found that the PGN levels were elevated in not only the circulation but also the local     The CFU assay showed increased CFU-G/M/GM of bone marrow cells cotreated with 6-OHDA and ML130 compared to those treated with ML130 alone (n = 14 per group). (i) 1 × 10 3 LSK cells from each group were cultured and tested for colonies in the LTC-IC assay. 6-OHDA blocked the beneficial effect of ML130 on PGN-induced reduction in the total number of colonies (n = 11 per group). (j) For the in vivo BrdU incorporation study, the bar graph in the (j) shows the percentages of BrdU + CD34 -LSK cells were decreased after he chemical sympathectomy induced by 6-OHDA irrespective of diabetic state by FACS analysis (n = 10 per group). Data represent mean ± SD. PGN: peptidoglycan; V: vehicle; LSK: Lin -/CD127 -/Sca-1 + c-Kit + ; LT-HSC: long-term repopulating hematopoietic stem cell; ST-HSC: short-term repopulating hematopoietic stem cell. * P < 0:05, * * P < 0:01; & compared to db/db+PGN; § compared to db/db+V; # compared to db/db+PGN+ML130; ※ compared to db/m+PGN+ML130; ¶ compared to db/db+PGN+ML130. 13 Stem Cells International finding innovatively demonstrates that the microbiota is associated with peripheral neuropathy and hematology in the context of diabetes and furthers our understanding of the precise mechanisms underlying diabetic bone marrow neuropathy and HSPC modulation.
Another novel finding of the study is that PGN affects bone marrow HSPCs via a NOD1-dependent pathway. Multiple proteins sense PGN and its fragments, including NOD1, TLR2, PGN recognition protein 1, and NOD-, LRR-, and pyrin domain-containing 3 [23]. We previously performed metatranscriptomic analysis and revealed that the enriched PGN biosynthesis pathways in the microbiota of diabetic mice were mainly responsible for the production of meso-diaminopimelic (meso-DAP) acid-containing muropeptides, which are specifically recognized by the NOD1 receptor [5]. NOD1 activation participates in multiple biological processes, such as innate immune regulation, chronic inflammation, and insulin resistance [25]. Consistent with our previous RNA sequencing data, our current study showed that a specific NOD1 inhibitor prevented PGNinduced bone marrow neuropathy and HSPC deprivation, further confirming the critical role of NOD1 in PGNmediated bone marrow defects in subjects with diabetes.

Conclusions
In summary, our study suggests for the first time the essential role of a microbial PAMP, PGN, in diabetes-induced HSPC depletion and impairment of their vasoreparative and reconstitution functions and highlights the importance of NOD1dependent autonomic neuropathy in this gut and bone marrow communication. These findings have expanded our understanding of the mechanisms underlying diabetesmediated bone marrow pathology and provide a novel therapeutic option for bone marrow rejuvenation for the prevention and treatment of diabetic vascular complications.

Data Availability
All data generated or analyzed during this study are included in this published article.