The BM is perfused by an extremely dense microvascular tree comprised by a variety of BM endothelial cell subtypes, which assemble into distinct vessels that largely differ in morphology, phenotype, and function (30, 31, 35–37). Following the archetypical structure of vascular trees, oxygenated blood enters the BM through arterial vessels, which branch into thinner arterioles. Arterial circulation transitions into large venous structures through a set of transitional vessels (also termed type H) in the proximity of endosteal surfaces (31, 32, 35). In the BM, venous thin-walled vessels construct a dense and highly interconnected network designated as sinusoidal (38) (Figure 1). Arterial vessels display low permeability, shielding the surrounding extravascular space from access to plasma-derived factors (36). Due to their anatomical localization at the interface between BM and bone surfaces, type H vessels critically influence bone remodeling (35). Sinusoids in turn are highly permeable and therefore serve as the main trafficking routes for cells in and out of BM tissues (39, 40).

Different subtypes of BMECs can be isolated and studied based on their specific phenotypic signature (29, 30, 36), the distinct expression of adhesion molecules, and the production of trophic hematopoietic factors, which has proven of relevance for their different biological functions (11, 41). As an example, sinusoidal ECs (SECs) among other angiocrine factors produce high levels of stem cell factor (SCF), CXCL12, and E-selectin, through which they critically participate in the maintenance and quiescence of HSPCs in perisinusoidal niches (42–45). Based on this molecular profile and the spatial interactions they establish, SECs have been further postulated as key regulators in different stages of hematopoietic cell development, such as the maintenance of mature megakaryocytes or the differentiation of immature B cells (7, 40, 46–48). An essential additional task of BMECs is the regulation of tissue regeneration and restoration of HSC homeostasis after extreme injury with ionizing radiation or chemotherapeutic agents, to which the fragile BM hematopoietic compartment is highly sensitive (29).

The mesenchymal compartment of the BM is not only engaged in hematopoietic regulation but also it contains a stem cell-based hierarchy required for bone development, homeostatic remodeling, and repair (49). The developmental roadmap by which multipotent mesenchymal stem cells (MSCs) give rise to chondrocytes, osteoblasts and adipocytes in situ within the BM is much less clearly defined than the hematopoietic hierarchy. Recent reports have provided new insight into the mesenchymal compartment through the use of novel genetically engineered mouse models to target and trace mesenchymal populations. These studies are gradually uncovering a rather complex scenario (50–52). The nuances of the system have been elegantly and thoroughly reviewed elsewhere and largely exceed the scope of this manuscript (6). For the purpose of this review, it is important to note that a general confusion has historically pervaded the field regarding the use of the term MSCs. Besides referring to the true bona fide skeletal stem cell and mesenchymal progenitors present in native BM tissues, the name MSC was indistinctively employed to designate the product of the in vitro clonogenic expansion of the adherent cell fraction isolated from different organs, which featured adipogenic, osteogenic, and chondrogenic potential in vitro (49, 53). The latter have been extensively characterized for their immunomodulatory properties in vitro and upon transplantation, which may be therapeutically exploited in the context of severe infectious and inflammatory challenge (54). However, in this manuscript, we will exclusively refer to those studies, which have addressed the effects of infections in MSCs or mesenchymal progenitor cells that reside in and can be prospectively isolated from BM tissues.

Mesenchymal stem cells are proposed to give rise in situ to a fibroblastic reticular cell network that densely populates BM and holds strong morphological and structural resemblance to stromal lattices found in other lymphoid organs, such as lymph nodes and spleen (23) (Figure 1). BM reticular cells, identified based on their expression of leptin receptor (Lepr), comprise the majority of progenitor cells that generate osteoblasts and adipocytes in steady state, as well as chondrocytes and reticular stroma in the context of bone repair (55–57). Fibroblastic stromal cells abundantly express factors that are absolutely indispensable for the hematopoietic function of the BM (55). For instance, they are the most prominent source of CXCL12 [thus designated as CXCL12-abundant reticular (CAR) cells] and SCF, through which they promote retention and maintenance of HSPCs in the BM (42–44). Beyond their regulation of early hematopoietic progenitors, at least a subset of Lepr⁺ reticular cells additionally secretes interleukin-7 (IL-7) and therefore contributes to different stages of lymphoid lineage development (47, 58, 59). Other less well-characterized subpopulations of mesenchymal origin have been described. These include periarteriolar cells defined in the NestinGFP mouse model by the highest expression of the transgene, and a subset of perivascular cells found mostly in trabecular bone regions, which express NG2 and PDGFRβ (33, 60).

Mature osteoblasts line the inner surfaces of bone and constitute one of the direct mature derivatives of mesenchymal reticular stromal cells (56, 61). Although their primary function is the production and secretion of mineralized matrix needed for continuous remodeling of bone surfaces, osteoblasts have been extensively studied for their regulatory crosstalk with the hematopoietic system. Initial studies pointed to a role of osteoblasts in providing a supportive HSC niche (62, 63). Nonetheless, recent evidence suggests that signals derived from the mature osteoblastic or endosteal niche are rather required for the maintenance of lymphoid progenitors and balanced T and B cell lymphopoiesis (43, 64–67). One of the alternative cellular fates of mesenchymal Lepr⁺ cells is differentiation into adipocytes (55, 56). Marrow adipogenesis is a highly complex phenomenon, the functional implications of which are only superficially understood. In general terms, adipocytes are regarded as negative regulators of hematopoiesis, which emerge mostly in the context of pronounced marrow injury and aging [(68); reviewed in Ref. (69)].

Cells of neural origin include glial GFAP⁺ non-myelinating Schwann cells (NMSCs) and adrenergic neurons, which are scarcely distributed throughout the BM interstitium (70, 71). NMSCs have been shown to participate in the maintenance of HSC quiescence through production of activated transforming growth factor-β (70). Adrenergic signals relayed by sympathetic nerves most likely play multiple unexplored and complex roles in BM hematopoietic function but have been studied for their control of the constant cellular trafficking in and out of the BM (71, 72). During homeostasis, HSCs are rhythmically released from BM into peripheral circulation in a pattern that is governed by the adrenergic-mediated control of CXCL12 expression levels (71). Likewise, continuous cell recruitment to the BM, which is dependent on cellular adhesive interactions with the microvasculature, oscillates following circadian cycles established by adrenergic signaling (72). Neural-derived signals have been additionally demonstrated to critically contribute to the complete and rapid structural regeneration of BM post-injury (73).

Akin to mature hematopoietic cells and HSPCs, certain BM stromal subsets have been shown to express specific receptors that render them susceptible to the direct action of microbial agents and inflammatory cytokines, as well as the molecular machinery needed to couple these signals into a defined intracellular activation program that alters their regulatory function. For instance, endothelial cells isolated from multiple organs, including the BM, consistently express TLR4 and its intracellular signaling protein, myeloid differentiation gene 88 (Myd88) (93). The integrity of this pathway in vascular cells is crucial for emergency granulopoiesis in the context of infections with Gram-negative bacteria. Using cell subset-specific deletion of Myd88 expression in mice, Boettcher et al. demonstrated that upon lipopolysaccharide (LPS)-mediated TLR4 stimulation, endothelial cells upregulate the principal granulopoietic cytokine, G-CSF to boost myeloid differentiation of hematopoietic progenitors (77). Although the specific contribution of BMECs with respect to those in vascular compartments of other organs was not determined, upregulation of G-CSF expression was most prominent in BMECs, suggesting the existence of at least a partial contribution of the local vascular compartment to overproduction of myeloid cells (90).

Albeit through different mechanisms, certain viral infections have been reported to similarly induce a skewing toward myeloid differentiation already at the level of HSPCs, through modulation of stromal-derived signals. Lymphocytic choriomeningitis virus (LCMV) is a natural murine pathogen, which has been widely employed in the discovery of general features of antiviral immune responses (94). During the early phases of acute LCMV infections, a myeloid HSPC fate is transiently enforced in order to accelerate production of myelomonocytic innate immune cells. The reactive shift in HSPC fate was shown to depend on the upregulation of IL-6 secretion by CAR cells, which in turn was triggered by IFNγ production by activated BM CD8 cytotoxic T cells (81). Of importance, previous studies had shown that acute LCMV infections induced transient BM aplasia, suppression of HSC function, and a pancytopenic state (95, 96). These effects were mediated, at least partially by the direct action of IFNα/β and IFNγ on hematopoietic cells. However, the possible contribution of stromal alterations in the context of LCMV infections had not been addressed in these studies. Notably, increased IL-6 production of CAR cells has also been observed upon infection with the rodent parasite Toxoplasma gondii (74). In this setting, induction of IL-6 not only augmented myeloid differentiation but also additionally elicited an anemic condition by stalling early stages of erythroid development in infected mice. Nevertheless, unlike LCMV infections, both IL-6 upregulation and anemia in this case were found to be independent of IFNγ, suggesting that multiple distinct mechanisms can operate to trigger local IL-6 secretion during different pathogenic inflammatory conditions (74).

Beyond altering the cytokine and growth factor milieu of the BM to instruct hematopoietic differentiation, infections or immunepathological reactions bring about (i) major changes in pro-migratory and retention cues provided by mesenchymal stromal cells and (ii) a breach in the barrier function and cell trafficking control exerted by BM vascular structures. Both effects critically contribute to the rapid mobilization of mature and immature hematopoietic cells from the BM to the periphery, which is a hallmark and a stereotypic host response to multiple pathogenic challenges (97–99). On the one hand, rapid deployment of mature cells, mostly of the myelomonocytic pool, increases the availability of circulating cells, which can be recruited to infected tissues for pathogen clearance (13, 14). In addition, mobilization of immature progenitors and HSCs leads to the transient establishment of extramedullary hematopoiesis (EMH) in other organs, mostly the spleen. EMH is a sign of dysfunction of the BM microenvironment, which is not only observed during infections but also in multiple stress-related conditions, and it is hypothesized to temporarily preserve early hematopoietic progenitors in a facultative supportive microenvironment (100, 101). However, a minor fraction of circulating HSPCs is found to traffic to peripheral tissues and differentiate in situ into immune cells, thereby contributing to local infectious control (102).

The mechanisms driving massive cellular exit from BM during infections are incompletely understood and will likely differ between the hematopoietic cell types mobilized and the type of pathogen. A general mechanism involves the interference of the CXCR4/CXCL12 axis, which is the major molecular pathway involved in BM retention described thus far. Expression and protein levels of CXCL12 in the BM are substantially reduced during certain bacterial infections, thus leading to release of HSPCs to peripheral circulation and lodging in spleen (76, 103). Given that CAR cells are the main source of CXCL12 in the BM and prominently express other adhesion molecules such as VCAM-1, it is highly likely that cell mobilization at least partially results from the erosion of the pro-retention function of this cell type. Supporting this notion, a recent study demonstrated that HSPC mobilization during Escherichia coli infection was mediated by increased G-CSF production, which has been shown to specifically reduce CXCL12 expression in CAR cells (104). Interestingly, in this infectious model, G-CSF secretion is triggered by the combined activation of TLR4 and NOD receptors in an undefined non-hematopoietic cell type, most likely endothelial as suggested by previous studies (76, 90). The attenuation of CXCL12-mediated retention signals does not only induce release of HSPCs from BM but also is hypothesized to contribute to rapid neutrophil egress during infections (105).

Besides this general mechanism, CAR cells additionally participate in BM cellular egress during infections in an active manner. Upon sensing of low levels of circulating LPS, BM perisinusoidal CAR cells upregulate expression of the pro-migratory chemokine CCL2, which is essential for emigration of CCR2⁺ inflammatory monocytes across sinusoidal vessels into systemic circulation (78). The resulting increase in frequencies of circulating monocytes was shown to be critical for efficient clearance of pathogen clearance of Listeria monocytogenes infections.

As observed for vascular and mesenchymal scaffolds in other lymphoid organs, pathogenic insults have been recently reported to selectively impact BM stromal cell survival (23, 106). Resulting cellular lesions can lead to imbalance in stromal composition and/or structural defects, which in turn have sizeable and distinct effects in the hematopoietic system. Deleterious actions in stroma may be induced by sharp increases in local pro-inflammatory cytokine levels. Using a murine model of sepsis induced by cecal ligation and puncture, Terashima and colleagues recently proved that mature, bone-lining osteoblast populations are strongly decreased during polymicrobial infection (75). Partial depletion of the osteoblast pool resulted in a paucity of IL-7 production in the BM microenvironment and consequently, a major loss of common lymphoid progenitors, which compromised T and B lymphopoiesis. In this case, the observed reduction in osteoblast numbers was independent of TLR signaling, but again at least partially mediated by abnormally high G-SCF levels (75). These observations are well in line with the previously reported negative effect of exogenously administered G-CSF on osteoblastic populations (104). Remarkably, infection-induced depletion of osteoblasts and its consequences could be mitigated through activation of osteoblastic proliferation using parathyroid hormone (75). Although, it should be cautioned that an IL-7-expressing niche in human BM tissues has not been described to date, these findings are highly relevant as they may contribute to mechanistically explain and potentially prevent the persistent lymphopenic state that frequently leads to fatal secondary infections in septic patients. Notably, an independent study using a similar model of polymicrobial systemic infection further revealed conspicuous signs of mesenchymal stromal cell damage in BM, which included a partial loss of structural integrity and the abundant emergence of adipocytes, indicative of the induction of aberrant differentiation programs in mesenchymal progenitor cells (107).

Bone marrow stromal cell subsets are also sensitive to the direct action of pathogenic virulence factors. For instance, administration of the bacterial second messenger bis-(3′–5′)-cyclic dimeric guanosine (c-di-GMP) was recently shown to strongly reduce endothelial and CAR cell numbers in BM and compromise their ability to produce relevant HSPC niche factors, such as CXCL12 or kit-L (80). Consequently, HSPCs were mobilized to the periphery, subsequently homing and proliferating in the spleen. c-di-GMP is known to trigger immune responses in a variety of cell types through the activation of the endogenous endoplasmic receptor stimulator of IFN genes (STING). Of importance, multiple cell types, including HSPCs and immune cells, express STING receptors. Hence, it cannot be ruled out that the damaging effect of c-di-GMP is at least to some extent mediated indirectly, through activation of inflammatory cascades in other cell types.

Finally, evidence suggests that stromal populations can be directly infected by certain intracellular pathogens this potentially leading to stromal cell death or dysfunction. Such is the case for viral agents, such as cytomegalovirus (CMV) and human immunodeficiency virus (HIV), which infect BM mesenchymal stromal cells and suppress their hematopoietic supportive capacity in vitro (see discussion below). In addition, it was recently reported that Mycobacterium tuberculosis displays specific tropism for the mesenchymal compartment of the BM, which acts as a reservoir for pathogen persistence and dissemination (79). However, from this study, it was unclear whether long-term latency in mesenchymal stromal cells had an impact in the bone forming or hematopoietic regulatory functions of MSCs.

While much work has focused in understanding how stromal cells influence the BM hematopoietic compartment, regulatory crosstalk is most likely bidirectional and significantly less is known on the potential mechanisms by which hematopoietic cells may shape their microenvironment. Understanding such reciprocal modulation may become especially crucial during inflammatory stress situations in which the hematopoietic content of the BM undergoes drastic qualitative and quantitative alterations due to massive egress and entry of immune cell subsets, as well as biased HSPC lineage differentiation.

In steady state, myeloid cells have been reported to regulate CAR cell functionality at least via two different mechanisms. BM macrophages provide cues that ensure production of sufficient levels of CXCL12 by CAR cells both in vitro and in vivo, through a yet unknown mechanism (91). Depletion of CD169⁺ macrophage populations consequently results in abnormally low CXCL12 levels and defective HSPC retention in BM. Macrophages have been further reported to play a completely unanticipated role in the indirect modulation of HSPC mobilization via a complex mechanism involving CAR cells, as well as neutrophils. Macrophages are responsible of the clearance of aged, dysfunctional circulating neutrophils, which takes place among other organs in the BM (108). For their elimination, as neutrophils age they are recruited to the BM parenchyma in waves, which follow circadian fluctuations and are thus synchronized by sympathetic nervous signals. The time window of maximal BM neutrophil entry within the regular light–dark cycles was observed to precede that of most active HSPC egress to circulation. Mechanistically, it was demonstrated that BM macrophages engulfing aged neutrophils are activated through cholesterol receptors and relay a signal to CAR cells that temporarily downmodulates their CXCL12 expression (92). As a consequence, waves of HSPC mobilization toward the blood are rhythmically generated. Collectively, the findings described above strongly suggest that the changing dynamics of BM-resident myelomonocytic cell subsets in the context of infections could potentially affect BM stromal function, a hypothesis that deserves further in-depth studies. On the basis of the evidence discussed above, it seems clear that the BM microenvironment and each of its individual components represent highly susceptible substrates for the indirect action of pathogenic stimuli in the hematopoietic system.