Hematopoiesis is the process of blood cell development that in vertebrates is initiated early during embryogenesis and may be divided into 3 phases or so-called distinct waves of hematopoiesis.

The first (or primitive) wave takes place in the yolk sac starting in mice at embryonic day 7.5 (E7.5) and generates unipotent blood cell types (1). The second (or pro-definitive) wave occurs in the yolk sac, embryo proper and allantois of the mouse embryo and gives rise to multipotent progenitors (2). The third wave of hematopoiesis represents definitive hematopoiesis and is dependent on the activity of hematopoietic stem cells (HSCs), which are the basic units of the adult hematopoietic system. HSCs generated in the embryonic aorta-gonad-mesonephros region first colonize the fetal liver (E10.5) and then shortly before birth (E16) migrate to the bone marrow (BM), where the majority of HSCs reside to sustain steady state hematopoiesis (3).

HSCs are multipotent, self-renewing cells capable of differentiating into all mature blood cell lineages over the lifespan of the animal. Lineage choice may be directed both intrinsically and extrinsically via activation of transcription factors or extrinsically by cytokines (4). The majority of HSCs are quiescent under steady-state conditions, and few HSCs cycle to sustain hematopoiesis (5). In order to maintain hematopoietic homeostasis and to prevent development of malignancies, the self-renewal and differentiation capacities of HSCs are tightly regulated. This is, at least partly, achieved by the specialized network of interactions between distinct cell types (6, 7) and secreted factors (8, 9) in the BM niche that maintains HSC activity in steady-state conditions.

However, in the case of systemic infections and pathological conditions, such as myeloablation after chemo- or radiotherapy, some HSCs may respond and exit their quiescent state. These ‘activated’ HSCs contribute to the pool of hematopoietic progenitor cells, which will undergo further differentiation in order to replenish the population of immune cells being in high demand at the sites of inflammation in the process of so-called emergency hematopoiesis. This is possible because hematopoietic stem and progenitor cells (HSPCs) express Toll-like receptors (10) and cytokine receptors (11) and, thus, can respond to inflammatory signals. Activation of TLR signaling in HSPCs not only drives myeloid cell differentiation (10), but also leads to production of cytokines, which regulate myeloid differentiation and HSPC proliferation (12). HSPCs may respond to cytokines released during inflammation either systemically or locally by cells in the hematopoietic microenvironment or BM niche. Indeed, it was shown that HSPCs express various cytokine receptors, including IL-1R (13), IL-6Rα, as well as both TNF receptors, TNFR1 and TNFR2 (12).

As mentioned above, proinflammatory cytokines are critical components of inflammation-induced myelopoiesis. However, inflammatory signals may also be implicated in the maintenance of homeostatic hematopoiesis. During embryonic development proinflammatory cytokines control HSPC specification in the pro-definitive wave of hematopoiesis (14). Moreover, proinflammatory cytokines may regulate adult hematopoiesis and maintain the balance between HSC dormancy and lineage commitment (15). This question is important because systemic and long-term anti-cytokine therapy is being applied to treat an increasing number of conditions, including autoimmune disorders. One of the proinflammatory cytokines implicated in hematopoiesis is TNF, which we discuss in the context of humanized mouse models in this Perspective.

TNF is a pleiotropic cytokine involved in inflammation, development of secondary lymphoid organs and immune regulation. TNF is produced as a transmembrane protein and can be proteolytically cleaved into a soluble form. TNF exerts its functions via two distinct receptors – TNFR1 and TNFR2 (16). Both receptors also may interact with soluble (LTα₃) and membrane-bound (LTα₂β₁) lymphotoxins, respectively (17). Interestingly, TNF may play a role both in embryonic and in adult hematopoiesis ( Figure 1 ).

The role of TNF during fetal hematopoiesis was mostly studied in zebrafish. These studies revealed that TNF derived from primitive neutrophils binds to TNFR2 on endothelial cells resulting in the upregulation of Notch ligand Jagged1a, which in turn binds to Notch1a receptor on the neighboring hemogenic endothelium triggering HSC fate specification. Moreover, TNF/TNFR2 axis also activates canonical NF-κB pathway in hemogenic endothelium, which triggers a transcriptional program to establish HSC generation (18). Of note, not only TNF but also other inflammatory stimuli such as TLR4-MyD88 signaling or G-CSF may lead to NF-κB activation required for HSC development (29). Interestingly, inflammatory signaling represents a highly conserved pathway regulating the HSC development. Studies in E9.5 mouse embryos revealed that hematopoietic cluster cells and endothelial cells respond to IFNγ and to a lesser extent to TNF stimulation. The most likely source of TNF in the mouse embryo is the population of primitive F4/80⁺ macrophages, similar to the situation in zebrafish embryo (30). However, the precise contribution of TNF to mammalian HSC development is not completely understood. Taken together, TNF signaling may be required for HSC emergence in the developing embryo via activation of evolutionarily conserved signaling pathways, but it might be partially redundant with other inflammatory stimuli.

Many studies have been performed to examine the role of TNF in the adult hematopoiesis; however, most of these have relied on cell culture and/or bone marrow chimeras, which could indirectly affect HSPC phenotype and functions. Another challenge in understanding the role of TNF in steady-state conditions is due to its capacity to induce systemic inflammation when administered in vivo which in turn may activate stress-induced hematopoiesis thus obscuring TNF contribution to hematopoiesis. Therefore, in this section we will focus on in vivo studies using gene-deficient mice. TNF deficiency did not alter the number of Lin^-Sca-1⁺c-kit⁺ (LSK) in the BM, consisting mostly of lineage-biased multipotent progenitors (19). In early studies TNFR1-deficient bone marrow was characterized by increased number of LSK (22). However, examination of purified LSKFLT3^- HSPCs revealed no differences in the numbers of HSPC isolated from the adult BM of TNFR1/TNFR2 double knockout mice (31). Hence, it is likely that TNF does not affect HSPC compartment under steady-state conditions in vivo.

Regarding the differentiation of HSCs to more committed progenitors, TNFR1-deficient mice showed a decrease in pre-B cell compartment and an increase in myeloid progenitors (32). Accordingly, TNF-deficient mice demonstrated an increase in the number of Gr-1⁺ neutrophils both in the BM and in peripheral blood (19). Transcriptome analysis of monocytes and their BM precursors revealed an increase in TNF expression upon differentiation of Ly6C^hi/Ly6C^int monocytes into Ly6C^lo monocytes (20). Therefore, TNF may control granulocyte number in the blood and BM and support monocytic differentiation in vivo.

Numerous studies on the role of TNF in HSPC functions relied on in vitro and in vivo colony formation assays together with the assessment of reconstitution potential, engraftment and survival abilities of multipotent progenitors upon transplantation into irradiated recipient mice. However, results obtained from these studies should be carefully interpreted since these setups may affect HSPC proliferation, survival, self-renewal and differentiation. We will discuss some experiments and their possible applicability to hematopoietic compartment.

Studies with competitive co-transplantation of TNFR1^-/-TNFR2^-/- CD45.2⁺ and wild-type (WT) CD45.1⁺ BM cells into lethally irradiated congenic CD45.1⁺CD45.2⁺ WT recipients showed enhanced activity of TNFR1^-/-TNFR2^-/- HSCs as determined by long-term reconstitution by TNFR-deficient HSCs following transplantation. Moreover, in vivo administration of TNF to WT mice led to a decrease in BM cellularity and to reduction in HSC cycling activity in a competitive transplantation assay (31). These data suggest that TNF blockade may be beneficial for post-transplantation reconstitution and also supports the idea that TNF may suppress HSC activity. In contrast, transplantation of bone marrow cells from 6 months old TNFR1-deficient mice into lethally irradiated recipients showed reduced repopulating ability of TNFR1^-/- BM cells as compared to WT cells (32). However, this effect was shown on non-purified HSPCs and under long-term transplantation conditions that may affect the outcome of the experiment. TNF does not inhibit expansion of highly purified Lin^-Sca-1⁺c-kit⁺Flk2^-CD150⁺CD48^- HSCs in medium supplemented with IL-3, SCF, TPO, EPO, GM-CSF, IL-11 and Flt3-L, or so-called cytokine-rich medium, regardless of TNF concentration. Under cytokine-poor conditions (medium supplemented with SCF and G-CSF only) HSC expansion rate upon addition of TNF was significantly decreased (21). Similar findings were reported by Pronk et al. when TNF was added to the culture of LSK cells (31). Importantly, studies by Yamashita et al. revealed that this effect was due to the inhibition of autophagy by TNF, which sensitized HSCs to cell death in cytokine-deprived environment (21). Taken together, HSCs appear to be resistant to TNF cytotoxicity but this resistance may be changed by the environmental stress.

Addition of exogenous TNF to bone marrow cultures of LSK resulted in reduced numbers of large-sized colonies, such as CFU-GEMMs (colony-forming unit − granulocyte, erythroid, macrophage, megakaryocyte) and CFU-GMs (granulocyte/macrophage colony forming units), in methylcellulose assay (22). LSK isolated from TNF-deficient mice gave rise to an increased number of splenic colony-forming units in lethally irradiated recipient mice (19). Experiments with TNFR1 and TNFR2 agonists demonstrated that TNFR2 is essential for TNF-mediated inhibition of colony formation of early Lin^-Sca-1⁺ progenitors, while addition of TNFR1 agonist had no significant effect on the number of Lin^-Sca-1⁺ proliferative clones (23). Thus, TNF may inhibit formation of CFU-GEMM and CFU-GM colonies and differentiation potential of multipotent progenitors, presumably via TNFR2.

Evaluation of TNF effects on myeloid lineage commitment revealed that addition of TNF or TNFR1 agonists resulted in a decrease in G-CSF-induced colony formation and in inhibition of G-CSF expression by BM cultures (23). This was further supported in vitro in TNF-deficient long-term bone marrow cultures, which were characterized by increased proliferative potential of granulocytic progenitors and increased numbers of CFU-GMs within sorted LSK population, as compared to WT BM cultures (19). Consistent with in vitro data, CFU-GM formation was reduced in WT-recipient mice reconstituted with TNF-deficient BM, suggesting a possible role of TNF, expressed by hematopoietic and not by stromal cells, in inducing cell death at the GMP stage (33). This is also in agreement with data showing elimination of GMPs by TNF in a dose-dependent manner (21). Next, TNF was demonstrated to block granulocytic differentiation and IL-3-dependent proliferation of granulocyte-committed cells (24). At the same time, an autonomous effect of TNF on monocytes was proposed. Specifically, TNF is required for monocyte survival in vivo (20). Thus, TNF stands at the crossroad of myeloid lineage commitment via negatively affecting granulocytic cell differentiation and driving the survival of monocytes. Interestingly, these findings suggest possibilities to control the differentiation of myeloid progenitors. For example, TNF was shown to directly upregulate a central transcription factor for myeloid lineage commitment, PU.1, in HSPC in vivo during acute inflammation (34). Moreover, in some abnormal hematopoiesis conditions, such as clonal hematopoiesis associated with aging, TNF blockade may help to overcome TET2-mutant HSPC skewing toward myeloid lineage and the formation of CFU-GMs (35).

Other studies revealed that addition of TNF to the LSK cultures negatively regulated both long-term and short-term reconstituting activity after transplantation of LSK into lethally irradiated recipient mice (36). However, TNF produced by BM microenvironment is required for long-term engraftment and survival of purified LSK in allogeneic and syngeneic recipients (37). This is in line with the fact that the engraftment of Lin^- BM cells from TNFR1- and TNFRs-deficient donors in wild-type recipients was defective suggesting a stimulatory role of TNF in successful progenitor engraftment. Interestingly, homing of the engrafted progenitors to the BM was primarily mediated by TNFR1 (25). On the other hand, total body irradiation and inflammation within BM was associated with elevated levels of TNF. Subsequently, TNF induced ROS accumulation in LSK leading to impairment of their reconstitution ability. Addition of TNFR1 antagonistic peptide to LSK cultures inhibited ROS accumulation suggesting that TNFR1 blockade prior to transplantation may lead to improved reconstitution capability (26). Altogether, more precise in vivo studies with defined protocols of total body irradiation and transplantation are needed to identify the role of TNF in the engraftment and survival of HSPCs. For example, TNF may contribute to regeneration of BM niche after HSC transplantation, since TNF-deficient mice displayed reduced number of BM endothelial cells upon myeloablation with a single injection of 5-fluorouracil. Furthermore, Gr1⁺CD115^- granulocyte-derived TNF promoted vascular regeneration following transplantation (27). In the context of hematological malignancies, such as myeloproliferation, inflammation may be implicated in the disruption of BM microenvironment. Flt3 ^ITD/ITD mice, harboring the most common somatic mutation in patients with acute myeloid leukemia, were shown to upregulate TNF production by endothelial cells in the BM niche, which subsequently may lead to suppression of HSC activity. Treatment of Flt3 ^+/+ and Flt3 ^ITD/ITD mice with Etanercept resulted in partial rescue of LSKCD150⁺CD48^- engraftment capacities (38).

As a consequence of inflammation, TNF may mobilize B cell progenitors from BM to peripheral tissues by suppressing their CXCL12-induced retention in the BM. This mobilization establishes extramedullary lymphopoiesis possibly needed for resolution of inflammation (28). Contribution of TNF to emergency myelopoiesis was clearly demonstrated by Yamashita et al. (21). In a model of poly(I:C)-induced inflammation, TNF, on one hand, induced NF-kB activity, which protected HSPCs from inflammation-induced necroptosis and, on the other hand, promoted myelopoiesis and induced apoptosis of GMPs (21). This effect may be protective in pathogenesis of some viral infections, in which case GMPs may act as the latent reservoirs for viruses in the BM and should be eliminated (39).

Defects in hematopoiesis may lead to the development of hematologic disorders. Although the exact role of TNF in different hematological diseases remains incompletely understood, elevated levels of TNF were found in patients suffering from myeloid leukemia (40) and myelodysplastic syndromes (41, 42), Fanconi anemia (43), Hodgkin’s disease (44) and Non-Hodgkin lymphoma (45).

Moreover, since TNF inhibitors are widely used to treat autoimmune disorders, it is important to address possible side effects of anti-TNF therapy. Indeed, hematological complications were reported in patients on TNF blockade ( Table 1 ). For example, a case report was published showing pancytopenia after treatment with Infliximab (47). Furthermore, side effects of Etanercept [which also neutralizes both LTα₃ and LTα₂β₁ (17)] treatment on hematopoietic cells have been reported including aplastic anemia (52), thrombocytopenia (55), and bone marrow aplasia with pancytopenia (64). Interestingly, a common phenomenon in patients receiving anti-TNF therapy is the development of neutropenia (81). Apart from that, some lymphoproliferative disorders were reported. For example, treatment of rheumatoid arthritis and inflammatory bowel disease (IBD) patients with Etanercept or Infliximab, respectively, led to formation of cutaneous and systemic T-cell lymphomas (75). A case of Hodgkin-type lymphoproliferative lesions was reported for an IBD patient treated with Infliximab for a 6-month period (79).

Altogether, TNF may play a crucial role in inflammation-induced hematopoiesis and may be implicated in the pathogenesis of some hematologic disorders. Anti-TNF therapy may lead to rare but severe side effects affecting hematopoietic compartment and resulting in the development of hematological complications and even malignancies. The exact mechanisms of these side effects are not well understood and should be addressed in the future using humanized mouse models.

To identify the effects and to evaluate the efficacy and safety of clinically available or novel human TNF (hTNF) inhibitors proper animal models should be generated and validated. Importantly, in spite of a conservative nature of TNF family of cytokines and their corresponding genes, most hTNF inhibitors do not block mouse TNF (82). Therefore, various panels of humanized mice that express hTNF and/or TNFRs are required to facilitate the research ( Table 2 ). Such preclinical models were first generated in 1991 by G. Kollias group, when the first mice with overexpression of human TNF were reported (83).

For many years TNF was mainly considered a proinflammatory cytokine with its role in host defense, but also with detrimental effects on autoimmunity. However, basic studies on TNF biology, as well as reported side effects in patients receiving anti-TNF therapy, highlighted its homeostatic functions in many physiological processes, including hematopoiesis. Regulation of TNF/TNFRs expression in various tissues modulates the cross-talk between immune and non-immune cells, which subsequently determines the outcome of TNF action.

Future investigation of pathological versus regulatory functions of TNF and deciphering its systemic and local effects in tissues may help to improve current therapeutic approaches. Therefore, mouse models with humanized TNF/TNFRs system represent a powerful tool to study side effects of anti-TNF therapy on hematopoiesis.

VG, KA, MD, and SAN designed research. VG and KA. performed research and analyzed data. VG, KA, AD, TY, MSD, and SN discussed the concept and wrote the manuscript. All authors contributed to the article and approved the submitted version.

Reverse genetics studies were supported by the grant 075-15-2019-1660 from the Ministry of Science and Higher Education of the Russian Federation, myelopoiesis studies were supported by RFBR grant 19-34-51030.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.