M‐CSF directs myeloid and NK cell differentiation to protect from CMV after hematopoietic cell transplantation

Abstract Therapies reconstituting autologous antiviral immunocompetence may represent an important prophylaxis and treatment for immunosuppressed individuals. Following hematopoietic cell transplantation (HCT), patients are susceptible to Herpesviridae including cytomegalovirus (CMV). We show in a murine model of HCT that macrophage colony‐stimulating factor (M‐CSF) promoted rapid antiviral activity and protection from viremia caused by murine CMV. M‐CSF given at transplantation stimulated sequential myeloid and natural killer (NK) cell differentiation culminating in increased NK cell numbers, production of granzyme B and interferon‐γ. This depended upon M‐CSF‐induced myelopoiesis leading to IL15Rα‐mediated presentation of IL‐15 on monocytes, augmented by type I interferons from plasmacytoid dendritic cells. Demonstrating relevance to human HCT, M‐CSF induced myelomonocytic IL15Rα expression and numbers of functional NK cells in G‐CSF‐mobilized hematopoietic stem and progenitor cells. Together, M‐CSF‐induced myelopoiesis triggered an integrated differentiation of myeloid and NK cells to protect HCT recipients from CMV. Thus, our results identify a rationale for the therapeutic use of M‐CSF to rapidly reconstitute antiviral activity in immunocompromised individuals, which may provide a general paradigm to boost innate antiviral immunocompetence using host‐directed therapies.


Introduction
The first months after hematopoietic cell transplantation (HCT) are characterized by profound immunosuppression, which leaves patients at high risk of viral infection or reactivation of common opportunistic viruses such as cytomegalovirus (CMV).The infection itself but also its subsequent treatment is associated with significant morbidity and mortality (Arber et al, 2003;Boeckh & Ljungman, 2009;Ahmed, 2011;El Chaer et al, 2016;Cho et al, 2019).Although vaccines against CMV are under development, they are not yet routinely available in the clinic (Plotkin, 2015).Moreover, antiviral treatments based on inhibition of viral replication are limited to specific viruses, can have significant bone marrow toxicity, and run the risk of variant development and breakthrough infections (Boeckh & Ljungman, 2009;Ahmed, 2011;El Chaer et al, 2016;Cho et al, 2019;Hill et al, 2021).Cell-based therapies are still not widely on hand and associated with high costs (Kaeuferle et al, 2019).Biologics stimulating the patient's general antiviral immune response could therefore be a welcome alternative or complementary treatment option but are currently unavailable.
Cytomegalovirus can lead to a diverse range of pathologies in immunocompromised humans (Griffiths et al, 2015;Griffiths & Reeves, 2021), and the closely related murine CMV (MCMV) has similar cellular tropism and kinetics (Krmpotic et al, 2003;Alexandre et al, 2014).The spleen is an early site for filtering blood-borne virus and initiating immune responses, whereas the liver is a principal site of viral infection after its decline in the spleen (Hsu et al, 2009).Type I interferons (I-IFNs) (Baranek et al, 2012), produced by plasmacytoid dendritic cells (pDCs) (Dalod et al, 2002;Zucchini et al, 2008), constitute a first line of defense against CMV with natural killer (NK) cells and cytotoxic T cells coming in as a critical second and third wave of the immune response that block viral replication by killing infected cells (Orange & Biron, 1996).Cytokines including IL-12 and IL-15 produced by conventional dendritic cells (cDCs) can indirectly contribute to viral defense by stimulating NK cell proliferation, activation, and effector function (Nguyen et al, 2002;Dalod et al, 2003;Baranek et al, 2012;Puttur et al, 2016).Other myeloid cells have been shown to have indirect and diverse roles in the response to CMV infection.Whereas Ly6C À CX3CR1 + patrolling monocytes act as carriers of CMV and can disseminate viral infection to distant organs throughout the body (Daley-Bauer et al, 2014), Ly6C + CCR2 + inflammatory monocytes can activate NK and cytotoxic memory CD8 + T cells during microbial infection, including MCMV (Soudja et al, 2012;Rovis et al, 2016).Culture models proved that the ability of macrophages to resist MCMV infection depends on signaling mechanisms via I-IFNs and type II IFNs (II-IFNs) (Presti et al, 2001;Strobl et al, 2005;Kropp et al, 2011), which might also be important in vivo.Myeloid-specific deletion of signal transducer and activator of transcription (STAT)1, a key transcription factor for mounting IFN responses, is also required for the early control of MCMV infection and spleen pathology but does not affect viral clearance (Gawish et al, 2019).Hence, the role of myeloid cells in MCMV infection appears multifaceted and complex.
Interestingly, in the myeloid STAT1 deletion model, the ability to combat early MCMV infection correlated with the ability to mount extramedullary hematopoiesis (Gawish et al, 2019).In this study we have specifically investigated the role of emergency hematopoiesis on MCMV infection under leukopenic conditions and report that M-CSF-induced myelopoiesis promotes rapid reconstitution of antiviral activity and protection from infection.Using a murine model of HCT and infection with lethal doses of MCMV, we observed that M-CSF treatment prompted antiviral immunity resulting in substantially improved survival and pathogen clearance in mice.Dissecting the mechanism underlying this M-CSF-mediated protection against MCMV infection, we identified a multistep differentiation program in which M-CSF-induced myelopoiesis further stimulated NK cell differentiation and activation via IL-15 and I-IFN mediators.Lastly, we observed that M-CSF also induced myelo-monocytic differentiation from human G-CSF-mobilized HSPCs, enhanced IL15Ra expression on monocytes, and increased functional NK cells numbers.

M-CSF protects HCT recipients from CMV viremia and mortality
Cytomegalovirus infection/reactivation remains a perilous threat during immunosuppression (Arber et al, 2003;Locatelli et al, 2016;Ljungman et al, 2017).MCMV is a natural pathogen in mice that recapitulates patho-mechanisms of human CMV infection (Krmpotic et al, 2003).To study the antiviral effects of M-CSF on MCMV under leukopenic conditions, we used a murine HCT model (Kandalla et al, 2016)

M-CSF treatment increases NK cell abundance, differentiation, and activation
Since NK cells are early antiviral effector cells, including during HCT (Ullah et al, 2016), we investigated whether M-CSF treatment influenced NK cells.We observed an increase in NK cell numbers in the spleen two weeks after M-CSF treatment both in uninfected mice and after infection (Fig 2A).Separate analysis of CD45.2 + recipient and CD45.1 + graft donor cells revealed that most of the NK cell increase arose from donor cells (Fig EV2A).Since M-CSF is short- lived (Koths, 1997), but increased NK cell numbers two weeks after application, we supposed the mechanism to act on NK cell progenitors.NK cell differentiation stages can be identified by differential expression of surface markers and transcription factors (Fig 2B) (Huntington et al, 2013;Vosshenrich & Di Santo, 2013;Serafini et al, 2015).NK cell progenitors express CD122, CD27, and NKG2D but not the mature markers NK1.1 and NKp46.We observed that M-CSF increased the number of donor-derived CD122 + CD27 + NK cell progenitors both in uninfected and in infected mice (Fig 2C).Consequently, we analyzed the NK cell maturation and differentiation status, which can be distinguished into CD11b À CD27 + immature, CD11b + CD27 + mature M1 and CD11b + CD27 À mature M2 NK cells (Fig 2B) (Kim et al, 2002;Hayakawa & Smyth, 2006;Chiossone et al, 2009).M-CSF treatment increased both donor-derived immature and mature M1 and M2 NK cells, particularly in infected mice (Fig 2D).A smaller increase of progenitor and mature cells was also observed for resident host NK cells (Fig EV2B).
This was further confirmed by gene expression analysis of stagespecific transcription factors (Fig 2B).Fourteen days after M-CSFsupported HCT and after an additional 1.5 days of MCMV or mock infection, spleen NK1.1 + cells showed increased expression of the immature NK cell transcription factors Ikaros, Id2, Runx3, Gata3, and Tbet as well as the mature NK cell transcription factor Eomes after exposure to MCMV (Fig 2E).Similar observations were made for host-derived NK cells (Fig EV2C).Whereas Ikaros and Gata3 were more strongly induced by M-CSF in uninfected mice, Tbet and Eomes were preferentially induced after infection (Fig 2E).Importantly, infection alone was insufficient for the observed inductions.
Together, this indicated that M-CSF leads to an increased number of NK cell progenitors and enhanced their differentiation along the NK cell lineage trajectory.

NK cells execute M-CSF-stimulated antiviral immunity
The major antiviral activity of NK cells is mediated by the production of inflammatory cytokines like IFNc, and perforin (PFR1)dependent delivery of granzyme B (GrB) into infected cells (Bukowski et al, 1984).Interestingly, M-CSF treatment increased the number of IFNc-(Fig 3A ) and  in infected mice in concert with enhanced mRNA levels for IFNG, GZMB and PRF1 as well as maturation and activation genes (CEBPA,MITF and XCL1;Figs 3C and EV2D).Consistently, M-CSF induced NK cell accumulation at infectious foci within the liver early after infection culminating in reduced numbers of MCMV-infected cells (Fig 3D).To determine whether antiviral NK cell activity was required for the protective effect of M-CSF, we depleted NK cells using anti-NK1.1 antibodies in M-CSF-treated and MCMV-infected HCT recipients (Fig 3E).NK cell depletion nearly abolished the increased survival of M-CSF-treated mice, demonstrating that a significant part of the protective effect of M-CSF against viral lethality depended on NK cells.

M-CSF-induced myelopoiesis is required for its antiviral effect
Since M-CSF has not been reported to act directly on the NK cell lineage, we investigated whether M-CSF's effects on the myeloid ◀ Figure 1.M-CSF protects HSPC recipients from CMV viremia and mortality.
A Leukopenia model to study MCMV viremia.B Survival of mice after HSPC transplantation (arrow), MCMV infection (stippled arrow) and treatment (arrowheads) with control PBS (ÀM-CSF; n = 7) or three doses of 10 lg mouse recombinant M-CSF (+M-CSF; n = 6).Transplanted, uninfected mice (n = 5) are shown as control.C Histopathology of MCMV-induced hepatitis.Assessment of inflammatory foci 4 days after infection of transplanted mice treated with M-CSF or control PBS.Example of hematoxylin and eosin (H&E)-staining and inflammatory foci (n = 4) (scale bar = 100 lm).D Histopathology of MCMV-induced hepatitis.Apoptotic (arrowheads) and necrotic (arrows) hepatocytes and quantification as median cell numbers per area (n = 4) (scale bar = 30 lm).E Histopathology of MCMV-induced hepatitis.Assessment of necrotic area 8 days after infection of transplanted mice (H&E).Percentage of affected areas (n = 4) (scale bar = 100 lm).F Histological analysis of infected hepatocytes (H&E); quantification per area (n = 4) (scale bar = 30 lm).G RT-qPCR-based quantitation of viral mRNA per 200 ng RNA (n = 5).A FACS examples and median of absolute number of total NK cells (CD19 À CD3 À Ly6G À NK1.1 + ) are shown (n = 5 mice per group, one independent experiment is shown but was confirmed twice).B Markers specific to differentiation and maturation stages of NK cells used in this analysis are indicated.C Median of absolute number of donor-derived NK progenitor cells (CD122 + CD27 + NK1.1 À Nkp46 À CD45.1 + ) are displayed (n = 5 mice per group, one independent experi- ment is shown but was confirmed twice).D FACS examples and median of absolute numbers of donor-derived immature NK cells, donor-derived M1 (CD11b + CD27 + ) and M2 NK cells (CD11b + CD27 À ) are shown (n = 5 mice per group, one independent experiment is shown but was confirmed twice).E Gene expression analysis of transcription factors expressed by NK cells in FACS-sorted, donor-derived NK1.

M-CSF drives myeloid IL-15 trans-presentation to promote antiviral competence
Since myelopoiesis and NK cell differentiation were required for the antiviral effect of M-CSF, we hypothesized that M-CSF-induced myelopoiesis could indirectly affect NK cell differentiation and antiviral activity.Indeed, anti-CD115-mediated depletion of myeloid cells resulted in reduced immature and mature NK cells (Fig 5A).To identify myeloid signals that could affect NK cells, we first focused on IL-15, a cytokine paramount for NK cell differentiation and effector functions (Nguyen et al, 2002;Budagian et al, 2006;Huntington et al, 2009;Boudreau et al, 2011).IL-15 can be produced and transpresented by IL15Ra on myeloid cells (Lucas et al, 2007;Castillo et al, 2009;Huntington et al, 2009;Patidar et al, 2016) including during MCMV infection (Fehniger et al, 2007;Baranek et al, 2012;Ghilas et al, 2021).M-CSF treatment resulted in swiftly increased IL-15 mRNA levels in spleens after MCMV infection (Fig 5B Next, we analyzed the effect of increased IL-15 signaling from Ly6C hi monocytes on the expression of IL-15 response genes in NK target cells.Like IL-15 signaling, which is engaged once the IL-15/ L15Ra complex binds to IL15Rb on target cells (Huntington et al, 2009;Baranek et al, 2012;Patidar et al, 2016), M-CSF treatment increased expression of the downstream genes IL15RB and of STATB5, JAK3, and E2F1-6 in NK cells (Fig 5E).To check whether IL-15-dependent myeloid cell to NK cell signaling was important for antiviral activity protecting HCT recipients from lethal MCMV infection, we compared IL15RA-KO GMPs incapable of trans-presenting IL-15 with WT GMPs.We observed 80% survival in WT GMPtransplanted mice after infection but no survival of IL15RA-KO GMP-transplanted mice (50,000 GMPs for each genotype), demonstrating that IL-15 signaling from myeloid cells was required for NK cell support (Fig 5F).Furthermore, M-CSF treatment resulted in no survival advantage in IL15RA-KO HCT recipient mice and was comparable to untreated WT HCT mice (Fig 5G), indicating that IL-15 signaling was acting downstream of M-CSF.Together, our data demonstrate that myeloid-derived IL-15 signaling was required for the antiviral effect derived from M-CSF-induced myelopoiesis.

M-CSF-induced I-IFN production stimulates IL-15-dependent antiviral immunity
Type I interferons contribute to the early antiviral immune response preceding NK cell activation (Degli-Esposti & Smyth, 2005;Liu et al, 2005;Baranek et al, 2012;Alexandre et al, 2014;Cocita et al, 2015) and thus, may constitute a rapid response mechanism that could prevent fatal viremia during leukopenia after HCT.MCMV infection was shown to increase IFNB1 mRNA in the spleen (Dalod et al, 2002;Zucchini et al, 2008).Consistently, we found enhanced IFNB1 mRNA levels in the spleen swiftly after MCMV infection, which were strongly further augmented with M-CSF treatment ( Together, this supported the notion that M-CSF treatment increased I-IFN levels during MCMV infection of HCT recipients by promoting a faster reconstitution of I-IFN-producing monocytes and pDCs.This also agrees with the observation that M-CSF-driven myelopoiesis can also stimulate pDC development (Fancke et al, 2008).The observed effects of both loss-and gain-of-function experiments targeted at myeloid cells (Fig 4C and D) or transplantation of GMPs, which also give rise to pDCs, thus support the notion of I-IFNs also contributing to antiviral immunity upon M-CSF administration after HCT.
Beyond its direct antiviral effects on infected cells, I-IFNs can also indirectly affect the antiviral immune response by activating NK cells or by stimulating IL-15 production in myeloid cells (Nguyen et al, 2002;Degli-Esposti & Smyth, 2005;Baranek et al, 2012).To investigate the relative importance of I-IFNs on myeloid cells, we injected IFNAR1-KO or WT GMPs at day 10 after HCT.IFNAR1 deficiency abolished the protective effect of GMP ◀ Figure 3. NK cell activity is required for the antiviral effect of M-CSF.Analysis performed 1.5 days (or 4 days in D) after MCMV or mock infection.
A NK cell activity in the spleen.FACS examples and median percentage of donor-derived NK1.1 + NK cells producing IFNc (n = 5-6 mice per group, two independent experiments).B FACS examples and median percentage of donor-derived NK1.1 + NK cells producing GrB (n = 5-6 mice per group, two independent experiments).C Gene expression analysis of activation and maturation-related factors in FACS-sorted, donor-derived NK1.Finally, we further queried the effect of M-CSF-driven myelopoiesis and IL15Ra signaling in human G-PBMCs on functional NK cell differentiation.We first analyzed CLPs, which encompass NK cell progenitors (Grzywacz et al, 2011).Interestingly, CLPs were enriched in M-CSF-treated G-PBMCs, both at days 5 and 9 (Fig 8D).Although the culture regime lacked exogenous IL-2, IL-15 or IL-21 and thus was not ideal for NK cell differentiation and survival, M-CSF-driven myelopoiesis resulted in significantly more NK cells (SSC-A low Lin À CD56 + CD16 + ) at day 9 of culture (Fig 8E).In line with the findings in murine cells, M-CSF treatment also increased the numbers of GrB-expressing NKs (Fig 8F ) significantly compared to IL-3-driven myelopoiesis on day 9 of in vitro culture, indicating that in contrast to IL-3, M-CSF treatment specifically stimulated functionally competent NK cell production also in human G-PBMCs.
Together, these findings indicated that the coordinated myeloiddriven NK cell differentiation and activation program initiated by M-CSF-mediated myelopoiesis in mice was translatable to the human context and was thus directly relevant for clinical conditions of HCT.
No adverse events of M-CSF after allogeneic HCT Next, we analyzed the effects of M-CSF on engraftment and recovery of transplant recipients in an allogeneic HCT model (Fig EV5A).Following allogeneic HCT and assessment according to previous reports (Alexander et al, 2014), we did not find any differences in the frequency of C57BL/6j CD45.1 + donor HSPC-derived CD11b + F4/80 + monocytes or inflammatory Ly6C HI monocytes after M-CSF treatment (Fig EV5B and C).Furthermore, disease scoring (Lai et al, 2012) showed no statistical difference between mice treated with M-CSF or PBS, going to the lowest possible score as early as 20 days following allogeneic HCT for both conditions (Fig EV5D).All mice survived M-CSF treatment and vehicle control after allogeneic HCT.We further showed that tri-lineage engraftment in the peripheral blood was not affected by M-CSF treatment at 4 and 12 weeks following allogeneic HCT (Fig EV5E).Merely few residual recipient BALB/c CD45.2 + cells were found ("alloHSPCs" in Fig EV5B), reflecting full bone marrow (BM) engraftment of CD45.1 + donor cells, which we confirmed at 12 weeks (Fig EV5F).These CD45.1 + donor cells were unaffected by M-CSF treatment concerning long-term engrafting HSPCs (KSL Flt3 À CD150 + CD48 À ) ◀ Figure 4. M-CSF-induced myelopoiesis is required for its antiviral effect.
A Splenic GMPs of control or M-CSF-treated, uninfected mice 14 days after HCT.B Splenic granulocytes (Ly6G + CD11b + ) and mononuclear phagocytes (Ly6G À CD11b + ) of control or M-CSF-treated, uninfected recipient mice 14 days after transplantation.C Analysis of M-CSF-dependent myeloid cells for its antiviral effect.Survival curve of MCMV-infected and PBS-control-treated (n = 10), M-CSF and Ig-control-treated Together, our data reveal no contraindication for the short-term treatment with M-CSF following allogeneic HCT, suggesting that it should be a safe and feasible cytokine to promote antiviral activity in standard protocols of allogeneic HCT.

Discussion
In this study we have identified the previously unknown protective effects of M-CSF-induced myelopoiesis against viral infection during the vulnerable leukopenic phase after HCT.We identified a coordinated differentiation program between myeloid and NK cells that plays a major role in reconstituting protection against viral infection and assigns a critical role to M-CSF-induced myelopoiesis in participating in antiviral immunity.
Immunocompromised individuals are prone to opportunistic infections including CMV viremia, but also to infection-induced morbidity and mortality as shown in mice (Arber et al, 2003).Here, we used a murine model of immunosuppression after HCT to investigate the protective antiviral effects of M-CSF-induced myelopoiesis preceding MCMV infection.HCT is an important major therapeutic strategy that involves a conditioning therapy by which the recipient's hematopoietic system is immunosuppressed to foster engraftment of donor HSPCs.Patients encounter severe immunodeficiency after HCT that leaves them highly vulnerable to opportunistic bacterial, fungal, and viral infection before the donor's hematopoietic system is sufficiently reconstituted.Although improvements have been made in prophylaxis and management, viral infection, and reactivation, such as CMV, still contribute significantly to morbidity and mortality after allogeneic HCT (Zaia, 1990;Ljungman et al, 2002;Boeckh & Ljungman, 2009).Unfortunately, available antiviral drugs are associated with numerous adverse effects (Ahmed, 2011;El Chaer et al, 2016).For example, ganciclovir severely compromises myelopoiesis, and thus further aggravates susceptibility to secondary infections (Goodrich et al, 1993;Boeckh et al, 1996;Salzberger et al, 1997) and enhances risk to secondary malignancy (de Kanter et al, 2021).Although progress has been made with the introduction of letermovir as a non-toxic antiviral agent, it might select for virus variants and virus breakthrough infections as well as late reactivation once cessation of prophylaxis occurs (Hill et al, 2021).Furthermore, as an agent targeting viral terminase complex it is limited to be used against CMV and is not effective against other viruses.Adoptive transfer protocols of lymphoid progenitors also have been proposed as a therapeutic strategy in refractory or high-risk cases (Kaeuferle et al, 2019).Cell therapy approaches, however, require complex logistics, which limits their availability and leads to high costs.Given the remaining clinical need for both acute and prophylactic antiviral treatments, the application of M-CSF may represent an attractive, cost-effective, and broadly applicable antiviral host-directed approach.
Several properties of M-CSF make it an ideal candidate for accelerating immunocompetence recovery in HCT recipients and present key advantages over other myeloid cytokines used in clinical practice.We showed before in murine models that M-CSF directly engages HSPCs and thus intervenes at the earliest point of the differentiation hierarchy to initiate the production of innate immune cells (Sarrazin et al, 2009;Mossadegh-Keller et al, 2013;Kandalla et al, 2016).M-CSF prophylaxis could therefore shorten the time of immune system reconstitution to reduce the risk of infections.Other cytokines, in particular G-CSF, are also used to stimulate immune functionality.However, in contrast to M-CSF, G-CSF can only act on already existing mature or late myeloid progenitor cells to activate their functional competence.Since these cells will only develop weeks after HCT, G-CSF will be ineffective in the early phase after HCT.By acting at the earliest point of the hematopoietic differentiation hierarchy, M-CSF can stimulate myelopoiesis swiftly after conditioning therapy.Consistent with this, we showed previously in vivo in mice that M-CSF but not G-CSF can stimulate the increased production of myeloid cells from HSPCs and protect from bacterial and fungal infections (Kandalla et al, 2016).Importantly, in previous murine studies M-CSF-induced myelopoiesis neither compromises stem cell numbers or activity (Sarrazin et al, 2009), nor comes at the expense of the generation of other blood cell lineages like platelets that are important for restoring blood clotting activity (Kandalla et al, 2016).Here, we report an additional advantage of M-CSF treatment by promoting rapid reconstitution of antiviral activity in vivo and protection from viral infection through a multistep myeloid and NK cell differentiation program.A significant advantage of the early action of M-CSF on HSPCs appears to be the stimulation of a combination of innate immune cells that are required to combat pathogens.Whereas G-CSF only stimulates granulocytes and their direct progenitors, M-CSF stimulates the production of (i) granulocytes, mediating cytotoxic bacterial killing, (ii) monocytes and macrophages, capable of pathogen control by ◀ Figure 5. Myeloid IL-15 trans-presentation is required for the antiviral activity of M-CSF.
A Splenic NK1.1 + , immature and mature NK cells of uninfected control or anti-CD115-treated mice 2 days after depletion and 14 days after HCT and M-CSF treatment.B Mice MCMV-or mock-infected 14 days after HCT and analyzed 1.5 days after.Il15 mRNA levels (RT-qPCR).C Mice MCMV-or mock-infected 14 days after HCT and analyzed 1.5 days after.Sorted, donor-derived monocytes and cDCs assessed by Fluidigm.D Mice MCMV-or mock-infected 14 days after HCT and analyzed 1.5 days after.Ly6C hi monocytes (left), IL15Ra-expressing, donor-derived Ly6C hi or Ly6C low monocytes (right).E Mice MCMVor mock-infected 14 days after HCT and analyzed 1.5 days after.Gene expression analysis of donor-derived NK cells by Fluidigm.phagocytosis and reactive oxygen production, and (iii) dendritic cells with the strongest antigen presentation activity that alerts the adaptive immune system.In this study, we now show that M-CSF also induced I-IFN-producing pDCs and indirectly stimulated NK cell differentiation and activation through induction of IL-15-producing monocytic cells, which together mediated strong antiviral activity.Allogeneic HCT harbors the risk of acute and chronic GvHD (Ferrara et al, 2009).To date, most pre-clinical studies indicated a  beneficial effect of M-CSF in allogeneic HCT.M-CSF treatment applied in mice just prior to transplantation induced host macrophages to engulf alloreactive T cells (Hashimoto et al, 2011).Conversely, depletion of resident macrophages by abrogating the M-CSF/CSF-1R-axis during the peri-transplantation period aggravated GvHD in vivo (Macdonald et al, 2010).With respect to chronic GvHD in the human setting, a retrospective analysis of over 50 Japanese bone marrow transplant recipients receiving M-CSF at or early after transplantation showed a lower risk to develop this complication (Kimura et al, 2012).This underpins the relevance of the timing of M-CSF administration for the differential effect on GvHD development.
Although the early application of M-CSF peri-transplantation mediated the antimicrobial effects described by us and lead to a lower rate of chronic GvHD, it must be noted that the therapeutic window appears to be smaller when the compound is applied later after engraftment of T cell-replete allografts in mice (Alexander et al, 2014).The M-CSF prophylaxis described by us targets NK cells and pDCs, whose protective functions during CMV infection are well described both in mice (Alexandre et al, 2014;Brinkmann et al, 2015) as well as in humans (Brinkmann et al, 2015).This is important for a fast antiviral response under immunosuppressed and leukopenic conditions since an antiviral T cell response cannot be mounted, especially in the context of T cell-depleted allografts.Under these circumstances, the development of engrafted T cells arising from donor HSPCs occurs much later than viral reactivation during immunosuppressive leukopenia.In line with this, a recent randomized trial demonstrated that allogeneic transplantation of ex vivo T cell-depleted hematopoietic grafts clearly reduced the incidence of GvHD but was associated with a significant increase of non-relapse mortality (Luznik et al, 2022).As the majority (> 50%) of these deaths were related to infectious complications, the use of M-CSF peri-transplantation may be a strategy to improve the overall outcome of T cell-depleted allogeneic HCT.
The effect on NK cells described in this study is mediated by M-CSF-induced myelopoiesis, in particular by monocytes.The role of monocytes and macrophages in CMV infection is multifaceted.On the one hand, rodent studies have shown they can be target cells for MCMV infection (Hanson et al, 1999;Hokeness et al, 2005;Daley-Bauer et al, 2012), thus serving as vehicles of CMV dissemination (Smith et al, 2004;Daley-Bauer et al, 2014).On the other hand, the observation that macrophage depletion in vivo increased MCMV burden (Hanson et al, 1999), also supports a protective role during CMV infection.This ambiguity might be dependent on the context of infection or on the specific monocyte subpopulation.Whereas Ly6C À CX3CR1 hi patrolling monocytes are involved in CMV dissemination in the mouse (Daley-Bauer et al, 2014), Ly6C + CCR2 + inflammatory monocytes can engage antiviral responses in early infection via direct or indirect mechanisms (Salazar-Mather et al, 2002;Hokeness et al, 2005;Soudja et al, 2012;Rovis et al, 2016;Gawish et al, 2019).In murine models, Ly6C + CCR2 + inflammatory monocytes could initiate differentiation of memory CD8 + T and NK cells into antimicrobial effector cells (Soudja et al, 2012) or showed direct iNOS-mediated antiviral effects (Rovis et al, 2016).I-IFN signaling is also important for recruitment of CCR2 + inflammatory monocytes via MCP-1/CCL2 in vivo (Salazar-Mather et al, 2002).Thus, mice deficient for MCP-1 or CCR2 showed a reduced accumulation of monocyte-derived macrophages and NK cells in liver, increased viral titers, widespread virus-induced liver pathology and reduced survival (Hokeness et al, 2005;Crane et al, 2009).Previously, murine studies showed that CD11c hi DC-derived IL-15 promoted NK cell priming (Lucas et al, 2007) and that inflammatory monocyte-derived IL-15 could stimulate NK cell differentiation (Soudja et al, 2012).In the immunosuppressed settings investigated here, Ly6C hi monocytes appeared to be more important than DCs for IL-15 presentation, since they expressed higher levels of IL15Ra required for IL-15 cross-presentation to NK cells (Lucas et al, 2007).◀ Figure 7. M-CSF supports terminal differentiation of monocyte-derived macrophages from human G-CSF-mobilized PBMCs.
A Cytospins at days 5 or 9 after in vitro differentiation without myelopoiesis-inducing cytokines (À), or with IL-3 (+3) or M-CSF (+M) (modified Giemsa).Images were acquired with a Zeiss AX10 benchtop microscope at 40× magnification.B Median fluorescence intensity of SSC-A (granularity) at seeding (d0), or after in vitro cytokine treatment without myelopoiesis-inducing cytokines (À, green circle), with IL-3 (+3, blue circle) or with M-CSF (+M, salmon circle) (data shown in technical triplicates from five biological donors).C Pseudocolor scatterplots of G-CSF-mobilized PBMCs after in vitro cytokine treatment shown for CD34 (left) and CD11b expression (right) on the x-axis and SSC-A on the y-axis without myelopoiesis-inducing cytokines (À, top row), with IL-3 (+3, middle row) or with M-CSF (+M, bottom row).D Frequency of CD34 + HSPCs at seeding (d0), or after in vitro cytokine treatment without myelopoiesis-inducing cytokines (À, green circle), with IL-3 (+3, blue circle) or with M-CSF (+M, salmon circle) (data shown in technical triplicates from five biological donors).E Frequency of CD11b + myeloid cells at seeding (d0), or after in vitro cytokine treatment without myelopoiesis-inducing cytokines (À, green circle), with IL-3 (+3, blue circle) or with M-CSF (+M, salmon circle) (data shown in technical triplicates from five biological donors).F Contour plots of HSPC populations comprising of CLPs (CD34 + CD45RA + CD38 À ) or GMPs (CD34 + CD45RA + CD38 + ) upon selection from G-CSF-mobilized PBMCs (d0) or after in vitro cytokine treatment at day 5: without myelopoiesis-inducing cytokines (À) versus IL-3 (+3) versus M-CSF (+M).G Frequency of GMPs at seeding (d0, empty circle) or without myelopoiesis-inducing cytokine treatment (À, green circle), with IL-3 (+3, blue circle) or with M-CSF (+M, salmon circle) (data shown in technical triplicates from five biological donors).H Frequency of HLA-DR + GMPs at seeding (d0, empty circle) or without myelopoiesis-inducing cytokine treatment (À, green circle), with IL-3 (+3, blue circle) or with M-CSF (+M, salmon circle) (data shown in technical triplicates from five biological donors).I M-CSF-stimulated differentiation of monocyte-derived macrophages (Mo-Macs, CD11b + CD66b À CD64 + ) (data shown in technical triplicates from five biological donors).J M-CSF-stimulated differentiation of monocyte-derived macrophages (Mo-Macs, CD11b + CD66b À HLA-DR + ) (data shown in technical triplicates from five biological donors).K Frequency of CD14 + cells at seeding (d0), or after in vitro cytokine treatment without myelopoiesis-inducing cytokines (À, green circle), with IL-3 (+3, blue circle) or with M-CSF (+M, salmon circle) (data shown in technical triplicates from five biological donors).L M-CSF-stimulated differentiation of monocyte-derived macrophages (Mo-Macs, CD14 + CD16 + ) (data shown in technical triplicates from five biological donors).(Nguyen et al, 2002;Lucas et al, 2007), we observed that both IL15RA-and IFNAR1deficiency in GMP-derived myeloid cells abolished their protective effect against MCMV, whereas ectopic IL-15 could rescue IFNAR1deficiency.This suggested that IL-15 induction in monocytes required I-IFNs that were mainly produced by M-CSF-induced pDCs.Together, our experiments revealed the surprising capacity of M-CSF to initiate a fully synchronized differentiation program and cytokine mediated crosstalk between different myeloid and NK cell lineages to provide effective antiviral prophylaxis during leukopenia following HCT-mediated immunosuppression.Compatible with this notion, M-CSF therapy could not only be beneficial for HCT protocols but also for other settings of severe leukopenia, e.g., after chemotherapy or in septicemia.Furthermore, the myeloid-mediated antiviral activity described here could also be more generally helpful to combat other viral infections.M-CSF-induced IL-15 might promote innate immune responses via the induction of NK cells without a coinciding T helper type 2 cell-dependent cytokine storm, as has been suggested for SARS-CoV2 in humans (Kandikattu et al, 2020).This renders M-CSF application an attractive, potentially antiviral treatment prophylaxis.However, further studies are needed to evaluate the clinical employability of M-CSF following HCT as a prophylaxis of CMV infection in humans.For this, phase I/II clinical trials will be needed to evaluate the addition of M-CSF to the currently licensed cytokine treatment options comprising of G-CSF and GM-CSF.
For all in vivo experiments, block randomization was used to establish identical or similar group sizes.

MCMV infection, viral loads, and histopathology
Two weeks after HCT, mice were injected intraperitoneally with 5,000 plaque-forming units (PFU) of MCMV K181 v70 in 200 ll PBS.Viral loads were measured by quantitative reverse transcription polymerase chain reaction (RT-qPCR) of Ie1 mRNA (Baranek et al, 2012) extracted from frozen tissues 36-40 h (1.5 days) or 72 h as reported previously (Cocita et al, 2015).Paraformaldehyde-fixed (4%), paraffin-embedded and hematoxylin and eosin (H&E)-stained liver sections were scored by a trained veterinary pathologist blinded to sample identity for indicated parameters.

Microfluidic real-time RT-PCR gene expression analysis
Total mRNA extraction from 50,000 FACS-sorted cells and cDNA synthesis were performed with lMACS one step T7 template kit (Miltenyi) and specific gene expression (primers in Table EV3) was detected according to Fluidigm protocols as previously described (Soucie et al, 2016) or by SybrGreen method (Mossadegh-Keller et al, 2013).Ct values were calculated by BioMark Real-time PCR Analysis software (Fluidigm) using the DDCt method and HPRT for normalization.

Statistical analysis
None of the mice or human samples were excluded from the experiments.Multiple statistical methods, including Student's t-test, Mann-Whitney U-test, log-rank (Mantel-Cox) test were used in this study depending on the data type, and the details can be found in the figure legends: test used and exact value of n.For in vivo experiments, block randomization was used to establish identical or similar group sizes.No statistical methods were used to predetermine sample sizes.Veterinary pathologists assessing and scoring liver sections were blinded to sample identity for indicated parameters.No further blinding was performed.Experiments were performed independently at least twice.
Data between two groups were analyzed with unpaired Student's t-tests.All statistical analyses were performed using GraphPad Prism (Version 9.4.0,GraphPad Software, La Jolla, CA, USA).All data were analyzed using the FlowJo software (Version 10.8.1; Tree Star, Ashland, OR, USA) and expressed as medians + individual data points or means AE SEM unless stated otherwise.P-values less than 0.05 were considered significant.

Study approval
Procedures involving animals and their care were conducted in accordance with institutional guidelines and internal laws and policies and approved by the Institutional Animal Care and Use Committee (APAFIS #17258-2018102318448168-v5 and APAFIS #36188-2022032912082580-v6).All mouse experiments were performed under specific pathogen-free conditions and animals were monitored daily for signs of morbidity.
The use of human samples was approved by the Ethical Review Committee of the TU Dresden (approval no.EK477112016 and EK393092016) and all human research conformed to the Declaration of Helsinki and the Department of Health and Human Services Belmont Report.Informed consent was obtained from all participants.
Viral infection or reactivation of lifelong dormant viruses in immunosuppressed patients accrues to a significant death toll.For example, reactivation of cytomegalovirus, an opportunistic virus of the Herpesviridae family, in patients that have undergone hematopoietic stem cell transplantation (HCT) is a major cause of acute mortality at the leukopenic phase when donor stem cell reconstitution has yet to occur.Unfortunately, licensed drugs against CMV are either insufficiently effective and rely on the viral replication machinery (letermovir) or have severe side effects including bone marrow toxicity and risk of posttreatment mutagenesis (ganciclovir).Hence, further alternative well-tolerated host-directed therapies are needed that rapidly confer broad-spectrum antimicrobial competence in severely immunocompromised patients following HCT.

Results
The administration of M-CSF around the time of transplantation revealed a novel mechanism to protect against lethal CMV infection in leukopenic mice following HCT.Here, M-CSF stimulated hematopoietic stem cells to overcome leukopenia by committing to monopoiesis.We show that in mice M-CSF engages critical host-directed initial mechanisms to reconstitute antiviral immune responses after HCT, which are type I interferon and natural killer (NK) cell-dependent.Mechanistically, NK cell-mediated protection from lethal CMV infection arose from IL-15 and type I interferon production by M-CSFdriven monopoiesis or plasmacytoid dendritic cell differentiation, respectively.Synergistically, these brought about NK cell differentiation and activity from recruited progenitors.As in mice, human G-CSF-mobilized stem and progenitor cells within PBMCs also responded to M-CSF administration with elevated monopoiesis, IL-15Ra expression, and functional NK cell generation.Lastly, we could show that M-CSF did not confer any adverse effects on long-term hematopoietic stem cell engraftment and hematopoietic lineage balance in the blood or acute post-transplantation symptoms following allogenic HCT.

Impact
Overall, our preclinical results support the efficacy and feasibility of M-CSF administration as a prophylactic host-directed therapy that confers rapid protection from lethal CMV infection.Moreover, our data suggests a paradigm by which M-CSF boosts innate antimicrobial immunocompetence, which may constitute a well-tolerated general host-directed therapeutic in states of severe immunosuppression.ii Information on CMV in transplanted patients provided by Cedars Sinai: https://www.cedars-sinai.org/health-library/diseases-and-conditions/c/cmv-and-transplant-patients.html.
. As shown in Fig 1A, mice received three injections of murine M-CSF or PBS at the time of HCT and were infected 14 days later with MCMV doses accounting for 80-90% lethality in untreated transplant recipients (Fig EV1A).Survival rates significantly increased from 14.3 to 83.3% in M-CSF-treated mice (Fig 1B).Mice receiving four treatments of murine M-CSF over several days (Fig EV1B) or human M-CSF (Fig EV1C) both showed similar but not further improved survival rates.Accordingly, we used three treatments at the time of transplant as optimal condition throughout the study, although it should be stressed that a single M-CSF treatment already improved survival rates to a similar extent (Fig EV1D).Macrophage colony-stimulating factor-treated mice showed less severe liver injury with a proclivity for scarcer inflammatory foci (Fig 1C), a reduction of apoptotic or necrotic hepatocytes (Fig 1D), and decreased necrotic areas after MCMV infection (Fig 1E).M-CSFtreated mice also showed a decreased viral load as shown by reduced number of infected hepatocytes (Fig 1F) and viral RNA copy numbers (Fig 1G).Together, these results demonstrated that M-CSF treatment protected HCT recipients from MCMV-induced tissue damage and lethality.
Data information: ***P < 0.0001 by Mantel-Cox test (B).*P < 0.05 by two-tailed Mann-Whitney U-test (C-G).All data are representative of at least two independent experiments.Source data are available online for this figure.

Figure 2 .
Figure 2. M-CSF treatment increases NK cell production, differentiation, and activation.Experimental set-up as in Fig 1A.Analysis of spleen NK cell populations.Mice were MCMV-or mock-infected (PBS control) 14 days after HCT (AE M-CSF support as indicated in Fig 1A).Analysis was performed 1.5 days after MCMV or mock infection.A FACS examples and median of absolute number of total NK cells (CD19 À CD3 À Ly6G À NK1.1 + ) are shown (n = 5 mice per group, one independent experiment is shown but was confirmed twice).B Markers specific to differentiation and maturation stages of NK cells used in this analysis are indicated.C Median of absolute number of donor-derived NK progenitor cells (CD122 + CD27 + NK1.1 À Nkp46 À CD45.1 + ) are displayed (n = 5 mice per group, one independent experi- ment is shown but was confirmed twice).D FACS examples and median of absolute numbers of donor-derived immature NK cells, donor-derived M1 (CD11b + CD27 + ) and M2 NK cells (CD11b + CD27 À ) are shown (n = 5 mice per group, one independent experiment is shown but was confirmed twice).E Gene expression analysis of transcription factors expressed by NK cells in FACS-sorted, donor-derived NK1.1 + NK cells (definitions of Fig 2A) by nanofluidic Fluidigm array real-time PCR.
lineage could indirectly impact on NK cell-mediated antiviral activity.M-CSF treatment can increase donor myelopoiesis in HSPCtransplanted mice (Mossadegh-Keller et al, 2013; Kandalla et al, 2016).Accordingly, M-CSF increased donor-derived GMPs, granulocytes, mononuclear phagocytes (Fig 4A and B), pDCs, and cDCs (see Fig EV3A-C) two weeks after HCT.To determine whether this was relevant to the antiviral effect of M-CSF, we used complementary loss-and gain-of-function approaches.We injected anti-MCSFR/CD115 antibody 12 days after HCT, which selectively eliminates M-CSF-dependent myeloid cells (Tagliani et al, 2011).Myeloid cell depletion completely abolished the protective effect of M-CSF treatment in MCMV-infected HCT recipients (Fig 4C).Affirmatively, these mice showed reduced GMPs, monocytes, cDCs, and pDCs 48 h after anti-CD115 myeloid depletion (Fig 4D).This indicated that myeloid cells were required for the M-CSF-dependent antiviral activity.For gain-of-function experiments, we transplanted GMPs Figure 2.
Fig 6A).During MCMV infection, I-IFNs are predominantly produced by pDCs.We observed that pDC numbers (Fig 6B) and I-IFN-producing pDCs (Fig 6C) were increased in the spleen of M-CSFtreated mice 14 days after HCT, particularly after MCMV infection.Monocytes also showed a strongly increased expression of IFNB1 and upstream transcription factors of the IRF family (Fig 5C).
Data information: Data are illustrated as mean AE SEM.A ratio-paired t-test was used.*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.Data are representative of five independent experiments.Source data are available online for this figure.
Consistent with the synergistic role of IL-15 and I-IFNs for NK cell activation demonstrated in murine studies

Figure 8 .
Figure 8. M-CSF-driven myelopoiesis induces IL15Ra expression by monocyte-derived macrophages and supports NK cell viability and cytokine competence in human G-CSF-mobilized PBMCs.