Lactate released by inflammatory bone marrow neutrophils induces their mobilization via endothelial GPR81 signaling

Neutrophils provide first line of host defense against bacterial infections utilizing glycolysis for their effector functions. How glycolysis and its major byproduct lactate are triggered in bone marrow (BM) neutrophils and their contribution to neutrophil mobilization in acute inflammation is not clear. Here we report that bacterial lipopolysaccharides (LPS) or Salmonella Typhimurium triggers lactate release by increasing glycolysis, NADPH-oxidase-mediated reactive oxygen species and HIF-1α levels in BM neutrophils. Increased release of BM lactate preferentially promotes neutrophil mobilization by reducing endothelial VE-Cadherin expression, increasing BM vascular permeability via endothelial lactate-receptor GPR81 signaling. GPR81−/− mice mobilize reduced levels of neutrophils in response to LPS, unless rescued by VE-Cadherin disrupting antibodies. Lactate administration also induces release of the BM neutrophil mobilizers G-CSF, CXCL1 and CXCL2, indicating that this metabolite drives neutrophil mobilization via multiple pathways. Our study reveals a metabolic crosstalk between lactate-producing neutrophils and BM endothelium, which controls neutrophil mobilization under bacterial infection.

I nnate immune neutrophils are the first cells to be mobilized and recruited from the bone marrow (BM), against bacterial infections 1 . During steady-state homeostasis, most murine neutrophils reside in the BM, and only a small fraction migrate to the circulation 2,3 . However, during inflammation, BM neutrophils are rapidly mobilized to the blood and recruited to peripheral organs in order to eradicate the invading pathogens 4,5 . Activation of pro-inflammatory signaling pathways involves stimulation of metabolic pathways in immune cells as a part of host defense responses 6 . These include increased oxygen consumption during myeloid induced phagocytosis, due to activation of NADPH oxidase (NOX) in order to generate oxygen free radicals 7 . Furthermore, studies mimicking bacterial infection also revealed increased glucose uptake 8 and a shift to glycolysis rather than oxidative phosphorylation, which is characteristic of proinflammatory macrophages 9 .
While there are multiple studies concerning immunemetabolism of macrophages, less is known about neutrophil metabolic control during the onset of acute inflammation. Neutrophils are highly glycolytic with limited mitochondrial respiration 10,11 . Lactate, a major byproduct of glycolysis, was found to be released by human neutrophils 12 and increased lactate levels are used as a marker for detecting sepsis in patients 13,14 . However, mechanistic reasoning for lactate production by BM neutrophils and its potential role during the early phase of acute inflammation remain elusive.
Here, we report that exposure to LPS or live Salmonella activates (within 4 h) BM neutrophils to produce and release lactate in both NOX-and hypoxia-inducible factor-1α (HIF-1α)dependent manners. The metabolite lactate preferentially mobilizes neutrophils by increasing BM vascular permeability upon activation of the lactate-receptor GPR81 expressed by BM endothelial cells. In addition, lactate also induces the release of the neutrophil attracting chemokines CXCL1 and CXCL2, and of the neutrophil mobilizing-cytokine granulocyte colony stimulating factor (G-CSF), which also involves GPR81-independent mechanisms. Consequently, lactate administration increases the defective LPS-induced mobilization of activated neutrophils in NOX-mutated mice, further demonstrating the critical roles of this metabolite in neutrophil mobilization during the early phase of bacterial infection.

Results
LPS increases lactate production by BM neutrophils. Neutrophils are predominantly glycolytic cells that produce reactive oxygen species (ROS) through the cytosolic enzyme NOX. This process is essential for microbial eradication and regulation of inflammation 15,16 . To better understand the metabolic consequences of BM neutrophil activation during the onset of acute inflammation, we treated wild-type (WT) mice with a low dose of LPS to mimic acute gram-negative bacterial inflammation. Our findings indicate that 4 h after LPS administration activated BM neutrophils (CD11b + /Ly6G high cells; Supplementary Fig. 1a) displayed increased glucose uptake (Fig. 1a), upregulated gene expression encoding the rate limiting glycolytic enzymes (hexokinase 1 (HK1) and phosphofructokinase 1 (PFKL); Fig. 1b) and downregulated levels of the TCA cycle genes ( Supplementary  Fig. 1b). Collectively, our findings suggest that BM neutrophils activate their glycolysis with very low rates of TCA cycle and oxidative phosphorylation during the onset of acute inflammation.
Next, we documented high production of ROS in BM neutrophils following LPS administration (Fig. 1c). Since ROS was shown to activate HIF-1α in macrophages 17 , we tested the impact of LPS on HIF-1α levels in BM neutrophils and found higher percentages of HIF-1α + neutrophils in the BM induced by LPS exposure (Fig. 1d). Moreover, we found that BM neutrophils express elevated levels of lactate dehydrogenase A (LDHA), a key glycolytic enzyme involved in the conversion of pyruvate to lactate, following systemic exposure to LPS (Fig. 1e). Notably, we found that selective depletion of neutrophils by neutralizing Ly6G antibodies resulted in lower levels of BM lactate (a functional output of LDHA activity) in mice injected with LPS (Fig. 1f,  Supplementary Fig. 1c). These data were supported by the observation that BM isolated neutrophils directly released high amounts of lactate following in vitro LPS stimulation (Fig. 1g,  Supplementary Fig. 1d). Taken together, our results demonstrate that LPS can directly induce glycolysis and oxidative bursts in BM neutrophils which lead to the production and release of lactate by these leukocytes during the early phase of acute inflammation. However, we cannot rule out that LPS administration can also indirectly activate BM neutrophils and drive their mobilization via its effects on other BM cell subsets.
Lactate cellular levels are tightly balanced by monocarboxylase transporters (MCTs) 12,18 . MCT1 mediates lactate influx, while MCT4 is expressed only in glycolytic cells and mediates lactate efflux 19,20 . In addition, lactate can bind and signal through its Gprotein coupled receptor GPR81 (also named hydroxycarboxylic acid receptor 1 (HCAR1)) and activates different signaling pathways 21,22 . We found that in steady state, MCT4 is highly expressed by BM neutrophils, while low expression of MCT1 and GPR81 was observed on the neutrophil surface (Fig. 1h, Supplementary Fig. 1e, f). Among BM myeloid cells, we found that MCT4 is preferentially expressed by neutrophils (LysM high / Ly6G high ) and to a lower extent by monocytes (LysM int /Ly6G neg ) ( Supplementary Fig. 1g). Moreover, LPS administration further upregulated MCT4 expression on BM neutrophils (Fig. 1i). Interestingly, we found that LPS-induced glucose uptake ( Supplementary Fig. 1h) and MCT4 expression ( Supplementary  Fig. 1i) also on PB neutrophils as well and not only on BM neutrophils. In addition, LPS treatment increased the levels of lactate ( Supplementary Fig. 1j) and reduced the glucose levels in the blood without changing the blood pH ( Table 1), suggesting that LPS-mobilized neutrophils are still glycolytic and most probably release lactate upon reaching vascular targets within inflamed peripheral organs to further promote their recruitment. Taken together, our results suggest that during the acute phase of LPS-induced inflammation, BM neutrophils uptake high amounts of glucose and convert it to lactate via glycolysis in the cytosol. Consequently, ROS and HIF-1α levels are increased to further activate the glycolytic pathway to produce and to release high levels of lactate via MCT4 (Fig. 1j).
Lactate preferentially induces BM neutrophil mobilization. Since our results indicate that BM neutrophils produce lactate during the onset of acute LPS-induced inflammation, we next investigated whether exogenous administration of this metabolite affects neutrophil mobilization from the BM, a hallmark of bacterial infection. Strikingly, injection of sodium lactate to WT mice (an upper physiological concentration range, Supplementary  Fig. 2a, b) led to reduction in BM neutrophil levels within 4 h (Fig. 2a, Supplementary Fig. 2c), accompanied by robust neutrophil mobilization to the peripheral blood (PB; Fig. 2b, Supplementary Fig. 2d) and recruitment to the liver in a dosedependent manner (Fig. 2c), without changing the blood pH (Table 1).
Notably, rapid neutrophil mobilization to the blood was detected as early as 2 h post lactate injection (Fig. 2d). We applied intravital two-photon laser-scanning microscopy in order to track live neutrophil mobilization over time. We generated chimeric   Supplementary  Fig. 2i).
Upon neutrophil activation, ROS are produced by NOX via a robust "oxidative burst" 25,26 . To assess if neutrophils acquire an activated phenotype in response to lactate, we measured ROS production and found increased levels of ROS high activated neutrophils in the circulation 4 h after lactate administration (Fig. 2f). Next, we found that the number of white blood cells (WBCs) was reduced in the BM following lactate injection ( Supplementary Fig. 3a). In addition, despite the increase in PB neutrophils, we detected reduction in the number of PB WBCs ( Supplementary Fig. 3a), suggesting a differential cell-specific response. We therefore determined the effect of lactate on other BM immune cells. Our results reveal that despite the expression of lactate transporters MCT4 and MCT1 and lactate receptor GPR81 on other BM immune cells ( Supplementary Fig. 3b-d; respectively), lactate administration acted preferentially on neutrophils (LysM high /Ly6G high cells), as we found no significant increase in the levels of PB monocytes or lymphocytes 4 h post lactate injection ( Supplementary Fig. 3e, f; respectively). These results suggest that in addition to the higher levels of lactate transporters, other pro-inflammatory factors are involved in the rapid neutrophil mobilization 4 h following lactate administration. Nevertheless, mobilization of monocytes to the circulation was observed later, 8 h post lactate injection ( Supplementary  Fig. 3g).
Finally, we examined the involvement of lactate production and release in neutrophil mobilization during acute inflammation. LPS administration increased the levels of PB neutrophils, in particular of activated ROS high neutrophils ( Fig. 2h-k). Pharmacological inhibition (Fig. 2g) of LDHA by specific inhibitors (sodium oxamate or FX11) or inhibition of MCT4 (lactate efflux) by α-cyano-4-hydroxycinnamic acid (CHC) markedly reduced the levels of either neutrophils or ROS high neutrophils in the blood ( Fig. 2h-k, respectively, see Methods for gating neutrophils in these specific experiments). Taken together, our results suggest that lactate preferentially induces rapid neutrophil mobilization to the peripheral blood and in particular of activated ROS high neutrophils, followed by other myeloid cells, such as monocytes later on.
Lactate production by neutrophils requires NOX/ROS signaling. The NOX/ROS axis in inflamed and activated neutrophils plays a key role in phagocytic defense against microbial pathogens. NOX also triggers ROS-mediated HIF-1α expression in macrophages 17 and activates HIF-1α in prostate cancer 27 . Thus, we investigated whether the NOX/ROS axis is also involved in LPS-induced metabolic activation of BM neutrophils by treating gp91phox −/− (NOX2 mutated) mice with LPS. We found a reduction in the levels of BM HIF-1α + neutrophils compared to their WT counterparts (Fig. 3a). Interestingly, NOX/ROS deficiency in gp91phox −/− mice impaired the molecular machinery imperative for lactate production and release by inflammatory BM neutrophils (indicated by the expression of LDHA and MCT4, and by BM lactate levels in vivo and in vitro) ( Fig. 3b-e, respectively). These results suggest that the NOX/ROS axis is essential for HIF-1α activity and lactate production by BM neutrophils during acute LPS-induced inflammation.
To determine whether NOX expression in neutrophils or in BM stromal cells is essential for lactate production, we established chimeric mice in which hematopoietic cells from the BM of WT mice were transplanted into gp91phox −/− mice and vice versa ( Supplementary Fig. 4a). As expected, we found that following LPS administration the NOX/ROS signaling in neutrophils and not in the stromal compartment is responsible for HIF-1α elevation ( Supplementary Fig. 4b) as well as lactate production and release ( Supplementary Fig. 4c, d). In line with these findings, we found that the NOX mutation impairs PB neutrophil activation predominantly in chimeric mice with NOX mutated in the hematopoietic compartment ( Supplementary Fig. 4e).
Consequently, LPS treatment failed to mobilize neutrophils in general and activated ROS high neutrophils in particular to the blood in gp91phox −/− mice compared to their WT counterparts (Fig. 3f, g). Importantly, lactate injection to bypass endogenous production and release of lactate by neutrophils, increased the levels of PB ROS high neutrophils in gp91phox −/− mice following LPS treatment (Fig. 3h). Taken together, our data reveal that NOX/ROS activities are upstream of lactate production in BM neutrophils since abnormal low metabolic rates were found in gp91phox −/− neutrophils during the onset of the acute inflammation. HIF-1α mediates lactate release and neutrophil mobilization. Activation of HIF-1α is essential for myeloid cell activation, infiltration to inflamed tissues and for the regulation of the glycolytic capacity in these cells 28,29 . HIF-1α regulates macrophage metabolism by enhancing the expression of glycolytic enzymes [30][31][32] . However, the involvement of HIF-1α in neutrophil-derived lactate production and release during the early phase of acute LPS-triggered inflammation has not been demonstrated.
To better understand the function of HIF-1α in lactate production by activated BM neutrophils, we generated mice with a myeloid-selective HIF-1α deficiency (deletion of exon 2 that disrupts the HIF-1α coding sequence; Supplementary Fig. 5a)   In addition, we also found a significant increased LDHA expression following LPS treatment in WT but not in HIF-1α −/+ or HIF-1α −/− mice ( Supplementary Fig. 5b, Fig. 4b, respectively). Another HIF-1α target is the lactate transporter MCT4 33 . In our myeloid-specific HIF-1α −/+ or HIF-1α −/− mice, LPS treatment up-regulated within 4 h MCT4 expression on BM neutrophils in a HIF-1α dependent manner (Supplementary Fig. 5c and Fig. 4c). Moreover, BM HIF-1α mutated neutrophils failed to release lactate ( Supplementary Fig. 5d, Fig. 4d, e) in contrast to WT BM neutrophils. Taken together, these results reveal that optimal neutrophil HIF-1α activity is essential for both high glycolysis activity and lactate production and release by activated BM neutrophils during the onset of acute LPS-triggered inflammation.
Next, we determined the role of HIF-1α in LPS-induced neutrophil mobilization to the circulation. We found lower levels of neutrophils in both the PB and in the liver of HIF-1α mutated mice treated with LPS ( Supplementary Fig. 5e, f, Fig. 4f, g) in comparison with their WT counterparts. Notably, lactate injection to HIF-1α mutated mice increased neutrophil levels in both PB and liver, similar to WT mice ( Supplementary Fig. 5g, h, Fig. 4h, i), thus placing HIF-1α upstream of lactate production.
Altogether, these findings indicate that HIF-1α directs metabolic programs in BM neutrophils, including glycolysis as well as lactate production and release, that are critical for neutrophil mobilization from the BM during the onset of acute inflammation. Moreover, our results show that disruption of one HIF-1α allele is sufficient to impair lactate production and release by activated BM neutrophils and to reduce their mobilization to the circulation.
Lactate increases BM endothelium permeability via GPR81. Others and we have shown that BM sinusoidal vessels are more permeable and serve as the exclusive trafficking site for both immature hematopoietic stem cells and mature leukocytes 24,[34][35][36] . To assess the involvement of BM sinusoidal "leakiness" in LPS-and lactate-induced neutrophil mobilization, we measured the penetration of Evans Blue Dye (EBD) from the circulation into the BM, a classical readout of BM vascular permeability. Indeed, lactate administration to WT mice increased BM vascular permeability within 4 h (Fig. 5a). Surprisingly, we identified that both sinusoidal (sBMECs; Endomucin + 37,38 , CD45 − /CD31 + /Sca-1 − 24,38 ) and arterial BM endothelial cells (aBMECs; Endomucin − 37,38 , CD45 − / CD31 + /Sca-1 + 24 ) express the lactate-receptor GPR81 (Fig. 5b, Supplementary Fig. 6a-c). To study the contribution of GPR81 signaling to BM vascular permeability and neutrophil mobilization, we treated mice with a specific GPR81 agonist (3,5-DHBA). This agonist reduced BM neutrophil numbers (Supplementary Fig. 6d) and increased their numbers in PB (Fig. 5c, Supplementary Fig. 6e) and liver ( Supplementary Fig. 6f) 4 h after administration. The GPR81 agonist also directly enhanced BM vascular permeability, as reflected by EBD penetration to the BM in WT mice (Fig. 5d). Importantly, using GPR81 −/− mice we found that GPR81 is essential for enhanced BM vascular permeability induced by lactate or LPS (Fig. 5e). Strikingly, we observed that selective depletion of neutrophils resulted in a reduction in BM vascular permeability in mice exposed to LPS (Fig. 5f) which was rescued by exogenous lactate treatment. These results suggest that lactate produced and released from BM neutrophils in the course of LPS-induced inflammation directly elevates BM vascular permeability by binding and signaling through endothelial GPR81.
Previously it was shown that lactate activates GPR81 and suppresses cAMP production by reducing adenylyl cyclase activity 21 . cAMP regulates endothelial permeability through its direct binding to Exchange protein directly activated by cAMP (Epac) 39,40 , which is critical for the maintenance of vascular endothelial cadherin (VE-cadherin)-mediated adherence junctions 41,42 . Consistent with a role of lactate/GPR81 axis in enhanced vascular permeability, we found that lactate, LPS or GPR81 agonist preferentially reduced VE-cadherin expression by the more permeable sBMECs in WT mice compared to their GPR81 −/− counterparts (Fig. 5g, Supplementary Fig. 6g). Interestingly, this reduction was not observed on the less permeable aBMECs that express higher levels of surface VE-cadherin to begin with 24 ( Supplementary Fig. 6h).
Notably, enforced disruption of VE-cadherin assemblies in GPR81 −/− mice with a VE-cadherin-function blocking antibody that impairs vascular integrity 41 , elevated PB neutrophil levels, similar to WT mice (Fig. 5h). These results further suggest that VE-cadherin expression and activity in the BM vasculature is modulated by GPR81 signaling triggered by lactate.
in combination with a selective Epac1 agonist 40 . We found that Epac1 agonist reduced lactate-induced BM vascular permeability (Fig. 5i) and neutrophil mobilization to the circulation (Fig. 5j, Supplementary Fig. 6i). Altogether, our results indicate lactatedriven activation of GPR81 signaling in sBMEC and suggest a downstream reduction in cAMP and Epac1 activation. This may be the mechanism that reduces VE-cadherin expression leading to increased BM vascular permeability and enhanced neutrophil mobilization during acute inflammation (Fig. 5k).
GPR81/lactate signaling mediates neutrophil mobilization. Next, we investigated the impact of lactate/GPR81 signaling on neutrophil mobilization. Pharmacological inhibition of GPR81 signaling 43 markedly reduced the effect of lactate on neutrophil mobilization from the BM to the PB and recruitment to the liver (Supplementary Fig. 7a-c; respectively). Furthermore, a significant lactate-induced reduction in BM neutrophil levels was observed in WT but not in GPR81 −/− mice (Fig. 6a), coupled with a large increase in neutrophil numbers in PB and liver of WT mice and a smaller but significant increase in the numbers of PB mobilized neutrophils in GPR81 −/− mice treated with lactate ( Fig. 6b, Supplementary Fig. 7d). Furthermore, LPS-induced mobilization of BM neutrophils was also lower in GPR81 −/− mice compared to their WT counterparts (Fig. 6c, d, Supplementary Fig. 7e). Importantly, this attenuated neutrophil mobilization following LPS treatment in GPR81 −/− mice was rescued by administration of blocking VE-cadherin antibodies 41 (Fig. 6e). This finding implicates LPS as a potent regulator of VE-cadherin dependent BM vascular permeability via its ability to trigger lactate release.

ARTICLE
To determine the relative contributions of vascular, stromal and hematopoietic GPR81 signaling ( Supplementary Fig. 1f, Supplementary Fig. 3d [44][45][46] ) to vascular permeability and neutrophil mobilization induced by lactate we generated chimeric mice in which hematopoietic cells from the BM of WT mice were transplanted into GPR81 −/− mice and vice versa ( Supplementary  Fig. 7f). The effects of lactate treatment to these chimeric mice demonstrated that lactate/GPR81 signaling in BM endothelial and stromal cells, but not in hematopoietic cells, control the increase in BM vascular permeability and neutrophil mobilization ( Fig. 6f-h). Moreover, to better reveal the specific function of endothelial lactate/GPR81 signaling in neutrophil mobilization, we generated inducible endothelial-specific GPR81 KO mice. We observed a significant lactate-induced reduction in BM neutrophil levels in GPR81 flox/flox /Cdh5 cre neg (WT counterparts) but not in endothelial-specific GPR81 KO mice (GPR81 flox/flox /Cdh5 cre pos ; Fig. 6i), coupled with a large increase in PB neutrophil levels of WT counterparts and a smaller but significant increase in PB mobilized neutrophils in endothelial-specific GPR81 KO mice treated with lactate (Fig. 6j).
The residual lactate-induced neutrophil mobilization we detected in both GPR81 −/− and in inducible endothelial- specific GPR81 KO mice (Fig. 6b, j) raised the possibility that lactate triggers additional pathways independent of GPR81 signaling which promote neutrophil mobilization from the BM. Since the chemokines CXCL1 and CXCL2 47-51 and the neutrophil-mobilizing cytokine G-CSF [51][52][53] contribute to the early stages of neutrophil mobilization and recruitment during acute inflammation, we next examined the potential effects of lactate concerning the release of these factors in BM and blood. Notably, we found that lactate treatment dramatically increased CXCL1 levels in both compartments in an endothelial-GPR81independent manner ( Supplementary Fig. 7g, h, Fig. 6k, l). We also observed a more moderate increase in the levels of CXCL2 in both BM and blood of WT mice treated with lactate ( Supplementary Fig. 7i, j). Lactate-induced elevation of CXCL1 and CXCL2 most probably downregulated surface expression of their receptor CXCR2 by PB neutrophils (Supplementary Fig. 7k). Importantly, we also found that lactate administration dramatically increased blood G-CSF levels in WT mice and a smaller but  (Fig. 6m). These results could explain the residual lactate-induced neutrophil mobilization we observed in the two models of GPR81 −/− mice and why only neutrophils are rapidly mobilized 4 h post lactate administration despite the increased BM vascular permeability. Collectively our results reveal that the lactate/GPR81 axis is a key regulator of BM vascular permeability required for optimal mobilization of BM neutrophils to the blood and their recruitment to inflamed organs. In addition to this axis, lactate also triggers the release of neutrophil mobilizing chemokines and G-CSF, key inducers of rapid neutrophil mobilization from the BM.

Salmonella increases lactate production by BM neutrophils.
Finally, to examine the clinical relevance of our findings, we infected mice with gram-negative SalmonellaTyphimurium and found higher levels of lactate in both BM and blood of WT mice (Fig. 7a,  b; respectively), similar to those prompted by LPS (Fig. 1f, Supplementary Fig. 1j). Next, we infected gp91phox −/− mice and compared their metabolic parameters with WT mice. Our findings indicate that following the onset of infection (4 h after bacterial challenge) WT BM neutrophils displayed high generation of ROS, elevated levels of HIF-1α + neutrophils as well as lactate production and release (Fig. 7c-f; respectively). In contrast, similar infection of gp91phox −/− mice did not increase the numbers of BM HIF-1α + neutrophils as observed in WT counterparts (Fig. 7d). Interestingly, the deficiency in NOX/ROS activities in infected gp91phox −/− mice resulted in impaired rates of lactate production and release from BM neutrophils infected with Salmonella (Fig. 7e, f). Consequently, WT mice infected with Salmonella also have higher levels of activated neutrophils in the blood and liver, as compared to gp91phox −/− or GPR81 −/− mice counterparts (Fig. 7g-i). Of note, although to a lower extent than in WT mice, elevated levels of liver neutrophils were also observed in Salmonella-infected gp91phox −/− or GPR81 −/− mice, suggesting that the NOX and GPR81 axes are major but not exclusive players in Salmonella-induced neutrophil recruitment.
Altogether, our data reveals that the regulatory mechanisms identified in the present work by which neutrophils respond to enforced LPS challenges are also activated during physiological bacterial infections with Salmonella, demonstrating its clinical relevance.

Discussion
In this study, we describe a role for the metabolite lactate implicated in the early phase of acute inflammation triggered by bacterial LPS. Our findings reveal that lactate acts as a critical regulator of rapid neutrophil mobilization from the BM to the circulation induced by LPS. We also found that LPS-induced inflammation enhances glycolytic activities in BM neutrophils that promote lactate production and release. Lactate released from inflammatory BM neutrophils triggers their mobilization acting on its receptor GPR81 functionally expressed by endothelial cells to locally increase vascular permeability. In addition, lactate facilitates rapid neutrophil mobilization from the BM also by elevating the levels of neutrophil-attracting chemokines CXCL1 and CXCL2, and by increasing the release of the cytokine G-CSF. A schematic model summarizing these processes is presented in Fig. 8.
Lactate is not only a waste byproduct of cell metabolism that merely accumulates at inflammatory sites 54 and at the tumor microenvironment, but instead is an effector molecule that contributes to inflammation and to both, onset and progression of cancer 55 . Recently, lactate was also identified as an effector metabolite that triggers multiple cell signaling pathways in various organs 21,22,56,57 . Furthermore, the role of lactate in regulation of immune cells is documented in different diseases, including cancer and multiple sclerosis 55,58-60 . Hence, inhibitors to lactate dehydrogenase and MCTs could be effective targets to attenuate undesirable neutrophil mobilization in pathological inflammatory processes and potentially also in malignancies facilitated by tumor helper neutrophils 61 . Since lactate can also boost TLR4 activation and transcription of pro-inflammatory genes through NF-κB signaling in innate immune cells like macrophages 18 , such inhibitors may be useful in controlling pathologies resulting in excess activation of these subsets.
the reduced levels of mobilized ROS high neutrophils in the blood of NOX −/− mice exposed to LPS.
Distinct metabolic pathways play pivotal roles in immune cell regulation and function. In particular, HIF-1α controls key glycolytic programs in many cell types and different hypoxic environments 32,33,66 . However, LPS can promote glycolysis and induce HIF-1α expression in macrophages even under nonhypoxic conditions. We report that LPS increased glucose uptake and triggered the activities of glycolytic enzymes in BM neutrophils via HIF-1α elevation. Indeed, using mice with a myeloidspecific defect in HIF-1α, we found that this transcription factor is critical for glycolysis and plays essential roles in lactate production and release by BM neutrophils during acute inflammatory responses, key checkpoints in neutrophil mobilization from the BM. Interestingly, in support of our results, intense exercise was reported to elevate plasma lactate levels and to enhance neutrophil-endothelial cell interactions in the muscle microvasculature in a ROS-dependent manner 66 . However, in that study lactate was not confirmed to act as a direct mediator of neutrophil-endothelial interactions. In addition, a recent study reported that facilitating the crosstalk of neutrophils with endothelial cells improves blood vessel regeneration, and hematopoietic recovery 67 . Our results demonstrate that during the early phase of acute inflammation, LPS activates BM neutrophils to produce and release lactate to the BM microenvironment. Lactate within the BM communicates with BM-blood endothelial vessels to open the junctions for neutrophil mobilization and recruitment through GPR81 signaling, to distant infected organs. LPS can mobilize neutrophils via direct and indirect mechanisms, therefore, we cannot rule out that the effects of LPS challenge may also involve other BM cell types which functionally express TLR4. Previously, the neurotransmitter neuropeptide Y (NPY) activation in endothelial cells (ECs) was found to reduce VE-cadherin expression along ECs junctions, resulting in increased vascular permeability and hematopoietic stem and progenitor cell (HSPC) mobilization 42 . Along these lines, we found that mechanistically, lactate signaling via its endothelial GPR81 receptor decreased VE-cadherin expression preferentially by the more permeable sinusoidal BM ECs, which enhances BM vascular permeability and neutrophil mobilization from the BM. Properties of the endothelial barrier are normally maintained by high levels of cAMP. This secondary messenger activates Epac1 and its main downstream GTPase, Rap-1 reported to stabilize VE-cadherin assemblies, which is critical for low vascular permeability. Endothelial barrier disruption by inflammatory stimuli such as bacterial LPS 68 or TNF-α 69 is partially associated with decreased cAMP and Epac1 activity as well as reduced VE-cadherin expression. These mechanisms may act in concert with the lactate-triggered GPR81 signaling axis found by us to acquire the threshold endothelial permeability, critical for neutrophil mobilization from the BM 24,34 .
G-CSF is a major neutrophil-mobilizing cytokine under both basal and stress conditions 2,51,70,71 . G-CSF is also a hematopoietic cytokine, produced by different BM cell types, including ECs 52 , osteoblasts 72 , and osteocytes 73 . In addition, G-CSF has multiple functions including regulation of neutrophil progenitor cell proliferation and differentiation 52 and the functional activation of neutrophils 74 . Several other neutrophil-mobilizing agents are responsible for rapid neutrophil mobilization during the early stage of acute inflammation, the most notable being CXCR2 ligands (CXCL1 and CXCL2 51,75 ). At peripheral sites of inflammation CXCL1 produced by inflamed ECs can also mediate neutrophil adhesion and intraluminal crawling, whereas CXCL2produced by neutrophils is implicated in neutrophil breaching of endothelial-cells junctions 50 . In addition to its direct effect on BM vascular permeability, lactate acts dramatically on BM neutrophil rapid mobilization by elevating the levels of circulating G-CSF and the chemokines CXCL1 and CXCL2, leading to downregulation of CXCR2 expression on blood neutrophils. A recent study 76 reported that G-CSF produced in response to Escherichia coli-challenge, inhibits CXCR2-mediated cellular signaling. The reduction in CXCR2 expression we observed in our model following lactate treatment can therefore, stem from both internalization induced by CXCL1/2 upregulation as well as from G-CSF induced inhibition. Importantly, lactate elevated CXCL1 levels independently of GPR81 signaling and increased G-CSF in both GPR81-dependent and -independent manners. These results may explain the residual lactate-induced neutrophil mobilization we observed in GPR81 −/− mice and why only neutrophils are rapidly mobilized 4 h following lactate treatment despite the increased BM vascular permeability. In light of our results, we hypothesize that lactate induces the release of CXCL1, CXCL2, and G-CSF preferentially via its influx transporter MCT1. In support of our results, a recent study showed that in addition to chemokines release, increased lung vascular permeability can also impact neutrophil trafficking 77 . Our model indicates that lactate acts as a pro-inflammatory factor that enhances neutrophil mobilization via increasing BM vascular permeability and elevating chemokines/G-CSF that are pivotal features of an acute inflammatory response.
In vivo treatments. Sodium-Lactate (Sigma) was injected intraperitoneally (i.p.; 30 or 50.5 mg dissolved in 200 μl PBS, pH-adjusted to~7.0) 4 h or 30 min before mice sacrifice and cells harvest. These lactate doses are non-lethal and do not cause acidosis (Table 1) or severe inflammatory side effects in strains applied for this study 21,55,82 . In order to induce acute inflammation, LPS (from Escherichia coli 0111:B4, L2630, Sigma) was injected i.p. (at 50 µg per injection 83 ) 4 h or 30 min before cells were harvested. This LPS dose is non-lethal and does not cause acidosis (Table 1) in strains used for this study.
We noticed that the number of vehicle injections affects the levels and phenotype of circulating neutrophils. For example, experiments that included four injections of the vehicles PBS or PBS with DMSO, like in Fig. 2. These extra injections dramatically increased the number of total PB neutrophils compared to a single PBS injection. We observed two clear populations of PB neutrophils (CD11b int /Ly6G high and CD11b high /Ly6G high ) following PBS injections but only the CD11b high /Ly6G high population was affected by LPS (increased to 15% and decreased in the presence of the inhibitors; Fig. 2h-k). The CD11b int /Ly6G high population, which increased only after PBS injections (without elevation of CD11b expression, a marker for primed and activated neutrophils following bacterial signals) disappeared after LPS treatment. GPR81 antagonist (3-hydroxy-butyrate; 3-OBA, Sigma, dissolved in PBS) was injected i.p. at 15 mM/mouse three times, at 7 A.M. together with lactate and two additional injections were performed 15 and 30 min following the first injection, to inhibit GPR81 signaling. GPR81 agonist (3,5-dihydroxybenzoic acid; 3,5-DHBA, TOCRIS, dissolved in PBS) was injected i.p. at 30 mg/kg 4 h or 30 min before cells were harvested. Epac1 agonist (8-CPT-2′-O-Me-cAMP; TOCRIS, dissolved in PBS) was injected i.p. at 1 mM/mouse twice, at 7 A.M. together with lactate and one additional injection was performed 1 h following the first injection, to inhibit the BM vascular permeability.
For quantifying the number of liver neutrophils, a similar piece of the liver (size-length: 1.7 × width: 1.2 cm) from the same lobe was taken from each mouse.  Fig. 8 Mechanisms of lactate-induced neutrophil mobilization from the BM. Our model suggests that enhanced lactate-produced by BM neutrophils during bacterial infection induces neutrophil mobilization by modulating metabolic signaling in BM endothelial cells. LPS binds TLR4 expressed on neutrophils that directly activates NADPH oxidase (NOX) and enhances glucose uptake via a glucose transporter 1a. Glucose in turn, is converted to pyruvate by the glycolysis pathway 1b. NOX activity leads to ROS production (2) which elevates HIF-1α expression. HIF-1α in turn, induces downstream expression of LDHA that converts pyruvate to lactate 3a. HIF-1α also up-regulates the lactate transporter MCT4 3b to allow lactate release (4). Under steady-state conditions, surface VE-cadherin on endothelial cells (ECs) is highly expressed, which maintains the endothelial barrier integrity with low permeability (a). During inflammation (b), lactate released from BM neutrophils binds to GPR81 on sinusoidal BM endothelial cells 5a, activates G i protein and thereby reduces cAMP/Epac1 activity 5b. Consequently, lactate via GPR81 signaling decreases surface VE-cadherin expression (5c), leading to higher BM vascular permeability (6). In addition, lactate elevates CXCL1 and G-CSF levels (produced by different cells including ECs 50,52 ) also in a GPR81independent manner 7a with more moderate increases in CXCL2 levels in WT mice (most probably produced by BM neutrophils 50 ). Lactate-induced elevation of CXCL1, CXCL2, and G-CSF downregulates surface CXCR2 expression on PB neutrophils 7b facilitating neutrophil mobilization (8). Taken together, LPS-induced lactate promotes rapid neutrophil mobilization from the BM to the blood (8) preferentially via BM GPR81/VE-cadherin-dependent (5)(6) and also by GPR81-independent (7) pathways.
Gr-1 and CD11b antibodies were used for BM neutrophils detection. Frequency of BM FSC high /SSC high cells reduction was observed following Ly6G depletion ( Supplementary Fig. 1c). In addition, depletion of~85% of BM neutrophils (Gr-1 high /CD11b high ) was also observed following depletion (Supplementary Fig. 1c).
Blood pH measurement. Measurements of blood pH and glucose levels in WT mice following PBS, lactate or LPS treatment, were performed by using i-STAT clinical analyzer (Abaxis). Peripheral blood was collected from the heart using syringes coated with lithium heparin. For the i-STAT analysis 100 μl of blood was immediately placed into a cartridge (i-STAT CG8, PET-VETBIOMED) to determine the parameters following the manufacturer's instructions. The i-STAT machine was calibrated using the Electronic Simulation module before the analysis of samples. Blood glucose levels were used as a control for the quality, and reliability of the machine (test). Both lactate and LPS treatments reduced the blood levels of metabolic glucose without affecting the blood pH (Table 1).
Neutrophil isolation. Flushed BM cells (from WT and mutated mice) were resuspended in 200 μl MACS buffer (2 mM EDTA, 0.5% bovine serum albumin in PBS) and BM neutrophils were isolated according to the protocol of "Neutrophil isolation kit, mouse" by Miltenyi. The yields were approximately 8 × 10 6 cells in total (for 6 bones of two legs), and the purity was >94%.
In vitro treatment for lactate production. BM isolated neutrophils (1 × 10 6 ) were seeded on 24-well plate in serum-free RPMI-1640 culture media and then stimulated with either PBS or 120 ng/ml LPS for 5 h. Cell-free supernatants from nonstimulated or LPS-stimulated cultured neutrophils were harvested at 5 h and analyzed for lactate content using an L-lactate assay (700510, Cayman).
For FACS staining, we used the fluorescence minus one (FMO) as a negative control, which contains all the fluorochromes except for the one that is being measured. The FMO in the manuscript is presented as "secondary only".
BM cell sorting. BM single-cell suspensions were sorted gating on CD45 + / CD11b +/ Ly6G high cells using SORP-FACSAriaII machine with a 100 μm nozzle. Dead cells were excluded based on DAPI incorporation reaching purities greater than 94%.
Cells (9000-35,000 cells per sample) were sorted into Eppendorf tubes containing 100 μl lysis buffer (RLT buffer (QIAGEN, 79216) with 1% 2-Mercaptoethanol). After sorting, lysis buffer was added to the samples to reach 2-4 times of the volume of the sorted cells. Cell lysis was achieved by vortex for 30 s. Samples were snap-frozen on dry ice and stored at −80°C until processed.
Bulk-seq library preparation. Samples were defrosted and washed with AMPure XP bead (BECKMAN COULTER, A63881) at 1:1 ratio. RNA libraries from the bulk tissues were prepared using mcSCRBseq protocol 84 with minor modifications. RT reaction was applied directly on the beads with a final volume of 20 µl. RNase free water (8.4 μl) was added to the beads and mixed with 9.6 μl reaction buffer (1× Maxima H Buffer, 1 mM dNTPs, 2 μM TSO* E5V6NEXT, 7.5% PEG8000, 20U Maxima H enzyme, 1 μl barcoded RT primer). Subsequent steps were applied as mentioned in the protocol. Library final concentration of 2.4 pM was loaded on NextSeq 550 (Illumina) sequencing machine aiming for 20 M reads per sample. Raw files were converted to FASTQ files using bcl2fastq 2.17 (Illumina), to obtain the UMI counts, fastq reads were aligned to the mouse reference genome GRCm38.94 (Ensembl) using zUMI package 85 with the following parameters: RD1 16 bp, RD2 66 bp with a barcode (i7) length of 8 bp. UMI counts for "coding_genes" were processed with the edgeR package running on R 3.5.1. Samples with less than 10k UMI counts were removed. Genes that were expressed in cpm >10 and in two or more samples were retained for the downstream analysis.
Data processing. The final cpm (counts per million) table included 14 mice (n = 8, WT: (n = 3, PBS; n = 5, LPS); n = 6, HIF-1α −/− (n = 3 per group)). We normalized the cpm values by the sum of all genes for each sample. We next divided the gene expression values in each sample by the sum for all genes that individually make up less than 1% of the sample's summed expression values. For each analyzed group, we computed the means and standard errors of the means over the different mice. Only genes with mean values higher than 5 × 10 −6 were considered.
Bacterial preparation. Salmonella typhimurium (strain SL1344) were kindly provided by Roi Avraham (Weizmann Institute, Israel). Salmonella working concentration of 2 × 10 8 cell/ml was prepared from an overnight inoculum in LB (Lysogeny broth) incubated at 37°C with shaking. The bacteria were washed and diluted with PBS to the desired concentration in 200 µl and was administered i.p. at 7 A.M. All the mice were sacrificed at 11 A.M. PBS alone served as vehicle control. This Salmonella dose is nonlethal in strains applied for this study.
L-Lactate measurement. The L-lactate assay was performed by using the L-lactate kit (700510, Cayman) which provides a fluorescence method according to the manufacturer's instructions.
For detecting BM L-lactate, six bones from each mouse were flushed with 1 ml PBS and BM extracellular fluid samples (supernatants) were prepared according to the manufacturer's instructions. BM supernatants of WT, NOX mutated (gp91phox −/− ), and HIF-1α mutated (heterozygous and homozygous) mice were collected 4 h post LPS, and BM supernatants of WT mice collected 4 h post lactate or Salmonella treatments. For detecting blood L-lactate, plasma was collected from WT mice following LPS, lactate or Salmonella treatments and prepared according to the manufacturer's instructions. The levels of BM lactate of WT mice treated with PBS varied between independent experiments (Figs. 1f, 7a, and Supplementary  Fig. 2a), since new L-lactate kits were used in each experiment. However, the trend following LPS or Salmonella treatment was similar in all the experiments.
In this assay, lactate dehydrogenase catalyzes the oxidation of lactate to pyruvate, along with the concomitant reduction of NAD + to NADH. NADH reacts with the fluorescence substrate to yield a highly fluorescent product, which is analyzed with an excitation wavelength of 530-540 nm and an emission wavelength of 585-595 nm. analyzed using an ImageStreamX (Amnis corp., part of EMD-Millipore, Seattle, WA) machine. Samples were visualized and analyzed for the expression of markers with IDEAS 6.2 software (Amnis). Single stained control cells were used to compensate fluorescence between channel images to avoid emission spectra overlap. Cells were gated for single cells with the area and aspect ratio features and for focused cells, using the Gradient RMS feature. Cells were then gated for the selection of positively stained cells based on their pixel intensity, as set by the cutoff with secondary antibody control staining. Three samples from three mice were analyzed to confirm biological repeats of observed data.
Establishment of chimeric mice. Recipient GPR81 −/− , gp91phox −/− , Sca-1 eGFP or WT mice were subjected to irradiation (950 cGY) 16-20 h before transplantation. Irradiated mice received 8-10 million donor total BM cells (filtered and resuspended in RPMI −/− ) and the cells were allowed to repopulate for 4 weeks before conducting experiments. Four weeks after transplantation, recipient (transplanted) mice were treated with lactate or LPS and sacrificed 4 h later at 11 A. M. to determine BM vascular permeability, neutrophil mobilization and metabolic changes in BM neutrophils. For neutrophil imaging, recipient Sca-1 eGFP mice were irradiated (as mentioned above) and received 8-10 million donor total BM cells (Catchup; Ly6G tdTomato). Four weeks after transplantation, chimeric mice were treated with PBS or lactate to track live neutrophil mobilization over time. Sca-1-eGFP transgenic mice were used to distinguish between Sca-1 low sinusoidal BMECs (sBMECs) from Sca-1 + arterial BMECs (aBMECs) 24 .
Intravital imaging by two-photon laser scanning microscopy. Chimeric mice were anesthetized with a mixture of 50 mg ketamine, 15 mg xylazine and 2.5 mg acepromazine per kg of body weight at the indicated time points after lactate or PBS administration. Mice were mounted in a custom-designed heated mouse holder and the skin above the calvaria was cut to reveal the underlying calvarial bone. The mouse was immobilized to the heated holder, a drop of PBS was applied to the skull, covered with a glass coverslip and placed under the microscope waterdipping objective (Zeiss 20 × 1.05 NA plan objective). Zeiss LSM 880 upright microscope fitted with Coherent Chameleon Vision laser was used. Images were acquired with a femtosecond-pulsed two-photon laser tuned to 940 nm. The microscope was fitted with a filter cube containing 565 LPXR to split the emission to a PMT detector (with a 579-631 nm filter for tdTomato fluorescence) and to an additional 505 LPXR mirror to further split the emission to two GaAsp detectors (with a 500-550 nm filter for GFP fluorescence). For in vivo intravital imaging experiments, images were acquired as 70-100 µm z-stacks with 3-5 µm steps between each z-plane (15-19 frames) at 30 s interval for 30-50 min. Static images were acquired as 30-100 µm z-stacks with 1-3 μm steps between each z-plane (20-50 frames). The zoom was set to 0.7, and images were acquired at 512 × 512 xy resolution. Static images were acquired following mice scarification 4 h post treatment. Sca-1 eGFP recipient mice reconstituted with Catchup donor BM (Ly6G tdTomato) cells were used to endogenously label blood vessels (Sca-1 marker distinguish between Sca-1 low sinusoidal BMECs (sBMECs) from Sca-1 + arterial BMECs (aBMECs) 24 ) and neutrophils, respectively throughout all intravital imaging experiments. The second harmonic generation (SHG) signal was used to detect the inner bone surface (endosteum) and detect the BM cavities.
Quantification of the total volume of neutrophils inside the sinus. Quantification of the total volume of neutrophils inside the sinus area over time was done using Imaris software [v9.5.1, Bitplane AG], as followed: Neutrophils express Ly6G tdTomato, while the sinusoidal-and arterial endothelial cells express Sac-1 eGFP with different intensity (sinusoid: Sac-1 low ; arterials: Sca-1 high 24 ). Sca-1 eGFP signal is very bright and bleeds through into the tdTomato channel. To enable detection of neutrophils only, first a new channel was generated by subtracting the Sca-1 eGFP channel from the Ly6G tdTomato channel and then manually masking the new channel in the sinus area using Imaris manual surface option. Imaris Surface was used to segment neutrophils inside the sinus automatically (using local contrast option, fix threshold, and discarding very small objects) and then measured their total volume for all time points.
In vivo EBD penetration to the BM. BM vascular endothelial barrier function was assessed using the EBD assay. Evans Blue (Sigma) dissolved with PBS was filtered and injected i.v. to mice. For the lactate, GPR81 agonist, Epac1 agonist, and LPS treated mice; 20 mg/ml EBD was injected 4 h before mice were sacrificed. For neutrophil depletion, the anti-Ly6G antibody was administered i.p. at 24 h prior to EBD and LPS injections. In each experiment, a non-injected mouse was used for control blank measurements. Subsequently, mice were perfused with PBS via the left ventricle to remove the intravascular dye. Femurs were removed and formamide was used for bone flushing, crushing and chopping. EBD was extracted in formamide by incubation and shaking of flushed and crushed fractions, overnight at 60°C. After 30 min centrifugation at 2000g, EBD in BM supernatants was quantitated by spectrophotometric analysis at 620 nm.
Immunofluorescence. Femurs fixed overnight at 4°C in 4% paraformaldehyde followed by 30% sucrose exchange for 2 days, were embedded in optimal cutting temperature compound (Sakura Finetek USA, Inc., Tissue-Tek) and 'snap-frozen' in N-methylbutane chilled in liquid nitrogen. Sections of 8μm thickness were generated with a cryostat (Leica) at −24°C with a tungsten carbide blade (Leica) and a CryoJane tape transfer system (Instrumedics). Sections were mounted on adhesive-coated slides (Leica), fixed in acetone and air-dried. Sections were permeabilized with 0.1% Triton for 8 min, blocked in 20% normal horse serum (Vector) and stained overnight at 4°C with primary antibodies. Then sections were incubated for 1 h with secondary antibodies and stained for 5 min in RT with Hoechst33342 (Molecular Probes). Endomucin was stained using rat anti-mouse Endomucin (1:50, clone V.7C7, Santa Cruz), followed by donkey anti-rat Alexa 488 (1:400, Jackson ImmunoResearch). Femoral sections from GPR81-RFP and WT mice were stained with rabbit anti-mouse RFP (1:100, MBL), followed by donkey anti-rabbit Rhodamin Red X conjugated antibody (1:200, Jackson ImmunoResearch).
Confocal imaging was performed using an upright Leica TCS SP8, equipped with internal Hybrid (HyD) detectors and Acusto Optical Tunable Filter (Leica microsystems CMS GmbH, Germany). Excitation of Rhodamin Red X was done at a wavelength of 561 nm by DPSS laser. Excitation of Alexa488 was done at a wavelength of 488 nm by Argon laser. Excitation of Hoechst33342 was done by Diode405 laser. Emission signal for Rhodamin Red X, Alexa488 and Hoechst was collected using the HyD detectors at the range of 588-640, 497-555, and 413-501 nm, respectively. Images were acquired using the 8 k Hz resonant/galvometric scanner in a 1024 × 1024 format using 20X air objective (HC PL APO 20×/0.75 CS2) and application zoom factor of 7 or 63× oil immersion objective (HC PL APO 63×/1.40 OIL CS2). Z stack acquisition was performed using 14 z steps while only one slice was selected for presentation. The presented slice has pixel width 0.077 µm, pixel height 0.077 µm, and voxel depth 0.999 µm. The acquired images were processed using ImageJ software to enhance the contrast.
Chemokine quantification in BM and plasma. CXCL1 and CXCL2 protein amount were measured in BM samples. CXCL1, CXCL2, and G-CSF were measured in plasma samples taken 4 h post lactate injection using commercial ELISA reagents, following the manufacturer's protocol (R&D Systems). Total protein content in BM fluids was quantified by Bradford assay. CXCL1/2 concentration in the BM fluids was normalized per protein content.
Statistical analyses. All statistical analyses were conducted with Prism 8.0c version ( * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, n.s. represents nonsignificant). All data are expressed and presented as mean ± standard error of the mean (s.e.m.) and all n numbers represent biological repeats. Unless indicated otherwise in figure legends, a Student's two-tailed unpaired t test was used to determine the significance of the difference between means of two groups. Oneway ANOVA or two-way ANOVA was used to compare means among three or more independent groups. Tukey or Bonferroni post hoc tests were used to compare all pairs of treatment groups when the overall P value was <0.05. All mutated or transgenic mice were assigned according to their genotype. Litter mates and sex-matched animals were used whenever possible. All other parameters are random. The authors were not blinded to allocation during experiments.