A neutrophil - B-cell axis governs disease tolerance during sepsis via Cxcr4

Sepsis is a life-threatening condition characterized by uncontrolled systemic inflammation and coagulation, leading to multi-organ failure. Therapeutic options to prevent sepsis-associated immunopathology remain scarce. Here, we established a model of long-lasting disease tolerance during severe sepsis, manifested by diminished immunothrombosis and organ damage in spite of a high pathogen burden. We found that, both neutrophils and B cells emerged as key regulators of tissue integrity. Enduring changes in the transcriptional profile of neutrophils, included upregulated Cxcr4 expression in protected, tolerant hosts. Neutrophil Cxcr4 upregulation required the presence of B cells, suggesting that B cells promoted tissue tolerance by suppressing tissue damaging properties of neutrophils. Finally, therapeutic administration of a Cxcr4 agonist successfully promoted tissue tolerance and prevented liver damage during sepsis. Our findings highlight the importance of a critical B-cell/neutrophil interaction during sepsis and establish neutrophil Cxcr4 activation as a potential means to promote disease tolerance during sepsis. Summary We show that a B cell/neutrophil interaction in the bone marrow facilitates tissue tolerance during severe sepsis. By affecting neutrophil Cxcr4 expression, B cells can impact neutrophil effector functions. Finally, therapeutic activation of Cxcr4 successfully promoted tissue tolerance and prevented liver damage during sepsis.


Introduction 1
Sepsis is a life-threatening condition triggered by severe infections with bacteria, viruses or fungi.

2
In spite of the successful use of antimicrobial therapies, mortality rates remain high with up to 3 50%, (1,2). The main determinant of sepsis-associated mortality is rarely the pathogen, but 4 instead the combination of dysregulated systemic inflammation, immune paralysis and hemostatic 5 abnormalities that together cause multi-organ failure (3). Upon pathogen sensing, ensuing 6 inflammation promotes the activation of coagulation, which in turn generates factors that further 7 amplify inflammation, thus creating a vicious, self-amplifying cycle. These events result in systemic 8 inflammation and the widespread formation of microvascular thrombi that together cause vascular 9 leak, occlusion of small vessels and ultimately multi-organ failure (4,5). Whether a patient 10 suffering from sepsis enters this fatal circuit of immunopathology or instead is able to maintain 11 vital organ functions and survives sepsis is not well understood (6-8).

13
The concept of "disease tolerance" describes a poorly studied, yet essential host defense strategy,

20
DNA damage response, tissue remodeling or oxidative stress (10). However, little is known about 21 the specific contribution of immune cells to disease tolerance during severe infections, and 22 therapeutic options to increase disease tolerance are limited due to a lack of knowledge about 23 detailed molecular and cellular tolerance mechanisms (6)(7)(8).

25
In this study, we investigated mechanisms of disease tolerance, by comparing tolerant and 26 sensitive hosts during a severe bacterial infection. While sensitive animals developed severe 27 coagulopathy and tissue damage during sepsis, tolerant animals were able to maintain tissue 28 integrity in spite of a high bacterial load. Tolerance was induced by the prior exposure of animals 29 to a single, low-dose of LPS and could be uncoupled from LPS-induced suppression of cytokine 30 responses. We provide evidence for a deleterious and organ-damaging interaction between B 31 cells and neutrophils during sepsis in sensitive animals, while in tolerant animals neutrophils and 32 B cells jointly orchestrated tissue protection during sepsis, which was associated with 33 transcriptional reprogramming of neutrophils and B cell dependent upregulation of neutrophil Cxcr4. Our data suggest that B cells can modulate the tissue damaging properties of neutrophils 1 by influencing neutrophil Cxcr4 signaling. Consequently, the administration of a Cxcr4 agonist 2 prevented sepsis-associated microthrombosis and resulting tissue damage, thereby exposing a 3 potential therapeutic strategy to foster tissue tolerance in severe sepsis.

LPS pre-exposure induces long-term tissue tolerance during Gram-negative sepsis 2
To establish a model for the study of tissue tolerance during sepsis we challenged mice 3 intravenously (i.v.) with a subclinical dose of LPS 1 day, 2 weeks, 5 weeks or 8 weeks, 4 respectively, prior to the induction of Gram-negative sepsis by intraperitoneal (i.p.) injection of the 5 virulent E. coli strain O18:K1. While LPS pretreatment 24h prior to infection significantly improved 6 pathogen clearance, any longer period (i.e. 2-8 weeks) between LPS administration and infection 7 did not affect the bacterial load when compared to control mice ( Figure 1A, Figure S1A).

8
Importantly though, all LPS pre-treated groups were substantially protected from sepsis-9 associated tissue damage, illustrated by the absence of elevated liver transaminase (ASAT and 10 ALAT) plasma levels ( Figure 1B). Thus, short-term (24h) LPS pre-exposure improved resistance 11 to infection and consequently tissue integrity, while long-term (2-8 weeks) LPS pre-exposure 12 enabled the maintenance of tissue integrity irrespective of a high bacterial load, which per 13 definition resembles disease tolerance.
14 To dissect the underlying mechanism of tissue tolerance, we thus performed all subsequent 15 experiments by treating mice with either LPS or saline two weeks prior to bacterial infection, 16 allowing us to compare tolerant with sensitive hosts. Mice were either sacrificed two weeks after 17 LPS pretreatment to assess changes in tolerant hosts prior to infection, or six to 18h after E. coli 18 infection to determine early (6h) or late inflammation and organ damage (18h), respectively, during 19 sepsis (Fig. 1C). Doing so, we observed that organ protection ( Figure 1B Figure S1B and 1C). A major cause of organ damage during sepsis is the disseminated activation 23 of coagulation, which is characterized by systemic deposition of micro-thrombi and substantial 24 platelet consumption, resulting in a critical reduction in tissue perfusion (4)(5)(6). While we discovered 25 a severe decline in platelet numbers upon E. coli infection in sensitive mice, tolerant mice 26 maintained significantly higher blood platelet counts ( Figure 1G) and, in sharp contrast to sensitive 27 animals, showed almost no deposition of micro-thrombi in liver ( Figure 1H and 1I) and lung 28 sections ( Figure S1D), indicating that tissue tolerance occurred systemic and not organ specific.

29
Considering that LPS exposure itself can impact coagulation factor levels and blood platelet 30 numbers (11,12), we importantly found similar platelet counts in sensitive and tolerant mice at the 31 onset of E. coli infection (2 weeks post LPS) ( Figure 1G). In addition, we did not detect any 32 indication for an altered coagulation potential in tolerant mice before sepsis induction, as both 33 groups showed a similar plasma thrombin generation potential prior to infection ( Figure 1J left panel, Figure S1E). However, compared to sensitive animals, the thrombin generation capacity 1 was only preserved in tolerant mice after infection (18h p.i.), suggesting that tolerance 2 mechanisms prevented sepsis-associated consumption coagulopathy (Figure 1J right panel and   3   1K). Taken together, low-dose LPS pretreatment prevented the formation of micro-thrombi and 4 induced a long-lasting state of tissue tolerance during subsequent sepsis.

29
These findings indicated that B cells, but not T cells, played an ambiguous role as they were 30 involved in both, sepsis-associated organ damage and the establishment of LPS-triggered tissue 31 tolerance. We then tested if splenectomy would replicate the protective effects of B cell deficiency 32 during sepsis and interestingly found that splenectomy was associated with reduced liver damage 33 in naïve, sensitive mice, but, in contrast to complete lymphocyte deficiency, not sufficient to abrogate LPS-induced tissue protection in tolerant animals ( Figure 2G and S2F). This suggested 1 that mature splenic B cells contributed to tissue damage during severe infections, while other, not 2 spleen derived, B cell compartments were instrumental in driving disease tolerance.

3
Given that B cells were shown to promote early production of proinflammatory cytokines such as 4 IL-6 during sepsis in a type I IFN dependent manner (13), we next investigated if LPS pretreatment    cytokine production during endotoxin tolerance in vitro (16,17). These data suggested that in 15 tolerant hosts, B cells contributed to tissue protection during sepsis, and that an LPS mediated 16 modulation of early inflammation is unlikely to explain these protective effects.

16
(H-I) IL-6 levels in plasma and liver of NaCl or LPS pretreated wildtype or Rag2 -/mice at 6h p.i.

29
it seemed counterintuitive at first, that the absence of neutrophils or B cells, respectively, 30 prevented tissue damage in a primary infection, while they at the same time seemed critical for 31 tissue protection in a model of LPS-induced tolerance. We thus hypothesized that B1 and B1-like 32 cells, in contrast to B2 cells, reduced neutrophil's tissue damaging effector functions. Using sIgM 33 deficient mice, enabled us to rule out a major role for IgM in tissue tolerance during sepsis, even though IgM was reported to exhibit anti-thrombotic functions in cardiovascular diseases (39) and 1 high plasma IgM levels positively correlate with a better outcome in human sepsis (24) and mouse 2 models (23). However, while sIgM deficiency did not prevent LPS-induced tolerance, naïve sIgM -3 /mice developed less organ damage during primary sepsis as compared to control animals. As 4 sIgM deficiency goes along with a decreased abundance of B2 and an increased abundance of 5 B1 cells (40) this further supported the notion of tissue damaging B2, and tissue protective B1 6 cells.

8
Since we discovered that LPS-induced protection was still observed in splenectomized animals,  to the periphery as well as their homing back to the bone marrow when they become senescent 27 (28, 29). Importantly, Cxcr4 signaling is essential, as Cxcr4 knockout mice die perinatally due to 28 severe developmental defects ranging from virtually absent myelopoiesis and impaired B 29 lymphopoiesis to abnormal brain development (43). Antagonizing SDF1/Cxcr4 signaling is 30 approved for stem cell mobilization from the bone marrow and is under extensive research in 31 oncology, as it is critical for tumor development, metastasis and tumor cell migration (44). More recently, Cxcr4 signaling was described to delay neutrophil aging and to protect from vascular effects of upregulated Cxcr4 on neutrophils in sepsis. Strikingly, activating, but not antagonizing, 1 Cxcr4 during sepsis induced tissue tolerance, suggesting that B cell driven regulation of Cxcr4 is 2 a potential mechanism of disease tolerance and thus might be an interesting therapeutic target 3 during severe sepsis.    Greenberger lysis buffer (300mMol NaCl, 30mMol Tris, 2mMol MgCl2, 2mMol CaCl2, 1% Triton X-21 100, 2% protease Inhibitor cocktail) (61), and supernatants were stored at -20°C. For RNA 22 isolation, lysates were stored in RLT buffer (Qiagen, containing 1% β-mercaptoethanol) at -80°C.

23
Pathogen burden was evaluated in organ homogenates by plating serial dilutions on blood agar 24 plates (Biomerieux), as previously described (57). Blood platelet counts were determined in freshly 25 isolated anticoagulated EDTA blood using a VetABC differential blood cell counter. Liver

4
CD4 + and CD8 + T cell depletion was performed by i.v. administration of anti CD4 (200μg/mouse) 5 or anti CD8 (400μg/mouse) antibodies 36h prior LPS treatment and repeated every three days

Cell transfers and splenectomy 13
Splenocytes were isolated from naïve WT C57BL/6 mice and i.v. injected into Rag2 deficient mice or LPS and two weeks later, challenged with E. coli as described above. Resting B cells were isolated from spleens of naïve UBC-GFP mice using magnetic beads (Milteny Biotec, Mouse B 18 cell isolation kit) and i.v. injected into Rag2 deficient mice (5 x 10 6 cells/mouse) after erythrocyte 19 lysis (ACK lysis buffer) two weeks and four days prior to LPS/NaCl treatment. After pretreatment 20 with NaCl or LPS transplanted animals were challenged with E. coli as described above. Mice 21 were splenectomized or sham operated as described previously (62) and after 1 week recovery, treated with NaCl/LPS and challenged with E. coli as described above.

23
In vitro thrombin-generation assay 24 Thrombin generation was assayed according to the manufacturer's instruction (Technoclone).

Flow cytometry 32
Splenocytes were isolated by passaging spleens through 70μm cell strainers and after erythrocyte followed by filtering through 70μm cell strainers. Cells were counted using a CASY cell counter 1 and after unspecific binding was blocked using mouse IgG (Invitrogen), cells were stained in PBS 2 containing 2% FCS using antibodies (see table) against mouse CD45, CD3, CD19, CD23, IgM, 3 CD21, CD43, CD11b and Ly-6G. This was followed by incubation with a Fixable Viability Dye 4 eFluor 780 (eBioscience) according to the manufacturer's instructions to determine cell viability.

5
After several washing steps, cells were fixed (An der Grub Fix A reagent) and analyzed via flow 6 cytometry using a BD LSRFortessa™ X-20 cell analyzer.   Liver sections (4 μm) were stained with H&E and analyzed by a trained pathologist in a blinded 2 fashion according to a scoring scheme, involving necrosis, sinusoidal-and lobular inflammation, 3 steatosis and endothelial inflammation (0 representing absent, 1 mild, 2 moderate, and 3 severe).

4
The sum of all parameters indicated the total histology score. After staining for fresh fibrin (MSB 5 stain, performed at the routine laboratory at Newcastle University), samples were scored for the 6 presence of microthrombi by a trained pathologist in a blinded fashion. NIMPR1 immunostaining 7 was performed on paraffin-embedded liver sections as described earlier (64). Briefly, antigen 8 retrieval was achieved using a citrate-based buffer at pH 6.0 (Vector laboratories), followed by 9 several blocking steps. Incubation with anti-NIMP-R14 antibody (Abcam) was performed at 4°C,

Statistical analysis 19
Statistical evaluation was performed using GraphPad Prism software except for statistical analysis 20 of RNA sequencing data, which was performed using R. Data are represented as mean ± SEM 21 and were analyzed using either Student´s t-test, comparing two groups, or one-way ANOVA analysis, followed by Tukey multiple comparison test, for more than two groups. Differences with 23 a p-value ≤ 0.05 were considered significant. For DEG, genes with an FDR-adjusted p value of < 24 0.1 were considered differentially expressed.

Declaration of interests 19
The authors declare no financial or commercial conflict of interest.                                  Data in (A-E) are from a single experiment (n = 4-5/group) and data in (F) are from a different 1 single experiment (n = 3-4/group). All data are presented as mean +/-SEM. * p ≤ 0.05.