key: cord-0291624-dv49g9kd
authors: Louwe, P. A.; Gomez, L. Badiola; Webster, H.; Perona-Wright, G.; Bain, C. C.; Forbes, S. J.; Jenkins, S. J.
title: Recruited macrophages that colonise the post-inflammatory peritoneal niche convert into functionally divergent resident cells
date: 2020-12-02
journal: bioRxiv
DOI: 10.1101/2020.11.30.404988
sha: 075aa12abb666710cf6283cb18b1fcec6162f5ab
doc_id: 291624
cord_uid: dv49g9kd

Inflammation generally leads to substantial recruitment of monocyte-derived macrophages. What regulates the fate of these cells and to what extent they can assume the identity and function of resident macrophages remains unclear. We compared the normal fate of inflammation-elicited macrophages in the peritoneal cavity with their potential under non-inflamed conditions and in the absence of established resident macrophages. Following mild inflammation, elicited macrophages persisted for at least 5 months but failed to fully assume a GATA6hi resident identity due to the presence of enduring resident cells. In contrast, severe inflammation resulted in ablation of resident macrophages and a protracted phase wherein the cavity was incapable of sustaining a resident phenotype, yet ultimately elicited cells acquired a mature GATA6hi identity. Elicited macrophages also exhibited divergent features resulting from inflammation-driven alterations to the peritoneal cavity micro-environment and environment-independent features related to origin and time-of-residency. Critically, one environment-dependent feature of inflammation-elicited macrophages irrespective of severity of inflammation was a failure to produce the chemokine CXCL13, which correlated with a progressive loss in accumulation of peritoneal B1 cells post-inflammation. Hence, rather than being predetermined, the fate of inflammation-elicited peritoneal macrophages appears largely regulated by environment, changes in which result in long-term alteration in function of the peritoneal macrophage compartment post-inflammation.

inflammation-elicited macrophages in the peritoneal cavity with their potential under non-23 inflamed conditions and in the absence of established resident macrophages. Following mild 24 inflammation, elicited macrophages persisted for at least 5 months but failed to fully assume 25 a GATA6 hi resident identity due to the presence of enduring resident cells. In contrast, severe 26 inflammation resulted in ablation of resident macrophages and a protracted phase wherein 27 the cavity was incapable of sustaining a resident phenotype, yet ultimately elicited cells 28 acquired a mature GATA6 hi identity. Elicited macrophages also exhibited divergent features 29 resulting from inflammation-driven alterations to the peritoneal cavity micro-environment 30 and environment-independent features related to origin and time-of-residency. Critically, 31 one environment-dependent feature of inflammation-elicited macrophages irrespective of 32 severity of inflammation was a failure to produce the chemokine CXCL13, which correlated 33 with a progressive loss in accumulation of peritoneal B1 cells post-inflammation. Hence, 34 rather than being predetermined, the fate of inflammation-elicited peritoneal macrophages 35

Inflammation radically alters the composition and function of the tissue macrophage 41 compartment, typically leading to substantial recruitment of monocytes from the blood and 42 activation or even loss of the tissue resident cells 1 . While these early cellular processes have 43 been well characterized across various tissues and models of inflammation, it remains poorly 44 understood how homeostasis within the macrophage compartment is reinstated post 45

inflammation and consequently what long-term effects inflammation may have on tissue 46 macrophage function. 47

In the steady-state resident macrophages across tissues share expression of core lineage-49 related genes upon which a tissue-specific transcriptional, epigenetic and functional identity 50 is overlaid 2-5 . These unique molecular identities are largely established upon exposure to 51 tissue-specific environmental signals that in turn drive expression of tissue-specific 52 transcription factors. Tissue resident macrophages also have diverse developmental origins 6,7 53 with many tissues containing self-renewing populations largely seeded during embryogenesis 54 and short-lived bone marrow (BM)-derived populations that seemingly inhabit distinct 55 anatomical regions 8, 9 . However, in the absence of embryonically-seeded macrophages, 56 circulating monocytes appear able to give rise to long-lived and transcriptionally and 57 functionally normal resident cells 10-14 , suggesting origin may not be a key determinant of 58 macrophage identity per se, but rather tissue-specific anatomically-restricted signals and cell 59 interactions constitute a 'niche' that controls macrophage longevity and gene expression. 60

Whilst a small number of seemingly ontogeny-related transcriptional differences may 61 distinguish embryonic and monocyte-derived resident macrophages present within the same 62 'niche' 12,14 , limited evidence suggests that even these may be gradually reprogrammed over 112

Here we studied the peritoneal cavity, a clinically relevant site commonly used to model 113 inflammatory processes, to investigate what regulates the fate of inflammatory macrophages 114 following sterile inflammation. Using adoptive transfer to unequivocally track inflammatory 115 macrophages and determine the degree to which the environment dictates the fate of these 116 cells, we demonstrate that macrophages infiltrating the cavity following mild inflammation 117 persist long-term but that competition with incumbent resident macrophages inhibits 118 effective acquisition of a mature resident phenotype. Consistent with this competition model, 119

severe inflammation, which caused ablation of incumbent resident macrophages resulted in 120 conversion of inflammatory macrophages to mature resident cells. We therefore reveal the 121 existence of a 'biochemical niche' for resident peritoneal macrophages. Competition for the 122 'niche' largely dictates the capacity of monocyte-derived cells to undergo conversion to 123 mature resident macrophages and a failure to compete retains them in a highly proliferative 124 and immunoregulatory state. 125 126 127

Results 128

To investigate what regulates the fate and phenotype of inflammatory and resident 130 macrophages following resolution of sterile peritoneal inflammation, we used a well-131 characterised model of intraperitoneal injection with low dose zymosan A (10µg/mouse), in 132 which both populations remain present following resolution of the neutrophilic phase 36,37,40 . 133 First, to definitively delineate incumbent resident cells from inflammatory macrophages 134 recruited during the acute phase of inflammation, we utilized an established method of injecting fluorescent PKH26-PCL dye intraperitoneally 24hrs before zymosan to exclusively 136 label peritoneal phagocytes present prior to inflammation 38 . Uptake of PKH26-PCL dye was 137 largely restricted to all resident LPM and most SPM (Supplementary Figure 1a) Figure 1e) , features known to differentiate 153 inflammatory from resident macrophages during resolution 31,36 . Thus, PKH26-PCL-labelling 154 faithfully delineated resident vs recruited macrophage subsets and importantly, this system 155 used a minimal number of surface antibodies thereby circumventing potential confounding 156 effects of adoptive transfer of antibody-coated cells. 157

Next, we used adoptive transfer to unequivocally determine the fate of these populations. 159

Specifically, Dye Hi F4/80 Hi resident macrophages (RMac Z10 ) and Dye Lo F4/80 Int inflammatory 160 macrophages (IMac Z10 ) were FACS-purified from C57BL/6 WT (CD45.2 + ) donor mice 3 days 161 after injection of low dose zymosan (Figure 1a ) and transferred intraperitoneally into 162 separate congenic WT (CD45.1/2 + ) host animals. The recipient mice had been pre-treated 3 163 days prior with an equivalent dose of zymosan to ensure labelled cells were transferred into 164 a similar environment (Figure ci) . Eight days post-transfer, transferred donor RMac Z10 and 165 IMac Z10 exhibited a similar degree of engraftment, defined as the number retrieved as a 166

proportion of those transferred, although this was somewhat greater for RMac Z10 (Figure 1d) . 167

Whereas transferred RMac Z10 remained predominantly MHCII Lo , IMac Z10 remained largely 168 MHCII Hi and continued to express marginally less F4/80 such that the two donor populations 169 were identified with relative accuracy using these markers (Figure 1e) . Critically, virtually all 170 transferred IMac Z10 expressed the LPM-specific transcription factor GATA6 but at markedly 171 lower levels than RMac Z10 (Supplementary Figure 2a,b) . The host CD11b + myeloid 172 compartment also contained a mixture of F4/80 Int/Hi MHCII Hi GATA6 + and F4/80 Hi MHCII Lo 173 GATA6 + macrophages, consistent with persistence of endogenous inflammatory and resident 174 macrophages, but also a minor fraction of GATA6 -F4/80 Lo MHCII Hi cells (Figure 1e ; 175 Supplementary Figure 2c) suggestive of newly generated SPM and/or CD11b + DCs. Hence, by 176 combining dye-labelling and adoptive transfer, we developed a robust system to identify and 177 fate map tissue resident and inflammatory macrophages in the context of peritoneal 178 inflammation. Furthermore, our data reveal that the distinct populations of MHCIIand 179 MHCII + peritoneal macrophages present following zymosan-induced peritoneal 180 inflammation 36 arise from persistence of tissue-resident macrophages established prior to 181 inflammation and monocyte-derived macrophages recruited at the onset of inflammation, 182 respectively. 183 184

We next explored what regulates the short-term fate of these cells. First, we transferred 186 RMac Z10 and IMac Z10 into naïve recipient mice (Figure 1cii ) to determine whether their 187 survival and phenotype is dictated primarily by the post-inflammation micro-environment. In of IMac Z10 adopted a more resident-like MHCII Lo phenotype (Figure 1i ) but failed to acquire 202 similar levels of GATA6 as RMac Z10 within this period (Supplementary Figure 2f) . Surprisingly, 203 although RMac Z10 also persisted better in the depleted environment, with an engraftment 204 efficiency nearer 100%, they were unable to expand to the same degree as IMac Z10 ( Figure 1h ). Notably, host macrophages also repopulated the cavity during this period, yet they 206 largely exhibited an MHCII Hi phenotype (Figure 1i ) resembling that of IMac Z10 in their native 207 inflamed environment, suggesting these cells likely derive from Ly6C + monocytes recruited to 208 the cavity post-depletion (Supplementary Figure 2e) . We also found that irrespective of 209 environment, nearly all IMac Z10 expressed the GATA6-independent LPM marker CD102 5 yet 210 few expressed Tim4 (Supplementary Figure 2g) . Together, these data suggest that while 211 IMac Z10 persist through the early phases of resolution, their survival and conversion to 212 MHCII Lo cells is largely regulated by the presence of competing resident macrophages. for 24 hours to determine whether MHCII expression by inflammatory macrophages is 219 responsive to retinoic acid or other omental factors. As CD102 and Tim4 did not appear to be 220 altered in any of the in vivo experiments we used these to identify resident (CD102 + /Tim4 + ) 221 and inflammatory CD102 + Tim4 -) macrophages post culture. Indeed, post-culture and 222 expression of these surface markers remained unchanged between treatments (Figure 1j ). 223

Culture with ATRA led to increased expression of the GATA6 responsive marker F4/80 5 by 224 CD102 + /Tim4 + resident and CD102 + Tim4inflammatory macrophages but not down-225 regulation of their MHCII expression. In contrast, culture with omental supernatant led to 226 downregulation of MHCII by CD102 + Tim4inflammatory macrophages (Figure 1k) Figure 3g) . Furthermore, 280 comparison with our published single cell RNAseq analysis 15 of LPM revealed that genes 281 expressed more highly in IMac Z10 overlapped exclusively with those expressed more highly 282 by LPM of recent monocyte origin in naïve female mice (e.g. Apoe, Retnla, and genes related 283 to MHCII presentation), whereas genes expressed more highly in RMac Z10 overlapped 284 exclusively with those expressed more highly by the most long-lived LPM (e.g. Table 4 ). Hence, these data suggest the majority of transcriptional 303 differences between IMac Z10 and RMac Z10 are determined by the post-inflammatory 304 environment or competition with incumbent resident macrophages for access to niche 305 signals, whereas a smaller number of differentially expressed genes may represent cell-306 intrinsic features related to origin. Specifically, enduring resident macrophages seemingly 307 prevent IMac Z10 transition to a mature GATA6 hi phenotype, thus retaining them in a 308 transcriptional state associated with steady-state monocyte-derived LPM. To determine whether colonizing inflammatory macrophages differ functionally and 355 behaviourally to established resident macrophages, we developed a gating strategy based on 356 a Tim4 + Sema4a -(R1) and Tim4 -Sema4a + (R3) profile to identify the majority of RMac Z10 and 357 IMac Z10 , respectively (Figure 3a,b) . Using this approach, we were able to track the major long-358 term changes in phenotype of the resident LPM pool triggered by inflammation without need 359 for dye-based fate-mapping (Supplementary Figure 4b) . In addition, to determine whether 360

IMac-derived LPM are functionally similar to LPM of recent monocyte origin recruited during 361 homeostasis, we confirmed that the Tim4 -Sema4a + gate identified the majority of Tim4 -362 MHCII + LPM in naïve mice (Supplementary Figure 4c) , which we previously validated to 363 identify newly monocyte-derived LPM 15 . 364 365 Consistent with our previous observations showing that LPM of recent monocyte-origin 366 proliferate more than established LPM during homeostasis 20 , the Sema4a + Tim4fraction of 367 LPM from naïve mice (subsequently referred to as Mo-LPM and RM-LPM respectively) 368 exhibited the highest level of proliferation, as determined by Ki67 expression (Figure 3c) . 369

Similarly, Sema4a + Tim4and Sema4 -Tim4 + defined-populations found 8wks post-zymosan 370

injection (subsequently referred to as Mo Z10 -LPM and RM Z10 -LPM respectively) exhibited the 371 same divergent pattern in proliferative activity (Figure 3c) . Furthermore, while both Mo Z10 -LPM and RM Z10 -LPM displayed typical macrophage morphology, the cytoplasm of RM Z10 -LPM 373 contained many more vacuoles (Figure 3d ) indicative of greater phagocytic activity. Indeed, 374

both Mo Z10 -LPM and Mo-LPM had appreciably lower side-scatter characteristics than their 375 RM counterparts, albeit higher than SPM (Figure 3e) . Moreover, examination of phagocytic 376 potential in vitro using pHrodo-labelled Escherichia coli particles revealed that Tim4 + LPM 377 from naïve mice and 8wks after inflammation were significantly more phagocytic than the 378 Tim4fraction (Figure 3f ). Of note, incubation at 37 o C for 1hr caused rapid acquisition of 379 surface Sema4a by Tim4 + macrophages thereby preventing analysis of Sema4a-defined 380 populations in this assay. Furthermore, re-analysis of our short-term transfer experiments 381 revealed that only Tim4 + recipient LPM acquired PKH26-PCL dye from donor RMac 382 irrespective of whether recipients were naïve or zymosan-injected (Supplementary Figure  383 4d,e) suggesting up-take of dying donor cells is restricted to Tim4 + cells. Lastly, to test 384 responsiveness to challenge, RM-LPM and Mo-LPM from naïve mice and RM Z10 -LPM and 385

Mo Z10 -LPM were purified 8wks post-zymosan injection, exposed in vitro to LPS and cytokine 386 and chemokine production assessed by multiplex assay. The overall response of Mo Z10 -LPM 387 and Mo-LPM compared to their RM counterparts was remarkably similar; both produced 388 higher levels of IL-10 and somewhat more IL-1b and GM-CSF and less CXCL10 and TNFa 389 (Figure 3g,h) , suggesting these are common features of monocyte-derived LPM. 390

Furthermore, direct comparison confirmed that despite some subtle differences, Mo-LPM 391 and Mo Z10 -LPM produced largely similar quantities of cytokines and chemokines, as did RM-392 LPM compared with RM Z10 -LPM (Supplementary Figure 4f,g) . Hence, together with our gene 393 expression profiling, these data suggest that recency-of-monocyte origin more strongly 394 influences the behaviour of LPM than prior experience of inflammation and that persistence 395 of inflammatory macrophages leads to the expansion of a normally minor subset of IL-10 producing monocyte-derived LPM present under homeostatic conditions. Finally, we found 397 that purified Mo Z10 -LPM transferred into naïve recipient mice that subsequently received LPS 398 produced less TNFa than transferred RM Z10 -LPM (Figure 3i,j) , confirming these cells also 399 respond differently to challenge in vivo. 400 401

In the mild model of sterile peritonitis studied so far, the initial macrophage 'disappearance 403 reaction' and inflammatory response that occurs is relatively limited and transient 36 surviving IMac Z1000 in the high dose environment adopted a F4/80 Hi GATA6 + profile by 8wks 424 following severe inflammation and almost none subsequently expressed CCR5 or FRb ( Figure  425   4c,d,e) . These results are consistent with a more mature phenotype, again reflecting more 426 rapid differentiation in the absence of competition from enduring resident macrophages. 427

Nevertheless, a shared deficiency of IMac-derived cells in both high and low-dose 428 environments was the failure to produce CXCL13, a GATA6 21,22 and Rxra 25 independent 429 feature of LPMs. Thus, these data suggest impaired CXCL13 expression by IMac arises from 430 long-term alterations in the LPM niche that occurs irrespective of inflammation severity and 431 retinoic acid production. IMac Z1000 and IMac Z10 derived cells also largely shared the propensity 432 to express the intrinsic marker of monocyte-derived LPM Sema4a (Figure 4e) , and to lack 433 expression of the environment-independent but time-dependent marker VSIG4 (Figure 4e) . 434

More surprisingly, IMac Z1000 retained high levels of MHCII, and largely expressed the 435 otherwise time-dependent marker Tim4 (Figure 4e) . Furthermore, donor-derived 436 macrophages following severe inflammation almost perfectly resembled the phenotype of 437 host cells, consistent with their likely uniform origin from inflammatory macrophages 438 Figure 5e) . In contrast, in the lower dose environment host macrophages 439 neither aligned with RMac Z10 nor IMac Z10 but corresponded to a mixed population of these 440 cells, (Figure 4f, Supplementary Figure 5e ) re-emphasising that phenotype is ontogeny-441 restricted in this environment. Consequently, the LPM compartment on the whole 8 weeks 442 after high dose zymosan differed markedly to that after low dose zymosan for each marker 443 assessed (Supplementary Figure 5e) .

As both F4/80 and MHCII expression by IMac Z10 were rapidly responsive to niche signals and 446 competition with LPM after low dose zymosan (Figure 1k, Figure 2d) Transient peritoneal inflammation has lasting consequences for the incidence and severity of 486 future disease 38,46 but the mechanisms underlying this effect remain largely unresolved. Here, 487 we demonstrate that inflammatory peritoneal macrophages recruited following sterile 488 inflammation persist long-term but in an aberrant state of activation largely due to an inability 489 to compete with incumbent macrophages for 'niche' signals and inflammation-driven 490 alterations in the peritoneal environment. In so doing, we reveal the existence of multiple overlapping biochemical 'niches' that control programming of resident peritoneal 492 macrophages and which are distinct from that controlling cell survival. 493

Like Liu et al 35 , our study suggests that the degree of replacement of LPM from the bone 494 marrow following inflammation depends on the magnitude of initial macrophage 495 disappearance. Furthermore, the increased survival of inflammatory macrophages following 496 transfer into macrophage-depleted recipients provides definitive evidence that incumbent 497 LPM impair survival of recruited cells. Hence, even without a defined physical niche, 498 monocyte contribution to resident macrophages within fluidic environments appears 499 subject to the same parameters of niche access and availability proposed by Guilliams and 500 Scott 16 . 501

Likewise, our data support a model whereby competition for access to a 'biochemical' niche 502 plays a critical role in determining the long-term transcriptional identity of inflammation-503 elicited macrophages. Specifically, inflammation-elicited macrophages that survived 504 following mild inflammation exhibited striking long-term differences to incumbent resident 505 macrophages including high MHCII and low GATA6 expression, yet more rapidly adopted a 506 GATA-6 hi MHCII lo resident-like phenotype and transcriptome following transfer into naive 507 macrophage-depleted mice. The failure of inflammation-elicited macrophages to down-508 regulate MHCII expression following transfer into intact naïve mice confirms this feature 509 arises from competition with incumbent resident macrophages for signals that regulate 510 MHCII expression. Consistent with this, inflammation-elicited macrophages rapidly 511 downregulated MHCII in vitro in response to RA-independent omental factors. Similarly, the 512 GATA6 lo phenotype of inflammation-elicited macrophages following mild inflammation 513 likely arises due to competition with enduring resident macrophages for retinoic acid, since more rapid acquisition of a GATA6 hi phenotype occurred following severe inflammation 515 concurrent with the ablation of resident macrophages. Critically, inflammation-elicited 516 macrophages gradually adopted features seemingly regulated by competition, suggesting 517 that with time these cells receive sufficient cues to acquire a mature resident phenotype. 518

Several features of inflammation-elicited macrophages appeared regulated by changes in the 519 cavity microenvironment post-inflammation. For example, severe inflammation led to 520 sustained expression of MHCII by inflammation-elicited macrophages despite natural ablation 521 of competing resident cells. Furthermore, the severely inflamed environment triggered MHCII 522 expression by adoptively transferred resident LPM that would otherwise remain largely 523 MHCIIin a non-inflamed or mild inflammation setting. Hence, it seems likely severe 524 inflammation triggers release of novel signals stimulatory for MHCII. 525 A small number of genes remained differentially expressed in inflammation-elicited 526 macrophages following transfer into macrophage-depleted mice, supporting the notion that 527 developmental origin influences macrophage identity 47 . These included features 528 reprogrammed over time (eg Timd4, VSIG4, Cd209b), and permanent 'legacy' features of 529 monocyte-derived cells (eg Sema4a, CD62L). The processes regulating these traits remains 530 unclear 47 . However, the rapid acquisition of Tim4 expression by inflammation-elicited LPM 531 following severe inflammation and the inhibition of this by transfer of Tim4 + resident 532 macrophages provides proof-of-principle that seemingly origin-related time-dependent 533 features of resident macrophages can be rapidly reprogrammed by appropriate niche signals. 534

While the panel of genes assessed here was relatively limited, our findings that gene 535 expression is largely dictated by competition with resident cells or the post-inflammatory 536 environment are likely to hold true on the transcriptome as a whole. For example, the overlap 537 between environment-dependent genes and GATA-6 regulated genes 5,21,22 suggests lower 538 GATA6 expression by inflammation-elicited macrophages controls a significant proportion of 539 their unique transcriptional profile. Similarly, the retarded expression of GATA6 by monocyte-540 derived LPM macrophages recruited during homeostasis also likely underlies a significant 541 degree of the distinct transcriptional clustering of these cells. Critically, these data reveal that 542 GATA6 expression does not act binarily but rather the level of expression has a critical role in 543 determining LPM identity, as predicted for transcription factors with many competing target 544 sites 48 . 545

One of our most intriguing findings was the difference in proliferative activity of incumbent 546 resident and inflammation-elicited LPM, with only the latter exhibiting the capacity to overtly 547 expand following transfer into macrophage-depleted mice. Notably, GATA6 directly regulates 548 proliferation of LPM 22 , suggesting this disparity may relate to differences in GATA6 549 expression. However, treatment with exogenous CSF1 or IL-4 stimulates recently-recruited 550 and incumbent resident macrophages to proliferate to equivalently high degrees 20,35,49 and 551 hence, the poor expansion of incumbent resident macrophages within the macrophage-552 deplete environment is not due to an intrinsic inability to proliferate. 553

Despite differentiating under inflamed conditions, persistent inflammation-elicited LPM bore 554 striking similarities to monocyte-derived LPM present under non-inflamed conditions. As well 555 as gene expression and proliferative activity, monocyte-derived LPM exhibited a largely 556 comparable response to LPS irrespective of condition of differentiation, most notably 557 characterized by increased production of IL-10 compared to embryonically-seeded LPM. 558

Other than IL-1b, the profile of cytokine production by monocyte-derived LPM was the opposite reported for GATA6-deficient LPM 50 , suggesting other factors control their 560 differential response to LPS. Hence, mild peritonitis does not lead to the existence of a unique 561 subset of LPM but rather the expansion of a subset present in homeostasis. As we have 562 previously shown that the abundance of Tim4monocyte-derived LPM gradually increases 563 with age 20 , it would appear that mild inflammation accelerates a process normally associated 564 with aging. 565 566 Our data also predict that inflammation-elicited LPM fail to express Cxcl13 due to 567 inflammation-induced loss of requisite niche signals. Notably, Cxcl13 expression by LPM is 568 sustained in vitro without addition of 'niche' factors 41 , potentially explaining why CXCL13 569 expression remains intact in incumbent LPM following mild inflammation. Hence, unlike the 570 reversible programme of gene expression controlled by GATA6 that is lost in the absence of 571 RA 41 , niche signals required to induce expression of CXCL13 in newly recruited macrophages 572 may not be needed to maintain expression in resident cells. 573 574 CXCL13-deficient mice are profoundly deficient in peritoneal B1 cells 44 yet CXCL13 is not 575 required for retention of B1 cells in the cavity 51 . We found the extent of inflammation and 576 consequently the ratio of CXCL13-producing resident macrophages to monocyte-derived 577 CXCL13 -LPM in the cavity correlates with a failure to accumulate peritoneal B1 cell with age. 578

As no other peritoneal lavage cells produce CXCL13 15,44 , our data suggest CXCL13 production 579 by peritoneal macrophages is required for continued recruitment of B1 cells from the 580 circulation. Notably, replacement of peritoneal LPM by monocytes and concurrent failure to 581 expand peritoneal B1 cells also occurs following abdominal surgery 15 . Hence, long-term 582 dysregulation of B1 cells is likely a general feature of peritoneal inflammation. Sterile peritoneal inflammation also led to elevated circulating natural antibody. Whereas splenic 584 and bone marrow B1 cells spontaneously secrete high levels of natural IgM, those in the 585 cavities do not 52 . Furthermore, levels of serum anti-PC IgM gradually drop with age 53 . Hence, 586

we speculate that the failure to accumulate B1 cells in the peritoneal cavity may allow their 587 re-entry into tissues permissive for antibody secretion such as fat associated lymphoid 588 clusters 54 . 589 Eight week-old female C57BL/6J CD45.1 + CD45.2 + or CD45.2 + C57BL/6J mice were exposed to 634 a single dose of 9.5 Gy g-irradiation under anaesthetic, with all but the hind legs of the animals 635 protected by a 2 inch lead shield. Animals were subsequently given 2-5×10 6 BM cells from 636 female congenic CD45.2 + C57BL/6J or C57BL/6J CD45.1 + CD45.2 + animals respectively by i.v. 

Omentum factors were generated by culturing the omentum from naïve mice in 1ml of 700 macrophage serum free media (GIBCO) for 5 days as described 5 after which medium was 701 collected, centrifuged at 300g and the supernatant collected and diluted in 1:2 in media. 702

Peritoneal cells were collected 11 days post low dose zymosan were collected as described 703 under sterile conditions and 5x10 5 plated and incubated for 2 hours at 37°C in cell culture 704 medium (RPMI, 10%FCS, 1% L-Glutamine and 1% Pen/strep supplemented with 20ng/ml 705 CSF1) after which medium was aspirated and cells were incubated in 250µl cell culture 706 medium with 250µl Omentum factors or macrophage serum free media with ATRA (Sigma, 707 1µm) or without for 24 hours. Then, medium was removed and plate was incubated with 708 5mM EDTA in ice cold PBS for 10 minutes on ice to collect cells. Wells were repeatedly washed 709 with ice cold 5mM EDTA PBS and wells were inspected using a microscope to confirm 710 negligible adherent cells remained. Cells were quantified and prepared for flow cytometry as 711 described. 712 713 Cytokine production assay 714 Cells were kept on ice at all time and all steps were carried out using sterile reagents in a 715 laminar flow hood using the sort protocol described. For each population of interest 1x10 3 716 cells per condition were sorted into 75µl sort medium (Folic acid deficient RPMI containing 717 20% FCS (Low endotoxin), 2% L-Glutamine and 2% Pen/strep). Cells were centrifuged at 100g 718 at 4 degrees for 5 minutes. The total mixture was then transferred into a 96 well plate 719 incubated at 37°C for 2 hours. Media was gently aspirated and cells were resuspended in 75µl 720 cell culture medium (Folic acid deficient RPMI supplemented with 1µg/ml Folic Acid (Sigma-721 Aldrich), 10% FCS (Low endotoxin), 1% L-glutamine and 1% Pen/Strep). Where indicated cells 722 received a final concentration of LPS of 1ng/ml (O11:B4, Sigma-Aldrich) in cell culture 723 medium or equivalent amount of dPBS. Cells were incubated for 14 hours and supernatant 724 was collected and analysed for cytokine release using the Legendplex Mouse Anti-Virus or 725

Mouse Inflammation panel according to the manufacturers protocol. Data was acquired using 726 an Attune flow cytometer and analysed using the Legendplex analysis software. 727 728 Phrodo phagocytosis assay 729 Cells were collected and 2x10 6 cells/sample were stained as described. Each sample was 730 washed twice with ice cold RPMI and was split into two tubes each and left on ice for 10 731 minutes. Then to each tube 10µl of Phrodo E.Coli particles was added and for each sample 732 one tube was incubated at 37°C and one at 4°C for 1hr. All samples were placed on ice and 733 washed once using 300µl Buffer C and were then resuspended in 300µl Buffer C. Cells were 734 analysed directly after finishing the protocol. Data is presented as normalized Phrodo mean 735 fluorescence intensity (MFI 37°C-MFI 4°C). overnight. Plates were then blocked with 100µl of blocking buffer (1% Casein in PBS; VWR) 742 for 1.5hr at room temperature, before serum samples were added at 1:100 dilution in 50µl 743 blocking buffer and incubated for 2hr at room temperature. Wells without antigen were used 744 as blank controls for each sample to measure non-specific antibody binding. Plates were then 745 incubated with 1:5000 HRP-conjugated anti-mouse IgG (Abcam) or 1:2000 anti-mouse IgM 746 (Southern Biotech) in blocking buffer for 1h at room temperature before addition of TMB 747 (Seracare). After 10 minutes the reaction was stopped with 0.16M sulphuric acid solution and 748 the OD450 value measured. Values for blank controls were then subtracted for each sample to 749 quantify antigen-specific antibody levels. Plates were washed twice with 0.1% Tween20 750 (Sigma-Aldrich) in PBS between all steps except before addition of TMB, when they were 751 

Supplementary Figure 1. A toolbox to investigate inflammatory macrophage fate 1133 (a) Dye labelling efficiency of F4/80 HI MHCII LO resident macrophages (grey) and

CD226 + small peritoneal macrophages (green) or CD226 -DCs or immature macrophages

24hrs after intraperitoneal administration of PKH26-PCL (n=3). **p<0.01 by determined one-way 1136 ANOVA with Tukey's multiple comparisons test

Experimental scheme for the adoptive transfer of unlabelled CD45.1/2 + peritoneal exudate cells 1138 (PEC) into the peritoneal cavity of CD45.2 + mice injected with PKH26-PCL intraperitoneally 24hrs 1139 prior

Representative PKH26-PCL labelling and quantification of donor F4/80 hi macrophages prior to 1141 transfer (top) and 2hrs post transfer (red; n=3) compared to recipient F4/80 Hi macrophages (black)

0001 determined by one-way ANOVA with Tukey's multiple comparisons test. 1143 (d) Non-host chimerism of F4/80 HI PKH26-PCL HI RMac in the naïve peritoneal cavity

Dye injection given 8wks after irradiation and zymosan injection given 24hrs 1146 thereafter

Expression of F4/80, MHCII and Tim4 by RMac(black), RMac Z10 (blue) and IMac Z10 3 days post 1148 10µg zymosan. ***p<0.001 ****p<0.0001 determined one-way ANOVA

Davies, L. C., Jenkins, S. J., Allen, J. E. & Taylor, P. R. Tissue-resident macrophages. 785 Nature immunology 14, 986-995, doi:10.1038/ni.2705 (2013 shown as log2 fold change in mean pg/ml over the mean RM-LPM using a box-and-whiskers plot.

(g) Secreted cytokine/chemokine profile collected from cultures of Mo-LPM or Mo Z10 -LPM (n=6), 1253 sourced from naïve or 8wks post zymosan mice, 14hrs after culture with LPS (1ng/ml). Results are 1254 shown as log2 fold change in mean pg/ml over the mean Mo-LPM using a box-and-whiskers plot. 1255 *p<0.05determined by repeated student's t test with Holm-Sidak correction.

For all experiments, data are presented as mean ± standard deviation with each symbol representing 1257 an individual animal. All data were pooled from at least 2 independent experiments.