Pre/pro-B cells generate macrophage populations PNAS PLUS during homeostasis and inflammation

Tatsiana Audzevicha, Rachael Bashford-Rogersb, Neil A. Mabbottc,d, Dan Framptone, Tom C. Freemanc,d, Alexandre Potocnikf, Paul Kellamb, and Derek W. Gilroya,1

aCentre for Clinical Pharmacology and Therapeutics, Division of Medicine, University College London, London WC1E 6JJ, United Kingdom; bDepartment of Medicine, Cambridge Biomedical Campus, University of Cambridge, Cambridge CB2 0XY, United Kingdom; cThe Roslin Institute, University of Edinburgh, Easter Bush, Midlothian EH25 9RG, United Kingdom; dRoyal (Dick) School of Veterinary Sciences, University of Edinburgh, Easter Bush, Midlothian EH25 9RG, United Kingdom; eDivision of Infection and Immunity, University College London, London WC1E 6AE, United Kingdom; and fInsitute of Immunology and Infection Research, University of Edinburgh, Edinburgh EH9 3FL, United Kingdom

Edited by Harinder Singh, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH and accepted by Editorial Board Member Ruslan Medzhitov March 31, 2017 (received for review October 6, 2016)

Most tissue-resident macrophages (Mφs) are believed to be derived publications (16, 18, 19), mature B cells do not differentiate into prenatally and are assumed to maintain themselves throughout Mφs. Instead, we found a population of cells within the early pro- life by self-proliferation. However, in adult mice we identified a B-cell/fraction B-cell compartment with non-rearranged B-cell progenitor within bone marrow, early pro-B cell/fraction B, that receptor (BCR) genes that exit bone marrow under steady state, differentiates into tissue Mφs. These Mφ precursors have non- circulate in peripheral blood, and populate serous cavities where rearranged B-cell receptor genes and coexpress myeloid (GR1, CD11b, they differentiate into Mφs, thereby contributing to the overall and CD16/32) and lymphoid (B220 and CD19) lineage markers. During pool of tissue Mφs alongside embryonically derived cells. Hence, steady state, these precursors exit bone marrow, losing Gr1, and in addition to local proliferation, we now further refine our un- enter the systemic circulation, seeding the gastrointestinal system derstanding of how tissue Mφs numbers are maintained during as well as pleural and peritoneal cavities but not the brain. While adulthood. Moreover, these data report cells with origins that in these tissues, they acquire a transcriptome identical to embryoni- are distinct from embryonic Mφ precursors but nonetheless dif- cally derived tissue-resident Mφs. Similarly, these Mφ precursors also ferentiate into cells with equivalent phenotypes. CELL BIOLOGY enter sites of inflammation, gaining CD115, F4/80, and CD16/32, and become indistinguishable from blood monocyte-derived Mφs. Thus, Results we have identified a population of cells within the bone marrow Mφs with a Potential B-Cell Origin Detected Using Mb1-iCre/Rosa26R-YFP Reporter Mice. Analysis of cells from the peritoneal cavity of naive early pro-B cell compartment that possess functional plasticity to dif- ++ ++ ferentiate into either tissue-resident or inflammatory Mφs, depend- wild-type mice revealed a population of F4/80 CD11b Mφs ing on microenvironmental signals. We propose that these precursors that also expressed CD19 (Fig. 1A). These cells constitute ∼0.8% represent an additional source of Mφ populations in adult mice dur- of total Mφs, suggesting a potential developmental relationship ing steady state and inflammation. between mononuclear phagocytes and B cells. Crossing Mb1-iCre (20) with Rosa26-YFP (21) mice allowed us to investigate this homeostasis | inflammation | macrophages potential further, because Mb1-dependent Cre-recombinase ex- pression is induced in B cells during the very early pro-B stage (20). Thus, tissue Mφs that express YFP must have been derived from B onocytes and macrophages (Mφs) maintain tissue ho- cells. In naive Mb1-iCre/Rosa26-YFP mice a significant proportion Mmeostasis and orchestrate immune-mediated responses to of Mφs expressed YFP, including ∼25% of peritoneal Mφs, 4–10% infection/injury. During embryonic hematopoiesis, monocytes arise of pleural cavity Mφs, and 1–2.5% of intestinal Mφs, with very few from myeloid precursors in fetal liver, whereas during adulthood they are derived from bone marrow progenitors. Circulating Significance monocytes migrate to sites of infection/injury where they differ- entiate into Mφs (1, 2) and under certain circumstances can also replenish Mφs in the colon (3). In contrast, the majority of resident In this report we provide evidence of a source of macrophage φ Mφ populations are embryonically derived and are proposed to (M ) populations that are derived from unique biphenotypic maintain their numbers by local proliferation after birth (4–9). early pro-B cells with non-rearranged B-cell receptors. These However, in the course of our studies we identified a population of early precursors give rise to either tissue resident- or monocyte- φ biphenotypic Mφs in the serous cavities of naive mice that coex- derived M s during homeostasis and inflammatory responses, press the B-lineage marker CD19 and the Mφ markers CD11b and thereby demonstrating functional plasticity depending on the environmental cues in adult mice. We suggest that these find- F4/80. Interestingly, revised models of hematopoiesis (10, 11) and φ evidence of developmental plasticity within hematopoietic precur- ings significantly advance and expand our understanding of M biology and hematopoiesis, the plasticity of hematopoietic pre- sors suggest a closer developmental relationship between myeloid φ and lymphoid lineages than previously appreciated (10–14). Con- cursors, and the heterogeneity of M subsets. sequently, because there is convincing evidence that lymphocyte-to- φ Author contributions: D.W.G. designed research; T.A. and A.P. performed research; R.B.-R., M differentiation can occur in vitro under artificial experimental N.A.M., D.F., T.C.F., and P.K. contributed new reagents/analytic tools; N.A.M., D.F., T.C.F., and settings (15–19), we determined whether the biphenotypic cells D.W.G. analyzed data; and D.W.G. wrote the paper. we observed in the peritoneal and pleural cavities and in the gas- The authors declare no conflict of interest. trointestinal system were derived from B lymphocytes or from This article is a PNAS Direct Submission. H.S. is a guest editor invited by the Editorial precursors within the B-cell lineage. This latter derivation would Board. challenge the existing view that local self-proliferation is the only Freely available online through the PNAS open access option. means by which tissue-resident Mφs maintain their numbers Data deposition: Microarray data are archived under ArrayExpress accession number throughout adulthood. E-MTAB-1878. Using a range of B-cell lineage-specific transgenic reporter mice 1To whom correspondence should be addressed. Email: [email protected]. and single-cell PCR, ImageStream/polychromatic flow cytometry, This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. and adoptive transfer studies, we show that, contrary to previous 1073/pnas.1616417114/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1616417114 PNAS Early Edition | 1of10 Downloaded by guest on September 29, 2021 + A YFP MφsinMb1-iCre/Rosa26R-YFP Mice Do Not Arise from Mature B Cells. The hematopoietic development of B cells includes seven – ′– Mac stages (A CandC F), which are defined by stage-specific cell- 53.0 CD19+ F4-80+ surface marker expression and BCR gene rearrangement status

F4/80 0.84 CD11b (22, 23). Rearrangement of Ig genes, which is initiated during 0" 0" bone marrow development at the pro-B-cell stage/fraction B stage and is completed in pre-B/fraction C′ B cells (23), leads to surface 0" 0" expression of BCRs of unique DNA sequence and enables B-cell F4/80 CD19 development and survival (24). This distinctive feature of B-cell + development was exploited to determine whether YFP Mφs B Gated Gated on on ssue-residenttissue-resident Mφs Mφs carry V-DJ rearrangements in their IgH locus as evidence of their origin from mature/pre B cells. Peritoneum Pleural Cavity Intestine Before embarking on single-cell PCR to determine whether + YFP Mφs carry V-DJ rearrangements, we ensured that the cell populations we intended to sort by FACS were single cells and not + YFP+ doublets or clusters of MφswithYFP B cells. Previous studies F4/80 F4/80 F4/80 24.5 YFP+ have reported biphenotypic murine peritoneal Mφs coexpressing 0" YFP+ 0" 1.97 0" 7.28 F4/80 and CD11b along with surface IgM, CD5, CD19, and/or B220(25–27) but have not considered the possibility that cell 0" 0" 0" YFP YFP YFP doublets/clusters might generate potential false-positive data. ImageStream analysis of the murine peritoneum identified three ++ ++ − − Spleen Liver Brain distinct cell types within the CD11b F4/80 Mφ gate: CD19 YFP − + + + (embryonic Mφs) and CD19 YFP and CD19 YFP cells (Fig. 2A). We suspect that these biphenotypic Mφ coex- CD11b CD11b YFP+ CD11b YFP+ pressing CD11b, F4/80, YFP, and CD19 are intermediates in YFP+ 0.2 0.3 the transition of B cells to Mφs. However, ImageStream also 0" 0" 0" 0.63 revealed significant numbers of doublets (Mφ/B-cell clusters) + + within this intermediate CD19 YFP Mφ population, with 0" 0" 0" YFP YFP YFP single biphenotypic cells constituting only ∼16.74% of the total + + φ ∼ φ CD19 YFP M s (Fig. 2B); thus 80% of the cells within the Fig. 1. M s with a potential B-cell origin detected in Mb1-iCre/Rosa26R-YFP biphenotypic population are aggregates of Mφs and B cells. See reporter mice. (A) Flow cytometric analysis of wild-type C57BL6 mice show- ++ ++ ing CD19 expression on peritoneal CD11b F4/80 Mφs. (B) Equivalent SI Appendix, Fig. S3.1 for further data on identifying clusters/ doublets. analysis of Mb1-iCre/Rosa26-YFP mice showing the YFP expression pattern in − − φ Therefore, we sorted CD19 YFP (embryonic Mφs) and M s from various tissues. Representative data are shown for 10 experiments − + + + carried out on naive animals. Mac, macrophage. CD19 YFP and intermediate CD19 YFP biphenotypic cells along with B cells to act as a positive control, bearing in mind that the intermediate or biphenotypic cells contain random clusters + YFP Mφs detectable in the spleen, liver, and brain (Fig. 1B); of Mφ/B-cell aggregates. On a single-cell level, although V-DJ φ rearrangements were detected in ∼71% of the single B cells an- gating strategies to discern tissue M populations are shown in SI + + alyzed, only ∼33% of the intermediate CD19 YFP Mφswere Appendix,Fig.S1.1. + − + V-DJ , and no rearrangements were found in CD19 YFP To determine B-cell lineage-restricted reporter expression in − − Mb1-iCre/Rosa26-YFP mice, we analyzed YFP expression in vari- B-cell–derived MφsorinCD19YFP tissue-resident embryonic Mφs(Fig.2C). We found a significant increase (from 33 to 84%) in ous hematopoietic progenitors of these animals as well as in ma- + + + ture leukocytes in various tissue compartments. Analysis revealed the frequency of V-DJ reactions in the CD19 YFP populations + ∼ that 97–99% of peripheral blood B cells were YFP ; only minimal (compared with 87% for B cells) when multiple cells (5, 10, or 20) YFP expression was detected on T cells and monocytes, and no per reaction were analyzed. We propose that this increase arose from contaminating B cells or B-cell/Mφ clusters in at least ∼33% YFP expression was detectable on neutrophils (SI Appendix,Fig. + + + of single-sorted B-Mφ intermediate CD19 YFP Mφs. As expec- S1.2A). In bone marrow we found ∼7% of YFP cells among − + − + + + ted, all single B cells expressed Cd79b, whereas CD19 YFP Lin IL7Rα Sca-1 c-Kit common lymphoid progenitors (CLP), but − − + − − + − + – φ no YFP cells were detected among Lin IL7Rα CD34 / CD16/32 B-cell derived M sandCD19YFP tissue-resident embryonic MφsexpressedEmr1 (encoding the MφsmarkerF4/80).Unlike common myeloid progenitors (CMP). Equally, no YFP expres- − − ++ ++ other Mφ subsets, coexpression of the B-cell–specific transcript sion was detectable in Lin IL7Rα Sca-1 c-Kit multilineage − − + ++ Cd79b and the Mφs-specific transcript Emr1 was detected in in- α + + progenitors (MLP) or in Lin IL7R CD34 CD16/32 termediate CD19 YFP Mφs on a single-cell level. Collectively, φ + granulocyte/M progenitors (GMP) (SI Appendix, Fig. S1.2B). these data show that YFP macrophages display no evidence of More specifically, the lymphoid-biased multipotent progenitor − − + VDJ recombination and that the VDJ recombination detected in (LMPP; Lin IL7Rα Flk-2hiCD34 ) was found to be negative (SI + + − − − CD19 YFP Mφs arises from sorting of doublets/clusters. Appendix, Fig. S1.2B). Finally, ∼0.15% of Lin CD4 CD8 early + Of note, single-cell PCR reported doublet rates of ∼30%, whereas thymic progenitors were YFP (SI Appendix, Fig. S1.2C). Gating ImageStream estimated ∼80%. This disparity arises from the differ- strategies for flow cytometric identification of hematopoietic ent flow cytometry algorithms used by the two methods for single-cell progenitors in bone marrow and T-cell progenitors in thymus are sorting. Nonetheless, both methods reliably identified the presence of + shown in SI Appendix, Fig. S1.3 and S1.4. contaminating B-cell/Mφ clusters within the intermediate CD19 + φ + + YFP M population; in the absence of stringent exclusion criteria, Phagocytosis of Apoptotic YFP B Cells by Mφs Does Not Generate YFP such clusters will generate false-positive results. Mφs. Immunofluorescent microscopy on FACS-purified peritoneal − cell populations and in vitro phagocytosis assays in which YFP Evaluation of B-Cell–Specific Reporter Mice Points Toward an Early Pro-B- + Mφs were cocultured with early/late apoptotic YFP B cells con- + φ + Cell Origin of YFP M s. We analyzed bone marrow of Cd19-Cre/ firmed that the YFP fluorescence detected in YFP Mφs from Rosa26-YFP mice using the gating strategy in SI Appendix,Fig. Mb1-iCre/Rosa26-YFP mice was caused by the cytoplasmic YFP S4.1A. In these animals, although surface CD19 is expressed from + expression in Mφs and not by the ingestion of apoptotic YFP B early pro-B-cell stage/fraction B onwards (SI Appendix,Fig.S4.1B), + cells (SI Appendix,Fig.S2). only ∼15% of this B-cell subset was YFP (Fig. 3A). However, Cre

2of10 | www.pnas.org/cgi/doi/10.1073/pnas.1616417114 Audzevich et al. Downloaded by guest on September 29, 2021 PNAS PLUS

A All Singlets Focused F4/80++CD11b++Mφs YFP+CD19+ B cells B BF YF CD19 F4/80 CD11B YFP+CD19+ P YFP+CD19+ Singlets Focused 0.12 42.6 B Mac B cells CD19-YFP- 1b

19 92.5 Doublets uency D1

D19 YFP+CD19+ q C C 0 non-Mac CD 0 0 re

F YFP+CD19- Aspect Ratio 4.65 000 YFP+CD19- Area Gradient RMS F4/80 YFP YFP YFP-CD19-

C %ofsingle Single cells Multiple cells Sample Macrophage Macrophage YFP+ number B YFP+ YFP+ YFP- DNA B YFP+ YFP+ YFP- CD19+ Mφs cell CD19+ CD19- CD19- Ladder,pb cell CD19+ CD19- CD19- Product size

V-DJ - 300 - ~300 bp 1 9.68 2 14.46 - 200 - 95 bp Cd79b - 100 - 3 26.09 Average 16.74 Emr1 - 200 - 150 bp - 100 -

:llew/slleC 1 1 1 1 1 1 1 1 1 1 1 1 :llew/slleC 20 10 5 20 10 5 20 10 5 20 10 5 V-DJ+ reactions: 32/45 15/45 0/45 0/45 39/45 38/45 0/45 1/45

Fig. 2. YFP+ MφsinMb1-iCre/Rosa26R-YFP mice do not arise from mature B cells. (A) ImageStream analysis of naive murine peritoneum. Dot plots showing CELL BIOLOGY ++ ++ − − − + + + + − − single focused cells in CD11b F4/80 Mφ gate identifying three Mφ populations: CD19 YFP , CD19 YFP , and CD19 YFP ; the non-Mφ (CD11b / F4/80 ) + + ++ ++ compartment contains CD19 YFP B cells. (B, Upper) Representative images of CD11b F4/80 Mφ populations and B cells. (Lower)Althoughconfirmingthe + presence of Mφs coexpressing CD11b, F4/80, YFP, and CD19, ImageStream reveals that a substantial number of doublets occur within rare populations of CD19 + + + YFP Mφs; approximately 16.74% of the cells within CD19 YFP Mφ gate are single biphenotypic cells. (C) Detection of V-DJ rearrangements and expression of B-cell– and Mφ-specific genes in single and multiple cells purified by FACS. At least 45 single and multiple (20/10/5) cells of each type, including peritoneal B-1/B-2 B cells and CD11b++F4/80++ Mφ populations (CD19−YFP−,CD19−YFP+,andCD19+YFP+) were analyzed for the presence of V-DJ rearrangements in genomic DNA, B-cell– + specific Cd79b transcripts, and Mφ-specific Emr1 transcripts. The frequency of V-DJ reactions shown below the graph demonstrates that, despite coexpression of + + + Cd79b and Emr1 in intermediate CD19 YFP Mφs, the number of V-DJ reactions detected on the single-cell level is low compared with B cells (15/45 versus 32/45). The + + + number of V-DJ reactions increased to B-cell levels (38/45) when multiple cells (20/10/5) of CD19 YFP Mφs were analyzed per reaction; these results suggest con- tamination by B cells.

expression increases at later stages of bone marrow development back-gating analysis revealed that these cells are found within in these animals than in the Mb1-iCre/Rosa26-YFP strain (28), early pro-B/fraction B cells in the bone marrow (Fig. 4C). The + resulting in ∼90% YFP labeling detectable in mature B/fraction F majority of these cells in bone marrow were also Gr1 , but only cells (Fig. 3A). Although YFP expression was found in ∼90% of ∼30% expressed Gr1 in blood; both bone marrow and blood cell peritoneal B cells from Cd19-Cre/Rosa26-YFP mice, very small populations expressed very little, if any, IL7Rα (SI Appendix, + numbers of YFP Mφs were found in this Cd19-Cre reporter strain; Fig. S4.3B). + similar data were obtained using Cr2-Cre reporter mice in which Therefore, the absence of YFP MφsinCd19-Cre/Rosa26-YFP − − YFP expression is also restricted to mature B-cell populations (Fig. mice and their presence in Rag2 / /Mb1-iCre/Rosa26-YFP mice + 3B). These data, along with those in Fig. 2C (no V-DJ rearrange- suggestthatYFP MφsinMb1-iCre/Rosa26-YFP mice might be + + + + + ments detectable in YFP+ Mφs), suggest that YFP Mφs found in derived from non-rearranged B220 CD43 CD19 YFP CD16/ ++ + Mb1-iCre/Rosa26-YFP mice are not derived from mature B cells. 32 CD11b cells within early pro-B/fraction B cells. Conse- + + Next we investigated whether YFP Mφs developed in Mb1-iCre/ quently, YFP Mφs are hereafter referred to as “pB-Mφs” for Rosa26-YFP mice that were also deficient in the Rag2 gene and cells derived from a population within early pro-B cells. In addi- + + + + ++ + therefore lacked mature B cells because of a developmental block tion, B220 CD43 CD19 YFP CD16/32 CD11b Bcellsthat + at the bone marrow pro-B-cell stage (29). The percentage of YFP give rise to pB-Mφs are called “pB-Mφprecursors.” − − Mφs in the peritoneal cavity of Rag2 / /Mb1-iCre/Rosa26-YFP mice + + was similar to that detected in B-cell–sufficient Rag2 / /Mb1-iCre/ Pro-B-Cell–Derived Mφs Are Generated in Response to Inflammation. Rosa26-YFP mice (Fig. 3B). These data and those presented in Fig. 2 To determine the dynamic relationship between pB-Mφprecursors + suggest that tissue-resident YFP Mφs are not derived from mature and pB-Mφs, we injected zymosan (31) into the peritoneum of B cells but instead arise from an immature B-cell population with Mb1-iCre/Rosa26-YFP mice. Zymosan triggers an acute resolving non-rearranged BCR genes, most likely within pro-B cells. Because inflammation, causing immediate changes in cellular trafficking a proportion of early pro-B/fraction B cells have not undergone within the peritoneum and in blood and bone marrow. In the naive − rearrangement (30), we suspect that it is this population that harbors peritoneum and as shown previously in Fig. 1B,YFP embryonic- + − the precursors that are an alternative source of Mφs in adulthood. derived Mφs and YFP pB-Mφs were identified within the B220 + + − ++ In support of this hypothesis we found B220 CD43 early B CD19 fraction; however, both populations were also CD16/32 . + + ++ cells in bone marrow (Fig. 4A) and in blood (Fig. 4B) that were Within B220 CD19 B cells a subset that coexpressed CD16/32 + + ++ + ++ + also CD19 YFP CD16/32 CD11b . High levels of CD16/ CD11b and YFP was identified (Fig. 5A). This latter pop- 32 and CD11b expression, which are commonly found in myeloid ulation was phenotypically similar to pB-Mφprecursors in bone cell populations (SI Appendix, Fig. S4.2), suggest that this pop- marrow and blood (Fig. 4 A and B). Four hours after zymosan + − − ulation might be a potential precursor of YFP Mφs. Indeed, injection, CD19 YFP embryonic Mφs and pB-Mφs transiently

Audzevich et al. PNAS Early Edition | 3of10 Downloaded by guest on September 29, 2021 CD11b++F4/80++ Mφ CD19+ B cell A B Mb1-iCre CD3-CD19-CD11c- Mb1-iCre

Early B220+ CD43+ Bone Marrow B cells Fr. A / pre-pro-B Fr. B/ pro-B Fr. C/ pro-B Fr. C’/ pre-B

Cd19-Cre

Late B220+ CD43- Bone Marrow B cells Fr.D/ pre-B Fr. E/ new-B Fr. F/ mature-B

Cr2-Cre

Cd19-Cre Rag2-/-/- /Mb1/Mb1-iCre Early B220+ CD43+ Bone Marrow B cells Fr. A / pre-pro-B Fr. B/ pro-B Fr. C/ pro-B Fr. C’/ pre-B

Rosa26-YFP

Late B220+ CD43- Bone Marrow B cells

Fr.D/ pre-B Fr. E/ new-B Fr. F/ mature-B

+ Fig. 3. Evaluations of B-cell–specific reporter mice point toward an early pro-B-cell origin of YFP Mφs. Cells and tissues from Mb1-iCre/Rosa26-YFP mice − − − − (Mb1-iCre), CD19-Cre/Rosa26-YFP mice (CD19-Cre), Rag2 / /Mb1-iCre/Rosa26-YFP mice (Rag2 / / Mb1-iCre), Cr2-Cre/Rosa26-YFP mice (Cr2-Cre), and Rosa26- YFP mice were collected and analyzed by flow cytometry. (A) Representative dot plots showing induction of YFP expression at later B-cell bone marrow developmental stages in CD19-Cre/Rosa26-YFP mice (CD19-Cre) than in Mb1-iCre/Rosa26-YFP mice (Mb1-iCre). B shows little YFP expression in Cd19-Cre and + + CD21 (Cr2-Cre) reporter strains, but expression in Rag2 / /Mb1-iCre/Rosa26-YFP mice. For the gating strategy see SI Appendix, Fig. S4.1 for representative dot plots showing YFP expression in peritoneal CD19+ B cells and CD11b++F4/80++ Mφs in the mouse strains named above. Fr, fraction; Mac, macrophage.

disappeared from the peritoneum, as is consistent with the “leu- downregulating the expression of Gr1 and CD43 but up-regulating kocyte disappearance phenomenon” (32), whereas the numbers of CD115, F4/80, and CD16/32 (Fig. 5D). As pB-Mφprecursors differ- peritoneal pB-Mφprecursors increased at onset but declined from entiated into Mφs from 24 h after zymosan administration, they 24 to 72 h after zymosan injection (Fig. 5B). The temporal profiles lost CD19 and B220 expression and acquired Mφ characteristics, of these cells are shown in Fig. 5C. including the expression of F4/80 and higher levels of CD11b and pB-Mφprecursors detected within the B-cell compartment at 4 h CD16/32 coincident with the increase in the numbers of pB-Mφs underwent phenotypic changes over the course of inflammation by from 24 h onwards (Fig. 5D).

4of10 | www.pnas.org/cgi/doi/10.1073/pnas.1616417114 Audzevich et al. Downloaded by guest on September 29, 2021 Consistent with the idea that biphenotypic cells within bone PNAS PLUS + A Bone Marrow marrow pro-B/fraction B give rise to YFP Mφ, the proportion of pB-Mφprecursors in bone marrow increased at 4 h postperitonitis and gradually returned to preinflammation levels by 72 h (Fig. 5D). Changes in peripheral blood pB-Mφprecursors correlated in- versely with the fluctuations in bone marrow pB-Mφprecursors, demonstrating a transient decrease at inflammatory onset and B220

B220 recovery by resolution (Fig. 5E). These data, taken together with an increase in peritoneal pB-Mφprecursors during the early stage of inflammation (Fig. 5C), reflect the generation and expansion of bone marrow pB-Mφprecursors triggered by inflammation followed by their migration via the blood stream into the peritoneum and CD19 CD43 their further differentiation into inflammatory pB-Mφ. We next tested the potential of pB-Mφprecursors to proliferate in either myeloid or lymphoid mixtures over 14 d in culture. Using CMPs and pro-B cells as controls, we found that pB- Mφprecursors grow only in a myeloid mixture and did not respond to growth factors that support lymphocyte growth (Fig. 6A); this result is consistent with the virtually undetectable levels of IL7Ra

CD19 on these cells (SI Appendix, Fig. S4.3C). We also tested the CD19 ability of FACS-purified peritoneal pB-Mφprecursors to differen- tiate into Mφs in vitro. We used pB-Mφprecursors collected at 4 h postperitonitis because they were quantitatively enriched during that phase compared with the naive peritoneum; naive perito- + − +/− − YFP YFP neal B cells (CD19 Gr1 CD11b F4/80 ) were used as a con- trol. After 6 d of cultivation in the presence of M-CSF, small and distinctively round pB-Mφprecursors differentiated into cells with characteristic Mφ morphology as they increased their cell size

and acquired F4/80 expression (Fig. 6 B and C). Significantly CELL BIOLOGY fewer cells acquired these Mφ characteristics when cultivated in the absence of M-CSF or in the presence of GM-CSF (Fig. 6C; for further comparisons see SI Appendix, Fig. S5.1). CD16-32 CD16-32 Early Pro-B Cells Reconstitute Tissue MφsinVivoFollowingLethal + Irradiation. FACS-purified bone marrow YFP early pro-B/ + + + fraction B cells comprising biphenotypic CD19 B220 YFP ++ + φprecursors CD11b CD11b CD16/32 CD11b pB-M (Fig. 4A and SI Appendix, Fig. S6.1) were injected into lethally irradiated wild-type mice along + B with CD45.1 total bone marrow from congenic wild-type mice to Blood improve survival rates following irradiation. The mature bone marrow B-cell/fraction F subset (SI Appendix,Fig.S6.1) in com- + bination with CD45.1 total bone marrow was used as a control. Data in Fig. 7A depict the B-cell and Mφ composition of the peritoneum of naive wild-type CD45.1 mice alongside Mb-1iCre/ Rosa26-YFP mice, which are on a CD45.2 background. Adoptive + transfer of CD45.1 bone marrow resulted in reconstitution of all lineages, whereas early pro-B cells contributed to the restoration of the B-cell lineage in the peritoneum (Fig. 7B) and spleen (Fig. 7C). Importantly, adoptively transferred biphenotypic pB- + Mφprecursors within early pro-B cells gave rise to YFP Mφsinthe peritoneum but did not contribute to myeloid populations in the spleen (Fig. 7C and SI Appendix,Fig.S6.2B and C). Although C YFP expression was found in roughly 7% of CLPs (SI Appendix, + Fig. S1.2B), no YFP mature T cells were found in the spleens of − + recipient mice (Fig. 7C); moreover, the transfer of Lin IL7Rα ++ ++ + Sca-1 c-Kit CLPs containing YFP CLPs to lethally irradiated + mice did not result in the appearance of YFP Mφs in these an- imals (Fig. 7D). Also, upon the transfer of Mb-1iCre/Rosa26-YFP + LMPPs into sublethally irradiated CD45.1 Rag2-null recipients no evidence of the reporter was found in the hosts (SI Appendix, Table S1), as is consistent with the absence of YFP reporter ex- pression in these cells (SI Appendix, Fig. S1.2B). These findings provide evidence that early pro-B cells and specifically bipheno- precursors typic pB-Mφ , and not mature B cells or contaminating CLPs, differentiate into Mφsinvivo. Fig. 4. (A and B) Biphenotypic populations of CD19+B220+CD43+YFP+CD16/ ++ + – φ 32 CD11b cells with B-lymphoid and myeloid characteristics identified in Transcriptomic Analysis of Pro-B-Cell Derived M s and Their Classic bone marrow (A)andblood(B). These cells are referred to as “pB-Mφprecursors.” Counterparts. In the first instance, it was necessary to discriminate + + + + ++ + (C) Quantification of CD19 B220 CD43 YFP CD16/32 CD11b pB-Mφprecursors between Mφ populations present in the naive peritoneum and in early B-cell bone marrow fractions. Values are expressed as the absolute those that appear during inflammatory resolution. To do so, we number of cells per one limb. Fr, fraction. used the phagocytic dye PKH26PCL (PKH-Red), which labels

Audzevich et al. PNAS Early Edition | 5of10 Downloaded by guest on September 29, 2021 4 hours peritonitis A pB-Mφ precursor B CD19 CD16-32 B cell

Mφ pB-Mφ

C

+ Fig. 5. YFP Mφs are generated from biphenotypic + + + + ++ + B220 CD43 CD19 YFP CD16/32 CD11b pro-B-cell– D derived precursors in response to inflammation. In- flammation was induced by i.p. injection of 0.1 mg zymosan. Tissues were collected at various time points after peritonitis induction and were analyzed by flow cytometry. (A) Gating strategy showing the identifi- + + + + ++ cation of B220 -CD19 Bcells,B220 -CD19 CD16/32 ++ − − − CD11b peritoneal pB-Mφprecursors,YFPB220 CD19 F4/80++CD16/32++CD11b++ Mφs, and YFP+B220−CD19− F4/80++CD16/32++CD11b++ pB-Mφs in naive perito- + + neum. (B) Dot plots showing the B220 -CD19 B peri- toneal compartment with an increased proportion of ++ ++ − CD16/32 CD11b pB-Mφprecursors and the B220 − CD19 compartment with a reduced frequency of ++ ++ F4/80 CD11b Mφs at 4 h after peritonitis induction. (C) Temporal profiles of peritoneal Mφs, pB-Mφs, and peritoneal pB-Mφprecursors during inflammation. (D) Cell-surface markers for the cells over time. (E and EF F) Temporal profiles of pB-Mφprecursors in bone marrow (E) and peripheral blood (F) during inflammation. Values are expressed as the percentage of total bone marrow or blood cells. Statistical evaluation was per- formed by one-way ANOVA using Bonferroni posttest (*P < 0.05, **P < 0.01, ***P < 0.001). BM, bone mar- row; MFI, mean fluorescence intensity.

resident peritoneal phagocytes (33) but not infiltrating pB- the global gene-expression profiles of individual cell populations Mφprecursors or peripheral blood monocytes (SI Appendix,Fig.S7.1). (30 datasets, n = 3 per group) by performing a sample-to-sample Embryonically derived Mφs and pB-Mφs were further designated correlation analysis using the BioLayout Express3D software tool as MφsTR-naive (P1) and pB-MφsTR-naive (P2), respectively, to (34) with a Pearson correlation threshold of r ≥0.97. This analysis highlight that they are tissue-resident cells from the naive peri- showed that datasets for monocytes, neutrophils, B-1 cells, and B-2 toneum. Injecting PKH-Red into Mb1-iCre/Rosa26-YFP mice B cells (these additional cells types were included for comparison before inflammation resulted in MφsTR-naive and pB-MφsTR-naive purposes) clustered together like-with-like in distinct regions of + becoming PKH-PCL . Injecting zymosan shortly after PKH-PCL the graph, whereas all Mφ populations formed a single, distinct and examining the cavity during resolution revealed four addi- component within this network (Fig. 8B). Within this component, tional inflammatory Mφ populations including MφsTR-naive that naive tissue-resident pB-MφTR-naive and MφTR-naive were located experienced 72 h of inflammation, hereafter called “MφsTR-inflam” like-with-like, as were inflammatory tissue-resident pB-MφTR-inflam (P3) and “monocyte-derived mo-Mφsinflam” (P4) because they and MφTR-inflam and inflammation-induced pB-Mφinflam and mo- − − are YFP PKH . In addition, we detected pB-MφsTR-naive that Mφinflam. These data suggest that pB-Mφs and classic Mφsthatare experienced 72 h of inflammation, which we designated “pB- embryonically derived share highly similar overall transcriptional MφsTR-inflam” (P5) and pB-Mφprecursors that infiltrated into the profiles. Network correlation and cluster analysis of transcript cavity in response to zymosan (evident at 4 h in Fig. 5B)and coexpression shown in Fig. 8C revealed gene clusters with similar that also experienced 72 h of inflammation, which we designated functions or cell-specific activities occupying specific cliques within “pB-Mφsinflam” (P6) (Fig. 8A). the network graph. Dataset S1 provides the contents of each of Gene-expression profiles for these naive and inflammation- these clusters, and the expression profiles of selected clusters experienced Mφ populations then were analyzed (see Materials across the 30 datasets are displayed in SI Appendix,Fig.S7.4and and Methods for microarray and sample preparation and SI Ap- S7.5. For example, cluster C2 and C4 contained many genes pendix,Fig.S7.2andS7.3for gating strategies). First, we compared expressed highly by B cells (SI Appendix,Fig.S7.5A), including

6of10 | www.pnas.org/cgi/doi/10.1073/pnas.1616417114 Audzevich et al. Downloaded by guest on September 29, 2021 A PNAS PLUS CMPs proB cells pB-Mφ precursors

Myeloid lymphoid Myeloid lymphoid Myeloid lymphoid

2130±103 out of ND ND 2178±445 out of 1875±397 out of 74±21 out of 2500 events (n=5 2500 events (n=5 2500 events (n=5 500 events (n=5 wells seeded wells seeded with wells seeded wells seeded with single cells) single cells) with single cells) with single cells)

B Day 1 Day 6 Day 1 Day 6

Day 1 Day 6 + M-CSF Day 1 Day 6 + M-CSF Macrophage precursor CELL BIOLOGY

pB-Mφ F4/80 CD19 DAPI

C

Fig. 6. In vitro assays for pro-B-cell–derived precursors’ differentiation. (A) Results of an in vitro assay in which single-cell populations were cultured for 10–14 d in medium supporting either myeloid or lymphoid growth. Results are shown as the number of events bearing either a lymphoid or myeloid phenotype in the respective media, as explained in Materials and Methods.(B) FACS-purified peritoneal Mφs and pB-Mφprecursors 4 h after peritonitis induction were cultured with or without M-CSF/GM-CSF and were analyzed by cell morphology using Rapid-Romanowsky stain and α-CD19, α-F4/80, and DAPI labeling. (Scale bars, 20 μm.) (C) Cell size was quantified using ImageJ. Statistical evaluation was performed by one-way ANOVA using Bonferroni posttest (*P < 0.05, **P < 0.01, ***P < 0.001). ND, not detected; n.s., not significant.

Cd19 and many Ig-encoding genes including IgH-V11 (SI Appen- very low levels of genes within the B-cell–related clusters (in- dix,Fig.S7.5B). Clusters C5, C17, and C50 were enriched with cluding those derived from pro-B cells) were detected in any of the genes known to be expressed highly by Mφs(SI Appendix,Fig. Mφ datasets (SI Appendix,Fig.S7.5A–C), and their expression S7.5C) (35). Genes within these clusters were expressed highly by levels were consistent with our postsort purity analysis data, which all monocyte and Mφ populations represented in this analysis, as indicated the presence of low levels of contaminating B cells compared with B cells and neutrophils (Fig. 6G). In contrast, only (0–10%) in some Mφ samples (SI Appendix, Fig. S7.2). Importantly

Audzevich et al. PNAS Early Edition | 7of10 Downloaded by guest on September 29, 2021 Peritoneum A Naïve mice B Irradiated mice (CD45.2) reconstituted with:

(CD45.1) Mb-1iCre/Rosa26-YFP Total BM from CD45.1 + YFP+ Total BM from CD45.1 + YFP+ (CD45.2) mature B from Mb1- iCre/ early pro-B from Mb1- iCre/ YFP YFP B cells Rosa26- mice (CD45.2) Rosa26- mice (CD45.2)

Mφs

Fig. 7. pB-Mφprecursors within the early pro-B-cell bone marrow fraction reconstitute peritoneal Mφs in lethally irradiated mice. FACS-purified early pro-B cells contain- ing YFP+CD16/32++CD11b biphenotypic pB-Mφprecursors C and mature B cells (control) from the bone marrow of Spleen Mb1- iCre/Rosa26-YFP mice (SI Appendix,Fig.S6.1)were Irradiated mice (CD45.2) reconstituted with: D mixed in a 1:10 ratio with total bone marrow cells from congenic CD45.1 donor mice and were injected i.v. into Total BM from CD45.1 + YFP+ early pro-B from Mb1- iCre/Rosa26- Peritoneum YFP mice (CD45.2) Irradiated mice (CD45.2) reconstituted with: lethally irradiated (950 rad) wild-type mice. Hematopoi- Total BM from CD45.1 + CLPs from Mb1- iCre/Rosa26-YFP mice (CD45.2) etic reconstitution was assessed in the peritoneum and B cells Mφs spleen 6 wk after the injection of donor bone marrow cells. (A and B) Representative plots showing the ex- pression of tracing markers on peritoneal cell pop- ulations of naive CD45.1 and Mb1-iCre/Rosa26-YFP mice (A) and the engraftment of donor CD45.1+ (all lineages) + and YFP (early pro-B/mature B) cells in the peritoneum of lethally irradiated mice (B). (C) Analysis of the re- constitution of hematopoietic lineages in spleen shows + Spleen engraftment of CD45.1 progenitors in B and T cells and Irradiated mice (CD45.2) reconstituted with: in the myeloid compartment, whereas YFP+ early pro-B Total BM from CD45.1 + CLPs from Mb1- iCre/Rosa26-YFP mice (CD45.2) B cells T cells Myeloid cells cells contribute exclusively to the reconstitution of the B-cell lineage and do not differentiate into myeloid or − + ++ ++ + T cells. (D)Lin IL7Rα Sca-1 c-Kit CLPs containing YFP CLPs were mixed with total bone marrow cells from congenic CD45.1 donor mice and were injected into ir- radiated wild-type mice. Engraftment was determined in the peritoneal B-cell and myeloid compartments as well as in the spleen. BM, bone marrow.

genes in clusters C55 and C71 were expressed highly by both types we provide evidence of a subset of early pro-B cells with non- of inflammation-induced Mφs, including pB-Mφinflam as well as rearranged BCR that differentiate into Mφs during homeostasis mo-Mφinflam, as compared with other Mφ populations (SI Appen- and inflammation. In this work we used a combination of B-cell– dix,Fig.S7.4F), whereas embryonic and pro-B-cell–derived Mφs specific reporter mice including Mb1-iCre/Rosa26-YFP, Cd19- TR-naive TR-naive − − from naive peritoneum (Mφ and pB-Mφ ) and after Cre/Rosa26-YFP,andRag2 / /Mb1-iCre/Rosa26-YFP mice, which φTR-inflam φTR-inflam inflammation (M and pB-M ) characteristically enabled fate-mapping studies to establish that a proportion of expressed genes included in clustersC6,C36,andC40(SI Ap- tissue-resident Mφs in serous cavities and intestine are derived pendix,Fig.S7.4G). Hence transcriptome analysis also reveals the – φ from a subpopulation of non-rearranged early pro-B cells coex- plasticity of the pro-B derived M precursors that give rise to two pressing B-cell and myeloid markers. distinct subsets of Mφs with greatly contrasting phenotypes dur- φTR-naive φinflam These findings are fundamentally distinct from earlier reports, ing homeostasis (pB-M ) and inflammation (pB-M ). which demonstrated a myeloid lineage switch occurring in early Genes in all the mononuclear phagocyte-related clusters were – – φ φ lymphoid progenitors in vitro (12, 36, 37) as well as B-cell to M highly expressed in pB-M s at levels similar to those expressed by α φ differentiation using enforced C/EBP expression (15, 16, 38). embryonic/monocyte-derived M s counterparts. No clusters were φ identified that contained unique genes expressed highly in pB-Mφs Bipotential progenitors that give rise to M s have been reported in postnatal bone marrow (39), but those cells are different from alone. Similarly, no genes were identified that were highly ex- − the pB-Mφprecursors reported here, because they are CD45R pressed in embryonic/monocyte-derived Mφs but not in pB-Mφs. + + − − + + − CD19 CD127 CD11b Gr1 and D-J . Moreover, AA4.1 B200 Together these data suggest that, despite the distinct origin of the − + pro-B-cell–derived Mφs, they share very similar transcriptional and Mac1 Ly6A bipotential progenitors in fetal liver (12) and adult therefore predicted phenotypic characteristics with classic Mφ bone marrow early progenitors with lymphoid and myeloid po- populations, i.e., embryonic/monocyte-derived Mφs. tential (EPLMs) expressing B220, c-Kit, IL-7Rα (CD127), Flt3 (CD135), and CD93, but not CD19 and NK1.1 (10), demonstrate Discussion both lymphoid and myeloid potential in vitro. Although the above Our study highlights the developmental plasticity of immature B reports and other studies used primarily artificial experimental cells and their role as an additional source of Mφs in vivo. Here systems (17–19), we demonstrate a biphenotypic population

8of10 | www.pnas.org/cgi/doi/10.1073/pnas.1616417114 Audzevich et al. Downloaded by guest on September 29, 2021 PNAS PLUS CD11b++F4/80++Mφs pression on mature myeloid cells after transfer of Mb-1iCre/ A - - - B CD3 CD19 CD11c pB-Mφinflam Naive Monocytes Rosa26-YFP LMPPs into alymphoid recipients would strongly Neutrophils argue that the generation of graft-derived myeloid cells in this model does not originate from a LMPP progenitor. Therefore, mo-Mφmo- inflam MφTR-naïve these progenitors are unlikely to account for selective YFP ex- P1 P2 pression in tissue-resident Mφ populations in the peritoneum, B-1 B cells P1 – MφTR-naive pleural cavity, and intestine. This assertion is reinforced by the TR-naive MφTR-inflam + + P2 – pB-Mφ absence of YFP Mφs in irradiated mice receiving YFP CLPs. 0 hours ++PKH-Red PKHPKH-Red Red The molecular signals that enable a myeloid fate in the subset of pB-MφTR-naïve early pro-B cells described here remains to be elucidated. How- B-2 B cells ever, it is likely that the demand for Mφs in tissues generates these pB-MφTR-inflam signals. Indeed, this process became evident during zymosan- induced peritonitis, in which local inflammatory factors triggered φprecursors 72 hours the expansion of bone marrow pB-M , which egressed into blood and accumulated in the peritoneum where they differenti- inflam P3 P5 ated into inflammatory pro-B-cell–derived Mφs(pB-Mφ ). P4 P6 Although these temporal changes are certainly only correlative, these data suggest that bone marrow pB-Mφprecursors seed the peritoneal and pleural cavity as well as the gastrointestinal system P3 – MφTR-inflam P4 – mo-Mφinflam during homeostasis and that these circulating blood precursors P5 – pB-MφTR-inflam P6 – pB-Mφinflam enter the sites of infection/injury and generate Mφsduringin- C flammation. Notably, despite their different origins, pB-Mφsare

C5. Monocyte/Mφ C7. Monocyte transcriptomically similar to embryonic Mφs during steady state/ C24. Ribosome homeostasis but are phenotypically distinct from those generated by pB-Mφprecursors during inflammation, thus emphasizing the precursors C16. Cell cycle functional plasticity of pB-Mφ , which is likely to be de- termined by the tissue environment and the inflammatory signals C4. B-1>B-2 cell they receive. C6. All Mφ> C10. Neutrophil> φprecursors inflam inflam mo-Mφ /pB-Mφ Other myeloid Consistent with the absence of IL7R on pB-M ,in CELL BIOLOGY

C2. B cell vitro assays revealed the propensity of these cells to differentiate

C3. B cell/Neutrophil down a myeloid lineage. Indeed, reconstitution of tissue-resident C8. All myeloid Mφs in lethally irradiated mice was observed by adoptively C1. Neutrophils C9. Monocyte/ transferring these early pro-B cells, but not LMPPs or CLPs. In Neutrophils terms of the origin of these cells, bone marrow pB-Mφprecursors as + + − Fig. 8. Pro-B-cell–derived Mφs possess transcriptome profiles similar to those well as the CD19 YFP V-DJ intermediate Mφs in the perito- of their embryonic and monocyte-derived counterparts. (A)Inflammationwas neum express CD19, as extensively verified by a number of induced in Mb-1.iCre/Rosa26-YFP mice by i.p. injection of 0.1 mg zymosan. methods including single-cell profiling. CD19 expression on Flow cytometry reveals YFP− embryonic MφsTR-naive (P1) and YFP+ pro-B-cell– these cells, which is strictly controlled by the transcription factor derived pB-MφsTR-naive (P2) in naive peritoneum. PHK-Red injected before PAX5, suggests their close developmental relationship with ++ ++ peritonitis labels resident phagocytes and reveals four CD11b F4/80 Mφ B cells (41). It must be emphasized that these precursors lack − − + subsets at resolution (72 h): YFP resident [embryonic, YFP PKH (MφsTR-inflam)] detectable Ig rearrangements, and they are unable to generate − − − (P3) and YFP inflammation-induced Mφs [monocyte-derived, YFP PKH – + B-lineage cells in B-cell promoting culture conditions. In that (mo-Mφsinflam)] (P4) along with two groups of pro-B-cell–derived YFP case, it is likely that these biphenotypic macrophage precursors Mφs generated during homeostasis, namely resident pB-Mφs[YFP+PKH+ + − are committed to the myeloid lineage but transiently express (pB-MφsTR-inflam)] (P5) and inflammation-induced pB-Mφs[YFP PKH (pB-Mφsinflam)] 3D some B-lineage markers. We believe that this mechanism is an (P6). The software tool BioLayout Express was used for the visualization and additional source of tissue Mφs and is an intrinsic part of normal analysis of Illumina Mus musculus 6v2 microarray data, which were variance hematopoiesis in vivo. In summary, the discovery of Mφ subsets stabilized (VST) and robust spline normalized (RSN) using the lumi R/Bio- φ Conductor package. Correlation networks, sample comparisons, and identifi- arising from pro-B cells reveals an alternative source of M s 3D during homeostasis and inflammation. These findings signifi- cation of coexpression modules were performed in BioLayout Express for φ 30 datasets (n = 3 per group). (B) Diagram of sample similarity analysis showing cantly advance and expand our understanding of M biology and cell type-specific clustering of datasets. Analysis was performed using a Pearson hematopoiesis, the plasticity of hematopoietic precursors, and correlation threshold of r ≥0.97 and the Markov clustering (MCL) algorithm the heterogeneity of Mφ subsets. with an inflation value of 3.0. Individual clusters of nodes (samples) were ar- bitrarily assigned a color. (C) Network graph generated using a Pearson cor- Materials and Methods relation threshold of r ≥0.95 and MCL clustering algorithm with an inflation Animals. Male, 6- to 8-wk-old C57BL/6 mice (Harlan), B6.129P2-Cd19 < tm1 value of 1.7 shows clusters of genes correlating in their expression profiles and (cre)Cgn/J, B6.Cg-Tg(Cr2-cre)3Cgn/J mice (Jackson Laboratories), Rosa26-YFP includes clusters of genes with similar function and/or genes with cell-specific mice (kindly provided by Ulla Dennehy, University College London, London), expression pattern. SI Appendix,TableS1provides the content of each of these B6.CD45.1 mice (kindly provided by B. Stockinger, National Institute for clusters. Medical Research, London), and B6.Mb1-iCre mice (kindly provided by M. Reth, Max Plank Institute of Immunobiology and Epigenetics, Freiburg, Germany) were housed and bred in pathogen-free conditions. All ex- within early pro-B cells that contributes significantly to Mφs periments were performed in compliance with United Kingdom Home populations in vivo, representing a normal facet of hematopoiesis. Office regulations. Although YFP expression was detected in some CLPs, very few mature T cells (<0.5%) or thymic progenitors (∼0.15%) Peritonitis and Resident Phagocytic Cell Labeling. Self-resolving peritonitis was + expressed YFP. We suspect that the great majority of YFP CLPs induced by i.p. injection of sterile zymosan A from Saccharomyces cerevisiae − + + + (Lin IL7Rα Sca-1 c-Kit ) are committed toward a B-cell dif- (Sigma) solution in PBS at a dose of 0.1 mg per mouse. Peritoneal cells were collected from naive animals at 4, 16, 24, 48, and 72 h after inflammation. ferentiation pathway because YFP expression within these CLPs For labeling of resident peritoneal phagocytes, 0.5 mL of 0.5 μM PKH-Red is enabled by Cd79a (Mb-1) promoter activation driven, in turn, fluorescent cell linker (Sigma) solution in diluent B (Sigma) was injected i.p. by E2A and EBF transcription factors, which initiate early B-cell before the induction of peritonitis according to the manufacturer’s lymphopoiesis (40). Indeed, the complete absence of EYFP ex- instructions.

Audzevich et al. PNAS Early Edition | 9of10 Downloaded by guest on September 29, 2021 Bone Marrow Adoptive Transfer. C57BL/6 mice (CD45.2) were lethally irradiated 700, and CD127(IL7Ra)-PE-Cy7 from eBioscience. The following biotinylated with single dose of 950 rad and were reconstituted with 2.5 × 106 donor bone antibodies for lineage markers were purchased from eBioscience: CD3(17A2), marrow cells. Each mouse received an i.v. injection of a mixed population of CD4(GK1.5), CD8(53-6.7), CD19(1D3), B220(RA3-6B2), IgM(R6-60.2), Gr1(RB6- + FACS-purified YFP bone marrow B cells from Mb1-iCre/Rosa26-YFP mice and 8C5), CD11b(M1/70), Ter119(TER119), and NK1.1.(PK136). Live/Dead viability + total CD45.1 bone marrow cells from a congenic wild-type strain at a dye was purchased from Invitrogen. Flow cytometry data analysis was per- + + + + + − 1:10 ratio. Total early pro-B cells (YFP B220 CD19 CD43 CD24 BP-1 ) con- formed with FlowJo vX.0.7 software (Tree Star) using fluorescence minus one + ++ ++ precursors taining biphenotypic YFP CD16/32 CD11b pB-Mφ or mature bone control samples as reference for setting gates. + + − marrow B cells (YFP B220highCD19 CD43 CD24low) were analyzed for their φ + potential to reconstitute tissue M s. CD45.1 total bone marrow cells were Statistical Analysis. Statistical evaluation of data was performed in GraphPad injected into animals to improve the survival rates following lethal irradia- Prism (GraphPad Software) by an unpaired t test in case of two groups and by tion. Hematopoietic reconstitution was analyzed by flow cytometry analysis one-way ANOVA using Bonferroni posttest. A P value less than 0.05 was in peritoneum and spleen at 2, 4, and 6 wk following irradiation and considered statistically significant (*P < 0.05, **P < 0.01, ***P < 0.001). Sta- adoptive transfer. tistical analysis of BCR sequencing and microarray data was performed using R project software. Flow Cytometry and Cell Sorting. Flow cytometry and cell sorting were per- For assessment of BCR rearrangements, detection of B-cell– and Mφ-specific formed on an LSR-II/LSR-Fortessa flow cytometer and FACSAria sorter (BD mRNA transcripts amplified by RT-PCR, microarray analysis, and hematopoietic Biosciences), respectively. For flow cytometry analysis, 0.5–1 × 106 of freshly cell isolation, please see SI Appendix, Supplementary Materials and Methods. isolated cells reconstituted in FACS buffer and incubated with Fc-Blocker (AbD Serotec) were stained in a final volume of 50 μLfor20–30 min and then were washed three times with 150 μL of FACS buffer. For FACS sorting, cells were ACKNOWLEDGMENTS. We thank Dr. Brigitta Stockinger of the National Institute for Medical Research for supplying Rag2−/− mice; Dr. Michael Reth resuspended at 10 × 106 cells/40 μL of FACS buffer. The following antibodies of the Max Plank Institute of Immunobiology and Epigenetics for supplying were used: CD11b-PerCP-Cy5.5/V450/Alexa-700, CD19-Alexa Fluor 700/PE, Mb-1iCre mice; Ulla Dennehy of University College, London for supplying CD23-PE-Cy7, CD21-APC, IgM-PE/PE-Cy7/Biotin, Gr1-APC/PE, B220-Alexa-700/ Rosa26-YFP mice; and Dr. Edward Tsao for advice and training in RNA prep- Horizon-V500, BP-1(Ly-51)-Biotin, CD24(HSA)-PE-Cy7, and Streptavidin-APC/ aration. The work carried out in this report was sponsored by a Wellcome Horizon-V450/Horizon-V500 from BD Pharmingen; CD19-FITC, B220-RPE, and Trust Senior Research Fellowship (to D.W.G.) and Institute Strategic Pro- F4/80-APC from AbD Serotec; MHC-II-PE/FITC/Alexa-700, F4/80-APC/PE-Cy7/ gramme grants from the Biotechnology and Biological Sciences Research PerCP-Cy5.5, CD115-PE/APC, CD16/32(FcγR)-PerCP-Cy5.5, CD34-Alexa Fluor Council (to N.A.M. and T.C.F.).

1. Ginhoux F, Jung S (2014) Monocytes and macrophages: Developmental pathways and 22. Hardy RR, Carmack CE, Shinton SA, Kemp JD, Hayakawa K (1991) Resolution and tissue homeostasis. Nat Rev Immunol 14:392–404. characterization of pro-B and pre-pro-B cell stages in normal mouse bone marrow. 2. Gordon S, Taylor PR (2005) Monocyte and macrophage heterogeneity. Nat Rev J Exp Med 173:1213–1225. Immunol 5:953–964. 23. Hardy RR, Hayakawa K (2001) B cell development pathways. Annu Rev Immunol 19: 3. Bain CC, et al. (2013) Resident and pro-inflammatory macrophages in the colon 595–621. represent alternative context-dependent fates of the same Ly6Chi monocyte pre- 24. Rolink A, Melchers F (1991) Molecular and cellular origins of B lymphocyte diversity. cursors. Mucosal Immunol 6:498–510. Cell 66:1081–1094. 4. Hoeffel G, et al. (2015) C-Myb(+) erythro-myeloid progenitor-derived fetal monocytes 25. Borrello MA, Phipps RP (1999) Fibroblast-secreted macrophage colony-stimulating give rise to adult tissue-resident macrophages. Immunity 42:665–678. factor is responsible for generation of biphenotypic B/macrophage cells from a sub- 5. Gomez Perdiguero E, et al. (2015) Tissue-resident macrophages originate from yolk- set of mouse B lymphocytes. J Immunol 163:3605–3611. + sac-derived erythro-myeloid progenitors. Nature 518:547–551. 26. Borrello MA, Palis J, Phipps RP (2001) The relationship of CD5 Blymphocytestomacrophages: – 6. Rosas M, et al. (2014) The transcription factor Gata6 links tissue macrophage phe- Insights from normal biphenotypic B/macrophage cells. Int Rev Immunol 20:137 155. notype and proliferative renewal. Science 344:645–648. 27. Takahashi K, et al. (1998) Effects of granulocyte/macrophage colony-stimulating 7. Okabe Y, Medzhitov R (2014) Tissue-specific signals control reversible program of factor on the development and differentiation of CD5-positive macrophages and localization and functional polarization of macrophages. Cell 157:832–844. their potential derivation from a CD5-positive B-cell lineage in mice. Am J Pathol 152: – 8. Schulz C, et al. (2012) A lineage of myeloid cells independent of Myb and hemato- 445 456. poietic stem cells. Science 336:86–90. 28. Rickert RC, Roes J, Rajewsky K (1997) B lymphocyte-specific, Cre-mediated muta- – 9. Ginhoux F, et al. (2010) Fate mapping analysis reveals that adult microglia derive from genesis in mice. Nucleic Acids Res 25:1317 1318. primitive macrophages. Science 330:841–845. 29. Shinkai Y, et al. (1992) RAG-2-deficient mice lack mature lymphocytes owing to in- ability to initiate V(D)J rearrangement. Cell 68:855–867. 10. Ceredig R, Rolink AG, Brown G (2009) Models of haematopoiesis: Seeing the wood for 30. Hardy RR, Carmack CE, Shinton SA, Kemp JD, Hayakawa K (2012) Resolution and the trees. Nat Rev Immunol 9:293–300. characterization of pro-B and pre-pro-B cell stages in normal mouse bone marrow. 11. Kawamoto H, Katsura Y (2009) A new paradigm for hematopoietic cell lineages: 1991. J Immunol 189:3271–3283. Revision of the classical concept of the myeloid-lymphoid dichotomy. Trends Immunol 31. Stables MJ, et al. (2011) Transcriptomic analyses of murine resolution-phase macro- 30:193–200. phages. Blood 118:e192–e208. 12. Cumano A, Paige CJ, Iscove NN, Brady G (1992) Bipotential precursors of B cells and 32. Sultan AM, Dunn CJ, Mimms PC, Giroud JP, Willoughby DA (1978) The leucocoyte macrophages in murine fetal liver. Nature 356:612–615. disappearance reaction in non-immune acute inflammation. J Pathol 126:221–230. 13. Wada H, et al. (2008) Adult T-cell progenitors retain myeloid potential. Nature 452: 33. Melnicoff MJ, Morahan PS, Jensen BD, Breslin EW, Horan PK (1988) In vivo labeling of 768–772. resident peritoneal macrophages. J Leukoc Biol 43:387–397. 14. Adolfsson J, et al. (2005) Identification of Flt3+ lympho-myeloid stem cells lacking 34. Freeman TC, et al. (2007) Construction, visualisation, and clustering of transcription erythro-megakaryocytic potential a revised road map for adult blood lineage com- networks from microarray expression data. PLOS Comput Biol 3:2032–2042. mitment. Cell 121:295–306. 35. Hume DA, Summers KM, Raza S, Baillie JK, Freeman TC (2010) Functional clustering 15. Heavey B, Charalambous C, Cobaleda C, Busslinger M (2003) Myeloid lineage switch and lineage markers: Insights into cellular differentiation and gene function from of Pax5 mutant but not wild-type B cell progenitors by C/EBPalpha and GATA factors. large-scale microarray studies of purified primary cell populations. Genomics 95: – EMBO J 22:3887 3897. 328–338. 16. Xie H, Ye M, Feng R, Graf T (2004) Stepwise reprogramming of B cells into macro- 36. Mikkola I, Heavey B, Horcher M, Busslinger M (2002) Reversion of B cell commitment – phages. Cell 117:663 676. upon loss of Pax5 expression. Science 297:110–113. 17. Almeida SR, et al. (2001) Mouse B-1 cell-derived mononuclear phagocyte, a novel cellular 37. Reynaud D, Lefort N, Manie E, Coulombel L, Levy Y (2003) In vitro identification of – component of acute non-specific inflammatory exudate. Int Immunol 13:1193 1201. human pro-B cells that give rise to macrophages, natural killer cells, and T cells. Blood 18. Popi AF, et al. (2009) Co-ordinated expression of lymphoid and myeloid specific 101:4313–4321. transcription factors during B-1b cell differentiation into mononuclear phagocytes in 38. Bussmann LH, et al. (2009) A robust and highly efficient immune cell reprogramming vitro. Immunology 126:114–122. system. Cell Stem Cell 5:554–566. 19. Popi AF, Osugui L, Perez KR, Longo-Maugéri IM, Mariano M (2012) Could a B-1 cell 39. Montecino-Rodriguez E, Leathers H, Dorshkind K (2001) Bipotential B-macrophage derived phagocyte “be one” of the peritoneal macrophages during LPS-driven in- progenitors are present in adult bone marrow. Nat Immunol 2:83–88. flammation? PLoS One 7:e34570. 40. Busslinger M, Nutt SL, Rolink AG (2000) Lineage commitment in lymphopoiesis. Curr 20. Hobeika E, et al. (2006) Testing gene function early in the B cell lineage in mb1-cre Opin Immunol 12:151–158. mice. Proc Natl Acad Sci USA 103:13789–13794. 41. Nutt SL, Urbánek P, Rolink A, Busslinger M (1997) Essential functions of Pax5 (BSAP) in 21. Srinivas S, et al. (2001) Cre reporter strains produced by targeted insertion of EYFP pro-B cell development: Difference between fetal and adult B lymphopoiesis and and ECFP into the ROSA26 locus. BMC Dev Biol 1:4. reduced V-to-DJ recombination at the IgH locus. Genes Dev 11:476–491.

10 of 10 | www.pnas.org/cgi/doi/10.1073/pnas.1616417114 Audzevich et al. Downloaded by guest on September 29, 2021