An Innate Response to Allogeneic Nonself Mediated byMonocytes1

Daniel Zecher,* Nico van Rooijen,† David M. Rothstein,*‡ Warren D. Shlomchik,‡

and Fadi G. Lakkis2*

The mammalian innate immune system has evolved diverse strategies to distinguish self from microbial nonself. How the innateimmune system distinguishes self-tissues from those of other members of the same species (allogeneic nonself) is less clear. Toaddress this question, we studied the cutaneous hypersensitivity response of lymphocyte-deficient RAG�/� mice to spleen cellstransplanted from either allogeneic or syngeneic RAG�/� donors. We found that RAG�/� mice mount a specific response toallogeneic cells characterized by swelling and infiltration of the skin with host monocytes/macrophages and neutrophils. Theresponse required prior priming with allogeneic splenocytes or skin grafts and exhibited features of memory as it could be elicitedat least 4 wk after immunization. Neither depletion of host NK cells nor rechallenging immunized mice with F1 hybrid splenocytesinhibited the response, indicating that the response is not mediated by NK cells. Depletion of host monocytes/macrophages orneutrophils at the time of rechallenge significantly diminished the response and, importantly, the adoptive transfer of monocytesfrom alloimmunized RAG�/� mice conferred alloimmunity to naive RAG�/� hosts. Unlike NK- and T cell-dependent allore-sponses, monocyte-mediated alloimmunity could be elicited only when donor and responder mice differed at non-MHC loci. Theseobservations indicate that monocytes mount a response to allogeneic nonself, a function not previously attributed to them, andsuggest the existence of mammalian innate allorecognition strategies distinct from detection of missing self-MHC molecules by NKcells. The Journal of Immunology, 2009, 183: 7810–7816.

I mmunity in most studied vertebrates is dependent on twointegrated immune systems, the innate and adaptive, whoseprincipal function is to protect the host against infection (1).

Cells of the innate immune system express germline-encoded pat-tern recognition receptors that detect conserved molecular patternson microbes (1, 2). Activation of these cells by microbes leads toan acute inflammatory response that limits the infection but doesnot necessarily eliminate it. Cells of the adaptive immune system,in contrast, express a highly diverse repertoire of Ag receptors,generated by RAG-mediated gene rearrangements, that specificallyrecognize microbial peptides presented to them in the context ofMHC molecules on activated innate cells. Once triggered by theinnate immune system, the adaptive system eliminates the microbeand provides long-term protection of the host against re-infection(immunological memory).

Vertebrate immunity, however, is not restricted to antimicrobialresponses. All vertebrates studied so far are capable of allorecog-nition, the ability to discriminate between self tissues and those of

other members of the same species (3, 4). Allorecognition in mam-mals is best exemplified by the rejection of transplanted cells, tis-sues, or organs (allografts) and by graft vs host disease after bonemarrow transplantation (5). It is also observed under natural con-ditions such as the maternal response to paternal alloantigens onfetal cells and the rejection of transmissible allogeneic tumors (6–9). In the adaptive immune system, T lymphocytes are the princi-pal cell type responsible for allorecognition due to their ability todetect molecular variations in MHC molecules and/or the endog-enous peptides bound to them (10).

How the innate immune system recognizes allogeneic nonself isless clear. Cells of the innate system are activated within minutesto hours after transplanting allogeneic tissues to humans or exper-imental animals and are present in large numbers at the feto-ma-ternal interface in the gravid uterus (11, 12). The prevailing viewis that innate immune cells, with the exception of NK cells, do notengage in specific allorecognition but instead respond to inflam-matory ligands released by stressed or dying cells at the time oftransplantation (13, 14). These ligands are not unique to allograftsbut are common to self and nonself tissues subjected to hypoxic orother forms of injury (15, 16). NK cells, in contrast, engage in“missing self recognition” in that they are inhibited by self-MHCmolecules but become more amenable to activation by stimulatoryligands when they encounter MHC-deficient or MHC-mismatchedtargets (17–19). Although NK cells contribute to innate allorecogni-tion of MHC-mismatched grafts (20, 21), they are not expected torecognize allogeneic tissues transplanted between MHC-matched in-dividuals. Therefore, a major gap in our understanding is whetherinnate cells other than NK cells do or do not distinguish between selfand allogeneic nonself and whether such allorecognition would bedependent on or independent of MHC polymorphisms.

Another impetus to explore lymphocyte-independent allore-sponses is the intriguing observation that the rejection of allo-geneic tissues is common to many invertebrate organisms that

*Thomas E. Starzl Transplantation Institute, Departments of Surgery, Immunology,and Medicine, University of Pittsburgh, Pittsburgh, PA 15261; †Department of Mo-lecular Cell Biology, Faculty of Medicine, Free University, 1100 DE Amsterdam, TheNetherlands; and ‡Departments of Medicine and Immunobiology, Yale UniversitySchool of Medicine, New Haven, CT 06520

Received for publication July 9, 2009. Accepted for publication October 7, 2009.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work was funded by National Institutes of Health Grant AI064343 (to D.M.R.,W.D.S., and F.G.L.). D.Z. was funded by a grant from the DeutscheForschungsgemeinschaft.2 Address correspondence and reprint requests to Dr. Fadi G. Lakkis, University ofPittsburgh, Thomas E. Starzl Transplantation Institute, W1548 Thomas E. Starzl Bio-medical Science Tower, 200 Lothrop Street, Pittsburgh, PA 15261. E-mail address:lakkisf@upmc.edu

Copyright © 2009 by The American Association of Immunologists, Inc. 0022-1767/09/$2.00

The Journal of Immunology

www.jimmunol.org/cgi/doi/10.4049/jimmunol.0902194

predate the evolution of T, B, and NK cells. The ubiquity ofrejection responses, particularly among colonial marine inver-tebrates, has led to the proposal that allorecognition evolvedbefore the emergence of vertebrate adaptive immune systems,characterized by polymorphic MHC molecules and re-arranginglymphocyte receptors, and that it serves to protect the organismagainst stem cell parasitism (22–25). The recent identificationof specific allorecognition systems in two colonial marine or-ganisms, Botryllus and Hydractinia (26, 27), and the character-ization of their role in preventing stem cell chimerism (28, 29),have confirmed these earlier suspicions. To investigate whetherthe mammalian immune system mounts an alloimmune re-sponse independent of lymphocytes, we studied the cutaneousreaction of RAG�/� mice to cells transplanted from allogeneicdonors. We show in this study that these mice mount an allo-immune response independent of T, B, and NK cells. The re-sponse requires prior priming with allogeneic nonself, is medi-ated by monocytes, exhibits features of memory, and is elicitedby MHC-unlinked genetic disparities between the donor andresponder mouse strain.

Materials and MethodsMice

C57BL/6 (B6), BALB/c, B6.C-H2d/bByJ (B6.C), C3H/HeJ, B6 RAG1�/�,BALB/c RAG1�/�, and congenic B6 SJL RAG1�/� (CD45.1�) mice werepurchased from The Jackson Laboratory. B6 and BALB/c RAG2�/� werepurchased from Taconic Farms. (C57BL/6 � BALB/c)F1 RAG1�/� micewere generated by breeding the parental strains at the University of Pitts-burgh animal facility. All mice were housed in a specific pathogen-freeenvironment and experiments were performed in accordance with the In-stitutional Animal Care and Use Committee of the University of Pittsburgh.RAG1�/� and RAG2�/� mice were used interchangeably and are referredto as RAG�/� mice throughout. RAG�/� mice were tested periodically byperforming serum IgM ELISA and flow cytometry on either blood orspleen cells to confirm absence of Abs, T cells, and B lymphocytes.

Abs and reagents

Biotin-conjugated mAbs against CD45.1 (clone A20), CD4 (L3T4), CD8b(Ly-3), CD19 (MB19–1), F4/80 (BM8), pan-NK (DX5), CD11b (M1/70),CD11c (HL3), I-Ad (AMS-32.1), Ter119 (Ly-76), and CD16/32 (2.4G2); Al-exa750-conjugated anti-CD45.2 (104); allophycocyanin-conjugated anti-F4/80(BM8); phycoerythrin-conjugated anti-CD115 (CSF1-R, ASF98); and PacificBlue-conjugated Streptavidin were obtained from eBioscience. FITC-conju-gated anti-Gr-1 (RB6–8C5) and purified anti-CD16/32 (2.4G2) were from BDPharmingen. Ethidium monoazide was from Invitrogen. Phycoerythrin-conju-gated polyclonal anti-NKp46 was from R&D Systems. NK cell-depleting Ab(PK136), neutrophil-depleting Ab (anti-Ly6G, 1A8), and the correspondingisotype control Abs were from BioXCell. Clodronate was a gift from RocheDiagnostics and was incorporated into liposomes by one of the co-authors(NvR) as described (30).

In vivo alloresponse

Five to 10-wk-old male mice were immunized by injecting 2 � 107 allo-geneic splenocytes s.c. in the neck or by applying an allogeneic tail skingraft to the dorsoventral aspect of the mouse. Unless otherwise indicated,RAG�/� mice were used as both cell donors and recipients to avoid con-tamination with T and B lymphocytes. In experiments where wild-type(wt)3 mice served as source of splenocytes, the splenocytes were purged ofT and B by incubation with anti-CD16/32 (Fc block) and biotin-conjugatedmAbs against CD8b, CD4, and CD19, followed by binding to streptavidin-coated microbeads and separation by AutoMACS (Miltenyi Biotec). Con-trol mice were either left unimmunized or were grafted with syngeneic tailskin. One or four weeks later, baseline thickness of the ear pinnae wasmeasured with a microcaliper (Mitutoyo), and the mice were rechallengedwith 5 � 106 allogeneic splenocytes in the dorsal surface of one ear andsyngeneic splenocytes in the opposite ear using a 28 1/2-gauge needle (BDBiosciences). Cells were suspended in 20 �l PBS before injection. Theensuing ear swelling was measured by an experienced operator at 6, 12, 24,and 48 h after rechallenge. Swelling is reported as either � ear swelling (ear

thickness at given time point minus baseline) or as allo-dependent earswelling (� swelling of ear challenged with allogeneic splenocytes minus� swelling of ear challenged with syngeneic splenocytes).

Analysis of ear-infiltrating leukocytes

Excised ears were pooled from two to three mice per group, separatedalong the cartilage, digested in RPMI 1640 plus 1 mg/ml collagenase IVplus 0.01 mg/ml DNase I plus 0.25 mg/ml hyaluronidase (all from Sigma-Aldrich) for 1 h at 37°C, and purified by gradient centrifugation (Lym-pholyte M, Cedarline Laboratories) before flow cytometric analysis. Fortissue analysis, 6-�m acetone-fixed sections were stained with biotin-con-jugated anti-CD45.1 and unconjugated anti-Ly6G followed by streptavi-din-conjugated Cy3 (Jackson ImmunoResearch Laboratories) and anti-ratAlexa488 (Invitrogen). Sections were counterstained with 4�,6-diamidino-2-phenylindole (Molecular Probes).

NK, macrophage, and neutrophil depletion

In vivo depletion of host NK, monocyte/macrophages, or neutrophils wasaccomplished by injecting 250 �g anti-NK1.1 (PK136) mAb i.p., 250 �lclodronate-loaded liposomes i.v., or 250 �g anti-Ly6G (1A8) mAb i.p.,respectively. Control mice received isotype control Ab or PBS-loaded li-posomes. Injections were given once, 16 h before rechallenging mice withsplenocytes. Depletion of NK cells before immunization was done analo-gous to the above protocol except that anti-NK1.1 mAb and correspondingisotype Ab were administered 7 days before immunization. Depletion ofspecific leukocyte populations was confirmed by flow cytometric analysisof blood, bone marrow, and spleen (supplemental Fig. S1).4

Sorting and transfer of monocytes

BALB/c RAG�/� mice were immunized with either syngeneic or alloge-neic (B6) splenocytes. Seven days later, spleens were removed and F4/80lowGr-1int monocytes were high-speed sorted using an ARIA instrument(BD Biosciences) to �95% purity. In brief, 2.0–2.5 � 105 monocytes fromeither group were injected i.v. via the tail vein into naive BALB/c RAG�/�

mice 24 h before measuring the alloresponse as described above.

Statistical analysis

All data are presented as mean � SEM and were analyzed by paired orunpaired Student’s t test or one-way ANOVA. Bonferroni multiple-com-parison test was applied when indicated. Significance was set at p � 0.05.

ResultsEvidence for an allospecific response in RAG�/� mice

To study alloresponsiveness in the innate immune system, we ex-amined the response of recombination activating gene-deficient(RAG�/�) mice, which lack mature T and B lymphocytes (31), toallogeneic splenocytes. BALB/c RAG�/� mice were immunizedwith splenocytes from genetically disparate C57BL/6 (B6)RAG�/� mice and, 7 d later, were rechallenged by injecting allo-geneic (B6) and syngeneic (BALB/c) RAG�/� splenocytes in op-posite ear pinnae. This assay is analogous to the delayed typehypersensitivity (DTH) reaction widely used to study T cell im-munity in vivo (32). As shown in Fig. 1a, rechallenging immu-nized RAG�/� mice with allogeneic cells caused significantlygreater ear swelling than rechallenge with syngeneic cells. Theresponse peaked as early as 6 h after rechallenge and declinedgradually thereafter. The magnitude and kinetics of the RAG�/�

response were different from that of wt mice. The latter exhibitedmore pronounced ear swelling that peaked at 24 h (Fig. 1b), con-sistent with the role of effector/memory T lymphocytes in medi-ating DTH to alloantigen in wt mice (33). We then asked whetherprior exposure to allogeneic cells (immunization or priming) isnecessary for eliciting the observed RAG�/� response. To ensurethat only the swelling response attributable to the presence of al-logeneic nonself is analyzed, swelling in the ear injected with syn-geneic cells was subtracted from that injected with allogeneic cells.The difference is reported as allo-dependent ear swelling in this

3 Abbreviations used in this paper: wt, wild type; DTH, delayed type hypersensitivity. 4 The online version of this article contains supplementary material.

7811The Journal of Immunology

and all subsequent experiments. As shown in Fig. 1c, naive (un-immunized) RAG�/� mice exhibited minimal allodependent earswelling (primary response) compared with previously immunizedmice (recall response). The response could be elicited even if themice were rechallenged with allogeneic splenocytes 4 wk afterimmunization, suggesting longevity of the response.

To investigate the antigenic specificity of the RAG�/� allore-sponse and to test whether it can be elicited by priming with non-hematopoietic tissue allografts, we placed either syngeneic(BALB/c), allogeneic (B6), or allogeneic “third-party” (C3H) tailskin grafts onto BALB/c RAG�/� recipients. Seven days later,transplanted mice were rechallenged with B6 and BALB/cRAG�/� splenocytes in the ears, and allodependent swelling wascalculated. Although minimal or no allodependent responses wereobserved in mice immunized with syngeneic skin allografts, al-lodependent swelling was readily detectable in mice immunizedwith allogeneic B6 skin and, to a lesser extent, in those immunizedwith “third-party” C3H skin (Fig. 1d). This finding is somewhatdifferent from T lymphocyte-mediated DTH responses in wt mice,which are elicited only if the host is rechallenged with cells fromthe same mouse strain it had encountered before (34). The resultsshown in Fig. 1, therefore, indicate that the innate immune systemresponds to allogeneic nonself independent of the adaptive system.This response requires prior priming with either hematopoietic ornonhematopoietic allografts and has features of memory, but doesnot display complete alloantigenic specificity in the combinationstested.

Leukocyte recruitment in response to allogeneic nonself

To begin to examine which cells participate in the innate al-loresponse of RAG�/� mice, we quantitated and phenotyped

host leukocytes that infiltrated the ear pinnae six and 12 h afterinjecting either syngeneic or allogeneic RAG�/� splenocytes.Host leukocytes infiltrating or residing in the ear were distin-guished from injected cells by the congenic marker CD45.1. Asshown in Fig. 2a, significantly more host leukocytes werepresent in ears injected with allogeneic than syngeneic spleno-cytes. Additional flow cytometric analysis revealed that the in-filtrating leukocyte population consisted mainly of neutrophils(F4/80�Ly6C�), monocytes (F4/80intLy6C�), and macro-phages (F4/80highLy6C�) (Fig. 2b). NK cells (NKp46�) werealso present but in smaller proportions (flow cytometry plots notshown). Whereas NK cell numbers were comparable, macro-phages, monocytes, and neutrophils accumulated to a larger ex-tent in ears injected with allogeneic as compared with syngeneicsplenocytes (Fig. 2, c and d).

The RAG�/� alloresponse is independent of NK cells

NK cells respond to cells or tissues with which they do not shareMHC molecules (missing self reaction) and, in addition, havebeen recently shown to mediate a memory response to chemicalhaptens independent of the adaptive immune system (19, 35).We therefore investigated whether NK cells are responsible forthe alloresponse of RAG�/� mice. Two independent experi-mental approaches failed to demonstrate a role for NK cells inthis response. First, BALB/c RAG�/� mice immunized with B6splenocytes reacted equally to (BALB/c � B6)F1 and B6RAG�/� splenocytes despite the fact that host NK activation isinhibited by the shared H-2d MHC allele expressed on F1 cells(Fig. 3a). Second, NK depletion of the host (B6 RAG�/� mousein this case) either throughout the experiment (data not shown)

FIGURE 1. Evidence for an allospecific response in RAG�/� mice. a, BALB/c RAG�/� mice were immunized with B6 RAG�/� splenocytes s.c. and1 wk later were rechallenged with allogeneic B6 RAG�/� (allo) and syngeneic BALB/c RAG�/� (syn) splenocytes in opposite ears. Ear thickness wasmeasured at the indicated time points. Baseline swelling (measured before splenocyte injection) was subtracted to determine � swelling; n 5–6/group �7 experiments (pooled data shown). b, Delayed-type hypersensitivity response of BALB/c wt mice immunized with B6 wt splenocytes and rechallengedwith B6 wt (allo) and BALB/c wt (syn) splenocytes as in a; n 6/group � 1 experiment. c, The RAG�/� alloresponse requires prior priming and exhibitsmemory. BALB/c RAG�/� mice were injected in opposite ears with BALB/c RAG�/� (syn) and B6 RAG�/� (allo) splenocytes either without priorimmunization (primary) or 1 or 4 wk after immunization with B6 RAG�/� splenocytes (recall). After calculating � swelling, the syngeneic response wassubtracted from the allogeneic response to determine allo-dependent swelling; n 5– 6/group � 8 experiments (pooled data shown). d, Priming ofRAG�/� mice can be achieved by skin grafting and exhibits coarse allospecificity. BALB/c RAG�/� mice were grafted with BALB/c, B6, or C3H(3rd party) wt tail skin. One week later, mice were rechallenged in the ears with syngeneic (BALB/c) or allogeneic (B6) RAG�/� splenocytes, andthe allo-dependent response was calculated as in c; n 3– 4/group � 3 experiments (pooled data shown). Bar graphs are mean � SEM; �, p � 0.05;��, p � 0.01; ���, p � 0.001.

7812 ALLOIMMUNITY MEDIATED BY MONOCYTES

or only before rechallenge (Fig. 3b) did not alter the allore-sponse to BALB/c RAG�/� spleen cells. We also tested thepossibility that injected (donor) NK cells mount a graft-vs-hostreaction that may contribute to the observed ear swelling.(BALB/c � B6)F1 mice were immunized with B6 RAG�/�

splenocytes and rechallenged 7 days later with either syngeneic(F1) or allogeneic (B6) RAG�/� splenocytes in the ear. In thissetting, host F1 cells express the B6 MHC allotype (H-2b) andinhibit activation of donor (B6) NK cells. As shown in Fig. 3c,F1 hosts mounted an allospecific response to B6 splenocytesequivalent to that of control BALB/c hosts which do not express

H-2b. This finding strongly argues against NK-mediated graft vshost reaction as a contributor to the ear swelling response.

Phagocytic cells and neutrophils mediate the recall phase of theRAG�/� alloresponse

To study the role of non-NK, innate immune cells in the allore-sponse of RAG�/� mice, we administered clodronate-loaded lipo-somes or anti-Ly6G lytic Ab to immunized mice to deplete phago-cytic cells or neutrophils, respectively. Depletion of either cell typebefore rechallenge blunted allodependent ear swelling by �50%,

FIGURE 2. Leukocyte recruitment in response to alloantigens: Leukocyte accumulation in ears of immunized CD45.1� B6 RAG�/� after rechallengewith either syngeneic or allogeneic CD45.2� Rag�/� splenocytes is shown. Twelve hours after rechallenge (unless indicated otherwise), cells wereextracted from ears and host CD45.1� leukocytes were enumerated and phenotyped by flow cytometry (a–c), or visualized by immunostaining (d). Naive unmanipulated ears from naive mice; NK NK cells (NKp46�), M macrophages (F4/80high Ly6C�); Mono monocytes (F4/80int Ly6C�), PMN polymorphonuclear neutrophils (F4/80neg Ly6C�). Data in a and c are pooled from three independent experiments. Bar graphs are mean � SEM; �, p �0.05; ��, p � 0.01; ���, p � 0.001. Sections in d were stained with pan-hematopoietic marker CD45.1 (red), neutrophil-specific mAb Ly6G (green) andcounterstained with nuclear marker 4�,6-diamidino-2-phenylindole (blue). Host neutrophils (CD45.1�Ly6G�) are yellow. Magnification �100 (�,�1000).

FIGURE 3. The RAG�/� alloresponse is independent of NK cells. a, Host NK inhibition does not suppress the alloresponse. BALB/c RAG�/� micewere immunized with B6 RAG�/� splenocytes and, 7 days later, were rechallenged with either B6 (allo) or (B6 � BALB/c)F1 (F1) RAG�/� splenocytesin one ear and syngeneic (BALB/c) RAG�/� splenocytes in the contralateral ear to calculate allo-dependent swelling as in Fig. 1c; n 6/group � 1experiment. b, Host NK cell depletion does not suppress the alloresponse. B6 RAG�/� mice were immunized with BALB/c Rag�/� splenocytes and treatedwith either NK-depleting mAb PK136 (�-NK1.1) or istoype control mAb (Isotype) 16 h before rechallenge; n 5/group � 2 experiments (pooled datashown). c, No evidence for an NK-mediated graft-vs-host reaction. F1 RAG�/� mice (F1 responder) were immunized with B6 RAG�/� splenocytes andrechallenged with allogeneic B6 RAG�/� and syngeneic F1 splenocytes in opposite ears. Allo-dependent swelling was compared with historical BALB/cRAG�/� controls (BALB/c responder); n 5/group � 1 experiment. Bar graphs are mean � SEM.

7813The Journal of Immunology

suggesting that both monocytes/macrophages and neutrophils con-tribute to the recall response (Fig. 4, a and b). Administration ofempty (PBS) liposomes did not influence the response. Depletionof monocytes/macrophages by clodronate-loaded liposomes and ofneutrophils by anti-Ly6G was confirmed by flow cytometry (sup-plemental Fig. S1).

Adoptive transfer of alloimmunity by primed monocytes

To identify which innate cell population is primed at the time ofimmunization and therefore confers long-lived alloimmunity toRAG�/� mice, an adoptive cell transfer approach was used. Weharvested spleen cells from BALB/c RAG�/� mice immunized 1wk earlier with allogeneic (B6 RAG�/�) splenocytes, isolated dis-tinct cell populations, and transferred them to naive BALB/cRAG�/� mice. One day later, alloresponsiveness of the adoptivehost was tested in the ear swelling assay. In initial experiments, aspleen cell population enriched for neutrophils and monocytes, butcontaining �2% NK cells, dendritic cells, or macrophages, con-ferred alloresponsiveness to naive RAG�/� hosts (data notshown). Because neutrophils have a short lifespan (�2 days) afterthey exit from the bone marrow (36), we then tested whethermonocytes are responsible for the long-lived alloresponsiveness

observed in our model. We found that F4/80intGr-1int monocytes,high-speed sorted from the spleens of alloimmunized RAG�/�

mice (Fig. 4c), conferred full alloreactivity to naive RAG�/� hosts(Fig. 4d). The monocytic lineage of the transferred population wasconfirmed by morphologic examination and by expression of thesurface molecules CD115 (M-CSF-R) and Ly6C (Fig. 4c). Lessthan 1% of the transferred population expressed the NK cellsmarker NKp46 (Fig. 4c). Control mice that received a similar cellpopulation from mice immunized with syngeneic splenocytes didnot mount an alloresponse (Fig. 4d). These findings demonstratethat monocytes mediate the innate alloresponse in RAG�/� mice.

The RAG�/� alloresponse is dependent on non-MHC disparitiesbetween donor and responder strains

MHC molecules are central to the recognition of allogeneic non-self by T lymphocytes. TCR, generated by RAG-mediated generearrangement, directly recognize allogeneic MHC molecules ontarget cells or, alternatively, react to MHC or non-MHC-derivedallopeptides bound to self-MHC molecules (5). Likewise, al-lorecognition by NK cells is dictated by MHC polymorphisms. NKcells are inhibited if they share the same MHC class I moleculeswith their targets but are released from inhibition if MHC class I

FIGURE 4. Role of non-NK innate cells in the RAG�/� alloresponse. a, Neutrophil depletion before rechallenge significantly reduces the alloresponse.BALB/c RAG�/� mice were immunized and rechallenged with allogeneic and syngeneic cells as in Fig. 1c. Sixteen hours before rechallenge, mice receivedneutrophil-depleting �-Ly6G mAb i.p. or were left untreated; n 6/group � 2 experiments (pooled data shown). b, Monocyte/macrophage depletion beforerechallenge significantly reduces the alloresponse. Experiment was performed as in a, except that mice received clodronate-loaded liposomes (CLL) orPBS-loaded control liposomes (PBS) i.v. 16 h before rechallenge; n 6/group � 1 experiment. c, Surface phenotype and morphologic characteristics ofsplenic monocytes sorted for adoptive transfer. Gated F4/80lowGr-1int cells are CD115�Ly6C�NKp46� (shaded histograms) and exhibit morphologicalcharacteristics characteristic of monocytes. Open histograms represent negative controls (for CD115 and Ly6C) or positive control (for NKp46). Pho-tomicrograph was obtained after cells were high-speed sorted (Giemsa stain; magnification �100). d, Adoptive transfer of sorted F4/80lowGr-1int splenicmonocytes from alloimmunized mice confers alloreactivity to naive mice. Naive BALB/c RAG�/� mice were injected i.v. with 2.0 - 2.5 � 105 syngeneicmonocytes sorted from RAG�/� mice that had been immunized with either allogeneic splenocytes (allo Mono) or syngeneic splenocytes (syn Mono) sevendays earlier. Twenty-four hours later, mice were challenged with syngeneic and allogeneic cells and allo-dependent swelling was determined. n 3–4/group � 2 experiments (pooled data shown). Bar graphs are mean � SEM; �, p � 0.05; ��, p � 0.01.

7814 ALLOIMMUNITY MEDIATED BY MONOCYTES

molecules are not shared (missing self hypothesis) (19). Microbialnonself recognition by nonlymphocytic innate cells, in contrast, isindependent of the MHC and is determined by germline-encodedpattern-recognition receptors (1, 2). We therefore investigatedwhether the allodeterminant(s) responsible for the monocyte-me-diated alloresponse of RAG�/� mice are linked or unlinked to theMHC. To do so, we immunized BALB/c RAG�/� mice (H-2d)with B6 RAG�/� splenocytes and rechallenged them in the earpinnae with either B6 (H-2b) or congenic B6.C (H-2d) splenocytes.The B6.C congenic strain is identical to the B6 strain except for a38 Mb genomic segment that encodes the H-2d MHC locus of theBALB/c strain. As shown in Fig. 5, ear swelling was comparablein mice rechallenged with either B6 or B6.C splenocytes. Con-versely, B6 RAG�/� mice did not respond to B6.C splenocyteswith which they differed at the MHC but shared all other geneticloci (Fig. 5). These findings indicate that the innate alloresponse ofRAG�/� mice is dependent on determinants not linked tothe MHC.

DiscussionWe investigated in this study whether innate immune cells otherthan NK cells distinguish self from allogeneic nonself. We foundthat T and B cell-deficient RAG�/� mice mount a specific re-sponse to allogeneic nonself independent of NK cells. The re-sponse required prior priming with allogeneic splenocytes or skingrafts and exhibited “memory,” features traditionally associatedwith adaptive immunity. Adoptive cell transfer and depletion stud-ies demonstrated that innate alloimmunity in our model is con-ferred by monocytes but that both neutrophils and monocytes/mac-rophages contribute to the effector phase of the response. Finally,we demonstrated that the response is elicited by MHC-unlinkeddisparities between donor and responder. Our findings provide di-rect evidence that monocytes mediate a response to allogeneicnonself, a function not previously attributed to them, and suggestthe presence of MHC-unlinked innate allorecognition systems inthe mouse.

Memory had long been assumed to be a defining feature ofadaptive immunity (37). This view, however, has been challengedby recent studies. In the mouse, NK cells activated by hapten,virus, or cytokines acquire classical features of memory that in-clude extended life span, self-renewal capacity, and ability tomount a robust recall response (35, 38, 39). A memory response to

pathogens has also been observed in invertebrate organisms thatlack mammalian-type adaptive immune systems (40). For exam-ple, Pham et al. (41) demonstrated specific and persistent immu-nity to bacterial pathogens in Drosophila that is mediated byphagocytes. Our finding that monocyte-mediated alloimmunity inRAG�/� mice is long-lived provides the first evidence that phago-cytic cells of the vertebrate innate immune system exhibit featuresof memory. This observation further blurs the divisions previouslythought to exist between innate and adaptive immunity.

The innate alloresponse characterized in our study is distinctfrom that mediated by NK cells as it is not influenced by MHCmatching; instead, it is dependent on mismatching at loci unlinkedto the MHC. It is not currently known what allodeterminants theseloci code for. Theoretically, these could be any of many genesoutside the MHC that are polymorphic among inbred mousestrains. Polymorphic, Ig domain-containing genes or gene familiesthat are expressed in cells of the innate immune system have beenidentified in mice and humans (42). Recently, a role for SIRP�polymorphisms in the rejection of hematopoietic stem cell xeno-grafts by the innate immune system has been reported (43). It is notknown, however, whether SIRP� also serves as an allodetermi-nant. Alternatively, the innate system of RAG�/� mice may rec-ognize differences in glycosylation patterns between mouse strainsrather than amino acid polymorphisms (44). With the exception ofwell-characterized glycomic differences between distant species,best exemplified by the lack of �-1,3-galactosyltransferase in ca-tarrhine primates (including humans) but its presence in most othermammals (45), glycomic differences between conspecifics havenot yet been carefully explored.

It has been proposed that the mammalian innate immune re-sponse to allografts is largely a response to endogenous inflam-matory ligands released by stressed or dying cells at the time oftransplantation (13, 14). These nonpolymorphic ligands, which in-clude uric acid and the nuclear protein high-mobility group box 1,among others, are common to both self and nonself tissues sub-jected to hypoxia and signal through innate pattern recognitionreceptors on innate immune cells (15, 16, 46). The pattern recog-nition receptors include Toll-like receptors and components of theinflammasome, all of which participate in the recognition of mi-crobial products as well. Our study does not exclude that theseubiquitous recognition pathways contribute to the initiation of theadaptive alloimmune response, but provides direct evidence thatinnate immune cells other than NK cells respond either directly orindirectly to non-MHC determinants. Although our data providestrong evidence that innate alloresponsiveness in RAG�/� mice ismediated by monocytes, they do not determine whether monocytesdirectly recognize allogeneic nonself. It is quite possible that othercomponents of the innate immune system recognize allogeneicnonself and then stimulate monocytes.

Specific allorecognition systems that predate the emergence oflymphocytes and the MHC have been identified in colonial marineinvertebrates (26, 47). These systems prevent fusion between ge-netically disparate members of the same species and thus, serve assafeguards against stem cell and germline parasitism (25, 28). It istherefore tempting to speculate that the vertebrate innate allorecog-nition phenomenon described in this study is an echo of ancient butessential allorecognition systems. It is conceivable that monocytesprovide innate immune surveillance that alerts the host to the pres-ence of allogeneic stem cells encountered under either natural orartificial conditions. In mice, blood monocytes patrol healthy tis-sues by crawling along resting endothelium and rapidly migrateinto sites of infection where they differentiate into phagocytic mac-rophages or Ag-presenting dendritic cells independent of either Tor NK cells (48). Monocytes that reside in the spleen have also

FIGURE 5. The RAG�/� alloresponse is dependent on non-MHC dis-parities between donor and responder strains: BALB/c RAG�/� mice im-munized with B6 RAG�/� splenocytes were rechallenged 7 days later withT- and B cell depleted B6 (H-2b) (solid bars) or congenic B6.C (H-2d)(hatched bars) wt splenocytes. Alternatively, B6 RAG�/� mice were im-munized with BALB/c RAG�/� splenocytes and rechallenged 7 d laterwith T and B cell-depleted B6.C (H-2d) wt splenocytes (open bars). B6.Cmice share the same MHC (H-2d) with BALB/c mice but differ at non-MHC loci. Allo-dependent ear swelling was calculated as in Fig. 1c. n 6/group � 1–2 experiments. Bar graphs are mean � SEM; �, p � 0.05; ��,p � 0.01; ���, p � 0.001.

7815The Journal of Immunology

been shown to rapidly deploy to sites of acute inflammation, forexample the infarcted myocardium, where they play a role in tissuerepair (49). In addition, recruitment of monocytes to the peritonealcavity of SCID mice has been observed in response to xenogeneicpig cells and to a significantly lesser extent to allogeneic cells (50).Examples of natural allogeneic encounters in vertebrates includeentry of fetal stem cells into the maternal circulation during preg-nancy, described in humans (6), and the transmission of allogeneictumors between members of the same species, documented in dogsand Tasmanian devils (8, 9). Artificial allogeneic encounters, incontrast, are represented by solid organ and hematopoietic stemcell (bone marrow) transplantation in humans and experimentalanimals. We hypothesize that in these situations monocytes serveas sentinels that continually alert the adaptive immune system tothe presence of allogeneic nonself and could inflict damage on thetarget tissue or organ.

AcknowledgmentsWe thank A. Williams for technical assistance; R. Hoffman and A. Larre-gina for technical advice; H. Shen for cell sorting; and L. W. Buss,S. D. Dellaporta, M. L. Nicotra, and R. Medzhitov for critical reading ofthe manuscript.

DisclosuresThe authors have no financial conflict of interest.

References1. Medzhitov, R. 2007. Recognition of microorganisms and activation of the im-

mune response. Nature 449: 819–826.2. Beutler, B. A. 2009. TLRs and innate immunity. Blood 113: 1399–1407.3. Klein, J. 1975. Function and evolution of the H-2 complex. In Biology of the

Mouse Histocompatibility-2 Complex. Springer-Verlag, New York, Heidelberg,Berlin, pp. 521–536.

4. Hildemann, W. H., and G. H. Thoenes. 1969. Immunological responses of Pacifichagfish, I: skin transplantation immunity. Transplantation 7: 506–521.

5. Sykes, M., K. Wood, and D. H. Sachs. 2008. Transplantation Immunology. InFundamental Immunology, 6th ed. W. E. Paul, ed. Woleters Kluwer/LippincottWilliams & Wilkins, Philadelphia.

6. Herzenberg, L., D. W. Bianchi, J. Schroder, H. Cann, and M. Iverson. 1979. Fetalcells in the blood of pregnant women: detection and enrichment by fluorescence-activated cell sorting. Proc. Natl. Acad. Sci. USA 76: 1453–1457.

7. Bonney, E. A., and P. Matzinger. 1997. The maternal immune system’s interac-tion with circulating fetal cells. J. Immunol. 158: 40–47.

8. Pearse, A. M., and K. Swift. 2006. Allograft theory: transmission of devil facial-tumour disease. Nature 439: 549.

9. Murgia, C., J. Pritchard, S. Kim, A. Fassati, and R. Weiss. 2006. Clonal originand evolution of a transmissible cancer. Cell 126: 477–487.

10. Felix, N. J., D. L. Donermeyer, S. Horvath, J. J. Walters, M. L. Gross, A. Suri,and P. M. Allen. 2007. Alloreactive T cells respond specifically to multiple dis-tinct peptide-MHC complexes. Nat. Immunol. 8: 388–397.

11. LaRosa, D. F., A. H. Rahman, and L. A. Turka. 2007. The innate immune systemin allograft rejection and tolerance. J. Immunol. 178: 7503–7509.

12. Silver, R., M. Peltier, and D. Branch. 2000. The immunology of pregnancy. InMaternal-Fetal Medicine, 5th ed. R. Resnik, and J. Iams, eds. Saunders, NewYork, pp. 89–110.

13. Gallucci, S., M. Lolkema, and P. Matzinger. 1999. Natural adjuvants: endoge-nous activators of dendritic cells. Nat. Med. 5: 1249–1255.

14. Kono, H., and K. L. Rock. 2008. How dying cells alert the immune system todanger. Nat. Rev. Immunol. 8: 279–289.

15. Chen, C. J., H. Kono, D. Golenbock, G. Reed, S. Akira, and K. L. Rock. 2007.Identification of a key pathway required for the sterile inflammatory responsetriggered by dying cells. Nat. Med. 13: 851–856.

16. Tsung, A., J. R. Klune, X. Zhang, G. Jeyabalan, Z. Cao, X. Peng, D. B. Stolz,D. A. Geller, M. R. Rosengart, and T. R. Billiar. 2007. HMGB1 release inducedby liver ischemia involves Toll-like receptor 4 dependent reactive oxygen speciesproduction and calcium-mediated signaling. J. Exp. Med. 204: 2913–2923.

17. Karre, K., H. G. Ljunggren, G. Piontek, and R. Kiessling. 1986. Selective rejec-tion of H-2-deficient lymphoma variants suggests alternative immune defensestrategy. Nature 319: 675–678.

18. Karlhofer, F. M., R. K. Ribaudo, and W. M. Yokoyama. 1992. MHC class Ialloantigen specificity of Ly-49� IL-2-activated natural killer cells. Nature 358:66–70.

19. Lanier, L. L. 2005. NK cell recognition. Annu. Rev. Immunol. 23: 225–274.20. Bix, M., N.-S. Liao, M. Zijlstra, J. Loring, R. Jaenisch, and D. Raulet. 1991.

Rejection of class I MHC-deficient hematopoietic cells by irradiated MHC-mis-matched mice. Nature 349: 329–331.

21. Kitchens, W. H., S. Uehara, C. M. Chase, R. B. Colvin, P. S. Russell, andJ. C. Madsen. 2006. The changing role of natural killer cells in solid organrejection and tolerance. Transplantation 81: 811–817.

22. Hildemann, W. H. 1974. Some new concepts in immunological phylogeny. Na-ture 250: 116–129.

23. Burnet, F. M. 1971. “Self-recognition” in colonial marine forms and floweringplants in relation to the evolution of immunity. Nature 232: 230–235.

24. Klein, J. 1999. Self-nonself discrimination, histoincompatibility, and the conceptof immunology. Immunogenetics 50: 116–123.

25. Buss, L. 1982. Somatic cell parasitism and the evolution of somatic tissue com-patibility. Proc. Natl. Acad. Sci. USA 79: 5337–5341.

26. De Tomaso, A. W., S. V. Nyholm, K. J. Palmeri, K. J. Ishizuka, W. B. Ludington,K. Mitchel, and I. L. Weissman. 2005. Isolation and characterization of a pro-tochordate histocompatibility locus. Nature 438: 454–459.

27. Nicotra, M., A. Powell, R. Rosengarten, M. Moreno, J. Grimwood, F. Lakkis,S. Dellaporta, and L. Buss. 2009. A hypervariable invertebrate allodeterminant.Curr. Biol. 19: 583–589.

28. Laird, D., A. De Tomaso, and I. Weissman. 2005. Stem cells are unit of naturalselection in a colonial ascidian. Cell 123: 1351–1360.

29. Poudyal, M., S. Rosa, A. E. Powell, M. Moreno, S. L. Dellaporta, L. W. Buss, andF. G. Lakkis. 2007. Embryonic chimerism does not induce tolerance in an in-vertebrate model organism. Proc. Natl. Acad. Sci. USA 104: 4559–4564.

30. Van Rooijen, N., and A. Sanders. 1994. Liposome mediated depletion of mac-rophages: mechanism of action, preparation of liposomes and applications. J. Im-munol. Methods 174: 83–93.

31. Mombaerts, P., J. Iacomini, R. S. Johnson, K. Herrup, S. Tonegawa, andV. E. Papaioannu. 1992. RAG-1 deficient mice have no mature B and T lym-phocytes. Cell 68: 869–877.

32. Sunday, M. E., J. Z. Weinberger, B. Benacerraf, and M. E. Dorf. 1980. Hapten-specific T cell responses to 4-hydroxy-3-nitrophenyl acetyl. J. Immunol. 125:1601–1605.

33. Loveland, B. E., and I. F. McKenzie. 1982. Delayed-type hypersensitivity andallograft rejection in the mouse: correlation of effector cell phenotype. Immunol-ogy 46: 313–320.

34. Smith, F. I., and J. F. Miller. 1979. Delayed-type hypersensitivity to allogeneiccells in mice, III: sensitivity to cell-surface antigens coded by the major histo-compatibility complex and by other genes. J. Exp. Med. 150: 965–976.

35. O’Leary, J. G., M. Goodarzi, D. L. Drayton, and U. H. von Andrian. 2006. T cell-and B cell-independent adaptive immunity mediated by natural killer cells. Nat.Immunol. 7: 507–516.

36. Mauer, A. M., J. W. Athens, H. Ashenbrucker, G. E. Cartwright, andM. M. Wintrobe. 1960. Leukokinetic studies, II: a method for labeling granulo-cytes in vitro with radioactive diisopropylfluorophosphate (DFP32). J. Clin. In-vest. 39: 1481–1486.

37. Janeway, C. A., and R. Medzhitov. 2002. Innate immune recognition. Annu. Rev.Immunol. 20: 197–216.

38. Sun, J. C., J. N. Beilke, and L. L. Lanier. 2009. Adaptive immune features ofnatural killer cells. Nature 457: 557–561.

39. Cooper, M. A., J. M. Elliott, P. A. Keyel, L. Yang, J. A. Carrero, andW. M. Yokoyama. 2009. Cytokine-induced memory-like natural killer cells.Proc. Natl. Acad. Sci. USA 106: 1915–1919.

40. Kurtz, J. 2005. Specific memory within innate immune systems. Trends Immunol.26: 186–192.

41. Pham, L. N., M. S. Dionne, M. Shirasu-Hiza, and D. S. Schneider. 2007. Aspecific primed immune response in Drosophila is dependent on phagocytes.PLoS Pathog. 3: e26.

42. Barclay, A. N., and M. H. Brown. 2006. The SIRP family of receptors andimmune regulation. Nat. Rev. Immunol. 6: 457–464.

43. Takenaka, K., T. K. Prasolava, J. C. Wang, S. M. Mortin-Toth, S. Khalouei,O. I. Gan, J. E. Dick, and J. S. Danska. 2007. Polymorphism in Sirpa modulatesengraftment of human hematopoietic stem cells. Nat. Immunol. 8: 1313–1323.

44. Marth, J. D., and P. K. Grewal. 2008. Mammalian glycosylation in immunity.Nat. Rev. Immunol. 8: 874–887.

45. Koike, C., M. Uddin, D. E. Wildman, E. A. Gray, M. Trucco, T. E. Starzl, andM. Goodman. 2007. Functionally important glycosyltransferase gain and lossduring catarrhine primate emergence. Proc. Natl. Acad. Sci. USA 104: 559–564.

46. Zhai, Y., X. Shen, R. O’Connell, F. Gao, C. Lassman, R. Busuttil, G. Cheng, andJ. W. Kupiec-Weglinski. 2004. Cutting edge: TLR4 activation mediates liverischemia/reperfusion inflammatory response via IFN regulatory factor 3-depen-dent Myd88-independent pathway. J. Immunol. 173: 7115–7119.

47. Nicotra, M. L., A. E. Powell, R. D. Rosengarten, M. Moreno, J. Grimwood,F. G. Lakkis, S. L. Dellaporta, and L. W. Buss. 2009. A hypervariable inverte-brate allodeterminant. Curr. Biol. 19: 583–589.

48. Auffray, C., D. Fogg, M. Garfa, G. Elain, O. Join-Lambert, S. Kayal, S. Sarnacki,A. Cumano, G. Lauvau, and F. Geissmann. 2007. Monitoring of blood vesselsand tissues by a population of monocytes with patrolling behavior. Science 317:666–670.

49. Swirski, F. K., M. Nahrendorf, M. Etzrodt, M. Wildgruber, V. Cortez-Retamozo,P. Panizzi, J. L. Figueiredo, R. H. Kohler, A. Chudnovskiy, P. Waterman, et al.2009. Identification of splenic reservoir monocytes and their deployment to in-flammatory sites. Science 325: 612–616.

50. Fox, A., J. Mountford, A. Braakhuis, and L. C. Harrison. 2001. Innate and adap-tive immune responses to nonvascular xenografts: evidence that macrophages aredirect effectors of xenograft rejection. J. Immunol. 166: 2133–2140.

7816 ALLOIMMUNITY MEDIATED BY MONOCYTES