Activation of the myeloid checkpoint inhibitor LILRB3 (ILT5) elicits profound immununomodulation

Despite advances in identifying the key immunoregulatory roles of the human leukocyte immunoglobulin (Ig)-like receptor (LILR) family members, the function of the inhibitory receptor LILRB3 (ILT5, CD85a, LIR3) remains unclear. Studies indicate a predominant myeloid expression; however, high homology within the LILR family and a relative paucity of reagents have hindered progress. To investigate its function and potential immunomodulatory capacity, a panel of LILRB3-specific monoclonal antibodies (mAb) was generated. LILBR3-specific mAb bound to discrete epitopes in either Ig-like domain two or four. LILRB3 ligation on primary human monocytes by agonistic mAb resulted in phenotypic and functional changes, leading to potent inhibition of immune responses in vitro, including significant reduction in T cell proliferation. Importantly, agonizing LILRB3 in humanized mice induced tolerance and permitted efficient engraftment of allogeneic cells. Our findings reveal powerful immunoregulatory functions of LILRB3 and identify it as an important myeloid immune checkpoint receptor.


Introduction
Molecules of the human LILR family, discovered over two decades ago (1, 2), are expressed on leukocytes and are commonly dysregulated in a wide range of pathologies (3,4). There are six activating (LILRA1-6) and five inhibitory (LILRB1-5) LILRs that together regulate immune responses (3). They all display two, or four, homologous C-2 type Ig-like extracellular domains, but differ in their transmembrane and cytoplasmic regions (2, 5).
The LILRBs are proposed to act as immune checkpoints serving to control and limit overt immune responses (3). In agreement with this, LILRB expression is increased in suppressive (also referred to as alternatively activated or M2) macrophages and tolerogenic dendritic cells Among the inhibitory LILRBs, LILRB3 (ILT5/LIR3/CD85a), containing 4 extracellular Iglike domains and 4 intracellular ITIM motifs, represents an attractive immunomodulatory target because of its relative restriction to, and high expression on, myeloid cells (3, 4).
However, due to the lack of specific reagents and model systems, its exact functions and immunoregulatory potential have not been fully explored. In this study, we addressed this by generating a bespoke panel of novel LILRB3-specific mAb, some of which were used to probe the function of LILRB3 in suitable preclinical platforms. Our data demonstrate that LILRB3 activation confers potent immunoinhibitory functions through reprograming and tolerizing of myeloid cells, and suggest that modulating its activity may provide exciting new treatment strategies in various disease settings, such as transplantation.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Generation and characterization of a panel of fully human LILRB3-specific mAb
To study the protein expression and function of LILRB3, LILRB3-specific antibodies were identified from a human antibody phage-display library, n-CoDeR (29,30). Initial alignment analysis of extracellular domains of LILRB1-5 indicated the presence of a limited number of conserved amino acid (a.a.) residues across the LILRB3 ectodomain ( fig. S1A), against which specific mAb could be generated. In this regard, phages binding to the 'target', ectodomain of LILRB3 protein (present in solution, coated to a plastic surface or expressed on cells), and not to the homologous (~65% extracellular homology) 'non-target' LILRB1 ectodomain protein, were selected ( Fig. 1A; fig. S1B). To increase specificity and yield, the cross-reactive phages were initially removed through a pre-selection (negative selection/depletion using 'non-target' proteins), followed by the selection itself (positive selection). Following each selection round, the selected clones were screened against the ectodomains of both LILRB1 and LILRB2 by flourometric microvolume assay technology (FMAT) and enzyme-linked immunosorbent assay (ELISA), and cross-reactive clones were further excluded from the panel. After three rounds of phage panning and enrichment, successful selection of clones specific for LILRB3 was reconfirmed by FMAT and ELISA, with target-specific phage converted to soluble scFv and screened further ( Fig. 1B and C).
Successful clones were selected based on binding to LILRB3 and lack of cross-reactivity to LILRB1 and LILRB2. Selected scFv clones (>200) were then sequenced and tested for binding against primary cells and LILRB transfectants using high throughput flow cytometry ( Fig. 1D). Subsequently, 46 candidate target-specific clones were converted to human IgG1 (hIgG1) and, in addition to screening against LILRB1-3 transfectants, to exclude those with potential broader LILR cross-reactivity, were screened against a larger panel of LILRexpressing cell lines (Fig. 1E). Due to cross-reactivity to one or more other LILR family members, as exemplified by clone A30 (Fig. 1E, bottom panel), 30 mAb clones were further excluded at this stage. In total a panel of 16 LILRB3-specific antibodies were identified for further study. These LILRB3-specific clones were further tested and confirmed to have no cross reactivity to the mouse orthologue, PIR-B (data not shown). A selection of these mAb were then fluorochrome-labelled and used to determine the LILRB3 expression profile of human peripheral blood; demonstrating predominant staining of monocytes and to a lesser extent granulocytes (Fig. 1F and G), in agreement with previous reports (2,3,5). The immunophenotyping also revealed that LILRB3 expression was significantly higher on circulatory CD14 +hi /CD16 -'classical' and CD14 + /CD16 +low 'intermediate' monocytes compared with the more inflammatory CD14 + /CD16 +hi 'non-classical' monocytes ( Fig. 1F and G).
The selected LILRB3 mAb were then tested for their specific binding properties. Surface plasmon resonance (SPR) analysis showed that all LILRB3-specific clones bound to recombinant LILRB3-hFc protein in a dose-dependent manner (as represented by A16; Fig.   2A) and displayed a range of affinities (Table 1). Interestingly, all mAb had similar association rates (~10 5 ), but varied in their dissociation rates by three orders of magnitude (~10 -3 -10 -6 ).
Cell surface epitope mapping studies were then performed and compared to a commercial mAb (clone 222821), with a series of LILRB3 extracellular domain (D) mutants displaying either all four extracellular domains (wild-type [WT]), three, two, or one domain, transiently transfected into HEK293F cells. Two distinct groups of mAb were identified: those that bound to the WT, D3 and D2 expressing cells (including clone 222821); and those that bound only the WT-transfected cells, indicating mAb binding within D2 (exemplified by A12) or D4 (exemplified by A1), respectively (Fig. 2B). Although conserved a.a. residues are present throughout the ectodomain (fig. S1A), the selected mAb were shown to bind either within D2 or D4 (6/16 and 10/16 clones, respectively; Table 1), perhaps indicating improved accessibility for these regions within the 3D structure. In agreement with this, subsequent blocking assays confirmed that a number of D2-binding mAb reduced the binding of the commercial mAb, suggesting a shared or related epitopes; whilst others did not, confirming binding to discrete epitopes (Fig. 2C and Table 1).
Subsequently, reporter cells transfected with a chimeric receptor expressing the extracellular domain of LILRB3, fused with the human CD3ζ cytoplasmic domain, were used to investigate whether the generated mAb were able to crosslink the receptor. Cross-linking results in the production of NFAT activation and the subsequent expression of GFP and is indicative of agonistic potential (31). Using these cells, we were able to identify two distinct groups of LILRB3 mAb, those with 'agonistic' activity capable of inducing signaling upon binding to the receptor (e.g., A1) and those which were inert (e.g., A28) (Fig. 2D).
Collectively, these data demonstrate that highly specific, fully hIgG1 mAb were raised against LILRB3, amenable for the comprehensive evaluation of LILRB3 function.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

LILRB3 ligation modulates T cell activation and proliferation
Accordingly, using a select number of mAb, we sought to investigate the immunomodulating effect of the LILRB3 mAb on cellular effector functions. LILRB1 has previously been shown to inhibit T cell responses; either by causing dephosphorylation of the CD3 signaling cascade, or competing with CD8 for HLA-I binding (32,33). Moreover, LILRBs can indirectly inhibit T cell responses by rendering antigen-presenting cells (APC), such as monocytes and DCs tolerogenic (13,16,34). To investigate the immunomodulatory potential of LILRB3 and its ability to regulate adaptive immune responses, we utilized a T cell proliferation assay incorporating fresh PBMCs isolated from healthy human donors, as before (35). Fcγ receptors (FcγRs) help mediate the effects of human IgG (36), therefore, to study the direct F(ab):LILRB3-mediated effects of the mAb on T cell proliferation, they were first deglycosylated to reduce FcγR-IgG interactions. SDS-PAGE confirmed a decrease in molecular weight of deglycosylated mAb compared to WT controls, indicative of successful deglycosylation (Fig. 3A). The mAb were then introduced to a T cell proliferation assay where CD3 and CD28 antibodies elicit cell clustering and CFSE dilution, indicative of a significant increase in CD8 + T cell proliferation, compared to non-treated controls ( Fig. 3B and C). Clone A1, shown to be an agonist (Fig. 2D), significantly inhibited CD8 + T cell proliferation in this assay when compared to the isotype control ( Fig. 3B and C). Other LILRB3-specific mAb had either no, or subtle effects, as represented by clones A16 and A28.
These data demonstrate that LILRB3 ligation by agonistic mAb suppresses T cell responses; whereas, other clones confer no inhibitory effects. Similar effects were also observed when considering CD3 + CD8 -T cells (predominantly CD3 + CD4 + T cells; Fig. 3C and not shown).
When the assay was repeated with isolated T cells, no inhibition was seen, confirming that APCs within the PBMC mixture, most likely monocytes, were responsible for the effects observed ( fig. S2), as expected, given the lack of expression of LILRB3 on T cells ( Fig. 1F and G).

LILRB3 ligation induces immune tolerance in humanized mice
Given these data showing that T cells could be suppressed following LILRB3 ligation on myeloid cells, we next investigated the possible effects of LILRB3 modulation in an allogeneic engraftment model using humanized mice, previously reconstituted with primary human fetal hematopoietic stem/progenitor cells (HSPC) (Fig. 4A). Characterization of was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 6, 2020. ; https://doi.org/10.1101/2020.06.04.135400 doi: bioRxiv preprint peripheral blood leukocytes and bone marrow of adult humanized mice demonstrated that LILRB3 was expressed on and restricted to the myeloid cells, but not lymphocytes, similar to humans ( Fig. 4B and fig. S3). We recently showed that allogeneic human lymphoma cells are readily rejected in humanized mice due to HLA mismatch (37). To test the potential of LILRB3 ligation to suppress the allogeneic immune response, adult humanized mice were treated with the agonistic LILRB3 mAb (A1) and the engraftment of allogeneic human B cell lymphoma cells, derived from an unrelated donor (37, 38), monitored overtime (Fig. 4A).
LILRB3 mAb treatment was able to induce a state of tolerance in vivo and led to a successful engraftment of the donor allogeneic cells (Fig. 4C). Accordingly, LILRB3-treated tumorbearing humanized mice subsequently succumbed to disease with high tumor burden, whereas, isotype control-treated mice readily rejected the lymphoma cells without morbidity ( Fig. 4D). These observations corroborate our in vitro functional assays and identify LILRB3 as a key regulator of immune tolerance in an allotransplant setting. Given the expression pattern of LILRB3 on myeloid but not lymphocytic cells in both the human PBMC and humanized mice, we sought to explore the effects of the LILRB3 mAb on these cells.

LILRB3 ligation leads to transcriptional modification and M2-skewing of human APCs
To investigate the pathways and factors involved in LILRB3-mediated immunosuppression, we next investigated the transcriptomic changes in monocytes following LILRB3 engagement. Short-term (~18 hour) in vitro treatment of freshly-isolated human peripheral CD14 + monocytes with the agonistic LILRB3 mAb (A1) caused a dramatic shift in their phenotype ( Fig. 5A), with the cells displaying an elongated morphology resembling immunosuppressed M2 macrophages (39). In accordance with this, RNAseq analysis revealed that ligation of LILRB3 on monocytes induced a signature resembling 'M2-skewed' immunosuppressive macrophages (Fig. 5B). Concurrently, the expression of genes associated with 'M1-skewed' immunostimulatory macrophages was downregulated in LILRB3-ligated monocytes ( Fig. 5B-C). These data were confirmed by qPCR for a number of the differentially regulated genes on a further 6 donors (Fig. 5D). As further evidence, we showed that the effects were dependent upon LILRB3 agonism as treatment of monocytes with a non-agonistic LILRB3 mAb (A28), despite binding the same domain, did not affect monocyte phenotype or gene expression (Fig. 5D). Gene set enrichment analysis (GSEA) of the RNAseq data showed a positive correlation with gene signatures reported for suppressive macrophages, e.g., oxidative phosphorylation (40). Conversely, LILRB3-ligated monocyte was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 6, 2020. ; https://doi.org/10.1101/2020.06.04.135400 doi: bioRxiv preprint gene signatures negatively correlated with those reported for inflammatory macrophages, e.g., IFN-γ and IFN-α responsive elements, as well as allograft rejection (Fig. 5E), in-line with our in vivo observations (Fig. 4). In summary, these data confirm that LILRB3 activation results in significant phenotypic and transcriptional alterations in human primary myeloid cells, leading to potent inhibition of downstream immune responses (Fig. 6).

Discussion
We previously demonstrated that ligation of LILRB1 on human monocytes induces a tolerogenic phenotype, hindering T cell responses (16,41). In this study, we investigated another inhibitory LILR family member, LILRB3, whose function, largely due to lack of suitable reagents and experimental systems, is not yet fully determined. Limited previous studies investigated the consequences of LILRB3 activation on granulocytes and have demonstrated its inhibitory function on neutrophils (42) and basophils (43) in culture. Here, we largely concentrated on myelomonocytic cells and the subsequent regulation of adaptive immune responses. We, therefore, initially generated and characterized an extensive panel of fully human mAb with specificity for LILRB3 through a number of stringent panning and selection processes. Those clones showing unexpected cross-reactivity to other human LILR family members were excluded. Immunoprofiling of circulatory leukocytes from healthy donors using these highly specific mAb confirmed the reported expression of LILRB3 on myelomonocytic and granulocytic cells, but not on lymphocytes (2,3,5). This pattern of expression on myeloid and granulocytic but not lymphoid cells was confirmed in a large cohort of independent donors (>50), suggesting that, despite the polymorphic nature of LILRB3 (2,8,44), the selected antibodies recognize many, if not all, variants, which is important for the development of these reagents for therapeutic applications. Subsequent analysis showed that the LILRB3 mAb displayed a range of affinities, albeit all within the nanomolar (nM) range, with similar on-rates, but off-rates differing over three orders of magnitude. K D values in the low nM range are generally considered to be viable drug candidates; rituximab, for example, has an 8 nM affinity for its target, CD20 (45). This suggests that the LILRB3 mAb generated here have potential as therapeutic agents. However, as LILRB3 shares high sequence homology (>95%) in its extracellular domain with LILRA6, there is a possibility that some LILRB3 mAb may also recognize shared epitopes on LILRA6, if co-expressed (8). These initial data might receive further evidence from other reagents as well as investigation as to whether LILRA6 protein is detectable in leukocyte subsets, e.g., using proteomics approaches similar to a recent study with neutrophils (42). Epitope mapping experiments revealed that the specific LILRB3 mAb reported here were generated against two specific ectodomains, either Ig-like domain two or four. Interestingly, none of the generated specific LILRB3 mAb bound to Ig-like domains one or three, suggesting that these domains may not contain epitopes that are unique for LILRB3, or more likely those unique residues are not exposed/accessible. Collectively, these data confirm that our LILRB3 mAb 1 1 will be useful tools for dissecting LILRB3's molecular mechanisms and may additionally have therapeutic benefits in relevant pathologies.
The ability of the LILRB3 mAb to influence T cell responses was variable: ranging from inhibition to indications of modest increases in proliferation, supportive of agonistic or blocking properties, respectively. Similar to LILRB1 (16,17,46), these effects are likely through manipulations of APCs, specifically monocytes, as they are the only cells expressing LILRB3 in the culture. In support of this, the agonistic LILRB3 mAb did not suppress T cell proliferation in the absence of monocytes. Binding epitopes influence the ability of mAb to modulate receptor function in many systems (36, 47) and so it was unsurprising to see LILRB3 mAb capable of differing functions. However, the D4-binding A1 mAb was a strong inhibitor of proliferation; whereas, other D4-binding mAb (e.g., A28) had no significant effect. Therefore, domain-specific epitopes did not seem to correlate directly with LILRB3 mAb-mediated effector cell functions and may not be predictive of LILRB3 mAb function per se. Further detailed analyses, e.g., surface alanine scanning mutagenesis (47) and/or structural studies, are required to define the specific extracellular epitopes engaged by the selected LILRB3 mAb and to investigate their influence on receptor activity.
Our observations demonstrating immunoinhibitory activities downstream of LILRB3 were further confirmed in the reconstituted humanized mouse model. In this system, where LILRB3 is present only on the hematopoietic cells, and predominantly monocytes, in the absence of appreciable numbers of neutrophils, ligation of LILRB3 with an agonistic LILRB3 mAb prior to injection of allogeneic lymphoma cells (37, 38) induced tolerance in vivo and enabled subsequent tumor engraftment. This demonstrates the capacity of LILRB3 to exert profound immunosuppressive effects that may be exploited in therapeutic settings, such as autoimmunity and transplantation, where transient induction of immune tolerance will be beneficial.
Although typically regarded as an orphan receptor, our earlier studies suggest that LILRB3 may associate with cytokeratin-associated proteins such as those exposed on necrotic cancer cells (31). Others have also identified angiopoietin-like protein 5 and bacteria, such as was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 6, 2020. ; https://doi.org/10.1101/2020.06.04.135400 doi: bioRxiv preprint Staphylococcus aureus, as a source of potential ligands (48,49). Therefore, our data provide a strong mechanism of action whereby such endogenous or pathogenic ligands may be able to subvert immune responses by ligating LILRB3 during an ongoing immune response.
To investigate the pathways and factors involved in LILRB3-mediated immunosuppression, we investigated the transcriptomic changes in isolated peripheral myeloid cells following LILRB3 activation. Over one hundred genes were differentially regulated in primary human monocytes following LILRB3 ligation, some of which are known to be modulated in M2 macrophages and TAMs (12, 50). Amphiregulin (AREG) was among the genes whose expression was significantly upregulated in LILRB3-ligated monocytes. AREG is an epidermal growth factor-like growth factor, responsible for inducing tolerance and immunosuppression, via various mechanisms including enhancement of Treg activity (51).
Furthermore, AREG is overexpressed in tumor-associated DCs (52) and suppressive/M2 macrophages (53) and has been suggested to play a crucial role in immunosuppression and cancer progression (54). Such LILRB3-inducible factors may be responsible for the suppression observed in our T cell assays. Our ongoing efforts aim to interrogate these findings further and define the mechanisms responsible for LILRB3-mediated suppression of immune responses at molecular and cellular levels, e.g., via siRNA knockdown of AREG in monocytes and validation of differentially regulated genes in the humanized mouse models.
A recent study investigating the mode of action of Glatiramer acetate (Copaxone), a peptidebased drug licensed in the late 1990's, used to treat patients with the relapsing-remitting form of multiple sclerosis that ameliorates autoimmunity, identified LILRB2 and LILRB3 as potential ligands (55). On the other hand, blocking human LILRB2 with antagonistic mAb on human myeloid cells is able to promote their pro-inflammatory activity and enhance antitumor responses in preclinical models (17); and a LILRB2 mAb (MK-4830) recently entered phase I clinical trials (NCT03564691) for advanced solid tumors. Furthermore, recent data by Zhang and colleagues suggest that LILRB4 signaling in leukemia cells mediates T cell suppression and supports tumor cell dissemination to distal organs (56). These recent compelling reports further support our findings, demonstrating that activation of human LILRB3 induces immunosuppression via reprogramming of myeloid cells (i.e., reducing M1like maturation and promoting suppressive function).
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 6, 2020.

3
In conclusion, our findings show that LILRB3 activation on primary human myeloid cells exerts potent immunoinhibitory functions and that LILRB3-specific mAb are potentially powerful immunomodulatory agents, with broad application ranging from transplantation to autoimmunity and beyond, where fine-tuning of immune responses through myeloid cell activity is desired.
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 6, 2020.

Ethics Statement
All research with human samples and mice was performed in compliance with institutional

Generation of LILRB3 antibodies
Generation of LILRB3-specific mAb was performed using the nCoDeR phage-display library (29). Three consecutive panning rounds were performed, as well as a pre-panning step. In the panning, human (h) Fc fusion proteins containing the extracellular domains of LILRB1 and 1 5 LILRB3 (LILRB-hFc) were used as 'non-target' or 'target', respectively. These proteins were produced in transiently transfected HEK293 cells followed by purification on protein A, as described previously (36). CHO-S cells transiently transfected to express the various LILRB proteins were also used as targets/non-targets in the panning.
In panning 1, BioInvent n-CoDeR® scFv were selected using biotinylated in-house produced recombinant LILRB3-hFc recombinant fusion proteins (captured with streptavidin-coated Dynabeads®) with or without competition or LILRB1-hFc coated to etched polystyrene balls (Polysciences, US) or plastic immunotubes. Binding phages were eluted by trypsin digestion and amplified on plates using standard procedures (57). The amplified phages from panning 1 were used for panning 2, the process repeated, and the amplified phages from panning 2 used in panning 3. In the third panning round, however, amplified phages from all 3 strategies were combined and selected against LILRB-expressing CHO-S cells, prior to making the final LILRB3-specific mAb selection.
Next, the LILRB3-positive scFv from the enriched phage repertoires from panning 3 were recloned to allow soluble scFv expression in E. coli. The soluble scFv fragments expressed by individual clones were tested for binding against LILRB-transfected CHO-S cells using FMAT, and recombinant LILRB protein by ELISA. This allowed the identification of clones binding specifically to LILRB3. Clones were then further reduced in a tertiary screen against CHO-S cells expressing LILRB1-3 and primary cells (PBMCs) using a high throughput flow cytometry screening system and data analyzed by TIBCO Spotfire® software (TIBCO, USA).
Clones showing specific patterns of binding to LILRB3 were sequenced, yielding LILRB3specific mAb.

Production of full-length IgG
The unique scFv identified above were cloned into a eukaryotic expression system allowing transient expression of full-length IgG in HEK293-EBNA cells. The antibodies were then purified from the culture supernatants using Protein A affinity chromatography as previously described (36).

Production of domain mutant constructs
Using wild-type LILRB3 cDNA isolated from a healthy donor PBMCs, a series of domain mutant DNA constructs were generated by overlap PCR to express 1, 2 or 3 LILRB3 Ig-like domains (with domains identified based on annotations in Uniprot) for comparison to WT LILRB3 (4 domains). The gene constructs were then cloned into pcDNA3.

Preparation of human leukocytes
PBMC were isolated from leukocyte blood cones (Blood Transfusion Services, Southampton General Hospital) by gradient density centrifugation using lymphoprep (Axis-Shield, UK) and used for subsequent experiments, as before (58).
Cells were stained for 30 minutes at 4 o C and then washed twice, first in 10% red blood cell (RBC) lysis buffer (Serotec, UK) for PBMCs or 1x Erythrolyse RBC Lysing Buffer (Biorad, UK) for whole blood, and then with FACS wash (PBS, 1% BSA, 10 mM NaN 3 ), before acquisition on a FACSCalibur or FACSCanto II (BD Biosciences, USA) and analyzed with FlowJo software (BD Biosciences, USA).
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 6, 2020. ; https://doi.org/10.1101/2020.06.04.135400 doi: bioRxiv preprint For assays to determine if mAb bound to similar cross-blocking epitopes 1 x 10 6 PBMCs were blocked with 2% human AB serum for 10 minutes and stained with 10 μ g/ml unconjugated LILRB3 mAb for 30 minutes at 4°C. The cells were then stained with directlyconjugated LILRB3 mAb (clone 222821; mouse IgG2a; R&D Systems, UK) for 20 minutes at 4°C, before washing and acquisition using a FACSCalibur.
For LILRB3 epitope mapping studies, LILRB3-domain mutant-transfected HEK293F cells were stained with the relevant LILRB3 mAb for 25 minutes at 4°C, washed twice, stained with an anti-human-PE secondary (Jackson ImmunoResearch, USA) for 20 minutes at 4°C, before washing and acquisition using a FACSCalibur.

Surface plasmon resonance (SPR)
SPR was performed with the Biacore T100 (GE Healthcare, UK) as per the manufacturer's instructions. LILRB3-hFc recombinant protein (extracellular LILRB3 domain with a hFc tag) was used as the ligand and immobilized by amine coupling onto a series S sensor chip (CM5).
Various LILRB3 mAb were used as 'analytes' and flowed across the chip, and SPR measured.

T cell Proliferation assay
PBMCs (1-2 x 10 7 ) were labelled with 2 µM carboxyfluorescein succinimidyl ester (CSFE) at room temperature for 10 minutes. An equal volume of FCS was then added to quench labeling for 1 minute, prior to washing. Cells were subsequently resuspended in serum-free CTL medium (Immunospot, Germany) and plated at 1x10 5 cells/well in a 96-well roundbottom plate (Corning, UK). Cells were then stimulated with 0.02 µg/ml CD3 (clone OKT3, was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Hematopoietic stem/progenitor cell (HSPC) isolation and generation of humanized mice
Humanized mice were generated, as described (37). In brief, human fetal livers were obtained from aborted fetuses at 15-23 weeks of gestation, in accordance with the institutional ethical guidelines (Advanced Bioscience Resources, Inc., USA). All women gave written informed consent for the donation of their fetal tissue for research. Fresh tissue was initially cut into small pieces and digested with collagenase VI (2 mg/ml; Roche) for 30 minutes at 37°C.
Single-cell suspensions were prepared by passing the digested tissue through a 100 μ m cell strainer (BD Biosciences, USA). HSPCs were purified using a CD34 + selection kit (Stem Cell Technologies, Canada); the purity of CD34 + cells was 90%-99%. Viability was determined through trypan blue exclusion of dead cells. All cells were isolated under sterile conditions and injected into NSG mice.
NSG mice were purchased from the Jackson Laboratories (Bar Harbor, USA) and maintained under specific pathogen-free conditions in the animal facilities at MIT. To reconstitute mice, newborn pups (less than 2 days old) were irradiated with 100 cGy using a Gamma radiation source and injected intracardially with CD34 + cells (~2 × 10 5 cells/recipient), as reported previously (37). Around 12 weeks later, human leukocyte cell reconstitution of PBMCs was determined by flow cytometry and calculated as follows: % human CD45 + cell / (% human CD45 + cell + % mouse CD45 + cell). Mice with ≥ 40% human CD45 + leukocytes were used in subsequent experiments.

In vivo allograft assay
Fully reconstituted humanized mice were injected with 200 µg LILRB3 mAb (clone A1) or an isotype-matched (hIgG1) control on day 0 and day 4, i.v. and i.p, respectively. On day 7, cohorts of mice were injected i.p. with 1 x 10 7 luciferase-positive human ' double-hit' B cell lymphoma cells (37, 38), derived from unrelated donors. Lymphoma cell growth was was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Mice with palpable tumors were sacrificed and Kaplan-Meier survival curves plotted.

Transcriptome analysis
To assess LILRB3-mediated transcriptional changes on monocytes, human peripheral blood monocytes were isolated from freshly prepared PBMCs taken from healthy donors using an

Quantitative PCR (qPCR)
Probe-based qPCR was used to amplify cDNA in 20μl reactions performed in triplicate for each sample condition in a PCR plate (Bio-Rad, UK), as per the manufacturer's protocol. The 96-well plate was run on a C1000 Thermal Cycler CFX96 Real-time System PCR machine was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 6, 2020. ; https://doi.org/10.1101/2020.06.04.135400 doi: bioRxiv preprint 0 (Bio-Rad, Kidlington, UK). The CFX manager software (Bio-Rad, Kidlington, UK), was used for data acquisition and analysis of gene expression initially recorded as cycle threshold values (Ct). The Ct values were normalised to housekeeping gene Glyceraldehyde 3phosphate dehydrogenase (GAPDH) and standardised to gene expression levels in isotype control-treated samples.

Statistics
Paired two-tailed T-tests were performed for both the immunophenotyping and T cell proliferation data; straight bars indicate median values. On bar graphs, where at least 3 experiments were performed, error bars represent standard deviation. Kaplan-Meier plots were analyzed by Log-rank test. One-way ANOVA with Bonferroni's multiple comparisons test were performed for qPCR data analysis. Statistical analysis was performed using GraphPadPrism (v6-8).
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 6, 2020. ) .
was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 6, 2020.  ,  T  h  e  O  r  p  h  a  n  I  m  m  u  n  e  R  e  c  e  p  t  o  r  L  I  L  R  B  3  M  o  d  u  l  a  t  e  s  F  c  R  e  c  e  p  t  o  r  -M  e  d  i  a  t  e  d  F  u  n  c  t  i  o  n  s  o  f  N  e  u  t  r  o  p  h  i  l  s  .   J  I  m  m  u  n  o  l   2  0  4  ,  9  5  4  -9  6  6  (  2  0  2  0  )  .  4  3  .  D  .  E  .  S  l  o  a  n  e   e  t  a  l  .   ,  L  e  u  k  o  c  y  t  e  i  m  m  u  n  o  g  l  o  b  u  l  i  n  -l  i  k  e  r  e  c  e  p  t  o  r  s  :  n  o  v  e  l  i  n  n  a  t  e  r  e  c  e  p  t  o  r  s  f  o  r  h  u  m  a  n  b  a  s  o  p  h  i  l  a  c  t  i  v  a  t  i  o  n  a  n  d  i  n  h  i  b  i  t  i  o  n  .   B  l  o  o  d   1  0  4  ,  2  8  3  2  -2  8  3  9  (  2  0  0  4  )  .  4  4  .  A  .  A  .  B  a  s  h  i  r  o  v  a   e  t  a  l  .   ,  D  i  v  e  r  s  i  t  y  o  f  t  h  e  h  u  m  a  n  L  I  L  R  B  3  /  A  6  l  o  c  u  s  e  n  c  o  d  i  n  g  a  m  y  e  l  o  i  d  i  n  h  i  b  i  t  o  r  y  a  n  d  a  c  was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 6, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020  was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 6, 2020. ; https://doi.org/10.1101/2020.06.04.135400 doi: bioRxiv preprint  was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 6, 2020. ; https://doi.org/10.1101/2020.06.04.135400 doi: bioRxiv preprint  was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 6, 2020. ; https://doi.org/10.1101/2020.06.04.135400 doi: bioRxiv preprint on hCD45 + peripheral blood hCD14 + myeloid cells; isotype control in pink, LILRB3 in blue. (C) The effect of agonistic LILRB3 mAb on engraftment of allogeneic cells in humanized mice. Age-and sex-matched humanized mice were injected with 200 µg LILRB3 mAb (clone A1) or an isotype-matched (hIgG1) control mAb (iso ctrl) on day 0 and 4, i.v. and i.p, respectively. On day 7, mice were injected i.p. with 1x10 7 non-autologous luciferase + human lymphoma cells. Lymphoma cell growth was monitored over time using an IVIS imager; and (D) humanized mice were sacrificed upon the development of signs of terminal tumor development. Survival data was analyzed using log-rank test (p < 0.01); representative data from 3 independent experiments (3 individual HSPC donors) shown (n = 3 mice/group). Transcriptomic analysis of LILRB3-treated monocytes reveals upregulation of M2-associated genes compered to controls. RNA was extracted from cells following mAb treatment (~18 hours) and subjected to RNA sequencing. Red depicts genes that were significantly upregulated and green depicts genes that were significantly downregulated compared to isotype control treated-cells (n = 4 independent donors). (C) Ligation of LILRB3 on primary human CD14 + monocytes induces M2-polarized genes. GSEA graph showing a significant enrichment for M2-polarizing genes in LILRB3-treated monocytes versus isotype control, respectively. UP; upregulated, NES; normalized enrichment score = -1.68; FWER; familywise-error rate p <0.001. (D) qPCR analysis of selected genes following LILRB3 ligation on monocytes using an agonistic LILRB3 mAb (A1), a non-agonistic LILRB3 mAb (A28) or an isotype control (iso). Data were normalized to GAPDH mRNA levels and standardized to the levels of isotype control-treated monocytes. Fold difference data were log 10 transformed. One-way ANOVA with Bonferroni's multiple comparisons test was performed, n = 5-6 independent donors (* p < 0.005). (E) GSEA analysis showing negative was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 6, 2020. ; https://doi.org/10.1101/2020.06.04.135400 doi: bioRxiv preprint Table 1. Summary of binding domain, blocking potential and affinity measurements ('Univalent' model of 1:1 binding assessed by SPR) for the selected LILRB3-specific mAb (hIgG1). was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which this version posted June 6, 2020. ; https://doi.org/10.1101/2020.06.04.135400 doi: bioRxiv preprint 1 protein coated on plastic polystyrene beads in the first two selection rounds, followed by cells expressing the target in selection around three using all phages from all three techniques used in panning two (C). Throughout the panning rounds, LILRB1 was used as a non-target and as a competitor for LILRB3. 50 nM biotinylated non-target was used in the pre-selection for strategies where biotinylated protein was used (A, B). 10 µg/ml non-target was used for strategies where coated protein was used (C). No pre-selection was required for the cell strategies (3C). Red border indicates completion with excess non-target protein was used. was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.