Leukocyte mono-immunoglobulin (Ig)–like receptor 5 (LMIR5)/CD300b is a DAP12-coupled activating receptor predominantly expressed in myeloid cells. The ligands for LMIR have not been reported. We have identified T cell Ig mucin 1 (TIM1) as a possible ligand for LMIR5 by retrovirus-mediated expression cloning. TIM1 interacted only with LMIR5 among the LMIR family, whereas LMIR5 interacted with TIM4 as well as TIM1. The Ig-like domain of LMIR5 bound to TIM1 in the vicinity of the phosphatidylserine (PS)-binding site within the Ig-like domain of TIM1. Unlike its binding to TIM1 or TIM4, LMIR5 failed to bind to PS. LMIR5 binding did not affect TIM1- or TIM4-mediated phagocytosis of apoptotic cells, and stimulation with TIM1 or TIM4 induced LMIR5-mediated activation of mast cells. Notably, LMIR5 deficiency suppressed TIM1-Fc–induced recruitment of neutrophils in the dorsal air pouch, and LMIR5 deficiency attenuated neutrophil accumulation in a model of ischemia/reperfusion injury in the kidneys in which TIM1 expression is up-regulated. In that model, LMIR5 deficiency resulted in ameliorated tubular necrosis and cast formation in the acute phase. Collectively, our results indicate that TIM1 is an endogenous ligand for LMIR5 and that the TIM1–LMIR5 interaction plays a physiological role in immune regulation by myeloid cells.
A growing number of studies have characterized a variety of paired activating and inhibitory receptors (Ravetch and Lanier, 2000; Klesney-Tait et al., 2006; Lanier, 2009). We have previously identified a leukocyte mono-Ig–like receptor (LMIR) mainly expressed in myeloid cells (Kumagai et al., 2003; Izawa et al., 2007; Yamanishi et al., 2008). The mouse LMIR family is also known as the CMRF-35–like molecule/myeloid-associated Ig-like receptor/dendritic cell–derived Ig-like receptor/CD300 family (Luo et al., 2001; Chung et al., 2003; Yotsumoto et al., 2003). LMIR1 and LMIR3 are immunoreceptor tyrosine-based inhibitory motif–containing inhibitory receptors, whereas other members are activating receptors that associate with immunoreceptor tyrosine-based activation motif–containing adaptor proteins (Luo et al., 2001; Chung et al., 2003; Kumagai et al., 2003; Yotsumoto et al., 2003; Izawa et al., 2007; Yamanishi et al., 2008). LMIR5 is a DAP12-coupled activating receptor predominantly expressed in myeloid cells (Yamanishi et al., 2008). However, the ligands for LMIR remained unknown. In this study, we identified T cell Ig mucin 1 (TIM1) as a ligand for LMIR5 by retrovirus-mediated expression cloning (Kitamura et al., 2003).
TIM1–4 are characterized as important regulators of immune responses associated with autoimmunity and allergic diseases (McIntire et al., 2001; Kuchroo et al., 2003, 2008). The TIM molecules are type 1 cell-surface glycoproteins, consisting of an N-terminal IgV domain and a mucin domain. TIM1/hepatitis A virus cellular receptor 1 (Kaplan et al., 1996)/kidney injury molecule–1 (KIM-1; Ichimura et al., 1998) is expressed in activated T cells and delivers a signal that enhances T cell activation and proliferation (Meyers et al., 2005; Umetsu et al., 2005). TIM1 can also interact with itself (Santiago et al., 2007b). In addition, a soluble form of KIM-1/TIM1 is released by shedding (Bailly et al., 2002). On the other hand, TIM4 is expressed in macrophages and dendritic cells and is a natural ligand for TIM1 (Meyers et al., 2005). Interestingly, TIM1 and TIM4 recognize phosphatidylserine (PS) and are critical for the efficient clearance of apoptotic cells (Kobayashi et al., 2007; Miyanishi et al., 2007; Santiago et al., 2007a; Ichimura et al., 2008). Recent reports have demonstrated that the narrow cavity built by the CC’ and FG loops of the Ig-like domain is a binding site for PS (Kobayashi et al., 2007; Santiago et al., 2007a,b). In addition, TIM1/KIM-1 expression is strongly induced in the injured kidney epithelial cells (Ichimura et al., 1998, 2008; Waanders et al., 2010), and confers a phagocytic phenotype on epithelial cells (Ichimura et al., 2008). TIM1 is also a marker for renal tubular damage (Waanders et al. 2010).
In the present study, using biological and biochemical analysis, we demonstrate that TIM1 and TIM4 are endogenous ligands for LMIR5. In addition, we generated LMIR5−/− mice and delineated the physiological significance of the LMIR5–TIM1 interaction by using an acute kidney injury model.
Cloning of the ligand for LMIR5
To identify the LMIR5 ligand, we generated an Fc fusion protein containing the extracellular domain of LMIR5 (LMIR5-Fc). Several hematopoietic cell lines were incubated with LMIR5-Fc, which stained A20 cells but not Ba/F3 cells, as determined by flow cytometric analysis, suggesting the expression of LMIR5 ligand in A20 cells (Fig. 1 A). To identify the surface protein bound by LMIR5-Fc, we used retrovirus-mediated expression cloning (Kitamura et al., 2003). A retrovirus cDNA library constructed from A20 cells was transduced via infection to Ba/F3 cells that were not stained by LMIR5-Fc (Fig. 1 A). The transfectants stained by LMIR5-Fc were sorted and expanded in culture. This cycle of sorting and expansion was repeated three times until LMIR5-Fc stained most cells (Fig. S1 A). After we obtained single-cell clones that were stained with LMIR5-Fc, we isolated TIM1 cDNA from most of these clones by PCR (Fig. S1 B and not depicted). We confirmed that a cell-surface glycoprotein TIM1 was expressed in A20 cells but not in Ba/F3 cells by using anti-TIM1 antibody (Fig. 1 A, bottom). When TIM1 was expressed in Ba/F3 cells, LMIR5-Fc strongly stained TIM1-transduced Ba/F3 but not parental Ba/F3 cells. In addition, LMIR5-Fc–staining levels were correlated with TIM1 expression at both surface protein and mRNA levels (Fig. 1, A and B). Collectively, these results indicated that LMIR5-Fc bound to TIM1. In accordance, pretreatment of LMIR5-Fc with 10 µg/ml anti-LMIR5 antibody, but not control antibody, abolished the binding of LMIR5-Fc to TIM1-expressing Ba/F3 cells (Fig. 1 C, left). Similarly, the preincubation of TIM1-expressing Ba/F3 cells with 100 µg/ml anti-TIM1 antibody (222414) suppressed this binding (Fig. 1 D, left). The binding of LMIR5-Fc to TIM1-expressing Ba/F3 cells was dose-dependently inhibited by anti-LMIR5 or anti-TIM1 antibody (Fig. 1, C and D, right). These observations strongly suggested that LMIR5 interacts directly with TIM1. Then, Fc fusion proteins containing the extracellular domains of LMIR1, 2, 3, and 4 (LMIR1/2/3/4-Fc) were generated. TIM1-expressing Ba/F3 cells were bound only by LMIR5-Fc among the LMIR-Fc fusion proteins (Fig. 1 E). When Ba/F3 cells transduced with Flag-tagged TIM1, 2, 3, and 4 were incubated with LMIR5-Fc, LMIR5-Fc bound to those cells transduced with TIM4 as well as TIM1 among the TIM family (Fig. 1 F). In support of this, coimmunoprecipitation experiments illustrated the physical interaction of LMIR5 with both TIM1 and TIM4 (Fig. 1 G). After generating Fc fusion proteins containing the extracellular domains of TIM1 or 4 (TIM1/4-Fc), we incubated LMIR5- or mock-transduced Ba/F3 cells with TIM1-Fc or TIM4-Fc, which stained LMIR5-transduced Ba/F3 cells more strongly as compared with parental Ba/F3 cells (Fig. S1 C), indicating that TIM1 or TIM4 bound to surface-expressed LMIR5. On the other hand, the fact that TIM1-Fc or TIM4-Fc stained parental Ba/F3 cells at significant levels suggested that ligands for TIM1 or TIM4 other than LMIR5 were expressed in Ba/F3 cells. Our results suggested that TIM1 and TIM4 are possible ligands for LMIR5.
The Ig-like domain of LMIR5 binds to that of TIM1 in the vicinity of the PS-binding site
To determine which region of LMIR5 was required for the interaction with TIM1, we generated LMIR5 deletion mutants (LMIR5 del1/2/3/4/5–Fc; Fig. 2, A and B). Notably, like LMIR5-Fc, LMIR5 del1/2/3–Fc bound to TIM1-expressing Ba/F3 cells, whereas LMIR5 del4/5–Fc lacking the C terminus of the Ig-like domains did not bind at all (Fig. 2 C). These results suggested that the intact Ig-like domain of LMIR5 is indispensable for the LMIR5–TIM1 interaction.
According to recent reports, TIM1 and TIM4 bind to PS through the highly conserved binding cleft (FG–CC’ cleft) of the Ig-like domains (Kobayashi et al., 2007; Miyanishi et al., 2007; Santiago et al., 2007a,b; Ichimura et al., 2008). To clarify the involvement of this region in LMIR5–TIM1 binding, we generated the TIM1 mutants W115A/F116A (WF/AA) or N117A/D118A (ND/AA), because these mutations in the metal ion–dependent ligand-binding site were reported to dampen TIM1–PS binding (Kobayashi et al., 2007; Santiago et al., 2007a). Interestingly, these substitutions completely abolished the binding of LMIR5-Fc to TIM1-expressing Ba/F3 cells (Fig. 2 D). Collectively, these findings indicated that LMIR5 bound to TIM1, presumably through the interaction of the Ig-like domain of LMIR5 with the structural domain formed by FG loops of TIM1. Related to this, we found different capabilities of anti-TIM1 antibodies to recognize TIM1 epitopes. Anti-TIM1 antibody (222414) detected TIM1 WT and the ND/AA mutant but not the WF/AA mutant, whereas anti-TIM1 antibody (RMT1-10) detected TIM1 WT and the mutants WF/AA and ND/AA (Fig. S2 A). Thus, anti-TIM1 antibody (222414) presumably recognizes the FG loop structure that is critically maintained by W115/F116 but not N117/D118, whereas anti-TIM1 (RMT1-10) antibody reacts outside this area of TIM1. Then, we compared the inhibitory effect of these antibodies on LMIR5–TIM1 binding. Unlike anti-TIM1 antibody (RMT1-10) or anti-Flag antibody, anti-TIM1 antibody (222414) pretreatment did inhibit LMIR5–TIM1 binding (Fig. S2 B). Collectively, these results suggested that the Ig-like domain of LMIR5 interacted with TIM1 in the region formed by the FG loop of the Ig-like domain, the structure of which was similar or close to the FG–CC’ cleft bound by PS, and that it was recognized by anti-TIM1 antibody (222414).
LMIR5 neither binds to PS nor affects TIM1- or TIM4-mediated phagocytosis of apoptotic cells
Given the close proximity of the PS- and LMIR5-binding regions in TIM1, we tested whether LMIR5 interacts with TIM1 through PS. To this end, we performed a protein–lipid overlay assay; although TIM1-Fc bound specifically to PS, as previously reported (Kobayashi et al., 2007; Miyanishi et al., 2007), we found no binding of LMIR5-Fc to PS or to other phospholipids (Fig. 3 A). Consistently, NIH3T3 cells transduced with TIM1 but not LMIR5 promoted phagocytosis of the apoptotic cells through recognition of PS (Fig. 3 B). Next, we asked if the LMIR5–TIM1 interaction affected the TIM1-mediated phagocytosis of the apoptotic cells. As expected, coincubation with anti-TIM1 antibody or TIM1-Fc significantly suppressed phagocytosis of the apoptotic cells in TIM1-expressing NIH3T3 cells (Fig. 3 C). In contrast, coincubation with LMIR5-Fc did not inhibit this phagocytosis (Fig. 3 C). Similarly, the interaction of LMIR5 with TIM4 did not affect TIM4-mediated phagocytosis of the apoptotic cells in peritoneal macrophages (Fig. 3, D and E). Collectively, interaction of neither TIM1 nor TIM4 with LMIR5 affected the phagocytosis of apoptotic cells through its recognition of PS, despite the finding that LMIR5 bound to TIM1 or TIM4 at close proximity to the PS-binding site.
The TIM1–LMIR5 interaction induces LMIR5-mediated activation of mast cells
As previously reported (Kitaura et al., 2003; Yamanishi et al., 2008), LMIR5-mediated activation of mast cells depended on LMIR5 expression levels (Fig. 4 A, left). Notably, stimulation with TIM1-Fc, but not human IgG1 as a control, induced significant levels of extracellular signal-regulated kinase (ERK) activation in LMIR5-transduced BM-derived mast cells (BMMCs), whereas no detectable levels of ERK activation were induced by TIM1-Fc in mock-transduced BMMCs (Fig. 4 A, right). Similarly, stimulation with TIM1-Fc or TIM4-Fc induced cytokine production of fetal liver–derived mast cells (FLMCs) transduced with LMIR5, but not mock-transduced cells (Fig. 4 B; and Fig. S3, A and B). In addition, DAP12 deficiency completely dampened cytokine production of LMIR5-transduced FLMCs stimulated by TIM1-Fc, but not PMA as a control (Fig. 4 B). These results indicated that TIM1-Fc or TIM4-Fc activated mast cells through interaction with LMIR5, which was dependent on the expression levels of LMIR5. As previously reported (Nakae et al., 2007), we found high expression levels of TIM3 as well as no detectable expression levels of TIM2 and TIM4 in mast cells (Fig. S4). However, TIM1 expression in mast cells was not confirmed in either protein or transcript levels (Fig. S4). Therefore, we reasoned that the involvement of the TIM1–TIM1/4 interaction in TIM1-Fc–induced activation of mast cells was negligible. We then performed co-culture of LMIR5- or mock-transduced FLMCs with TIM1-transduced Chinese hamster ovary (CHO) cells, demonstrating higher levels of IL-6 released into the supernatants of the former co-culture as compared with the latter (Fig. 4 C, left). In addition, we found higher levels of IL-6 released into the supernatants in the co-culture of LMIR5-expressing FLMCs with TIM1-transduced CHO cells as compared with mock-transduced CHO cells (Fig. 4 C, right). These results indicated that interaction of LMIR5 with surface-expressed TIM1 as well as soluble TIM1 induced the LMIR5-mediated activation of mast cells.
In vivo evidence that the LMIR5–TIM1 interaction induces the accumulation of neutrophils
To investigate the physiological role of TIM1 as a ligand for LMIR5, we generated LMIR5-deficient mice (Fig. S5 A). We confirmed gene targeting by genomic PCR (Fig. S5 B) and the complete absence of LMIR5 expression in BM neutrophils by flow cytometry (Fig. 5 C). Semiquantitative RT-PCR analysis demonstrated no significant difference of LMIR1–4 transcript levels in BM cells between WT and LMIR5−/− mice (Fig. S5 C). LMIR5−/− mice were born at the expected Mendelian ratio and showed no obvious abnormalities. In addition, WT and LMIR5−/− mice did not reveal major differences in the myeloid and lymphoid development of the BM, spleen, thymus, peripheral blood, and peritoneal cells (Fig. S6).
To address the physiological significance of the LMIR5–TIM1 interaction, we chose a model for kidney ischemia/reperfusion injury (IRI; Kelly et al., 1996; Ichimura et al., 1998, 2008; Wu et al., 2007; Lech et al., 2009; Waanders et al., 2010), which is the known in vivo model for TIM1 induction. Consistent with previous reports (Ichimura et al., 1998, 2008; Waanders et al., 2010), TIM1 mRNA levels rapidly increased at day 1 after IRI and diminished through days 2–4 in the IRI kidneys of WT mice, whereas they were maintained at low levels before and after IRI in contralateral kidneys (Fig. 5 A, left). TIM4 mRNA was undetectable in IRI or contralateral kidneys (unpublished data). Notably, LMIR5 mRNA levels increased through days 1 and 2, and thereafter decreased through days 3 and 4 in IRI but not control kidneys of WT mice (Fig. 5 A, right). Further examination showed that in IRI kidneys of WT mice, TIM1 was highly expressed in CD45− (nonhematopoietic) cells as compared with CD45+ (hematopoietic) cells, whereas LMIR5 was predominantly expressed in CD45+ cells (Fig. 5 B). Together with previous findings (Kelly et al., 1996; Ichimura et al., 1998, 2008; Wu et al., 2007; Lech et al., 2009; Waanders et al., 2010), these results indicated that LMIR5-expressing neutrophils were recruited in IRI kidneys after the up-regulation of TIM1 expression in the renal tubular cells. We then compared the IRI kidneys from WT mice with those from LMIR5−/− mice. Flow cytometric analysis delineated that the percentage of neutrophils in IRI kidneys was lower in LMIR5-deficient mice compared with WT mice (Fig. 5 D). On the other hand, in vitro migration assays demonstrated a comparable ability of WT and LMIR5−/− neutrophils to migrate toward chemoattractants such as MIP-2, C5a, KC, and fMLP (Fig. S7 A). To further test if the LMIR5–TIM1 interaction was involved in the neutrophil accumulation in the kidney IRI, we used dorsal air pouch experiments as an in vivo neutrophil recruitment model. Importantly, both TIM1-Fc–induced neutrophil migration and cytokine production in dorsal air pouches were dampened by LMIR5 deficiency, although the response to LPS was comparable between both mice (Fig. 5, E and F). These results suggested that the LMIR5–TIM1 interaction contributed to the accumulation of neutrophils in kidney IRI.
LMIR5 deficiency ameliorates renal tubular damage induced by kidney IRI
Next, we examined if LMIR5 deficiency attenuated the renal damage induced by IRI. As previously reported (Kelly et al., 1996; Wu et al., 2007; Lech et al., 2009), higher amounts of IL-6 and MCP-1 were produced in IRI kidneys compared with contralateral kidneys of WT mice (Fig. 6 A and Fig. S7 B). Notably, LMIR5 deficiency attenuated the increase of cytokine production in IRI kidneys, whereas it did not affect TIM1 expression in IRI kidneys (Fig. 6 A and Fig. S7 B). In WT mice, IL-6 transcripts were equivalently expressed in CD45+ and CD45− cells of IRI kidneys, whereas MCP-1 transcripts were highly expressed in CD45+ cells (Fig. 6 B). These results indicated a contributory role of CD45+ cells in cytokine/chemokine production of IRI kidneys. Histological analysis of IRI kidneys showed severe tubular damage of WT mice, as indicated by widespread tubular necrosis in the outer medulla and cast formation in the inner medulla at 1 d after IRI (Fig. 6 C). On the other hand, tubular damage was significantly ameliorated by LMIR5 deficiency (Fig. 6 C). Related to this, the number of neutrophils infiltrating the interstitial compartments of IRI kidneys was lower in LMIR5-deficient mice compared with WT mice (Fig. 6, D and E). We found no tubular damage as well as negligible numbers of neutrophils in contralateral kidneys of WT or LMIR5−/− mice (Fig. 6, C and E; and not depicted). Importantly, immunohistological examination displayed the frequent colocalization of LMIR5-expressing neutrophils and TIM1-expressing tubular epithelial cells in IRI kidneys of WT mice (Fig. 6 F). Altogether, LMIR5 deficiency ameliorated ischemia-induced renal tubular damage associated with neutrophil accumulation.
Identification of ligands for immune receptors is indispensable for delineation of their biological functions. Recent studies led us to postulate that paired immune receptors have acquired the ability to recognize both endogenous and exogenous ligands (Shiratori et al., 2004; Satoh et al., 2008). The same hypothesis could be applied to LMIR/CD300 family members, but their ligands remained unknown.
In the present study, we identified TIM1 and TIM4 as endogenous ligands for LMIR5. Because TIM1 and TIM4 play an important role in mediating uptake of apoptotic cells through recognition of PS, we were curious about the specific functions of LMIR5 in a similar context. In fact, the Ig-like domain of LMIR5 bound to TIM1 in the vicinity of the PS-binding site within the Ig-like domain of TIM1 (Fig. 2). However, the LMIR5–TIM1/4 interaction did not hamper the TIM1/4-mediated clearance of apoptotic cells (Fig. 3, C–E). One possible explanation is that TIM1 or TIM4 binds to PS at a higher affinity in comparison to LMIR5. Alternatively, the binding site of TIM1 or TIM4 to LMIR5 might be close, but not identical, to that to PS. In addition, several lines of evidence (Fig. 3, A and B) led us to conclude that LMIR5 is not directly involved in the clearance of apoptotic cells. Because LMIR5 is a DAP12-coupled activating receptor, it is plausible that the LMIR5–TIM1/4 interaction induces activation of LMIR5-expressing mast cells. However, in in vitro experiments, myeloid cells were not activated by the interaction of endogenous LMIR5 and TIM1. As one possible explanation, we assumed that biological events induced by the LMIR5–TIM1 interaction require in vivo environmental factors such as cytokines/chemokines, cell–cell interaction, or cell–extracellular matrix interaction. We also postulated that the biological outcomes induced by the LMIR5–TIM1 interaction might be evident in a pathological situation where soluble TIM1 and/or surface-expressed LMIR5 increase at high levels. To test this, we generated LMIR5-deficient mice and used a mouse model of acute kidney injury, IRI, where TIM1 is up-regulated in the IRI kidney. Intriguingly, LMIR5 deficiency attenuated the acute kidney damage, characterized by tubular necrosis and cast formation. In addition, the recruitment of neutrophils in IRI kidneys, reportedly associated with tissue damage (Kelly et al., 1996; Wu et al., 2007; Bolisetty and Agarwal, 2009; Lech et al., 2009), was suppressed in LMIR5-deficient mice. Notably, IRI induced the marked up-regulation of TIM1 expression in epithelial tubular cells, followed by the recruitment of neutrophils to IRI kidneys. Considering that a large amount of soluble TIM1 is released in the ischemic kidneys, it was possible that LMIR5-expressing myeloid cells interacted with soluble TIM1 or surface TIM1 in epithelial cells. In support of this, histological examination displayed the frequent colocalization of LMIR5-expressing neutrophils and TIM1-expressing epithelial cells in IRI kidneys. Taking these observations together, we assume the relevant mechanism to be as follows. First, soluble TIM1 released from or surface TIM1 expressed by renal tubular cells activates LMIR5-expressing myeloid cells, including resident macrophages/monocytes, mast cells, neutrophils, and dendritic cells through the LMIR5–TIM1 interaction. Then, production of cytokines/chemokines and/or soluble TIM1 promotes the neutrophil recruitment, leading to the renal tubular damage. This scenario was supported by the finding that the neutrophil accumulation induced by TIM1-Fc, a mimic form of soluble TIM1, was dampened by LMIR5 deficiency in the dorsal air pouch experiments. Because the migration of BM neutrophils toward chemoattractants was not affected by LMIR5 deficiency in the in vitro assay (Fig. S7 A), we reasoned that the impaired recruitment of neutrophils in LMIR5-deficient mice was presumably caused by the lack of LMIR5–TIM1 interaction, but not to the defective migratory function of LMIR5-deficient neutrophils. Of note, neutrophils failed to migrate toward TIM-Fc alone in the in vitro assay (unpublished data), suggesting the requirement of additional signals supplied by the surrounding cells in vivo. We concluded that soluble TIM1 induced the neutrophil accumulation via LMIR5 under in vivo conditions through both direct and indirect mechanisms. However, TIM1-Fc also binds to LMIR5-deficient neutrophils as well as parent Ba/F3 cells (Fig. S1 C and not depicted), suggesting the existence of unidentified ligands for TIM1. Although recent studies implied multiple mechanisms in the pathology of kidney IRI (Kelly et al., 1996; Wu et al., 2007; Bolisetty and Agarwal, 2009; Jang and Rabb, 2009; Lech et al., 2009), the present paper shows that the LMIR5–TIM1 interaction was involved in the tubular damage in the acute phase after IRI.
In conclusion, we provide evidence that TIM1 is a physiological ligand for LMIR5 and that the LMIR5–TIM1 interaction is pivotal in neutrophil accumulation related to tissue damage in kidney IRI. Blocking the LMIR5–TIM1 interaction might be a novel therapeutic strategy for acute renal tubular damage.
MATERIALS AND METHODS
Cells and mice.
DAP12−/− mice were used as previously described (Kaifu et al., 2003). Animal experiments were conducted in accordance with the guidelines of and with permission provided by the Animal Care and Use Committee of the University of Tokyo (approval no. 20-8). LMIR5−/− mice were generated as previously described (Murata et al., 2004). We used LMIR5−/− mice that had been backcrossed for at least eight generations with C57BL/6 mice (Charles River).
Antibodies and other reagents.
Anti-LMIR5 polyclonal antibody and anti-TIM1 mAb (222414) were obtained from R&D Systems. Biotinylated anti–mouse TIM1 (RMT1-10 or RMT1-4) mAbs were obtained from eBioscience. Anti–mouse TIM1 (RMT1-17), TIM2 (RMT2-14), TIM3 (RMT3-23), and TIM4 (RMT4-54) mAbs have been previously described (Nakayama et al., 2009).
Gene expression analysis.
Generation of Fc fusion proteins.
The cDNA fragments corresponding to the extracellular domains of mouse LMIR1/2/3/4/5, LMIR5 deletion mutants, TIM1, or TIM4 were used. The Fc fusion proteins were purified as previously described (Satoh et al., 2008).
In some experiments, cells were stained with 1 µg/ml of Fc fusion proteins or human IgG1 followed by 10 µg/ml of PE-conjugated F(ab′)2 donkey anti–human IgG. Flow cytometric analysis was performed with a FACSCalibur (BD) equipped with CellQuest software (BD) and FlowJo software (Tree Star, Inc.), as previously described (Izawa et al., 2007; Yamanishi et al., 2008).
Retrovirus-mediated expression cloning.
Expression cloning was performed as previously described (Kitamura et al., 2003). In brief, plasmids (provided by H. Arase, Osaka University, Osaka, Japan) from the cDNA library were transfected into PLAT-E packaging cells (Morita et al., 2000). Infected Ba/F3 cells were stained by LMIR5-Fc. After three rounds of enrichment using a FACSAria (BD), single-cell clones were obtained. The integrated cDNA was recovered by PCR and sequenced.
PIP-strip assay was performed as previously described (Kobayashi et al., 2007). PIP strips were purchased from Echelon Bioscience.
Phagocytosis assay was performed as previously described (Kobayashi et al., 2007; Miyanishi et al., 2007). CFSE-labeled apoptotic thymocytes were coincubated with peritoneal macrophages in the presence of 10 µg/ml of the Fc fusion proteins indicated in the figures for 30 min. Alternatively, CFSE-labeled live or apoptotic U937 cells were coincubated with NIH3T3 transfectants in the presence of 20 µg/ml of the antibodies or Fc fusion proteins indicated in the figures for 45 min. The percentage of CSFE+ cells was measured by flow cytometric analysis.
Immunoprecipitation and Western blotting.
Transfected Ba/F3 cells were incubated with 20 µg/ml LMIR5-Fc, human IgG1, anti-Flag mAb, or mouse IgG1. Cell lysates were immunoprecipitated by using protein A–sepharose. To detect phosphorylation of ERK1/2, transfected BMMCs preincubated with 20 µg/ml TIM1-Fc or human IgG1 for 30 min on ice were stimulated in anti–human IgG antibody–coated plates at 37°C for 7 min, as previously described (Kumagai et al., 2003; Izawa et al., 2007; Yamanishi et al., 2008).
Measurement of cytokines.
For co-culture assay, 5 × 104 CHO cells transduced with TIM1 or mock transduced were co-cultured with 2 × 105 FLMCs transduced with LMIR5 or mock transduced for 24 h. Concentrations of IL-6 in the culture supernatants and those of IL-6 and MCP-1 in the renal extract were measured by ELISA, as previously described (Izawa et al., 2007; Wu et al., 2007; Yamanishi et al., 2008). Protein levels of cytokines/chemokines in the renal extract were corrected for the total amounts of protein.
Neutrophil infiltration into mouse dorsal air pouch.
Air pouches were formed on the dorsum of mice as previously described (Sin et al., 1986). In brief, 7 ml of sterile air was injected subcutaneously into the back of mice on days 0 and 3. On day 6, 100 µg TIM1-Fc or of control Fc, or 1 mg LPS dissolved in 1 ml of PBS was injected into the air pouches. 4 h after the injection, the air pouches were lavaged. Total cells in the lavage fluid were counted and the percentages of Gr-1+/CD11b+ neutrophils were estimated by FACS analysis. IL-6 levels in the lavage fluid were measured by ELISA.
Induction of renal IRI.
In brief, the left renal pedicle of male mice was exposed and clamped for 45 min with a microaneurysm clamp via flank incision, as previously described (Lech et al., 2009). Mice were sacrificed at 24, 48, and 96 h after reperfusion. The IRI and contralateral kidneys were collected for histological analysis, flow cytometry analysis, and measurements of cytokines and chemokines.
Formalin-fixed kidneys were embedded in paraffin and stained with hematoxylin and eosin by standard methods. Neutrophils were detected using rat anti–mouse neutrophil mAb (clone 4/7; AbD Serotec), as previously described (Wu et al., 2007). Neutrophils in the outer medulla were counted in 10 consecutive high-power fields (400×) from each section. Immunofluorescence staining was performed as previously described (Morikawa et al., 2004). In brief, 6-µm-thick frozen sections were stained with primary antibodies: anti-LMIR5 antibody, anti-TIM1 (RMT1-17) mAb, rat anti–mouse neutrophil mAb, and the appropriate secondary antibodies (Jackson ImmunoResearch Laboratories, Inc.). All sections were counterstained with DAPI (Invitrogen).
Data are shown as means ± SD. Statistical significance was determined by the Student t test, with P < 0.05 considered statistically significant.
Online supplemental material.
Fig. S1 shows that TIM1 was identified as the ligand for LMIR5 by retrovirus-mediated expression cloning. Fig. S2 shows differential blocking effects of anti-TIM1 antibodies on the TIM1–LMIR5 interaction. Fig. S3 shows that TIM4-Fc as well as TIM1-Fc induced LMIR5-mediated activation of mast cells. Fig. S4 shows no detectable expression levels of TIM1 and TIM4 in BMMCs, FLMCs, or neutrophils. Fig. S5 shows gene targeting of LMIR5. Fig. S6 shows normal development of myeloid cells and lymphocytes in LMIR5−/− mice. Fig. S7 shows relative gene expression levels of IL-6, MCP-1, TIM1, or LMIR5 in the postischemic kidneys from WT or LMIR5−/− mice. Table S1 shows the gene-specific primers used in this study.
We thank Dr. H. Arase for providing plasmids. We also thank Drs. N. Watanabe and Y. Ishi for cell sorting, and K. Kimura for histological analysis. We are grateful to Dr. D. Wylie for her excellent language support.
This study was supported by grants from the Ministry of Education, Culture, Sports, Science, and Technology (MEXT), Japan, and was in part supported by the Global Center of Excellence Program “Center of Education and Research for the Advanced Genome-Based Medicine–for personalized medicine and the control of worldwide infectious diseases,” MEXT, Japan. Y. Yamanishi is supported by a postdoctoral fellowship from the Japan Society for the Promotion of Science.
Dr. T. Kitamura serves as a consultant for R&D Systems. The authors declare no further financial conflicts of interest.