|Home | About | Journals | Submit | Contact Us | Français|
The T cell immunoglobulin mucin (TIM) proteins regulate T cell activation and tolerance. Both TIM-4, expressed on human and mouse macrophages and dendritic cells, and TIM-1 specifically bind to phosphatidylserine (PS) on the surface of apoptotic cells and do not bind to any other phospholipid tested. TIM-4+ peritoneal macrophages, TIM-1+ kidney cells, as well as TIM-4 or TIM-1 transfected cells efficiently phagocytose apoptotic cells and phagocytosis can be blocked by TIM-4 or TIM-1 mAbs. TIM proteins have a unique binding cavity made by an unusual conformation of the CC′ and FG loops of the TIM IgV domain and mutations in this cavity eliminated PS binding and phagocytosis. TIM-4 mAbs that block PS binding and phagocytosis map to epitopes in this binding cavity. These results show that TIM-4 and TIM-1 are immunologically restricted members of the group of receptors that recognize PS, critical for the efficient clearance of apoptotic cells and prevention of autoimmunity.
The T cell immunoglobulin mucin (TIM) gene family was identified as molecules related to higher risk of development of asthma and allergic disease in mice (McIntire et al., 2001). Multiple studies have now confirmed the association of the TIM gene family with development of autoimmunity and allergic diseases (reviewed in (Kuchroo et al., 2006)). TIM-1 and TIM-3 are preferentially expressed on distinct mouse T cell subsets with TIM-1 being on Th2 (Umetsu et al., 2005) and TIM-3 on Th1 (Monney et al., 2002). The TIM gene family includes three members in human and eight in mouse. Human TIM-1, 3, and 4 appear to be the functional orthologues to mouse TIM-1, 3 and 4 (Kuchroo et al., 2003). Mouse TIM-5-8 are inferred from genomic sequences and have not yet been characterized. All TIM family proteins are type I cell-surface glycoproteins and share common structural motifs including an immunoglobulin variable (IgV) domain with 6 cysteines, a mucin-like domain, transmembrane domain and a cytoplasmic domain. All TIMs except for TIM-4 contain a tyrosine-kinase phosphorylation motif in the cytoplasmic domain. The TIMs differ in the length of the mucin domain and the number of O-linked glycosylation sites (Kuchroo et al., 2003).
The crystal structures of murine TIM-1, 2, 3 IgV domains have been described (Cao et al., 2007; Santiago et al., 2007). TIM-1 and TIM-3 IgV domains display a distinctive cleft formed by the CC′ and FG loops of the GFC β-sheet. This is the face commonly used by Ig superfamily receptors for ligand binding. This cavity is conserved within the TIM family, except for TIM-2. We have found that in TIM-1 and TIM-4, a narrow cavity built by the CC′ and FG loops, is a binding site for phosphatidylserine (PS). In the accompanying paper, Santaigo et al. describe the structure of the co-crystal of PS and mTIM-4 (Santiago et al., submitted).
The redistribution of PS to the outer leaflet of the plasma membrane is a key signal for recognition of apoptotic cells by phagocytes (Fadok et al., 1992; Verhoven et al., 1995). Rapid removal of apoptotic cells by phagocytes is critical for the maintenance of tolerance and prevents inflammation and autoimmune responses against intracellular antigens released from the dying cells (Savill and Fadok, 2000). Several receptors have been identified that participate in the recognition of PS on the surface of apoptotic cells and deficiencies of these molecules in knockout mouse models lead to a failure to maintain self-tolerance and development of autoimmunity (Asano et al., 2004; Cohen et al., 2002).
While several reports of the murine TIM family focused on the interaction between APC and CD4 T cells and regulation of Th1 and Th2 balance and tolerance, there are no data on TIM function in phagocytosis or innate immunity. Here we report that TIM-4 is expressed on mature macrophages in both human and mouse. Our binding assays and the TIM-4 crystal structure of Santiago et al. (Santiago et al., submitted) showed binding of PS to TIM-4 in the cavity formed by the CC′ and FG loops. Anti-TIM-4 mAbs that blocked binding to PS substantially reduced phagocytosis of apoptotic cells by mouse peritoneal macrophages. TIM-4 or TIM-1 transfected NIH-3T3 cells and TIM-1 expressing human kidney epithelial cell lines had increased phagocytosis of apoptotic cells that could be blocked by specific mAbs.
We prepared Ig fusion proteins with the extracellular domains or the IgV domains of TIM-1, 2, or 4 in order to identify their functional ligands. TIM-4-Ig and TIM-1-Ig bound weakly to all long-term cultured cell lines tested (data not shown). We noticed that a minor population of each cell line had a much higher intensity of staining with TIM-4-Ig or TIM-1-Ig. Based on the side scatter and forward scatter gating of the cells, the brightly staining cells appeared to be dead cells (Fig. 1A); however, the intensity of staining was much larger than the typical increase in background staining frequently observed on dead cells. We prepared Jurkat cells with a moderate proportion of apoptotic cells by crosslinking CD95 (anti-Fas, IgM), and stained with TIM-4-Ig or TIM-1-Ig as well as annexin V-FITC and Propidium Iodide (PI) to visualize early and late apoptotic cells. TIM-4-Ig bound weakly to live Jurkat cells but strongly to a structure on early and late apoptotic Jurkat cells (Fig 1B).
To further characterize the binding of TIM-1 and TIM-4 to apoptotic cells, we prepared populations with a high proportion of early apoptotic cells. Treatment of Jurkat cells with anti-CD95 mAb, followed by centrifugation on Ficoll to remove late apoptotic cells, gave a population with more than 90% early apoptotic cells as judged by Annexin V-FITC/PI staining (Fig. 1C). We analyzed the binding of TIM-Ig to these apoptotic cells or to untreated cells. Binding of TIM-Ig was performed in the absence of annexin-V-FITC/PI to eliminate any competition or complication by annexin-V binding to PS. HTIM-4-Ig, hTIM-1-Ig, mTIM-4(IgV)-Ig, and mTIM-1(IgV)-Ig bound slightly to live Jurkat cells but strongly to apoptotic Jurkat cells (Fig. 1E, F).
Red blood cells (RBC) can be induced to undergo an apoptotic-like process, termed eryptosis, by incubation with ionomycin and phorbol ester. Eryptotic RBC expose PS on the RBC surface (Fig. 1D). Human TIM-4-Ig, hTIM-1-Ig, mTIM-4(IgV)-Ig, and mTIM-1(IgV)-Ig did not bind detectably to fresh RBC but bound strongly to eryptotic RBC (Fig. 1E). In contrast, mTIM-2(IgV)-Ig, or control Ig did not bind to any cell tested. Similar binding results with these TIM-Ig were seen with several mouse (300.19, FL5.12) and human (Jurkat, U937) cell lines following induction of apoptosis by various methods (cytokine deprivation, etoposide, Fas-mediated) (data not shown). Binding was not species restricted. In contrast, freshly isolated human T cells did not bind any of the TIM-Ig (data not shown).
The binding of TIM-4 and TIM-1 fusion proteins containing the IgV but not the mucin domain indicates that binding to apoptotic cells can be mediated by the IgV domain alone. By pre-incubating TIM-Ig with TIM mAbs known to recognize the IgV domain, we tested the capacity of TIM mAbs to block the interaction with apoptotic cells. The interaction of hTIM-4-Ig with apoptotic cells was blocked by hTIM-4 mAb 9F4 but not by mAb 4E11 (Fig. 1G). The interaction of hTIM-1-Ig was blocked by hTIM-1 mAb 3D1 and moderately blocked by hTIM-1 mAb 1D12 (Fig. 1H).
Exposure of PS on the outer leaflet is a major difference between live and apoptotic cells, so we tested whether TIM-1 and TIM-4 recognize PS using PS-coated plates in a solid phase ELISA. Murine TIM-1(IgV)-Ig and mTIM-4(IgV)-Ig bound to PS-coated plates in a concentration-dependent manner; however, mTIM-2(IgV)-Ig did not bind (Fig 2A). Human TIM-1-Ig, hTIM-4-Ig, mTIM-1(IgV)-Ig, and mTIM-4(IgV)-Ig bound to plates coated with PS, but not to plates coated with phosphatidylinositol (PI), phosphatidylcholine (PC) or phosphatidylethanolamine (PE) (Fig 2B). Murine TIM-2(IgV)-Ig did not bind to any of these phospholipids, even at high concentrations of TIM-2-Ig.
To determine the binding of TIM proteins to other phospholipids, we employed a protein-lipid overlay assay using nitrocellulose membranes on which various phospholipids had been spotted (PIP strip). Figure 2C shows that mTIM-1 and mTIM-4 specifically bound to PS but not to the other phospholipids except for a slight binding to cardiolipin. These findings indicate that human and mouse TIM-1 and TIM-4 specifically recognize PS via the IgV domain of the TIM molecule.
Recognition of cell surface PS is a critical signal for clearance of dead cells by phagocytes (Fadok et al., 1992; Verhoven et al., 1995). To study the function of TIM-1 and TIM-4 in phagocytosis of apoptotic cells, we transfected hTIM-1 or hTIM-4 into NIH-3T3 cells. Stable hTIM-1 or hTIM-4-3T3 expressed moderate levels of each molecule on their cell surfaces (Suppl. Fig. 1). Flow cytometry was used to measure uptake of CMTMR-labeled live or apoptotic U937 cells or PKH67-labeled fresh or eryptotic RBC by transfected 3T3 cells (Fig. 3A, B, E). Live U937 cells or fresh RBC did not bind and were not phagocytosed by TIM-1, TIM-4, or untransfected 3T3 cells. In contrast, hTIM-1 or hTIM-4-transfected 3T3 cells rapidly and efficiently phagocytosed apoptotic U937 (Fig. 3B) or eryptotic RBC (Fig. 3E). Untransfected 3T3 cells slowly and inefficiently phagocytosed apoptotic U937 or eryptotic RBC. Electron microscopy and three-dimensional imaging by confocal microscopy showed that uptake of apoptotic U937 by TIM-1 or TIM-4 transfected cells was truly phagocytosis as the apoptotic cells were completely engulfed and internalized (Fig. 3C, F–I, and data not shown). Labeling of apoptotic U937 with the pH-sensitive dye, CypHer5E (Beletskii et al., 2005), showed that the phagocytosed U937 was compartmentalized into an acidic compartment such as a phagosome (Fig. 3D and Suppl. Fig. 5). Using eryptotic RBC, we saw a similar enhancement of phagocytosis in TIM-1 or TIM-4 transfected 3T3 cells. (Suppl. Fig. 2).
To test whether PS is the molecule recognized by TIM-1 and TIM-4 in phagocytosis of apoptotic cells, we prepared phospholipid liposomes containing various phospholipids and tested their capacity to inhibit phagocytosis by TIM transfectants. Liposomes containing equal amounts of PS and PC (PS/PC) reduced phagocytosis by both TIM-1 and TIM-4 transfectants in a concentration-dependent fashion with complete inhibition at 10 μM of total phospholipids (Fig. 3J). Inhibition of phagocytosis was not observed with liposomes containing another anionic lipid, PI, nor with PC or PE (Fig. 3J, K). These observations show that recognition of PS on the plasma membrane by hTIM-4 and hTIM-1 depends not only on the anionic charge of PS but also on highly specific recognition of PS structural features.
TIM-4 mRNA expression has been reported in mouse spleen and lymph node (Meyers et al., 2005; Shakhov et al., 2004) but which cells display surface TIM-4 is unclear. We found that resident mouse peritoneal cells express higher levels of TIM-4 mRNA than splenocytes (Fig. 4A). To identify the cell surface expression of TIM-4, we used two-color staining with F4/80 to identify peritoneal macrophages (M) and either of two mTIM-4 mAbs. Peritoneal macrophages expressed high levels of TIM-4 but no detectable TIM-1 (Fig. 4B). TIM-4 staining was present in moderate abundance on both CD8α+ and CD8α− subsets of CD11c+ DCs, consistent with the initial identification of TIM-4 mRNA expression in CD11c+ cells (Meyers et al. 2005), but was not detectable on plasmacytoid DCs (Fig. 4C) or GM/CSF-stimulated bone-marrow derived DC (data not shown). TIM-4 was not detected on unstimulated CD19+ B cells (Fig. 4C) or B cells stimulated with LPS or anti-CD40 mAb (data not shown). TIM-4 was not detected on resting or activated T cells but TIM-1 was expressed on activated T cells ((Umetsu et al., 2005) and data not shown).
Human lymphoid tissues such as tonsil, thymus and spleen, as well as testis expressed TIM-4 mRNA by qRT-PCR (Fig. 4D). Kidney, liver and lung expressed low levels of TIM-4 mRNA, and peripheral blood mononuclear cells expressed very low levels. To determine whether human M express TIM-4, we prepared human splenic M either by positive selection using CD14 MACS microbeads or by adherence to tissue culture plastic (see Materials and Methods). Splenic adherent and CD14+ cells expressed high levels of TIM-4 mRNA whereas non-adherent and CD14− cells expressed little TIM-4 mRNA (Fig. 4E). Human splenic CD14highCD11clowCD11b+ cells, consisting of M and immature dendritic cells (DC), expressed high levels of TIM-4 mRNA but CD14− CD11chighCD11b−, consisting of mature DC in human spleen expressed little TIM-4 mRNA (data not shown). Human peripheral blood monocytes did not express detectable levels of cell surface TIM-4 (data not shown).
To further characterize the expression of TIM-4 in human tissue, we performed immunohistochemistry of tonsil (Fig. 4F) and spleen (Fig. 4G) with goat anti-hTIM-4 polyclonal antibody. Tingible-body M located in germinal centers of tonsil or white pulp of spleen showed expression for TIM-4. Tingible bodies are the remnants of cells phagocytosed by macrophages. There was no staining with control goat IgG (Fig. 4F, G). The specificity of TIM-4 staining was further confirmed by showing that absorption of the TIM-4 antibody on TIM-4 transfected 300.19 cells removed the reactivity whereas absorption on untransfected 300.19 cells did not (Fig. 4F). These results show that macrophages in human and mouse express TIM-4.
To assess the involvement of TIM-4 in phagocytosis of apoptotic cells, PM were pre-incubated with increasing amounts of 21H12 mAb or isotype control. PM were then co-cultured with PKH67-labeled apoptotic thymocytes for 30 min, and phagocytosis measured by flow cytometry (Fig. 5A, B). Approximately 44 % of the PM phagocytosed apoptotic thymocytes and this percentage was not affected by isotype control antibody. In contrast, 2 or 10 μg/ml of mTIM-4 mAb 21H12 reduced the percentage of PM that engulfed apoptotic cells to 13.6%. These results suggested that TIM-4 is one of the receptors used by mouse PM to recognize and phagocytose apoptotic cells.
Some human kidney cell lines show natural surface expression of TIM-1. We confirmed that 769P cells (Vila et al., 2004), derived from a renal clear cell carcinoma, express TIM-1 on their cell surface using hTIM-1 antibody 3D1 (Fig. 6A). In contrast, 769P cells did not express TIM-4. To test whether 769P cells phagocytose apoptotic cells via TIM-1, we cultured 769P cells with apoptotic U937 cells or eryptotic RBC and measured phagocytosis. During a two hour incubation, 769P cells phagocytosed apoptotic U937 (41.9%), but not live U937 (0.35%) (data not shown). The internalization of apoptotic cells was confirmed by 3-dimensional imaging by confocal microscopy (Fig. 6B). 769P cells also phagocytosed eryptotic RBC much more efficiently than fresh RBC (Fig. 6C). Using the TIM-1 mAbs 3D1 and 1D12 that blocked the binding of TIM-1-Ig to apoptotic Jurkat cells (Fig. 1H), we found that binding and phagocytosis of apoptotic cells was blocked by these TIM-1 antibodies in a concentration-dependent manner (Fig. 6D and Suppl. Fig. 3). These studies show that 769P cells can recognize and phagocytose apoptotic cells through TIM-1 on their cell surface.
The crystal structure of mTIM-4 showed that PS binds in the cavity built up by the CC′ and FG loops of TIM-4 and suggests this cavity is important for biological function (Santiago et al., submitted). To test the function of this site in phagocytosis of apoptotic cells, we prepared alanine substitutions of each of the 4 amino acids comprising this cavity at the FG loop (W119-D122). Hydrophobic W119 and F120 are located at the top of the FG loop. N121 and D122 are at the bottom of this cavity and trap positively charged ions. Each mutant was transfected into NIH-3T3 cells and transfectants identified with polyclonal goat anti-hTIM-4 antibody (Fig 7A). Alanine substitutions in W119-D122 did not affect binding by the 4E11 hTIM-4 mAb. In contrast, the W119A, F120A, or N121A but not D122A substitutions destroyed the epitope on TIM-4 recognized by the 9F4 mAb. These results show that 9F4 mAb recognizes the FG loop of hTIM-4 whereas the 4E11 mAb reacts outside of this area. This epitope mapping is consistent with the capacity of 9F4, but not 4E11, to reduce phagocytosis of eryptotic RBC by hTIM-4 transfected NIH-3T3 cells (Fig. 7B).
The phagocytic capacity of untransfected, wild-type, and mutant TIM-4 transfected 3T3 cells was compared (Fig. 7C, D and Suppl. Fig. 4)). Wild-type hTIM-4 transfected 3T3 efficiently phagocytosed eryptotic RBC with 70% of the cells phagocytosing an RBC after 90 min while untransfected 3T3 phagocytosed eryptotic RBC poorly with only 17% of the cells phagocytosing an RBC after 90 min. All 3T3 lines transfected with mutant TIM-4 (W119A, F120A, N121A and D122A) had reduced phagocytic activity equivalent to untransfected 3T3. These results show that the PS binding cavity composed of W119 to D122 is necessary for the recognition and phagocytosis of apoptotic cells by TIM-4.
In this study we have shown that TIM-4 and TIM-1 recognize PS on the surface of apoptotic cells. Transfection of TIM-4 or TIM-1 into NIH-3T3 cells greatly augmented their capacity to phagocytose apoptotic cells. This phagocytic activity was completely blocked by liposomes containing PS but not PC, PI, or PE. We identified mAbs against mouse and human TIM-4 and TIM-1 that would block binding to PS and phagocytosis. All the blocking mAbs mapped to the IgV domain on which the PS binding cavity is located. Consistent with a role in recognition of apoptotic cells, TIM-4 is highly expressed on phagocytic cells such as peritoneal macrophages, tingible-body macrophages, and dendritic cells. Tingible-body macrophages are the cells that engulf apoptotic cells in lymphoid tissue (Smith et al., 1991).
PS comprises about 7% of the phospholipid of a cell but is compartmentalized to the inner leaflet of the cell membrane. The redistribution of PS to the external surface of the plasma membrane is a key element of apoptotic cell recognition and is a molecular cue that dying cells should be phagocytosed (Fadok et al., 1992; Verhoven et al., 1995). For example, healthy Jurkat cells expose about 0.9 attomol (600,000 molecules) per cell of PS which rises to about 240 attomol/cell late in apoptosis (Borisenko et al., 2003). Macrophages inefficiently phagocytose cells expressing this low, healthy level of PS and phagocytosis increases only after a threshold level of about 5 attomol/cell is passed and plateaus above 25 attomol/cell (Borisenko et al., 2003).
Several receptor combinations have been identified that participate in the recognition of PS on the cell-surface of apoptotic cells including Milk fat globule EGF-factor 8 (MFG-E8) (Hanayama et al., 2002) and Growth arrest specific gene 6 (GAS6) (Scott et al., 2001). These are two component systems composed of a soluble PS “tethering” molecule such as MFG-E8 or GAS6 and a membrane bound “tickle” molecule such as αVβ3 or αVβ5 integrin or Mer responsible for recognition of the “tether” and internalization and signaling. A PS receptor on the surface of macrophages has been described; however, the candidate molecule cloned by phage display (Fadok et al., 2000) has been discredited (Williamson and Schlegel, 2004). TIM-4 is a likely candidate for this macrophage PS receptor with the IgV domain comprising the “tether” that binds PS and the extended mucin domain providing a “tickle” to the TIM transmembrane and cytoplasmic domains. Recognition of PS by TIM-4 and TIM-1 was highly specific whereas other PS binding molecules such as MFG-E8 (Hanayama et al., 2002), GAS6 (Nagata et al., 1996; Nakano et al., 1997), β2-glycoprotein (Balasubramanian et al., 1997) and lectin-like oxidized low-density lipoprotein receptor 1 (Oka et al., 1998) bind not only PS but other anionic phospholipids.
Phagocytosis of apoptotic cells is evolutionarily ancient and PS-recognition molecules such as MFG-E8 and GAS6 are broadly expressed in tissues. In contrast, expression of TIM-4 is highly restricted to professional APC and testis, suggesting an immunologically restricted function in recognition of apoptotic cells. Given the redundant systems that recognize apoptotic cells, the strong effects of TIM-1 or 4 mabs or fusion proteins are impressive.
The binding of TIM-1 or TIM-4-Ig to apoptotic cells was not species-restricted and apoptosis could be induced by methods including Fas-mediated, etoposide, cytokine deprivation, and starvation of cells, or Ca/ionomycin induction of eryptosis in RBC. Using strips coated with a variety of phospholipids, we showed that TIM-4 and TIM-1 bound to PS and not other phospholipids. All healthy long-term cell lines showed a low level of binding of TIM-1 and TIM-4, consistent with their low but detectable level of exposure of cell surface PS as measured by anti-PS mAb or AnnexinV-FITC (data not shown). In contrast, TIM-1 and TIM-4 did not bind to T cells from freshly isolated blood (data not shown). This is consistent with the fact that in vivo, cells exposing PS are rapidly phagocytosed whereas in vitro cell lines are not monitored as stringently since professional phagocytic cells are not present.
The co-crystal structure of PS bound to TIM-4 (Santiago et al., submitted) shows a direct interaction not mediated by intervening molecules. The specific binding of PS to TIM-4 is due to the unique structure of TIM IgV domains. The four non-canonical cysteines found in all TIM family members form disulfide bonds that reverse and fix the folded conformation of the long CC′ loop onto the GFC β sheet (Cao et al., 2007; Santiago et al., 2007). Instead of the relatively flat binding surface found in most Ig superfamily ligand recognition surfaces, the CC′FG loops form a narrow cavity with dimensions of approximately 7 × 9 × 11 angstroms. The outer lips of the cavity are composed of hydrophobic amino acids positioned to interact with the hydrophobic fatty acid region of PS. The bottom of the cavity is composed of charged amino acids that capture a metal ion and coordinate with the charged head of the PS. The recognition of PS by macrophages has been shown to be stereospecific for the L-isomer of serine and the D-isomer is not recognized (Hoffmann et al., 2005). The co-crystal structure of the L-isomer of PS with TIM-4 shows this stereospecificity and the D-isomer would reverse the charge interactions within the binding pocket and be repulsed because of the juxtaposition of positive to positive and negative to negative.
Mutation of any of the amino acids (119–122, WFND) that line this binding cavity eliminates binding to PS and phagocytosis. The 9F4 mAb that blocks TIM-4 binding to PS maps to this epitope. The structures of mouse and human TIM-1 and TIM-4 are very highly conserved in this region whereas TIM-2 is completely different (--AF where -indicates a gap in the aligned sequences). This likely explains the similar recognition of PS by TIM-4 and TIM-1 and the different ligand binding of TIM-2. Clearly, TIM-2 is an outlier in the TIM family and binding to proteins including Semaphorin4A and h-Ferritin has been reported. TIM-1 and TIM-4 have been reported to be ligands for each other based on the binding of TIM-1-Ig to TIM-4 transfectants and vice versa (Meyers et al., 2005); however, Biacore studies with highly purified protein showed the TIM-1:TIM-4 interaction was of very low affinity (Sizing et al., 2007). In our assays of TIM-mediated phagocytosis, the apoptotic cell did not express TIM-1 or TIM-4, thereby excluding a TIM/TIM interaction. Some of the low level, universal binding of TIM proteins observed by ourselves and others (Cao et al., 2007; Wilker et al., 2007) may reflect the health of the cell and its level of exposure of PS. In addition, some apparent TIM-1/TIM-4 interactions could reflect a “bridge” with two TIM proteins binding to an exosome or membrane fragment via exposed PS. Such a bridge has been shown to be formed by MFG-E8 between an exosome and an apoptotic cell.
TIM-1 was also cloned as kidney injury molecule 1 (KIM-1) and is expressed minimally by healthy kidney but in post-ischemic kidney is highly unregulated on the luminal side of dedifferentiated tubule epithelial cells (Ichimura et al., 1998). While the function of TIM-1 in recovery from ischemic kidney injury is not yet clear, our results show that TIM-1 on a renal cell carcinoma line mediates phagocytosis of apoptotic cells and suggests one of the functions of TIM-1 on dedifferentiated tubular epithelial cells may be recognition of dead cells after ischemic kidney injury and clearance of dead cells to reconstitute the tissue. In addition, TIM-1 ectodomain has been shown to be shed from the injured kidney (Han et al., 2002) and may regulate the recognition of apoptotic cells either by opsonization or blockade. While TIM-1 can mediate phagocytosis in a large kidney cell, it may function on a small T cell more to receive signals via small exosomes or to sense the activation status of the T cell and the health of the APC and surrounding tissue.
In vivo administration of either TIM-4-Ig or TIM-1-Ig resulted in hyperproliferation of T cells and enhancement of T cell cytokine production (Meyers et al., 2005). The administration of TIM-1 agonistic mAb enhanced T cell activation and proliferation and induced increased airway inflammation and blocked development of respiratory tolerance (Umetsu et al., 2005). TIM-1 associates with T cell receptor and crosslinking with TIM-1 mAb leads to engagement of tyrosine signaling motifs in the cytoplasmic domain of TIM-1 and higher production of Th2 cytokines (Binne et al., 2007; de Souza et al., 2005). A common perplexing feature of in vivo treatment with TIM mAbs or fusion proteins is the finding that at a time when T cells should have returned to quiescence, T cells taken from the treated animal and assayed in vitro are still proliferating and producing cytokines. During a primary T cell response in vivo, T cells expand exponentially and then 90% or more die by apoptosis and are phagocytosed. T cell activation leads to a rise of exposed PS on the surface of the T cell and this PS is associated with lipid rafts and the immunological synapse (Fischer et al., 2006). While some in vitro activated T cells express high levels of PS and progress to activation induced cell death, the majority express an intermediate level of PS, survive, and restore PS exposure to normal levels. TIM-mediated recognition of PS on activated T cells may have a role as T cells pass through this “near-death” experience. The extended T cell response caused by TIM blockade may be due to blockade of the phagocytosis of activated T cells exposing PS. Rapid removal of apoptotic cells by phagocytes is critical for the maintenance of tolerance and prevents inflammation and autoimmune responses against intracellular antigens released from the dying cells (Savill and Fadok, 2000) (Wu et al., 2006). Disruption of this process may explain the genetic association of the TIM gene family with auotoimmunity and asthma.
While the involvement of TIM-4 in phagocytes seems clear, the cytoplasmic domain of TIM-4 does not contain any tyrosine signaling motifs and the signaling pathway requires further study. In summary, TIM-1 and TIM-4 have unique structures that let them look death in the eye and give it a molecular kiss.
Case materials were obtained from the Brigham & Women’s Hospital and Dana-Farber Cancer Institute, Boston, MA, in accordance with institutional policies.
All mouse experiments were approved by the Animal Care and Use Committee of the Dana-Farber Cancer Institute or Childrens Hospital, Boston.. C57BL/6 and BALB/cByJ mice were purchased from Jackson Laboratories. To collect resident peritoneal cells, mice were sacrificed and the peritoneal cavities were infused with 3 ml of ice-cold PBS + 0.3% BSA + 0.03% EDTA, cells recovered, and washed three times in media.
BALB/cByJ mice were immunized subcutaneously with human TIM-1(IgV)-Ig or TIM-4(IgV)-Ig in CFA and boosted multiple times with the FP in PBS or IFA. One day after final boost, lymph node cells were fused with NS1 myeloma cells, cloned, and the hybridomas were screened by cell surface staining of human TIM-1 or TIM-4 transfected 300.19 cells. 3D1 (mouse IgG1) and 1D12 (mouse IgG1) for hTIM-1; 9F4 (mouse IgG1) and 4E11 (mouse IgG2b) for hTIM-4 were chosen. Anti-mTIM-4 mAbs (21H12, rat IgG1 and QT3, rat IgG2a) were made by immunization of a Lewis strain rat using a similar protocol and will be described in detail elsewhere (Karisola et al, manuscript in preparation). Anti-mouse TIM-1 (3B3, rat IgG2a) has been described (Umetsu et al., 2005). After purification with protein A-Sepharose, antibodies were conjugated to PE by Zymed Laboratories (South San Francisco, CA).
TIM-Ig FP consist of the complete TIM extracellular domain (indicated as TIM-Ig) or the signal and IgV domain of the TIM (indicated as TIM (IgV)-Ig) linked to the hinge-CH2-CH3 domains of mouse IgG2a (with four point mutations blocking Fc receptor and complement binding) in the pEF6 vector. The mouse TIM-1(IgV) fused to a mouse IgG2a Fc tail (Umetsu et al., 2005) and human TIM-1 fused to a human IgG1 Fc tail (Tami et al., 2007) have been described. Other TIM-Ig were made in the same fashion as mTIM-1(IgV)-Ig (Umetsu et al., 2005). Ig FP were produced in stably transfected CHO cells and purified from conditioned media by protein A-Sepharose chromatography.
NIH-3T3 cells were transfected by electroporation with a pEF6 plasmid containing hTIM-1, or wild type or mutant hTIM-4 cDNAs (W119A, F120A, N121A, D122A). Cells were selected with blasticidin, sorted by flow cytometry for hTIM-1 with 3D1-PE or for hTIM-4 with 9F4-PE or 4E11-PE.
Mutations of a hTIM-4 cDNA in pEF6 were made by oligonucleotide directed mutagenesis. For example, the D122A mutant was made by PCR amplifying the vector/amino terminal end and the middle region of hTIm-4 with overlapping primer pairs, 5′-GCTCGGATCCACTAGTCCAGTGTG with 5′-TTACAGCGTTGAACCAGCCAGGCACTTCTATG and 5′-CTGGCTGGTTCAACGCTGTAAAGATAAACGTG with 5′-TGTGGCTTCCTCCGGAAGGGTGCTTGGGGTTA, respectively. The two PCR fragments were incubated with Spe I and BspE I digested Tim-4 plasmid, joined by Infusion reaction (Clontech, Mountain View, CA), transformed into E. coli and the sequence confirmed.
Immunostaining for TIM-4 was performed on formalin-fixed paraffin embedded human tissue sections following microwave antigen retrieval in 10 mM citrate buffer, pH 6.0, using a standard indirect avidin-biotin horseradish peroxidase (HRP) method and diaminobenzidine color development, as previously described (Dorfman et al., 2003). Immunostaining with goat anti-human TIM-4 polyclonal antibody (R&D Systems) was compared with that of goat IgG control antibody diluted to identical protein concentration for all cases studied, to confirm staining specificity. The specificity of TIM-4 immunostaining was also compared after absorption of the TIM-4 antibody on TIM-4 transfected 300.19 cells or untransfected 300.19 cells.
HTIM-4-transfected 3T3 cells were incubated with apoptotic U937 for 90 minutes, washed, and fixed in 0.1M cacodylate buffer (pH 7.4) containing 2.5% formaldehyde, 5% glutaraldehyde, and 0.06 % picric acid. Fixed cells were embedded in Epon araldite and 2.4 volumes of dodecinyl succinic anhydride, propylene oxide and hardened at 60°. Thin sections (1 μm or less) were examined with a JEM-1200 EX electron microscope.
Total RNA was extracted from cells or organs with Trizol (Invitrogen). Human thymus, liver, kidney, lung and testis RNAs were from multiple-tissue cDNA panels (Clontech). Reverse transcription was done using Taqman reverse transcription reagents (Applied Biosystems) with Oligo(dT) as primer, following the manufacturer’s protocol. The mouse and human TIM-4 and GAPDH primer-probe sets were purchased from Applied Biosystems. Taqman Universal PCR Master Mix and the 7500 Real-Time PCR System (Applied Biosystems) were used for PCR. Comparative threshold cycles (CT) was used to determine gene expression. For each sample, the TIM-4 CT value was normalized with the formula ΔCT = CTTIM-4 − CTGAPDH. For relative expression, the mean ΔCT was determined, and relative TIM-4 expression was calculated with the formula 2−ΔCT.
Cells from human spleen were obtained by physical disruption followed by filtration through a 70 μm nylon cell strainer (BD Bioscience). Mononuclear cells were separated by Ficoll-Hypaque (GE-Healthcare) and washed with PBS with 2% FBS. CD14+ and CD14− cells were purified through positive selection by MACS Sort magnetic beads in MACS LS Separation columns (Milteny Biotec). The percentages of CD14+ cells were analyzed by flow cytometry using FITC-conjugated anti-human CD14 mAb (BD Pharmingen) and were greater than 70% in CD14+ cells and less than 0.9% in CD14− cells. Adherent splenic cells were prepared by culture on tissue culture plates for 2 hrs at 37 °C and non-adherent cells were removed by washing with PBS.
For determination of cell surface mTIM-4 expression on freshly isolated cells, cells were preincubated with anti-Fc receptor mAb (2.4G2, ATCC) for 15 min to block Fc receptors. Peritoneal cells were stained with anti-F4/80-APC (BD Pharmingen). CD11c+ cells were purified by positive selection from spleen using MACS CD11c microbeads (Miltenyi Biotec), and stained with PE-Cy5-anti-CD8α (clone 53-6.7). B cells were stained with anti-CD19 (FITC) (BD PharMingen). Cells were costained with PE-conjugated anti-mTIM-4 mAb or isotype control, washed and analyzed. For determination of TIM-Ig binding, cells were stained with TIM-Ig or isotype control in PBS + 2% FCS + 0.02% NaN3 for 30 min on ice, washed twice, and then stained with PE or APC conjugated goat F(ab′)2 anti-mouse IgG2a or PE conjugated anti-human IgG (Southern Biotechnology), depending on FP isotype. To determine if anti-TIM mAbs blocked binding of TIM-Ig, the indicated TIM-Ig was pre-incubated with anti-TIM mAb or isotype control for 30 min, then used to stain cells as described above. Flow cytometry was performed on a Cytomics FC500 (Beckman-Coulter) and data were processed with FlowJo software (Tree Star).
Jurkat cells were cultured for 3 hours in the presence of agonistic anti-human CD95 mAb (7C11, IgM) (Robertson et al., 1995). U937 human monocyte cell line was incubated with 50 μM etoposide (Sigma) for 5h. Thymocytes from C57BL/6 mice were incubated with 10 μM dexamethasone (Sigma) in RPMI for 3 hours. Human red blood cells (RBC) from fresh human blood were separated from plasma, platelets and PBMC by Ficoll-Hypaque centrifugation. After centrifugation, neutrophils and upper 10% of RBC were removed by pipetting. The remaining RBC were washed twice and incubated in RPMI containing 4 μM ionomycin (Sigma) and 1 mM CaCl2 for 3 hr as described previously (Lang et al., 2005). Apoptosis and PS expression on the outer leaflet of the plasma membrane was detected by staining with AnnexinV-FITC and propidium iodide kit (BD Pharmingen). After treatment, U937 were washed 3 times and labeled with 2–5 μM of CellTracker Orange (CMTMR, Molecular Probes, FL2 channel) and thymocytes or RBC with 1–2 μM PKH67 (Sigma, FL1 channel) according to the manufacturer’s instructions.
Small unilamellar liposomes containing the indicated phospholipids(s) (Avanti Polar Lipids) were made as described previously (Fadok et al., 2000). A slash symbol (PS/PC) indicates a 50:50 molar mix of the indicated lipids. The lipids were mixed in chloroform, dried, resuspended in PBS, mixed, and sonicated for 3 min.
1×106 resident peritoneal cells were cultured in 24 well plates for 2 hours and washed with PBS three times. Anti mTIM-4 mAb or isotype control was added to the cells in media and pre-incubated for 5 min before the addition of 1 × 106 PKH67 labeled apoptotic thymocytes. After 30 min at 37 °C, cells were washed twice with cold PBS, then detached from the plates by treatment with 0.05% trypsin and 0.02% EDTA (Cellgro). Collected cells were stained with PE conjugated anti-CD11b mAb (BD Pharmingen) and analyzed by flow cytometry. CD11b+ cells that were also PKH67+ were counted as phagocytic.
Phagocytosis by TIM transfected NIH-3T3 cells or the human renal clear cell carcinoma 769P cell line (ATCC) was measured by first culturing 5×104 cells in 24 well plates for 2 days. Cells were then co-incubated with labeled live or apoptotic U937 or fresh or eryptotic RBC for 45 to 90 min at 37 °C. Plates were quickly washed three times with cold PBS with 0.5 mM EDTA to remove non-adherent U937 or RBC, then detached by treatment with 0.05% trypsin and 0.02% EDTA (Cellgro) and analyzed by flow cytometry. TIM 3T3 or 769P cells were gated by forward and side scatter. To examine whether liposomes containing PS blocked phagocytosis, cells were pre-incubated with phospholipid liposomes for 10 min and then incubated with eryptotic RBC. Where indicated, cells were pre-incubated with anti-TIM mAb or isotype control for 30 min before the addition of eryptotic RBC.
Phagocytosis was also evaluated by confocal and fluorescent microscopy. 5×104 TIM transfected 3T3 cells or 769P cells were cultured for 2 days in 4 well Lab-Tek II Chamber Slides (Nalge Nunc). Slides were washed one time with RPMI without serum and cells labeled with 2–5 μM of CellTracker Green (CMFDA, Molecular Probes) in RPMI according to manufacturer’s instructions. Cells were co-incubated with CMTMR-labeled U937 for 45 to 90 min. Slides were quickly washed three times with cold PBS with 0.5 mM EDTA and fixed with 3% paraformaldehyde. Samples were analyzed on a Nikon TE2000-U inverted Microscope with EZ-C1 software (Nikon) or an Olympus CKX41.
Apoptotic U937 were washed 3 times with PBS, labelled with 2 μM CypHer5E Mono NHS ester (GE Healthcare) in PBS for 20 min at room temperature, washed, incubated with hTIM-4 transfected 3T3 cells for 100 min, and washed. Localization of apoptotic bodies from neutral cell surface (colourless) into acidic intracellular endosomes (deep red) was detected by excitation with a 635 nm laser and emission fluorescence was collected using the fv1000 spectral scanner. Images were manipulated in Volocity (Improvision Inc, Waltham, MA) for presentation in movie format. Photoshop and NIH ImageJ* were used to composite images.
PIP strips on which the indicated phospholipids had been spotted were purchased from Echelon Bioscience. Dot-blot experiments were carried out according to manufacturer’s protocol. Strips were incubated overnight in StartingBlock (TBS) Blocking Buffer (Pierce) at 4 °C then transferred to 2 μg/ml of TIM-Ig in blocking buffer with 0.5 mM CaCl2 for 2 hours. The PIP strip was then washed three times in TBST (140 mM NaCl, 10 mM Tris-HCl, 0.1% Tween20) with 0.5 mM CaCl2 before incubating with HRP-conjugated rat anti-mouse IgG2a mAb (BD Pharmingen). Antibody binding was detected using ECL plus Western blotting detection reagents (Amersham Bioscience) and visualized using Kodak X-Omat 2000A processor.
The solid phase ELISA for TIM binding to phospholipids was carried out as described (Hanayama et al., 2002). In brief, the indicated phospholipids in methanol (5 μg/ml, 100 μl) were added to polystyrene ELISA plates and air-dried. The plates were then blocked with 2% BSA in PBS. TIM-Ig or anti-PS mAb (clone 1H6, mIgG2a, Upstate) were diluted in PBS, added to the wells, and incubated at room temperature for 2 h. The plates were washed with PBS with 0.05% Tween 20 before incubation with goat F(ab′)2 anti-mouse IgG2a biotin and Streptavidin-HRP (Southern Biotechnology) for mTIM-1, -2, -4 and hTIM-4-Ig, or goat anti-human IgG-HRP (Jackson Immunoresearch) for hTIM-1-Ig. The absorbance at 450nm was measured with Spectra Max190 (Molecular Devices).
Supplemental Figure 1. Surface expression of human TIM-1 and TIM-4 on transfected NIH 3T3 cells. Cells were stained with 3D1, 1D12, 9F4 or 4E11 and analyzed by flow cytometry. Open curves represent staining with PE-conjugated anti-hTIM mAb and filled gray curves represent isotype control staining.
Supplemental Figure 2. 3T3-hTIM-1 or 3T3-hTIM-4 phagocytose eryptotic RBC. 3T3, 3T3-hTIM1 or 3T3-hTIM4 were loaded with PKH67 (green). Cells were co-cultured with PKH26 labeled eryptotic RBC (red) for 90 min, washed, and visualized with a fluorescent microscope. 3T3-hTIM1 or 3T3-hTIM4 phagocytose multiple RBC.
Supplemental Figure 3. Anti-hTIM-1 mAb reduced the binding of apoptotic U937 to 769P. The TIM-1+TIM-4− human kidney cell line, 769P, in 24 well plates was pre-incubated with (A) isotype control or (B) 10 μg/ml of 3D1 anti-hTIM-1 mAb. U937 were made apoptotic by 5 hr incubation with 50 μM etoposide, washed, then cultured with 769P for 60 min. After washing out unattached U937, the binding of apoptotic U937 to 769P was analyzed by light microscopy. Original magnifications were 400X. Binding of apoptotic U937 to 769P (yellow arrows) was blocked by 3D1 mAb.
Supplemental Figure 4. 3T3, or 3T3 transfected with WT or mutant hTIM-4 were labeled with CMFDA (green). Cells were co-cultured with PKH26 labeled eryptotic RBC (red) for 90 min, washed, and visualized with a fluorescent microscope.
Supplemental Figure 5 Movie. Apoptotic U937 were labelled with pH sensitive dye CypHer5E and incubated with hTIM-4 transfected 3T3 cells for 100 min, and washed. Localization of apoptotic bodies from neutral cell surface (colourless) into acidic intracellular endosomes (deep red) was detected by excitation with a 635 nm laser and emission fluorescence was collected using the fv1000 spectral scanner.
We express our thanks to Michael Robertson and Jerry Ritz for anti-Fas mAb and to Barbara Seaton for helpful discussions on PS receptors. This work was supported by NIH P01 AI054456 (To GF, DU, GK, RDK) and NIH R01 NS045937 (GF).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.