Transglutaminase Type II Is a Key Element in the Regulation of the Anti-Inflammatory Response Elicited by Apoptotic Cell Engulfment¹

Phagocytosis is a phylogenetically ancient process that plays a key role in essential basic functions such as macromolecular uptake, host defense, and degradation of senescent cells (1, 2). Monocyte/macrophages are professional phagocytes that have evolved a specialized and complex machinery that provides for highly efficient ingestion of pathogens and initiation of the immune response (3, 4). Macrophages also play an important role in the recognition and clearance of apoptotic cells. An essential feature of this physiological process is the absence of an inflammatory response (5). Macrophages perform the functions of recognition, binding, and internalization of apoptotic cells without undergoing activation, as is the case in the clearance of infectious organisms (6). Activation consists of morphological and behavior changes in macrophages that result in the secretion of preformed and/or newly synthesized constituents, such as cytokines and chemokines. These, in turn, switch on the inflammatory response (7). In contrast, macrophages that ingest apoptotic cells fail to secrete chemoattractants such as IL-8 and MCP-1 (8, 9). Instead, these cells release TGF-β1, platelet-activating factor, and PGE₂, factors that dampen the inflammatory response (6, 10).

The efficient clearance of apoptotic cells is also necessary to prevent an immune response that may be triggered by self-Ags, which are normally not exposed on living cells (11, 12). It is widely accepted that a persistent state of inflammation in various organs is a consequence of the defective clearance of apoptotic cells (13, 14) and contributes to the development of autoimmune disorders (15, 16, 17, 18).

The process of apoptotic cell removal includes a complex molecular machinery on both dying cells and phagocytes (reviewed in Refs.19 and 20) and is based on two steps: recognition/binding (tethering) and engulfment (tickling) (21). In the engulfment pathway, several different mammalian proteins have been identified (CrkII, Dock180, Rac, and Elmo) (22, 23) that appear to be involved in the internalization process by promoting cytoskeletal rearrangement. However, the exact cascade of events as well as many of the intermediate factors involved are still unknown.

We recently reported that the ablation of transglutaminase type II (TG2)³ in mice results in defective clearance of apoptotic cells associated with the development of splenomegaly, autoantibodies, and glomerulonephritis (24). TG2 is a multifunctional enzyme that belongs to a large family of calcium-dependent transamidating acyltransferase. TG2 catalyzes posttranslational modifications of proteins at the level of glutamine and lysine residues and is involved in a variety of biological functions, including cell death, extracellular matrix stabilization, and signaling (25). The activation of TG2 during apoptosis leads to the assembly of detergent-insoluble polymeric protein structures (25). In addition, TG2 interaction with major cytoskeletal components results in cytoskeleton rearrangements (26, 27), although as a G protein it can also have a role in cell signaling (28).

To gain further insight into the reported role of TG2 in apoptosis-dependent phagocytosis and the prevention of inflammation/autoimmunity, we performed an in vitro and in vivo study on the macrophage response after exposure to apoptotic cells. Using highly sensitive cytokine protein arrays, we examined the production of macrophage cytokine under unstimulated conditions. Our results demonstrate that phagocytosis of apoptotic cells is followed by rapid decrease in proinflammatory cytokine release, a phenomenon that is impaired in the absence of TG2.

Wild-type (WT) and TG2 knockout (TG2^−/−) C57/BL6 mice (29) were used. Male mice were used for all experiments. Animals were 4–6 mo of age. All care and procedures were performed according to approved protocols and in accordance with institutional guidelines.

The effect of the absence of TG2 on the clearance of apoptotic cells was investigated in in vivo and in vitro experiments. For in vivo experiments, we induced apoptosis in the liver by injection of a single dose of lead nitrate (PbNO₃). Equal numbers (n = 12) of WT and TG2^−/− mice were treated with PbNO₃; animals injected with vehicle only served as controls. Separate groups of WT and TG2^−/− animals were used as the source of peritoneal macrophages for in vitro studies.

Apoptosis was induced in vivo in the livers of WT and TG2^−/− mice by a single i.v. injection of PbNO₃ (10 μmol/100 g body weight; CarloErba) dissolved in physiological saline. The injection of PbNO₃ produces a proliferative response and a doubling of liver weight in 3 days. This is followed by an involution phase characterized by a massive induction of apoptotic cell death and active clearance of the apoptotic cells, reaching a maximum on day 5 after treatment (30). Restoration of pretreatment liver weight occurs after 2–3 days.

Animals were killed at different times after PbNO₃ injection, and liver samples were collected. For histological examination, small fragments of liver tissue were formalin-fixed, embedded in paraffin, and stained with H&E. For the cytokine analysis, blood sera were collected and stored at −70°C.

Peritoneal macrophages were isolated from WT and TG2^−/− mice. Macrophages were obtained by peritoneal lavage with PBS and enriched by adherence selection for 1 h at 37°C. Cells were cultured in RPMI 1640 medium supplemented with 10% FBS, 2 mM glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin at 37°C in 5% CO₂ for 48 h before use. Macrophages were plated in 2-well chamber slides (5 × 10⁵/well) for light microscopic analysis, in 24-well plates containing coverslips (1 × 10⁶/well) for scanning electron microscopy, and in 6-well plates (3 × 10⁶/well) for cytokine assays. To study apoptotic cell uptake, human apoptotic lymphocytes were used. PBLs were isolated by differential centrifugation using Ficoll-Paque, resuspended, and cultured in RPMI 1640 medium supplemented with 10% FBS, 2 mM glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin at 37°C in 5% CO₂. After 1 h, nonadherent cells were collected; the majority consisted of lymphocytes, as determined by morphological examination of cytospin preparation. Most monocytes adhered to the culture flasks and remained behind. These lymphocytes were then cultured for 24 h before use.

Apoptotic cell death was induced either by incubation with cycloheximide (10⁻² M, 12 h) or by exposure to UV irradiation at 254 nm for 10 min, followed by 3 h of culture. The percentage of apoptotic cells was quantified by flow cytometric analysis using annexin V and propidium iodide staining. Under the conditions used, cells were <5% propidium iodide positive and were 50–80% annexin V positive. Experiments were performed using UV- and cycloheximide-treated lymphocytes independently.

For the in vitro phagocyte interaction assays, apoptotic lymphocytes were washed twice, resuspended in fresh medium, and then added to the macrophage monolayer (3/1). The cells were allowed to interact at 4°C (binding) or at 37°C (engulfment) for different intervals of time. Phagocytosis of apoptotic cells was determined under a light microscope by counting 400 macrophages and was expressed as the percentage of macrophages containing apoptotic bodies. For cytokine analysis, cell culture supernatants were centrifuged at 2000 rpm to remove particulate and debris, then stored at −70°C.

Macrophages were fixed with 2.5% glutaraldehyde in 0.1 M Millonig’s phosphate buffer at 4°C for 1 h. After washing in Millonig’s phosphate buffer, cells were dehydrated in increasing acetone concentrations and then critical point-dried using liquid CO₂. Samples were sputter-coated with gold. Scanning electron microscopy was conducted using the Stereoscan 240 scanning electron microscope (Cambridge Instruments).

Cytokines released by macrophages were measured in the culture medium or in mouse sera using a Mouse Cytokine Array I Kit (Panomics) consisting of 22 different cytokine and chemokine Abs spotted in duplicate onto a membrane.

The membranes were processed according to the instructions of the manufacturer. Briefly, membranes were incubated with 4 ml of 1× blocking buffer at room temperature for 1 h, washed twice with 1× wash buffer II, and incubated with 2 ml of sample (mouse serum or macrophage culture medium) for 2 h at room temperature. After decanting the samples, membranes were washed three times with 4 ml of 1× wash buffer I (5 min/wash), followed by two washes with 1× wash buffer II (5 min/wash). The membranes were then incubated with 1.5 ml of diluted biotin-conjugated Abs at room temperature for 1 h, and washing steps were repeated as described above. Membranes were incubated for 1 h at room temperature with a dilution of streptavidin-conjugated peroxidase. After a thorough wash, the membranes were exposed to the mixed detection buffers (Panomics), according to the instructions of the manufacturer, for 5 min in the dark before imaging. Spots were visualized using ECL (Amersham Biosciences). Membranes exposed to Kodak X-OMAT radiographic film were then processed. Each film was scanned, and spot densities were measured with Scion Image for Windows (National Institutes of Health). The densities were exported into Excel, and the background intensity was subtracted before analysis.

TGF-β expression was evaluated by Western blot, immunohistochemical analysis, and ELISA.

Western blot analyses of TGF-β was performed on liver samples from WT and TG2^−/− mice, both untreated and 3 or 5 days after PBNO₃ injection. Frozen tissue fragments were homogenized with lysis buffer (50 mM Tris-HCl, pH 8, containing 120 mM NaCl, 0.2 mM EDTA, 1% Nonidet, and protease inhibitor mixture). Aliquots of total protein extracts were run on 10% SDS-polyacrylamide gels and electroblotted onto nitrocellulose membrane overnight at 4°C in 25 mM Tris, 192 mM glycine, and 20% (v/v) methanol. Membranes were then incubated for 1 h at ambient temperature with rat anti-mouse mAb specific for TGF-β1 (BD Pharmingen) and with the appropriate secondary HRP-conjugated rabbit anti-rat Ab (Sigma-Aldrich). Detection was achieved using a preformed streptavidin-HRP complex (Amersham Biosciences), and the signal was developed using ECL detection system (Amersham Biosciences).

For immunohistochemistry, formalin-fixed, paraffin-embedded, thin liver sections from WT and TG2^−/− mice (untreated or 3 or 5 days after PBNO₃ injection) were used. Endogenous peroxidase activity was blocked by 3% H₂O₂ for 5 min. After rinsing in PBS, sections were incubated with normal goat serum for 5 min. Rat anti-mouse TGF-β mAb (BD Pharmingen) was applied at 1/10 dilution in PBS for 1 h at room temperature. A peroxidase-conjugated rabbit anti-rat IgG (Sigma-Aldrich) was used as secondary Ab. The immunoreaction product was revealed using aminoethylcarbazole (Biogenex) as chromogen substrate and 0.01% H₂O₂. Sections were counterstained in Mayer’s acid hemalum.

The TGF-β concentrations in mouse sera and in the macrophage culture supernatants were determined by ELISA, using Quantikine immunoassays (R&D Systems). Samples were acid treated (to activate latent TGF-β) and neutralized before ELISA, according to the manufacturer’s instructions.

Values were expressed as the mean ± SD. Analysis was performed by two-tailed Student’s t test for comparisons between two groups. Statistical significance was set at p < 0.05.

We recently reported that the ablation of TG2 in mice is associated with the defective clearance of apoptotic cells (24). Light microscopic observations of the liver after apoptosis induction by lead nitrate treatment revealed that the deficiency in apoptotic cell clearance observed in the TG2^−/− mice (Fig. 1,c) was paralleled by an increased inflammatory response, as evidenced by the presence of large infiltrates of blood cells in the parenchimal tissue (Fig. 1,d). This phenomenon was completely absent in control mice (Fig. 1, a and b).

To investigate the role of TG2 in the phagocytosis of apoptotic cells, in a first set of experiments we performed in vitro phagocytosis assays. Cultured peritoneal macrophages from WT mice exposed to apoptotic lymphocytes displayed a very efficient ability to recognize and phagocyte apoptotic cells, and we observed that the vast majority of macrophages had internalized two or more apoptotic cells after 20 min of incubation (Fig. 2,A, a and b). When macrophages isolated from TG2^−/− mice were cocultured with apoptotic lymphocytes, phagocytosis was impaired. We observed a marked reduction in their general ability to ingest apoptotic cells as well as in the average number of apoptotic cells ingested (Fig. 2,A, c and d). The percentages of macrophages that contained apoptotic lymphocytes after 30 min of exposure in WT and TG2^−/− mice were 80.2 ± 7.5 and 49.5 ± 3.6%, respectively. It is important to note that this deficiency appeared to be dependent on macrophages; in fact, no difference was observed in phagocytic efficiency when lymphocytes from TG2^−/− mice were used (not shown). To discriminate between recognition (binding) and engulfment capacities, we performed an interaction assay, incubating apoptotic lymphocytes with macrophages at 4°C for 30 min. Under these conditions, binding without internalization occurred, as indicated by the peripheral association of apoptotic cells with macrophages. Macrophages from WT and TG2^−/− mice exhibited similar binding capacities (one macrophage can bind one or more apoptotic cells; Fig. 2 Ba), demonstrating that the lack of TG2 does not affect the ability to recognize and bind apoptotic cells, but only the capacity to internalize them.

To investigate whether the different phagocytic capacity observed in TG2^−/− macrophages was related to a reduced rate of interaction with apoptotic cells, we conducted phagocytosis analysis at various time intervals. As reported in Fig. 2,Bb, our results showed that the percentage of TG2^−/− macrophages containing apoptotic cells was significantly lower and never reached WT levels. We also performed a time-course analysis of phagocytosis monitored by scanning electron microscopy. Cultured macrophages from WT mice were characterized by membrane ruffling with microvilli of variable lengths over the entire cell surface (Fig. 3). When apoptotic lymphocytes were added to the culture medium, they were rapidly bound by phagocytes, which immediately started to surround them (Fig. 3,a). Internalization of apoptotic cells was completed in 15–20 min, demonstrating the ability of macrophages to clear apoptotic cells (Fig. 3,b). At this stage, the cells showed globular protrusions, indicating the presence of engulfed corpses, whereas macrophages displayed a more flattened phenotype at longer time intervals (1 h after exposition of apoptotic cells), suggesting that the phagocytic process was completed (Fig. 3,c). In contrast, macrophages from TG2^−/− mice showed completely different behavior. In fact, apoptotic lymphocytes were noted to adhere to the surface of macrophages after a few minutes of coincubation (Fig. 3,d); however, this first step, i.e., the tether mechanism, was not properly followed by the engulfment phase in the majority of macrophages, even after longer intervals (Fig. 3, f and g). Interestingly, increasing the exposure time (1–2 h; during which time, numerous apoptotic cells were still attached to macrophage surface), the shape of the macrophages changed to an activated phenotype, consisting of a round shape and extensive membrane ruffling (Fig. 3 g).

Evaluation of the cytokine expression profile in macrophage culture media or mouse sera was performed using a highly sensitive cytokine Ab array method, enabling the simultaneous detection of low concentrations of multiple cytokines in one assay (picogram per milliliter range). The map of the 22 cytokines spotted on the membranes is reported in Fig. 4 A. The detectable cytokines in our experimental systems were first evaluated by experiments using untreated WT and TG2^−/− mice, both in vitro (on macrophage culture medium) and in vivo (on blood serum). The results reported in Fig. 4,B show that ∼50% of all available cytokines on the filters were detectable, even though some were at a very low level (Fig. 4 B).

Among the cytokines released in the supernatant by cultured macrophages (Fig. 5), four proinflammatory cytokines and chemokines showed major changes in response to apoptotic cell exposure: IL-12p70, RANTES, MCP-1, and MCP-5 (Fig. 5,B). The modification in the levels of these cytokines was not due to the FBS present in the culture medium, because incubation with medium alone did not produce any detectable signal; in addition, no cytokines were detected in the supernatants when lymphocytes (viable or apoptotic) were incubated alone (data not shown). Thus, the cytokine expression profile reported in Fig. 5,A reflects the macrophage production. Interestingly, the secretion of all cytokines considered was strongly inhibited 20 min after exposure of WT macrophages to apoptotic cells (Fig. 5).

Of note, the unstimulated macrophages obtained from TG2^−/− mice produce similar amounts of cytokines as those detected in WT macrophages (Fig. 5,A), but upon exposure to apoptotic cells, we found a completely different response (Fig. 5,B). In fact, the above-described cytokines were still released in considerable amounts after exposure of the macrophages to apoptotic lymphocytes (3-fold higher compared with WT). In addition, the expression of IL-6 was detected only when TG2^−/− macrophages were exposed to apoptotic cells (Fig. 5 A).

To evaluate whether the impaired clearance of apoptotic cells observed in the livers of TG2^−/− mice was reflected by cytokine production, we analyzed the sera obtained from control and lead nitrate-injected mice (Fig. 6) by the cytokine Ab arrays method. Of note, the cytokine profile in mice sera (Fig. 6,A) was comparable to that observed with isolated macrophages (compare with the profile in Fig. 5,A). However, in addition to IL-12p70, RANTES, MCP-1, and MCP-5, two other cytokines, IL-12 and soluble TNFR type I (sTNF-RI), showed significant differences (Fig. 6). For each cytokine considered, similar values were found in WT and TG2-null control mice; once again, as observed in isolated macrophages, opposite trends were detected in response to apoptotic cell exposure (Fig. 6,A). In fact, although a general reduction in the production of inflammatory cytokines (in particular, the chemokines MCP-1 and MCP-5 completely disappeared) was detected in WT serum 5 days after PbNO₃ injection (when there was the maximal phagocytic activity of apoptotic cells) (24), the phenomenon was completely absent in TG2^−/− mice. Interestingly, the proinflammatory cytokine IL-12 and sTNF-RI levels were significantly higher even when compared with those in untreated control animals (Fig. 6 B).

It has been proposed that the anti-inflammatory immunosuppressive mechanism induced during macrophage ingestion of apoptotic cells leading to the production of anti-inflammatory cytokines is regulated by TGF-β. In view of the fact that TG2 has been shown to participate in the maturation of TGF-β (31), we have analyzed the expression of this cytokine in the livers of WT vs TG2^−/− mice after lead nitrate treatment. Western blot analysis, performed on control and PbNO₃-treated animals (Fig. 7 A), showed minimal expression in WT liver under normal conditions. An increased amount of TGF-β was observed after 3 days of treatment, which further increased on day 5. Interestingly, TG2^−/− mice showed a higher steady-state TGF-β levels with respect to WT controls, and this difference was further enhanced after PbNO₃ treatment.

It has been demonstrated that macrophages are the major source of TGF-β after in vivo injection of apoptotic cells into peritoneum or lung (10). Consequently, we carried out immunohistochemical localization of TGF-β in PbNO₃-treated livers (Fig. 7,B). Interestingly the enzyme showed a different localization in WT and TG2^−/− mice. In WT livers, immunostaining was located at the sinusoidal level, where Kupffer cells (the tissue macrophages of the liver) are normally present (Fig. 7,B, a and c); in contrast, in TG2^−/− TGF-β was detected in hepatic parenchyma, particularly in hepatocytes facing the portal tract and in areas containing inflammatory infiltrates (Fig. 7,B, b and d). To verify by ELISA whether this increased TGF-β production was reflected in the level of the circulating cytokines, we measured the level of TGF-β in the blood of control vs PbNO₃-treated mice (Fig. 7,Ca). The basal level of circulating TGF-β in WT animals was not modified by apoptosis induction in the liver. Interestingly, in TG2^−/− mice, differently from what was observed in liver, the basal level of circulating TGF-β was lower compared with that in WT mice (50%) and did not change after PbNO₃ treatment despite the accumulation of the cytokine in the liver (Fig. 7 Ca).

The measurement of TGF-β levels in the culture medium of WT macrophages incubated with apoptotic cells (Fig. 7 Cb) showed a significant increase after 60 min of exposure to apoptotic cells (no significant modifications were observed at 20 min). However, this increase was significantly less pronounced in TG2^−/− macrophages, suggesting a defect in the regulation of this cytokine in the absence of TG2.

An emerging feature of chronic inflammatory diseases is the persistent presence of apoptotic cells resulting from impaired phagocytosis (16, 18). We show in this study that macrophage uptake of apoptotic cells is dramatically affected in the absence of TG2, whereas recognition and binding are not. In addition, we demonstrated that the defective clearance of apoptotic cells is associated with altered proinflammatory cytokine release. In particular, we show that the ablation of TG2, which is necessary for the correct uptake of apoptotic cells, also affects macrophage inactivation. Furthermore, our results indicate that the binding of apoptotic cells is not sufficient to inhibit the release of inflammatory cytokines. This suggests that the second step of phagocytosis (i.e. internalization) is required. This finding is in agreement with a recent study concerning inflammation in atherosclerosis, which demonstrates that the impairment of engulfment, and not binding, of apoptotic cells by macrophages caused by the presence of oxidized lipoproteins may promote inflammation (32).

A large body of work deals with the inhibition of cytokine release from macrophages that have ingested apoptotic cells (6, 33, 34). By contrast, Kobayashi et al. (35) reported that the interaction of phagocytes with apoptotic cells leads to the production of proinflammatory cytokines. Additional investigation by the same authors led to the conclusion that the production of inflammatory cytokines by macrophages can be induced by exposure to late apoptotic cells (36). This event normally does not occur in vivo, because apoptotic cells are cleared as soon as they appear, thus leading to an almost null response (37). Our work provides new insight into this controversy. In fact, considering that the macrophage cytokine expression levels are generally very low, methods used to measure their expression after apoptotic cell exposure need to be conducted under stimulatory conditions (i.e., after LPS treatment) and/or after prolonged incubation with target cells (6, 33, 34). For the first time, we analyzed cytokine release modulation under unstimulated conditions and after a very short time of exposure to apoptotic cells. Our results show that the basal level of cytokine expression is immediately inhibited by the internalization of apoptotic cells both in vivo and in vitro.

Another controversial issue concerns the level (transcriptional or translational) of regulation of cytokine release (38). Our results are consistent with a post-transcriptional regulation, because we found that the shutoff of inflammatory cytokine release occurs very rapidly. We hypothesize that the regulation could involve the stability or degradation of cytokine transcripts after the inhibition of their secretion. Indeed, it has been reported that production of inflammatory cytokines by macrophages requires activation of the p38 MAPK pathway. This promotes the stability and translability of inflammatory cytokine transcripts (39, 40), whereas selective inhibitors of p38 MAPK prevent proinflammatory cytokine release (41). In keeping with this, it is interesting to note that TG2 activity is required for activation of p38 MAPKs (42).

We previously reported that mice lacking TG2 are prone to develop inflammatory pathologies (24, 43). The data reported in this study demonstrate that no major differences in the basal levels of cytokines exist between WT and TG2^−/− mice, with the exception of circulating TGF-β. However, our results indicate that a strong impairment of phagocytic clearance of apoptotic cells could account for the development of inflammatory disorders. In fact, we showed that the chemokines MCP-1 and MCP-5 are not attenuated in the absence of TG2. This event might result in the selective recruitment of monocytes into inflammatory sites, as observed in vivo in the livers of TG2^−/− mice after apoptosis induction, where many inflammatory cells infiltrate the parenchimal tissue. Furthermore, we showed that in knockout mice, the expression of both IL-12 and sTNF-RI significantly increased. Interestingly, the dysregulation of IL-12 has been described in a wide range of autoimmune-prone mouse models (44, 45, 46).

The down-regulation of proinflammatory mediator release by macrophages in response to apoptotic cell ingestion has been largely attributed to TGF-β stimulation (10, 38). Accordingly, our data showed that in the livers of WT mice, TGF-β expression increases during the clearance of apoptotic cells, although it is minimally produced under normal conditions. Immunohistochemical analysis showed that the site of TGF-β expression is at the level of sinusoidal cells. Surprisingly, we found that TGF-β also increases in the livers of TG2^−/− mice, and the up-regulation is even higher than that observed in WT livers. These results might appear to contrast with the unrepressed production of proinflammatory cytokines observed in TG2^−/− mice. It is noteworthy that TGF-β in knockout mice is localized not in the sinusoidal compartment, but, rather, in the parenchimal cells. A similar expression pattern has been reported for the liver during inflammatory conditions, such as fibrosis in chronic liver disease and autoimmune hepatitis (47, 48), where a close correlation exists between this abnormal localization of TGF-β and the inflammatory activity in the tissue, thus confirming the ability of hepatocytes localized near areas of inflammatory infiltrates to synthesize TGF-β (47, 48). Our findings demonstrating a similar TGF-β localization confirm that the defective phagocytosis of apoptotic cells caused by TG2 ablation results in the imbalance of the inflammatory regulation at multiple levels. Particularly controversial is the role of the systemic level of TGF-β in inflammatory diseases. Local and circulating effects of TGF-β could be divergent (49, 50). In fact, due to the different systemic effects exerted by TGF-β and particularly its essential role in the maintenance of self-tolerance, its stimulation and secretion must be tightly regulated (51). Our data support this view; in fact, the clearance of apoptotic cells locally restricted to the liver does not affect the circulating level of TGF-β.

It has been reported that TGF-β is the only cytokine to be up-regulated in macrophages after the ingestion of apoptotic cells and that this is responsible for the proinflammatory cytokine down-regulation (38). It should be noted that in those studies TGF-β release was evaluated after a prolonged time of incubation with apoptotic cells. By contrast, the results of our experiments, performed a short time after apoptotic cell exposure, suggest that the inhibition of proinflammatory cytokines occurs rapidly and independently from TGF-β. These results are also in agreement with the hypothesis that TGF-β could act at a later time, amplifying the immunosuppressive response (52). As far as the role of TG2 on TGF-β expression is concerned, it must be stressed that the absence of TG2 causes a dysregulation of the steady-state level of the enzyme. In fact, compared with WT mice, TG2^−/− mice displayed an increased level of TGF-β in liver and a lower amount in blood. The discrepancy between liver and circulating TGF-β levels suggests an impaired release of TGF-β. Regarding this point, it is important to note that TG2 may affect the vesicle traffic that is necessary for cytokine secretion, through modification of major components of the cytoskeleton (26, 53).

In conclusion, our study demonstrates that TG2 ablation in mice impairs the ability of macrophages to engulf dying cells, potentiating the susceptibility to inflammatory pathologies. Our data have profound clinical relevance, considering the recent proposal to use TG2 inhibitors for therapeutic applications in autoimmune and inflammatory diseases.

We thank all the staff of the animal facility at Regina Elena Institute for their technical assistance.

The authors have no financial conflict of interest.