Bone marrow transplantation reveals the in vivo expression of the mitochondrial uncoupling protein 2 in immune and nonimmune cells during inflammation.

The mitochondrial uncoupling protein 2 (UCP2) is expressed in spleen, lung, intestine, white adipose tissue, and immune cells. Bone marrow transplantation in mice was used to assess the contribution of immune cells to the expression of UCP2 in basal condition and during inflammation. Immune cells accounted for the total amount of UCP2 expression in the spleen, one-third of its expression in the lung, and did not participate in its expression in the intestine. LPS injection stimulated UCP2 expression in lung, spleen, and intestine in both immune and non-immune cells. Successive injections of LPS and dexamethasone or N-acetyl-cysteine prevented the induction of UCP2 in all three tissues, suggesting that oxygen free radical generation plays a role in UCP2 regulation. Finally, both previous studies and our data show that there is down-regulation of UCP2 in immune cells during their activation in the early stages of the LPS response followed by an up-regulation in UCP2 during the later stages to protect all cells against oxidative stress.

UCP2 1 belongs to a newly discovered subgroup of mitochondrial carrier proteins related to the well known UCP1 from brown adipose tissue (for review, see Ref. 1). Unlike UCP1, UPC2 and its related family member UCP3 do not seem to be involved in cold-adapted thermogenesis. Studies on Ucp2 Ϫ/Ϫ mice revealed two intriguing phenotypes (for review, see Ref. 2). Zhang et al. (3) found that Ucp2 Ϫ/Ϫ mice exhibit higher serum insulin and lower blood glucose levels compared with wild-type mice. In vitro, pancreatic islets from Ucp2 Ϫ/Ϫ mice had increased insulin secretion in response to glucose, which was correlated to a higher ATP level (3). Arsenijevic et al. (4) found an improvement of the macrophage oxidative burst activity in Ucp2 Ϫ/Ϫ mice. Macrophages isolated from Ucp2 Ϫ/Ϫ mice infected with Toxoplasma gondii produced more reactive oxygen species (ROS) and had a higher toxoplasmacidal activity compared with wild-type mice (4). This enhanced macrophage oxidative activity was also confirmed in another mouse model. LDL R Ϫ/Ϫ mice, which are susceptible to atherosclerosis when fed an atherogenic diet, were transplanted with bone marrow cells from either Ucp2 Ϫ/Ϫ or Ucp2 ϩ/ϩ littermate mice. In this mouse model of atherosclerosis, the lack of UCP2 accelerated the development of unstable atherosclerotic plaques, showing a protective role for UCP2 in atherosclerosis (5). Altogether, these results suggest that UCP2 acts as a mild uncoupler, controlling both ATP synthesis as well as the production of ROS (4,6,7). However, conflicting data have been obtained regarding the respiratory coupling state of Ucp2 Ϫ/Ϫ and Ucp3 Ϫ/Ϫ mice mitochondria (8 -12). Although the uncoupling activity of all UCPs has been established in vitro in proteoliposomes (13)(14)(15) and yeast or mammalian mitochondria (16,17), the in vivo activity of UCP2 and UCP3 is still a matter of debate.
Another approach to understanding the physiological roles of UCP2 has been to study the expression of the protein in wildtype mice. The detection of UCP2 protein is difficult for two main reasons. First, UCP2 is at least 160-fold less abundant in spleen mitochondria than UCP1 is in brown adipose tissue. Second, most of the anti-peptide antibodies cross-react with other mitochondrial carriers and, therefore, are misleading. Nevertheless, with the help of the Ucp2 Ϫ/Ϫ mice as negative controls, a highly sensitive anti-UCP2 polyclonal antibody suitable for in vivo detection of UCP2 has been characterized. UCP2 was found to be expressed in spleen, lung, intestine, and, at lower levels, in white adipose tissue (18).
Taking as given the immunologic phenotype of Ucp2 Ϫ/Ϫ mice (4), the induction of the protein after LPS treatment (18), and the presence of immune cells in almost all organs where UCP2 protein was found (18), we set out to determine the precise contribution of immune cells in the expression of UCP2. In the absence of antibodies suitable for in situ detection of UCP2, bone marrow transplantation between Ucp2 Ϫ/Ϫ and Ucp2 ϩ/ϩ mice was performed. Levels of UCP2 in all groups of mice were assessed at basal conditions after LPS treatment. To provoke liver injury, mice were also fed an atherogenic diet. Finally, we investigated the effects of N-acetyl-cysteine and dexamethasone on the expression of UCP2 in the LPS model.

MATERIALS AND METHODS
Animals and Treatments-Studies on mice were performed in agreement with the institutional CNRS guidelines defined by the European Community guiding principles and by the French decree No. 87/848 of October 19, 1987. Authorization to perform animal experiments was given by the French Ministry of Agriculture, Fisheries and Food (permit A92580 was issued February 2, 1994 and 92-148 was issued May 14, 2002). All mice were 7-10 weeks old. Ucp2 Ϫ/Ϫ mice have been described previously. The mice were transferred onto a C57BL/6J genetic background (99.2%). C57BL/6J mice were purchased from Elevage Janvier (Orleans, France). Mice were injected intraperitoneally with various concentrations of LPS from Escherichia coli serotype 055:B5 (Sigma) and left with free access to food and water. Rectal temperature was measured with a thermocouple probe thermometer (CHY 508BR, Bioseb, Chaville, France). Mice were killed 14 h after injection. Atherogenic diet was purchased from UAR (Epinay sur Orge, France) and consisted of an A03 diet complete with 15% cocoa butter, 1.25% cholesterol, and 0.5% sodium cholate. Medullar aplasia was induced by 9.5 Gy total body irradiation. Mice were injected intravenously under anesthesia with bone marrow cells (1.2 ϫ 10 6 ) extracted from the femur and tibia of either Ucp2 ϩ/ϩ or Ucp2 Ϫ/Ϫ mice. The Ucp2 ϩ/ϩ mice that received bone marrow from Ucp2 Ϫ/Ϫ mice were named Ucp2 ϩ/ϩ tKO, and the Ucp2 Ϫ/Ϫ mice that received bone marrow from Ucp2 ϩ/ϩ were named Ucp2 Ϫ/Ϫ tWT. Macrophage engraftment of hemapoietic tissues after bone marrow transplantation requires one month, whereas the engraftment of Kuppfer cells in liver or of resident lung macrophages requires 3-6 months (19,20). Therefore, transplanted mice were allowed to recover for 4 months to ensure that the engraftment of resident macrophages was close to completion. Transplanted mice were either fed an atherogenic diet for 12 additional weeks or treated with LPS.
Morphometric Analyses-The aortic sinus and aorta of the mice were frozen. Serial sections of the aortic sinus were assayed for lipid deposition with oil red O. Immunohistochemistry was performed as described previously (21) using a monoclonal rat anti-mouse macrophage antibody, clone MOMA-2 (BIOSOURCE International).
Biochemical Methods: Isolation of Mitochondria and Western Blot Analysis-All steps were carried out at 4°C. Fresh tissues were minced in TES buffer (10 mM Tris, pH 7.5, 1 mM EDTA, 250 mM sucrose) supplemented with the following protease inhibitors: 1 mM benzamidine, 4 g/ml aprotinin, 1 g/ml pepstatin, 2 g/ml leupeptin, 5 g/ml bestatin, 50 g/ml sodium-tosyl-phechloromethyl ketone, and 0.1 mM phenylmethylsulphonyl fluoride. Minced tissue was carefully disrupted in a Thomas' potter at low speed rotation. Unbroken cells and nuclei were removed by two successive centrifugations of the homogenate at 750 ϫ g for 10 min. Mitochondria were collected after centrifugation of the supernatant at 10,000 ϫ g for 20 min, and protein content was assayed using a bicinchoninic acid kit (Sigma). The anti-UCP2 antibody (hUCP2-605) and the Western blotting conditions have been described previously (18). Anti-cytochrome c oxydase subunit I (COX I), monoclonal antibody (1D6), and anti-mitochondrial porin monoclonal antibody (20B12) were purchased from Molecular Probes (Leiden, The Netherlands) and used at 0.5 g/ml. Direct recording of the chemiluminescence was performed with the charge-coupled device camera of the GeneGnome instrument, and quantification was achieved using GeneSnap software (Syngene, Ozyme, Saint Quentin en Yvelines, France).
Statistical Analysis-Data are expressed as mean Ϯ S.E. Comparisons between groups were made by using a one-way analysis of variance. In the figures, the following correspondences may be found: *, p Յ 0.05; **, p Յ 0.01; and ***, p Յ 0.001. p Յ 0.05 was considered as statistically significant. Fig. 1 shows the induction of UCP2 protein 14 h after LPS injection. In contrast to the LPS serotype 0111:B4 used in our previous study (18), the LPS serotype 055:B5 increased UCP2 not only in lung but also in spleen and duodenum (Fig. 1A). At an LPS concentration of 5 mg/kg, UCP2 levels increased 2.7fold in spleen (p Յ 0.001). UCP2 levels in the lung and duodenum were increased 2.8-and 2.3-fold, respectively, at an LPS concentration of 10 mg/kg; however, these levels of expression were consistently less than those seen in the spleen (Fig. 1B), which is the predominant organ of UCP2 expression in the conditions examined. Given the importance of the liver in the metabolic response to endotoxin and previous results obtained by Faggioni et al. (22) and Cortez-Pinto et al. (23), we examined the induction of UCP2 in liver by LPS treatment. At a high concentration of this LPS serotype (10 mg/kg), UCP2 increased to a hardly detectable level by our antibody and reached 25% of the basal level of UCP2 in lung ( Fig. 2A). The protein band detected below UCP2 is still present in liver mitochondria from Ucp2 Ϫ/Ϫ (Fig. 2B, lane 4) and is, therefore, not UCP2.

LPS Serotype 055:B5 Induces UCP2 in Multiple Organs-
Bone Marrow Transplantation Reveals the Contribution of Immune Cells to UCP2 Expression-Expression of UCP2 mRNA in liver has been described in mice fed a fish oil diet (24), LPS-treated mice (22), and genetically deficient ob/ob mice (25)(26)(27). In these animal models, liver inflammation can occur, and it is possible, therefore, that the expression of UCP2 observed is due to resident or infiltrating macrophages. To test this hypothesis, bone marrow transplantation was performed. Ucp2 ϩ/ϩ mice, Ucp2 Ϫ/Ϫ mice, Ucp2 ϩ/ϩ tWT-transplanted mice, and Ucp2 ϩ/ϩ tKO-transplanted mice were fed an atherogenic diet or were treated with LPS. After LPS treatment (10 mg/kg), UCP2 protein was induced in liver from Ucp2 ϩ/ϩ and Ucp2 ϩ/ϩ -tKO mice ( Fig. 2A), demonstrating that LPS induced UCP2 in hepatocytes. After 6 weeks of an atherogenic diet containing cholate, the formation of atherosclerotic plaques in both the Ucp2 Ϫ/Ϫ and Ucp2 ϩ/ϩ groups of mice was observed (Fig. 2B). Consistent with our previous study on transplanted LDL R Ϫ/Ϫ mice fed a cholate-free atherogenic diet (5), the mean size of the atherosclerotic plaques was increased four times in Ucp2 Ϫ/Ϫ mice compared with Ucp2 ϩ/ϩ (Fig. 2B; p ϭ 0.0011). In contrast to the LPS model of inflammation, the atherogenic diet induced UCP2 protein in liver mitochondria from Ucp2 ϩ/ϩ and Ucp2 ϩ/ϩ -tWT mice but not in liver mitochondria from Ucp2 ϩ/ϩ tKO mice FIG. 1. LPS induces UCP2 in lung, spleen, and intestine. Increasing amounts of LPS (NaCl as control) were intraperitoneally injected into mice (five mice per condition). A, Western blot analysis of UCP2 content in spleen, lung, and duodenum mitochondria. UCP2 was detected with 0.1 g/ml hUCP2-605 antibody. The antibody against the cytochrome c oxidase subunit I protein was used to estimate the amount of mitochondrial proteins. B, relative amount of UCP2 protein in all three tissues in basal or stimulated conditions. UCP2 level in lung of control mice was taken as reference. (Fig. 2C). This result shows that the relatively high level of UCP2 protein (50% of the level of UCP2 in control lung) detected in response to an atherogenic diet is associated with immune cells and not with hepatocytes. To confirm this observation, frozen sections of liver were immunostained with the MOMA-2 macrophage-specific antibody (Fig. 2D). Resident macrophages were present in both Ucp2 ϩ/ϩ control mice and in Ucp2 ϩ/ϩ tWT-transplanted mice, showing that 4 months after bone marrow transplantation, repopulation of the liver by immune cells has occurred. As revealed by the strong MOMA-2 staining, the atherogenic diet in the presence of cholate triggered a massive infiltration of macrophages in the liver of Ucp2 ϩ/ϩ tWT mice (Fig. 2D).
In spleen, expression of UCP2 almost disappeared in the Ucp2 ϩ/ϩ tKO mice (Fig. 3A), showing that UCP2 protein is more than 90% associated with immune cells in this organ (Fig. 3D). In contrast, expression of UCP2 in duodenum disappeared in Ucp2 Ϫ/Ϫ tWT mice, demonstrating the expression of UCP2 within intestine cells only (Fig. 3B). Lung exhibited an intermediary phenotype. A significant amount of UCP2 protein is present in the lung of Ucp2 Ϫ/Ϫ tWT (Fig. 3C), corresponding to 30% of the signal found in untransplanted mice. To verify that the transplantation was completed, wild-type mice were also transplanted with wild-type bone marrow (Ucp2 ϩ/ϩ tWT), and the level of UCP2 in spleen was measured in all groups of mice. As shown in Fig. 3D, the levels of UCP2 protein in the spleen of Ucp2 ϩ/ϩ tWT and Ucp2 Ϫ/Ϫ tWT mice were comparable to the level of UCP2 in the spleen of untransplanted wild-type mice.
LPS injection stimulated UCP2 protein in spleen of Ucp2 Ϫ/Ϫ tKO mice only (data not shown). However, in the lung, stimulation of UCP2 expression occurred in both groups of mice (Fig. 4), showing that 14 h after injection, both immune and non-immune cells up-regulate UCP2. In intestine, a weak UCP2 protein signal appeared in the Ucp2 Ϫ/Ϫ tKO mice treated with LPS (data not shown), indicating that, in addition to the induction of UCP2 by intestine cells, a small proportion of immune cells contributed to UCP2 induction by LPS. Effects of N-Acetyl-cysteine and Dexamethasone after LPS Injection-To confirm that UPC2 responds to the septic shock induced by LPS, mice were first injected intraperitoneally with LPS (5 mg/kg) and 1 h later, with either N-acetyl-cysteine (150 mg/kg) or dexamethasone (1 mg/kg). As expected, the high concentration of LPS induced a severe hypothermia, with body temperature decreasing to 35°C (Fig. 5A). Both N-acetyl-cysteine and dexamethasone significantly restored the body temperature (p Յ 0.01 or 0.001) and prevented UCP2 induction in spleen, lung, and duodenum (Fig. 5, B and C).

Bone Marrow Transplantation Reveals New Expression
Pattern of UCP2-In spleen, UCP2 expression is thoroughly restricted to immune cells, which confirms that UCP2 is expressed in macrophages and lymphocytes as proposed previously (4,11). However, immune cells do not contribute to the UCP2 protein detected in intestine. The lung exhibits an intermediary phenotype. In liver, expression of UCP2 can be ascribed to either hepatocytes or to resident and infiltrating immune cells, depending on the inflammatory model. The pres-FIG. 2. Induction of UCP2 in liver mitochondria upon LPS treatment or atherogenic diet. Ucp2 ϩ/ϩ mice were irradiated and transplanted with bone marrow from Ucp2 Ϫ/Ϫ littermate mice, whereas Ucp2 Ϫ/Ϫ mice were transplanted with bone marrow from Ucp2 ϩ/ϩ littermate mice. Four months after transplantation, Ucp2 ϩ/ϩ mice, Ucp2 Ϫ/Ϫ mice, Ucp2 ϩ/ϩ tWT-transplanted mice, and Ucp2 ϩ/ϩ tKO-transplanted mice were subjected to LPS injection or to an atherogenic diet. A, Western blot analysis of UCP2 induction in liver mitochondria after 10 mg/kg LPS or NaCl injection. B, effect of the atherogenic diet on Ucp2 Ϫ/Ϫ and Ucp2 ϩ/ϩ mice. Serial sections of the aortic sinus were assayed for lipid deposition with oil-red. The arrow indicates the atherosclerotic plaques (L, lumen). Lesion sizes were also measured on both groups of mice. C, Western blot analysis of the expression of UCP2 in liver after the atherogenic diet. COX I antibody was used to normalize the amount of mitochondrial protein loaded. D, representative examples of MOMA-2 immunostaining in red of frozen liver sections from Ucp2 ϩ/ϩ mice and Ucp2 ϩ/ϩ tWT mice on normal diet and from Ucp2 ϩ/ϩ tWT after the atherogenic diet. ence of cholate in the atherogenic diet, which causes a chronic inflammation of the liver, triggers macrophage infiltration, whereas LPS injection, which induces a severe septic shock for a short period of time, is probably not sufficient to allow a massive infiltration of immune cells. All together, these data showed that immune cells are a dominant site for UCP2 expression, which should be taken into account when measuring UCP2 expression in vivo.
Anti-inflammatory Drugs Confirm the Involvement of UCP2 during Inflammation-Dexamethasone has been shown to significantly reduce neutrophil infiltration in bronchoalveolar lavage fluid and, consequently, the oxidative burst activity in the lung of LPS-treated mice (28). It also inhibits, via the NFB regulation pathway, the production of cytokines and chemokines such as TNF␣, interleukine-1␣ and -1␤, and macrophage inhibitor protein-1␣ (28,29). In contrast, NAC exerts a weaker effect on cytokine regulation (28, 30), but it attenuates several FIG. 4. LPS induction of UCP2 in the lung of translanted mice. Ucp2 ϩ/ϩ tKO mice and Ucp2 Ϫ/Ϫ tWT mice were intraperitoneally injected either with LPS (10 mg/kg) or with NaCl as control. Lung mitochondria were prepared, and the expression of UCP2 was measured by Western blot in all groups of mice. A, immunoblot decorated with the hUCP2605 and the COX I antibody to normalize the amount of mitochondrial protein loaded. B, quantification of the induction of UCP2 by LPS in nonimmune cells (Ucp2 ϩ/ϩ tKO mice) and in immune cells (Ucp2 Ϫ/Ϫ tWT mice). NaCl-injected mice were used as references. pathophysiologic changes in lung tissue, as reported by Bernard et al. (31) in a sheep model of adult respiratory distress syndrome. The inhibiting effects of dexamethasone and NAC on UCP2 induction by LPS suggest that free radicals may directly regulate the expression of UCP2 either at the translational level as described by Pecqueur et al. (18) or at the transcriptional level. In addition, the possibility that UCP2 regulation is determined by cytokine levels cannot be excluded. Systemic LPS injection in mice triggers an increase in blood TNF␣ level in less than 2 h, which is followed by bronchopulmonary hyperactivity and an accumulation of neutrophils in the microvasculature of the lung (29). UCP2 induction occurs just after the peak of blood cytokine levels. However, Menon et al. (32) reported that TNF␣ is not required for the up-regulation of UCP2 mRNA levels observed in the genetically obese ob/ob mice, and Nakatani et al. (33) showed that PPAR␣ activators up-regulated liver Ucp2 mRNA by means of a TNF␣independent pathway.
Toward a Two-step Model of in Vivo Regulation of UCP2 in Immune Cells-UCP2 expression in immune cells upon LPS treatment has been studied in vitro by several groups. We initially described a down-regulation of UCP2 mRNA in cultured peritoneal macrophages 4 h after LPS treatment (4). Lee and et al. (34) observed the same effect with LPS treatment and furthermore demonstrated that leptin-deficient mice exhibited a constitutive down-regulation of UCP2 in macrophages. Kizaki et al. (35) also described a down-regulation of UCP2 mRNA and protein in macrophages 6 h after LPS treatment.
Both groups suggested that this should result in an increase in reactive oxygen species, triggering the transduction cascade of inducible nitric oxide synthase expression and the production of cytokines and prostanoids. Consistent with this proposal, Chen et al. (36) described a 15-fold decrease of UCP2 mRNA in bone marrow-derived dentritic cells after LPS treatment and a 9-fold increase in the expression of inducible nitric oxide synthase. In addition to these in vitro studies, bone marrow transplantation revealed an increase of UCP2 protein in immune cells 16 h after LPS treatment. Two explanations are possible: either (i) UCP2 is inducted into immune cells, or (ii) immune cells are recruited into the organ. In the spleen, the first explanation seems more likely, whereas in the lung and duodenum, a recruitment of immune cells cannot be excluded.
Taken together, previous studies and our data support a two-step model of UCP2 regulation consisting of an early phase and a late phase response to LPS (Fig. 6). In the early phase of the LPS response, UCP2 down-regulation within macrophages allows an increased production of ROS, which leads to the activation of macrophages and to the secretion of pro-inflammatory cytokines such as TNF␣ and interleukin-1. This is consistent with the observation that Ucp2 Ϫ/Ϫ macrophages are over-activated after infection with T. gondii (4). At a later stage, 12-16 h after LPS injection, pro-inflammatory cytokines have induced a systemic oxidative stress, and there is a need to counteract the toxic effects of inflammation and over-stimulation of immune cells. Up-regulation of UCP2 expression may be seen as an adaptive response to reduce the production of ROS in immune cells in a negative feedback regulatory cycle (Fig. 6) and also in non-immune tissues such as the lung and duodenum. Finally, these data suggest the interesting possibility that UCP2 may serve as either a pro-oxidant or an anti-oxidant, depending on its mitochondrial concentration. At low levels, UCP2 participates in the activation of immune cells, enhancing their oxidative burst activity. At higher levels, UCP2 appears to serve a cytoprotective effect, guarding against an excess of oxygen free radicals. FIG. 6. Two-step model of UCP2 regulation in the LPS model. LPS decreases the expression of UCP2 in immune cells and increases the expression of inducible nitric oxide synthase and cytokines leading to septic shock. In response to this oxidative stress, all target organs and also the immune cells stimulate the expression of UCP2 to prevent an excess of free radicals. Dexamethasone and NAC decrease the cytokine production and the oxidative stress, respectively, which inhibited the induction of UCP2 by LPS.