Insulin resistance and type 2 diabetes are associated with low HDL-cholesterol (HDL-c) levels, which would impair reverse cholesterol transport (RCT). A promising therapeutic strategy is to raise HDL with cholesteryl ester transfer protein (CETP) inhibitors, but their effects on RCT remains to be demonstrated in vivo. We therefore evaluated the effects of CETP inhibitor torcetrapib in CETP-apolipoprotein (apo)B100 mice made obese and insulin resistant with a 60% high-fat diet. High-fat diet over 3 months increased body weight and homeostasis model of insulin resistance index by 30% and 846%, respectively (p < 0.01 for both vs. chow-fed mice). Total cholesterol (TC) increased by 46% and HDL-c/TC ratio decreased by 28% (both p < 0.05). Compared to vehicle, high-fat-fed mice treated with torcetrapib (30 mg/kg/day, 3 weeks) showed increased HDL-c levels and HDL-c/TC ratio by 41% and 37% (both p < 0.05). Torcetrapib increased in vitro macrophage cholesterol efflux by 22% and in vivo RCT through a 118% increase in 3H-bile acids fecal excretion after 3H-cholesterol labeled macrophage injection (p < 0.01 for both). Fecal total bile acids mass was also increased by 158% (p < 0.001). In conclusion, CETP inhibition by torcetrapib improves RCT in CETP-apoB100 mice. These results emphasize the potential of CETP inhibition to prevent cardiovascular diseases. Clin Trans Sci 2011; Volume 4: 414–420
Dyslipidemia is a major characteristic of insulin-resistant and type 2 diabetic patients and contributes to the increased cardiovascular risk in these individuals. Although LDL-cholesterol therapies are known to reduce the risk of cardiovascular events,1 additional therapies are required to further prevent cardiovascular diseases. As HDL-cholesterol (HDL-c) levels are inversely correlated with cardiovascular risk2 and are known to be lower in diabetic dyslipidemia, a considerable interest has focused on the development of novel therapies that raises HDL-c.3
The cardioprotective effect of HDL-c would be related, at least in part, to its major role in macrophage-to-feces reverse cholesterol transport (RCT). This physiological process mediates the transport of excess cholesterol by HDL from peripheral tissues back to the liver for uptake of cholesteryl esters by the scavenger receptor class B type I (SR-BI). In the liver, cholesteryl esters are then metabolized as free cholesterol or bile acids for further excretion into the bile and ultimately in the feces.4 In humans, another RCT pathway is represented by the cholesteryl ester transfer protein (CETP), which transfers cholesteryl esters from HDL to apolipoprotein (apo) B-containing lipoproteins VLDL/LDL for further hepatic uptake by the LDL-receptor. Increased CETP activity is thought to drive the lowering of HDL-c levels in diabetic dyslipidemia.5 Hence, dyslipidemia would impair RCT and inhibition of CETP thus represents a relevant therapeutic strategy to raise HDL-c levels and potentially improve RCT. However, the CETP pathway is thought to be a major route for human RCT,6 since it represents approximately 80% of the plasma cholesteryl esters output.7 Hence, one of the major concerns with the development of CETP inhibitors is whether inhibiting the CETP pathway would alter the rate of RCT in vivo.
As no clinical data are available regarding the effects of CETP inhibition on human RCT, the use of animal models expressing CETP represents a relevant strategy to evaluate the effects of CETP inhibitors. Although the molecular mechanisms of RCT have been extensively investigated in mice, substantial differences exist between mouse and human RCT (e.g., mouse does not express CETP). However, introducing the expression of human CETP in mouse has been shown to induce a more human-like profile.8,9 In this line, recent studies in transgenic mice that express both human apoB100 and human CETP (CETP-apoB100 mice) have demonstrated the relevance of this animal model to evaluate the effects of compounds raising HDL-c, such as niacin10 and CETP inhibitor torcetrapib.11 On another note, mouse fed a 60% high-fat diet has been described as a model of obesity and type 2 diabetes.12
Hence, to investigate the effects of CETP inhibition on RCT in the face of insulin resistance and dyslipidemia, we used CETP-apoB100 mice fed a 60% high-fat diet over 3 months and evaluated the effects of the CETP inhibitor torcetrapib in this animal model.
Animals and diet
All animal protocols were approved by the local ethical committee (Comité regional d’éthique de Midi-Pyrénées). Seven-week-old male human CETP/human apoB100 (CETP-apoB100) transgenic mice (Taconic, Lille Skensved, Denmark) were housed in plastic cages (seven animals/cage) containing wood shavings and maintained in a room with a 12-hour light cycle with free access to food and water. Animals were adapted to these conditions and fed a rodent chow diet (Purina chow 5001, Research Diets, New Brunswick, NJ, USA) for 1 week. This rodent chow diet was defined as the control diet. Mice were then fed the control diet (n= 7) or a 60% high-fat diet (n= 14) pelleted by Research Diets (diet D12492) over 3 months.
After 3 months of diet, mice were food deprived for 3 hours at 08:00 am and blood (25 μL/EDTA) was collected from the tail tip to measure blood glucose and plasma total cholesterol. High-fat-fed mice were randomized to create two homogenous groups with the same mean values for total cholesterol (1.78 g/L) and blood glucose (188 mg/dL). Mice were then treated for 3 weeks with either vehicle (1% DMSO/99% of a saline solution of 0.5% hydroxypropyl methylcellulose/1% Tween 80) or torcetrapib 30 mg/kg, once daily. To evaluate the effects of high-fat diet on biochemical parameters, chow-fed mice were also treated with vehicle over 3 weeks. After 3 weeks of treatment, mice were food deprived for 3 hours at 08:00 am and blood (50 μL/EDTA) was collected from the tail tip and a pool of plasma (1 pool/group) was kept aside for fast protein liquid chromatography (FPLC) profiles and macrophage-to-plasma cholesterol efflux. Mice were then fasted overnight and blood (50 μL/EDTA) was collected from the tail tip for biochemical analysis.
To evaluate the effects of CETP inhibition on RCT, mice fed the high-fat diet recovered over 2 days after blood collection and then underwent the in vivo macrophage-to-feces RCT experiment, as described below.
In vitro macrophage-to-plasma cholesterol efflux
J774 macrophages were seeded in triplicates in 24-well plates (1.5 × 106 cells/well) and incubated for 24 hours in complete RPMI medium. Oxidized LDL at a final concentration of 50 μg/mL were incubated over 0.5 hour at 37°C with 1 μCi/mL 3H-cholesterol in RPMI. Macrophages were then 3H-cholesterol labeled/cholesterol-loaded for 24 hours using this labeling medium. After the 24-hour incubation, cells were washed and equilibrated for 4 hours with RPMI medium supplemented with 0.2% BSA. Cellular cholesterol efflux was then stimulated by adding serum-free medium containing 2.5% of pooled plasma from each group of mice (vehicle or torcetrapib-treated mice). After 4 hours, medium was collected and cell lipids were subsequently extracted with 250 μL isopropyl alcohol. For each well, 3H-cholesterol released in the medium and present in the total cell lipid extract was measured by liquid scintillation counting. Background cholesterol efflux (determined in the absence of plasma) was subtracted. The percentage of cholesterol efflux was calculated as the amount of 3H-cholesterol recovered in the medium divided by the total label (cells + medium) in the wells multiplied by 100. Data presented are the mean of two experiments, which were performed in triplicate.
In vivo macrophage-to-feces RCT
Preparation of J774 cells and in vivo RCT study were performed as previously described.13 J774 cells obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA), were grown in suspension in RPMI/HEPES supplemented with 10% FBS and 0.5% gentamicin in suspension in Nalgene Teflon flasks. Cells were radiolabeled with 5 μCi/mL 3H-cholesterol and cholesterol loaded with 50 μg/mL oxidized LDL over 48 hours. Radiolabeled cells were then washed with RPMI/HEPES and equilibrated for 4 hours in fresh RPMI/HEPES supplemented with 0.2% BSA and gentamicin. Cells were pelletted by low-speed centrifugation and resuspended in MEM/HEPES prior to injection into mice.
The day of the experiment, mice were weighed and placed in individual cages. 3H-cholesterol-labeled and oxidized LDL-loaded J774 cells (2.5 × 106 cells containing 10 × 106 dpm in 0.5 mL minimum essential medium) were then injected intraperitoneally. High-fat-fed mice had free access to food and water, and were treated daily with vehicle (n= 7) or torcetrapib 30 mg/kg twice daily (n= 7) over 48 hours. Blood (50 μL/EDTA) was collected from the tail tip at 24 and 48 hours to measure radioactivity released into the plasma (10 μL counted in a liquid scintillation counter). Radioactivity in HDL at 48 hours after injection was also measured after phosphotungstate/MgCl2 precipitation. After 48 hours, mice were weighed, sacrificed by cervical dislocation, exsanguinated and liver was harvested from each animal and weighed. Samples of liver were then flash-frozen in liquid nitrogen and then stored at −80°C prior to western blot analysis. An approximately 50-mg piece of liver was also kept and homogenized using an ultrasounds probe in 500 μL water. The liver homogenate was then used for hepatic 3H-cholesterol and 3H-bile acids extraction as described for the feces analysis below.
Feces were collected over 48 hours and then stored at 4°C before extraction of cholesterol and bile acids. Fecal cholesterol and bile acids extraction was performed as previously described.13 The total feces collected from 0 to 72 hours were weighed and soaked in Millipore water (1 mL water per 100 mg feces) overnight at 4°C. The following day, an equal volume of absolute ethanol was added, and the mixtures were homogenized. To extract the 3H-cholesterol and 3H-bile acid fractions, 1 mL of the homogenized samples was combined with 1 mL ethanol and 200 μL NaOH. The samples were saponified at 95°C for 1 hour and cooled to room temperature, and then 3H-cholesterol was extracted two times with 3 mL hexane. The extracts were pooled, evaporated, resuspended in 500 μL toluene, and then 200 μL were counted in a liquid scintillation counter. To extract 3H-bile acids, the remaining aqueous portion of the feces was acidified with concentrated HCl and then extracted two times with 3 mL ethyl acetate. The extracts were pooled together, evaporated, resuspended in 500 μL ethyl acetate, and 200 μL were counted in a liquid scintillation counter. For both cholesterol and bile acids extract, the remaining volume was evaporated and resuspended in ethanol and commercial kits were used to determine fecal total cholesterol (Biomerieux, Marcy l’Etoile, France) and total bile acids (Bioquant, San Diego, CA, USA) mass.
Results were expressed as a percentage of the radioactivity injected recovered in plasma, liver, and feces. The plasma volume was estimated as 3.5% of the body weight.
Blood glucose was monitored using a glucometer (Roche Diagnostics, Meylan, France). Plasma insulin was determined with a commercial kit (Mercodia, Uppsala, Sweden). Total cholesterol and triglycerides were assayed using commercial kits (Biomerieux, Marcy l’Etoile, France). HDL-c was determined using the phosphotungstate/MgCl2 precipitation method as previously described.13 Non-HDL-c and HDL-c/total cholesterol ratio were calculated from HDL-c and total cholesterol values. Plasma CETP activity was measured by fluorescence using a commercial kit (Roarbiomedical, New York, NY, USA). FPLC lipoprotein profiles (total cholesterol) using pooled plasma from vehicle- or torcetrapib-treated mice were analyzed, as previously described (Briand et al., 2010). FPLC fractions corresponding to LDL and HDL were pooled into three (LDL) or four (HDL) aliquots. ApoB100, A-I, and E composition was then evaluated with coomassie staining after resolving LDL on a 3–8% Tris-Acetate gel and HDL on a Nu-PAGE 4–12% Bis-Tris gels (Invitrogen, Carlsbad, CA, USA) under reducing conditions.
Western blot analysis
Western blot analysis was performed on liver from four individuals of each treatment group (vehicle or torcetrapib). Fifty microgram of total protein were resolved by 3–8% Tris-Acetate or 4–12% Bis-Tris gels (Invitrogen) and transferred to nitrocellulose membrane (Invitrogen). Commercial antibodies were used to detect the following proteins: SR-BI and ATP-binding cassette A1 (Novus, Littleton, CO, USA) and LDL-receptor (Biovision, Mountain View, CA, USA). Antibody for βactin (Abcam, Cambridge, MA, USA) was used as a loading control. Blots were visualized with the ECL chemiluminescent detection system (GE Healthcare, Piscataway, NJ, USA).
Data are presented as mean ± SEM. An unpaired Student’s t-test was used to compare chow-fed versus high-fat-fed mice or vehicle- versus torcetrapib-treated mice. A p < 0.05 was considered significant.
High-fat diet induces obesity, insulin resistance, and dyslipidemia in CETP-apoB100 mice
The effects of the 60% high-fat diet on body weight and biochemical parameters are shown in Table 1. As expected, mice fed the high-fat diet showed a 29% higher body weight (p < 0.001 vs. chow-fed mice). Blood glucose and plasma insulin levels were significantly increased by 37% and 551%, respectively. This induced an 846% increase of the HOMA-IR index (p < 0.001). Total cholesterol levels and non-HDL-c levels were significantly increased by 45% and 70% in mice fed the high-fat diet (p < 0.01 vs. chow-fed mice). This resulted in a 28% decrease of the HDL-c/total ratio (p < 0.05 vs. chow). Moreover, CETP activity was increased by 39% in mice fed the high-fat diet (p < 0.01 vs. chow).
Table 1. Effects of high-fat diet and torcetrapib in CETP-apoB100 mice (mean ± SEM, n= 7 per group).
Chow + vehicle
High fat + vehicle
High fat + torcetrapib
*p < 0.05, **p < 0.01, ***p < 0.001 chow + vehicle versus high fat + vehicle; †p < 0.05, ††p < 0.01, †††p < 0.001 high fat 1 vehicle versus high fat 1 torcetrapib.
Body weight (g)
33.2 ± 1.0
43.0 ± 2.0***
42.4 ± 1.3
Blood glucose (mg/dL)
142.8 ± 5.6
196.4 ± 10.7**
176.8 ± 4.3
Plasma insulin (μU/mL)
32.0 ± 4.4
208.5 ± 16.6***
101.2 ± 24.7††
11.3 ± 2.2
106.6 ± 10.4***
44.5 ± 11.3††
Total cholesterol (g/L)
1.01 ± 0.06
1.47 ± 0.14**
1.30 ± 0.06
0.39 ± 0.04
0.43 ± 0.05
0.61 ± 0.04††
0.61 ± 0.08
1.04 ± 0.10**
0.69 ± 0.04††
HDL-c/total cholesterol ratio
0.40 ± 0.05
0.29 ± 0.01*
0.47 ± 0.02†††
0.79 ± 0.10
0.80 ± 0.07
0.70 ± 0.06
CETP activity (pmol/h/μL)
43.3 ± 4.0
60.2 ± 3.3**
36.6 ± 1.6†††
The overall data indicate that CETP-apoB100 mice fed the high-fat diet are obese, insulin resistant, and dyslipidemic.
Torcetrapib improves dyslipidemia in CETP-apoB100 mice under high-fat diet
The effects of CETP inhibition by torcetrapib were measured in mice fed a high-fat diet (see Table 1). After 3 weeks of treatment, torcetrapib decreased blood glucose by 10%, although not significantly. Unexpectedly, plasma insulin levels were reduced by 51% (p < 0.01 vs. vehicle), which led to a 58% reduction of the HOMA-IR index (p < 0.01 vs. vehicle). CETP inhibition with torcetrapib significantly increased HDL-c levels by 42% (p < 0.01 vs. vehicle) and raised HDL-c/total cholesterol ratio by 61% (p < 0.001). As expected, CETP activity was significantly reduced by 39% with torcetrapib (p < 0.001 vs. vehicle).
To further evaluate the effects of torcetrapib, total cholesterol lipoprotein profiles were measured by FPLC using pooled plasma from nonfasted mice (see Figure 1). Torcetrapib tended to reduce the increased LDL pic by the high-fat diet (Figure 1A). As expected, CETP inhibition by torcetrapib increased the levels of HDL and led to enlarged particles. Apo’s composition was evaluated by coomassie staining of FPLC fractions corresponding to LDL and HDL (see Figure 1B). Expectedly, high-fat diet tended to increase LDL-apoB100 bands, but no major effect was seen with torcetrapib. High-fat diet tended to increase the levels of apo A-I and somewhat apo E. As CETP inhibition further enlarged the HDL particles, the increase in apo A-I and also apo E was more evident in mice treated with torcetrapib.
Taken together, these data suggest that torcetrapib treatment improves dyslipidemia in obese insulin-resistant CETP-apoB100 mice.
Torcetrapib promotes macrophage-to-feces RCT in obese insulin-resistant CETP-apoB100 mice
To test whether CETP inhibition by torcetrapib promotes macrophage-to-feces RCT, 3H-cholesterol-labeled/oxLDL-cholesterol-loaded macrophages were incubated in vitro with plasma collected from vehicle- or torcetrapib-treated mice. As shown in Figure 2A, macrophage-to-plasma cholesterol efflux was increased by 22% with plasma from torcetrapib-treated mice (p < 0.01 vs. vehicle). To evaluate the effect of CETP inhibition in vivo, mice were then injected with 3H-cholesterol-labeled/oxLDL-cholesterol-loaded macrophages. As shown in Figure 2B, 3H-tracer appearance in plasma was not different at time 24 and 48 hours after injection. However, when plasma HDL was isolated at time 48 hours (Figure 2C), 3H-tracer appearance was increased by 38% in mice treated with torcetrapib (p < 0.05 vs. vehicle). As shown in Figure 2D, hepatic 3H-cholesterol recovery after 48 hours was also increased by 45% (p < 0.05 vs. vehicle), while hepatic 3H-bile acids were reduced by 55% (p < 0.05 vs. vehicle). No difference was seen in 3H-cholesterol fecal excretion (Figure 2E). However, torcetrapib treatment significantly increased fecal 3H-bile acids excretion by 118% (p < 0.01 vs. vehicle). To confirm the effects of torcetrapib on fecal excretion, the mass of fecal total cholesterol and bile acids was then measured. As shown in Figure 2F, fecal cholesterol mass remained unchanged. However, fecal total bile acids mass increased by 158% with torcetrapib (p < 0.001 vs. vehicle).
Overall, these data indicate that CETP inhibition with torcetrapib promotes macrophage-to-feces RCT through an increased fecal bile acids excretion in CETP-apoB100 mice.
Torcetrapib treatment does not affect liver ABCA1, SR-BI, and LDL-r expression
To test whether the increased rate of RCT could be related to hepatic transporter and receptors involved in lipoprotein/cholesterol metabolism, the expression of ATP-binding cassette A1, SR-BI, and LDL-receptor was evaluated by western blot. As shown in Figure 3, no change was observed with torcetrapib treatment, suggesting that improvement of RCT by CETP inhibition might be more related to its effect on plasma lipoprotein metabolism.
Our data demonstrate that CETP inhibition with torcetrapib promotes macrophage-to-feces RCT in obese insulin-resistant CETP-apoB100 mice. One of the major concern that raised with the development of CETP inhibitors is whether the enlargement of HDL (due to cholesteryl esters enrichment) are functional particles to promote RCT.14,15 In the CETP-apoB100 mouse model, CETP inhibition with torcetrapib also resulted in the enlargement of HDL particles (see Figure 1), which contained higher levels of both apo A-I and apo E. As in vitro macrophage cholesterol efflux was significantly increased with plasma from torcetrapib-treated mice, these changes in apo composition might have beneficial effects on the HDL quality and capacity to promote cholesterol efflux. Since apo A-I-containing particles are efficient in removing cholesterol from macrophages,16 increased levels of apo A-I may contribute to the upregulation of in vitro macrophage cholesterol efflux. Meanwhile, apo E may have a substantial role in the upregulation of macrophage-to-plasma cholesterol efflux. In another study, HDL isolated from patients treated with the CETP inhibitor anacetrapib showed an increased efflux potential, which was concomitant with higher content of apo E, but not apo A-I.17 As well, HDL from subjects with homozygous CETP deficiency18 or moderately hyperlipidemic patients treated with torcetrapib19 have increased apo E content, which enhances the ability to promote macrophage cholesterol efflux.
The assessment of in vivo RCT also indicated an increased 3H-tracer appearance in the HDL fraction. This effect could be both related to increased macrophage-cholesterol efflux and changes in HDL particles turnover. For technical reasons, we were not able to evaluate the effects of CETP inhibition on HDL catabolism (large volume of plasma from vehicle- and torcetrapib-treated mice would be required to radiolabel HDL particles prior injection into other mice). In a larger animal model, other investigators demonstrated a reduced turnover of HDL particles from rabbits treated with a CETP inhibitor, while the removal of HDL-cholesteryl esters was not compromised by CETP inhibition.20 Although this was not investigated in this study, it is possible that CETP inhibition by torcetrapib could have reduced HDL-cholesteryl ester turnover, which would contribute to increase HDL-c levels and result in higher 3H-tracer appearance in the HDL fraction.
At the hepatic level, an increased 3H-cholesterol recovery was observed after radiolabeled macrophage injection. However, no major change was seen regarding the expression of hepatic receptors SR-BI and LDL-receptor. Hence, upregulation of RCT upon CETP inhibition by torcetrapib might be more related to change in HDL composition rather than expression of hepatic receptors. Actually, since apoE is an effective ligand for LDL-receptor,21 enlarged/apo E-rich HDL itself may promote the uptake of HDL-derived cholesterol. Although this remains to be determined in the CETP-apoB100 mouse model, previous studies performed in the perfused rat liver showed that hepatic uptake of apo E-rich HDL was 10-fold higher compared to apo A-I-rich HDL.22 Moreover, in vitro and in vivo studies in mice demonstrated that hepatic uptake of HDL isolated from hyperlipidemic patients treated with torcetrapib was significantly enhanced.23
In regard with the last step of RCT, CETP inhibition by torcetrapib-stimulated macrophage-to-feces RCT through an increased bile acids excretion (Figure 2D and E). Although torcetrapib treatment was not associated with increased fecal sterol excretion in humans with low HDL-c levels,24 the increased bile acids excretion by torcetrapib has been already described by others in a hamster model.25 However, the mechanisms leading to higher bile acids excretion remain unknown. In this study, we observed a significant reduction of 3H-bile acids in the liver. Hence, it is possible that enlarged/apo E-rich particles would drive HDL-derived cholesterol toward bile acids synthesis for further biliary and fecal excretion. This hypothesis is supported by previous studies in cultured rat hepatocyte, in which stimulation of bile acid synthesis was found to be dependent on both cholesteryl esters and apo E content of the HDL particle.26
Taken together, these data suggest that the change in HDL composition upon CETP inhibition might be beneficial at multiple steps of macrophage-to-feces RCT in CETP-apoB100 mice.
Unexpectedly, another benefit of CETP inhibition with torcetrapib in the obese insulin-resistant CETP-apoB100 mice was the reduction of hyperinsulinemia and, to a lesser extent, blood glucose, which led to a significant reduction of the HOMA-IR index. In mice treated with CETP inhibitor torcetrapib, higher apo A-I content in HDL could have contributed to the improvement of insulin resistance. The effects of apo A-I on glucose homeostasis have been demonstrated in vitro27-29 as well as in vivo in mouse.27 Apo A-I stimulates phosphorylation of both AMPK and ACC and thus elevates glucose uptake in vitro,27,28 while apo A-I-deficient mice have both increased plasma glucose and insulin, and exhibit impaired glucose tolerance.27 On the contrary, apo A-I overexpression in mice appears to slightly improve insulin sensitivity.30 As the reduction of insulin resistance was unexpected in this study, further in vivo experiments would be required to definitely demonstrate a beneficial effect of CETP inhibition on glucose homeostasis, such as in vivo hyperinsulinemic euglycemic clamp and tissue glucose uptake. On another note, it should be reminded that the clinical development of the CETP inhibitor torcetrapib was halted due to off-target effects (e.g., increased blood pressure), independent of CETP inhibition.31,32 In this study, we cannot exclude that modulation of glucose metabolism could be due to torcetrapib itself. Hence, the effects of other CETP inhibitors should be also evaluated in this animal model to rule out this possibility.
Overall, our data suggest that CETP inhibition promotes RCT in a mouse model of obesity and insulin resistance with a human-like lipoprotein profile. Despite the failure of torcetrapib in the clinical setting, the development of other CETP inhibitors is still ongoing.33,34 While larger clinical trials will help to demonstrate the relevance of CETP inhibition in reducing cardiovascular risks, preclinical experiments are also needed to test whether a given CETP inhibitor has the potential to promote RCT. While CETP inhibitors, including torcetrapib, have shown some positive effects in normolipidemic animal models,25 it remains important to evaluate whether CETP inhibition is still beneficial in the face of diabetic dyslipidemia for better translation toward the clinical setting. These data suggest that the obese insulin-resistant CETP-apoB100 mouse could be a useful model in this line of investigations.
Conflict of Interest
François Briand, Quentin Thieblemont, and Thierry Sulpice are employees of Physiogenex. Agnès André and Khadija Ouguerram have no conflict of interest.
This work was supported by Oseo and the Région Midi-Pyrénées, France and the European fund Fonds Européen de Développement Régional.