Mycobacterium canettii Infection of Adipose Tissues

Adipose tissues were shown to host Mycobacterium tuberculosis which is persisting inside mature adipocytes. It remains unknown whether this holds true for Mycobacterium canettii, a rare representative of the M. tuberculosis complex responsible for lymphatic and pulmonary tuberculosis. Here, we infected primary murine white and brown pre-adipocytes and murine 3T3-L1 pre-adipocytes and mature adipocytes with M. canettii and M. tuberculosis as a positive control. Both mycobacteria were able to infect 18–22% of challenged primary murine pre-adipocytes; and to replicate within these cells during a 7-day experiment with the intracellular inoculums being significantly higher in brown than in white pre-adipocytes for M. canettii (p = 0.02) and M. tuberculosis (p = 0.03). Further in-vitro infection of 3T3-L1 mature adipocytes yielded 9% of infected cells by M. canettii and 17% of infected cells by M. tuberculosis (p = 0.001). Interestingly, M. canettii replicated and accumulated intra-cytosolic lipid inclusions within mature adipocytes over a 12-day experiment; while M. tuberculosis stopped replicating at day 3 post-infection. These results indicate that brown pre-adipocytes could be one of the potential targets for M. tuberculosis complex mycobacteria; and illustrate differential outcome of M. tuberculosis complex mycobacteria into adipose tissues. While white adipose tissue is an unlikely sanctuary for M. canettii, it is still an open question whether M. canettii and M. tuberculosis could persist in brown adipose tissues.


INTRODUCTION
In mammals, brown adipose tissue (BAT) and white adipose tissue (WAT) are mainly composed of mature adipocytes, pre-adipocytes and immune cells (Ouchi et al., 2011). WAT and BAT differ in cell morphology, tissue distribution and physiological functions (Gomez-Hernandez et al., 2016): WAT constitutes the main energy reserve of the organism while BAT ensures thermogenesis in hibernating animals and newborns (Cannon and Nedergaard, 2004;Gomez-Hernandez et al., 2016) and remains metabolically active in adults (Nedergaard et al., 2007).
Adipose tissues have been previously supposed to act as long-term sanctuaries sheltering Mycobacterium tuberculosis, the major agent of human tuberculosis worldwide (Neyrolles et al., 2006;Kim et al., 2011;Agarwal et al., 2014Agarwal et al., , 2016Rastogi et al., 2016). In vitro experiments were conducted using murine 3T3-L1 mature adipocytes (Neyrolles et al., 2006;Kim et al., 2011;Rastogi et al., 2016), and 3T3-L1 pre-adipocytes (Neyrolles et al., 2006), but also with murine primary pre-adipocytes and mature adipocytes obtained from WAT (Agarwal et al., 2014). In immune-competent mice, intravenously and intra-nasally inoculated M. tuberculosis disseminates in WAT of visceral, subcutaneous, peri-renal and mesenteric adipose depots (Agarwal et al., 2014. In these models, the burden of M. tuberculosis in WAT depots stagnated or decreased with time (Agarwal et al., 2014). Indeed, compelling data show that M. tuberculosis adopts in mature adipocytes and WAT a non-replicating, dormant state characterized by an arrest of multiplication, accumulation of intra-cytosolic lipid inclusions (ILIs), resistance to anti-mycobacterial drugs (Neyrolles et al., 2006;Agarwal et al., 2014) and up-regulation of dosR and icl genes (Rastogi et al., 2016). The infection of mature adipocytes was therefore proposed as a suitable model to study the accumulation of neutral lipids into ILIs within mycobacteria (Santucci et al., 2016).
Among the M. tuberculosis complex (MTBC), Mycobacterium canettii is a peculiar member specifically diagnosed in dozens of tuberculosis patients with reported contacts to the Horn of Africa (Aboubaker Osman et al., 2015). M. canettii differs from the other members of the MTBC by processing a larger 4.48 ± 0.05 Mb mosaic genome and producing cordless and smoothlooking mycobacteria (Gutierrez et al., 2005;Koeck et al., 2011;Supply et al., 2013;Boritsch et al., 2016a) with an ability of intra-species horizontal gene transfer (Boritsch et al., 2016b). M. canettii infection mainly presents as lymph node and pulmonary tuberculosis (Koeck et al., 2011;Aboubaker Osman et al., 2015). However, M. canettii pulmonary tuberculosis is unique in being seemingly non-contagious (Koeck et al., 2011).
In order to further examine the in-vitro interactions of M. canettii with natural adipose tissues, we developed in-vitro experimental models using primary murine white and brown pre-adipocytes; and further investigated the interactions between M. canettii and murine 3T3-L1 pre-adipocytes and mature adipocytes using M. tuberculosis H37Rv as a positive control.  (Alibaud et al., 2011). Fluorescent strains were cultured in Middlebrook 7H10 broth supplemented with 10% OADC and 50 µg/mL kanamycin (Sigma). Prior to infection, mycobacteria were resuspended in phosphate buffered saline (PBS), shaken on a vortex mixer for 10 min with glass ball to disperse clumps and centrifuged for 1 min at 300 g to remove residual clumps. The supernatant was then dispersed by expelling the suspension 10 times through a sterile 25-gauge needle attached to a 1-mL syringe. Calibration was then performed according to Mcfarland standard confirmed by counting mycobacteria after Ziehl-Neelsen staining (RAL diagnostics, Martillac, France). All experiments using these mycobacteria were performed in a biosafety level 3 laboratory of the Faculté de Médecine, Aix-Marseille Université, France.

Primary Murine Pre-adipocyte Culture
The experiments conducted on mice was approved by the Institutional Animal Care and Use Committee of Aix-Marseille University "C2EA-14, " France and registered by the "Ministère de l'Enseignement Supérieur et de la Recherche" under reference n • 2015092415474605. Mice were handled according to the rules of Décret N • 2013-118, Février 7, 2013, France. Inguinal WAT and inter-scapular BAT were collected from ten 6-week-old Balb/cByj mice. Pre-adipocytes cells were isolated from tissues as described elsewhere (Aune et al., 2013) and then cultured in complete culture medium Dulbecco's Modified Eagle Medium DMEM/F12 (Invitrogen, France) supplemented with 10% heat-inactivated fetal bovine serum (FBS) and were plated at a concentration of 5 × 10 4 cells/ml/well in 12-well plates and incubated at 37 • C under a 5% CO 2 atmosphere.

3T3-L1 Cell Lines
Murine embryonic fibroblasts 3T3-L1 (ATCC, lot 62158491) were cultured in DMEM (Invitrogen, France) supplemented with 10% heat-inactivated FBS at 37 • C under a 5% CO 2 atmosphere. Prior to confluence, cells were plated at 5 × 10 4 cells/well in a 12-well sterile plates and incubated as described above. Six days after plating, pre-adipocytes reach confluency and stop dividing. Adipocyte differentiation was initiated upon confluency by adding 1 µg/mL insulin (Sigma), 0.5 mM isobuthyl-methylxanthine (Sigma), and 1 µM dexamethasone (sigma) into the culture medium. After 48 h of induction, the medium was replaced by DMEM-10% FBS containing 1 µg/mL of insulin and maintained each 2 days in order to achieve a full differentiation as previously described (Neyrolles et al., 2006). Adipocyte differentiation was checked by BODIPY 493/503 staining (Sigma) of lipid droplets over the 2-week experiments.

Infection
Pre-adipocytes and mature adipocytes were infected with a multiplicity of infection (MOI) of 1 per cell at day 6 after plating and at day 10 post-induction respectively. Accordingly, 1 × 10 5 cells per well were infected with 1 × 10 5 mycobacteria. After 4-h incubation, cells were washed 3 times with serumfree DMEM to remove all extracellular bacteria. Only mature adipocytes were then incubated with 200 µg/mL amikacin for 2 h and then washed thrice. The last wash was plated to validate the correct elimination of extracellular mycobacteria. The inoculated cultures were incubated with fresh complete medium up to 7 days for primary cultures and 12 days for 3T3-L1 cell lines at 37 • C under a 5% CO 2 atmosphere. At various time points, infected cells were lysed in PBS containing 0.1% Triton X-100 and the lysate was plated at 10-fold dilutions from 10 −1 to 10 −4 onto Middlebrook 7H10 agar plates incubated at 37 • C under a 5% CO 2 atmosphere. The number of colonies enumerated after 20-day of incubation was used to estimate the number of colony-forming units (CFUs) per 10 5 cells. All experiments were conducted in triplicate: three independent plates the same day for primary cell infection experiments and three replicates at different days with different bacterial cultures for 3T3-L1 cells infection experiments.

Intracellular Localization of Mycobacteria
Ziehl-Neelsen staining (RAL diagnostics, Martillac, France) was performed on infected cells (MOI 1:1 and 5:1) at day 3 p.i. Observations were acquired using the slide scanner Axio Scan.Z1 (Zeiss) (Magnification 20X) with a color Hitachi tri-CCD (1,800 × 1,200 pixels) camera. The number of intracellular mycobacteria per one cell was recorded on image acquisition by ImageJ software.
M. canettii mCherry was used to localize intracellular bacteria using confocal microscopy. Infected adipocytes (MOI 10) were fixed using 4% formaldehyde at day 3 p.i. Lipid droplets were stained with BODIPY 493/503 (Sigma). The images were acquired using a confocal microscope (Zeiss, Spinning disk with a Yokogawa head and an emCCD camera 512 × 512 pixels) at magnification 63X/1.4 oil objective.

Processing for Electron Microscopy
Cells were fixed at room temperature with 2.5% glutaraldehyde in Na-cacodylate buffer 0.1 M (pH 7.2) containing 0.1 M sucrose, 5 mM CaCl 2 , and MgCl 2 5 mM, washed with complete cacodylate buffer and postfixed for 1 h at room temperature with 1% osmium tetroxide in the same buffer without sucrose (de Chastellier, 2008). They were washed with buffer, scratched gently, concentrated in agar up to 2% with cacodylate buffer and processed for 1 h at room temperature with 1% uranyl acetate in maleate buffer. The samples were dehydrated in a graded series of ethanol solutions and gradually incorporated in Spurr resin. Thin sections (80 nm thick) were stained with 1% uranyl acetate in distilled water and then with lead citrate before being observed by electron microscopy.

Nile Red Staining
Adipocytes were infected with M. canettii at MOI 5:1 on 12well plate. At day 7 p.i. all wells were lysed in 500 µL water containing 0.1% Triton X-100. The pooled lysate was sonicated and centrifuged at 9,000 g for 25 min. The M. canettii cells were washed thrice with 0.1% Triton X-100 and the pellet was resuspended in 500 µL of PBS containing 0.05% Tween 80. Nile red staining was performed as previously described (Christensen et al., 1999). The stained smears were subjected to high resolution confocal (Zeiss AiryScan head, magnification 63X, Numerical aperture = 1.4), with higher resolution (1.6X) than conventional confocal.

Lactate Dehydrogenase (LDH) Assay
LDH, a soluble cytoplasmic enzyme released into extracellular space when the cell membrane is damaged, was used as an estimator of cell lysis (Korzeniewski and Callewaert, 1983). Culture supernatants of uninfected or infected mature adipocytes with M. canettii or M. tuberculosis (MOI 1) were collected at day 3, 7, and 12 post-inoculation. The release of LDH from negative controls and infected cells was measured spectrophotometrically at 340 nm on a Cobas 8000 (Roche, Meylan, France).

Statistical Analyses
Statistics were performed using the SigmaPlot13 software. The distribution of the variables was assessed statistically with the Shapiro-Wilk test. The equality of the variance was checked using the Brown-Forsythe test. All data were normally distributed and were expressed using means ± standard deviation. The statistical significance was performed using the Student T-test. The Chisquare test was used to compare rates and proportions with Yates correction. A p < 0.05 was regarded as significant.

M. canettii and M. tuberculosis within Primary Murine Pre-adipocytes
Pre-adipocytes purified from murine inguinal WAT and interscapular BAT were ex vivo inoculated with M. canettii or M. tuberculosis for 4 h p.i. For all experiments, extracellular mycobacteria were removed by extensive washing with serumfree medium, 4 h p.i. At day 3 p.i. Ziehl-Neelsen staining was performed to assess the ability of mycobacteria to infect primary murine pre-adipocytes. While negative control cells remained free of detectable mycobacteria, microscopic observations clearly showed intracellular M. canettii and M. tuberculosis in white and brown pre-adipocytes (Figure 1).
The ratio of infected cells relative to the total number of observed cells as well as the average number of intracellular bacilli per one cell were measured on a random selection of 10 microscope fields on stained infected cells at day 3 p.i. (see Supplementary Data S1). White pre-adipocytes infected at MOI 1:1 yielded 19% and 18% of infected cells with 9 ± 6 M. canettii bacilli per cell and 7 ± 5 M. tuberculosis organisms per cell, respectively. Likewise, 22 and 19% of brown preadipocytes were infected with 16 ± 7 M. canettii organisms/cell and 11 ± 4 M. tuberculosis organisms/cell, respectively. The ratio of infection did not significantly differ between brown and white pre-adipocytes (p = 0.7 for M. canettii and p = 0.8 for M. tuberculosis, χ 2 test) but the number of bacilli in brown pre-adipocytes was significantly higher than in white pre-adipocytes (p = 0.02 for M. canettii and p = 0.03 for M. tuberculosis).
A similar intracellular behavior was observed for M. canettii and M. tuberculosis during the 7-day experiment in primary white and brown pre-adipocytes: both mycobacteria survived and multiplied with approximately one log gain.

M. canettii Is Internalized into 3T3-L1 Pre-adipocytes and Mature Adipocytes
To explore M. canettii interactions with 3T3-L1 pre-adipocytes and mature adipocytes compare to M. tuberculosis (Neyrolles et al., 2006;Kim et al., 2011), the quantification of the ratio of infected cells/total cells and the number of intracellular bacilli/cell were performed on a random selection of 10 microscope fields after Ziehl-Neelsen staining of infected cells at day 3 p.i. (see Supplementary Data S1). For pre-adipocytes, 20% of cells were infected with 7 ± 5 M. canettii organisms/cell while 40% of cells were infected with 12 ± 6 M. tuberculosis organisms/cell which is significantly higher compared to M. canettii infection (p < 0.001, χ 2 test). Under the same conditions, only 9% and 17% of adipocytes were infected with M. canettii and M. tuberculosis, respectively. The ratio of infected cells was significantly higher for pre-adipocytes compared to mature adipocytes (p = 0.002 for M. canettii and p < 0.001 for M. tuberculosis, χ 2 test).
To confirm the intracellular location of M. canettii within 3T3-L1 mature adipocytes, cells were inoculated with M. canettii CIPT 140010059 mCherry (CSURP3621) at MOI = 10. Three days p.i. infected cells were treated with BODIPY 493/503 stain and fixed. Confocal microscopy observations showed the presence of 15 ± 6 intracellular M. canettii inside adipocytes, as compared to negative controls which remained free of detectable mCherryfluorescence ( Figure 3A). We noticed that approximately 30% of intracellular M. canettii organisms were localized in close contacts with cytoplasmic lipid bodies (LBs) in 100% of the observed adipocytes (Figure 3A white squares).
Further electron microscopy observation of adipocytes infected at a MOI 5:1 for 3 days demonstrated the presence of M. canettii within adipocytes in close contact with adipocyte LBs, corroborating observations performed by confocal microscopy (Figure 3B, upper panel). Intracellular M. canettii bacilli were loaded with small electron-dense vesicles resembling ILIs ( Figure 3B, upper panel). We also found that some preadipocytes that failed to reach full differentiation were infected by M. canettii which had weakly accumulated ILIs (Figure 3B, lower panel).

Intracellular Outcome of Internalized M. canettii
Using the 3T3-L1 continuous cell line which is more robust and easier to manipulate than primary cells, it was possible to compare the successive steps of infection over 12 days in three independent experiments conducted at different days, comparing pre-adipocytes and mature adipocytes. After thorough dispersal of the mycobacterial aggregate, cells were infected with M. tuberculosis, used as positive control, or M. canettii at a MOI 1:1 for 4 h followed by extensive elimination of extracellular mycobacteria. All results were compared with non-infected cells manipulated in parallel, which remained negative in culture during the 12-day experiment. At various time points 0, 3, 7, and 12 days p.i. cells were lysed and viable bacterial content was measured by plating cell lysates onto Middelbrook agar 7H10 and scoring CFUs/10 5 cells (Figure 4) (see Supplementary Data S1).
As for intracellular behavior, both mycobacteria continuously replicated into pre-adipocytes during the 12-day experiment and gained two logs between day 0 and day 12 (Figure 4). In mature adipocytes, M. tuberculosis and M. canettii showed two different kinetics: intracellular M. tuberculosis inoculum stabilized between day 3 and 12 p.i. (4 ± 3.66 CFUs gain), whereas M. canettii replicated with a 2 × 10 1 ± 2.5 × 10 1 CFUs gain between day 3 and 12 p.i. which is significantly higher than M. tuberculosis (p = 0.01) (Figure 4) but this load remains lower than in pre-adipocytes (Figure 4).
The arrest of M. tuberculosis's replication inside adipocytes has been correlated with a heavy accumulation of ILIs leading to a dormant state (Neyrolles et al., 2006;Kim et al., 2011;Agarwal et al., 2014). To investigate whether this holds true for M. canettii, intra-adipocyte bacilli were extracted at day 7 p.i. after Triton X-100 cell lysis and sonication. While Middlebrook 7H10 agar-grown M. canettii were free of detectable ILIs, Nilered staining showed that intracellular M. canettii yielded ILIs (Figure 5). Intracellular M. canettii presented variable numbers of small ILIs corresponding to 60% ILI +1 and 40% ILI +2 profiles which were previously described (Caire-Brandli et al., 2014).

Intracellular M. canettii Induce Mature Adipocyte Cytolysis
Microscopic examination indicated that M. canettii infection yielded no cytolysis on mature 3T3-L1 adipocytes up to day 7 p.i. but cellular debris were observed at day 12 p.i. while a smaller amount of cellular debris was observed in uninfected control cells. Microscopic observation of M. tuberculosis-infected cells clearly showed cellular debris at day 7 p.i. A cytolytic effect of mycobacteria was further investigated by measuring the lactatedehydrogenase (LDH) in the supernatant of infected and noninfected control adipocytes (see Supplementary Data S1). We observed a progressive increase of the LDH concentration in the three culture supernatants. However, the increase of LDH concentration was significantly higher in M. canetti-infected adipocytes than in negative control adipocytes at day 3 p.i. (p = 0.002) and day 12 p.i. (p < 0.001) (Figure 6). In M. tuberculosis-infected cells, high rates of LDH concentration measured throughout the 12-day experiment were significantly higher compared to uninfected cells and M. canettii infected cells at each time point (Figure 6). Positive and high correlation was noted between the rates of LDH release and the count of CFUs in the case of M. canettii and M. tuberculosis infection (r = 0.90 for M. canettii and r = 0.88 for M. tuberculosis).

DISCUSSION
Data here reported indicate that M. canettii is able to infect, survive and replicate in primary murine white and brown preadipocytes which are constantly detected in adipose tissues Adipocytes were infected at MOI 5:1 with M. canettii for 4 h. At day 3 post-inoculation, the supernatant was removed and infected cells were fixed for electron microscopy. M. canettii organisms were inside adipocytes in close contact with adipocytes LBs as indicated by the red arrow and accumulated ILIs (upper panel). Pre-adipocytes that failed to reach full differentiation were infected also with M. canettii (lower panel). (Cawthorn et al., 2012) and 3T3-L1 pre-adipocytes and mature adipocytes after in-vitro and ex-vivo experimental infection. Differentiation of 3T3-L1 cells was induced so as to obtain in vitro model of mature white adipocytes (Morrison and McGee, 2015). Data were authenticated by their reproducibility over a triplicate experiment in the presence of negative controls. All experiments with M. tuberculosis, here used as a positive control, were in agreement with those previously reported (Neyrolles et al., 2006;Kim et al., 2011;Agarwal et al., 2014). We extended for the first time these observations to brown pre-adipocytes.
The primary and 3T3-L1 pre-adipocytes also phagocytized M. canettii and M. tuberculosis as previously shown with other particles (Cousin et al., 1999). More precisely, the amount of intracellular mycobacteria was significantly higher in brown than in white pre-adipocytes, corroborating previous observation of the earlier deposition of M. bovis in the BAT after intravenous administration in mice (Mauss and Levy, 1972). In vivo, the rich vascular network of BAT (Shimizu et al., 2014) could facilitate the tissue infection by M. canettii. Furthermore, the burden of intracellular mycobacteria was higher in pre-adipocytes than in mature adipocytes. Pre-adipocytes are fibroblastlike cells lacking cytoplasmic lipid droplets thus capable of rapidly phagocytosing (Hoffman and Dow, 2016); while mature adipocytes have lost this phagocytic activity (Cousin et al.,FIGURE 6 | Quantitative analysis of LDH release. Culture supernatants of infected adipocytes were collected at day 3, 7, and 12 post-inoculation. LDH concentration was assessed in uninfected adipocytes (blue), M. canettii infected cells (orange), and M. tuberculosis H37Rv infected cells (gray). Data shown are the mean + standard deviation of three independent measurements. 1999). Likewise, foamy macrophages loaded with LBs lost the ability to phagocytize mycobacteria compared to undifferentiated non-foamy macrophages (Peyron et al., 2008). These data indicate that pre-adipocytes are preferential targets for MTBC mycobacteria.
In conclusion, our results offer new pieces of information regarding the subtle interplay between MTBC mycobacteria and their hosts: infection of pre-adipocytes and adipocytes is a probable common feature of the MTBC at large, being shared by distantly related M. tuberculosis and M. canettii (Gutierrez et al., 2005), nevertheless, the intracellular outcome depends on the mycobacteria. While WAT is an unlikely sanctuary for M. canettii, the question remains open as for BAT. In the current quest for a yet unknown animal reservoir for M. canettii, our observations suggest that any mammal harboring BAT is a candidate reservoir, suggesting new field of investigations.

AUTHOR CONTRIBUTIONS
SC and MD, conceived and supervised the experiments; Fé.B, performed the experiments; Fa.B, conducted the mice experiment; IP, conducted the electron microscopic observations; PW, was helpful for all confocal observations; SC and MD, and Fé.B analyzed the data and drafted the manuscript. All authors reviewed and approved the manuscript.

FUNDING
This study was financially supported by URMITE, IHU Méditerranée Infection, Marseille, France; CNRS and by the A * MIDEX project (n • ANR-11-IDEX-0001-02) funded by the "Investissements d'Avenir" French Government program, managed by the French National Research Agency (ANR).