Mycobacterium abscessus glycopeptidolipids inhibit macrophage apoptosis and bacterial spreading by targeting mitochondrial cyclophilin D

Mycobacterium abscessus (MAB) is a species of nontuberculous mycobacteria (NTM) and a major causative pathogen of pulmonary diseases especially in patients with cystic fibrosis. MAB infection is notoriously difficult to treat because of its intrinsic or inducible resistance to most antibiotics. The rough (R) morphotype of MAB, lacking cell surface glycopeptidolipids (GPLs), is associated with more severe and persistent infection than the smooth (S) type; however, the mechanisms underlying the R type’s virulence and the relation with GPLs remain unclear. In this study, we found that R-type MAB is much more proapoptotic than the S type, as a result of GPL-mediated inhibition of macrophage apoptosis. Polar GPLs inhibited an apoptotic response (induced by proapoptotic stimuli) by suppressing ROS production and the cytochrome c release and by preserving mitochondrial transmembrane potential. Furthermore, GPLs were found to be targeted to mitochondria and interacted with cyclophilin D; their acetylation was essential for this interaction. Finally, GPLs inhibited the intracellular growth and bacterial spreading of R-type MAB among macrophages via apoptosis inhibition. These findings suggest that GPLs limit MAB virulence by inhibiting apoptosis and the spread of bacteria and therefore provide a novel insight into the mechanism underlying virulence of MAB.

Mycobacterium abscessus (MAB) is a rapidly growing nontuberculous mycobacterium (NTM) and is a common causative agent of lung diseases especially in cystic fibrosis patients. 1,2 MAB infection is notoriously difficult to treat because of its intrinsic or inducible resistance to most antibiotics. 3,4 Although the isolation rate of NTM species such as M. avium (MAV), MAB, and M. kansasii in clinical samples is rapidly increasing, little is known about their virulence factors and their pathogenicity at the cellular level.
MAB has two morphological forms (morphotypes): smooth (S) and rough (R). The major difference between these types is the loss of surface-associated glycopeptidolipids (GPLs) in the R type. 5,6 S-type MAB appears to be predominant in the environment and is mostly associated with an initial infection, whereas the R morphotype is associated with more severe and persistent infection. 7,8 Nonetheless, it is still unclear why the R type is more virulent than the S type.
GPLs are produced by most NTM species [9][10][11] and share a common diglycosylated N-linked long fatty acyl chain linked to a tetrapeptide-amino-alcohol core but show differences in modifications consisting of attached rhamnose and 6-deoxytalose. The GPL core forms apolar or nonspecific GPLs (nsGPLs) that are produced by all NTM species, whereas polar, serotype-specific GPLs (ssGPLs) vary depending on the strain. 12,13 MAB produces unique polar GPLs that are diglycosylated on alaninol and di-O-acetylated on 6-deoxytalose. 12,13 MAB GPLs mask underlying bioactive cell wall lipids, thereby delaying the immunity activation during the initial colonization stages, 14 indicating their role as a virulence factor. In addition, GPLs facilitate MAB colonization by preventing Toll-like receptor signaling in airway epithelial cells. 15,16 Nevertheless, the precise pathogenic role of GPLs during MAB infection remains unclear.
Macrophages are the primary target and an important reservoir of mycobacteria in lungs, as indicated by the fact that virulent mycobacteria can survive and replicate within macrophages. Apoptotic cell death is regarded as an innate intracellular response designed to limit multiplication of intracellular pathogens. 17,18 A high intracellular burden of virulent M. tuberculosis induces apoptosis with rapid progression to necrosis as a mode of mycobacterial escape. 19 It was also suggested that M. avium uses apoptotic macrophages as a tool for cell-to-cell spreading in the host. 20 We also have reported that highly virulent clinical MAB strains induce significantly more cell death than nonvirulent strains. 21 On the other hand, the pathogenic role of cell death in a virulent MAB infection remains largely unknown. Recently, a study on a zebrafish infection model indicated that R-type MAB induces more macrophage apoptosis than the S type and the virulence correlates with a loss of GPL production and massive production of a serpentine cord. 22 In the present study, we examined the precise role of GPLs in virulence and why the R type is more virulent than S type. Here, we used extracted GPLs, studied their targeting to mitochondria, and their function in macrophage apoptosis.

Results
R-type MAB induces macrophage apoptosis significantly more than the S type. To elucidate the relation between apoptosis and MAB virulence, we compared the apoptotic response of macrophages infected with either S or R morphotype. In the first group, RAW 264.7 cells were infected with MAB at multiplicity of infections (MOIs) of 1 and 5, and then their cell death pattern was analyzed by fluorescenceactivated cell sorting (FACS) with annexin V/propidium iodide (PI) staining from 1 to 4 h with 1 h intervals. The R type induced apoptosis in a time-and dose-dependent manner: significantly more apoptosis than the S type ( Figure 1a). The S morphotype did not induce substantial apoptosis at MOI of 1 and induced prominent apoptosis only at MOI of 5 at 4 h after infection. The cell death progressed rapidly in R-typeinfected RAW 264.7 cells. The difference in the cell death induced by R and S types was not due to a difference in bacterial clumping as indicated by acid fast bacilli (AFB) staining of both bacterial types prepared for infection (Supplementary Figure 1a), and there was no difference in the macrophage infection rate between the two types (Supplementary Figure 1b). Next, MAB-induced apoptosis was confirmed in RAW 264.7 cells (by a TUNNEL assay) that were infected with either of the types of MAB for 2 h and washed three times with PBS. At 6 h after final washing, the R type induced significantly more apoptosis in RAW 264.7 cells than the S type ( Figure 1b). Similar results were observed in primary cells: bone marrow-derived macrophages (BMDMs; Figure 1b). To rule out an effect of extracellularly growing bacteria, BMDMs that were MAB infected for 2 h were treated with kanamycin to kill extracellular bacteria, were washed three times with PBS, and then were cultured in a medium containing kanamycin. At 48 h post infection, the R type induced significantly more apoptosis as compared with the S type (Supplementary Figure 2a).
We then tested whether MAB induced apoptosis in a caspase-dependent manner. Western blot analysis revealed the presence of cleaved forms of caspase 3, caspase 9, and PARP in RAW 264.7 cells infected with the R type ( Figure 1d). Caspase 3 and 9 activities were significantly higher in R-typeinfected than in S-type-infected macrophages (Figure 1c). R-type-induced apoptosis was inhibited by caspase 3-and 9-specific inhibitors ( Figure 1e). Overall, these results indicated that the R type induces robust caspase-dependent apoptosis in macrophages when compared with the S type's effects.
Polar GPLs of the S type inhibit intrinsic apoptosis of macrophages. Several apoptosis-regulating proteins and glycolipids of M. tuberculosis have been reported. 23,24 In an assay designed to identify the apoptosis-inducing component of MAB, we found that heat killing of the bacteria did not result in a significant decrease in R-type-induced macrophage apoptosis (Supplementary Figure 2b) or caspase activities (Figure 1c). In addition, there was no difference in cell survival effects between culture filtrate proteins from R-and S-type MAB (Supplementary Figure 3). These results suggested that lipids rather than proteins are the apoptosismodulating factors of MAB. Therefore, we hypothesized that the differential apoptotic response induced by the two morphotypes may be due to a difference in the amount of cell wall lipids. As expected, a major difference in total cell wall lipids between S and R types was observed in thin-layer chromatography (TLC) analysis ( Figure 2a) and two-dimensional TLC (Supplementary Figure 4) in the GPL region. Total lipids and purified GPLs from R-type MAB showed a few spots in the GPL region. The acetone-precipitated pellet (APP), where GPLs were removed, yielded the same pattern in both morphotypes. To quantitatively compare the GPL amounts between both types, we purified mycolic acid, total lipid extract, GPL, and APP from the bacteria (Supplementary Figure 5). The amount of the mycolic acid in both types was same, but the GPLs in S type were 3.1-fold higher than R type when the GPLs were purified from 10 mg of the total lipid extracts of each type (Supplementary Table 1). Some spots in the GPL region on TLC were analyzed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF/MS). We confirmed that the small spots of APP that appeared in the GPL region were not lipids of the GPL fraction (Supplementary Figure 6).
Next, we analyzed the potential participation of GPLs in the inhibition of macrophage apoptosis. GPLs did not induce any apoptosis (data not shown). Total lipids and purified GPLs from the S type, but not APP or total lipids of the R type, inhibited staurosporine (STS)-induced apoptosis in a dose-dependent manner (Figure 2b and Supplementary Figure 7). Furthermore, GPLs inhibited apoptosis induced by oligomycin or H 2 O 2 but did not inhibit apoptosis induced by TNF-α plus cycloheximide (Supplementary Figure 7). These results suggested that GPLs modulate the intrinsic apoptotic response of macrophages.
To further elucidate the compounds participating in the modulation of apoptosis, the GPLs were fractionated into polar and apolar GPLs (Figure 2c). Polar GPLs, but not apolar GPLs, significantly inhibited oligomycin-induced apoptosis of macrophages ( Figure 2d). In fact, R-type GPLs consisted of only a small amount of apolar GPLs that were not involved in the inhibition of apoptosis. GPLs deacetylated by alkaline lysis (dGPLs) showed a loss of their antiapoptotic effects (Supplementary Figure 8). These results suggested that polar GPLs are mainly responsible for inhibiting macrophage apoptosis, and their acetylation is necessary for this activity.
GPLs inhibit ROS production and a cytochrome c release from mitochondria and preserve mitochondrial transmembrane potential (ΔΨm). It is well known that enhanced production of reactive oxygen species (ROS) can cause apoptosis. We examined the involvement of cellular ROS in GPL-mediated inhibition of apoptosis in RAW 264.7 cells. GPLs significantly inhibited STS-, H 2 O 2 -, or oligomycininduced cellular ROS production in a dose-dependent manner ( Figure 3a). Next, we evaluated the effects of GPLs on ΔΨ m by flow cytometric analysis using 3,3′-dihexyloxacarbocyanine (DiOC 6 ) staining. Pretreatment of RAW 264.7 cells with GPLs prevented disruption of ΔΨ m by oligomycin, STS, or H 2 O 2 ( Figure 3b). Mitochondria play a central role in the response to apoptotic stimuli and are the major source of ROS in the cell. In macrophages, GPLs dramatically inhibited mitochondrial ROS production induced by oligomycin, a GPLs are targeted to mitochondria. Next, we determined where GPLs are localized in GPL-treated cells. GPLs were found to be colocalized with mitochondria in GPL-treated cells as evidenced by confocal microscopy (Figure 4a). We also found that a considerable proportion of GPLs in S-type-infected macrophages colocalized with mitochondria ( Figure 4b). Only a few anti-GPL antibody-immunostained mitochondria were observed in R-type-infected cells when compared with S-type-infected cells, because of the low GPL content of the R-type MAB. Given that GPLs were not As expected, the anti-GPL antibody reacted with purified GPL spots used as a positive control and did not react with the APP spots serving as a negative control (Figure 4c). A positive reaction with the anti-GPL antibody was detected in both the cytosolic and mitochondrial fractions of the cells treated with GPLs. These findings revealed that GPLs are efficiently transported to mitochondria in S-type MAB-infected macrophages.
GPLs interact with cyclophilin D, a component of the MPT pore. It is important to identify the host molecules that physically and functionally interact with GPLs for elucidation of their apoptosis-modulating mechanism. Therefore, we tested whether GPLs interact with pro-or anti-apoptotic molecules and mitochondrial permeability transition (MPT) pore components. Immunoprecipitation with the anti-GPL antibody revealed that GPLs bind to cyclophilin D (Figure 5a), a modulator of MPT pore opening. 25 GPLs immunoprecipitated with the anti-cyclophilin D antibody were subjected to dot blotting with serial dilution that confirmed that cyclophilin D is coimmunoprecipitated with GPLs ( Figure 5a). Furthermore, confocal microscopic analysis of GPL-treated cells revealed that GPLs colocalized with cyclophilin D (Figure 5b). Although dGPLs were colocalized with mitochondria, their colocalization rate was significantly lower than that of GPLs (Figure 5b). In contrast, dGPLs did not interact with cyclophilin D because the amount of cyclophilin D that was  immunoprecipitated with dGPLs was much smaller than that immunoprecipitated with GPLs ( Figure 5c). These results indicated that GPLs interact with the MPT pore component cyclophilin D, thus stabilizing mitochondrial membrane transition, and their acetylation may be crucial for the interaction with cyclophilin D.
GPLs inhibit the apoptotic response induced by R-type MAB. To confirm the roles of GPLs during infection, we determined whether GPLs could inhibit the R-type-induced apoptotic response in macrophages. The GPLs indeed inhibited R-type-induced macrophage apoptosis (Figure 6a). It is well known that cyclosporin A specifically binds cyclophilin D, leading to inhibition of MPT and apparently blocking apoptosis. 25,26 Just as GPLs, cyclosporin A inhibited R-type-induced macrophage apoptosis (Figure 6a). ROS production induced by the R-type MAB was also significantly inhibited after pretreatment with GPLs ( Figure 6b). Moreover, GPLs prevented ΔΨm disruption induced by R-type MAB (Figure 6c). R-type MAB induced activation of caspases 3 and 9 in macrophages (Figure 1a). We measured caspase activity in macrophages infected with red fluorescent protein (RFP)-expressing R-type MAB. Activated caspase 3 specifically bound to the green-fluorophore-conjugated inhibitor and showed bright spots in a confocal image (Figure 6d). These spots were significantly reduced by cotreatment with GPLs. These results suggested that the stronger apoptotic response induced by R-type MAB in macrophages when compared with the S type may be caused by the lack of GPLs.
GPLs inhibited the intracellular growth and the spreading of R-type MAB via apoptosis inhibition. It is well known that the R type is more virulent than the S type in terms of MAB infection, as proven in studies on survival of MABinfected mice and bacterial growth in lungs. 8 At the cellular level, the intracellular survival rate of the mycobacteria in macrophages allows for indirect assessment of their virulence. Therefore, we analyzed the relation between GPLmediated inhibition of apoptosis and MAB pathogenicity by determining the effect of GPLs on intracellular survival of the R and S types in macrophages. BMDMs were infected with MAB for 2 h, incubated for additional 2 h with kanamycin to kill extracellular bacteria, washed three times with PBS, and then intracellular growth was determined. At 24 h after infection, the growth of the R type within macrophages was significantly greater when compared with that of the S type (Figure 7a). Moreover, pretreatment with GPLs or cyclosporin A significantly suppressed both the intracellular growth and extracellular release of the R type (Figures 7a and b). In contrast, pretreatment with oligomycin significantly enhanced both intracellular and extracellular growth of the R type. In contrast, the growth of the S type was not affected by GPLs or oligomycin pretreatment. It is reported that M. avium uses apoptotic macrophages as a tool for spreading in the host. 20 Therefore, we analyzed the correlation between apoptosis and the spread of bacteria. In the first assay, BMDMs infected with GFP-expressing R-or S-type MAB for 2 h were added with uninfected BMDMs. All To confirm the effect of GPLs on the spread of bacteria, RAW 264.7 cells infected with RFP-expressing R-type MAB for 2 h were added with uninfected cells stained with a blue dye. FACS analysis revealed that the proportion of R-typeinfected cells among blue-stained RAW 264.7 cells increased in a time-dependent manner but was significantly inhibited by GPL pretreatment (Figure 7d). Collectively, these data suggest that apoptosis induced by R-type MAB may contribute to the spread of bacteria.

Discussion
We have previously reported that highly virulent clinical MAB strains induce significantly more cell death than nonvirulent strains. 21 It was recently reported that MAB of the S type is less proapoptotic than the R type. 27 In the present study, we found that GPLs treatment inhibit the apoptotic response via mitochondrial targeting, thereby suppressing the growth of R-type MAB, resulting in inhibition of spreading of MAB among host cells. The data suggest that the low virulence of S-type MAB is associated with the apoptosis-inhibitory activity of the GPLs that are located on the outermost surface.
A variety of key events of cell death take place in mitochondria, including the cytochrome c release, loss of ΔΨ m , ROS production, and participation of apoptosis-regulating factors. 28 We found that GPL-mediated inhibition of apoptosis resulted from downregulation of mitochondrial ROS (and of the release of cytochrome c) and was caused by protection from a ΔΨ m loss. Moreover, GPLs were found to be localized to the mitochondrial compartment in macrophages infected with MAB as well as in macrophages treated with GPLs, although the mechanism of GPL internalization by the host cells remains unknown. Mitochondrial damage has been suggested to play a critical role in the outcome of infection of macrophages with M. tuberculosis. 29 We recently demonstrated that the M. tuberculosis HBHA protein is targeted to the mitochondria and induces mitochondria-dependent apoptosis in macrophages. 30 In the present study, MAB GPLs inhibited macrophage apoptosis via inhibition of mitochondrial damage induced by apoptotic stimuli. Neisseria meningitides porin B interacts with mitochondrial porin VDAC, resulting in enhanced cell survival. 31 Of note, our findings showed that MAB GPLs, just as cyclosporin A, interact with cyclophilin D, leading to inhibition of MPT and of macrophage apoptosis. Inhibitors of MPT pore opening such as cyclosporin A appear to block cell death. 25,26 In our experimental model, MAB GPLs prevented disruption of ΔΨ m induced by R-type MAB and by proapoptotic stimuli. MAB GPLs and cyclosporin A equally inhibited MAB-induced macrophage apoptosis. These results suggest that MAB GPLs interfere with the regulatory role of cyclophilin D in MPT pore opening, although further research into the underlying mechanisms is needed.
It is known that NTM species produce unique polar GPLs. MAB produces specific polar GPLs that are diglycosylated and diacetylated on the core GPL structure. In the present study, the polar GPL region was found to be involved in the inhibition of macrophage apoptosis. It is reported that proper acetylation and methylation of M. avium GPLs are required for TLR signaling and activation of the innate immune response. 32 In the present study, acetylation of MAB GPLs was found to be necessary for both the inhibitory activity and interaction with cyclophilin D because dGPLs did not bind to cyclophilin D and did not have any antiapoptotic effect.
Macrophage death is an important characteristic of hostmycobacteria interactions. Therefore, the properties of mycobacteria that modulate apoptosis have been extensively studied. Most reports have indicated that the induction of apoptosis by M. tuberculosis and M. avium is inversely proportional to bacterial virulence. 33 Nevertheless, the function of apoptosis in infection is still a subject of debate. 34 RD1-deficient M. tuberculosis cannot trigger apoptosis and shows an impaired ability to colonize new uninfected cells, suggesting that apoptosis promotes cell-to-cell spread of bacteria. [34][35][36] The importance of ESX-1-dependent apoptosis in the spread of M. marinum among host cells has been shown in the zebrafish model. 37 In addition, apoptosis appears to act as a virulence mechanism for M. leprae. 38,39 Therefore, induction of apoptosis can apparently serve different functions according to the mycobacterial strain and its infection course. In the present study, inhibition of apoptosis by GPLs suppressed the growth of the virulent R type; however, induction of apoptosis by oligomycin enhanced bacterial growth. Furthermore, the increased apoptosis was found to be associated with spreading of the bacteria among host cells. Recently, a significant increase in the proportion of apoptotic macrophages was demonstrated in R-type-infected zebrafish as compared with S-type-infected macrophages. 22 These authors also suggested that macrophage apoptosis is a key event for the release of extracellular MAB. In the present study, extracellular R-type MAB growth rapidly increased starting at 24 h after infection when compared with S-type MAB. Extracellular growth of R-type MAB was enhanced by treatment with an apoptosis inducer and was suppressed by GPLs or cyclosporin A. In agreement with these results, cyclosporin A suppressed intracellular growth of M. tuberculosis in macrophages by inhibiting necrosis of the infected macrophages. 40 Nonetheless, in our experimental model, GPLs and cyclosporin A appeared to attenuate the increase in the number of annexin V-positive macrophages induced by R-type MAB and inhibited both extracellular and intracellular growth of the bacteria. Therefore, this discrepancy may due to a difference in the growth rate and apoptotic response between M. tuberculosis and MAB. Further research on the various characteristics of cell death modulated by GPLs and cyclosporine A is needed. Taken together, our findings reveal that R-type-induced apoptosis contributes to the spread of MAB and provides novel insights into the cellular pathogenic mechanism of R-type MAB and why it is more virulent than the S type. In line with our results, another study indicated that M. avium uses apoptotic macrophages as a vector for spreading. 20 However, further studies are needed to investigate whether GPLs can also inhibit alveolar macrophages apoptosis in vivo that can suppress the spread and growth of MAB R type and attenuate the virulence.

Materials and Methods
Animal experiments. Male inbred C57BL/6 mice (5-6 weeks old) were used for preparation of the BMDMs. All the animal experiments were approved by the institutional research and ethics committee at Chungnam National University (permission number: CNU-00405). All the animal procedures were performed in accordance with the guidelines of the Korean Food and Drug Administration.
Bacterial culture and GPL purification. M. abscessus ATCC19977 and the isogenic R type were grown to mid-log phase in the Middlebrook 7H9 medium containing OADC (BD, Franklin Lakes, NJ, USA), were washed three times, aliquoted, and frozen at − 80°C until use. Bacterial total lipids and GPLs were purified, solubilized, and confirmed as described previously. [41][42][43] Total lipids were extracted from M. abscessus with a chloroform/methanol mixture (2 : 1, v/v) by ultrasonication for 20 min and were phase-separated with 5% distilled water by centrifugation. GPLs were purified from total lipid extracts by acetone precipitation. The purified lipids were separated by TLC (Millipore, Billerica, MA, USA) in chloroform/methanol (9 : 1, v/v) and detected by spraying with 20% H 2 SO 4 and heating at 200°C for 15 min. Two-dimensional TLC and visualization of phospholipids and amine groups were performed by means of ninhydrin and Dittmer's solutions. Purified GPLs were separated into two fractionspolar and apolar GPLsby scratching silica off a GPL-loaded TLC plate. For treatments, the lipid antigens were solubilized in distilled water containing 0.025% of Tween-80 (Sigma-Aldrich, St. Louis, MO, USA) using sonication.