Mycobacterium tuberculosis (M. tuberculosis), causes tuberculosis (TB) and is a major public health problem (World Health Organization [WHO], 2015). The ability of M. tuberculosis cells to survive under host and antibiotic stress partly explains why M. tuberculosis is a successful human pathogen. Hence, cellular adaptations conferring stress tolerance in M. tuberculosis and in related species are an active area of research (Kieser and Rubin, 2014).

Investigations into cell biology of mycobacteria have revealed several unique characteristics in growth and division, which contributes to their survival under stress conditions (Thanky et al., 2007; Hett and Rubin, 2008; Kieser and Rubin, 2014). One such cellular structure is the complex cell envelope of mycobacteria (Brennan and Nikaido, 1995). Electron microscopy has revealed the ultrastructure of cell envelope layers in mycobacteria (Takade et al., 1983; Hoffmann et al., 2008; Zuber et al., 2008; Vijay et al., 2012). The cell envelope is essential for M. tuberculosis survival as it acts as a permeability barrier for the entry of antibiotics and also modulates host immune response (Jarlier and Nikaido, 1994; Briken et al., 2004; Torrelles and Schlesinger, 2010). Therefore, it is also an important drug and vaccine target (Chatterjee, 1997; Abrahams and Besra, 2016; Tima et al., 2017). The composition of cell envelope layers has been determined using cell envelope mutants (Etienne et al., 2002, 2005) and antibiotic treatments which inhibit the envelope synthesis in mycobacteria (Mdluli et al., 1998). These studies have advanced our understanding of the cell envelope role as a permeability barrier and in inhibiting phagocytosis of mycobacteria by macrophages (Mdluli et al., 1998; Etienne et al., 2002, 2005).

Another feature revealed by electron microscopy was the accumulation of intracytoplasmic lipid inclusions in mycobacteria under different host infection model systems (Peyron et al., 2008; Caire-Brandli et al., 2014; Barisch and Soldati, 2017a). In an in vitro human granuloma model of infection, M. tuberculosis cells accumulated lipid inclusions during infection of lipid loaded macrophages called foam cells (Peyron et al., 2008). Similarly, M. avium accumulated host-derived lipids as inclusions in foam cells and exhibited a thin cell envelope (Caire-Brandli et al., 2014). Recently, M. marinum was also found to have lipid inclusions derived from host lipids during the infection of Dictyostelium (Barisch and Soldati, 2017a). These studies have identified triacylglycerols as the major lipid in mycobacterial lipid inclusions derived from host cells (Peyron et al., 2008; Daniel et al., 2011; Caire-Brandli et al., 2014; Barisch and Soldati, 2017a). M. tuberculosis and M. smegmatis can also accumulate lipid inclusions containing triacylglycerols under in vitro stress conditions independent of host cells (Garton et al., 2002; Anuchin et al., 2009; Deb et al., 2009). Several studies have shown that M. tuberculosis uses diverse host carbon sources such as cholesterol, pyruvate, and glucose (Pandey and Sassetti, 2008; Marrero et al., 2013; Baker et al., 2014). Utilization of such diverse carbon sources by M. tuberculosis contributes to its pathogenesis and persistence in the host (Pandey and Sassetti, 2008; Marrero et al., 2013; Baker et al., 2014).

Importantly, the accumulation of lipid inclusions in M. tuberculosis was associated with persistence, antibiotic tolerance, cavitation, and poor treatment outcome (Deb et al., 2009; Russell et al., 2009; Daniel et al., 2011; Hammond et al., 2015; Kayigire et al., 2015; Sloan et al., 2015). It is possible that this is due to growth arrest of M. tuberculosis and loss of antimicrobial functions by foamy macrophages leading to persistent infection (Peyron et al., 2008; Daniel et al., 2011; Caire-Brandli et al., 2014). This phenomenon may lead to clinical complications, such as relapse of infection and the emergence of antibiotic-resistant M. tuberculosis (Cohen et al., 2013; Sebastian et al., 2017). Thus, intracytoplasmic lipid inclusions and the cell envelope are important for the survival of M. tuberculosis. The understanding of these cellular characteristics and their adaptations to stress in clinical M. tuberculosis isolates is limited. This understanding is vital for the development of novel therapeutic targets. In our previous study, we have observed that host stresses influenced cell length distribution in a large set (n = 158) of clinical M. tuberculosis isolates (Vijay et al., 2017). In this study we investigated the accumulation of lipid inclusions and cell envelope ultrastructure of M. tuberculosis in sputum by transmission electron microscopy (TEM). The ultrastructure of M. tuberculosis in sputum was compared with the ultrastructure of clinical M. tuberculosis isolates and H37Rv in liquid culture, and under conditions of oxidative stress, iron deficiency, and exposure to the antibiotic isoniazid.

We analyzed the lipid inclusions and cell envelope layers in clinical M. tuberculosis isolates ex vivo in sputum representing the host environment. We then compared this with the ultrastructure of clinical M. tuberculosis isolates and H37Rv in liquid culture and under different in vitro stress conditions. This revealed the accumulation of intracytoplasmic lipid inclusions in clinical M. tuberculosis isolates as a cellular adaptation in sputum, liquid culture, and under stress conditions. Analysis of six clinical M. tuberculosis isolates revealed isolate-to-isolate variation in the extent of lipid inclusions in mid-log culture and its increased accumulation under stress conditions. The thickness of M. tuberculosis cell envelope ETL was significantly reduced under different stress conditions. Formation of an intracytoplasmic compartment in M. tuberculosis cells was also observed under iron deficiency.

Mycobacterium tuberculosis cells with lipid inclusions have been associated with foamy macrophages and unfavorable treatment outcome in tuberculosis patients (Garton et al., 2002; Peyron et al., 2008; Kayigire et al., 2015; Sloan et al., 2015). In the present study, clinical M. tuberculosis isolates displayed lipid inclusions even in liquid culture, which was not observed in the laboratory strain H37Rv. Similarly, M. avium and M. marinum also do not accumulate lipid inclusions in macrophages and the extracellular environment, respectively (Caire-Brandli et al., 2014; Barisch and Soldati, 2017a). This indicates that accumulation of lipid inclusions is a more prominent cellular adaptation in clinical M. tuberculosis isolates compared to laboratory strains of mycobacteria. Supporting this, we also observed increased accumulation of lipid inclusions under both oxidative stress and sub-inhibitory concentration of isoniazid only in clinical M. tuberculosis isolates. Isoniazid can also induce oxidative stress and may therefore link these findings (Timmins and Deretic, 2006). It will be interesting to study how other antibiotic treatments influences the accumulation of lipid inclusions in clinical M. tuberculosis isolates, as its accumulation may have a role in M. tuberculosis persistence to antibiotics (Hammond et al., 2015; Kayigire et al., 2015; Sloan et al., 2015).

We observed increased accumulation of lipid inclusions in M. tuberculosis cells at 500 μM DFO compared to 100 μM DFO-treated cells under iron deficiency. DFO concentration-dependent accumulation of lipid inclusions were found in both clinical M. tuberculosis isolates and H37Rv. Supporting these observations it has also been reported that iron deficiency and oxidative stress can induce lipid accumulation in mycobacteria, which depends on host foamy macrophages (Bacon et al., 2007; Peyron et al., 2008). Host oxidative stress generates oxidized low-density lipoproteins, and oxygenated mycolic acids present in M. tuberculosis; both can trigger the differentiation of host macrophages into foamy cells (Peyron et al., 2008; Palanisamy et al., 2012). This in turn facilitates the accumulation of lipid inclusions in M. tuberculosis cells and provides a protective niche for its survival. Our host stress models were based on in vitro culture lacking foamy macrophages. Hence, accumulation of lipid inclusions in our host stress models in M. tuberculosis cells may have derived lipids from oleic acids present in the culture media, as seen in case of M. smegmatis (Garton et al., 2002; Anuchin et al., 2009).

Oxidative stress was also a co-factor in all of the stress conditions where we observed the increased accumulation of lipid inclusions in M. tuberculosis cells (Rodriguez and Smith, 2003; Timmins and Deretic, 2006). Transcriptional adaptation of M. tuberculosis in macrophages and under in vitro stress conditions strongly correlates with the ultrastructural adaptations observed here, indicating that under host stress M. tuberculosis shifts to a fatty acid-based metabolism (Schnappinger et al., 2003). Enzymes involved in fatty acid metabolism are also essential for in vivo growth and virulence (Munoz-Elias and McKinney, 2005; Reed et al., 2007). The accumulation of lipid inclusions is implicated in M. tuberculosis cell division arrest and induction of antibiotic tolerant dormant phenotype (Daniel et al., 2011; Caire-Brandli et al., 2014). This needs to be reinvestigated as our study shows that lipid inclusions per se may not inhibit cell division in M. tuberculosis. We observed M. tuberculosis cells with lipid inclusions growing in mid-log culture and under iron deficiency, similar to the growth observed in M. marinum with lipid inclusions (Barisch and Soldati, 2017a). It is possible that accumulation of lipids being a cellular adaptation that can facilitate M. tuberculosis entry into, and survival during dormancy (Barisch and Soldati, 2017b).

The unique triple layered cell envelope, reported in several laboratory mycobacterial strains and in clinical strains of M. tuberculosis (Takade et al., 1983; Brennan and Nikaido, 1995; Velayati et al., 2009; Vijay et al., 2012), was also observed in all of the clinical M. tuberculosis isolates in the present study. The ultrastructure of triple layered cell envelope from our study was also similar to the cell envelope ultrastructure of M. tuberculosis processed by cryofixation and rapid freeze substitution (Yamada et al., 2010, 2015). We also observed tearing of resin around M. tuberculosis cells in sputum, as observed in TEM images of M. marinum granulomas and M. tuberculosis cells (Bouley et al., 2001; Vijay et al., 2014). The thickness of the triple layers under mid-log growth conditions was consistent across the six clinical M. tuberculosis isolates and H37Rv used here. However, under stress conditions like sputum, oxidative stress, and iron deficiency, we observed a significant reduction in the thickness of cell envelope ETL, although the extent of this reduction varied between the two sputum samples despite a similar bacterial load. These findings suggest that the ETL can be reduced in thickness under host stress, which may vary from patient to patient. This needs to be investigated in a greater number of patients and correlated with aspects such as severity of tuberculosis symptoms and persistence to understand the clinical significance of such adaptations.

The ETL is mainly composed of lipids like mycolic acids (Mdluli et al., 1998; Wang et al., 2000) and transcriptional analysis of M. tuberculosis cells under host stress also indicate cell envelope remodeling and fatty acid degradation (Schnappinger et al., 2003). Cell envelope lipids are also involved in host immune modulation and virulence of M. tuberculosis strains (Karakousis et al., 2004; Makinoshima and Glickman, 2005). It has also been observed that under different stress conditions M. tuberculosis loses acid fastness due to loss of cell envelope lipids and it is associated with dormancy and antibiotic tolerance (Bhatt et al., 2007; Deb et al., 2009). We also observed reduced acid fast staining and M. tuberculosis cells with acid fast stained cytoplasmic beads in oxidative stress and iron deficiency. Such cells were also observed in some sputum samples and under isoniazid treatment, these observations strongly correlate with the ultrastructural adaptations such as reduced ETL and accumulation of lipid inclusions in our study. Further investigations are needed to understand the role of reduced cell envelope lipids on the accumulation of intracytoplasmic lipid inclusions in M. tuberculosis.

Reduction in the envelope lipids may enhance the permeability of cell envelope and influence the susceptibility of M. tuberculosis to antibiotics. Cell envelope modifications and enhanced antibiotic susceptibility in M. smegmatis have been observed under iron deficiency (Pal et al., 2015). Triacylglycerol is also a component of M. tuberculosis cell envelope and loss of acid fastness is observed under iron deficiency and in hypoxia (Rastogi et al., 2017). Such cell envelope modifications accompany non-replicative persistence and antibiotic tolerance of M. tuberculosis in vitro (Rastogi et al., 2017). These observations indicate the influence of host factors on cellular adaptations in M. tuberculosis and antibiotic susceptibility. As there are multiple host factors and complex interactions influencing antibiotic susceptibility, this needs to be investigated further to identify the factors that can enhance susceptibility to antibiotics. There was significant variation in the accumulation of lipid inclusions in clinical M. tuberculosis isolates in mid-log culture. Such differences between M. tuberculosis isolates may have a clinical significance in persistence against host stress and antibiotics. Hence, variations in cellular adaptations need to be correlated with persistence and antibiotic tolerance among clinical M. tuberculosis isolates to understand its role in treatment failure.

Beijing lineage was shown to accumulate triacylglycerides and has triacylglyceride synthase gene (Rv 3130c) upregulated during in vitro growth (Reed et al., 2007). This gene is a member of DosR regulon, and some of the regulon genes are constitutively overexpressed in Beijing lineage (Domenech et al., 2017). DosR and WhiB3 have been shown to modulate lipid accumulation in M. tuberculosis (Singh et al., 2009), and also contribute to bacilli adaptation to hypoxia and redox stresses, respectively (Park et al., 2003; Saini et al., 2004; Singh et al., 2009). These proteins may play a role in the accumulation of lipid inclusions under oxidative stress and iron deficiency in clinical M. tuberculosis isolates. The mechanism of formation of lipid inclusions in mycobacteria also involves interactions with host lipid droplets and membrane phospholipids (Barisch and Soldati, 2017b). Thus, host stresses may induce significant cell biological adaptations in clinical M. tuberculosis isolates; its molecular mechanism needs to be further investigated.

In addition to reduction in the thickness of ETL and accumulation of lipid inclusions in M. tuberculosis cells, we also observed intracytoplasmic compartments under iron deficiency. These compartments were approximately 200 nm in size and were specifically observed in all M. tuberculosis isolates cultured under 500 μM DFO. It is possible that these compartments are mesosomes as observed in bacteria treated with antibiotics (Santhana Raj et al., 2007; Li et al., 2008). Studies have also shown the formation of intracellular compartments which accumulate H₂O₂ under cellular damage (Ebersold et al., 1981; Li et al., 2008; Xin et al., 2014). Mesosomes and other such intracellular structures are considered as ultrastructural artifact induced under chemical fixation and dehydration process, and these are not observed under cryo-electron microscopy lacking such fixation (Pilhofer et al., 2010). In this study we have used primary fixation with osmium tetroxide for 1 h, and post-fixation with glutaraldehyde for 2 h. There is a possibility of such chemical fixation inducing the formation of intracellular structures, specifically under stress conditions. Cellular adaptations under iron deficiency may increase the probability of formation of such structures during chemical fixation, as we observed them only in M. tuberculosis cells under iron deficiency. In cryoelectron microscopy cells are imaged at frozen-hydrated state without chemical fixation or dehydration of cells and can avoid much of the fixation artifacts (Pilhofer et al., 2010). Cytoplasmic structure termed as stack has been reported in slow growing Pseudomonas deceptionensis M1 by TEM and also confirmed by cryo-electron microscopy (Delgado et al., 2013). If confirmed to be a true cellular structure by cryoelectron microscopy and specific for M. tuberculosis in iron deficiency. These compartments probably may have a role in iron storage.

Iron limitation has been a common host defense encountered by M. tuberculosis; hence, it has evolved mechanisms to sequester iron from the host by using siderophores like mycobactin (Rodriguez and Smith, 2003; Ratledge, 2004). Inside M. tuberculosis cells bacterioferritins BfrA and BfrB function as iron storage proteins. Recent observations have shown that BfrB can be encapsulated by the protein encapsulin to form nanocompartments in vitro (Reddy et al., 2012; Contreras et al., 2014). We observed arrangement of units with size ∼20 nm inside these intracytoplasmic compartments in M. tuberculosis cells, which is similar in size to the encapsulin observed in vitro (Contreras et al., 2014). These observations suggest that these intracytoplasmic compartments may be encapsulin-based nanocompartments in M. tuberculosis. They may be used to isolate excess of iron from generating oxidative cellular damage or a similar protective function under stress (Contreras et al., 2014). Iron storage has been essential for M. tuberculosis survival and virulence, hence has also been a potential novel drug target (Pandey and Rodriguez, 2012). It is important to investigate further the nature of these intracytoplasmic compartments in avirulent laboratory strains by cryo-electron microscopy and its role in M. tuberculosis survival under iron deficiency. Recent observations further implicate survival of M. tuberculosis in iron deficiency and accumulation of lipid inclusions to antibiotic tolerance and persistence (Baron et al., 2017; Kurthkoti et al., 2017).

In summary, we were able to demonstrate the major cellular adaptations of clinical M. tuberculosis isolates to host and antibiotic stress conditions. Further investigation of these cellular adaptations and their role in M. tuberculosis survival under stress is important. These will aide in our understanding of the ability of M. tuberculosis cells to persist during host and antibiotic stress. The variations in cellular response among clinical M. tuberculosis isolates may be associated with the persistence and treatment outcome among patients.

SV, NT, GT, NP, and EJ conceived and designed the experiments. SV, HH, and DT did the experiments. SV and AP did TEM analysis. SV, NT, GT, EJ, and AP analyzed and interpreted the data. SV, HH, DT, NP, NT, GT, AP, and EJ drafted and revised the manuscript and approved the final version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.