Mycobacterial Cultures Contain Cell Size and Density Specific Sub-populations of Cells with Significant Differential Susceptibility to Antibiotics, Oxidative and Nitrite Stress

The present study shows the existence of two specific sub-populations of Mycobacterium smegmatis and Mycobacterium tuberculosis cells differing in size and density, in the mid-log phase (MLP) cultures, with significant differential susceptibility to antibiotic, oxidative, and nitrite stress. One of these sub-populations (~10% of the total population), contained short-sized cells (SCs) generated through highly-deviated asymmetric cell division (ACD) of normal/long-sized mother cells and symmetric cell divisions (SCD) of short-sized mother cells. The other sub-population (~90% of the total population) contained normal/long-sized cells (NCs). The SCs were acid-fast stainable and heat-susceptible, and contained high density of membrane vesicles (MVs, known to be lipid-rich) on their surface, while the NCs possessed negligible density of MVs on the surface, as revealed by scanning and transmission electron microscopy. Percoll density gradient fractionation of MLP cultures showed the SCs-enriched fraction (SCF) at lower density (probably indicating lipid-richness) and the NCs-enriched fraction (NCF) at higher density of percoll fractions. While live cell imaging showed that the SCs and the NCs could grow and divide to form colony on agarose pads, the SCF, and NCF cells could independently regenerate MLP populations in liquid and solid media, indicating their full genomic content and population regeneration potential. CFU based assays showed the SCF cells to be significantly more susceptible than NCF cells to a range of concentrations of rifampicin and isoniazid (antibiotic stress), H2O2 (oxidative stress),and acidified NaNO2 (nitrite stress). Live cell imaging showed significantly higher susceptibility of the SCs of SC-NC sister daughter cell pairs, formed from highly-deviated ACD of normal/long-sized mother cells, to rifampicin and H2O2, as compared to the sister daughter NCs, irrespective of their comparable growth rates. The SC-SC sister daughter cell pairs, formed from the SCDs of short-sized mother cells and having comparable growth rates, always showed comparable stress-susceptibility. These observations and the presence of M. tuberculosis SCs and NCs in pulmonary tuberculosis patients' sputum earlier reported by us imply a physiological role for the SCs and the NCs under the stress conditions. The plausible reasons for the higher stress susceptibility of SCs and lower stress susceptibility of NCs are discussed.


Name
Reference or Source

Bacterial strains
Mycobacterium smegmatis mc 2 155 William R. Jacobs (Snapper et al., 1990) Mycobacterium The 64% and 66% percoll fractions were mostly containing SCs and hence they were termed short-sized cells' fraction 1 & 2 (SCF1 & SCF2) based on their frequency of length distribution ( Figure 1E,F), respectively. The 78% percoll fraction was mostly composed of NCs and was called normal-sized cells' fraction (NCF). All the other percoll fractions of Msm cells were called mixed-sized cell fractions (MCF). under DIC (bottom lines) are indicated in the second column; their respective proportions are mentioned in the third column; the proportion of outliers which are much above or below the average length ± SD is indicated in the fourth column. These are cells, which were also counted to determine the average length, but having lengths much lower or much higher than the average length of the cells in the respective fraction, in spite of having same density as that of average sized cells. n ≥ 300 cells from each fraction. The lines in the first column indicate the percoll fractions; the average lengths and SD of cells in the SCF1, SCF2, and NCF fractions as measured under bright-field (BF) are indicated in the second column; their respective proportions are mentioned in the third column; the proportion of outliers which are much above or below the average length ± SD is indicated in the fourth column. These are cells, which were also counted to determine the average length, but having lengths much lower or much higher than the average length of the cells in the respective fraction, in spite of having same density as that of average sized cells.

__________________________________________________
The percoll step-gradient used for Mtb cells was 60-76%. The 68-76% percoll fractions of the Mtb sample contained very few or no cells. The 60%, 62% and 64% Mtb percoll fractions were also mostly composed of shorter cells based on their frequency of length distribution ( Figure   8C). However, since the average size of cells from 60% and 62% were comparable, they were mixed together and termed Mtb SCF1. Subsequently, the 64% Mtb percoll fraction was termed as Mtb SCF2. The 66% percoll fraction was mostly composed of longer cells and was called Mtb NCF ( Figure 8C). The lines in the first column indicate the percoll fractions; the average lengths and SD of cells in the SCF1, SCF2, and NCF fractions as measured under bright-field (BF) (top lines) and

Supplementary
under DIC (bottom lines) are indicated in the second column; their respective proportions are mentioned in the third column; the proportion of outliers which are much above or below the average length ± SD is indicated in the fourth column. These are cells, which were also counted to determine the average length, but having lengths much lower or much higher than the average length of the cells in the respective fraction, in spite of having same density as that of average sized cells. n ≥ 300 cells from each fraction.

CFU Determination of Msm and Mtb SCF1, SCF2 and NCF
The cells in the Msm 64% (SCF1), Msm 66% (SCF2) and Msm 78% (NCF), obtained following analytical scale percoll gradient centrifugation, were resuspended in 400 µl of 1x PBS or 0.5% Tween 80 or Middlebrook 7H9 medium, while the cells in the NCF were further diluted 250 times with Middlebrook 7H9 medium (as mentioned under 'MATERIALS AND METHODS'). Subsequently, 200 µl from each of the respective cell suspensions was added into 25 ml Middlebrook 7H9 medium taken in 100 ml flask, to obtain 10 3 cells/ml of cell density, followed by exposure to stress. In order to obtain 10 5 cells/ml of the preparative scale percoll gradient fractionated SCF and NCF [which was visually made (by dilution with the medium) to the same cell density as that of SCF (the 400 µl of SCF1 + SCF2 mixture prepared)], 100 µl from each of the respective cell suspensions was added into 25 ml Middlebrook 7H9 medium taken in 100 ml flask followed by exposure to the stress (as mentioned under 'MATERIALS AND METHODS'). It may be noted here that the visual comparison of cell density of the fractions for matching cfu were earlier verified for the accuracy of cfu values by plating performed multiple times using independent samples prepared from multiple cultures at different times by different people. Following the addition of cells from the cell suspensions, 100 µl was taken from each of these samples from the 25 ml Middlebrook 7H9 medium, at 0 hr (before exposure) and at the time mentioned post-exposure to the stress agents, for serial dilution followed by plating to determine their cfu.
Since the cells in the percoll fractions were found to elongate, when kept in PBS or Middlebrook 7H9 medium after removal of percoll, it was not possible to take cell count of the fractions or to determine cfu (takes 3 days), to obtain almost equal cell number of the SCF1, SCF2 and NCF for the exposure to stress agents. Therefore, a large number of sets (n = 80 sets) of the SCF1, SCF2, and NCF samples, which were prepared independently on multiple occasions, were plated to find out the consistency and reproducibility of the size range of the cells that get fractionated into 64%, 66%, and 78% percoll fractions, corresponding to SCF1, SCF2, and NCF samples, respectively. From these experiments, the consistent volumes of the SCF1, SCF2, and NCF, which reproducibly gave consistent cfu, were found out and used.
The average (± standard deviations) observed for the Msm cfu in the SCF1, SCF2 and NCF were as follows: 110.4 (± 43.3), 202.6 (± 80.03) and 124.6 (± 94.7). Though these variations were observed in the cfu, the technical triplicates within each set of experiment were consistent and the nature and the trend of the response of the individual fractions were reproducible and consistent.
The cell density of the Mtb SCF2 (64%) and Mtb NCF (66%) was visually made (by dilution with the medium) to the same cell density as that of Mtb SCF1 (60 + 62%). In order to obtain 10 4 cells/ml of Mtb SCF1, SCF2 and NCF, for exposure to stress, 100 µl each of the respective cell suspensions was added into 25 ml Middlebrook 7H9 medium taken in 100 ml flask followed by exposure to the stress.
The average (± standard deviations) observed for the Mtb cfu in the SCF1, SCF2 and NCF were as follows: 42.8 (± 7.6), 57.2 (± 8.5) and 45.4 (± 21.9). Even though these variations were observed in the cfu, as observed in Msm cell samples, the technical triplicates within each set of experiment were consistent and the nature and the trend of the response of the individual fractions were reproducible and consistent.
It was not possible to obtain 100% enrichment for either the SCs or the NCs of Msm or Mtb cells in any percoll fraction for the following probable reasons. The cell size heterogeneity of Msm mother cells varied from on an average of 4 µm (normal size) to 9 µm (longer size) in length. With the cells elongating prior to division, the length varied from 8 µm to 18 µm! Therefore, although difficult to determine, it is possible that the longer cells whose sizes are higher than the average cell size (+) SD in the 64% Msm SCF1 might have come from the asymmetric division of longer mother cells. Similarly, the shorter cells whose sizes are lesser than the average cell size (-) SD in the 78% Msm NCF might have come from the asymmetric division of shorter mother cells. Similar explanations may as well be applicable to the enrichment of Mtb cells in the respective percoll fractions.
We speculate that the low proportions of normal-sized cells in the SCF and the short cells in the NCF could probably be due to the comparable buoyant density of these respective minor population of the cells to that of the larger proportion of the cells in the respective fraction. Nevertheless, majority of the shorter cells got fractionated into the percoll fractions of low buoyant density. Differential buoyant density of mycobacterial cells has been found in M. tuberculosis cells subjected to multiple stress conditions and has been suggested to be due to differential lipid (triglyceride) content (Deb et al., 2009). Thus, the heterogeneity in the population seemed to be not confined to cell size alone, but to density as well, indicating that the high levels of heterogeneity found in mycobacterial population (Deb et al., 2009;McCarthy, 1974;Khomenko, 1987;Anuchin et al., 2009;Ghosh et al., 2009;Ryan et al., 2010;Farnia et al., 2010;Markova et al., 2012;Aldridge et al., 2012) seem to be based on several parameters, operating through multiple mechanisms, under diverse growth and stress conditions.
Equal cell density (10 3 , 10 4 , or 10 5 cells/ml) of the respective Msm and Mtb SCF1, SCF2, and NCF cells were exposed individually to a range of concentrations of rifampicin and isoniazid (antibiotic stress) and H2O2 (oxidative stress) (Milano et al., 2001), for different durations. Likewise, the Msm SCF1, SCF2 and NCF cells with the same cell density (10 3 cells/ml) were also exposed to 7.5 mM NaNO2 (pH 5) (nitrite stress) (Colangeli et al., 2009) for 30 min (as mentioned under 'MATERIALS AND METHODS'). The percentage survival of the different samples, in terms of cfu, against the four stress agents was determined by plating the respective stressed cells and the unstressed cells on stress-agent-free plates. Using these experimental rationale and strategy, we investigated whether SCF1, SCF2, and NCF cells showed differential survivability against these stress agents.

Rationale for the Range of Rifampicin, Isoniazid and H 2 O 2 Concentrations Used
In order to find out the robustness of the stress response exhibited by Msm SCF1, SCF2 and NCF cells, the cells were exposed to a range of rifampicin, isoniazid and H2O2 concentrations. Subsequently, the response of the cells in these percoll fractions to acidified nitrite stress, was also determined. However, the concentrations of these stress conditions which resulted in very less lethality or those which effected in very high lethality to the cells, were not selected for the stress exposure. The range of concentrations of rifampicin, isoniazid and H2O2 used are those that effect survivability between 0% and 100% survival. For example, since the survivability of the Msm NCF cells was ~80% (data not shown) when exposed to < 25 µg/ml rifampicin and < 10% survival during 100 µg/ml rifampicin exposure, these extreme rifampicin concentrations were not used for the experiments. In the same manner, exposure of Msm NCF cells to < 2.5 µg/ml isoniazid concentration showed ~80% survival (data not shown). In contrast, the exposure of Msm SCF cells to 15 µg/ml isoniazid concentration resulted in < 10% survival. Likewise, the Msm NCF cells when exposed to 0.4 mM H2O2, the percentage survival was > 90%, while the exposure of the cells to 1 mM H2O2 resulted in < 10% survival. Hence, these extreme concentrations of rifampicin, isoniazid and H2O2 and were not used for the experiments. Similarly, exposure of Msm NCF cells to 10 mM NaNO2 (pH 5) was observed to be lethal (data not shown). Hence, the Msm cell samples were exposed to 7.5 mM of acidified sodium nitrite for 30 min. The durations of exposure to the stress agents for the chosen range of concentrations were also standardised keeping the range of survival in view.