Variation in Mycobacterium tuberculosis genotype and molecular phenotype influence clinical phenotype of Pulmonary tuberculosis and Tuberculous Meningitis infection in host

2.1 Ethics statement

All studies on human subject including collection of sputum and CSF samples from PTB and TBM cases were approved by Ethical Committee of Dr. G. M. Taori Central India Institute of Medical Sciences (CIIMS), Nagpur. Informed consents were obtained from all particpants after oral explanation of the study to them. All protocols for animal experiments, including infection, dosing were approved by Ethical committee of CIIMS and National JALMA Institute of leprosy and other mycobacterial diseases, Agra (JALMA), Uttar Pradesh, India.

2.2 Study setting and participant recruitment

An clinical observational and molecular study was conducted at CIIMS, Nagpur, a frontline tertiary care neurology hospital in Central India. The hospital serves approximately 60000 patients per year, 70% of which are neurological cases. A total of 1458 patients with TB admitted to CIIMS from 2003 to 2013 were included in the study using predefined clinical inclusion criteria as per standards of TB care in India [10]. For each medical case file, the patient’s history, physical findings, radiographical information and reports of laboratory investigations were assessed to obtain the necessary information relating to diagnosis of TB. For each patient, demographic information (age, gender), lifestyle factors (smoking habits and alcohol use) and clinical characteristics were recorded. Clinical characteristics included, co-morbid conditions (diabetes mellitus, hypertension, malignancies, other infections encountred in the past one year), previous treatment for TB, and history of known contact with an active TB case. Radiological investigations included chest radiographs and computerized tomography (CT) of PTB and TBM cases.

Of the 1458 cases eligible for inclusion, 632 TBM cases died, emigrated or were lost to follow-up within the first year of diagnosis. 68 cases did not give consent for the study. 81 cases were excluded due to lack of sample Cerebrospinal fluid (CSF) or sputum for analysis. 677 cases of TB were included in the final analysis. Among these, 144 cases were diagnosed with bacterial meningitis and thus excluded. Of the remaining 533 cases, further case exclusions included 36 TBM cases with concurrent PTB and 17 PTB cases with lymph node enlargement / pleural pathology/ abnormal CT/MRI. Finally, 333 patients were classified as clinically confirmed TBM cases and 157 patients were classified as clinically confirmed PTB cases. Among these cases, those with negative cultures, HIV positive results and age ≤ 16 years of age at the time of diagnosis were excluded. Thus a total of 53 acid fast bacilli (AFB)/culture positive TBM cases and 72 AFB/culture positive PTB cases were selected, among which 10 archieved cultures for each group cases were processed for further analysis in each group. The study flow design is mentioned in Figure 1.

• Download figure
• Open in new tab

Figure 1.

Study flow diagram for participant recruitment and experimental analysis

2.3 Sputum and CSF collection

2-3 ml of early morning sputum was collected in sterile sputum containers and immediately refrigerated at 4-8 °C until use. All samples were processed within 24-48 hrs after collection. CSF was collected by standard lumbar puncture techniques. 3-5mL of CSF was collected in sterile vaccutainer tubes from TBM patients admitted to the inpatient ward of CIIMS. Detection of AFB bacilli in both sputum and CSF was confirmed using Ziehl-Neelsen (ZN) staining and culturing in BacT/ALERT 3D automated culture system.

2.4 Culturing

2-3mL of sputum and CSF samples were cultured in Middle brook 7H9 liquid medium (Difco, USA) along with oleic acid, albumin, dextrose and catalase (OADC, Biomeriux, France) enrichment and antibiotic supplements in BacT/ALERT culture bottles (Biomeriux, France). These were incubated at 37°C in the BacT/ALERT 3D system (Biomerieux, France) for 28-35 days

2.5 Gas chromatographic analyses of Fatty Acid Methyl Esters (GC FAME)

For GC-FAME, 2-3 mL broth from positive cultures of both strains (n=10, each strain) were subcultured into a 100 mL flask containing Dubos liquid media (Hi Media Lab, India), supplemented OADC enrichment (BD Biosciences, USA) and 0.5 % glycerol and incubated for 10-15 days at 37°C until growth reached O.D₆₀₀ 1.0. The fatty acids were extracted and methylated to form fatty acid methyl esters (FAME) as per procedure described elsewhere [11]

2.6 Whole genome shotgun sequencing

Whole genome shotgun sequencing of selected MTB isolates (S3 and C3 group) was performed on an IlluminaMiSeq platform with more than 100X sequence coverage. For microbial identification, DNA was amplified with a 27F primers set and sequenced using ABIBigDye terminator chemistry. The sequences were BLAST against nr data bases. The pair-end sequencing library was prepared using illuminaTruSeq DNA Library Preparation Kit. Library preparation was started with gDNA fragmentation of 1µg DNA, followed by pair-end adapter ligation. The ligated product was purified using 1X Ampure Beads. The purified ligated product was subjected to size-selection at ∼400-650 bp. The size-selected product was PCR amplified as described in the kit protocol. The amplified libraries were analyzed in Bioanalyzer 2100 (Agilent Technologies) using High Sensitivity (HS) DNA chip as per manufacturer’s instructions. After obtaining the concentration of the library with Qubit Fluorometer and its mean peak size of Bioanalyzer profile, 10 pM of the library was loaded onto MiSeq for cluster generation and sequencing. Sequencing was done using the 2 x 250 bp (500 cycles) chemistry to generate the raw data.

2.7 Genome Analysis

After sequencing, the genome was submitted to the Rapid Annotation using Subsystem Technology (RAST) server for functional annotation and genome comparison [12]. The genome data was submitted to National Center for Biotechnology Information (NCBI) with C3 and S3 reference number CP023170 and CP023169, respectively. Orthologous Average Nucleotide Identity Tool [13] was used to measure the overall similarity between four of the genomes (MTB strain H37Rv, CCDC5079, C1 and S1 strain). Mauve software was used for genome alignment and finding nucleotide level differences between the two genomes [14]. The genomes were further compared with a reference genome i.e. H₃₇Rv (accession no. NC_018143) strain for S3 and CCDC5079 (accession no. NC_021251) for C3 using Briggs software, where H₃₇Rv & CCDC5079 were taken as the reference genome [15]. Pfam was employed for domain analysis of the exclusive proteins found in each genome, for which the protein sequence of each protein in fasta format was submitted to the pfam server [16]. PE-PGRS and PE-PPE sequences were manually extracted, filtered, sorted and aligned using Geneious software [17]. Geneious software was also used for finding the least conserved region in the 4 MTB strains displaying deletion and SNP variation. Determination of 3D structure of the protein was carried by Swiss-Model server (homology based modelling) and visualized by UCSF Chimera using pdb file generated by the Swiss-model server [18-19]

2.8 Monocyte preparation and MTB infection

The human monocytic leukemia cell line, THP-1 was obtained from National Center for Cell Sciences (NCCS) Pune, India and cultured in RPMI 1640 (Gibco) supplemented with 2mM L-glutamine and 10 % fetal bovine serum (Sigma). Monocytes were exposed to respective MTB strains at a multiplicity of infection (MOI) in 1:1 ratio. After 2 hours of infection, the cells were gently washed three times with RPMI 1640 and treated with gentamycin (50ug/mL) to kill extracellular bacteria followed by centrifugation at 600 × g for 10 min to selectively spin down cells. Cells were resuspended in complete medium and cultured for 48 hours. Cell supernatant was collected at 48 hours and centrifuged for 10 min at 1000 g to get rid of cells and debris. The supernatant was stored at −20°C until assayed for cytokines. Each experiment was repeated in triplicate.

2.9 Cytokines estimation

Cytokines (IL-6, TNF-α, and VEGF) were assayed in cell supernatants by quantitative ELISA assay (Bender Med Systems, USA) as per manufacturer’s instructions.

2.10 Experimental mice

Inbred female BALB/c mice aged 6–10 weeks were procured from the National Institute of Immunology, New Delhi, India. Animals were housed in a Biosafety Level 3 facility at the JALMA, Agra, India. Before experiments all animal were acclimatized to housing condition for 14 days and fed with water and food under asceptic conditions.

2.11 Development of MTB infection and estimation of mycobacterial burden

For development of TB infection, two groups of experimental mice (n=12) were infected intravenously through the tail vein with 2×10⁷ CFU with both MTB strains. A control group of mice (n=10) without infection was separately maintained. Mice from each group (n=4) were sacrificed at 30 and 50 days after infection for CFU analysis and histopathology as per protocol described elsewhere [20].

2.12 Statistical Analysis

The frequencies (percentage) of demographic, clinical and risk factors were measured on a nominal scale. Comparison between disease groups were performed using the Chi square test in MedCalc statistical software (version 10.1.2.0) and a difference with p<0.05 was considered to be significant. All experimental results are expressed as mean ± standard deviation (SD) and presented on linear scale except for CFU, where logarithmic scale has been used. All statistical analysis were performed using Prism 6 version 6.01, GraphPad software, (San Diego, CA). Difference in the cytokines level, CFU analysis between experimental and control group was determined by t test. Test for proportion was used for comparison of total mortality in the experimental and control group. P value < 0.05 was considered statistically significant

3.1 Baseline characteristics

Out of 1458 eligible participants, 323 cases satisfying the primary inclusion criteria were considered for the study. The baseline characteristics of the study population is presented in Table 1. The median age of PTB patients (58.8 years) was lower than that of TBM patients (63.9 years) (p < 0.001). We also examined the age distributions stratified by TB type. Our results confirmed the tendency for TBM and PTB to occur at age ≥ 40 years. The overall male to female ratio of TB cases was 1.7 (302/ 178). For TBM and PTB patients, the male to female ratio was 2 (216/107) and 1.2 (86/71) respectively. The difference was however not statistically significant. History of TB in the past was more common among PTB patients (5.13% vs. 1.2%). The proportion of patients who had a history of contact with a known case of TB (26.8% vs. 6.8%) was significantly higher among PTB patients. A significantly higher proportion of PTB cases had a cough and fever as compared to TBM cases (P<0.0001). On the other hand, a higher proportion of TBM cases had headache, neck stiffness, seizures, limb weakness and reduced level of consciousness (P<0.01). The mean duration of illness was also significantly higher in TBM cases as compared to PTB cases (P<0.0001). As compared to PTB patients a significantly higher proportion of TBM patients had concomitant diabetes mellitus (3.8% vs. 9.6%) (p < 0.05). A higher percentage of PTB cases had malignancies as compared to PTB cases, which did not reach statistical significance. The proportion of patients who were hypertensive (17.9% vs. 7.6%) was significantly higher among the TBM cases as compared to PTB cases (p < 0.05 ). The proportion of patients who were either currently smoking or ex-smokers was significantly higher among PTB patients (54.1%) compared to TBM patients (23.8%) (P < 0.0001). Similarly, a higher proportion (i.e. 34.4%) of PTB patients were either currently consuming alcohol or former alcohol drinkers (Table 1), and this difference was statistically significant (p < 0.05)

3.2 MTB isolates C1 and S1 exhibited differential growth pattern and fatty acid composition

Both C3 and S3 showed a difference in growth patterns. C3 isolates were associated with comparably higher growth times (43 days) as compared to S3 (29 days) which demonstrated rapid growth in BacT/ALERT 3D system (Figure 2a). The positive cultures from system (n=10, each) were then collected and used for GC-FAME analysis to study the unique fatty acid profiles of each strain. FAME profiles generated through GC-MS, subsequently compared using a microbial identification system are represented in Figure 2 b. Both MTB strains were associated with distinct FAME profiles. S3 showed significantly (P < 0.0001) higher levels of 10Me-18:0 TBSA FAME, with mean retention time of 15.955±5.27, compared to C3 (8.83±2.97). All C3 strain were associated with higher expression profiles of FAME 18:1w9c (17.555±7.35) compared to S3, although the variations were not significant statistically (Figure 2 c).

• Download figure
• Open in new tab

Figure 2:

a) Growth pattern of two MTB isolates form CSF (C3) and Sputum (S3) samples cultured in BacT/Alert 3D system b) GC-FAME profiles of two MTB isolates along with fatty acid and alcohol designation. Arrows indicate differences in FAME profile of CSF isolate from sputum isolate. c) Characteristics of alcohol, fatty acid after GC FAME analysis associated with of MTB sputum and CSF isolates

3.3 Whole genome sequencing showed distinct genomic variation in derived from TBM and PTB cases

Total high quality data of 462 Mb and 1.08 Gb were generated for MTB S3 and C3 isolates using Illumina Miseq technology respectively. The mapping was performed on both isolates against MTB genomes using bwa 0.7.5a. This resulted in 1005066 mapped reads for C3 and 2518638 mapped reads for S3 with more than 99% of genome coverage on respective mycobacterium genomes (Table 2). The total gene sequences were extracted from consensus using the .gff file (from NCBI database) with the help of in-house perl scripts. This was followed by synteny analysis between C3 and S3. The C3 genome was identified as having a total of 4,406,296 bp with 4,204 genes and the S3 strain with 4,411,638 bp and 4,111 genes (Table 2). The mapping and alignment of the genome with representative genomes shows that C3 is 99.93% similar to MTB CCDC5079 (East—Asian/ Beijing family) and S1 is 99.96% similar to MTB H₃₇Rv (Euro-American) (Figure 3a). The Blast analysis carried out by Brigg clearly indicate the high level of similarity existing between our and representative genomes (Figure 3b). The genes annotated by RAST subsystem shows differences in number of annotation (supplementary table 1) of which 24 genes can be seen in C3 and S3 out of which 9 genes were exclusively present in C3 strain and 17 genes exclusively present in S3 strain (Table 3). Comparative analysis in Mauve assist in highlighting the region where major differences in the genome exist (Figure 4). Phage PhiRv1 genes (phage major capsid protein, phiRv1 prohead protease, probable phiRv1 phage protein, phage associated DNA helicase, putative phiRv1 phage protein, probable phiRv1 integrase and partial REP13E12 repeat protein) were found to be present in S3 but absent in C3 strain (Figure 4a). Five CRISPR genes (Cas2, Cas1, Csm6, Csm5 and Csm4) were also located in S3 which were absent from C3 strain (Figure 4a). A large deletion in C3 genome was also found at two of the locus where PE-PGRS virulence associated protein was present in S3 (Figure 4b).

• Download figure
• Open in new tab

Figure 3.

a) shows heatmap generated with OrthoANi values calculated from the OAT software. It clearly shows the genome closeness between C3 and CCDC5079 strain of MTB and S3 and H37Rv strain of MTB. The tree also shows that the two strain under study belongs to two different lineages of MTB. b) BLAST analysis of 4 Mycobacterium tuberculosis strains through BRIGG. Strain H37Rv was taken as reference genome. The green ring represent H37Rv vs S3, blue represent H37Rv vs C3 and red ring represent H37Rv vs CCDC5079

• Download figure
• Open in new tab

Figure 4.

shows the comparative analysis between strain C3 and S3 at gene level generated through progressive alignment using Mauve software. a) indictes the genomic region where phage and CRISPR gene are present in S3 but not in C3. b.) The large deleted region in C3 is shown at the locus where PE-PGRS genes are present in S3. c) shows region where KASIII enzyme in present in C3 but absent in S3. d) variation of PE-PGRS and e) PE-PPE element existing in the two genome of M. tuberculosis strain C3 and S3. Geneious software was used for extraction, sortment and alignment of these elements. Finally the plot was generated using MS-Excel. More variation exist in case of PE-PGRs element than PE-PPE elements

3.4 Strain variations occur in membrane forming and membrane associated proteins

Genetic differences between the two strains were revealed when it was discovered that 3-oxoacyl-[acyl-carrier-protein (ACP)] synthase III was located in C3 but absent in S3 genome (Figure 4 c). Analysis in pfam confirmed that the two domains present in the protein belongs to ACP synthase III family. A back search in the genome through RAST confirmed that no other copy of this gene exists in the genome of either strains. Swiss-model server predicted the 3-dimensional structure of the MtfabH (3-oxoacyl-[acyl-carrier-protein (ACP)] synthase III) protein which in comparison with other MtfabH protein, shows 100% similarity (Supplementary Figure S1). When lipoproteins present in both the genomes were compared with the help of Pfam server, we discovered that lipoprotein LppE was truncated at the N-terminal in C3 and both the protein of LppA (Lipoprotein A) were also truncated in C3 at N-terminal (Supplementary Figure Figure S2 a). One of the desaturase related genes was found to be deactivated in the C3 genome by the insertion of the mobile element (Supplementary Figure S2 b).

3.5 PE-PGRS and PE-PPE are hotspot sites for mutations

The number and length of the PE-PGRS and PE-PPE proteins vary with the species and the strain. The plot shows that a large a amount of variation exists between both genomes in the case of the PE-PGRS element, with much less variation seen for PE-PPE (Figure 4d & e). S3 consisted of a PE-PGRS element of longer length than C3. Comparison of all the four MTB genome strains through geneious software showed that the site of PE-PGRS and PE-PPE element is a hotspot site of mutation with a large amount of deletions, single nucleotide variation (SNP) and insertions present in this region (Supplementary Figure S3).

3.6 Infection in THP-1 cell lines by both C3 and S3 showed a differential cytokine pattern

To study differential inflammatory response, cytokine expression profile induced by both C3 and S3 strains was studied using the THP-1 cell line model. Figure 5 abc shows mean cytokine levels in cell supernants of THP-1 cell lines after infection with both MTB strains. The C3 strain induced significantly higher levels (P>0.05) of inflammatory cytokines including TNF-α and VEGF compared to the S3 strain. Both strains were, however associated with similar expression levels of IL-6

• Download figure
• Open in new tab

Figure 5:

Cytokine profile a) TNF, b) VEGF and c) IL-6 in THP cell supernatants induced by C3 and S3 Strain. Monocytes were exposed with MTB at a multiplicity of infection (MOI) of one bacterium per cell ie. in1:1 ratio. Cell supernatant was collected at 48 hours and assayed for cytokines. Each experiment were repeated in triplicates. * represent statistically significant (P>0.05) and ** highly significant values (P>0.001). d & e)

3.7 Infection by the C3 strain in BALB/c mice was associated with more pronounce extra pulmonarydissemination and TBM development compared to S3 strain

To study pathogenesis and strain specific dissemination of both MTB strains, groups of mice were sacrificed at 30 and 50 days after infection and organs were isolated to examine lung and brain pathology. After isolation, organs were first visually examined for observation of amount of edema/swelling in the brain and granulomatous lesions in the lungs and brain (figure 6 abc). Mice infected with the CCDC5079 (C3) strain showed prominent swelling/edema of the brain in the left cerebral hemisphere at 30 days, which was further increased with progression in infection compared to control mice. Visual examination of brain of mice infected with the H₃₇R_v strain however revealed no apparent abnormality with its morphology similar to the control mouse group. Examination of the lungs of mice infected with both MTB strains showed chronic pathology with multi-lobed lungs observed at 30 days. Lungs of mice infected with H₃₇R_v (S3) strain showed progressive lung pathology with visibly evident granulomatous lungs at 50 days (6a). On the contrary, lungs of CCDC5079 infected mice showed gross reduction in pathology compared to H₃₇R_v at 50 days (6b). No abnormality was observed in lungs and brains of control mice (a). Histopathological examination of the brain sections of both strain groups showed swelling of neurons along with lymphocytic infiltration which was progressively more prominent at 50 days in the CCDC5079 strain infected mice. Edema of brain parenchyma was also evident in brains of infected mice at both given time points after infection compared to H₃₇R_v strain and control group (6b&c) which showed normal brain pathology with no evident infiltration and edema development at both 30 and 50 days. Examination of the lung sections revealed chronic pathology with high lymphocytic infiltration and an alveolar space mostly filled with inflammatory cells and edematous fluid in lung parenchyma of the pulmonary strain at 30 and 50 days post infection. Mice infected with CCDC5079 strain showed similar lung pathology at 30 days, but were associated with gross reduction in lung pathology with low infiltration and a largely preserved alveolar space with fewer cells filled with edematous fluid with progress in infection towards 50 days (6b&c).

• Download figure
• Open in new tab

Figure 6.

Representative photographs & histological examination lungs and Brain tissue of BALB/c mice 30 days and 50 days of a) control & mice post infection with b) MTB H₃₇R_v (S3) and c) MTB CCDC5079 (C3) strain. Mice (n=4) were sacrificed 30 days and 50 days after intravenous infection with MTB strain. Lungs and brains of both mice group were examined visually as well by histological sectioning for gross pathological changes compared to control. Mice infected with S₃ strain were associated with chronic lung pathology with high lymphocytic infiltration at both time points after infection. on contrary mice with infection of C3 strain showed chronic brain pathology with development of edema and swelling of neurons and lymphocytic infiltration. Both Strains of mice were associated with different organ pathology. Arrows indicate gross lung and brain pathology mice with MTB infection. Bacillary load in d) lungs and e) brains of mice infected with and C3 strain. 30 and 50 days post infection, mice (n=4) were sacrificed and organs were isolated, homogenized and serially diluted. 10 fold dilution of lung and brain homogenates were then were plated onto Middlebrook 7H11 medium with 0.05% Tween-80 and glycerol.. Bacillary burden in both organ are represented as log₁₀ CFU. (**) represent significant (p<0.05) and (***) represents highly significant (p<0.001) values. f) Indicates % mortality rates associated with S3 and C3 strain post infection in mice

CFU analysis of both organs revealed significantly (P<0.05) higher bacterial load in the lungs (7.65±0.2) of H₃₇R_v strain infected mice at 30 days post infection, compared to CCDC5079 strain (7.2±0.2) (Figure 6d). Analysis of CFU load in the brain showed significantly higher (P<0.001) mycobacterial load associated with mice infected with CCDC5079 compared to H₃₇R_v strain at both given time points after infection (Figure 6e). Analysis of mortality rates associated with both strains revealed, significantly high mortality (60%) rate in CCDC5079 strain compared to H₃₇R_v (20%) post infection (Figure 6f)