Whole genome sequence analysis

A small nuclear family from Northeast Brazil was identified with leprosy cases in three generations over a two-year period (Fig 1A). Unusually, monozygotic twins of age 22 months were both affected by leprosy with nearly identical distribution of skin lesions. The early age of onset for the twins together with the high prevalence of leprosy in the family suggested that host genetic factors may be involved in the familial clustering of the disease. Genomic DNA was obtained for six family members and used for whole genome sequencing (WGS; Fig 1B). The mean base coverage of all samples was 31±12 fold, with 91.5% of the genomes covered at least 10-fold (S1 Table). In total, nearly 8.4 million single nucleotide variants (SNVs) and short indels, including 44,679 coding and splice-site variants, were identified in the six family members. To systematically identify candidate leprosy susceptibility variants, we applied seven variant filtering strategies based on allele frequencies, mode of inheritance and allele distributions within affected family members according to age-at-diagnosis (S1 Fig). Next, variant-level and gene-level metrics were used to prioritize variants according to their in-silico predicted impact on protein function. In the final screening step, variants were prioritized according to the physiological function of their tagged proteins (see Materials and Methods).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Missense variants in LRRK2 and NOD2 detected in a family with early onset leprosy in monozygotic twins.

(A) Timeline of leprosy diagnosis among the affected family members, indicating the sample ID, leprosy subtype based on WHO classification and age-at-diagnosis. No patient developed leprosy reactions during treatment and a five-year follow-up. (B) Pedigree of the studied family indicating leprosy phenotype and genotypes of candidate variants LRRK2 N551K (rs7308720), LRRK2 R1398H (rs7133914), and NOD2 R702W (rs2066844) in individuals with whole genome sequencing (WGS) data. Men and women are represented by boxes and circles, respectively. Leprosy patients (regardless of the clinical subtype) are indicated by filled symbols, while an unknown phenotype is indicated by a symbol with diagonal stripes. Monozygosity is represented by a horizontal line linking siblings. PB: Paucibacillary; MB: Multibacillary.

https://doi.org/10.1371/journal.ppat.1011260.g001

Considering the low prevalence of leprosy and assuming recessive models of inheritance, we selected a minor allele frequency (MAF) of < 20% in any of the HapMap populations as cut-off (S1 Fig) [15,16]. We detected a total of 30 segregating non-synonymous and frameshift indels variants in 24 genes as leprosy risk candidates (S2 Table). Of these, five SNPs were predicted in-silico to encode protein-damaging variants: rs7271712 (T397M) in SLC17A9, rs7308720 (N551K) and rs7133914 (R1398H) in LRRK2, rs61740826 (C406Y) in ZNF678 and rs2229531 (V200M) in ACP5 (S2 Table). The twins and their father were homozygous for N551K in LRRK2; and only the early onset twins were homozygous for LRRK2 R1398H, SLC17A9 T397M, ZNF678 C406Y and ACP5 V200M. Among these variants, LRRK2 N551K and R1398H had previously been shown to impact on LRRK2 protein activity and are established protective factors for CD and PD [7,17]. Conversely, the remaining three genes carrying homozygous missense variants in the twins–SLC17A9, ZNF678 and ACP5 –had no known link to infectious diseases. Hence, the two LRRK2 variants were considered the top candidates contributing to the early onset leprosy phenotype.

Under dominant models, 40 SNVs and Indels with MAF lower than 10% were identified as leprosy risk candidates, including eight variants prioritized as protein-damaging in silico (S1 Fig and S2 Table). Of the genes tagged by novel mutations only the CR1 gene had prior evidence of a common variant impacting on human infectious disease risk, including leprosy [18,19]. The novel CR1 E1674G amino acid change may deserve further attention in future studies. In addition, one variant (NOD2 R702W, rs2066844) was a co-dominant risk factor for CD [20]. Variants near or within NOD2 have previously been associated with leprosy and T1R [7,21]. A follow-up of NOD2 variants revealed three additional amino acid changes that had been excluded in the filtering approaches for segregation in the family: P268S, A612T and A725G (S2 Fig). The A725G variant was likely benign and there was no strong evidence for a risk effect of this amino acid mutations in common immune or infectious disease. In contrast, NOD2 612T, which was present in the affected father and grandmother but not in the twins, and P268S, which was considered to be too common to have a dominant effect on leprosy risk, had been associated with risk of CD [17,22].

We also searched for deletion structural variants (DSVs) that might contribute to leprosy susceptibility in the family (S1 Fig and S3 Table). In the DSV analysis, we identified 55 deletions with length ranging from 0.9 Kb to 43.4 Kb that overlapped coding regions of protein-coding genes. From this, four DSVs passed filtering for segregation in the family. DSVs overlapping FSTL4, C18orf32 and PARVB were found in the three generations of the family where the affected individuals were heterozygous for the variants (S3 Table). Only the twins and their affected father carry the fourth candidate DSV, which overlapped with the C9orf50 gene. There was no known involvement of these genes in mycobacterial or immune-mediated diseases.

Since both LRRK2 and NOD2 had been implicated in susceptibility to leprosy, CD and PD, and the corresponding proteins interact in-vivo in Paneth cells, we considered these two genes high priority candidates for exerting a joint effect on early onset leprosy susceptibility [10,23]. This choice did not exclude the possibility that additional susceptibility variants segregated in the family. LRRK2 N551K and R1398H have global MAFs of 8.7% and 8.5%, respectively. However, their frequencies and linkage disequilibrium (LD) pattern vary among populations. In South Asians, both variants are of low frequency (MAF 4%) and present strong LD (r² = 1), while in American and African populations they are more common (MAF between 14% and 16%), but with lower LD (r² = 0.66 and 0.18, respectively) (S4 Table). NOD2 R702W has global MAF of 2.6% and its MAF ranges from 0% in East Asian to 4.3% in European populations (S4 Table). Ancestry estimated based on principal component analysis of the family members of the present study and unrelated individuals from the five HapMap populations suggested a high African (AFR) ancestry for the grandmother (ID2) and shared American (AMR) and AFR ancestry for the remaining family members including the twins (S3 Fig). A reliable estimate of the genetic ancestry of family members is necessary to obtain the probability of encountering the NOD2 and LRRK2 genotypes observed in the family. Based on the allele frequencies in the AMR or AFR population and assuming complete LD between the two LRRK2 variants, the probability of carrying the LRRK2 and NOD2 genotype combination of the twins (LRRK2 N551K/R1398H homozygous + NOD2 R702W heterozygous) was estimated at approximately 0.09%-0.10% in AMR and 0.03% in AFR individuals (S4 Table). Applying an approximate r² = 0.5 between LRRK2 N551K and R1398H for both population leads to an approximate estimate of 0.05% in AMR and 0.01% in AFR populations. Incidence of leprosy in the Northern Brazilian study population is 1.6x10^-4 (Brazilian Health Ministry, Leprosy Epidemiological Record 2022). Countrywide, only 3% of all incident leprosy cases are children younger than 4 years. Combining genotype frequency with age-adjusted incidence suggests that the likelihood of finding additional early onset cases with the same genotypes can be estimated at approximately 1x10^-9.

LRRK2 variants affect ROS production and apoptosis in response to mycobacteria

To functionally validate the findings from the WGS analysis, we explored the impact of the N551K and R1398H LRRK2 variants on the cellular response to mycobacteria (Fig 2). The human and mouse LRRK2 proteins are highly conserved with approximately 90% amino acid sequence identity [24]. Hence, we applied CRISPR/Cas9 technology to knock-in the homozygous candidate LRRK2 variants in mouse RAW264.7 macrophages either individually or as double mutants (DM). A Lrrk2 knock-out (KO) cell line was also generated as control. The expression of LRRK2 wild-type (WT) and corresponding mutant proteins in RAW264.7 cells is shown in Fig 2A. Oxidative burst, the generation of reactive oxygen intermediates (ROS), is an important mechanism by which intracellular mycobacterial growth is controlled [25]. Compared to LRRK2 WT, the LRRK2 R1398H mutation significantly reduced the production of ROS by RAW264.7 cells in response to live M. leprae infection (P < 0.001 at 1-2h post-infection [p.i.], P < 0.01 at 4-6h p.i.) while the N551K variant had no significant effect (Fig 2B). Interestingly, the LRRK2 DM had a similar effect on ROS production as the R1398H mutation (P < 0.001 at 1-2h p.i., P < 0.05 at 4h, p.i.), which suggested that the effect of the N551K/R1398H double mutation was mainly due to R1398H (Fig 2B). Consistent with a previous report, relative to LRRK2 WT expressing cells, the production of ROS was significantly lower in LRRK2 KO cells in response to infection (P < 0.01 at 1h p.i., P < 0.001 at 2-6h p.i., Fig 2B) [26,27]. We also investigated if mutant LRRK2 affected ROS production in response to Bacillus Calmette-Guérin (BCG) and we obtained similar results to those in response to M. leprae infection (Fig 2B). From these results, we concluded that BCG is a valid surrogate for M. leprae to study the effect of LRRK2 and NOD2 mutations on the host response to mycobacterial challenge.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Effects of LRRK2 variants on respiratory burst and apoptosis in response to pathogens.

(A) Protein levels in RAW264.7 cells of LRRK2 WT, LRRK2 mutants [N551K, R1398H and DM] and no LRRK2 protein (KO). The result is a representative of three independent experiments done by Western blot analysis. The expression of GAPDH is shown as a loading control. (B) Kinetics of reactive oxygen species (ROS) production in cells with different Lrrk2 genotypes after stimulation with live M. leprae (left panel) or BCG (right panel). Results present the mean ± SE for three independent experiments (done in triplicates). (C-E) Effect of LRRK2 mutations on apoptosis in response to BCG compared to uninfected cells. (C) Flow cytometry for apoptosis, which is a representative of two independent experiments (done in triplicates) with similar results. D-E Percentages of live cells (D) and total apoptotic cells (E) were calculated in uninfected (grey bars) and BCG-infected cells (black bars) expressing different LRRK2 variants. (D-E) Data is presented as mean ± SE (n = 6). The two experiments are represented as circles and squares. ** 0.001 ≤ P < 0.01; * 0.01 ≤ P < 0.05.

https://doi.org/10.1371/journal.ppat.1011260.g002

Apoptosis is part of the innate immune response against mycobacteria [28]. Mutant LRRK2 variants or loss of LRRK2 expression have been linked to apoptotic cell death [29–31]. Hence, we tested if LRRK2 N551K, R1398H or DM affected BCG-induced apoptosis. As shown in Fig 2C–2E, in uninfected cells, mutant LRRK2 or absence of LRRK2 protein had no effect on naturally occurring apoptosis by RAW264.7 cells. However, following infection with BCG, apoptosis was significantly increased in LRRK2 KO cells compared to LRRK2 WT cells. In contrast, cells expressing LRRK2 R1398H displayed significantly reduced BCG-induced apoptosis compared to LRRK2 WT cells while LRRK2 N551K had no significant impact on the extent of BCG-induced apoptosis. Finally, expression of LRRK2 DM reduced BCG-induced apoptosis to a similar level as LRRK2 R1398H, suggesting that the effect of the double mutation on BCG-induced apoptosis is mainly due to the R1398H mutation (Fig 2C–2E). These results demonstrated how the same LRRK2 amino acid substitution can score as gain or loss of function variant depending on the read-out assay employed.

To investigate if LRRK2 heterozygous R1398H affects the cellular function, we generated a RAW264.7 cell line with heterozygous R1398H. We assessed the impact of homozygous 1398H/H (HOM) and heterozygous 1398R/H (HET) on apoptosis and ROS production in response to BCG challenge as compared to cells expressing WT LRRK2. In the absence of BCG stimulation, there was no difference across the genotype groups for ROS production (S4A Fig, 0h) and total apoptosis (S4B Fig, non-infected). Upon BCG challenge, we found strong reduction of ROS production (S4A Fig, 2-6h) and apoptosis (S4B Fig, BCG-infected) in HOM cells compared to WT cells. In direct comparison, HET and WT cells displayed no significant differences. However, a trend test across genotypes was significant suggesting a level of co-dominant control for the assay read-outs (S4 Fig).

Effect of LRRK2 variants and NOD2 R702W on ROS production and apoptosis in response to BCG

Since the twins and their father are heterozygotes for NOD2 R702W (Fig 1B), we investigated if NOD2 R702W and LRRK2 DM have a synergistic impact on ROS production and apoptosis. As mouse Nod2 does not carry an arginine (R) at position 702, it was not possible to introduce the human mutation into the mouse gene. To study LRRK2-NOD2 protein interactions, we opted to express the human NOD2 WT and mutant NOD2 proteins in the genome-edited RAW264.7 cells. Hence, plasmids expressing flag-tagged NOD2 WT, its variant R702W or an empty vector were introduced into RAW264.7 cells carrying LRRK2 WT, LRRK2 DM or not expressing LRRK2 protein (KO). No significant difference in NOD2 WT and NOD2 R702W protein expression was observed across cell lines (Fig 3A). Transfected cells were infected with BCG (MOI 10:1) and kinetics of ROS production were established. Overexpression of both NOD2 WT and NOD2 R702W proteins increased the ROS production in all three LRRK2 variant cell lines upon infection with BCG (Fig 3B). When overexpressed in LRRK2 WT cells, NOD2 R702W mediated significantly lower BCG-induced ROS production compared to NOD2 WT (P < 0.001 at 2-6h p.i.). When introduced in LRRK2 DM cells mutant NOD2 also mediated a lower induction of ROS (P < 0.05 at 2-4h p.i.). However, the R702W effect in LRRK2 WT cells was larger compared to LRRK2 DM cells (Fig 3B). When overexpressed in LRRK2 KO cells, NOD2 R702W did not significantly reduce ROS production relative to the NOD2 WT (Fig 3B). Taken together, our results showed that LRRK2 and NOD2 jointly modulated ROS production following BCG infection. However, biological significance of the effect will require additional study. Contrary to ROS production, overexpression of NOD2 WT or NOD2 R702W had no significant impact on apoptosis on any LRRK2 background (Fig 3C). A representative flow cytometry figure of the apoptosis experiments is presented in S5 Fig.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Effects of LRRK2 and NOD2 variants on respiratory burst and apoptosis in response to BCG infection.

(A) RAW264.7 cells carrying LRRK2 WT, LRRK2 DM and no LRRK2 protein (KO) were transfected with plasmids expressing NOD2 WT or mutant NOD2 (R702W). One representative result of three independent experiments is shown. (B) Effect of NOD2 WT and R702W on reactive oxygen species (ROS) production after infection with BCG on the background of different LRRK2 genotypes. BCG-induced ROS production in cells transfected with NOD2 WT, NOD2 R702W or an empty plasmid are presented for cell lines with LRRK2 WT (Left panel), LRRK2 DM (middle panel) and LRRK2 KO (right panel). The graphs present one representative experiment (mean ± SD) of three independent experiments, each one done in triplicates. (C) Effect of LRRK2 and NOD2 variants on apoptosis in response to BCG. Percentage of total apoptotic cells, including cells with early and late apoptosis, was calculated in uninfected and BCG-infected cells, which is shown in the figure. The illustrated result presents one representative experiment (mean ±SD) of two independent experiments (done in triplicates). Significance of difference between LRRK2 WT+NOD2 WT and the genotype carried by the early onset leprosy twins (LRRK2 DM+NOD2 R702W) is indicated in red. B-C *** P < 0.001; ** 0.001 ≤ P < 0.01; * 0.01 ≤ P < 0.05; ns: non-significant. BCG: Bacillus Calmette–Guérin; DM: double-mutant; KO: knock-out; WT: Wild-type.

https://doi.org/10.1371/journal.ppat.1011260.g003

LRRK2 variants and NOD2 R702W reduce NOD2-dependent RIP2 phosphorylation and NF-κB activity

We used co-immunoprecipitation (Co-IP) to determine if NOD2 interacts with LRRK2 in RAW264.7 macrophages and how the LRRK2 and NOD2 variants segregating in the study family affect this interaction. To test the impact of macrophage activation on the possible LRRK2/NOD2 interaction, we used the NOD2 ligand N-glycolyl muramyl dipeptide (MDP) as trigger [32]. Co-IP revealed that transfected NOD2 did interact with endogenous LRRK2, and this interaction was independent of the LRRK2 variant and MDP stimulation but sensitive to the NOD2 variant (Fig 4A). The interaction between LRRK2 and NOD2 proteins was confirmed by co-localization analysis with laser confocal microscopy (S6 Fig). Consistent with Co-IP, these co-localization results demonstrated that NOD2 R702W strongly diminished the interaction between LRRK2 and NOD2 in RAW264.7 macrophages (S6 Fig). To validate these results in human cells, we obtained PBMC from two family members and conducted a co-localization analysis of LRRK2 and NOD2 in monocyte-derived macrophages (MDMs) of the father and one of the twin girls as well as from a control subject (Fig 4B and 4C). MDMs were differentiated with GM-CSF and co-localization was analyzed after 24h of stimulation with N-glycolyl-MDP. In concordance to the results obtained with RAW cells, we observed a lower proportion of LRRK2-NOD2 co-localization in the macrophages from the father and daughter (NOD2 R702W heterozygote), compared to the control subject who was homozygous for the R702 reference allele (Fig 4B and 4C). Combined, our data showed impaired LRRK2-NOD2 interaction due to the NOD2 R702W variant in both murine and human macrophages.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Effects of LRRK2 and NOD2 variants on LRRK2-NOD2 interaction, NOD2-dependent RIP2 phosphorylation and NF-κB activity.

(A) Protein-protein interaction between endogenous LRRK2 and transfected NOD2 by Co-immunoprecipitation (Co-IP). Twenty-four hours post-transfection, cells were left untreated or treated with N-glycolyl MDP (10 μg/ml) for another 24 hours. Cell lysates were prepared and immunoprecipitated with a rabbit monoclonal antibody against LRRK2. Immunoprecipitants were analyzed by Western blot analysis with antibodies directed against LRRK2 and the FLAG tag of the fused FLAG-NOD2. The expression of LRRK2 and NOD2 using the same antibodies was also analyzed in the total lysate, where GAPDH expression was used as a loading control. LRRK2 KO RAW264.7 cell line transfected with a plasmid expressing NOD2 WT was used as a negative control. (B-C) Representative confocal images of LRRK2 and NOD2 co-localization in monocyte-derived macrophages. (B) Co-localization of LRRK2 (green) and NOD2 (red) was identified (yellow, merged) in macrophages of two family members (father and twin) with heterozygosity (R/W) for NOD2 p.702 and a control subject with homozygosity (R/R) for the common reference allele. Nuclei were stained with DAPI (blue). Scale bar represents 20μm in the figures and 10μm in the insets. (C) Quantification of total co-localization between LRRK2 and NOD2 was estimated from 40 cells by LAS X Software (Leica). Significance was calculated using one-way ANOVA, comparing each condition with the control using the Kruskal-Wallis test. Each dot represents a cell, the band in the box plot indicates the median, the box indicates the first and third quartiles and the whiskers indicate ± 1.5× interquartile range. (D) Effects of NOD2 mutation on MDP-induced phosphorylation of RIP2 at Ser 176 in cells with different LRRK2 genotypes. Twenty-four hours post-transfection, cells were left untreated or treated with different concentrations of MDP for another 24 hours. Phosphorylation of RIP2 (p-RIP2) was analyzed by immunoblotting with a specific antibody against phosphorylated at Ser 176 of RIP2. (E) NOD2 expression in the cells from panel D. The displayed results are representative of three independent experiments. The expression of GAPDH is shown as a loading control. (F) Effects of LRRK2 and NOD2 variants on NF-κB activity. The three RAW264.7 cell lines with different LRRK2 genotypes were transfected with NOD2 plasmid (WT, R702W or empty vector) together with NF-κB firefly-Luc plasmid and Renilla-Luc plasmid pRL-TK (internal control). Twenty-four hours after electroporation, cells were left untreated or treated with 10 μg/ml of N-glycolyl MDP for another 24 hours. Cell lysates were subjected to luciferase assays. Results are expressed as relative luciferase activity (fold change), as compared with the luciferase activity of LRRK2 KO cells transfected with empty vector in the absence of N-glycolyl MDP. Results are presented as mean ± SD of a representative experiment (done in triplicate) of three independent experiments. Significance of difference between LRRK2 WT+NOD2 WT and the genotype carried by the early onset leprosy twins (LRRK2 DM+NOD2 R702W) is indicated in red. *** P < 0.001; ** 0.001 ≤ P < 0.01; * 0.01 ≤ P < 0.05; ns: non-significant.

https://doi.org/10.1371/journal.ppat.1011260.g004

NOD2 activation by MDP results in the activation of the obligate NOD2 kinase RIP2 (alias RIPK2) [33]. Upon interaction with NOD2, RIP2 becomes auto phosphorylated at two main sites, S176 and Y474. Phosphorylation of RIP2 results in the recruitment of TAK1 and subsequent activation of NF-κB [33]. We investigated the effect of the NOD2/LRRK2 interaction on RIP2 phosphorylation by Western blot using a RIP2 S176-specific antibody. RIP2 phosphorylation increased after transfection with NOD2 WT and NOD2 R702W constructs in an N-glycolyl MDP concentration dependent manner irrespective of the cellular LRRK2 background (Fig 4D). However, relative to NOD2 WT, RIP2 phosphorylation mediated by NOD2 R702W was substantially weaker irrespective of the LRRK2 variant carried by the cells even though no significant difference in the expression of NOD2 WT and R702W proteins was observed in these cells (Fig 4D and 4E). Compared to LRRK2 WT, LRRK2 DM reduced NOD2-driven phosphorylation of RIP2 (S7 Fig). Strikingly, LRRK2 KO resulted in decreased N-glycolyl MDP-induced RIP2 phosphorylation no matter if the cells were overexpressing NOD2 WT or NOD2 R702W (Figs 4D and S7). Combined, our results identified LRRK2 as important part of the NOD2 signaling cascade and demonstrated that the LRRK2 DM and NOD2 R702W mutations additively reduced RIP2 phosphorylation relative to wild-type proteins.

Next, we performed luciferase reporter assays to detect the combined impact of NOD2/LRRK2 variants on NF-κB activation. Consistent with the effect on RIP2 phosphorylation, N-glycolyl MDP-induced NF-κB activation increased in all three LRRK2 variant expressing cell lines overexpressing NOD2 WT or NOD2 R702W. In both LRRK2 WT and LRRK2 DM cells, NOD2 R702W triggered lower N-glycolyl MDP-induced NF-κB activation compared to NOD2 WT (Fig 4F). Cells devoid of LRRK2 or expressing LRRK2 DM displayed a strong trend of lower NF-κB activation relative to LRRK2 WT cells for both wild-type and mutant NOD2 (Fig 4F). Consequently, LRRK2 DM and NOD2 R702W had a cumulative effect on the reduction of N-glycolyl MDP-induced NF-κB activation and the variants carried by the early onset leprosy twins reduced NF-κB activation by approximately half compared to the wild-type variants expected in the general population (Fig 4F).

Combined effects of LRRK2 DM and NOD2 R702W on cytokine production in response to BCG infection

Next, we asked to what extent LRRK2 DM and NOD2 R702W modulate cytokine/chemokine production. The release of four key mediators (MCP-1, TNF, IL-10, and IL-6) into the supernatant of cell cultures was measured before and after infection with BCG or stimulation with MDP. As expected, infection with BCG induced a stronger cytokine/chemokine response than stimulation with N-glycolyl MDP only (Fig 5). Moreover, the additive effect of NOD2 (WT or R702W variant) overexpression was more pronounced for MDP stimulation. In the absence of transfected NOD2, BCG infection triggered secretion of MCP-1 and TNF while stimulation with N-glycolyl MDP did not, which is consistent with its role as NOD2 ligand. N-glycolyl MDP-triggered secretion of IL-6 remained below the limit of detectability. Release of MCP-1, TNF and IL-6 was substantially reduced by both LRRK2 DM and NOD2 R702W relative to wild-type proteins. There was no significant difference between LRRK2 DM and LRRK2 KO cells, except for the BCG-triggered secretion of MCP-1, which was lower for LRRK2 DM compared to LRRK2 KO (Fig 5A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Effects of LRRK2 and NOD2 on cytokine secretion in response to BCG infection or stimulation with N-glycolyl MDP.

Cell culture supernatant concentrations for MCP-1, TNF, IL-10 and IL-6 in cells expressing LRRK2 WT, LRRK2 DM (N551K+R1398H) or devoid of LRRK2 (KO) and transfected with NOD2 wild-type (WT), NOD2 R702W or empty vector. Secreted cytokines were measured following (A) infection with live bacillus Calmette–Guérin (BCG) or (B) stimulation with N-glycolyl MDP. The concentrations of IL-6 triggered by N-glycolyl MDP stimulation were below the detection limit and are not shown. A-B Results are presented as mean ± SD of a representative experiment done in triplicate of three independent experiments. Significance of difference between LRRK2 WT+NOD2 WT and the genotype carried by the early onset leprosy twins (LRRK2 DM+NOD2 R702W) is indicated in red. *** P < 0.001; ** 0.001 ≤ P < 0.01; * 0.01 ≤ P < 0.05; ns: non-significant.

https://doi.org/10.1371/journal.ppat.1011260.g005

An unexpected observation was the impact of LRRK2 on the release of the anti-inflammatory cytokine IL-10. RAW264.7 cells expressing LRRK2 DM displayed strongly reduced secretion of IL-10, which was not significantly increased following NOD2 overexpression. Strikingly, absence of LRRK2 resulted in an increased release of IL-10 for both BCG and N-glycolyl MDP stimulation relative to either wild-type or mutant LRRK2-expressing cells even in the absence of overexpressed NOD2 (Fig 5). This observation together with the NOD2-independent impact of LRRK2 mutations on apoptosis activity suggested a pronounced immune regulatory role of LRRK2 (Fig 2). Collectively, our data show that NOD2 signaling is dependent on the interaction with functional LRRK2. While these observations deserve additional study, the important observation in the context of the present paper was the additive effect of both LRRK2 and NOD2 variants shared by the twins on the release of major mediators of the immune response.

Ethics statement

All participants, or their legal representative, provided written informed consent independently and agreed to donate specimens for this study. This study was approved by Research Ethics boards of the Pontifícia Universidade Católica do Paraná (CEP 169.382 and 1.709.543) and the Federal University of Piauí (CEP 657.779).

Study participants

A three generational family comprising seven family members–including four leprosy cases–was enrolled from Piauí state, Northeast Brazil. The index case was the father who was diagnosed with multibacillary (MB) leprosy in 2008 at the age of 27 years and a second time in 2011. Following the first leprosy diagnosis of the father, household contact tracing was conducted. As part of this follow-up, both twin girls were diagnosed with paucibacillary (PB) leprosy at the age of 22 months by two independent experienced leprologists. In 2011, the grandmother was diagnosed with PB leprosy at the age of 49 years. All individuals in the family had been BCG vaccinated. No patient developed leprosy reactions during treatment and a five-year follow-up. No additional mycobacterial diseases were detected in any family member. Both the paternal aunt (ID3) and the mother (ID5) of the twins remained unaffected. However, only the mother had been in prolonged contact with the three leprosy cases. Therefore, only the mother was included in the variant filtering approaches as unaffected control.

Whole genome sequencing

WGS was performed for six family members (ID2 to ID7 from Fig 1B) on HiSeq 2500 platform (Illumina) to generate paired-end 150 bp reads at the Genome Quebec and McGill Innovation Centre. Quality assessment of the raw data was performed using FastQC v0.11.4 software (Babraham Bioinformatics; http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). The reads were mapped to human genome reference GRCh37+decoy using the BWA-mem algorithm on BWA v0.7.12 [106]. Mapped reads were sorted according to their genomic coordinate position and PCR duplicates were flagged using Picard v1.134 (Broad Institute, GitHub Repository; https://github.com/broadinstitute/picard). Next, local realignment around indels and base recalibrations were performed using GATK v3.5 [107]. Quality assessment of the mapped reads was performed using QualiMap v2.1.1 [108]. GATK HaplotypeCaller was used to call SNVs and short indels for each sample, followed by GenotypeGVCFs for the six samples together. Variant Quality Score Recalibration from GATK, using default parameters was used to reduce the amount of false positive [109]. Genotypes with genotype quality (GQ) lower than 20 were removed. Finally, the variants were annotated using wANNOVAR (2016) [110].

Deletion structural variants (DSVs) in autosomal chromosomes were detected using Genome STRiP v2.0 SV pipeline with default parameters [111]. For that, a total of 25 high-coverage WGS samples from the 1000G database were included in this step to run together with the 6 samples from the studied family [112]. Next, GATK was used to apply hard filtering to remove low quality deletions as follow: GSELENGTH < 200; GSCLUSTERSEP ≤ 2.0 or NA; GSM1 ≤ 0.5 or ≥ 2.0 or NA; GLINBREEDINGCOEFF < -0.2; GSNONVARSCORE ≥ 13.0; GSDUPLICATESCORE ≥ 0 or DOSAGE_CORRELATION ≥ 0.5; call rate ≥ 80%. Overlap of the DSVs to protein-coding genes was annotated using GeneOverlap command on Genome STRiP.

To identify the population structure of the studied family, principal component analysis (PCA) was performed using PLINK v1.9 [113]. For that, genotypes from VCF files were converted to PLINK format using BCFtools (http://github.com/samtools/bcftools). Autosomal variants were pruned based on LD (window size of 50kb, step size of 5 and variance inflation factor of 1.5) and MAF > 10% as implemented in PLINK. Variant pruning was done using the samples from the 1000 Genomes Consortium (1000G) representing the five super populations worldwide: African/African American, Admixed American/Latin, East Asian, South Asian and European [15]. From the 428,824 pruned variants, A/T and C/G SNPs and variants absent from the studied family were excluded. In total, 237,150 variants were used for PCA analysis including 2,504 unrelated individuals from 1000G and the six family members from the present study.

Candidate variant detection from WGS data

Five filtering steps (A to E) were used to select variants based on A) the variant location within protein-coding genes (coding or splice-site variants), B) its type (missense, nonsense, frameshift indels or splice-site variant), C) inheritance model (dominant or recessive), D) variant frequency in population samples from the database, and E) age-at-diagnosis of the affected family member. To apply filtering step D, we searched the variant frequency in the following population samples from the 1000G and Exome Aggregation consortium (ExAC) databases: African/African American, Admixed American/Latin, East Asian, South Asian and European (Non-Finnish European from ExAC) [15,16]. Using different parameters/thresholds in steps C to D, we implemented seven custom filtering approaches to identify candidate variants (S1 Fig). Filtering approaches #1 to #4 were designed to detect variants in the recessive model, while filtering approaches #5 to #7 to detect variants in the dominant model (S1 Fig). Different thresholds of variant frequencies were used in the four approaches following the recessive model, while in the dominant model, we selected variants that were not reported in the 1000G nor ExAC databases (See S1 Fig). Based on the age-at-diagnosis of the leprosy patients in the family (Fig 1A), in filtering step E we searched for variants present in i) all the affected family members, regardless of the age-at-diagnosis (Filtering approaches #1 and #5 in S1 Fig); ii) the father and the twins, which are the cases younger than 30 years (Filtering approaches #2 and #6 in S1 Fig) and iii) only in the twin girls, that are the early-onset cases in the family (Filtering approaches #3, #4 and #7 in S1 Fig). An additional search for candidate variants segregating in the family was done employing known leprosy genes [10]. We included in this analysis a total of 86 genes previously associated with leprosy in GWAS and/or target association studies up to January 2023 (S5 Table). Filtering steps were applied to identify candidate variants in these genes as implemented in the WGS data, but with a less stringent threshold for variant frequency in databases in step D (S1 Fig). For known leprosy genes, thresholds of MAF < 20% and MAF < 10% in the four populations from 1000G/ExAc were used in the recessive and dominant models, respectively (S1 Fig). Genotype validation of candidate variants was done as described in the S1 Method.

Once candidate variants were identified, variant-level and gene-level metrics based on computational prediction were used to prioritize the variants that are most likely to have an impact on the protein structure and function. For that, we used PolyPhen-2 v2.2.2r398 and CADD v 1.4 as variant-level metrics, and GDI (2016) as gene-level metric [114–116]. We focused on variants that presented the three following criteria: i) it has scaled CADD score ≥ 20 (Top 1% most deleterious), ii) it is a missense variant with PolyPhen-2 HumVar score > 0.446 (possibly or probably damaging), a nonsense variant, frameshift indel or splice-site variant and iii) it is located in a gene with GDI score < 13.84 (medium or low GDI). Among compound heterozygous variants (Filtering approach #4 in S1 Fig), we prioritized genes with GDI score < 13.84 where both variants reached criteria one and two. Linkage disequilibrium (LD) estimates between LRRK2 N551K and R1398H were performed using Haploview software v4.2 based on genotyping data from the five populations from 1000G [15,117]. Estimated frequencies of LRRK2 R1398H and NOD2 R702W combined genotypes were obtained assuming Hardy-Weinberg equilibrium for both SNPs and using MAF from gnomAD database [118].

For detection of candidate DSVs, deletions reported as overlapping with exon, CDS or gene were selected. Then, we applied the same filtering approaches as used for SNV and short indels for the recessive and dominant model (S1 Fig). DSV with call-rate < 80% in the 31 samples were excluded. We searched whether the candidate deletions were present on Database of Genomic Variants (DGV) catalog and kept those that were found in less than 50% of the sequenced samples from 1000 Genomes database (S1 Fig) [15,119].

Genome-editing with CRISPR/Cas9

1. Synthesis of gRNAs: The gRNAs for generation of LRRK2 N551K, R1398H were synthesized by using GeneArt precision gRNA synthesis kit (Thermo Fisher) according to the manufacturer’s instruction. Prior to making gRNAs, 34-nucleotide forward and reverse target DNA oligonucleotides were designed using the CRISPR search and design tool (Thermo fisher) and synthesized (S5 Table). Then the gRNA DNA templates were PCR assembled and gRNAs were synthesized by in vitro transcription. The gRNAs were purified and their concentrations were measured. TrueGuide Synthetic sgRNA for generation of Lrrk2 KO cell line was purchased from Thermo Fisher (Assay ID: CRISPR206078_SGM).

2. Electroporation: One day prior to transfection, RAW264.7 cells were split into a new flask with fresh growth medium such that the cells reach 70–90% confluent the following day. On the day of electroporation, cells were washed with PBS (without Ca²⁺ and Mg²⁺), digested with 0.25% trypsin-EDTA for 8–10 min at 37°C. After neutralization with growth medium, the cells were counted and appropriate amounts of cells (1 x 10⁵ cells per transfection) were transferred to a 1.5 ml microcentrifuge tube. The cells were washed once with PBS by centrifugation at 500g for 5 min. At the same time as preparation of cells for electroporation, 2μg Cas9 protein and 400 ng gRNA were mixed in 10 μl of resuspension buffer R and incubated at room temperature for 10 min. Prepared cells were re-suspended in the buffer R containing Cas9-gRNA complex and 50 pmol of donor HDR templates (S6 Table) was added. Cell mixture was transferred into a 10 μl Neon tip with Neon pipette and electroporation were performed using the parameters as following: pulse voltage 1680 V, pulse width 20 ms and pulse number 1. After electroporation, cells from two Neon tips were immediately mixed into prewarmed 1 ml growth medium in a well of 12-well plate and cultured for 4 days.

3. Restriction fragment length polymorphism (RFLP) assay: Genomic DNA was extracted from RAW264.7 cells transfected with Cas9-gRNA and donor HDR templates. Genomic DNA was then PCR amplified with primers flanking the donor target region (see PCR primer sequences in S6 Table). The amplification was carried out with AmpliTaq Gold 360 master mix (Thermo Fisher), using the following cycling condition: 95°C for 10 min for initial denaturation; 40 cycles of 95°C for 30s, 60°C for 30s and 72°C for 35s; and a final extension at 72°C for 7 min. Then, 1 μg PCR products were digested with 10 U of BstUI at 60°C or AvaI at 37°C overnight and resolved on 1.2% agarose gel.

4. Single cell clone analysis: Single cells were prepared by digestion of cells with 0.25% trypsin-EDTA. Cells were counted and serially diluted to 2 x 10⁴cells/ml, 5 x 10²cells/ml and 5 cells/ml. Next, 200 μl of 5cells/ml was dispensed to each well of 96-well plates using a multichannel pipette. Plates were incubated at 37°C in a 5% CO₂ incubator.

5. Screening knock-in mutation (N551K and R1398H): Genomic DNA was isolated from single clones. The donor target region was PCR amplified with AmpliTaq Gold 360 master mix (Thermo Fisher) (see PCR primer sequences in S6 Table). PCR amplicons were sequenced using standard Sanger sequencing.

6. Screening KO: Cell lysates were prepared from single cell clones and western blot analysis were used to screen knockout clones.

Cell culture and NOD2 transfection

LRRK2 WT, LRRK2 DM or LRRK2 KO RAW264.7 cells were maintained in DMEM medium supplemented with 10% fetal bovine serum and 1% streptomycin-penicillin, and incubated in a humidified atmosphere containing 5% CO₂ at 37°C. The cells were passaged every 3 days. Plasmid pcDNA3.1+/C-(K)DYK-NOD2 encoding Flag tagged human wild-type NOD2 was purchased from GenScript. A plasmid pcDNA3.1+/C-(K)DYK-NOD2 R702W encoding NOD2 R702W variant was generated using a QuickChange XL site-directed mutagenesis kit (Agilent Technologies) according to the manufacturer’s instructions. Primers used for generation of NOD2 R702W mutation were 5’-ggcctggcgccagagcagggcct-3’ and 5’-aggccctgctctggcgccaggcc-3’. The NOD2 R702W mutation was confirmed by Sanger sequencing. Plasmids pcDNA3.1+/C-(K)DYK-NOD2, pcDNA3.1+/C-(K)DYK-NOD2 R702W and pcDNA3.1+/C-(K)DYK empty vector (negative control) were transfected into the three RAW264.7 cells lines using a Neon transfection system (Thermo Fisher Scientific). Briefly, 1 × 10⁶ cells were suspended in 100 μL of buffer R containing 15 μg of plasmid and electroporated at 1,680 V for 20 ms and 1 pulse.

Preparation of BCG-Russia and M. leprae

BCG Russia culture was maintained in middlebrook7H9 medium supplemented with 10% ADC, 0.1% Tween 80, and 0.2% glycerol at 37°C on a roller. On the day of infection, appropriate amount of log-phase BCG-Russia were transferred to a 50 ml conical tube and pelleted by centrifugation at 2500 rpm for 6 min. The supernatant was removed, and the cell pellet was washed once with 1x PBS and re-suspended in complete DMEM medium. To break up large aggregates into single cells, the re-suspended BCG was treated in water bath sonication for 20s x 5 times, followed by passing the BCG through a 22 1/2-G needle 8 times. The remaining bacterial clumps were removed by centrifugation for 5min at a centrifugation force of 100g. Bacterial load was determined by plating serial 10-fold dilutions of BCG on Middlebrook 7H10 agar plate (supplemented with 10% OADC) and counting colonies after incubation for at least 3 weeks. Viable M. leprae was obtained from the National Hansen’s Disease Program, Health Resources and Services Administration, Baton Rouge, LA, USA.

ROS detection

Cells were seeded in 96-well plates at a concentration of 3×10⁴ cells per well and stimulated with IFN-γ (100ng/ml) for 24 hours and then infected with BCG-Russia or M. leprae at a MOI of 10:1. At indicated time points following infection, intracellular ROS was detected using ROS-ID total ROS detection kit (Enzo life science) according to the manufacturer’s instruction. For that, cells were carefully washed with 200 μl/well of 1× wash buffer. Following wash buffer removal, 100 μl/well of ROS detection mix (4 μl of 5mM oxidative stress detection reagent /10 ml of 1× wash buffer) was added prior to incubation of plates in a humidified incubator (37°C, 5% CO₂) for 30 min and reading were acquired at wavelength 488/520nm on a plate reader. The experiment was performed three times, each in triplicate.

FACS analysis of apoptosis

LRRK2 WT or CRISPR/Cas9-edited RAW264.7 cells (mock transfected, transfected with NOD2 constructs or empty vector) were separately cultured at a concentration of 6×10⁵/well in a 6-well plate for 16–18 hrs. The cells were then infected with BCG (MOI 10:1) or left uninfected for 24 hours prior to apoptosis detection. The Annexin V staining was done according to the manufacturer’s instruction (Biolegend). Briefly, the cells were first detached from the culture plates and washed twice with 2mL of 1X azide-free and serum/protein-free PBS at RT. The supernatant was discarded. Then, 0.5μl of Zombie Aqua fixable viability dye (ZA-FVD) was added to 100μL of cells in 1X azide-free and serum/protein-free PBS and incubated in dark for 30 minutes at 2°C. After incubation, the cells were washed twice with 1X azide-free PBS+0.2% BSA. The cells were washed once with 1X azide-free PBS+0.2% BSA and then once with 1X Annexin V Binding Buffer (BD Biosciences). The cells were resuspended in 1X Binding Buffer at 2x10⁶ cells/ml. Next, 5μl of Annexin V-APC were added to 100μL of the cell suspension and incubated in dark for 15 minutes at room temperature. After incubation, the cells were washed twice with 2ml of 1X Binding Buffer. The supernatant was discarded. The cells were resuspended in 200μl of 1 X Binding Buffer and immediately collected by flow cytometry with BD FACSCanto II (BD Biosciences). The data was analyzed on FlowJo v10.4.2 (FlowJo, LLC) with viability and Annexin V single stains as FMOs.

Western Blot analysis

Equivalent amounts of total cellular lysates were separated on 4% to 12% Tris-Glycine gels (Invitrogen) and electrophoretically transferred to polyvinylidene difluoride membranes (Millipore, Bedford, MA). The membranes were blocked with 5% BSA in TBS-T (Tris-buffered saline-0.1% Tween 20) for 1h at room temperature (RT), and then followed by incubation with primary antibodies overnight at 4°C. A rabbit monoclonal antibody against LRRK2 (Abcam) was used at 1:1,000 dilution. A mouse anti-GAPDH monoclonal antibody (Thermo Fisher) was used at a 1:10,000 dilution. A mouse monoclonal antibody against FLAG (Sigma) was used at 1: 2,500. A rabbit anti-phospho-RIP2 (S176) (abm) was used at 1:1,000. After incubation, membranes were washed 5 times for 5min with TBS-T and were further incubated with appropriate secondary antibodies coupled to horseradish peroxidase for 1h RT. Upon extensive washing, the membrane was developed with enhanced chemiluminescence detection reagents (Bio-Rad), followed by imaging using a ChemiDoc Touch imaging system (Bio-Rad).

Co-immunoprecipitation

For co-immunoprecipitations, 24 hours after electroporation, cells were treated with or without 5μg/ml of N-glycolyl MDP for another 24 hours, and then cells were lysed in Pierce IP lysis/wash buffer supplemented with 1X protease inhibitor mixture (Thermo Fisher scientific). Cell lysates were centrifuged at 13,000xg for 10 min to remove cellular debris. Next, 800μg of the total cell lysates was incubated with 5μg of the rabbit anti-LRRK2 antibody (Abcam) overnight at 4°C with rotation. Then, 500μl of Pierce protein A/G magnetic beads was added and incubated at room temperature for 1h with rotation. The beads were collected with a magnetic stand and washed three times with IP lysis/wash buffer and once with ultrapure water. The proteins bound to the beads were eluted and analyzed by Western blot, which were detected with a mouse anti-LRRK2 (1:1,000; EMD Millipore) or mouse anti-FLAG (1:2,500; Sigma) primary antibody and IP-specific secondary antibody (anti-mouse IgG for IP, Abcam).

Co-localization with laser confocal microscopy in RAW cells

Transfected or untransfected cells were seeded in 8-well chambers. Twenty-four hours after transfection, cells were treated with or without 5 μg of N-glycolyl MDP 2 for another 24 hours. Cells were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100. Samples were blocked with blocking buffer [5% BSA, 2.52 mg/ml glycine in PBST (PBS + 0.1% Tween 20)] and incubated with the primary antibodies diluted in blocking buffer–rabbit anti-LRRK2 1:500 and mouse anti-FLAG 1:250 (for NOD2)–overnight at 4°C. Cells were washed with PBS for 3 x 5 minutes and incubated with secondary antibodies prepared in a diluted (1:5) blocking buffer [Alexa Fluor 488 goat anti-mouse IgG (H+L) 1:1000 and Alexa Fluor 594 donkey anti-rabbit IgG (H+L) 1:1000] for 1 hour. Cells were washed with PBS for 3 x 5 minutes and nuclei were stained with DAPI. Images were obtained by confocal microscopy. Colocalization between LRRK2 and NOD2 was measured from 25–30 cells by Zeiss 2012 ZEN confocal software.

Co-localization analysis in human monocyte-derived macrophages

Primary monocytes were obtained from peripheral blood monocytes from a healthy volunteer and two members of the studied family: the father and one of the twin girls (Fig 1). The blood, once collected, was transferred from the collection tube to a 50ml falcon tube (Corning) and diluted in sterile PBS (1:1) (Thermo Fisher Scientific). Once diluted, 25ml of blood was slowly transferred to another 50ml falcon tube containing 20mL of Ficol Paque (GIBCO) and centrifuged at 800g for 30 min at 25°C without brake. After centrifugation, peripheral blood mononuclear cells (PBMC) were separated in a new 50 ml falcon tube and washed with sterile PBS by centrifugation at 700g for 10 min at 25°C, and again in 20 ml by centrifugation at 500g for 10 min at 25°C. Finally, cells were resuspended in 2ml of RPMI medium supplemented with 2% FBS (GIBCO), 2mM of L-glutamine (L-GLU) (GIBCO) and 100 μg/ml of Penicillin-Streptomycin (PEES) (GIBCO). Cells were counted in a Neubauer chamber and 2x10⁶ PBMC (approximately 2x10⁵ monocytes) were plated per well, in a 24-well plate containing coverslips, in a final volume of 500μl of RPMI medium supplemented by 2% FBS, 2mM of L-GLU and 100 μg/ml of PEES for 2h at 37°C. After 2 hrs, non-adherent cells were gently washed 2 times with warm PBS and then incubated with 1ml of RPMI medium supplemented by 10% FBS, 2mM of L-GLU, 100 μg/mL of PEES and 100ng of GM-CSF (PrepoTech). After 3 days, half of the medium volume was removed and completed with another 500 μl of RPMI medium supplemented by 10% FBS, 2mM of L-GLU, 100μg /ml of PEES and 100ng of GM-CSF. On day 6 the cells were treated with 10ug/ml of N-glycolyl-MDP for another 24 hrs.

After that, coverslips with cells were fixed with 4% paraformaldehyde (Sigma-Aldrich) solution for 24 hrs and then 2% paraformaldehyde solution for another 48h. Then the coverslips were submitted to the immunofluorescence labeling protocol, which consists of a step of 30 minutes in a blocking solution: PBS supplemented by 5% normal goat serum (NGS) (Life technologies ref: PCN5000), 2% BSA (Sigma ref. A9418) and 0.1% triton X-100 (Amersham Biosciences, Ref: 17-115-01). This step was followed by a 1-hr incubation with the primary antibodies against LRRK2 (AB133474) at a dilution of 1:50 and against NOD2 (MA116611) at a dilution of 1:100. After the incubation time with the primary antibodies, the coverslips were washed 3 times in PBS. Then the coverslips were incubated with the secondary antibodies, Alexa 633 Goat anti-Rabbit (ThermoFisher Scientific) (1:500) and Alexa 546 Goat anti-Mouse (ThermoFisher Scientific) (1:400) for 30 minutes. After incubation with the secondary antibodies, coverslips were incubated with DAPI for 2 minutes and then washed 3 times in PBS. Coverslips were mounted on glass slides with ProLong Gold Antifade Mountant (ThermoFisher Scientific) and sealed with Permount Mounting Medium (Fisher Scientific). Images were obtained by confocal microscopy. Colocalization between LRRK2 and NOD2 was measured from 40 cells by LAS X Software (Leica).

Luciferase reporter assays

Luciferase reporter assays were performed using Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer’s instructions. Briefly, 1×10⁶ LRRK2 WT, DM or KO RAW 264.7 cells were suspended in 100μl of a solution (buffer R) containing 10μg of pcDNA3.1+/C-(K)DYK-NOD2, pcDNA3.1+/C-(K)DYK-NOD2 R702W or empty vector control, 3μg of NF-κB firefly-Luc plasmid, and 1μg of Renilla-Luc plasmid pRL-TK. Cells were transfected by electroporation using the Neon Transfection System (Thermo Fisher Scientific) at 1,680V for 20 ms and 1 pulse. Cells from each transfection reaction were plated in a 24-well plate with 1×10⁵ cells per well. Twenty-four hours after electroporation, cells were treated with 10μg/ml of N-glycolyl MDP for another 24 hours or left untreated; following this, cell lysates were prepared by using Passive Lysis Buffer (Promega). Luciferase activity was determined from a 20-μL cell lysates and measured on the microplate reader. Firefly luciferase activity was normalized to Renilla luciferase activity.

Cytokine and chemokine measurements

For cytokine and chemokine measurements, twenty-four hours after electroporation cells were incubated with or without BCG-Russia (MOI 10:1) for another 24 hours. Cell culture supernatants were collected and centrifuged to remove debris. Cytokines and Chemokines were detected using Milliplex Map (EMD Millipore, St. Charles, MO, USA) multiplex magnetic bead-based antibody detection kits according to the manufacturer’s instructions. A 6-plex premixed kit was used, which included analytes IL-1β, IL-6, IL-10, IL-12 (p40), MCP-1 and TNF. Assay was read using xPONENT 3.1 acquisition software and MAGPIX instrument (Luminex Corporation, Toronto, ON, Canada). Data was analyzed using Mulliplex Analyst software v4.2 (EMD Millipore) and presented as picogram of cytokine per milliliter of supernatants (pg/ml).

Statistical analysis

Statistical analysis was conducted using GraphPad Prism 5 (GraphPad Software, California USA, www.graphpad.com). Kinetics of ROS production were analyzed by comparing the mutant cells with the wild-type cells at each timepoint using two-way ANOVA with Bonferroni correction. In apoptosis, NF-κB activation and cytokine secretion assays, one-way ANOVA and post hoc t test with Bonferroni correction was used to compare means between two groups. Co-localization analysis was done using one-way ANOVA, comparing each condition with the control using Kruskal-Wallis test. Adjusted P < 0.05 was used as significance threshold, which is represented by asterisks in the figures. Data visualization for bar plots and box plots was done using ggplot2 package in R version 3.6.3 (https://www.R-project.org/" https://www.R-project.org/).