Graphical abstract

   graphic file with name fx1.jpg
   [71]Open in a new tab

Highlights

     * •
       rs194800 at 16p21 is associated with MAC disease susceptibility and
       PRKCB expression
     * •
       PRKCB exacerbates M. avium infection and promotes intracellular
       survival of M. avium
     * •
       PRKCB inhibits the fusion between lysosomes and mycobacterial
       phagosomes
     * •
       PRKCB inhibitor has therapeutic potential against M. avium
       infection
     __________________________________________________________________

   Zheng et al. identify that 16p21 is the susceptibility locus for
   pulmonary MAC disease, and the risk allele of rs194800 promotes PRKCB
   expression. PRKCB blocks the fusion between lysosomes and mycobacterial
   phagosomes to promote the intracellular survival of M. avium, thus
   increasing susceptibility to MAC disease.

Introduction

   Non-tuberculous mycobacterial (NTM) pulmonary disease (NTM-PD) is a
   severe progressive illness caused by non-tuberculous mycobacteria that
   is increasing in incidence and prevalence
   worldwide.[72]^1^,[73]^2^,[74]^3 Similar to the case for tuberculosis,
   NTM disease requires complicated treatment lasting more than 12 months.
   Unfortunately, treatment outcomes are unfavorable due to these bacteria
   responding poorly to multiple anti-mycobacterial drugs. There is thus a
   need to elucidate the pathogenesis and discover therapeutic strategies
   to control NTM infection.

   Non-tuberculous mycobacteria are ubiquitous environmental
   microorganisms, the human exposure to which is likely to be widespread.
   However, only a fraction of the population will develop disease related
   to these bacteria. This suggests that, in addition to microbial
   pathogenic or environmental factors, host factors also play a crucial
   role in the development of NTM disease. Previous studies have reported
   that underlying structural lung diseases are host risk factors for NTM
   disease.[75]^4^,[76]^5^,[77]^6 Nevertheless, many patients suffering
   from NTM disease had no known risk factors, implying that host genetic
   factors are also involved in the pathogenesis of NTM disease.

   To date, only two genome-wide association studies (GWASs) have been
   conducted in NTM-PD, one in the Japanese population and the other in
   the Korean population.[78]^7^,[79]^8 This indicates a pronounced
   paucity of GWAS in this disease context compared with those in the two
   other mycobacterial diseases
   tuberculosis[80]^9^,[81]^10^,[82]^11^,[83]^12^,[84]^13 and
   leprosy.[85]^14^,[86]^15^,[87]^16^,[88]^17^,[89]^18 Meanwhile,
   family-based whole-exome sequencing revealed an association between the
   TTK protein kinase gene (TTK) and susceptibility to
   NTM-PD.[90]^19^,[91]^20 However, in these published studies, only a few
   loci with genome-wide significance were identified and no follow-up
   functional studies on possible causal genes were performed. It is thus
   valuable to conduct a GWAS for NTM-PD in diverse populations to obtain
   a deeper understanding of the genetic basis of NTM-PD.

   In this work, we performed a two-stage GWAS in the Chinese population
   and discovered a significant susceptibility locus at 16p21 (rs194800,
   p[meta] = 7.85 × 10^−13) associated with NTM-PD, and the subtyping
   analysis showed that the 16p21 locus retained genome-wide significance
   mainly in Mycobacterium avium complex (MAC) patients. The rs194800
   variant regulated the expression of the protein kinase C beta (PRKCB)
   gene and was associated with the severity of NTM-PD. Furthermore,
   Prkcb-deficient mice were found to be more resistant to M. avium
   infection. In vitro, Prkcb was shown to enhance the intracellular
   survival of M. avium by inhibiting the fusion between lysosomes and
   mycobacterial phagosomes in macrophages. The inhibition of PRKCB by
   inhibitors impaired M. avium growth in vitro and in vivo. Our findings
   offer valuable insights into the genetic causes and pathogenesis of MAC
   disease and provide a promising target for host-directed therapy for
   MAC disease.

Results

GWAS

   To identify loci conferring susceptibility to NTM-PD in the Chinese
   population, we performed genome-wide analysis of 430 NTM cases and 475
   healthy controls (discovery stage, [92]Table S1). Here, 51 variants at
   chromosome 16p21 reached genome-wide significance (p < 5 × 10^−8,
   [93]Figures 1A and 1B). We next took forward the most significant
   independent variants (p < 1 × 10^−5, [94]Table S2) for replication in a
   sample of 202 NTM cases and 319 healthy controls ([95]Tables S1 and
   [96]S2). Meta-analysis of the discovery and replication datasets
   identified one locus with genome-wide significance at chromosome 16p21
   indexed by rs194800 (p[meta] = 7.85 × 10^−13; odds ratio [OR] = 2.02,
   [97]Tables S2 and [98]S3), which is in the intergenic region between
   CHP2 and PRKCB ([99]Figure 1C). Further fine-mapping of the 16p21
   region using the sum of single effects (SuSiE) method identified 22
   variants with a posterior inclusion probability (PIP) greater than 0.01
   ([100]Table S4).[101]^21 Among these, rs194800 had the highest PIP of
   0.045, suggesting its potential role in disease susceptibility. After
   adjusting for the effects of confounding variables (age, gender, and
   BMI), rs194800 remained significant in the NTM-PD, MAC, and
   M. abscessus (MAB) analyses ([102]Table S5). We also identified a
   suggestive locus (p < 5 × 10^−6) at chromosome 2p24.2 indexed by
   rs2881207 (p[meta] = 3.70 × 10^−6; OR = 0.34), which is located in the
   downstream region of the RDH14 gene ([103]Figure S1A). No evidence for
   heterogeneity across stages (p[het] > 0.05) was observed for either
   rs194800 or rs2881207 ([104]Table S2). In addition, the suggestive
   signals on chromosome 6p21.32 implicated the human leukocyte antigen
   (HLA) region as a potential risk factor for NTM-PD ([105]Figure S1B).

Figure 1.

   [106]Figure 1
   [107]Open in a new tab

   Plots showing putatively significant loci in the NTM GWAS

   (A) Manhattan plot for the GWAS conducted on NTM-PD in Han Chinese
   population. The x axis represents chromosomal position, and the y axis
   represents the −log10 (p values) of genome-wide SNPs analyzed in the
   GWAS. The horizontal red line represents the genome-wide significance
   threshold of p = 5.0 × 10^−8.

   (B) Quantile-quantile (Q-Q) plot for genome-wide association analysis
   of NTM-PD in discovery stage. The horizontal axis shows −log10
   transformed expected p values, while the vertical axis indicates −log10
   transformed observed p values. The blue line represents the expected
   −log10(p values) under the null hypothesis.

   (C) Regional plot for the 16p21 susceptibility region. The most
   significant variant, represented in purple diamond, serves as a lead
   variant. −log[10]p values are shown for variants for the region 200 kb
   on either side of the lead variant. The color of each variant
   represents its pairwise linkage disequilibrium (LD) with the lead SNP,
   which is quantified by the r^2 values. These r^2 values were determined
   using the 1000 Genomes ASN data (November 2014). The recombination rate
   (second y axis) is plotted in light blue color. The genes within the
   relevant region are annotated and displayed as arrows. See also
   [108]Figure S1 and [109]Tables S1–S7.

   To further confirm the role of host genetic factors in different types
   of NTM infections, we divided the NTM-PD patients into three groups,
   namely, pulmonary MAC patients, pulmonary Mycobacterium kansasii (MAK)
   patients, and pulmonary MAB patients, and conducted a subtyping
   analysis. For 311 MAC cases and 794 healthy controls, the 16p21 locus
   retained genome-wide significance and the most significant variant was
   rs2520016 (p = 5.35 × 10^−10, [110]Table S2), followed by rs194800 (p =
   1.96 × 10^−9, [111]Table S2). Both rs2520016 and rs194800 showed
   significant associations in MAB cases (p = 1.72 × 10^−4 and
   5.29 × 10^−6, respectively; [112]Table S2), but not in MAK patients
   (p > 0.05, [113]Table S2), indicating that chromosome 16p21 was
   primarily a susceptibility locus for MAC disease.

   Furthermore, we investigated rs109592, rs11646605, and rs849177
   identified in reported GWASs on NTM disease in our
   dataset.[114]^7^,[115]^8 We first analyzed the association of these
   SNPs in total NTM samples and then conducted separate analyses in the
   MAC, MAB, and MAK groups. The variants rs109592 and rs11646605 were
   significantly associated with susceptibility to NTM disease, MAC
   disease, and MAB disease in our samples, and rs849177 was not
   replicated in our samples ([116]Table S6).

   We explored the association of rs194800 and rs2520016 with
   susceptibility to TB based on our previous GWAS of TB in the Chinese
   Han population.[117]^11 These single-nucleotide polymorphisms (SNPs)
   exhibited marginal evidence of associations with TB (p < 0.10,
   [118]Table S7).

Genetic variation is associated with gene expression and clinical
characteristics

   The SNPs rs194800 and rs2520016 are located in the intergenic region
   between CHP2 and PRKCB ([119]Figure S2A). To investigate the potential
   biological significance of the two variants, we conducted an expression
   quantitative trait locus (eQTL) mapping analysis using peripheral blood
   mononuclear cells (PBMCs) obtained from NTM patients and healthy
   individuals. For NTM patients, carriers of the rs194800 C allele
   exhibited higher PRKCB expression than T-allele carriers, whereas no
   such association was observed in the normal subjects, suggesting that
   allele-specific expression of PRKCB might be induced by NTM infection
   ([120]Figure 2A). However, the rs194800 variant had no impact on CHP2
   expression ([121]Figure 2B) and rs2520016 had no correlation with PRKCB
   or CHP2 expression ([122]Figures 2C and 2D). Additionally, we performed
   eQTL analysis using Open Targets Genetics,[123]^22 GTEx
   (Genotype-Tissue Expression, v.10),[124]^23 ImmuNexUT (Immune Cell Gene
   Expression Atlas from the University of Tokyo),[125]^24 and
   scQTLbase.[126]^25 Open Targets Genetics revealed that rs2520016 and
   rs194800 were consistently associated with the expression of the PRKCB
   gene across various biological contexts, including different states of
   monocytes and macrophages such as naive, lipopolysaccharides
   (LPS)_stimulated, and interferon gamma (IFNG) exposed ([127]Table S8).
   ImmuNexUT demonstrated significant eQTL associations between both
   variants and PRKCB in classical monocytes ([128]Table S9). Similarly,
   analysis from scQTLbase further supported these findings, showing
   significant eQTL associations between these SNPs and PRKCB expression
   in various immune cell types, including CD14^+ monocytes and
   conventional dendritic cells ([129]Table S9). Moreover, GTEx v.10
   analysis suggested that these variants might also influence CHP2
   expression in lung tissue ([130]Table S9). Collectively, these analyses
   underscore the substantial role of these variants in modulating PRKCB
   expression. Furthermore, we generated a promoter reporter construct
   containing a 1,000-bp fragment with the SNP rs194800 T or C allele
   ([131]Figure 2E) and performed a dual-luciferase assay. The C allele
   exhibited significantly higher promotion of luciferase expression than
   the T allele in HEK293T cells ([132]Figure 2F). To confirm whether the
   rs194800-containing region has the features of an enhancer, we have
   examined the H3K4me1, H3K27ac chromatin modifications, assay for
   targeting accessible-chromatin with high-throughout sequencing
   (ATAC-seq) signal, and DNase cluster distribution in the regions
   surrounding rs194800 with RegulomeDB and the Encyclopedia of DNA
   Elements (ENCODE, the ENCODE Project Consortium) datasets in University
   of California, Santa Cruz (UCSC) Genome Browser. The obvious epigenetic
   signals of H3K4me1 and H3K27ac near the rs194800 locus strongly
   supported the presence of enhancer activity in the rs194800-containing
   region ([133]Figure 2G). The high signals of ATAC-seq signal and DNase
   clusters near rs194800 reinforced the probability of the existence of
   enhancer activity ([134]Figure 2G). Overall, these findings suggested
   that the risk allele of rs194800 might promote allele-specific
   expression of PRKCB.

Figure 2.

   [135]Figure 2
   [136]Open in a new tab

   Association of rs194800 and rs2520016 with PRKCB and CHP2 mRNA
   expression in PBMCs

   Genomic DNA and mRNA were isolated from PBMCs from NTM-PD patients (n =
   93) and healthy controls (n = 96). CHP2 and PRKCB mRNA expression was
   measured via quantitative reverse-transcription PCR (RT-PCR) and
   normalized to GAPDH.

   (A–D) The association of PRKCB and CHP2 mRNA expression with rs194800
   genotype (A and B) and rs2520016 genotype (C and D). Each symbol
   represents an individual (mean ± SEM).

   (E) Schematic of the vector constructs.

   (F) Luciferase reporter plasmids were transfected into HEK293T cells,
   and relative luciferase activities were assayed by using Renilla
   luciferase activity for normalization (mean ± SEM).

   (G) Overview of the H3K27ac and H3K4me1chromatin modifications, ATAC,
   and DNase cluster distribution in the regions surrounding rs194800
   supported from the UCSC Genome Browser. Rs194800 is indicated by the
   black vertical line.

   Data are representative of one experiment with at least three
   independent biological replicates in (A)–(D) and (F). Two-tailed
   unpaired Student’s t test (A–D, F) was used for statistical analysis.
   See also [137]Figure S2 and [138]Tables S8 and [139]S9.

   Next, we examined the association between the rs194800 polymorphism and
   various clinical indicators in NTM patients ([140]Figure S2B). The
   patients carrying the rs194800 C allele had lower plasma levels of
   soluble interleukin-2 receptor (sIL-2R), tumor necrosis factor alpha,
   and interleukin (IL)-1β than those carrying the T allele
   ([141]Figures S2C–S2E). The rs194800 C-allele carriers had a higher
   proportion of lung fields with lesions than those without the C allele
   ([142]Figure S2F). Meanwhile, the proportion of hemoptysis
   ([143]Figure S2G) and the proportion of symptom scores greater than 4
   or less than 4 ([144]Figure S2H) showed no association with rs194800.
   Taken together, the rs194800 C allele was associated with the severity
   of NTM infection.

Prkcb-deficient mice are more resistant to M. avium infection

   Because the 16p21 locus retained genome-wide significance only in MAC
   disease, we assessed the effect of PRKCB on the M. avium infection.
   Wild-type (WT) and Prkcb^−/− mice were infected with M. avium for
   4 weeks. Subsequently, histopathological impairment and bacterial
   burden in the lung were examined. The Prkcb^−/− mice exhibited reduced
   infiltration of immune cells and fewer inflammatory lesions compared
   with WT mice ([145]Figures 3A and 3B). For colony-forming unit (CFU)
   assay, Prkcb^−/− mice displayed decreased bacterial burden in their
   lungs ([146]Figure 3C). Moreover, we detected PRKCB expression in PBMCs
   and found that PRKCB mRNA expression was significantly elevated in
   NTM-PD patients compared with that in healthy controls, especially in
   those with MAC infection ([147]Figure 3D). Furthermore, we analyzed
   PRKCB protein expression immunohistochemically in sections of lung
   tissue surgically excised from NTM-PD patients, with adjacent normal
   tissue serving as a control. PRKCB was expressed prominently in MAC and
   MAK patients, whereas it was nearly undetectable in MAB patients
   ([148]Figures 3E–3J). Strong PRKCB staining from a necrotizing
   granuloma of MAC patients was located at the periphery of granulomatous
   lesions but not in necrotic areas ([149]Figure 3K), suggesting that
   PRKCB involved in MAC disease by regulating cellular functions rather
   than by promoting tissue necrosis. In addition to its high expression
   in granuloma, PRKCB is also present in small amounts in the epithelium
   of healthy lung ([150]Figure 3K). Together, Prkcb exacerbates M. avium
   infection and plays a negative regulatory role in host anti-M. avium
   immunity.

Figure 3.

   [151]Figure 3
   [152]Open in a new tab

   Prkcb-deficient mice are more resistant to M. avium infection

   (A) Histopathology of lung sections from WT (n = 6) or Prkcb^−/− (n =
   7) mice infected with M. avium by intranasal infection for 4 weeks.
   ^#1–3 indicate three representative lung sections for each group. The
   quantitation of inflammatory areas is shown in each section.

   (B) Quantitation of inflammatory areas in the lungs of mice infected
   with M. avium as in (A) (mean ± SEM).

   (C) Bacterial CFU in the lungs of the mice infected with M. avium as in
   (A) (mean ± SEM).

   (D) Real-time qPCR detection of PRKCB expression in PBMCs from controls
   (n = 96) or M. avium complex pulmonary disease (MAC-PD) patients (n =
   48), M. abscessus pulmonary disease (MAB-PD) patients (n = 30), and
   M. kansasii pulmonary disease (MAK-PD) patients (n = 9) (mean ± SEM).

   (E–G) Immunohistochemical examination of PRKCB expression in granulomas
   from MAC patients (n = 2) (E), MAK patients (n = 2) (F), MAB patients
   (n = 2) (G), and adjacent normal tissue control (scale bars, 200 μm
   [top] and 20 μm [bottom]). One representative field for each group was
   showed, and positive cells are brown.

   (H–J) Quantification of PRKCB expression by immunostaining from the MAC
   group (H), MAK group (I), and MAB group (J) (mean ± SEM). 3 fields per
   group for quantitation of PRKCB-positive cells.

   (K) Representative PRKCB staining in a necrotizing granuloma of lung
   sections from MAC patients. Blue dashed lines delineated the granuloma.
   Yellow dashed lines delineated area surrounding necrotic zone within
   the granuloma. The right panels showed PRKCB staining in the granuloma.
   The middle and bottom left panels showed PRKCB staining in the normal
   lung. The red arrows indicated positive cells. (scale bars, 2000 μm
   [left top], 200 μm [left middle], 50 μm [left bottom], 200 μm [right
   top], 100 μm [right middle], and 50 μm [right bottom]). Images shown
   are representative of those observed in 2 MAC patients.

   Data are representative of one experiment with at least three
   independent biological replicates in (C) and (D). Two-tailed unpaired
   Student’s t test (B, D, H–J) was used for statistical analysis.
   Two-sided Mann-Whitney U test (C) was used for statistical analysis.

PRKCB promotes intracellular survival of M. avium through inhibiting
phagosomal acidification in macrophages

   To further explore the function of PRKCB during M. avium infection, we
   conducted transcriptomic analysis on the lung tissue of WT and
   Prkcb^−/− mice infected with M. avium for 4 weeks by using RNA
   sequencing (RNA-seq). A number of significantly differentially
   expressed genes (DEGs) were identified ([153]Figures S3A and S3B), and
   pathway enrichment analyses on these DEGs revealed that the innate
   immune system was the most significant pathway ([154]Figure S3C),
   suggesting that PRKCB might play a crucial role during M. avium
   infection through regulating innate immunity. Macrophages are the
   primary host innate immune cells, and their effectiveness in containing
   or eliminating intracellular non-tuberculous mycobacteria is essential
   for host anti-NTM immunity.[155]^26 Therefore, we analyzed the effect
   of PRKCB on the clearance of non-tuberculous mycobacteria through
   macrophages. Primary murine peritoneal macrophages from WT and
   Prkcb^−/− mice were infected with M. avium, MAB, and MAK for 3 h and
   24 h and then subjected to a CFU assay. The intracellular NTM loads
   were similar at 3 h post-infection, indicating that PRKCB had no
   effects on the phagocytosis of non-tuberculous mycobacteria by
   macrophages. However, the intracellular amounts of M. avium and MAK
   were dramatically decreased at 24 h post-infection in Prkcb^−/−
   macrophages ([156]Figures 4A–4F), suggesting PRKCB as a host factor
   that aided the intracellular survival of NTM. It has been reported that
   phagosomal acidification is essential for the clearance of
   non-tuberculous mycobacteria.[157]^27^,[158]^28 We thus next examined
   whether PRKCB promoted intracellular survival of NTM by inhibiting
   phagosomal acidification. For this, the macrophages from WT and
   Prkcb^−/− mice were pretreated with Bafilomycin A1 (BafA1), an
   inhibitor of vacuolar ATPase (V-ATPase), and the intracellular survival
   of M. avium was examined. The inhibitors rescued the promoting effect
   of PRKCB on the intracellular survival of M. avium ([159]Figures 4G and
   4H), suggesting that PRKCB promoted intracellular survival of M. avium
   through inhibiting V-ATPase-mediated phagosomal acidification. By
   LysoTracker, an acidotrophic fluorescent dye accumulated in acidic
   organelles, phagosomal acidification was assessed in primary murine
   peritoneal macrophages from WT and Prkcb^−/− mice infected with
   M. avium for 6 h. The results showed that Prkcb deficiency promoted
   phagosomal acidification ([160]Figures 4I and 4J). We treated
   macrophages with BafA1 and examined phagosomal acidification. BafA1
   eliminated the difference in phagosomal acidification between WT and
   Prkcb^−/− mice ([161]Figures 4I and 4J), suggesting that PRKCB
   suppresses phagosomal acidification via V-ATPase. In contrast to the
   results of M. avium infection, PRKCB promoted phagosomal acidification
   to inhibit intracellular survival of M. tuberculosis through V-ATPase
   ([162]Figures S4A–S4D), suggesting that there was no shared
   pathogenesis mechanism mediated by PRKCB between NTM-PD and TB.

Figure 4.

   [163]Figure 4
   [164]Open in a new tab

   PRKCB promotes intracellular survival of NTM through inhibiting
   phagosomal acidification in macrophages

   (A–F) Intracellular CFU and bacterial survival in WT or Prkcb^−/− mice
   peritoneal macrophages infected with M. avium (A and B), M. kansasii (C
   and D), and M. abscessus (E and F) for the indicated times (MOI = 5)
   (mean ± SEM).

   (G and H) Intracellular CFU and bacterial survival in WT or Prkcb^−/−
   mice peritoneal macrophages treated with BafA1 (100 nM, 2 h) and
   infected with M. avium (MOI = 5) (mean ± SEM).

   (I) LysoTracker staining of macrophages from WT and Prkcb^−/− mice
   treated with BafA1 (100 nM, 2 h) and infected with M. avium for the 6 h
   (MOI = 5) (scale bar, 5 μm). M. avium were labeled with fluorescein
   isothiocyanate (FITC), and DAPI staining was used to identify nuclei.

   (J) Quantitative analysis of fluorescence by ImageJ as in (I) counting
   at least 200 random cells from each sample (mean ± SEM).

   Data are representative of one experiment with at least three
   independent biological replicates in (A)–(J). Two-tailed unpaired
   Student’s t test (A–H, J) was used for statistical analysis. See also
   [165]Figures S3 and [166]S4.

PRKCB interacts with VPS16 and ATP6V1G

   To uncover the key effectors of PRKCB inhibition on phagosomal
   acidification, we performed a label-free proteomic analysis. The
   differential peptides were primarily enriched in vesicle fusion,
   protein localization to vacuole or lysosome, and lysosomal or vacuolar
   transport signaling ([167]Figure 5A). We examined the association of
   PRKCB with various components of these pathways by
   co-immunoprecipitation and identified the interaction of PRKCB with
   vacuolar protein sorting-associated protein 16 homolog (VPS16)
   ([168]Figure 5B). We further confirmed the interaction by
   overexpressing PRKCB and VPS16 ([169]Figure 5C). Colocalization of
   PRKCB with VPS16 was found by immunofluorescence ([170]Figure 5D).
   VPS16, a common subunit of the class C core endosomal vacuole tethering
   (CORVET) and homotypic fusion and vacuole protein sorting (HOPS)
   complex, is reported to function as a tether protein involved in
   phagosome-lysosome fusion.[171]^29 To examine the association of PRKCB
   and V-ATPase machinery, the interaction between PRKCB and various
   subunits of V-ATPase complex was analyzed using immunoprecipitation,
   and the results revealed that PRKCB interacted with the G subunit
   ([172]Figures 5E and 5F). The colocalization of PRKCB with ATP6V1G was
   also identified ([173]Figure 5G). The interaction of PRKCB with VPS16
   and ATP6V1G implied that the V-ATPase and class C VPS complexes might
   be closely associated during phagosome maturation and endosome-lysosome
   fusion. In order to examine this hypothesis, we detected the
   interaction of VPS16 with ATP6V1G and found that VPS16 interacted with
   ATP6V1G ([174]Figures 5H and 5I). The colocalization of VPS16 with
   ATP6V1G was also observed ([175]Figure 5J), suggesting that VPS16 and
   ATP6V1G might mediate the interaction between the class C VPS complex
   and V-ATPase machinery.

Figure 5.

   [176]Figure 5
   [177]Open in a new tab

   PRKCB interacts with VPS16 and ATP6V1G

   (A) Schematic diagram (upper) showing the proteomic analysis of
   overexpression of PRKCB or vector in HeLa cells. The top enriched
   signaling pathways are shown (bottom).

   (B) Immunoprecipitation and immunoblot of lysates of HeLa cells
   transfected with plasmids encoding FLAG-PRKCB and detected with the
   indicated antibody.

   (C) Immunoprecipitation and immunoblot of lysates of HeLa cells
   transfected with plasmid encoding FLAG-PRKCB or hemagglutinin
   (HA)-VPS16.

   (D) Immunofluorescence of HeLa cells transfected with plasmids encoding
   FLAG-PRKCB or HA-VPS16 (scale bar, 5 μm). DAPI staining was used to
   identify nuclei.

   (E) Immunoprecipitation and immunoblot of lysates of HeLa cells
   transfected with plasmids encoding FLAG-PRKCB or HA-tagged subunits of
   V-ATPase.

   (F) Immunoprecipitation and immunoblot of lysates of HeLa cells
   transfected with plasmids encoding FLAG-PRKCB or HA-V1G.

   (G) Immunofluorescence of HeLa cells transfected with plasmids encoding
   FLAG-PRKCB or HA- V1G (scale bar, 10 μm). DAPI staining was used to
   identify nuclei.

   (H) Immunoprecipitation and immunoblot of lysates of HeLa cells
   transfected with plasmids encoding HA-VPS16 and detected with the
   indicated antibody.

   (I) Immunoprecipitation and immunoblot of lysates of HeLa cells
   transfected with plasmids encoding FLAG-VPS16 or HA-V1G.

   (J) Immunofluorescence of HeLa cells transfected with plasmids encoding
   FLAG-VPS16 or HA-V1G (scale bar, 5 μm). DAPI staining was used to
   identify nuclei.

PRKCB blocks the interaction between Vps-C and V-ATPase complexes

   Because PRKCB interacted with VPS16 and ATP6V1G, and VPS16 was
   recruited to ATP6V1G, we asked whether PRKCB binding to VPS16 and
   ATP6V1G affected the association of VPS16 and ATP6V1G during M. avium
   infection. To address this question, we examined endogenous interaction
   using VPS16 antibodies as the bait and detected endogenous ATP6V1G in
   WT and Prkcb^−/− macrophages infected with M. avium. Endogenous
   interactions of VPS16 and ATP6V1G could be detected in both the
   uninfected and infected WT macrophages ([178]Figure 6A), suggesting
   that interaction between ATP6V1G and the class C VPS regulated
   endosomal-lysosomal fusion in resting cells and phagosome-lysosome
   fusion in infected cells. Compared to WT macrophages, the interaction
   of VPS16 and ATP6V1G was significantly upregulated in Prkcb^−/−
   macrophages ([179]Figure 6A), suggesting that PRKCB inhibited the
   interaction of the class C VPS complex and V-ATPase complex. Disruption
   of the interaction between VPS16 and ATP6V1G by PRKCB could be observed
   using ATP6V1G antibody as the bait and detected VPS16 ([180]Figure 6B).
   Next, we detected the colocalization of VPS16, ATPV1G, and M. avium
   within WT and Prkcb^−/−macrophages. The results showed that VPS16 and
   ATP6V1G colocalized to the mycobacterial phagosome within macrophages
   and the colocalization enhanced in Prkcb^−/− macrophages compared to WT
   macrophages ([181]Figures 6C and 6D), suggesting that PRKCB blocked the
   fusion of lysosomes and bacilli-containing phagosomes. Furthermore,
   blocking of Vps16 with small interfering RNA (siRNA) attenuated Prkcb
   deficiency-mediated enhancement of M. avium-induced phagosomal
   acidification ([182]Figures 6E–6G), suggesting that PRKCB inhibited
   phagosomal acidification through VPS16. Knockdown of Vps16 in WT and
   Prkcb^−/− macrophages diminished the PRKCB-induced increase in the
   intracellular survival of M. avium in macrophages ([183]Figures 6H and
   6I). It has been reported that M. avium could evade macrophages’
   killing through secreting effector molecules into the cytoplasm
   to interfere with phagosomes maturation.[184]^30^,[185]^31 To determine
   whether M. avium proteins interacted with PRKCB or VPS16 to involve in
   the fusion of phagosomes and lysosomes, we isolated the M. avium
   phagosomes and performed a mass spectrometric analysis to find M. avium
   proteins bound to phagosome according to the methods of Danelishvili
   and Bermudez ([186]Figures S5A and S5B).[187]^28 Thirteen M. avium
   proteins were identified in our samples including M. avium proteins of
   alpha (MAV_1525) and beta (MAV_1527) subunits of ATP synthase reported
   in the Danelishvili and Bermudez study. We examined the interaction of
   PRKCB and the two reported M. avium proteins MAV_1525 and MAV_1527
   using immunoprecipitation techniques. Additionally, we investigated the
   interaction of VPS16 and these M. avium proteins. Furthermore, we
   assessed the interaction of PRKCB or VPS16 and the two other M. avium
   proteins MAV_4671 and MAV_4996, which have been reported to interact
   with host proteins.[188]^27^,[189]^28 The results indicated that three
   out of the four M. avium proteins, namely MAV_1525, MAV_1527, and
   MAV_4996, were found to interact with VPS16. However, none of the four
   M. avium proteins showed any interaction with PRKCB ([190]Figures S5C
   and S5D), suggesting that the VPS16-mediated phagosomal fusion pathway
   was crucial for the host to eliminate M. avium. The pathway was
   utilized not only by host molecules like PRKCB but also by M. avium
   proteins. Together, these results demonstrate that PRKCB inhibits
   phagosomal acidification through blocking the class C VPS- and V-ATPase
   complexes-mediated phagosome-lysosome fusion during M. avium infection.

Figure 6.

   [191]Figure 6
   [192]Open in a new tab

   PRKCB blocks the interaction between Vps-C and V-ATPase complexes
   during infection

   (A and B) Endogenous interactions of VPS16 with ATP6V1G in WT or
   Prkcb^−/− mice peritoneal macrophages infected with M. avium for the
   indicated time and immunoprecipitated with VPS16 antibody (A) or V1G
   antibody (B). Rabbit immunoglobulin G (IgG) was used as negative
   control.

   (C) Localization of M. avium, VPS16, and V1G was detected in WT or
   Prkcb^−/− mice peritoneal macrophages infected with M. avium for 6 h
   with immunofluorescence staining (scale bar, 5 μm). DAPI staining was
   used to identify nuclei.

   (D) Quantitative analysis of puncta area by ImageJ as in (C) counting
   at least 200 random cells from each sample (mean ± SEM).

   (E) The efficiency of siRNA knockdown of vps16.

   (F) LysoTracker staining of macrophages from WT and Prkcb^−/− mice
   transfected with control siRNA or siRNA targeting mouse vps16 for 48 h
   and infected with M. avium for 6 h (MOI = 5) (scale bar, 5 μm).
   M. avium were labeled with FITC, and DAPI staining was used to identify
   nuclei.

   (G) Quantitative analysis of fluorescence by ImageJ as in (F) counting
   at least 200 random cells from each sample (mean ± SEM).

   (H and I) Intracellular CFU and bacterial survival in vps16 knockdown
   primary peritoneal macrophages from WT and Prkcb^−/− mice and infected
   with M. avium for the indicated times (MOI = 5) (mean ± SEM).

   Data are representative of one experiment with at least three
   independent biological replicates in (C)–(I). Two-tailed unpaired
   Student’s t test (D, E, G, H, and I) was used for statistical analysis.
   See also [193]Figure S5.

PRKCB inhibitor has therapeutic potential against M. avium infection

   PRKCB inhibitors, particularly ruboxistaurin (LY-333531, RBX), have
   been the subject of extensive research.[194]^32^,[195]^33^,[196]^34 The
   reported randomized and multicenter clinical trials have demonstrated
   that ruboxistaurin can reduce the development and/or progression of
   diabetic microvascular complications, with phase 3 trials having been
   completed. Recent study has also indicated that ruboxistaurin is a
   potential therapeutic option for COVID-19.[197]^35 To explore the
   therapeutic effects of PRKCB inhibitors on NTM disease, peritoneal
   macrophages were pretreated with one of two effective PRKCB inhibitors,
   enzastaurin (ENZ) and ruboxistaurin, followed by infection with
   M. avium. Both inhibitors were capable of inhibiting the intracellular
   survival of pathogens in macrophages, with ruboxistaurin exhibiting a
   superior inhibitory effect ([198]Figures 7A and 7B). Ruboxistaurin
   exerted its effect by targeting PRKCB in vitro ([199]Figures 7C and
   7D). In vivo, mice were intranasally infected with M. avium for 2 weeks
   and subsequently administered an intragastric solution of ruboxistaurin
   at a dosage of 1 mg/kg for an additional 4 weeks ([200]Figure 7E). In
   WT mice, treatment with the inhibitor significantly alleviated
   pathological damage in the lung tissue ([201]Figures 7F and 7H) and
   reduced bacterial burden ([202]Figure 7I) compared with the findings in
   the untreated group, suggesting that PRKCB inhibitor may hold
   therapeutic potential against M. avium infection. The difference in
   pathological damage ([203]Figures 7G and 7H) and bacterial burden
   ([204]Figure 7I) was eliminated in Prkcb^−/− mice treated with
   ruboxistaurin, suggesting that ruboxistaurin played a role through
   targeting PRKCB.

Figure 7.

   [205]Figure 7
   [206]Open in a new tab

   PRKCB inhibitor ameliorates M. avium infection

   (A and B) Intracellular CFU in mice peritoneal macrophages treated with
   enzastaurin (10 μM, 24 h) or ruboxistaurin (10 μM, 24 h) and infected
   with M. avium for 24 h (MOI = 5) (mean ± SEM).

   (C and D) Intracellular CFU and bacterial survival in WT and Prkcb^−/−
   mice peritoneal macrophages treated with ruboxistaurin (10 μM, 24 h)
   and infected with M. avium for the indicated times (MOI = 5) (mean ±
   SEM).

   (E) Schematic diagram of the procedure to evaluate the effect of PRKCB
   inhibitor on WT mice (n = 7) and Prkcb^−/− mice (n = 6) intranasally
   infected with M. avium for 2 weeks and intragastric administrated with
   ruboxistaurin (1 mg/mL) or vehicle solution for another 4 weeks.

   (F and G) H&E staining of lung sections of WT mice (F) and Prkcb^−/−
   mice (G) infected with M. avium as in (E). ^#1–3 indicate three
   representative lung sections for each group. The quantitation of
   inflammatory areas is shown in each section.

   (H) Quantitation of inflammatory areas in the lungs of mice infected
   with M. avium as in (E) (mean ± SEM).

   (I) Bacterial CFU in lungs of WT mice and Prkcb^−/− mice as in (E).
   RBX, ruboxistaurin; ENZ, enzastaurin.

   Data are representative of one experiment with at least three
   independent biological replicates in (A)–(D), (H), and (I). Two-tailed
   unpaired Student’s t test (A–D, H) was used for statistical analysis.
   Two-sided Mann-Whitney U test (I) was used for statistical analysis.

Discussion

   In this two-stage GWAS of NTM-PD, we identified that the locus rs194800
   on chromosome 16p21 was significantly associated with NTM-PD risk,
   especially for MAC disease. We further demonstrated that the rs194800 C
   allele was associated with PRKCB expression and the severity of NTM
   infection. Follow-up functional studies showed that PRKCB exacerbated
   M. avium infection in mice and promoted the intracellular survival of
   M. avium in macrophages through inhibiting the fusion of lysosomes and
   phagosomes. Finally, PRKCB inhibitors made mice more resistant to
   M. avium infection. This GWAS of NTM-PD attempted to identify the
   causal gene and mechanism behind the GWAS-related findings. These
   findings suggest that PRKCB is a strong causal gene involved in the
   pathogenesis of MAC disease and that treatment with a PRKCB inhibitor
   may serve as a therapeutic strategy against MAC disease.

   Non-tuberculous mycobacteria are ubiquitous in the environment.
   However, despite widespread exposure to them, only a few people develop
   NTM disease. Candidate gene studies have identified NRAMP1, CFTR,
   IL-10, and IL-28B as susceptibility genes for
   NTM-PD.[207]^36^,[208]^37^,[209]^38^,[210]^39 To date, only two GWASs
   of NTM-PD have been performed. In 2021, Ho Namkoong and colleagues
   conducted a GWAS in a Japanese population with pulmonary MAC infection
   and identified chromosome 16p21, specifically rs109592, as having the
   strongest association with MAC disease susceptibility. They further
   validated rs109592 in Korean and European populations. Interestingly,
   in our work, we also found the same susceptibility region on chromosome
   16p21 in the Chinese population. Of note, in our study, the strongest
   association with NTM-PD was rs194800 but not rs109592. We also analyzed
   rs109592 in our samples and found that it had a significant association
   with susceptibility to NTM-PD in the Chinese population. Linkage
   disequilibrium analysis revealed that rs109592 and rs194800 were
   located within the same linkage equilibrium region (r^2 > 0.80 based on
   the 1000G European [EUR] and East Asian [ASN] data). Notably, two GWASs
   conducted in different populations determined that chromosome 16p21
   exhibited the strongest association with susceptibility to MAC disease,
   suggesting that this region may contain causative variants and genes
   responsible for driving susceptibility to MAC disease. However, further
   investigation is warranted, including larger sample sizes and
   international collaborative GWAS across diverse populations.

   Because clinical manifestations of NTM-PD are similar to TB, we
   investigated the SNPs responsible for NTM-PD susceptibility in TB
   patients, and the role of PRKCB in Mtb infection. None of the SNPs
   showed an association with TB, and PRKCB exerted opposing regulatory
   effects on the intracellular survival of NTM and Mtb, suggesting that
   the chromosome 16p21 region might be specific to NTM-PD susceptibility
   and there is no genetic sharing mechanism or shared pathogenesis
   mediated by PRKCB between NTM-PD and TB.

   Within the chromosome 16p21 region, CHP2 and PRKCB are the two closest
   neighboring genes. In Ho Namkoong’s study, it was emphasized that CHP2
   was a strong candidate gene involved in the pathogenesis of pulmonary
   MAC disease, while PRKCB was considered a potential gene alongside CHP2
   based on multi-tissue eQTL analysis. In our study, rs194800 exhibited
   eQTL effects on PRKCB. Analysis of enhancer activity indicated that the
   region containing the variant rs194800 had allele-specific enhancer
   activity, and the risk allele C of rs194800 might correspond to higher
   enhancer activity as compared to the non-risk allele. However, whether
   rs194800 is a functional SNP warrants further investigation. Notably,
   PRKCB has also been identified as a genetic susceptibility locus for
   primary biliary cholangitis and systemic lupus
   erythematosus.[211]^40^,[212]^41 Taking together, rs194800 is a
   probable causative variant for NTM-PD and PRKCB may also be a candidate
   effector gene contributing to such diseases.

   PRKCB is a member of the protein kinase C (PRKC) family and functions
   as a serine/threonine-protein kinase.[213]^42 Increasing evidence
   suggests that PRKCB regulates B cell polarity and activation and
   autophagy levels[214]^43^,[215]^44^,[216]^45 and is involved in the
   pathogenesis of several diseases.[217]^46^,[218]^47^,[219]^48 However,
   the function of PRKCB in NTM-PD remains unknown. Our study demonstrates
   that PRKCB involves in the pathogenesis of NTM-PD. PRKCB weakened the
   host’s ability of clearance of M. avium. The persistent M. avium might
   induce excessive or prolonged inflammatory responses to cause
   pathological damage to the lungs.

   Although it has been long observed that inhibition of
   phagosome-lysosome fusion and alteration of phagosomal pH may
   contribute to the intracellular survival of M. avium,[220]^49 the
   underlying mechanism remains poorly understood. V-ATPase is recruited
   to the phagosomal membrane during phagosome maturation for luminal
   acidification and the class C VPS complex, as the membrane-tethering
   factor is involved in this process.[221]^50^,[222]^51 Our studies
   demonstrated that VPS16 and subunit G of V-ATPase were key players to
   mediate phagosome-lysosome fusion during M. avium infection. Vesicle
   fusion in the endocytic pathway depended on two membrane-tethering
   protein complexes, HOPS and CORVET. The subunits of VPS11, VPS16,
   VPS18, and VPS33 associate to form the stable VPS-C core complex for
   HOPS and CORVET.[223]^52^,[224]^53 VPS16 is required for the fusion of
   intracellular compartments with lysosomes.[225]^54^,[226]^55^,[227]^56
   Our research revealed that PRKCB interacted with VPS16 and V1G,
   suggesting that this interaction may potentially lead to the specific
   localization of PRKCB at the phagosome-lysosome interface. This
   specific localization hindered the fusion of V-ATPase with the
   phagosome, resulting in inhibition of mycobacterial phagosome
   acidification. Our findings show a mechanism for explaining how
   M. avium prevents lysosomal fusion with mycobacterial phagosome in
   macrophages, enabling its survival within phagosomes. However, whether
   PRKCB depends on its kinase activity to block the interaction between
   VPS16 and ATP6V1G needs to be explored further.

   The limited efficacy of current treatments for NTM disease poses a
   significant challenge for disease management. One of the main reasons
   for the inadequate success of NTM disease treatment is the resistance
   of NTM pathogens to multiple anti-mycobacterial drugs.[228]^57 There is
   thus an urgent need for novel therapeutic strategies that target the
   host rather than the NTM pathogens themselves.[229]^26 Ruboxistaurin, a
   highly specific inhibitor of PRKCB,[230]^58 has been extensively
   studied in phase 3 trials for diabetic retinopathy and has shown good
   tolerance in patients.[231]^59 Our study found that ruboxistaurin could
   inhibit the intracellular survival of non-tuberculous mycobacteria in
   macrophages and reduced the bacterial burden and pathological
   impairments in the lung tissue of mice infected with M. avium through
   specifically targeting PRKCB. These findings suggest that targeting
   PRKCB could be a promising host-directed therapy against MAC disease.
   Importantly, ruboxistaurin has been proven to be safe, making it a
   promising therapeutic strategy for MAC disease. Together, these
   findings not only support the significance of PRKCB in M. avium
   infection but also reveal a potential therapeutic approach for managing
   MAC disease.

   Identifying the causal genes and mechanisms behind GWAS hits poses
   various challenges. Building on our GWAS, we performed functional
   studies to identify rs194800 as a probable causative variant and PRKCB
   as a candidate effector gene involved in the pathogenesis of MAC
   disease. These findings enhance our understanding of the genetic basis
   of MAC disease and contribute to our knowledge of its pathogenesis.
   Moreover, targeting PRKCB may hold promise as a therapeutic approach
   for MAC disease.

Limitations of the study

   The present study suggests that the rs194800-containing region has the
   characteristics of an enhancer through bioinformatics analysis.
   Additional functional experiments, such as constructing rs194800
   knockin cells or mice, could provide further evidence to confirm that
   rs194800 is a functional SNP. In addition, the mechanism by which PRKCB
   binding to VPS16 and ATP6V1G affects phagosome-lysosome fusion needs
   further exploration. The role of PRKCB kinase activity in this process
   is necessary to evaluate. Further experiments with PRKCB kinase
   activity mutants, combined with analyses such as phosphoproteome
   analysis of PRKCB, could provide additional mechanistic insights.

Resource availability

Lead contact

   Further information and requests for resources and reagents should be
   directed to and will be fulfilled by the lead contact, Wei Sha
   (Shfksw@tongji.edu.cn).

Materials availability

   Plasmids generated in this study are available from lead contact upon
   request.

Data and code availability

     * •
       RNA-seq data have been deposited at BioSample database, and GWAS
       summary statistics data have been deposited at [232]bio-x.com
       (gwas.bio-x.cn) and [233]figshare.com and are publicly available as
       of the date of publication. Accession numbers are listed in the
       [234]key resources table.
     * •
       This paper does not report original code.
     * •
       Any additional information required to reanalyze the data reported
       in this paper is available from the [235]lead contact upon request.

Acknowledgments