Polymorphisms in PI3K/AKT genes and gene‑smoking interaction are associated with susceptibility to tuberculosis

Abstract Background Phosphatidylinositol 3-kinase (PI3K) and protein kinase B (AKT) are involved in the clearance of Mycobacterium tuberculosis (MTB) by macrophages. Aim This study aimed to investigate the effects of polymorphisms in the PI3K/AKT genes and the gene-smoking interaction on susceptibility to TB. Methods This case-control study used stratified sampling to randomly select 503 TB patients and 494 control subjects. Logistic regression analysis was used to determine the association between the polymorphisms and TB. Simultaneously, the marginal structure linear dominance model was used to estimate the gene-smoking interaction. Results Genotypes GA (OR 1.562), AA (OR 2.282), and GA + AA (OR 1.650) at rs3730089 of the PI3KR1 gene were significantly associated with the risk to develop TB. Genotypes AG (OR 1.460), GG (OR 2.785), and AG + GG (OR 1.622) at rs1130233 of the AKT1 gene were significantly associated with the risk to develop TB. In addition, the relative excess risk of interaction (RERI) between rs3730089 and smoking was 0.9608 (95% CI: 0.5959, 1.3256, p < 0.05), which suggests a positive interaction. Conclusion We conclude that rs3730089 and rs1130233 are associated with susceptibility to TB, and there was positive interaction between rs3730089 and smoking on susceptibility to TB.


Introduction
TB is an infectious disease, primarily caused by Mycobacterium tuberculosis (MTB).MTB is a slow-growing, acid-resistant bacterium transmitted mainly through the respiratory tract.The current global TB epidemic situation remains critical.According to the latest report released by the World Health Organisation, the Global TB Report 2021, the number of latent TB infections is expected to reach 2 billion worldwide, and the number of new TB cases in 2020 will be approximately 9.9 million, affecting more men than women.With an estimated 842,000 new cases of TB in 2020 (833,000 in 2019), China carries the second-highest TB burden in the world after India (2.59 million).
Globally, an estimated one-third of the population is infected with MTB, and approximately 5-10% of these infected individuals eventually develop clinical TB disease (Abel et al. 2014).Family aggregation and twin studies strongly affirm the importance of host genetic factors in determining the outcome of MTB infection (Dallmann-Sauer et al. 2018;Aravindan 2019).
The genetic factors may be influenced by environmental or other host factors, resulting in gene-environment interaction phenomena (Kaelin and McKnight 2013).According to the available literature (Chen et al. 2015;Jafari et al. 2016;Cai et al. 2019), the main genes associated with susceptibility to TB include the human leukocyte antigen (HLA) gene, natural-resistance associated macrophage protein1 (NRAMP1), the solute carrier family 11 member 1 (SLC11A1) gene system, the vitamin D receptor (VDR) gene, the mannan-binding lectin (MBL) gene, and the nitric oxide synthase 2A (NOS2A) gene, etc.The PI3K/AKT genes have been associated with human immunodeficiency virus, hepatitis C, hepatitis B, influenza, and other infectious diseases (Noguchi and Kinowaki 2008;Dunn and Connor 2012;Fruman et al. 2017).
PI3K regulates signalling and intracellular vesicle trafficking by phosphorylating intracellular inositol lipids, thus playing an important role in signal transduction during cell growth, proliferation, apoptosis, motility, and differentiation (Nicholson and Anderson 2002;McCubrey et al. 2006).PI3Ks can be divided into three classes.Class I PI3K produces 3-phosphatidylinositol lipids and directly activates signal transduction pathways, while class II and class III PI3Ks are regulators of membrane trafficking in endocytic pathways and are involved in endocytic cycling and autophagy (Bilanges et al. 2019).As a key component of the immune system, macrophages can recognise, respond to and react to MTB infection.Autophagy is the main mechanism of MTB clearance by macrophages (Hmama et al. 2015).There are three isoforms of AKT, AKT1, AKT2 and AKT3, which are downstream targets of PI3K (Hers et al. 2011).AKT can be activated by PI3KCI, thereby opposing the inhibitory effect of the TSC1/2 heterodimer on Rheb which leads to mTORC1 activation, resulting in inhibition of autophagy (Janku et al. 2011).AKT is activated by the phosphorylation of mTORC2, which further promotes the inhibition of autophagy (Sun et al. 2005).The PI3K/AKT pathways activate MTB recombinant leucine-responsive regulatory protein (rLrp) to inhibit the production of pro-inflammatory cytokine and downregulate macrophage antigen presentation (Wang et al. 2014;Hmama et al. 2015).In summary, we hypothesise that polymorphisms in the genes of the PI3K/AKT may be associated with the risk of TB development.Therefore, we designed a case-control study to analyse the association between thirteen SNPs and five genes of the PI3K/AKT pathways and TB susceptibility.
Smoking is an important behavioural risk factor for the development of TB, increasing susceptibility, severity, and mortality (van Zyl Smit et al. 2010;Silva et al. 2018), and studies have found a negative association between smoking and TB treatment success outcomes (Khan et al. 2015).The inhaled cigarette smoke increases the risk of TB by compromising respiratory immune function (Leung et al. 2007).In addition, smoking increases macrophage lysosomal fragmentation, which prevents effective clearance of MTB (Berg et al. 2016).Since TB is a typical multifactorial disease, the development of TB is influenced not only by genetic and individual behavioural factors but also by complex interactions between genes and the act of smoking.Therefore, this study also investigated the effects of the interaction between PI3K/ AKT gene polymorphisms and smoking on TB susceptibility.

Participants and study design
In this case-control study, we adopted a multistage sampling method to extract the study subjects, and the details were reported in our previous publication (Wang et al. 2021).To acquire samples for the case group, four county-level centres for disease control and prevention in Qidong County, Yueyanglou District, Yueyang County, and Hongjiang City were selected from 122 counties/cities/districts in Hunan Province.Newly registered TB patients were randomly selected from the four centres.As for the control group, healthy individuals were randomly selected from 14 community health service centres in Kaifu District, Changsha, and permanent residents of Xin'ansi Community health service centre, by the method of gender-age frequency matching.
All participants (>18 years of age) were selected from the Han Chinese population and were unrelated.
Inclusion criteria: (1) Case groups were diagnosed by the unified TB diagnostic criteria formulated by the Ministry of Health of China: sputum culture positive for MTB and chest X-ray showing characteristic TB lesions, or sputum culture negative for MTB but with signs of active TB on chest X-ray, strong positive PPD test and obvious TB-related clinical signs (e.g.cough, haemoptysis or haematochezia, chest tightness, shortness of breath, chest pain, fatigue, etc.; Ministry of Health of China 2023).(2) Inclusion as controls required a confirmed history of MTB exposure (Bacillus Calmette-Guérin (BCG) scar: the average diameter of PPD induration was ⩾10 mm; without BCG scar or no history of BCG vaccination: the average diameter of PPD induration was 5 mm), and normal chest X-rays.
Exclusion criteria: Patients who presented with conditions that can increase the risk of contracting TB were excluded.Such conditions included HIV infection, severe malnutrition, primary immune deficiency, long-term use of corticosteroids, immunosuppressive therapy, cancer, diabetes, and organ transplant.

Size of sample
In this study, the minor allele frequency (MAF) was 0.085, the estimated odds ratio (OR) was 1.8, α = 0.05 and β = 0.20, and we estimated the need for 464 case and control groups according to the sample size estimation formula for a 1:1 unmatched case-control study.

Environmental information collection
Environmental information of the study participants was collected using an epidemiological questionnaire.The basic contents of the questionnaire were collected through face-to-face enquiries between the investigator and the study subjects, which included demographic characteristics and lifestyle information (such as sex, age, height, weight, education, marital status, smoking, tea drinking, alcohol consumption, and BCG vaccination).The related study variables were as follows: (1) history of BCG vaccination: diameter of the stick mark ≥3 mm; and (2) smoking behaviour: smoking more than one cigarette a day for more than a year.

Collection of blood specimens
We collected 5 ml blood samples from study subjects using ethylenediaminetetraacetic acid (EDTA) anticoagulation tubes and stored them at −80 °C.The peripheral leukocyte genome was extracted from the whole blood samples using a blood DNA kit.

Selection basis of SNPs of PI3K/AKT genes
The selection of polymorphic loci was based on (1) SNPs in the PI3K/AKT related to TB susceptibility (as published on PubMed); and (2) SNPs in the PI3K/AKT genes associated with susceptibility to other diseases reported in previous literature.We used the dbSNP database (http://www.ncbi.nlm.nih.gov) to exclude those SNPs with minor allele frequencies less than 5% in East Asia.For each target gene, tagging SNPs were selected based on the pairwise r 2 ≥ 0.8.For specific information on genes, please refer to the supplementary documents (S1).

Genotyping
In this study, mass spectrometry technology was used to genotype the rs3730089, rs3756668, rs7236272, rs2699887, rs6443624, rs9838117 of PI3K gene and the rs1130233, rs2498786, rs2494752, rs2494750, rs2304186, rs7254617 of AKT gene.Detection of polymorphic loci: (1) Polymerase chain reaction (PCR) system (5 μL), containing 0.1 μL of dNTP (25 mM), 0.2 μL of Hotstar (5 U/uL), 0.4 μL of MgCl2 (25 mM), 1.0 μL of PCR primer mixture, 1.8 μL of ddH2O, 0.5 μL of 10X PCR buffer, and 1 μL of DNA (20-50 ng/μL), was used for DNA amplification; (2) Shrimp Alkaline Phosphatase (SAP) digestion was performed using an SAP digestion reaction system (2 μL), including 0.17 μL of SAP buffer, 1.53 μL of ddH2 and 0.3 μL of SAP enzyme; (3) A single base extension reaction system (2 μL), including 0.2 μL of iplex buffer, 0.2 μL of terminator mixture, 0.619 μL of ddH 2 O, 0.94 μL of extension primer mixture and 0.041 μL of iplex enzyme, was used for single base extension reaction; (4) Resin purification: the clean resin was electroplated to the resin plate (6 mg) for purification, and the resin extension product was transferred to the 384-well spectrum chip (Sequenom) for positioning; (5) Chip sampling: The Massarray nano-distributor RS1000 was started to transfer the purified resin extension product to the 384-well spectral chip; (6) Mass spectrometry detection and quality control: The matrix-assisted laser desorption ionisation time-of-flight mass spectrometry (MALDI-TOF MS) was used to detect the molecular weight of the extension product, and then MassArrayTYPER4.0 was used to determine the genotype of each SNP site by determining the difference in molecular weight of the extension product.

Statistical analysis
Epidata 3.0 was used to create the database and assign values to each variable.SPSS 22.0 was used to analyse general demographic data and environmental factors.Frequencies and percentages were used for the statistical description of categorical data.Chi-squared (χ2) tests were used to compare differences in distribution between case and control groups (p < 0.05, considered significant).The Hardy-Weinberg equilibrium was used to test whether the frequency of each genotype in the polymorphic locus of the control group differed from those in the ideal state.Due to the low frequency of homozygous mutant genotypes at SNP sites in this study, genotypes were dichotomised according to the dominant inheritance model.Multivariate unconditional logistic regression analysis was performed using the occurrence of TB as the dependent variable and six loci of the PI3K gene and six loci of the AKT gene as independent variables.Sex, age, marital status, educational background, body mass index (BMI), smoking status, alcohol drinking and tea drinking were used as covariates to exclude possible confounding risk factors.To account for multiple comparisons, the p-value from the multivariate unconditional logistic regression was adjusted using false discovery rate (FDR) correction to effectively control the false positive rate, and the corrected P-value (QFDR) ≤0.05 was considered statistically significant.Multifactor Dimensionality Reduction (MDR) method was used for gene-gene interaction.In order to avoid the decrease of statistical efficiency, sites with p < 0.05 after multiple comparisons were selected for subsequent analysis.A marginal structural linear odds ratio model was used to analyse gene-smoking interactions, and the relative excess risk due to interactions (RERI) was used to quantitatively estimate the magnitude of additive interactions.RERI is calculated as follows: RERI = RR 11 -RR 10 -RR 01 + RR 00 , and RERI > 0 suggested positive interactions.

Ethics declarations
All protocols were approved by the Xiangya School of Public Health Central South University Ethics Review Committee (XYGW-2018-11).Experiments were performed by relevant named guidelines and regulations.Informed consent was obtained from all participants.

Results
In total, 503 TB patients and 494 controls participated in this study.Previously, we showed that the two groups exhibited no statistically significant difference (p > 0.05) in terms of sex, age, marital status, educational background, and alcohol consumption, while differences in BMI, history of BCG vaccination, smoking status and tea consumption were statistically significant (p < 0.05) (Wang et al. 2020).The univariate analysis showed that the rs3730089 of the PI3KR1 gene was significantly associated with the risk to develop TB (p < 0.001).The Hardy-Weinberg equilibrium test showed that all six polymorphism sites of PI3K genes in the control group reached genetic balance (p > 0.05) (Table 1 and  Table S2).
After adjusting for the covariates of sex, age, marital status, educational background, BMI, smoking status, alcohol consumption, tea drinking, and BCG vaccination, multivariate unconditional logistic regression analysis showed that rs3730089 of the PI3KR1 gene was associated with an increased risk of TB.Compared with the genotype GG, genotypes GA, AA, and GA + AA at rs3730089 of PI3KR1 were significantly associated with the risk to develop TB (GA: OR (95%CI) = 1.562 (1.191-2.047),AA: OR (95%CI) = 2.282 (1.325-3.929),GA + AA: OR (95%CI) = 1.650 (1.274-2.138)).Genotypes GA, AA, and GA + AA at rs3756668; genotypes AG, GG, and AG + GG at rs7236272; genotypes GA, AA, and GA + AA at rs2699887; genotypes CA, AA, and CA + AA at rs6443624; and genotypes GT, TT, and GT + TT at rs9838117 of the PI3K gene did not show statistically significant association with the risk of TB (Table 1).The univariate analysis showed that the rs1130233 of the AKT1 gene was significantly associated with the risk to develop TB (p < 0.001).The Hardy-Weinberg equilibrium test showed that all the seven polymorphism sites of AKT genes in the control group reached genetic balance (p > 0.05; Table 2 and Table S2).
Multiple comparison results showed that the corrected P values of rs3730089 locus of PI3KR1 gene and rs1130233 locus of AKT1 gene were <0.05.Therefore, Multifactor Dimensionality Reduction (MDR) was used to analyse the effect of gene-gene interaction on TB susceptibility for rs3730089 and rs1130233.The results showed that the combination of the GG genotype of rs3730089 and the TT genotype of rs1130233 was a low-risk combination (OR = 0.530, p < 0.001).The combination of the GG + AA genotype of rs3730089 and the TC + CC genotype of rs1130233 was a high-risk combination (OR = 2.013, p < 0.001).This suggests that there may be an interaction between the rs3730089 locus of PI3KR1 and the rs1130233 locus of AKT1 that affects the risk of developing TB (Table 3).
We examined the impact of the interactions between PI3KR1/AKT1 and smoking on the incidence of TB using marginal structural linear odds models.Adjusted for the covariates of sex, age, marital status, educational background, BMI, tea drinking, alcohol drinking, and history of BCG vaccination, the RERI between rs3730089 in PI3KR1 and smoking was found to be 0.9608 (95% CI:0.5959, 1.3256), which suggests a positive interaction.In addition, the RERI between rs1130233 of AKT1 and smoking was −0.2321 (95% CI: -0.6618, 0.1976), which suggests no interactions (Table 4).

Discussion
In this study, we have found that rs3730089 in PIK3R1 and rs1130233 in AKT1 were associated with TB susceptibility, and the interaction of rs3730089 in PI3KR1 with smoking had a positive effect on the risk of TB.
Our results suggest that the rs3730089 mutations (GA, AA, GA + AA) in PI3KR1 significantly increased the risk of TB in the Chinese population.A previous population-based study reported a significant association between rs3730089 and type 2 diabetes, accompanied by an impact on phenotypic characteristics, such as obesity, insulin resistance, and lipid parameters in a Turkish population (Karadoğan et al. 2018).Another case-control study in a US population reported that the polymorphism of rs3730089 in PIK3R1 was associated with an increased risk of colon cancer (Li et al. 2008).PIK3R1 gene is located on chromosome 5 and encodes a p85α regulatory subunit.The rs3730089 variation was derived from a missense mutation in exon 6 of PIK3R1 (Jamshidi et al. 2006;Arcaro and Guerreiro 2007).Functional studies have shown that the rs3730089 polymorphism in the PIK3R1 gene leads to reduced p85α protein expression (Almind et al. 2002), and the p85α regulatory subunit not only acts as an important suppressor of the PI3K signalling pathways (Luo and Cantley 2005) but also inhibits the p110δ catalytic subunit of PI3K (Yu et al. 1998).The p110δ catalytic subunit affects macrophage production by cytokines, which are key regulators of the immune response to MTB infection (Liu et al. 2006;Yu et al. 2015).In addition, much evidence suggests that PI3K participates in autophagy by interacting with various regulatory proteins to form multiple complexes (Nascimbeni et al. 2017).MTB can largely evade elimination by the body's immune response by preventing the fusion of MTB phagocytic macrophages with lysosomes (Vergne et al. 2005).Therefore, the mutation of rs3730089 in PI3KR1 may increase susceptibility to TB by manipulating autophagy and affecting cytokine secretion.We found that the mutation of rs1130233 (AG, GG, AG + GG) in AKT1 significantly increases the risk of TB in the Chinese population.However, another study in a Chinese Han population did not find any association of rs1130233 with TB (Wang et al. 2010).That study analysed 101 TB patients and 106 controls, whereas our study involved 503 patients and 494 controls.We speculate that the sample size of that study may have been insufficient to detect significant differences in genotype frequencies.Another study conducted in Europe linked AKT1 rs1130233 with reduced survival in patients with pancreatic ductal adenocarcinoma (Avan et al. 2014).AKT1 is located on human chromosome 14 (14q32.32)and encodes a 56 kDa protein consisting of 480 amino acid residues (Kim et al. 2012).Rs1130233 is located on exon 8 and belongs to the promoter region of AKT1 (Piao et al. 2015).The polymorphism of rs1130233 in AKT1 can affect the expression and translation of AKT1 mRNA and thus AKT1 protein levels (Avan et al. 2014).AKT1 was shown to be an autophagy-associated kinase whose activation prevents the maturation of phagosomes (Xiang et al. 2020), and some studies have reported killing of MTB in macrophages by inhibiting AKT1 to induce phagosome maturation (Kuijl et al. 2007).Therefore, the mutation of rs1130233 in AKT1 can affect susceptibility to TB by affecting autophagy in the immune response.
In our study, adjusted for a range of possible important confounders, a marginal structural linear odds model analysis indicated a positive interaction between the mutation of rs3730089 in PI3KR1 and smoking on the risk of developing TB.These results suggest that smoking further increases the risk of TB in the GA and AA genotypes at the rs3730089 locus.Cigarette smoke contains a large number of toxic and harmful substances such as nicotine and carbon monoxide, which can lead to impaired function of lung macrophages, thereby affecting autophagy and apoptosis (Sopori 2002;Birrell et al. 2008), and PI3K/AKT are among the genes associated with the autophagy regulatory pathway.Since genetic mutations cannot be altered, the results of this study suggest that tobacco control may not only reduce the risk of tuberculosis by smoking itself, but also by reducing the positive interaction between smoking and rs3730089 of the PI3KR1 gene, which may provide some ideas for tuberculosis prevention.
We have found that the polymorphisms rs3730089 in the PI3KR1 gene and rs1130233 in the AKT1 gene were associated with the risk of TB development, but the underlying biological mechanism remains to be deciphered by further experimental animal studies.Moreover, the Chinese population was selected for this study and should be replicated for validation in other populations.The controls in this study were derived from a single community, which may produce some selection bias.Despite certain limitations of our study, its results are reliable.These findings can help identify populations at risk for TB and enable targeted preventive measures for high-risk populations.

Conclusion
We conclude that the mutation of rs3730089 (GA, AA, GA + AA) in the PI3KR1 gene and rs1130233 (AG, GG, AG + GG) in the AKT1 gene are associated with susceptibility to TB in a Chinese population, and there was a significant positive interaction between rs3730089 in the PI3KR1 gene and smoking on susceptibility to TB.
HWE-P: Hardy-Weinberg equilibrium-p value.*p < 0.05.**The p value was adjusted by the method of false discovery rate (fDR).TB: Tuberculosis.a p < 0.05.# Adjusted for the covariates of sex, age, marital status, educational background, body mass index, smoking status, alcohol drinking, tea drinking, and BCg vaccination.TB: Tuberculosis.

Table 2 .
AKT gene polymorphism vs. tuberculosis incidence.Adjusted for the covariates of sex, age, marital status, educational background, body mass index, smoking status, alcohol drinking, tea drinking, and BCg vaccination.TB: Tuberculosis.

Table 3 .
impact of interactions between rs3730089 and rs1130233 on susceptibility of tuberculosis.Adjusted for the covariates of sex, age, marital status, educational background, body mass index, smoking status, alcohol drinking, tea drinking and BCg vaccination.TB: Tuberculosis.
# #Adjusted for the covariates of sex, age, marital status, educational background, body mass index, tea drinking, alcohol drinking, and BCg vaccination.a p < 0.05, RERic, crude relative excess risk of interaction; RERi a , adjusted relative excess risk of interaction.