Comprehensive analysis of protein acetyltransferases of human pathogen Mycobacterium tuberculosis

TB remains a major threat for global health largely due to vast mortality and morbidity. In 2017, an estimated 17 million people were infected by TB and 1.6 million died from this disease (WHO, 2018). The increase of incidences with MDR_TB (multidrug resistant TB) and XDR_TB (extensively drug-resistant TB) further exacerbates the difficulty of TB patients’ treatment. It is imperative to find new targets and new anti-TB drugs. M. tuberculosis is the causative agent of TB and has developed the ability to survive in a dormant state for long time under stress conditions within its human host or to resist various antibiotics [1]. Rv0262c, a conserved aminoglycoside 2′-N-acetyltransferase (AAC(2′)-Ic) in mycobacteria, can acetylate all known aminoglycosides including ribostamycin, neomycin B, gentamicin, and tobramycin bearing 2′ amino group [2,3]. Moreover, Rv0262c is capable of performing O-acetylation in kanamycin A and amikacin at 2′ position [4]. Rv3225c, a putative phophotransferase containing GNAT (GCN5-related acetyltransferase) domain in N terminus, was found to have low level of aminoglycoside-modifying activity conferring resistance to aminoglycoside antibiotic in mycobacteria [5]. Eis (rv2416c, enhanced intracellular survival gene), one lysine N-acetyltransferase, was found its overexpression conferring M. tuberculosis kanamycin resistance [6]. About 80% of clinical isolates harboring eis promoter mutations exhibited low-level kanamycin resistance [7]. In addition, Eis has an unexpected function of acetylating capreomycin thereby deactivating the drug [8]. In 2012, Green et al. screened three small molecule libraries and found 25 inhibitors that display specific and strong inhibitory activity against Eis in vitro [9]. Thus, these acetyltransferases has emerged as targets for inhibitor design.

Integrity of cell envelope is crucial for M. tuberculosis survival, virulence, and persistence [10]. PG (Peptidoglycan), mycolic acids, and AG (arabinogalactan) are the major constituents of cell envelope in M. tuberculosis [11]. GlcNAc (Amino sugar N-acetylglucosamine), a critical component of PG and UDP-GlcNAc (an activated type of amino sugar), is an indispensable precursor for different cell wall components [11]. Bi-functional enzyme GlmU (N-acetylglucosamine-1-phosphate uridyltransferase), encoded by rv1018c gene, contains both acetyltransferase and uridyltransferase domains [12,13]. GlmU is involved in the final two steps of UDP-N-acetyl-d-glucosamine (UDP-GlcN) biosynthesis process [12,13]. The deletion of glmU gene can change the cell wall structure, and GlmU is necessary for mycobacterial survival in THP-1 cells and guinea pigs [14]. In addition, GlmU interacting with IL-8 can facilitate the pathogen M. tuberculosis entry into human neutrophils [15]. In this work, we found the presence of acetyltransferases with different substrates by analyzing the M. tuberculosis proteome. Comparative proteomic analyses showed that there are homologues of several M. tuberculosis acetyltransferases in opportunistic and non-pathogenic mycobacteria. We also found that one acetyltransferase can be lysine acetylated, succinylated and glutarylated, and many acetyltransferases with their neighboring genes are conserved in mycobacteria.

All characterized and predicted acetyltransferases in the M. tuberculosis H37Rv proteome were identified by searching the keywords: ‘acetyltransferase’ and ‘acetylase’ in the NCBI (National Center for Biotechnology Information). The detailed workflow was shown in Supplementary Figure S1. The proteomes of the 14 mycobacterial species downloaded from NCBI ftp were used in this study (Supplementary Table S3). The genomic map was created by using the DNAplotter tool [16].

M. tuberculosis H37Rv acetyltransferases found in the above step were compared with the other 13 species as shown in Supplementary Table S4 for the identification of homologous protein using Blast. Two proteins were treated as homologous: identity value ≥ 50%, query coverage ≥ 70%.

The following bioinformatics analysis followed previous report [17]. The Antigenic Index for all acetyltransferases was predicted through the VaxiJen v2 webserver, and the cut-off value was set 0.4 [18]. GlobPlot (http://globplot.embl.de/) website was conducted for searching the globularity and disorder in the acetyltransferases protein sequences [19]. The ProtParam tool of ExPASy (http://web.expasy.ogr/protparam/) was performed to predict the GRAVY (Grand Average of Hydropathicity), aliphatic indices, instability indices, and in vivo half-life of these proteins [20,21].

Acetyltransferases and their neighboring genes were screened to analyze their co-occurrence and predict functional associations through TB database (http://www.tbdb.org/) [22].

All public transcriptomic and proteomic studies data were downloaded from the internet, and these data were used to retrieve the expression patterns of acetyltransferases under different conditions including oxygen-depleted model, nutrient starvation model, phagosome model, acid-nitrosative multi-stress, and mice model [23–27]. At least a 2-fold decrease or increase in the expression intensity of acetyltransferases is treated as significant. All post-translation modifications (PTMs) data including acetylation, succinylation, and glutarylation about acetyltransferases were obtained from our published articles [28–30].

All acetyltransferases including characterized, possible, probable, or hypothetical proteins were searched in the proteome of M. tuberculosis H37Rv, and a total of 47 acetyltransferases was found (Supplementary Table S1). Among these acetyltransferases, 9 are amino-acid acetyltransferase, 4 are antibotic acetyltransferases, 6 are other acetyltransferases, and the others are putative proteins without known functions. These proteins are mapped on a circle map with their Rv numbers (Figure 1A), representing 1.2% of the proteome of M. tuberculosis. Functional catagories of all the 47 acetyltransferases were retrieved from TubercuList database [31], and these proteins were divided into six categories based on their functionality as shown in Figure 1B. Twenty-four acetyltransferases belong to the intermediary metabolism and respiration category while eleven acetyltransferases belong to lipid metabolism category. These results not only show the variability of acetyltransferases, but also point to their probable functional importance.

To explore whether these acetyltransferase and their homologues play an important role in the pathogenicity and metabolic processes of mycobacterium, these 47 acetyltransferases of M. tuberculosis were compared with various representative pathogenic, opportunistic and non-pathogenic mycobacteria species (a total 12) (Figure 2). M. bovis was found to have the highest number of homologous acetyltransferases in M. tuberculosis, while M. leprae had the lowest number of homologues. When moving from pathogenic to opportunistic to non-pathogenic bacteria, we found a reduction in the number of M. tuberculosis H37Rv homologues. On average strict pathogens have around 40 acetyltransferases; however, opportunistic and non-pathogenic mycobacteria have 37 and 38, respectively. Notably, the reduced number may be due to not synthesized in the species or the technical limitations resulting in unidentified acetyltransferases. Hence, the existence of these acetyltransferases in different mycobacterial species should be validated by experimental approaches.

MTB complex including M. tuberculosis, M. bovis, M. canetti, M. africanum, M. microti, M. pinnipedii, M. mungi, M. caprae, and M. suricattae can cause tuberculosis in humans or other organisms [32]. Among the 47 putative acetyltransferases, Rv0032 (BioF2) annotated as putative 8-amino-7-oxononanoate synthase 2 with N-Acyltransferase superfamily region was only present in the MTB complex, but absent in other pathogenic, opportunistic pathogenic or non-pathogenic species of mycobacterium (Supplementary Table S2). Biotin, a critical micronutrient, is an essential enzyme cofactor for biotin-dependent enzymes such as pyruvate carboxylase and acyl-CoA carboxylase in M. tuberculosis [33]. Two BioF proteins are present in the M. tuberculosis genome, namely, BioF1 encoded by rv1569 gene and BioF2 [34]. BioF1 is KAPA (7-Keto-8-aminopelargonic acid) synthase and is responsible for the first reaction step in the biotin biosynthesis pathway [33]. Using genome-wide genetic screens method, Sassetti et al. had found that M. tuberculosis bioF mutants (A and B) resulted in rapid clearance in the early phase of infection, and showed significantly decreased growth rates in mouse lung and spleen [35]. The growth of transposon insertion inactivated BioF1 M. tuberculosis was slowed, but returned to normal when supplemented with biotin [35]. However, the role of bioF2 in M. tuberculosis biotin metabolism remains unclear and the potential characteristics of acetyltransferase need to be verified by the experimental results. Notably, though BioF2 was only present in MTB complex, it does not mean that it is a specific enzyme to MTB just based on the current information. Furthermore, the other acetyltransferases that are shared between MTB complex and other mycobacterial species might also have essential functions in MTB complex, which need to be verified in the further study.

We used a VaxiJen tool to analyze antigenicity of 47 acetyltransferases, and found 40 possible antigens and 7 non-antigenic proteins, which indicated that most of acetyltransferases proteins (85%) in M. tuberculosis are antigenic in nature, reflecting the immunomodulatory nature of these acetyltransferases. High antigenic regions within M. tuberculosis proteins have been reported involved in both humoral and T-cell responses in vitro or in clinical samples [36]. Mycolic acids, one important component of the unique cell wall in mycobacteria, constitute a bracket for lipid antigens to stimulate CD1-restricted T cells [37]. In 2004, Bhakta et al. demonstrated that Rv3566c (arylamine N-acetyltransferase, nat) is involved in the biosynthetic pathway of mycolic acids and complex lipids [38]. nat deletion M. bovis BCG strain showed multiple phenotypes such as postponed entry into log phase, changed morphology and lipid composition of cell wall, and increased intracellular killing rate by mouse macrophage cell line RAW264.7 [38].

Due to few virulence genes identified in M. tuberculosis [39]; therefore, it is urgent to find and identify the novel virulence genes. Virulent M. tuberculosis H37Rv strain is the main pathogen that can cause tuberculosis, while H37Ra strain loses the pathogenicity to a large extent [40]. The avirulent M. tuberculosis H37Ra strain is different from virulent M. tuberculosis H37Rv strain in many aspects including genomic insertion, deletion, and frame shift [40]. These two strains are the good material to find the potential virulence gene in M. tuberculosis. We analyzed protein sequence differences of acetyltransferases between H37Rv and H37Ra strain, and found that two acetyltransferases including Rv3027c and Rv3423.1 have a partial or whole deletion in the sequence length (Figure 3). rv3027c gene codes for possible GCN5-related N-acetyltransferase while its homologue in H37Ra, MRA_3058, has an N-terminal deletion (36aa), resulting in a 19.8% decrease in the hydrophilicity of the protein. Globular domains for proteins have special functions, whose deletion or addition might cause function loss or gain [41]. Comparing with Rv3027c, MRA_3058 has almost 45 amino acids increase in globular domain (Figure 3). Rv3423.1, one hypothetical protein, has no homologue in H37Ra. Recently, Joes et al. demonstrated that Rv3423.1 is a new histone acetyltransferase, which can acetylate histone H3 of host cell at the lysine 9/lysine 14 positions [42]. In addition, Rv3423.1 protein can be detected in the culture filtrate of virulent M. tuberculosis but not avirulent strains [42]. This indicates that these two acetyltransferases may play a critical role in the virulence of M. tuberculosis H37Rv and might underlie the attenuation of M. tuberculosis H37Ra.

M. leprae, the causative agent of leprosy, has much reduced genome than M. tuberculosis [43]. Both M. tuberculosis and M. leprae belong to prototypical intracellular pathogens that have evolved ways to survive in the intracellular phagosomes [44]. To explore whether these 47 acetyltransferases involve in the same cellular process, we compared the quantity difference of acetyltransferases between M. tuberculosis and M. leprae. Among the 47 acetyltransferases present in M. tuberculosis, 25 acetyltransferases have homologs while 22 proteins did not show any homology in M. leprae. We identified eight acetyltransferases, namely Rv1074c, Rv1135A, Rv1867, Rv3523, Rv3556c, Rv3566c, Rv2669 and Rv3027c, which are present in all other mycobacteria but absent in M. leprae. Rv1074c was identified from M. tuberculosis culture filtrate proteins [45]. Rv1867 was identified as potential tuberculosis drug target by comparative analyses of genomes from M. tuberculosis and human [46]. Rv3523 was predicted as cholesterol metabolism related protein [47]. Rv3556c is found to be expressed in S7 clinical strain but inhibited in laboratory H37Rv when bacteria encounter hypoxia condition [48]. Recent study had demonstrated that kstR2 (a TetR-type transcriptional repressor) and rv3556c are de-repressed by cholesterol in M. Tuberculosis [49], implying the role in the control of cholesterol utilization. Rv1135Awith the condensing enzyme activity can catalyze a decarboxylating or non-decarboxylating Claisen-like condensation reaction and may involve in the synthesis and degradation of fatty acids [50].rv2669 gene was found to be up-regulated in senX3–regX3 (two-component regulatory system) M. tuberculosis mutant strain [51]. This mutant strain showed the growth defect in the macrophages, and was attenuated in both immunocompetent and immunodeficient mice [51]. M. tuberculosis mutants in Rv3027c were attenuated for growth in macrophage [52].

Virulence factor database (VFDB) is a source for providing the current knowledge about virulence factors from different bacterial pathogens (http://www.mgc.ac.cn/VFs/) [53]. To understand whether these different acetyltransferases are related with virulence in M. tuberculosis, VFDB was searched, resulting two acetyltransferases of M. tuberculosis, Rv1347c (lysine N-acyltransferase MbtK) and Rv2416c (enhanced intracellular survival protein). It was previously shown that Rv1347c is involved in the mycobactin biosynthesis [54]. Seven acetyltransferases also belong to the category of essential genes according to the bibliometric approach [55]. Correspondence of acetyltransferases with virulence factors and essential genes are shown in Figure 4A and Supplementary Table S1. Essential genes are indispensable for survival. These essential genes are: rv1653 (arginine biosynthesis bifunctional protein ArgJ), rv2215 (dihydrolipoamide acyltransferase, DlaT), rv3341 (homoserine O-acetyltransferase), rv3546 (acetyl-CoA acetyltransferase), rv1018c (bifunctional protein GlmU), rv2747 (amino-acid acetyltransferase), and rv1347c. Apart from Rv3546, other acetyltransferases are essential genes validated through high density mutagenesis experiment [56].

PTM (protein post-translational modification) functions in the regulation of different cellular processes in bacteria and eukaryotes [57]. Recently, our group identified 658 acetylated proteins with 1128 acetylation sites, 626 succinylated proteins with 1545 succinylation sites, and 24 glutarylated proteins with 41 glutarylation sites by using proteomics methods. To investigate whether three PTMs including acetylation, succinylation and glutarylation occur in these acetyaltransferases, we compared acetyltransferases with our previous data. The comparison showed that only one acetyltrasferase, Rv2215, can be acetylated, succinylated and glutarylated (Figure 4B and Supplementary Table S5). Rv2215 is dihydrolipoyllysine-residue acetyltransferase component of pyruvate dehydrogenase complex M. tuberculosis H37Rv rv2215 mutant showed significant growth defect in vitro and more sensitive to nitrosative stress but not heat or INH [58]. In addition, rv2215 mutant survival in bone marrow-derived macrophages from C57BL/6 and iNOS -/- mice was decreased [58]. Previous studies had demonstrated that the activities of metabolic enzymes such as acetyl-CoA synthetase (ACS) in M. tuberculosis can be regulated by PTMs [59,60]. Therefore, we guess that the enzyme activity of Rv2215 may be regulated by these PTMs. it is very necessary to study the role of these three PTMs in the regulation of Rv2215 by site-specific mutagenesis experiments in vitro and in vivo.

To explore the occurrence and genome organization of these acetyltransferases, we performed analysis in the TB database (http://www.tbdb.org/) website. The results showed that the adjacent genes (at least one) of most acetyltransferases exist in other mycobacteria, even Rhodococcus and Streptomyces. For example, the adjacent genes of Rv1653c including Rv1652, Rv1654, Rv1655, and Rv1656 are highly conserved in M. leprae TN, M. smegmatis str. mc²155, M. avium 104, Rhodococcus sp.RHA1, and Streptomyces avermitilis MA-4680 (Supplementary Figure S2). However, the adjacent genes of several acetyltransferases are not conserved. For instance, there are no homologous proteins of the adjacent genes for Rv2416c.

To determine whether these acetyltransferases function during infection in M. tuberculosis, available data from previous transcriptomic and proteomic studies were used to analyze those acetyltransferases that show differential expression in infection models or under in vitro conditions mimicing aspects of infection. The analysis results are shown in Supplementary Table S2. Eighteen acetyltransferases showed differential expression in at least one of the experimental conditions while five acetyltransferases were essential for infection in mice model.

Seven acetyltransferases (Rv3523, Rv3546, Rv3556c, Rv1347c, Rv2416c, Rv1505c, and Rv3535c) are induced for infection in the BM macrophages model. In 2000, Wei et al. selected the M. smegmatis transformants containing an M. tuberculosis H37Rv plasmid library in human histocytic macrophage-like U-937 cell line, and found an enhanced intracellular survival gene (Eis), Rv2416c [61]. Both Eis enzymes from M. tuberculosis and M. smegmatis acetylated capreomycin and several lysine-containing compounds, and this acetylation of capreomycin was found at the ɛ-amine of the β-lysine side chain [61]. Recently, Sowajassatakul et al. found that overexpression of eis could occur without a mutation in the promoter region and be detectable in amikacin- and kanamycin-resistant M. tuberculosis clinical strain [62]. Ye et al. demonstrated that Eis proteins have the ability to acetylate many arylalkylamines, and are a novel family of arylalkylamine N-acetyltransferase AANAT (EC 2.3.1.87) [63]. Eis_Mtb from M. tuberculosis prefers to acetylate histamine and octopamine, while Eis_Msm from M. smegmatis uses tyramine and octopamine as substrates [63]. In addition, M. tuberculosis Eis can modulate autophagy, inflammation, and cell death through redox-dependent signal pathway to repress host innate immune defenses [64]. Kim et al. showed that M. tuberculosis Eis can acetylate DUSP16 (dual-specificity protein phosphatase 16)/MKP-7 (mitogen-activated protein kinase phosphatase-7) at Lys55, to inhibit JNK-dependent autophagy, phagosome maturation, and ROS (reactive oxygen species) generation. It will be very interesting to study the physiological role of the other essential acetyltransferases.

By in silico methods, acetyltransferases were predicted from the available complete proteome of M. tuberculosis. This is the first study establishing whole map of M. tuberculosis acetyltransferases and their evolution, antigen, and genomic context. We have identified 47 acetyltransferases in M. tuberculosis. Several acetyltransferases have well established roles in virulence, antibiotic resistance, and metabolism.

Aminoglycoside acetyltransferases, members of the GNAT superfamily, can confer resistance to aminoglycoside antibiotics in bacteria [65]. CysE (Serine acetyltransferase), one enzyme involved in L-cysteine biosynthesis pathway, exists in bacteria and plants, but not in humans [66]. CysE can catalyze the chemical reaction transferring the acetyl from AcCoA (acetyl-CoA) to L-Ser (L-serine). Rv2335 was identified and characterized as an serine acetyltransferase in M. tuberculosis [67], and M. smegmatis homologous cysE knockout mutant strain showed significant morphological changes and suppressed the bacteria growth [68]. NCBI blast results showed that homologous proteins of these acetyltransferases are not found in human and other mammals. This implies that those acetyltransferases can become the potential drug targets.

In the present study, we found that Rv2215 can be acetylated, succinylated, and glutarylated. It have been demonstrated that lysine acetylation in protein catalyzed by specific acetyltransferases play an important role in bacterial metabolism, stress response, and virulence [69,70]. Rv0998, one GNAT family member having a cyclic nucleotide binding domain, can acetylate several mycobacterial fatty acyl CoA ligases (FACLs), thereby repressing FACLs enzyme activities and regulating fatty acid metabolism [71]. Therefore, it will be very interesting to study whether Rv0998 is the key enzyme responsible for acetylating DlaT protein. In 2015, Tran et al. using high-throughput screen method identified two compounds (6624116, 5655606) that can inhibit the acetyltransferase activity of GlmU in M. tuberculosis and increase the anti-TB activity combination with other anti-TB drugs [72]. In 2016, Garzan et al. discovered two Eis inhibitors as kanamycin adjuvants to kill drug-resistant M. tuberculosis through high-throughput screening [73]. Later, they developed pyrrolo (1,5-a) pyrazine-based analogues as new effective inhibitors of Eis [74]. These inhibitors may become promising anti-TB drugs. However, some of these acetyltransferases were classified into hypothetical proteins without any predicted function. Therefore, it will be interesting to further study the functional role of these acetyltransferases in the adaptation of the M. tuberculosis in in vivo systems.

Designed the experiments: L.X. and J.X; Acquisition of data: L.X., W.Y., X.F., and J.X.; Analysis and interpretation of data: L.X., W.Y., X.F.; Draft of the manuscript: L.X. and J.X.; Critical revision of the manuscript for intellectual content: L.X., W.Y., X.F., and J.X.

The authors declare that there are no competing interests associated with the manuscript.

This work was supported by National Natural Science Foundation [grant numbers 81301394, 81371851, 81511120001, 81071316 and 81271882]; Key Project of Science and Technology Research of Henan Provincial Department of Education [grant number 20A180001]; New Century Excellent Talents in Universities [grant number NCET-11-0703]; Collaborative Innovation Center of Infectious diseases [grant numbers PXM2015_014226_000058 and 3500-115215]; Excellent Ph.D. thesis fellowship of Southwest University [grant numbers kb2010017 and ky2011003]; National Megaprojects for Key Infectious Diseases [grant number 2008ZX10003-006]; the Fundamental Research Funds for the Central Universities [grant numbers XDJK2012D007, XDJK2013D003, XDJK2014D040 and XDJK2016D025]; Graduate research and innovation project of graduate in Chongqing [grant number CYS14044]; The Chongqing Municipal Committee of Education for postgraduates excellence program [grant number YJG123104]; and the undergraduates teaching reform program [grant number 2013JY201].

AcCoA

acetyl-CoA

AG

arabinogalactan

CysE

serine acetyltransferase

DUSP16

dual-specificity protein phosphatase 16

Eis

enhanced intracellular survival gene

FACLs

fatty acyl CoA ligases

GlmU

N-acetylglucosamine-1-phosphate uridyltransferase

GNAT

GCN5-related acetyltransferase

MKP-7

mitogen-activated protein kinase phosphatase-7

PG

peptidoglycan

PTM

protein post-translational modification

PTMs

post-translational modifications

TB

tuberculosis