Universal stress protein Rv2624c alters abundance of arginine and enhances intracellular survival by ATP binding in mycobacteria

The universal stress protein family is a family of stress-induced proteins. Universal stress proteins affect latency and antibiotic resistance in mycobacteria. Here, we showed that Mycobacterium smegmatis overexpressing M. tuberculosis universal stress protein Rv2624c exhibits increased survival in human monocyte THP-1 cells. Transcriptome analysis suggested that Rv2624c affects histidine metabolism, and arginine and proline metabolism. LC-MS/MS analysis showed that Rv2624c affects the abundance of arginine, a modulator of both mycobacteria and infected THP-1 cells. Biochemical analysis showed that Rv2624c is a nucleotide-binding universal stress protein, and an Rv2624c mutant incapable of binding ATP abrogated the growth advantage in THP-1 cells. Rv2624c may therefore modulate metabolic pathways in an ATP-dependent manner, changing the abundance of arginine and thus increasing survival in THP-1 cells.

regulon 11,12 , suggesting that universal stress proteins modulate latent infection. The M. tuberculosis universal stress protein Rv2623 is known to be involved in mycobacterial persistence 13 , and Rv1996 in isoniazid susceptibility 14 . The exact physiological roles of the other mycobacterial universal stress proteins, however, are unknown.
A microarray study has shown that rv2624c mRNA is highly induced by nitrosative stress, a host innate immune system-induced natural stressor against infection 15 . In addition, Rv2624c has been shown to be an immunogenic protein, the antibody against Rv2624c being detected in serum from TB patients serum 16 . These studies suggest that Rv2624c affects the infected host cell immune response. Knockout of a usp gene is compensated for by other universal stress proteins in mycobacteria 17 , while overexpression of a universal stress protein provides information about its function as overexpression increases its effect over that of other universal stress proteins. For example, overexpression of rv1996 increased isoniazid susceptibility 14 . Overexpression of universal stress protein rv2623 arrested growth of both M. smegmatis and M. tuberculosis 13 . In this study, we overexpressed rv2624c in mycobacteria and investigated the physiological functions of Rv2624c. We showed that Rv2624c increases the survival of mycobacteria in hypoxia and bacillary-infected THP-1 cells. Transcriptome analyses suggested that Rv2624c affects histidine metabolism and arginine and proline metabolism. Furthermore, LC-MS/MS analysis showed that Rv2624c affects the abundance of arginine, a modulator of both mycobacteria and infected THP-1 cells. In addition, biochemical analysis showed that Rv2624c is a nucleotide-binding universal stress protein. The growth advantage in THP-1 cells is ATP dependent, as an Rv2624c mutant incapable of ATP-binding abrogated the growth advantage of M. smegmatis mc 2 155 cells overexpressing Rv2624c.

Results and Discussion
Overexpression of Rv2624c increases mycobacterial survival in the Wayne mycobacterial model of latency. In the M. tuberculosis genomic context, rv2624c neighbors rv2623 but is in the opposite orientation ( Fig. 1A) 17 . Rv2623 is the first mycobacterial universal stress protein identified that has been shown to be involved in M. tuberculosis persistence in vivo and to regulate mycobacterial growth in vitro 13 . Like Rv2623, Rv2624c has been shown to be highly induced under NO and hypoxia, and within infected macrophages 9,11,12 . However, the function of Rv2624c has not been reported as yet.
We first examined whether overexpression of rv2624c affects mycobacterial growth. Consistent with a previous study showing that Rv2623 overexpression arrests growth 11 , the pMV261-Rv2623/mc 2 155 strain grew slowly compared with the control M. smegmatis mc 2 155 strain harboring an empty vector (pMV261/mc 2 155) (Fig. 1B). In contrast, under the experimental conditions used here, growth of pMV261-Rv2624c/mc 2 155 was comparable with pMV261/mc 2 155, indicating that overexpression of rv2624c did not affect mycobacterial growth (Fig. 1B).
As universal stress proteins play roles in latent infection, we investigated the effects of Rv2624c in the Wayne mycobacterial latency model in which mycobacterial cultures are grown under gradual oxygen concentration depletion to force bacilli to enter a physiologically latent state. Bacteria (OD 600 of 0.01) were seeded into vials with a headspace: liquid ratio of 1:2 supplemented with the O 2 probe methylene blue, sealed and incubated with slow stirring (Fig. 1C). No difference in viability was detected on the first day after inoculation (Fig. 1C). Differences in the viability of pMV261-Rv2624c/mc 2 155 under hypoxia compared to pMV261/mc 2 155 were observed on days 3, 5, and 7 after inoculation (Fig. 1C). These results suggest that Rv2624c affects viability under hypoxic conditions. Overexpression of rv2624c in M. smegmatis increases survival in human monocyte THP-1 cells. Formation of granulomas is a hallmark of the host response to infection with M. tuberculosis. The hypoxic microenvironment within granulomas is an important factor leading to mycobacterial latency [18][19][20][21][22] . Rv2624c activity is required for survival under hypoxia, suggesting that latency gene rv2624c likely modulates the host immune system. In addition, studies have shown that Rv2624c is an immunogenic protein 16,23 . Macrophages are central innate immune cells that play a key role in the host response against pathogens. Using a macrophage-killing assay, we evaluated the intracellular survival of mycobacterial strains in the macrophage-like cell line THP-1 after a synchronized infection (Fig. 1D). After extensive washing, cells were lysed and spread on 7H10 medium (T 0 ) to determine how many bacilli had infected the cells. As shown in Fig. 1D, the initial numbers of pMV261/mc 2 155 bacteria infecting THP1 cells was comparable to that of pMV261-Rv2624c/mc 2 155, indicating there was no statistical difference between strains in their entry into host cells. At 4 h post-infection, the number of pMV261-Rv2624c/mc 2 155 CFUs (8.23 × 10 5 ± 2.4 × 10 4 ), was significantly higher than that of pMV261/mc 2 155 CFUs (2.4 × 10 5 ± 3 × 10 4 ). This result suggests that overexpression of rv2624c increased mycobacterial survival in THP-1 host cells. In contrast, overexpression of Rv2623 attenuated growth in THP-1 cells (1.2 × 10 5 ± 1.5 × 10 4 CFUs) (Fig. 1D). Those results showed that overexpression of rv2623 decreased survival and overexpression of rv2624c increased survival in human monocyte THP-1 cells. mc 2 155 and pMV261-Rv2624c/mc 2 155 cells have distinct metabolic patterns. Transcriptional reprogramming plays important roles in bacterial responses to various stressors. To elucidate why pMV261-Rv2624c/mc 2 155 cells have a growth advantage over pMV261/mc 2 155 cells under the tested conditions (hypoxia and infection of THP-1 host cells), we examined changes in mRNA expression using RNA sequencing. Comparing the RNA of pMV261-Rv2624c/mc 2 155 cells to that of pMV261/mc 2 155 cells showed modest expression differences. Two hundred and eighty-six were upregulated more than 2-fold and 33 genes were upregulated more than 4-fold, respectively (p < 0.01), while 11 genes were downregulated. As different genes cooperate to perform their biological functions, pathway-based analysis gives insight into the Differential Expressed Genes (DEGs) involved in biological functions. KEGG Pathway Analysis 24 showed that "arginine and proline metabolism" (Q = 0.004), "tryptophan metabolism" (Q = 0.013) and "histidine metabolism" (Q = 0.013) pathways are significantly enriched in DEGs compared to the whole M. smegmatis genome. As listed in Table 1, 15 DEGs are involved in histidine metabolism (pathway ID ko00380) and 19 upregulated genes are involved in arginine and proline metabolism. Interestingly, using KEGG-User Data Mapping ( Fig. 2A), four genes were shown to be involved in "histidine metabolism", which includes the amino acid synthesis pathway from histidine to glutamate, and seven genes were shown to be involved in "arginine and proline metabolism", an arginine synthesis pathway. Arginine is an important modulator of infected macrophages and the pathogens which have infected them 25,26 , and arginine as an adjuvant for chemotherapy for active tuberculosis improves clinical outcomes 27 . There are many catabolic pathways for pathogens to degrade arginine, including the arginine deiminase pathway and the arginase pathway, the former supporting growth under anaerobic conditions and the latter supporting growth under aerobic conditions. Therefore, arginine metabolism is essential for both the host and the pathogen, and competition for arginine and thus may determine the outcome of infection 28 . Some pathogens have been shown to alter their arginine-dependent metabolic activities when infecting host cells. For example, M. marinum induces argS, an arginine metabolism gene, when it is inside host cells 29 . Results from our transcriptome analysis suggest that differences in the abundance of arginine may be important in mycobacterial survival within infected macrophages.

Overexpression of Rv2624c increases the abundance of arginine. To examine whether Rv2624c
affects arginine synthesis, we measured the abundance of arginine in pMV261 and pMV261-Rv2624c bacterial cells using mass spectrometry. As metabolites are easily altered by simple experimental procedures such as centrifugation, we used the filter-culture approach described by de Carvalho et al. 30 to extract metabolites without significantly altering metabolite profiles. We then compared the abundance of arginine in pMV261/mc 2 155 and pMV261-Rv2624c/mc 2 155 using LC-MS/MS. The structure of arginine and its two major fragments at m/z 70 and 116 is shown in Fig. 2B. The abundance of arginine was statistically higher in pMV261-Rv2624c/mc 2 155 than in pMV261 (Fig. 2C).
Rv2624c is a nucleotide-binding universal stress protein. Universal stress proteins are an ancient protein family, present in organisms from bacteria to plants, but absent in animals. Based on their ATP-binding ability, universal stress proteins have been assigned to two subclasses: one whose members do not bind nucleotides and the other whose members bind nucleotides 31 . A previous study has shown that mycobacterial universal stress protein Rv2623 is an ATP-binding protein and that the ATP binding of Rv2623 regulates bacillary growth 13 . Alignment of universal stress proteins, including the well-studied Rv2623, predicted that Rv2624c is an ATP-binding protein (Fig. 3), suggesting Rv2624c might have ATP-dependent biological functions. In addition, transcriptome analysis suggested that overexpression of Rv2624c affects the mycobacterial histidine, and arginine and proline metabolic pathways, which both use ATP. To determine if Rv2624c has ATP-dependent biological functions, we investigated the biochemical characteristics of Rv2624c protein to elucidate whether it affects growth in infected host cells in an ATP-dependent manner. M. tuberculosis Rv2624c was cloned and expressed in E. coli. SDS-PAGE analysis of purified His 6 -tagged Rv2624c protein showed a single band around the predicted molecular mass of ~30 kDa that was confirmed by immunoblotting to be Rv2624c (data not shown). Gel filtration analysis of native His 6 -Rv2624c proteins showed that the majority of the purified Rv2624c had a molecular mass of ~30 kDa (Fig. 4A), indicating that Rv2624c is a monomer under native conditions. Similar to Rv2623 13 , purified Rv2624c proteins from E. coli bound tightly to ATP and ADP after extensive purification steps ( Figure S1), indicating that Rv2624c is a nucleotide-binding protein. To gain insight into the function of ATP binding, the amino acid sequence of Rv2624c was compared with that of Rv2623 (Fig. 3). In a previous study on Rv2623, G117 and D15 were identified as key amino acids for ATP binding 13 . When we mutated the corresponding amino acids in Rv2624c (D17, equivalent to D15 in Rv2623, and G109, equivalent to G117 in Rv2623) to E and A, HPLC analysis of nucleotides extracted from Rv2624c D17E and Rv2624c G109A showed that both mutations had abrogated the protein's ATP-binding ability (Fig. 4B). We next compared the growth of mycobacterial strains overexpressing Rv2624c, Rv2624c D17E and Rv2624c G109A . When measuring the OD 600 values, no growth differences were detected among the strains (data not shown). Consistent with this result, the size of mycobacterial colonies was similar (Fig. 5A). In contrast, strains overexpressing Rv2623 showed arrested growth 13 and formed small colonies  5A). We also examined the survival percentage of these mycobacterial strains in infected macrophage cells.
In contrast to their growth on 7H10 medium, we observed statistically significant differences in growth between mc 2 155 cells overexpressing Rv2624c and mc 2 155 cells overexpressing Rv2624c D17E (p < 0.001), but not mc 2 155 overexpressing Rv2624c G109A . As in the macrophage killing assay studies, the percentage of M. smegmatis strains overexpressing mutant Rv2624c D17E that survived in infected THP-1 cells was reduced (Fig. 5B). Growth in THP-1 cells infected with cells expressing the G109A mutant was comparable with that of cells expressing wild-type Rv2624c (Fig. 5B). The distinct effects exhibited by the wild type, and the G109A and D17E mutants defective in ATP binding suggest a correlation between survival in infected host cells (THP-1 cells) and ATP binding. Studies on mycobacterial universal stress proteins appear to indicate that they are involved in energy-related biological functions, e.g. Rv2623 regulates bacillary growth by ATP binding 13 , and Rv1996 regulates NAD/NADH-related isoniazid susceptibility 14 . Rv1636, another mycobacterial universal stress protein, has also recently been shown to bind ATP and is predicted to function in energy (i.e., ATP)-dependent pathways 32 . Based on the results of this study, and our understanding of the literature, we suggest that mycobacterial universal stress proteins regulate different energy-related pathways and are thus potential drug targets for antibiotics selection.

Conclusions
We have shown that Rv2624c overexpression in mycobacteria increases their survival in the Wayne mycobacterial latency model and in human monocyte THP-1 cells. Transcriptome analysis suggests that Rv2624c affects histidine metabolism, and arginine and proline metabolism. Further studies using LC-MS/MS showed that Rv2624c modulated the abundance of arginine, a modulator of both mycobacteria and infected macrophages. In addition, biochemical characterization showed that Rv2624c is a nucleotide-binding universal stress protein and modulates metabolic pathways in an ATP-dependent manner, changing the abundance of arginine and thus increasing survival in THP-1 cells.   Construction of the rv2624c-overexpressing mycobacterial strain and E. coli expression strain. The full-length rv2624c gene was amplified from M. tuberculosis H37Rv genomic DNA using sense (Rv2624c-F) and antisense primers (Rv2624c-R) ( Table S1). The PCR reaction conditions were 98 °C for 30 s, followed by 30 cycles of 98 °C for 10 s, 60 °C for 30 s and 72 °C for 30 s, and then 72 °C for 10 min. The PCR-amplified fragments were purified, digested and cloned into E. coli expression vector pQE80L (Qiagen, Frankfurt, Hesse-Darmstadt, GER), named pQE-Rv2624c for expression of His 6 -Rv2624c, and mycobacterial vector pMV261, named pMV261-Rv2624c. pQE-Rv2624c was transformed into BL21 (DE3) for expression of in-frame 6xHis-tag recombinant Rv2624c and pMV261-Rv2624c was transformed into M. smegmatis wild-type strain mc 2 155 for overexpression of Rv2624c in mycobacteria.   In vitro growth kinetics. To determine the effect of Rv2624c overexpression and its mutations on growth, M. smegmatis strains harboring wild-type or mutant pMV261-Rv2624c (Rv2624cD17E and Rv2624cG109A) were grown in Middlebrook 7H9 medium containing 50 mg/L kanamycin. Log phase cultures (OD 600 of 0.8-1.0) of mycobacterial strains were diluted into 7H9 medium to OD 600 of 0.5. To generate a growth curve, cells were re-inoculated and OD 600 was measured at the indicated times. To compare the size of colonies of different mycobacterial strains, cells were 10x serially diluted (1:10), spotted (5 μ L) onto 7H10 medium supplemented with kanamycin (50 mg/L) and incubated at 37 °C. Photographs were taken after 3 days of incubation. The experiments were performed in triplicate.

Establishment of the
Macrophage killing assay. The macrophage killing assay was performed as previously described 33 .
Briefly, human monocyte THP-1 cells (ATCC TIB-202) were differentiated by treatment with 100 μ g/L phorbo-12-myristate-13-acetate (PMA; Sigma) for 24 h. Infection with various mycobacterial strains was carried out at a multiplicity of infection (MOI) of 10:1 for 4 h at 37 °C and 5% CO 2 . Unphagocytized bacilli were removed by extensively washing with PBS three times and the samples were incubated for 1 h with RPMI1640 containing 10 mg/L gentamycin to kill extracellular bacteria. The infected cells were lysed with PBS containing 0.05% Tween 20, and then diluted and plated on LB agar. The CFU number is indicated as the infected number of bacilli (T0). Next, infected THP-1 cells were incubated for 4 h, and then collected and diluted on LB agar. The corresponding post-infection CFUs were counted as the number of surviving bacilli (T1). The mc 2 155 strain with empty vector pMV261 was used as a negative control.
RNA isolation, RNA-sequencing and data analysis. Bacterial cultures (OD 600 of 2) of pMV261-Rv2624c/mc 2 155 and pMV261/mc 2 155 (negative control) were collected and total RNA was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. Construction and sequencing of the above two strains was performed by BGI-Shenzhen (Shenzhen, China). Briefly, rRNA was removed from extracted RNA from the various mycboacterial strains using a Ribominus Transcriptome Isolation Kit (Thermo Fisher Scientific, Waltham, MA, USA). NEXTflex RNA Fragmentation Buffer (Bioo Scientific, Austin, Texas, USA) was added to cleave mRNA into short fragments. Using these short fragments as templates, random hexamer primers were used to synthesize the first-strand cDNA. Second-strand cDNA was synthesized using buffer, dATPs, dGTPs, dCTPs, dUTPs, RNase H and DNA polymerase I. Short fragments were purified with a QIAQuick PCR extraction kit (Qiagen) and resolved by electrophoresis for end reparation and addition of poly(A). After ligating the synthetic short fragments to sequencing adapters, UNG enzyme was used to degrade second-strand cDNA, and the product was purified with a MiniElute PCR Purification Kit (Qiagen). The library was sequenced using an Illumina HiSeq2000. Clean reads were mapped to the reference genome and gene sequences using SOAP2. Mismatches of no more than five bases were allowed in the alignment. Gene coverage is the percentage of a gene covered by reads and is equal to the ratio of the number of bases in a gene covered by unique mapping reads to the number of total bases in that gene. The RPKM method (reads per kilobase per million reads) 34 was used to calculate gene expression. The algorithm for identifying differentially expressed genes between two samples was developed by BGI Shenzhen, and the false discovery rate (FDR) control was used for correct p values. In this study, differential expression was indicated when the false discovery rate was ≤ 0.001 and the ratio was larger than 2. Analysis of differentially expressed genes was further carried out using the KEGG Pathway analysis 24 .