Expression and production of soluble Mycobacterium tuberculosis H37Rv mycosin-3.

Mycobacteria encode five type VII secretion system (T7SS) or ESX for nutrient acquisition and virulence. Mycosins are membrane-anchored components of ESX with serine protease activity but an unidentified substrate range. Establishing the substrate specificity of individual mycosins will help to elucidate individual ESX functions. Mycosin-1 and -3 orthologues from two environmental mycobacterial species, Mycobacterium smegmatis and Mycobacterium thermoresistibile, have been heterologously produced, but mycosins from Mycobacterium tuberculosis (Mtb) remain to be studied. Here we describe the successful production of Mtb mycosin-3 as a first step in investigating its structure and function.

Introduction

Pathogenic Gram negative bacteria use a range of secretion systems to secrete virulence factors or transport the factors into host cells to manipulate the host immune system [1]. Type VII secretion systems (T7SSs) are restricted to mycobacteria and some other high GC Gram-positive bacteria [2], [3]. Mycobacterium tuberculosis (Mtb), the etiological agent of tuberculosis, has five T7SSs, denoted as ESX-1 to -5 presumably evolved by gene duplication [4]. ESX-1 and -5 are critical to virulence in pathogenic mycobacteria [5], and ESX-3 participates in mycobactin-mediated iron acquisition [6], [7]. ESX-5 was recently found to additionally function in nutrient acquisition [8]. ESX-1, 3 and 5 are correspondingly essential for Mtb growth in vitro [9], [10]. The roles of ESX-2 and -4 are not yet clear. The close association of ESXs with fundamental biological processes has resulted in much research interest in T7SS.

Details of T7SS secretion have not been fully elucidated including the highly conserved mycosin components. Analysing mycosins may therefore help to unravel their functions. Mycosin-5 was not co-isolated with the central, double membrane spanning complex consisting of EccB, EccC, EccD and EccE, indicating a weak association in vivo [11]. Mycosins share a conserved catalytic triad of aspartate, histidine and serine with subtilisin-like serine proteases [12]. Screening experiments, however, did not identify mycosin substrates [12]. Recently, mycosin-1 (MycP1) was found to cleave EspB twice upon secretion [13] to potentially facilitate its maturation for host target interaction. This is, however, unlikely to be the only mycosin substrate, as the gene espB is unique to ESX-1. ESX-1 substrate secretion is dependent on mycosin-1 but removing its enzymatic activity unexpectedly increases secretion [13]. Mycosin-1 may thus ensure Mtb persistence by balancing immune detection and virulence [13].

Mycosins have an N-terminal secretion signal followed by a potential “pro-peptide”, a catalytic domain, a proline-rich linker and a hydrophobic transmembrane region (Fig. 1). While removal of the “pro-peptide” was originally proposed to be required for enzymatic activation [12], this was found not to affect its protease activity [14], [15], [16]. In addition, crystal structures of mycosin-1 from M. smegmatis and M. thermoresistibile and mycosin-3 (MycP3) from M. smegmatis suggest that the “pro-peptide” wraps around the catalytic domain possibly to stabilize it. The “pro-peptide” has hence been renamed the “N-terminal extension region” [14], [15], [16]. The MycP1 orthologue of M. smegmatis inefficiently cleaves EspB in vitro possibly due to other ESX-1 components being absent [14]. However, mycosin-1 orthologues from M. smegmatis and M. thermorsistibile are unlikely to be involved in virulence in these saprophytic species despite an amino acid sequence identity of 70% with Mtb protein. Production of recombinant mycosin-1 or -3 from M. tuberculosis is problematic. Although the role of mycosin-3 remains enigmatic, it is essential to M. tuberculosis survival in vitro [9], [17] making it a potential anti-TB drug target [15], [18].

Fig. 1

The primary structure of M. tuberculosis mycosin-3 (MycP3). Single underline: signal peptide; double underline: N-terminal extension; dashed underline: proline-rich linker; wave underline: hydrophobic transmembrane region; white on black: catalytic triad, Asp95-His126-Ser342.

In this study, the gene mycP from M. tuberculosis H37Rv was cloned and expressed. Extensive effort was made to optimize the construct for soluble mycosin-3 production to increase yield and stability. This report may aid efforts to study mycosin-3 with respect to substrate screening, functional characterization, enzyme kinetics and crystal structure determination.

Materials and methods

Media, plasmids and bacteria strains

Lysogeny broth (LB) was used to culture all Escherichia coli strains including XL-1 blue (Promega), BL21 (DE3) pLysS (Promega), Arctic Express (Agilent Technologies), Origami II (Novagen), and Rosetta gami II (Novagen). E. coli expression vector pGEX-6P-1 (GE Health Science), pCOLD (Takara), and pET-28a (Novagen), were used to produce mycosin-3 fusion proteins respectively with an N-terminal glutathione S-transferase (GST)-, an N-terminal His6-, and a C-terminal His6-tag.

Cloning of MycP3 constructs

A range of truncated mycP constructs were generated by polymerase chain reaction (PCR) using specific primer pairs (Table 1), Phusion DNA polymerase (ThermoScientific), and either Mtb H37Rv gemomic DNA (gift from Rob Warren) or codon-optimized mycP gene (GeneArt) as template. PCR products were ligated into pJET2.1 cloning vector (ThermoScientific) using T4 DNA ligase (Promega). The recombinant pJET2.1 vector was propagated using E. coli XL-1 Blue strain and restriction digested to provide an insert for ligation into expression vectors. The recombinant host-specific expression vectors were electroporated into E. coli expression host cells.

Table 1

Primers used to generate the starting M. tuberculosis mycP construct (pET-28a construct was not codon-optimized), and eight codon-optimized mycP constructs (Constructs A to I), Expression hosts and vectors are as listed.

Construct Name	Encoded Amino Acid Sequence	Expression Host	Expression Vector	Primer Sequences and Their Restriction Sites
Construct A	Ile26-Asn430	E. coli BL21 (DE3) pLysS	pET-28a	forward: 5′-CCATGGCGATCGGGCCGCCGG-3′ (NcoI)
				reverse: 5′-CTCGAGGTTGCGCGGTGTGGTG-3′ (XhoI)
			pGEX-6P-1	N/A (restricted directly from the synthetic construct)
Construct B	Arg51-Asn430		pET-28a	forward: 5′-CCATGGAACGCGGTGCGTGCAG-3′ (NcoI)
				Construct A reverse primer
			pGEX-6P-1	forward: 5′-GGATCCCGCGGTGCGTGCAG-3′ (BamHI)
				Construct A reverse primer
Construct C	Gly52-Leu401		pGEX-6P-1	forward: 5′-GGATCCGGTGCATGTAGCG-3′ (BamHI)
				reverse: 5′-CTCGAGTCACAGCTGCCAGGTC-3′ (XhoI)
Construct D	Ser57-Leu401		pGEX-6P-1	forward: 5′-GGATCCGGTGTTATTCCGG-3′ (BamHI)
				Construct C reverse primer
Construct E	Gly62-Leu401		pGEX-6P-1	forward: 5′-GGATCCGGTACAGATCCGG-3′ (BamHI)
				Construct C reverse primer
Construct F	Val67-Leu401		pGEX-6P-1	forward: 5′-GGATCCGTTCCGACCCCGAG-3′ (BamHI)
				Construct C reverse primer
Construct G	Ser71-Leu401	E. coli Arctic Express and BL21 (DE3) pLysS	pGEX-6P-1	forward: 5′-GGATCCCAGACCATGCTG-3′ (BamHI)
				Construct C reverse primer
			pCOLD	forward: 5′-CATATGCAGACCATGCTGAATC-3′ (NdeI)
				Construct C reverse primer
Construct H	Leu77-Leu401	E. coli BL21 (DE3) pLysS	pGEX-6P-1	forward: 5′-GGATCCCTGCCAGCAGCATG-3′ (BamHI)
				Construct C reverse primer
Construct I	Ile24-Leu401	E. coli Origami II and Rosetta gami II	pGEX-6P-1	forward: 5′-GGATCCATTGGTCCGCCTGTTG-3′ (BamHI)
				Construct C reverse primer

*Underlined sequences are restriction sites as indicated in brackets.

Production and purification of MycP3 protein

An overnight 50 mL starter culture was prepared from a single E. coli transformant colony. An aliquot of the starter culture was transferred to 1 L LB medium, allowed to grow to mid-log phase (OD600 nm=0.6 to 0.8) at 37 °C and induced with 0.1–0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) at 16, 25 and 30 °C for 18 h for test expression. Arctic express and pCOLD transformants were induced with 0.1 mM IPTG at 13 °C and expression continued for 40 h.

For the GST fusion protein purification, cultures were harvested by centrifugation at 3000g at 4 °C for 10 min. The pellet was resuspended in phosphate buffered saline (PBS) and sonicated using a probe sonicator (MSE) at an amplitude of 20 µm for 5 cycles of 30 s with 30 s incubation on ice. The soluble and insoluble fractions of the cell lysate were separated by centrifugation at 14,000g at 4 °C for 45 min. The supernatant was agitated with 2 mL PBS-equilibrated glutathione agarose beads (ABT) at 4 °C on a roller mixer for 1 h for fusion protein coupling. The mixture was poured into a drip column and the flowthrough eluted under gravity. The column was washed with 20 column volumes (CV) of PBS. The fusion protein was eluted with elution buffer: 25 mM Tris–HCl, pH 7.4, 150 NaCl, 15 mM reduced glutathione.

For the purification of the His6-tagged fusion proteins, the E. coli cell pellet was resuspended in lysis buffer (25 mM Tris–HCl, pH 7.4, 150 mM NaCl, 10 mM imidazole) and lysed by a tissue and cell homogenizer FastPrep-24 (MP Biomedical) at a speed of 6 m/s for 5 cycles of 30 s alternating with 30 s incubation on ice. The soluble fraction was separated as described above and the target protein was coupled to Ni-NTA beads (Qiagen). The beads were loaded into a drip column and washed with 20 CV wash buffer (25 mM Tris–HCl, pH 7.4, 150 NaCl, 20 mM imidazole) and eluted with elution buffer: wash buffer with 250 mM imidazole.

Results

Identifying the catalytic domain of MycP3 is critical prior to cloning and production experiments. MycP3 has a signal peptide, Met1-Ala25, and an N-terminal extension, Ile26-Gln50, N-terminal of the presumed catalytic domain, Arg51-Leu401, followed by a linker, Pro402-Asn430, and a transmembrane region, Val431-Glu461 (Fig. 1). Two truncated constructs of mycP were amplified from Mtb H37Rv genomic DNA and cloned into pET-28a E. coli expression vector to produce C-terminally His6-tagged fusion protein: Construct A encoding Ile26-Asn430, lacks the signal peptide and the transmembrane region. Construct B encodes Arg51-Asn430, which additionally lacks the “pro-peptide” or “N-terminal extension”. Neither construct produced sufficient fusion protein in E. coli BL21 (DE3) for visualization on SDS-PAGE.

E. coli codon usage differs significantly from that of mycobacteria [19]. Mtb mycP DNA encoding Ile26-Asn430 was correspondingly codon-harmonized for E. coli expression. Corresponding constructs A and B were cloned into plasmid pGEX-6P-1. The encoded GST MycP3 fusion proteins produced small quantities of insoluble protein (Figs. S1 and S2 for construct A). Small amounts of soluble fusion protein proved unstable, prone to rapid proteolysis and precipitation. Chaperone proteins, such as DnaK (69 kDa) and heat shock chaperonin (60 kDa) were co-produced and identified by mass spectrometry (results not shown, service provided by Central Analytical Facility, Stellenbosch University). Triton X-100 treatment removed the chaperones. However, MycP3 was thereupon rapidly degraded (results not shown). Lowering the expression temperature to 16, 25 and 30 °C and varying IPTG concentrations from 0.1 to 1 mM did not improve the solubility or stability of the fusion protein (Fig. S2).

As inherently disordered region (IDR) of the MycP3 fusion protein may limit its solubility and stability, the protein was analysed using PrDOS [20]. This identified Met1-Pro37 and Pro405-E461 as potentially disordered (Fig. 2). The region Pro402 to Glu461 was excluded from analysis as it encompasses the proline-rich linker and the C-terminal transmembrane α-helix [12]. A set of MycP3 constructs removing residues Ile26 to Gln50 in steps of five to six amino acids was generated: Construct C encodes Gly52-Leu401, D: Ser57-Leu401, E: Gly62-Leu401, F: Val67-Leu401, G: Ser71-Leu401, and H: Leu77-Leu401. The six constructs were cloned into pGEX-6P-1 for GST fusion protein production. Protein production levels for constructs C and D were low while those for E to H were higher (Fig. S3). However, the GST–MycP3 fusion proteins were invariably insoluble (Fig. S4). Gene expression at 16, 25 and 30 °C similarly did not improve the solubility of the resulting protein. Construct G was selected for further optimization of MycP3 production as the elimination of a proline/valine-rich portion (Gly52-Pro70) could improve the yield.

Fig. 2

Disordered region probability plot for Mtb H37Rv MycP3 with a prediction false positive rate of 5%, where the disordered region prediction software DrDOS identified two disordered regions, Met1-Pro37 and Pro405-E461.

MycP3 contains four cysteines, Cys54, Cys123, Cys205 and Cys249. As a secreted protein, MycP3 stability could be disulfide bond dependent. Construct I, encoding MycP3 Ile26-Leu401 was produced in E. coli expression strains, Origami II and Rosetta Gami II with oxidizing cytosol. However, the resulting protein was again insoluble (results not shown).

The vector pCOLD and the strain Arctic Express are designed to increase the solubility of produced protein [21]. The co-production of pCOLD encoded cold-shock chaperones alongside Construct G did, again, not increase the protein solubility in both BL21 (DE3) and Arctic Express cells (Fig. S5). Unexpectedly, though, expression of Construct G in pGEX-6P-1 vector in Arctic Express produced significant amounts of soluble GST MycP3 Ser71-Leu401 fusion protein (Fig. 3) that could be cleaved by PreScission Protease (Roche) to release MycP3 Ser71-Leu401 (Fig. S6).

Fig. 3

Test expression and solubility test for Construct G expressed in E. coli Arctic Express strain. Protein production level are high in both soluble and insoluble fractions. M: Protein Marker; E: Empty (bands present overflow from other lanes); P: Pellet; S: Supernatant.

Discussion

Production of Mtb proteins in E. coli has repeatedly been found to be challenging possibly due to its GC rich (>65%) genome and a distinct codon usage. Mtb proteins further incorporate more glycines, alanines, prolines and arginines than E. coli and the organism also has post-translational modification machinery that E. coli lacks [22]. Although saprophytic M. smegmatis is occasionally suitable for Mtb protein production, only a handful of successful cases of protein production have been reported [23].

In this study, a range of mycP constructs were generated with variable truncations and tags. Truncations can eliminate inherently disordered regions whereas a soluble tag may increase the solubility of an attached cargo. Correspondingly, GST–MycP33 fusion proteins proved more soluble than His6-tagged counterparts. Lower IPTG concentrations and production temperatures limit heat shock protein production, proteolytic degradation, protein aggregation and improve protein stability [24]. MycP3 fusion protein, however, remained insoluble except when produced in Arctic Express. Although rarely used for Mtb H37Rv protein production [23], Arctic Express successfully produced GST-tagged MycP3 fusion proteins in significant quantity at low temperature despite lacking an N-terminal extension that stabilizes the catalytic domain [14], [15], [16].

The crystal structures of three mycosin orthologues from environmental mycobacterial species with high sequence identity to Mtb proteins have been solved. This includes MycP1 from M. smegmatis and M. thermoresistibile, as well as MycP3 from M. smegmatis [14], [15], [16]. An amino acid sequence alignment [25] of MycP1 and MycP3 from Mtb and M. smegmatis, as well as MycP1 from M. thermoresistibile reveals that MycP1 from M. smegmatis and M. thermoresistibile are closely related (alignment score: 78.8) while Mtb MycP1 is significantly different (alignment score 72.0) (Table 2). MycP3 from Mtb and M. smegmatis are even more distantly related. A model for Mtb MycP3 was generated using MycP3 from M. smegmatis and Swiss Model [26]. The two structures demonstrate no significant differences (Fig. S7). A QMEANZ-score of −1.45 [26], however, indicates some uncertainty in the model structure. The low sequence alignment and QMEANZ-score may indicate a difference in substrate specificity especially as ESX-1 is virulence associated in Mtb while it is essential for DNA transfer in M. smegmatis [27], [28]. Investigating the function of Mtb mycosins thus remain important.

Table 2

Sequence identity between mycosin orthologues from Mtb, M. smegmatis, and M. thermoresistibile

SeqA Name	Length	SeqB Name	Length	Alignment Score
Msmeg_0083|MycP1	449	M.therm KEK_05522|MycP1	443	78.8
Mtb Rv3883c|MycP1	446	Msmeg_0083|MycP1	449	72.0
Mtb Rv3883c|MycP1	446	M. therm KEK_05522|MycP1	443	71.1
Mtb Rv0291|MycP3	461	Msmeg_0624|MycP3	459	59.0
Msmeg_0624|MycP3	459	M. therm KEK_05522|MycP1	443	44.0
Mtb Rv3883c|MycP1	446	Msmeg_0624|MycP3	459	44.0
Mtb Rv0291|MycP3	461	M. therm KEK_05522|MycP1	443	43.8
Msmeg_0624|MycP3	459	Msmeg_0083|MycP1	449	43.2
Mtb Rv0291|MycP3	461	Msmeg_0083|MycP1	449	42.8
Mtb Rv0291|MycP3	461	Mtb Rv3883c|MycP1	446	42.4

To distinguish the enzyme specificity of different mycosins is critical to understand the function of different ESXs. In conclusion, this study reports the successful attempts at producing soluble MycP3 from Mtb H37Rv. It is hoped that this information may facilitate future structural and functional studies. Mycosins are evolutionarily only distantly related to other substilisins implying that they could be attractive drug targets especially as MycP1 regulates secretion and processes the secreted virulence factor from ESX-1 whereas MycP3 appears to be involved in iron or possibly even heme acquisition [29].

Author contributions

WS and NG conceived and supervised the study; ZF, WS and NG designed experiments; ZF performed experiments; ZF and WS analysed data; ZF, WS and NG wrote the manuscript.