A haloarchaeal VapBC toxin-antitoxin system: regulatory mechanism and role in establishing population heterogeneity to facilitate survival of subpopulations under stress conditions

ABSTRACT

Toxin-antitoxin (TA) systems are ubiquitously distributed in bacteria and archaea and have been implicated in various functions such as plasmid maintenance, phage defense, biofilm formation, stress response, and persistence, but their regulatory mechanisms and roles in haloarchaea are largely unknown. In this study, we found that four of the eight putative VapBC TA systems of Natrinema gari J7-2 are functional in strain J7-2 and Haloferax volcanii. Among them, the NgVapBC1 system is composed of the antitoxin NgVapB1 and the toxin NgVapC1, and the two proteins could form a NgVapBC1 complex. The NgVapBC1 system could be transcriptionally autoregulated, wherein a pseudo-palindromic sequence in the vapBC1 operon promoter acts as a negative cis-regulatory element, and the NgVapBC1 complex is a stronger repressor than NgVapB1. The qRT-PCR and in vivo toxicity analyses showed that strain J7-2 maintains a high vapB1/vapC1 mRNA ratio through truncation of the vapBC1 operon transcript within the toxin-coding region for preventing the synthesis of excess toxin NgVapC1 to release toxicity abnormally. Mutational analyses showed that differential start codon usage by vapB1 and vapC1 and a −4 nt overlap of their stop and start codons contribute to maintaining a higher vapB1/vapC1 mRNA ratio, and the −4 nt overlap-mediated translational coupling of vapB1 and vapC1 would enable both post-transcriptional and translational regulation of the NgVapBC1 system. The phenotypic comparison of strain J7-2 and its ΔvapBC1 mutant revealed that the NgVapBC1 system contributes to population heterogeneity of decline-phase rather than exponential-phase strain J7-2 and facilitates the survival of subpopulations under stress conditions.

IMPORTANCE

Although TA systems are regarded as versatile modulators of prokaryotic cell fate, their bona fide physiological roles remain elusive. Moreover, so far, there is no report on the regulatory mechanism and function of haloarchaeal VapBCs, which constitute the major group of TA systems in haloarchaea. Our results demonstrate that a VapBC system of the haloarchaeon Natrinema gari J7-2 is regulated at multiple levels and mediates the establishment of population heterogeneity, thereby facilitating the survival of the strain J7-2 population as a whole in diverse and changing environments. This study provides the first experimental evidence of the regulatory mechanism of the biological role of the VapBC system in conferring fitness advantages to haloarchaea.

INTRODUCTION

Toxin-antitoxin (TA) systems are ubiquitous and abundant in prokaryotes and consist of a toxin that inhibits an essential cellular process and an antitoxin that neutralizes its cognate toxin. The genes encoding antitoxin and toxin are usually organized as a bicistronic operon and commonly located in mobile genetic elements (MGEs), including plasmids, prophages, integrative and conjugative elements, or transposons, or constitute small genomic islets by themselves in chromosomes, often in multiple copies (1). TA systems are currently classified into eight types (I–VIII) based on the nature and mode of action of the antitoxin, while type II systems, of which both toxin and antitoxin are proteins, are probably the most prevalent (1, 2). Type II TA systems are generally autoregulated at the transcriptional level through a mechanism of conditional cooperativity: at low toxin/antitoxin ratios, the TA complex cooperatively binds to the promoter region to repress TA operon transcription, whereas at high toxin/antitoxin ratios, the excess toxin destabilizes the TA-operator complex to induce transcription (3, 4). Type II TA systems could also be regulated at the post-transcriptional level by limited cleavage of toxin mRNA (5, 6). In terms of post-translational regulation of type II systems, a long-lasting paradigm suggests that stressful conditions would stimulate preferential degradation of the antitoxin in the TA complex and promote the liberation of the toxin; however, recent studies challenge this view and suggest that one likely mechanism for toxin activation is conformational changes in the TA complex due to the binding of a third protein (1, 2, 7).

The virulence-associated proteins B and C (VapBC) system represents the largest subgroup of the type II TA systems (2, 8). VapC toxins belong to the PilT N-terminal (PIN) domain family with endoribonuclease activity (9), and several of them have been shown to cleave unique tRNA, rRNA, and mRNA species (10–12). VapB antitoxins consist of an N-terminal DNA-binding domain that autoregulates the TA operon and a C-terminal region that obstructs the VapC active site in the VapBC complex to neutralize the toxin (9, 13, 14). Diverse functions have been assigned to TA systems including VapBCs, such as plasmid maintenance, integrative MGE stabilization, phage defense, biofilm formation, stress response, and persistence, although the roles of TA systems in stress response and persistence remain controversial (2, 7, 15, 16). The elucidation of the biological functions of TA systems is still an ongoing challenge.

For archaea, VapBCs have been intensively studied in thermoacidophilic archaea and implicated in stress responses. For instance, Saccharolobus solfataricus (formerly Sulfolobus solfataricus) possesses 22 VapBCs, and the inactivation of the vapBC6 operon by targeted gene disruption leads to heat-shock sensitivity and a heat-dependent reduction in growth rate, suggesting that VapBC6 could regulate thermophilicity of the archaeon (11, 17). Based on the evidence that three of the 12 VapCs of Metallosphaera prunae could degrade rRNA in vitro, it has been proposed that this thermoacidophilic archaeon employs VapBCs for post-transcriptional regulation under uranium stress (18). In Sulfolobus acidocaldarius, which has 18 VapBCs, gene deletion analyses show that VapBC14 is involved in regulating biofilm formation (19), while VapBC4 not only participates in regulating biofilm formation but also plays an important role in heat stress adaptation (12). Given that thermoacidophilic archaea lack versions of bacterial heat shock proteins (HSPs), including Hsp100, Hsp90, and the DnaK/DnaJ/GrpE system (20), it has been suggested that other strategies, such as TA systems, are involved in handling thermal stress (16). In addition, VapCs from hyperthermophilic archaea, such as Pyrobaculum aerophilum (21, 22), Archaeoglobus fulgidus (23), and Pyrococcus horikoshii (24), represent the earliest reported PIN domain proteins that have been structurally characterized by X-ray crystallography, which have provided valuable information essential for understanding the structure-function relationship of VapBCs (9).

Much less is known about the role of TA systems in haloarchaea. The MazF toxins of the MazEF type II TA system from Haloquadra walsbyi (25) and Methanohalobium evestigatum (26) act as sequence-specific mRNA interferases and could exert toxicity when expressed in Escherichia coli, but their physiological roles in haloarchaea remain unclear. The CreTA type VIII TA system functions as an addiction module that prevents the loss of its cognate CRISPR-Cas system in Haloarcula hispanica (27). The temperate haloarchaeal virus SNJ2 encodes a functional HicAB type II TA system, of which the HicA toxin is toxic to the host cell in the absence of the HicB antitoxin, but this TA system does not participate in the SNJ2 life cycle transition (28). To date, there is no report on the function of the VapBC system in haloarchaea, although VapBCs constitute the major group of haloarchaeal TA systems (29).

Haloarchaea thrive in hypersaline environments (e.g., solar salterns, salt lakes, and salt deposits) and are usually exposed to environmental stresses associated with changes in salinity, temperature, radiation, nutrient availability, etc. Notably, both haloarchaea and thermoacidophilic archaea lack Hsp100 and Hsp90, but the former possess the DnaK/DnaJ/GrpE system that is absent in the latter (20). It is interesting to investigate whether exerting the function of the VapBC system could serve as an alternative strategy for handling stresses in haloarchaea, as proposed for thermoacidophilic archaea (16).

The haloarchaeon Natrinema gari J7-2 was isolated from a salt mine in China (30, 31). Nnm. gari strains have also been isolated from Thai traditional salty fish sauce and used for degrading histamine in high salt-fermented fishery products (32). We have previously characterized three extracellular subtilases (SptA (33, 34), SptC (35), and SptE (36)) and three HtrA proteases (NgHtrA, NgHtrB, and NgHtrC) (37) from strain J7-2 and found that these proteases are important for the growth, survival, and the stress response of the haloarchaeon. For instance, SptA not only contributes to the growth-phase transition of strain J7-2 but also facilitates the viable cells to scavenge the dead cell-derived proteins under nutrient-limiting conditions (38–40). As an extracytoplasmic protein control system, the HtrA proteases are essential for the survival of strain J7-2 in response to stresses such as heat, high salinity, and toxic substances (37). Additionally, thermal and osmotic stresses promote the expression of DnaK in Natrinema sp. J7, implying the role of this chaperone in the stress response of the haloarchaeon (41). By analyzing the complete genome sequence of strain J7-2 (30), eight putative vapBC loci were identified on the chromosome of this haloarchaeon. Recently, one of the predicted antitoxins, namely NJ7G_2729, was found to potentially participate in cell division (42). It is unclear whether and how the eight predicted VapBCs actually function as TA systems and also play a role in the survival and stress response of strain J7-2. In this study, four of the eight putative VapBC systems of Nnm. gari J7-2 are experimentally confirmed to be functional in strain J7-2 and Hfx. volcanii, and we focused on NgVapBC1 to investigate its regulatory mechanism and physiological role. NgVapBC1 was found to be regulated at multiple levels and contribute to the establishment of population heterogeneity, which confers fitness advantages to strain J7-2.

RESULTS

Bioinformatics analysis identified eight putative VapBCs in Nnm. gari J7-2

Eight VapBC-encoding loci (Table 1) were identified on the chromosome of strain J7-2 by the online tool TAfinder (43). According to the prediction by IslandViewer (44), the locus encoding NJ7G_2321/4368 resides on a ~9.3 kb MGE, while those encoding NJ7G_3854/3855 and NJ7G_3856/3857 reside on another ~36.9 kb MGE; they show high sequence identity (80%–95%) to homologous TA pair loci on plasmids of other haloarchaea (Table 1). The loci encoding NJ7G_0345/0346, NJ7G_0582/0581, and NJ7G_3796/3795 also share high sequence identity (>70%) with plasmid-borne TA pair loci of other haloarchaea (Table 1), and the former two have a lower GC content than the average of the chromosome (Fig. S1), implying that the three vapBC loci are also located on MGEs. It appears that at least five of the eight vapBC loci may be propagated via horizontal gene transfer.

TABLE 1.

Putative VapBCs in strain J7-2 and their homologs in haloarchaeal MGEs

	Putative VapBCs in strain J7-2			Homologous TA pairs in haloarchaeal MGEsa
	Antitoxin	Toxin	Feature of gene locus region	Identityb (%)	MGE (GenBank no.)	Strain
1	NJ7G_0345	NJ7G_0346	Lower GC content than the average of the chromosome	79	Plasmid pNMAG02 (CP001934)	Natrialba magadii ATCC 43099
2	NJ7G_0582	NJ7G_0581	Lower GC content than the average of the chromosome	71	Plasmid pHFX3 (CP172598)	Haloferax sp. S1W
3	NJ7G_2321	NJ7G_4368	A MGE (~9.3 kb) with six transposase/integrase genes	95	Plasmid unnamed1 (CP101874)	Natrinema thermotolerans A29
4	NJ7G_2439	NJ7G_2438
5	NJ7G_2729	NJ7G_2730	NJ7G_2729 potentially participates in cell division (42)
6	NJ7G_3796	NJ7G_3795c		74	Plasmid pHALXA01 (CP002840)	Halopiger xanaduensis SH-6
7	NJ7G_3854c	NJ7G_3855	A MGE (~36.9 kb) with one transposase/integrase gene	80	Plasmid pHsi55 (CP073370)	Haloarcula marismortui ATCC 33800
8	NJ7G_3856	NJ7G_3857	Co-locates with NJ7G_3854/3855 in the same MGE	92	Plasmid pNPA200 (CP040638)	Natrinema pallidum BOL6-1

^a The information of haloarchaeal MGEs with TA pairs that show highest sequence identities to corresponding TA pairs of strain J7-2 is shown.

^b Nucleotide sequence identity between the TA pair-encoding region of strain J7-2 and its homolog in haloarchaeal MGE.

^c A 1-bp deletion in the gene causes an N-terminal truncation of NJ7G_3795, and a 1-bp insertion in the gene leads to an N-terminal truncation of NJ7G_3854.

Among the eight putative VapCs, NJ7G_3795 is an N-terminal truncated form of VapC, while the other seven all possess three or four conserved acidic residues in the active site of VapC toxins (Fig. 1A). Although these putative VapCs show very low sequence identity (~10%–19%) to known homologs such as PAE0151 from P. aerophilum (22) and MtVapC26 from Mycobacterium tuberculosis (45) (Fig. 1A), the overall structures of all of them except NJ7G_3795, predicted by AlphaFold 3 (46), are well superimposed with the crystal structure of PAE0151 (22) (Fig. 1B).

Fig 1.

Primary and tertiary structures of putative VapCs of strain J7-2. (A) Amino acid sequence alignment of the eight VapCs of strain J7-2, MtVapC26 (WP_070895652, 5X3T), and PAE0151 (Q8ZZP3). The asterisks indicate four conserved acidic residues in the active site of VapC toxins. (B) Superimpositions of predicted structure models of VapCs of strain J7-2 with the crystal structure of PAE0151 (2FE1). The conserved acidic residues in the active site are indicated.

According to their similarities to known homologs in primary and tertiary structures, the putative VapBs of strain J7-2 can be classified into three groups, which contain a ribbon-helix-helix (RHH) domain, a SpoVT/AbrB domain, and a photosynthetic reaction center (PRC) barrel domain, respectively (Fig. 2). For the members of the first group, despite sharing only ~10%–17% sequence identity with MtVapB26 of M. tuberculosis (Fig. 2A), NJ7G_0345, NJ7G_2321, NJ7G_2439, and NJ7G_3856 all have an N-terminal domain that is well superimposed with the DNA-binding RHH domain of MtVapB26 (45) (Fig. 2B). NJ7G_3854 is an N-terminal truncation of RHH domain-containing VapB, with a size only about half those of its homologs (Fig. 2A). In the second group, NJ7G_0582 and NJ7G_3796 also show very low sequence identity (~15%–17%) to BsVapB of Bacillus subtilis, but their N-terminal domains are well superimposed with the DNA-binding SpoVT/AbrB domain of the latter (47) (Fig. 2B). For NJ7G_2729, it shares 41% sequence identity with the C-terminal YlmC domain (AfCdpB2-C) of the cell division protein B2 (CdpB2) of A. fulgidus (Fig. 2A) and adopts a PRC barrel structure highly similar to that of the latter (48) (Fig. 2B).

Fig 2.

Primary and tertiary structures of putative VapBs of strain J7-2. (A) Amino acid sequence alignment of the VapBs of strain J7-2 with MtVapB26 (WP_003910158), BsVapB (WP_080332445), and AfCdpB2-C (WP_010877866), respectively. (B) Superimpositions of predicted structure models of VapBs of strain J7-2 with the structure models of MtVapB26 (5X3T), BsVapB (2K1N), and AfCdpB2-C (8QZO), respectively.

Heterologous expression and gene deletion analyses revealed four functional VapBCs

To investigate whether the putative VapBCs are functional or not, the eight putative VapC toxins were heterologously expressed in Hfx. volcanii, driven by a tryptophan-inducible promoter P_tnaA (49). Similar to Hfx. volcanii containing the blank vector pSY1 (50), the strain expressing NJ7G_2730, NJ7G_3795, NJ7G_3855, or NJ7G_3857 grew well on the Hv-Cab (51) agar plate with tryptophan (Fig. 3A). By contrast, the strains harboring the expression plasmids for NJ7G_0346, NJ7G_0581, NJ7G_2438, and NJ7G_4368, respectively, displayed severe growth defects under the same conditions (Fig. 3A), suggesting that these four VapCs are toxic to Hfx. volcanii cells. When grown in Hv-Cab liquid medium, the strains expressing the four toxic VapCs also showed growth defects, as evidenced by their lower cell densities (Fig. 3B) and significantly lower viable cell numbers (Fig. 3C) compared with the control strain containing pSY1. Nevertheless, the strain expressing NJ7G_4368 displayed a better growth than those expressing NJ7G_0346, NJ7G_0581, or NJ7G_2438 (Fig. 3B), suggesting that the former exhibits a weaker toxicity than the latter three in Hfx. volcanii.

Fig 3.

Functional analyses of VapBCs of strain J7-2 in Hfx. volcanii. (A–C) Effects of VapCs of strain J7-2 on the growth of Hfx. volcanii. Hfx. volcanii strain carrying the blank vector (pSY1) or the expression plasmid for target protein was streaked on Hv-Cab agar plate without (−) or with (+) 1 mM tryptophan (Trp) and grown at 37°C for 4 days (A). Target strains were cultivated at 37°C in 18% Hv-Cab liquid medium supplemented with 1 mM tryptophan and novobiocin (0.3 μg/mL). At the indicated time points, samples were taken and subjected to OD₆₀₀ measurement (B) and viable cell counting on Hv-Cab agar plate at 37°C (C). The data are expressed as means ± SDs from biological triplicates (***, P < 0.001; **, P < 0.01). (D) Neutralization of VapCs by VapBs in Hfx. volcanii. Hfx. volcanii strain carrying the expression plasmid for antitoxin alone, toxin alone, or both antitoxin and toxin was streaked on novobiocin (0.3 μg/mL)-containing 18% Hv-Cab agar plates without (−) or with (+) 1 mM tryptophan and grown at 37°C for 4 days. Schematic representation of promoter (P_fsptA or P_tnaA)-preceded recombinant genes on expression plasmids is shown.

The functional relationship between toxin and its cognate antitoxin was investigated by co-expressing them in Hfx. volcanii, where the expression of VapC and VapB were driven by an inductive P_tnaA and a constitutive promoter P_fsptA (50), respectively (Fig. 3D). In contrast to the growth-defective strain harboring the expression plasmid only for the VapC toxin, the strain carrying the co-expression plasmid for NJ7G_0345/NJ7G_0346, NJ7G_0582/NJ7G_0581, NJ7G_2439/NJ7G_2438, or NJ7G_2321/NJ7G_4368 was able to grow on the Hv-Cab agar plate with tryptophan (Fig. 3D). These results suggest that NJ7G_0345, NJ7G_0582, NJ7G_2439, and NJ7G_2321 could act as VapB antitoxins to neutralize their cognate VapC toxins.

We next performed gene deletion analysis of the VapBCs in strain J7-2 using the endogenous CRISPR-Cas system-based genome-editing method (52). The single gene-deletion mutants for the antitoxins NJ7G_2729, NJ7G_3796, NJ7G_3854, and NJ7G_3856, whose cognate VapCs (NJ7G_2730, NJ7G_3795, NJ7G_3855, and NJ7G_3857) did not show toxicity to Hfx. volcanii (Fig. 3A), could be obtained (Fig. S2). However, we were unable to obtain single gene-deletion mutants for the antitoxins NJ7G_0345, NJ7G_0582, NJ7G_2439, and NJ7G_2321 (data not shown), whose cognate VapCs (NJ7G_0346, NJ7G_0581, NJ7G_2438, and NJ7G_4368) were toxic to Hfx. volcanii (Fig. 3A), implying that these four VapBs are essential for the survival of strain J7-2 by neutralizing their cognate VapC toxins. In support of this possibility, these four essential VapB genes and their cognate VapC genes could be deleted simultaneously to generate double gene-deletion mutants (Fig. S2).

The above-mentioned results demonstrate that NJ7G_0345/NJ7G_0346, NJ7G_0582/NJ7G_0581, NJ7G_2439/NJ7G_2438, and NJ7G_2321/NJ7G_4368 are functional in both Hfx. volcanii and strain J7-2. Based on the evidence that three of the four functional VapBCs of strain J7-2 possess an antitoxin with a RHH domain (NJ7G_0345, NJ7G_2439, and NJ7G_2321; Fig. 2B), we chose NJ7G_0345/NJ7G_0346 as a representative for further study. Hereafter, NJ7G_0345 and NJ7G_0346 were named NgVapB1 and NgVapC1, respectively, and the TA system constituted by them was named NgVapBC1.

NgVapB1 neutralizes NgVapC1 by forming a complex

The double gene-deletion mutant ΔvapBC1, in which the coding regions for NgVapB1 and NgVapC1 were deleted (Fig. 4A and S2), was used as the host to investigate the interaction between NgVapB1 and NgVapC1. By using pSHS (52) as the vector, we constructed the expression plasmid for either or both NgVapB1 and NgVapC1, driven by a constitutive promoter (P₀₅₆₆) of a high-abundance blue copper domain protein (NJ7G_0566) of strain J7-2 (53) and inductive P_tnaA, respectively (Fig. 4B). Similar to that in Hfx. volcanii (Fig. 3D), the expression of NgVapC1 alone in ΔvapBC1 led to a severe growth defect, while the expression of NgVapB1 or co-expression of NgVapB1 and NgVapC1 did not affect the growth of the mutant strain (Fig. 4B), suggesting that NgVapB1 could neutralize the toxicity of NgVapC1 in vivo.

Fig 4.

The interaction between NgVapB1 and NgVapC1. (A) Schematic view of vapBC1 operon and adjacent genes in strain J7-2 and the deletion region in the mutant ΔvapBC1. (B) In vivo interaction between NgVapC1 and NgVapB1. ΔvapBC1 strain carrying the blank vector (pSHS) or the expression plasmid for NgVapB1, NgVapC1, or both NgVapB1 and NgVapC1 was streaked on mevinolin (5 μg/mL)-containing 18% Hv-Cab agar plate without (−) or with (+) 1 mM tryptophan (Trp) and grown at 37°C for 4 days. Schematic representation of promoter (P₀₅₆₆ or P_tnaA)-preceded recombinant genes on expression plasmids is shown. (C) In vitro pull-down assay for the interaction between NgVapB1 and NgVapC1. The cellular proteins of E. coli harboring the expression plasmid for NgVapB1 or His-tagged NgVapC1H before (lanes 1 and 4) and after (lanes 2 and 5) IPTG induction, the inclusion bodies (lane 6) and refolded sample (lane 7) of NgVapC1H, as well as the eluted fraction from the nickel-affinity column for separation of the sample containing NgVapB1 (lane 3), NgVapC1H (lane 8), or both NgVapB1 and NgVapC1H (lane 9) were subjected to SDS-PAGE. (D) Predicted structural model of the NgVapBC1 complex. The four acidic residues in the active site (Asp5, Glu43, Asp112, and Asp133) of NgVapC1 and the residue Arg98 of NgVapB1 are shown. Dotted lines represent the ion pairs formed between the side chain of Arg98 and those of Asp5 and Asp112.

We further used an in vitro pull-down assay to confirm the interaction between NgVapB1 and NgVapC1. After expression in E. coli, recombinant NgVapC1 with a C-terminal His-tag (NgVapC1H) existed in inclusion bodies, which could be converted into soluble form after refolding and then purified by Ni-NTA affinity chromatography (Fig. 4C, lane 8). Untagged NgVapB1 was produced in E. coli as soluble form, which alone failed to bind the Ni-NTA resin (Fig. 4C, lane 3) but could be co-eluted with NgVapC1H from the column with a molar ratio of approximately 1:1 by band intensity quantification (Fig. 4C, lane 9), suggesting that the two proteins can interact with each other to form a stable NgVapBC1 complex. The structural prediction of this complex by AlphaFold 3 reveals that the three C-terminal α-helices of NgVapB1 bind to NgVapC1 and the side chain of Arg98 from NgVapB1 electrostatically interacts with those of Asp5 and Asp112 in the active site of NgVapC1 to inhibit the activity of the toxin (Fig. 4D), which is a common interaction mode conserved in VapBC complexes (9).

A pseudo-palindromic sequence (PPS) in the vapBC1 operon promoter functions as a negative cis-regulatory element for vapBC1 expression

It has been reported that VapB antitoxins and VapBC complexes could bind to pseudo-palindromic, inverted repeats in their own operon promoters and negatively autoregulate their own transcription (9, 13). In addition to the highly conserved cis-acting elements (TATA box and WW motif) of haloarchaeal promoters (54), the promoter (P₀₃₄₅) region of vapBC1 operon of strain J7-2 also contains a PPS (GTGCTAC-N4-GTATTAC; PPS1) comprising a pair of inverted repeats (IR1f/IR1r), which flank the WW motif and are well conserved in promoter regions of vapBC1 operon homologs from 31 haloarchaea (Fig. 5A and S3). We used electrophoretic mobility shift assays (EMSAs) to investigate whether PPS1 can bind NgVapB1 or the NgVapBC1 complex in vitro. Recombinant NgVapB1 with a C-terminal His-tag (NgVapB1H) was produced in E. coli in a soluble form and purified (Fig. S4A). The DNA fragment covering the region of P₀₃₄₅ (117 bp) and its mutated fragments, including P_0345M1r, P_0345M1f, and P_0345M1fr with mutation of either or both IR1f and IR1r and P_0345M1i with mutation of the 4-bp spacer between them (Fig. 5B), were used for EMSA analysis. The increase in NgVapB1H concentration led to a band shift for P₀₃₄₅ (Fig. S4B), indicating that this antitoxin could bind to its own promoter. In the presence of 0.1 µM NgVapB1H, a band shift was observed for P₀₃₄₅ but not for P_0345M1r, P_0345M1f, or P_0345M1fr; at 0.2 µM NgVapB1H, P₀₃₄₅ and P_0345M1f, as well as P_0345M1i, exhibited pronounced band shifts, while P_0345M1r and P_0345M1fr only showed a weak band shift; when NgVapB1H concentration increased to 0.4 µM, P_0345M1r and P_0345M1fr also exhibited pronounced band shifts as P₀₃₄₅ and P_0345M1f did (Fig. S4C). These results reflect the important role of PPS1, particularly IR1r, in the binding of NgVapB1. Our attempts to probe the interaction between the NgVapBC1 complex and its native promoter via EMSA failed because the purified NgVapBC1 complex (Fig. 4C, lane 9) tended to form aggregates (data not shown), similar to the case of the FitAB complex (10).

Fig 5.

The effect of PPS1 on the vapBC1 operon promoter activity. (A) The sequence logo of promoter regions of NgVapBC1 and its 31 homologs from haloarchaea. The conserved inverted repeat motifs of PPS1 are indicated by arrows. The TATA box and WW motif are indicated. (B) Sequences of the NgVapBC1 promoter region (P₀₃₄₅) and its mutants. The mutated and unaltered nucleotides are shown in letters and dashes, respectively. The inverted repeats (IR1f and IR1r) are indicated by arrows. The TATA box and WW motif are underlined. (C) Promoter activity assay. Strain J7-2 and ΔvapBC1 carrying the blank vector (pSHS) or the expression plasmid for BgaH, driven by the indicated promoter, were grown in 23% MGM at 37°C to mid-exponential phase, followed by BgaH activity assay of cell extracts. The data are expressed as means ± SDs from biological triplicates (***, P < 0.001; **, P < 0.01; ns, no significance).

We then investigated the role of PPS1 in controlling vapBC1 expression in vivo, with the halophilic β-galactosidase (BgaH)-encoding bgaH (55) as a reporter gene. When grown in modified growth medium containing 3.1 M NaCl (23% MGM), strain J7-2 harboring the bgaH expression plasmid driven by P₀₃₄₅ exhibited much lower BgaH activity than ΔvapBC1 carrying the same plasmid (Fig. 5C), indicating that NgVapB1 and NgVapC1 repress the promoter activity of P₀₃₄₅ in strain J7-2. By contrast, there was no significant difference in BgaH activity between strain J7-2 harboring the bgaH expression plasmid driven by P_0345M1r or P_0345M1fr and ΔvapBC1 carrying the same plasmid; meanwhile, the difference in BgaH activity between strain J7-2 carrying the bgaH expression plasmid driven by P_0345M1f and ΔvapBC1 carrying the same plasmid is much less significant than that driven by P₀₃₄₅ (Fig. 5C). These results suggest that PPS1 acts as a negative cis-regulatory element and its mutation could relieve the repression effect of NgVapB1 and NgVapC1 on the promoter activity. We noticed that there was also no significant difference between strain J7-2 and ΔvapBC1 in expression level of BgaH driven by P_0345M1i (Fig. 5C), implying the spacer between IR1f and IR1r is involved in the function of PPS1. In ΔvapBC1 that lacks NgVapB1 and NgVapC1, the expression level of BgaH driven by the mutant (P_0345M1r, P_0345M1f, P_0345M1fr, or P_0345M1i) was lower than that driven by P₀₃₄₅ (Fig. 5C), suggesting that the mutation of PPS1 or the WW motif affects the promoter activity.

The NgVapBC1 complex is a stronger repressor than NgVapB1 for autoregulation of the vapBC1 operon

To investigate whether NgVapB1 and NgVapBC1 complex differ from each other in their capacities to repress vapBC1 expression, we constructed the plasmid for co-expression of P₀₃₄₅-driven BgaH with NgVapB1H or with both NgVapB1H and C-terminal His-tagged NgVapC1H (Fig. 6A). In an effort to ensure an excess of antitoxin over toxin to prevent possible toxic effect of NgVapC1 on host cells, the strong constitutive promoter P₀₅₆₆ and a relatively weak dnaK promoter (P_dnaK) of Hfx. volcanii (Fig. S5) were employed to drive the expression of NgVapB1H and NgVapC1H, respectively (Fig. 6A). The ΔvapBC1 strain expressing BgaH alone exhibited much higher BgaH activity than the strain co-expressing BgaH and NgVapB1H, while the strain co-expressing BgaH, NgVapB1H, and NgVapC1H showed almost no BgaH activity (Fig. 6A). These results suggest that NgVapB1 could repress the promoter activity of the vapBC1 operon and NgVapBC1 complex has a higher repression activity than NgVapB1 alone.

Fig 6.

Comparison of the capacities between NgVapB1 and NgVapBC1 complex to repress P₀₃₄₅ promoter activity (A) and dose-dependent effect of NgVapBC1 complex on P₀₃₄₅ promoter activity (B). ΔvapBC1 strains carrying the blank vector (pSHS) or the expression plasmid for BgaH alone or along with NgVapB1H or both NgVapB1H and NgVapC1H were grown at 37°C to mid-exponential phase in 23% MGM (A) or in 18% Hv-Cab with different concentrations of tryptophan (Trp) (B), followed by BgaH activity assay and anti-His-tag immunoblot analysis (B, inset) of cell extracts. The arrowheads indicate the positions of NgVapC1H (C1) and NgVapB1H (B1). Schematic representation of promoter (P₀₃₄₅, P₀₅₆₆, P_dnaK, or P_tnaA)-preceded recombinant genes on expression plasmids is shown. The data are expressed as means ± SDs from biological triplicates (***, P < 0.001; **, P < 0.01; *, P < 0.05; ns, no significance).

When P₀₃₄₅-driven BgaH was co-expressed with tryptophan-inducible P_tnaA-driven NgVapB1H and NgVapC1H in ΔvapBC1, the measured BgaH activity decreased with the increase of inducer concentration, accompanied by the increase in the amounts of NgVapB1H and NgVapC1H as evidenced by anti-His tag immunoblot analysis (Fig. 6B). By contrast, the BgaH activity exhibited by the strain expressing P₀₃₄₅-driven BgaH alone was not affected by the change of tryptophan concentration (Fig. 6B). These results imply that the expression level of NgVapBC1 complex positively correlates with its ability to repress vapBC1 operon promoter in vivo.

NgVapB1 is more resistant to proteolytic degradation when co-expressed with NgVapC1

The expression plasmid for NgVapB1H alone or the co-expression plasmid for both NgVapB1H and NgVapC1H was transferred into strain J7-2 or ΔvapBC1, and the resulting strains showed similar growth curves when grown in 23% MGM (Fig. 7A). Immunoblot analysis revealed that when the strains entered the stationary phase (8 days), NgVapB1H suffered severe proteolytic degradation by unidentified cellular protease(s) (Fig. 7B). We noticed that after 13-day cultivation, NgVapB1H molecules were almost completely degraded in strain J7-2 or ΔvapBC1 when expressed alone but were partially retained when co-expressed with NgVapC1H (Fig. 7B), implying that the formation of NgVapBC1 complex endows the antitoxin with increased resistance to proteolysis.

Fig 7.

In vivo proteolytic degradation of NgVapB1. The target strains were grown in 23% MGM at 37°C. At the indicated time points, samples were taken and subjected to OD₆₀₀ measurement (A) and anti-His-tag immunoblot analysis (B). Schematic representation of promoter (P₀₅₆₆ or P_dnaK)-preceded recombinant genes on expression plasmids are shown, and the data are expressed as means ± SDs from biological triplicates (A). The arrowheads indicate the positions of NgVapC1H (C1) and NgVapB1H (B1) (B).

It was previously reported that overexpression of Lon protease in E. coli leads to lethality, partly because the degradation of the antitoxin of a type II TA system (YefM-YoeB) by this protease liberates the toxin activity (56). Accordingly, we overexpressed strain J7-2 LonB protease (NgLonB) with a C-terminal His-tag (NgLonBH) in this haloarchaeon, driven by promoter P_sptA (50). However, the overexpression strain (J7-2/NgLonBH) and the control strain (J7-2/pSHS) showed similar growth curves (Fig. S6), suggesting that active NgVapC1 has not been released. We noticed that NgLonBH suffered degradation over time (Fig. S6, inset), which is consistent with a previous report on the Lon protease of Caulobacter crescentus (57).

Differential start codon usage by vapB1 and vapC1 and a −4 nt overlap of their stop and start codons contribute to the prevention of abnormal release of toxicity by the toxin

In the vapBC1 operon, there is a −4 nt overlap (GTGA) of the vapB1 stop codon (TGA) and the vapC1 start codon (GTG), and this overlap motif is conserved in haloarchaeal vapBC operons homologous to vapBC1 (Fig. 8A). Moreover, ATG and GTG are utilized as start codons by vapB1 and vapC1, respectively (Fig. 4A), and this start codon usage bias is also conserved in haloarchaeal vapB (Fig. 5A) and vapC (Fig. 8A). To investigate the impact of the differential start codon usage and the −4 nt overlap motif on NgVapBC1 regulation, we constructed vapBC1 operon mutants with different start codon usage patterns for vapB1 and vapC1 (Fig. 8B). When grown in 23% MGM, ΔvapBC1 strains carrying the expression plasmid for wild-type vapBC1 operon (strain AG) or its mutant (strain AA, GA, or GG) showed a similar growth curve to the control strain carrying pSHS by cell density quantification (Fig. 8C). Nevertheless, when entering the decline phase (e.g., 14 days), strains AA, GA, and GG showed significantly lower viable cell numbers than strain AG, while strain AG and the control strain displayed a similar viable cell number throughout the whole growth period (Fig. 8D). These results suggest that the use of ATG by vapB1 and GTG by vapC1 as start codons contributes to preventing the release of toxicity by NgVapC1.

Fig 8.

Role of start codon usage bias in vapBC1 operon expression. (A) The sequence logo of the regions around the start codon of NgVapC1 and its 16 homologs from haloarchaea. (B) Schematic representation of P₀₅₆₆-preceded recombinant genes with different start codons in expression plasmids. (C–H) Effect of differential start codon usage on expression of NgVapB1 and NaVapC1. The target strains were grown in 23% MGM at 37°C. At the indicated time points, samples were taken and subjected to OD₆₀₀ measurement (C), viable cell counting on 23% MGM agar plate at 37°C (D), determination of transcript levels by qRT-PCR analysis (E, F, and G), and anti-His-tag immunoblot analysis (H). For panel E, the vapB1/vapC1 mRNA ratio was calculated directly using the measured transcript levels of vapB1 and vapC1. For panels F and G, the values of the samples of AA, GA, and GG are relative to that of the sample of AG. For panel H, the arrowhead indicates the position of NgVapC1H (C1). The data are expressed as means ± SDs (***, P < 0.001; **, P < 0.01; *, P < 0.05; ns, no significance).

Although vapB1 and vapC1 are organized as a bicistronic operon, the qRT-PCR analysis of strain AG revealed a vapB1/vapC1 mRNA ratio of 13.1, 31.2, and 45.1 at exponential (3 days), stationary (8 days), and decline (14 days) phases, respectively (Fig. 8E), indicating that the vapBC1 transcript suffers from truncation of vapC1 mRNA. Under the same conditions, vapB1/vapC1 mRNA ratios of strains AA and GA were lower than those of strains AG and GG (Fig. 8E). Compared with strain AG, strain AA showed higher mRNA levels of vapB1 (Fig. 8F) and vapC1 (Fig. 8G), as well as a higher level of NgVapC1H (Fig. 8H). These results demonstrate that the use of ATG and GTG as start codons by vapB1 and vapC1, respectively, correlates with maintaining a high vapB1/vapC1 mRNA ratio and a lower level of vapC1 mRNA to avoid abnormal accumulation of excess toxin. Moreover, in each of the strains, the relative amounts of NgVapC1H (Fig. 8H) are roughly correlated with the relative levels of vapC1 mRNAs (Fig. 8G), reflecting the importance of differential start codon usage by vapB1 and vapC1 in controlling the levels of transcripts to modulate the amounts of antitoxin and toxin. Notably, regardless of which start codon (ATG or GTG) was used by vapB1, the use of GTG rather than ATG by vapC1 as the start codon led to lower mRNA levels of both vapB1 and vapC1 (Fig. 8F and G), suggesting that the start codon of the toxin is involved in the stability of the vapBC1 transcript.

We next investigated whether differential start codon usage influences the synthesis of NgVapB1 and NaVapC1 by separate monocistronic genes and probed the role of the −4 nt overlap in regulation of the NgVapBC1 system. The plasmid for co-expressing vapB1 with ATG or GTG start codon and vapC1 with GTG start codon, individually driven by P_tnaA, was constructed and transferred into ΔvapBC1, yielding the strains AG’ and GG’ (Fig. 9A). When grown in Hv-Cab medium with mevinolin and tryptophan, strain AG’ displayed a similar growth curve and viable cell number to the control strain carrying pSHS throughout the cultivation period, while strain GG’ exhibited a remarkable growth defect (Fig. 9B and C). At the exponential phase, strain AG’ showed a higher vapB1/vapC1 mRNA ratio (Fig. 9D) and a higher NgVapB1H/NaVapC1H ratio than strain GG’ (Fig. 9E), indicating that the replacement of the start codon ATG of vapB1 by GTG decreases the relative amount of the antitoxin to promote the release of toxicity by the toxin.

Fig 9.

Effect of start codon on ratio of NgVapB1 to NaVapC1 synthesized by separate monocistronic mRNAs. (A) Schematic representation of P_tnaA-preceded recombinant genes with different start codons in expression plasmids. (B–E) Effect of start codon on expression of NgVapB1 and NaVapC1. The target strains were grown at 37°C in 18% Hv-Cab supplemented with 1 mM tryptophan and 5 μg/mL mevinolin. At the indicated time points, samples were taken and subjected to OD₆₀₀ measurement (B), viable cell counting on Hv-Cab agar plate at 37°C (C), determination of transcript levels by qRT-PCR analysis (D), and anti-His-tag immunoblot analysis (E). The arrows in panel B indicate the sampling time points (the exponential phase) for qRT-PCR analysis (D) and anti-His-tag immunoblot analysis (E). For panel E, the relative amounts of NgVapB1H and NgVapC1H were determined by quantifying the band density (inset), and the arrowheads indicate the positions of NgVapC1H (C1) and NgVapB1H (B1). The data are expressed as means ± SDs from biological triplicates (***, P < 0.001; **, P < 0.01; *, P < 0.05; ns, no significance).

We noticed that the vapB1/vapC1 mRNA ratios of exponential-phase strains AG’ (1.7) and GG’ (0.8) expressing the separate monocistronic genes (Fig. 9D) were much lower than those of strains AG (13.1) and GG (7.7) expressing the bicistronic genes with the −4 nt overlap (3 days, Fig. 8E). These results suggest that the −4 nt overlap motif in the vapBC1 operon plays an important role in maintaining a higher vapB1/vapC1 mRNA ratio for prevention of the synthesis of excess toxin to release toxicity abnormally.

The NgVapBC1 system contributes to population heterogeneity of the decline-phase strain J7-2 and facilitates the survival of some subpopulations under stress conditions

To probe the physiological role of the NgVapBC1 system in strain J7-2, we compared the growth and survival of strain J7-2 and ΔvapBC1 under different conditions. Although strain J7-2 and ΔvapBC1 displayed similar growth curves by cell density quantification, the former showed a significantly higher variability among biological replicates in viable cell number than the latter at the decline phase (Fig. 10A; 16–18 days). The exponential (3 days) and decline (16 days) phase cells of strain J7-2 and ΔvapBC1 were collected and subsequently cultivated at different temperatures (37°C–53°C) and salinities (3.1–4.3 M NaCl), under nutrient-limiting condition (DM), and in the presence of the antibiotic novobiocin, respectively (Fig. 10B). Compared with ΔvapBC1, the decline, rather than exponential-phase strain J7-2, still showed a higher variability among biological replicates in viable cell number under these conditions (Fig. 10B). These results suggest that the NgVapBC1 system contributes to population heterogeneity of the decline-phase strain J7-2.

Fig 10.

The role of NgVapBC1 system in growth and survival of Strain J7-2. Strain J7-2 and ΔvapBC1 were grown in 23% MGM at 37°C, and samples were taken at the indicated time points for OD₆₀₀ measurement (inset in panel A) and viable cell counting on 23% MGM agar plates at 37°C (A). The exponential (3 days) and decline (16 days) phase cells were serially diluted and spotted onto MGM agar plates with different NaCl concentrations, 5-fold diluted MGM agar plates with 3.1 M NaCl (DM), or 23% MGM agar plates with 0.4 μg/mL novobiocin (Nov) and then cultivated at the indicated temperatures for 7 days before viable cell counting (B). For UV sensitivity, serially diluted exponential (3 days) and decline (16 days) phase cells were spotted onto 23% MGM agar plates and exposed to UV at 80 J/m², followed by cultivation at 37°C for 7 days in darkness before viable cell counting (C). The data in panels A and C were used for drawing correlation plots between the number of viable cells in the exponential (3 d) and decline (16 d) phase cultures and their survival after UV radiation (D). Significant differences between means (or medians) and variances of two groups are indicated in black and red, respectively (***, P < 0.001; **, P < 0.01; *, P < 0.05; ns, no significance).

The exponential- and decline-phase cells of strain J7-2 and ΔvapBC1 were subjected to UV radiation. In contrast to the exponential-phase cells, the decline-phase strain J7-2 showed a higher variability among biological replicates in survival rate than the decline-phase ΔvapBC1, and some biological replicates of the decline-phase strain J7-2 displayed higher levels of survival rate than those of ΔvapBC1 after UV exposure (Fig. 10C). These results suggest that the NgVapBC1 system-mediated population heterogeneity of strain J7-2 facilitates subpopulations to survive under stress conditions.

After UV exposure, the decline-phase cells of both strain J7-2 and ΔvapBC1 displayed higher levels of survival rate than their exponential-phase cells (Fig. 10C); moreover, the numbers of viable cells in the exponential-phase cultures of strain J7-2 and ΔvapBC1 are moderately and negatively correlated with their survival rates (R² = ~0.67) after UV exposure (Fig. 10D), implying that less metabolically active cells are more resistant to UV stress. Notably, for the decline-phase cultures, the viable cell number of strain J7-2 is highly and negatively correlated with its survival rate (R² = ~0.82) after UV exposure, whereas there is no correlation between the viable cell number and the survival rate (R² = ~0.02) of ΔvapBC1 under the same conditions (Fig. 10D). These results suggest that the NgVapBC1 system participates in the reduction of metabolic activity of decline-phase cells to improve their stress resistance.

We postulated that the population heterogeneity of decline-phase strain J7-2 may correlate with cell-to-cell fluctuation in active NgVapC1 level, which will lead to variation in metabolic activity and stress resistance of different subpopulations. To test this possibility, we employed the tryptophan-inducible promoter P_tnaA to drive the expression of NgVapC1H in ΔvapBC1 (Fig. 11A). With the increase in tryptophan concentration, the growth of ΔvapBC1 harboring the expression plasmid for NgVapC1H (ΔvapBC1/NgVapC1H) was inhibited to a greater extent and the amount of cellular NgVapC1H was increased (Fig. 11A). In the absence of tryptophan, ΔvapBC1/NgVapC1H still showed a delay in growth compared with the control strain ΔvapBC1/pSHS (Fig. 11A), probably due to leaky expression from the P_tnaA promoter. These results reflect the toxic effect of NgVapC1H on the metabolic activity of the cells. The 5-day cultures containing different concentrations of tryptophan were subjected to viable cell counting and UV treatment. The results showed that the numbers of viable cells (Fig. 11B) in the cultures are negatively correlated with their survival rates after UV exposure (Fig. 11C), in particular ΔvapBC1/NgVapC1H. These results validate the role of NgVapC1 in improving the stress resistance of strain J7-2 by reducing cellular metabolic activity.

Fig 11.

Effect of NgVapC1 on growth and survival of strain J7-2. The target strains were grown at 37°C in 18% Hv-Cab with 5 μg/mL mevinolin and different concentrations of tryptophan (Trp), and samples were taken at the indicated time points for OD₆₀₀ measurement (A), anti-His-tag immunoblot analysis (A, inset; C1 indicates the position of NgVapC1H), and viable cell counting on Hv-Cab agar plates at 37°C (B). The cells of 5-day cultures were spotted onto Hv-Cab agar plates and exposed to UV at 80 J/m², followed by cultivation at 37°C for 7 days in darkness before viable cell counting (C). The data are expressed as means ± SDs (n = 10; ***, P < 0.001; *, P < 0.05; ns, no significance).

DISCUSSION

Our results show that four of the eight putative VapBCs of Nnm. gari J7-2 are functional. The NgVapBC1 system was further characterized in terms of its regulation at multiple levels and role in the establishment of population heterogeneity of the haloarchaeon.

Nonfunctional and functional VapBCs

For the four nonfunctional VapBCs of strain J7-2, all of their toxins except NJ7G_3795 possess the four active-site residues of VapC toxins, but none of them show detectable toxicity in our assays. Similarly, the overexpression of some M. tuberculosis-derived VapCs with an intact active site in M. bovis does not result in growth inhibition (58). The mechanism for the nontoxicity of these VapCs remains to be further elucidated. The NJ7G_2729-GFP fusion protein was found to be recruited to the FtsZ ring in Natrinema sp. J7, implying that NJ7G_2729 participates in cell division (42). Considering that NJ7G_2730 and NJ7G_2729 are organized in an operon, we postulate that NJ7G_2730 may be involved in cell division. Additionally, the toxin MtvT of a plasmid-encoded type II TA system MtvTA exhibits no detectable toxicity, but the MtvTA system could regulate plasmid conjugative transfer and bacterial virulence in Pseudomonas aeruginosa (59). It is possible that in strain J7-2, the putative VapCs without cell toxicity, instead of acting as toxins, may actually exert other functions.

Besides NgVapBC1, the other three putative VapBCs (NJ7G_0582/NJ7G_0581, NJ7G_2439/NJ7G_2438, and NJ7G_2321/NJ7G_4368) of strain J7-2 are experimentally confirmed to be functional in strain J7-2 and Hfx. volcanii. The evidence that NJ7G_4368 exhibits a weaker toxicity to Hfx. volcanii than NgVapC1, NJ7G_0581, or NJ7G_2438 may reflect a functional diversity among these VapBCs. Nevertheless, regulatory mechanisms and physiological roles of the three functional VapBCs, as well as the possible relationship among the multiple TA systems of strain J7-2, remain to be elucidated.

Transcriptional regulation

The antitoxin NgVapB1 or the NgVapBC1 complex negatively autoregulates its own transcription by binding to the pseudo-palindromic sequence PPS1 in the vapBC1 operon promoter of strain J7-2. PPS1 is downstream of the TATA box recognized by the TATA-binding protein (TBP) and flanks the WW motif that possibly interacts with the transcription factor B (TFB) and RNA polymerase (RNAP) in haloarchaea (54). Because the formation of the TBP-TFB-DNA complex is required for recruiting RNAP to form the preinitiation complex (PIC) (60), the occupation of PPS1 by NgVapB1 or the NgVapBC1 complex would affect PIC formation necessary for vapBC1 operon transcription. The conservation of PPS1 in vapBC operon promoters of haloarchaea emphasizes the role of this negative cis-regulatory element in autoregulation of haloarchaeal VapBC systems. To the best of our knowledge, the present results are the first evidence of the PPS-mediated autoregulation for archaeal type II TA systems. In bacteria, vapBC operons also employ PPSs for autoregulation (9, 13). While haloarchaeal vapBC operon PPSs are downstream of the TATA box, bacterial vapBC operon PPSs generally cover the −10 and/or −35 region, which can be occupied by the VapBC complex to prevent the interaction of the core promoter elements with RNAP holoenzyme (4, 61, 62). Despite the difference in the position of PPS to the TATA box in archaea and −35/−10 regions in bacteria, both archaeal and bacterial vapBC operons employ PPSs for binding toxin-antitoxin complexes to block the initiation of transcription, suggesting that the transcriptional autoregulation mechanism of VapBC systems is conserved across the two domains of life.

Post-transcriptional and translational regulation

Strain J7-2 maintains a high vapB1/vapC1 mRNA ratio through truncation of the vapBC1 transcript within the toxin-coding region, allowing the production of an excess of NgVapB1 to prevent abnormal release of toxicity by NgVapC1. Transcripts with truncation of the toxin-coding region have been reported for bacterial type II TA systems. In the case of the MazEF system of E. coli, the toxin MazF cleaves the mazEF transcript specifically at ACA sites, leading to a high mazE/mazF mRNA ratio due to the presence of more ACA sites in the coding region of MazF (6). For the Kis-Kid system of plasmid R1, its truncated transcript originates from limited degradation of the toxin Kid-coding sequence containing a stem-loop structure, which is located in the 5′ region of kid and may act as a protective barrier against degradation of upstream mRNA by 3′–5′ exoribonucleases (5). Additionally, the RNA-seq data reveal toxin-coding region truncated transcripts of the DinJ-YafQ and YafNO systems of E. coli, but the underlying mechanism remains unclear (63). In strain J7-2, a single-base mutation (GTG → ATG) only in the start codon of vapC1 causes a significant change in the vapB1/vapC1 mRNA ratio, suggesting that the NgVapBC1 system is unlikely to maintain a higher vapB1 mRNA level through a mechanism similar to MazEF (6) or Kis-Kid (5).

The −4 nt overlap (GTGA) of the vapB1 stop codon and the vapC1 start codon in the vapBC1 operon is conserved in homologous vapBC operons of haloarchaea. It was recently reported that a −4 nt overlap (ATGA or GTGA) of the stop and start codons is the most common spacing motif in unidirectional gene pairs of prokaryotes and is required for translational coupling via termination-reinitiation (TeRe), whereby the same ribosome (or at least the small subunit) that terminates translation of the upstream gene reinitiates translation of the downstream gene that could not be expressed in the absence of translational coupling (64, 65). We found that separate monocistronic vapB1 and vapC1 generate a much lower vapB1/vapC1 mRNA ratio than the bicistronic vapBC1 operon, emphasizing the importance of translational coupling for modulating the antitoxin/toxin mRNA ratio. Moreover, the replacement of the start codon GTG of vapC1 by ATG does not disrupt the integrity of the −4 nt overlap (GTGA → ATGA) but could improve the vapBC1 transcript stability and decrease the vapB1/vapC1 mRNA ratio. Consistent with this, ATG is more efficient than GTG not only in initiating translation but also in binding the ribosome to protect the transcript against RNase-mediated degradation in haloarchaea (50, 66). Given that the vapBC1 operon employs the TeRe mechanism for translation, the truncated vapBC1 transcript probably results from a weaker protective effect of the ribosome on the downstream toxin-coding region. Specifically, the translation of vapC1 depends on the completion of the translation of upstream vapB1, and during translation, the polysome-associated vapB1 mRNA is resistant to RNase; by contrast, the use of a “weak” start codon GTG by vapC1 leads to less efficient translation, and thus, the vapC1 mRNA is less occupied by ribosomes and more sensitive to RNase-mediated degradation. Moreover, translational coupling could facilitate the coordinated translational regulation of overlapping gene pairs to ensure a precise stoichiometry of interacting proteins (64, 65). In this context, the −4 nt overlap-mediated translational coupling of vapB1 and vapC1 allows both post-transcriptional and translational regulation of the NgVapBC1 system.

Post-translational regulation

It has long been held that type II antitoxins are more sensitive to proteolysis than their cognate toxins, promoting the liberation of the latter (1, 56). However, accumulating evidence suggests that the activation of toxins in TA complexes is unlikely due to preferential degradation of the antitoxin because antitoxins are protected from degradation after forming stable TA complexes with toxins (1, 2, 7). In agreement with this, NgVapB1 could form a complex with NaVapC1, and it suffers severe proteolytic degradation when expressed alone in strain J7-2 but becomes more resistant to proteolysis when co-expressed with the toxin. Moreover, the evidence that overexpression of LonB protease in strain J7-2 does not result in lethality raises the possibility that active NgVapC1 could not be liberated through proteolytic degradation of NgVapB1 by LonB, in contrast to the E. coli YefM-YoeB system that is activated via Lon-dependent degradation of the antitoxin (56). Notably, haloarchaeal LonB proteases differ from their bacterial homologs in that the former are membrane-associated while the latter are typically soluble (67). It is presently unclear whether membrane association of LonB limits its action on NgVapB1. Additionally, we cannot rule out the possibility that NgVapB1 is inherently resistant to LonB. Some non-proteolytic mechanisms have been proposed for toxin activation of type II TA systems of M. tuberculosis, such as chaperone-mediated conformational change in HigBA complex (68), thermal remodeling of ParDE1 complexes (69), and phosphorylation of VapB to inhibit the formation of VapBC complex (70). Further studies are needed to investigate whether these non-proteolytic mechanisms are applicable to the NgVapBC1 system.

Physiological roles

It was reported that the deletion of vapBC6 in S. solfataricus or vapBC4 in S. acidocaldarius leads to increased thermal sensitivity, indicating critical involvement of VapBCs in response to thermal stress in thermoacidophilic archaea (11, 12). By contrast, the exponential-phase cells of strain J7-2 and ΔvapBC1 do not display phenotypic differences under different stress conditions, implying that NgVapBC1 does not function as a bona fide stress-responsive system. Instead, NgVapBC1 contributes to the population heterogeneity of the decline-phase strain J7-2. The establishment of heterogeneous populations with distinct toxin activities and metabolic capacities has been observed for several type II TA systems of bacteria (71). TA systems are tightly regulated at multiple levels and implemented in regulatory networks, of which all components can be differentially regulated at all levels to facilitate or hinder phenotypic switches via toxin activation (71, 72). When strain J7-2 cells enter the decline phase, the loss of cellular homeostasis would disturb the regulatory network of the NgVapBC1 system to trigger stochastic activation of NgVapC1, leading to cell-to-cell variation in toxin activity and metabolic capacity.

Our results suggest that the establishment of population heterogeneity of strain J7-2 is associated with NgVapBC1 and could confer fitness advantages to the haloarchaeon. Previous reports show that VapC toxins cleave RNAs (tRNA, rRNA, and/or mRNA) to inhibit translation, thereby slowing down cellular metabolism and growth (10–12). In general, slow-growing cells are more stress-tolerant than fast-growing ones (73). Indeed, the growth rate of strain J7-2 is negatively correlated with the survival rate of this haloarchaeon under UV stress. For the decline-phase strain J7-2, some subpopulations with a higher level of active NgVapC1 would grow more slowly but survive better under stress conditions; in contrast, other subpopulations with a lower level of active NgVapC1 are less stress-resistant but may contribute more to population restoration once the stress is relieved. Meanwhile, we do not exclude the possibility that the toxin-mediated inhibition of translation of certain proteins (e.g., DNA replication-related proteins) may affect the genome stability of strain J7-2 to generate genetic heterogeneity, which also enables the population to better withstand different stresses. In this context, the NgVapBC1-mediated growth and metabolic heterogeneity facilitate the survival and multiplication of the strain J7-2 population as a whole in diverse and changing environments. Given that VapC toxins act on RNAs, further single-cell transcriptomics analysis is expected to reveal the RNA substrates and the underlying mechanism responsible for NgVapBC1-mediated population heterogeneity of strain J7-2.

MATERIALS AND METHODS

Strains and growth conditions

The bacterial and haloarchaeal strains used in this study are listed in Table S1 in the supplemental material. Unless otherwise indicated, Nnm. gari J7-2 and its derivatives were grown aerobically at 37°C with shaking at 185 rpm in 23% MGM supplemented with 5 µg/mL mevinolin when necessary, as described previously (39). The Hv-Cab medium (51) was used for cultivating Hfx. volcanii WFD11, strain J7-2, and its derivatives expressing the target gene driven by the tryptophan-inducible promoter P_tnaA at 37°C in the presence of the indicated concentration of tryptophan. When necessary, novobiocin (4 μg/mL) was added into the Hv-Cab medium. E. coli DH5a, E. coli JM110, and E. coli BL21(DE3) were used as hosts for plasmid construction, demethylation, and expression, respectively, and were grown at 37°C in Luria-Bertani medium supplemented with kanamycin (30 µg/mL) as needed. Solid medium was prepared by adding 1.5% (wt/vol) agar.

Plasmid construction and mutagenesis

The plasmids and primers used in this study are listed in Tables S2 and S3, respectively. The E. coli expression plasmids were constructed by inserting the target gene fragment, amplified from the strain J7-2 genome by PCR, into the NdeI-NcoI site of pET26b (Novagen). The plasmids for genome editing of strain J7-2 were constructed as described previously (52). Briefly, a mini-CRISPR preceded by promoter P_sptA (50) was synthesized by PCR using partially overlapping primers (Table S3). The homologous arms flanking the target gene were amplified from the strain J7-2 genome and connected by overlapping PCR to generate a donor DNA. Using the GenRec Assembly Master Mix Kit (generalbiol, Anhui, China), the P_sptA-preceded mini-CRISPR and the donor DNA were inserted into the BamHI-XbaI site of pSHS, yielding the genome-editing plasmid. To construct the vectors for expressing recombinant proteins in haloarchaea (Hfx. volcanii WFD11, strain J7-2, and its derivatives), different promoters were amplified from the genome of strain WFD11 (P_dnaK and P_tnaA) or strain J7-2 (P_sptA, P_fsptA, P₀₅₆₆, and P₀₃₄₅), and connected with the target gene fragment amplified from the strain J7-2 genome or bgaH amplified from plasmid pST1 (50) by overlapping PCR, followed by insertion into the KpnI-NcoI site of pSY1 (50) or the BamHI-XbaI site of pSHS (52). The sequences of all recombinant plasmids were verified by DNA sequencing.

Construction of gene-deletion mutants of Nnm. gari J7-2

Deletion of target genes in strain J7-2 was carried out using the endogenous CRISPR-Cas system-based genome-editing method as described previously (52) with minor modifications. Briefly, a non-methylated genome-editing plasmid was isolated from E. coli JM110 and then transferred into strain J7-2 through polyethylene glycol-mediated transformation, followed by plating on 18% MGM agar plates. For screening of the target mutant, randomly selected colonies were cultivated in liquid 23% MGM, and the cells were recovered by centrifugation and lysed with sterile deionized water, followed by PCR analysis with external and internal primer pairs specific for the target gene. For curing of the plasmid, the PCR-verified mutant was subjected to 2–3 rounds of subculturing (2 days for each round) in liquid 23% MGM, and the absence of the plasmid was confirmed by PCR analysis. Finally, the mutants were validated by DNA sequencing of the PCR product of the target region of the genome.

Expression in E. coli, purification, and refolding of recombinant proteins

The expression, purification, and refolding of recombinant proteins were carried out as described previously (40) with modifications. Briefly, after induction with isopropyl-β-D-thiogalactopyranoside (IPTG), E. coli BL21(DE3) cells expressing the target protein were suspended in 50 mM Tris-HCl buffer (pH 8.0) containing 3 M KCl and disrupted by sonication on ice, followed by centrifugation (13,400 × g for 10 min) to collect soluble and insoluble fractions. The soluble fraction containing NgVapB1H was subjected to Ni-NTA affinity chromatography on a Ni²⁺-charged Chelating Sepharose Fast Flow (GE Healthcare) column, and the elution fraction was dialyzed against 50 mM Tris-HCl buffer (pH 8.0) containing 3 M KCl overnight at 4°C, followed by centrifugation (13,400 × g, 10 min) to recover the soluble fraction. The insoluble fraction containing NgVapC1H was solubilized in 50 mM Tris-HCl (pH 8.0) containing 8 M urea, and then diluted 1:100 with 50 mM Tris-HCl buffer (pH 8.0) containing 3 M KCl for protein refolding. After incubation at room temperature (~25°C) for 30 min, the dilution was centrifuged at 13,400 × g for 15 min to recover the soluble fraction as refolded NgVapC1H. Protein concentration was determined using the Bradford method with bovine serum albumin (BSA) as a standard.

β-Galactosidase activity assay

The β-galactosidase activity assay of strains carrying an expression plasmid for β-galactosidase was determined using o-nitrophenyl-β-D-galactopyranoside (Meilunbio, Dalian, China) as the substrate as described previously (39).

SDS-PAGE and immunoblot analyses

Protein samples were precipitated with 20% (wt/vol) trichloroacetic acid and washed with acetone before being subjected to Tris-tricine SDS-PAGE analysis. The anti-His tag monoclonal antibody (1:10,000, Proteintech, Chicago, USA) and the horseradish peroxidase-conjugated goat anti-mouse IgG secondary antibody (1:10,000, Biodragon, Jiangsu, China) for immunoblot analysis as described previously (40), and the signals were detected with enhanced chemiluminescence (ECL) reagent (Biodragon). The images were photographed using the ChemiDoc MP imaging system (Bio-Rad, USA) with an exposure time of 0.5 or 2 min.

Determination of mRNA level

Total RNA extraction from cells was performed using Triquick Reagent (Solarbio, China) according to the manufacturer’s instructions. Genomic DNA removal and reverse transcription were conducted using PrimeScript RT Reagent Kit with gDNA Eraser (Takara Bio Inc, Beijing) according to the manufacturer’s instructions. Using the gene-specific primers (Table S3), the qRT-PCR was carried out in a 20 μL reaction mixture as described previously (50). The qRT-PCR results were calculated using the threshold cycle (2^-ΔΔC_T) method (74). To quantify the transcript levels of target genes, 16S rRNA C_T values were used to normalize the C_T values of target transcripts.

Electrophoretic mobility shift assay

The EMSA assay was carried out as described previously (75) with modifications. In brief, target DNA fragments for EMSA were PCR-amplified from strain J7-2 genomic DNA using the sequence-specific primers (Table S3) and purified by DNA agarose gel electrophoresis using a Gel Extraction Kit (Omega). Purified NgVapB1H (0.1–0.4 μM) was incubated with target DNA fragment (100 nM) in binding buffer (20 mM Tris-acetate [pH 6.0], 3 M KCl, 2 mM dithiothreitol, 2 mM EDTA, 15 mM MgCl₂, 57 mM β-mercaptoethanol, 100 µg/mL BSA, and 10% [wt/vol] sorbitol) at 30°C for 30 min. Subsequently, a 20 µL reaction mixture was applied to a 6.5% native PAGE gel in 1× TBE buffer (89 mM Tris, 89 mM borate, and 2 mM EDTA, pH 8.3), and the electrophoresis was run at 120 V for 2 h at 4°C. The DNA bands on the gels were stained with ethidium bromide and visualized by the BioDocAnalyze Gel Imaging System (Analytik Jena).

Bioinformatic analyses

The structural models of proteins were predicted by using the AlphaFold3 server (46). Molecular graphics and analyses were performed with PyMOL (Version 2.4.0, Schrödinger, LLC) with the source code available at https://github.com/schrodinger/pymol-open-source. Sequence logos were generated using WebLogo (https://weblogo.threeplusone.com/).

Statistical analysis

All statistical analyses were performed using GraphPad Prism 9.0 (Graphpad Software, Inc., San Diego, CA, USA). Normality was assessed using the Shapiro-Wilk test. For normally distributed data, unpaired t-test and Welch’s t-test were applied for those without and with significant differences in standard deviations, respectively, to determine the significant difference between the means of the two groups; meanwhile, the F-test was used to determine the significant difference between variances of the two groups. For non-normally distributed data, the Mann-Whitney U test was used to determine the significant difference between medians of the two groups, while the Brown-Forsythe test was used to determine the significant difference between variances of two groups. In all statistical comparisons, differences were considered statistically significant at P < 0.05.

DATA AVAILABILITY

The genome sequence of Natrinema gari J7-2 has been deposited in GenBank under the accession number CP003412.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aem.00747-26.

Supplemental material. aem.00747-26-s0001.pdf.

Tables S1 to S3; Fig. S1 to S6.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.