A Review of the Functional Annotations of Important Genes in the AHPND-Causing pVA1 Plasmid

Acute hepatopancreatic necrosis disease (AHPND) is a lethal shrimp disease. The pathogenic agent of this disease is a special Vibrio parahaemolyticus strain that contains a pVA1 plasmid. The protein products of two toxin genes in pVA1, pirAvp and pirBvp, targeted the shrimp’s hepatopancreatic cells and were identified as the major virulence factors. However, in addition to pirAvp and pirBvp, pVA1 also contains about ~90 other open-reading frames (ORFs), which may encode functional proteins. NCBI BLASTp annotations of the functional roles of 40 pVA1 genes reveal transposases, conjugation factors, and antirestriction proteins that are involved in horizontal gene transfer, plasmid transmission, and maintenance, as well as components of type II and III secretion systems that may facilitate the toxic effects of pVA1-containing Vibrio spp. There is also evidence of a post-segregational killing (PSK) system that would ensure that only pVA1 plasmid-containing bacteria could survive after segregation. Here, in this review, we assess the functional importance of these pVA1 genes and consider those which might be worthy of further study.


Introduction
Acute hepatopancreatic necrosis disease (AHPND) is a bacterial disease that causes severe damage in shrimp farming [1]. In 2009, the first outbreak of AHPND was seen in China, and the disease quickly spread to Vietnam, Malaysia, Thailand, Philippines [1], and Bangladesh [2]. According to a recent report, it is now also present in the USA [3].
The original AHPND-causing bacteria was identified as a specific strain of Vibrio parahaemolyticus [4][5][6][7][8][9][10]. V. parahaemolyticus is a Gram-negative halophilic bacterium that can be found in marine environments [11]. In the past, V. parahaemolyticus was already recognized as a causal agent for human acute gastroenteritis in raw, contaminated sea foods [11], and it contains two hemolysin virulence factors, thermostable direct hemolysin (tdh), and TDH-related hemolysin (trh), that are both confirmed to cause damage

The AHPND-Causing Plasmid pVA1
The AHPND-causing plasmid, pVA1, was first discovered by next-generation sequencing (NGS). By comparing the DNA sequences between V. parahaemolyticus strains that caused AHPND (3HP, 5HP, and China) and those that did not (S02), a 69-kb plasmid, pVA1, was found in the 3HP, 5HP, and China strains but not in the S02 strain [15][16][17][18][19]. Subsequently, the pVA1-encoded binary pore-forming PirAB vp toxins were identified as the key factors in the AHPND pathology [5], and in addition to specific strains of V. parahaemolyticus, other Vibrio species, such as V. harveyi, V. campbellii, V. owensii, and V. punensis, have also become AHPND-causing agents [20][21][22][23]. All of these AHPND-causing Vibrio spp. harbor pVA1-related plasmids that contain the pirA and pirB toxin genes [5,[20][21][22][23]. Recently, a non-Vibrio bacterium, Microccocus luteus, was also reported to carry the pirA and pirB toxin genes [24]. Another report confirmed that pVA1 pirAB vp genes could be transferred from an AHPND-causing V. parahaemolyticus to a non-Vibrio and non-pathogenic Algoriphagus sp. [25]. These developments suggest that pVA1-related plasmids could be transmitted between different species, which would make it more difficult to control AHPND. It is already known that pVA1-related plasmids contain a number of genes for horizontal gene transfer, plasmid transmission, and maintenance, including the post-segregational killing (PSK) system. Meanwhile, components of type II and III secretion systems may further facilitate the toxic effects of pVA1-containing bacterial spp. Therefore, in addition to the toxic pirA and pirB genes, these plasmid-related genes ( Figure 1; Table 1) are also worthy of further study. In this review, we will discuss the functional roles of these genes/proteins, and hopefully, open new areas for AHPND research.

PirA vp (ORF 51) and PirB vp (ORF50)
The products of these two genes were named based on their sequence homology to a binary toxin family: Photorhabdus insect-related (Pir)-like proteins. As mentioned above, PirA vp and PirB vp have been confirmed as the major factors that cause damage to the hepatopancreatic tissues of shrimps, and this is considered the key symptom of AHPND [4][5][6][7][8][9][10]. Although some reports showed PirB vp alone has the ability to cause cell damage [5,6], both PirA vp and PirB vp are necessary to achieve full toxicity. The pirA vp and pirB vp genes are both part of the same operon [5], which means that they are regulated and expressed synchronously.     Structural analysis showed that there was a functional relationship between PirA vp /PirB vp and the Bacillus thuringiensis Cry toxin [5,9]. Functionally, Cry toxins act as pore-forming toxins and achieve cytotoxic effects via three domains: the N-terminal pore-forming domain I, the middle receptor binding domain II and C-terminal sugar/receptor binding domain III [43][44][45]. The typical cytotoxic mechanism of the Cry toxin is as follows: domain III first recognizes the N-Acetylgalactosamine (GalNAc) that is present on several receptors (e.g., aminopeptidase N, APN). Subsequently, domain II also binds to the recognized receptor. Then, the α1 helix of domain I is hydrolyzed, and this triggers the oligomerization of the Cry toxin to form pores on the cell membrane [43][44][45].
It was found that PirA vp and PirB vp have a Cry toxin-like folding [5,9]. The N-terminal and C-terminals domain of PirB vp (PirB vp N and PirB vp C) have a similar folding to the pore-forming domain I and receptor binding domain II of Cry toxin, respectively. The PirB vp N has an alpha-helical bundle structure, and this is called an inside-out membrane fold. The inside-out membrane fold consists of a hydrophobic α-helix surrounded by multiple amphipathic α-helices, and it is often found in other pore-forming toxins such as colicin and Cry toxins [9]. The structural characteristics of this domain allow it to switch between soluble (hydrophobic α-helix hides inside; inactivated) and transmembrane (hydrophobic α-helix exposed; activated and toxic) forms [9]. Meanwhile, the PirB vp C contains an antiparallel β-barrel jelly-roll topology. Despite the divergence of their amino acid sequences, PirB vp C and Cry domain II share a similar antiparallel β-barrel jelly-roll topology, suggesting that PirB vp C may also be involved in receptor binding. There is also a recent report which showed that alpha amylase-like protein, which is a 1,4-α-d-glucan glucanohydrolase targeted by the Cry toxin, has the ability to interact with PirB vp [46]. However, this result was based on far western and mass spectrometry analysis [46] and will need to be confirmed by other methods. Shrimp receptors for PirB vp C, thus still remain to be further investigated.
Like PirB vp C, PirA vp also has a jelly-roll topology. The structural analysis further shows that PirA vp has a substrate-binding pocket [9]. Although the precise targets of PirA vp have not yet been identified, the structural similarity of PirA vp to Cry domain III implies that these targets may include carbohydrates such as GalNAC [9].
Given their structural similarities, the cytotoxic activation of PirA vp /PirB vp should be similar to that of the Cry toxins. If so, then the first step for PirA vp /PirB vp would be to assemble into a Cry-like three-domain complex. In 2019, Lin et al. proposed binding models for PirA vp /PirB vp by using crosslinking based-and hydrogen-deuterium exchange-mass spectrometry. As shown in Figure 2, The PirA vp /PirB vp heterodimer model has all the functional domains that are necessary for its cytotoxicity: i.e., binding to the carbohydrate/receptor by PirA vp and PirB vp C would be followed by the creation of an uncontrollable pore by PirB vp N on the cell membrane to induce cell death. However, the binding affinity of PirA vp and PirB vp was found to be relatively low (~7 µM) in vitro [13], suggesting that PirA vp and PirB vp would not form a stable complex when the two proteins interact directly. It is possible, however, that the interaction with the receptor/sugar may improve the stability of the PirA vp /PirB vp complex. We also note that in contrast to the structural characteristics of PirA vp and PirB vp C discussed above, in this model, PirA vp and PirB vp C were matched respectively to the Cry receptor binding domain II and sugar-binding domain III (Figure 2 lower images). This suggests that the real functions of PirB vp C and PirA vp may be reversed. Interestingly, a recent report further supports this idea by finding that PirB vp , but not PirA vp , is able to target glycosaminoglycans like GalNH 2 and GlcNH 2 [47]. Clearly, more work needs to be done before we fully understand how these toxins, in fact, attach to the shrimp cell membrane.

Transposases (ORF15, ORF48, ORF55, ORF57 and ORF68)
DNA transposons are mobile DNA elements that can move from one DNA molecule to another. Transposases are a type of enzyme that can bind to the inverted sequences on both ends of a transposon, digest, and release the transposon, and promote translocation of this DNA fragment to another location on the genome [26,48]. This provides a mechanism to deliver genetic information into other chromosomes and confer new functions to a gene or replace a defective gene.

Transposases (ORF15, ORF48, ORF55, ORF57 and ORF68)
DNA transposons are mobile DNA elements that can move from one DNA molecule to another. Transposases are a type of enzyme that can bind to the inverted sequences on both ends of a transposon, digest, and release the transposon, and promote translocation of this DNA fragment to another location on the genome [26,48]. This provides a mechanism to deliver genetic information into other chromosomes and confer new functions to a gene or replace a defective gene.
In pVA1, a total of five transposase genes were found (ORF15, 48, 55, 57, and 68). Apart from ORF57, which only contains a zinc-binding domain of the IS91 transposase family, the other four genes are complete transposases. ORF15, 48, and 55 share 100% sequence identity, and as members of the IS5 family of DDE transposases, they all contain a conserved DNA hydrolysis DDE (Asp-Asp-Glu) motif. The remaining transposase, ORF68, belongs to the Rpn family of recombinationpromoting nuclease/putative transposases. All four of these transposases have the ability to trigger the transfer of DNA transposons.
In previous reports, it has been shown that the pirAB vp genes can be deleted/transferred by gene transposition. For instance, in the pVA1 plasmid isolated from V. parahaemolyticus strain 3HP, it was found that the ORF49-54 gene cluster, which includes the pirAB vp genes (ORF50 and ORF51) was In pVA1, a total of five transposase genes were found (ORF15, 48, 55, 57, and 68). Apart from ORF57, which only contains a zinc-binding domain of the IS91 transposase family, the other four genes are complete transposases. ORF15, 48, and 55 share 100% sequence identity, and as members of the IS5 family of DDE transposases, they all contain a conserved DNA hydrolysis DDE (Asp-Asp-Glu) motif. The remaining transposase, ORF68, belongs to the Rpn family of recombination-promoting nuclease/putative transposases. All four of these transposases have the ability to trigger the transfer of DNA transposons.
In previous reports, it has been shown that the pirAB vp genes can be deleted/transferred by gene transposition. For instance, in the pVA1 plasmid isolated from V. parahaemolyticus strain 3HP, it was found that the ORF49-54 gene cluster, which includes the pirAB vp genes (ORF50 and ORF51) was flanked by two transposase genes in opposite directions (ORF 48 and ORF55) with inverted repeats in their terminals. It was also found that the gene cluster consisting of ORF 48-ORF54 was lost in the pVA1 type plasmid isolated from V. parahaemolyticus strain M2-36, leaving only ORF55 behind [5]. A gene cluster containing pirAB vp has also been inserted into another AHPND-causing plasmid, pVH, isolated from V. owensii [49]. The same transposase genes with inverted repeats were also found up-and down-stream of this cluster [49]. Another report describes a pVA1 plasmid from a natural V. parahaemolyticus strain (XN87), which contains a mutated pirA vp gene that is interrupted by a transposon gene fragment [50]. This results in an out-of-frame insertion of pirA vp and further affects the downstream gene expression of pirB vp [50]. Furthermore, whole-genome sequencing of 40 V. parahaemolyticus isolated from shrimp hepatopancreas and aquaculture water in Malaysia indicated that pirAB vp genes are prone to deletion [51]. Taken together, these results demonstrate the importance of this DNA transposon in the transfer, deletion, and mutation of the pirAB vp -containing gene cluster. Screening other bacteria or plasmids for the presence of this gene cluster will also be useful for identifying potential AHPND-causing pathogens. Conjugation is a process used by bacteria to transfer their genetic material to a recipient bacterium. This horizontal gene transfer facilitates bacterial evolution, including the dissemination of antibiotic resistance genes [52]. During conjugation, a single-stranded DNA molecule is first produced, and the plasmid undergoes DNA replication and formation of the protein-DNA complex, which is processed by the DNA transfer and replication (Dtr) system. The coupling protein (CP) further delivers this protein-DNA intermediate to the trans-membrane channel, where it is actively secreted through the channel and the exocellular pili, which are produced by the mating pair formation (Mpf) system. The Mpf/CP conjugation system is a kind of type-IV secretion system (T4SS), and there are twelve components essential for the Mpf system, TraF, TrbB, TrbC, TrbD, TrbE, TrbF, TrbG, TrbH, TrbI, TrbJ, TrbK, and TrbL [30,31]. On the pVA1 plasmid, all of these genes have been found except for trbJ and trbK. In addition, ORF11 is annotated as a VirB1-like gene, which is an important component that facilitates the formation of type-IV secretion systems (T4SS) [32]. This mechanism is, therefore, available for pVA1-related plasmids or genes to use for transferring between different bacterial cells.

Antirestriction Proteins (ORF32 and ORF35)
Among different species of bacteria, conjugative plasmids play a crucial role in spreading a variety of genes between various bacterial species. However, the transfer of DNA in naturally occurring vectors is tightly regulated by the immigration system, or restriction-modification (R-M) system, of the host [56]. The bacterial R-M system acts as an immune system that attacks the foreign DNA in the cell [56]. To overcome this restriction barrier of the host, some donor plasmids encode antirestriction proteins such as Ard (alleviation of restriction of DNA). On the pVA1 plasmid, there are two putative antirestriction protein genes, ORF32 and ORF35, which respectively have deduced amino acid sequences with similarity to ArdB and ArdC.
Ard genes are commonly found in transposons and conjugative plasmids in various prokaryotes [57]. The Ard antirestriction proteins are of three types, ArdA, ArdB, and ArdC, all of which act as inhibitors of the type I R-M system [33,34,[58][59][60][61]. ArdA and ArdB proteins are small and acidic, and they both include a small region of similarity that consists of 14 residues designated as the "antirestriction" domain [58]. However, while ArdA has a DNA-like shape and charge distribution, suggesting it acts as a DNA-mimic [59,62], ArdB proteins have a novel structural fold, and do not show any DNA-like properties [60]. Meanwhile, ArdC proteins can also bind to single-stranded DNA, and they have been observed to protect single-stranded but not double-stranded plasmid DNA against the activity of type II restriction endonuclease (HhaI) in vitro [34]. In their role as type I R-M system inhibitors, ArdB and ArdC presumably act to prevent pVA1 from being digested during plasmid transmission. Conversely, if their anti-restriction activity can be blocked, the transferred pVA1 plasmid should then become susceptible to attack by the bacterium's type I R-M system. This is an issue that deserves further investigation.

Secretion System (ORF3, 86, 89, and 90)
In addition to the type IV secretion system mentioned above (i.e., ORF11, VirB1-like gene), four other ORFs (ORF3, 86, 89, and 90) belonging to type II and III secretion systems are also found on the pVA1 plasmid. V. parahaemolyticus has been reported to contain not only two type-III secretion systems (T3SS1 and T3SS2) and two type-VI secretion systems (T6SS1 and T6SS2) [63,64], but also two T2SSs and two T2/4SSs [15]. These secretion systems are all related to the virulence of V. parahaemolyticus. For example, T3SS1 contributes to the secretion of effector proteins that are cytotoxic to HeLa cells, and T3SS2 is related to the enterotoxicity in the rabbit ileal loop test [65]. T6SS1 has been reported as an antibacterial system and is associated with the AHPND-causative strains [66]. In addition, T6SSs also contribute to the adhesion of V. parahaemolyticus to host cells [67], while T2SS is commonly employed by Gram-negative bacteria to transport a variety of molecular cargos [68]. Since PirA vp and PirB vp are both secreted toxins [5], the pVA1 secretion systems may play an important role during toxin release.
The secretion systems of drug-resistant bacteria were considered as the drug targets [69,70]. For example, Acinetobacter baumannii T2SS is responsible for the secretion of multiple enzymes, and its inactivation reduces the in vivo fitness of A. baumannii, as well as increasing its sensitivity to the human complementary system [70]. Recently, several small molecules, such as salicylidene acyl hydrazides and N-Hydroxybenzimidazoles, have been confirmed as T3SS inhibitors [69]. Moreover, high-throughput screening (HTS) based on secreted lipase activity was also developed to identify small molecule inhibitors of the T2SS [70]. Similarly, a search for inhibitors of the secretion systems of AHPND-causing V. parahaemolyticus might also identify molecules with the potential for controlling the damage caused by PirAB vp toxins.

pndA (ORF7)
A bacterial toxin-antitoxin system is a class of the plasmid addiction system (PAS) or postsegregational killing (PSK) system that is used to prevent the loss of plasmids from bacterial populations. For examples, in the hok-sok system of R1 plasmid, the srnB-srnC system of F plasmid and the pndA-pndB system of R438 plasmid, highly stable "killer" mRNAs that encode toxin proteins (Hok, SrnB, and PndA) are synthesized while small, unstable antisense RNAs (sok, srnC and pndB) are also produced. When coupled with host RNase II, these antisense RNAs facilitate the degradation of the killer mRNAs, and suppress the expression of toxin proteins [35]. However, after cell segregation, the production of antisense RNAs in the plasmid-free cells ceases, and due to their low stability, the levels of antisense RNAs rapidly decreases. Meanwhile, the killer mRNAs are still present. This results in the production of the encoded toxin proteins, each of which contains a single transmembrane domain, which is used to kill the plasmid-free cells from the inside [71,72]. The presence of a pndA gene (ORF7) in the pVA1 plasmid suggests that the PSK system is being used to ensure the presence of pVA1 in daughter cells. However, to date, the antisense RNA sequences for the pndA gene (i.e., pndB) have not been determined. Clearly, this will need to be further investigated to ascertain whether or not AHPND-causing bacteria actually include a viable pVA1-derived PSK system.

DNA Methyltransferase (ORF47 and ORF56)
DNA methylation is catalyzed by DNA methyltransferases (MTases, or methylases), which can transfer the methyl group from S-adenosyl methionine to the C-5 or N-4 positions of cytosine, or the N-6 position of adenine [73]. In bacteria, it is well known that DNA methylation protects the host's own DNA from degradation by restriction-modification systems that are used to digest foreign DNA such as plasmids, transposons, and viral DNA [74,75]. In these cases, each MTase has its cognate restriction enzyme. However, some MTases (e.g., Dam and Dcm) do not have corresponding restriction enzymes, and instead of participating in restriction-modification systems, they are involved in the regulation of gene expression and virulence (for a review, please see [76]). ORF47 on the pVA1 plasmid is predicted to be an N-6 DNA MTase that is similar to Dam. Dam is essential for Vibrio cholerae virulence [36]. It also regulates the finP gene on the pSLT virulence plasmid of Salmonella and affects the formation of F-type pilus that is responsible for plasmid transfer via conjugation [77]. It should, therefore, be worth investigating whether this predicted N-6 DNA MTase of pVA1 works like Dam, and further affects the virulence of the plasmid-harboring bacteria.

Conclusion Remarks
By incorporating the pVA1 plasmid, Vibrio parahaemolyticus becomes an AHPND-causing pathogen that is fatal to shrimps. Although the major virulence of the pVA1 plasmid is due to the PirA vp /PirB vp toxins, there are still~40 important ORFs with functional annotations ( Table 1). The ORFs discussed here are involved in toxin secretion, gene transposition, plasmid maintenance, bacterial conjugation, and a post-segregational killing (PSK) system. As discussed above, some gene products, like the conjugation factor and the secretion system, could be potential drug targets for combatting AHPND. Other genes, such as those for anti-restriction proteins, should also be worth investigating for their functional roles in the spread of the AHPND-causing plasmid among Vibrio spps. Taken together, we hope the information reviewed here will be useful for suggesting new research avenues for increasing our understanding of the mechanism of this disease.