Dissecting the Arginine and Lysine Biosynthetic Pathways and Their Relationship in Haloarchaeon Natrinema gari J7-2 via Endogenous CRISPR-Cas System-Based Genome Editing

ABSTRACT The evolutionary relationship between arginine and lysine biosynthetic pathways has been well established in bacteria and hyperthermophilic archaea but remains largely unknown in haloarchaea. Here, the endogenous CRISPR-Cas system was harnessed to edit arginine and lysine biosynthesis-related genes in the haloarchaeon Natrinema gari J7-2. The ΔargW, ΔargX, ΔargB, and ΔargD mutant strains display an arginine auxotrophic phenotype, while the ΔdapB mutant shows a lysine auxotrophic phenotype, suggesting that strain J7-2 utilizes the ArgW-mediated pathway and the diaminopimelate (DAP) pathway to synthesize arginine and lysine, respectively. Unlike the ArgD in Escherichia coli acting as a bifunctional aminotransferase in both the arginine biosynthesis pathway and the DAP pathway, the ArgD in strain J7-2 participates only in arginine biosynthesis. Meanwhile, in strain J7-2, the function of argB cannot be compensated for by its evolutionary counterpart ask in the DAP pathway. Moreover, strain J7-2 cannot utilize α-aminoadipate (AAA) to synthesize lysine via the ArgW-mediated pathway, in contrast to hyperthermophilic archaea that employ a bifunctional LysW-mediated pathway to synthesize arginine (or ornithine) and lysine from glutamate and AAA, respectively. Additionally, the replacement of a 5-amino-acid signature motif responsible for substrate specificity of strain J7-2 ArgX with that of its hyperthermophilic archaeal homologs cannot endow the ΔdapB mutant with the ability to biosynthesize lysine from AAA. The in vitro analysis shows that strain J7-2 ArgX acts on glutamate rather than AAA. These results suggest that the arginine and lysine biosynthetic pathways of strain J7-2 are highly specialized during evolution. IMPORTANCE Due to their roles in amino acid metabolism and close evolutionary relationship, arginine and lysine biosynthetic pathways represent interesting models for probing functional specialization of metabolic routes. The current knowledge with respect to arginine and lysine biosynthesis is limited for haloarchaea compared to that for bacteria and hyperthermophilic archaea. Our results demonstrate that the haloarchaeon Natrinema gari J7-2 employs the ArgW-mediated pathway and the DAP pathway for arginine and lysine biosynthesis, respectively, and the two pathways are functionally independent of each other; meanwhile, ArgX is a key determinant of substrate specificity of the ArgW-mediated pathway in strain J7-2. This study provides new clues about haloarchaeal amino acid metabolism and confirms the convenience and efficiency of endogenous CRISPR-Cas system-based genome editing in haloarchaea.

hyperthermophilic archaea (11,12). It is of great interest to dissect the functions of genes involved in arginine and lysine biosynthesis of haloarchaea and their possible evolutionary relationships.
The CRISPR-Cas (clustered regularly interspaced short palindromic repeats with Cas) system is a unique RNA-guided acquired immune system in prokaryotes, and has been widely used in genome editing and gene function analysis in diverse organisms (23). However, heterologous CRISPR-Cas systems have the disadvantages of being toxic to some bacterial cells and low activity in certain hosts, particularly extremophiles (24,25). An alternative strategy for genome editing is to employ endogenous CRISPR-Cas systems, which are encoded by most archaeal genomes and approximately half of bacterial genomes (26). Endogenous CRISPR-Cas system-based genome editing was first established in Streptococcus pneumoniae (27) and has been successfully applied in many bacteria and archaea (28)(29)(30)(31)(32)(33). For haloarchaea, the use of endogenous type I-B CRISPR-Cas system for genome editing in Haloferax volcanii is hindered by its high tolerance to CRISPR-mediated self-targeting, likely because the high copy numbers of chromosome in this haloarchaeon mediate accurate genome repair via homologous recombination (34). In contrast, the native type I-B CRISPR-Cas system of the polyploid haloarchaeon Har. hispanica has been successfully harnessed for genome editing, highlighting the convenience and efficiency of endogenous CRISPR-Cas system-based genome editing in haloarchaea (35).
Nnm. gari J7-2 was isolated from a salt mine in China (36), and three extracellular subtilisin-like proteases (SptA, SptC, and SptE) of strain J7-2 have been characterized (37)(38)(39)(40). Gene deletion analysis shows that the major extracellular protease SptA is not necessary for cell survival but facilitates the growth of strain J7-2 by degrading protein substrates into peptides and amino acids, which serve as nutrients for living cells (41). Meanwhile, strain J7-2 is capable of growth on synthetic media without amino acid supplements, and the biosynthetic pathways for 20 amino acids have been predicted in its genome, representing an ideal model for functionally dissecting amino acid biosynthesis pathways in haloarchaea (20). The strain J7-2 genome also contains three CRISPR loci (CRISPR1, CRISPR2, and CRISPR3), and the CRISPR1 locus is preceded by a full set of cas genes of the type I-B CRISPR-Cas system (20). Previously, a chemical FIG 1 Lysine and arginine biosynthetic pathways in prokaryotes. The four subpathways of the DAP pathway are indicated (numbered 1 to 4). The proteins sharing the same color (cyan, red, blue, green, brown, or purple) in different pathways are homologs. mutagenesis of strain J7-2 generated arginine and lysine auxotrophs with mutations in argC and dapD, respectively, to develop a genetic manipulation system for this haloarchaeon (18). In this study, an endogenous CRISPR-Cas system-based genome editing method for strain J7-2 was established and used for dissection of the biosynthetic pathways of arginine and lysine, focusing on the functions of ArgW and ArgX in arginine biosynthetic pathway of haloarchaea. The possible evolutionary relationship between arginine and lysine biosynthetic pathways in haloarchaea is discussed.

RESULTS
Harnessing the endogenous CRISPR-Cas system for genome editing in strain J7-2. The type I-B CRISPR-Cas system of strain J7-2 contains eight cas genes (cas1 to cas8), and the CRISPR1 array is comprised of 37 repeats and 36 spacers serving as the main repository of guiding RNAs ( Fig. 2A). To validate the function of the CRISPR-Cas system of strain J7-2, a plasmid challenge assay was performed by using invader plasmids carrying a spacer-matching artificial protospacer and a protospacer adjacent motif (PAM) (Fig. 2B). The trinucleotide TTC, a functional PAM verified in Hfx. volcanii (42) and Har. hispanica (43), was used as the PAM, and three spacers of the CRISPR1 array (spacer1-36, spacer1-18, and spacer1-1) of strain J7-2 were chosen as the protospacers to construct invader plasmids p1-36, p1-18, and p1-1, respectively, using the plasmid pYC-SHSmcs (41,44) (here, abbreviated pSHS) as a vector (Fig. 2B). Subsequent transformation of strain J7-2 by the three invader plasmids resulted in an ;100-fold decrease in transformation rate compared with that by the parental plasmid pSHS (Fig. 2C). It was suggested that only those sequences that led to an at least 100-fold decrease in transformation rates in the plasmid challenge assay could be regarded as a functional PAM, because transformation rates cannot be determined very accurately (42,45,46). In this context, TTC is a functional PAM and the type I-B CRISPR-Cas system is active in strain J7-2.
Besides TTC, additional 14 trinucleotides were also tested for their potential as a functional PAM, with spacer1-36 being used as a protospacer in the constructed invader plasmids. The plasmid challenge assay showed that, among the 14 trinucleotides only CTC and TTT triggered a decrease in transformation rate by more than 100-fold FIG 2 Challenging the CRISPR-Cas system of strain J7-2 with invader plasmids. (A) Schematic representation of the type I-B CRISPR-Cas system of strain J7-2. The Cas genes (cas1 to cas8) and CRISPR1 are drawn to scale as arrows. The accession numbers of the Cas genes are shown below the arrows. Rectangles and diamonds represent the spacers and direct repeats, respectively. (B) Schematic representation of the invader plasmid. The invader DNA fragment comprising the PAM and protospacer was inserted into the vector pSHS to construct the invader plasmid for transformation of strain J7-2. (C) Transformation efficiencies of strain J7-2 by invader plasmids carrying the same PAM (TTC) and different protospacers matching spacer1-36, spacer1-18, and spacer1-1 of CRISPR1, respectively; (D) Transformation efficiencies of invader plasmids carrying the same protospacer (p1-36) and different trinucleotides as PAMs. The values are expressed as means 6 standard deviation (SD) from three independent experiments (***, P , 0.001; **, P , 0.01; *, P , 0.05; n.s., no significance; calculated by Student's t test) (C and D), and the raw data are listed in Table S4. ( Fig. 2D) compared with pSHS without PAM and protospacer (Fig. 2C), indicating that CTC and TTT are also functional PAMs for strain J7-2.
The gene crtB, which encodes a phytoene synthase required for biosynthesis of carotenoids (e.g., b-carotene and bacterioruberin) in haloarchaea (47), has been used as the target gene to establish the endogenous CRISPR-Cas system-mediated gene knockout technique for Har. hispanica, and the DcrtB deletion mutant (white) could be easily distinguished from the wild type (red) based on colony color phenotype (34). Accordingly, crtB (NJ7G_1607) was selected as the target gene to develop the endogenous CRISPR-Cas system-based gene editing technique in strain J7-2.
An artificial mini-CRISPR, comprising two repeats separated by a spacer matching a PAM (TTC)-preceded sequence (protospacer) within crtB, as well as a donor DNA fragment were inserted into pSHS, yielding the self-targeting plasmid pKC; the mini-CRISPR is under the control of the sptA promoter (48) (Fig. 3A). Similar to the case of Har. hispanica (34), the transformation of strain J7-2 by pKC led to an ;10-fold decrease in transformation rate compared with that by pSHS (Fig. 3B), reflecting the cytotoxicity of the self-targeting CRISPR. Among 462 pKC transformants, 42.2% and 57.8% formed white and red colonies, respectively (Fig. 3C). By PCR analysis using the primers co-f and co-r (Fig. 3A), all of the 10 randomly selected white colonies showed the expected ;1-kb amplicon (Fig. 3D), suggesting that crtB has been deleted in these strains. Colony 4 ( Fig. 3D) was chosen for curing of the self-targeting plasmid, and its subculture was subjected to PCR analysis using the primer pairs ci-f/ci-r and s1-f/s1-r, matching target sequences within crtB in the strain J7-2 genome and the plasmid pKC, respectively (Fig. 3A), showing that neither crtB nor pKC is present in this strain (DcrtB mutant). The lawn color phenotype (Fig. 3E) and DNA sequencing result (Fig. 3F) also confirmed that crtB has been deleted from the genome in The artificial mini-CRISPR on the plasmid pKC contains two repeats (green diamonds) separated by a spacer (red rectangle) matching a PAM (TTC [in blue])-preceded protospacer within crtB on strain J7-2 genome. The mini-CRISPR is driven by the sptA promoter (P sptA ). The donor (U1 and D1) and the positions of the primers (co-f, co-r, ci-f, ci-r, s1-f, and si-r) used for PCR analysis are shown. (B) Comparison of the transformation rates of strain J7-2 by plasmids pSHS and pKC. The transformants were grown on 18% MGM plates containing 5 mg/mL mevinolin, and the values are expressed as means 6 SD from three independent experiments (***, P , 0.001; calculated by Student's t test). The raw data are listed in Table S5 the DcrtB mutant. These results indicate that the endogenous CRISPR-Cas system could be harnessed for genome editing in strain J7-2.
Genes and enzymes involved in arginine and lysine biosynthesis in strain J7-2. Strain J7-2 contains a full set of predicted genes involved in arginine biosynthesis (Fig. 4A); the products of argW, argX, argB, argC, argD, and argE are predicted to convert glutamate to ornithine, and those of argF, argG, and argH are predicted to convert ornithine to arginine (Fig. 4B). We noticed that ArgW, ArgX, ArgB, ArgC, ArgD, and ArgE of strain J7-2 share high sequence identities not only with those of the arginine biosynthetic pathway in other haloarchaea such as Har. hispanica (;63 to 78%) but also with the lysine biosynthesis enzymes of T. thermophilus (;36 to 54%); meanwhile, these enzymes are homologous to the bifunctional enzymes of S. acidocaldarius and T. kodakarensis involved in biosynthesis of both lysine and ornithine, with sequence identities of ;25 to 53% ( Table 1). The LysW of T. thermophilus (TtLysW), S. acidocaldarius (SaLysW), or T. kodakarensis (TkLysW) possesses four or three Zn 21 -binding Cys residues and a EDWGE motif containing the C-terminal Glu residue to covalently link the amino group of AAA or glutamate (6,11,12,49), and these crucial residues are conserved in the Nnm. gari J7-2 ArgW (NgArgW) (Fig. 4C). The six residues of the active site in LysX and ArgX of T. thermophilus (TtLysX), S. acidocaldarius (SaLysX and SaArgX), and T. kodakarensis (TkLysX) are conserved in the strain J7-2 ArgX (NgArgX) (Fig. 4D). NgArgX also possesses a GSWGR motif that recognizes the EDWGE motif of LysW/ ArgW proteins (11, 12) ( Fig. 4C and D). Additionally, the predicted overall structures of NgArgW and NgArgX, generated by either RoseTTAFold (50) or SWISS-MODEL (51), are well superimposed with the crystal structures of their homologs (TkLysW and TkLysX) of T. kodakarensis (12) (Fig. 4E). These pieces of evidence raise the possibility that the predicted arginine biosynthetic pathway of strain J7-2 may be a bifunctional route for biosynthesis of not only arginine but also lysine via the AAA pathway, but the strain J7-2 genome lacks the genes encoding the enzymes (LysS, LysTU, homoisocitrate dehydrogenase, and LysN) required for de novo biosynthesis of AAA from 2-oxoglutarate ( Fig. 4B). Additionally, in LysX/ArgX proteins a 5-amino-acid signature motif and two highly conserved residues (Arg and Ala) are involved in substrate recognition (12), while the 5-amino-acid signature motif of NgArgX is different from those of SaArgX, SaLysX, TtLysX, and TkLysX ( Fig. 4D) but is highly conserved in haloarchaeal ArgX proteins (see Fig. S1A in the supplemental material).
By using the crystal structure of the LysW-ADP-glutamate-bound complex of ArgX from Sulfolobus tokodaii (StArgX) (11) as the template, the homology modeling of NgArgX shows that four of the six hydrogen bonds formed between glutamate and the substrate recognition residues of StArgX are conserved in the glutamate-binding model of NgArgX (Fig. S1B). It seems that the substrate recognition mechanism of NgArgX is similar to that of StArgX, although the 5-amino-acid signature motifs of the two enzymes are different.
In strain J7-2, a gene cluster (dapA, dapB, dapD, hyp4048, lysA, dapF, and dapE) and two genes (ask and asd) at other loci are predicted to constitute the DAP pathway for lysine biosynthesis (Fig. 4A), although the gene encoding DapC is missing (Fig. 4B). These genes, except dapC and hyp4048, are conserved in other haloarchaea such as Hfx. volcanii and Har. hispanica (Fig. 4A). It is known that the products of four genes (ask, asd, dapC, and dapE) of the DAP pathway are evolutionarily related to the arginine biosynthetic enzymes encoded by argB, argC, argD, and argE, respectively (1, 3). In strain J7-2, however, Ask, Asd, and DapE share only low sequence identities (15.3 to 21.7%) with their counterparts (ArgB, ArgC, and ArgE) in the arginine biosynthetic pathway (Fig. 4B), reflecting a high evolutionary divergence between the two sets of enzymes.
argW and argX are essential for biosynthesis of arginine rather than lysine in strain J7-2. The endogenous CRISPR-Cas system-based genome editing method was employed to delete argW and argX of strain J7-2 simultaneously or separately, generating the DargWX, DargX, and DargW mutant strains. To probe the role of the C-terminal residue Glu 54 of NgArgW, we constructed the plasmid pWE54A, carrying a spacer matching a CTC-preceded protospacer at the 39 end of argW and a donor DNA with a codon mutation leading to the replacement of Glu 54 by Ala (E54A), and used it to construct the argW(E54A) mutant strain (Fig. 5A). All of the four mutant strains were verified by DNA sequencing (Fig. 5B). In contrast to strain J7-2, the four mutants were unable to grow on the synthetic medium (SM) plate without amino acid supplements or supplemented only with lysine, but grew when supplemented with arginine ( Fig. 5C). By using pSHS as the vector, we constructed the complementary plasmid expressing either or both NgArgW and NgArgX driven by the promoter (P 0566 ) of a high-abundance blue copper domain protein (NJ7G_0566) of strain J7-2 (52) (Fig. 5E, bottom). Similar to strain J7-2, the DargWX/ArgWX, DargX/ArgX, DargW/ArgW, and argW(E54A)/ArgW complementary strains were able to grow on SM plates either with or without arginine or lysine (Fig. 5C). In liquid SM, the DargWX/ArgWX complementary strain rather than the DargWX mutant harboring the blank vector pSHS could grow and show a similar growth profile to strain J7-2 carrying pSHS under the same cultivation condition (Fig. 5D). These results suggest that argW and argX are indispensable for biosynthesis of arginine rather than lysine and that the C-terminal residue Glu 54 of NgArgW is crucial for arginine biosynthesis in strain J7-2.
The results of anti-His tag immunoblot analyses showed that recombinant NgArgX with a C-terminal His tag could be detected in the cell extracts of DargWX/ArgWX and DargX/ArgX strains grown in liquid SM; however, NgArgW was not detected in the extract of the DargW/ArgW strain (Fig. 5E), probably due to a low expression level of this protein.
Subsequently, the cell extract of the DargW/ArgW strain was subjected to nickel-nitrilotriacetic acid (Ni-NTA) affinity chromatography to concentrate the recombinant NgArgW with an N-terminal His tag for immunoblot analysis, revealing multiple bands with apparent molecular weights (AMWs [;25 to 40 kDa]) much higher than predicted molecular weight of recombinant NgArgW (6.7 kDa) (Fig. 5E). Considering that NgArgW contains four Cys residues, we reasoned that the observed higher-AMW forms of NgArgW may have resulted from the formation of intermolecular disulfide bonds during the electrophoresis. To test this possibility, recombinant NgArgW with a C-terminal His tag was produced in E. coli and purified by Ni-NTA affinity chromatography, followed by SDS-PAGE analysis in the presence of 8 M urea. Before treatment with b-mercaptoethanol (b-ME), purified NgArgW showed an AMW of ;14 kDa; after b-ME treatment, a portion of the 14-kDa NgArgW molecules were converted into monomeric forms (;7 kDa) (Fig. 5F). Besides the 14-kDa NgArgW, higher-AMW forms of NgArgW were also detected by immunoblot analysis (Fig. 5F), which suggests the formation of intermolecular disulfide bonds between two or more NgArgW molecules during the electrophoresis, even when the protein sample was pretreated with b-ME. In contrast to the 14-kDa NgArgW, the 7-kDa NgArgW monomer was not revealed by anti-His tag immunoblot analysis (Fig. 5F). When the blotted proteins were visualized by using enhanced chemiluminescence (ECL), the 7-kDa NgArgW monomer could be detected (Fig. 5F). These results confirm that recombinant NgArgW and NgArgX have been expressed to fulfil their functions in the complementary strains.
The arginine and lysine biosynthetic pathways are independent of each other in strain J7-2. To investigate whether the predicted DAP pathway is functional in strain J7-2 and to probe possible interconnection between the lysine and arginine biosynthetic pathways, four genes of the two pathways were chosen to construct the DdapB, Dhyp4048, DargB, and DargD mutants via the endogenous CRISPR-Cas system-based genome editing method, and the resulting mutants were confirmed by DNA sequencing (Fig. 6A).
The DdapB mutant was unable to grow on the SM plate without amino acid supplements but grew well with lysine supplement; in contrast, the DdapB/DapB complementary strain was able to grow on the SM plate in either the absence or presence of lysine (Fig. 6B). Meanwhile, recombinant DapB could be detected in DdapB/DapB by immunoblot analysis (Fig. 6D). In the DAP pathway gene cluster of strain J7-2, the gene hyp4048 encodes a hypothetical protein (NJ7G_4048) (Fig. 4A). However, the Dhyp4048 mutant grew as well as strain J7-2 on the SM plate without amino acid supplements (Fig. 6C). These results suggest that dapB but not hyp4048 is necessary for lysine biosynthesis via the DAP pathway in strain J7-2.
Despite their evolutionary relationships, the DAP pathway enzymes Ask, Asd, and DapE of strain J7-2 share only 15.3 to 21.7% sequence identities with their counterparts (ArgB, ArgC, and ArgE) in the arginine biosynthetic pathway (Fig. 4B), implying that the two sets of enzymes have become highly specialized for corresponding pathways during evolution. In support of this, the DargB mutant strain was unable to grow on the SM plate without arginine supplement; in contrast, the DargB/ArgB complementary strain grew well under the same condition (Fig. 6B), and recombinant ArgB was detected in the DargB/ArgB strain by immunoblot analysis (Fig. 6D). These results suggest that the loss of argB in DargB cannot be compensated for by its counterpart ask in the DAP pathway of strain J7-2.
Because dapC is missing in haloarchaea, it is assumed that haloarchaeal ArgD is synonymous with DapC and may act as not only an N-acetylornithine aminotransferase for arginine biosynthesis but also an N-succinyl-L, L-diaminopimelate aminotransferase for lysine biosynthesis (19). To test this possibility, the DargD mutant was analyzed for its amino acid requirements for growth. On SM plates, the DargD mutant could grow only in the presence of arginine but did not grow without amino acid supplements or with lysine supplement, while the DargD/ArgD complementary strain did not require amino acid supplements for growth (Fig. 6C). Immunoblot analysis also revealed the recombinant ArgD in the DargD/ArgD strain (Fig. 6D). These results indicate that ArgD participates only in arginine biosynthesis and that the function of the missing DapC is accomplished by a yet unknown aminotransferase in strain J7-2.
Although the strain J7-2 genome lacks the genes encoding the enzymes for de novo biosynthesis of AAA from 2-oxoglutarate (Fig. 4B), there is a possibility that strain J7-2 may utilize external AAA to synthesize lysine via the arginine biosynthetic pathway, of which the enzymes share high sequence identities with their counterparts of the LysW-mediated AAA pathways in T. thermophilus, S. acidocaldarius, and T. kodakarensis (Table 1). To test this possibility, the DdapB mutant strain, in which the DAP pathway has been blocked, was inoculated on SM plates supplemented with different concentrations of AAA, but no growth of this mutant was observed, whereas the DdapB/DapB complementary strain grew well under same conditions (Fig. 6E). Therefore, the arginine biosynthetic pathway of strain J7-2 cannot act as a bifunctional route for biosynthesis of both arginine and lysine, and lysine is synthesized only via the DAP pathway.
NgArgX is a key determinant of substrate specificity of the ArgW-mediated arginine biosynthetic pathway. It was reported that the 5-residue signature motif defines substrate specificity of LysX/ArgX proteins for glutamate and AAA, respectively (11, 12) (Fig. 4D). In order to investigate whether the substrate specificity of NgArgX could be altered and used for lysine biosynthesis from AAA, we constructed four variants (NgArgXa, NgArgXc, NgArgXd, and NgArgXf) of NgArgX by replacing some or all of the residues in the signature motif with the corresponding residues of monofunctional SaLysX or bifunctional TkLysX; an additional four residues adjacent to the signature motif were also replaced by corresponding residues of SaLysX, yielding the variants NgArgXb and NgArgXe (Fig. 6F). However, the DdapB strains carrying the expression plasmids for the six mutated NgArgX proteins, respectively, could grow on SM plates supplemented with lysine but not on that supplemented with AAA (Fig. 6G), although the six variants could be produced in the DdapB mutant grown in modified growth medium with 18% total salts (18% MGM) by immunoblot analysis (Fig. 6D). Moreover, all of the six variants could be produced in the DargX mutant grown in 18% MGM (Fig. 6D), while four of them (NgArgXa, NgArgXb, NgArgXc, and NgArgXd) could rescue the growth defect of the DargX mutant on SM plates without amino acid supplements (Fig. 6G), indicating that the substrate specificity of the four variants has not been altered by modification of the signature motif. Notably, the variants NgArgXe and NgArgXf could not rescue the growth defect of the DargX mutant on SM plates without amino acid supplements (Fig. 6G), suggesting that the mutations in the signature motif of the two variants not only are unable to endow them with the ability to act on AAA but also lead to the loss of their specificity for glutamate.
We next investigated whether NgArgX and its variant NgArgXa can catalyze the covalent linkage of glutamate and/or AAA with NgArgW in vitro. Recombinant NgArgW with an N-terminal His tag was purified from E. coli, while NgArgX and NgArgXa were expressed and partially purified from the DargWX mutant (Fig. S2). NgArgX or NgArgXa was mixed with NgArgW and the substrate glutamate or AAA, and after the reaction the mixture was treated with trypsin, followed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis. When glutamate was used as the substrate, a peptide corresponding to the NgArgW-derived C-terminal peptide 45 APELEEDWGE 54 plus Glu (1,303.54 Da) was identified in the reaction mixture containing either NgArgX or NgArgXa, and the covalent linkage of the substrate glutamate with the C-terminal residue Glu 54 of NgArgW was verified by LC-MS/MS analysis ( Fig. 7A and C). In contrast, when AAA was used as the substrate the peptide 45 APELEEDWGE 54 plus AAA (theoretical mass of 1,317.63 Da) was not identified in the reaction mixture containing either NgArgX or NgArgXa (Fig. 7B and D). These results suggest that both NgArgX and NgArgXa specifically act on glutamate rather than AAA.

DISCUSSION
Endogenous CRISPR-Cas system-based genome editing in Nnm. gari J7-2. In this study, the CRISPR-Cas system of Nnm. gari J7-2 has been verified to be functional and successfully harnessed for genome editing in this haloarchaeon. Because CRISPR-Cas systems are widely distributed in prokaryotes, endogenous CRISPR-Cas system-based genome editing is emerging as a new strategy for editing bacterial and archaeal genomes, but very few reports are available on successfully exploiting haloarchaeal CRISPR-Cas systems for genome editing. For instance, Hfx. volcanii is highly tolerant to self-targeting of the genome by its native CRISPR-Cas system, and exploiting endogenous CRISPR-Cas systembased genome editing in Hfx. volcanii is impeded, most likely due to an accurate genome repair via homologous recombination in this polyploid haloarchaeon (34). Nevertheless, endogenous CRISPR-Cas system-mediated gene repression can be achieved in Hfx. volcanii after deleting the genes encoding the nuclease Cas3 for target cleavage and the RNase Cas6 for pre-CRISPR RNA (pre-crRNA) processing (53). In contrast to the case of Hfx. volcanii, CRISPR-mediated self-targeting is highly cytotoxic in the polyploid haloarchaeon Har. hispanica, and endogenous CRISPR-Cas system-based genome editing could be readily achieved with a high efficiency (35). The discrepancy in tolerance to CRISPR-mediated selftargeting between Hfx. volcanii and Har. hispanica may derive from a possible difference in their chromosome copy numbers, DNA repair systems, and/or CRISPR-Cas systems (34,35). Differing from Hfx. volcanii but similar to Har. hispanica, Nnm. gari J7-2 is sensitive to CRISPR-mediated self-targeting, facilitating the establishment of endogenous CRISPR-Cas system-based genome editing in this haloarchaeon.
It was observed that when deleting crtB by the endogenous CRISPR-Cas systembased genome editing method, only 42.2% of strain J7-2 transformants formed white colonies, while nearly all the colonies of Har. hispanica transformants showed a white phenotype (35), implying that the gene-editing efficiency in strain J7-2 is lower than that in Har. hispanica. Besides a possible difference in the activities of CRISPR-Cas systems between the two haloarchaea, the copy number of the self-targeting plasmid may be important for the gene-editing efficiency. In this study, the plasmid pSHS, which is present at about 1 copy per chromosome in Nnm. gari cells (44), was employed to construct the self-targeting plasmid pKC containing the mini-CRISPR and the donor DNA. Due to the relatively low copy number of pKC, the amount of the donor DNA and the expression level of the self-targeting crRNA may not be high enough for highly efficient knockout of crtB in the chromosome. Nevertheless, the use of a low-copy self-targeting plasmid for gene editing has the advantage of an efficient plasmid curing, as evidenced by the finding that the DcrtB mutant lacking the self-targeting plasmid could be obtained after only two rounds of subculturing. In Har. hispanica, the endogenous CRISPR-Cas system-based genome editing system has been optimized in terms of the selection of PAM and the promoter of the mini-CRISPR, as well as the size of the donor DNA (35). Additionally, it was reported that the length of the spacer in the mini-CRISPR is related to the gene-editing efficiency of the endogenous CRISPR-Cas system in Clostridium tyrobutyricum (30). Future optimization of the corresponding elements of the self-targeting plasmid is expected to improve the endogenous CRISPR-Cas system-based genome editing efficiency in strain J7-2.
Arginine and lysine biosynthesis in Nnm. gari J7-2. By using the endogenous CRISPR-Cas system-based genome editing system, several genes predicted to be involved in the biosynthesis of arginine and lysine in strain J7-2 were genetically dissected. In contrast to the wild-type strain J7-2, the DargW, DargX, DargB, and DargD mutant strains display an arginine auxotrophic phenotype. Meanwhile, the mutagenesis of argC or argH in Hfx. volcanii or Nnm. gari J7-2 also generates an arginine auxotroph (18,21,22). These data confirm that the ArgW-mediated arginine biosynthetic pathway is functional and indispensable for arginine biosynthesis in haloarchaea. In bacteria, N-acetylglutamate synthase (ArgA) or ornithine acetyltransferase (ArgJ) is responsible for the acetylation of the a-amino group of glutamate to prevent intramolecular cyclization of intermediates during arginine biosynthesis (2). In hyperthermophilic archaea such as S. acidocaldarius and T. kodakarensis, the a-amino group of glutamate is covalently linked to the g -carboxyl group of the C-terminal Glu residue of the carrier protein SaLysW or TkLysW to avoid intramolecular cyclization (11,12). Similar to hyperthermophilic archaea, the haloarchaeon Nnm. gari J7-2 employs the carrier protein NgArgW to modify the a-amino group of glutamate. In support of this, the argW(E54A) mutant strain, in which the C-terminal residue Glu 54 of NgArgW is replaced by alanine, shows an arginine auxotrophic phenotype. It appears that the modification of the a-amino group of glutamate with a carrier protein rather than an acetyl group is a common strategy employed by archaea for arginine biosynthesis.
It was noticed that the recombinant NgArgW, which has four Cys residues conserved in LysW-like proteins, tends to form intermolecular disulfide bonds by SDS-PAGE analysis. However, because the cytoplasm is normally maintained in a reducing state due to the presence of antioxidants such as thioredoxins, glutaredoxins, and glutathione and its analogs (54,55), disulfide bond formation between NgArgW molecules is unlikely to occur in the cytoplasm. Moreover, in the crystal structure of TtLysW or TkLysW, a zinc atom is coordinated with the four Cys residues (12,49). Accordingly, the four Cys residues in NgArgW most likely participate in Zn 21 binding rather than form disulfide bonds in vivo. In contrast, during the SDS-PAGE analysis, the structure of NgArgW is disrupted, allowing the exposure of Cys residues to form disulfide bonds under oxidative conditions. Although resembling S. acidocaldarius and T. kodakarensis in employing the LysW/ ArgW-mediated pathway for arginine biosynthesis, strain J7-2 differs from the two hyperthermophilic archaea in that it synthesizes lysine via the DAP pathway rather than the AAA pathway, as evidenced by the finding that the DdapB mutant strain shows a lysine auxotrophic phenotype. Similar to the case of strain J7-2, the isotopic labeling study on Har. hispanica (16) and the mutagenesis of lysA in Hfx. volcanii (21) show that the two haloarchaea utilize the DAP pathway for lysine biosynthesis. While Har. hispanica synthesizes lysine via the dehydrogenase subpathway (16), Nnm. gari J7-2 employs the succinylase subpathway for lysine biosynthesis since the mutation of dapD specific for this subpathway leads to lysine auxotrophy in this haloarchaeon (18). Although the genes encoding DapC and diaminopimelate dehydrogenase (Ddh), specific for the succinylase and dehydrogenase subpathways, respectively, have not been identified yet, it is evident that the DAP pathway is commonly employed by haloarchaea for lysine biosynthesis.
In Gram-negative bacteria, the DAP pathway not only is responsible for lysine biosynthesis but also plays a central role in cell wall synthesis since DAP is an essential precursor for peptidoglycan biosynthesis (3). For those bacteria lacking the DAP pathway, such as T. thermophilus and Deinococcus radiodurans (56), their cell wall does not have DAP but contains ornithine, an intermediate of the arginine biosynthetic pathway (57,58). In the cases of S. acidocaldarius and T. kodakarensis, they also lack the DAP pathway but employ the LysW-mediated AAA pathway for lysine biosynthesis, and they have only a single S-layer as their cell wall (59,60). It has been proposed that the maintenance of the DAP pathway for lysine biosynthesis, at least in bacteria, might be correlated with the appearance of the extant structure of their cell wall (3). However, this proposal is not applicable to haloarchaea, which have the S-layer as the only component of their cell wall (61) but retain the DAP pathway for lysine biosynthesis. It is presently unclear whether the haloarchaeal DAP pathway has any function other than lysine biosynthesis. The possibility that the intermediates of the DAP pathway may be involved in some cellular processes in haloarchaea cannot be excluded.
The relationship between the arginine and lysine biosynthetic pathways in Nnm. gari J7-2. Despite the evolutionary relationship between the arginine and lysine biosynthetic pathways, our results suggest that the biosynthetic pathway for arginine and the DAP pathway for lysine are independent of each other in strain J7-2. First, the DargW, DargX, DargB, and DargD mutant strains show only an arginine auxotrophic phenotype, while the DdapB mutant strain shows only a lysine auxotrophic phenotype. Second, the loss of the gene argB, which is necessary for arginine biosynthesis, cannot be compensated for by its counterpart ask in the DAP pathway. Third, the enzyme ArgD, which is assumed to be synonymous with DapC of the DAP pathway in haloarchaea (19), is not involved in lysine biosynthesis in strain J7-2. Finally, the strain J7-2 genome lacks the genes for de novo biosynthesis of AAA, and the DAP pathway-blocked DdapB strain is unable to grow in the presence of AAA; the replacement of the 5-aminoacid signature motif responsible for substrate specificity of NgArgX by that of SaLysX or TkLysX cannot endow the DdapB mutant with the ability to biosynthesize lysine from AAA. In Hfx. volcanii, the mutagenesis of the genes involved in arginine (argB, argC, argD, and argH) and lysine (lysA) biosynthesis also generates arginine and lysine auxotrophs, respectively (21,22). In this context, the two sets of enzymes of the arginine and lysine biosynthetic pathways in haloarchaea have been highly specialized during evolution.
As in other haloarchaea, the gene dapC in the DAP pathway has not been identified in the strain J7-2 genome. Our results show that the ArgD of strain J7-2 participates only in biosynthesis of arginine rather than lysine. Similarly, the mutagenesis of argD in Hfx. volcanii generates an arginine auxotroph (21). These data suggest that haloarchaeal ArgD is a monofunctional enzyme specific for the arginine biosynthetic pathway and exclude the possibility that this enzyme is synonymous with DapC in haloarchaea (19). This is in contrast to the case of ArgD from E. coli, which acts as a bifunctional enzyme to catalyze the transamination reactions in both the arginine biosynthetic pathway and the DAP pathway (14). Given the fact that the succinylase subpathway is functional in strain J7-2, there should be a yet unknown aminotransferase gene to fulfil the function of the missing dapC; however, this unknown gene is likely to share a very low sequence identity with known dapC or may be a nonorthologous gene and remains to be identified in the future.
Determinant role of NgArgX in substrate specificity of the ArgW-mediated arginine biosynthetic pathway. Strain J7-2 utilizes the ArgW-mediated pathway to synthesize arginine but not lysine, although the enzymes of this pathway are homologs of those of the bifunctional LysW-mediated pathways for both lysine and arginine (or ornithine) in S. acidocaldarius and T. kodakarensis (11,12). Considering that SaLysW could act as a common carrier protein of glutamate and AAA to mediate the biosynthesis of both arginine and lysine in S. acidocaldarius (11), the carrier protein NgArgW is unlikely to be crucial for determination of the preference of the ArgW-mediated pathway for glutamate rather than AAA as the substrate in strain J7-2. A more likely scenario is that the catalytic enzyme or enzymes in the ArgW-mediated pathway of strain J7-2 have become highly specific for glutamate and its derivatives to biosynthesize arginine, wherein the first enzyme of the pathway, NgArgX, plays a key role in determining the pathway specificity. For LysX/ArgX proteins from hyperthermophilic archaea, the replacement of two or four residues in the signature motif of bifunctional TkLysX by monofunctional LysX-or ArgX-type residues increases substrate preference of the variant for AAA or glutamate, respectively (12): the replacement of two residues in the signature motif of SaLysX (Asn-Thr) with those of SaArgX (Gly-Phe) leads to substrate preference for glutamate over AAA, contrasting distinctly with the substrate specificity of wild-type SaLysX (11). In the case of NgArgX, the replacement of the two residues (Ala-Leu) in its signature motif by those of SaLysX (Asn-Thr), however, cannot shift the substrate preference of the variant NgArgXa from glutamate to AAA by in vitro analysis. Meanwhile, the expression of the NgArgX variants with a SaLysX-or TkLysX-like signature motif cannot endow the DdapB strain with the ability to biosynthesize lysine from AAA. Moreover, the variants NgArgXe and NgArgXf with the 5-residue signature motif of SaLysX and TkLysX, respectively, are unable to act on not only AAA but also glutamate. It is evident that NgArgX has been evolved to be highly specific for glutamate rather than AAA and is a key determinant of substrate specificity of the ArgW-mediated arginine biosynthetic pathway.
In summary, the endogenous CRISPR-Cas system in strain J7-2 has been harnessed for genome editing and used for functional analysis of the genes involved in arginine and lysine biosynthesis. The results showed that arginine and lysine are produced in strain J7-2 via the ArgW-mediated pathway and the DAP pathway, respectively, and the two pathways are independent of each other. To the best of our knowledge, this is the first report on the arginine biosynthetic pathway mediated by the ArgW/LysW system in haloarchaea. In contrast to the bifunctional LysW-mediated pathways for both lysine and arginine (or ornithine) in hyperthermophilic archaea, the ArgW-mediated pathway in strain J7-2 is a monofunctional pathway for arginine biosynthesis, wherein NgArgX is a key determinant of substrate specificity of this pathway. Given that arginine and lysine biosynthesis enzymes evolved from ancestral enzymes, haloarchaeal enzymes responsible for arginine and lysine biosynthesis have become functionally specialized during evolution.

MATERIALS AND METHODS
Strains and growth conditions. The strains used in this study are listed in Table S1 in the supplemental material. Natrinema gari J7-2 (CCTCC [AB91141]) and its derivatives were grown at 37°C in 18% MGM supplemented with mevinolin (5 mg/mL) when necessary, as described previously (37). The Hv-Min medium (62) (63). E. coli DH5a and E. coli JM110 were used as hosts for plasmid construction and demethylation, respectively, and E. coli BL21(DE3) was used as the expression host; they were grown at 37°C in Luria-Bertani (LB) medium supplemented with ampicillin (100 mg/mL) as needed. Solid medium was prepared by adding 1.5% (wt/vol) agar.
Plasmid construction. The plasmids and primers used in this study are listed in Tables S2 and S3, respectively. Plasmids for invader challenging assays were constructed according to the method of Li et al. (43). Briefly, a sticky fragment comprising a designed PAM sequence and a subsequent spacer-matching protospacer was generated by annealing two different-sized primers, and then inserted into the BamHI-XbaI site of the vector pSHS (see Fig. S3 in the supplemental material) (41,44). Plasmids for genome editing were constructed by inserting a mini-CRISPR preceded by the sptA core promoter (41) and a donor DNA into the BamHI-XbaI site of pSHS using the Sangon Biotech Ready-to-use seamless cloning kit (Sangon Biotech, Shanghai, China) according to the manufacturer's protocol. The promoter-preceded mini-CRISPR was synthesized by PCR using a set of partially overlapping primers (Fig. S4). The upstream (U) and downstream (D) homologous arms of the target sequence were separately amplified from the strain J7-2 genome and connected by overlapping PCR, yielding the donor DNA. In the genome-editing plasmid for introduction of point mutations into the strain J7-2 genome, the donor DNA contained a codon mutation. To construct complementary plasmids, the target gene and the promoter of the gene encoding a high-abundance blue copper domain protein (NJ7G_0566) (52) were separately amplified from the strain J7-2 genome, connected by overlapping PCR, and then inserted into the BamHI-XbaI site of pSHS. The E. coli expression plasmid was constructed by inserting the target gene into the BamHI-NcoI site of pET15b. The sequences of all recombinant plasmids were verified by DNA sequencing.
Plasmid challenge assay. The plasmids were demethylated in E. coli JM110 and then transferred into strain J7-2 using the polyethylene glycol (PEG)-mediated transformation method according to the Halohandbook online protocol (https://haloarchaea.com/wp-content/uploads/2018/10/Halohandbook _2009_v7.3mds.pdf). The transformants were plated on 18% MGM agar plates containing 5 mg/mL mevinolin, and the plates were incubated at 37°C for 5 days. Colonies were counted and expressed as CFU.
Construction and screening of the mutants of Nnm. gari J7-2. After transformation of strain J7-2 with the demethylated genome-editing plasmid, the transformants were plated and grown on 18% MGM agar plates. For screening of the target mutant, randomly selected colonies were cultivated in liquid 18% MGM; cells were recovered by centrifugation and lysed with double-distilled water (ddH 2 O), followed by PCR analysis with external and internal primer pairs of the target gene. For curing of plasmid, the PCR-verified mutant was subjected to 2 to 3 rounds of subculturing (2 days for each round) in liquid 18% 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.
Recombinant protein expression and purification. E. coli BL21(DE3) cells harboring an expression plasmid for NgArgW were cultivated in LB medium at 37°C until the optical density at 600 nm (OD 600 ) reached ;0.7. Production of recombinant NgArgW was induced by the addition of 0.4 mM isopropyl-b-D-thiogalactopyranoside (IPTG), and cultivation continued at 37°C for 4 h. Cells were suspended in 20 mM HEPES-NaOH (pH 7.5) containing 0.5 M NaCl and disrupted by sonication on ice. The soluble fraction was recovered by centrifugation at 13,000 Â g for 10 min at 4°C and then subjected to Ni-NTA affinity chromatography on an Ni 21 -charged Chelating Sepharose Fast Flow column (GE Healthcare Bio-Sciences) to purify the Histagged recombinant protein. The DargWX strain carrying an expression plasmid for NgArgX or its variant NgArgXa was grown at 37°C in 18% MGM. The mid-log-phase cells were recovered by centrifugation and suspended in 20 mM HEPES-NaOH (pH 7.5) containing 3.0 M KCl, followed by sonication on ice. The Histagged recombinant proteins were purified by Ni-NTA affinity chromatography in the presence of 3.0 M KCl. The concentration of purified samples was determined using the Bradford method with bovine serum albumin as a standard. The amount of target protein in the partially purified sample was further assessed by band intensity measurement following SDS-PAGE.
SDS-PAGE and immunoblot analyses. SDS-PAGE was performed with 12% polyacrylamide gel using a Tris-glycine or Tris-tricine buffer system. Protein samples were precipitated with 20% (wt/vol) trichloroacetic acid (TCA) and washed with acetone before being subjected to SDS-PAGE analysis. In some cases, 8 M urea was included in the loading buffer and the gel for urea-SDS-PAGE. The anti-His tag monoclonal antibody (1:10,000) (Novagen) and the alkaline phosphatase (AP)-conjugated goat anti-mouse IgG secondary antibody (1:5,000) (Abbkine, China) were used for immunoblot analysis, and the immunoreactive proteins were visualized using the BCIP/NBT (5-bromo-4-chloro-3-indolylphosphate-nitroblue tetrazolium) chromogen kit (Solarbio, Beijing, China) as described previously (64). In some cases, horseradish peroxidase (HRP)conjugated goat anti-mouse IgG was used as the secondary antibody (1:5,000) (Sangon Biotech, China), and the signals were detected using Pierce ECL Western blotting substrate (Thermo Fisher Scientific, USA). The images were photographed using the GE Amersham Imager AI680 with an exposure time of 1 or 5 min.
In vitro activity assay for NgArgX and its variant. The activity of NgArgX or its variant to catalyze the covalent linkage of glutamate or AAA to NgArgW was determined according to the method described by Ouchi et al. (11), with modifications. Briefly, the reaction was carried out at 45°C for 12 h in a mixture containing 10 mg/mL NgArgX or its variant, NgArgW, 10 mM glutamate or AAA, 20 mM ATP, 1 mM MgCl 2 , 0.1 mM ZnSO 4 , and 3 M KCl in 100 mM HEPES (pH 7.5). The reaction mixture was subjected to in-solution digestion by trypsin as described previously (52). The trypsin-digested peptides were determined by nano-LC-MS/MS using an Easy-NanoLC system coupled online with the Q Exactive-HF mass spectrometer (Thermo Scientific, San Jose, CA). The peptide sequences were identified using Proteome Discoverer 2.5 software (Thermo Scientific) to determine the chemical structure of the reaction product.
Protein structure prediction. The structural models of NgArgW and NgArgX were predicted by RoseTTAFold (50) or SWISS-MODEL (51). Chimera software (65) was used for visualization of the predicted structure.
Data availability. The genome sequence of Natrinema gari J7-2 has been deposited in GenBank under accession no. CP003412.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 1.2 MB.

ACKNOWLEDGMENTS
We thank Zhangying Wang (Wuhan University, China) for technical help, and we thank Xiaolu Zhao and Wenda Huang (Wuhan University, China) for help with LC-MS/MS analysis.