S. islandicus M.16.4 and its derivatives described in Table 1, were grown in tissue culture flasks (Falcon) at 73°C to 76°C without shaking. DY (dextrin-tryptone) media (pH 3.45) was used in all cases and contained the following components (per 1 L of Milli-Q H₂O): basal salts (K₂SO₄, 3.0 g; NaH₂PO₄, 0.5 g; MgSO₄, 0.145 g; CaCl₂ ⋅ 2H₂O, 0.1 g), 20 μl trace mineral stock solution (3.0% FeCl₃, 0.5% CoCl₂ ⋅ 6H₂O, 0.5% MnCl₂ ⋅ 4H₂O, 0.5% ZnCl₂, and 0.5% CuCl₂ ⋅ 2H₂O), 0.2% (wt/vol) dextrin, and 0.1% (wt/vol) tryptone. Plate media was prepared by pre-warming 2x DY media, supplemented with 20 mM of MgSO₄ and 7 mM CaCl₂ ⋅ H₂O, and mixing an equal volume of a freshly boiling 1.7% (wt/vol) Gelrite solution. To grow uracil and agmatine auxotrophic strains, DY media was supplemented with a final concentration of 20 μg/ml of uracil and 50 μg/ml of agmatine. For 5-FOA counterselections, a final concentration of 50 μg/ml was also added to DY plate media. For general molecular cloning manipulations Escherichia coli (NEB 5-alpha competent E. coli) was utilized on LB media. Ampicillin (100 μg/ml) was added to media when required.

The mating assays for S. islandicus strains were performed through selection of agmatine prototrophy and 5-FOA resistance. Two genotypes are needed in order to perform the mating assay: a strain with functional lacS, argD, and pyrEF loci, and a strain with deletions in these three markers (Figure 1).

In all mating assays, both partner strains were grown to a mid-log phase, as detailed above, in either DY media or DY media supplemented with uracil and agmatine. Once strains reached their mid-log phase, their optical density (600 nm) was measured and equalized using warm DY media to an OD₆₀₀ of 0.3. The two strains were then mixed in equal parts and incubated at room temperature for 15 min. The co-incubates were then incubated under shaking conditions (180 rpm) at 76°C overnight. Following incubation, the mixed cultures were washed to remove agmatine using centrifugation and equalized to an OD₆₀₀ of 0.5. Finally, cultures underwent serial dilutions and were plated in either selective DY media containing 5-FOA and uracil (10⁰ and 10^–1 dilutions) or non-selective DY media containing agmatine and uracil (10^–4 and 10^–5 dilutions). After cells are plated, plates are incubated for 7–14 days at 76°C. Once colonies are observed on plates, an X-gal solution (2 mg/ml) was sprayed on both selective and control plates to determine original background. After the genotype is confirmed through PCR, the colonies are counted and analyzed to determine the recombination frequency of both the tester strain and either the wild-type or mutants with all three genetic markers.

Recombinant frequencies are calculated through the following formula: recombinantfrequency=recombinantsG1controlG1, where recombinantsG1 represents the cfu/ml of colonies on selective media of a particular genotype. ControlG1 represents the cfu/ml of colonies counted on non-selective media of the same genotype, G1. Colonies are verified through PCR amplification using primers: argD_chk-F/R & M16_pyrEFII-F/R (Supplementary Table 1).

Gene knockout shuttle vectors were constructed using NEBuilder Hifi assembly kit (NEB, USA) and according to manufacturer’s protocol. IDT G-Blocks were synthesized for the traG spacer and recombination template. The shuttle vectors were cloned in NEB 5-alpha competent E. coli (High Efficiency; NEB, USA) cells and plated on LB with ampicillin (100 μg/ml). Shuttle vectors were subsequently isolated using the QIAprep Spin Miniprep Kit (Qiagen, USA) as per manufacturer’s instructions.

Once knockout plasmids were isolated, plasmids were electroporated into competent RJW004 cells and incubated in incubation buffer for 1 hr (Zhang et al., 2013a). Transformations were then plated onto DY plates and incubated for 10–14 days. After colonies appeared on media, X-gal staining and PCR amplification was performed on colonies to confirm vector presence in the strain. Once colonies were confirmed, they were grown in liquid DY media to mid-log phase.

These strains were subsequently plated onto counter-selective DY media containing uracil, agmatine, and 5-FOA at concentrations mentioned above. Plates were incubated for 10–14 days. After incubation, colonies were stained with X-gal and picked for PCR amplification of target genes. Subsequently verified colonies were streaked for purification and subsequent DNA extraction and whole-genome sequencing.

Further manipulation of RJW004-ΔtraG was needed to obtain the proper genotype for mating assays (Figure 4). The argD, pyrE, and pyrF genes were inserted through purification and electroporation of M.16.4-derived amplicons (Zhang and Whitaker, 2018). Because the source is isogenic to RJW004, no homology modifications were introduced into the amplification of the genes. Enrichment for 14 days was necessary for efficient isolation of pyrEF⁺ colonies. Genomic DNA was extracted and sent to sequencing for further verification.

Conjugation of pM164 into S. islandicus M.16.2 was performed according to Prangishvili et al. (1998) with minor modifications and a selection scheme (Supplementary Figure 1). In general, S. islandicus RJW004 was co-incubated with S. islandicus M.16.2 at a donor: recipient ratio of 1:10,000 for 48 h under shaking conditions in liquid DY media containing uracil and agmatine at 76°C. After incubation, the co-culture was diluted using serial dilutions to 10^–3 of the original incubation and plated on DY media to select against RJW004. After incubation for 10 days, colonies were picked and verified for pM164 in the chromosome and original background through PCR amplification of pM164 region and the strain verification, respectively. M.16.2p was then purified through streaking and sent to sequencing using Illumina short reads. Additionally, circularization of pM164 was assessed through amplification of plasmid distal regions (Supplementary Figure 3).

SNP analyses were performed on recombinant isolates from three independent tester/M.16.2p mating assays. Colonies were isolated and transferred to liquid DY media until mid-log phase. Subsequently, DNA extraction and Illumina short read sequencing were performed on 30 RJW004 recombinants. Sequencing files were then analyzed for M.16.2p-specific donor SNPs through Breseq analysis (Deatherage and Barrick, 2014). SNPs were then mapped onto an M.16.4 reference for analysis. Alignment for Figure 3 was performed using Clinker (Gilchrist and Chooi, 2021). Variant positions in core genome were assessed using Spine and Nucmer (Kurtz et al., 2004; Ozer et al., 2014).

Statistical analysis was performed on mating assays in all cases. Comparisons were done utilizing ANOVA when experiments involved more than two assays. In cases where two assays were compared a t-test was performed to assess significant differences. In all cases p-value-0.05 is utilized.

The natural isolates utilized in this study were isolated from the Mutnovsky region in Kamchatka Russia in 2010. We focus on one of two closely related but distinct groups (the Red group) observed to be diverging over time in a manner that represents incipient speciation from within the same hot spring (Cadillo-Quiroz et al., 2012). The Red group is composed of three natural isolates which are closely related and have shown evidence of high levels of recombination between them. At the nucleotide level, the Red group possess 6,052 variant positions across core genome and an ANI (average nucleotide identity) of above 99.75% in all cases. We include one Blue group strain, M.16.27, for comparison of transfer among incipient species.

To test recombination frequencies between these strains we utilized a tester strain, RJW004, which is an M.16.4 background strain possessing three gene deletions (ΔargD, ΔpyrEF, and ΔlacS). This strain is crossed with wild-type strains in mating assays described in Figure 1 (Zhang et al., 2013a,b). The lacS gene is not selected in this assay but used to differentiate donor and recipient genome backgrounds on selective plates through X-gal staining. Selection on uracil/5-FOA (5-Fluoroorotic acid) containing plates without agmatine selects for strains in wild type background (blue) that receive the ΔpyrEF deletion locus from the tester (argD⁺pyrEF^–lacS⁺), and the tester strain (white) that receives the argD locus from the WT (argD⁺pyrEF^–lacS^–) (Figure 1). In the M.16.4 strain the selected markers (ΔpyrEF, argD) are separated by 62Kb. To differentiate deletion from spontaneous mutations in each experiment WT and tester strains are plated on selective plates and the number of spontaneous mutations is subtracted from total.

Figure 1 shows bi-directional recombination when the tester strain is crossed with WT strains from the same population. Figure 2 shows the recombination profiles of the above-mentioned natural isolates for our two selective markers. The recombination frequency of the ΔpyrEF locus from the tester to WT strains is consistent, independent of the WT background, yet slightly higher in the cross with M.16.27. In contrast, argD⁺ recombinants showed substantial variation with M.16.4 having the highest recombination frequency, M.16.27 having the lowest frequency, and the M.16.2 assay possessing the lowest donor frequency of the red isolates (Figure 2).

Interestingly, an integrated conjugative plasmid, named pM164, was found 8086bp from the highly transferred argD marker in M.16.4 and M.16.40 (Figure 3). PCR amplification has also shown evidence of excision in M.16.4 along with the integrated version, providing evidence that pM164 is an active pNOB8-like element. Because the argD recombination frequency was associated with pM164, in both M.16.4 and M.16.40, we hypothesized that pM164 may well be driving the transfer of proximal genes like argD. We note that M.16.27 has a divergent plasmid integrated in a different position relative to markers and relatively closer (∼72 kb) to pyrEF than to argD. This plasmid, however, lacks important plasmid components such as repA, parA, and trbE.

In bacterial conjugation models traG (M164_1618), encoding for a VirD4-type conjugation component, is essential for plasmid transfer as it mediates the interactions between the mating pore and the DNA transfer system (Arends et al., 2013; Kohler et al., 2013). The traG gene was identified through comparisons to pNOB8 and its genomic characterization (She et al., 1998). TraG in Sulfolobus spp. contains a conjugation-specific type IV secretion system (T4SS) domain found in bacterial conjugation proteins such as VirD4, TrwB, and TraG (She et al., 1998). When compared, three motifs are present that correspond to bacterial conjugation proteins such as TrwB in E. coli and with the archaeal HerA helicase (Supplementary Figure 2). Thus, we hypothesized that a traG deletion in pM164 would abrogate transfer and thus the associated high frequency recombination of argD. We created a marker-less in-frame deletion of the pM164 traG (M164_1618) in RJW004, along with the appropriate marker insertions, to create an argD⁺pyrEF⁺lacS^–traG^– strain by using the endogenous CRISPR-Cas system-based genome editing (Li et al., 2015). Because this strain is lacS^–, it was paired with RJW007, an argD^–pyrEF^–lacS⁺ tester, to yield selectable recombinant genotypes (Table 1). An M.16.4 (wt)/RJW004 (tester) mating assay was performed for comparison. We found a significant decrease in argD recombination frequencies when traG was not present in the argD donor strain (Figure 4). Background levels of marker exchange were maintained in these crosses, however, the bias for the argD marker was absent.

We tested whether pM164 is mobilizable and can independently transfer its high recombination phenotype to a naïve strain. The red isolate, M.16.2, did not possess a conjugative plasmid but has an identical att site and selective marker locations, as such we chose to conjugate pM164 into M.16.2. The conjugation protocol was adapted from Prangishvili et al. (1998) and is diagramed in Supplementary Figure 1 (see “3. Materials and methods”). First, ΔtraG-RJW004 was utilized as a pM164 donor to test the importance of traG to plasmid transfer in S. islandicus and found that only 1/99 recipient cells possessed pM164. Thus, much like in bacteria, traG is essential for conjugation in S. islandicus (Waters et al., 1992).

Using traG+ donor we were able to obtain M.16.2 transconjugants possessing pM164. Transconjugants were verified for strain identity and the appropriate pM164 insertion event using PCR amplification of the att site (Supplementary Figure 1). 99% of colonies surveyed possessed pM164 in the insertion site, showing strong evidence of insertion site specificity and high efficiency of transfer (She et al., 2004). Highly efficient transfer is characteristic of conjugative plasmids in Sulfolobus, as has been observed previously by Prangishvili et al. (1998). Of the colonies verified to possess pM164, three were purified and sent for sequencing to further confirm the strain and the presence of pM164 in the appropriate locus.

We mated M.16.2p, a pM164 transconjugant of the parental strain M.16.2, with the naive strain to assess the recombination frequencies of the two selectable markers (Figure 5). Through sequencing we confirmed that M.16.2p does not possess any other MGE other than pM164 and further confirmed that pM164 is integrated into the correct position. We show that high frequency recombination of argD follows insertion of pM164 into the appropriate tRNA-adjacent attachment site. Thus, we show direct evidence of an integrated plasmid influencing core genome recombination dynamics in Archaea. These results provide evidence that indeed pM164 and not the strain is the deciding factor in localized high frequency recombination of argD.

To gain further understanding into the mechanism of pM164 transfer, we sequenced 10 RJW004 recombinants from three independent M.16.2p mating assays for a total of 30 isolates. Each isolate represents independent recombination events within each experiment. We assess chromosomal regions that are transferred from M.16.2p to our tester strain, RJW004 (Figure 6A). The M.16.2p donor locus that stretches between argD and the pM164 integration site were differentiated from the recipient (RJW004) locus through SNP analysis (Figure 6). Because M.16.2 is highly similar to M.16.4, SNP positions scattered throughout the chromosome can be utilized as markers for recombination. We find consistent recombination tracts leading from the selected marker argD to pM164 in an uninterrupted manner. We also find a slope in recombinant SNPs upstream of argD that is not observed on the other end of pM164. From these data we infer that pM164 transfers unidirectionally while integrated in the chromosome, in a manner similar to Hfr strains in E. coli. In addition, we also found that 17/30 isolates possess unselected recombination events throughout the chromosome.

We find that long unselected donor tracts do not correlate with genomic distance from pM164. Thus, we hypothesize that although pM164 has a localized effect made apparent by the high frequency of unselected donor SNPs surrounding the argD/att region, the ME/Ced system most likely accounts for most of the intercellular mobilization of DNA throughout the chromosome. The localized high frequency effect is most likely due to unidirectional integrated plasmid transfer from a plasmid origin toward argD, these tracts are then resolved through homologous recombination between the chromosomes, defining the recombination tract lengths. These data, along with the essentiality of plasmid transfer through a TraG-mediated mechanism, show that in S. islandicus integrated plasmids can consistently and reproducibly create high-frequency marker exchange in nearby chromosomal loci.