Species-Specific Recognition of Sulfolobales Mediated by UV-Inducible Pili and S-Layer Glycosylation Patterns

Type IV pili can be found on the cell surface of many archaea and bacteria where they play important roles in different processes. The UV-inducible pili system of Sulfolobales (Ups) pili from the crenarchaeal Sulfolobales species are essential in establishing species-specific mating partners, thereby assisting in genome stability. With this work, we show that different Sulfolobus species have specific regions in their Ups pili subunits, which allow them to interact only with cells from the same species. Additionally, different Sulfolobus species have unique surface-layer N-glycosylation patterns. We propose that the unique features of each species allow the recognition of specific mating partners. This knowledge for the first time gives insights into the molecular basis of archaeal self-recognition.

not produce archaella or Aap pili (two other type IV pili-like structures present on the cell surface) and was found to aggregate normally upon UV induction ( Fig. 1B; see Fig. S1A). The absence of other surface structures enabled unambiguous analysis of Ups pili using electron microscopy. We used our previously established "pop-in pop-out" approach (37) to exchange both upsA and upsB in this background strain MW501 with the orthologous genetic region from S. tokodaii (from the start codon of upsA until the stop codon of upsB), resulting in strain MW135 ( Fig. 1A; Table 1). Upon UV induction, S. acidocaldarius MW135 was still found to produce Ups pili (Fig. S1B); however, interestingly, it showed little to no cellular aggregation (Fig. 1B). To test if this S. acidocaldarius Ups-hybrid strain was able to recognize and, therefore, aggregate with Molecular Basis of Archaeal Self-Recognition ® S. tokodaii cells, fluorescence in situ hybridization with species-specific probes was performed on mixed S. acidocaldarius/S. tokodaii strains after UV induction. A positive control with a mixture of background strain S. acidocaldarius MW501 and S. tokodaii confirmed previously observed species-specific aggregation (Fig. 1C, first panel). The negative control in which a S. acidocaldarius ΔupsAB strain (MW143) was mixed with S. tokodaii revealed, as expected, no aggregation of the S. acidocaldarius ΔupsAB strain and normal aggregation of S. tokodaii (Fig. 1C, second panel). Interestingly, cells from S. acidocaldarius MW135 interacted with S. tokodaii cells and, thereby, formed mixed species aggregates (Fig. 1C, third panel). This suggests that the S. acidocaldarius cells expressing S. tokodaii Ups pilin subunits were now recognizing and, therefore, interacting with S. tokodaii cells.
To find putative species-specific regions in the pilin subunits involved in speciesspecific recognition, alignments were made using UpsA and UpsB amino acid sequences, from several Sulfolobales (see Fig. S2A). Additionally, the relationship between UpsA and UpsB homologs was studied by creating a phylogenetic tree (Fig. S2B, Text S1). A region with low conservation was revealed in UpsA (Fig. S2A, amino acid 84 to 98 for S. acidocaldarius, red box). To test whether this region plays a role in cell-cell recognition, the region of low conservation in S. acidocaldarius UpsA (amino acid 84 to 98) was exchanged with the corresponding part from S. tokodaii UpsA (amino acid 80 to 101) (resulting in strain MW137) ( Fig. 1A; Table 1). Similar to what was observed for the S. acidocaldarius mutant in which upsA and upsB were exchanged completely (MW135), S. acidocaldarius MW137 still formed Ups pili ( Fig. S1B) but showed little to no UV-inducible aggregation with itself (Fig. 1B). Instead, it was found to aggregate with S. tokodaii (Fig. 1C, fourth panel). This observation strongly suggests that the nonconserved region (exchanged in MW137) defines the species specificity during cellular aggregation.
The role of glycosylation in species specificity. The fact that Sulfolobus Ups wild-type strains are able to form mating pairs with Ups-deletion strains (33) suggests that factors other than Ups pili play a role in species-specific recognition. All Sulfolobales harbor an S-layer containing two proteins, namely, SlaA and SlaB. Both proteins are heavily glycosylated, and the cells are thereby fully covered in an extensive extracellular glycan layer (38)(39)(40). We, therefore, suggested that Ups pili might recognize glycosylated proteins and, thereby, initiate cellular interactions. To confirm this hypothesis, UV-induced aggregation assays were performed in the presence of monosaccharides that are also part of the S. acidocaldarius N-glycan chain (Glc 1 Man 2 GlcNAc 2 QuiS, containing glucose, mannose, N-acetylglucosamine, and the Sulfolobus-specific sulfoquinovose residues) (39) (Fig. 2). The addition of N-acetylglucosamine or glucose did not result in altered cellular aggregation ( Fig. 2A and B); however, in the presence of mannose, cell aggregates were significantly smaller (Fig. 2B). To verify that the observed reduced aggregation was not caused by a lower expression of the pilin genes, we performed quantitative PCR (qPCR) on cDNA from cells isolated after the addition of mannose. We could not observe any differences in upsA transcript levels between cells that were or were not incubated with mannose, independent of UV treatment (see Fig. S3). Thus, we assume that Ups pili expression is not affected by the addition of mannose. Our results, therefore, suggest that mannose molecules partially saturate the binding sites of the Ups pili and, thereby, inhibit interactions between pili and the glycan chains on the S-layer of the host cell, resulting in reduced aggregation.
Defining the glycosylation pattern of S. tokodaii S-layer proteins. Our hypothesis that S-layer glycosylation is important for species specificity suggests that different Sulfolobus species have different glycosylation patterns. So far, the glycan structure of S. tokodaii is unknown. To analyze the glycan structures on the S-layer of S. tokodaii, N-glycans were released from isolated S-layer by hydrazinolysis. Using matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF-MS) profiling, one main N-glycan species and two other low-abundant species could be identified in both positive (see Fig. S4A) and negative ion mode (Fig. S4B). The structures of N-glycans were proposed based on mass-to-charge ratio of each N-glycan ions ob-van Wolferen et al.  Table 2) as well as its tandem mass spectrometry (MS 2 ) fragmentation pattern (Fig. 3). The three N-glycan species were identified as QuiS 1 Hex 4 HexNAc 2 , QuiS 1 Hex 3 HexNAc 2 , and QuiS 1 Hex 4 HexNAc 1 ( Table 2). To determine the linkages between the sugars in the deduced N-glycan species, linkage analysis (41) was performed on the permethylated N-glycans released from S-layer proteins. The various types of linkages observed on each monosaccharide and their relative abundances on the N-glycans are shown in Fig. S5. The most plausible position of this linkage in the glycan chain can be observed on the right-side column in Fig. S5. Based on this linkage information, sequential mass spectrometry (MS n ) determination of glycan branching ( Fig. 3) and the glycan masses ( Fig. S4), the N-glycan glycoforms, and their isomers were deduced (see Fig. S6). Fig. 4 schematically shows the most prominent glycan structures from S. acidocaldarius (39), Sulfolobus solfataricus (42), and S. tokodaii (this study). In agreement with our hypothesis, the core of these structures is similar, whereas the terminal saccharides differ. A typical sulfated sugar residue is present in all three  Sulfolobus glycan structures. Using liquid chromatography (LC)-MS profiling on the tryptic digest of S-layer proteins SlaA and SlaB, several different glycopeptides could indeed be observed (Text S1, Fig. S7 and S8, respectively). Determination of the binding site in UpsA. We know that a S. acidocaldarius mutant in which both Ups pilin subunits are deleted does not aggregate upon UV stress (32). Here, we could successfully complement this phenotype by expressing the upsAB genes from a maltose-inducible plasmid ( Fig. 5) (ΔupsAB ϩ upsAB). Using site-directed mutagenesis on this plasmid, we, moreover, created point mutations within the abovedescribed region of interest of UpsA (black squares in Fig. S2A), namely, D85A, N87A, N94A, and Y96A. All mutants still produced Ups pili upon UV induction (Fig. S1B). Interestingly, when expressing UpsA in which the poorly conserved residues D85 or Y96 were mutated to alanine, UV-induced aggregation was significantly reduced. On the  other hand, mutation of conserved N87 or N94 showed wild-type aggregation (Fig. 5). These results suggest that the region of low conservation within UpsA is specifically adapted to the glycan structure of the same species to ensure species-specific aggregation.

DISCUSSION
Both bacterial and archaeal T4P have shown to be essential for surface adherence. Given the fact that bacterial T4P are strongly related to pathogenicity, their mode of binding has primarily been studied for pathogenic bacteria, such as P. aeruginosa, V. cholerae, Neisseria, and enteropathogenic Escherichia coli species. However, nonpathogenic bacteria and archaea also carry several T4P involved in adhesion, which are studied in far less detail. The crenarchaeal Sulfolobales carry the following three types of T4P: archaella, involved in swimming motility (21); Aap pili, involved in attachment to diverse surfaces (25,26); and Ups pili, mediating intraspecies cellular aggregation and DNA exchange (30,32,33,36). During this study, we have examined the role that Ups pili play in the formation of Sulfolobus mating partners. In particular, we focused on the role that pilin subunit UpsA plays in cell recognition.
The Ups pilus is formed by two pilin subunits, UpsA and UpsB, which are both thought to be major pilin subunits that build up mixed pili structures (32). We revealed that UpsA is involved in species-specific cellular interactions, and we were able to alter this specificity by exchanging (parts of) the pilin subunit with that of another species (Fig. 1). The binding of bacterial surface structures to other cells is often based on interactions with sugars (43,44). Surface-exposed glycans can be found on cells from all domains of life where they display an enormous range of different structures that are often highly specific to certain species (45). Glycans are, therefore, perfect anchors to bind specific host or partner cells.
The glycosylation ratio of S-layer protein SlaA from S. acidocaldarius was found to be extremely high compared with that of S-layer proteins from euryarchaeal species. This high glycosylation density is thought to be an adaptation to the high temperature and acidic environment that Sulfolobus species live in (39). Recently, the extensive lawn of glycans on top of the S-layer of S. solfataricus was visualized by cryo-electron microscopy (EM), emphasizing the general importance of Sulfolobus N-glycosylation in the formation of cellular interactions with anything that is present in the extracellular environment (40). In Saccharomyces cerevisiae, surface-exposed lectins can bind to surface-exposed sugars in a calcium-dependent manner, thereby forming cellular aggregates, a process which is called flocculation (46). This behavior can be inhibited by saturating the binding of the lectins through the addition of loose sugars to the Molecular Basis of Archaeal Self-Recognition ® medium (47) (Fig. 2). In similar experiments with S. acidocaldarius, we found that mannose has an inhibiting effect on UV-induced cellular aggregation. Since two outer mannose residues are present in the S. acidocaldarius N-glycan tree, binding of Ups pili to this side of the glycan tree is probable. When analyzing the N-glycans of S. tokodaii, we could indeed find differences in the outer part of the N-glycan structure compared with that of S. acidocaldarius (39) and S. solfataricus. (42) (Fig. 4). As observed for Eukarya (48), the core or the glycan structure is similar in all three species, whereas the outer residues differ. Our results, thereby, suggest that UpsA contains a specific binding pocket that is able to bind specific sugar moieties of the N-glycans presented on the S-layer of distinct Sulfolobus species (Fig. 6).
Among the euryarchaeal Haloferax species, glycosylation was found to be essential for cell fusion (49), emphasizing the importance of glycosylation in archaeal cellular recognition in general. It is unclear if pili or other types of lectin molecules are involved in cellular interactions that initiate Haloferax fusion events. Similar to our findings, different Haloferax species are also known to be differentially glycosylated (50), leading to semispecific cell-cell recognition (49). Cell fusion between different Haloferax species could also be observed but with far lower efficiency (51). In addition, under different environmental conditions, Haloferax glycosylation patterns change, leading to more or less favorable N-glycans for mating (52). One could envision that low-frequency interactions between different Sulfolobus species also occur and might occasionally lead to horizontal gene transfer (35), thereby playing an important role in genome evolution. In a single hot spring in Kamchatka, Russia, two different groups of Sulfolobus islandicus strains were found to be present. Despite their coexistence, it was postulated that S. islandicus species mainly exchange DNA within these groups (53). It is likely that N-glycan patterns and Ups pili between the species are different, serving as a barrier to  gene transfer. This behavior might be seen as the two groups diverging into different species. Bacterial T4P are dynamic structures that can polymerize and depolymerize, allowing cycles of pili extension and retraction and enabling cells to pull themselves toward other (host) cells and surfaces (1). So far, nothing is known about the dynamics of Ups pili, and it is unclear how interactions between Ups pili and glycosylated S-layer result in the formation of mating pairs. We assume that for the cells to exchange DNA, tight cellular interactions have to be formed that enable direct cell surface contact. If cellular interactions between Sulfolobus cells are initiated with the tip of an Ups pilus, the distance created by the Ups pili themselves will have to be overcome. Unlike bacteria, archaea do not carry homologs of the PilT ATPase allowing retraction of T4P (54). Recently, pili retraction has been observed in certain bacterial T4P systems that lack a retraction ATPase (55,56); in addition, retraction could be observed in certain PilT deletion mutants (although with far lower force) (57)(58)(59). One could, therefore, imagine that PilT-independent retraction by multiple Ups pili will create a collective force that is strong enough to pull cells together and form tight aggregates. It can, of course, not be excluded that so far unidentified retraction ATPases are involved in retraction of Ups pili. Alternatively, one could imagine scenarios in which Ups pili are degraded extracellularly and, thereby, shortened until cell-cell contact is established, or they could be flexible enough to completely bend toward the cell surface.
The Ced system that is involved in DNA transfer among Sulfolobales can also be found in several crenarchaea that do not encode Ups pili (35,36); it is, therefore, likely these species have developed a different mechanism to initiate cellular interactions.

Given the importance of glycosylation in cell-cell interactions in both euryarchaeal
Haloferax and crenarchaeal Sulfolobus species, glycosylation is likely to play a role in these interactions.
This study has given molecular insights in the cellular recognition mechanism of the described crenarchaeal Ups system (30,32,33). Our current model suggests that upon DNA damage, Ups pili are formed; the UpsA pilin subunits contain a species-specific glycan-binding pocket in pilin subunit UpsA that can bind glycans presented on cells from the same species (Fig. 6). This system allows the formation of species-specific cellular connections prior to DNA exchange via the Ced system (36). In that way, only DNA from the same species is exchanged and used for DNA repair via efficient homologous recombination. This proposed cellular recognition mechanism in Sulfolobales promotes the exchange of genomic DNA to cells from the same species, thereby playing an important in role genome integrity and the maintenance of species. Coevolution of N-glycosylation and pilin subunit UpsA might play an important role in speciation.

Bioinformatics.
UpsA homologs from several different species were aligned using ClustalW; this alignment was used to create maximum likelihood tree for phylogenetic analysis, as described in the supplemental methods section.
Culture conditions. Sulfolobus acidocaldarius strains and derived mutants (Table 1) were grown aerobically at 75°C in basic Brock medium (60), supplemented with 0.1% NZ amine, 0.2% dextrin, and 20 g/ml uracil, and adjusted to pH 3.5 with sulfuric acid. For solid media, the medium was supplemented with 1.5% gelrite. Plates were incubated for 5 to 6 days at 75°C. Sulfolobus tokodaii strain 7 was grown aerobically at 75°C in basic Brock medium (60), supplemented with 0.1% NZ amine and 0.4% dextrin, and adjusted to pH 3.5 with sulfuric acid. E. coli competent cells DH5␣ and ER1821 (New England BioLabs [NEB]) used for cloning and methylation, respectively, of plasmid DNA were grown in LB medium (10 g/liter tryptone, 5 g/liter yeast extract, and 10 g/liter NaCl) at 37°C supplemented with the appropriate antibiotics. The growth of cells was monitored by optical density measurements at 600 nm.
qPCR. To test the effect of sugars on the transcription of upsA, we isolated RNA from cells with or without the addition of mannose (both with and without UV induction). cDNA was synthesized and qPCR performed as described in the supplemental methods.
Deletion, exchange, and complementation of genes in S. acidocaldarius. To construct deletion and pilin exchange mutants, up-and downstream flanking regions of the genes of interest (approximately 600 bp) were amplified with primers listed in Table S1. Overlap PCR was performed to connect the up-and downstream fragments. To replace (parts of) upsA and upsB from S. acidocaldarius with their homologs from Sulfolobus tokodaii, synthetic DNA was ordered (GenScript) consisting of S. acidocaldarius upsAB flanking regions and (parts of) S. tokodaii upsAB genes (Table S1). The PCR product and synthetic Molecular Basis of Archaeal Self-Recognition ® DNA fragments were subsequently cloned into pSVA406, resulting in the plasmids listed in Table S1. All plasmids contain a pyrEF cassette allowing selection on plates without uracil. The plasmids were methylated in E. coli ER1821 containing pM.EsaBC4I (NEB) (61) and transformed into S. acidocaldarius MW501 (Δfla/Δaap) ( Table 1) (37). This uracil auxotrophic background strain lacks Aap pili and archaella, allowing easy EM analysis. Integrants were selected on plates lacking uracil and grown in 24-well plates for 2 days in the same medium. Subsequently, cultures were plated and grown for 5 days on second selection plates containing uracil and 100 g/ml 5-fluoroorotic acid (FOA) to select for clones in which the plasmid looped out by homologous recombination. Obtained colonies were tested by PCR for successful deletion/replacement of the genes. Correctness of strains was confirmed by DNA sequencing. Strains that were made during this study are listed in Table 1.
For complementation of a ΔupsAB mutant (MW143), the DNA region comprising upsA and B was amplified using primers listed in Table S1 and cloned into pSVA1450 under the control of a maltoseinducible promoter resulting in plasmid pSVA1855 (Table S1). This plasmid was subsequently used as a template to introduce point mutations into upsA (D85A, N87A, N94A, and Y96A) ( Table 1) using two overlapping primers per mutation (Table S1). Resulting plasmids were then transformed via electroporation into MW143 as described previously (37). Cultures were grown without the addition of uracil. Expression of (mutated) UpsA and B was induced by the addition of 0.2% maltose.
UV treatment and aggregation assays. UV light treatment was performed as described in reference 30; a total of 10-ml culture (grown to an optical density at 600 nm [OD 600 ], 0.2 to 0.3) was treated with a UV dose of 75 J/m 2 (254 nm; UV cross-linker; Spectroline) in a plastic petri dish. For FISH experiments, S. acidocaldarius and S. tokodaii were first mixed in equal amounts. For complementation the ΔupsAB strain, expression of UpsAB (derivatives) was additionally induced with 0.2% maltose. For sugar assays 20 mM mannose, glucose, or N-acetylglucosamine was added after UV induction. Afterward, cultures were put back at 75°C for 3 h. Samples taken at different time points were analyzed with phase-contrast microscopy. To quantify aggregated cells after induction with UV, 5 l of cell culture (diluted to OD 600 of 0.2 by adding in extra medium and swirling the culture) was spotted and dried on a microscope slide covered with a thin layer of 2% agarose in Brock minimal medium. All pipetting steps were done carefully using tips with their points cut off. Cells were visualized with phase-contrast microscopy (Axio Observer.Z1; Zeiss). Total amounts of free and aggregated cells (Ն3) were counted for at least three fields per strain using the ImageJ cell counter. Percentages of cells in aggregates were subsequently calculated. Aggregate sizes were calculated by dividing the total amount of aggregated cells with the amount of aggregates.
Fluorescence in situ hybridization. For FISH experiments, 10 l of a mixed UV-induced (described above) culture was spotted and dried on a glass slide. To fix the cells, 10 l of 37% formaldehyde was spotted on top of the cells, and they incubated for 20 min at room temperature. Afterward, formaldehyde was removed and the cells were washed for 10 min with a drop of 1ϫ phosphate-buffered saline (PBS). Glass slides were subsequently dried at room temperature. Cells were permeabilized by incubating the slides 3 min in 50%, 80%, and 96% ethanol, respectively. After drying the slides, 10 l of hybridization buffer (900 mM NaCl, 20 mM Tris HCl [pH 8.0], and 10% formamid) mixed with 50 ng/l FISH probes (for S. acidocaldarius and S. tokodaii) (Table S1) was spotted on the cells. Slides were incubated in the dark at 46°C for 1.5 h for hybridization. Subsequently, the cells were washed by incubating the slides for 10 min in wash buffer (450 mM NaCl and 20 mM Tris HCl [pH 8.0]) at 48°C. Slides were then dipped in ice cold water and dried. For microscopy, 1ϫ PBS was spotted on the cells and a coverslip was added. Cells were examined using fluorescence microscopy (Axio Observer.Z1; Zeiss).
Electron microscopy analysis. Ups pili on S. acidocaldarius cells were visualized with transmission electron microscopy (TEM). For this, 5 l of culture was taken 2 h after UV induction and directly dropped onto a carbon᎑coated copper grid (SF162-3; Plano), without any prior centrifugation steps. After 2 min of incubation, the fluid was taken off with Whatman paper, and subsequently, 5 l of 2% uranyl acetate was dropped onto the grid for negative staining and incubated for 30 s. The grid was washed once with water and dried. Transmission electron microscopy images were recorded using the Talos L120C (Thermo Scientific) microscope equipped with a 4,000 by 4,000 Ceta CMOS camera. Acceleration voltage was set to 120 kV and magnification to 2.27 Å/pixel. S-layer isolation and N-glycan analysis. A cell pellet from a 1.5-liter S. tokodaii strain 7 culture with an OD 600 of about 0.4 was used to isolate S-layer as described previously (39). The cells were resuspended and incubated while shaking for 60 min at 37°C in 30 ml of buffer A (10 mM NaCl, 0.5% sodium lauroylsarcosine, and a small amount of DNase). Samples were centrifuged for 45 min in an Avanti J-26 XP centrifuge (Beckman Coulter) at 21,000 ϫ g (rotor JA-25.50), yielding a brownish tan pellet harboring unsolubilized cell debris with a white top layer (the S-layer fraction). After centrifugation, the white layer on top of the pellet was resuspended in 1 ml buffer B (10 mM NaCl and 0.5% SDS) and incubated for 30 min at 37°C; subsequently, it was spun down in a tabletop centrifuge at maximum speed for 10 min. The latter step was repeated until the pellet was completely white (thereby solubilizing and removing most of the remaining lipids and membrane proteins). The purified S-layer fraction was washed several times with water and then stored in water at 4°C. N-linked glycans were released from the purified S-layer and analyzed as described in the supplementary methods.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. TEXT S1, DOCX file, 0.03 MB. We thank Małgorzata Ajon (University of Groningen) for technical support.