AglH, a thermophilic UDP-N-acetylglucosamine-1-phosphate:dolichyl phosphate GlcNAc-1-phosphotransferase initiating protein N-glycosylation pathway in Sulfolobus acidocaldarius, is capable of complementing the eukaryal Alg7

AglH, a predicted UDP-GlcNAc-1-phosphate:dolichyl phosphate GlcNAc-1-phosphotransferase, is initiating the protein N-glycosylation pathway in the thermoacidophilic crenarchaeon Sulfolobus acidocaldarius. AglH successfully replaced the endogenous GlcNAc-1-phosphotransferase activity of Alg7 in a conditional lethal Saccharomyces cerevisiae strain, in which the first step of the eukaryal protein N-glycosylation process was repressed. This study is one of the few examples of cross-domain complementation demonstrating a conserved polyprenyl phosphate transferase reaction within the eukaryal and archaeal domain like it was demonstrated for Methanococcus voltae (Shams-Eldin et al. 2008). The topology prediction and the alignment of the AglH membrane protein with GlcNAc-1-phosphotransferases from the three domains of life show significant conservation of amino acids within the different proposed cytoplasmic loops. Alanine mutations of selected conserved amino acids in the putative cytoplasmic loops II (D100), IV (F220) and V (F264) demonstrated the importance of these amino acids for cross-domain AlgH activity in in vitro complementation assays in S. cerevisiae. Furthermore, antibiotic treatment interfering directly with the activity of dolichyl phosphate GlcNAc-1-phosphotransferases confirmed the essentiality of N-glycosylation for cell survival.


Introduction
All living cells exhibit an outer surface covered with an array of glycans. These glycans are either loosely attached or covalently linked to surface proteins or lipids. Protein glycosylation is one of the most common posttranslational protein modifications found in all three domains of life (Larkin and Imperiali 2011). In particular, protein N-glycosylation is widely distributed in Eukarya and Archaea whereas it is rarely found in Bacteria. In these systems, the biosynthesis of N-linked oligosaccharides is initiated by the transfer of a sugar(-1-phosphate) residue from a nucleotideactivated sugar onto the lipid carrier dolichyl phosphate (DolP) or undecaprenyl phosphate (UndP), respectively. The fully assembled lipid-linked glycan is then transferred to a specific Asn (N) residue within a target protein.
Eukaryotic protein N-glycosylation is initiated by the GlcNAc-1-phosphate transferase Alg7/Dpagt1 (Lehrman 1991;Mclachlan and Krag 1992), which converts UDP-GlcNAc and DolP into UMP and DolPP-GlcNAc. This step is essential in Eukarya and defined mutations in the human Dpagt1 cause severe clinical phenotypes, leading to lethal diseases (Jaeken and Matthijs 2007;Wurde et al. 2012). DolPP-GlcNAc acts as a primer for the elongation Abstract AglH, a predicted UDP-GlcNAc-1-phosphate: dolichyl phosphate GlcNAc-1-phosphotransferase, is initiating the protein N-glycosylation pathway in the thermoacidophilic crenarchaeon Sulfolobus acidocaldarius. AglH successfully replaced the endogenous GlcNAc-1-phosphotransferase activity of Alg7 in a conditional lethal Saccharomyces cerevisiae strain, in which the first step of the eukaryal protein N-glycosylation process was repressed. This study is one of the few examples of cross-domain complementation demonstrating a conserved polyprenyl phosphate transferase reaction within the eukaryal and archaeal domain like it was demonstrated for Methanococcus voltae (Shams-Eldin et al. 2008). The topology prediction and the alignment of the AglH membrane protein with GlcNAc-1-phosphotransferases from the three domains of life show significant conservation of amino acids within the different proposed cytoplasmic loops. Alanine mutations of selected conserved amino acids in the putative cytoplasmic loops II (D 100 ), IV (F 220 ) and V (F 264 ) demonstrated the 11th International Congress on Extremophiles

Communicated by A. Driessen.
This article is part of a special feature based on the 11th International Congress on Extremophiles held in Kyoto, Japan, September 12-16, 2016. of the N-glycans by specific glycosyltransferases (GTases), sequentially transferring individual sugars from corresponding nucleotide-or lipid phosphate-activated precursors (Burda and Aebi 1999).
The bacterial protein N-glycosylation pathway from Campylobacter jejuni starts with the transfer of diacetamido bacillosamine-1-P from UDP-2,4-diacetamido bacillosamine onto the lipid carrier UndP by the phosphotransferase PglC (Glover et al. 2006). Similar to the initiation of protein N-glycosylation, the biosynthesis of several other bacterial glycoconjugates starts at the cytoplasmic site of the cell membrane with the formation of an UndPPlinked monosaccharide. Examples include the biosynthesis of O-antigen polymers (Meier-Dieter et al. 1992;Samuel and Reeves 2003;Schmidt et al. 1976), the capsular antigens (Masson and Holbein 1985;Troy et al. 1975;Whitfield 2006), lipopolysaccharide (Alexander and Valvano 1994;Schmidt et al. 1976), teichoic acids (Brown et al. 2008;Ginsberg et al. 2006;Mancuso and Chiu 1982) and the peptidoglycan (Bouhss et al. 2008;Bugg and Brandish 1994;Typas et al. 2012). The initiation step is meditated by different classes of membrane-bound UDP-hex(NAc)-1-phosphate:polyprenyl phosphate sugar-1-phosphotransferase enzymes, which use UndP as an acceptor but differ in their specificity for the nucleotide sugar donor. The bacterial enzyme WecA (formerly known as Rfe) catalyses the first step of biosynthesis of many LPS O-antigens and capsular K-antigens by the transfer of GlcNAc-1-P residue from UDP-GlcNAc onto UndP (Amor and Whitfield 1997).
In Archaea, N-glycosylation pathways for three different euryachaea and one crenarchaeon have been described [for review see (Jarrell et al. 2014)]. In contrast to Eukarya and Bacteria, most of the euryarchaeal N-glycans characterised to date, i.e. Hfx. volcanii, Haloarcula marismortui, Methanococcus voltae, and Pyrococcus furiosus, are assembled on DolP lipid carrier (Calo et al. 2011;Chang et al. 2015;Guan et al. 2010;Larkin et al. 2013), with two exceptions. Methanothermus fervidus assemble their N-glycans on DolPP (Hartmann and König 1989) and Hbt. salinarum, which synthesizes two distinct N-glycans, assembles one on DolP and the second on DolPP (Paul and Wieland 1987). A recent comparison of archaeal lipid-linked oligosaccharides revealed a difference between eury-and crenarchaeota. Indeed, selected species of the euryarchaeota possessed DolP-linked glycans, whereas the crenarchaeota Pyrobaculum calidifontis and Sulfolobus solfataricus assemble their N-glycans on DolPP (Taguchi et al. 2016). Different steps of the N-glycosylation pathway of the closely related crenarchaeon Sulfolobus acidocaldarius have been studied in detail (Meyer and Albers 2013). Here, a tribranched hexasaccharide, composed of two GlcNAc, two terminal Man, one sulfoquinovose and one terminal Glc are found as the N-glycan linked to the S-layer protein (Peyfoon et al. 2010). Interestingly, the basal structure of this archaeal N-glycan (GlcNAc 2 Man) resembles this of the eukaryal one and the biosynthesis might rely on homologs to the eukaryal ones.
Unlike in euryarchaeota, the initiation step of the protein N-glycosylation process in thermophilic crenarchaeota has not yet been elucidated. In the present study, we identified a candidate enzyme for the initiation step of the protein N-glycosylation pathway in S. acidocaldarius by homology searches. The identified AglH (Saci0093) is able to restore N-glycosylation in a conditional lethal Saccharomyces cerevisiae alg7 mutant. Various attempts to delete aglH in S. acidocaldarius were unsuccessful, and the use of tunicamycin, a specific inhibitor of UDP-HexNAc-1phosphate:polyprenol phosphate HexNAc-1-phosphotransferases, as well as the treatment with bacitracin interfering with the regeneration of DolP, revealed the essentiality of AglH in S. acidocaldarius.

Growth studies with bacitracin and tunicamycin
Pre-cultures grown to an OD 600 of 0.4-0.5 were used to inoculate 5 ml of fresh Brock medium (0.1% w/v NZ amine and 0.1% w/v dextrin) to a start OD of 0.01. Directly with inoculation or after reaching exponential growth phase, different concentrations of the antibiotics bacitracin (0.6, 1.2, and 2.4 mM) and tunicamycin (2, 6, 16, and 32 µg/ ml) were added. Growth was monitored by measuring the OD 600 over a time period of 80 h.

Construction of a ΔaglH deletion
A markerless deletion mutant would represent an important tool to analyse whether the predicted UDP-GlcNAc-1-phosphate:polyprenyl phosphate GlcNAc-1-phosphotransferase (AglH) indeed catalyses the first step of the protein N-glycosylation process. A gene deletion plasmid was used for homologous recombination as described previously (Wagner et al. 2009). To construct the deletion plasmid, 800-1000 bp of sequence up-and downstream of saci0093 was PCR amplified. ApaI and BamHI restriction sites were introduced at the 5ʹ ends of the upstream forward primer (1842) and the downstream reverse primer (1845), respectively. The upstream reverse primer (1843) and the downstream forward primer (1844) were each designed to incorporate 15 bp of the reverse complement strand of the other primer, resulting in a 30 bp overlapping stretch. The up-and downstream fragments were fused by overlapping PCR, using the 3ʹ ends of the up-and downstream fragments as primers. The resulting PCR product was cloned into plasmid pSAV407 using the restriction enzymes ApaI and BamHI, yielding the plasmid pSVA1229. The plasmid was transformed into Escherichia coli DH5α cells, which were subsequently plated on LB agar containing 50 mg ml −1 ampicillin. The plasmid was confirmed by sequencing. To avoid restriction in S. acidocaldarius, the plasmid was methylated by transformation in E. coli ER1821 cells containing pM.EsaBC4I, which expresses a methylase (obtained from NEB).

Cloning of aglH FLAG into a S. cerevisiae expression vector
To verify the proposed enzymatic function of AglH we reasoned that aglH might be able to complement a conditional lethal yeast alg7 mutant, as has previously been demonstrated for the human Alg7 (Eckert et al. 1998) and the AglH of M. voltae (Shams-Eldin et al. 2008). We, therefore, cloned aglH fused to a FLAG-tag sequence into the pRS426-MET (Mumberg et al. 1995) expression vector. The aglH gene was PCR amplified with forward primer (1755) and reverse primer (1756), introducing EcoRI and XhoI restriction sites, respectively. The 1756 primer incorporated the FLAG-tag sequence upstream of the XhoI restriction site. The digested PCR fragment was cloned into the expression vector pRS426-MET (Mumberg et al. 1995) linearised with the restriction enzymes EcoRI and XhoI, creating plasmid pSAV1212 ( Table 3). The plasmid was transformed into E. coli DH5α, and was confirmed by sequencing.

Oligonucleotide-directed mutagenesis of aglH FLAG
Plasmid pSVA1212 was used as a template for the oligonucleotide-directed mutagenesis of aglH FLAG . In general, a PCR with a specific forward and reverse primers (Table 2) carrying the nucleotide exchange were used to amplify the gene. The resulting PCR products were digested with DpnI to exclude the methylated template plasmid pSVA1229. After purification, each construct was transformed into E. coli DH5α, and platted on selective LB plates containing 50 mg ml −1 ampicillin. The exchange of the nucleotides was confirmed by sequencing of the entire aglH FLAG gene.

Transformation and selection of S. acidocaldarius mutants
Competent cells were obtained according to the protocol of Kurosawa and Grogan (Kurosawa and Grogan 2005). Aliquots containing 400-600 ng of methylated plasmids (pSVA1229 or pSVA1246) were mixed with 50 µl of competent MW001 cells and incubated for 5 min on ice. Electroporation was performed using a Gene pulser II (Bio-Rad, USA) with the input parameters 1.25 kV, 25 mF, 1000 W in 1 mm cuvettes. Immediately after the pulse, 50 µl of a 2 × concentrated recovery solution (1% sucrose, 20 mM β-alanine, 20 mM malate buffer pH 4.5, 10 mM MgSO 4 ) was added and the samples were incubated for 30 min at 75 °C under mild shaking conditions (150 rpm). Prior to plating, additionally 100 µl of heated 2x-concentrated recovery solution was added, and 2 × 100 µl was spread on gelrite plates containing Brock medium supplemented with 0.1% NZ-amine and 0.1% dextrin, lacking uracil. After incubation for 5-7 days at 75 °C, large brownish colonies were used to inoculate 50 ml of first selection Brock medium containing 0.1% NZ-amine. After 3 days of incubation at 75 °C, each culture was screened for the presence of the integrated plasmid by PCR. Positively tested isolates were inoculated in Brock medium (0.1% NZ-amine and 0.1% dextrin) and grown until an OD 600 of 0.4 was reached. 40 µl of aliquots were then spread on second selection plates, where the recombination step was selected by the presence of 5-fluoroorotic acid (5-FOA) and uracil. After incubation for 5-7 days at 75 °C, differently sized colonies were picked and streaked onto new second selection plates to ensure single colony formation. Colonies were screened by PCR for the presence or the deletion of saci0093 (aglH), using the outer primers (1842 and 1845).

Immunostaining of the AglH FLAG point mutations expressed in S. cerevisiae
Cell pellets of the same OD 600 from 5 ml of SGR medium cultures of YPH499-HIS-GAL-ALG7 transformed with the AglH complementation plasmids (Table 3) were resuspended in 1 ml of buffer A (100 mM NaCl, 100 mM Tris-HCl, 1 mM EDTA, pH 8) and lysed by 20 min sonification with an intensity of 60% and an interval 20 s (Bandelin Sonopuls). Unbroken cells were removed by centrifugation at 3000×g at 4 °C for 20 min. The supernatant was centrifuged in a Beckman Coulter Optima Max-XP Ultracentrifuge at 120,000×g at

Identification of AglH, a predicted UDP-GlcNAc-1-phosphate:dolichyl phosphate GlcNAc-1-phosphotransferase
To identify the enzyme required for the first step of the protein N-glycosylation pathway in the thermoacidophilic crenarchaeon S. acidocaldarius, its genome was analysed for the presence of eukaryotic, bacterial, and archaeal homologues of the UDP-GlcNAc-1-phosphate:dolichyl phosphate GlcNAc-1-phosphotransferase Alg7/Dpagt1, WecA, and AglH, respectively. This search identified saci0093 (SACI_RS00435) encoding a predicted UDP-GlcNAc-1-phosphate:dolichyl phosphate GlcNAc-1-phosphotransferase, which shares 27% amino acid sequence identity with the yeast Alg7 and the human Dpagt1, 31% with AglH from the euryarchaeon M. voltae and 26% with WecA of E.coli. Saci0093 is located in a gene locus downstream of genes whose products are predicted to be involved in the early steps of isoprenoid lipid biosynthesis (Fig. 1). The gene saci0091 (SACI_RS00425, idi) codes for an isopentenyl diphosphate delta-isomerase, which is necessary for the biosynthesis of isoprenoid compounds via the mevalonate pathway (Boucher et al. 2004). The corresponding protein from S. shibatae is an isopentenyl-diphosphate delta-isomerase that catalyses the  (Nakatani et al. 2012;Yamashita et al. 2004). Saci0092 (SACI_RS00430, gds) encodes a bifunctional geranylgeranyl pyrophosphate synthase that synthesizes both farnesyl pyrophosphate (FPP) and geranylgeranyl pyrophosphate (GGPP) (Ohnuma et al. 1998;Ohnuma et al. 1996). GGPP is an intermediate for the biosynthesis of many isoprenoid compounds like carotenoids, geranylgeranylated proteins, as well as archaeal ether-linked membrane lipids. Furthermore, GGPP acts as a precursor in the proposed DolP biosynthesis pathway . AglH from S. acidocaldarius shows high similarity in primary sequence and transmembrane topology with eukaryal and bacterial UDP-GlcNAc-1-phosphate:polyprenyl phosphate GlcNAc-1-phosphotransferases Topology prediction of AglH (Saci0093) in combination with the known primary sequence revealed high levels of similarities with eukaryal as well as bacterial GlcNAc-1-phosphotransferases. AglH possesses ten transmembrane (TM) helices separated by five internal and four external hydrophilic loops (Fig. 2). A similar topology was shown for the eukaryal Alg7 enzyme with ten predicted TM domains, while bacterial WecA typically exhibits ten to eleven TM domains (Anderson et al. 2000;Lehrer et al. 2007 and archaeal orthologues illustrated high amino acid similarity within each of the five cytoplasmic loops, indicating a possibly conserved function across the three domains of life (Fig. 3).
In the cytoplasmic loop (CL) I, a conserved D 39 xxK/N motif is detected (Figs. 2, 3). In the CL II of Saci0093, a D 99 D 100 xxN 103 motif is present, similar to the conserved motif of eukaryal and bacterial DDxxD motifs (Fig. 3). This conserved motif is potentially involved in Mg 2+ cofactor binding (Amer and Valvano 2002;Lloyd et al. 2004;Xu et al. 2004). Within the fifth TMD, the CL III, and the six TMD, a long N 158 xxN 161 xxxxG 165 xxG 168 xxxxG 173 motif is present. The two conserved aspartic residues have been mutated in the E. coli WecA DD156/159GG , which resulted in only 3% of the wt transferase activity (Amer and Valvano 2002). The impaired function in vivo and the reduction of enzyme activity in vitro suggest the importance of this motif for the catalytic activity. This motif, in combination with the DDxxK/N motif of the second cytoplasmic loop, resembles the characteristics of a Walker B motif (Amer and Valvano 2002). The large predicted CL V is thought to be involved in the recognition of the carbohydrate moiety of the UDP-GlcNAc donor. In the respective loop of bacterial WecA, a conserved histidine residue is located within a less conserved short-sequence motif HIHH (Amer and Valvano 2001;Anderson et al. 2000). Replacement of the highly conserved His 279 residue with serine rendered WecA unable to restore O7 production in an E. coli wecA::Tn10 strain (Lehrer et al. 2007). Unlike bacteria, eukaryal and archaeal orthologues are missing the HIHH motif. If the functional interpretation for WecA is correct, this raises the question of how the carbohydrate moiety of the UDP-GlcNAc donor is being recognized in these two domains of life (Fig. 3).
Collectively, the sequence features for AglH are entirely consistent with it being a UDP-GlcNAc-1-phosphate:polyprenyl phosphate GlcNAc-1-phosphotransferase participating in the N-glycosylation pathway.

AglH is essential for viability in S. acidocaldarius
Based on the obvious conservation of the different motifs within the cytoplasmic loops as well as the characteristic topology profile of AglH (Saci0093), we propose that algH indeed encodes a UPD-GlcNAc-1-P transferase initiating the protein N-glycosylation process by transferring GlcNAc-1-P from UDP-GlcNAc to DolP. To verify his hypothesis, we designed the plasmid pSVA1229, incorporating the up-and downstream region enclosing ΔaglH, to create a markerless deletion mutant of aglH in S. acidocaldarius by homologues recombination. Integration of this plasmid in S. acidocaldarius MW001 was confirmed by PCR using primers encompassing the flanking regions of the target gene. A second homologous recombination, enforced by the addition of 5-fluoroorotic acid, resulted in the segregation of the plasmid, generating either the wildtype stain or the preferred ΔaglH deletion strain. Screening of more than 150 different colonies by PCR using primers derived from the flanking regions revealed only the presence of wild-type genetic organization. This result suggests that AglH and N-glycosylation process is essential for the archaeon. Polar effects cannot be completely excluded; however, they are unlikely, given the fact that the integration of the plasmid could be achieved and this did not alter the archaeal phenotype. Furthermore, all essential genes for isoprenoid lipid biosynthesis are located upstream of the aglH. The segregation of the plasmid would not disturb the genetic neighbourhood, as the integrants did not show any evident alteration in growth.
To underline the essentiality of AglH activity, we explored the growth of S. acidocaldarius in the presence of antibiotics interfering specifically with the initiation of the N-glycosylation. Tunicamycin, a nucleoside antibiotic, resembles UDP-HexNAc residue and specifically inhibits UDP-HexNAc:polyprenol-P HexNAc-1-P transferases by blocking the active site (Esko and Bertozzi 2009). Bacitracin interferes with the dephosphorylation of isoprenyl pyrophosphate hindering the recycling of new isoprenyl phosphate lipid carrier (Stone and Strominger 1971). The addition of both antibiotics, after reaching early exponential growth phase, showed a dose-dependent reduction of growth of S. acidocaldarius (Fig. 4a, b). The addition of 0.3 mM bacitracin (t d = 10.7 ± 0.6 h) did not significantly reduce the growth compared to the wt doubling time (t d = 9.2 ± 0.9 h), whereas the addition of 1.2 and 2.4 mM bacitracin decreased the growth two-and fourfold (t d = 18.5 ± 2.5 h; t d = 42.1 ± 4 h), respectively. In addition, the addition of tunicamycin reduced the growth rate of S. acidocaldarius. Here, 2 µg/ml (t d = 13.6 h ± 1.6) and 6 µg/ml (t d = 16.4 h ± 1.2) led to a significant reduction of growth. Furthermore, the addition of 16 µg/ml increased the doubling time to 40.1 h ± 1.3 for the next 20 h. No growth was detected after 20 h in the samples treated with 16 µg/ml tunicamycin and with the two highest bacitracin concentrations. Interestingly, cells appeared larger compared to wt strain. Thus, the increased OD within the first hours may be attributed to an enlarged cell size. At 32 µg/ ml no growth could be detected (Fig. 4b). The effect of the antibiotics inhibiting N-glycosylation and thereby inhibiting cell growth was the most obvious when the antibiotics were directly added after inoculation (Fig. 4c). Here, first the OD 600 slightly increased to 0.1-0.15, then no growth could be detected over 70 h. Based on the reduced stability of the antibiotics (i.e. tunicamycin) at the growth condition (pH 3, 75 °C), the cells recover growth after 70 h. c Cell growth without antibiotics (filled circle), with 1.2 mM bacitracin (filled square), or 6 µg/ml tunicamycin (filled triangle), antibiotics were added directly after inoculation of the culture, indicated by black arrow

AglH restores growth of a conditional lethal alg7 yeast mutant in a complementation assay
To confirm the function of AglH, we tested whether this gene is able to complement a conditional lethal yeast alg7 mutant, as has been successfully shown for the human alg7 (Eckert et al. 1998) and aglH of M. voltae (Shams-Eldin et al. 2008). The full length algH gene was cloned into a yeast shuttle vector pRS426Met (Mumberg et al. 1995), containing the selective marker ura3 + , resulting in the plasmid pSVA1212. After transformation of this plasmid into the conditional lethal yeast alg7 mutant, YPH499-HIS-GAL-ALG7 (Mazhari-Tabrizi et al. 1999), transformants were plated on SGR as well as SD medium lacking histidine and uracil. Using plates containing either SGR medium (containing galactose) or SD medium (lacking galactose), in which the endogenous yeast alg7 is expressed or repressed, respectively. Cells transformed with the empty pRS426Met vector served as a negative control and displayed growth only on galactose-containing plates, i.e. conditions under which endogenous alg7 was expressed (Fig. 5). Under alg7 repression conditions, this strain failed to grow. Strains containing the plasmid carrying the human alg7 (positive control) (Eckert et al. 1998) or the aglH gene displayed sustained growth on glucose-containing plates, demonstrating the ability of the gene product to functionally replace the essential function of the yeast Alg7 enzyme (Fig. 5). These results showed that S. acidocaldarius AglH is indeed functional in S. cerevisiae and able to suppress the lethal phenotype of alg7 by catalysing the transfer of GlcNAc-1-P onto DolP.

Replacement of conserved amino acids resulted in a functional loss of AglH
The complementation assay was employed to verify the functional importance of distinct conserved amino acid residues in the predicted cytoplasmic loops (CL) of SaciAglH. The residues mutated were D 39 and K 42 in CL I, D 100 and N 103 in CL II, K 218 and F 220 in CL IV and F 264 in CL V; all were replaced individually by alanine. Mutations were introduced in the plasmid containing the aglH (Saci0093) gene fused to sequences encoding a FLAG epitope tag. The AglH-FLAG fusion protein is fully functional and allows growth of the conditional lethal yeast mutants on SD medium (Fig. 6). Mutations in CL I (D 039 A and K 042 A) had no effect on the function of the AglHFLAG protein. Also N 103 A and K 218 A mutations in CL II and CL III did not affect growth of the yeast mutant. In contrast, the strain complemented with the D 100 A showed no growth under conditions that repressed the yeast alg7 gene, implying that D 100 could be essential for catalysis. The F 220 A and F 264 A mutations in CL IV and CL V, respectively, had different effects. While the F 220 A mutant was substantially impaired in activity as indicated by very poor growth in the complementation assay (Fig. 6), the F 264 A mutant appeared inactive. To investigate the function of the C-terminal part of the enzyme, a stop codon was introduced at amino acid position 218 AglH Δ218-329 .The insertion of a stop codon prevented the ability to functionally complement the Alg7 phenotype. These results showed that the amino acids D 100 , F 220 and F 264 in AglH are important for successful complementation of the lethal phenotype of YPH499-HIS-GAL-ALG7.

Fig. 5
Rescue of a conditional lethal alg7 yeast mutant by the thermophilic aglH from S. acidocaldarius. YPH499-HIS-GALprom-ALG7 was transformed with the pRS426-METt plasmids carrying either the human ALG7 (Hs-alg7) or the archaeal Saci0093 (Saci-aglH). Transformed cells were streaked onto plates containing minimal medium lacking histidine and uracil and containing either galactose or glucose

Exchange of conserved amino acid residues did not change the expression level of AglH FLAG
To ensure that the AglH FLAG derivatives were expressed and to eliminate the possibility of reduced protein stability due to the amino acid replacements, protein expression was investigated. The membrane fraction of each of the mutated AglH FLAG transformed strains was examined by immuno blotting (Fig. 7). AglH was detected in the wild-type and all mutated AglH FLAG versions, showing that the lack of function observed in Fig. 7 for the D 100 A, F 220 A, and F 264 A mutations is not caused by reduced protein expression or degraded protein. The observed molecular weight of Agl-H FLAG corresponds to the calculated molecular weight of 36.6 kDa.

Discussion
Over the last decade, research on archaeal N-glycosylation pathways has revealed the diversity of the glycosylation process in terms of the involvement of specific GTases, as well as the structural and compositional complexity of archaeal N-glycans (Jarrell et al. 2014). In addition, these studies demonstrated a significant role of glycosylation for the stability of S-layer proteins as well as for the stability and the assembly of the archaellum, the archaeal motility structure (Calo et al. 2010;Yurist-Doutsch et al. 2008). In the thermoacidophilic crenarchaeon family Sulfolobales, most if not all surface-exposed proteins (Palmieri et al. 2013), including the S-layer protein , the archaellin Meyer et al. 2011), and sugar binding proteins (Elferink et al. 2001), are post-translationally modified by N-glycosylation. Defects in N-glycan biosynthesis resulted in a significant effect of the growth under elevated salt concentrations as well as a reduced motility .
To understand the N-glycosylation in S. acidocaldarius in more detail, we searched for the enzyme initiating this process. We identified aglH (saci0093), coding for a UDP-GlcNAc-1-phosphate:dolichyl phosphate GlcNAc-1-phosphotransferase, which shows high similarities in the overall topology as well as in the amino acid sequence with the eukaryotic Alg7, bacterial WecA, and archaeal homologs (Fig. 3). Based on these structural and sequence 1 3 similarities, we analysed whether AglH acts also as a functional homolog by complementation of a conditional alg7 yeast deletion mutant. Such complementation has been successfully demonstrated previously with the euryarchaeal AglH (Mv1751) from M. voltae (Shams-Eldin et al. 2008).
Here, we showed that also the thermophilic AglH enzyme complements defects in the biosynthesis of pyrophosphatelinked GlcNAc-isoprenoid units of Eukarya. Although the AglH from S. acidocladarius is adapted to high temperatures of around 75 °C, this enzyme is still able to complement the conditional lethal alg7 yeast mutant restoring the N-glycosylation process at 28 °C.
In addition to the ability of AglH to function in Eukarya, we could demonstrate that the substitution of conserved amino acid residues D 100 , F 220 , and F 264 leads to loss of function in the in vivo complementation assay in yeast (Fig. 6). The inability of the AglH D 100 A mutant to complementing the conditional yeast mutant is consistent with the substitution of the two aspartic acids DD 90/91 GG in WecA of E. coli, where the mutant was impaired in its activity in vitro and in an in vivo complementation system (Amer and Valvano 2002). It is proposed that the DDxxN/D motif and the conserved motif (NxxNxxxGxxGLxxG) of CL III catalyse the formation of the diphosphate linkage of DolPP-GlcNAc (Fig. 3), as the two conserved motifs show sequence similarity to the Walker B motif (Amer and Valvano 2002). The substitution of the highly conserved F 220 located in the CL IV leads to a drastic reduction in the ability of the AglH to functionally replace Alg7 in yeast. Although little growth could be detected on glucose-containing plates (non-permissive), which is most likely in response to a small amount of Alg7 remaining from the pre-culture on SRG medium, a significantly reduced activity of the AglH F 220 A is obvious. The data for this mutant align well with the corresponding F 249 L in eukaryotic UDP-GlcNAc:dolichol-P GlcNAc-1-P transferase (GTP), which resulted in a loss of more than half of the enzymatic activity (Dal Nogare et al. 1998). The function of the less conserved F 264 , is unclear, but the whole CL V might be involved in the in recognizing the nucleotide-activated sugar donor. Apart from proposed function of the CL V, the overall basic loop might be involved in the oligomerization or interaction with cytosolic GTases. Recently, Alg13/ Alg14, which catalyses the second step in eukaryotic N-glycan biosynthesis, was found to interact with Alg7. This interaction tethers the soluble Alg13/Alg14 GTases to the membrane of the ER (Lu et al. 2011), which allows the enhanced biosynthesis of the N-glycan by clustering the GTase reactions. As S. acidocaldarius possesses also a chitobiose core as the N-linking unit at the reducing terminus of its N-glycan like Eukarya (Peyfoon et al. 2010;Zahringer et al. 2000), a comparable assembly process might also be present in this archaeon. However, bioinformatics searches for an Alg13/Alg14 ortholog in S. acidocaldarius failed to identify an N-acetylglucosamine transferase. Deletion of aglH was not possible, implicating an essential role of this gene for the viability of S. acidocaldarius. This result is consistent with the inability to delete the key enzyme AglB of the N-glycosylation process, catalysing the transfer of the oligosaccharide onto target proteins . Treatment with antibiotics directly interfering with the function of AglH function furthermore strengthened the importance of AglH for cell survival (Fig. 4). Studies of the membrane-bound pyrophosphatase SepP (Saci1025) in S. acidocaldarius, recycling DolPP back into DolP precursor, revealed the necessity of this DolP cycle (Meyer and Schafer 1992). Here, supplementation of bacitracin interfering with the function of SepP into the growth medium led to the death of the cells, consistent with our findings (Fig. 4). Also treatment with tunicamycin has been previously shown to result in a reduced growth, a gradual increase in cell size, and cell death in S. acidocaldarius (Hjort and Bernander 2001). Furthermore, treatment of tunicamycin resulted also in a reduction in the relative amount of glycosylated proteins (Grogan 1996), underlining that this process is indeed important for generating N-glycans. However, at that time, no analyses on the lipid-linked glycans confirming the presence of DolPP-linked glycans in S. acidocaldarius or other crenarchaeota have been performed. Analyses of the lipid-linked glycans in euryarchaea revealed the presents of DolP rather than DolPP-linked oligosaccharide (Chaban et al. 2009;Guan et al. 2010;Taguchi et al. 2016). In Hfx. volcanii and in M. voltae, the enzymes AglK and AglJ have been shown to initiate the N-glycosylation process by creating DolP-linked glycans, respectively Chaban et al. 2009;Larkin et al. 2013). Interestingly, besides AglK, a homolog of AglH/Alg7/Dpagt1 has been identified in M. voltae (Chaban et al. 2006). A complementation assay in yeast, as it was done in this study, revealed that this AglH is able to replace the function of the eukaryal Alg7, which transfers GlcNAc-1-P from UDP-GlcNAc to DolP, yielding DolPP-GlcNAc (Shams-Eldin et al. 2008). Therefore, it was previously proposed that AglH initiates the N-glycosylation in M. voltae. However, lipid analyses as well as a detailed biochemical characterisation of AglK and AglH revealed that DolPP-Glc-NAc in M. voltae is not used for the N-glycan assembly, as only DolP-linked oligosaccharides have been detected (Larkin et al. 2013). The study furthermore showed that AglK and AglC, the second enzyme of the N-glycosylation pathway, synthesize DolP-GlcNAc-Glc-2,3-diNAcA. This DolP-linked disaccharide is transferred by AglB onto a target peptide in vitro, proving that DolP not only can act as an acceptor but also is used as the glycan donor in M. voltae (Larkin et al. 2013).
Here, we proposed that in contrast to euryarchaea the N-glycan in crenarchaeon Sulfolobus is assembled on DolPP, similar to Eukarya. In fact, two recent studies confirmed the presence of the DolPP-linked N-glycans in Sulfolobales (Guan et al. 2016;Taguchi et al. 2016). Interestingly the study of Taguchi et al. underlined the difference of the lipid carrier between cren-and euryarchaea, as in all analysed euryarchaea DolP-linked oligosaccharides were present, whereas in the crenarchaea P. calidifontis and S. solfataricus lipid-linked oligosaccharides were assembled on DolPP backbone (Taguchi et al. 2016). The use of DolPP like in Eukarya emphasises the close evolutionary relationship between Crenachaeota and Eukarya and strengthens the proposed origin of Eukarya from within the TACK super phylum, comprising also the crenarchaeota (Archibald 2008;Guy and Ettema 2011). Interestingly, a newly characterised archaeal phylum, termed Lokiarchaeota, proposed to be one of the deepest branched archaea forms a monophyletic group with eukaryotes in phylogenomic analyses. Within the genomes of these Lokiarchaeota subunits of the eukaryal oligosaccharyltransferase complex are found, i.e. OST3/OST6 (Spang et al. 2015), which had not been found in Archaea so far. This indicates that the archaeon, from which the eukaryotic cell emerged, might have possessed an already highly developed protein N-glycosylation pathway.
With this study, we have identified the enzyme that catalyses the first step of the protein N-glycosylation process in S. acidocaldarius. This will facilitate a more detailed understanding of the archaeal version of this posttranslational modification. Future analyses will be directed to identify additional components of the thermophilic crenarchaeal protein N-glycosylation process.