Archaeal glycoproteins present a variety of N-linked glycans not seen elsewhere. The ability to harness the agents responsible for this unparalleled diversity offers the possibility of generating glycoproteins bearing tailored glycans, optimized for specific functions. With a well-defined N-glycosylation pathway and available genetic tools, the haloarchaeon Haloferax volcanii represents a suitable platform for such glyco-engineering efforts. In Hfx. volcanii, the S-layer glycoprotein is modified by an N-linked pentasaccharide. In the following, S-layer glycoprotein N-glycosylation was considered in cells in which AglD, the dolichol phosphate mannose synthase involved in addition of the final residue of the pentasaccharide, was replaced by a haloarchaeal homologue of AglJ, the enzyme involved in addition of the first residue of the N-linked pentasaccharide. In the engineering strain, the S-layer glycoprotein is modified by a novel N-linked glycan not found on this reporter from the parent strain. Moreover, deletion of AglD alone and introduction of the AglJ homologue from Halobacterium salinarum, OE2528R, into the deletion strain resulted in increased biosynthesis of the novel 894 Da glycan concomitant with reduced biogenesis of the pentasaccharide normally N-linked to the S-layer glycoprotein. These findings justify efforts designed to transform Hfx. volcanii into a glyco-engineering ‘workshop’.
It is now clear that Archaea, like Eukarya and Bacteria, are capable of N-glycosylation, the covalent addition of glycan moieties to select asparagine (Asn) residues of target proteins. Indeed, archaeal glycoproteins reveal diversity in the composition of their N-linked glycan moieties not seen elsewhere (for review, see Eichler and Adams, 2005; Calo et al., 2010a). Combining those agents responsible for generating the broad diversity seen in N-linked glycan composition together with the stability of archaeal proteins in the face of physically challenging surroundings (Danson and Hough, 1998) could lead to the creation of glycoproteins designed to function optimally under specific environmental conditions. Such efforts are of biotechnological significance in light of reports assigning archaeal glycosylation roles in enhancing protein solubility in hypersaline conditions (Mengele and Sumper, 1992), in offering protection to proteins exposed to highly acidic surroundings (Zahringer et al., 2000) and in contributing to thermoarchaeal protein stabilization (Albers et al., 2004). Specifically, these and other desired traits could be enhanced upon introducing non-native glycans into proteins of interest.
With a relatively well-delineated N-glycosylation pathway and appropriate tools for genetic manipulation available (Allers and Mevarech, 2005), efforts to exploit Hfx. volcanii for the design of non-native N-linked glycans have begun. In initial glyco-engineering efforts, the ability of AglD homologues from Halobacterium salinarum, Haloarcula marismortui and Haloquadratum walsbyi to restore missing function to Hfx. volcaniiΔaglD cells confirmed the ability of introduced non-native glycosyltransferases to participate in the Hfx. volcanii N-glycosylation pathway (Calo et al., 2010b). In the present study, we have considered the effects of introducing a haloarchaeal homologue of AglJ, namely that predicted dolichol phosphate hexose synthase involved in adding the first residue of the N-linked pentasaccharide decorating the S-layer glycoproteins (Kaminski et al., 2010), into cells lacking Hfx. volcanii AglD, namely the mannosyl dolichol phosphate synthase involved in adding the final pentasaccharide residue (Guan et al., 2010). We report that the S-layer glycoprotein in Hfx. volcaniiΔaglD cells transformed to express Hbt. salinarum OE2528R presents an N-linked glycan profile distinct from that seen in the parent strain. These results offer further support to efforts aimed at developing Hfx. volcanii into a ‘glyco-factory’ capable of generating differentially glycosylated proteins.
The introduction of Hbt. salinarum OE2528R into Hfx. volcanii cells lacking AglD does not replace missing AglD function but affects N-glycosylation
Earlier efforts demonstrated the ability of haloarchaeal homologues of Hfx. volcanii AglD to restore that activity lost in Hfx. volcaniiΔaglD cells (Calo et al., 2010b). To now assess the impact of transforming Hfx. volcaniiΔaglD cells to express a homologue of AglJ, assigned as the other dolichol phosphate hexose synthase involved in N-glycosylation in this species (Kaminski et al., 2010), on S-layer glycoprotein N-glycosylation, Hbt. salinarum OE2528R (61% identical, 74% similar to Hfx. volcanii AglJ) was introduced into the deletion strain. Expression of the haloarchaeal AglJ homologue in Hfx. volcanii cells deleted of aglD was confirmed by immunoblot via detection of the Clostridium thermocellum cellulose-binding domain (CBD) fused to Hbt. salinarum OE2528R at the gene level, using anti-CBD antibodies. In the transformed strain, a band corresponding to the combined molecular weight of the plasmid-encoded AglJ homologue and the CBD tag was detected (approximately 50 kDa) (Fig. 2A). In a control sample, prepared from ΔaglD cells transformed with plasmid-encoded CBD-AglD, an approximately 85 kDa band was observed, corresponding to the molecular weight of the fusion protein.
To investigate whether Hbt. salinarum OE2528R expressed in Hfx. volcaniiΔaglD cells could functionally replace absent AglD, the transformed strain was grown in the presence of [2-3H] mannose, given the identification of mannose as the final residue of the N-linked pentasaccharide decorating the S-layer glycoprotein in the parent strain (Guan et al., 2010). In these experiments, the extent of [2-3H] mannose incorporation into the N-linked glycan was determined by SDS-PAGE and fluorography. Such analysis revealed the incorporation of radiolabel into the S-layer glycoprotein of cells of the parent strain and of ΔaglD cells transformed to express CBD-tagged AglD but not in cells lacking AglD (Fig. 2B), confirming the presence of mannose in the N-linked pentasaccharide decorating the S-layer glycoprotein and the importance of AglD for addition of this residue, as well as the fact that the presence of the CBD moiety does not interfere with enzyme function. In the case of Hfx. volcaniiΔaglD cells transformed to express CBD-tagged Hbt. salinarum OE2528R, no [2-3H] mannose-derived radioactivity was associated with the S-layer glycoprotein, reflecting that mannose is not an element of any glycans decorating the S-layer glycoprotein in this strain.
In cells lacking AglD, the S-layer glycoprotein migrates faster in SDS-PAGE than in the parent strain and can no longer be stained by PAS glycostain (Kaminski and Eichler, 2010). Previous studies revealed that both changes are reversed upon complementation of the mutant cells with a plasmid-encoded CBD-tagged version of AglD, as well as with CBD-tagged versions of the AglD homologues, Hbt. salinarum OE1482, Har. marismortui rrnAC1873 and Hqr. walsbyi HQ1489 (Calo et al., 2010b). However, in the case of Hfx. volcaniiΔaglD cells transformed to express CBD-tagged Hbt. salinarum OE2528R, only partial restoration of S-layer glycoprotein behaviour was observed. Specifically, although S-layer glycoprotein SDS-PAGE migration in the transformed strain was returned to that position seen for this protein from the parent strain (Fig. 2C, top panel), PAS glycostaining of the S-layer glycoprotein was not restored (Fig. 2C, bottom panel). In a control experiment, where Hfx. volcaniiΔaglD cells transformed to express CBD-tagged Hbt. salinarum OE2546F, a homologue of Hfx. volcanii AglI (51% identity, 68% similarity), neither SDS-PAGE migration nor PAS glycostaining of the S-layer glycoprotein was restored to what is seen in the native strain (not shown). As such, the effect on S-layer glycoprotein behaviour seen with the Hfx. volcaniiΔaglD strain transformed to express the AglJ homologue, Hbt. salinarum OE2528R, is unlikely an artefact of transformation but rather could reflect the presence of a modified version of the N-linked glycan decorating the S-layer glycoprotein, distinct from what is seen in cells of the parent or ΔaglD strains.
A novel N-linked glycan decorates the S-layer glycoprotein in Hfx. volcanii ΔaglD cells expressing Hbt. salinarum OE2528R
The pentasaccharide covalently linked to select Asn residues of the Hfx. volcanii S-layer glycoprotein comprises a hexose, two hexuronic acids, a methyl ester of a hexuronic acid and a mannose (Abu-Qarn et al., 2007; Guan et al., 2010; Magidovich et al., 2010). To determine whether Hfx. volcaniiΔaglD cells transformed to express the AglJ homologue, Hbt. salinarum OE2528R, modified the composition of this pentasaccharide or present a novel N-linked glycan, the S-layer glycoprotein from the transformed cells was digested with trypsin and analysed by liquid chromatography-electrospray ionization/mass spectrometry/mass spectrometry (LC-ESI/MS/MS). In agreement with earlier studies (Calo et al., 2010b), such analysis of a sample prepared from cells of the parent strain revealed the existence of a doubly charged [M+2H]2+ ion peak at m/z 1224.97, corresponding to the N-terminal 1ERGNLDADSESFNK14 fragment, with its Asn-13 residue modified by the pentasaccharide (Sumper et al., 1990; Abu-Qarn et al., 2007) (Fig. 3A). No such peak was detected in Hfx. volcaniiΔaglD cells transformed to express Hbt. salinarum OE2528R (Fig. 3D). However, both parent strain cells (Fig. 3B) and ΔaglD cells transformed to express Hbt. salinarum OE2528R (Fig. 3E) present a doubly charged [M+2H]2+ ion peak at m/z 1143.44, corresponding to the Asn-13-containing glycopeptide modified by the first four subunits of the pentasaccharide attached at this position (Sumper et al., 1990; Abu-Qarn et al., 2007; Guan et al., 2010). The correct assembly of the first four residues of the N-linked pentasaccharide at the level of the protein target [and, by extension, at the level of the dolichol phosphate carrier on which the tetrasaccharide is assembled (Guan et al., 2010)] verifies that native AglJ function is not compromised upon introduction of Hbt. salinarum OE2528R, an AglJ homologue, into the ΔaglD cells. Still, since the level of the m/z 1143.44 peak shown is normalized to that level seen in the parent strain, it is clear that less of the tetrasaccharide-charged glycopeptide is found in the transformed strain. In contrast, the absence of AglD had no effect on the level of this peak (not shown). As such, it appears that the decrease in the level of the m/z 1143.44 peak represents a Hbt. salinarum OE2528R-related effect on the processing of the pentasaccharide N-linked to the S-layer glycoprotein. More striking, however, is the observation that Hfx. volcaniiΔaglD cells transformed to express Hbt. salinarum OE2528R present a doubly charged [M+2H]2+ ion peak at m/z 1238.51 not seen in the parent strain (Fig. 3C and F).
To determine whether the doubly charged [M+2H]2+ ion peak at m/z 1238.51 corresponds to a S-layer glycoprotein-derived glycopeptide, tandem mass spectrometry was performed. Fragmentation of the m/z 1238.51 species resulted in doubly charged product ion peaks at m/z 1150.74, 1062.78 and 981.67, corresponding to the neutral loss of one hexuronic acid (i.e. a loss of 176 Da), two hexuronic acids (i.e. a loss of 352 Da), and two hexuronic acids and a hexose (i.e. a loss of 514 Da) respectively (Fig. 4A). Examination of the LC-ESI/MS profile indeed revealed that as was the case for the doubly charged [M+2H]2+ ion peak at m/z 1238.51, the engineered strain presents a doubly charged [M+2H]2+ ion peak at m/z 1150.49 not seen in the parent strain (Fig. 4B, top panels). A doubly charged [M+2H]2+ ion peak at m/z 1062.48 is also seen in the transformed strain. This peak is also noted in the parent stain, albeit to a much lower extent (Fig. 4B, bottom panels). No doubly charged [M+2H]2+ ion peak at m/z at 981.46 was detected in either strain. These results thus point to the S-layer glycoprotein-derived Asn-13-containing tryptic fragment (1580.7 Da) from Hfx. volcaniiΔaglD cells transformed to express Hbt. salinarum OE2528R as being modified by a 894 Da glycan comprising an N-linking 380 Da moiety of unknown structure, a hexose and two hexuronic acid residues (Fig. 4A, inset).
The novel N-linked 894 Da glycan is present inΔaglD cells but is substantially augmented in the presence of Hbt. salinarum OE2528R
The detection of a doubly charged [M+2H]2+ ion peak at m/z 1062.48 in cells of the parent strain implies that these cells, to some extent, modify the S-layer glycoprotein with the precursor of the N-linked 894 Da glycan containing the 380 Da moiety of unknown character and a hexose. Since the complete 894 Da glycan was not detected, it was speculated that the presence of AglD somehow interferes with the biosynthesis of the complete glycan in the parent strain. To test this hypothesis, the levels of the doubly charged [M+2H]2+ ion peak at m/z 1062.78, 1150.74 and 1238.51 in cells of the parent and ΔaglD strains were normalized to the levels of these peaks detected in ΔaglD cells transformed to express Hbt. salinarum OE2528R.
Comparison of the level of the doubly charged [M+2H]2+ ion peak at m/z 1062.78 in the three Hfx. volcanii strains revealed a slight increase in the ΔaglD strain (Fig. 5A, middle panel), relative to what is seen in the parent strain (top panel), but still only close to half of that amount seen in ΔaglD cells transformed to express Hbt. salinarum OE2528R (bottom panel). On the other hand, the ΔaglD strain contained just close to 50% of the doubly charged [M+2H]2+ ion peak at m/z 1150.74 seen in ΔaglD cells transformed to express Hbt. salinarum OE2528R (Fig. 5B, middle and bottom panels); only a negligible amount of this peak was detected in the parent strain (bottom panel). Finally, 10-fold more of the doubly charged [M+2H]2+ ion peak at m/z 1238.51 was detected in ΔaglD cells (Fig. 5C, middle panel) than was present in the parent strain (top panel), while ΔaglD cells transformed to express Hbt. salinarum OE2528R (bottom panel) in turn presented 10-fold more of this peak that did cells of the deletion strain. It thus appears that whereas the presence of AglD prevents the assembly of the complete 894 Da glycan N-linked to the S-layer glycoprotein, the replacement of AglD by Hbt. salinarum OE2528R substantially augments such assembly.
The ability to harness the various glycosyltransferases and other saccharide-processing enzymes responsible for generating the wide variability seen in the composition of N-linked glycans decorating archaeal glycoproteins (Eichler and Adams, 2005) carries enormous potential for the biosynthesis of glycoproteins bearing tailored glycans. With a relatively well-delineated N-glycosylation pathway (Calo et al., 2010a) and the availability of genetic tools for manipulating this strain (Allers and Mevarech, 2005), Hfx. volcanii offers a suitable platform for hosting such chimeric glycosylation systems. Here, the effects of introducing a Hbt. salinarum homologue of Hfx. volcanii AglJ, OE2528R, into the aglD deletion strain further justify the suitability of Hfx. volcanii as a glyco-engineering ‘workshop’.
The results show that introduced Hbt. salinarum OE2528R did not simply mediate the addition of a novel fifth residue to the glycan N-linked to the S-layer glycoprotein in place of the absent mannose residue normally found at this position. Instead, the presence of Hbt. salinarum OE2528R in the aglD deletion strain compromised the amount of N-linked tetrasaccharide corresponding to the first four residues of the pentasaccharide normally generated. This shows that while introduced Hbt. salinarum OE2528R has an effect on Hfx. volcanii N-glycosylation, it does not compete with AglJ, responsible for charging a dolichol phosphate with the first residue of the pentasaccharide (Kaminski et al., 2010). Rather, Hfx. volcaniiΔaglD cells transformed to express Hbt. salinarum OE2528R decorated the S-layer glycoprotein with a novel 894 Da N-linked glycan comprising two hexuronic acids, a hexose and a 380 Da component of unknown composition, with preliminary evidence suggesting this moiety to include an additional hexuronic acid residue. The presence of unidentified residues in N-linked glycans decorating archaeal glycoproteins has also been reported in Methanococcus voltae, where both the S-layer glycoprotein and flagellin are modified by an N-linked trisaccharide of defined composition that can include an additional, as yet unidentified, 220 or 262 Da residue (Chaban et al., 2009). The same glycan, like the N-linked glycan decorating glycoproteins in Methanococcus maripaludis, also includes several highly modified residues (Voisin et al., 2005; Kelly et al., 2009).
Although the novel 894 Da N-linked glycan noted in ΔaglD cells transformed to express Hbt. salinarum OE2528R was not detected on the S-layer glycoprotein from cells of the Hfx. volcanii parent strain, the S-layer glycoprotein was modified by an early 542 Da precursor of this glycan in parent strain cells. More of this and more advanced precursors of the 894 Da N-linked glycan were shown to decorate the S-layer glycoprotein of cells deleted of aglD. Indeed, trace amounts of the complete glycan were detected in cells of the ΔaglD strain, albeit at levels some 20-fold less than seen in ΔaglD cells transformed to express Hbt. salinarum OE2528R. At the same time, the expression of Hbt. salinarum OE2528R in ΔaglD cells led to a decrease in the level of the tetrasaccharide-bearing precursor of the pentasaccharide decorating the S-layer glycoprotein in the parent and ΔaglD strains. These observations suggest that in the absence of AglD, the activities of other enzymes involved in N-glycosylation, namely those involved in the biogenesis of the 894 Da glycan, are encouraged. This effect is, moreover, enhanced when Hbt. salinarum OE2528R is introduced into the deletion strain. It would thus appear that the presence or absence of AglD affects the relative importance or two different N-glycosylation pathways. Introduced Hbt. salinarum OE2528R, seemingly catalysing a reaction performed to a much lesser extent by a native, as yet unidentified N-glycosylation enzyme, subsequently shifts this balance substantially. Recently, examination of the N-linked glycan decorating M. maripaludis pili distinguished it from that glycan N-linked to the S-layer glycoprotein and flagellin in the same species through the presence of an additional hexose branch, pointing to the ability of Archaea to generate different N-linked glycans in a single species (Ng et al., 2011). In the case of the Hfx. volcanii S-layer glycoprotein, the results presented here point to different N-linked glycans being added to the same protein as a function of the presence or absence of certain N-glycosylation pathway enzymes. The Hbt. salinarum S-layer glycoprotein is also modified by two distinct N-linked glycans, although nothing is known of the biosynthesis of these glycans (Lechner and Wieland, 1989).
Based upon analysis of the dolichol phosphates upon which N-linked glycans are assembled in Hfx. volcanii (Guan et al., 2010; Kaminski et al., 2010), the existence of different N-linked glycans in this species is not unexpected. Preliminary experiments observed differences in the profile of the glycan-charged dolichol phosphate pool in Hfx. volcanii cells including or lacking AglD (Z. Guan and J. Eichler, unpubl. obs.). Earlier analysis of the Hfx. volcanii dolichol phosphate pool reported the presence of species modified by glycans that included sulfated or phosphorylated hexoses and the deoxyhexose, rhamnose (Kuntz et al., 1997), although none of these glycans was shown to be N-linked to the S-layer glycoprotein, the major glycoprotein in this species (Sumper et al., 1990; Abu-Qarn and Eichler, 2006). It is conceivable that these dolichol phosphate-bound glycans contribute to the 380 Da component of the novel 894 Da N-linked glycan reported here.
In conclusion, the finding that a non-native homologue of a Hfx. volcanii N-glycosylation pathway glycosyltransferase can modify the glycosylation of a reporter glycoprotein in a novel manner offers encouragement for efforts to develop Hfx. volcanii into a glyco-engineering platform. At the same time, this study also makes it clear that much remains to be learned about N-glycosylation in Hfx. volcanii.
Strains and growth conditions
The Hfx. volcanii parent strain WR536 (H53) (obtained from M. Mevarech, Tel Aviv University), the same strain deleted of aglD and the deletion strain transformed to express C. thermocellum cellulose-binding domain-tagged Hbt. salinarum OE2528R were grown in complete medium containing 3.4 M NaCl, 0.15 M MgSO4·7H2O, 1 mM MnCl2, 4 mM KCl, 3 mM CaCl2, 0.3% (w/v) yeast extract, 0.5% (w/v) tryptone, 50 mM Tris-HCl, pH 7.2, at 40°C (Mevarech and Werczberger, 1985).
The preparation of Hfx. volcanii strains deleted of aglD was previously reported (Abu-Qarn and Eichler, 2006), as was preparation of the aglD deletion strain transformed to express CBD-AglD (Calo et al., 2010b). Haloferax volcaniiΔaglD cells transformed to express CBD-tagged Hbt. salinarum OE2528R were generated by introducing plasmid pWL-CBD-OE2528R into the deletion strain. To generate plasmid pWL-CBD-OE2528R, genomic DNA was isolated from Hbt. salinarum R1 (obtained from D. Oesterhelt, Max Planck Institute) and OE2528R was PCR-amplified using primers OE2528for (CCCAGATCTATGAGCGAGGAGTACGAG) and OE2528rev (CCCGGTACCTCACTTGTTCAGGCGC), designed to introduce BglII and KpnI sites at the 5′ and 3′ ends of the amplified sequence respectively. The amplified fragment was then ligated into plasmid pWL-CBD-AglD (Plavner and Eichler, 2008), pre-treated with BglII and KpnI so as to remove the aglD insert.
LC-ESI/MS/MS analysis of the Hfx. volcanii S-layer glycoprotein was performed as previously described (Calo et al., 2010b).
[2-3H]-mannose radiolabelling of the S-layer glycoprotein
For [2-3H]-mannose radiolabelling of the S-layer glycoprotein, 100 µl aliquots of Hfx. volcanii cells were incubated with 6 µl of [2-3H]-mannose (23.8 mCi mmol−1; PerkinElmer, Boston, MA). One hour later, the protein content of the cells was precipitated with 15% (w/v) trichloroacetic acid and separated by SDS-PAGE. The incorporation of radiolabel was revealed following fluorography and exposure to film.
Proteins bearing the CBD moiety were detected by immunoblot, using polyclonal antibodies raised against the C. thermocellum cellulose-binding domain (obtained from Ed Bayer, Weizmann Institute of Science; 1:10 000). Antibody binding was detected using goat anti-rabbit horseradish peroxidase (HRP)-conjugated antibodies (1:2500, Bio-Rad, Hercules, CA) and an ECL enhanced chemiluminescence kit (GE Healthcare, Piscataway, NJ). Periodic acid-Schiff reagent (PAS) glycoprotein staining was performed as described previously (Dubray and Bezard, 1982).
The authors thank Rachel G. Lichtenstein and Yael Eilam for assistance in the early stages of the research and Christian Raetz for support of Z.G. J.E. is supported by the Israel Science Foundation (Grant 30/07). The mass spectrometry facility in the Department of Biochemistry of the Duke University Medical Center and Z.G. are supported by the LIPID MAPS Large Scale Collaborative Grant No. GM-069338 from NIH.