Exploring the Unique N-Glycome of the Opportunistic Human Pathogen Acanthamoeba*

Background: Acanthamoeba is a facultative parasite of humans capable of causing keratitis or encephalitis. Results: Novel N-glycan modifications, including core mannosylated fucose, mannose 6-phosphate, and multiple pentose residues, were observed. Conclusion: Acanthamoeba is predicted to possess all ER glycosyltransferases involved in N-glycosylation as well as unusual fucosyl- and pentosyltransferases in the Golgi. Significance: Examining Acanthamoeba glycosylation may yield insights into virulence factors for this genus of therapeutically resistant parasites. Glycans play key roles in host-pathogen interactions; thus, knowing the N-glycomic repertoire of a pathogen can be helpful in deciphering its methods of establishing and sustaining a disease. Therefore, we sought to elucidate the glycomic potential of the facultative amoebal parasite Acanthamoeba. This is the first study of its asparagine-linked glycans, for which we applied biochemical tools and various approaches of mass spectrometry. An initial glycomic screen of eight strains from five genotypes of this human pathogen suggested, in addition to the common eukaryotic oligomannose structures, the presence of pentose and deoxyhexose residues on their N-glycans. A more detailed analysis was performed on the N-glycans of a genotype T11 strain (4RE); fractionation by HPLC and tandem mass spectrometric analyses indicated the presence of a novel mannosylfucosyl modification of the reducing terminal core as well as phosphorylation of mannose residues, methylation of hexose and various forms of pentosylation. The largest N-glycan in the 4RE strain contained two N-acetylhexosamine, thirteen hexose, one fucose, one methyl, and two pentose residues; however, in this and most other strains analyzed, glycans with compositions of Hex8–9HexNAc2Pnt0–1 tended to dominate in terms of abundance. Although no correlation between pathogenicity and N-glycan structure can be proposed, highly unusual structures in this facultative parasite can be found which are potential virulence factors or therapeutic targets.

encephalitis, the latter being nearly always fatal in patients with underlying co-morbidities. Acanthamoeba keratitis is nowadays commonly associated with the wearing of contact lenses. Indeed, although high amoebal densities brought into the eye are more common but not limited to individuals wearing contact lenses, contact lens wearers often have microlesions in the cornea, which facilitate the invasion of the amoebae into deeper layers. The potential for affecting humans is intensified due to its ubiquitous presence in various environmental sources from which the amoeba can be isolated, including public water supplies, bottled water, and the atmosphere (1,2). Acanthamoeba trophozoites are capable of encystment, a process triggered by harsh environmental conditions during which a double cyst wall is formed and metabolism is reduced to form dormant cysts, which often renders treatment of Acanthamoeba keratitis ineffective. Furthermore, the denomination of species in the Acanthamoeba genus was highly disputed as it was based on morphological details subject to variability depending on environmental factors (3); therefore, a system of 12 genotypes was established in 1996 (4), which has meanwhile been expanded to 17 genotypes (5,6). The high genetic variability of the genus is reflected by the fact that no single strain originates either from multiple patients (unless family members) or from more than one environmental source.
A vital step in the process of invading a host is often the attachment of parasites to epithelia, which in several cases is associated with lectin-oligosaccharide interactions. Examples of such include the host-pathogen interactions of other protozoa, including the dysentery-causing obligate parasite Entamoeba histolytica and the sexually transmitted Trichomonas vaginalis. Indeed, a lectin-based invasion mechanism is known for E. histolytica, specifically involving attachment to the colonal epithelia via a glycosylated Gal-and GalNAc-binding lectin (7), whereas in the case of T. vaginalis, epithelial host galectin-1 binds to the lipoglycan of the parasite (8). In the case of the establishment of Acanthamoeba keratitis, a potentially glycosylated mannose-binding lectin of the parasite mediates attachment to the host cornea (9).
In this study we screened the N-glycomic potential of eight strains, including four clinical isolates, representing five genotypes of the genus Acanthamoeba, prior to more in-depth analyses on one strain (4RE). Although there are various earlier reports on a partial characterization of lipophosphonoglycan (10) and the presence of a N-acetylglucosaminyl-1-phosphotransferase (11) as well as various radiolabeling (12), lectin binding (12)(13)(14)(15) or monosaccharide composition studies (16), our present data are the first regarding the actual structures of glycosylated macromolecules of this genus. As it has been shown that many protozoal parasites synthesize highly unique N-glycans (17), it can be hypothesized that the investigation of the N-glycosylation of parasitic and nonparasitic Acanthamoeba strains may yield insights into virulence factors.
Glycan Preparation-N-Glycans were prepared from axenically grown Acanthamoeba by enzymatic release of glycans from partially purified glycopeptides using either PNGase 2 A or F (Roche Applied Science) as described previously for Dictyostelium (28). Generally, cells were taken up in formic acid (up to 5%) denatured at 100°C for 10 min and lysed using a glass homogenizer; the lysate was then subjected to proteolysis over night using 1 mg of pepsin per g wet weight of cells. Thereafter the proteolysate was applied to Dowex 50 cation-exchange resin, and material eluted with 0.5 M ammonium acetate (pH 6) was desalted (Sephadex G-25) prior to addition of either PNGase A or F (Roche Applied Science). The released N-glycans, which did not bind a second Dowex 50 column, were pyridylaminated overnight prior to gel filtration (Sephadex G-15) (29); further analysis by MALDI-TOF MS and HPLC was performed as described below.
Glycan Analysis-The N-glycome of the 4RE strain was fractionated by either reversed phase (RP) HPLC (Agilent Hypersil ODS, 4 ϫ 250 mm; gradient of 0.3% methanol/min using 0.1 M ammonium acetate (pH 4) as buffer; 1.5 ml/min) or normal phase (NP) HPLC (Tosoh TSKgel Amide-80, 4.6 ϫ 250 mm; gradient from 71.25% to 61.75% acetonitrile over 20 min followed by a gradient from 61.75% to 47.5% acetonitrile for a further 45 min using 0.01 M ammonium formate (pH 7) as buffer; 1 ml/min). Glycans were, respectively, detected by fluorescence with excitation/emission wavelengths of 320/400 in the case of RP-HPLC or 310/380 nm in the case of NP-HPLC (29,30). Selected fractions were subjected to two-dimensional HPLC with NP-HPLC in the first dimension and RP-HPLC in the second. Both columns were calibrated in terms of glucose units using a pyridylaminated dextran hydrolysate standard; comparisons were also made with the elution times of pyridylaminated Man 5 GlcNAc 2 and Man 9 GlcNAc 2 (Takara) as well as the endoplasmic reticulum mannosidase I digestion product (Man 8 GlcNAc 2 ) of the latter. For comparative examination of the properties of a ␤1,2-xylosylated plant glycan (Man 3 GlcNAc 2 Xyl), a relevant standard was isolated after PNGase F digestion of an pepsinized extract of white beans.
Further analysis by MS was performed after treatment overnight with 0.2 l of either jack bean ␣-mannosidase (Sigma-Aldrich) and/or bovine kidney ␣-fucosidase (Sigma-Aldrich) or recombinant endoglycosidase H (Roche Applied Science) or Aspergillus ␣1,2-mannosidase (Prozyme) in 50 mM ammonium acetate buffer (pH 5) at 37°C. Monoisotopic MALDI-TOF MS was performed using a Bruker Ultraflex or Autoflex Speed TOF-TOF instrument with 2,5-dihydroxybenzoic acid or 6-aza-2-thiothymine as matrix; MS/MS was performed by laser-induced dissociation. Spectra were processed with the manufacturer's software (Bruker Flexanalysis 3.3.80) using the SNAP algorithm with a signal/noise threshold of 6 for MS (unsmoothed) and 3 for MS/MS (smoothed four times). Thereafter, mass spectra were analyzed manually; after an initial assessment of each entire spectrum on the basis of known glycan masses and putative glycan series, each individual HPLCpurified fraction was subjected to MS and MS/MS to determine the presence of key fragments (e.g. those corresponding to core fucose, pentosylated and methylated residues) and fragment series. These manual analyses validated the proposed calculated composition and allowed the prediction of structures, which in part could be further verified by exoglycosidase digestion or HPLC elution times. For semi-quantitative comparison, the relative intensities of the 10 most abundant glycans from each strain were calculated with reference to the MS peak intensity of the Hex 9 HexNAc 2 glycan (set as 100).
Permethylation of the major two-dimensional HPLC-purified N-glycan was achieved by a modification of the "Methylation with Sodium Hydroxide and Methyl Iodide in Dimethyl Sulfoxide with the Presence of Water" protocol from Ciucanu and Costello (31). Dry glycans were dissolved in dimethyl sulfoxide with a trace of water; after addition of powdered NaOH dissolved in dimethyl sulfoxide and subsequent shaking, deuterated iodomethane (Sigma-Aldrich) was added to a final concentration of 3 mM, and the mixture was incubated at room temperature for 10 min. The liquid phase was transferred to a fresh reaction tube where the reaction was quenched with water and neutralized with 0.1 M HCl. Isolation of the perdeuteromethylated products was achieved by repeated extraction with chloroform. The sample was taken up in 50% methanol in 1 mM sodium acetate, and 2,5-dihydroxybenzoic acid was used as matrix for MALDI-TOF MS and MS/MS analysis.

RESULTS
Glycomic Screening of Eight Acanthamoeba Strains-To examine the glycomic potential of Acanthamoeba, we initially tested cell extracts of trophozoites of eight selected strains (see Table 1) by Western blotting with a range of antibodies and lectins. From these blots we concluded that Acanthamoeba is able to synthesize glycosidic structures with varying monosaccharide composition. In particular, we first investigated the ability of its proteins to bind to anti-horseradish peroxidase (anti-HRP), an antibody raised against a common plant glycan structure (32), which recognizes epitopes in a range of plant, invertebrate, and protist organisms (33). Positive results from the anti-HRP blots led to the conclusion that most of the selected strains synthesize either core ␣1,3-fucose and/or ␤1,2xylose epitopes on glycoproteins across the molecular mass range (Fig. 1A). Reactivity to fucose-recognizing lectins AAL and LCA, a GalNAc-recognizing lectin (VVA) and galactoserecognizing lectins ECL and RCA, was also widespread among the strains, whereas generally trace reactivity toward WGA was observed (supplemental Fig. 1). Furthermore, reactivity to a single-chain antibody specifically recognizing mannose 6-phosphate residues was apparent particularly toward proteins of Ͻ50 kDa (Fig. 1B). To gain further insight into the N-glycosylation potential of these Acanthamoeba strains, it was therefore necessary to isolate the protein-bound N-glycans from cell pellets and use chromatographic and mass spectrometric means of analysis.
MALDI-TOF MS and MS/MS of N-glycans released by PNGase F were employed to screen the eight strains for the TABLE 1 Acanthamoeba strains used in this study Genotypes T4, T5, T6, T7, and T11 are based on the phylogeny of rRNA sequences, whereas the morphological groups are based on cyst morphologies. Pathogenic (P) and non-pathogenic (NP) strains are defined under consideration of whether the isolate was from a clinical source (with associated disease) or an environmental source (including, in the case of 4RE, a lens case of a subject with no keratitis).  range of structures present; these data, summarized in Table 2 and supplemental Fig. 2, suggest that Acanthamoeba is capable of producing N-glycans not previously found in protozoans. The major structures in most strains are large mannosidic glycans with 7 to 10 hexoses with or without a pentose residue (Hex 7-10 HexNAc 2 Pnt 0 -1 ). Little difference was apparent when comparing PNGase A and F digests in terms of the degree of modification with deoxyhexose residues (supplemental Fig. 2); furthermore, monosaccharide composition data indicated the presence of xylose and fucose (data not shown). Therefore, we conclude that the pentose residues are, in part, ␤1,2-xylose residues capable of binding anti-HRP, whereas the deoxyhexose residues are assumed to be, in part, core ␣1,6-fucose residues.

Strain
Other than varying amounts of larger glycans (m/z 2200 -3000) of differing composition, strains of the genotype T4 (ATCC 30234, Neff, IBU, and PAT06) exhibit similar glycomic profiles. The glycan profile of strain 11DS shows large similarity to those of genotype T4 strains, which correlate with the fact that this T6 strain exhibits many traits associated with genotype T4 (20). Analysis of strain 72/2 (genotype T5), the only group III strain tested, revealed N-glycosylation similar to those of the T4 strains, with a shift to larger structures. A completely different pattern was detected for strain Pb30/40 (genotype T7, morphological group I), which synthesizes glycans of smaller mass with the composition Hex 5-7 HexNAc 2 Pnt 0 -5 . These characteristics render Pb30/40 unique among all eight trophozoite samples examined and fit very well with the fact that group I is only very distantly related to group II and group III (34).
The overall profile of the glycans from group II strain 4RE (genotype T11) revealed close similarity to the T4 strains, with the exception of additional rather large structures, ranging in mass from 2400 to 3100 Da, containing 10 to 13 hexoses, 2 pentoses, and maximally 1 fucose and/or methyl groups. Due to these peculiarities but also the relative simplicity of its glycans in terms of numbers of nonhexose residues, the PNGase F-released glycans of this strain were chosen for further analysis.
Complete N-Glycome Digestions-The complete pyridylaminated N-glycan pool of the 4RE strain was subject to various enzymatic treatments prior to MALDI-TOF MS (Fig. 2). Jack bean ␣-mannosidase, Aspergillus ␣1,2-mannosidase, and endoglycosidase H were employed. Jack bean mannosidase digestion resulted in a significant shift in the N-glycome with the smallest glycans being Hex 1 HexAc 2 Pnt 0 -1 (m/z 665 and 797); especially the glycans of up to Hex 9 HexNAc 2 Pnt 1 were affected by this treatment, whereas the largest glycans of between 2200 and 3100 Da were rather resistant, and maximally one hexose was removed from such large glycans (e.g. from the largest glycan of m/z 3056). Endoglycosidase H removed all oligomannosidic glycans but not pentosylated ones; Aspergillus ␣1,2-mannosidase had, like the jack bean mannosidase, a significant effect, resulting in a large increase in Hex 5 HexNAc 2 , compatible with the presence of three or four ␣1,2-linked mannose residues on Hex 8 -9 GlcNAc 2 , but had no effect on the larger glycans. In the case of other strains, combined jack bean mannosidase and endoglycosidase digestion results in similar shifts in the overall spectra, but again the larg- The data from initial glycomic screening are summarized to show m/z values and the predicted composition of the 10 major N-glycans derived from PNGase F digests of glycopeptides from each of eight different Acanthamoeba strains; MS/MS was employed to predict the composition. As a means of semi-quantitative comparison among strains we calculated relative peak intensities of nine major glycans compared with the intensity of Hex 9 HexNAc 2 (m/z 1983), a glycan present in all strains (the relative intensity of this glycan is normalized to be 100). The glycans are abbreviated in the form of H v N w P x F y Me z , where H represents hexose (⌬m/z 162), N N-acetylhexosamine (⌬m/z 203), P pentose (⌬m/z 132), F fucose (⌬m/z 146), and Me a methyl group (⌬m/z 14). The genotypes (T4, T5, T6, T7, and T11) and the pathogenic/nonpathogenic (P or NP) status of each strain are also indicated. est glycans appeared to be also generally unaffected (supplemental Fig. 3), suggestive of the presence of terminal residues other than mannose (e.g. glucose, methylated hexose, or pentose). For further structural information, the pyridylaminated N-glycans of the 4RE strain were fractionated by NP-and RP-HPLC (supplemental Figs. 4 and 5) prior to further mass spectrometric analyses.
Oligomannosidic Glycans-Among the common, but not most abundant, N-glycans of the 4RE strain are a series of endoglycosidase and mannosidase-sensitive structures (Hex 5-9 HexNAc 2 ); apparently these can be mainly digested down to a Man 1 GlcNAc 2 stub with m/z 665 as seen in the overall jack bean mannosidase-treated profile or to Man 5 GlcNAc 2 (m/z 1313) with the fungal ␣1,2-specific enzyme (Fig. 2). The putative Man 8 GlcNAc 2 species elutes at approximately 5 glucose units on RP-HPLC, just prior to the putative Man 9 GlcNAc 2 (Supplemental Fig. 4B); this elution order, comparable with literature values (35) and observed when analyzing Man 9 GlcNAc 2 and its endoplasmic reticulum ␣-mannosidase I digestion product (36), would be compatible with the major Man 8 GlcNAc 2 isomer being Man8B. The putative Man 5 GlcNAc 2 in the untreated samples eluted at 7.2 glucose units (supplemental Fig. 4B), thereby co-eluting with a com-mercial standard and indicating that this glycan corresponds to the normal isomer resulting from Golgi processing.
Core Fucosylation and Hexose Capping of Fucose-Among the range of glycans from the 4RE strain whose mass indicated the presence of deoxyhexose residues, one of the least complex is Hex 6 HexNAc 2 Fuc 1 (m/z 1621 as [MϩH] ϩ ). In addition, fucose residues were found attached to glycans with additional pentoses and methyl groups (see Table 2 for details). Many fucosylated structures from the 4RE strain gave rise not to the m/z 446 ion (GlcNAc 1 Fuc 1 -PA) during fragmentation, but to one of m/z 608 (Fig. 3A). This fragment indicates a hexose being bound to the core fucose. The Hex 6 HexNAc 2 Fuc 1 glycan (released with PNGase F) was sensitive to jack bean ␣-mannosidase treatment, which resulted in loss of up to five hexose residues (smallest product being of m/z 811) and, upon MS/MS of three of the products, a loss of the m/z 608 fragment ion accompanied by a dominance of an m/z 446 fragment indicative of core fucosylation (Fig. 3B shows the MS/MS of the m/z 973 product). Fucosidase treatment had no effect unless preceded by incubation with ␣-mannosidase; a number of products, the smallest of which is of m/z 665, lacked any MS/MS hallmarks of fucosylation as judged by the presence of the m/z 300 fragment (Fig. 3C). However, Aspergillus ␤-galactosidase had no effect on this glycan (data not shown). Under the assumption that the  DECEMBER 21, 2012 • VOLUME 287 • NUMBER 52 jack bean mannosidase contains no further hexosidase activities, these data would indicate that the structure with m/z 1621, based on a typical Man 5 GlcNAc 2 glycan, is decorated with a core ␣1,6-fucose capped with an ␣-linked mannose. Similar data on the mannosidase sensitivity of the core hexosylfucose motif were obtained for Hex 6 HexNAc 2 Fuc 1 from the ATCC 30234 strain. A number of larger glycans from some Acanthamoeba strains are also predicted to carry such a unique epitope (see below).

N-Glycans of Acanthamoeba
Phosphorylation of Mannose Residues-The reactivity of Acanthamoeba lysates with the scFV M6P-1 antibody (Fig. 1B) directed us to examine for the presence of potentially phosphorylated glycans in the 4RE strain. A phosphorylated hexose would give rise to a fragment of m/z 243 in positive-ion mode; this fragment, in combination with two other potentially phosphorylated fragments of m/z 405 (Hex 2 Phos) and 1069 (Hex 3 HexNAc 2 Phos-PA), was found when analyzing a glycan of m/z 1717. Thus, this species has a predicted composition of Hex 7 HexNAc 2 Phos-PA (Fig. 4).
Methylation of Hexose Residues-MS/MS of a number of glycans isolated from the 4RE strain revealed fragments of m/z 177 and 339, which are suggestive of methylation of hexose residues. Examples of methylated glycans include pentosylated and fucosylated structures, including higher molecular mass glycans of the form Hex 10 -13 HexNAc 2 Pnt 2 Fuc 0 -1 Me.
Other example fragments include m/z 542 and 841 (Hex 2 HexNAc 1 Me 1 and Hex 2 HexNAc 2 Me 1 -PA) as shown here for glycans with the composition Hex 8 HexNAc 2 Me 1 and Hex 9 HexNAc 2 Pnt 2 Me (Fig. 5). Based on these fragments, we presume that the methylated residue in these cases is one of the ␣-linked mannose residues attached directly to the core ␤-mannose.
Structural Analysis of Pentosylated Glycans-Whereas in other strains a greater degree of pentosylation is possible (Table  2), maximally two pentose residues were observed in the glycans of the 4RE strain. The major structures (Hex 8 -10 -HexNAc 2 Pnt 1 ) are assumed to be novel compared with the known literature; bearing in mind that all examined glycans of this composition elute in multiple RP-HPLC peaks, we conclude that Acanthamoeba is capable of synthesizing several isomers of these structures. Various putatively pentosylated fragments, such as m/z 295 (HexPnt), 336 (HexNAcPnt), 635 (HexNAc 2 Pnt-PA), 797, 959, 1121, and 1283 (Hex 1-4 -HexNAc 2 Pnt-PA) were detected. These could be indicative of the presence of pentose close to or in the core region, but are ambiguous; thereby, pentose could in theory be either on the ␤-linked mannose of the glycan core, as found in plants (which present fragments of m/z 797 and 959; see below), or on the second GlcNAc of the chitobiose unit, as found on N-glycans of T. vaginalis (which present fragments of m/z 336 and 635 (29)), or on a terminal ␣-mannose as in a microalga (37); rearrangements during fragmentation can also not be ruled out (38). This large degree of ambiguity compounded by the presence of many isomers makes it difficult to arrive at definitive structural propositions. However, the MS/MS spectrum of a glycan of Hex 9 HexNAc 2 Pnt 2 Me (Fig. 5B) shows the presence of a pentosylated fragment containing a methylated hexose (m/z 674; Hex 2 HexNAc 1 PntMe) as well as a fragment of m/z 797 (HexHexNAc 2 Pnt-PA), but not of m/z 635 (HexNAc 2 Pnt-PA); these data suggest strongly that the xylose is linked to the ␤-linked mannose of this particular glycan.
The most dominant quasimolecular ion in the whole glycome spectrum of 4RE is of m/z 1931 in its protonated form, corresponding to Hex 8 HexNAc 2 Pnt 1 -PA. Species of this composition are enriched in the NP-HPLC fraction IX and were further purified by RP-HPLC. Digestion of the resulting major two-dimensional purified glycan with jack bean mannosidase resulted in a ladder of products (compare Fig. 6, A and B). MS/MS of the untreated glycan and its products revealed a fragment of m/z 635 which would be compatible with a composition of HexNAc 2 Pnt 1 -PA (Fig. 6, C and D) and differs from the fragmentation pattern of an authentic Man 3 GlcNAc 2 Xyl 1 glycan from beans (Fig. 6E). Perdeuteromethylation of the twodimensional HPLC-purified Hex 8 HexNAc 2 Pnt 1 -PA resulted in ions of m/z 2510 and 2527 as well as their sodium adducts 2532 and 2549 (Fig. 6F). It appears that, in this and other perdeuter-   DECEMBER 21, 2012 • VOLUME 287 • NUMBER 52

JOURNAL OF BIOLOGICAL CHEMISTRY 43197
omethylation experiments, the pyridylamino function may be either mono-or dipermethylated, resulting in a major fragment of m/z 367 and a minor one of m/z 384; furthermore, a fragment of m/z 787 would be compatible with a perdeuteromethylated form of HexNAc 2 Pnt 1 -PA, and that of m/z 2143 would result from loss of the reducing-terminal GlcNAc-PA (Fig. 6G). As this form of Hex 8 HexNAc 2 Pnt 1 is the dominant structure in 4RE, addition of pentose to the distal core GlcNAc is the major form of pentosylation in this strain, whereas xylosylation of the core mannose is relatively rare.
Analysis of Larger N-Glycans-Larger glycans containing xylose and often deoxyhexose and/or methyl residues are also found in the 4RE strain ( Table 2). MS/MS of two-dimensional HPLC-purified fractions of these glycans suggested the presence, in many cases, of hexosylated core fucose as judged by the presence of the m/z 608 fragment (Fig. 7, A-F), similar to the case of the Hex 6 HexNAc 2 Fuc glycan described above. Two of the largest structures, occurring in the same NP-HPLC fraction (XIV), had predicted compositions of Hex 12 HexNAc 2 Pnt 2 Fuc 1 and Hex 13 HexNAc 2 Pnt 2 Fuc 1 Me (m/z 2858 or 3034 as [MϩH] ϩ ). Attempts at digesting these glycans with glycosidases were largely unsuccessful, which could be due to putative terminal glucosylation or pentosylation as well as steric hindrance; however, a mannosidase-mediated unveiling of the core fucose was observed for these glycans, as shown by the loss of the m/z 608 MS/MS fragment and its replacement by one of m/z 446. Subsequent fucosidase digestion resulted, in turn, in the loss of the m/z 446 fragment and the appearance of a strong one of m/z 300 (Fig. 7, G-I).
Multiple Pentosylation in a Nonpathogenic Strain-Glycans of the nonpathogenic Pb30/40 strain, the only example here of a group I Acanthamoeba, were of interest due to the high degree of pentosylation and the apparent lack of fucose and methyl groups. Indeed, up to five pentose residues on a Man 7 GlcNAc 2 structure are postulated; complete glycome digestion suggested that a number of mannose residues could be released from such glycans to yield Hex 4 HexNAc 2 (supplemental Fig. 3). Therefore, the question was as to how four or five pentose residues could be attached to a putative Man 4 GlcNAc 2 "core." From the MS/MS data of a purified form of one of these glycans (Fig. 8), we assume that two pentose residues are associated with the chitobiose core and that two are more peripheral. Considering that a neutral loss of 299.5 (GlcNAc-PA) was observed from the Hex 7 HexNAc 2 Pnt 4 glycan and that the m/z 767 (putatively HexNAc 2 Pnt 2 -PA) fragment was slightly more abundant than the m/z 563 fragment (putatively HexNAc 1 Pnt 2 -PA), we assume that two pentose residues may be associated with the distal (second) core GlcNAc but that some degree of rearrangement during MS/MS occurs to result in artifactual pentosylation of the reducing terminus. The traces of potential dipentose-containing small fragments (Pnt 2 , HexPnt 2 and HexNAcPnt 2 ; m/z 265, 427, and 468) would be compatible with modifications on an internal GlcNAc and on a mannose. As five mannose residues were removed by jack bean ␣-mannosidase and two by Aspergillus ␣1,2-mannosidase from the purified glycan, we assume the peripheral dipentosylation is present on an inner mannose but is not directly attached to the core ␤-mannose.

Various Modifications of the Core GlcNAc in Pathogenic
Strains-During our survey, it became obvious that many strains express N-glycans with fragments characteristic of core fucosylation. Depending on the strain, fragments of m/z 446 (HexNAcFuc-PA), 608 (HexHexNAcFuc-PA), and 740 (HexHexNAcPntFuc-PA) were observed. The modification of a glycan from 4RE with a mannosylfucosyl motif (indicated by a m/z 608 fragment) has been described above; however, for some strains only an m/z 446 fragment was observed, but in other cases one of m/z 740 is apparent (Fig. 9). However, as both nonpathogenic strains such as 4RE (Fig. 3) and ATCC 30234 (data not shown) and pathogenic strains such as 11DS, 1BU, 72/2, and PAT06 contain different degrees of capping of core fucose, there is no apparent correlation between the core modifications and pathogenicity. Nevertheless, the HexHexNAcPntFuc-PA core is particularly novel and was only present in the group II strains.

DISCUSSION
The Glycomic Potential of Acanthamoeba-The genus Acanthamoeba presents an interesting target for glycan analysis for several reasons, one being that the N-glycomes of the only two other amoebozoans studied so far (E. histolytica, Dictyostelium discoideum) have revealed novel features (28,39). The facultative nature of the pathogenicity of Acanthamoeba sets it apart from Entamoeba, an obligate parasite, and Dictyostelium, which is nonparasitic. Thus, glycomic data on Acanthamoeba are of both phylogenetic and biomedical interest. Although a high abundance of larger structures in the pathogenic morphological group III strain 72/2 was detected, significant quantities of large glycans are also observed in both pathogenic (11DS and 1BU) and nonpathogenic group II strains (4RE and ATCC 30234). However, the glycomes of the pathogenic PAT06 and nonpathogenic Neff, also belonging to group II, are more dominated by unmodified or monopentosylated oligomannosidic glycans. The rather simple glycome of the Pb30/40 strain, featuring dipentose motifs, is in keeping with it belonging to group I, which is phylogenetically rather distant, in terms of 18 S rDNA sequences, to the groups II and III (40). Thereby, the glycan structures found are indeed novel and unlike those in other amoebae, but there is no obvious strong link with genotype or pathogenicity.
Both Familiar and Unfamiliar Glycan Modifications-The mass spectrometric and blotting data did suggest the presence of two mammalian-type glycan modifications: mannose 6-phosphate and core ␣1,6-fucose. In the 4RE strain, chosen here for further characterization, one glycan of low abundance was found to present a series of fragments suggestive of phosphorylation (m/z 243, 405); indeed, previous data have shown that Acanthamobea lysates display UDP-N-acetylglucosamine: glycoprotein N-acetylglucosamine-1-phosphotransferase activity (41). Many of the glycans were core-fucosylated as indicated by the fragment of m/z 446 before or after exoglycosidase digestion. These glycans were released by PNGase F and so are not of the core ␣1,3-fucosylated type found in Dictyostelium (28); together with the sensitivity of this core fucose to bovine ␣-fucosidase, reactivity to LCA and detection of a relevant fucosyltransferase activity (data not shown), our data indicate that the fucose is ␣1,6-linked to the reducing terminus.
Although the mammalian type of core fucosylation is present, a number of Acanthamoeba N-glycans carried hexose residues attached to the core fucose (fragment of m/z 608); we initially assumed the epitope in Acanthamoeba to be of the same nature (galactosylation of core fucose) as that found in nematodes, cephalopods, a gastropod, and a planaria (30,(42)(43)(44). This Gal-Fuc epitope confers sensitivity toward a toxic fungal lectin (CGL2) in Caenorhabditis elegans (45), and it is of special interest that Acanthamoeba ATCC 30234 is also sensitive to this lectin (46). However, our analyses indicate that the core fucose is uniquely capped with ␣-linked mannose. Thereby, the basis for the toxicity of the CGL2 toward Acanthamoeba may be a broader specificity of this lectin than previously thought.
Acanthamoeba synthesizes novel pentosylated structures, which are extremely difficult to characterize given their uncommon nature and their frequently low abundance. Monopentosylated oligomannosidic N-glycans, reminiscent in terms of composition with some structures recently observed in the pathogenic fungus Cryptococcus neoformans (47), are very common in the investigated Acanthamoeba strains and are often the single most abundant N-glycan synthesized, but probably occur as multiple isomers. Based on MS/MS data, we hypothesized that the pentose is most often attached to the second GlcNAc of the core due to the significant occurrence of a fragment of m/z 635; this is also verified by analysis of a perdeuteromethylated structure. This location for pentosylation has been observed in the protist Trichomonas and the microalga Porpyridium (29,37). Only for a methylated glycan could we see clear evidence for a pentosylation of the core ␤-mannose in the 4RE strain, which would be the location compatible with the anti-HRP cross-reactivity toward amoebal proteins (48). Therefore, we conclude that xylosylation of the ␤-mannose may be at a relatively low level in Acanthamoeba or that other pentosylated structures cross-react with anti-HRP.
In addition to the monopentosylated glycans, a variety of multiply pentosylated structures were observed. These are of a very diverse nature with predicted compositions ranging from Hex 5 HexNAc 2 Pnt 2 to Hex 8 HexNAc 2 Pnt 6 Fuc 1 Me. Taking all of the pieces of evidence together, we assume that Acanthamoeba is capable of transferring pentose residues to the ␣-linked peripheral mannoses as in microalgae (37); the amount of the glycans available that carry more than one pentose limits the possibilities for their investigation, but we present evidence for dipentose motifs in the Pb30/40 strain. Also, glycans displaying pentose bound to the hexosylfucosyl motif were detected in some strains, but not in the 4RE strain.
The occurrence of methyl groups leads to Acanthamoeba N-glycans that bear resemblance to structures previously described for mollusks and some other species. Possibly Acanthamoeba is capable of synthesizing methylated N-glycans displaying similarities to those described for mollusks, a planarian, and a microalga. Biomphalaria, for example, is capable of attaching 3-O-Me to either of the mannoses that are attached to the ␤-linked mannose in glycans carrying a core xylose (49), whereas Lymnaea only methylates the ␣1,3-linked mannose (50). The planarian Dugesia japonica, on the other hand, modifies terminal mannoses with methyl groups (44). In the microalga Porphyridium, internal mannose residues of glycans, with the composition Hex 7-9 HexNAc 2 Pnt 0 -2 Me 3 , were found to be methylated (37). A glycan of the composition Hex 5 HexNAc 2 Me 1 from strain 72/2 is reminiscent of a structure of the same composition found in the clam Hippopus hippopus; the clam glycan carries a 6-methylated ␣1,6-linked terminal mannose residue (51). Most methylated glycans in Acanthamoeba are pentosylated and fucosylated; however, the exact nature and location of the methyl groups in Acanthamoeba still need to be determined and leave plenty of work for further studies.
Biosynthesis of N-Glycans in Amoebae-All Acanthamoeba strains assessed produce comparatively large structures, apparently based on the common oligomannosidic eukaryotic structures Man 7-9 GlcNAc 2 and containing up to four ␣1,2-mannose residues. Maximally 13 hexose residues, one of which is attached to a core fucose, were present in the largest glycans, but hardly any glycans smaller than Hex 4 HexNAc 2 are detected. Therefore, in terms of N-glycan precursor assembly, Acanthamoeba is possibly closer related to organisms such as yeast, slime molds, animals, and plants than to other parasitic protozoans. Samuelson et al. (52) hypothesized that the protozoans which were obligate parasites, such as E. histolytica, T. vaginalis, and trypanosomatids, probably lost different glycosyltransferases required for N-glycan precursor assembly in the course of evolution. This can make sense as obligate parasites tend to be genetically reduced and dependent on the host, whereas a free-living Acanthamoeba does not require a host and is highly biochemically autonomous. A tBLASTn search of whole genome shotgun sequences from the Neff strain (data not shown) suggests that all 14 standard alg genes, including those encoding the complete set of ER mannosyl-and glucosyltransferases, are present in Acanthamoeba. After transfer, glucosidase and mannosidase activities are also predicted based on the structural and genomic evidence, although the significant amounts of Hex 10 -12 HexNAc 2 suggest that deglucosylation in endoplasmic reticulum is incomplete, but the lack of a commercially available glucosidase II has limited our possibilities of proving that the glucose "caps" are indeed present.
In contrast to the apparent dominance of oligomannosidic glycans, no complex or hybrid type glycans were detected. These findings led us to believe, that Acanthamoeba lacks GlcNAc-transferase I (GlcNAc-TI) because its activity is necessary for the formation of complex glycans in eukaryotes (53). This would be similar to the case of Dictyostelium, where the GlcNAc-TI is also thought to be absent (28), and it has been hypothesized that a classical GlcNAc-TI is also probably absent in trypanosomatids and trichomonads (29). On the other hand, tBLASTn results suggest that the Neff strain genome may encode proteins with homology to characterized GlcNAc-TI and Golgi mannosidase II proteins (data not shown). Nevertheless, both a xylosyltransferase and a fucosyltransferase activity were detected by us, which accept Man 5 GlcNAc 2 as a substrate (data not shown) and so, therefore, are GlcNAc-TI-independent.

CONCLUSION
Our data would indicate that Acanthamoeba is the third amoebozoan (Entamoeba and Dictyostelium also belonging to this group) to express unusual N-glycans. These three organisms are the only amoebozoans from which extensive data on N-glycans are available up to date. It is therefore still too early to draw conclusions about this entire group of organisms; elucidating the glycomic potential of further amoebozoans though has the potential to yield new insights into variations of the common eukaryotic N-glycosylation pathway. We further predict that Acanthamoeba utilizes the common eukaryotic pathway for N-glycan synthesis, as it is apparently capable of synthesizing the complete lipid linked oligosaccharide precursor dolichol-PP-Glc 3 Man 9 GlcNAc 2 , which can be deduced from the presence of large oligomannosidic structures such as Man 9 GlcNAc 2 and the putatively glucosylated Hex 10 -12 Hex-NAc 2 -based species, as well as the presence of all standard alg genes in its genome. This trait might be a distinguishing factor between obligate and facultative protozoal pathogens. In addition, there must be novel genes encoding xylosyltransferases and fucosyltransferases or homologues of already characterized enzymes of this type which display altered substrate specificity. Taken together, these data present Acanthamoeba as a potentially pathogenic organism capable of synthesizing a large number of highly uncommon N-glycans. Certainly, these glycans and the genes responsible for their synthesis will be the topic of further exciting studies.