Comparative Analysis of Protein Glycosylation Pathways in Humans and the Fungal Pathogen Candida albicans

Protein glycosylation pathways are present in all kingdoms of life and are metabolic pathways found in all the life kingdoms. Despite sharing commonalities in their synthesis, glycans attached to glycoproteins have species-specific structures generated by the presence of different sets of enzymes and acceptor substrates in each organism. In this review, we present a comparative analysis of the main glycosylation pathways shared by humans and the fungal pathogen Candida albicans: N-linked glycosylation, O-linked mannosylation and glycosylphosphatidylinositol-anchorage. The knowledge of similarities and divergences between these metabolic pathways could help find new pharmacological targets for C. albicans infection.


Introduction
Although evolutionarily distant, humans and microorganisms of the Candida genus are closely related from a health perspective. C. albicans is a commensal organism that colonizes mucosal surfaces of the digestive tract and oral and vaginal cavities, and is able to cause superficial or systemic infections (candidiasis), particularly in the light of immunological host defects [1]. Nonetheless, other Candida species including C. glabrata, C. krusei, C. parapsilosis and C. tropicalis have also emerged as important causative agents of candidiasis. Intact glycosylation pathways in both, the human host and the fungal pathogen, are important, if not essential, for their development; thus, the knowledge of commonalities and divergences of these metabolic processes, as well as their functions, could help define pharmacological targets to suppress the pathogenicity of Candida and other fungal pathogens.

The N-Linked Glycosylation Pathway
The N-glycosylation pathway involves attachment of glycans to the amide nitrogen atom in the side chain of asparagine (Asn) residues of eukaryotic, archaeal, and bacterial glycoproteins. The best described model where the eukaryotic Nglycosylation pathway has been characterized in detail is the baker yeast Saccharomyces cerevisiae [2]. Through the years this model has helped to identify and characterize various human and fungal orthologs involved in this pathway.
The synthesis of the dolichol-linked glycan and its transfer to proteins are identical in both, human cells and C. albicans [3,4] (see Table 1 and Figure 1). In fact, these processes are quite conserved among eukaryotic cells and there are only a handful of organisms where these stages are slightly different, such as trypanosomatids, some protists, and the fungal pathogen Cryptococcus neoformans [5,6].
The eukaryotic N-linked glycosylation pathway is divided in two sequential stages: (a) synthesis in the rough endoplasmic reticulum (rER) of the dolichol-linked glycan precursor Dol-PP-GlcNAc 2 Man 9 Glc 3 and its transfer to a nascent protein and (b) the N-linked glycan processing and maturation in the rER and Golgi (Figure 1). Both stages require the action of different glycosyltransferases (GTs) and an adequate supply of donor substrates, which can be, depending on Once transferred, the Glc 3 Man 9 GlcNAc 2 glycan is trimmed by the action of glucosidases and enters a quality control checkpoint performed by the CNX/CRT cycle. Once it passes this checkpoint, it is trimmed by mannosidase MAN1B1 to generate a Man 8 GlcNAc 2 structure. At this point divergence occurs with C. albicans that synthesizes high-mannose glycans. In humans, the Man 8 GlcNAc 2 structure is further demannosylated to Man 5 GlcNAc 2 by Golgi mannosidases type I (MAN1A, MAN1B, and MAN1C). This N-linked glycan suffers further demannosylation and glycosylation processing by type II mannosidases (MAN2A1, MAN2A2), N-acetylgalactosaminyl transferases (GlcNAcT), galactosyltransferases (GalT), fucosyltransferases (FUT) and sialyltransferases (STs). In humans, a glucosidase-independent trimming of Glc 3 Man 9 GlcNAc 2 takes place, generating a Man 8 GlcNAc 2 structure. In addition, lysosomal targeting of glycoproteins through modification with phosphate groups is only found in human cells.
the GT family, nucleotide-activated sugars or dolicholactivated sugars. The lipid dolichol used as a carrier in the first stage of N-glycosylation is a polymer of isoprene units (CH 3 -C(CH 3 )=CH-CH 2 -) predominantly of 14-17 units in the baker yeast [7], 19-22 in humans [8], and of undetermined length in Candida. Dolichol is modified in the rER cytosolic face by human and C. albicans orthologue GTs Alg7, Alg13/14, Alg1, Alg2, and Alg11, using the nucleotide sugars UDP-GlcNAc and GDP-Man as donor substrates to synthesize the Dol-PP-GlcNAc 2 Man 5 intermediate. This intermediate is then flipped from the cytosol to the rER lumen, where synthesis proceeds by GTs Alg3, Alg9, Alg12, Alg6, Alg8, and Alg10 that use the dolichol-linked sugars Dol-PP-Glc and Dol-PP-Man as donors to synthesize the glycan precursor Dol-PP-GlcNAc 2 Man 9 Glc 3 .
International Journal of Microbiology 3 The flippase that translocates Dol-PP-GlcNAc 2 Man 5 has not been identified yet in any organism, although a critical accessory protein in yeast, Rft1, and its human ortholog have been proposed to participate in this process [9,10]. Nonetheless, a recent work in Trypanosoma brucei, an early diverging eukaryote, has pointed that Rft1 is not the much sought after flippase, although it is critical for allowing maturation of the Dol-PP-GlcNAc 2 Man 5 intermediate once it is flipped to the rER lumen [11]. A putative ortholog of Rtf1 has been also found in C. albicans (Table 1).
Once synthesized, the Dol-PP-GlcNAc 2 Man 9 Glc 3 precursor glycan is transferred en bloc by the oligosaccharyl transferase complex (OST) to Asn residues by linkage to carboxamide nitrogens. The Asn residues targeted for Nlinked glycosylation are located, with rare exceptions, within the consensus sequence Asn-X-Ser/Thr (where X is any amino acid except proline) [12]. However, not all consensus sequences are N-linked glycosylated, because this protein modification is a co-translational process, and thus other factors are involved in selecting consensus sequences, such as accessibility of OST to the consensus sequence during the unfolded state of the protein. The OST complex has not been characterised in detail in C. albicans; however, the fungus encodes all the subunit orthologs found in S. cerevisiae OST, which is comprised of nine different transmembrane subunits: Wbp1, Swp1, Stt3, Ost1, Ost2, Ost3, Ost4, Ost5, and Ost6, where Stt3 is the catalytic subunit [13] (Table 1). Mammalian equivalents to yeast/C. albicans OST subunits are known and include: ribophorin I (Ost1) and II (Swp1), OST48 (Wbp1), defender against apoptotic cell death or DAD1 (Ost2), N33 (Ost3), magnesium transporter 1 (Ost6), and OST4 (Ost4) [14][15][16], (Table 1). In addition, two Stt3 protein orthologs (STT3A and STT3B) have been identified in plants, insects, and vertebrates [15,17,18]. The human STT3A isoform is primarily responsible for cotranslational modification of sequons when the nascent polypeptide enters the rER lumen. The STT3B isoform is less competent for cotranslational glycosylation, but mediates the posttranslational modification of skipped glycosylation sites in unfolded proteins [19]. The mammalian OST has been found in three complexes that exhibit different ribosome affinities and subunit compositions: OSTC(I), OSTC(II), and OSTC(III) [16]. Furthermore, two additional components found in the mammalian OST complex have been reported: KCP2 and DC2 [16,20].
Once transference onto the protein is achieved, the pathway continues with the processing and maturation stage. Processing is carried out, in both human and C. albicans, by rER enzymes: the mannosyl oligosaccharide glucosidase I (MOGS/Cwh41) that removes the outermost 1,2glucose unit, and the mannosyl oligosaccharide glucosidase II which trims the following 1,3-glucose residue exposing the Glc 1 Man 9 GlcNAc 2 epitope [21] (Figure 1). In humans/ Candida, glucosidase II is a heterodimer composed of two subunits, the hydrolytic -subunit (GANAB/Rot2) and thesubunit (GLU2B/Gtb1), see Table 1.
The Glc 1 Man 9 GlcNAc 2 epitope is a key point of ER quality control of glycoproteins, as it binds to the calnexin/calreticulin (CNX/CRT) lectin that is a folding sensor associated to ERp57. At this point, glucosidase II removes the last glucose residue and, if correctly folded, the glycoprotein exits the rER after the 1,2-mannosidase removes one Man residue from the middle branch of the N-linked glycan core, generating GlcNAc 2 Man 8 ( Figure 1). If the protein is misfolded, the glycan is reglucosylated by the action of the UGGT1 glucosyltransferase in humans and its ortholog Kre5 in C. albicans [22]. UGGT is a conformational sensor, regenerating the acceptor substrate for the calnexin/calreticulin lectin, starting a new deglucosylation step by glucosidase II. This cycle continues until the protein is correctly folded or targeted for ER-associated degradation [23].
In contrast to Candida, humans code for an endomannosidase (MANEA) located in the Golgi/ERGIC compartment that provides a glucosidase I and II independent pathway for N-linked glycan maturation. MANEA is able to remove the inner most Glc residue along with the Man residue attached to it, generating the GlcNAc 2 Man 8 structure ( Figure 1).

The Fate of GlcNAc 2 Man 8 in Humans.
The further processing of the GlcNAc 2 Man 8 structure is the divergence point between humans and C. albicans ( Figure 1). In humans, the N-linked glycans are processed by Golgi-resident mannosidase IA, IB, and IC, which have different hydrolytic patterns but all generate Man 5 GlcNAc 2. This glycan is then acted upon by glycosyltransferase GlcNAcT-I to generate a GlcNAcMan 5 GlcNAc 2 structure that is acted upon by type II -mannosidases. The type II -mannosidases include the Golgi mannosidase II (MAN2A1), and in some cell types, additional mannosidases MAN2A2 and mannosidase III have been described as bypassing enzymes when mannosidase II fails to hydrolyse the N-linked glycan core [24][25][26]. The type II -mannosidases remove the terminal 1-3Man and 1-6Man residues allowing addition of a second GlcNAc residue to give way to complex glycans. The GlcNAc residues can be extended with additional monosaccharide linkages involving galactose, fucose, or sialic acid residues. Furthermore, the hybrid and complex N-linked glycans found in humans may exist with two or more GlcNAc-bearing branches or antennae. In forming multiantennary N-linked glycan structures, GlcNAc residues may be added to the trimannosyl core by six different GlcNAc transferases (I-VI) [27]. If type II -mannosidases do not act or GlcNAcT-III bisects the GlcNAcMan 5 GlcNAc 2 structure, hybrid glycans are then generated ( Figure 1) [28].
In animals, N-linked glycans are terminated by sialic acid [29] in 2,3-, 2,6-, or 2,8-linkages by specific sialyltransferases [30]. In humans, sialic acid is mostly of the Nacetylneuraminic acid form, in contrast to most mammalian species, where a mixture of N-glycolylneuraminic acid and N-acetylneuraminic acid is generally found. Sialic acid can be further modified by acetylation or sulphation [31]. This monosaccharide in view of its terminal position, linkages, and negative charge has been an important element in the evolution of animal glycan function [32]. Although 2,3-and 2,6-sialic acid have been identified in the cell wall of C. albicans [33,34], no ortholog to vertebrate sialyltransferase or ability to synthesize sialic acid has been characterized in this fungus [35]. However, evidence of sialic acid synthesis has been reported in Aspergillus fumigatus [36] and C. neoformans, where sialyltransferase activity has been identified [37].
Another frequent modification of human N-linked glycans not seen in C. albicans is 1,6-core fucosylation of the first GlcNAc residue, as well as terminal fucosylation on Gal or GlcNAc residues [38]. Nonetheless, fucose has been identified as a component of the cell wall of C. albicans [33] and binds the UEA-I lectin that is specific for L-fucose, more particularly to 1,2-fucose. UEA-I binding was associated with increased adherence to epithelial cells [39]. Recently mass spectrometry identified 1,6-fucose residues in oligomannosylated N-linked glycans of the fungi Cantharellus cibarius [40]. This raises the question on how this type of glycans are presented in the surface of mushrooms, as no FUT8 family member of fucosyltransferases responsible for this linkage has been identified in yeast nor mushrooms [38].
Although little is known about fucosylation and sialylation mechanisms in C. albicans or fungi in general, more information is hinting at their role in pathological human host interactions through molecular mimicry.
Furthermore, human N-glycans can be phosphorylated to target glycoproteins to the lysosomes, through interaction with the Man-6-P receptor [41]. Phosphorylation occurs International Journal of Microbiology 5 by modification of the GlcNAc 2 Man 8 structure by a UDP-GlcNAc-dependent GlcNAc-1-phosphotransferase ( Figure 1) [42]. A GlcNAc phosphodiester is added to N-linked glycans on one of three mannose residues on the arm with the 1,6linkage to the core mannose. A second phosphodiester can then be added to the other side of the N-glycan or onto other mannose residues (Figure 1). Afterwards, the phosphodiester glycosidase in the trans-Golgi removes the GlcNAc to generate Man-6-P residues. The phosphate residues partially block the action of processing mannosidases, maintaining the Nglycans in an oligomannosyl form. Some hybrid N-linked glycans with Man-6-P can also be found. This sorting system of soluble proteins does not exist in C. albicans or other yeast species, but interestingly they contain the gene MRL1, which seems to be the ortholog of that encoding the human Man-6-P receptor [43].

The Fate of GlcNAc
albicans. In C. albicans, the N-linked glycan GlcNAc 2 Man 8 is modified by proteins that have no human orthologs (see Table 2). The GlcNAc 2 Man 8 core is recognised by Och1, an 1,6mannosyltransferase that adds the first mannose residue of the N-linked glycan outer chain [44] (Figure 1). This mannose residue works as a molecular primer to build the 1,6-mannose backbone, which in S. cerevisiae is elongated by the M-Pol I complex (a heterodimer composed of Mnn9 and Van1) that adds 3 to 7 mannose residues [45] and then by M-Pol II, a multimeric complex composed of Mnn9, Anp1, Mnn10, Mnn11, and Hoc1 [46,47] (Figure 1). Both, in vivo and in vitro studies have shown that Mnn10 and Mnn11 contribute to most of the 1,6-mannosyltransferase activity of M-Pol II [47]. Thus far, there is only experimental evidence about Mnn9 role in C. albicans [48]; however, the encoding genes for all members of both complexes are present within C. albicans genome and it is likely they work as described in the baker yeast.
Parallel to this process, the 1,6-mannose polymer works as a molecular scaffold where branches of 1,2-mannose residues are added by Mnn5 [49,50]. These are further elongated by the mannosyltransferases Mnt4 and Mnt5 [51], and members of the MNN2-like gene family [52]. In S. cerevisiae, the branches are terminated with 1,3-mannose residues added by action of the mannosyltransferase Mnn1 [53]. In C. albicans, these mannose residues are also present and are likely to be incorporated to glycans via the same protein [54]; however, it is most frequent that the 1,2mannose branches are further decorated and capped with 1,2-mannose units [55]. The 1,2-mannosylation is characteristic of this pathogenic yeast species and is carried out by members of the BMT gene family [56].
Another decoration attached to the 1,2-mannose branches is the phosphomannan, which is a mannose residue attached to the N-linked glycan by a phosphodiester bond ( Figure 1). This phosphorylation is not related to that found in humans and is partially synthesized by phosphomannosyltransferases Mnt3 and Mnt5 [51]. The identity of the enzymes involved in the addition of the rest of the phosphomannan remains unknown, although it is likely that members of the MNN4-like family contribute to this activity [57,58]. As the N-linked glycan, the phosphomannan can be further 1,2-mannosylated with up to 14 1,2-mannose units [59], by action of Bmt2, Bmt3, and Bmt4 [56].

Functions of N-Linked Glycans in Humans.
The N-linked glycans associated to glycoproteins participate in the calnexin/calreticulin ER quality control system of glycoprotein folding [60]. Furthermore, N-linked glycans are involved in protein stabilization and trafficking and serve as moieties recognized by receptors, thereby modulating binding by increasing or decreasing affinity. The N-linked glycans play from trivial to essential roles in glycoprotein function and are involved in most, if not all, cellular processes. There is clear evidence that this posttranslational modification is essential for homeostasis in multicellular organisms as has been demonstrated by clinical phenotypes, mostly multisystemic, of patients affected by congenital disorders of glycosylation, indicating that N-linked glycan integrity is required for normal tissue function [61].

Functions in C. albicans.
The N-linked mannans are essential for C. albicans viability, as demonstrated by treatment with tunicamycin, a drug that inhibits the action of Alg7 during N-linked glycan core synthesis [62]. Furthermore, they are quite important for cell fitness: defects in either the processing step by rER -glycosidases or elongation by Golgi mannosyltransferases lead to longer duplication times, swollen cells, inability to perform proper cell separation, abnormal colony morphology, and impaired ability to undergo dimorphism [21,44,51,63]. These pleiotropic defects are likely consequences of loss of the cell wall plasticity: mutant cells with defects in the N-linked mannan biosynthesis have rearrangements in the wall composition, including low mannan levels and high chitin and glucan contents, which led to increasing the sensitivity to cell wall perturbing agents such as tunicamycin, Congo red, Calcofluor white, hygromycin B, and caffeine [4,21,44,[50][51][52]63]. Protein modification by N-linked mannans also modulates protein secretion, but surprisingly in a negative form, as shown in mutants lacking rER -glycosidases, which display increased cell wall protein content [21]. In addition, biofilm formation seems to depend on N-linked mannans, as shown by the inability of tunicamycin-treated cells to form this kind of microbial consortiums [64]. Finally, and most important, Nlinked mannosylation is required for normal cell adhesion and virulence [4,21,44,50,51,63,[65][66][67][68][69]. The extent to which fucosylation and sialylation play a role in pathogenicity through the adhesion to the host surface, particularly extracellular matrix components, still requires further characterization in C. albicans.

The O-Linked Mannosylation Pathway
In contrast to C. albicans that only synthesizes O-linked mannosyl glycans (O-Man), six additional types of O-linked glycans are found in humans, and are classified based on the first sugar attached to the amino acid residue: GlcNAc, GalNAc, galactose, xylose, glucose, or fucose. The O-Man glycans were identified on brain proteoglycans more than 30 years ago [70], and the O-mannosylation of -dystroglycan ( DG) has been the most studied. In contrast to C. albicans, human O-Man glycans contain only one mannose residue (linked to the protein) and are extended with other monosaccharides (Figure 2). In C. albicans, the O-Man glycans are composed of up to five mannose residues [71]. Most of the mammalian O-mannosyl glycans are variations of the common tetrasaccharide core NeuAc 2-3Gal 1-4GlcNAc 1-2Man 1-Ser/Thr, although branched structures with 2,6-di-substituted mannose (GlcNAc-linked 1,2 and 1,6) have been described in brain glycoproteins [72]. In humans, the first mannose residue is added in the rER by protein-O-mannosyl-transferase 1 (POMT1) [73] and 2 (POMT2) [74], homologous to C. albicans Pmt4 and Pmt2, respectively (Table 3). Both enzymes perform their function in an essential complex that uses Dol-P-Man as sugar donor [75]. Another O-linked glycan structure (GalNac 1-3GlcNAc 1-4Man 1-Ser/Thr) has been reported in DG [77] (Figure 2). This structure is further phosphorylated in the 6-position of O-mannose by the action of the Protein-O-mannose kinase (POMK) [78]. The LARGE and LARGE2 bifunctional glycosyltransferases act on the phosphomannose structure producing repeating units of [-3-xylose-1,3-glucuronic acid-1-] (Figure 2) [79,80]. Two other proteins, FUKUTIN and FKRLP, with glycosyltransferase characteristics, are also involved in the formation of human O-Man glycans, but their function remains unknown.
As in humans, in C. albicans, this pathway starts in the rER and finalises in the Golgi complex. The synthesis  The mannose residue can also be phosphorylated by the action of POMK, allowing further modification by disaccharide repeats of xylose and glucuronic acid synthesized by LARGE and LARGE2. Fukutin and FKRP genes are needed for correct glycosylation, but their roles remain to be clearly defined. In C. albicans, mannosylation involves the addition of the first mannose residue by the action of Pmt1-5, and that is further extended with four additional mannose residues by the action of Mnt 1 and Mnt2.
begins with the addition of one -linked mannose residue to Ser or Thr residues via an ester bond. This reaction takes place in the rER lumen and is catalysed by the proteinmannosyl transferases that use Dol-P-man as sugar donor [81]. This enzyme activity is performed by a family composed of five members that are subclassified in three groups: the Pmt1 (Pmt1/5), Pmt2 (Pmt2/6), and Pmt4 subfamilies [82] ( Figure 2). The proteins encoded by these subfamilies do not have redundant activity in vivo, as each member has specific substrates [81][82][83]. In addition, these enzymes interact among them generating protein-protein interactions. In S. cerevisiae, Pmt1 interacts in vivo with Pmt2, and combined disruption of PMT1 and PMT2 results in more than 90% less enzyme activity in vitro [84]. Another predominant complex includes Pmt5 and Pmt3, but in the absence of Pmt5, Pmt3 can form a complex with Pmt1, and Pmt2 can form a complex with Pmt5 when Pmt3 is disrupted [84]. Pmt1, Pmt5, and Pmt6 have no human orthologs (Table 2).
Once the glycoproteins are transported to the Golgi complex, the O-linked glycans are further elongated by the Golgi 1,2-mannosyltransferases Mnt1 and Mnt2 that have redundant activities to fully elongate the glycans [71,85]. This mannan structure can also be phosphomannosylated, and in fact, the phosphosugar attached to O-linked mannans represents about 20% of total cell wall phosphomannan content [21]. However, the machinery involved in this process is different of that described for N-linked mannans, as Mnt3 and Mnt5 do not add phosphomannose to O-linked mannans [51]. Mnt proteins have no human orthologs ( Table 2).
As mentioned before, sialic acid has been described in C. albicans [33,36], and sialidase treatment has been shown to increase binding of the peanut agglutinin that has specificity for the Gal 1,3GalNAc sequence present in human Core 1 O-glycans which has not been described in C. albicans. This suggests that sialic acid could be part of C. albicans O-linked mannans and the presence of uncharacterized galactosyltransferases.

Functions in Humans.
The best characterized mammalian O-linked Man glycoprotein is DG (Figure 2). This protein is a glycosylated peripheral membrane protein involved in linking the cytoskeleton of neurons and muscle cells to the basal lamina through interactions with extracellular proteins; glycosylation of DG is essential for its function [86]. To date, mutations in seven glycosyltransferase or glycosyltransferase-like genes have been reported to affect the O-linked mannosylation pathway and are causative for various forms of autosomal recessive congenital muscular dystrophies associated with variable brain and ocular abnormalities [87]. As it was mentioned earlier, O-Man glycans in humans are not highly mannosylated structures; they only possess a single Man residue. This divergence is functionally important in the immune systems recognition of pathogenic yeast and fungal microorganisms, including C. albicans. Highly mannosylated structures, as those found in yeast and fungi, are recognized as foreign by both circulating antibodies and elements of the complement system, including both the classical and alternative pathways [88].

Functions in C. albicans.
Loss of PMT2 or combined disruption of PMT1 and PMT4 led to nonviable cells, indicating that O-linked mannosylation is essential for growth and cell viability [81]. In addition, incomplete O-linked mannan elaboration has been associated with rearrangements in the cell wall composition, increasing sensitivity to cell wall perturbing agents, defects in morphogenesis, reduced tissue adhesion, defective biofilm formation, and virulence attenuation [71,81,[89][90][91].
The O-linked mannans are key cell wall elements during the C. albicans sensing by immune cells. This cell wall component is sensed by TLR4 receptor [88], and loss of either O-linked mannans or TLR4 receptor has a negative impact on cytokine production by human PBMCs [88], on the proinflammatory response of oral epithelial cells [92], and on yeast killing by human polymorphonuclear cells [93]. Indeed, TLR4 −/− knockout mice are more susceptible to infections caused by C. albicans due a defective immune response against the fungus [94,95]. Furthermore, it has been demonstrated that simultaneous stimulation of dectin-1 and either TLR2 or TLR4 significantly enhances cytokine production in both human monocytes and macrophages [96,97]. Therefore, it has been hypothesized that recognition of O-linked mannans plays a pivotal role, along with 1,3-glucan sensing, in the establishment of a protective anti-Candida immune response.
However, the relevance of O-linked mannans during C. albicans sensing is not the same for different kinds of immune cells, as yeast cells lacking both MNT1 and MNT2, and therefore expressing truncated O-linked mannans at the cell wall surface [71], are as good as the wild type control cells to stimulate binding and cytokine production by human dendritic cells [98]. Moreover, there are some C. albicans strains whose immune sensing is independent of recognition via TLR4, suggesting that the fungus might be able to modulate the production of this cell wall component [99].
Next, GlcNAc-PI is deacetylated by PIG-L/Gpi12, generating GlcN-PI [108], which requires crossing the rER membrane to continue the synthetic pathway within the lumen. This transport, as in the other protein modification described above, is carried out by a rER flippase. Then, inositol acylation takes place due to the acyltransferase activity of PIG-W/Gwt1, being the donor acyl-CoA [109,110]. GPI mannosylation takes place using Dol-P-Man as mannose donor and begins with action of the mannosyltransferase PIG-M/Gpi14 in complex with PIG-X/Pbn1 that adds the first mannose ( 1,4linked) to GlcN [111]. The second ( 1,6-linked) and third PIG-F (Gpi11) Figure 3: The GPI anchorage pathway. GPIs are glycolopids that act as a membrane anchor for many cell surface proteins and are composed of an inositol molecule that is sequentially modified in the ER with a Man3GlcN glycan and phosphoethanolamine groups. Numbers show sequential steps for the synthesis of a Man3GlcN glycan bearing GPI. A GPI-transamidase complex acts upon the phosphoethanolamine group linked to the terminal mannose to add the surface protein. The C. albicans Smp3 adds a fourth mannose residue that is essential for protein transfer; but in humans it is not essential for transamidation and its expression appears to be restricted to brain and colon.
( 1,2-linked) mannose units are transferred by PIG-V/Gpi18 and PIGB/Gpi10 mannosyltransferases, respectively [112][113][114][115]. The enzyme Smp3 catalyses the addition of a fourth mannose residue ( 1,2-linked) to Man-3 of the glycan core, being an essential step in yeast and Candida cells, as it is required for subsequent attachment of phosphoethanolamine [116,117]. Mammalian cells mostly transfer trimannosyl-GPIs to proteins and do not require the addition of a fourth mannose residue, but a human ortholog of the yeast mannosyltransferase Smp3 that adds a fourth, 1,2-linked Man to trimannosyl GPI precursors has been identified, displaying high expression in brain and colon, suggesting that Man 4 -GPIs elaboration could be tissue-specific [117]. In C. albicans, Smp3 and is essential for viability and has been proposed to be a potential antifungal target [118]. An Etn-P unit can be attached to the first mannose of the glycan core as a side branch by PIG-N/Mcd4 and also to the second mannose by a complex of PIG-F/Gpi11 and PIG-G/Gpi7 [119][120][121]. Finally, a moiety of Etn-P is added to the third mannose of the core, being this residue the one bound to the protein through an amide link. A transferase association between PIG-O/Gpi13 and PIG-F/Gpi11 is responsible for this step [122,123]. The GPI synthetic pathway is highly conserved, but GPIs can be further modified in the lipid and glycan moieties depending on genus, species, and protein type [124].
The GPI-anchored proteins have a C-terminal sequence that directs the attachment of a GPI anchor. The removal of the C-terminal GPI signal sequence and its replacement with GPI on the lumen of rER are catalyzed by the GPI transamidase (GPIT), which is a complex consisting of the membrane proteins PIG-K/Gpi8, GAA-1, PIG-S/Gpi-17, PIG-T/Gpi16, and PIG-U/Cdc91 [125,126]. In the first step of the GPIT-catalyzed reaction, the GPI signal sequence is cleaved and the newly generated -carbonyl group is attached via a thioester linkage to the PIG-K subunit of GPIT. Nucleophilic attack on the activated carbonyl by the amino group of the terminal EtN-P residue of GPI regenerates GPIT and yields a GPI-anchored protein.
After transfer, the inositol group introduced before mannosylation of the GPI precursor is removed in humans and yeast by the orthologous PGAP1/Bst1 deacylase ER proteins [127]. Yeast is able to remodel the shorter acyl chains of the diacylglycerol shortly after transfer to either baselabile C 26:0 /C 26:0 diacylglycerols or to a base-stable ceramide consisting of C 18:0 phytosphingosine and a hydroxy-C 26:0 fatty acid. The remodeling initiates with the removal by PGAP3/Per1 of the acyl chain at the sn-2 position of the diacylglycerol [128,129]. PGAP3-dependent removal of unsaturated fatty acyl chains at the sn-2 position occurs predominantly in the Golgi, whereas Per1 activity is located in the rER.
Next, a C 26:0 acyl chain is introduced at sn-2 by the O-acyltransferase Gup1 that is the only enzyme involved in GPI anchor synthesis in C. albicans that has no human ortholog [130]. The mammalian PGAP2 protein is involved in the subsequent introduction of a saturated (C 18:0 ) fatty acid at sn-2 [131]. Mutations in the yeast gene that encodes a homolog of PGAP2, CWH43, albeit a much larger protein, cause cell wall abnormalities consistent with defects in cell surface anchorage of GPI proteins. In mammals, remodeling at sn-2 requires prior inositol deacylation by PGAP1 [129]. The PGAP3-and PGAP2-dependent remodeling activities, in turn, are necessary for the GPI-anchored proteins to associate with lipid rafts.

Functions in Humans.
The obvious role for GPI-anchors is the attachment of proteins to cell surface. Examples include cell surface receptors (e.g., folate receptor, CD14), cell adhesion molecules (e.g., neural cell adhesion molecule), cell surface hydrolases (e.g., alkaline phosphatase), and complement regulatory proteins (e.g., decay-accelerating factor [CD55]). Human diseases arise by failures in this posttranslational process, stressing its importance for proper function of human cells. The paroxysmal nocturnal hemoglobinuria is a consequence of lower surface expression of GPI-proteins, due to clonal acquired mutations in PIG-A [132]. Inherited mutation in the promoter region of PIG-M impairs the binding of the transcription factors, resulting in abrogation of GPI mannosylation, leading to propensity to venous thrombosis and seizures [133]. Other congenital diseases involving defective PIG anchoring have been recently described [134].

4.2.
Functions in C. albicans. As in human cells, GPI synthesis is essential for S. cerevisiae growth [101,135]. C. albicans has 115 putative GPI-proteins, with diverse predicted functions, including adhesion to host tissues [136]. C. albicans GPI7 leads to an aberrant cell wall composition with increased chitin content and less protein abundance [137], while cells lacking Smp3 mannosyltransferase are nonviable [118].
Recently, it has been demonstrated that defects in GPI synthesis affect hypha growth [138]. Yadav et al. propose that Gpi2 and Gpi19 subunits of the GPI-GnT complex regulate ergosterol synthesis and RAS signaling, which explains the influence of GPI synthesis in the dimorphic switch [139]. Adhesins of the Als family are known to be GPI-proteins [140], so it is not surprising that virulence is attenuated in GPI mutants.

The Glycosylation Pathways as Potential Drug Targets against Fungal Infections
The information gathered in the last decades about the human glycosylation pathways has helped to differentiate the normal processes from those found in neoplastic cells, and these are now explored as potential strategies to treat cancer [141,142]. Since protein glycosylation is a key process for C. albicans fitness and virulence attributes [4,143], it is assumed that the development of inhibitors for any of the glycosylation pathways may assist in treatment of candidiasis. Tunicamycin is one of the oldest N-linked glycosylation inhibitors that has been thoroughly characterised over the last decades. It affects the elaboration of the N-linked glycan core [144], and tunicamycin-treated cells of C. albicans lose the viability [21] and the ability to generate biofilms [64], making this molecule a potential anti-C. albicans drug. However, the UDP-N-acetylglucosamine, dolichol phosphate GlcNAc-1-P transferase, the molecular target of this compound, is equally sensitive in both human and fungal cells [64]. A promising strategy for treatment of C. albicans infections could be found in the rER glucosidase inhibitors, which have been used in the experimental control of some viruses [145][146][147][148][149]. We have previously shown that C. albicans cells require processing of the N-linked glycan core in order to elongate their N-linked mannans, and loss of either glucosidase I or II led to virulence attenuation [21]. Since the human cells have a bypassing strategy for glucosidase trimming via the endomannosidase enzyme activity, it is feasible to conceive the potential low cytotoxicity of these drugs on the host cells. Celgosivir [150], PBDNJ0804-a deoxynojirimycin derivative [149], and CM-10-18 [147] are rER glucosidase I inhibitors that could show anti-C. albicans activity.
The O-linked mannosylation pathway can also be targeted for drug design. It was recently demonstrated that the rhodanine-3-acetic acid derivative OGT2468 is a PMT inhibitor in S. cerevisiae [151] and that it likely affects the same biosynthetic pathway in C. albicans. Whether this compound affects or not the elaboration of O-linked Man glycoprotein in human cells remains to be addressed.
The enzymes involved in GPI synthesis are also potential targets to develop new antifungal drugs. Gepinacin and E1210 were found to inhibit the fungal acyltransferase Gwt1, impairing the growth of fungal pathogens. Despite the functional similarity, gepinacin has no effect on the mammalian ortholog PIG-W. Assays on C. albicans cells treated with gepinacin indicate that they overexpose -glucans on the wall surface, which triggers a better macrophage response [152][153][154]. Development of Smp3 or Gup1 inhibitors would be of value in view of their nonessential nature or absence in humans, respectively.

Conclusions
Metazoa (animals) and fungi derive from a common ancestor that existed ∼1 billion years ago, nonetheless the basis of protein glycosylation pathways is strikingly conserved in spite of this period. In this review, we can look at the common bases and differences that emerge when comparing glycosylation mechanisms in C. albicans and humans.
The study of C. albicans glycosylation machinery is an important step to identify pharmacological targets to treat local or systemic candidiasis. Ideal pharmacological targets are represented by those elements only present in Candida. In the N-linked glycosylation pathway at least 16 GTs participate in mannan synthesis and are not present in humans (see Table 2), making this pathway an attractive alternative for drug design. Thus far, some promising approaches have been done with glucosidase inhibitors, but their toxicity in human cells remains to be addressed. In addition, rER-mannosidase inhibitors could be used as an alternative approach, as fungal cells only contain one mannosidase class I within the rER, and its loss is associated with virulence attenuation [21].
The above data indicate that fungal glycosylation pathways are promising for inhibitory compound screening that are species specific, both because of the presence of many nonhomologous proteins identified in C. albicans, particularly in the N-glycosylation pathway, and also because of the presence of homologous proteins that have a low degree of identity. Further studies should focus on developing compounds to inhibit the essential functions of glycosylation pathways taking into account these facts.

Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.