Whoa man! Unexpected protein O-mannosylation pathways in mammals

Abstract

The recent expansion of well-characterized O-mannosylated mammalian proteins beyond the archetypical example of α-dystroglycan has inspired new interest in the possibility of additional functional roles of this modification. In an effort to explore those roles, a new study now serendipitously uncovers the existence of an alternative pathway to the well-described POMT (protein O-mannosyltransferase) family of O-mannosyltransferases.

Introduction

The O-linked glycosylation of proteins initiated with O-mannose (O-Man) has been known to exist in mammals for more than 30 years (1–3), but until recently, the only well-documented O-mannosylated mammalian protein was α-dystroglycan (α-DG).² This protein is known to be heavily decorated with O-linked glycans, including both mucin-type O-GalNAc (N-acetylgalactosamine) structures and elaborated O-Man glycans, which are introduced by a series of enzymes beginning with the protein O-mannosyltransferases POMT1 and POMT2. Given that aberrant O-mannosylation has now been implicated in a wide range of human diseases, including congenital muscular dystrophy (CMD), cancer, and arenavirus infection (1–3) and the reported abundance of O-Man glycans (approximately one-third of O-glycans in mouse brain) (4, 5), efforts by multiple groups have been expanding the O-Man glycoproteome (6, 7). However, although progress has been made in identifying O-mannosylated proteins, the precise functional roles and the further elaboration of the O-Man on proteins other than α-DG have been largely unexplored (6). Larsen and colleagues (8) intended to shed light on this topic using a combination of innovative gene knock-out strategies in mammalian cells and mass spectrometry but instead report the surprising finding that O-Man transfer to proteins is more complex than anticipated.

Mutations resulting in defective glycosyltransferases that generate the extended O-Man structures on α-DG result in a subset of CMDs called dystroglycanopathies (1–3), and interest in understanding these diseases has led to most of our knowledge about mammalian O-mannosylation pathways. The process begins with the highly conserved POMT1/POMT2 enzyme complex catalyzing the transfer of mannose from the activated sugar–donor molecule (9), dolichol-phosphate mannose, in the endoplasmic reticulum to Ser or Thr residues on target proteins (Fig. 1); mutations in the POMT1 and POMT2 genes have been identified in the most severe form of CMD known as Walker–Warburg syndrome. Until now, these enzymes were thought to be the only protein-modifying, initiating O-mannosyltransferases in mammals.

Figure 1.

Summary of the evolving O-mannosylation pathways. Left, the classical POMT1/POMT2-dependent O-mannosylation of α-DG. The classical pathway diverges with the activity of either POMGNT1 or POMGNT2 generating various core structures. Nucleocytoplasmic O-mannosylation (center) has also recently been described in yeast (10). Larsen and colleagues (8) describe the identification of POMT1/POMT2-independent O-mannosylation of proteins other than α-DG (right) that appear to not be extended but warrant further investigation. Green circles represent α-linked O-mannose attached to Ser or Thr residues on proteins, and blue squares represent N-acetylglucosamine with the linkage indicated.

Larsen et al. (8) sought to delineate the functions of recently identified O-Man glycans on cadherins (cdh) and protocadherins (pcdh) (6, 7) by carefully orchestrated gene disruption strategies in CHO and human HEK293 cells. Using concanavalin A– lectin weak affinity chromatography (ConA-LWAC) for O-Man glycopeptide enrichment followed by detection by mass spectrometry, the authors quantitatively compared the O-Man glycoproteomes of various knock-out cell lines. The authors initially probed the O-Man glycoproteome by ConA-LWAC enrichment of glycopeptides derived from CHO cells deficient in POMGNT1, the second enzyme in the classical O-mannosylation pathway (CHO^PGNT1), expected to result in a subpopulation of unextended O-Man glycopeptides (Fig. 1). In agreement with previous studies (6, 7), glycopeptides derived from cdhs and pcdhs were identified in addition to glycopeptides derived from α-DG. Interestingly, similar analyses of wild-type (WT) CHO cells, expected to possess mature, extended O-Man glycans, resulted in the detection of unextended O-Man peptides from most of the proteins identified in CHO^PGNT1, suggesting that the authors were able to capture O-Man proteins in an intermediate stage of modification and/or these select proteins are not elongated by POMGNT1 or POMGNT2. Further, the authors analyzed CHO^PGNT1 that were also deficient in POMT1, POMT2, or both. In these mutant cell lines, they were unable to detect any α-DG O-Man glycopeptides, but did identify glycoproteins including cdhs and pcdhs. This is consistent with O-mannosylation of α-DG being dependent on both POMT1 and POMT2. However, it uncovered the surprising finding that other O-Man–modified proteins were not dependent on POMT1 or POMT2.

In an effort to increase the number of O-Man proteins identified and validate their findings, the authors replicated their experiments in genetically modified HEK293 cells, termed SimpleCells (HEK293^SC) that lack the ability to extend O-GalNAc glycosylation due to a gene deletion. Experiments carried out using the HEK293^SC genetic background combined with different knock-out combinations of POMGNT1, POMT1, and/or POMT2 resulted in similar findings as those observed in modified CHO cell lines. This finding greatly strengthened their argument that certain proteins, such as cdhs, do not require the classical POMTs to initiate O-mannosylation. Furthermore, glycosylation site mapping of recombinant mouse E-cadherin, Pcdhα-C2, and Pcdhγ-A4 expressed in POMT1/POMT2-deficient HEK293 cells again identified sites of unextended O-mannose modification. A quantitative MS approach was then pursued by differential labeling of tryptic peptides using stable dimethyl isotopes. These experiments provided further support that O-mannosylation of cdhs and pcdhs were not significantly affected by POMT1 and POMT2 disruption.

The authors cumulatively demonstrate the requirement of POMT1/2 for O-mannosylation of α-DG and a select few other proteins while providing strong evidence for the existence of novel, alternative, mammalian O-mannosylation machinery specific for cdhs, pcdhs, and IPT/TIG (immunoglobulin, plexins, transcription factors-like/transcription factor immunoglobulin) domain-containing proteins. The authors state in their conclusion that they have preliminary data identifying novel mammalian protein O-mannosyltransferases. These exciting findings raise a multitude of questions including: How does the POMT1/2 enzyme complex exhibit substrate specificity? Is there any interplay between POMT1/2 O-mannosylation and the yet-to-be-described POMT(s)? Is their candidate enzyme(s) conserved in model organisms to facilitate study? Is their recently identified candidate enzyme responsible for all non-POMT1/2–mediated O-mannosylation or are there multiple (domain-specific?) POMTs to be discovered? Because it appears that POMT1/2-independent O-mannosylation is not extended, how is O-Man of these target proteins escaping modification by POMGNT1/2? And finally, what are the functional implications of O-Man on proteins other than α-DG? Given the domain-specific nature of these potentially not extended O-Man additions, analyzing whether the primary function of modification is to facilitate protein folding in the endoplasmic reticulum is warranted. In conclusion, Larsen and colleagues (8) utilize novel genetically modified cell lines and innovative analytical tools to gain unexpected insight into the mammalian O-mannosylation pathway, which continues to surprise scientists with its evolving complexity.