Genome-wide evolutionary conservation of N-glycosylation sites

Abstract

Although posttranslational protein modifications are generally thought to perform important cellular functions, recent studies showed that a large fraction of phosphorylation sites are not evolutionarily conserved. Whether the same is true for other protein modifications, such as N-glycosylation is an open question. N-glycosylation is a form of cotranslational and posttranslational modification that occurs by enzymatic addition of a polysaccharide, or glycan, to an asparagine (N) residue of a protein. Examining a large set of experimentally determined mouse N-glycosylation sites, we find that the evolutionary rate of glycosylated asparagines is significantly lower than that of nonglycosylated asparagines of the same proteins. We further confirm that the conservation of glycosylated asparagines is accompanied by the conservation of the canonical motif sequence for glycosylation, suggesting that the above substitution rate difference is related to glycosylation. Interestingly, when solvent accessibility is considered, the substitution rate disparity between glycosylated and nonglycosylated asparagines is highly significant at solvent accessible sites but not at solvent inaccessible sites. Thus, although the solvent inaccessible glycosylation sites were experimentally identified, they are unlikely to be genuine or physiologically important. For solvent accessible asparagines, our analysis reveals a widespread and strong functional constraint on glycosylation, unlike what has been observed for phosphorylation sites in most studies, including our own analysis. Because the majority of N-glycosylation occurs at solvent accessible sites, our results show an overall functional importance for N-glycosylation.

FIG. 1.

Fraction of unconserved asparagine (N) residues in glycoproteins identified from (A) any of five mouse tissues and (B) each of the five mouse tissues. Error bars indicate standard errors. P values are from χ² tests. N⁺ sites are experimentally identified glycosylated sites, N⁰ sites are all other N sites in the same proteins, and N⁻ sites are those N⁰ sites that are not followed by canonical glycosylation motifs.

FIG. 2.

Proportion of conserved canonical glycosylation motifs. Error bars indicate standard errors.

FIG. 3.

Variation in substitution rate in the neighborhood of mouse N sites. Error bars indicate standard errors.

FIG. 4.

Substitution patterns of mouse N⁺ and N⁰ sites in (A) mouse–rat, (B) mouse–human, and (C) mouse–chicken comparisons. The dot indicates a deletion. P values are from χ² tests.

FIG. 5.

Correlation between the number of N⁺ sites per glycoprotein and the substitution rate ratio between N⁺ and N⁰ sites.

FIG. 6.

N-glycosylation at solvent accessible and inaccessible sites. (A) Frequency distributions of solvent accessibility scores for N⁺ and N⁰ sites. (B) Substitution rate ratio between N⁺ and N⁰ sites varies with solvent accessibility. *P < 0.05; **P < 0.01. P values are from χ² tests. (C) Fraction of unconserved canonical motifs. P values are from χ² tests.

FIG. 7.

Fraction of unconserved asparagine (N) residues depends on glycosylation, whereas that of unconserved serine (S), threonine (T), and tyrosine (Y) residues does not depend on phosphorylation. The comparison is made between mouse proteins that are subject to both glycosylation and phosphorylation and their rat orthologs. The number of residues considered in each sample is provided in each bar. Error bars indicate standard errors.