Figure 1.
NXS/T site densities and genome compositions.
(A,B) Actual NXS/T densities versus predicted densities from amino acid content in cytosolic and secretory proteins from species with high coverage genomes in Ensembl. Linearity analyzed by Pearson correlation. (C) Fraction of total Asn in NXS/T sites for each protein (mean ± SEM) in intracellular (cytosolic) and signal-sequence containing (secretory) proteins of five well annotated species. (D) NXS/T sites were binned by nearest neighboring, <20 then by increments of 10 amino acids. The difference between (sec) and cytosolic (cyt) bins bins reveals greater than expected proximity (skewness). The area under the line is ∼6.8% of all sites. (E) β1,6GlcNAc-branched N-glycans on lymphocytes from human, chimpanzee, C57/B6 strain mice, and zebra fish by L-PHA-FITC staining and FACS analysis. Each point is from an individual animal, except zebra fish where spleens were pooled.
Figure 2.
Encoding structure and simulations of NXS/T site gain and loss.
(A) Codon chart highlighting the coding structure of NXS/T. Asn is A-rich with 14 neighbors (red), while 10 Ser/Thr codons have 31 codon neighbors (blue), for 89% and 53%, probabilities of nonsynonymous mutation, respectively. There are 64 codons, of which 61 encode amino acids and 3 encode a signal for termination of a peptide sequence. The 10,416 possible XNXS/T, NXXS/T and NPS/T sequences are 3.01%, 1.59% and 0.03% of all possible coding 9-mer sequences (613), respectively. (B) Asymmetry in the two major paths of loss and gain. The coding 9-mers indicated in the XNXS/T and NXXS/T ovals are the subset of all coding 9-mer (% in brackets) are one mutation away from NXS/T. The solid arrows is the more likely pathway of interchange, while the broken arrows indicate the less probably pathway. The green arrows indicate how positive and purifying selection of NXS/T sites disrupts equilibrium in near-motifs. (C) The extracellular domains of 30 human secretory genes were serially mutated and subject to random selection used in the EvolveAGene 3.06 program (EVO-simulated, Y-axis). Simulated site gain in a set of human secretory proteins correlated with values calculated manually based on near-site numbers and paths into NXS/T (X-axis). (D) Site loss by EVO-simulation is proportional to actual sites in the initial sequences, and losses resulted in the expected ratio of XNXS/T to NXXS/T near-sites. Site gain by EVO-simulation correlated with (E) Asn but not Ser+Thr content, (F) A nucleotide content.
Figure 3.
Selection of sites, and departure from neutrality in near-NXS/T and A nucleotide.
(A–C) For human, mouse, and pufferfish, secretory and cytosolic proteins are grouped by actual NXS/T number, and near-sites were counted and expressed as density (mean/100 amino acids ±SE). XNXS/T (X≠P) densities of secretory and cytosolic proteins for human and mouse (grey shaded area) are significantly different by paired t-test. NPS/T represented only 0.03% of near-sites and was not significantly difference. Data for proteins with 0 to 15 sites is shown and represents 97.3% and 98.9% of all human sequences in secretory (n = 4440) and cytosolic (n = 15,626) proteins, respectively. (D) Difference in % (A) content for genes grouped by NXS/T number and expressed as secretory - cytosolic. Pearson correlation for mouse and human is p<0.0001. Note that a positive slope is expected in the absence of selection as exemplified by pufferfish.
Figure 4.
NXS/T multiplicity correlates with the evolutionary rates of secretory proteins.
(A) Human-mouse and human-pufferfish amino acid identities for gene orthologues grouped by human NXS/T number, linear regression slopes are indicated, * Pearson correlation p<0.05. (B) Apparent evolutionary rates (dN/dS) of human-mouse orthologues grouped by site number. (C) Human SNPs in cytosolic and secretory proteins represented as N/S. (D) Slopes from NXS/T multiplicity versus amino acid identity (secreted-cytosolic)/separation time from human), were graphed for 52 animal species (Fig. S7)]. Pearson R2 = 0.40; Spearman correlation R = 0.696. (E) Amino acid substitution rates are proportional to NXS/T site number. The slopes from human-mouse and human-pufferfish % identity were used to calculate substitutions.
Figure 5.
Acetyl-Lys site multiplicity and protein evolution.
(A) human proteins (n = 2058) grouped by number of acetyl-Lys sites (4802 total sites) from ref [31]. A-nucleotide content in acetyl-Lys modified proteins is greater than expected (grey). The expected values are based on the regression slope for Lys content versus A-nucleotide content in all cytosolic proteins. Mean (A) content is 27.7% in acetyl-Lys modified proteins and 25.8% in non-modified cytosolic proteins (yellow) (P<0.01). Higher than average (A) content in acetyl-Lys modified proteins may reflect “sequence conditioning” due to acetyl-Lys experimentation and hitchhiking of adaptive sequences by linkage (B) human-mouse amino acid identity and (C) evolutionary rates (dN/dS) for orthologues grouped by site number. P value is for Pearson correlation. (D) Protein interactions from Biogrid (mean ±SE) grouped by number of acetyl-Lys sites, or NXS/T sites in secretory proteins.
Figure 6.
Variation in NXS/T sites and selective constraint across codons in CD28 and CTLA-4.
Alignments for (A) CD28 and (B) CTLA4 along with line graphs depict the dN/dS ratio (ω) across sites as estimated by the M8 (PAML) and REL (HYPHY) models, for the extracellular domains of each gene. Glycosylation (NXS/T) sites are highlighted in black while XNXS/T and NXXS/T near-sites are highlighted in red and blue, respectively. Vertical dashed black lines are used to indicate the position of highly conserved NXS/T sites on the line graph. The horizontal dashed grey lines indicate neutral evolution at ω = 1, where sites below and above this line are under negative and positive selection, respectively. Sites estimated to be at ω >1 with a posterior probability over 95% are indicated with an asterisk. Site numbering follows human, with sites absent in the human sequence (due to insertions in other sequences in the alignment) marked with dashes (−). The M8 and REL estimates for ω broadly agree. Overall, CD28 shows higher levels of ω and a greater number of NXS/T and near sites, as compared to CTLA4. An increase in ω is often observed flanking the NXS/T sequences (more so in CD28 than CTLA4). Abbreviations–Calli., Callithrix; Micro., Microcebus; Dipodo., Dipodomys; Spermo., Spermophilus; Orycto., Oryctolagus; Ailuro., Ailuropoda; Mono., Monodelphis; Sarco., Sarcophilus; Ornitho., Ornithorhynchus; Taenio., Taeniopygia.
Figure 7.
Site repositioning and hitchhiking of conditional neutral sequence.
(A) Sequence compositions of CD28/CTLA-4 and TIE1/TIE2 paralogs, as well as the mean for secretory and cytosolic proteins. (B,C) The potential for site gain in the extracellular portions of the human genes was determined by simulated mutagenesis for CD28/CTLA-4 and TIE1/TIE2 paralogs. Sequences without near-sites (w/o) were created by replacement with GTA (Val) CCA (Pro) CTA (Leu), for five mutation steps away from NXS/T. Each bar is the mean ± SE of 10,000 runs with branch lengths in increments indicated by the upper limits on the X-axis (ie. mutations/position). (D) The (A) content in the intra- and extra- cellular portions of TIE1 and TIE2, grouping 53 species by clades. (E) NXS/T density comparing extracellular portions TIE1 and TIE2 by clades (Table S6).