A computational approach for deciphering the organization of glycosaminoglycans

Abstract

Background: Increasing evidence has revealed important roles for complex glycans as mediators of normal and pathological processes. Glycosaminoglycans are a class of glycans that bind and regulate the function of a wide array of proteins at the cell-extracellular matrix interface. The specific sequence and chemical organization of these polymers likely define function; however, identification of the structure-function relationships of glycosaminoglycans has been met with challenges associated with the unique level of complexity and the nontemplate-driven biosynthesis of these biopolymers.

Methodology/principal findings: To address these challenges, we have devised a computational approach to predict fine structure and patterns of domain organization of the specific glycosaminoglycan, heparan sulfate (HS). Using chemical composition data obtained after complete and partial digestion of mixtures of HS chains with specific degradative enzymes, the computational analysis produces populations of theoretical HS chains with structures that meet both biosynthesis and enzyme degradation rules. The model performs these operations through a modular format consisting of input/output sections and three routines called chainmaker, chainbreaker, and chainsorter. We applied this methodology to analyze HS preparations isolated from pulmonary fibroblasts and epithelial cells. Significant differences in the general organization of these two HS preparations were observed, with HS from epithelial cells having a greater frequency of highly sulfated domains. Epithelial HS also showed a higher density of specific HS domains that have been associated with inhibition of neutrophil elastase. Experimental analysis of elastase inhibition was consistent with the model predictions and demonstrated that HS from epithelial cells had greater inhibitory activity than HS from fibroblasts.

Conclusions/significance: This model establishes the conceptual framework for a new class of computational tools to use to assess patterns of domain organization within glycosaminoglycans. These tools will provide a means to consider high-level chain organization in deciphering the structure-function relationships of polysaccharides in biology.

Figure 1. Heparin lyase cleavage matrices for HS disaccharide-disaccharide linkages.

(A) Estimated cleavage probabilities for heparin lyase I. (B) Estimated cleavage probabilities for heparin lyase III. Each cell in (A) and (B) represents a disaccharide-disaccharide linkage with an indicated probability of cleavage by the specific heparin lyase. Turquoise cell  =  linkage not cleaved by lyase. Crosshatched cell  =  linkage prohibited by rules of biosynthesis.

Figure 2. Schematic diagrams of chainmaker and chainbreaker routines for generating HS chains.

(A) Chainmaker routine: (1) creates a base chain of N units of G0A0; (2) selects a disaccharide randomly from a pool based on experimental composition and places the disaccharide at a random position on the chain; (3) checks the positional constraints for disaccharide placement; (4) repeats steps 2–3 until placement occurs; (5) repeats steps 2–4 until all disaccharides from the pool are placed on the chain. The completed chain moves to the chainbreaker routine. (B) Chainbreaker routine: (6) receives the completed chain from the chainmaker routine; (7) initiates the first digestion (heparin lyase I or III) by comparing a random probability with the cleavage probability of each bond; (8) determines the broken bonds and releases disaccharides with cleavage on both sides; (9) repeats steps 7–8 for the second digestion; (10) cleaves all remaining bonds and releases all remaining disaccharides in the final digestion (heparin lyase II); (11) compares the released disaccharides from the first two digestions with the experimental breakdown constraints. If the constraints are satisfied, the successful chain moves to storage.

Figure 3. General patterns of domain organization predicted for fibroblast HS and epithelial HS.

(A) Chemical structure of the disaccharide containing 2-O-sulfated iduronic acid that defines the highly sulfated domain. All other disaccharides, including the remaining sulfated disaccharides, belong to the less sulfated domain. (B) Schematic diagram of HS chains showing domain organization. Yellow blocks are highly sulfated domains; blue blocks are less sulfated domains. Chain lengths are 250 disaccharides. Minimum block length is 1 disaccharide. (C) Size distribution of less sulfated domains in HS chains. N = 250 disaccharides and M = 100 chains. Although the distribution is shown for less sulfated domains up to 80 disaccharides in length, a few longer domains are present in both sets of HS chains. For epithelial HS, less sulfated domains extend to 87 disaccharides; for fibroblast HS, less sulfated domains extend to 113 disaccharides. Dashed lines indicate average sizes: E = 11.4±0.2 disaccharides (average ±95% confidence limits) for epithelial chains and F = 20.0±0.4 disaccharides for fibroblast chains.

Figure 4. Average domain sizes predicted for fibroblast HS and epithelial HS chains.

(A) Uncertainty in average domain size of less sulfated domains as a function of chain number. N = 50 and 250 disaccharides; M = 5–400 chains. % Uncertainty  =  [(±95% confidence limits)/(average domain size)] ×100. (B) Average domain size as a function of chain length for fibroblast HS. (C) Average domain size as a function of chain length for epithelial HS. N = 25–450 disaccharides and M = 100 chains for (B) and (C). Error bars show 95% confidence interval. (D) Schematic diagram of chains showing predicted domain patterns. Yellow blocks are highly sulfated domains; blue blocks are less sulfated domains. Each chain is above the critical length, and domain sizes are equilibrium values. For fibroblast HS, less sulfated domain  = 20.1±0.1 disaccharides (average ±95% confidence limits) and highly sulfated domain  = 1.22±0.02 disaccharides. For epithelial HS, less sulfated domain  = 11.2±0.2 disaccharides and highly sulfated domain  = 1.38±0.02 disaccharides.

Figure 5. Examples of average Fourier power spectra for HS chains.

Average power spectrum for fibroblast HS based on (A) 20 chains, (C) 40 chains, and (E) 60 chains. N = 256 disaccharides. Average power spectrum for epithelial HS based on (B) 20 chains, (D) 40 chains, and (F) 60 chains. N = 128 disaccharides. Each spectrum is normalized with respect to the highest response and is shown for the first half of the symmetric trace (k = 1, …, N/2).

Figure 6. Specific motif for inhibition of elastolysis predicted for fibroblast HS and epithelial HS.

(A) Chemical structure of the disaccharide pair that defines the cluster domain. Disaccharides excluded from the cluster domain belong to the connector domain. (B) Schematic diagram of HS chains showing domain organization with specific motif for inhibition of elastolysis. Yellow blocks are cluster domains; blue blocks are connector domains; dark blue blocks are connector domains that meet the size requirement for effective elastase inhibition. Chain lengths are 250 disaccharides. Minimum block length is 1 disaccharide. (C) Size distribution of connector domains in HS chains. N = 250 disaccharides and M = 100 chains. Crosshatched bars indicate lengths of connector domains (2–20 disaccharides) for effective elastase inhibition. Although the distribution is shown for connector domains up to 100 disaccharides in length, longer domains are present in both sets of HS chains. For fibroblast HS, connector domains extend to 226 disaccharides with an average size of 52±5 disaccharides (average ±95% confidence limits). For epithelial HS, connector domains extend to 184 disaccharides with an average size of 36±3 disaccharides. (D) Inhibition of elastolysis by GAG preparations. Relative rate  = (elastin digestion with inhibitor)/(elastin digestion without inhibitor). Bar height equals the average of duplicate readings; error bar shows the propagation-of-error estimate using standard errors. Control  =  no inhibitor; Hep  =  commercial heparin (17–19 kDa); HS  =  commercial heparan sulfate (8–10 kDa); Fibro HS  =  HS preparation from rat pulmonary fibroblasts; Epi HS  =  HS preparation from rat pulmonary epithelial cells. Reaction conditions: [Inhibitor]  = 5.0 µg/mL; [HNE (human neutrophil elastase)]  = 120 nM; [Elastin]  = 0.93 mg/mL; buffer  =  Dulbecco's phosphate-buffered saline without calcium and magnesium salts; temperature  = 37°C; time  = 4 hours; volume  = 1.073 mL.