The activity of HSulf2 on various bovine HS substrates was examined by disaccharide analysis using size-exclusion liquid chromatography-mass spectrometry (SEC-LC/MS), a technique that has previously been applied to the analysis of HS disaccharides from mammalian organs 
. The bovine samples included HS from lung, intestine, aorta, and two fractions from kidney that were purified using two different concentrations of salt (1.1 M or 1.25 M NaCl) for elution from anion exchange chromatography 
. The samples were digested exhaustively with HSulf2 and then subjected to exhaustive depolymerization using a mixture of heparin lyases I, II, and III. These conditions generate an array of disaccharides, designated subsequently using the structural code of Lawrence and Esko 
. This alphanumeric code (see Fig. S1
) consists of four digits that designate the structures of the uronic acid and glucosamine that comprise each disaccharide. The first position specifies whether the uronic acid is saturated (U) or Δ4,5
-unsaturated (D). The second digit specifies whether the uronic acid is unmodified (0) or modified by 2O-
sulfation (2). The third digit specifies whether the glucosamine is N-
acetylated (A) or N-
sulfated (S). The fourth digit specifies whether the glucosamine is unmodified (0) or modified by 6O-
sulfation (6). For example, D2S0 specifies the structure delta-uronic acid-2-sulfate-N-
sulfoglucosamine and U0A6 specifies the structure uronic acid-N
The SEC LC/MS method generated disaccharide analysis profiles consistent with those from other methods while producing additional information on NRE disaccharides and low abundance disaccharides containing an unmodified glucosamine primary amino group. For example, the SEC LC/MS method measured the abundance of the D2S6 disaccharide in rat brain as 4.2±0.1% 
. This value is consistent with the 3.9-5.4% range obtained by others for rat brain HS 
. Another group has found rat cerebrum and cerebellum HS to have between 6–8% of D2S6 
. Similar abundances are observed in mouse brain 
. It is likely that the small differences in results obtained using SEC LC/MS and other methods arise from the samples and the biases of the tissue extraction methods used in each laboratory.
The abundances of HS Δ4,5
-unsaturated disaccharides before and after digestion with HSulf2 as determined using SEC LC/MS are shown in . The relative abundances of control or HSulf2 treated disaccharides were calculated from peak area integration of EICs of disaccharides eluting from the SEC column. As expected, HSulf2 treated HS displayed a decrease in the abundance of D2S6, the principal disaccharide substrate of the Sulfs 
Disaccharide abundances before and after HSulf2 digestion.
The relative changes in the abundance of two pairs of isobaric disaccharides was determined using tandem MS, following the rationale of previously developed methods 
. The data show a small decrease in the D2A0/D0A6 pair due to HSulf2 digestion (). The D0A6 abundance of this pair, as determined using tandem MS, decreased in all samples except for intestine. These results demonstrate that D0A6 is a target for HSulf2. The percentage of D0S6 in the D2S0/D0S6 isomer pair decreased dramatically resulting from HSulf2 treatment, as determined using tandem mass spectrometry (). The data are consistent with the conclusion that HSulf2 converts D2S6 to D2S0 and D0S6 to D0S0. The overall increase in abundance of the D2S0/D0S6 pair is due to the increase in the abundance D2S0. The abundances of D0A0 are the same before and after HSulf2 treatment for all organs. This result demonstrates that HSulf2 digestion does not bias the activity of the heparin lyase enzymes used to depolymerize HS.
Changes in the profile of internal, Δ4,5-unsaturated disaccharides from the HSulf2 treated bovine organ HS samples are shown in . The differences are displayed as a natural log fold-change between control and HSulf2 treated samples. The abundance of each disaccharide, averaged over the five HS sources, is shown as a percentage of all Δ4,5-unsaturated disaccharides on the graph. As mentioned, the trisulfated disaccharide, D2S6, decreased in abundance for all organ HS populations. This change, however, occurred in an organ specific manner, with the greatest decrease occurring for bovine 1.25 M kidney HS, and the smallest decrease occurring for bovine aorta HS. The abundance of D2A6, a disaccharide found in NA/NS domains, was also significantly decreased in the organ HS populations, with the exception of HS from bovine intestine. Two other 6O-sulfate containing disaccharides, D0A6 and D0S6, occur as isomer pairs (with D2A0 and D2S0, respectively) in the MS mode.
Fold change plot of the abundances of Δ4,5-unsaturated disaccharides from control and HSulf2 treated bovine HS samples.
Saturated disaccharides resultant from the NRE of the bovine HS chains were also quantified using SEC-LC/MS, and differences in the abundance before and after treatment with HSulf2 are shown in . The abundance for each disaccharide, averaged over the five HS sources, is shown as a percent of all saturated disaccharides on the graph. This information can be used to evaluate the changes induced by HSulf2 at HS termini. The change in the amount of U2S6 (the saturated counterpart of D2S6) observed at the NRE after HSulf2 treatment differed from the change in D2S6 observed from the internal regions of the HS chains (). More specifically, the changes observed for U2S6 for bovine intestine and lung were much greater than for D2S6. Overall, saturated disaccharides represent a small fraction of the total disaccharides liberated by heparin lyases. The low MS ion abundances for saturated disaccharides preclude tandem MS experiments used for differentiation of disaccharides that occur as isomer pairs. Despite this, there are readily apparent differences in the profile of the U0S6/U2S0 isobar at the NRE compared to internal regions after HSulf2 treatment. The abundance of U0S6/U2S0 is lower for all samples following HSulf2 digestion, and this is accompanied by an increase in the abundance of U0S0. Again, these changes are organ specific. U2A6 was lower in abundance at the NRE for both of the bovine kidney samples, and only slightly less abundant for intestine, aorta and lung HS. An experiment using inactivated HSulf2 showed no changes in the abundances of HS disaccharides following lyase digestion, relative to samples in which no HSulf2 was added. These results show that the experimental digestion and workup conditions did not cleave or otherwise modify the HS chains.
Fold change plot of the abundances of saturated disaccharides from control and HSulf2 treated bovine HS samples.
Maccarana et. al.
have previously analyzed the same pool of bovine HS samples 
by profiling disaccharides produced from nitrous acid depolymerization. Under the conditions they used, disaccharides originating from NS domains were liberated from the HS chains. These data can be used to determine the profile of disaccharides that reside within these highly sulfated domains. We combined this information with the current data to determine the change in degree of sulfation for the NS domains of each of the bovine HS substrates that would occur upon treatment with HSulf2, as shown in . The black bars show the change in NS domain degree of sulfation that would occur if all NS domain 6O-
sulfates were liberated by HSulf2. The changes in degrees of sulfation of NS domains that occur based on the current disaccharide analysis () are shown in the white bars in . In all cases, the observed change in degree of sulfation does not reach the maximum possible value. The patterns for decrease in NS domain degree of sulfation are organ specific. Additionally, there are differences between the patterns observed for the extent of change in degree of sulfation when comparing the maximum vs. observed scenarios. For example, the order of percent decrease in degree of sulfation for the observed scenario is aorta < intestine < lung <1.1 M kidney <1.25 M kidney while the order for the maximum scenario is intestine < aorta < lung/1.1 M kidney <1.25 M kidney.
Changes in degree of sulfation of bovine HS NS domains due to HSulf2 digestion.
Of additional interest was the editing of the fine structure of other mammalian HS substrates by HSulf2. These experiments would determine whether the trends we observed for the editing of bovine HS by HSulf2 were also true for HS isolated by different approaches and from different animals. To this end, murine kidney and liver HS, known to be among the most sulfated of HS populations from this organism 
, were analyzed. Murine liver has been shown to contain a very high proportion of D2S6 
, measured previously as high as 17.0% and currently as 19.4% of total disaccharide content. Changes in the profile of Δ4,5
-unsaturated disaccharides produced from exhaustive heparin lyase digestion of murine HS with and without HSulf2 treatment are shown in . The abundance of each disaccharide, averaged over the two HS sources, is shown as a percentage of all Δ4,5
-unsaturated disaccharides on the graph. Interestingly, the pattern of change in the disaccharides after HSulf2 treatment is very similar to that observed for bovine HS (). Tandem MS revealed only modest desulfation of D0A6 from the murine samples (), which was observed as 11.1% for liver and 5.6% for kidney. As observed for the bovine HS samples, significant desulfation of D0S6 occurred (). The decreases in abundance of this disaccharide were 55.8% and 51.5% for liver and kidney, respectively.
Fold change plot of the abundances of Δ4,5-unsaturated disaccharides from control and HSulf2 treated murine HS samples. Control or HSulf2 treated murine HS samples were analyzed using SEC-MS.
HSulf2 induced similar changes to the NRE of murine HS as it did to the NRE of bovine HS, as shown in . The abundance of each disaccharide, averaged over the two HS sources, is shown as a percentage of all saturated disaccharides on the graph. For murine kidney and liver HS, the fold change of U2S6 detected at the NRE after HSulf2 treatment is similar to that of the corresponding internal disaccharide, D2S6 (). The decrease in the abundance of the U0S6/U2S0 isobar observed at the NRE of bovine organ HS chains is also present for the murine HS preparations. U2A6 decreased in abundance following HSulf2 treatment in the case of murine kidney, but not for murine liver.
Fold change plot of the abundances of saturated disaccharides from control and HSulf2 treated murine HS samples.
To further characterize the changes induced at the NRE of HS by HSulf2, the total abundance of Δ4,5-unsaturated or saturated disaccharides was compared before and after treatment of the samples with HSulf2. The mass spectral abundances from the SEC separations are expressed as a fold change for both bovine () and murine () HS samples. Small increases in the abundance of Δ4,5-unsaturated disaccharides were detected following HSulf2 treatment. Comparatively large increases were observed, however, for saturated disaccharides.
Influence of HSulf2 digestion on Δ4,5-unsaturated and saturated disaccharide abundances following lyase depolymerization.
To assess the possible origin of the increase in saturated structures, lyase resistant tetrasaccharides were examined. The SEC-UV traces showed small increases in absorbance in the range of elution time that corresponds to elution of tetrasaccharides (Fig. S2
). The SEC LC/MS data detected one major tetrasaccharide composition in all samples, [1,1,2,3,0] (the tetrasaccharide composition is given as [ΔHexA, Hex, GlcN, Sulfate, Acetate]), along with its saturated counterpart, [0,2,2,3,0]. The abundances of these structures were considerably higher in the bovine samples, and were too low in the murine samples for accurate quantitation. This is likely explained by the amount of HS injected onto the LC/MS system (higher for bovine samples than for murine samples, according to total ion counts) and the intensity of the tetrasaccharides was proportional to the overall signal for all disaccharides observed in the LC/MS datasets. With the exception of bovine intestine, an increase in the abundance of [1,1,2,3,0] and [0,2,2,3,0] was observed for all samples after treatment with HSulf2, as shown in . The abundances of [1,1,2,3,0] and [0,2,2,3,0] are observed to increase in the bovine HS populations treated with HSulf2, except for bovine intestine HS. Interestingly, when compared to its internal counterpart, [0,2,2,3,0] displayed a greater fold-change increase in abundance.
These lyase-resistant oligosaccharides represent a small percentage of domains in HS chains. It is known that disaccharide repeats containing a 3O
-sulfated GlcN residue resist heparin lyase cleavage 
. Such residues are required for anticoagulant function of heparin/HS. The changes in the abundances observed in that result from HSulf2 digestion indicated that the presence of 6O
-sulfate groups influences the susceptibility of the oligosaccharides to lyase digestion. The SEC LC/MS system is able to detect tetrasaccharides, while oligosaccharides of dp6 or longer cannot be detected because of their highly polydisperse nature. The data in and suggest that HSulf2 digestion changes the susceptibility of HS chains to lyase digestion. The increases in abundances for dp2 and dp4 in the HSulf2 digested samples indicate that the overall susceptibility of the HS chains to lyase digestion is increased by the removal of 6O-
sulfate groups. The data also show that the extent to which saturated dp2 and dp4 increase in abundance is greater than for their internal counterparts. This further suggests that HSulf2 displays increased activity towards the NRE than the internal oligosaccharides, taken as an average. The data do not rule out that there may be particular internal domains toward which HSulf2 is highly active. Thus, it appears that NRE domains have increased susceptibility towards HSulf2 relative to the average internal domains and that this pattern is present in most organs samples studied.
shows the total percentage of 6O-sulfate released from each of the bovine or murine HS preparations as a function of the total abundance of 6O-sulfates in each chain. In general, as the 6O-sulfate content increases, so does the amount of 6O-sulfate released. However, while murine liver, murine kidney, and bovine 1.25 M kidney have similar total 6O-sulfate content, there is considerable variation in the amount of 6O-sulfate released. The ability of HSulf2 to edit HS sulfation may be assessed in more detail by examining the 6O-desulfation of D2S6, as shown in . This figure shows the percent of 6O-desulfation of D2S6 as a function of the overall D2S6 content for a given preparation. The two bovine kidney preparations exhibit the highest degree of 6O-desulfation despite the fact that their initial D2S6 content is much lower than that of either murine kidney or murine liver HS. This indicates that D2S6 is desulfated in a manner dependant on the domain context in which it resides.
Effect of HSulf2 digestion on abundances of lyase-resistant tetrasaccharides in exhaustive digests of bovine HS.
Comparison of the extent of HSulf2 mediated release of 6O-sulfate from bovine and murine HS samples.