Discovery of exolytic heparinases and their catalytic mechanism and potential application

Heparinases (Hepases) are critical tools for the studies of highly heterogeneous heparin (HP)/heparan sulfate (HS). However, exolytic heparinases urgently needed for the sequencing of HP/HS chains remain undiscovered. Herein, a type of exolytic heparinases (exoHepases) is identified from the genomes of different bacteria. These exoHepases share almost no homology with known Hepases and prefer to digest HP rather than HS chains by sequentially releasing unsaturated disaccharides from their reducing ends. The structural study of an exoHepase (BIexoHep) shows that an N-terminal conserved DUF4962 superfamily domain is essential to the enzyme activities of these exoHepases, which is involved in the formation of a unique L-shaped catalytic cavity controlling the sequential digestion of substrates through electrostatic interactions. Further, several HP octasaccharides have been preliminarily sequenced by using BIexoHep. Overall, this study fills the research gap of exoHepases and provides urgently needed tools for the structural and functional studies of HP/HS chains.


Reviewer #2 (Remarks to the Author):
This manuscript described by Zhang et al. deals with molecular identification, structure determination, and potential application of exotype heparinases (exoHeps). Recombinant proteins classified into subgroup 2 of polysaccharide lyase family 15 (PL15_2) were identified to be heparin lyases exolytically acting on heparin oligosaccharides rather than heparin and haparan sulfate polysaccharides, although these proteins have previously been annotated to such enzymes due to the presence of heparinase II/III-like motif. Based on determination of crystal structure of the Bacteroides intestinalis PL15_2 exoHep, exoHeps were found to include the additional N-terminal small β-sheet domain distinct from other structurally resembled heparinases (Hep II and Hep III). This β-sheet domain contributes to formation of L-shaped active cleft probably involved in the exolytic activity. Substrate specificity of the exoHep using various heparin oligosaccharides was thoroughly investigated, indicating that this enzyme become a powerful tool for determination of structure of heparin, while physiological function of the enzyme in the bacterial cells remains to be clarified.
The work is technically sound, and advances the knowledge in this field, indicating the results obtained here are considered to be informative and valuable. On the other hand, several significant concerns are raised as follows: Major points:

1) Exotype heparinase
The authors mentioned that all identified heparinases belong to the family of endolytic lyases (lines 74-76), while there are few reports describing the endolytic activity of heparinases. Some heparinases (e.g. the enzyme from Bacteroides stercoris HJ-15) have been found to produce unsaturated heparin disaccharides as main products from "heparin polysaccharides", although it is unclear whether these disaccharides are final products or not. In this work, the authors determined the mode of action (exotype) of PL15_2 enzymes by using "heparin oligosaccharides (DP13)" as a substrate, though these enzymes can act on heparin polysaccharide (lines 158-159). The authors should also show time-dependent degradation profile (similar to Figure 1) by using heparin polysaccharides as a substrate. If PL15_2 enzymes show an exotype activity, no other products except for unsaturated disaccharides will be observed at any reaction time (min and h). In Figure S2, a peak at retention time of around 18 min was observed only in the presence of enzymes. What is the peak? The authors specify the product.

2) Structural determinants for exolytic activity
The authors mentioned that the additional N-terminal small β-sheet domain is essential to exolytic activity (lines 328-329). If the truncated enzyme with a lack of the small β-sheet domain shows an "endolytic" activity but not loses the enzyme activity, this mention will be understood. Regarding exo/endotype polysaccharide lyases, similar studies have previously been reported in PL11 lyases (JBC, 284, 10181-10189, 2009). In this paper, exotype polysaccharide lyase has been converted to endotype lyase through identification of structural determinants of these lyases for mode of action. Thus, the authors should demonstrate structure-based conversion of exotype to endotype or vice versa in this work. The readers of this Journal are interested in the mode of action of the PL13, PL21, and PL12 haparinases with an addition of small β-sheet domain.

Minor points:
3) Line 24: "seemingly" is in contradiction with lines 74-76. 4) Line 29: "is essential to the exolytic activity" should be changed to "is essential to the enzyme activity". 5) Line 50: "N-acetyl" should be changed to "N-acetyl". N is in an italic form.  (Table 1). Did the authors examine the effect of EGTA on the enzyme activity? The significance of calcium ions should be discussed. 10) Line 262-272: The role of "exit structure" by negatively charged residues was described, while this is not direct evidence for release of the product from "exit". The authors are encouraged to show the enzyme activity of the mutants with acidic residues (Asp and Glu) to basic and acidic (Glu and Asp) residues. 11) Line 271: " Fig S5a" should be changed to " Fig S10a". 12) Line 324: (I, II and II) to (I, II and III). 13) Lines 363-373: The authors mentioned that L-shaped tunnel include the entrance by positively charged residues and the exit by negatively charged residues. In relation to 10), replacement of acidic residues with basic residues in the exit structure may provide useful information on the role of the exit. The positively charged exit region of the mutants would exhibit the affinity with the product, indicating the role of the exit as pathway for release of the product. 14) Line 436: sequence to sequences.

21) Figure1
: What is small peak "Di"? The authors should isolate the product and determine its molecular weight. Figure S2: The authors should show the time-dependent degradation profile. What is the peak at retention time of around 18 min?

Reviewer #3 (Remarks to the Author):
The authors discovered a novel Hepase with exolytic activity, exoHep. This exoHep showed high activity against hepase III resistant heparin fractions, relatively low activity against the heparin and extremely weak activities towards HS. The authors confirmed the exolytic cleavage and substrate specificity using hepase III resistant heparin and tetrasaccharide. This enzyme degrades highly sulfated HP from the reducing end. This is very unusual as nearly all enzymes acting exolytically on carbohydrates act from the non-reducing end.
The authors elucidated the protein structure and protein-heparin interactions through crystallization of SeMet labeling followed by single wavelength anomalous dispersion phasing.
Finally, the authors used this exoHep successfully sequenced the 5 octasaccharide. They proposed the mechanism of the exolytic mode of exoHep and strategy for exo-sequencing using exoHep.
This new exoHep provides a new tool for HP/HS sequencing.
The manuscript is well written.
1. The most unusual feature of these exo enzymes is that they act from the reducing end. While the author's data suggest this is the case they should apply a second method to confirm. The author's should test whether it is possible to disrupt enzyme action by first reducing the hemiacetal at the oligosaccharides reducing end or by reductively aminating the reducing end with 2-aminoacridone. This would help confirm that action does take place from the reducing end.
2. The authors should also use RI detector instead of UV 232nm used in Fig 1? This way they can see both oligosaccharide and disaccharide changing at the same time.
3. What's the purity of the five octasaccharides? Did author confirm the octasaccharide chain using another sequencing method for comparison to the exoHep sequencing method? 4. Figure 5, the color of exoHep is too light. It is suggested to use a darker background.
Our responses to the reviewers' comments have been described below one by one referring to the pages and locations of the marked manuscript in yellow.

Reviewer # 1 (the main findings of the study):
The manuscript provides a substantial contribution to the limited tool set for the analysis of heparin/HS. Currently most labs use three enzymes to degrade polymers of 20-200 disaccharides in length. Considering bacteria edit heparin / HS chains to gain entry into the cells, there is huge potential for the discovery of enzymes to sequence GAG chains which is relatively unexplored. This manuscript provides a stepping stone for that requirement, in identifying multiple heparin exoenzymes, which can to utilised in the overall goal of glycosaminoglycan sequencing to further understand the interplay of numerous biological processes. The tetrasaccharide sequencing and analysis is excellent, characterisation of the enzymes using multiple analytical methods is of a high standard, the only weakness in the manuscript is the octasaccharide sequencing, which would benefit from MS/MS or NMR data to confirm interpretation of the HPLC disaccharide analysis. Overall, this is a manuscript that would be of high interest to the heparin field.

Our response:
Thanks a lot. We have changed as suggested.
2. Line 37 -Please add "family" belonging to the glycosaminoglycan family.

Our response:
Thanks. We have added as suggested.

Our response:
Thanks. We have corrected it.
4. Line 55 -Can the authors remove the number "32" as I'm assuming this is excluding the free amines, which is perfectly acceptable, however, the theoretical disaccharide number is debated. It is also unlikely that all 32 exist, so a theoretical number should include all combinations.

Our response:
Thanks. We have removed and corrected as suggested.
5. Figure 1, please add a chromatogram for the starting heparin dp13 material. Please also run a disaccharides with 2 sulfates, one sulfate and 0 sulfates as a calibration curve for easier interpretation to the reader. Possibly add dp2, dp4, dp6, dp8, dp10 calibration SEC chromatogram. Please added SEC separation within the title.

Our response:
Thanks for your concerns. As we know, the saturated HP DP13 does not have specific absorbance at 232 nm, and thus no obvious signal corresponding to it is observed in the chromatograms. To show the presence of substrate in Fig. 1, we have showed the SEC chromatogram of the saturated HP DP13 detected at 210 nm. In addition, we have added a SEC calibration curve of size-defined HP oligosaccharides including HP UDP2 (disaccharides with 3 sulfates, 2 sulfates, one sulfate and 0 sulfate), UDP4, UDP6, UDP8, UDP10, UDP12 and UDP14 in the Supplement data (the new Supplementary Fig. S3). At the meantime, we also added SEC separation in the corresponding figures as you suggested.
6. Line 518 -As previously reported -there is no reference Our response: Thanks. We have added the reference. 11. General enquiry, could the authors comment on the chromatography performance of the ProPAC PA1 column vs the Pack Polyamide II column. This is of generally interest, and can be placed in methods, but not necessary.

Our response:
Thanks for your concerns. Based on our experience, compared with the Pack Polyamine II column, the ProPAC PA1 column shows better capacity for the separation of HP/HS oligosaccharides in particular bigger oligosaccharides such as hexa-and octasaccharides.
However, considering the fact that the Pack Polyamine II column is much cheaper than the ProPAC PA1 column and good enough to separate HP/HS disaccharides and tetrasaccharides, thus when we analyze small HP/HS oligosaccharides we prefer to use Pack Polyamine II column rather than ProPAC PA1 column.
12. Could the authors please provide an explanation on whether P8-4 could have the UA-GlcNS6S in the middle of the sequence for example, ∆4HexUA2S1-4GlcNS6S1-UA1-4GlcNS6S1-4HexUA2S1-4GlcNS6S1-4HexUA2S1-4GlcNS6S. Perhaps MS/MS data or NMR data should be provided to complement the HPLC data.

Our response:
Thanks for your concerns. We have performed the MS/MS analysis of all five HP octasaccharides to complement the HPLC sequencing data of (the new supplementary 13. Please upload raw Mass Spec data files for intact dp8 oligosaccharides.

Our response:
Thanks for your concern. We have upload the raw Mass Spec data files for each intact dp8 oligosaccharide to figshare and can be accessed by the link https://figshare.com/s/3c029e519df2d0b1625d.
14. S12 ( c ) The Dp4 shows that one disaccharide is observed, therefore must contain two GlcNAc disaccharides, in table S5 there is only 1 disaccharide. Could the authors please clarify there chromatograms further.

Our response:
Thanks a lot. Based on your concerns we noticed this problem. To clarify this question, we have repeated this experiment and found that the DP4 composed of two disaccharides showed that the octasaccharide ∆4HexUA1-4GlcNAc3S1-UA1-4GlcNS6S1-4HexUA2S1-4GlcNS6S1-4HexUA1-4GlcNS6S was the main component of P8-3.
15. S12 -for each disaccharide profile, it is perhaps possible to determine the reducing and non-reducing end disaccharides in some cases, but it is unclear on how the two middle disaccharides are deduced to generate the overall sequence. MS/MS or NMR data could easily supplementary this disaccharide analysis to provide confident octasaccharide sequences.

Our response:
Thanks for your concerns. As mentioned above, we have added the MS/MS data of each sequenced octasaccharide to complement the HPLC data.

Reviewer # 2 (the main findings of the study):
This Substrate specificity of the exoHep using various heparin oligosaccharides was thoroughly investigated, indicating that this enzyme become a powerful tool for determination of structure of heparin, while physiological function of the enzyme in the bacterial cells remains to be clarified.
The work is technically sound, and advances the knowledge in this field, indicating the results obtained here are considered to be informative and valuable. On the other hand, several significant concerns are raised as follows: Reviewer #2 (Revisions needed): (Comments)

Exotype heparinase
The authors mentioned that all identified heparinases belong to the family of endolytic lyases (lines 74-76), while there are few reports describing the endolytic activity of heparinases. Some heparinases (e.g. the enzyme from Bacteroides stercoris HJ-15) have been found to produce unsaturated heparin disaccharides as main products from "heparin polysaccharides", although it is unclear whether these disaccharides are final products or not. In this work, the authors determined the mode of action (exotype) of PL15_2 enzymes by using "heparin oligosaccharides (DP13)" as a substrate, though these enzymes can act on heparin polysaccharide (lines 158-159). The authors should also show time-dependent degradation profile (similar to Figure 1) by using heparin polysaccharides as a substrate.
If PL15_2 enzymes show an exotype activity, no other products except for unsaturated disaccharides will be observed at any reaction time (min and h). In Figure S2, a peak at retention time of around 18 min was observed only in the presence of enzymes. What is the peak? The authors specify the product.

Structural determinants for exolytic activity
The authors mentioned that the additional N-terminal small β-sheet domain is essential to exolytic activity (lines 328-329). If the truncated enzyme with a lack of the small β-sheet domain shows an "endolytic" activity but not loses the enzyme activity, this mention will be understood. Regarding exo/endotype polysaccharide lyases, similar studies have previously been reported in PL11 lyases (JBC, 284, 10181-10189, 2009). In this paper, exotype polysaccharide lyase has been converted to endotype lyase through identification of structural determinants of these lyases for mode of action. Thus, the authors should demonstrate structure-based conversion of exotype to endotype or vice versa in this work.
The readers of this Journal are interested in the mode of action of the PL13, PL21, and PL12 heparinases with an addition of small β-sheet domain.

Our response:
Thanks for your good suggestions. As mentioned in the manuscript, the deletion of the Nterminal small β-sheet domain will leading to the thorough inactivation of BIexoHep, meaning that the additional N-terminal small β-sheet domain is essential to the activity of BIexoHep. Additionally, we also tried to link the small β-sheet domain to the PL13, PL21, and PL12 heparinases and test the activities of the engineered enzymes. However, we found that the grafting resulted in the thorough inactivation of the PL13 (E-HepI) and PL21 (E-HepII) heparinases ( figure A and B below). In contrast, the grafting mutation did not cause the inactivation of PL12 heparinases (E-HepIII) (figure B), but the action pattern of E-HepIII was not affected obviously (figure C) compared with that of the wild HepIII ( figure   D). These results indicate that the additional N-terminal small β-sheet domain is not the decisive factor for the exolytic activity of the PL15_2 family enzymes. While, interestingly, we obtained some other progresses in the followed mutation experiments. We attempted to mutate the acidic residues Asp in the exit tunnel to uncharged Asn or the basic residue His so that the chargeability of the exit tunnel was converted to uncharged or positively   3. Line 24: "seemingly" is in contradiction with lines 74-76.

Our response:
Thanks a lot. We have revised the corresponding sentence.
4. Line 29: "is essential to the exolytic activity" should be changed to "is essential to the enzyme activity".

Our response:
Thanks for your suggestion. We have revised the sentence as you suggested.
5. Line 50: "N-acetyl" should be changed to "N-acetyl". N is in an italic form.

Our response:
Thanks a lot. We have corrected it.

Our response:
Thanks for your concern. We have renamed exohep to BIexoHep in order to indicate its source (Bacteroides intestinalis).

Our response:
Thanks for your concern. The three proteins BRexoHep, BXexoHep, and PHexoHep possess the sequence similarities of 60.57%, 47.79% and 31.13% with BIexoHep (old name exoHep), respectively, and have the conserved active residues found in all exoHepases, but they have no activity on all tested polysaccharides. Frankly, we do not know the exact reason why these three proteins are inactive, which may be due to some difference in three dimensional structure compared with the active exoHepases. In fact, the current method to find new enzymes or other functional proteins based on the homology comparison of primary sequences often leads to the cloning and expression of many proteins with no activity or unknown function. Thus, it is not enough to predict the activity of a protein based on the primary structure alone, and further structural and functional study is necessary.  (Table 1). Did the authors examine the effect of EGTA on the enzyme activity? The significance of calcium ions should be discussed.

Our response:
Thanks for your concern. As shown in Table 1 10. Line 262-272: The role of "exit structure" by negatively charged residues was described, while this is not direct evidence for release of the product from "exit". The authors are encouraged to show the enzyme activity of the mutants with acidic residues (Asp and Glu) to basic and acidic (Glu and Asp) residues.

Our response:
Thanks for your good suggestion. We have made a series of mutants by replacing the acidic residues Asp 70 , Asp 281 and Asp 335 in the exit with residue His or Asn, respectively, and the enzyme activities of the mutants were affected to varying degrees (supplementary

Our response:
Thanks a lot. We have corrected this mistake.

Our response:
Thanks a lot. We have corrected it.
13. Lines 363-373: The authors mentioned that L-shaped tunnel include the entrance by positively charged residues and the exit by negatively charged residues. In relation to 10), replacement of acidic residues with basic residues in the exit structure may provide useful information on the role of the exit. The positively charged exit region of the mutants would exhibit the affinity with the product, indicating the role of the exit as pathway for release of the product.

Our response:
Thanks for your good suggestions. Based on your suggestion, we have mutated three acidic residues Asp 70 , Asp 281 and Asp 335 together in the exit tunnel to the basic residue His or Asn, which converted the electrical property of the exit tunnel from negative to positive or uncharged, and the corresponding mutants BIexoHep-D70H-D281H-D335H and BIexoHep-D70N-D281N-D335N both show too low enzyme activities to be accurately measured (supplementary Fig. S14a), which provide biochemical evidence to the role of the exit as pathway for release of the product, as you kindly suggested.
Our response: Thanks. We have corrected it.
Our response: Thanks. We have added the molar extinction coefficient.
Our response: Thanks. We have corrected it.

Line 633: Remove 2003?
Our response: Thanks for your concern. The "2003" are part of the title and thus cannot be deleted.

Our response:
Thanks a lot. We have inserted the space.
Our response: Thanks. We have removed the dot.

Our response:
Thanks for your concern. The authors used "delta" in the title and thus we cannot change it to "∆".

Figure1
: What is small peak "Di"? The authors should isolate the product and determine its molecular weight.
Our response: Thanks. The "Di" is an abbreviation for HP disaccharides that can be determined by 22. Figure S2: The authors should show the time-dependent degradation profile. What is the peak at retention time of around 18 min?

Our response:
Thanks for your suggestion and concern. We have shown the time-dependent degradation profile of the enzymes, as shown in the new Supplementary Fig. S4. The peak at retention time of around 18 min in supplementary Figure S2 represents polysaccharide/large oligosaccharide and enzyme protein that eluted in the void volume, which is often detected in the size exclusion chromatogram, and we specified the peaks in supplementary Fig. S2 as suggested.

Reviewer # 3 (the main findings of the study):
The authors discovered a novel Hepase with exolytic activity, exoHep. This exoHep showed high activity against hepase III resistant heparin fractions, relatively low activity against the heparin and extremely weak activities towards HS. The authors confirmed the exolytic cleavage and substrate specificity using hepase III resistant heparin and tetrasaccharide. This enzyme degrades highly sulfated HP from the reducing end. This is very unusual as nearly all enzymes acting exolytically on carbohydrates act from the nonreducing end.
The authors elucidated the protein structure and protein-heparin interactions through crystallization of SeMet labeling followed by single wavelength anomalous dispersion phasing.
Finally, the authors used this exoHep successfully sequenced the 5 octasaccharide. They proposed the mechanism of the exolytic mode of exoHep and strategy for exo-sequencing using exoHep.
This new exoHep provides a new tool for HP/HS sequencing.
The manuscript is well written.

Reviewer #3 (Revisions needed):
(Comments) 1. The most unusual feature of these exo enzymes is that they act from the reducing end.
While the author's data suggest this is the case they should apply a second method to confirm. The author's should test whether it is possible to disrupt enzyme action by first reducing the hemiacetal at the oligosaccharides reducing end or by reductively aminating the reducing end with 2-aminoacridone. This would help confirm that action does take place from the reducing end.

Our response:
Thanks for your good suggestion. Based on your suggestion, we labeled the reducing end of the HP DP13. As the results shown in Supplementary Fig. S5, the 2-AB-labeling at the reducing ends of HP DP13 completely inhibited the action of the PL15_2 Hepases compared with the quick degradation of unlabeled HP DP13 by these enzymes (Fig. 1a-e), indicating that the introduction of 2-AB group at the reducing end of HP chain hindered the action of the PL15_2 Hepases and further confirmed that these enzymes are exolytic Hepases that act from the reducing ends of substrates.
2. The authors should also use RI detector instead of UV 232 nm used in Fig 1? This way they can see both oligosaccharide and disaccharide changing at the same time.

Our response:
Thanks for your concern and suggestion. As you suggested, we can detect both oligosaccharide and disaccharide changing by using RI detector, but this way will make the results very confusion due to the appearance of many peaks corresponding to oligosaccharides with different size, which will seriously disrupt the judgment of the exo-or endolytic characters of the enzymes. In fact, compared with RI detector the biggest advantage of UV232 nm is the ability to distinguish the unsaturated saccharides produced by enzyme from the saturated substrates such as the saturated HP DP13 or polysaccharide, otherwise we cannot judge the peaks from the reducing ends or nonreducing ends of the test substrates. Taken together, the application of UV232 nm can make the judgment of substrate-degradation mode of the enzymes much easier, and the only disadvantage is difficult to detect the changing of the saturated substrates as you concerned. To show the presence of substrate in Fig. 1, we have showed the SEC chromatogram of the saturated HP DP13 detected at 210 nm that is much more sensitive than RI detector and can save precious sample a lot.
3. What's the purity of the five octasaccharides? Did author confirm the octasaccharide chain using another sequencing method for comparison to the exoHep sequencing method?
Our response: Thanks for your concerns. As shown in the manuscript, the purity and compositions of the 4. Figure 5, the color of exoHep is too light. It is suggested to use a darker background.

Our response:
Thanks a lot. We have changed the color as you suggested.