A SARS-CoV-2 antibody curbs viral nucleocapsid protein-induced complement hyperactivation

Although human antibodies elicited by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) nucleocapsid (N) protein are profoundly boosted upon infection, little is known about the function of N-reactive antibodies. Herein, we isolate and profile a panel of 32 N protein-specific monoclonal antibodies (mAbs) from a quick recovery coronavirus disease-19 (COVID-19) convalescent patient who has dominant antibody responses to the SARS-CoV-2 N protein rather than to the SARS-CoV-2 spike (S) protein. The complex structure of the N protein RNA binding domain with the highest binding affinity mAb (nCoV396) reveals changes in the epitopes and antigen’s allosteric regulation. Functionally, a virus-free complement hyperactivation analysis demonstrates that nCoV396 specifically compromises the N protein-induced complement hyperactivation, which is a risk factor for the morbidity and mortality of COVID-19 patients, thus laying the foundation for the identification of functional anti-N protein mAbs.

With the rise of escape mutants, alternate approaches in diagnostic and therapeutic development efforts need consideration. This work has implications in diagnostic and therapeutic areas and provides important data linked to quick recovery, important at this stage of the global pandemic. The data overall is suited for the readership of Nature Communications and can be accepted upon fulfilment of the requested changes. The suggestions are provided to help the authors improve the analyses of certain sections of this work.
Line 110: Please rephrase to 'To take advantage of patient ZD006' to 'To maximize analysis of patient ZD006 samples' or similar.
Line 163-164: Were the mAbs purified as their respectively identified IgG clonotypes or all as IgG1? Kindly mention briefly in the results section while pointing to Methods or other relevant section.
Section 'Complex structure of mAb with N-NTD': This is an important and interesting section of the manuscript, that could provide valuable insights to the readership with additional analysis.
• For example, authors should briefly mention here, by which method they solve the crystal structure, what was used as their homology model, or if it was done using direct methods (if so details of it), if they have renumbered the antibody according to the Kabat nomenclature, etc. • Authors should analyze and comment on how the constriction of the linker region by the binding mAb-nCoV396 impede viral function. • Do other N protein targeting mAbs show similar mechanisms of action in published data for SARS2, SARS, MERS etc? • The interactions of the variable region of the mAb should be further elaborated and compared to other published structural data.
Extended data table 4: Please mention the descriptions of the statistical terms under the table. Is I/ really that or <I>/<>? Also, please list data collection parameters and further refinement statistics on the Statistics Table (instead of the Methods), like Ramachandran score, for proper evaluation of data quality.
Lines 210-: The authors hypothesize and biophysically characterize cross-interaction between mAb-nCoV396 and two other related coronaviruses. Is there evidence of cross-neutralization of the same viruses with this mAb? Extended figure 2: Please explain the special characters in the consensus sequence and "TT" label on the N-NTD sequence. Figure 3: There is considerable movement between the C-terminal region of the unbound and mAbbound N-NTD states. The authors suggest allosteric modulation on the full-length N-protein to be a consequence of this movement. Have they (or others) experimentally observed such changes that links directly to function? How does the enlarged RNA binding pocket modulate viral function?
Extended data table 1: It would be very helpful if the number of analyzed heavy chain-light chain pairs isolated from the respective recovered patients would be tabulated on this table.
Discussion: Comprehensive analysis needs to be included comparing this present data to other publications and available data against the N-protein of SARS CoV2 and related coronaviruses like SARS or MERS. Structural, immunological and biophysical data available for relevant comparisons should be discussed.
Reviewer #3 (Remarks to the Author): General Comments: In this manuscript, Kang et al utilize the convalescent plasma of 6 patients who recovered from SARS-CoV2 infection to profile the antibody responses in the early recovery phase of infection. They subsequently focus on a patient who had a rapid recovery (ZD006), and identify that this patient had dominant antibody responses to the SARS-CoV2 N-protein. They conduct detailed structural analyses for these antibodies, which provide novel insights into how one of these antibodies (nCoV396) binds to the RNA binding domain of the N-protein and subsequently, provide evidence (using a virus-free in vitro system) how a specific monoclonal antibody against the N protein may affect N protein-induced complement hyper activation. This manuscript advances the field in helping understand the interaction of the mAb to the N protein RNA binding domain and how it may potentially modulate disease severity, in which lies its conceptual innovation. Their in vitro systems provide technical innovation, although have inherent limitations of a virus free system. Overall, the statistical analysis is appropriate and valid. Reproducibility of some of the functional assays could be improved as detailed below.
MAJOR COMMENTS: 1. The authors use a C2 internal quenching fluorescent peptide-based analysis for ex vivo complement activation. However, this system needs greater descriptive detail. For example, is this system truly representative of MASP-2 protease mediated cleavage of C2, and if so, how [in terms of design, validation]? 2. Towards the end of the results section, the authors state "in conclusion, these results demonstrate that ...not only by facilitating Vmax of MASP-2 catalytic activity but also enhancing substrate binding specificity in the reactions". Is there an independent way to validate these findings, and how can they be certain this is the case? For example, is there a positive control for MASP-2 catalytic activity that has been utilized? Either that, or another assay for MASP-2 mediated cleavage would be important. Along those lines, is there a way to assess how the binding of nCoV396 to the N-protein modulates inflammation (including but not restricted to markers of complement activation) either in vivo or ex vivo? 3. The authors use five other serum samples from autoimmune disease donors. These samples require additional description. For example, what is 'autoimmune disease' and why were these chosen? How do they compare to healthy donors? And how do existing levels of complement activation markers and serine proteases (including MASP-2) in these specimens modulate the functional analysis? 4. The authors selected nCoV396, nCoV416 and nCOV457 for production of recombinant Fab antibodies for purposes of functional and structural characterization. It is unclear why they subsequently chose to focus on nCoV396 for purposes of testing complement hyper activation. Do they have data on the two other antibodies and do those results affect the conclusions? MINOR COMMENTS: 1. Page 4 -the authors state "a recent retrospective observational study of ....revealed that complement disorder was associated with"....Please be more specific regarding this observation. 2. Page 4 -A recent preprint study....it is good that the authors have acknowledged this study is a preprint study. It is somewhat concerning this study was uploaded on a preprint server in March 2020 but has yet not been published, although is widely cited. To that extent, at the end of this paragraph, the authors should add a sentence about how the current study can help further advance the role of N-protein inducing complement activation. They should also subsequently address this in detail in the discussion, possibly via focusing on the RNA binding domain of the N-protein. Do the authors suggest that this domain is what is contributing to complement-mediated hyper activation? Or how do they put the findings of their current study, into perspective with the two prior studies (references 19 and 21)? 3. The authors mention that ZD004 and ZD006 had only minimal levels of antibody response to the S protein, but much higher titers to the N protein. Hence, how likely is the current schema proposed valid in the human population? The authors may want to consider incorporating references such as McAndrews et al. (https://insight.jci.org/articles/view/142386), among others, into their Discussion to address this issue. 4. Regarding the antibody repertoires, how did the authors ensure that there was no cross-reactivity of monoclonal Abs to other viral proteins? Additionally, among 32 mAbs that bound to N-FL, 13 Abs bound to N-NTD and one Ab bound to N-CTD, but which part of the protein did the remaining 18 Abs bind to? Response: We thank this reviewer for appreciating the novelty of the first structure of SARS-CoV-2 N protein bound to a human monoclonal antibody. We are also grateful that the reviewer highlighted the coupling of structural work with ex vivo functional assays in this study.
1. The authors picked nCoV454 as one of the seven mAbs that bind to N-NTD, but actually it did not show N-NTD binding (in Figure 2b). Please clarify.
Response: We apologize for this mistake and thank this reviewer for suggestions to correct the manuscript.
NCoV454 indeed binds to full-length N protein instead of the N-NTD. We have updated the sentence involving nCoV454 at line 164 -165 in the revised manuscript.  Response: We thank this reviewer for considering our contribution to be very exciting and our experimentation to be greatly sound and it is suited for Nature Communications.
1. Line 110: Please rephrase to 'To take advantage of patient ZD006' to 'To maximize analysis of patient ZD006 samples' or similar.
Response: We concur with the reviewer's suggestion and have rephrased this sentence.
2. Line 163-164: Were the mAbs purified as their respectively identified IgG clonotypes or all as IgG1?
Kindly mention briefly in the results section while pointing to Methods or other relevant section.
Response: We thank the reviewer's valuable suggestion. All mAbs were produced as IgG1 antibodies regardless of their original Ig isotypes (Reference 30 in manuscript). We have added this sentence 'All of the purified antibodies were produced as IgG1 antibodies regardless their original Ig isotypes' at Line 123-124 and a slightly more detailed description in the Method section.
3. Section 'Complex structure of mAb with N-NTD': This is an important and interesting section of the manuscript, that could provide valuable insights to the readership with additional analysis. (2) Authors should analyze and comment on how the constriction of the linker region by the binding mAb-nCoV396 impede viral function.
Response: We are grateful that this reviewer appreciates novel aspects of the overall structural stabilization in the context of the linker region's constriction. From the reported NTD structures of SARS-CoV (45-181aa; PDB: 1SSK), MERS (1-164aa; PDB:4UD1), and MHV (60-197aa PDB:3HD4) N protein, we found that they all contain the regions which are corresponding to 162-170 residues of SARS-CoV-2.
These results indicate that 162-170 residues are indispensable for the structure and function of N protein.
We also found that 162-170 is the recognized region for the antibodies binding to NTD or full-length of N protein through N-derived epitope peptide ELISA analysis (Supplementary Table 1 Table 1

).'
Furthermore, we also verified that other two antibodies (nCoV454 and nCoV457) with the same binding site as nCoV396 could inhibit the complement hyperactivation caused by SARS-CoV-2 N protein. The experimental results are shown in Supplementary Fig. 3a, b. Therefore, we believe that the region of 162-170 is very important for the virus function. Response: To our knowledge, the ability of N-specific mAbs in COVID-19 against hyperactivation of the complement system is firstly reported by our group. Our supplementary data supports that not only the nCoV396 is capable of inhibition of N-induced complement hyperactivation, but also other two mAbs (nCoV454 and nCoV457) can work on it (Please see the response for Reviewer #2 point 3- (2)). To further validate whether these mAbs work against other highly pathogenic relative N proteins in the ex vivo complement system, we next perform the ex vivo assays in the presence of SARS-CoV N-protein. As shown in Supplementary Fig. 3c, d To evaluate the potential epitope of other mAbs, we perform the epitope screening assays with continually N-derived peptides (Please see the response for Reviewer #2 point 3-(2)). As shown in Supplementary ## R-work = ∑hkl ||Fobs (hkl) |-|Fcalc (hkl) || / ∑hkl |Fobs(hkl) |. R-free is calculated in an identical manner using 10% of randomly selected reflections that were not included in the refinement.

Lines 210-:
The authors hypothesize and biophysically characterize cross-interaction between mAb-nCoV396 and two other related coronaviruses. Is there evidence of cross-neutralization of the same viruses with this mAb?
Response: We found that mAb-nCoV396 has a high affinity with SARS-CoV N protein and MERS-CoV N protein by SPR (Extended Data Figure 2b). Besides, we repeated the ex vivo assays with SARS. We found that SARS-CoV N protein can also induce complement hyperactivation, and mAb-nCoV396 can inhibit this induction (Please see the response for Reviewer #2 point 3- (2)). Reproducibility of some of the functional assays could be improved as detailed below.
Response: We thank this reviewer for appreciating the novelty of our report about the monoclonal antibody (nCoV396) against the N protein can curb N protein-induced complement hyperactivation. We are very grateful to the reviewer for providing valuable comments on our article, and we have made a serious response to these comments.  Fig. 4). Taken together, our works reveal that these systems truly representative of MASP-2 protease mediated cleavage of C2.
2. Towards the end of the results section, the authors state "in conclusion, these results demonstrate that ...not only by facilitating Vmax of MASP-2 catalytic activity but also enhancing substrate binding specificity in the reactions". Is there an independent way to validate these findings, and how can they be certain this is the case? For example, is there a positive control for MASP-2 catalytic activity that has been utilized? Either that, or another assay for MASP-2 mediated cleavage would be important. Along those lines, is there a way to assess how the binding of nCoV396 to the N-protein modulates inflammation (including but not restricted to markers of complement activation) either in vivo or ex vivo?
Response: As described in response to Reviewer #3 critique point 1, we purified the catalytic domain of  To test whether the two other antibodies have similar functions with nCoV396, we next verified the function of nCoV457 in ex vivo assays. Consistently, nCoV457 has a similar but weaker inhibitory effect to nCoV396, which may be due to nCoV457 and N protein's weaker affinity than nCoV396 (Supplementary Fig. 3a, b). Response: We have supplemented the experiment that SARS-CoV-2 N protein can directly improve the enzymatic activity efficiency of the catalytic domain of MASP-2 purified in vitro (Supplementary Fig S4).
These results complement the ex vivo data in the article, and together they can prove that N protein is contributing to complement-mediated hyperactivation. From our perspective, we found that N protein directly interacted with MASP-2, induced the activity of MASP-2, and served as a potent target for antibody therapy of COVID-19. patients than in non-ICU patients. 9 Consistently with their results, our study found several COVID-19 patients infected with SARS-CoV-2 have higher antibody titers to N protein than S protein.
Due to the small sample size, it cannot be said that this is a common phenomenon in the human population. We concur with the reviewer's suggestion and have edited the sentence at Line 321-324. 4. Regarding the antibody repertoires, how did the authors ensure that there was no cross-reactivity of monoclonal Abs to other viral proteins? Additionally, among 32 mAbs that bound to N-FL, 13 Abs bound to N-NTD and one Ab bound to N-CTD, but which part of the protein did the remaining 18 Abs bind to?
Response: To test whether the antibody nCoV396 or other mAbs have the potential to cross-react with other highly pathogenic coronavirus, we perform the ex vivo and in vitro assays in the context of another relative N protein. As shown in Supplementary Fig. 3c, d etc. Therefore, we focus our subsequent studies on the monoclonal antibodies that bind to N-NTD, N-CTD or full-length protein in this project. Nevertheless, we agree with the reviewer that it is worth checking whether these monoclonal antibodies are bound to other regions of the nucleocapsid protein in the future work once we can express these flexible regions.

REVIEWER COMMENTS
Reviewer #1 (Remarks to the Author): I understand the difficulty to characterize the other monoclonal antibodies, which recognize SARS-CoV-2 N protein but don't bind to N-NTD or N-CTD. The authors adequately addressed the raised comments and questions and improved the manuscript.
Reviewer #2 (Remarks to the Author): The authors have addressed the concerns raised and the improvements suggested in the review and the edited manuscript is now suitable for publication.
Reviewer #3 (Remarks to the Author): The authors have provided appropriate responses to most of the comments that I had raised in my original review. The addition of Supplementary Figure 4 is helpful to understand their system, designed to probe MASP-2 mediated C2 cleavage. Addition of text regarding nCoV457 is also helpful to understand why nCoV396 was selected, as is a clarification regarding the patients with autoimmune disease. The comments below are primarily to help the scientific community better understand the manuscript, and to facilitate reproducibility -MINOR COMMENTS 1. The methodology described in response to Reviewer 3, question 1 needs to be incorporated into the Methods (either main manuscript or Supplement). If it has already been reported in this level of detail included in the response to reviewer comments, I apologize, as I was not able to find it. In this situation, please do point me to where it is in the manuscript. Having the level of detail that has been provided in the response to the reviewer would be necessary to incorporate in the manuscript to ensure reproducibility by the scientific community. 2. Reviewer 3 question 3 -I am confused about the answer and the inclusion of patient samples in different experiments is still somewhat hard to follow. It appears the authors are referring to Extended Table 6, instead of Extended  Figure 6 is helpful, this includes only one patient and control. They should include at least n=3 in both groups, ideally more. A minor change is that the legend of Figure 4 needs to be updated to 'patients with abnormal C3 or C4 values'. To that extent, the last figure in Figure 4E (bottom right) is confusing; it does not include the patients from 4b and 4d; nor a healthy control. If this could be clarified why it was done this way, or ideally, the data provided, that would be helpful. 3. Reviewer 3 question 4 -While nCoV457 was compared to nCoV396, why was the same thing not done for nCoV416? It appears nCoV454 was tested, which makes this response somewhat confusing. 4. While the authors have added the references of Hachim et al and Sun et al in lines 321-4, they should also add the sentence "due to the small sample size..." (that they wrote in their response) to the main manuscript, and cite the McAndrews manuscript, to demonstrate some equipoise.
5. The answer to Reviewer 3, Minor Comment#4 is satisfactory, but needs to be included in the main manuscript under a paragraph on limitations.

REVIEWERS' COMMENTS
Reviewer #3 (Remarks to the Author): The authors have provided appropriate responses to most of the comments that I had raised in my original review. The addition of Supplementary Figure 4 is helpful to understand their system, designed to probe MASP-2 mediated C2 cleavage. Addition of text regarding nCoV457 is also helpful to understand why nCoV396 was selected, as is a clarification regarding the patients with autoimmune disease. The comments below are primarily to help the scientific community better understand the manuscript, and to facilitate reproducibility.

Response:
We thank reviewer's opinions and accept his/her suggestions which key clarifications need addressing in the revised version of the manuscript. The appropriate sentences have been edited from the revised version.
MINOR COMMENTS 1. The methodology described in response to Reviewer 3, question 1 needs to be incorporated into the Methods (either main manuscript or Supplement). If it has already been reported in this level of detail included in the response to reviewer comments, I apologize, as I was not able to find it. In this situation, please do point me to where it is in the manuscript. Having the level of detail that has been provided in the response to the reviewer would be necessary to incorporate in the manuscript to ensure reproducibility by the scientific community. Response: We concur with the reviewer's recommendation that the response to Reviewer 3 -Question 1 should be incorporated into the Methods. For clarifying, we have added the methodology description in the line 494-500 of the Methods part.
2. Reviewer 3 question 3 -I am confused about the answer and the inclusion of patient samples in different experiments is still somewhat hard to follow. It appears the authors are referring to Extended Table 6, instead of Extended  Figure 6 is helpful, this includes only one patient and control. They should include at least n=3 in both groups, ideally more. A minor change is that the legend of Figure 4 needs to be updated to 'patients with abnormal C3 or C4 values'. To that extent, the last figure in Figure 4E (bottom right) is confusing; it does not include the patients from 4b and 4d; nor a healthy control. If this could be clarified why it was done this way, or ideally, the data provided, that would be helpful.

Response:
We must apologize for our mistakes at referring a wrong table. A corrected sentence has been edited from the revised version.
We understood the confusing nomenclature of Patient ID in the main text, more likely due to the discontinuous nomenclature. In order to describe the content of our experiment more clearly, we (1) To clarify the results in Fig.4b and Supplementary Fig. 5a (2) Serum-07 is used to prove that different concentration SARS-CoV-2 N protein can promote the enzymatic activity of MASP2 toward C2 substrate, corresponding to Fig. 4d result. (3) Serum-08 is used to prove that different concentration SARS-CoV-2 N protein antibody (nCoV396) can inhibit the enzymatic activity of MASP-2 induced by N protein, corresponding to Fig. 4f result.
(4) Serum-09 to -11 are used to expand of the sample size to prove that SARS-CoV-2 N protein antibody (nCoV396) can inhibit the ability of MASP-2 to cleave C2 induced by N protein, corresponding to Fig. 4g result.
In addition, to clarify the independent experimental number with reviewer's critique, we now have added three health control vs patients paired data in Supplementary Fig. 5a (the previous version Supplementary Fig. 6). The results are consistent with the previous conclusions, the serum MASP-2 activity to C2 of the healthy donors are much weaker than those from abnormal serologic samples.  Fig. 4b and 4d, which may cause misunderstandings by readers. We have modified Fig. 4g (the previous version Fig. 4e bottom right) to address this issue. We analysis five serum sample from biologically independent donors (n=5) with abnormal serologic C3 values. (serum-08 to -12). And we use Michaelis-Menten equation to calculate the Vmax (with experimental data from Fig. 4f (serum-08), Supplementary Fig. 5b (serum-09 to-11) and 7a (serum-12)). Each sample was performed in triplicates and mean values ± SEM of Vmax are presented. Two-sided Kruskal-Wallis test with Dunnett's multiple comparisons test was used for comparing the Vmax of groups. The significant reference is 0.05. Finally, we have concluded that the Vmax of reactions increases in the presence of the N protein, whereas declines in the presence of both the mAb nCoV396 and the N protein. These clarifications have been added into Fig. 4 legend.
3. Reviewer 3 question 4 -While nCoV457 was compared to nCoV396, why was the same thing not done for nCoV416? It appears nCoV454 was tested, which makes this response somewhat confusing.

Response:
As reviewer mentioned, we did the same experiment for nCoV416. As the epitope recognized by nCoV416 is the same as nCoV454 and nCoV457, we did not include the result in Supplementary table in previous version. At the latest version, we have completed the Supplementary Table 12 with  nCoV416 result. 4. While the authors have added the references of Hachim et al and Sun et al in lines 321-4, they should also add the sentence "due to the small sample size..." (that they wrote in their response) to the main manuscript, and cite the McAndrews manuscript, to demonstrate some equipoise.

Response:
We are very grateful for your suggestions. We have added the corresponding sentence in the line 312-315 and cited the McAndrews literature into the latest manuscript.
5. The answer to Reviewer 3, Minor Comment#4 is satisfactory, but needs to be included in the main manuscript under a paragraph on limitations.

Response:
We are very willing to accept your pertinent suggestions, in the discussion part of the article we add the following content in the line 368-377: "For the other 18 antibodies, however, it is labor intense, time consuming, unpredictably to express the rest of the nucleocapsid portion separately due to the nature of the protein. These regions belong to disorder or flexible parts of the protein, although we have worked on these several times. The comprehensive studies have suggested that the nucleocapsid compact functional domains are its NTD and CTD. These two domains play several vital roles in viral RNA recognition, viral genomic RNA packing, high-order structure formation of viral ribonucleoproteins (RNP), etc. Therefore, we focus our subsequent studies on the monoclonal antibodies that bind to N-NTD, N-CTD or full-length protein in this project. Nevertheless, we believe that it is worth checking whether these monoclonal antibodies are bound to other regions of the nucleocapsid protein in the future work."