Information

How to filter out the antibody light chains into kappa and lambda types on the PDB website?

How to filter out the antibody light chains into kappa and lambda types on the PDB website?



We are searching data for your request:

Forums and discussions:
Manuals and reference books:
Data from registers:
Wait the end of the search in all databases.
Upon completion, a link will appear to access the found materials.

The antibody light chains can either be kappa or lambda types. I am searching homologous sequences on the RCSB PDB website. Is there a way to filter out these two types?


The instructions are as follows.

  1. Go to the RCSB PDB Webite.
  2. Under the search bar, clickBrowse by Annotations.
  3. Click the blue tab labeledSCOP.
  4. Click the white arrow next to the folder titled All Beta Protiens.
  5. Click the white arrow next to the folder titled Immunoglobulin-like beta-sandwich (… ).
  6. Click the white arrow next to the folder titled Immunoglobulins (Superfamily).

This is where it splits, based on your specific search.

For Light Chain, Constant Domain:

  1. Click the white arrow next to the folder titled C1 set domains (antibody constant domain-like).
  2. For Kappa Type, Click the folder (not the arrow) titled Immunoglobulin light chain kappa constant domain, CL-kappa.
  3. For Lambda Type, Click the folder (not the arrow) titled Immunoglobulin light chain lambda constant domain, CL-lambda.

For Light Chain, Variable Domain:

  1. Click the white arrow next to the folder titled V set domains (antibody variable domain-like).
  2. For Kappa Type, Click the folder (not the arrow) titled Immunoglobulin light chain kappa variable domain, VL-kappa (… ).
  3. For Lambda Type, Click the folder (not the arrow) titled Immunoglobulin light chain lambda variable domain, VL-lambda (… ).

This should get you to the list of proteins that you are looking for.


Detection of oligoclonal IgG kappa and IgG lambda bands in cerebrospinal fluid and serum with Hevylite™ antibodies. comparison with the free light chain oligoclonal pattern

Oligoclonal IgG bands in cerebrospinal fluid that are absent in serum indicate intrathecal IgG synthesis and are a sensitive marker of CNS inflammatory diseases, in particular multiple sclerosis. It may be of interest to determine whether these bands are predominantly IgGκ or IgGλ.

Methods

We have used Hevylite™ antibodies and developed a technique for detection of oligoclonal IgGκ and IgGλ bands by means of isoelectric focusing followed by immunoblotting. The same technique was used for oligoclonal free κ and free λ detection. Among several techniques tested, affinity immunoblotting appears to be the most sensitive it can detect less than 1 ng of IgGκ or IgGλ paraprotein. We compared oligoclonal IgG profiles with those of oligoclonal IgGκ and IgGλ. There was good agreement concerning the presence or absence of intrathecal synthesis. We observed the ratios between oligoclonal IgGκ and IgGλ bands, and they did not always match the ratios between free κ and free λ bands. We were also able to detect antigen-specific CSF-restricted oligoclonal IgGκ and IgGλ bands in neuroborreliosis. It remains to be determined subsequently by a clinically-oriented prospective study, whether predominant IgGκ/IgGλ or free κ/free λ can be observed more frequently in particular diseases with oligoclonal IgG synthesis.

Discussion

Very sensitive detection of oligoclonal IgGκ and IgGλ bands in cerebrospinal fluid with Hevylite antibodies is feasible detection of antigen-specific IgGκ or IgGλ is possible as well. In particular situations, e.g. when difficulties arise in distinguishing between oligoclonal and monoclonal pattern, the test may be of considerable clinical value.


2.2. The identification of serum monoclonal proteins

In parallel with the clinical and scientific observations of the role of Bence Jones protein, electrophoretic techniques for protein separation were evolving and eventually entered clinical laboratories. Perlzweig et al. [56] were the first to report hyperproteinaemia in MM in 1928. Shortly after in 1930, Tiselius reported the homogeneity of certain serum globulins using a technique he devised, termed moving boundary electrophoresis


Diagnosis

Diagnostic testing for AL amyloidosis involves blood tests, urine tests and biopsies. Blood and/or urine tests can indicate signs of the amyloid protein, but only bone marrow tests or other small biopsy samples of tissue or organs can positively confirm the diagnosis of amyloidosis. Some tests are only done once to establish a diagnosis of AL amyloidosis, while others will be repeated to monitor disease progression and response to therapy.

Blood and Urine Tests

Blood and urine tests should be performed to help verify the diagnosis. They can also aid in discovering which organs are involved and how much they are compromised. These tests may include:

  • A 24-hour urine collection to look at the level of protein in your urine sample. Excess protein in the urine may be an indication of kidney involvement.
  • The level of ALP (an enzyme called “alkaline phosphatase”) in your regular blood workup.
  • Blood tests to look for stress and strain on the heart are useful in many forms of heart disease, including AL amyloidosis. The cardiac biomarkers that are used include troponin T or troponin I, and NT-proBNP (which stands for N-terminal pro-brain natriuretic peptide) or BNP (brain natriuretic peptide). Different laboratories use one versus the other.
  • Tests for abnormal antibody (immunoglobulin) proteins in the blood include the Freelite Assay, which shows the level of kappa and lambda light chains in a separate blood test. The Freelite Assay test is often referred to as FLC, which is an abbreviation for free light chains.
  • Another test for abnormal immunoglobulin can be done with blood and/or urine. It is called “immunofixation electrophoresis.”

These blood and urine tests can help with the diagnosis and used often while monitoring response to treatment.

Echocardiogram and Imaging

The echocardiogram (also called “echo”) is an ultrasound of the heart. A doctor can look for amyloid deposits in the heart, while viewing the size and shape of it and the location and extent of any impact of amyloid.

Recently, other imaging tests for the heart have also shown to be useful. One test is the MRI (magnetic resonance imaging), and, in this instance, is also referred to as CMR (for cardiac magnetic resonance). Pyrophosphate scanning, a nuclear medicine test, is also used to evaluate whether an unusual type of abnormality of heart muscle function (“cardiomyopathy”) is present. Current data suggests this scan may be useful in distinguishing different types of amyloid heart disease.

Tissue Biopsy

A tissue biopsy involves the removal of a small sample of tissue to find evidence of amyloid deposits. Any kind of tissue or organ biopsy must be sent to a lab for microscopic examination, where the tissue is stained with a dye called “Congo-red stain.” After putting it under a microscope, amyloid protein is discovered if it turns an apple-green color, resulting in a diagnosis of amyloidosis. Possible areas for less invasive biopsies include:

  • A fat pad biopsy (from under the skin in the abdomen)
  • Labial salivary gland biopsy (the inner lip) and,
  • Skin or bone marrow.

With combinations of these blood and urine tests and tissue biopsies, a positive diagnosis can be confirmed in a high percentage of patients.

Bone marrow aspirate and biopsy

There are two types of bone marrow tests that may be performed. These involve the removal of some liquid bone marrow (a bone marrow aspirate) and/or the removal of a 1 – 2 cm core of bone marrow tissue in one piece (a bone marrow biopsy). These samples can help to determine the percentage of amyloid producing plasma cells, and when tested in the lab they can assist in identifying whether the abnormal plasma cells are producing kappa or lambda light chains.

Organ Biopsy

If amyloidosis is still suspected but biopsies of the bone marrow, fat pad, lip or skin sites turn up negative, then a surgical biopsy of the organ that is indicating symptoms should be performed and sent to a lab. Biopsy samples may be taken from the:

If any biopsy result shows a positive diagnosis for amyloidosis, then it is essential to also determine the accurate type of amyloid protein that is involved. In this case, the type of AL amyloidosis must be confirmed, showing a bone marrow disorder with light chain involvement, also known as a “plasma cell dyscrasia.”


Materials and Methods

Production of rabbit IgG

The genes encoding rabbit immunoglobulins were chemically synthesized (Genewiz, NJ, USA) followed by subcloning into pcDNA3.4 using In-fusion cloning kit (Takara Bio, Shiga, Japan). For the removal of cysteines in the kappa chains, mutagenesis using PrimeSTAR Max DNA Polymerase (Takara Bio) was conducted. IgGs were produced using Expi293 expression system (Thermo Fisher Scientific, MA, USA) by co-transfection of two vectors encoding heavy and light chains, respectively. IgGs were isolated by protein A affinity chromatography using rProtein A Sepharose Fast Flow (GE Healthcare, IL, USA) followed by the final purification by size-exclusion chromatography (SEC) using HiLoad 16/600 Superdex 200 pg or HiLoad 26/600 Superdex 200 pg column (GE Healthcare). The concentration of antibodies and the antigen peptide were determined by the absorbance at 280 nm.

Measurement of thermal stability

Thermal stability of antibodies was determined by DSC at the concentrations of 12–30 μM in phosphate buffered saline (PBS). Data collection was performed in a VP-Capillary DSC instrument (Malvern Instruments, UK) at a scanning rate of 1.0 or 0.33°C min −1 between 20 and 100°C. Rescans were conducted in standard protocol provided by the manufacturer.

Measurement of binding activity

The thermodynamic parameters of the interaction between antibody and antigen were examined by isothermal titration calorimetry (ITC) in an iTC200 instrument (Malvern). Antigen peptide (Arg-Pro-His-Phe-Pro-Gln-Phe-pSer-Tyr-Ser-Ala-Ser: pSer represents phosphoserine) was chemically synthesized (Scrum Inc, Tokyo, Japan). To the solution of antibody (A4, G11: 2 μM, A3: 3 μM) in the cell, 1.5 μL of the solution of antigen peptide (A4, G11: 40 μM, A3: 60 μM) was titrated per injection for 25 times at the interval of 160 s with the stirring rate at 1000 rpm. The measurements were conducted in PBS at 25°C. Measurements were performed three times in the same procedure.

Data analysis

Both DSC and ITC data were analyzed by Origin 8.0 (Originlab Corp., MA, USA). For the DSC measurements, the analysis was conducted in non-two state model with two denaturation processes. For the ITC measurements, the analysis was conducted as 1:1 binding model.


Discussion

Antibodies directed to NTD and to RBD can neutralize with high potency (less than 0.01 μg/mL IC50). While RBD shows many non-overlapping sites of vulnerability to antibody ( Barnes et al., 2020a Brouwer et al., 2020 Lv et al., 2020 Pinto et al., 2020 Yuan et al., 2020a ), NTD appears to contain only a single site of vulnerability to neutralization. As discussed above, one reason for this may be the high glycan density on NTD, with 8 N-linked glycans in � residues, a density of one glycan per � residues, and few glycan-free surfaces that can be easily recognized by the immune system. A second reason may be the restricted approach angle that we observed for all known NTD-directed neutralizing antibodies, including the seven reported here, which all approach spike from �ove.” We note in this context that competition analysis indicates other NTD-directed antibodies capable of recognizing spike and forming a separate competition group to be non-neutralizing ( Liu etਊl., 2020a )𠅊nd the other large surface on NTD that is exposed on spike faces toward the viral membrane. This surface is mostly glycan free, and antibodies binding to it would be required to approach from �low.” Thus, unlike RBD, where neutralizing antibodies appear to have diverse approach angles, the presence of only a single-NTD site of vulnerability may relate to the requirement to approach from above.

In addition to satisfying requirements stemming from the restricted approach angle, the higher relative prevalence of NTD-supersite-directed antibodies is likely to stem from increased immunogenicity due to both the lower relative glycan density of the supersite and the flexible nature of the N3 hairpin and N5 loop primary recognition regions, as well as their ability to assume distinct conformations that allow for recognition by diverse antibodies. In the case of the multi-donor VH1-24 antibody class, which arises from the most prevalent VH gene utilized (Figure S1A), two additional NTD-directed neutralizing class members have recently been identified: FC05 and CM25 ( Voss etਊl., 2020 Wang etਊl., 2021 ). We found VH1-24 to be the most negatively charged human VH gene (Figure S5B). Such negative electrostatic potential complements the highly electropositive nature of the NTD supersite that we observe here ( Figureਆ D). Thus, multiple factors, including epitope glycosylation and flexibility, restrictions on approach angle, and paratope charge complementarity, can contribute to the prevalence of antibodies targeting the NTD supersite.

Although the approach to the spike from above observed for all NTD-directed neutralizing antibodies is consistent with a neutralization mechanism based on steric hindrance of spike interaction with ACE2 receptor at the cell membrane, there is currently no evidence for competition between NTD-directed antibodies and ACE2 ( Liu etਊl., 2020a ). A plausible alternative model would be for antibody recognition of the NTD supersite to impede spike function in mediating fusion of virus and host cell membranes. Indeed, protease-resistance analysis of MERS spike in complex with MERS NTD-directed neutralizing antibody 7d10 showed that 7d10 binding prevented increased protease sensitivity associated with the prefusion-to-postfusion transition ( Zhou etਊl., 2019 ). A conformational stabilization mechanism could also explain how an antibody that binds only one subunit per spike trimer could achieve effective neutralization. Further studies will be required to understand mechanisms of neutralization for antibodies that recognize the NTD supersite.

With respect to vaccine implications, our results clearly identify the NTD site of vulnerability most likely to elicit neutralizing antibodies. There are many ways that this information can be incorporated into vaccine design, including the inclusion of NTD along with RBD in vaccine formulations, the multivalent display of the NTD supersite on nanoparticle immunogens, and epitope-focusing through the creation of scaffolds displaying the N3 β-hairpin and other regions of recognized by NTD-directed neutralizing antibodies.

With respect to the therapeutic potential of NTD-directed antibodies, these target a site that is remote from those targeting RBD sites and thus should provide complementary neutralization to RBD-directed antibodies and require distinct escape pathways. The fact that all, or a great majority, of NTD-neutralizing antibodies target a single site, however, suggests there may be little utility to utilizing combinations of NTD-directed neutralizing antibodies.

Finally, new mutant SARS-CoV-2 strains, particularly those emerged in the UK and South Africa (strains B.1.1.7 and B.1.351, respectively), are concerning due to increased transmissibility, and these strains escape most NTD-directed neutralizing antibodies. B.1.1.7 includes NTD deletion mutations D69-70 and D144, and strain B.1.351 includes NTD mutations D242-244 and R246I. Consistent with our findings, the mutated positions including 144, 242-244, and 246 are all within the NTD supersite. While the deletion at 69-70 is outside of the supersite, it forms part of the hairpin N2 loop of NTD its deletion could significantly impact the conformation of the NTD supersite. Notably, only three residues were shared among the eight NTD-directed neutralizing antibody epitopes analyzed here: Y144, R246, and L249 (Table S5). Interestingly, two of these three residues are the exact residues mutated in emerging variants of concern (� and R246I), and L249 is likely affected by �-244. Thus, the flipside of a single supersite is that variation of the supersite may induce resistance against most of the antibodies targeting the site𠅊nd be selected for among emerging variants.


Results

Analysis of the Light Chain Variable Region Genes in Zebra Finch

Using the human (kappa and lambda), chicken (lambda), and frog (sigma) IGL variable sequences as queries, we identified 21 IGL variable region genes located in a cluster in chromosome 15 of the zebra finch genome. The genomic location of these variable region genes in zebra finch is given in supplementary table S1 ( Supplementary Material online). The sequence comparison with functional sequences of human, chicken, and frog indicated that among the 21 IGL variable region genes, only one sequence is functional because it contains a complete coding sequence without frameshift mutations and/or internal stop codons, two conserved Cys residues in FR1 and FR3 regions and a proper RSS ( fig. 1). The genomic structure of the single functional IGL variable region gene in zebra finch is shown in supplementary figure S1 ( Supplementary Material online). The remaining IGVL genes either lacked the proper leader and/or RSS or were truncated in their 5′ or 3′ ends ( supplementary fig. S2 , Supplementary Material online), like the chicken IGVL pseudogenes ( Reynaud et al. 1987). Only four sequences contained internal stop codons ( supplementary fig. S2 and Supplementary Data , Supplementary Material online).

Alignment of IGL variable sequences. The randomly chosen reference sequences of known kappa (three IGVK sequences from human), lambda (three IGVL sequences from human and one IGVL from chicken), and sigma (the IGVS from frog) isotypes were taken from Das et al. (2008). The cladistic molecular markers that distinguish the three isotypes (κ, λ, and σ) are highlighted. The lengths of the RSS spacer sequences are given. The numbering of the amino acid positions in the V-segment is based on human IGVK and IGVL sequences, and the gaps relative to the frog IGVS sequences are indicated by “a” and “b” ( Das et al. 2008).

Alignment of IGL variable sequences. The randomly chosen reference sequences of known kappa (three IGVK sequences from human), lambda (three IGVL sequences from human and one IGVL from chicken), and sigma (the IGVS from frog) isotypes were taken from Das et al. (2008). The cladistic molecular markers that distinguish the three isotypes (κ, λ, and σ) are highlighted. The lengths of the RSS spacer sequences are given. The numbering of the amino acid positions in the V-segment is based on human IGVK and IGVL sequences, and the gaps relative to the frog IGVS sequences are indicated by “a” and “b” ( Das et al. 2008).

Analysis of IGL Joining and Constant Region Genes in Zebra Finch

From the similarity search using as queries, five functional IGL constant sequences (one IGCK and two IGCL from human, one IGCL from chicken, and one IGCS from frog), we identified only a single functional IGL constant–encoding gene in the zebra finch genome. This gene is located 4.5 kb downstream of the single functional variable region gene. ESTs and cDNA sequences confirm the presence of a single functional IGL constant–encoding gene in zebra finch because the identified ESTs and cDNA sequences align—with almost 100% identity and no gaps—to a single genomic position in the zebra finch genome that corresponds to the position of the single IGCL functional gene ( supplementary table S2 and Supplementary Data , Supplementary Material online). To identify the IGL joining region gene in the zebra finch genome, we scanned for the conserved RSS in the 4.5 kb region between the functional variable and constant region genes because the joining region gene is too short (usually 12 amino acids in length) to be identified by Blast searches. Once the potential RSS was identified in this 4.5 kb region, we translated the nucleotide sequences at the 3′ end of the RSS into amino acids and compared the translated sequence with the human, chicken, and frog IGL joining region sequences, which were identified in our previous study ( Das et al. 2008). Using this method, we identified a single functional IGL joining region gene in zebra finch.

Identification of Zebra Finch IGL Isotype

To characterize the isotype of the zebra finch IGL sequences, we used the cladistic molecular markers, which we previously described ( Das et al. 2008). Like the IGVL sequences of other tetrapods, the only functional variable region sequence of zebra finch lacks Ser or Thr at position 7 and does not possess a bulky aromatic residue (Phe or Tyr) at position 53 ( fig. 1). The tetrapods IGVL sequences generally have a fairly conserved DEAD (Asp–Glu–Ala–Asp) motif in the FR3 region ( Das et al. 2008). However, like chicken IGVL sequence (i.e., DEAV at position 64–67), the Asp residue is substituted for Val at position 67 ( fig. 1).

Consistent with the variable region sequence, the molecular markers in the joining and constant region sequences in zebra finch also categorize them as lambda light chain sequences ( figs. 2 and 3). In addition, like mammalian lambda light chain genes, the RSS sequences flanking the single functional variable region gene and the joining region gene are interrupted by a 23-bp and a 12-bp spacer, respectively, in the zebra finch IGL locus. Hence, the molecular markers in the variable, joining, and constant region sequences in zebra finch indicate that like chicken ( Sanders and Travis 1975) and duck ( Magor et al. 1994), the zebra finch genome encodes only the lambda isotype of Ig.

Alignment of IGL joining region sequences. The randomly chosen reference sequences of known kappa (three IGJK sequences from human), lambda (three IGJL sequences from human and one IGJL from chicken), and sigma (two IGJS from frog) isotypes were taken from Das et al. (2008). The cladistic molecular markers that distinguish the three isotypes (κ, λ, and σ) are highlighted ( Das et al. 2008). The lengths of the RSS spacer sequences are shown.

Alignment of IGL joining region sequences. The randomly chosen reference sequences of known kappa (three IGJK sequences from human), lambda (three IGJL sequences from human and one IGJL from chicken), and sigma (two IGJS from frog) isotypes were taken from Das et al. (2008). The cladistic molecular markers that distinguish the three isotypes (κ, λ, and σ) are highlighted ( Das et al. 2008). The lengths of the RSS spacer sequences are shown.

Alignment of IGL constant sequences. The reference sequences of known kappa (one IGCK sequence from human), lambda (two IGCL from human and one IGCL from chicken), and sigma (one IGCS from frog) isotypes were taken from Das et al. (2008). The numbering of the amino acid positions in the C-segment is based on Das et al. (2008). The gap relative to IGCS sequence of frog and the gaps relative to zebra finch IGCL sequence are characterized by alphabets. The cladistic molecular markers that distinguish the three isotypes (κ, λ, and σ) are highlighted.

Alignment of IGL constant sequences. The reference sequences of known kappa (one IGCK sequence from human), lambda (two IGCL from human and one IGCL from chicken), and sigma (one IGCS from frog) isotypes were taken from Das et al. (2008). The numbering of the amino acid positions in the C-segment is based on Das et al. (2008). The gap relative to IGCS sequence of frog and the gaps relative to zebra finch IGCL sequence are characterized by alphabets. The cladistic molecular markers that distinguish the three isotypes (κ, λ, and σ) are highlighted.

Strikingly, the zebra finch IGL constant region sequence can be distinguished from the tetrapod IGL constant region sequences because it contains a unique insertion of a six amino acid stretch (QQQSST) ( fig. 3). This insertion is confirmed by the fact that all ESTs/cDNAs sequences, when translated, include the additional six amino acids (see supplementary table S2 and Supplementary Data , Supplementary Material online). To further analyze this insertion, we generated a 3D model of the zebra finch IGCL protein and mapped the insertion on the 3D model ( fig. 4). Our analysis predicts that the QQQSST insertion extends a loop located in the region between the IGVL and IGCL by approximately 9 Å ( fig. 4 supplementary fig. S4 , Supplementary Material online). Although this extended loop is predicted to reduce the distance between the Ig light and the heavy chains by almost 6 Å ( supplementary fig. S4 , Supplementary Material online), it does not seem to affect the contact between the different Ig chains. This supposition deserves to be experimentally tested.

The amino acid insertion of the zebra finch IGCL protein sequence is mapped on a 3D model. The theoretical 3D model for the zebra finch IGCL (shown in light cyan) was constructed using as template the experimentally resolved structure of a human IGL lambda chain (shown in green PDB accession code: 1A8J). The predicted and the template structures were structurally aligned to highlight the insertion region (shown in red). The upper panel shows a cartoon representation and the lower shows a surface representation of the two structures.

The amino acid insertion of the zebra finch IGCL protein sequence is mapped on a 3D model. The theoretical 3D model for the zebra finch IGCL (shown in light cyan) was constructed using as template the experimentally resolved structure of a human IGL lambda chain (shown in green PDB accession code: 1A8J). The predicted and the template structures were structurally aligned to highlight the insertion region (shown in red). The upper panel shows a cartoon representation and the lower shows a surface representation of the two structures.

Genomic Organization of the IGL Locus in Zebra Finch

In the kappa- and sigma-encoding locus of most tetrapods, multiple joining region genes are present in a cluster, followed by a single constant gene, whereas in the lambda-encoding locus, joining and constant genes occur as IGJL–IGCL blocks, which usually have multiple copies. Only chicken has one IGJL–IGCL block ( Das et al. 2008). The IGL locus in zebra finch contains one functional IGVL gene, multiple pseudo-IGVL genes, and only one IGJLIGCL block ( fig. 5), like the chicken locus ( Reynaud et al. 1985, 1987).

Schematic diagrams of the genomic organizations of lambda light chain loci in human, horse, chicken, and zebra finch (not to scale). Rods above and below the lines indicate genes located on opposite strands based on the particular genome sequence. Long rods show functional genes, and short rods indicate pseudogenes. The positional information of human, horse, and chicken lambda chain genes are taken from Das et al. (2008).

Schematic diagrams of the genomic organizations of lambda light chain loci in human, horse, chicken, and zebra finch (not to scale). Rods above and below the lines indicate genes located on opposite strands based on the particular genome sequence. Long rods show functional genes, and short rods indicate pseudogenes. The positional information of human, horse, and chicken lambda chain genes are taken from Das et al. (2008).

The comparison of the IGL locus between the chicken and the zebra finch shows that in these species both the number and the position of the IGL genes are very similar ( fig. 5), despite the fact that these species have diverged more than 100 Ma ( Brown et al. 2008). The main difference between the two loci is the presence of a few pseudogenes with reverse orientation in chicken. In contrast to the general conservation of IGL locus observed between the two bird species, comparison of the IGL locus between different mammalian species (i.e., human–horse) that have diverged approximately 100 Ma showed many differences both in the constant and variable regions ( fig. 5). In the constant region, both human and horse contain seven IGJLIGCL blocks, but the distribution of functional genes and pseudogenes are different between the two species ( fig. 5). In the variable region, both the number and the distribution of functional genes and pseudogenes also vary between the human (32 functional and 42 variable region pseudogenes) and the horse (25 variable region functional and 20 pseudogenes) lambda loci.


Discussion

We set out to biophysically characterize all human six light and nine heavy chain isotypes and subtypes with regards to recombinant production and antigen-binding kinetics. Our attempts with bacterial expression of these isotypes did not result in detectable protein productions, even when applied for whole IgG1 as previously reported 21 . Given the human nature of the antibodies requiring post-translational modification (a factor we wanted to rule out), the failure of bacterial expression was inconsequential. The reasons for the failed bacterial expression may be attributed to the different bacterial plasmid used, as well as the different variable sequences.

To our knowledge, our study is one of the first comprehensive comparison of all known human heavy and light chain isotypes and subtypes utilizing two model antibodies against the same antigen simultaneously. It should be noted that the numbering of the Cλ families do not run in order because some Cλ families (e.g. Cλ4 & Cλ5) were later found to be pseudogenes or theoretical genes 17 . In fact, to solely investigate the effects of the CL, we kept the original VL, forming Vκ– Cλ hybrid variants.

On the heavy chain isotypes, IgM expectedly displayed distinct oligomerization to predominantly pemtameric/hexameric forms with some intermediates 10 even without co-transfection of J-chain plasmids (Fig. 1). Such oligomerization for IgM is immunologically superior in activating the complement system than monomeric forms 11 , making IgM variants suitable candidates for targeting circulatory metastatic cancer cells that would be abated by immobilization of antigens via agglutination. Although IgG4 was previously reported to exhibit Fab arms exchange in vivo 16,26 , such exchange would not have an effect in our monoclonal expression methods, maintaining the validity of our biophysical characterizations.

The comparison of the association and dissociation rates of the CH variants to their commercial counterparts (Figs 2 and 3) shows the general feasibility of CH isotype switching for both Trastuzumab and Pertuzumab, and possibly for all therapeutic antibodies. Given that the immune functions of secreted IgD remain enigmatic and the poor protein production profiles of the Trastuzumab and Pertuzumab IgD variants, there may be need to reconsider the use of secreted IgD as a therapeutic variant. The fact that murine IgD receptor does not bind to human IgD 27 while the human IgD receptor can bind both human IgD and IgA1 with the similar O-glycans modification at the hinge region 28 , makes IgD difficult to study using animal models. To remove the confounding factor of glycosylation, we produced all our CH variants in human HEK cell-lines over animal CHO or COS cells.

Interestingly, the avidity effects of Pertuzumab IgM reversed the general trend of poorer binding over Trastuzumab. While this only happened for the Pertuzumab model, and thereby we are cautious to generalize this to all antibodies, such observations do advocate for the use of IgM as a potential CH variant when working with weaker binding antibody candidates. Although IgG has been the predominant choice for therapeutic antibodies, IgM may yet be more suitable than IgG when it can increase overall binding strength and its superior agglutination effects when it comes to circulating antigens. However, the use of IgM as a therapeutics drug included a historical setback 29 , and the lack of a suitable animal model further present as major obstacles. While there are multiple receptors for IgM such as Fcα/μR and FcμR, and that human Fcα/μR binds both human and mouse IgM and IgA, it is expressed on different cell types compared to mice. In human, only pre-germinal centre B-cells (IgD + /CD38 + cells) express the receptors, while the murine counterparts are only expressed on circulating and resident B cell population 30 . Cross reactivity of the murine Fcα/μR to human IgM and IgA is not yet established, rendering this receptor unsuitable for animal model investigations. Nonetheless, a recently discovered receptor - FcμR(TOSO/FAIM3) 31,32 - present on both human and mice, have demonstrated cross-species IgM reactivity. There are however also differences in the receptor localisation. It is expressed in both human and murine spleen and thymus, but also on peripheral blood leukocytes for humans, and bone marrow and lymph nodes for mice 31,32 .Given such differences, it is therefore unlikely that the mouse model would be suitable for IgM investigations, especially for localization experiments.

Our other CH variants, IgA1, IgA2, and IgE showed similar dissociation equilibrium constants, association and dissociation rate readings to the IgG1 subtype when pre-loaded to Protein L sensors, making them good possible candidates for mucosal cancers and AllergoOncology (IgE). There are hints of avidity effects when we compared the results to preloaded antigen (Her2), Trastuzumab isotype variants displayed more avidity effects where the IgM and IgD variants had more fluctuations in the association constant, but not for the Pertuzumab isotype variants. Similarly, variability in the dissociation rates between the CH variants, being more pronounced on both antibody models, had differences where for Trastuzumab, IgE, IgA2 and IgD showed 1-log difference when bound to pre-loaded Her2, whereas for Pertuzumab, the differences in the Kd were more pronounced for IgM, IgA1 and IgA2.

Other than to narrow it down to isotype-CDR-linked effects, we were unable to determine the cause of these avidity differences. Nonetheless, the pre-loading of the antibody variants on the Protein L have shown that the V-regions of these isotype variants were functional in order to be recognized by Protein L and also bind Her2.

Since many Her2 + cancers occur in ductal areas, and our results did not exclude the suitability of IgE and IgA isotype variants, the availability of such Trastuzumab/Pertuzumab IgE and IgA variants may be ideal for such localities. These variants can be used in conjunction with reduced dosage of blood-based IgM/IgG variants to check circulating cancer metastasis, and in the process, reduce cardiac side effects associated with high doses of Trastuzumab IgG1 alone. However, the lack of a mouse Ig receptor that responds to human IgA 12 and IgE 15 are perhaps the main reasons to why therapeutic antibodies of such isotypes are not yet common, and may not be approved for clinical trials in the near future.

Although the results of the IgG variants are similar, various human IgG subtypes (IgG2 and IgG4) exhibited different protection properties in mice against Crytococcus neoformans 22 (encapsulated yeast). Since all human IgG subtypes can cross placenta 33 , therapeutics of this isotype is not likely to be suitable for use during pregnancy, thus advocating for the use of other isotypes to reduce possibilities of affecting the unborn and thus the withholding of targeted therapies during pregnancies.

As proof of concept in different localization, our preliminary experiments (not shown) in nude mice showed that IgG stayed longer in systemic circulation compared with the other isotypes, which were cleared in the kidneys and livers within the first few days. However this result is inconclusive as localization experiments in mice are severely limited given that mice Ig receptors for the other isotypes do not interact with the human Ig variants.

Nevertheless, the comparison of the different CH variants showed a trend of general better binding measurements when using the human IgG Fc capture AHC biosensors (Fig. 5) over that of Protein L biosensors (Fig. 3). Since our previous work on Trastuzumab light chains 19 showed that VL frameworks can affect protein L binding, we sought to rule out interferences elicited by Protein L to the Her2 binding site by using AHC instead of Protein L when focusing on light chain analysis (Fig. 3).

It remains enigmatic to why the human Ig system has 5 families of Cλ but only one Cκ. Although studies have shown that free floating light chains are associated with inflammation responses and autoimmunity 34 , the role of the light chain has been fairly elusive with the lack of known specific Igκ and Igλ receptors,. Light chains have been found to affect the in vivo half-life of antibodies (where huIgG2λ had a shorter half-life than huIgG2κ in mice) 35 , as well as play a role in antigen dissociation 19 . For this reason, it may be that the effects of CL on antibody half-life and antigen-binding is not solely dependent only on CL-CH, but must include the other regions as whole heavy and light chains. This demonstrates the need for detailed analysis of how the variable and constant regions of the antibodies of both chains come together to affect antigen-binding, production and also half-lives.

Our CL findings are consistent with Montaño et al. 35 (where the CL did not elicit significant effects), but differed from Ponomarenko et al. (whose team found the CL variants to display different bindings to cyclic and linear peptide epitopes). Since our antigen is the recombinant extracellular portion of Her2, and Ponomarenko et al. observed differences based on the conformation of the antigen, it is possible that CL variation effects may be antibody-antigen interaction dependent 18 and cannot be generalized to all antibodies. It is likely that the structural and rigidity of the antibody, influenced by modification in the light chain as hinted by Toughiri et al. 20 , may underlie the mechanism for the differences. Nonetheless, it is clear that for recombinant extracellular Her2 binding, both Trastuzumab and Pertuzumab CL variants did not show significant effects in our assays.

Currently, there remain significant hurdles for CH and CL variants of therapeutic antibodies to be adopted in actual clinical use given that pre-clinical animal experiments in this area are severely limited. Without suitable methods to assess both the potential benefits and immunopathology of antibody isotypes/subtypes as therapeutics (e.g. IgE anaphylaxis, IgA nephropathy, IgM rheumatoid factor, Hyper-IgD syndrome), it may be decades before CH isotype swapping are adopted as the next generation of antibody therapeutics.


  1. 1. Diamandis EP, Christopoulos TK (1991). The biotin-(strept) avidin system: principles and applications in biotechnology. Clin Chem 37: 625–636. pmid:2032315
    • View Article
    • PubMed/NCBI
    • Google Scholar
  2. 2. Holliger P, Hudson PJ (2005). Engineered antibody fragments and the rise of single domains. Nat Biotechnol 23: 1126–1136. pmid:16151406
    • View Article
    • PubMed/NCBI
    • Google Scholar
  3. 3. Hoogenboom HR (1997). Designing and optimizing library selection strategies for generating high-affinity antibodies. Trends Biotechnol 15: 62–70. pmid:9081300
    • View Article
    • PubMed/NCBI
    • Google Scholar
  4. 4. Adams GP, Weiner LM (2005). Monoclonal antibody therapy of cancer. Nat Biotechnol 23: 1147–1157. pmid:16151408
    • View Article
    • PubMed/NCBI
    • Google Scholar
  5. 5. Wu AM, Senter PD (2005). Arming antibodies: prospects and challenges for immunoconjugates. Nat Biotechnol 23: 1137–1146. pmid:16151407
    • View Article
    • PubMed/NCBI
    • Google Scholar
  6. 6. Adams GP, Schier R, McCall AM, Simmons HH, Horak EM, Alpaugh RK, Marks JD, Weiner LM (2001). High affinity restricts the localization and tumor penetration of single-chain fv antibody molecules. Cancer Res 61: 4750–4755. pmid:11406547
    • View Article
    • PubMed/NCBI
    • Google Scholar
  7. 7. Yang L, Mao H, Wang YA, Cao Z, Peng X, Wang X, Duan H, Ni C, Yuan Q, Adams G (2009). Single chain epidermal growth factor receptor antibody conjugated nanoparticles for in vivo tumor targeting and imaging. Small 5: 235–243. pmid:19089838
    • View Article
    • PubMed/NCBI
    • Google Scholar
  8. 8. Smith GP (1985). Filamentous fusion phage: novel expression vectors that display cloned antigens on the virion surface. Science 228: 1315–1317. pmid:4001944
    • View Article
    • PubMed/NCBI
    • Google Scholar
  9. 9. McCafferty J, Griffiths AD, Winter G, Chiswell DJ (1990). Phage antibodies: filamentous phage displaying antibody variable domains. Nature 348: 552–554. pmid:2247164
    • View Article
    • PubMed/NCBI
    • Google Scholar
  10. 10. Clackson T, Hoogenboom HR, Griffiths AD, Winter G (1991). Making antibody fragments using phage display libraries. Nature 352: 624–628. pmid:1907718
    • View Article
    • PubMed/NCBI
    • Google Scholar
  11. 11. Winter G, Griffiths AD, Hawkins RE, Hoogenboom HR (1994). Making antibodies by phage display technology. Annu Rev Immunol 12: 433–455. pmid:8011287
    • View Article
    • PubMed/NCBI
    • Google Scholar
  12. 12. Hoogenboom HR, Chames P (2000). Natural and designer binding sites made by phage display technology. Immunol Today 21: 371–378. pmid:10916139
    • View Article
    • PubMed/NCBI
    • Google Scholar
  13. 13. Bratkovič T (2010). Progress in phage display: evolution of the technique and its applications. Cell Mol Life Sci 67: 749–767. pmid:20196239
    • View Article
    • PubMed/NCBI
    • Google Scholar
  14. 14. Hawkins RE, Russell SJ, Winter G (1992). Selection of phage antibodies by binding affinity: mimicking affinity maturation. J Mol Biol 226: 889–896. pmid:1507232
    • View Article
    • PubMed/NCBI
    • Google Scholar
  15. 15. Daugherty PS, Chen G, Olsen MJ, Iverson BL, Georgiou G (1998). Antibody affinity maturation using bacterial surface display. Protein Eng 11: 825–832. pmid:9796833
    • View Article
    • PubMed/NCBI
    • Google Scholar
  16. 16. Luginbühl B, Kanyo Z, Jones RM, Fletterick RJ, Prusiner SB, Cohen FE, Williamson RA, Burton DR, Plückthun A (2006). Directed evolution of an anti-prion protein scFv fragment to an affinity of 1 pM and its structural interpretation. J Mol Biol 363: 75–97. pmid:16962610
    • View Article
    • PubMed/NCBI
    • Google Scholar
  17. 17. Lonberg N (2008) Human monoclonal antibodies from transgenic mice. Therapeutic Antibodies: Springer. pp. 69–97.
  18. 18. Milstein C (1999). The hybridoma revolution: an offshoot of basic research. Bioessays 21: 966–973. pmid:10517870
    • View Article
    • PubMed/NCBI
    • Google Scholar
  19. 19. Lonberg N, Huszar D (1995). Human antibodies from transgenic mice. International reviews of immunology 13: 65–93. pmid:7494109
    • View Article
    • PubMed/NCBI
    • Google Scholar
  20. 20. Sergeeva A, He H, Ruisaard K, St John L, Alatrash G, Clise-Dwyer K, Li D, Patenia R, Hong R, Sukhumalchandra P, You M, Gagea M, Ma Q, Molldrem J (2016). Activity of 8F4, a T-cell receptor-like anti-PR1/HLA-A2 antibody, against primary human AML in vivo. Leukemia 30: 1475–1484. pmid:27055866
    • View Article
    • PubMed/NCBI
    • Google Scholar
  21. 21. Kawashima Y, Kurima K, Pan B, Griffith AJ, Holt JR (2015). Transmembrane channel-like (TMC) genes are required for auditory and vestibular mechanosensation. European Journal of Physiology 467: 85–94. pmid:25074487
    • View Article
    • PubMed/NCBI
    • Google Scholar
  22. 22. Tiller KE, Tessier PM (2015). Advances in antibody design. Annu Rev Biomed Eng 17: 191–216. pmid:26274600
    • View Article
    • PubMed/NCBI
    • Google Scholar
  23. 23. Corti D, Lanzavecchia A (2013). Broadly neutralizing antiviral antibodies. Annu Rev Immunol 31: 705–742. pmid:23330954
    • View Article
    • PubMed/NCBI
    • Google Scholar
  24. 24. Yuan Q-A, Simmons HH, Robinson MK, Russeva M, Marasco WA, Adams GP (2006). Development of engineered antibodies specific for the Müllerian inhibiting substance type II receptor: a promising candidate for targeted therapy of ovarian cancer. Mol Cancer Ther 5: 2096–2105. pmid:16928831
    • View Article
    • PubMed/NCBI
    • Google Scholar
  25. 25. Ohaegbulam KC, Assal A, Lazar-Molnar E, Yao Y, Zang X (2015). Human cancer immunotherapy with antibodies to the PD-1 and PD-L1 pathway. Trends Mol Med 21: 24–33. pmid:25440090
    • View Article
    • PubMed/NCBI
    • Google Scholar
  26. 26. Kuroda D, Shirai H, Jacobson MP, Nakamura H (2012). Computer-aided antibody design. Protein Eng Des Sel 25: 507–521. pmid:22661385
    • View Article
    • PubMed/NCBI
    • Google Scholar
  27. 27. Marshall SA, Lazar GA, Chirino AJ, Desjarlais JR (2003). Rational design and engineering of therapeutic proteins. Drug Discov Today 8: 212–221. pmid:12634013
    • View Article
    • PubMed/NCBI
    • Google Scholar
  28. 28. Caravella JA, Wang D, Glaser SM, Lugovskoy A (2010). Structure-guided design of antibodies. Current computer-aided drug design 6: 128–138. pmid:26845329
    • View Article
    • PubMed/NCBI
    • Google Scholar
  29. 29. Lippow SM, Wittrup KD, Tidor B (2007). Computational design of antibody-affinity improvement beyond in vivo maturation. Nat Biotechnol 25: 1171–1176. pmid:17891135
    • View Article
    • PubMed/NCBI
    • Google Scholar
  30. 30. Barderas R, Desmet J, Timmerman P, Meloen R, Casal JI (2008). Affinity maturation of antibodies assisted by in silico modeling. Proceedings of the National Academy of Sciences 105: 9029–9034.
    • View Article
    • Google Scholar
  31. 31. Karanicolas J, Kuhlman B (2009). Computational design of affinity and specificity at protein–protein interfaces. Curr Opin Struct Biol 19: 458–463. pmid:19646858
    • View Article
    • PubMed/NCBI
    • Google Scholar
  32. 32. Chennamsetty N, Voynov V, Kayser V, Helk B, Trout BL (2009). Design of therapeutic proteins with enhanced stability. Proceedings of the National Academy of Sciences 106: 11937–11942.
    • View Article
    • Google Scholar
  33. 33. Miklos AE, Kluwe C, Der BS, Pai S, Sircar A, Hughes RA, Berrondo M, Xu J, Codrea V, Buckley PE (2012). Structure-based design of supercharged, highly thermoresistant antibodies. Chem Biol 19: 449–455. pmid:22520751
    • View Article
    • PubMed/NCBI
    • Google Scholar
  34. 34. Der BS, Kluwe C, Miklos AE, Jacak R, Lyskov S, Gray JJ, Georgiou G, Ellington AD, Kuhlman B (2013). Alternative computational protocols for supercharging protein surfaces for reversible unfolding and retention of stability. PLOS ONE 8: e64363. pmid:23741319
    • View Article
    • PubMed/NCBI
    • Google Scholar
  35. 35. Pulito VL, Roberts VA, Adair JR, Rothermel AL, Collins AM, Varga SS, Martocello C, Bodmer M, Jolliffe LK, Zivin RA (1996). Humanization and molecular modeling of the anti-CD4 monoclonal antibody, OKT4A. The Journal of Immunology 156: 2840–2850. pmid:8609403
    • View Article
    • PubMed/NCBI
    • Google Scholar
  36. 36. Lazar GA, Desjarlais JR, Jacinto J, Karki S, Hammond PW (2007). A molecular immunology approach to antibody humanization and functional optimization. Mol Immunol 44: 1986–1998. pmid:17079018
    • View Article
    • PubMed/NCBI
    • Google Scholar
  37. 37. Qiao C, Lv M, Li X, Geng J, Li Y, Zhang J, Lin Z, Feng J, Shen B (2013). Affinity maturation of antiHER2 monoclonal antibody MIL5 using an epitope-specific synthetic phage library by computational design. J Biomol Struct Dyn 31: 511–521. pmid:23003339
    • View Article
    • PubMed/NCBI
    • Google Scholar
  38. 38. Jacobs TM, Yumerefendi H, Kuhlman B, Leaver-Fay A (2014). SwiftLib: rapid degenerate-codon-library optimization through dynamic programming. Nucleic Acids Res 43: e34. pmid:25539925
    • View Article
    • PubMed/NCBI
    • Google Scholar
  39. 39. Ponsel D, Neugebauer J, Ladetzki-Baehs K, Tissot K (2011). High affinity, developability and functional size: the holy grail of combinatorial antibody library generation. Molecules 16: 3675–3700. pmid:21540796
    • View Article
    • PubMed/NCBI
    • Google Scholar
  40. 40. Chaparro‐Riggers JF, Polizzi KM, Bommarius AS (2007). Better library design: data‐driven protein engineering. Biotechnology journal 2: 180–191. pmid:17183506
    • View Article
    • PubMed/NCBI
    • Google Scholar
  41. 41. Pantazes R, Maranas C (2010). OptCDR: a general computational method for the design of antibody complementarity determining regions for targeted epitope binding. Protein Eng Des Sel 23: 849–858. pmid:20847101
    • View Article
    • PubMed/NCBI
    • Google Scholar
  42. 42. Li T, Pantazes RJ, Maranas CD (2014). OptMAVEn–A New Framework for the de novo Design of Antibody Variable Region Models Targeting Specific Antigen Epitopes. PLOS ONE 9: e105954. pmid:25153121
    • View Article
    • PubMed/NCBI
    • Google Scholar
  43. 43. Dunbrack RL Jr., Cohen FE (1997). Bayesian statistical analysis of protein side-chain rotamer preferences. Protein Sci 6: 1661–1681. pmid:9260279
    • View Article
    • PubMed/NCBI
    • Google Scholar
  44. 44. Pantazes RJ, Maranas CD (2013). MAPs: a database of modular antibody parts for predicting tertiary structures and designing affinity matured antibodies. BMC Bioinformatics 14: 168. pmid:23718826
    • View Article
    • PubMed/NCBI
    • Google Scholar
  45. 45. Gray JJ, Moughon S, Wang C, Schueler-Furman O, Kuhlman B, Rohl CA, Baker D (2003). Protein–protein docking with simultaneous optimization of rigid-body displacement and side-chain conformations. J Mol Biol 331: 281–299. pmid:12875852
    • View Article
    • PubMed/NCBI
    • Google Scholar
  46. 46. Poosarla VG, Li T, Goh BC, Schulten K, Wood TK, Maranas CD (2017). Computational de novo design of antibodies binding to a peptide with high affinity. Biotechnol Bioeng 6: 1331–1342.
    • View Article
    • Google Scholar
  47. 47. Lapidoth GD, Baran D, Pszolla GM, Norn C, Alon A, Tyka MD, Fleishman SJ (2015). AbDesign: An algorithm for combinatorial backbone design guided by natural conformations and sequences. Proteins 83: 1385–1406. pmid:25670500
    • View Article
    • PubMed/NCBI
    • Google Scholar
  48. 48. Le Grand SM, Merz KM (1993). Rapid approximation to molecular surface area via the use of Boolean logic and look‐up tables. J Comput Chem 14: 349–352.
    • View Article
    • Google Scholar
  49. 49. Chothia C, Lesk AM (1987). Canonical structures for the hypervariable regions of immunoglobulins. J Mol Biol 196: 901–917. pmid:3681981
    • View Article
    • PubMed/NCBI
    • Google Scholar
  50. 50. Al-Lazikani B, Lesk AM, Chothia C (1997). Standard conformations for the canonical structures of immunoglobulins. J Mol Biol 273: 927–948. pmid:9367782
    • View Article
    • PubMed/NCBI
    • Google Scholar
  51. 51. Martin AC, Thornton JM (1996). Structural families in loops of homologous proteins: automatic classification, modelling and application to antibodies. J Mol Biol 263: 800–815. pmid:8947577
    • View Article
    • PubMed/NCBI
    • Google Scholar
  52. 52. Morea V, Tramontano A, Rustici M, Chothia C, Lesk AM (1998). Conformations of the third hypervariable region in the VH domain of immunoglobulins. J Mol Biol 275: 269–294. pmid:9466909
    • View Article
    • PubMed/NCBI
    • Google Scholar
  53. 53. North B, Lehmann A, Dunbrack RL Jr. (2011). A new clustering of antibody CDR loop conformations. J Mol Biol 406: 228–256. pmid:21035459
    • View Article
    • PubMed/NCBI
    • Google Scholar
  54. 54. Adolf-Bryfogle J, Xu Q, North B, Lehmann A, Dunbrack RL Jr. (2015). PyIgClassify: a database of antibody CDR structural classifications. Nucleic Acids Res 43: D432–438. pmid:25392411
    • View Article
    • PubMed/NCBI
    • Google Scholar
  55. 55. Lefranc MP, Giudicelli V, Ginestoux C, Jabado-Michaloud J, Folch G, Bellahcene F, Wu Y, Gemrot E, Brochet X, Lane J, Regnier L, Ehrenmann F, Lefranc G, Duroux P (2009). IMGT, the international ImMunoGeneTics information system. Nucleic Acids Res 37: D1006–1012. pmid:18978023
    • View Article
    • PubMed/NCBI
    • Google Scholar
  56. 56. Das R, Baker D (2008). Macromolecular modeling with rosetta. Annu Rev Biochem 77: 363–382. pmid:18410248
    • View Article
    • PubMed/NCBI
    • Google Scholar
  57. 57. Mandell DJ, Coutsias EA, Kortemme T (2009). Sub-angstrom accuracy in protein loop reconstruction by robotics-inspired conformational sampling. Nat Methods 6: 551–552. pmid:19644455
    • View Article
    • PubMed/NCBI
    • Google Scholar
  58. 58. Das R (2013). Atomic-accuracy prediction of protein loop structures through an RNA-inspired ansatz. PLOS ONE 8: e74830. pmid:24204571
    • View Article
    • PubMed/NCBI
    • Google Scholar
  59. 59. Wang C, Bradley P, Baker D (2007). Protein-protein docking with backbone flexibility. J Mol Biol 373: 503–519. pmid:17825317
    • View Article
    • PubMed/NCBI
    • Google Scholar
  60. 60. Tyka MD, Keedy DA, Andre I, Dimaio F, Song Y, Richardson DC, Richardson JS, Baker D (2011). Alternate states of proteins revealed by detailed energy landscape mapping. J Mol Biol 405: 607–618. pmid:21073878
    • View Article
    • PubMed/NCBI
    • Google Scholar
  61. 61. Conway P, Tyka MD, DiMaio F, Konerding DE, Baker D (2014). Relaxation of backbone bond geometry improves protein energy landscape modeling. Protein Sci 23: 47–55. pmid:24265211
    • View Article
    • PubMed/NCBI
    • Google Scholar
  62. 62. Nivon LG, Moretti R, Baker D (2013). A Pareto-optimal refinement method for protein design scaffolds. PLOS ONE 8: e59004. pmid:23565140
    • View Article
    • PubMed/NCBI
    • Google Scholar
  63. 63. Smith CA, Kortemme T (2008). Backrub-like backbone simulation recapitulates natural protein conformational variability and improves mutant side-chain prediction. J Mol Biol 380: 742–756. pmid:18547585
    • View Article
    • PubMed/NCBI
    • Google Scholar
  64. 64. Kuhlman B, Dantas G, Ireton GC, Varani G, Stoddard BL, Baker D (2003). Design of a novel globular protein fold with atomic-level accuracy. Science 302: 1364–1368. pmid:14631033
    • View Article
    • PubMed/NCBI
    • Google Scholar
  65. 65. Röthlisberger D, Khersonsky O, Wollacott AM, Jiang L, DeChancie J, Betker J, Gallaher JL, Althoff EA, Zanghellini A, Dym O (2008). Kemp elimination catalysts by computational enzyme design. Nature 453: 190–195. pmid:18354394
    • View Article
    • PubMed/NCBI
    • Google Scholar
  66. 66. Jiang L, Althoff EA, Clemente FR, Doyle L, Röthlisberger D, Zanghellini A, Gallaher JL, Betker JL, Tanaka F, Barbas CF (2008). De novo computational design of retro-aldol enzymes. science 319: 1387–1391. pmid:18323453
    • View Article
    • PubMed/NCBI
    • Google Scholar
  67. 67. Richter F, Leaver-Fay A, Khare SD, Bjelic S, Baker D (2011). De novo enzyme design using Rosetta3. PLOS ONE 6: e19230. pmid:21603656
    • View Article
    • PubMed/NCBI
    • Google Scholar
  68. 68. Guntas G, Purbeck C, Kuhlman B (2010). Engineering a protein-protein interface using a computationally designed library. Proc Natl Acad Sci U S A 107: 19296–19301. pmid:20974935
    • View Article
    • PubMed/NCBI
    • Google Scholar
  69. 69. Fleishman SJ, Corn JE, Strauch EM, Whitehead TA, Karanicolas J, Baker D (2011). Hotspot-centric de novo design of protein binders. J Mol Biol 413: 1047–1062. pmid:21945116
    • View Article
    • PubMed/NCBI
    • Google Scholar
  70. 70. Fleishman SJ, Whitehead TA, Ekiert DC, Dreyfus C, Corn JE, Strauch EM, Wilson IA, Baker D (2011). Computational design of proteins targeting the conserved stem region of influenza hemagglutinin. Science 332: 816–821. pmid:21566186
    • View Article
    • PubMed/NCBI
    • Google Scholar
  71. 71. Canutescu AA, Dunbrack RL Jr. (2003). Cyclic coordinate descent: A robotics algorithm for protein loop closure. Protein Sci 12: 963–972. pmid:12717019
    • View Article
    • PubMed/NCBI
    • Google Scholar
  72. 72. Raha K, Wollacott AM, Italia MJ, Desjarlais JR (2000). Prediction of amino acid sequence from structure. Protein Sci 9: 1106–1119. pmid:10892804
    • View Article
    • PubMed/NCBI
    • Google Scholar
  73. 73. Leaver-Fay A, O'Meara MJ, Tyka M, Jacak R, Song Y, Kellogg EH, Thompson J, Davis IW, Pache RA, Lyskov S, Gray JJ, Kortemme T, Richardson JS, Havranek JJ, Snoeyink J, Baker D, Kuhlman B (2013). Scientific benchmarks for guiding macromolecular energy function improvement. Methods Enzymol 523: 109–143. pmid:23422428
    • View Article
    • PubMed/NCBI
    • Google Scholar
  74. 74. Davies HTO, Crombie IK, Tavakoli M (1998). When can odds ratios mislead? Bmj 316: 989–991. pmid:9550961
    • View Article
    • PubMed/NCBI
    • Google Scholar
  75. 75. Ihaka R, Gentleman R (1996). R: A language for data analysis and graphics. J Comp Graphical Stat 5: 299–314.
    • View Article
    • Google Scholar
  76. 76. R Core Team (2015) R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing.
  77. 77. Sircar A, Kim ET, Gray JJ (2009). RosettaAntibody: antibody variable region homology modeling server. Nucleic Acids Res 37: W474–479. pmid:19458157
    • View Article
    • PubMed/NCBI
    • Google Scholar
  78. 78. Weitzner BD, Kuroda D, Marze N, Xu J, Gray JJ (2014). Blind prediction performance of RosettaAntibody 3.0: grafting, relaxation, kinematic loop modeling, and full CDR optimization. Proteins 82: 1611–1623. pmid:24519881
    • View Article
    • PubMed/NCBI
    • Google Scholar
  79. 79. Weitzner BD, Jeliazkov JR, Lyskov S, Marze N, Kuroda D, Frick R, Adolf-Bryfogle J, Biswas N, Dunbrack Jr RL, Gray JJ (2017). Modeling and docking of antibody structures with Rosetta. Nature protocols 12: 401–416. pmid:28125104
    • View Article
    • PubMed/NCBI
    • Google Scholar
  80. 80. Nayeem A, Vila J, Scheraga HA (1991). A comparative study of the simulated‐annealing and Monte Carlo‐with‐minimization approaches to the minimum‐energy structures of polypeptides:[Met]‐enkephalin. J Comp Chem 12: 594–605.
    • View Article
    • Google Scholar
  81. 81. Lewis SM, Kuhlman BA (2011). Anchored design of protein-protein interfaces. PLOS ONE 6: e20872. pmid:21698112
    • View Article
    • PubMed/NCBI
    • Google Scholar
  82. 82. Liao HX, Lynch R, Zhou T, Gao F, Alam SM, Boyd SD, Fire AZ, Roskin KM, Schramm CA, Zhang Z, Zhu J, Shapiro L, Program NCS, Mullikin JC, Gnanakaran S, Hraber P, Wiehe K, Kelsoe G, Yang G, Xia SM, Montefiori DC, Parks R, Lloyd KE, Scearce RM, Soderberg KA, Cohen M, Kamanga G, Louder MK, Tran LM, Chen Y, Cai F, Chen S, Moquin S, Du X, Joyce MG, Srivatsan S, Zhang B, Zheng A, Shaw GM, Hahn BH, Kepler TB, Korber BT, Kwong PD, Mascola JR, Haynes BF (2013). Co-evolution of a broadly neutralizing HIV-1 antibody and founder virus. Nature 496: 469–476. pmid:23552890
    • View Article
    • PubMed/NCBI
    • Google Scholar
  83. 83. Padavattan S, Schirmer T, Schmidt M, Akdis C, Valenta R, Mittermann I, Soldatova L, Slater J, Mueller U, Markovic-Housley Z (2007). Identification of a B-cell epitope of hyaluronidase, a major bee venom allergen, from its crystal structure in complex with a specific Fab. J Mol Biol 368: 742–752. pmid:17374540
    • View Article
    • PubMed/NCBI
    • Google Scholar
  84. 84. Lehmann A, Wixted JH, Shapovalov MV, Roder H, Dunbrack RL Jr., Robinson MK (2015). Stability engineering of anti-EGFR scFv antibodies by rational design of a lambda-to-kappa swap of the VL framework using a structure-guided approach. mAbs 7: 1058–1071. pmid:26337947
    • View Article
    • PubMed/NCBI
    • Google Scholar
  85. 85. Adolf-Bryfogle J, Dunbrack Jr RL (2013). The PyRosetta Toolkit: a graphical user interface for the Rosetta software suite. PLOS ONE 8: e66856. pmid:23874400
    • View Article
    • PubMed/NCBI
    • Google Scholar
  86. 86. Kleffner R, Flatten J, Leaver-Fay A, Baker D, Siegel JB, Khatib F, Cooper S (2017). Foldit Standalone: a video game-derived protein structure manipulation interface using Rosetta. Bioinformatics: btx283.
    • View Article
    • Google Scholar
  87. 87. James LC, Roversi P, Tawfik DS (2003). Antibody multispecificity mediated by conformational diversity. Science 299: 1362–1367. pmid:12610298
    • View Article
    • PubMed/NCBI
    • Google Scholar
  88. 88. Jha RK, Leaver-Fay A, Yin S, Wu Y, Butterfoss GL, Szyperski T, Dokholyan NV, Kuhlman B (2010). Computational design of a PAK1 binding protein. J Mol Biol 400: 257–270. pmid:20460129
    • View Article
    • PubMed/NCBI
    • Google Scholar
  89. 89. Sammond DW, Bosch DE, Butterfoss GL, Purbeck C, Machius M, Siderovski DP, Kuhlman B (2011). Computational design of the sequence and structure of a protein-binding peptide. J Am Chem Soc 133: 4190–4192. pmid:21388199
    • View Article
    • PubMed/NCBI
    • Google Scholar
  90. 90. Stranges PB, Kuhlman B (2013). A comparison of successful and failed protein interface designs highlights the challenges of designing buried hydrogen bonds. Protein Sci 22: 74–82. pmid:23139141
    • View Article
    • PubMed/NCBI
    • Google Scholar
  91. 91. Clark LA, Boriack‐Sjodin PA, Eldredge J, Fitch C, Friedman B, Hanf KJ, Jarpe M, Liparoto SF, Li Y, Lugovskoy A, Miller S, Rushe M, Sherman W, Simon K, Van Vlijmen H (2006). Affinity enhancement of an in vivo matured therapeutic antibody using structure‐based computational design. Protein Sci 15: 949–960. pmid:16597831
    • View Article
    • PubMed/NCBI
    • Google Scholar
  92. 92. Wei Q, Dunbrack Jr RL (2013). The role of balanced training and testing data sets for binary classifiers in bioinformatics. PLOS ONE 8: e67863. pmid:23874456
    • View Article
    • PubMed/NCBI
    • Google Scholar
  93. 93. Alford R, Leaver-Fay A, Jeliazkov J, O'Meara M, DiMaio F, Park H, Shapovalov M, Renfrew P, Mulligan V, Kappel K, Labonte JW, Pacella MS, Bonneau R, Bradley P, Dunbrack Jr RL, Das R, Baker D, Kuhlman B, Kortemme T, Gray JJ (2017). The Rosetta All-Atom Energy Function for Macromolecular Modeling and Design. Journal of chemical theory and computation 13: 3031–3048. pmid:28430426
    • View Article
    • PubMed/NCBI
    • Google Scholar
  94. 94. Shapovalov MV, Dunbrack RL Jr. (2011). A smoothed backbone-dependent rotamer library for proteins derived from adaptive kernel density estimates and regressions. Structure 19: 844–858. pmid:21645855
    • View Article
    • PubMed/NCBI
    • Google Scholar
  95. 95. Weitzner BD, Dunbrack RL Jr., Gray JJ (2015). The origin of CDR H3 structural diversity. Structure 23: 302–311. pmid:25579815
    • View Article
    • PubMed/NCBI
    • Google Scholar
  96. 96. Weitzner BD, Gray JJ (2017). Accurate Structure Prediction of CDR H3 Loops Enabled by a Novel Structure-Based C-Terminal Constraint. J Immunol 198: 505–515. pmid:27872211
    • View Article
    • PubMed/NCBI
    • Google Scholar
  97. 97. Fleishman SJ, Leaver-Fay A, Corn JE, Strauch EM, Khare SD, Koga N, Ashworth J, Murphy P, Richter F, Lemmon G, Meiler J, Baker D (2011). RosettaScripts: a scripting language interface to the Rosetta macromolecular modeling suite. PLOS ONE 6: e20161. pmid:21731610
    • View Article
    • PubMed/NCBI
    • Google Scholar
  98. 98. Renfrew PD, Choi EJ, Bonneau R, Kuhlman B (2012). Incorporation of noncanonical amino acids into Rosetta and use in computational protein-peptide interface design. PLOS ONE 7: e32637. pmid:22431978
    • View Article
    • PubMed/NCBI
    • Google Scholar
  99. 99. Berkholz DS, Shapovalov MV, Dunbrack RL Jr., Karplus PA (2009). Conformation dependence of backbone geometry in proteins. Structure 17: 1316–1325. pmid:19836332
    • View Article
    • PubMed/NCBI
    • Google Scholar
  100. 100. Henikoff S, Henikoff JG (1992). Amino acid substitution matrices from protein blocks. Proc Natl Acad Sci USA 89: 10915–10919. pmid:1438297
    • View Article
    • PubMed/NCBI
    • Google Scholar
  101. 101. Davis IW, Arendall WB 3rd, Richardson DC, Richardson JS (2006). The backrub motion: how protein backbone shrugs when a sidechain dances. Structure 14: 265–274. pmid:16472746
    • View Article
    • PubMed/NCBI
    • Google Scholar
  102. 102. Jacobs TM, Williams B, Williams T, Xu X, Eletsky A, Federizon JF, Szyperski T, Kuhlman B (2016). Design of structurally distinct proteins using strategies inspired by evolution. Science 352: 687–690. pmid:27151863
    • View Article
    • PubMed/NCBI
    • Google Scholar
  103. 103. Chaudhury S, Lyskov S, Gray JJ (2011). PyRosetta: a script-based interface for implementing molecular modeling algorithms using Rosetta. Bioinformatics 26: 689–691.
    • View Article
    • Google Scholar
  104. 104. Kilambi KP, Pacella MS, Xu J, Labonte JW, Porter JR, Muthu P, Drew K, Kuroda D, Schueler-Furman O, Bonneau R, Gray JJ (2013). Extending RosettaDock with water, sugar, and pH for prediction of complex structures and affinities for CAPRI rounds 20–27. Proteins 81: 2201–2209. pmid:24123494
    • View Article
    • PubMed/NCBI
    • Google Scholar
  105. 105. Marze NA, Jeliazkov JR, Roy Burman SS, Boyken SE, DiMaio F, Gray JJ (2016). Modeling oblong proteins and water-mediated interfaces with RosettaDock in CAPRI rounds 28–35. Proteins 85: 479–486. pmid:27667482
    • View Article
    • PubMed/NCBI
    • Google Scholar
  106. 106. Schueler‐Furman O, Wang C, Baker D (2005). Progress in protein–protein docking: Atomic resolution predictions in the CAPRI experiment using RosettaDock with an improved treatment of side‐chain flexibility. Proteins: Structure, Function, andBioinformatics 60: 187–194.
    • View Article
    • Google Scholar
  107. 107. Honegger A, Plückthun A (2001). Yet another numbering scheme for immunoglobulin variable domains: an automatic modeling and analysis tool. J Mol Biol 309: 657–670. pmid:11397087
    • View Article
    • PubMed/NCBI
    • Google Scholar
  108. 108. Gertsman BB (2013) Epidemiology Kept Simple: An Introduction to Traditional and Modern Epidemiology (3rd Edition). Oxford: John Wiley and Sons.
  109. 109. Padavattan S, Schirmer T, Schmidt M, Akdis C, Valenta R, Mittermann I, Soldatova L, Slater J, Mueller U, Markovic-Housley Z (2007). Identification of a B-cell epitope of hyaluronidase, a major bee venom allergen, from its crystal structure in complex with a specific Fab. J Mol Biol 368: 742–752. pmid:17374540
    • View Article
    • PubMed/NCBI
    • Google Scholar
  110. 110. Lawrence MC, Colman PM (1993). Shape complementarity at protein/protein interfaces. J Mol Biol 234: 946–950. pmid:8263940
    • View Article
    • PubMed/NCBI
    • Google Scholar

Subject Areas

For more information about PLOS Subject Areas, click here.

We want your feedback. Do these Subject Areas make sense for this article? Click the target next to the incorrect Subject Area and let us know. Thanks for your help!

Is the Subject Area "Antigens" applicable to this article? Yes No

Is the Subject Area "Antibodies" applicable to this article? Yes No

Is the Subject Area "Enzyme-linked immunoassays" applicable to this article? Yes No

Is the Subject Area "Monte Carlo method" applicable to this article? Yes No

Is the Subject Area "Medical risk factors" applicable to this article? Yes No

Is the Subject Area "Databases" applicable to this article? Yes No

Is the Subject Area "Computer software" applicable to this article? Yes No

Is the Subject Area "Biochemical simulations" applicable to this article? Yes No


STAR★methods

KEY RESOURCES TABLE

Reagent or resourceSourceIdentifier
Antibodies
Goat anti-Human IgG (H+L) secondary antibody, HRPThermo Fisher ScientificCat#31410 RRID: AB_228269
Mouse anti-human IgG Fab antibody (HRP)GenScriptCat#A01855-200
Mouse anti-human CD20-PECy7BD PharMingenCat#560735 RRID: AB_399985
APC Mouse anti-human CD19BD PharMingenCat#555415
Anti-CD27-PEBD BiosciencesCat#555441 RRID: AB_395834
Anti-hCD32BD PharMingenCat#557333
Anti-HBs H004Wang etਊl., 2020bN/A
CR3022ter Meulen etਊl., 2006N/A
Anti-N polyclonal antibodyGu etਊl., 2020N/A
Bacterial and virus strains
E.਌oli Trans5α chemically Competent CellsTransGen BiotechCat#CD201-01
Authentic SARS-CoV-2 virus, nCoV-SH01 (GenBank: <"type":"entrez-nucleotide","attrs":<"text":"MT121215.1","term_id":"1819735426","term_text":"MT121215.1">> MT121215.1)Wu etਊl., 2020bN/A
Chemicals, peptides, and recombinant proteins
Streptavidin HRPBD BiosciencesCat#554066
Streptavidin APCBD BiosciencesCat#554067 RRID: AB_10050396
Streptavidin PEeBioscienceCat#12-4317-87
Human BD Fc BlockBD PharMingenCat#564220
SARS-CoV-2 S protein (RBD)GenScriptCat#Z03479
SARS-CoV-2 S1 proteinGenScriptCat#Z03501
Recombinant 2019-nCoV S2 protein (C-Fc)NovoproteinCat#DRA48
SARS-CoV-2 Spike protein (ECD, His & Flag tag)GenScriptCat#Z03481
Insect-C-His NPGenScriptCat#Z03480
Recombinant 2019 nCoV Spike S (amino acid 14-1212)Kactus BiosystemsCat#COV-VM5SS
Recombinant 2019 nCoV Spike RBDKactus BiosystemsCat#COV-VM4BD
RNAsin Plus RNase inhibitorPromegaCat#N2615
4 × dNTPS (100 mM)Solarbio Life SciencesCat#PC2300
DNase/RNase-Free waterSolarbio Life SciencesCat#R1600
PBS (10 × ), pH 7.2-7.4Solarbio Life SciencesCat#P1022
1 M Tris-HCl, pH 9.0Solarbio Life SciencesCat#T1160
IGEPAL CA-630SigmaCat#I8896
Dimethyl SulfoxideSigmaCat#2650
Bovine Serum AlbuminWeiAo Biotech, ShanghaiCat#WH3044
Fetal Bovine SerumGEMINICat#900-108
UltraPure SucroseMacklin BiochemicalCat#S824459
Cresol Red sodium saltMacklin BiochemicalCat#C806031
ABTS Chromogen / substrate solution for ELISAThermo Fisher ScientificCat#00-2024
UltraPure 0.5M EDTA, pH 8.0InvitrogenCat#15575-038
Hank’s Balanced Salt Mixture (D-Hanks)Solarbio Life SciencesCat#H1045-500
EZ TransLife iLAB Bio Technology, ShanghaiCat#AC04L082
TRIzol LS ReagentThermo Fisher ScientificCat# 10296010
Critical commercial assays
LS magnetic columnsMiltenyi BiotechCat#130-042-401
CD19 MicroBeads, humanMiltenyi BiotechCat#130-097-055
EZ-Link Sulfo-NHS-LC Biotin, No weight formatThermo Fisher ScientificCat#39257
BirA Biotin-Protein Ligase KitAvidityCat#BIRA500
Zeba Spin Desalting Columns, 7K MWCOThermo Fisher ScientificCat#89889
Superscript III Reverse TranscriptaseThermo Fisher ScientificCat#18080044
HotStarTaq DNA PolymeraseQIAGENCat#203209
Protein G Sepharose 4 Fast FlowGE HealthcareCat#17061805
Pierce™ IgG Elution bufferThermo Fisher ScientificCat#21004
Histopaque-10771SigmaCat#10771
AgeI-HFNew England BioLabsCat#R3552L
BsiwI-HFNew England BioLabsCat#R3553L
XhoINew England BioLabsCat#R0146L
SalI-HFNew England BioLabsCat#R3138S
T4 DNA polymeraseNew England BioLabsCat#M0203L
One Step PrimeScript RT-PCR KitTakaraCat#RR064B
Luciferase Assay SystemPromegaCat#E1501
Experimental models: cell lines
HEK293F cell line( Wu etਊl., 2020b N/A
HEK293T cell line( Xia etਊl., 2020a N/A
Expi293 Expression SystemThermo Fisher ScientificCat#A14635
Huh-7 cell line( Xia etਊl., 2020a N/A
Vero-E6 cell line( Xia etਊl., 2020a N/A
Raji cell line( Jaume etਊl., 2011 N/A
Oligonucleotides
Random PrimersThermo Fisher ScientificCat#48190011
Recombinant DNA
IG㬱 expression vectorvon Boehmer etਊl., 2016N/A
IGκ expression vectorvon Boehmer etਊl., 2016N/A
IGλ expression vectorvon Boehmer etਊl., 2016N/A
pNL4-3.luc.REXia etਊl., 2020aN/A
pcDNA3.1-SARS-CoV-1-SXia etਊl., 2020aN/A
pcDNA3.1-SARS-COV-2-SXia etਊl., 2020aN/A
IG㬱-GRLR expression vectorRobbiani etਊl., 2019N/A
Software and algorithms
PRISMGraphPadhttps://www.graphpad.com
IgBlastYe etਊl., 2013https://www.ncbi.nlm.nih.gov/igblast/
IMGT/V-QUESTBrochet etਊl., 2008http://www.imgt.org/IMGT_vquest/vquest
Others
Sterile 50 ml Disposable Vacuum Filtration SystemMillipore SigmaCat#SCGP00525
Amicon Ultra-4 Centrifugal Filters Ultracel-30KMerck Millipore Ltd.Cat#UFC803096
Ultrafree-MC Centrifugal filter units, 0.22uM GV DURAPOREMerck Millipore Ltd.Cat#UFC30GV0S
Pipet-Lite Multi Pipette L12-20XLS+RAININCat#17013808
General Long-Term Storage Cryogenic TubesNalgeneCat#5000-1020
SimpliAmp Thermal CyclerThermo Fisher ScientificCat# <"type":"entrez-protein","attrs":<"text":"A24812","term_id":"80001","term_text":"pir||A24812">> A24812
500 mL Bottle Top Vacuum Filter, 0.20 μm PoreThermo Fisher ScientificCat#566-0020
ACCUSPIN Tubes Sterile, 50ml CapacitySigmaCat#A2055-10EA

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Qiao Wang ([email protected]).

Materials availability

All unique reagents generated in this study are available from the Lead Contact with a completed Materials Transfer Agreement. Sharing of antibodies with academic researchers may require a payment to cover the cost of generation and a completed Material Transfer Agreement.

Data and code availability

The published article includes all datasets generated or analyzed during this study. Original data have been deposited Mendeley data: https://data.mendeley.com/datasets/bjpky4bzsd/1.

Experimental model and subject details

Human subjects

Volunteer recruitment and blood draws were performed at the Zhoushan Hospital under a protocol approved by the Zhoushan Hospital Research Ethics Committee (2020-003). Experiments related to all human samples were performed at the School of Basic Medical Sciences, Fudan University under a protocol approved by the institutional Ethics Committee (2020-C007). Study participants, 16 convalescent donors, whose infections have been confirmed by PCR, and 8 unexposed naive donors. All donors ranged in age from 7-67 with a mean of 37, and the female:male ratio was 14:10 (Figure S1B).

Cell lines

Human embryonic kidney 293T (HEK293T) cells, human hepatoma Huh-7 cells and African green monkey kidney Vero-E6 cells ( Xia etਊl., 2020a ) were maintained in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS), 100 U/ml penicillin, and 100 mg/ml streptomycin. Raji cells (human Burkitt’s lymphoma B lymphoblast) ( Jaume etਊl., 2011 ), were maintained in RPMI 1640 supplemented with 10% FBS, 100 U/ml penicillin, and 100 mg/ml streptomycin. All cell lines were cultured at 37ଌ in 5% CO2. Human embryonic kidney 293F (HEK293F) suspension cells were cultured using HEK293 serum-free OPM-293-CD05 medium (OPM Biosciences) at 37ଌ in 5% CO2 with shaking at 100 rpm.

Viruses

The authentic SARS-CoV-2 virus, nCoV-SH01 (GenBank: <"type":"entrez-nucleotide","attrs":<"text":"MT121215.1","term_id":"1819735426","term_text":"MT121215.1">> MT121215.1) used in this study was isolated from infected patients at the Biosafety Level 3 (BSL-3) laboratory at the Shanghai Medical College, Fudan University ( Wu etਊl., 2020b ). The SARS-CoV-2 virus was propagated in Vero-E6 cells. Concentrated virus stock was aliquoted and stored at liquid nitrogen. One aliquot of cell line-passaged authentic SARS-CoV-2 virus, originally launched from patient serum and stored at �ଌ, was thawed for in vitro cell infection experiments.

Bacteria

E.਌oli Trans5α (TransGen Biotech) were cultured at 37ଌ with shaking at 230 rpm.

Method details

Collection of human samples

Samples of peripheral blood were collected from SARS-CoV-2 patients at the Zhoushan Hospital in Zhejiang province. Serum samples were heat inactivated for 60 minutes at 56ଌ, separated by centrifugation of coagulated whole blood, and aliquoted for storage at �ଌ. After a 400 mL blood draw from donor #16, human peripheral blood mononuclear cells (PBMCs) were isolated using a cell separation tube with frit barrier. The isolated PBMCs were resuspended in 90% heat-inactivated FBS supplemented with 10% dimethylsulfoxide (DMSO) and cryopreserved in liquid nitrogen.

Antibodies

All the cloned human monoclonal antibodies, their GRLR version, and the previously reported monoclonal antibody CR3022 ( ter Meulen etਊl., 2006 ) were prepared by transient transfection of mammalian HEK293F cells as previously reported ( Wu etਊl., 2020b ).

ELISA

The ELISA binding of serum samples or purified IgG antibody fractions from serum samples or recombinant IgG antibodies against SARS-CoV-2 proteins, including S-ECD-, RBD-, S1-, S2-, and N-proteins (see details in Key resources table) was measured as previously reported ( Wang etਊl., 2020b ). Briefly, ELISA plates were first coated with 10 μg/ml of antigen in phosphate buffered saline (PBS) overnight at 4ଌ, and then blocked with 2% bovine serum albumin (BSA) in PBS. The serum or 1 st antibody was serially diluted 1:3 in PBS (maximum concentration, 1:10 for serum, 10 μg/ml for monoclonals) for eight dilutions in total, and added for incubation for one hour at room temperature. Visualization was with HRP-conjugated goat anti-human IgG (Thermo Fisher Scientific) or HRP-conjugated mouse anti-human IgG Fab (GenScript). The area under the curve (AUC) was calculated for each antibody by analysis using PRISM software to evaluate the antigen-binding capacity. ELISA assays using RBD mutants was performed as described above except coating RBD-Fc at a concentration of 2 μg/ml, and using wild-type RBD as a reference for normalization.

Competition ELISA

Competition ELISAs were performed as described previously ( Wang etਊl., 2020b ). Briefly, plates were coated with 2 μg/ml SARS-CoV-2 RBD or 2 μg/ml SARS-CoV-2 S-ECD and incubated with 15 μg/ml 1 st blocking antibody/proteins (60 μg/ml for antibody CR3022) for two hours. Biotinylated 2 nd antibodies/proteins (0.25 μg/ml) (15 μg/ml for antibody CR3022) were directly added for 30 minutes at room temperature. Detection was performed with streptavidin-HRP (BD Biosciences). PBS buffer substituted for the 1 st blocking antibody was used as a reference for normalization, while the anti-HBs antibody H004 ( Wang etਊl., 2020b ), which could not block the binding of the 2 nd antibodies, served as a negative control.

Preparation of SARS-CoV-2 and SARS-CoV-1 pseudotyped virus

The pseudotyped viruses were produced as previously reported ( Xia etਊl., 2020a ). Briefly, plasmids pNL4-3.luc.RE (the luciferase reporter-expressing HIV-1 backbone) and pcDNA3.1-SARS-CoV-1-S/pcDNA3.1-SARS-CoV-2-S (encoding for the S-protein of SARS-CoV-1 or SARS-CoV-2) were co-transfected into HEK293T cells using the transfection reagent VigoFect (Vigorous Biotechnology, Beijing). The supernatant containing the released pseudotyped particles was harvested at 72 hours post-transfection. After centrifugation, the supernatant was collected, aliquoted, and frozen at �ଌ. The production of the SARS-CoV-2 pseudovirus mutants was performed as described above except using the plasmids of pcDNA3.1-SARS-CoV-2 with the corresponding mutations (V341I, F342L, V367F, R408I, A435S, G476S, and V483A) in the S-protein ( Ou etਊl., 2020 ). These plasmids were constructed using the plasmid of pcDNA3.1-SARS-CoV-2-S as a template by a site-directed mutation kit (Yeasen Biotech, Shanghai).

In vitro neutralization assay by pseudotyped SARS-CoV-1 and 𢄢 viruses

In vitro SARS-CoV-1 and SARS-CoV-2 pseudovirus infection was performed as previously described ( Xia etਊl., 2020a ). Briefly, 1 × 10 4 /well Huh-7 cells were seeded in 96-well plates in DMEM supplemented with 10% FBS. The seeded cells were cultured for an additional eight hours before infection. To quantitate the neutralization capacity, the human serum (maximum concentration, 1:20), polyclonal antibodies purified from human serum (maximum concentration, 50 μg/ml), or monoclonal antibodies (maximum concentration, 10 or 1 or 0.625 μg/ml) was serially diluted 1:2 in DMEM medium for nine dilutions in total. Subsequently, the diluted antibodies or serum samples were incubated with SARS-CoV-1 or 𢄢 pseudoviruses for 30 minutes at 37ଌ before added onto Huh-7 cells for infection. For the neutralization blocking experiments, different antigens (RBD, S1, S2 or S-ECD proteins) were incubated at different concentrations, respectively, with 5 μg/ml purified IgG from donor #16 for one hour at 37ଌ before incubation with SARS-CoV-2 pseudovirus. After incubation for half an hour, the mixture was finally added to the Huh-7 cells for infection. After incubation for 12 hours, the supernatant was replaced with fresh DMEM medium supplemented with 2% FBS. The cell supernatant was removed after culture for further 48 hours, and the cells were lysed for luciferase activity measurement using a Firefly Luciferase Assay Kit (Promega) and luminometer according to the manufacturer’s instructions.

The absolute luciferase values were measured and the relative values were calculated by normalizing to the virus-only control well in the same lane. For example, the absolute luciferase value in a pseudovirus-only control well (considered as reference) was 5 × 10 4 , while adding one neutralizing serum sample might reduce this to 1 × 10 4 . Therefore, the normalized luciferase values were calculated as 100% in the pseudovirus-only control and 20% for this neutralizing serum. Since many aspects, such as pseudovirus concentration, cultured cell concentration, status of the cells, immunofluorescence reading, and etc., varied dramatically between different plates and different tests, normalization is necessary for combining data for comparison. For the serum neutralization assays ( Figures 1 F and 1G), the reciprocal of the serum dilution that resulted in 50% inhibition compared with pseudovirus alone was reported as the 50% neutralization titer (NT50).

In vitro neutralization assay by authentic SARS-CoV-2 virus

In vitro authentic SARS-CoV-2 neutralization assay was performed using Vero-E6 as previously reported ( Chi etਊl., 2020 ). Briefly, 1 × 10 4 /well Vero-E6 cells were seeded in 96-well plates. After culture for 24 hours, the 1:4 serially diluted antibodies (maximum concentration, 5 μg/ml) were mixed with 0.1 MOI (multiplicity of infection) authentic SARS-CoV-2 virus and incubated at 37ଌ for 30 minutes. This mixture was subsequently added into the cultured Vero-E6 cells. The supernatants were collected after further culture for two days for quantitative reverse transcription PCR and the cells were analyzed by immunofluorescence.

For immunofluorescence, the cells were fixed in 4% paraformaldehyde in PBS for 20 minutes, washed with PBS and permeabilized with 0.1% Triton X-100 in PBS at room temperature. After blocking with 3% BSA, the cells were incubated with anti-N polyclonal antibody ( Gu etਊl., 2020 ) at a dilution of 1:1000 overnight at 4ଌ and visualized with donkey anti-mouse IgG Alexa Fluor 488 (Thermo Fisher Scientific). Nuclei were stained with DAPI. Cells were imaged using an Eclipse Ti-S inverted fluorescence microscopy (Nikon).

For quantitative reverse transcription PCR, the viral RNA was extracted from the collected supernatant using Trizol LS (Thermo Fisher Scientific) and used as templates for quantitative PCR analysis by One-Step PrimeScrip RT-PCR Kit (Takara) following the manufacturer’s instructions. The primers and probe used were listed in Table S2. The PCR amplicon by SARS-CoV-2-N-F and SARS-CoV-2-N-R primers was inserted into pUC57 plasmid for standard curve generation. The program of the quantitative reverse transcription PCR was performed using the Mastercycler ep realplex Real-time PCR System (Eppendorf) as followed: 95ଌ 5 minutes 40 cycles of 95ଌ 10 s, 50ଌ 30 s, 72ଌ 30 s.

In vitro assay to detect antibody-dependent viral entry

In vitro SARS-CoV-2 pseudovirus ADE assays was performed using Raji cells as previously reported ( Jaume etਊl., 2011 ). Briefly, 3 × 10 4 Raji cells were seeded in each well of 96-well plates coated with 0.01% poly-L-lysine in PBS and cultured for 24 hours. The antibodies were serially diluted 1:2 (maximum concentration, 100 μg/ml) in RPMI 1640 for nine dilutions in total, and were incubated with the SARS-CoV-2 pseudovirus for 30 minutes. The mixture was applied onto the Raji cells and cultured for 60 hours. The measurement of luciferase activity was performed as described above using a Firefly Luciferase Assay Kit (Promega). The absolute luciferase activity values from all the wells were normalized to the luciferase activity value obtained with 2 μg/ml of antibody XG043 and expressed as the fold change in luciferase activity. Two replicates of XG043 (2 μg/ml) were performed on each plate and the average luciferase activity value of these two replicates was considered as reference (100% relative luciferase activity, the dotted lines in Figures 6 A, 6D, and 6E Figure S6B). Since many factors (virus concentration, cell concentration, immunofluorescence reading, etc.) vary between different plates or different rounds of experiments, normalization is necessary for comparing the luciferase activity values from different plates. The reason for choosing XG043 as the reference is simply because that XG043 was the first identified to induce ADE in our studies.

For the experiment to block the antibody-dependent viral entry, different concentrations of anti-hCD32 (BD PharMingen) were incubated with the Raji cells for 30 minutes at 37ଌ. Then, the mixture of 2 μg/ml antibody XG005 and SARS-CoV-2 pseudovirus was added to the treated Raji cells. The plates were incubated at 37ଌ for 60 hours before the measurement of luciferase activities as described above.

For in vitro Raji cell-dependent ADE assays using authentic SARS-CoV-2 virus, cultured Raji cells were incubated with the mixture of authentic SARS-CoV-2 virus and monoclonal antibodies (final concentration 4 μg/ml), XG038, XG016 and XG005, respectively. After 6, 24 or 48 hours incubation, the Raji cells were collected for RNA extraction and quantitative reverse transcription PCR analysis. SARS-CoV-2 N-protein RNA copy numbers were calculated using a standard curve composed of seven prepared N-protein DNA samples with 10-fold serial dilutions.

Protein production

The codon optimized wild-type cDNA of SARS-CoV-2 receptor-binding domain (RBD) (amino acid 330�) together with an Avi tag (GLNDIFEAQKIEWHE) was synthesized (GENEWIZ), and cloned into pACgp67 vector with a C-terminal 8 × His tag for purification. The SARS-CoV-2 RBD was expressed using the Bac-to-Bac baculovirus system. Extracted bacmid DNA was then transfected into Sf9 cells using Cellfectin II Reagent (Invitrogen). The low-titer viruses were harvested and then amplified to generate high-titer virus stock. The supernatant containing the secreted RBD without glycosylation was harvested 72 hours after infection and the RBD protein was captured by Ni-NTA resin (GE Healthcare) and purified. SDS-PAGE analysis revealed over 95% purity of the purified recombinant protein.

For the site-directed mutagenesis and expression of RBD mutants, SARS-CoV-2 RBD fragment (residue 319-541) and its mutants were synthesized (GenScript), fused with the human IgG1 Fc fragment, and cloned into mammal expression vector pSecTag. The plasmid was transfected into HEK Expi293 cells and incubated at 37ଌ for four days. Supernatant was harvest for further purification by Protein G resin according to the manufacturer’s protocol.

Single cell sorting of RBD- or S-ECD-binding memory B cells

S-ECD protein (GenScript) expressed and purified from recombinant baculovirus-infected insect Sf9 cells was chemically biotinylated using EZ-Link Sulfo-NHS-LC-Biotin kit (Thermo Fisher Scientific) as manufacturer’s instructions. Avi-tagged RBD expressed in baculovirus-infected insect Sf9 cells and Avi-tagged S-ECD expressed in mammalian HEK293T cells (Kactus Biosystems) were biotinylated using BirA Biotin-Protein Ligase kit (Avidity). The excess of unbound biotin was removed by using Zeba Spin Desalting column (Thermo Fisher Scientific). For each sample, the bait protein-PE and bait protein-APC were prepared by incubating 3 μg of biotinylated RBD or 25 μg of biotinylated S-ECD proteins with streptavidin-PE (eBioscience) or streptavidin-APC (BD Biosciences), respectively.

Purification of B cells, two-fluorescent-dye labeling of bait protein-binding B cells and single cell sorting experiments were performed as previously described ( Escolano etਊl., 2019 Robbiani etਊl., 2017 Wang etਊl., 2020b ). Briefly, PBMCs thawed and washed with RPMI medium were incubated with CD19 MicroBeads (Miltenyi Biotec) for positive selection of B lymphocytes. Sequential incubation at 4ଌ with human Fc block (BD Biosciences), bait protein-PE/APC (10 μg/ml for RBD, 60 μg/ml for S-ECD), and anti-CD20-PECy7 (BD Biosciences) was performed, followed by the single-cell sorting of CD20-PECy7 + bait protein-PE + bait protein-APC + memory B cells into 96-well plates using a FACSAria II (BD Biosciences). The single-cell sorted B cells were stored at �ଌ.

Antibody cloning, sequencing and production

Antibody cloning from the sorted single cells and the production of monoclonal antibodies were done as previously reported ( Robbiani etਊl., 2017 Wang etਊl., 2020b ). The sequences of primers for the 1 st / 2 nd round of nested PCR were listed in Table S2. Amplified PCR products from each single cell were loaded onto 2% agarose gel for electrophoresis and purified for Sanger sequencing. All the sequencing result of heavy and kappa/lambda light chains were analyzed by IMGT/V-QUEST ( Brochet etਊl., 2008 ) and IgBlast ( Ye etਊl., 2013 ), and the V(D)J gene segment and CDR3 sequences of each antibody were determined. The selected antibodies were subjected to vector construction and antibody expression as previously described ( von Boehmer etਊl., 2016 ).

Clustering analysis

Relative luciferase activities measured in neutralization or ADE assays or both were used for unsupervised hierarchical clustering analysis with the statistical scripting language R, using log-transformed data, Euclidean correlation coefficients for a distance metric, and ward.D2 clustering. A heatmap and cluster dendrogram tree were created using the Pretty Heatmaps (pheatmap and hclust) R packages.

Quantification and statistical analysis

The detailed results of statistical analysis are shown in the Result and Figure Legends. The Shapiro-Wilk test and Fisher’s F test were employed to check for normality and homogeneity of variances, respectively, prior to performing the comparison. Student’s t test was performed for RBD ELISA ( Figureਁ A), while Wilcoxon Rank Sum test was used for other ELISAs ( Figures 1 B�) and comparisons of ADE AUC ( Figureਇ E) due to their non-normal distribution. In order to determine whether there is a statistically significant difference of the ADE AUC and IC50 values of Cluster-X, -Y, and -Z antibodies, the nonparametric test (Dunn’s Kruskal-Wallis multiple comparison) was performed ( Figures 7 B and 7C). Fisher’s exact test was performed to assess the statistical significance based on the exact distribution of the frequencies of RBD Group-IV antibodies in three antibody clusters ( Figureਇ D). Correlation was evaluated by Spearman’s rank correlation method (Figures S6E–S6G). The area under the ELISA curves (ELISA AUC) ( Figures 1 A� and ​ and3A�), 3 A�), the half-maximal neutralizing titer (NT50) for serum neutralization assays ( Figures 1 F𠄱H), the 50% inhibitory concentration (IC50) values calculated for antibody neutralization capacities ( Figures 4 A, 4F, and ​ and5C), 5 C), the area under the ADE curve (ADE AUC) ( Figureਆ C), and enhancing power values (Figure S6D) were calculated in PRISM software as previously reported ( Bardina etਊl., 2017 Robbiani etਊl., 2019 Wang etਊl., 2020b ).