15 confirmed COVID-19 patients of 2 weeks convalescence were obtained from Fifth Medical Center of Chinese PLA General Hospital, Beijing, China. The clinical information of these patients was listed in Supplementary Table 1. 20 confirmed COVID-19 patients of 6 months convalescence were obtained from General Hospital of Central Theater Command of PLA, Wuhan, China.

ELISA plates were coated with S1 protein at 0.01 mg/mL in coating buffer (0.1 mol/L Na${}_2$CO${}_3$: 0.1 mol/L NaHCO${}_3$ = 3:7) at 4 °C overnight. After standard washing and blocking, diluted sera (1:100) were applied to each well. After a 1 h incubation at 37 °C, plates were washed and incubated with 0.25 $\mu$g/mL goat anti-human Ig antibodies conjugated with HRP (SouthernBiotech, Birmingham, AL, USA) for 1 h at 37 °C. TMB was used as the substrate, and the reaction was ceased by 2 mol/L H${}_2$SO${}_4$. The absorbance at 450 nm was measured by a microplate reader.

The pseudovirus neutralization assays were performed using HEK293-hACE2 cell lines, which express human ACE2. 5 $\times$ 10${}^4$/well cells were added to the well and cultured at 37 °C for 8 h. Serum diluted at 1:150 with PBS was incubated with TCID50 of 5 $\times$ 10${}^4$ TU SARS-CoV-2 pseudovirus (purchased from Future Biotherapeutics, Suzhou, Jiangsu, China) at 37 °C for 1 h. Then the mixed solution of virus and serum was added to the cell and cultured for 18 h at 37 °C supplied with 5% CO${}_2$. The cells were harvest and washed with PBS and detected by flow cytometry.

B cells were isolated from fresh or previously frozen PBMCs by immunomagnetic positive selection according to the manufacturer’s protocol (EasySep™ Human CD19 Positive Selection Kit II, STEMCELL). Purified B cells were eluted and washed with PBS containing 2% (v/v) fetal bovine serum (FBS) and 1 mM/L EDTA.

According to the manufacturer’s instructions for sorted B cells, total RNA was extracted using the RaPure Total RNA Micro Kit (Magen, R4012-02, Guangzhou, Guangdong, China), then the cDNA was synthesized by 5’-RACE using SMARTer® RACE 5’/3’ Kit (Takara, 634859, Mountain View, CA, USA) and generated a complete cDNA copy with the additional specific sequence at the 5’ end (universal linker sequence).

The upstream primer was targeted on the universal linker sequence from 5’-RACE, and the downstream primer targeted constant-region for IGHG, IGHA, IGHM, IGHD, IGK, and IGL genes in both of first-round and second-round PCR. Especially, barcodes were added to the second-round PCR primers that is convenient to distinguish BCRs from different individuals. PCR program for both rounds were: 5 cycles at 94 °C for 30 s, 72 °C for 3 min, 5 cycles at 94 °C for 30 s, 70 °C for 30 s, 72 °C for 3 min, and 25 cycles at 94 °C for 30 s, 68 °C for 30 s, 72 °C for 3 min (first-round PCR) and 5 cycles at 94 °C for 30 s, 72 °C for 90 s, 5 cycles at 94 °C for 30 s, 70 °C for 30 s, 72 °C for 90 s, and 25 cycles at 94 °C for 30 s, 68 °C for 30 s, 72 °C for 90 s (second-round PCR). The amplified DNA products were recovered from the agarose gel using a DNA Recovery Kit and sent to Novogene company for sequencing.

In addition, PCR was also performed to detect the expression of GAPDH. PCR condition for both cycles were: 95 °C for 5 min, 30 cycles at 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 30 s, followed by a final extension at 72 °C for 7 min. The PCR products were separated on 1.5% agarose gel by electrophoresis. The primers used for PCR are listed in Supplementary Tables 3,4,5.

We measured the concentration of PCR products with barcode and mixed the same amount of PCR products from COVID-19 patients (products of COVID-19 patients of 2 weeks convalescence were mixed as one sample while products of COVID-19 patients of 6 months convalescence were mixed as another one sample). The mixed PCR products were separated with 2% agarose gel. Gel extraction was performed and products were used for Pacbio sequencing. The obtained data was split according to barcodes by using the FASTX-toolkit (version 0.0.13, LongIsland, NY, USA). IMGT/High V-QUEST (version 1.7.1, Montpellier, France) was used for sequence annotation to determine the V(D)J genes and sequence of the variable region.

All data were analyzed by normality and lognormality tests to identify whether the data belong to a normal distribution, which was decided by the p-value of the Shapiro-Wilk test. Unpaired t-test was used in the condition of normal distribution, or Mann-Whitney test was used in the non-normal distribution (* p 0.05, ** p 0.01, *** p 0.005, **** p 0.0001). These are all executed in GraphPad Prism.

COVID-19 patients at 2 weeks of convalescence were discharged from Fifth Medical Center of Chinese PLA General Hospital from February 13 to March 3, 2020 were used in this study. Eligible participants consisted of 7 female and 7 male patients who ranged from age 39 to 74, with a median of 64 years of age. All COVID-19 patients were confirmed to be SARS-CoV-2 infected by respiratory RT-PCR tests. According to the guidelines for diagnosis and management of COVID-19 released by National Health Commission of China, discharge criteria were listed as follows: (1) Normal body temperature for more than three days; (2) Significant improvement in respiratory symptoms; (3) Pulmonary imaging shows a substantial reduction in pneumonia; (4) Two consecutively negative SARS-CoV-2 RT-PCR assays of respiratory tract specimens with a one-day sampling interval. A peripheral blood sample was collected in a BD Vacutainer and the plasma was separated by centrifugation at 2000 rpm for 10 min (Supplementary Table 1).

COVID-19 patients with 6 months convalescence were discharged from General Hospital of Central Theater Command of PLA, Wuhan, China.

The serum samples of COVID-19 patients 2 weeks, and 6 months, after recovery were obtained, and the level of specific IgM, IgG and IgA against the spike protein of SARS-CoV-2 were measured by ELISA. Serum from 27 healthy subjects were used as negative controls. The results showed a higher level of specific IgM and IgA against SARS-CoV-2 spike protein in recovered COVID-19 patients 2 weeks when compared to patients at the 6 month mark. However, the IgG level was elevated in COVID-19 patients 6 months after recovery, while no differences were observed in viral-specific IgM and IgA levels between COVID-19 patients 6 months after recovery and healthy controls (Fig. 1A). Subsequently, serum samples from 26 COVID-19 patients after 2 weeks of convalescence and 139 COVID-19 patients after 6 months of convalescence were used to analyze neutralization activity by blocking SARS-CoV-2 pseudovirus infection of HEK293 cells overexpressing human ACE2. The results showed that 27% of 2 weeks convalescent COVID-19 patients and 37.4% of the 6 months convalescent COVID-19 patients showed more than 25% blocking activity. Moreover, 20% of 6 months convalescent COVID-19 patients displayed greater than 50% blocking activity (Fig. 1B). Of note, approximately 42% of 6 months convalescent COVID-19 patients showed elevated binding of the pseudovirus with HEK293 cells overexpressing human ACE2; however, the reason for this observation is not clear. We also purified IgA and IgG from the 2-week convalescent serum and analyzed the neutralizing activity. Results showed that, 84.4% of IgG and 71.9% of IgA taken from these COVID-19 patients showed more than 25% blocking activity, and 65.6% of IgG and 50% of IgA in COVID-19 patients showed more than 50% blocking activity (Fig. 1C).

Fig. 1.

S-specific antibodies in the serum of COVID-19 patients 2-weeks and 6-months convalescence and the neutralizing activity. (A) The concentration of S-specific IgM, IgG and IgA in serum of COVID-19 2-weeks and 6-months convalescence by ELISA assay. Unpaired t-test, data are presented as mean $\pm$ SEM, *p 0.05, **p 0.01, ***p 0.001, ****p 0.0001. (B) The neutralizing activity of serum of COVID-19 patients 2-weeks and 6-months convalescence. Serum of healthy controls and COVID-19 patients 2-weeks and 6-months convalescence were pre-incubated with SARS-CoV-2 pseudovirus labeled with GFP and then incubated with HEK293T overexpressed human ACE2. FACS was used to detect the fluorescence signal of GFP. Unpaired t-test, data are presented as mean $\pm$ SEM, *p 0.05, **p 0.01, ***p 0.001, ****p 0.0001. (C) The neutralizing activity of IgG and IgA purified from serum of COVID-19 patients 2-weeks and 6-months convalescence. Unpaired t-test, data are presented as mean $\pm$ SEM, *p 0.05, **p 0.01, ***p 0.001, ****p 0.0001.

Viral infection can trigger virus-specific memory B-cell production, and this response can linger longer in the body post-infection. While different from naive B-cells, memory B-cells can express not only virus-specific IgM and IgD immunoglobulins, but also highly effective, high affinity virus-specific IgA, IgG or IgE antibodies [4, 20, 21].

We first collected the peripheral Blood Mononuclear Cell (PBMC) from 35 COVID-19 donors at different stages of recovery, including 15 donors 2 weeks after recovery, 10 donors with neutralizing antibodies 6 months after recovery, 10 donors without neutralizing antibodies 6 months after recovery, and 11 healthy donors were used as controls. CD19${}^{+}$ B-cells were sorted via immunomagnetic beads, and total RNA were extracted from these cells. Considering the diversity of BCR variable regions and a biased amplification problem caused by common use of multiple RT-PCR steps, we used a 5’-RACE reverse transcription kit to generate a complete cDNA copy with additional specific sequence at the 5’end of the amplicon. We designed the upstream primer to target 5’ terminal specific nucleotides sequences, and downstream primers to target BCR constant regions in 5 classes of BCR heavy chain, or the 2 types of BCR light chains. In addition, because the length of 5’-RACE products exceeded limits of some next-generation sequencing platforms, we used the PacBio sequencing technology which can sequence the full length of the BCR variable region (Fig. 2A).

Fig. 2.

Study design and the schematic diagram of 5’RACE. (A) Schematics of the experimental design for Ig V(D)J sequencing. PBMC were obtained from healthy people and COVID-19 patients and CD19${}^{+}$ B cells were sorted by immunomagnetic positive selection and total RNA was extracted for Ig variable region amplification. SMARTer IIA Oligonucleotide was random add to the 5’ end of RNA. Then universal upstream primer to target 5’ end specific nucleotides sequences and the downstream primers to target constant region of Ig. After two round of PCR, complete variable region was amplified and sequenced by pacbio sequencing. (B) VDJ family gene used in healthy control and COVID-19 patients by bubble chart. X, Y, Z axis represented for VH, JH and DH family gene separately, and the size of bubble represented for the proportion of the VHDHJH gene. (C) Frequency of the VHDHJH gene in healthy people (blue) and COVID-19 patients (red).

We first amplified and obtained the V(D)J transcripts of four classes of IgH (except IgE), and VJ transcripts of two types of IgL (Ig$\kappa$ and Ig$\lambda$) (Supplementary Fig. 1). Using the PacBio sequencing platform and IMGT/HighV-QUEST (version 1.7.1), a website that analyzes thousands of immunoglobulin gene nucleotide sequences, we obtained a total of 157,245 BCR transcript reads from the 35 patients we studied. As expected, sufficient diversity was observed in the BCR transcripts that were obtained from each individual’s B-cells, and detailed data are shown in Supplementary Table 2.

Next, we analyzed whether the relative abundance and inter-status variability of the V, D, J gene family was different between patients and healthy people (Fig. 2B,C). Our results showed that although there is no significant difference in the overall landscape and inter-status variability of each V(D)J gene family, some dominant recombination of V, D and J families are significantly different between patients and healthy donors. For example, the VH3DH3JH4 recombination pattern was dominant in both COVID-19 patients and healthy subjects. In contrast, some increased VDJ recombination patterns were more frequently observed in COVID-19 patients than in healthy controls, such as the increased VH4DH6JH6 and VH1DH6JH6 patterns usually found in IgG from COVID-19 patients, but not in healthy controls. Increased VH7DH6JH4 and VH4DH3JH2 patterns were usually observed in IgA from COVID-19 patients, and increased VH7DH3JH6 was observed in IgD from COVID-19 patients.

Antibodies commonly display large levels of diversity within the same individual or among different individuals. Theoretically, there are few convergent clonotype antibodies in an individual or between different individuals under physiological conditions. However, after an antigenic challenge, the convergent clonotypes antibodies or memory B-cells can be increased. In this study, we first analyzed the clonal proliferation of B-cells from each individual, as a whole, by analysis of the percentage of Top 10 CDR3 sequences (Fig. 3A). We found that B-cells expressing different types of Ig showed a certain degree of clonal bias within each individual patient, an even in healthy individuals, because they were previously infected by pathogens other than the SARS-CoV-19 virus. As expected, higher clonal bias of IgH chain and Ig$\lambda$ were found in COVID-19 convalescent patients, especially in patients 2 weeks after recovery, while Ig$\kappa$ showed no difference between healthy subjects and COVID-19 patients. We noted that IgG displayed an even higher clonal bias in COVID-19 patients at 6 months, rather than 2 weeks, after recovery, and that this finding is quite different from other IgH chains.

Fig. 3.

Clonal expansion and common CDR3 shared among the COVID-19 patients. (A) The top 10 CDR3 ration in COVID-19 patients compared with healthy people. COVID-19 6m-A represents for patients with neutralizing antibodies and COVID-19 6m-B represents for patients without neutralizing antibodies. Unpaired t-test, data are presented as mean $\pm$ SEM, *p 0.05, **p 0.01, ***p 0.001, ****p 0.0001. (B) COVID-19 patients specific CDR3 sequences of Ig heavy chain that are shared between every individual in COVID-19 patients 2 weeks and 6 months after recovery. Red cell represents for the CDR3 exists in this individual while blue cells represent for the CDR3 does not exists in this individual. (C) COVID-19 patients specific CDR3 sequences of Ig light chain that are shared between every individual in COVID-19 patients 2 weeks and 6 months after recovery.

In addition, many convergent clonotypes were frequently observed among the COVID-19 patients. Interestingly, we found a time-shift appearance of convergent clonotype pattern between different Ig classes in COVID-19 patients after 2 weeks versus 6 months of recovery (Fig. 3B,C). Shown is the shared convergent clonotypes that were significantly increased in IgD, IgA, IgG and Ig$\kappa$ in COVID-19 patients 6 months rather than 2 weeks after recovery. Of note, IgM showed an increased proportion of shared convergent clonotypes in COVID-19 patients 2 weeks after recovery when compared to the 6 month convalescence group. These results suggest that at least 2 weeks after COVID-19 convalescence, expansion of B-cell clones is dominated by the expression of IgM, which is converted to IgG and IgA at the 6 month time point. In contrast with COVID-19 patients, healthy subjects do not share CDR3 sequence among individuals.

In this study we analyzed the ratio of top 5 V(D)J recombination patterns in COVID-19 patients and healthy subjects (Fig. 4A), and found that the most frequently used V(D)J recombination pattern in IGH genes were different among individuals, while Ig$\lambda$ showed conserved recombination pattern in both healthy subjects and COVID-19 patients. The percentage of the IgG-biased V(D)J recombination pattern displayed increases in COVID-19 patients 6 month after recovery when compared to healthy subjects, while Ig$\lambda$ showed a decreased ratio of biased VJ recombination patterns in COVID-19 patients versus healthy subjects.

Fig. 4.

V(D)J recombination pattern in healthy subjects and COVID-19 patients. (A) The top 5 V(D)J rearrangement patterns that are used by B cells of healthy people and COVID-19 patients (left) and the percentage of top 5 V(D)J rearrangement in healthy people and COVID-19 patients. Unpaired t-test, data are presented as mean $\pm$ SEM, *p 0.05, **p 0.01, ***p 0.001, ****p 0.0001.

We found that, among the individuals, significant diversity in recombination patterns within IgM, IgA, IgG, IgD as well as Ig$\kappa$, but not in Ig$\lambda$ (Fig. 5A). This finding indicates that, under physiological conditions, the variable region of Ig$\lambda$ is conserved in different individuals. In contrast, convergent recombination patterns were frequently observed among the COVID-19 patients. Interestingly, we found a time shift change of appearing convergent recombination patterns between different Ig classes in COVID-19 patients after recovery of 2 weeks to 6 months. These shared convergent recombination pattern were significantly increased in IgM and Ig$\kappa$, and moderately increased in IgA in COVID-19 patients after a 2-week recovery. However, when 6-month recovery COVID-19 patients were analyzed, increased proportions of shared convergent recombination pattern were converted into IgD, IgG and IgA class immunoglobulins. These results suggest that at least 2 weeks into COVID-19 convalescence, the expansion of the B-cell clone is dominated by the expression of IgM and IgA, but not IgG.

Fig. 5.

Shared V(D)J recombination pattern in healthy subjects and COVID-19 patients. (A) Common V(D)J recombination pattern between individuals of healthy subjects and COVID-19 patients 2 weeks and 6 months after recovery. (B) COVID-19 patients specific V(D)J recombination pattern that are shared between every individual in COVID-19 patients 2 weeks after recovery. Heats maps show the expression frequency of Ig heavy chain and light chain. (C) The percentage of the most shared V(D)J recombination pattern in every patient by column charts.

We next analyzed and compared the shared V(D)J patterns between the different classes of Ig, which exist only among COVID-19 patients after 2 weeks of recovery (Fig. 5B,C). We found that the IgD class frequently utilizes IGHV3-74\IGHD6-13\IGHJ4 (5/15), whereas IgA tend to use IGHV4-39\IGHD3-22\IGHJ2 (9/15). IgG were likely to use IGHV1-8\IGHD3-22\IGHJ4 of (6/15), and Ig$\lambda$ usually use IGLV3-21\IGLJ6 (8/15). Note that the IGHV1-18\IGHD3-10\IGHJ6 pattern was shared by IgM class immunoglobulins in all COVID-19 patients (15/15), and the IGKV2D-29/IGKJ1 of Ig$\kappa$ was shared by 14 of 15 COVID-19 patients. In addition, we also analyzed for differences in the convergent clonotypes between moderate and severe individuals, and found no significant difference between the recovered moderate and severe individuals (data not shown).

According to the classical clonal selection theory, different types of Ig produced by one B-cell all present the same V(D)J combination and sequence. In other words, different types of Ig detected from the same B-cell cluster should present the same V(D)J combination pattern. However, our previous finding, revealed by single cell sequencing, was that one B-cell can express more than one classes of Ig and, moreover, different classes of Ig express their own unique variable region sequence, which is somewhat at odds with the classical concept of Ig class switching [19]. In this study, we further addressed if the same population of B-cell-derived Ig classes display different V(D)J recombination patterns (Fig. 6A). Results showed that approximately 80% of V(D)J recombination patterns were uniquely expressed in one Ig heavy chain within the same healthy individual (IgM: 77.39 $\pm$ 11.79; IgD: 78.25 $\pm$ 11.21; IgA: 78.16 $\pm$ 11.87; IgG: 78.37 $\pm$ 11.45). In contrast, the percentage of uniquely expressed V(D)J recombination patterns were decreased in COVID-19 patients, especially in COVID-19 patients 2 weeks after recovery (COVID-19 2 weeks: IgD: 34.54 $\pm$ 10.56; IgA: 43.57 $\pm$ 7.521; IgG: 35.57 $\pm$ 10.84; COVID-19 6 months: IgD: 60.55 $\pm$ 8.2; IgA: 54.01 $\pm$ 8.991; IgG: 51.66 $\pm$ 8.883). It should be noted that the percentage of unique V(D)J recombination patterns in IgM were similar in COVID-19 patients 2 weeks after recovery and healthy subjects (2weeks: 74.14 $\pm$ 9.214) (Fig. 6A). The percentage of uniquely expressed CDR3 of each Ig in the same individual was significantly decreased in COVID-19 patients especially in the 2 week convalescent patient (2 weeks: IgM: 92.93 $\pm$ 3.037; IgD: 72.93 $\pm$ 13.64; IgA: 82.75 $\pm$ 8.926; IgG: 74.4 $\pm$ 8.53; 6 months: IgM: 98.25 $\pm$ 0.9742; IgD: 97.98 $\pm$ 0.7597; IgA: 96.49 $\pm$ 1.441; IgG: 96.3 $\pm$ 1.48; healthy controls: IgM: 1.0; IgD: 1.0; IgA: 1.0; IgG: 1.0) (Fig. 6B).

Fig. 6.

VDJ recombination pattern and CDR3 sequences of different Ig classes in the same individual of healthy subjects and COVID-19 patients. (A) The percentage of VDJ recombination pattern that exist only in one type of Ig in the same individual. Unpaired t-test, data are presented as mean $\pm$ SEM, *p 0.05, **p 0.01, ***p 0.001, ****p 0.0001. (B) The percentage of CDR3 sequence that exist only in one type of Ig in the same individual. Unpaired t-test, data are presented as mean $\pm$ SEM, *p 0.05, **p 0.01, ***p 0.001, ****p 0.0001. (C) Common CDR3 sequences shared between Ig heavy chains in COVID-19 patients compared with healthy controls. Unpaired t-test, data are presented as mean $\pm$ SEM, *p 0.05, **p 0.01, ***p 0.001, ****p 0.0001.

We further analyzed if some convergent clonotypes were shared by different classes of Ig in the same individual (Fig. 6C). As expected, unlike the healthy donors that do not share convergent clonotypes among the different classes of Ig in an individual, we found that the same clonotype patterns frequently shared different classes of Ig in COVID-19 patients. For example, IgM showed increased frequency of the same CDR3 sequence with IgD, IgG and IgA in 2 weeks of convalescence COVID-19 patients, while IgD showed increased frequency of the same CDR3 sequence with IgG and IgA in COVID-19 patients with 6 months convalescence.

In theory, memory B-cells at 2 weeks and 6 months of convalescence may exhibit the same or similar V(D)J or VJ rearrangement patterns. We first excluded the V(D)J and VJ rearrangements shared with healthy individuals, then compared the VDJ and VJ rearrangements of different IgH and Ig$\lambda$ between COVID-19 patients with 2 weeks versus 6 months of convalescence (Fig. 7A,B). We found that identical V(D)J patterns shared by different COVID-19 patients at 2 weeks and 6 months of convalescence were easy to identify in all types of Ig heavy and light chains, while shared CDR3 exist only in IgG and Ig light chains. In detail, IGHV1-18\IGHD3-10\IGHJ6 of IgM (26/35), IGHV3-23\IGHD6-19\IGHJ4 of IgD (20/35), IGHV3-74\IGHD1-26\IGHJ4 of IgA (19/35), IGHV4-39\IGHD2-2\IGHJ6 of IgG (21/35) was frequently shared by COVID-19 patients at 2 weeks and 6 months of convalescence. Of note, IGKV2D-29\IGKJ1 of Ig$\kappa$ was shared by 28 of 35 COVID-19 patients.

Fig. 7.

Common V(D)J recombination pattern and CDR3 sequences that are shared by COVID-19 2 weeks and 6 months after recovery. (A) Common V(D)J recombination pattern exist only in COVID-19 patients and shared by COVID-19 2 weeks and 6 months after recovery. (B) Common CDR3 sequences exist only in COVID-19 patients and shared by COVID-19 2 weeks and 6 months after recovery. (C) Common V(D)J recombination pattern shared by COVID-19 2 weeks and 6 months after recovery with neutralizing antibodies and do not exist in healthy subjects and COVID-19 6 months after recovery without neutralizing antibodies. (D) Common CDR3 sequences shared by COVID-19 2 weeks and 6 months after recovery with neutralizing antibodies and do not exist in healthy subjects and COVID-19 6 months after recovery without neutralizing antibodies.

This work also identified numerous V(D)J and CDR3 sequences that are shared by COVID-19 patients convalescent for 2 weeks and 6 months who demonstrate neutralizing antibodies but not in COVID-19 patients 6 months convalescent without neutralizing antibodies nor in healthy subjects (Fig. 7C,D). These sequences may contribute to screening neutralizing antibodies.

We first prepared an affinity chromatography column using the spike (S) protein of SARS-CoV-2. IgG was purified by protein G chromatography from serum taken from 15 2-weeks convalescent COVID-19 patients. Following this, the IgG from each individual was purified by S protein affinity chromatography. Next, we identified IgG which can block the binding between pseudovirus and ACE2 overexpressing HEK293 cells, and the peptide profile of IgG with viral blocking activity was analyzed by protein mass spectrometry (Fig. 8). Subsequently, peptide sequence of IgG was mapped to the V(D)J sequence of recovered COVID-19 patients obtained in this study. Significantly, we found that some IgG variable domain peptides could map with the V(D)J sequence of recovered COVID-19 patients (Fig. 8), which suggested that the BCR repertoire library obtained in this study can be used to screen for effective neutralizing antibodies against SAR-CoV-2 or to prepare a phage antibody library following SARS-CoV-2 infection.

Fig. 8.

Scheme to verify the presence of neutralizing antibody sequences in the BCR repertoire obtain in this study. Purified IgG from the serum of COVID-19 patients by protein G and identified the neutralizing IgG by pseudovirus blocking experiment and then the peptide information was obtained by LC-Mass/Mass. After mapping to BCR repertoire, we obtained the amino acid sequence of neutralizing antibodies.