SARS-CoV-2 Cysteine-like Protease Antibodies Can Be Detected in Serum and Saliva of COVID-19–Seropositive Individuals

The identification of the link between a novel β-coronavirus strain, named severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), and a fatal respiratory illness, COVID-19, formally recognized as a pandemic by the World Health Organization on March 11 (1, 2) has led to a rush by health systems all over the world to develop and implement testing for viral infection. The rapid cloning and sequencing of the viral genome permitted the development of PCR-based assays for the detection of viral nucleic acids that have become a key strategy for both clinical diagnosis and epidemiological monitoring studies. However, besides identifying individuals with active infection, it is also necessary to know which patients, either symptomatic or asymptomatic, have developed an Ab response to the virus. Several reasons make SARS-CoV-2 serology tests crucial. First, PCR testing is not 100% efficient (3–5). Second, testing for viral RNA cannot detect evidence of past infection, which will be crucial for epidemiological efforts to assess how many people have been infected in any given area. In addition, this will allow definition of the infection fatality rate and help with management of the epidemic. Third, assays to measure Ab responses and determine seroconversion, although not appropriate to detect acute infections, are, however, valuable sources of information on the quality of the response exerted by different individuals developing different clinical manifestations. Moreover, if different isotypes and viral Ags are included in assays testing different time points after the onset of the disease, information of clinical importance will be produced. Finally, quantitative and qualitative assays of Ab responses can aid in the identification of factors that correlate with effective immunity to SARS-CoV-2, the duration of these immune responses, and may also aid in the selection of donors from whom preparations of convalescent serum/plasma can be generated for therapeutic use.

Multiple Ab tests to detect exposure to SARS-CoV-2 are becoming available. The majority of these assays have been optimized to detect IgG and, in some cases, IgM Abs using different viral Ags, with the Spike (S) protein and the nucleocapsid protein (NP) of SARS-CoV-2 being the more widely used (6, 7). These proteins are key elements of the viral particle and are expected, by analogy with other coronaviruses, to be highly immunogenic. However, the immunogenicity of other viral proteins (28 are encoded in the viral genome) has been little explored. In this article, we share our study of the Ab response to the main viral protease (Mpro, or 3CLPro) elicited after viral infection. Although this protein is not exposed in the viral particle, Mpro carries out a critical role in viral replication. Like other β-coronaviruses, SARS-CoV-2 is a positive-sense RNA virus that expresses multiple proteins as a single polypeptide chain, and Mpro cleaves the 1ab polyprotein to release mature proteins for the virus. Because this activity is essential for the viral life cycle, Mpro structure and function have been studied intensively (8); in particular, Mpro has been suggested as a target for specific inhibitors that might act as potent antiviral agents (9). However, to our knowledge, no study on the antigenicity of this protease has been reported.

To increase the possibilities of diagnosing COVID-19 patients, in this article, we report the use of an ELISA test involving the assay of seroreactivity to three different SARS-CoV-2 Ags, including the protease Mpro. These data demonstrate that individuals who have been infected with SARS-CoV-2 make high titer Ab responses to Mpro and that assays for seroreactivity to this protein sensitively and specifically discriminate between infected and noninfected individuals. Further, whereas most available tests assess for SARS-CoV-2–specific IgM and IgG Abs, in this study, we also explored the presence of IgA Abs in the sera tested. Whereas, in general, assays for IgM Abs resulted in a high background that limited the sensitivity of the ELISA, testing for IgA seropositivity provided very clean data, with low background and high signal, therefore providing a very good tool to complement IgG assays.

Interestingly, considerable significant amounts of IgA Abs specific for Mpro, as well as the receptor binding domain (RBD) and NP, were also frequently found in serum of COVID-19–infected individuals, and the amounts of IgA and IgM Abs could be related with disease severity.

Surprisingly, IgG Abs specific for SARS-CoV-2 Ags were also readily detectable in the saliva of these patients and, in this case, the titer of protease-specific Abs was higher than for the other two proteins tested. Because the nasal and buccal mucosa are key sites of viral infection and replication, the presence of Abs in saliva may be an important feature of the virus-specific immune response, but this observation may also allow the development of a rapid, completely noninvasive assay for COVID-19 seropositivity.

A gene encoding SARS-CoV-2 Mpro from the Wuhan-Hu-1 strain (ORF1ab polyprotein residues 3264–3569, GenBank code: MN908947.3) was amplified by PCR using the oligos 5′-GACCCATGGCTTCAGCTGTTTTTCAGAGTGGTTT-3′ and 5′-GACCTCGAGTTGGAAAGTAACACCTGAGCATT-3′, digested with NcoI and XhoI and ligated into the vector pET22b (Novagen) linearized with the same restriction enzymes.

Oligonucleotides 5′-GATCCATGGCTTCTGATAATGGTCCGCAAAATCAGCGTAATGCA-3′ and 5′-CAGGTCGACAGGCTCTGTTGGTGGGAATG-3′ were used to amplify the NP of SARS-CoV-2. The amplification product was then digested with NcoI and SalI and ligated into the pET26b vector (Novagen) digested with NcoI and XhoI.

The integrity of all constructs was verified by sequencing at Eurofins Genomics.

Recombinant viral proteins were expressed in the Escherichia coli strain BL21 Star (DE3) pLysS (Thermo Fisher Scientific).

SARS-CoV-2 Mpro protein was expressed by transforming this plasmid into the E. coli strain BL21 Star (DE3) pLysS. Transformed clones were precultured overnight at room temperature in 50 ml of 1 × Luria–Bertani medium with ampicillin (150 μg/ml) and chloramphenicol (34 μg/ml). The overnight culture was then inoculated into 1 l of 1 × Luria–Bertani medium (150 μg/ml ampicillin and 34 μg/ml chloramphenicol), and the culture was grown at 37°C with agitation until the OD₆₀₀ reached 0.6 when isopropyl-d-thiogalactoside was added to 1 mM to induce overexpression of the Mpro gene. The same protocol was followed to produce the NPs except that kanamycin (150 μg/ml) was used instead of ampicillin for antibiotic-mediated selection.

After overnight culture at 22°C for NP, 3 h at 37°C for Mpro, bacteria were harvested by centrifugation at 9500 × g, 4°C for 15 min, and the pellets were washed by resuspension in 150 ml of TES buffer (20 mM Tris pH 8, 2 mM EDTA, 150 mM NaCl) and recentrifugation. Washed pellets were either processed immediately or stored frozen for later use.

Fresh, or thawed, cell pellets were resuspended in ice-cold 50 mM NaH₂PO₄ buffer (pH 8), 500 mM NaCl, 10 mM imidazole (I2399; Sigma-Aldrich), 0.1% Sarkosyl, and 5% glycerol (pH 8.0). Lysozyme was then added (to 0.25 mg/ml), as were PMSF, Leupeptin, and Pepstatin A (all to a final concentration of 1 mM) and DNase I (2 μg/ml). Bacteria were lysed by sonication (three cycles of 30 s with 30 s rest on ice between pulses), and soluble proteins were separated by centrifugation of the lysed cells at 14,000 × g at 4°C for 45 min.

The 6-histidine tagged proteins were purified from the lysate using Nickel Affinity Cartridges, 5 ml (Agarose Bead Technologies S.L.). The bacterial supernatant was loaded on the column at a flow rate of 1 ml/min, followed by washing with five column volumes of 50 mM NaH₂PO₄ buffer, 500 mM NaCl, and 10 mM imidazole and then five column volumes of 50 mM NaH₂PO₄ buffer, 500 mM NaCl, and 25 mM imidazole. Recombinant proteins were eluted using a linear gradient of imidazole ranging from 25 to 250 mM over five column volumes (a representative SDS-PAGE analysis of the eluted fractions is shown in Supplemental Fig. 1A). The proteins were then further purified by gel filtration using a 10/30 Superdex 75 Increase column (Cytiva) pre-equilibrated in 20 mM HEPES, 1 mM EDTA, and 300 mM NaCl (pH 7.5). The gel-filtration analysis indicated that the SARS-CoV-2 Mpro protein purified as a dimer.

The cDNA region coding for the RBD (residues 334–528) defined in the structure of the S protein (Protein Data Bank Identification 6VSB) was amplified for expression in mammalian cells. The fragment was cloned in frame with the IgK leader sequence, an HA-tag (YPYDVPDYA) and a thrombin recognition site (LVPRGS) at its 5′ end, and it was followed by a second thrombin site, the TIM-1 mucin domain and the human IgG1 Fc region at the 3′ end. The recombinant cDNA was cloned in a vector derived from the pEF-BOS (10) for transient expression in HEK293 cells and in the pBJ5-GS vector for stable protein production in CHO cells following the glutamine synthetase system (11). The inclusion of the TIM-1 mucin domain enhanced protein expression.

Mammalian RBD (mRBD) fused to the mucin domain and the Fc region (mRBD-mucin-Fc) was initially purified from cell supernatants by affinity chromatography using an IgSelect column (GE Healthcare). The mucin-Fc portion and the HA-tag were released from the mRBD protein by overnight treatment with thrombin at room temperature. The mixture was run through a protein A column to remove the mucin-Fc protein, and mRBD was further purified by size-exclusion chromatography with a Superdex 75 column in HBS buffer (25 mM HEPES and 150 mM NaCl, pH 7.5). The concentration of purified mRBD was determined by absorbance at 280 nm.

A recombinant baculovirus expressing the RBD domain was generated using a pFastBac Dual-derived plasmid harboring the RBD coding sequence kindly provided by Dr. F. Krammer (6). HighFive (Thermo Fisher Scientific) cell cultures were infected with the recombinant virus at a multiplicity of infection of 3 PFU per cell and maintained in TC-100 medium (Thermo Fisher Scientific) for 72 h. Thereafter, cell medium was harvested and clarified by centrifugation (4300 × g for 10 min) and filtration through a 0.45-μm filter. Supernatant was loaded onto a chelated Nickel Affinity Cartridge, 5 ml (Agarose Bead Technologies S.L.), at a flow rate of 1.5 ml/min and eluted with a linear gradient of 500 mM Imidazole in Tris-saline buffer (pH 7.5). Fractions were analyzed by SDS-PAGE, and those containing RBD were pooled together and concentrated using an Amicon Ultra-15 Centrifugal unit with a 10-kDa cutoff membrane (Millipore). The concentrated protein was loaded onto a Superdex 75 10/300 Increase gel-filtration (GE Healthcare) equilibrated with PBS. The peak fractions were analyzed by SDS-PAGE and pooled together for further analysis.

For Ag coating, 100 μl/well of recombinant proteins diluted in 0.1 M borate-buffered saline (pH 8.8), at concentrations of 0.5 μg/ml (NP and MPro) and 1 μg/ml (RBD), was pipetted into wells of a 96-well Maxisorp Nunc-Immuno plate that was then incubated overnight at 4°C. Coating solutions were then aspirated, and the ELISA plates were washed three times with 200 μl of PBS 0.05% Tween 20 (PBS-T) and then dried before blocking with PBS-casein (1× PBS blocker; Bio-Rad) for 1 h at room temperature. The plates were washed again with PBS-T, and 100 μl of patient serum/plasma sample diluted in PBS-casein, 0.02% Tween 20 (as indicated) was added and incubated for 2 h at room temperature. The plates were washed again, and 100 μl/well of the indicated detection Ab (AffiniPure Rabbit Anti-Human IgM, Fcμ fragment specific; AffiniPure Rabbit Anti-Human Serum IgA, α-chain specific; AffiniPure Rabbit Anti-Human IgG, Fcγ fragment specific from Jackson Labs or anti-human [Fab]’2 HRPO-labeled Ab from Thermo Fisher Scientific) was added and incubated for 1 h at room temperature. The plates were washed with PBS-T four times and incubated at room temperature in the dark with 100 μl/well of Substrate Solution (OPD, Sigma-Aldrich; prepared according to the manufacturer’s instructions) (typically for 3 min). Fifty microliters of stop solution (3 M H₂SO₄) was then added to each well, and the OD (at 492 nm) of each well was determined using a microplate reader.

Negative controls included wells coated just with blocking buffer and serum samples collected from donors before 2019.

Graphics and statistical analysis were performed with Graph Pad Prism 8 Software (GraphPad Software, www.graphpad.com) and Stata 14.0 for Windows (Stata Corp LP, College Station, TX). Quantitative variables following a nonnormal distribution were represented as median and interquartile range and the Mann–Whitney U test was used to test for statistically significant differences. Variables with a normal distribution were described by mean ± SD, and differences between groups were assessed with Student t test. Qualitative variables were described as counts and proportions, and χ² or Fisher exact test was used for comparisons. Correlation between quantitative variables was analyzed using the Pearson correlation test.

Severity of COVID-19 was established as previously described (12). In this case, to determine differences in titers of Abs between groups of severity, the Cuzick test (which assesses trends across ordered groups) was employed.

Because several variables might contribute to differences in ELISA titers, we used multivariable linear analysis using generalized linear models (glm command of Stata) in which the dependent variable was ELISA titers of each isotype against each protein. The first model included age, gender, and time from symptoms onset, followed by backward stepwise approach removing all variables with a p value >0.15 to obtain the best model for each protein and isotype. Then, the variable of interest (severity, anosmia, or IL-6 serum levels) was forced in the model.

To determine the capacity of the different ELISA to discriminate between pre–COVID-19 sera and those sera obtained from patients with SARS-CoV-2, as determined by positive PCR from nasopharyngeal exudates, receiver operating characteristic (ROC) analysis was performed, using the roctab command of Stata 14.1 (College Station). Each cutoff point was selected based on the best trade-off values between sensitivity, specificity, and the percentage of patients correctly classified. ROC curves and area under the curve (AUC) were also obtained.

This study used samples from the research project “Immune response dynamics as predictor of COVID-19 disease evolution. Implications for therapeutic decision-making” approved by La Princesa Health Research Institute Research Ethics Committee (register no. 4070). Some experiments included patients from “Study of the lymphocytic response against SARS-COV-2, in different situations of host health and COVID-19 severity (InmunoCOVID)” approved by the Hospital La Paz Committee (PI-4101). All experiments were carried out following the ethical principles established in the Declaration of Helsinki. All included patients (or their representatives) were informed about the study and gave written informed consent.

Thirty-six COVID-19 patients, diagnosed by PCR, were recruited for the study. Nine of them presented active infection by SARS-CoV-2 at the moment of the study, whereas the rest had no detectable levels of the virus. Ten patients required hospitalization, of which six were admitted to the ICU (Table I). Thirty-three serum samples from patients presenting a monoclonal gammopathy, allergic disease, or rheumatoid arthritis, collected before June 2019 (pre–COVID-19), were used as negative controls. All samples were stored frozen before use.

Twelve donors with high Ab titers in serum were selected to measure specific IgG and IgA against SARS-CoV-2 in saliva. For this purpose, new saliva samples were collected from these patients (and also from 11 healthy donors), aliquoted, and immediately frozen. Prior to use, saliva samples were thawed, centrifuged at 400 × g, and diluted 1/2, 1/4, and 1/10 in 1× PBS with 1% casein (Bio-Rad) and 0.02% Tween 20 supplemented with Complete Protease Inhibitor Cocktail (Roche).

Since this study evaluated, for the first time (to our knowledge), whether coronavirus-infected individuals could generate an Ab response against the Cys-like protease, Mpro, other SARS-CoV-2 proteins, commonly used in serology tests, were produced for comparison. Mpro and NP were expressed in E. coli, and two different constructs of the RBD of the S protein were used: one was expressed by transfection in mammalian cells (mRBD) and a second was produced by baculovirus infection of insect cells (iRBD-His). All the proteins, except mRBD, had a histidine-tag and they were purified on Ni²⁺-NTA columns followed by size-exclusion chromatography (Supplemental Fig. 1A–D).

Before testing a large number of sera from COVID-19 patients and healthy donors, experiments were designed to optimize coating and dilution conditions. These data already revealed that COVID-19 patient sera contained high titers of Mpro-specific Abs. Ab reactivity to the viral protease reached saturation at relatively low concentrations and discriminated efficiently between individuals who had been infected with SARS-CoV-2 and those that had not been exposed to the virus (Fig. 1A). Serum dilutions from 1/50 to 1/1600 covered a broad range of reactivity to Mpro, from almost no recognition to saturation (reached at 1/100 dilution). It was also possible to detect low titers of Abs of the IgM and IgA isotypes in these patients (Fig. 1B), suggesting that, in subsequent experiments, a large screening of patient samples should be performed, including the three Ig subclasses. Coating titration experiments further confirmed the specificity of the assay (Fig. 1C). The IgG reactivity against the protease Mpro in COVID-19 patients was comparable, or in certain cases stronger, to the reactivity against RBD; however, no differences were noticed between the RBD recombinant proteins expressed in either mammalian cells or baculovirus (Supplemental Fig. 1E). These initial experiments suggested that the humoral response against the three viral proteins can be heterogeneous between different patients.

To further validate the assay, additional controls were performed, such as monitoring the background in plates with no viral Ag coating and testing sera collected before the COVID-19 pandemic (Supplemental Fig. 2).

A cohort of 36 COVID-19 patients (PCR+) and 33 healthy donors was recruited at La Princesa University Hospital, Madrid (Table I), and ELISA assays were performed to detect Mpro- as well as RBD- and NP-specific Abs of the IgG, IgA, and IgM subclasses in sera (Fig. 2).

Titration of the serum samples was carried out over a dilution range of 1/50–1/3200, and these experiments showed that assay for seropositivity to all three Ags discriminated between COVID-19–positive and –negative donors, as shown in dot plots comparing different dilutions (Supplemental Fig. 3). Fig. 2 summarizes the absorbance data from all the sera samples. To estimate the cutoff value, the sensitivity, and the specificity parameters for each Ag/Ig isotype pair, ROC analyses were performed (Fig. 3, Table II). The best AUC values were obtained with the measurement of IgG Abs specific for Mpro and NP (AUCs = 0.9945 and 0.9927, respectively). The sensitivity and specificity were above 90% for detection of IgG Abs of the three proteins tested, with values of sensitivity and specificity for Mpro of 97 and 100%, respectively. AUC values above 0.85 were obtained for the other isotypes (IgA, IgM). Measurement of anti-IgA Abs appeared to discriminate less accurately between pre–COVID-19 sera and COVID-19 sera; however, this is not because of a lack in sensitivity for this isotype. Instead, because background levels with IgA were very low and the signal clearly positive in some patients, the lack of detection suggests that certain COVID-19–positive patients have circulating IgA, whereas other COVID-19–positive patients lack IgA in peripheral blood. Whether the presence of IgA in periphery has any relationship with clinical aspects needs to be explored further in larger cohorts of patients.

Comparison between proteins showed some heterogeneity in the capacity of different donors to produce Abs, especially for IgM and IgA subclasses. Nonlinear polynomial regression showed a better correlation between the detection of Abs against NP and Mpro compared with NP and RBD or Mpro and RBD (Fig. 4A). Only one COVID-19 donor failed to make a full Ab response.

Further analyses were performed to explore the correlations between the titers of the different Abs in serum and clinical parameters. Interestingly, a trend for higher titer Ab responses was found in patients with more-severe disease (Fig. 4B), being more pronounced for IgM against Mpro and IgG against RBD. However, several other variables also contributed to the heterogeneity in Ab response, mainly age and time since the onset of symptoms (Table III). After adjustment for these possibly confounding factors, IgA anti-RBD was observed to be significantly higher in critical patients compared with patients with mild disease. In addition, critical patients showed a trend of higher IgM and IgA anti-Mpro titers than patients with mild COVID-19. Furthermore, intense IgM and IgA responses against the three proteins were significantly associated with higher serum IL-6 levels (data not shown).

Therefore, the use of SARS-CoV-2 Mpro, in combination with other Ags already described for serology tests, provided outstanding specificity and sensitivity for patient identification. IgG titrated further than IgA or IgM, indicating that, as expected, the IgG subclass is more abundant in serum. Assay for IgM Abs had a lower signal/noise ratio, and in many of the SARS-CoV-2–negative sera, a significant background could be observed for IgM. In contrast, SARS-CoV-2–specific IgA Abs were not detected in healthy donors, but were clearly present in 27 out of the 36 sera tested from COVID-19 patients.

To compare the kinetics of Mpro Ab response with that of other SARS-CoV-2 proteins, a follow-up of 14 patients was performed, selecting seven patients with a high titer the first month after the onset of symptoms and seven patients with a medium-low titer at the same time point. Ab levels were compared for each donor (Fig. 5A), and the percentage of change was also calculated (Fig. 5B). IgG concentration decreased slightly in most of the patients; however, Abs against the three proteins could still be detected 4 mo after the onset of symptoms. In contrast, IgA and IgM levels clearly decreased and, in certain patients, reached background levels.

One patient showed a marked increase in IgG in the sample obtained 4 mo after the onset of symptoms. In this case, the first sample was obtained on the first week of symptoms, probably when the immune response was still not fully activated. The patient had then a severe disease, explaining the increase in Ab levels later in time.

Saliva samples were collected from 11 healthy donors and 12 COVID-19 patients at the University Hospital La Princesa (Madrid) and tested in ELISA assays over a range of dilutions (1/2–1/10). IgG recognizing the three viral Ags tested could be observed in COVID-19 patients, with the strongest responses being those specific for the viral protease Mpro (Fig. 6). IgA and IgM responses were detected in only one of the COVID-19–infected individuals. This saliva sample was collected 59 d after the confirmation of a SARS-CoV-2–positive PCR test, and the patient had very mild disease.

The results presented in this article describe the detection of Abs against the SARS-CoV-2 protease, Mpro, in serum from COVID-19 patients. The titers of Mpro-specific Abs were comparable to those produced against SARS-CoV-2 NP and somewhat higher than the Ab responses to the RBD fragment of the S glycoprotein, both of which are generally considered immunogenic coronavirus proteins. These high titer Ab responses in serum were accompanied by the detection of Mpro-specific IgG Abs in saliva, providing a new opportunity for completely noninvasive diagnostic tests.

For IgG Abs in sera, the titers of NP and Mpro-specific Abs correlate very well with each other (r = 0.94 and p < 10⁻⁴) and also with anti-RBD responses (r = 0.89 and p < 10⁻⁴). In contrast, whereas NP- and Mpro-specific Ab titers also correlate well for IgA and IgM responses (r values >0.9), the correlation with IgM and IgA for RBD is much weaker (r values around 0.6). One plausible possibility is that the Ab responses to internal Ags, Mpro and NP, correlate well because production of Abs against these proteins requires either viruses with a broken membrane or release of viral material from infected cells.

The correlation with clinical data and symptoms onset reveals that Abs have higher titers as the severity of the disease increases. Although the sample size is not large, this correlation was significant and independent of age and time from the beginning of symptoms for anti-RBD IgA and almost significant for anti-Mpro IgM and IgA. The retrospective design of our study does not allow determination of whether these increased levels are cause or consequence of more-severe disease and what the basis of its relationship is with higher levels of IL-6 detected in critical patients. In this regard, it is surprising that IgM persisted at high levels in patients’ sera for more than a month after the beginning of symptoms.

The finding that the protease Mpro can be antigenic opens a new series of questions on the biology of this protein that is an important target for the development of antivirals to block SARS-CoV-2 replication. Mpro is key for cleavage and activation of the first polypeptide translated postinfection, but the protein has not been found in the virion. So, most probably, the generation of Abs directed against Mpro occurs at the end of the viral life cycle when intracellular Ags are released from the infected cell. It is not clear whether Abs specific for Mpro might interfere with viral replication directly; however, B cells producing these Abs would likely efficiently internalize and present this Ag to stimulate T cell recognition of peptides from intracellular proteins. NP and Mpro Ags used in these assays were produced in bacteria, so lack glycosylation. In mammalian cells, there is no clear evidence that these proteins are glycosylated, although they have potential N-glycosylation sites. Thus, it might be possible that carbohydrates could mask or alter Ab recognition of certain epitopes. In any case, the epitopes that are available in the recombinant proteins are sufficient for patient Ab recognition of these Ags.

Despite some reports suggesting that serum Ab against SARS-CoV-2 declined rapidly (J. Seow, C. Graham, B. Merrick, S. Acors, K. J. A. Steel, O. Hemmings, A. O'Bryne, N. Kouphou, S. Pickering, R. Galao et al., manuscript posted on medRxiv, DOI: 10.1101/2020.07.09.20148429; 12), our data agree with a recent report on more than 1000 patients in which seropositivity of Abs against N, RBD, and S was also maintained 4 mo after diagnosis (13). In the cohort presented in this article, the concentration of IgG in serum remained clearly detectable 4 mo after the onset of the disease. It is interesting to note that during this time period the Ab response to each of the three proteins tested seems to have similar kinetics. More follow-up experiments will be performed as longer-term samples become available. Whereas RBD, NP, and Mpro can be used for detection of seropositive patients as well as the full S protein (which, in general, provides high signal) (6), detection of its RBD fragment has the benefit of allowing an estimation of the effective immune response because the Abs against this fragment of the S protein are more likely to be virus-neutralizing (14).

Sequence analysis suggests that it is unlikely that the response detected against NP and Mpro is due to cross-reactivity between coronavirus-specific Abs. Whereas COVID-19 Mpro has 96% homology with the main protease of SARS-CoV, which emerged in China in 2003, the similarity with other coronaviruses is much lower. All the samples analyzed in this study came from hospitals in Spain, where no cases of SARS-CoV-1 have been reported. The similarity between the Cys-like proteases (Mpro) of different coronaviruses (SARS-CoV-2, HCovNL63, HCoVOC43, and HCov229E) is only ∼40% with changes and similarities distributed along the whole sequence (Supplemental Fig. 4).

A remarkable observation is that SARS-CoV-2–specific Abs can be detected in the saliva of seropositive individuals. Two major Ab classes are found in saliva: secretory IgA, synthesized locally by plasma cells in salivary glands, and IgG that is mainly derived from serum via gingival crevices (15). In our experiments, salivary SARS-CoV-2 Abs were mainly IgG rather than IgA; only one out of 12 individuals with SARS-2–specific IgA was observed, corresponding to a donor that had recovered from the disease 1 mo before the saliva test. These experiments do not specifically detect secretory IgA in saliva because the Ab used was raised by immunization with IgA purified from serum, and these Abs often do not recognize IgA2 well. The fact that three patients had very high levels of IgA in serum and did not have IgA in saliva suggests that the IgA detected in saliva probably is not related to serum IgA. The observation that COVID-19–positive, but not COVID-19–negative, individuals contain robustly detectable levels of SARS-CoV-2 NP- and Mpro-specific Abs in saliva is interesting because the development and validation of a saliva-based assay for SARS-CoV-2 seropositivity would represent a practical, noninvasive alternative to blood-based assays for COVID-19 diagnostic testing that might complement saliva-based nucleic acid tests for SARS-CoV-2 nucleic acid.

We thank the director of the National Centre for Biotechnology-Spanish National Research Council (CNB-CSIC), M. Mellado, for coordination; Luis Enjuanes and Sonia Zúñiga (CNB-CSIC) for SARS-CoV-2 DNA; Florian Krammer (Mount Sinai School of Medicine) for the plasmid for iRBD expression; César Santiago, Antonio J. Varas, Juan R. Rodríguez, and José F. Rodríguez (CNB-CSIC) for the production and purification of iRBD; and R. Delgado (Hospital 12 de Octubre, Madrid) for sera sample selection.

J.M.R.F., J.M.C., H.T.R., and M.V.-G. are inventors on the European patent “Assay for the detection of the Cys-like protease (Mpro) of SARS-CoV-2” (EP20382495.8). I.G.-A. received personal fees from Lilly and Sanofi; personal fees and nonfinancial support from Bristol Myers Squibb; personal fees and nonfinancial support from Abbvie; research support, personal fees, and nonfinancial support from Roche Laboratories; and nonfinancial support from Merck Sharpe & Dohme, Pfizer, and Novartis, not related to the submitted work. The other authors have no financial conflicts of interest.