A CRISPR-based ultrasensitive assay rede nes “undetectable” SARS-CoV-2 antibodies in kidney transplant recipients after COVID-19 vaccination

Yanan Tang Sichuan University Turun Song West China Hospital, Sichuan University Lu Gao Sichuan University Saifu Yin West China Hospital, Sichuan University Min Ma West China Hospital, Sichuan University Yun Tan Sichuan University Lijuan Wu West China Hospital, Sichuan University Tao Lin West China Hospital, Sichuan University Feng Li (  i-scu@outlook.com ) Sichuan University https://orcid.org/0000-0002-2616-5343


Introduction
Coronavirus disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was declared a pandemic by the World Health Organization (WHO) on 11th March 2020. 1 The urgent requirements for sensitive and speci c assays to detect viral infections and subsequent immune responses at both the individual and population levels have offered tremendous opportunities for accelerating the development and validation of new diagnostic techniques for clinical uses. CRISPRbased point-of-care tests (POCTs) are among the most attractive tools for diagnosing  infections via the detection of viral RNA under resource-limited conditions. 2,3 Currently, both CRISPR-Cas13 speci c high-sensitivity enzymatic reporter unlocking (SHERLOCK) and Cas12-mediated DNA endonuclease-targeted CRISPR trans reporter (DETECR) have been approved by the United States Food and Drug Administration (FDA) for commercial uses through Emergency Use Authorization (EUA). [3][4][5] Despite remarkable success as nucleic acid tests (NATs), CRISPR-Cas systems have not been exploited for the sensitive detection of anti-SARS-CoV-2 antibodies, another important class of biomarkers for estimating the seroprevalence of infection, monitoring the progress of pandemic, and assessing the protective humoral immunity of infected individuals and recipients of vaccination. [6][7][8] Although many conventional immunoassays, such as enzyme-linked immunosorbent assay (ELISA) and chemiluminescence immunoassay (CLIA), have been deployed for measuring anti-SARS-CoV-2 antibodies in the general population, [9][10][11][12] the mismatch between seroconversion and effectiveness in immunocompromised patients suggests that current assays might underestimate the humoral response in this population. 13,14 Therefore, simple, ultrasensitive assays with limit-of-detections (LODs) comparable with those of CRISPR-based NATs are critically needed for evaluating the adaptive humoral immune response and infection rates in immunocompromised subpopulations. 7,[15][16][17] Kidney transplant recipients (KTRs) are an especially vulnerable patient population because of the therapeutic immunosuppression that impairs their immune responses to the COVID-19 vaccine. 18 Several studies have reported that only 4-48% of KTRs have detectable anti-spike IgGs after receiving two vaccine doses. 18 On the other hand, spike speci c T cell and B cell-mediated immune responses were still found in KTRs with a negative antibody response after the administration of either mRNA or inactivated COVID-19 vaccines. [19][20][21][22] Recent studies have also revealed that KTRs who had a weak response (below the positivity limit) after the second dose were more likely to develop an antibody response after the third dose compared with those without an antibody response. 23 Collective clinical observations indicate that KTRs might have weak humoral response not detectable by the current techniques. An ultrasensitive antibody test capable of measuring "undetectable" anti-SARS-CoV-2 antibodies de ned by conventional techniques will offer valuable information for better understanding the immunity of KTRs and the design of therapeutic immunosuppression and vaccination programs for this population, particularly identifying those who can bene t from additional or booster doses of COVID-19 vaccines.
Herein, we describe the development of an Ultrasensitive CRISPR-based Antibody Detection (UCAD) assay that is 10,000 times more sensitive than the commercial ELISA kits for the detection of anti-SARS-CoV-2 spike protein receptor binding domain (anti-RBD) IgG and IgM in undiluted human serum samples. UCAD was clinically validated using 65 anti-SARS-CoV-2 positive serum samples collected from healthy participants after two doses of inactivated COVID-19 vaccines as positive controls and 17 human serum samples collected before the pandemic as negative controls. The clinical speci city of UCAD was further validated against 55 clinically identi ed COVID-19 negative serum samples collected during the pandemic. We nally deployed UCAD for the ultrasensitive detection of anti-RBD IgG and IgM in clinical serum samples collected from a cohort of 85 KTRs who received two doses of inactivated COVID-19 vaccine. Using UCAD, we detected 54 (63.5%) anti-RBD IgG positive (IgG + ) and 67 (78.8%) anti-RBD IgM positive (IgM + ) serum samples previously de ned to be "undetectable" using the standard CLIA. As expected, the levels of anti-RBD IgG or IgM in KTRs were found to be signi cantly lower than those in healthy controls. By rede ning the detectable antibody levels in KTRs, we were able to reclassify KTRs into anti-RBD-positive (IgG + /IgM + , IgG + /IgM -, IgG -/IgM + ) and anti-RBD-negative (IgG -/IgM -) groups. A further comparison between the two groups revealed signi cant differences in the populations of type-2 T helper (Th2) cells, Th17 cells, and plasmablasts responding to vaccination.

Results
Design and operation of UCAD assay. The idea of UCAD is to convert the detection of anti-SARS-CoV-2 into the production of a predesigned CRISPR-Cas12a targetable double-stranded DNA (dsDNA) barcode, so that the recombinase polymerase ampli cation (RPA) and subsequent indiscriminate single-stranded DNase (ssDNase) activity of Cas12a can be unleashed by the antibody (Fig. 1a). To achieve this goal, we rst designed a dsDNA barcode containing a 20-nucleotide (nt) binding domain to the CRISPR guide RNA (crRNA) (Fig. 1b). To disrupt the binding and activation of Cas12a in the absence of the antibody, we truncated the dsDNA barcode at both the target strand (TS) and nontarget strand (NTS) at the middle of the crRNA binding sequence. The TS and NTS probes were then conjugated with the SARS-CoV-2 spike protein RBD and an anti-human IgG (or IgM) antibody, respectively (Fig. 1c). A short complementary domain was left between the two probes to initiate antibody-induced proximity hybridization (Fig. 1c). Driven by a T4 polymerase, the complete dsDNA barcode forms via primer extension, which serves as a surrogate for the ampli ed detection of anti-RBD. Experimentally, the optimal length of the complementary domain between the TS and NTS probes was determined to be 6 nt ( Supplementary Fig.  S1). With this design, the melting temperature (T m ) was estimated to be only ~10°C, so that no stable duplex could be formed at 37°C (Fig. 1c). In the presence of anti-RBD, the binding of anti-human IgG (or IgM) and RBD to the same antibody brings the two DNA probes into proximity and thus leads to the formation of a stable duplex with an estimated T m of 46°C (Fig. 1c).
The assay protocol of UCAD consists of three simple steps (Fig. 1a), including 1) antibody-speci c primer extension to produce dsDNA barcodes; 2) RPA ampli cation; and 3) the cleavage of uorophore-quencher (FQ) labeled ssDNA reporters mediated by CRISPR-Cas12a. Reactions at each step were all performed homogenously at a constant 37 ºC in a single test tube without the need for any separation steps. By carefully optimizing the assay conditions ( Supplementary Fig. S2 & S3), we were able to detect as low as 10 aM anti-RBD human monoclonal antibody, an LOD comparable to those of CRISPR-based NATs (Fig. 1d, Supplementary Fig. S4). The high sensitivity and low background of UCAD were also con rmed using polyacrylamide gel electrophoresis (PAGE) analysis ( Supplementary Fig. S5). UCAD was also determined to be highly speci c, evidenced by testing against anti-SARS-CoV-2 nucleocapsid (N) human monoclonal and anti-RBD rabbit polyclonal antibodies at concentrations ranging from 10 aM to 1 pM ( Fig. 1e and Supplementary Fig. S6). Throughout the concentration ranges, detectable uorescence signals could only be achieved in the presence of the target anti-RBD human mAb but not the two closely related interfering antibodies.
We next examined whether UCAD would be effective for detecting anti-RBD IgG and IgM in human serum samples. The kinetic curves in Fig. 1f and 1g demonstrate that UCAD could effectively detect both anti-RBD IgG and IgM in certi ed IgG/IgM positive human serum samples obtained commercially, whereas very little uorescence increases were shown in the negative samples. The sensitivity of UCAD was further compared with standard ELISA using a series of human serum samples prepared by diluting the IgG/IgM positive serum using the negative serum with dilution factors ranging from 1 to 100,000. UCAD was able to detect anti-RBD IgG ( Fig. 1h and Fig. S7a) and IgM ( Fig. 1i and Fig. S7b) throughout the dilution series. In contrast, the standard ELISA kit worked only for 10-fold diluted serum samples ( Fig. 1h and 1i), suggesting that our UCAD assay is at least 10,000 more sensitive than ELISA.
Clinical validation of UCAD. We rst validated UCAD against 65 positive clinical human sera that were previously tested by West China Hospital of Sichuan University using a standard total anti-RBD CLIA assay and 17 pre-pandemic sera collected in 2019 before the outbreak of COVID-19. All positive samples were collected from healthy participants who had 2 doses of inactivated COVID-19 vaccine. Both IgG and IgM levels against RBD were determined using UCAD ( Fig. 2b and 2c). Signi cant differences were found between the positive and pre-pandemic cohorts (p < 0.0001, n = 82) ( Fig. 2d and 2e). To further evaluate the clinical speci city of UCAD, we also obtained a set of sera collected from 55 unvaccinated healthy participants during the pandemic, who have been con rmed to be anti-RBD negative using the standard CLIA test. Comparing with the pre-pandemic cohorts, neither IgG nor IgM levels were signi cantly different in this cohort ( Fig. 2d and 2e). The 55 negative sera were combined with the 17 pre-pandemic sera as a new negative group to determine the optimal cutoff values for serum anti-RBD IgG and IgM via receiver operating characteristics (ROC) curve analysis (Fig. 2f). The binary outcomes (positive or negative) for anti-RBD IgG and IgM were then combined and compared with the standard CLIA using a confusion matrix (Fig. 2g). The clinical sensitivity and speci city of UCAD (n = 137) were determined to be 100% and 97.2%, respectively. Altogether, within the total 72 pre-pandemic and CLIA-con rmed negative samples, only two samples (one for IgG test and one for IgM test) were found to be false positive with slightly higher uorescence than the cutoff values and were identi ed to be outliers using Dixon's Q test.
As a CRISPR-based ultrasensitive assay, UCAD also holds great promise as a POCT for detecting anti-SARS-CoV-2 antibodies under resource-limited conditions. We next engineered a lateral ow readout for UCAD by replacing the uorophore-quencher labeled ssDNA reporter with one labeled with FAM and digoxin (Dig) at the 3' and 5' ends, respectively. We then tested UCAD with a lateral ow readout for the detection of anti-RBD IgG and IgM in clinical sera (more details are provided in Supplementary Section S2). Consistent with our uorescence-based readout, we detected anti-RBD IgG and IgM in 15 positive clinical sera con rmed by both UCAD and CLIA and 7 pre-pandemic negative sera using the naked eye ( Supplementary Fig. S8).
Detection of anti-RBD IgG/IgM in vaccinated KTRs. We studied a cohort of 85 KTRs who had received 2 doses of inactivated COVID-19 vaccine using UCAD (Fig. 3a). All KTRs received a triad immunosuppression regimen composed of mycophenonate mofetil (500 mg bid), steroids (5-10 mg qd), and tacrolimus (TAC) with target trough levels of 5 ng/mL to 8 ng/mL. More details on demographic and therapeutic information are listed in Supplementary Tables S2 & S3. Sex and age were found to have no signi cant impact on the levels of anti-RBD IgG and IgM determined using UCAD ( Supplementary Fig.   S9).
Serum samples were collected 10 to 60 days (median = 26 days) after the second dose of vaccine.
Consistent with previous reports, 18 only 5 out of 85 KTRs (5.9%) were found to be anti-RBD positive using the standard CLIA assays (Supplementary Table S2). In contrast, a total of 73 KTRs (85.9%) were determined to be anti-RBD positive using UCAD, including 54 IgG + and 67 IgM + (Fig. 3b and 3c).
For both anti-RBD IgG and IgM, UCAD signals of the KTR cohort were signi cantly higher than those of the negative controls (n = 157, p < 0.0001) ( Fig. 3e and 3f). As expected, substantially reduced humoral immunity was observed in KTRs in response to vaccination, evidenced by the much lower levels of anti-RBD IgG and IgM than those of the vaccinated healthy cohort (n = 150, P < 0.0001) ( Fig. 3e and 3f). Several studies suggest that the compromised immunity of KTRs to vaccination was a result of therapeutic immunosuppression. 18,24 We also observed a negative linear correlation between anti-RBD IgM levels and the trough levels of TAC (r = -0.2519, p = 0.034) using Pearson correlation analysis (Fig. 3g), whereas the levels of IgG were not statistically correlated with TAC trough levels (Fig. 3h).
Upon UCAD analysis, we were able to classify the 85 KTRs into the anti-RBD positive group (IgG + /IgM + , IgG + /IgM − or IgG − /IgM + , n =73) and negative group (IgG − /IgM − , n = 12). Through Pearson correlation analysis, we found that the levels of IgG in the positive group were positively correlated with the percentage composition of lymphocytes in white blood cells (WBCs) counted before (r = 0.2536, p = 0.0316, Fig. 4b) and after vaccination (r = 0.2477, p = 0.0387, Fig. 4c). As lymphocytes are instrumental for antibody production, the positive correlation between IgG level and lymphocyte percentage indicated the activation of humoral immunity in KTRs after COVID-19 vaccination. This was also consistent with previous observations made in COVID-19 vaccinated healthy populations. 25,26 Flow cytometric and statistical analysis of blood samples collected from 7 IgG + /IgM + (Supplementary  (Fig. 4e, 4f & Supplementary Fig. S31-S33), suggesting that the abnormality of B cell differentiation and reduced antibody production in anti-RBD negative KTRs might also be related to T cell regulation. [24][25][26] Discussion Here, we present UCAD, a simple and ultrasensitive assay that is 10,000 times more sensitive than standard ELISA for the detection of serum IgG and IgM against SARS-CoV-2 RBD. Clinically, UCAD was determined to be 100% sensitive and 97.2% speci c based on 137 positive and negative COVID-19 sera collected from the general population. More importantly, in a cohort of 85 vaccinated KTRs, UCAD successfully identi ed 68 anti-RBD positive sera that were previously de ned to contain an "undetectable" level of anti-RBD using the conventional CLIA test (Supplementary Table S2), though uorescence signals of KTRs were signi cantly lower than those of the healthy participants. By reclassifying seroconversion positive and negative KTRs using UCAD, we found signi cant differences in the levels of plasmablasts, Th2 cells, and Th17 cells. These critical clinical ndings enabled by UCAD echo the recent clinical observations that the humoral response in KTRs might be underestimated by current tests, 11,12,16 may initiate more in-depth studies to better understand the humoral immunity of KTRs and shed light on the heterogeneity in response to COVID-19 vaccination. UCAD may also be used directly to guide the design of therapeutic immunosuppression and vaccination programs for KTRs. For example, to identify individuals who are more likely to develop an antibody response after the third booster dose of COVID-19 vaccine and high-risk individuals who are irresponsive to vaccine that convalescent plasma or COVID-19 speci c immunoglobulin should be applied as preventive strategy. 27 UCAD also represents a technical advance and paradigm shift of CRISPR-Dx from solely NATs to ultrasensitive protein analysis. The past three years have witnessed the remarkable success of CRISPR-Dx as fast and cost-effective POC NATs accessible for patients in low-resource settings. Because UCAD is an "add-on" to existing CRISPR-Dx with a one-step conversion from protein to DNA barcode in a homogeneous solution, it is readily adaptable to existing detection platforms for SHERLOCK or DETECTR without compromising any sensitivity or simplicity. For example, we show here that the uorescent readout of UCAD could be readily converted to a later ow test with a visual readout (Supplementary Section S2 and Fig. S8). Therefore, compared to existing ultrasensitive protein detection techniques, such as single molecular array (Simoa) 28,29 or proximity ligation assays (PLA) 15,16,30 , UCAD works isothermally at 37 ºC without the need for any specialized equipment or tedious operation and washing steps, and thus is particularly attractive for use in resource-constrained conditions, including the opportunity for self-monitoring.
In summary, this work demonstrates the development and clinical uses of UCAD as an ultrasensitive protein assay capable of rede ning "undetectable" anti-SARS-CoV-2 antibody levels in sera of KTRs. In future studies, we aim to use this method to track temporal changes in antibody levels in vaccinated individuals to gain better understanding of the immunity of healthy people and KTRs. We will also further engineer UCAD as a broadly applicable POCT for the ultrasensitive detection of clinically important proteins beyond anti-SARS-CoV-2.

Methods
Sample Collection. This study was approved by the Ethics Committee of West China Hospital of Sichuan University (NO: 2021-110), where all clinical serum samples (n = 222) were obtained and analyzed using a standard CLIA test. All research was performed in accordance with relevant guidelines and regulations. All participants have provided informed consent. Blood samples were collected following the standard procedure at West China Hospital of Sichuan University. Sera were prepared using serum separator tubes and incubated at room temperature for 1 h. The clot was removed by centrifugation at 1000 × g for 10 min. The supernatant was collected, aliquoted and stored at -80°C upon usage. Peripheral blood mononuclear cells (PBMCs) were separated from EDTA anticoagulated blood samples of 85 KTRs through density gradient separation, 31 and the WBCs were counted and differentiated by Sysmex hematology XN modular system.
Preparation of DNA probes for UCAD. All DNA probes and reporters were purchased from Sangon Biotech Co., Ltd. (Shanghai, China) and were puri ed by high-performance liquid chromatography (HPLC).
Sequences and modi cations of all DNA or RNA probes are listed in Table S1. For lateral ow strip-based visual readout, the RPA amplicons were incubated with 40 nM Cas12a, 40 nM crRNA and 40 nM FAM-Dig ssDNA reporter in 20 µL 1× NEBuffer™ 2 at 37°C for 30 min, and then loaded on the sample pad of the lateral ow strip with 30 µL of 4×SSC buffer. Once developed, the results on the lateral ow strips were read directly using the naked eye or captured using a digital camera.
Detection of anti-RBD using standard ELISA test. A commercial ELISA kit (BGI genomics, Shenzhen, China) was used to measure anti-RBD IgG and IgM in certi ed COVID-19 positive or negative human serum samples included in the ELISA kit. The assay was performed according to the instructions provided by the manufacturer. Brie y, 10 µL human serum sample (or diluted serum sample) was mixed with 100 µL diluent and incubated in the ELISA microplate well at 37°C for 30 min. The microplate was then washed 5 times with 350 µL washing buffer. Then, 100 µL IgG or IgM enzymatic working solution was added to the microplate and incubated at 37°C for 30 min. The microplate was washed again with 5 × 350 µL washing buffer. To this microplate, 50 µL substrate A and 50 µL substrate B were added and incubated at 37°C for 10 min in darkness before the addition of 50 µL quenching solution. The OD value (450-620 nm) of each sample was measured immediately after quenching the reaction using Cytation 5 multimode microplate reader.
Flow cytometric analysis of white blood cells. Seven IgG + /IgM + KTRs, ve IgG − /IgM − KTRs and ve vaccinated healthy participants were recruited for the subpopulation analysis of WBCs by Fluorescence Activated Cell Sorter (FACS). 2 ml EDTA anticoagulated clinical blood samples from each patient were collected. After lysing red blood cells as instructed, 31 the WBC subsets were characterized by a BD FACSCanto™ II Cytometer after uorescently labeling the surface antigens. 32 34 epitopes on the WBC surface were labeled using uorophore-tagged antibodies, such as FITC, PE, PerCP-CY5.5, PC7, APC, APC-H7, Paci cBlue and BV-421. The 34 epitopes were CD45 CD3 CD4 CD8 CD25 CCR7 CD45RA HLA-DR CD19 CD56 CD127 IgG1 IgD CD57 PD-1 CD28 CD27 CD38 CD86 CD123 7-AAD CD11c Lin CTLA-4 CCR10 CCR6 CCR4 CXCR3 CD160 LAG-3 Tim-3 CD15 CD33 CD11b CD14. Then labeled cells were sorted by a BD FACSCanto™ II Cytometer with antibody panels limited to 8-colors. Data were analyzed by BD FACSDiva TM with a standardized gating strategy following the ONE-Study protocol (Table S4). 33,34 Data and statistical analysis. All graphing and statistical analysis were performed in Excel and GraphPad Prism. The unpaired t-test was performed in Prism to compare UCAD signals of anti-RBD IgG and IgM between two cohorts. For comparisons between three or more cohorts, ordinary one-way ANOVA was performed. Dixon's Q test was used to identify outliers. Receiver operating characteristics (ROC) curve analysis was used in Prism to determine the cutoff values of UCAD for positive and negative results. A confusion matrix was performed using Excel to evaluate the performance of UCAD in comparison with the standard CLIA. Nonparametric Spearman correlation analysis was used in Prism to determine the correlation R value between UCAD signals and sampling time. Parametric Pearson correlation analysis was performed to evaluate the linear correlations between UCAD signals and clinical parameters, such as TAC trough levels and lymphocyte levels. For all analyses, a two-tailed P value < 0.05 was considered to be statistically signi cant.  cohort (n = 65) were signi cantly higher than those of the pre-pandemic and pandemic negative cohorts (p < 0.0001). The pre-pandemic and pandemic negative cohorts had no signi cant difference (p = 0.1090 for IgG and p = 0.2023 for IgM using unpaired t-test) and were thus combined as one negative group (n = 72) to determine cutoffs for distinguishing positive and negative results in UCAD. (f) ROC curves of the UCAD assay for detecting anti-RBD IgG and IgM in 137 human serum samples (positive: n = 65, negative: n = 72). Optimal cutoff uorescence values were selected through ROC analysis: 3007 a.u. at 40 min for anti-RBD IgG (speci city = 100%, sensitivity = 98.61%) and 1973 a.u. at 40 min for anti-RBD IgM (speci city = 87.69%, sensitivity = 97.22%). (g) Evaluation of the clinical sensitivity and speci city of UCAD compared to the standard CLIA test using a confusion matrix (n = 72 for negative sera and n = 65 for positive sera).