Establishment of HLA class I and MICA/B null HEK-293T panel expressing single MICA alleles to detect anti-MICA antibodies

Pre- and post-transplantation anti-MICA antibody detection development are associated with an increased rejection risk and low graft survival. We previously generated HLA class I null HEK-293T using CRISPR/Cas9, while MICA and MICB genes were removed in this study. A panel of 11 cell lines expressing single MICA alleles was established. Anti-MICA antibody in the sera of kidney transplant patients was determined using flow cytometric method (FCM) and the Luminex method. In the 44 positive sera, the maximum FCM value was 2879 MFI compared to 28,135 MFI of Luminex method. Eleven sera (25%) were determined as positive by FCM and 32 sera (72%) were positive by the Luminex method. The sum of total MICA antigens, MICA*002, *004, *009, *019, and *027 correlation showed a statistically significant between the two methods (P = 0.0412, P = 0.0476, P = 0.0019, P = 0.0098, P = 0.0467, and P = 0.0049). These results demonstrated that HEK-293T-based engineered cell lines expressing single MICA alleles were suitable for measuring specific antibodies against MICA antigens in the sera of transplant patients. Studies of antibodies to MICA antigens may help to understand responses in vivo and increase clinical relevance at the cellular level such as complement-dependent cytotoxicity.

www.nature.com/scientificreports/ The MICA gene is highly polymorphic, and exposure to mismatched MICA antigens can induce allogenic responses that are associated with transplant failure [15][16][17] . Pretransplant anti-MICA antibodies have been demonstrated to be risk factors for allograft rejection and lower graft survival [17][18][19][20][21][22][23] . Also, the post-transplantation development of anti-MICA antibodies is associated with acute rejection and allograft loss 21,22,24 . In an earlier study, a single MICA gene was expressed and used in a CR1 cell line that does not express HLA, but it was found that the CR1 cell line also expressed its own endogenous MICA gene 25 . An insect-derived cell line, High Five cells, was tried to be used, but it was judged unsuitable because of the nonspecific heterologous antibody response to carbohydrate molecules 26 . Since there is no human cell line in which the MICA/B gene is not fully expressed, the study of anti-MICA antibody assay for single MICA antigen using cell lines could not be continued. Currently, the Luminex method using ELISA or beads using soluble recombinant protein has been mainly developed and used clinically. Recently, because antibody measurement using Luminex assay is too sensitive, the clinical relevance is being questioned 27,28 . To determine the titer of a clinically meaningful antibody, it must be compared with the antibody response to MICA expressed on the cell surface. Therefore, in the previous study, a single MICA gene was transduced into HEK-293T cell line from which HLA class I had been completely removed and used in this study after MICA was completely removed.
We previously generated the HEK-293T cell line from which the HLA class I gene was removed using CRISPR/Cas9 29 . In this study, we generated the cell line from which MICA and MICB genes were removed. Based on this cell line, a panel of cell lines that expresses 11 types of MICA alleles frequent in Koreans was established. The presence of anti-MICA alloantibodies in the sera of kidney transplant patients was determined by flow cytometry method (FCM) using this panel of cell lines expressing single MICA alleles and FCM was also compared with Luminex method for single antigen.

Results
Establishment of MICA cell lines expressing single MICA antigen. We selected the H1E-25 HEK-293T cell line from which HLA class I was deleted to rule out alloantibody reactions against HLA 29 . Three plasmids encoding the Cas9 protein and gRNA to target MICA exon 2 (MAE2), MICB exon 2 (MBE2), and MICA/B exon 3 (MABE3) were used for complete deletion of the MICA/B molecule (Fig. 1A). These three plasmids were co-transfected with H1E-25 HEK-293T cells. Cells that were successfully transfected and expressed fluorescence were separated into single cells using fluorescence-assisted cell sorting (FACS) and cultured into 96-well plates. For the 48 clones growing, the cell lines with predicted deletions were investigated via PCR of the regions containing exon 2 and exon 3 (Fig. 1B). There were 12 heterozygous and six homozygous detectable changes (DCPs) in MICA and 14 heterozygous and 13 homozygous DCPs in MICB. We found that exon 2-3 of MICA/B was completely deleted in clones H1ME-5, -15, and -26 (Fig. 1C). MICA was cut at the predicted cut sites, and 267 bp were deleted from all three clones. However, in the exon 3 site of MICB, a total of 265 bp was deleted from all three clones by cutting 24 bp ahead of the predicted cut site.
Although MICA and MICB were weakly expressed in the H1E-25 HEK-293T cell line, it was confirmed that they were not expressed at all in the three clones from which both MICA and MICB were removed. In the subsequent experiments, the selected H1ME-5 HEK-293T cell line was used ( Fig. 2A). To establish a cell line panel expressing single MICA alleles, H1ME-5 HEK-293T cells were transduced with lentiviruses encoding 11 MICA alleles, MICA*002, *004, *007, *. 008, *009, *010, *011, *012, *019, *027, and *049. MICA-positive cells were sorted and cultured on day 6 after transduction. FACS analysis confirmed the high expression of 11 single MICA alleles (Fig. 1B). To investigate whether these cell panels were suitable for measuring anti-MICA antibodies in the patient serum, the serum determined as positive and the serum determined as negative were subjected to flow cytometric method (FCM) after fluorescence staining (Fig. 2C). Negative serum (13,101 MFI) was not measured in all cells expressing a single MICA allele in the same manner as H1ME-5, which did not express MICA. Positive serum (17,815 MFI) responded only to certain MICA alleles such as MICA*002, *011, and *049. These results indicate that the non-specific response of serum in the cell panel based on H1ME-5 HEK-293T cells is very low.
Responses of FCM and Luminex method to MICA alleles in patient serum. In this study, 11 MICA antigens used for FCM were selected to be present at a frequency of 2% or more in Koreans (Table 1). A total of 15 MICA antigens were included in the Luminex method, but six antigens were not reported in Koreans, and two antigens had a frequency of less than 1%. Seven antigens were detected by both the FCM and Luminex methods. Because MICA*008 and *027 or *009 and *049 alleles showed the same amino acid sequences of For detection of deletion mutation, targeted PCR were performed using specific primer pairs that included exons 2 and 3 of MICA/B genes, and genotypes were analyzed using gel electrophoresis. The black arrow indicates the PCR product size of the wild-type 293T control sample. The gray arrow indicates the PCR product size of the predicted deletion between exons 2 and 3 by designed gRNAs. When the PCR product size was different from that in control cells, we designated the detectable changes in PCR products as deletions (shorter products), insertions (longer products), or lack of amplification. When PCR products were detected as single or double bands, they were regarded as homozygous or heterozygous. (C) Nucleotide sequences between exons 2 and 3 in the selected MICA/B null clones (clones 5, 15, and 26). The target guide RNA sequence (in gray color), cut site (inverted triangle), and removed sequence (dot) were aligned to the nucleotide sequence from wild-type HEK-293T cell line.   Luminex method using sera of kidney transplant patients (n = 64). Results were displayed in the order of sera with weak response to strong response. As for the Luminex method, 33 sera with a response of 1247 MFI or higher were determined as positive, and, as for the FCM, 11 sera with a response of 237 MFI or higher were determined as positive. The 64 sera collected from kidney transplant patients used in this study consisted of 44 positive sera and 20 negative sera using the Luminex screening kit (mixed) (Fig. 3). The cut-off value for the FCM test was calculated as mean + 4 SD based on the data on the allele antigen with the largest deviation from 20 negative sera and determined as 237 MFI (mean = 59, SD = 45). Similarly, the cut-off value for the Luminex method was determined to be 1247 MFI (mean = 283, SD = 241). Among the negative sera, the maximum values for FCM and Luminex methods were 209 and 822 MFI, respectively.
After testing the positive 44 sera using the FCM and Luminex methods, the results from FCM were sorted from the highest value based on the allele with the highest response in each serum, and the results from the Luminex method were listed corresponding to FCM (Fig. 3). The maximum value of FCM was 2879 MFI, which was approximately 1/10 of the 28,135 MFI of the Luminex method. Eleven sera with more than this cut-off value were determined to be positive by FCM. Based on the above cut-off value, there were 11 (25%) and 32 (72%) positive serum samples by FCM and Luminex method, respectively. Correlation between the results of FCM and Luminex method. The correlation between the results of the two assays was analyzed in 11 serum samples that were positive in the FCM assay (Fig. 5). First, the total response in each serum sample was compared for each of the seven individual antigens present in both assays. MICA*002, *004, *009, *019, and *027 correlation showed a statistically significant (P = 0.0412, P = 0.0476, P = 0.0019, P = 0.0098, P = 0.0467, and P = 0.0049, respectively). However, the MICA*007 and *012 alleles showed no significant correlation.
Specificities to single MICA alleles of 11 positive sera by FCM. When analyzing the response pattern to each antigen for 11 positive sera in the FCM test, six sera were positive for more than five antigens, three were positive for two antigens, and three were positive for a single antigen (Table. 2). Two sera positive for a single antigen showed specific reactions to MICA*007 and *049, respectively. Two of the three sera positive for the two antigens reacted equally to MICA*002 and *011. In the sera positive for eight antigens, two sera responded identically to MICA*004, *007, *008, *009, *010, *012, and *027. MICA*008 and *027, which had the same ECD amino acid sequence, showed the same reaction (Fig. 5). However, MICA*009 and *049, which have the same ECD amino acid sequence, did not show the same reaction. The reactions to MICA*010 and *011, which existed in Koreans but were not included in the Luminex method, were 36% and 27%, respectively.

Discussion
In the initial study on allogeneic antibodies against MICA antigens, a cell line expressing a single MICA antigen was generated using m-HMy2. The C1R cell line weakly expresses the HLA-C gene but does not express HLA-A, HLA-B, or MICA. It was used for the detection of antibodies against MICA antigen reaction through CDC assay or for identification of shared epitopes by antibody absorption in patient sera 24,30 . The K562 cell line, which does not express HLA, expresses a single HLA antigen and is used for FCM or CDC tests. However, MICA was expressed and was not used for studies on a single MICA antigen 31 . High Five insect cells that do not express human alloantigens were also used to generate single MICA-expressing cell lines, but the recombinant glycoproteins on the cell surface were found to cause cross-reactions with antibodies in human serum 26 . As it was not possible to obtain a cell line expressing a single MICA antigen, recombinant soluble proteins with extracellular domains of each MICA antigen produced in Escherichia coli were mainly used for ELISA or Luminex method from the beginning 15,32 . We previously used the CRISPR/Cas9 to remove all the HLA-A, -B, and -C genes of the HEK-293T cell line to establish a cell line without allogeneic response to HLA 29 . In this study, MICA and MICB genes were removed to establish a cell line without both HLA and MICA/B genes (Fig. 2). This cell line confirmed that non-specific reactions did not occur in either the positive or negative sera of all patients and was used as internal negative control in this study. www.nature.com/scientificreports/ As DNA typing methods for MICA have been developed precisely, more MICA alleles have been reported worldwide, and the distribution of the MICA allele in Koreans has also been well investigated 33 . In this study, 11 types of MICA alleles with a frequency of 2% or more in Koreans were selected to generate cell lines expressing a single MICA antigen for FCM. Compared to the currently clinically used Luminex method, antibody responses to MICA*010 and *011 antigens, which were not included in the Luminex method, were found to be 36% and 27%, respectively (Table 2). MICA*001, *005, *015, *018, *028, and *046 antigens included in the Luminex method were not present in Koreans, but reactive antibodies were present (Fig. 4). This phenomenon is known to be due to cross-reactions between MICA antigens that share common epitopes 30 . For example, an antibody against MICA*001 cross-reacts with MICA*001, *012: 01, *018: 01, *007: 01, *002: 01, and *017 alleles belonging to antigen group 1, which shares the same epitope. In FCM, six out of 11 antibodies in the positive sera showed strong cross-reactivity against five or more antigens. There was no significant correlation between *007 and *012 alleles in the serum correlation study for the antigens detected by FCM and Luminex method. Comparing the results of FCM and Luminex is thought to be influenced by various factors. For FCM, Anti-human IgG Fc-FITC was used, and for LABscreen MICA single antigen, PE-Conjugated Goat Anti-Human IgG was used to confirm the MFI of each serum. Although the amount of antigen used in FCM and Luminex method was not accurately compared, it is thought that there is a relationship between the target amount and the measurement of antibody. Also, it is thought that the sensitivity of the measuring equipment used for FCM and Luminex will also have an effect (Fig. 5). Since MICA*008 and *027 alleles had the same extracellular domain, *008 was not included in the Luminex method, but antibody responses to both antigens were measured by FCM as expected (Table 2). However, MICA*009 and *049 alleles differ in one amino acid sequence in the intracytoplasmic region in exon 6, and the antibody responses are inconsistent, even though the extracellular domains are identical. These results may be due to the fact that MICA antigens are expressed on the cell surface in FCM, but, in the Luminex method, they are recombinant soluble proteins attached to beads. Previous studies on anti-HLA antibody suggest that conformational changes can occur due to the attachment of recombinant soluble HLA molecules to the beads 31 . In addition, when antibodies are measured using cell lines, non-specific responses to cell surface molecules other than MICA cannot be excluded. More sophisticated studies are needed in the future on these issues.
Initially, in the field of organ transplantation, tests to measure antibodies for histocompatibility mainly used the CDC assay and FCM, but the sensitivity increased dramatically as solid phase assays such as ELISA or Luminex method were developed. These sensitive tests can increase the early detection of patients at risk for transplant rejection, but low levels of antibodies may not be relevant in clinical events. In this study, the Luminex test was more sensitive than the FCM test, and only few serum positives by the Luminex method were found to be positive by FCM (Fig. 3). All 11 serum samples positive by FCM showed more than 10,000 MFI by the Luminex method. Pre-transplant donor-specific HLA antibodies measured by both the CDC and Luminex methods in Korean kidney transplant patients were identified as risk factors for microvascular inflammation in allograft biopsy, but anti-MICA antibodies measured using the Luminex method did not reach statistical significance 34 . In a study on prevalence of MICA antibodies in renal biopsies after transplantation, the MFI value was significantly higher in patients with interstitial fibrosis and tubular atrophy (IFTA) II or III (n = 3, median ± SE, 21,919.0 ± 2581.0) than in patients without IFTA II or III (n = 9, median ± SE, 500.0 ± 155.8) (P = 0.009) 35 .
The results of this study confirmed that the HEK-293T cell line, in which both HLA and MICA/B genes were removed using CRISPR/Cas9 technology, was suitable for measuring antibodies against single MICA antigens in the serum of transplant patients. In FCM, antibodies were tested only for IgG, but simultaneous measurement of IgM and other immunoglobulins and other subclass antibodies may be required. Studies on antibodies against MICA antigens at the cellular level such as CDC and antibody-dependent cytotoxicity in NK cells may help to better understand responses in vivo and increase clinical relevance. Following this study, studies to measure antibodies against MICB antigens are being conducted, and it is expected that the same approach may be applied on various non-HLA antigens in the future.  Production of lentivirus expressing single MICA alleles. Commercially available kits were tested for PBMC RNA isolation using NucleoSpin RNA (Macherey-Nagel, GmbH, Duren, Germany), according to manufacturer protocols were followed for the kit. cDNA was synthesized using the SuperiorScript III cDNA Synthesis kit (Enzynomics, Daejeon, Korea). Amplification of each MICA allele was carried out using PCR with one forward and two reverse primers: 5′-ATG GGG CTG GGC CCG GTC TT-3′ (forward), 5′-CTA GGC GCC CTC AGT GGA GC-3′ (reverse 1), and 5′-CTA GGT GCC CTC AGT GGA GC-3′ (reverse 2). PCR was carried out in 30 μL reaction mixture in 200-μL PCR tubes (Axygen, Hangzhou, China), containing 50 ng genomic DNA, distilled water, 10 × PCR reaction buffer (Kapa Biosystem, MA, USA), 0.6 μL of each 2.5 mM dNTPs, 10 μM primer sets, and 1.5 U of Taq DNA polymerase (Kapa Biosystem). The restriction enzymes BspE1 (R108S; Enzynomics) and Sal1 (R009S; Enzynomics) were used to remove copGFP from the pCDH vector (#CD523A-1; SBI, Palo Alto, CA, USA). The EZ-Fusion Cloning kit (EZ015M; Enzynomics) was used for cloning. Plasmid sequencing was performed using Sanger sequencing (Cosmo Genetech). To produce lentiviruses encoding each molecule (MICA*002, *004, *007, *008, *009, *010, *011, *012, *019, *027, and *049), 5 × 10 6 HEK-293T cells/10 were seeded in T75 flasks. Twenty-four hours later, 10 μg of a cloned MICA pCDH plasmid and lentivirus packaging plasmids (5 μg pMD2.G and 5 μg psPAX2, cat nos. #12259, #12260; Addgene, Cambridge, MA, USA) were cotransfected into HEK-293T cells using Lipofectamine reagent (Invitrogen). At 48 h after transfection, lentiviral supernatants were harvested and filtered through 0.45-μM filters. FCM for detection of anti-MICA antibodies in sera of patients. The recipient serum (5 μL) was incubated with 20 μL of the single MICA cell lines adjusted to 2 × 10 5 /mL at room temperature for 30 min. Anti-human IgG Fc-FITC (F031501; DAKO, Tokyo, Japan) was used. After two washes, 2 μL of anti-human IgG (FITC-F (ab′) 2 anti-human IgG, DAKO) was cross-reacted in a dark room for 30 min. After washing three times, 500 μL of PBS was added to make a cell suspension, and the fluorescence response patterns were compared using FACS Canto or Fortessa.

Generation of cell lines
Luminex method. MICA antibody screening was performed using LABScreen MICA Single Antigen-Group 1 (One Lambda Inc.), LABScreen Mixed (One Lambda Inc.) and analyzed using a LABScan 100 flow analyzer (Luminex 100 System) according to the manufacturer's instructions. The cut-off value of LABScreen Mixed was set to a ratio > 2.7 provided by the manufacturer, and this ratio was calculated by normalizing the bead result value to a negative control value. It was determined through this, and among the total 64 samples, 44 MICA antibodies were determined to be positive and 20 were determined to be negative. The same 64 sam-