We retrospectively reviewed data from 154 recipients who underwent LDLT at Kyushu University between January 1999 and November 2021. Patients with COVID-19 before the study were excluded. As a control, we retrospectively reviewed data from 31 LDLT donors who underwent liver resection at Kyushu University between January 2008 and November 2021 and 19 healthy volunteers who were clinical staff at Kyushu University. We collected sera from recipients and controls after the second and third doses of the SARS-CoV-2 vaccine. Patients who received only Pfizer-BioNTechBNT162b2 for the first, second, and third time were included in this study. Patients who had COVID-19 during the study period were excluded. In Japan, it was recommended that the second vaccine should be taken 21 days or later after the first vaccination and that the third vaccine should be taken at about six months after the second vaccination. The timing of vaccination of patients in this study was optional. The median time between first and second vaccination was 21 days (range: 21-36 days), and the median time between second and third vaccination was 244 days (range: 160-276 days). Since it takes approximately one week for antibody production after vaccination (Dagan et al., 2021), blood samples were taken when patients visited our hospital one week or later after the vaccine was administered. The median time from the second and third vaccinations to serum collection were 83 (range: 7–261 days) and 95.5 (range: 14-159 days) days, respectively. The study protocol was approved by the Institutional Review Board of Kyushu University Hospital (approval number 2020-639). This study was conducted in accordance with the Declaration of Helsinki, 1996. An optout approach was employed to obtain informed consent from our patients and personal information was protected during data collection.

Serum SARS-CoV-2 IgG antibody levels were measured using an ELISA kit (E-EL-E-602, Elabscience, Houston, Texas, USA) according to the manufacturer’s instructions. A total of 100 μL serum samples were diluted 1:20 in sample dilution buffer and incubated at 37°C for 45 min. After washing, 100 μL horseradish peroxidase-conjugated receptor-binding domain antigen working solution was added, and the mixtures were incubated at 37°C for 30 min. After washing, 90 μL of substrate reagent was added, and the mixtures were incubated at 37°C for 15 min. Fifty μL stop solution was added, and the plate was read immediately at 450 nm. The cut-off value was calculated as 0.13 + negative control. The patients, whose antibody titers were below the cut-off value, were defined as non-responder. It is necessary to correct for the difference in absorbance between wells in order to compare antibody titer in each well, because this kit was originally a qualitative detection one. Hence, we divided the absorbance by positive control, and calculated comparative absorbance values.

TMPRSS2-expressing Vero E6 (VeroE6/TMPRSS2) cells were obtained from the Japanese Collection of Research Bioresources Cell Bank (JCRB1819) and maintained in low-glucose Dulbecco’s modified Eagle’s medium (DMEM; Sigma-Aldrich, St. Louis, MO, USA) with 10% fetal bovine serum (FBS) (Biowest, Bradenton, France) and G418 (Nacalai Tesque, Kyoto, Japan). HEK293-C34 cells were gifted by Y Matsuura at Osaka University and maintained in high-glucose DMEM (Nacalai Tesque) with 10% FBS and 10 μg/ml blasticidin (solution) (InvivoGen, California, USA), and the exogenous expression of ACE2 and TMPRSS2 was induced by the addition of doxycycline hydrochloride (1 μg/ml) (Sigma-Aldrich). All the above cells were cultured at 37°C under 5% CO₂.

Recombinant SARS-CoV-2 was generated by circular polymerase extension reaction (CPER) as previously described (Torii et al., 2021). Briefly, nine DNA fragments encoding the partial genome of SARS-CoV-2 (strain WK-521, PANGO lineage A; GISAID ID: EPI_ISL_408667) were prepared by PCR using PrimeSTAR GXL DNA polymerase (Takara). A linker fragment encoding the hepatitis delta virus ribozyme, bovine growth hormone poly A signal, and cytomegalovirus promoter was also prepared using PCR. The corresponding SARS-CoV-2 genomic regions, PCR templates, and primers used for this procedure are summarized in Table S1 . Ten DNA fragments were mixed and used for CPER (Torii et al., 2021). To prepare green fluorescent protein (GFP)-expressing replication-competent recombinant SARS-CoV-2, we used fragment 9, in which the GFP gene was inserted into the ORF7a frame instead of the authentic F9 fragment ( Table S1 ) (Torii et al., 2021).

To prepare rBeta S-GFP, the fragment of the viral genome corresponding to the region of fragment 8 ( Table S1 ) was subcloned from a Beta isolate (strain hCoV-19/Japan/TY8-612-P1/2021; GISAID ID: EPI_ISL_1123289). rBA.1 S-GFP was gifted from K Sato at Tokyo University (Yamasoba et al., 2022). Nucleotide sequences were determined using a DNA sequencing service (Fasmac), and the sequence data were analyzed using ApE.

To produce recombinant SARS-CoV-2 (seed viruses), CPER products were transfected into HEK293-C34 cells using TransIT-LT1 (Takara) according to the manufacturer’s protocol. One day post-transfection, the culture medium was replaced with high-glucose DMEM (Nacalai Tesque) containing 2% FBS, 1% PS, and 1μg/ml doxycycline. At 6–10 days post-transfection, the culture medium was harvested and centrifuged, and supernatants were collected as seed viruses.

The chimeric recombinant SARS-CoV-2 [rB.1.1 S-GFP (Ancestral), rBeta S-GFP(Beta), and rBA.1 S-GFP(Omicron)] ( Figure S1A ) were amplified in Vero E6/TMPRSS2 cells, and the culture supernatants were harvested and stored at −80°C until use. Infectious titers in the culture supernatants were determined using 50% tissue culture infective doses (TCID₅₀). The culture supernatants of cells were inoculated onto VeroE6/TMPRSS2 cells in 96-well plates after serial 10-fold dilution with low-glucose DMEM containing 2% FBS and 1 mg/ml G418, and the infectious titers were determined 96 h post-infection (hpi). All experiments involving SARS-CoV-2 were performed in biosafety level-3 laboratories following standard biosafety protocols approved by Hokkaido University.

7.5×10³ VeroE6/TMPRSS2 cells per well were seeded in 96-well plates and maintained in high-glucose DMEM containing 10% FBS and 1% PS. The cells were then incubated overnight. The next day, each serum was serially diluted 3-fold in the culture medium with a first dilution of 1:10 (final dilution range of 1:10 to 1:21,870). The diluted serum was incubated with 140 TCID₅₀ of the chimeric recombinant SARS-CoV-2 at 37°C for 1 h. Next, the mixture of chimeric recombinant SARS-CoV-2 and serum was added to VeroE6/TMPRSS2 cells in the 96-well plate. At 1 hpi, the cells were washed and replaced with high-glucose DMEM containing 10% FBS and 1% PS.

GFP fluorescence was detected using the ECLIPSE Ts2 (Nikon) after incubating the plates at 37°C for 34–36 h. Then, the luminance of GFP was calculated using Image J. A GFP signal with a luminance value >150 in one field of view was considered positive ( Figure S1B ). The neutralizing antibody titer was defined as the minimum serum dilution at which the GFP signal was positive. The neutralization titer of each serum sample was determined using duplicate assays. Data were initially plotted using GraphPad Prism 9 software (GraphPad Software).

For RT-qPCR, after incubating the plates at 37°C for 34–36 h, the viral RNA was extracted from the supernatants using the PureLink^® RNA Mini Kit (Invitrogen). The sample was used as a template for RT-qPCR performed according to the manufacturer’s protocol using the One Step PrimeScript™ III RT-qPCR Mix (Takara), primers and probe (see Table S1 ). Fluorescent signals were acquired using the StepOnePlus™ Real-Time PCR System (Applied Biosystems). The assay for each serum sample was performed in duplicate, and the 50% neutralization titer (NT₅₀) was calculated using Prism 9 software.

Categorical variables, presented as numbers and percentages, were compared using Pearson’s chi-square test. Based on their distributions, continuous variables were presented as medians with ranges and compared using Student’s t-test. Any variable in the univariate analysis identified as significant (p<0.05) was considered a candidate for the multivariate logistic regression. Ineffectiveness against SARS-CoV-2 vaccination examined by ELISA was used to establish a univariate and multivariate logistic regression model. All statistical analyses were performed using JMP Pro 15 software (SAS Institute, NC, USA) and R software version 3.6.2.

The median control age of the controls and LDLT recipients were 37 (range: 21–60) and 66 (range: 29–84 years) years, respectively. The number of males in the control group was 33 (64.7%). The median duration between vaccination and antibody measurement in the control and LDLT groups were 95.5 (range: 8–261 days) and 78.5 (range: 7–205) days, respectively. Patient characteristics of the LDLT recipients are shown in Table 1 . The LDLT recipients were older and more frequently female than the controls (p<0.0001 and p=0.0172, respectively). The duration between vaccination and antibody measurement was shorter in the LDLT recipients than in the controls (p<0.0001).

SARS-CoV-2 IgG antibody titers in LDLT recipient and control sera were determined using ELISA. The median neutralizing antibody titer was lower in the recipients than in the controls (0.46 vs 1.00, p<0.0001, Figure 1A ). Neutralizing antibodies were induced in all controls by the second vaccine dose, while 39 LDLT recipients (25.3%) had no neutralizing antibodies. We compared the characteristics of LDLT recipients with and without induced antibody production (responders and non-responders, respectively) ( Table 2 ). Non-responders were older and had higher rates of ABO incompatibility, Hepatocellular carcinoma (HCC) history, and mycophenolate mofetil (MMF) treatment, and higher amount of ascites than responders. The number of immunosuppressant therapies is higher in non-responders than in responders. Figure 1B shows the association of various immunosuppressant doses with the level of antibody titers. Multivariate analysis identified age at vaccination, presence of ascites at vaccination, multiple immunosuppressive treatments, and MMF treatment as the factors involved in non-response to the second vaccine dose ( Table 3 ). Additionally, we investigated the correlation between the spike protein IgG antibody titer and the dose of each immunosuppressants and found that the IgG antibody titer was significantly lower in patients receiving a high MMF and prednisolone dose ( Figure 1B ).

To measure the neutralizing activity of sera, we developed a high-throughput neutralizing activity evaluation system. We previously generated a superfolder GFP (sfGFP)-carrying recombinant SARS-CoV-2 using an innovative reverse genetics method ( Figure S1A ) (Torii et al., 2021). This recombinant SARS-CoV-2 was mixed with diluted vaccinated serum and inoculated into the VeroE6/TMPRSS2 cells. The expression of GFP in VeroE6/TMPRSS2 cells was observed by fluorescent microscopy at 34 hpi. The percentage of cells expressing GFP increased with increasing serum dilution rate ( Figure 2A ). Simultaneously, we collected the supernatants of VeroE6/TMPRSS2 cells and quantified viral RNA using RT-qPCR. Similar to the GFP detection system, the quantity of viral RNA also increased at a high serum dilution rate ( Figure 2B ). To confirm the effectiveness of the neutralization assay by GFP detection, we compared NT₅₀ calculated by RT-qPCR and neutralizing activity titer calculated by GFP detection (see Methods) in 15 vaccinated serum samples. NT₅₀ were found to be well correlated with the titer of neutralizing antibody using GFP fluorescent ( Figure 2C ). These results suggest that the simplified method established in this study can be used for neutralization assays.

The neutralizing activities of sera from controls and LDLT recipients against rB.1.1 S-GFP, rBeta S-GFP, and rBA.1 S-GFP were examined. As shown in Figure 3A , the dilution rates of the recipient sera were approximately 3-fold lower against the Ancestral strain and approximately 5-fold lower against the Beta strain than those of the control sera. However, the dilution rate against the Omicron strains was comparable between the two groups. As shown in Figure 3B , the dilution rates of the control and recipient sera against rBeta S-GFP and rBA.1 S-GFP were significantly lower than those against rB.1.1 S-GFP. These results suggest that after the second vaccine dose, the dilution rates against the Ancestral and Beta variants were low in recipient sera compared to those in control sera; however, the dilution rate against the Omicron variant was null in both groups.

In addition, we examined the correlation between the comparative absorbance value and the dilution rates. In Figure S2 , there were significant correlations in Ancestral and Beta strains, while there was no significant correlation in Omicron strain.

We investigated the titers of neutralizing antibodies using ELISA. The titer after the third vaccine dose was elevated in both controls and LDLT recipients ( Figure 4A ). The increase in the median neutralizing antibodies titer from the second to the third vaccine dose was approximately 1.2-fold and more than 2-fold in the controls and LDLT recipients, respectively. Furthermore, approximately 25% of LDLT recipients had no neutralizing antibodies after the second dose, but the rate decreased to only about 5% after the third dose. Next, we evaluated the dilution rates of the control and recipient sera against rB.1.1 S-GFP and rBA.1 S-GFP after the third vaccine dose ( Figure 4B ). No significant difference was observed in the dilution rates against rB.1.1 S-GFP between the controls and LDLT recipients. Additionally, the dilution rate against rBA.1 S-GFP was comparable to that against rB.1.1 S-GFP in both groups ( Figure 4C ).

In addition, we examined the correlation between the comparative absorbance value and the dilution rates. In Figure S3 , there were significant correlations in both Ancestral and Omicron strains after third vaccination.