S309-CAR-NK cells bind the Omicron variants in vitro and reduce SARS-CoV-2 viral loads in humanized ACE2-NSG mice

ABSTRACT

Recent progress on chimeric antigen receptor (CAR)-NK cells has shown promising results in treating CD19-positive lymphoid tumors with minimal toxicities [including graft versus host disease (GvHD) and cytokine release syndrome (CRS) in clinical trials. Nevertheless, the use of CAR-NK cells in combating viral infections has not yet been fully explored. Previous studies have shown that CAR-NK cells expressing S309 single-chain fragment variable (scFv), hereinafter S309-CAR-NK cells, can bind to SARS-CoV-2 wildtype pseudotyped virus (PV) and effectively kill cells expressing wild-type spike protein in vitro. In this study, we further demonstrate that the S309-CAR-NK cells can bind to different SARS-CoV-2 variants, including the B.1.617.2 (Delta), B.1.621 (Mu), and B.1.1.529 (Omicron) variants in vitro. We also show that S309-CAR-NK cells reduce virus loads in the NOD/SCID gamma (NSG) mice expressing the human angiotensin-converting enzyme 2 (hACE2) receptor challenged with SARS-CoV-2 wild-type (strain USA/WA1/2020). Our study demonstrates the potential use of S309-CAR-NK cells for inhibiting infection by SARS-CoV-2 and for the potential treatment of COVID-19 patients unresponsive to otherwise currently available therapeutics.

IMPORTANCE

Chimeric antigen receptor (CAR)-NK cells can be “off-the-shelf” products that treat various diseases, including cancer, infections, and autoimmune diseases. In this study, we engineered natural killer (NK) cells to express S309 single-chain fragment variable (scFv), to target the Spike protein of SARS-CoV-2, hereinafter S309-CAR-NK cells. Our study shows that S309-CAR-NK cells are effective against different SARS-CoV-2 variants, including the B.1.617.2 (Delta), B.1.621 (Mu), and B.1.1.529 (Omicron) variants. The S309-CAR-NK cells can (i) directly bind to SARS-CoV-2 pseudotyped virus (PV), (ii) competitively bind to SARS-CoV-2 PV with 293T cells expressing the human angiotensin-converting enzyme 2 (hACE2) receptor (293T-hACE2 cells), (iii) specifically target and lyse A549 cells expressing the spike protein, and (iv) significantly reduce the viral loads of SARS-CoV-2 wild-type (strain USA/WA1/2020) in the lungs of NOD/SCID gamma (NSG) mice expressing hACE2 (hACE2-NSG mice). Altogether, the current study demonstrates the potential use of S309-CAR-NK immunotherapy as an alternative treatment for COVID-19 patients.

INTRODUCTION

The emergence of new SARS-CoV-2 variants, notably the B.1.1.529 variant (ancestral Omicron, BA.1/BA.1.1), at the end of 2021 caused significant threats to vaccine protection, with a series of mutants escaping from most currently available monoclonal antibodies (1–3). This resulted in an instability of the healthcare infrastructure due to a drastic increase in COVID-related hospitalizations worldwide. Although the clinical symptoms caused by B.1.1.529 are not as severe compared to those caused by the wild-type SARS-CoV-2 (4), unvaccinated and immunocompromised individuals have been shown to be at high risk of developing severe respiratory symptoms. While a few therapeutics have been developed or repurposed to treat COVID since the beginning of the pandemic (5), cell-based immunotherapy has not been seriously explored in treating COVID-19 patients with comorbidities.

As the first line of host defense, natural killer (NK) cells provide intrinsic protection against virally infected cells and tumor cells. NK cells represent approximately 5% to 25% of the total human peripheral blood lymphocytes, and severe COVID patients have significantly reduced NK cell numbers in the peripheral blood. Still, a similar NK cell frequency was observed in the bronchoalveolar lavage fluid (BALF) compared to that of healthy controls (6–9). In addition, the NK cells from severe COVID patients have been shown to undergo significant phenotypic and functional changes, including downmodulation of NK cell-activating receptors, upregulation of NK cell-inhibitory receptors, and poor cytotoxic functions (6, 10, 11). It has also been shown that SARS-CoV-2-infected cells downregulate NKG2D ligands to escape NK cell immunosurveillance (12,13). Importantly, the immunoglobulin G (IgG) persists approximately 3 months post SARS-CoV-2 infection (14), whereas chimeric antigen receptor (CAR)-NK cells can last for more than a year based on a recent clinical trial (15), which suggests a prolonged protection and potential prevention strategy for a future outbreak. Our lab has previously established a superior NK cell expansion system using peripheral blood mononuclear cells (PBMCs), yielding an average of a 10⁴-fold increase in NK cell expansion with high NK cell purity and memory-like phenotypes (16). Our development in NK cell expansion provides a platform for “off-the-shelf” NK cell and CAR-NK cell products and paves the way to engineering long-lived memory-like NK cells. We have thus proposed the use of CAR-NK cells expressing the S309 single-cell fragment variable (scFv) domain (17) to increase NK cell number in severe COVID patients (18), prolong the protection period, specifically target SARS-CoV-2-infected cells, prevent subsequent infection by future SARS-CoV-2 variants, and broaden the use of cell-based immunotherapy in diseases.

Monoclonal neutralizing antibody (NAb) S309 (19) was originally isolated from a SARS patient and shown to broadly neutralize the circulating isolates of both SARS-CoV-1 and SARS-CoV-2 (20). The S309 antibody binds outside the ACE2-binding site and is categorized as class 3 NAbs against SARS-CoV2 (21–23). Of note, S309 NAb has been redesigned as a monoclonal antibody therapy for the treatment of COVID-19 (sotrovimab, VIR-7831). Monoclonal S2E12 NAb was subsequently discovered and isolated from a convalescent COVID-19 patient and shown to protect hamsters against the SARS-CoV-2 challenge (19). Unlike S309 NAb, S2E12 NAb neutralizes SARS-CoV-2 by binding to the receptor-binding motif (RBM) and belongs to the class I NAb (19). Critically, although many currently available monoclonal antibodies, including S2E12 NAb, have lost their neutralization ability against B.1.1.529 variants, the activity of S309 NAb has largely been retained (2, 24, 25). Overall, there is a consensus that S309 NAb can broadly neutralize most, if not all, SARS-CoV-2 variants of concerns (VOCs).

In this study, we showed the binding ability of S309-CAR-NK cells against SARS-CoV-2 variants by comparing them with S2E12-CAR-NK cells and determined the cytotoxic functions against A549 cells expressing mutant spike proteins. In addition, we tested the efficacy of S309-CAR-NK cells in vivo using humanized angiotensin-converting enzyme 2 (ACE2) knock-in NOD Scid Gamma (NSG) mice (henceforth hACE2-NSG mice) challenged with SARS-CoV-2 wild-type (strain USA/WA1/2020). Collectively, our results demonstrate the potential application of CAR-NK cells in treating SARS-CoV-2, preparing for a future outbreak, and other infectious diseases in the future.

RESULTS

Generation of different B.1.617.2, B.1.621, and B.1.1.529 pseudotyped virus (PVs)

We previously reported the efficacy of S309-CAR-NK cells against SARS-CoV-2 wild-type in vitro (17). In this study, we wanted to evaluate whether S309-CAR-NK cells are effective at recognizing other SARS-CoV-2 variants. We first produced different variants of SARS-CoV-2 PVs by transfecting 293T cells with required plasmids (Table 1) and performed a 1:2 serial dilution for PVs on 293T-hACE2 cells; note that PVs without the expression of the spike protein served as a negative control (Fig. 1A). By incubating SARS-CoV-2 PVs with 293T-hACE2 cells followed by anti-spike staining, we observed that SARS-CoV-2 WTPVs had less binding to 293T-hACE2 cells compared to B.1.617.2 (Delta) and B.1.621 (Mu) (Fig. 1B). Since the PVs contain the green fluorescent protein (GFP), therefore cells that are infected by PVs will express the GFP; we subsequently evaluated the difference in SARS-CoV-2 variant infection by incubating different variants of SARS-CoV-2 PVs with 293T-hACE2 cells for 72 hours followed by flow cytometry (Fig. 1C). We observed an increase in the GFP-positive 293T-hACE2 population when 293T-hACE2 cells were incubated with B.1.621 PV (Fig. 1C). However, qRT-PCR results showed an insignificant increase in RNA levels (Fig. 1D). Collectively, we show that we can produce different variants of SARS-CoV-2 PVs for later experiments.

TABLE 1.

SARS-CoV-2 spike-containing plasmids for SARS-CoV-2 PV production

SARS-CoV2 PVs	Plasmid names	Mutations
SARS-CoV2 S (B.1.617.2)	pLV-spike V8-Delta (from Invivogen)	T19R, T95I, G142D, E156G, F157, R158, L452R, T478K, D614G, P681R, and D950N
SARS-CoV2 S (B.1.617.2)	pCAG-dUTR-Mu (from Dr. Wenhui Hu)	T95I, Y144S, Y145N, R346K, E484K, N501Y, D614G, P681H, and D950N
SARS-CoV2 S (B.1.1.529)	pCDNA3.1-SARS-CoV2- Flag-S-Flag-B.1.1.529 (from Dr. Shan-Lu Liu)	A67V, T95I, G142D, V143, Y144, Y145, N211I, L212V, V213P, R213E, P214, E215, G339D, R346K, S371L, S373P, S375F, K417N, N440K, G446S, S477N, T478K, E484A, Q493R, G496S, Q498R, N501Y, Y505H, T507K, D614G, H655Y, N679K, P681H, N764K, D786Y, N856K, Q954H, N969K, and L981F

Fig 1.

Characterization of SARS-CoV-2 PVs. (A) Representative histogram of PVs titering against 293T-hACE2 cells. 293T-hACE2 cells were incubated with different dilutions of SARS-CoV-2 PV variants for 2 hours at 37°C. Cells were collected for anti-spike staining followed by flow cytometry. (B) Different dilutions of PVs. 293T-hACE2 cells were incubated with different dilutions of SARS-CoV-2 PV variants for 2 hours at 37°C. (C) Quantification of PV-infected 293T-hACE2 cells using flow cytometry. PV-infected 293T-hACE2 cells were checked for GFP expression. TU/mL is calculated as [(number of 293T-hACE2 cells seeded) * (percent of 293T-hACE2 cells expressing GFP)] / (volume of SARS-CoV-2 PVs used). (D) Quantification of PV-infected 293T-hACE2 cells using qRT-PCR. 293T-hACE2 cells were incubated with SARS-CoV-2 PV variants for 72 hours at 37°C. RNA was isolated from adherent cells for qRT-PCR. Error bars represent ± SEM. The experiment was repeated three times. One-way ANOVA was employed in panels B, C, D, and E. ns, P ≥ 0.05; *, P ≤ 0.05; **, and P ≤ 0.01.

Establishment of primary CAR-NK cells bearing S309 or S2E12 neutralizing antibody

We previously established S309-CAR-NK cells and demonstrated their efficacy in targeting cells expressing wild-type (WT) spike and spike-bearing D614G, E484K, and N501Y mutations in vitro (17). Here, we generated primary CAR-NK cells expressing S2E12 scFv (hereinafter S2E12-CAR-NK cells) as well as CAR-NK cells expressing S309 (S309-CAR-NK cells). In brief, we codon-optimized the V domains of the heavy and light chains (V_H and V_L), respectively, of S309 or S2E12 and subcloned the scFv into the SFG vector followed by the human IgG (CH2-CH3) tag in the extracellular domain, CD28 transmembrane domain, and CD28, 4–1BB, CD3ζ, and IL-15 secretion intercellular domains (Fig. 2A). We observed that both S309-CAR-NK cells and S2E12-CAR-NK cells have similar levels of expressions of CAR in both mean fluorescence intensities (MFI) and transduction efficiency in percentage (%) determined by flow cytometry (Fig. 2B). We were able to consistently expand and generate SARS-CoV-2-specific-CAR-NK cells with an average of 90% NK cell purity and 70% CAR expressions from the PBMCs of eight donors (Fig. 2C). Therefore, we successfully generated both S309-CAR-NK cells and S2E12-CAR-NK cells.

Fig 2.

CAR-NK cells specifically bind to different variants of SARS-CoV-2 PVs via the scFv. (A) Schematic of CAR-NK constructs in the SFG vector. (B) Representative dot plot showing CAR expression on expanded primary NK cells. Untransduced NK cells and CAR-NK cells were stained with anti-CD56, anti-CD3, and anti-hIgG followed by flow cytometry. (C) The NK cell purity and CAR expression from each CAR-NK cell expansion were determined by flow cytometry. Each dot represents a biological replication. Unpaired t-tests were used, ns, P ≥ 0.05. (D) CAR-NK cells effectively bind to SARS-CoV-2 PVs, except for CoV-2 Sο. Untransduced NK cells or CAR-NK cells or 293T-hACE2 were incubated with SARS-CoV-2 PV variants for 2 hours at 37°C, followed by flow cytometry. (E) CAR-NK cells effectively bind to the His-Sο protein. Untransduced NK cells or CAR-NK cells were incubated with His-Sο for 1 hour at 4°C, followed by flow cytometry. (F) Quantification of effective binding to His-Sο by CAR-NK cells. Error bars represent ± SEM. Three biological experiments were carried out. One-way ANOVA was employed. ns, P ≥ 0.05; ****, P < 0.0005. Representative dot plots of S309-CAR-NK cells or S2E12-CAR-NK cells binding to His-SWT or His-Sο protein. Bar graph showing the binding of effector cells to recombinant spike proteins from three technical replicates. One-way ANOVA was employed. ns, P ≥ 0.05; *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; and ****, P ≤ 0.0001.

S309-CAR-NK cells bind to b.1.617.2, B.1.621, and B.1.1.529 PVs

After the generation of both S309-CAR-NK cells and S2E12-CAR-NK cells, we next examined whether the CAR-NK cells expanded from PBMCs could recognize different variants of SARS-CoV-2 PVs. We previously demonstrated that S309-CAR-NK cells effectively bind to SARS-CoV-2 WT PVs (17). By using the previously established methodology (Fig. S1), we confirmed that both S309-CAR-NK cells and S2E12-CAR-NK cells were effective at binding to the spike protein of SARS-CoV-2 B.1.617.2 (Delta), B.1.621 (Mu), and B.1.1.529 (Omicron) PVs (Fig. 2D). Our data showed that both S309-CAR-NK cells and S2E12-CAR-NK cells were unable to bind SARS-CoV-2 B.1.1.529 PVs; however, the cells are shown to bind to the recombinant spike protein of B.1.1.529 containing His tag (His-Sο) using His-SWT as a positive control (Fig. 2F). Collectively, our results showed similar binding activity of S309-CAR-NK cells and S2E12-CAR-NK cells for the spike protein derived from B.1.617.2, B.1.621, and B.1.1.529.

S309-CAR-NK cells and S2E12-CAR-NK cells effectively prevent PVs of SARS-CoV-2 variants from infecting 293T-hACE2 cells

We next designed a competitive binding assay to address the ability of CAR-NK cells to prevent SARS-CoV-2 from infecting cells expressing ACE2 receptors (Fig. S2). We first diluted expanded primary NK cells, S309-CAR-NK cells, and S2E12-CAR-NK cells and preincubated them with SARS-CoV-2 PVs. Given that NK cells have high basal killing against 293T-hACE2 cells, we therefore incubated 293T-hACE2 cells with the SARS-CoV-2 PV/NK cell mixture for 30 minutes at 4°C to minimize the cytotoxicity of NK cells against 293T-hACE2 cells before assessing the binding of SARS-CoV-2 PVs. By increasing the ratio of effector:293T-hACE2 cells, we observed an inverse relationship between SARS-CoV-2 PV binding to S309-CAR-NK cells or S2E12-CAR-NK cells versus 293T-hACE2, indicating that both S309-CAR-NK cells and S2E12-CAR-NK cells can competitively bind and prevent SARS-CoV-2 wild-type PV from binding to and infecting 293T-hACE2 cells (Fig. 3A). Though not statistically significant, we observed that S2E12-CAR-NK cells neutralized SARS-CoV-2 WT PVs more efficiently than S309-CAR-NK cells (Fig. 3B). We subsequently tested the SARS-CoV-2 variants, including B.1.617.2, B.1.621, and B.1.1.529, and observed that both S309-CAR-NK cells and S2E12-CAR-NK cells effectively neutralized all pseudotyped SARS-CoV-2 variants (Fig. 3C through F). To ensure that S309-CAR-NK cells and S2E12-CAR-NK cells are specific to SARS-CoV-2, we included CAR-NK cells expressing immunoglobulin kappa light chain (kappa-CAR-NK cells) as an additional negative control (Fig. S3). Unexpectedly, we observed a limited competitive binding of kappa-CAR-NK cells against SARS-CoV-2 (Fig. S3). We also included the sera obtained from healthy and unvaccinated convalescent COVID-19-infected donors from the first SARS-CoV-2 wave to further validate the reliability of our assay (Fig. S4). Consistent with other studies, the convalescent COVID-19-infected donor plasma highly neutralized other SARS-CoV-2 variants, except for B.1.1.529, while the plasma obtained from healthy donors poorly neutralized SARS-CoV-2 (Fig. S4). Our results showed that both S309-CAR-NK cells and S2E12-CAR-NK cells retained the competitive binding ability against the tested SARS-CoV-2 variants (Fig. 3). Hence, we have established a reliable CAR-NK cell competitive binding assay that showed the binding ability of both S309-CAR-NK cells and S2E12-CAR-NK cells against B.1.617.2, B.1.621, and B.1.1.529 variants.

Fig 3.

S309-CAR-NK cells competitively bind to B.1.617.2, B.1.621, and B.1.1.529 PVs (A) Representative dot plots showing S309-CAR-NK cells and S2E12-CAR-NK cells against SARS-CoV-2 SWT PV. (B–F) Briefly, various dilutions of untransduced PBNK cells, S309-CAR-NK cells, or S2E12-CAR-NK cells were preincubated with PV for 2 hours at 37°C. The cell/PV mixture was subsequently incubated with a fixed number of 293T-hACE2 cells for 30 minutes at 4°C. Cells were stained with anti-CD45 and anti-spike, followed by flow cytometry analysis. The competitive binding percentage (%) is calculated as 100 – ((% of CD45^- Spike⁺ population)/(% of 293T-hACE2 incubated with SARS-CoV-2 PV alone) ×100) for SARS-CoV2 WT (B), b.1.617.2 (C), B.1.621 (D), and B.1.1.529 PVs (E). S309-CAR-NK cells and S2E12-CAR-NK cells have unchanged neutralization against SARS-CoV-2 Sο. Error bars represent ±SEM of at least n = 3 replicates. Student’s t-tests were employed to compare the binding ability of S309-CAR-NK or S2E12-CAR-NK with untransduced PBNK. ns, P ≥ 0.05; *, P ≤ 0.05; # indicates P ≤ 0.05 for S309-CAR-NK cells vs untransduced PBNK; however, ns, P ≥ 0.05, for S2E12-CAR-NK cells vs untransduced PBNK. One-way ANOVA was employed with subsequent Student’s t-tests to compare the binding abilities among different SARS-CoV-2 variants. ns, P ≥ 0.05; *, P ≤ 0.05; **, P ≤ 0.01.

S309-CAR-NK cells have higher killing functions compared to S2E12-CAR-NK cells against A549 cells expressing the spike protein of B.1.617.2, B.1.621, and B.1.1.529 variants

Next, we evaluated the cytotoxic functions of S309-CAR-NK cells and S2E12-CAR-NK cells against A549 cells expressing the spike protein of B.1.617.2, B.1.621, and B.1.1.529 variants, using the A549 cell line as a negative control and A549 expressing spike from SARS-CoV-2 wild-type as a positive control. The spike expressions of the engineered A549 cell lines were confirmed by flow cytometry (Fig. 4A). Using the degranulation assay, we showed that both S309-CAR-NK cells and S2E12-CAR-NK cells degranulated against all target cell lines expressing the spike protein, which was higher than that of the untransduced PBNK cells (Fig. 4B and C). These results showed that the cytolytic functions of both S309-CAR-NK cells and S2E12-CAR-NK cells are unaltered at targeting the spike protein derived from different SARS-CoV-2 variants. Although S309 and S2E12 NAbs recognize different epitopes of RBD, we found similar capabilities of S309-CAR-NK cells and S2E12-CAR-NK cells in degranulation (Fig. 4C). Unexpectedly, we observed high levels of degranulation in untransduced PBNK cells in some donors (Fig. 4C).

Fig 4.

S309-CAR-NK cells effectively target A549 cell lines expressing the spike protein of B.1.617.2, B.1.621, and B.1.1.529 variants (A) Confirmation of spike expressions on the A549 target cell line by flow cytometry. (B) Representative contour plots showing degranulation of different effector cells against respective A549 cell lines. (C) Bar graph showing the percentage (%) of effector cells expressing the CD107a molecule from n = 6 biological donors. One-way ANOVA was employed, ns, P ≥ 0.05; *, P ≤ 0.05; **, P ≤ 0.01. (D) Representative graphs showing the specific lysis (%) of effector cells against different spike-expressing A549 cells. (E) Bar graph showing the specific lysis in percentage (%) of effector cells against A549 cells expressing different spike variants at the highest effector:target (E:T) ratio. n = 3 technical replicates. One-way ANOVA was employed, ns, P ≥ 0.05; *, P ≤ 0.05; **, P ≤ 0.01.

We next performed a chromium release assay to validate our results. As expanded primary NK cells possess an intrinsic recognition against A549, a lung carcinoma cell line, we blocked the NKG2D activating receptor before performing the chromium release assay to accurately quantify the cytolytic functions of S309-CAR-NK cells and S2E12-CAR-NK cells against the spike protein. Consistently, we found that S309-CAR-NK cells killed A549 cells expressing the spike protein of B.1.617.2, B.1.621, and B.1.1.529 variants efficiently, compared to that of S2E12-CAR-NK cells (Fig. 4D and E). These results altogether indicate that S309-CAR-NK cells may effectively and specifically target SARS-CoV-2-infected cells.

S309-CAR-NK cell-treated hACE2-NSG mice have reduced SARS-CoV-2 viral loads in the lungs

We next evaluated the efficacy of S309-CAR-NK cells in hACE2-NSG mice. We first challenged hACE2-NSG mice with 1 × 10⁵ plaque-forming units (PFU) of SARS-CoV-2 wild type (strain USA/WA1/2020) intranasally (24), immediately followed by intravenous (i.v.) administration of PBS vehicle control or 5 × 10⁶ expanded primary S309-CAR-NK cells with a total of three doses for each group (Fig. 5A); mice were closely monitored for their body weights for 10 days. Interestingly, we did not observe a significant clinical deterioration as determined by the body weight in either group (Fig. 5B). On day 10 post-infection, all mice were sacrificed for the collections of lungs and brains. Using the SARS-CoV-2 N gene molecular beacon probe, we showed that S309-CAR-NK cell-treated mice had reduced SARS-CoV-2 viral copies in the lungs, with an average of ten fold reduction (Fig. 5C). The left panel of Fig. 5C is normalized to the N protein of SARS-CoV-2 viral particles, whereas the right panel of Fig. 5C is normalized to the N protein of SARS-CoV-2 viral particles followed by GAPDH normalization for the lung tissues. Both analysis methods show significantly decreased SARS-CoV-2 viral copies in S309-CAR-NK cell-treated hACE2-NSG mice. Altogether, our in vitro data are consistent with those of the SARS-CoV-2 wild-type-challenged hACE2-NSG mice, showing that S309-CAR-NK cells are effective at reducing the viral loads of SARS-CoV-2 in the lungs, suggesting that CAR-NK cell therapy may be useful for COVID-19 treatment.

Fig 5.

Efficacy of S309-CAR-NK cells in reducing the SARS-CoV-2 viral loads in the lungs of hACE2-NSG mice. (A) Schematic experimental design. Briefly, hACE2-NSG mice were challenged with 1 × 10⁵ plaque-forming units (PFU) of SARS-CoV-2 intranasally, immediately followed by 5 × 10⁶ of S309-CAR-NK cells in 50 µL of PBS retroorbitally. Intravenous injections of two additional doses on days 2 and 4 post-infection. The study was terminated on day 10 for the collection of lung tissues. (B) Body weight loss of hACE2-NSG mice showed no significant differences between the two groups. (C) Collected lung tissues were processed via TRIzol for RNA extraction. The SARS-CoV-2 N gene was used for the molecular beacon probe. The GAPDH gene was used for normalization to calculate the fold change, where log₂Fc = 2^-ΔCt. n = 5 per group. Error bars ±SD, Student’s t-tests were employed, *, P ≤ 0.05.

DISCUSSION

Studies showed that the B.1.1.529 (Omicron) variant causes less severe symptoms compared to the previous VOCs (4). However, it is not yet known whether the relative mildness is due to the intrinsic properties of the B.1.1.529 variant or the prevalence of vaccine protection. Together, it indicates the relevance and imperative to explore alternative treatments for severe COVID-19 patients.

Although we showed decreased SARS-CoV-2 viral loads in the lungs of hACE2-NSG mice, the reduction of viral loads in treated mice was not as remarkable as we anticipated. Our results can be explained as follows:

1. We observed a more drastic decrease in body weight in the untreated group compared to the S309-CAR-NK cell-treated group from days 4 to 6 post-infection, indicating the protective mechanism of S309-CAR-NK cells against SARS-CoV2; however, all mice were quickly recovered by the increase in body mass after day 6. Furthermore, there were no deaths during the infection in the untreated group, suggesting that SARS-CoV-2-challenged hACE2-NSG mice may have mostly naturally eradicated SARS-CoV-2 viral particles by the day of termination or day 10 post-infection.

2. Unlike the commonly used K18-hACE2 mice in C57BL/6J background that uses the keratin 18 promoter to overexpress the hACE2 in the upper and lower respiratory tract (18, 26), the hACE2 knockin (KI)-NSG mice are generated by replacing mouse ACE2 using the endogenous Ace2 promoter (27). Studies show that by day 4 post-infection, SARS-CoV-2 viral copies are significantly reduced in the lungs compared to the nasal concha. In addition, 25% of the infected hACE2-KI-NSG mice have no SARS-CoV-2 viral copies detected in the lungs, indicating that the upper respiratory tract is more infected in hACE2-KI-NSG mice, while K18-hACE2 mice show more infection in the lower respiratory system. Given that the lungs were collected in our animal studies, our data are consistent with previously published data, showing a natural eradication of SARS-CoV-2 viral loads in the lungs of some hACE2-NSG mice.

In addition to the B.1.617.2 (Delta), B.1.621 (Mu), and B.1.1.529 (Omicron), we did not test other variants in this study; however, we provide the “proof-of-concept” of developing “off-the-shelf” CAR-NK therapy for infectious diseases and preparing an additional strategy for future coronavirus-mediated pandemic outbreak.

MATERIALS AND METHODS

Antibodies

All antibodies used are listed in Table S2

Plasmids

The sources of the original plasmids were obtained as shown in Table 1. All primers used for cloning purposes to engineer A549-expressing spike cell lines are listed in Table S1.

Cell culture

Adherent cell lines including 293T (purchased from ATCC), A549 (from Dr. Wei-Xing Zong at Rutgers University Ernest Mario School of Pharmacy), A549-expressing spike protein of B.1.617.2, B.1.621, and B.1.1.529 variants were cultured in D-10 media [complete DMEM (Corning) containing 10% fetal bovine serum (Corning) and 100 U/mL penicillin–streptomycin (Gibco)]. 293T-hACE2 cells (from Dr. Abraham Pinter at the Public Health Research Institute) were cultured in D-10 media containing 1 µg/mL puromycin. Primary NK cells and CAR-NK cells were cultured in R-10 media (RPMI-1640 (Corning) with 10% fetal bovine serum, 100 U/mL penicillin–streptomycin) supplemented with 200 U/mL recombinant hIL-2 and 5 ng/mL recombinant hIL-15 (Peprotech).

CAR-NK construct cloning

The DNA fragments of both S309-specific and S2E12-specific scFv domains were codon-optimized by GENEWIZ. To subclone into the SFG expression vector, both the PCR products and SFG plasmid were digested with SalI and Pfl23II digestion enzymes. All constructs were verified by sequencing.

Primary NK cell expansion and CAR-NK cell generation

All human-related work was approved by the Rutgers University Institutional Review Board (IRB). Briefly, PBMCs were isolated from buffy coats purchased from the New York Blood Center (NYBC) using lymphocyte separation medium by centrifugation. To expand primary NK cells, 5 $\times$ 10⁶ PBMCs were cocultured with 10 $\times$ 10⁶ 221-mIL21 cells with R-10 media supplemented with 200 U/mL recombinant hIL-2 and 5 ng/mL recombinant hIL-15 in a G-REX well. The medium was changed every 3–4 days for NK cell maintenance. To generate S309-CAR-NK cells or S2E12-CAR-NK cells, 293T cells were transfected using the lentiviral system described previously for 48–72 hours at 37°C. On day 4, expanded primary NK cells were transduced with the harvested lentiviral supernatant in a 24-well plate coated with RetroNectin at 50 µg/well. Transduced cells were collected after 48–72 hours at 37°C and continued culturing with R-10 media supplemented with 200 U/mL recombinant hIL-2 and 5 ng/mL recombinant hIL-15 in a G-REX well. The medium was changed every 3–4 days for NK and CAR-NK cell maintenance.

SARS-CoV-2 PV production

Plasmids used to generate different variants of SARS-CoV-2 PVs are listed in Table 1. SARS-CoV-2 PVs were produced as previously described (17). Briefly, 293T cells were transfected with the indicated S mutant vector using a lentiviral 4-plasmid system for 72 hours at 37°C; the empty virion PV only contained VSV-G (control envelope protein), pLP1 (pCMV/HBG-Gag/Pol), pLP2 (pRSV-Rev), and pCMV-luciferase-ecoGFP. SARS-CoV-2 PVs were diluted at 1:2 dilutions and incubated with 293T-hACE2 for 2 hours at 37°C before staining with anti-spike (SinoBiological), followed by an appropriate fluorophore-conjugated secondary antibody for flow cytometry.

SARS-CoV-2 PV infection assay

SARS-CoV-2 PVs were incubated with 5 $\times$ 10⁴ 293T-hACE2 cells overnight at 37°C. The medium was changed to D-10 media without penicillin–streptomycin and continued culturing for an additional 72 hours at 37°C. Infected 293T-hACE2 cells were assessed using flow cytometry to measure GFP expression. Transduction units per mL (TU/mL) were calculated as [(number of 293T-hACE2 cells seeded) * (percent of infected 293T-hACE2 cells)]/volume of SARS-CoV-2 PVs.

SARS-CoV-2 PV binding assay

SARS-CoV-2 PVs were incubated with 1 $\times$ 10⁵ untransduced NK cells, S309-CAR-NK cells, S2E12-CAR-NK cells, or 293T-hACE2 cells (positive control group) for 2 hours at 37°C. Cells in media only were used as negative controls. 293T-hACE2 cells were collected and stained with anti-spike (SinoBiological) followed by the fluorophore-conjugated secondary antibody. NK cells were stained with anti-CD56, anti-CD3, anti-hIgG, and anti-spike followed by the secondary antibody. The binding efficiency was assessed by flow cytometry.

SARS-CoV-2 protein-binding assay

Briefly, 1 $\times$ 10⁵ untransduced PBNK cells or CAR-NK cells were incubated with either 100 ng of His-SWT (SinoBiological) or His-Sο (SinoBiological) in 100 µL of PBS containing 0.5 mM MgCl₂ and 0.9 mM CaCl₂ on ice for 1 hour. Cells were washed thrice with PBS and stained with anti-CD56, anti-CD3, anti-hIgG, and anti-His, followed by flow cytometry.

SARS-CoV-2 PV competitive binding assay using CAR-NK cells

293T-hACE2 cells were seeded at 1 × 10⁵ cells in a 96-well plate at 37°C overnight. Untransduced PBNK cells, S309-CAR-NK cells, S2E12-CAR-NK cells, or human sera were serially diluted in R-10 media and incubated with various SARS-CoV-2 PVs for 2 hours at 37°C. The SARS-CoV-2 PV/cell mixture or SARS-CoV-2 PV/human serum mixture was subsequently incubated with preseeded 293T-hACE2 cells at 4°C for 30 minutes to reduce NK cell killing against target cells. 293T-hACE2 cells incubated with SARS-CoV-2 PV alone at 4°C for 30 minutes were used as positive controls. All cells were collected, centrifuged, and stained with anti-spike and anti-CD45 before flow cytometry. Cells were gated for the CD45^- spike⁺ population, where its percentage was divided by the percentage of 293T-hACE2 cells incubated with SARS-CoV-2 PV alone.

Generation of stable A549 cell lines expressing the spike protein from B.1.617.2, B.1.621, and B.1.1.529 variants

All primers and plasmids used for the subcloning purpose are listed in Table S1. Plasmids for each variant were purchased from Invitrogen, and the PCR products were obtained and subcloned into the SFG vector using In-Fusion followed by transformation and sequencing for the correct clone. A549 cells were subsequently transduced using RetroNectin (purified in-house) as previously described (17). A549 cells expressing the spike protein of B.1.617.2, B.1.621, and B.1.1.529 variants were then sorted using anti-spike 72 hours post-transduction.

CD107a assay

Effector cells were cocultured with target cells at a 1:1 ratio for 2 hours at 37°C in the presence of GolgiStop. Cells were subsequently washed and stained for anti-CD56, anti-CD3, anti-hIgG, and anti-CD107a, followed by flow cytometry.

Chromium release assay

1 × 10⁵ cells /mL of target cells was incubated with 10 µCi of Cr⁵¹ for 2 hours at 37°C. Excess Cr⁵¹ was washed off. Target cells were subsequently cocultured with different ratios of effector cells for an additional 4 hours at 37°C. Thirty microliters of the supernatant was transferred to 96-well Luma plates (PerkinElmer) to air-dry overnight. Values were read using a MicroBeta plate reader (PerkinElmer). To calculate the (%) specific lysis,${\displaystyle (\frac{\mathrm{v}\mathrm{a}\mathrm{l}\mathrm{u}\mathrm{e}-\mathrm{s}\mathrm{p}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{e}\mathrm{o}\mathrm{u}\mathrm{s}\mathrm{r}\mathrm{e}\mathrm{l}\mathrm{e}\mathrm{a}\mathrm{s}\mathrm{e}}{\mathrm{a}\mathrm{r}\mathrm{e}\mathrm{l}\mathrm{e}\mathrm{a}\mathrm{s}\mathrm{e}-\mathrm{s}\mathrm{p}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{e}\mathrm{o}\mathrm{u}\mathrm{s}\mathrm{r}\mathrm{e}\mathrm{l}\mathrm{e}\mathrm{a}\mathrm{s}\mathrm{e}})\times 100.}$

Animal studies

The SARS-CoV-2 challenge experiment was carried out in a BSL3 facility at Rutgers Public Health Research Institute. hACE2-KI-NSG mice were generated by replacing the mouse ACE2 with human ACE2 using the endogenous Ace2 promoter in-house at Rutgers, New Brunswick. Twenty- to 32-week-old male and female mice were randomized but not blinded for the challenge of 10⁵ PFU of SARS-CoV-2 (USA/WA1/2020) intranasally, immediately followed by 100 µL of PBS or 5 × 10⁶ S309-CAR-NK cells per mouse, three doses in total every other day. Mice were weighed every day at approximately similar times. On day 10 of the experiment, all mice were sacrificed for the collections of lungs and brains, where tissues were kept in TRIzol or 4% PFA for 72 hours in BSL3 before further processing.

qRT-PCR for SARS-CoV-2-challenged mice

Lung tissues were homogenized with a Beadbeater (VWR) followed by RNA purification for qRT-PCR analysis. qRT-PCR assays were carried out as described previously. Briefly, each reaction contains 100 nM CoV-N forward primer, 500 nM CoV-N reverse primer, 250 nM CoV-N molecular beacon probe, 1  ×  TaqPath 1-step RT-qPCR Master Mix (Thermo Fisher Scientific), and 2 µL of the RNA template. The one-step qRT-PCR reactions were run for 10 min at 53°C, 2 min at 95°C followed by 15 sec at 95°C, and 1 min at 58°C for 45 thermal cycles. Mouse GAPDH was used as a housekeeping control gene using SYBR Green for qRT-PCR with 2 min at 95°C followed by 95°C for 15 sec and 60°C for 50 sec for 45 thermal cycles.

Flow cytometric analysis

Flow cytometry data were collected using the following cytometers: LSRII (BD Biosciences), LSRFortessa (BD Biosciences), or Accuri C6 (BD Biosciences). Data were analyzed by FlowJo analysis software.

Statistical analysis

All experiments were performed at least three times. All statistical analyses were conducted using Prism, version 8 (GraphPad Software).

DATA AVAILABILITY

The raw data supporting the conclusions will be made available by the authors without undue reservation.

ETHICS APPROVAL

All animal studies were approved by the Rutgers Institutional Animal Care and Use Committee (IACUC).

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/jvi.00038-24.

Supplemental material. jvi.00038-24-s0001.docx.

Tables S1 to S3; Fig. S1 to S4.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.