Expanded clinical-grade NK cells exhibit stronger effects than primary NK cells against HCMV infection

Abstract

Cytomegalovirus (CMV) reactivation remains a common complication and leads to high mortality in patients who undergo allogeneic hematopoietic stem cell transplantation (allo-HSCT). Early natural killer (NK) cell reconstitution may protect against the development of human CMV (HCMV) infection post-HSCT. Our previous data showed that ex vivo mbIL21/4-1BBL-expanded NK cells exhibited high cytotoxicity against leukemia cells. Nevertheless, whether expanded NK cells have stronger anti-HCMV function is unknown. Herein, we compared the anti-HCMV functions of ex vivo expanded NK cells and primary NK cells. Expanded NK cells showed higher expression of activating receptors, chemokine receptors and adhesion molecules; stronger cytotoxicity against HCMV-infected fibroblasts; and better inhibition of HCMV propagation in vitro than primary NK cells. In HCMV-infected humanized mice, expanded NK cell infusion resulted in higher NK cell persistence and more effective tissue HCMV elimination than primary NK cell infusion. A clinical cohort of 20 post-HSCT patients who underwent adoptive NK cell infusion had a significantly lower cumulative incidence of HCMV infection (HR = 0.54, 95% CI = 0.32–0.93, p = 0.042) and refractory HCMV infection (HR = 0.34, 95% CI = 0.18–0.65, p = 0.009) than controls and better NK cell reconstitution on day 30 post NK cell infusion. In conclusion, expanded NK cells exhibit stronger effects than primary NK cells against HCMV infection both in vivo and in vitro.

Introduction

Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is an effective treatment for malignant hematologic disease [1]. Human cytomegalovirus (HCMV) is a virus that causes chronic infection and maintains long-term latency in healthy people but can be reactivated in the organ/tissue/cells of hosts by immunosuppressive treatment. Therefore, HCMV reactivation remains a common complication and increases mortality in patients who undergo allo-HSCT [2–7]. Antivirals targeting CMV, such as ganciclovir, foscarnet, and letermovir, have improved outcomes among HSCT patients at risk for CMV disease. However, the rate of CMV infection after transplantation is still 37.5–80% in all patients, even with antiviral prophylaxis against CMV [8–10], including in patients who undergo haploidentical HSCT [11, 12]. In addition, the long-term use of antiviral agents may cause CMV strains to become drug-resistant and has severe side effects, including marrow suppression and renal impairment. Therefore, the development of safe and effective methods for CMV prophylaxis is needed.

Natural killer (NK) cells are major components of the innate immune system. Purified NK cells can inhibit CMV replication in vitro [13]. The absence of NK cells in vivo has been shown to increase virus synthesis and exacerbate virus-induced hepatitis in both immunocompetent and immunodeficient mice [14–16]. Furthermore, in murine cytomegalovirus (MCMV)-infected mice, virus-specific “memory” NK cells proliferate and reside in organs for several months, and they can rapidly undergo a switch to a reactivated state, which is characterized by degranulation and cytokine production. Adoptive transfer of these “memory” NK cells can contribute to MCMV elimination [17]. Patients with selective deficiencies in NK cells have a higher risk of recurrent severe herpesvirus infection [18, 19]. When a patient lacking T cells is infected with HCMV, the viral load is negatively correlated with the number of NK cells and the cytokine levels in serum, which indicates a causal relationship between NK cell function and HCMV elimination [20]. However, whether adoptive transfer of NK cells can clear HCMV in vivo is unknown.

Human NK cells are the first lymphocytes to recover after allo-HSCT [21, 22]. The proportion of NK cells that were responsive to HCMV stimulation in vitro at day 45 tended to be lower in patients who subsequently developed HCMV infection than in patients who did not, indicating that early NK cell reconstitution may be protective against the development of HCMV infection post-HSCT [23]. Miller et al. found that CMV reactivation after allogeneic transplantation promotes a lasting increase in the number of educated NKG2C⁺ NK cells with potent function, which may contribute to the control of HCMV infection long after transplantation [24]. Furthermore, our previous data showed that the rapid reconstitution of IFN-γ-secreting NK cells at day 15 posttransplantation predicted a lower risk of HCMV reactivation [25, 26]. Moreover, donor-recipient KIR ligand matching might promote the NK cell licensing process, thereby increasing NK cell-mediated protection against HCMV reactivation after haploidentical HSCT [26]. Although the clinical findings obtained from these recipients indicate the importance of NK cell function in the control of HCMV infection, surprisingly little is currently known about whether NK cells are responsible for preventing HCMV infection posttransplantation. NK cell adoptive transfer is a promising method for cancer immunotherapy. However, considering the limited numbers of NK cells in the peripheral blood of healthy donors, currently, most adoptively infused NK cells used clinically are ex vivo expanded NK cells [27, 28]. Our previous reports showed that clinical-grade membrane-bound IL-21 (mbIL21)/4-1BBL-expanded NK cell products had higher expression of activating receptors, such as NKp30, NKG2D, and DNAM-1, and therefore had higher cytotoxicity against leukemia cells [29]. Moreover, our previous study showed that the adoptive transfer of expanded NK cells could induce HCMV clearance in the tissue in humanized HCMV-infected mice [30]. However, whether ex vivo mbIL21/4-1BBL-expanded NK cells have a stronger function than primary NK cells against HCMV infection is still unknown. In this study, we investigated the anti-HCMV properties of expanded NK cells and primary NK cells both in vitro and in a mouse model. Furthermore, we performed a clinical trial to explore the safety and efficacy of the adoptive transfer of ex vivo expanded NK cells to patients for protection against CMV infection.

Methods

Ex vivo expansion of NK cells

The mbIL21/4-1BBL-expressing NK cells were generated and maintained as previously described [29]. Fresh isolated peripheral blood mononuclear cells (PBMCs) were obtained from 20 donors in a clinical trial and cocultured with irradiated K562-mbIL21-41BBL APCs at a 1:1-cell ratio in RPMI 1640 medium containing 1000 IU/ml human IL-2 (Huaxin Biotechnology Company, China) and autologous serum for 7 days. The medium was refreshed daily. NK cells that were expanded for 14 d and 21 d were used for in vitro and in vivo experiments and infused into patients in a clinical trial.

Cell line and virus stocks

The MRC-5 human fibroblast line was cultured in complete Dulbecco’s Modified Eagle Medium (DMEM; Sigma-Aldrich) supplemented with 1% nonessential amino acids (Sigma‒Aldrich). PBMCs were isolated from healthy donors by density gradient centrifugation in Ficoll and maintained in complete RPMI 1640 medium (Sigma‒Aldrich). NK cells were then purified by magnetic cell sorting with an NK cell isolation kit (Miltenyi Biotec). The purity of the cells was equal to or greater than 95%, as determined by flow cytometry. All cells were cultured in a humidified atmosphere at 37 °C and in 5% CO₂.

The HCMV wild-type strain AD169 (Shanghai Medical College, Fudan University) was propagated using MRC-5 cells. For the preparation of purified HCMV stocks, cells were infected with AD169 at a multiplicity of infection (MOI) of 0.05. Supernatants containing extracellular virus were collected and stored. The titer of AD169 stocks was determined by standard plaque assays.

Antibodies and flow cytometry

For cell surface staining, NK cells were washed, resuspended in 100 µl of phosphate-buffered saline (PBS) and incubated with the appropriate monoclonal antibodies (mAbs) at room temperature in the dark for 15 min. The fluorochrome-conjugated monoclonal antibodies used to determine the NK cell phenotype before and after expansion are shown in Table S1. All the data were acquired using an LSRFortessa flow cytometer (BD Biosciences) and were analyzed with FlowJo v10 (TreeStar).

Enzyme-linked immunosorbent assay (ELISA)

The cell culture supernatants were collected after 24 h of incubation in each experimental condition. The concentrations of perforin (Abcam, UK), IFN-γ (SinoBiology, China) and TRAIL (SinoBiology, China) were measured by ELISA according to the manufacturer’s instructions.

Anti-HCMV function assay

MRC-5 cells were preinfected with AD169 for 1 h at an MOI of 2. Then, the cells were cocultured with 1000 IU/ml IL-2-prestimulated primary or ex vivo expanded NK cells at an effector-to-target ratio of 5:1 and were treated with GolgiPlug (1 µl/ml, BD Biosciences) and anti-CD107α antibody (3 µl/ml, BD Biosciences) for 4 h. anti-CD3-BV510 and anti-CD56-BUV737 antibodies were used for surface staining. Then, PBMCs were fixed and permeabilized with a fixation/permeabilization kit (BD Bioscience), which was followed by intracellular staining with anti-IFN-γ antibody. The cytotoxicity and cytokine secretion of NK cells were evaluated based on CD107α and IFN-γ secretion. During the blocking antibody experiment, ex vivo expanded NK cells were prestimulated with 2 µl/ml anti-DNAM-1 (559786; BD Biosciences), 4 µl/ml anti-NKp30 (325223; BioLegend, San Diego, CA), 2 µl/ml anti-NKG2D (320813; BioLegend), 2.5 µl/ml anti-NKG2A (#131411, R&D System) or 10 µl/ml anti-NKG2C (clone #134522, R&D System) and incubated overnight at the indicated concentrations.

Virus inhibition and reinfection assay

AD169-infected MRC-5 cells (MOI = 0.5, 1 h) were cocultured with 1000 IU/ml IL-2-prestimulated primary or ex vivo expanded NK cells as described for the cytotoxicity assay. Supernatants were collected on day 1 and day 5, and HCMV levels were quantified using the CMV Nucleic Acid Assay Kit (Liferiver, China) and analyzed with an ABI Prism 7300 qPCR machine.

The supernatants obtained on day 5 during the HCMV proliferation inhibition assay were collected, normalized and incubated with MRC-5 cells at the same MOI for 1 h. Then, the cells were cultured for another 5 days. On day 5, the supernatant was collected. In addition, the cells were digested with 0.05% trypsin-EDTA, and DNA was extracted with DNAzol and dissolved in TE buffer. Both DNA and supernatants were used to quantify HCMV levels.

Mouse model

Six- to eight-week-old female NSG mice (NOD-Prkdcscid IL2Rγnull) were purchased from the Nanjing Model Animal Center (Nanjing, China), sublethally irradiated (150 cGy by X-irradiation) and injected with G-CSF-mobilized PBMCs (1 × 10⁶) obtained from HCMV-seropositive donors at 0 weeks via the tail vein. At 2 weeks after engraftment, the mice received an intraperitoneal (i.p.) injection of 1 × 10⁶ AD169-infected MRC-5 cells. At 4 weeks after engraftment, the mice were subjected to adoptive transfer of 1000 IU/ml IL-2-prestimulated primary NK cells overnight (1 × 10⁷) or ex vivo expanded NK cells (1 × 10⁷), with or without additional adoptive transfer of ex vivo expanded NK cells (1 × 10⁷) at 5 weeks after engraftment. All of the mice received i.p. injection of 50,000 U of IL-2 every other day after NK cell infusion.

To test the expansion of NK cells, mice were sacrificed at day 14 after NK cell infusion. The spleen, liver, lungs and peripheral blood were harvested, and NK cell percentages were analyzed by flow cytometry. The fluorochrome-conjugated monoclonal antibodies that were used are shown in Table S2. Tissue HCMV elimination was analyzed based on HCMV DNA detection by in situ hybridization using an HCMV probe (BOND CMV Probe, PB0614). Mice were subjected to adoptive transfer of ex vivo expanded NK cells from four healthy donors. More than five mice per group were harvested at each time point.

NK cell enrichment, RNA sequencing and data analysis

To detect the anti-HCMV functions of NK cells in vivo, mice were sacrificed at day 14 after NK cell infusion, and liver and lung tissue cells were ground. PBMCs were harvested using Percoll (Solarbio, China) and stained with mouse CD45-Percp, human CD45-V500, human CD3-APC-H7 and human CD56-PE-Cy7 antibodies on ice in the dark for 30 min. After washing with PBS, the cells were resuspended, and then CD3-CD56⁺ NK cells were sorted to a purity of >98% using a BD SymphonyS6 cell sorter (BD Biosciences). The sorted NK cells were analyzed using RNA-seq. Briefly, total RNA was extracted using a TRIzol reagent kit according to the manufacturer’s protocol (Invitrogen, Carlsbad, CA, USA) after the cells were lysed. RNA concentrations were measured using Qubit. Next-generation RNA-seq libraries were constructed with qualified RNA samples using the NEBNext® Ultra™ RNA Library Prep Kit for Illumina. The libraries were sequenced on an Illumina NovaSeq platform to generate 150 bp paired-end reads according to the manufacturer’s instructions, yielding ~8 G raw data per sample. Low-quality reads and adapter sequences were removed, and clean reads were retained. Finally, the fragment per kilobase of transcript per million mapped reads (FPKM) and count values were used for the following analysis. Differential expression analysis between two different groups was performed using the DEGSeq R package (1.20.0). Genes/transcripts with a false discovery rate (FDR) below 0.05 and an absolute fold change ≥2 were considered differentially expressed genes/transcripts. Principal component analysis (PCA) was performed with the R package gmodels (http://www.r-project.org/). Gene Ontology (GO) enrichment analysis and hierarchy relation analysis of the DEGs were performed with the R package “topGO”.

Patients and study design

Twenty patients with malignant hematologic disease who underwent haploidentical transplantation at the Peking University People’s Hospital between March 2020 and March 2021 were prospectively enrolled in the clinical trial. The inclusion criteria were as follows: (1) patients with acute leukemia (AL), myelodysplastic syndrome (MDS), multiple myeloma (MM) or lymphoma; (2) patients and donors who were 16–65 years old; (3) patients and donors who were HCMV-seropositive (IgG-positive, IgM-negative) before HSCT; (4) patients who were negative for HCMV by day 20 ± 3 day post-HSCT; (5) patients who did not develop grade 1–4 aGVHD at 20 ± 3 day post-HSCT or whose aGVHD was under control and who received a dose of corticosteroid (methylprednisolone) <0.5 mg/kg/day before 72 h after NK cell infusion; and (6) patients who agreed to participate in this study and whose donors could offer peripheral blood to expand NK cells at a defined time point. The exclusion criteria were as follows: (1) patients with serious organ dysfunction; (2) patients with uncontrolled infection; (3) patients who developed uncontrolled acute graft-versus-host disease (aGVHD); (4) patients who had participated in other clinical trials within the previous month; and (5) patients whose donors were HBV-, HCV- or HIV-positive. All patients and donors provided written, informed consent, and the Institutional Review Board of the Peking University Institute of Hematology approved the study. This interventional study was registered at www.clinicaltrials.gov as #NCT04320303.

A total of 20 patients were infused with ex vivo mbIL21/4-1BBL-expanded NK cells from their HSCT donors at +20 day ± 3 day and +27 day ± 3 day posttransplantation. Moreover, subcutaneous injection of IL-2 (4 × 10⁵ IU/M²) was performed three times a week for 3 weeks after the first adoptive NK cell infusion for patients according to individual treatment plans. For data analysis, patients were further divided into two cohorts: ten patients in cohort A did not receive IL-2, while ten patients in cohort B did receive IL-2. Fresh peripheral blood samples were collected from patients at day 0 before NK cell infusion and 1, 2, 3, 4 and 8 weeks postinfusion. PBMCs were isolated by density gradient centrifugation using lymphocyte separation medium (GE Healthcare, Milwaukee, WI). The fluorochrome-conjugated monoclonal antibodies used are shown in Table S3.

Transplant protocol, HCMV monitoring and treatment

Haploidentical transplant protocols, including conditioning regimen, mobilization, stem cell infusion, and GVHD prophylaxis, were performed as described previously [1, 31, 32]. Briefly, all patients received myeloablative regimens. Pretransplant conditioning included treatment with cytarabine (4 g/m²/day on day –10 to day –9), busulfan (3.2 mg/kg/day intravenously (IV) on day –8 to day –6), cyclophosphamide (1.8 g/m²/day on day –5 to day –4), semustine (250 mg/m² on day –3) and rabbit ATG (thymoglobulin; IMTIX-Sangstat, Lyon, France, 2.5 mg/kg/day IV on day –5 to day –2). GVHD prophylaxis included cyclosporine (CSA), mycophenolate mofetil (MMF) and short-course methotrexate (MTX), as previously described [31, 32]. Grafts comprised granulocyte-colony stimulating factor (G-CSF)-mobilized bone marrow and blood cells, as described previously [31, 32].

HCMV infection monitoring was performed twice weekly based on HCMV DNA levels in the serum using real-time polymerase chain reaction (PCR) assay kits from PG Biotech Co. Ltd. (Shenzhen, China). Patients with HCMV infection (HCMV ≥ 1 × 10³ copies/ml for over 1 week) and refractory HCMV infection (HCMV infection lasting for over 2 weeks) were diagnosed according to previously published criteria [7]. These patients began first-line therapy with ganciclovir (5 mg/kg intravenously, twice daily). They received a combination of foscarnet (60 mg/kg intravenously, twice daily) and immunoglobulin when HCMV became refractory.

Retrospective control cohort matching

The retrospective control group included 134 consecutive patients with acute myeloid leukemia, acute lymphoid leukemia, chronic myeloid leukemia, or myelodysplastic syndrome who underwent haploidentical HSCT from May 2015 to May 2017 at Peking University Institute of Hematology from a previous clinical trial with the same conditioning regimen but without NK cell infusion. Patients subjected to NK cell infusion were propensity score matched to patients in the retrospective control group. Sex, age, diagnosis, disease risk index, conditioning regimen and GVHD status were subjected to propensity analysis via logistic regression models. In each cohort, each patient who underwent NK cell infusion was matched to three patients in the control group. Therefore, 60 patients were enrolled as a comparison cohort. These 60 patients comprised 30 patients in cohort C who were matched with cohort A and 30 patients in cohort D who were matched with cohort B.

Statistical analysis

All data were analyzed using SPSS 22.0 (SPSS, Inc., Chicago, IL, USA) and GraphPad Prism 8.0 software. Descriptive statistical analysis was performed to evaluate the variables related to the characteristics of patients. The chi-square test and Fisher’s exact test were used to compare count variables and for group comparisons. Measurement variables were compared using Student’s t test or a nonparametric test for comparing two groups. Student’s t test was used to analyze data with a normal distribution. A nonparametric test was used to analyze data without a normal distribution (the Mann–Whitney U test for comparisons of two groups). One-way analysis of variance (ANOVA) was used for comparisons of multiple groups. Overall survival (OS) and disease-free survival (DFS) rates were estimated with the Kaplan‒Meier method. The results with p values < 0.05 were considered statistically significant. p values are described as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

Results

Expanded NK cells showed stronger activation than primary NK cells in response to HCMV-infected fibroblasts

The percentage of NK cells increased to more than 80% after expansion, and this value was ~10% in healthy donors (Fig. 1A). A heatmap was generated to show the characteristics of receptor expression (Fig. 1B).

Fig. 1.

Expanded NK cells showed strong activation against HCMV-infected fibroblasts. A Purity of NK cells expanded for 14 days and 21 days compared with that of primary NK cells (priNKs). B The heatmap shows the expression of all activating receptors, chemokine receptors, adhesion molecules, inhibitory receptors, etc., on priNKs and expanded NK cells (exNKs). C–H Representative FACS image and histogram of receptor expression on priNKs and exNKs. I The levels of IFN-γ, perforin and TRAIL in the cell culture supernatant were measured by ELISA. PriNKs or exNKs were stimulated with HCMV-infected MRC-5 cells as well as IL-2 for 24 h. J, K Levels of CD107α and IFN-γ secretion among priNKs and exNKs with or without blockade of the activating receptors DNAM-1, NKp30 and NKG2D. The data are representative of six independent experiments. *p < 0.05, **p < 0.01. The error bars show the SEM

The expression of activating receptors, including NKG2D, NKp30, NKp44, NKp46, CD25, CD27, CD69, CD94, CD158d and 41BB, on expanded NK cells (exNKs) was increased (Fig. 1B, C). However, CD16 expression was decreased on exNKs (Fig. 1D). Cell chemokine receptors and adhesion molecule receptors, including CXCR3, CXCR4, CX3CR1, CCR4, DNAM-1, and CD62L, showed consistently increased expression (Fig. 1E, F). Moreover, inhibitory receptors (PD-1, TIM-3, CTLA-4, TIGIT and CD96) showed increased mean fluorescence intensities (MFIs; Fig. 1G). The inhibitory receptor NKG2A also showed increased expression (Fig. 1H). We quantified the levels of IFN-γ, perforin and tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) in the supernatant of NK cells stimulated with HCMV and found increased secretion of IFN-γ and perforin by exNKs (Fig. 1I).

ExNKs showed better anti-HCMV activity than primary NK cells (priNKs), as they secreted more CD107α and IFN-γ (Fig. 1J). We wondered whether the enhanced activation of exNKs was mediated by increased expression of activating receptors. However, there was no difference in activation when NKG2D, NKp30 and DNAM-1 were blocked separately. Interestingly, when the activating receptors NKG2D, NKp30 and DNAM-1 were blocked simultaneously, CD107α and IFN-γ secretion decreased (Fig. 1K). Hence, NKG2D, NKp30 and DNAM-1 may simultaneously enhance exNK activation in response to HCMV-infected cells.

Expanded NK cells showed better abilities to inhibit HCMV propagation than primary NK cells in vitro

To explore whether NK cells could inhibit HCMV propagation in host cells, AD169-infected MRC-5 cells were incubated for ~5 days with or without either priNKs or exNKs. As shown in Fig. 2A, on day 1, AD169-infected MRC-5 cells without NK cells showed mostly normal cell morphology. This morphology was disrupted by the addition of exNKs and partly impaired by the addition of priNKs. On day 5, AD169-infected MRC-5 cells without NK cells showed pathological morphology, and the cells were larger and round with Owl’s eye inclusions. There were some suspended pathological cells, some cells were lysed, and lysate granules were seen in the culture medium. Regarding AD169-infected MRC-5 cells treated with exNKs, all of the target cells were affected, and no cells could be seen under a microscope. Regarding AD169-infected MRC-5 cells treated with priNKs, only cell debris was observed. HCMV copy numbers were significantly higher on day 5 than on day 1 in AD169-infected MRC-5 cells, which decreased with exNK addition. There was no difference on days 1 and 5 after priNK addition (Fig. 2B). Therefore, exNKs and priNKs both inhibited HCMV propagation, but exNKs showed stronger inhibition.

Fig. 2.

Expanded NK cells showed enhanced abilities to inhibit HCMV transmission and propagation and reduce reinfection capacity in vitro. A Morphology of AD169-infected MRC-5 cells on day 1 and day 5 with or without priNK and exNK treatment. B CMV-DNA copy numbers on day 1 and day 5 in the supernatant of AD169-infected MRC-5 cells with or without priNK and exNK treatment. C CMV-DNA copies on day 5 in the supernatant of AD169-infected MRC-5 cells with or without NK cell treatment. D CMV DNA copy fold changes from day 1 to day 5 in the supernatant of AD169-infected MRC-5 cells with or without NK cell treatment. Supernatant (E) and intracellular (F) CMV copy numbers on day 5 for reinfection capacity evaluation. *p < 0.05, **p < 0.01, ***p < 0.001. The error bars show the SEM

Next, we cultured AD169-infected MRC-5 and NK cells in Transwell plates to prevent cell‒cell contact. On day 1, the MRC-5 cells in the priNK Transwell group exhibited more normal cell morphology than those in the effector-target cell contact group, while the cells in the exNK Transwell group were round (Fig. 2A). On day 5, the cell morphology of the Transwell group was similar to that of the cell contact group.

We compared HCMV copy numbers on day 5 among different groups and found that the number of HCMV copies in the NK cell addition groups was lower than that in the control group (Fig. 2C). The number of HCMV copies in the exNK groups was significantly lower than that in the priNK groups regardless of whether the assays were performed with cell‒cell contact or Transwells. Moreover, the number of HCMV copies in the Transwell groups was higher than that in the cell contact group for both priNKs and exNKs. We also calculated the fold change in HCMV copy numbers from day 5 to day 1. The control group (AD169-MRC-5) showed a nearly 2000-fold increase. As mentioned above, AD169-MRC-5 cells cocultured with priNKs and exNKs showed no increase in HCMV copy numbers (Fig. 2D). Moreover, the fold change in the number of HCMV copies in the Transwell groups was also higher than that in the cell contact group.

Therefore, the Transwell plates partially reduced the capacity of the NK cells to control HCMV transmission, indicating that the effect was mediated by both cell-to-cell contact and other soluble factors secreted by preactivated NK cells. In addition, exNKs and priNKs inhibited HCMV propagation in vitro, but exNKs showed better abilities than priNKs.

To explore the effect of NK cells on HCMV infection capacity, the supernatant of AD169-infected MRC-5 cells cocultured with or without NK cells on day 5 was used to further infect MRC-5 cells at the same MOI. The number of copies of HCMV both in the supernatant and in the cytoplasm after 5 days of reinfection with the supernatant from AD169-infected MRC-5 cells cultured without NK cells was significantly higher than that after reinfection with the supernatant of AD169-infected MRC-5 cells cultured with NK cells (Fig. 2E, F). Hence, both priNKs and exNKs can inhibit HCMV propagation by not only killing HCMV-infected cells but also decreasing HCMV infection capacity.

Expanded NK cells showed a strong ability to clear CMV infection, unlike primary NK cells, in a humanized HCMV mouse model

Next, we established an HCMV-infected humanized mouse model (Fig. 3A). The preirradiated NSG mice were transplanted with G-CSF-mobilized PBMCs from HCMV-IgG⁺ donors. After 2 weeks of humanized immune reconstitution, the HCD45⁺ cell percentage in the peripheral blood was determined, and more than 2% of the mice were selected to be further i.p. injected with AD169-infected MRC-5 cells to establish an HCMV-infected humanized mouse model. The results showed that adoptively transferred NK cells had the ability to migrate to the spleen, liver, and lungs and persisted in these target organs for 14 days, while mice subjected to exNK infusion showed more NK cells in organs and peripheral blood than mice subjected to priNK cell infusion (p < 0.05, p < 0.0001, p < 0.01, p < 0.05 for the liver, lungs, spleen and peripheral blood, respectively) (Fig. 3B). Findings regarding HCMV infection in the tissue on day 14 after NK cell infusion are shown in Fig. 3C. There was a higher incidence of HCMV clearance in the liver (76.5% vs. 33.3%, p = 0.032) and lungs (82.4% vs. 44.4%, p = 0.046) in mice subjected to exNK infusion than in those subjected to priNK cell infusion. However, the ability of exNKs to clear HCMV in the spleen was not as strong as that in the liver and lungs. The incidence of HCMV clearance in the spleen also tended to be higher in mice subjected to exNK infusion (47.1% vs. 33.3%) (Fig. 3C).

Fig. 3.

Expanded NK cells showed a stronger ability to clear HCMV infection in a humanized mouse model than primary NK cells. A Experimental design for the HCMV-infected humanized mouse model and the priNK and exNK infusion strategies. B The percentages of NK cells in the liver, lungs, spleen and peripheral blood on day 14 after priNK (n = 17) and exNK (n = 17) infusion. C Representative image of HCMV RNA in situ hybridization in tissue on day 14 after NK cell infusion. The cells stained brown are cells infected with HCMV. D Experimental design for the HCMV-infected humanized mouse model and exNK infusion strategies. E The percentages of NK cells in the liver, lungs, and spleen on day 14 after one (n = 17) or two (n = 17) exNK infusions. F Quantification of HCMV⁺ cells in the liver, spleen and lungs of mice. G Expression of the Fc receptor CD16 and the exhaustion receptors TIGIT, TIM-3, and PD-1 on NK cells in the liver, lungs and spleen on day 14 in mice given one or two exNK infusions. The data are representative of four independent experiments. Four to six mice per group were evaluated at each time point. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. The error bars show the SEM

We then tried to enhance NK cell function by infusing exNKs one more time on day 7 after the first exNK infusion. The strategy is shown in Fig. 3D. The mice in the two NK cell infusion groups showed a higher percentage of NK cells in the liver and lungs (Fig. 3E). Mice given two NK cell infusions showed complete HCMV clearance in the liver and lungs and 53% HCMV elimination in the spleen (Fig. 3C, F).

Moreover, the results showed increased expression of CD16 and decreased expression of the exhaustion receptors Tim-3, TIGIT and PD-1 on NK cells in the groups given two infusions, which may support the stronger anti-HCMV effect of two infusions (Fig. 3G). We further used antibodies against PD-1, Tim-3 or TIGIT in the humanized HCMV mice that were subjected to one NK cell infusion to study whether this infusion was helpful for clearing CMV infection. The results showed that the receptors on NK cells were effectively blocked (Fig. S1). HCMV elimination tended to be enhanced with the use of these blockers. The CMV elimination rate increased from 33 to 50% in the liver, 66 to 83% in the lung, and 17 to 33% in the spleen with the use of the anti-PD-1 antibody. In addition, we also evaluated CMV elimination with blockade of Tim-3 or TIGIT. The results were similar to those for PD-1 blockade. However, the findings were not statistically significant, which may have been because of the limited sample size. Further studies could be performed in the future.

Therefore, exNKs showed a stronger ability to eliminate HCMV infection than priNKs. This ability could be further enhanced by administering an injection at an additional time point postinfusion.

Expanded NK cells exhibited an activated state after infusion characterized by cytotoxicity pathway activation in mice

To verify the anti-HCMV function of infused NK cells in mice, we performed RNA sequencing using NK cells from the lungs and livers of mice infused one infusion or two infusions of exNKs on day 7 and day 14. PCA plots showed that exNKs were similar on days 7, 14, and 2 of infusion but very different from those before infusion (Fig. 4A). Moreover, there were few differences between NK cells in the liver and those in the lungs, indicating that there were few tissue-specific NK cells. Significantly higher numbers of differentially expressed genes (DEGs) were observed on day 7 in the single infusion and double infusion groups than on day 14 in the single infusion group, suggesting that NK cell reactivity was similar between the double infusion group and the single infusion group on day 7, while there was a greater difference from the single infusion group on day 14 (Fig. 4B, C).

Fig. 4.

NK cells that expanded postinfusion exhibited an activated state with cytotoxicity pathway enrichment in mice. A Principal component analysis (PCA) plots of NK cells extracted from the livers and lungs of mice and exNKs. PCA was performed with the R package gmodels. B Venny was used to analyze the numbers of differentially expressed genes (DEGs) between NK cells that were extracted from mice given two infusions or one infusion and exNKs before infusion. C Heatmap of the expression of the DEGs among exNKs in the livers of mice and before infusion. D GO biological process analysis showed enrichment of NK-mediated immunity and cytotoxicity processes. E Heatmap of the expression of the top DEGs related to both NK cell-mediated immunity and cytotoxicity processes. N = 3 for each group

GO analyses were performed based on the DEGs and revealed that gene signatures associated with NK cell-mediated immunity and cytotoxicity processes were more enriched in NK cells in livers from mice subjected to two infusions and from mice subjected to one infusion on day 7 and day 14 than in exNKs (Fig. 4D). The top differentially expressed genes are displayed in Fig. 4E. The expression of NCR1, HLA-E, and HLA-F was significantly upregulated, and these molecules are associated with the function of NK cells. NCR1 encodes the natural cytotoxicity receptor NKp46, and its upregulation led to stronger NK cell cytotoxicity than exNK cytotoxicity after infusion. HLA-E and HLA-F are ligands of the activating receptors NKG2C and KIR3DS1, the activation of which initiates a downstream antiviral immune response. As the CD96 receptor is inhibitory, CD96 downregulation also caused NK cells to be activated after infusion. The results of the DEG analysis and the enriched pathways in mouse lungs were consistent with the findings in the liver (Fig. S2). We further validated the expression levels of HLA-E, HLA-F, NKp46 and CD96 with FACS, and the results were consistent with the GO analysis (Fig. S3).

NK cell infusion prevented CMV infection and promoted CMV clearance posttransplantation

Based on the efficacy of exNKs against HCMV in vitro and in humanized mouse models, as well as our previous studies on the safety of exNK infusion in humans [29], we registered a clinical trial to assess whether adoptive transfer of exNKs in patients post-HSCT could protect against HCMV infection and promote anti-HCMV immune cell reconstitution. The detailed characteristics of a total of 20 patients in the NK cell infusion cohorts and 60 matched patients in the retrospective control cohorts are shown in Table 1. All 20 patients in the NK cell infusion cohort were engrafted successfully. Ten (50.0%) patients developed aGVHD at a median time of 26 (11–73) days after allo-HSCT. No infusion-related side effects were observed post-NK cell infusion. Three patients relapsed, and all of the patients were alive by the end point of observation. There were no significant differences in the occurrence of death, relapse or aGVHD between the NK cell infusion cohorts and the control cohorts. Therefore, adaptive NK cell infusion was safe and well tolerated, with no serious side effects related to infusion.

Table 1.

Clinical characteristics and outcomes in patients in NK cell infusion cohorts and control cohorts

Characteristics	NK infusion cohort	Control cohort	p value
	(n = 20)	(n = 60)
Age (median, years, range)	40 (20–64)	36 (18–63)	0.248
Sex (M/F)	15/5	41/19	0.573
Disease type n (%)			0.083
AML	14 (70.0)	32 (53.3)
ALL	1 (5.0)	13 (21.7)
MDS	4 (20.0)	15 (25.0)
MM	1 (5.0)	0 (0.0)
Disease status, high risk, No. (%)	5 (25.0)	24 (40.0)	0.227
Donor-recipient sex, n (%)			0.978
Male to male	12 (60.0)	34 (56.7)
Male to female	3 (15.5)	11 (18.3)
Female to male	3 (15.5)	8 (13.3)
Female to female	2 (10.0)	7 (11.7)
Donor-recipient relation, n (%)			0.330
Parent to child	5 (25.0)	21 (35)
Child to parent	10 (50.0)	26 (43.3)
Sibling to sibling	4 (20.0)	13 (21.7)
Unrelated	1 (5.0)	0 (0.0)
Donor-recipient blood type, n (%)			0.859
Match	12 (60.0)	38 (63.3)
Minor mismatch	4 (20.0)	8 (13.3)
Major mismatch	3 (15.0)	12 (20.0)
Bidirectional mismatch	1 (5.0)	2 (3.3)
aGVHD, n (%)
Patients who developed aGVHD, n (%)	10 (50.0)	25 (41.7)	0.228
1–2° aGVHD	10 (100.0)	19 (76.0)
3–4° aGVHD	0 (0.0)	6 (24.0)
Time of aGVHD after HSCT (median, days)	26 (11–73)	23.5 (10–62)	0.573
CMV infection
Patients who developed CMV infection, n (%)	13 (65.0)	50 (83.3)	0.156
Time of CMV infection after HSCT (median, days)	34 (24–84)	32.5 (19–58)	0.132
Peak titer of CMV DNA copy (103/ml), median	4.79 (1.52–47.4)	7.79 (1.18–191)	0.057
Total CMV persistence time (days, median, range)	14 (5–24)	17 (7–56)	0.013
Patients who developed refractory CMV infection (>2 weeks), n (%)	6 (30.0)	39 (65.0)	0.006
CMV infection cycles (median, range)	1 (1–2)	1 (1–4)	0.197
EBV infection
Patients who developed EBV infection, n (%)	1 (5.0)	18 (30.0)	0.049
Relapse
Patients who relapsed, n (%)	3 (15.0)	5 (8.3)	0.667
Hematological relapse, n (%)	2 (10.0)	3 (5.0)
Molecular relapse, n (%)	1 (5.0)	2 (3.3)
Death, n (%)	0 (0.0)	4 (6.7)	0.554

Bold values indicate statistally difference

AML acute myeloid leukemia, ALL acute lymphoblastic leukemia, MDS myelodysplastic syndrome, MM multiple myeloma, aGVHD acute graft-versus-host disease, CMV cytomegalovirus, EBV Epstein-Barr virus, NK cell Natural Killer cell, HSCT hematopoietic stem cell transplantation

First, we compared the incidences of HCMV infection and refractory HCMV infection among the 20 patients subjected to NK cell infusion and the 60 patients in the control cohort. As shown in Table 1, 13 patients developed HCMV infection, and 6 patients developed refractory HCMV infection. The cumulative incidences of HCMV (HR = 0.54, 95% CI = 0.32–0.93, p = 0.042) and refractory HCMV infection (HR = 0.34, 95% CI = 0.18–0.65, p = 0.009) for patients in the NK cell infusion cohort were significantly lower than those of the control cohort (Fig. 5A). Moreover, the total HCMV persistence time was shortened (14 (5–24) days vs. 17 (7–56) days, p = 0.013) (Fig. 5B). The peak titer of the HCMV DNA copy number also showed a decreasing trend (4.79 (1.52–47.4) vs. 7.79 (1.18–191), 10³ copies/ml, p = 0.057) (Table 1).

Fig. 5.

NK cell infusion prevented CMV infection and promoted CMV clearance posttransplantation. A Kaplan‒Meier curves of the cumulative incidences of HCMV infection and refractory HCMV infection in patients subjected to NK cell infusion (n = 20) and the control cohort (n = 60). B HCMV infection persistence time for patients subjected to NK cell infusion (n = 20) and the control cohort (n = 60). C, D Kaplan‒Meier curves of the cumulative incidences of HCMV infection and refractory HCMV infection and HCMV infection persistence time for patients in the NK cell infusion cohort treated with or without IL-2. n = 10 in cohort A, and n = 10 in cohort B. *p < 0.05. The error bars show the SEM

In addition, we compared the incidences of HCMV infection and refractory HCMV among patients treated with (cohort B) or without (cohort A) IL-2 maintenance post-NK cell infusion. Details of the patients in the two cohorts are shown in Table S4. The cumulative incidence of refractory HCMV infection showed a trend of being lower in cohort B than in cohort A (HR = 5.91, 95% CI = 1.19–29.4, p = 0.065) (Fig. 5C). Patients in cohort B showed no significant difference in HCMV persistence time from patients in cohort A (17 (5–24) days vs. 8.5 (7–14) days, p = 0.097) (Table S4 and Fig. 5D).

Hence, exNK infusion reduced the occurrence of HCMV infection and, more importantly, the occurrence of refractory HCMV infection in patients post allo-HSCT, which suggested that exNKs might prevent HCMV infection and promote HCMV clearance. Notably, combined IL-2 treatment might improve the efficacy of NK cell infusion, and the mechanism needs to be further evaluated.

Adoptive NK cell infusion promoted NK cell reconstitution posttransplantation

We monitored the quantitative and phenotypic reconstitution of NK cells dynamically in peripheral blood until 2 months after NK cell infusion (Fig. S4). Both the percentage and absolute number of NK cells were significantly increased and peaked at approximately the 2nd–3rd week after NK cell infusion and then decreased. The activating receptors NKG2C, NKG2D and DNAM-1, but not NKp30, exhibited increased expression. The proliferation capacity of NK cells in vivo was increased in the first week post-NK cell infusion, as indicated by KI-67 expression. Moreover, increased intracellular levels of granzyme B and perforin were detected. The percentage of the NKG2A⁺ NK cell subset decreased, while that of the KIR⁺ NK cell subset increased, which suggested that adoptive NK cell infusion could promote NK cell proliferation, activation and maturation post allo-HSCT.

We compared the quantities and characteristics of NK cells between NK cell infusion cohorts and retrospective control cohorts at 30 d posttransplantation (Fig. 6A). The quantity of NK cells was much greater in the group given NK cell infusion than in the control group. The activating receptors NKp30 and NKG2D but not DNAM-1 showed higher expression, and the absolute number of cells in the NKG2C⁺ NK cell subsets was also greater than that in the control group. The results above indicated the presence of more mature NK cell phenotypes in patients subjected to exNK infusion than in the control group.

Fig. 6.

Adoptive NK cell infusion promoted NK cell reconstitution posttransplantation. A Comparison of phenotypic reconstitution of NK cells between the NK cell infusion cohort (n = 20) and the control cohort (n = 60) on day 30 post-HSCT. B Dynamic changes in NK cell reconstitution in the group with refractory HCMV infection (n = 6) and the group with nonrefractory HCMV infection (n = 14) in the NK cell infusion cohort. *p < 0.05, ***p < 0.001, ****p < 0.0001. The error bars show the SEM

Then, we analyzed NK cell reconstitution in patients with or without refractory CMV infection. Patients with refractory CMV infection showed a significantly lower NK cell percentage on day 7 after infusion (Fig. 6B). Moreover, NK cell numbers lower than the cutoff value of 17.9 × 10⁴ cells/µl by day 7 were associated with a higher incidence of refractory CMV infection (0% vs. 66.7%, p = 0.011). Decreased NKp30 expression on NK cells for 1 week postinfusion was related to a weaker clinical response of refractory patients (Fig. S5).

Finally, we compared NK cell reconstitution between cohorts treated with or without IL-2. Patients in cohort B exhibited a higher percentage and absolute number of NK cells, a higher absolute number of NKG2C⁺ NK cells and higher expression of NKp30, which revealed that NK cells in patients may have stronger function when stimulated with IL-2 (Fig. S6).

Discussion

There is little direct evidence from previous studies on the use of NK cells to protect against HCMV infection in vivo. Here, we innovatively evaluated the anti-HCMV function of priNKs and exNKs both in vitro and in vivo. ExNKs showed higher activating receptor expression and stronger function than priNKs in both the killing of HCMV-infected cells and the inhibition of HCMV propagation in vitro. In addition, we verified the superior ability of exNKs to eliminate HCMV infection in a mouse model. Moreover, we demonstrated the safety and efficacy of adoptive transfer of exNKs into patients posttransplantation to protect against CMV infection in a clinical trial, and the clinical efficacy could be optimized through combination with IL-2.

We first explored the potential function of ex vivo IL-21/4-1BBL-expanded NK cells by analyzing membrane receptor expression and comparing it with that of priNKs. Indeed, exNKs showed significantly stronger proliferation and expression of activating receptors, chemokine receptors and adhesion molecules than priNKs; these differences resulted in strong anti-HCMV function of exNKs, which was verified by weakened exNK cell function after simultaneous blockade of the activating receptors. We found that blocking DNAM-1, NKp30, and NKG2D alone did not influence anti-HCMV function, so these activating receptors seemed to synergize to enhance exNK function, which has not been reported before. Unexpectedly, some inhibitory receptors were also upregulated. Studies have reported that one limitation of current strategies for NK cell expansion ex vivo using irradiated feeder cells is that NK cells become exhausted after rapid proliferation and differentiation [33, 34]. We further blocked NKG2A in vitro to study the effect of exNKs on HCMV. The results showed consistent function, with no decrease in CD107α or IFN-γ secretion compared with that in the control group. Therefore, the input of inhibitory signals mediated by the increased expression of inhibitory receptors might be overcome by enhanced activating signals. A recent study on the characteristics of ex vivo exNKs illuminated that the superior expansion and cytotoxicity of exNKs was related to enriched metabolic pathways. Reactivation of NK cell metabolic pathways could result in enhanced cytotoxicity and reduced exhaustion [35], which also supported our findings that exNKs showed strong anti-HCMV function despite increased inhibitory receptor expression. We fully assessed exNK function against HCMV-infected cells through not only direct cytotoxicity assays but also assays to test the capacity of exNKs to inhibit HCMV propagation and decrease HCMV infection ability. Using a Transwell system, we further demonstrated that the mechanism by which NK cells inhibit HCMV propagation is mediated by both cell-to-cell contact and soluble factors secreted by preactivated NK cells. Thomas et al. reported that fibroblasts were resistant to HCMV infection if they were exposed to supernatants from cocultures containing priNKs, but not if they were exposed to NK cell-free supernatants, indicating that IFN-γ plays an important role in this induced resistance to HCMV infection [36], which also supports our findings.

Our humanized mouse model innovatively revealed that adoptive infusion of exNKs induced stronger effects against HCMV infection than infusion of priNKs in vivo, which was supported by the higher NK cell percentage and enhanced HCMV elimination in the tissue in the exNK infusion group. The results from RNA sequencing further demonstrated that NK cells were activated, with increased expression of NCR, HLA-E, and HLA-F and decreased expression of CD96. A previous study reported that NK cells were exhausted after exposure to HCMV. We tried to maintain NK cell function and reverse NK cell exhaustion by infusing exNKs an additional time. Two infusions enhanced NK cell function, as demonstrated by higher NK cell percentages, promoted complete liver and lung HCMV elimination and decreased HCMV levels in the spleen. Moreover, inhibitory receptor expression was lower after this treatment than after one infusion, which revealed the reversal of NK cell exhaustion after two infusions.

Donor-derived NK cell adoptive infusion after ex vivo expansion has been reported to be safe, with minimal toxicity, and is effective for the treatment of some tumors, such as AML, MDS, and non-Hodgkin’s lymphoma (NHL) [27, 29, 37, 38]. Nevertheless, there has been little research on the effect against viral infection. Ciurea et al. conducted a clinical trial in which ex vivo IL-21-expanded donor-derived NK cells were infused into patients suffering high-risk myeloid malignancies before and after haploidentical HSCT [39]. It was observed that the incidence of HCMV infection was lower than that in the retrospective control group. However, different doses of NK cells were infused among patients in the cohort. It was unclear whether the exact dosage of NK cell infusion was effective for HCMV reactivation. In addition, donor HCMV serostatus, which also affects HCMV reactivation after HSCT, was not reported. We fully elucidated the anti-HCMV function of exNKs in vitro in our study as well as the mechanisms by which exNK infusion reduces HCMV infection post-HSCT. Our previous studies indicated that patients were susceptible to HCMV reactivation 2–6 weeks after HSCT, and the median time was 30 days [4, 7, 40]. Patients in our clinical cohorts were infused with NK cells at 20 days post-HSCT and were determined to be negative for HCMV infection before infusion. The lower and delayed incidence of HCMV infection in these patients than in retrospective cohorts demonstrated that exNK infusion could prevent HCMV infection after HSCT, but the efficacy was limited. Interestingly, there was a significant decrease in the occurrence of refractory HCMV infection, which reflects the therapeutic efficacy of HCMV clearance. Our results showed that the mechanism underlying the effectiveness of NK cell infusion against HCMV infection post-HSCT involved effectively promoting HCMV clearance by significantly decreasing the occurrence of refractory HCMV infection and shortening the time of HCMV persistence, as well as decreasing HCMV-DNA copy numbers.

Conventional NK cells live less than 10 days in humans [41]. Exposure to CMV infection clearly induces NK cell exhaustion, which is characterized by reduced effector function [42]. In patients who have undergone HSCT, HCMV was found to have a substantial impact on NK cell homeostasis and maturation by inducing the expansion of adaptive NK cells exhibiting dysfunctional features, exhausted phenotypes and increased levels of the inhibitory receptors PD-1 and LAG-3 [43], which may limit the efficacy of infused NK cells. Previous studies reported that interleukin-2 (IL-2) could enhance NK cell proliferation and stimulate NK cell activation and cytotoxicity in vitro [44]. Patients who underwent HSCT and received low-dose IL-2 exhibited selective expansion and increased activation receptor expression of NK cells in vivo [45, 46]. Our previous open-label randomized trial revealed that the absolute numbers of CD56^bri and CD56^dim NK cells were higher in the IL-2 arm than in the control arm, and both the proportions and absolute numbers of NK cells were increased 1 week after IL-2 therapy [45]. Therefore, we combined IL-2 treatment with NK cell infusion, adapting the treatment according to the individualized treatment plan for each patient. Finally, 10 patients received IL-2 treatment, while 10 patients did not. We innovatively demonstrated the important role of IL-2 in optimizing exNK function in vivo. The combination of NK cell infusion with IL-2 significantly enhanced HCMV clearance compared with NK cell infusion alone. Patients who received IL-2 injection exhibited a trend of reduced occurrence of refractory HCMV infection, shortened persistence time and lower peak HCMV-DNA copy numbers than patients in either the retrospective cohort or single NK cell infusion cohort. Hirakawa et al. detected the levels of p-STAT5, p-STAT3, p-AKT and p-ERK, which are involved in the NK cell functional signaling pathway, in NK cells after IL-2 stimulation in vitro [46]. The authors reported that administration of low-dose IL-2 upregulated Ki67 and CTLA4 expression and enhance the proliferation of CD56^brightCD16^– NK cells in vivo. We also detected higher percentages and absolute numbers of NK cells as well as higher expression of the activating receptor NKp30 and inhibitory KIRs on NK cells in patients in cohort B than in patients in cohort A at different time points after infusion. Therefore, IL-2 may stimulate NK cell expansion and enhance NK cell function in vivo. Previous studies showed that HCMV-specific cytotoxic CD8 + T cells (CMV-CTLs) could expand in culture with HCMV in an IL-2-dependent manner [47] and exhibited cytotoxicity against HCMV-infected fibroblasts. There has been no study on the effect of IL-2 on CMV-CTLs in vivo. Wang et al. reported that IFN-γ expression was considerably increased in CD8+ T cells obtained from liver biopsies of patients with chronic hepatitis B following IL-2 stimulation [48]. Our previous studies demonstrated that the adoptive transfer of CMV-specific T cells promotes quantitative and functional recovery of CMV-specific T cells to guard against refractory CMV infection after haplo-SCT. The CMV-CTL counts and percentages of nonrefractory patients were significantly higher than those of refractory patients at 2 months after HSCT [10]. Therefore, with IL-2 stimulation, CMV-CTLs may act synergistically with NK cells to prevent or eliminate HCMV infection.

However, it is still unclear how infused NK cells protect against HCMV and enhance the reconstitution process. Moreover, our approach for monitoring reconstituted NK cells in vivo cannot distinguish infused NK cells from endogenous NK cells, which limited the subsequent study of the mechanism by which infused NK cells enhance the reconstitution process. The increased KI-67 expression indicated that NK cell proliferation was increased in vivo. However, whether exNKs proliferated actively after infusion or endogenous reconstituted NK cells were actively stimulated to proliferate is also unclear.

In conclusion, this study provides direct evidence that exNKs have stronger effects against HCMV than priNKs both in vitro and in vivo. The results suggest the efficacy of exNK infusion in reducing the occurrence of HCMV infection and refractory HCMV infection in patients post-HSCT. The mechanisms by which exNKs protect against HCMV infection and target HCMV-infected cells need to be further investigated.

Author contributions

Q-NS conducted the in vitro experiments and animal experiments and performed the statistical analyses. X-XY conducted flow cytometry assays and facilitated in vitro experiments. Z-LX, Y-HC, T-TH, Y-YZ, ML, Y-QS, YW, L-PX and X-HZ performed the clinical examinations. X-JH and X-YZ designed the study and interpreted the data. X-JH, X-YZ and Q-NS wrote the manuscript. All authors read and approved the final manuscript.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41423-023-01046-5.