Phenotypic and functional characterisation of locally produced natural killer cells ex vivo expanded with the K562-41BBL-mbIL21 cell line

We characterised the expansion, phenotype and functional activity of natural killer (NK) cells obtained for a clinical trial. Nineteen expansion procedures were performed to obtain NK cell products for 16 patients. NK cells were expanded ex vivo from haploidentical donor peripheral blood mononuclear cells in the presence of the locally generated feeder cell line K-562 with ectopic expression of 4-1BBL and mbIL-21. The median duration of expansion was 18 days (interquartile range 15–19). The median number of live cells yielded was 2.26 × 109 (range 1.6–3.4 × 109) with an NK content of 96.6% (range 95.1–97.9%). The median NK cell fold expansion was 171 (range 124–275). NK cell fold expansion depended on the number of seeded NK cells, the initial level of C-myc expression and the initial number of mature and immature NK cells. The majority of expanded NK cells had the phenotype of immature activated cells (NKG2A + , double bright CD56 +  + CD16 +  + , CD57-) expressing NKp30, NKp44, NKp46, NKG2D, CD69, HLA-DR and CD96. Despite the expression of exhaustion markers, expanded NK cells exhibited high cytolytic activity against leukaemia cell lines, high degranulation activity and cytokine production. There was a noted decrease in the functional activity of NK cells in tests against the patient’s blasts. In conclusion, NK cells obtained by ex vivo expansion with locally generated K562-41BBL-mbIL21 cells had a relatively undifferentiated phenotype and enhanced cytolytic activity against cancer cell lines. Expansion of NK cells with feeder cells yielded a sufficient quantity of the NK cell product to reach high cell doses or increase the frequency of cell infusions for adoptive immunotherapy. Registered at clinicaltrials.gov as NCT04327037.


Introduction
Adoptive cell therapy represents a promising approach for cancer immunotherapy.Natural killer (NK) cells were identified more than 40 years ago and have been characterised as innate immunity cells with the ability to kill tumours and virally infected cells without previous sensitisation [1].Several ongoing clinical trials are investigating NK cell immunotherapy for the treatment of haematological cancers and solid tumours.The design and results of clinical trials conducted in past years have been reviewed in several publications [2][3][4].The high innate anticancer activity of cells and low incidence of toxicity after infusions [5] are advantages of NK immunotherapy.It has been suggested that NK cell dosing might be an important determinant of therapeutic response and disease outcomes [6].
Today, there are different technologies and sources to obtain NK cells: immunomagnetic isolation from peripheral blood, differentiation with cytokines from umbilical cord blood or induced pluripotent stem cells.Another way to obtain NK cells is expansion with stimulatory/feeder cells, such as irradiated peripheral blood mononuclear cells (PBMCs), EBV-transformed lymphoblastoid cell lines or genetically engineered feeder cells (artificial antigen presenting cells [aAPCs]).The current variety of sources and approaches for isolation, activation and expansion of NK cells for immunotherapy has been described in detail in reviews [4,7].Engineered feeder cells based on the HLAnegative K-562 cell line expressing costimulatory ligand 4-1BBL and membrane bound (mb) cytokines interleukin (IL)-15 or IL-21 are commonly used to expand NK cells ex vivo for experimental therapy in clinical trials [8].In contrast to mbIL15, mbIL21-expressing aAPCs promote improved proliferation of human NK cells with longer telomeres, less senescence, a similar killer cell immunoglobulin-like receptor (KIR) repertoire and expression of activation receptors [9].Thus, better expansion of NK cells with aAPCs expressing mbIL21 produces more cells; consequently, a higher cell dose could be applied during treatment.
We previously developed feeder K562 cell lines expressing 4-1BBL-mbIL-15 and 4-1BBL-mbIL-21 and compared NK cell expansion stimulated by these lines in vitro.We found > 55 times higher NK expansion in the presence of mbIL-21 compared with mbIL-15 [10].In this paper, we characterised the expansion, phenotype and functional activity of NK cells obtained for the clinical trial NCT04327037.For this purpose, NK cells were expanded ex vivo in the presence of the locally generated aAPC line K-562 modified to express 4-1BBL and mbIL-21.

Donors
Peripheral blood samples of 16 haploidentical donors (8 females and 8 males) with a median age of 36 years (range 24-50) were the source of PBMCs for NK cell expansion.The inclusion criteria for donors were as follows: HLAhaploidentical relative donor; age > 18 years; negative for HIV and hepatitis B and C; and absence of pregnancy or breastfeeding.
This study was performed in line with the principles of the Declaration of Helsinki.All donors and legal representatives of the patients provided written informed consent.The study protocol was approved by the local Ethics Committee (approval number 30/12/19) and was registered at clinicaltrials.gov(NCT04327037).

NK expansion and activation
For expansion, 50-250 ml of whole heparinised peripheral blood from the selected donors was withdrawn by venipuncture.PBMCs were isolated by gradient centrifugation.Donor NK cells were expanded and activated by co-culturing the PBMCs with the sub-lethally irradiated (100 Gy) feeder cell line K-562 modified for expression of 4-1BBL and mbIL-21 (K562-41BBL-mbIL21), which was obtained in our laboratory [10].(The incubation ratio was 1:2.)For cultivation, we used RPMI-1640 medium supplemented with 50 IU/ml of IL-2 and 10% foetal bovine serum (FBS).On day 7, PBMCs were re-stimulated with feeder cells in a 1:1 ratio.Cultivation was initiated in 75 cm 2 T-flasks.On day 7 of expansion, the cells were split and cultivated in 175 cm 2 T-flasks; the medium was changed every 3-4 days.At the end of culturing, NK cells were washed twice in an isotonic sodium chloride solution prior to the formulation of the final infusion product or freezing.The reagents for NK expansion are described in Supplementary Table 1.
Cell counts and viability were controlled by microscopic analysis with trypan blue staining.Fold expansions were calculated by dividing the absolute number of yielded live cells at the end of culturing by the respective number of seeded cells on day 0.
Quality control testing included cell viability, immunophenotyping, functional tests, bacteriological control, feeder cell contamination and oncogenic transformation controls.

Functional tests
All functional tests were performed on the final cell products using flow cytometry.

Flow cytometry cytotoxic test against leukaemia cell lines
Immortalised cultures of K-562 cells (ATCC CCL-243, chronic myeloid leukaemia), Raji cells (ATCC CCL-86, Burkitt lymphoma) and Jurkat cells (ATCC TIB-152, acute T cell leukaemia) transduced with the GFP gene were used as the target cells when expanded NK cells were used as the effector cells.The effector-to-target ratios 1:1, 5:1 and 10:1 were tested.The assay was performed for 1 h.The number of dead cells was determined by staining with 7-AAD.The cytotoxicity of NK cells was expressed as the fraction (%) of dead target cells.A representative example of cytotoxic test analysis is shown in Supplementary Fig. 1.

Flow cytometry cytotoxic test against recipient blasts
In cases when recipient blasts were available, we performed cytotoxic tests in which donor NK cells were used as effector cells and corresponding blasts were used as target cells.CFSE staining was used to label the target cells.The effector-to-target ratios 1:1, 5:1 and 10:1 were tested.The cytotoxic assay was performed for 4 h.The number of dead target cells was determined by staining with 7-AAD.Target cells were incubated alone to measure basal cell death (control).The cytotoxic activity of NK cells was expressed as % specific lysis calculated with the following formula: % dead blasts test − % dead blasts control ∕ 100% − % dead blasts control × 100%, where % dead blasts test is the percentage of dead blasts in tests with effector cells and % dead blasts control is the percentage of dead blasts in the control sample without effector cells.

Interferon gamma (IFNγ) production
Intracellular IFNγ production by NK cells was tested after stimulating donor NK cells with K-562 cells or the corresponding recipient blasts at a 1:1 ratio in the presence of brefeldin A. NK cells without target cells stimulation were used as the control.After 4 h of incubation, the cells were stained with anti-CD3 and anti-CD56 antibodies, followed by fixation and permeabilisation and staining with anti-IFNγ.A representative example of this assay is shown in Supplementary Fig. 2.

Degranulation assay
NK cells were incubated with anti-CD107a in the presence of K-562 cells or the corresponding recipient blasts to induce degranulation with an effector-to-target ratio of 1:1.NK cells without target cells stimulation were used as the control.Monensin was used to prevent the internalisation of CD107a.After 4 h of incubation, the cells were stained with anti-CD3 and anti-CD56 antibodies.An example of NK cell degranulation analysis is shown in Supplementary Fig. 3.

Microbiological control
To detect possible contamination of NK cell products with mycoplasma, we used multiplex polymerase chain reaction (PCR) with primers described in Supplementary Table 3, according to published recommendations [11].
The sterility of the cell products was tested with the BACT/ALERT automated microbial detection system (bioMERIEUX) and BACT/ALERT PF Plus culture bottles at day 0, before 7 days of infusion and in the final cell products.

BCR-ABL, cMYC and hTERT expression
To verify the absence of feeder cells in the final NK cell products, the transcript level of the BCR-ABL chimeric oncogene, which is a characteristic of the K-562 cell line, was determined.The established limit of detection was 10 −6 K-562 cells.
To check for the lack of oncogenic transformation of expanded cells, the transcript levels of cMYC and hTERT were analysed in PBMCs before cultivation and in the final NK cell products.BCR-ABL, cMYC and hTERT transcripts were assessed by real-time quantitative PCR using specific primers and a TaqMan probe.Primers and probes were selected independently.The sequences are described in Supplementary Table 3. Gene expression was calculated using the delta Ct method, with the GUS housekeeping gene used for normalisation.

Statistical analysis
The data describing cell growth kinetics are presented as the median and interquartile range.The data describing NK cell phenotype and activity are presented as the mean ± standard deviation (SD).The Wilcoxon matched pairs test was used to compare the expression of NK cell receptors before and after cultivation.Spearman's rank correlation coefficient (r s ) was calculated to analyse the strength of the association between parameters.A P value < 0.05 was considered statistically significant.

Growth kinetics of PBMC subsets
Nineteen expansion procedures were performed to obtain NK cell products for 16 patients.The detailed characteristics of the patients and donors are given in Supplementary Table 4.
During the expansion period, there was a decrease in the T cell fraction and an increase in the NK cell fractions.Figure 1 presents the growth kinetics of NK and T cell subsets during expansion.
At the end of culturing, the median fold expansion of NK cells was 171 (range 124-275), with a median purity of 96.6% (range 95.1-97.9%)and a median viability of 84.5% (range 81.0-86.0%).The median fold expansion of the total T cell fraction was 0.8 (range 0.6-1.7).However, among T cell subpopulations, there was an increase in the absolute number of CD3 + CD56 + cells (the median fold expansion in the final cell products was 5.8 (range 2.2-20.4)and a decrease in the CD3 + CD56-cells (the median fold expansion was 0.4 (range 0.3-0.7).The fold changes in NK cells and subpopulations of T cells during cultivation are shown in Fig. 2.
Detailed characteristics of cell populations before and after expansion are presented in Supplementary Tables 5  and 6.
We analysed the impact of donor age, absolute content of cells seeded for expansion (NK, T, NKT), levels of C-myc and hTERT expression, and levels of expression of activation/maturation/exhaustion markers on NK cell fold expansion.NK cell fold expansion depended on the initial seeded NK cell absolute number (r s = − 0.63, p = 0.005), duration of cell expansion (r s = 0.45, p = 0.06), the level of C-myc expression measured in cells before cultivation (r s = 0.62, p = 0.007), initial number of mature CD57 + (r s = − 0.58, p = 0.02) and CD56dimCD16 + (r s = − 0.74, p = 0.001), and number of immature CD56brightCD16 + cells (r s = 0.67, p = 0.004) NK cells.

Expanded NK cells have a phenotype of less differentiated activated cells
During expansion, stimulatory signals from feeder cells changed the expression of NK cell receptors.Figure 3 shows the expression of some activating NK cell receptors, receptors associated with cell activation or exhaustion and stages of maturation before and after culturing.At the end of culturing, the expanded NK cells had the phenotype of activated cells.After the period of expansion, the expression of the activating receptors NKp30, NKp44, NKp46 and NKG2D increased significantly (Fig. 3a).The expression of the other markers of activation, such as CD69, HLA-DR and CD96, also increased (Fig. 3b).The expression of the activating receptors NKG2C and CD25 did not differ before or after expansion.During expansion, NK cells also increased the expression of some markers of exhaustion.The percentage of NK cells expressing LAG-3 and TIM-3 increased significantly at the end of expansion compared with the initial levels (Fig. 3c).In addition, after the period of expansion, there was a redistribution of cells within the subsets, reflecting the stages of NK maturation (Fig. 3d).Before cultivation, more differentiated NK cells predominated, while after expansion, the population of NK cells consisted of cells with a more immature phenotype (higher level of NKG2A; almost all cells were CD56bright and loss of CD57 expression).Representative examples of NK cell immunophenotypes before and after expansion are shown in Supplementary Figs.4-9.

Expanded NK cells have high functional activity against leukaemia cell lines but lower function activity against recipient blasts
We evaluated the ability of expanded NK cells for degranulation and IFNγ production after stimulation with K-562 cells and recipient blasts (Fig. 4).
After stimulating expanded NK cells with K-562 cells, the mean level of degranulated NK cells was 72.1% ± 14.9% (Fig. 4a).However, stimulating NK cells with recipient blasts did not result in significant degranulation, and the mean level of CD107a expression was not different from that of the control samples (Fig. 4b).Only in two cases was there an increase in the percentage of degranulated NK cells.
There were similar results when IFNγ production was analysed.The mean level of IFNγ-producing NK cells after stimulation with K-562 cells was 23.9 ± 10.1% (Fig. 4c), whereas, except for one tested case, recipient blasts did not stimulate cytokine production (Fig. 4d).
The cytotoxic activity of expanded NK cells was assessed against leukaemia cell lines and recipient blasts (Fig. 5).The mean level of NK cytotoxicity at 5:1 was 86.9% ± 12.7% against K-562 cells, 66.6% ± 18.6% against Raji cells and 43.6% ± 18.9% against Jurkat cells.For seven patients, it was possible to test the cytotoxic activity of expanded donor NK cells against recipient blasts.The cytotoxicity against blasts was lower in comparison with cell lines.The mean cytotoxicity against blasts at the effectorto-target ratio of 5:1 was 23.8% ± 15.3%.

Quality control
None of the NK cell products showed microbiological contamination.
By the end of NK expansion, BCR-ABL expression was undetectable, a finding that indicates the eradication of the feeder cells in the final NK cell cultures.After the period of expansion, cMYC expression decreased compared with the initially seeded cells, except for one case where the level of cMYC expression was insignificantly higher.The level of hTERT expression decreased at the end of NK expansion in comparison with the initial level in all cases except two, which showed an increased level.However, the levels of C-Myc and hTERT expression before expansion were measured in the total population of PBMCs cells, whereas after expansion, it was predominantly the NK cell population.
The level of hTERT expression in the initial cell culture and the final NK cell product had a significant negative correlation with donor age (r s = − 0.6, p = 0.009 and r s = − 0.57, p = 0.02, respectively).Moreover, we found a negative significant correlation between the level of hTERT expression in the final NK cells and the duration of cell expansion (r s = − 0.57, p = 0.02).
cMYC and hTERT expression before and after cultivation are presented in Supplementary Table 7.

Discussion
Cellular immunotherapy is at the forefront of progress in cancer research.The success of CAR T cells has impacted the development of novel alternative cell anticancer strategies.NK cells possess similar functions to T effector cells but with decreased toxicity.Early studies demonstrated the feasibility and safety of the adoptive transfer of allogeneic NK cells [12][13][14].Clinical response rates in clinical studies were variable; the therapeutic response depended on the diagnosis, stage of disease, timing of immunotherapy, NK alloreactivity, cell dose and frequency of infusions [12,13,15,16].
We previously reported the creation of our own K-562based cell line transduced to express 4-1BBL and mbIL-21 [10].In our small-volume experiments, the median fold expansion of NK cells at 3 weeks was > 21 000 (range 3 000-300 000).Other groups have reported > 10 000 fold expansion of NK cells using a feeder cell line based on the myeloid OCI-AML3 cell line (at 5 weeks) [21] and the B-lymphoblastoid 721.221 cell line (at 3 weeks) [22] transduced to express mbIL-21.
When comparing feeder lines based on mbIL-15 and mbIL-21, mbIL21 supports greater proliferation of NK cells than mbIL15 [9,10,23].Thus, the expansion of NK cells with feeder cells expressing mbIL-21 creates the possibility of generating substantially more NK cells from a single withdrawal of peripheral blood.In our study, at the end of culturing, the median fold expansion of NK cells was 171 (range 124-275), with minimal expansion of total T-cells-0.8(range 0.6-1.7)-whichwere predominantly CD3 + CD56 + cells.If we were to use all our obtained cells for our patients as a single infusion, the NK cell dose would be 111 × 10 6 (range 59-134 × 10 6 ) cells/ kg, and the T cell dose would be 2.6 × 10 6 (range 1.3-4.9× 10 6 ) cells/kg.The fold order of NK cell expansion depends on cell source, culture systems for cell propagation (flasks, G-Rex, rocking bioreactors or automated closed systems), number of restimulations with APCs, cytokines and ways to calculate the fold of expansion.For instance, NK expansion efficiency varied from tens [16] to thousands [20] of folds, even when the same APC line (Clone 9. mbIL21) was used.According to our results, NK cell fold expansion depended on the number of seeded NK cells, the initial level of C-myc expression, and the initial number of mature and immature NK cells.We speculate that the number of immature NK cells with a high ability to proliferate is the main factor affecting the efficacy of NK cell expansion.The starting number of CD56dimCD16 + NK cells and the level of C-myc expression could be used as markers of proliferative activity and as criteria for donor selection for NK cell propagation.
An important consideration for the clinical application of any cell product is its safety profile.The main concern of adoptive transfer of allogeneic NK cell products is T cell contamination and consequent graft versus host disease (GVHD).Therefore, the elimination of CD3 + cells before or after expansion is a key step in many protocols.This step is especially important when NK expansion is based on stimulation with high doses of cytokines (IL-2, IL-15 or IL-21) that also stimulate the proliferation of T cells.However, when expanding NK cells using a feeder line, low doses of IL-2 (10-100 IU/ml) are usually used [9,[24][25][26].We used 50 IU/ml of IL-2 and the median purity of NK cells in the final cell products was 96.6%, despite the low initial content of NK cells and expansion without prior CD3 depletion.Importantly, GVHD has not been reported in most studies utilising allogenic feeder-based expanded NK cells.
Therefore, we can conclude that the application of genetically modified feeder cells for NK cell expansion from whole blood without leukapheresis and magnetic selection represents an opportunity to obtain an NK cell product with good expansion, high NK purity and low production cost.Cost is an especially important factor that delays the development and application of cell therapy technologies in countries with limited financial resources.However, the prospect of using serum-free media and G-Rex flasks yields safer NK cells in a shorter time with minimal manipulation.According to our results, the median duration of NK expansion was 18 days (range [15][16][17][18][19]; using G-Rex would shorten the expansion period to 10-12 days. In previous studies, NK cells expanded using gene-modified feeder lines have shown high expression of activating receptors, as well as in vitro antitumour cytotoxicity against different tumour cell lines [9,21,23].We also demonstrated increased expression of the activating receptors NKp30, NKp44, NKp46 and NKG2D, as well as CD69, HLA-DR and CD96.In addition, NK cells acquired a less differentiated phenotype.After expansion, almost all NK cells of CD56bright expressed NKG2A and did not express CD57.CD57 is a marker of mature, terminally differentiated NK cells with high cytotoxicity but decreased proliferative activity [27].However, under specific conditions of cell stimulation, CD57 + human NK cells acquire the CD57-phenotype.Using mbIL-21 for NK cell stimulation decreased CD57 expression [21,28].At the end of culturing, the majority of NK cells expressed CD16; indeed, > 75% of NK cells had the double-bright (CD56bright/CD16bright) phenotype.Previous studies have shown that stimulating NK cells with a combination of cytokines (IL-2, IL-12 and IL-15), as well as mbIL-21-or mbIL-15-expressing feeder lines, produces CD16 + CD56bright NK cells with high cytotoxicity [9,29,30].
NK cell expansion resulted in the increased expression of immune checkpoints LAG-3 and TIM-3, which are commonly associated with T cell exhaustion.The molecular mechanisms underlying NK cell exhaustion are not defined as clearly as they are for T cells.Unlike T cells, TIM-3 and TIGIT are expressed at high levels in unstimulated and functional NK cells [31].A better understanding of the role of immune checkpoints in NK effector function would be beneficial for designing immunotherapeutic approaches for cancer patients.
Despite the expression of exhaustion markers, expanded NK cells were functionally active.They exhibited high cytolytic activity against leukaemia cell lines, high degranulation activity and production of cytokines after stimulation with K-562 cells.However, expanded NK cells were less active in tests performed against primary recipient blasts.Low cytotoxic NK cell activity against recipient blast cells might be explained by various peculiarities of antigen expression on blast cells (e.g.MIC A/B, ULBP1-6 or HLA class I molecules) but not by the dysfunction of NK cells.Moreover, the prevalence of death ligand-mediated mechanisms of cytotoxicity over granule exocytosis could be an explanation of low NK cytotoxicity against patient's blast cells in a short in vitro test.Thus, expanded NK cells are capable of both IFNγ production and cytotoxicity and have phenotypic characteristics associated with both mature and immature NK cells.Our results support the uncoupling of phenotypic markers of maturation and canonical functions of NK cell subsets in expanded ex vivo NK cells [30].
Next-generation strategies of NK cell immunotherapy focus on improving NK cell persistence in vivo and redirecting specificity through cytokine armouring, bi/trispecific killer engagers or genetic modification [32].The success of CAR T cells in B-cell leukaemia and lymphoma stimulated the development of CAR-NK therapy.Unlike CAR T cells, adoptive transfer of CAR NK cells was not associated with the development of cytokine release syndrome, neurotoxicity or graft-versus-host disease [33].The safety of allogeneic NK cell immunotherapy promoted the production of "off the shelf" CAR NK cells to circumvent cost and timing constraints seen in CAR T cell manufacturing.Until recently, genetic modification of NK cells was unsuccessful due to the relative resistance of NK cells to traditional methods of viral transduction.However, using retroviral vectors or lentiviral vectors pseudotyped with a modified baboon envelope glycoprotein significantly improved NK cell transduction efficiency, promoting CAR NK studies [34].Compared to CAR T cells, CAR NK cell production with feeder cells does not require activation beads or cytokines (IL-7, IL-15, IL-21), resulting in reduced cost for CAR NK cells.

Conclusion
We demonstrated that NK cells generated by ex vivo expansion with the locally generated K562-41BBL-mbIL21 feeder line have a relatively undifferentiated phenotype and enhanced cytolytic activity against cancer cell lines.The expansion of NK cells with the feeder cells used allows a sufficient quantity of the NK cell product to reach high cell doses, increase the frequency of cell infusions, or genetically modify to produce CAR NK cells.

Fig. 1
Fig. 1 Growth kinetics of natural killer (NK) cells and T lymphocytes during expansion.a-Changes in NK and T subset composition (%); babsolute number of T and NK cells in culture during expansion.Data are presented as the median and interquartile range

Fig. 2 Fig. 3
Fig. 2 Fold increase in natural killer (NK) cells and subpopulations of T cells during expansion.a-Fold expansion for NK cells; b-fold expansion for T cell subpopulations.Data are presented as the median and interquartile range

Fig. 4
Fig. 4 Degranulation and interferon gamma (IFNγ) production after stimulating expanded natural killer (NK) cells with K-562 cells and recipient blasts.a-Level of degranulated cells after stimulating NK cells with K-562 cells; b-level of degranulated cells after stimulating NK cells with recipient blasts; c-level of IFNγ-producing

Fig. 5
Fig. 5 Cytotoxic activity of expanded natural killer (NK) cells.a-Cytotoxicity against the K-562, Raji and Jurkat cell lines; b-cytotoxicity against recipient blasts.All data are presented as the mean ± SD