UCB was collected from healthy donors (n=38) with prior written informed maternal consent, under HTA license 22513 (Anthony Nolan Cord Blood Bank) and with ethical approval of 20/EM/0028 East Midlands Derby NHS Research Ethics Committee UK. CBUs not suitable for banking and cryopreservation, were diverted to processing laboratories, either in the same location or shipped overnight to our research laboratories at a separate location. Upon arrival at the banking facility, fresh CBUs were mixed and 1.5ml was withdrawn from the blood bag using a syringe with Luer lock (OriGen) and sterile connection device (Genesis); the bag was then sealed and transported/stored at 4°C prior to processing. The 1.5ml sample of whole UCB was used to conduct a complete blood count using the XN-1000 hematology analyzer (Sysmex).

Adult peripheral blood (APB) was collected from healthy volunteers (n=5) following prior written informed consent, with ethical approval 19/LO/141 of London Bridge NHS REC, UK. Samples were stored at 4°C prior to processing for 44h.

UCB or APB mononuclear cells (CBMCs and PBMCs, respectively) were isolated from fresh whole blood via centrifugation over a Ficoll-Paque PLUS 1.077g/ml gradient (Cytiva) in Leucosep tubes (Greiner Bio) followed by erythrocyte lysis with Pharmlyse (BD Bioscience).

Depletion of CD3⁺ cells from CBMCs and PBMCs was undertaken following mononuclear cells isolation. Mononuclear cells were incubated with CD3 antibody-conjugated magnetic microbeads (Miltenyi Biotec) then passed through an LD column in the QuadroMACS separator (Miltenyi Biotec), where the negative fraction was collected. The process was later replicated in both the semi-automated AUTOMACS (Miltenyi Biotech), and the TyTo fluorescent activated cell sorter (FACS; Miltenyi Biotech) using an APC conjugated anti-CD3 antibody (Miltenyi Biotech).

Cells were centrifugated at 400xg for 10 min (room temperature), the supernatant was discarded, then cells were resuspended at 8x10⁶ cells/ml in Cryostor CS10 (STEMCELL Technologies) and aliquoted into cryo vials at 0.5mL/vial. Vials of cells were frozen at -1°C/min to -80°C in a CoolCell (Corning), then transferred to the vapor phase of a liquid nitrogen tank.

Once isolated, CD3^- CBMCs were either cultured immediately with no cryopreservation step, or cryopreserved cells were thawed and seeded in culture. Cells that were isolated <24h post UCB collection (19h ±3.3; n=13) are referred to as immediately isolated, whereas cells where CD3^- CBMCs was delayed by approximately 2 days (44h ±7.7; n=25) due to delivery between collection center and processing center are referred to as delayed isolations.

CD3^- CBMC were seeded at 1x10⁶ cells/ml in NK MACS medium (Miltenyi Biotech) with 10% heat inactivated human AB serum (Sigma-Aldrich), 1% penicillin streptomycin (pen/strep; Corning), supplemented with 1000IU/ml IL-2 and 100ng/ml IL-15 (Peprotech). Cells were cultured for 21 days at 37°C with 5% CO₂. Culture medium was hemi-depleted every 48–72 hours and fresh medium and cytokines added. These are minimal, non-optimized, conditions for NK cell growth.

CD3^- PBMCs were always cultured immediately after isolation (no cryopreservation), in the same conditions as UCB derived cells, except a lower concentration of IL2 was used (500 IU/ml) in accordance with previous studies (14).

Specific target cell lysis of K562 cells was measured at various effector to target (E:T) cell ratios following coculture with NK cells expanded for 7,14 or 21 days, for 3h.

The Toxilight Bioassay kit (Lonza) was used according to manufacturer’s guidelines to calculate the percentage of target specific lysis. Relative bioluminescence was read out on the Varioskan LUX (Thermo Fisher Scientific).

Specific lysis is calculated using the equation below,

RLU= relative light unit measure of bioluminescence

minimum death RLU = the bioluminescence of the target only control culture

test RLU = bioluminescence of the test co-culture of NK:K562- bioluminescence of NK alone

maximum death RLU= bioluminescence of fully lysed control target (K562 cells)

Direct cytotoxicity and ADCC of NK cells against Raji cells was measured using this protocol, except two conditions were tested per CBU. One where Raji target cells were pre-incubated with CD20 antibody Obinutuzumab and a no antibody control.

Expanded UCB NK cells that were cryopreserved at day 22–24 of expansion were thawed and rested in culture (37°C with 5% CO₂) for 7 days in NK MACS medium with 10% heat inactivated AB human serum, 1% pen/strep, supplemented with 500IU/ml IL-2 and 100ng/ml IL-15 before enhanced cytotoxicity characterization. Their cytotoxicity against five different target leukemic cell lines was measured: K562, Raji, THP-1, KG-1a and MOLM-13. Target cells were first stained with CellTraceViolet (C34557; Invitrogen) in PBS for 15 min at 37°C, then washed in co-culture medium. As with the bioluminescence assay, NK cells and target cells were seeded at a range of E:Ts alongside target only and effector only controls, then they were incubated for 4h or 24h at 37°C with 5% CO₂. After the incubation, cells were stained for flow cytometry (gating strategy Supplementary Figure 1) to assess target cell viability and NK cell CD107a expression. Target specific lysis was calculated using the following equation (15):

UCB and APB cells were washed with FACS buffer (PBS buffer with 3% FBS), then stained for 15 min at 4°C in the dark with one of the antibody cocktails made up in FACS buffer (Supplementary Table 1). After staining, cells were washed twice with FACS buffer. Cell viability was assessed with either LIVE/DEAD Fixable Aqua Dead Stain (L34957; Invitrogen) in PBS for 25 min at 4°C, or Zombie NIR Fixable Viability Kit (423106; Biolegend) in PBS for 10 min at room temperature. For intracellular staining panel, cells were fixed at this point with BD Cytofix/Cytoperm kit (554714) according to manufacturer’s instructions, and then stained with antibody cocktail 3a. Lastly, all cells were washed and resuspended in FACS buffer and acquired on the BD Fortessa Flow Cytometer (Cocktails 1 and 2) or the ID7000 Spectral Analyser (Cocktails 3 and 3a, 4 & 5). Data was analyzed using Flowjo version 10.10.00or ID7000 Software (gating strategy in Supplementary Figure 2).

To monitor NK cell proliferation during expansion, NK fold expansion was calculated at days 7, 14 and 21 of culture using the following calculation (16):

All statistical analysis was conducted on Prism 10 version 10.2.3. For all datasets a Shapiro-Wilk test was applied to determine whether data was normally distributed, so that parametric or non-parametric testing could be applied where appropriate. When assessing the variance between two groups two-tailed T tests or Mann Whitney U tests were used, for parametric and non-parametric data, respectively. Whilst for three or more groups, one-way ANOVAs with Tukey’s multiple comparisons tests were applied. Paired or unpaired tests were used for matched and unmatched samples, respectively. Paired statistical tests were used when comparing the n=13 NK cells from immediately isolated starting material (including cryopreserved and non-cryopreserved), whilst unpaired tests were used for comparisons between the cells from delayed isolated n=20something units and the immediately isolated ones. When comparing between different groups at different time points two-way ANOVAs were used with Tukey’s multiple comparisons tests. For correlational analysis to measure the relationship between two variables Pearson or Spearman correlation was used, depending on whether data was parametric or non-parametric. Simple linear regression was conducted to identify potential causal relationships between two variables. Relevant p values and r values are referenced where appropriate in the text, as well as means quoted with ± standard deviation. Significant p-values are denoted with *p<0.05, **p ≤ 0.01, ***p ≤ 0.001 and ****p ≤ 0.0001.

UCB units not suitable for banking (below TNC threshold for transplant for example) were used to isolate CD3^- CBMCs. This process is easily achieved at the banking site, as units arrive. Post Ficoll yields were 1.71x10⁸ ± 6.1x10⁷CBMCS/CBU, and the CD3 depleted fraction yield post isolation was 4.46x10⁷ ± 1.45x10⁷ CD3^- CBMCs/1x10⁸ cell input, with a purity (%CD3^-) of 86.1% ± 9.4 CD3^-. The proportion of CD3⁺ cell carry over was minimal immediately post isolation when magnetic beads were used (0.91±0.89%), but higher when using FACS (12.2±7.6%). Following cryopreservation, the CD3+ carryover remains at similar levels on D0 (day of thaw) and decreases during the 21 days of culture (Supplementary Figure 3).

Using a basic feeder free expansion system consisting of NK culture media supplemented with IL-2 and IL-15, CD3^- CBMCs were cultured for 21 days resulting in an increase in the CD3^-CD56⁺ NK cell population from 12.8% ± 6.6% at day 0, to 96.8% ± 3.3% at day-21 (Figure 2A). As expected, CD3^- PBMCs (99.8% ± 0.1% CD3^-) had a significantly smaller starting proportion of CD3^-CD56⁺ NK cells, 5.9% ± 3.5% (p<0.01; Figure 2A), but they too achieved a high proportion of CD3^-CD56⁺ NK cells, 98.4% ± 0.2% by day 21. Over the culture period the phenotype of the cell population became more homogeneous, with most cells becoming CD3^-CD56⁺ NK cells and the total cell number increased considerably. However, substantial inter-donor variability in the fold expansion and the cytotoxic potential of the UCB derived NK cells was identified.

Through our characterization of UCB-derived NK cells we observed substantial variability between the ex vivo performance of CD3^- CBMCs from different CBUs. Comparison with equivalent cells expanded from APB (albeit with a smaller cohort) also suggested that this effect may be more prevalent in UCB. To identify possible factors that may be related to this variance, correlational analysis was used to assess the relationships between ex vivo NK cell performance indicators and CB unit/donor characteristics (n=25; Figure 4A; Supplementary Figure 4A).

Results from this analysis identified a significant relationship between decreased platelet density width (PDW) and increased day-21 NK fold expansion (p<0.05; r=-0.582; Figure 4A), whilst increased red cell distribution width (RDW) and the number of nucleated red blood cells (NRBCs) correlated with day-21 target specific lysis against K562 (cytotoxicity; p<0.05; r=0.551; Figure 4A, p<0.01; R = 0.559; Figure 4A respectively).

UCB NK cell performance indicators were also compared between categorical factors relating to the UCB donor/unit, (n=25; Figures 4B–G). Whilst maternal parity and fetal sex were not associated with any effects on either NK cell fold expansion or cytotoxicity in our study (p>0.05), there was a trend for NK cells derived from CBU collected from caesarean sections (CS-both elective and emergency, see Supplementary Figure 5 for breakdown) to achieve a higher fold expansion by day 21 of culture (72.7-fold ±37.5), compared to those isolated from CBU collected from vaginal deliveries (VD; 40.9-fold ±21.0), however this did not reach statistical significance in this cohort (p<0.07; Figure 4B; subtypes of delivery plotted in Supplementary Figure 5A).

In addition to CBU/donor characteristics our results highlighted a significant relationship between time of CBU collection and CD3^- CBMC isolation, and subsequent NK fold expansion (p<0.05; r=-0.457; Figure 4A). To investigate this, we compared expansion characteristics between CD3^- CBMCs that were isolated <24h post UCB collection (19h ±3.3; termed immediate isolation; n=13) to cells isolated 44h ±7.7 post collection (termed delayed isolation; n=25). Following 7 days of expansion, immediately isolated, UCB NK cells demonstrated increased fold expansion (7.2-fold ±4.3) compared to NK cells from unmatched delayed isolation CD3^- CBMCs (3.3-fold ±2.5; p<0.05; Figure 5A).

Since banking cryopreserved isolated cells for future use would be highly desirable from a manufacturing point of view, we also tested NK cell expansion from immediately isolated CD3^- CBMCs that had been cryopreserved. Our results showed that CD3^- CBMCs that were immediately isolated and cryopreserved had significantly higher fold expansion (12.2-fold ±6.9) compared to unmatched delayed and, interestingly, also matched immediately isolated CD3^- CBMCs that were expanded without a cryopreservation step, using the same protocol within the first 7 days (3.3-fold ±2.5; p<0.05; Figure 5A).

When UCB NK cells from the “optimal starting material” (cryopreserved CD3^- CBMCs isolated immediately) were expanded in culture for 21-days, the fold expansion was significantly higher compared to cells expanded in the same system from non-optimal starting material (non-cryopreserved matched CD3^- CBMCs; delayed isolation-unmatched units) at each of the timepoints tested, day-7 (12.2-fold ±6.9; p<0.001), 14 (142.8-fold ±74.24; p<0.001) and 21 (283.8-fold ±126.5; range from 101 – 487-fold; p<0.0001; Figure 5B). Phenotypically these two groups had similar expression of CD69 and NKp46 (Figures 5D–H). However, expanded NK cells from the optimal starting material had higher expression of CD16 (Figure 5C), and NGK2D (Figure 5F), but lower expression of NKG2A (Figure 5E) and NKp44 (Figure 5G), with the latter statistically significant at all time points.

Despite the differences in expansion and phenotype between the two processing modalities, the cytotoxicity demonstrated by NK cells against the K562 cell line were comparable at each E:T ratio at day-21 (p>0.05; Figure 5I). Although, NK cells expanded from the optimal starting material showed higher K562 specific lysis at day 7 (p<0.05), by days 14 and 21 it was comparable between the two groups (at 10:1 effector-to-target ratio; Figure 5J).

A repeat of the correlational analysis between ex vivo NK cell performance indicators and CBU/donor characteristics was conducted, this time on units isolated immediately (<24h) and cryopreserved before expansion. Results show that despite PDW, RDW and NRBCs showing similar variability in this cohort (Supplementary Figure 4B), these characteristics no longer significantly affect parameters of NK cell performance in this group (p>0.05; Figure 6A). This finding suggests that prolonged extracorporeal exposure to red blood cells and platelets, particularly when activated, may impair NK cell performance, and that early isolation can mitigate this effect.

Additionally, once we corrected for the influence that time to isolation had on cell quality, we found that CBUs obtained from CS, expanded significantly better (387.2 ± 107.5 fold) than those from VD (219.1 ± 91.66 fold; p=0.0117; Figures 6A, B). The specific K562 lysis capacity (cytotoxicity), however, was equivalent (Figure 6E). A trend for CBUs from multiparous births to expand better was also observed but did not reach statistical significance (Figure 6C). Fetal sex had no influence on either expansion or specific lysis, regardless of time to isolation (Figures 6D, G).

Although the data presented thus far was collected from experiments where CD3^- CBMCs were isolated using a manual magnetic bead/column system (QuadroMACS separator, Miltenyi Bio), results from NK cell expansions conducted in parallel using a GMP compliant closed isolation process based on fluorescence activated cell sorting (Tyto, Miltenyi Bio) found that CD3^- CBMC starting material from matched UCB units, behaved similarly (p<0.001; R² = 0.7281; Figure 7A). Furthermore, NK cell fold expansion from CD3^- CBMC starting material that was grown in basic feeder free culture medium was predictive of NK cell fold expansion performance in matched CBUs expanded using a feeder cell expansion system, even if the absolute numbers were different (P<0.01; R² = 0.6090; Figure 7B). This was also true for day 21 specific lysis of K562 cells by NK cells expanded in either system (Figures 7C, D).

Following the ex vivo expansion described above, cells from 8 of the 13 CBUs used above and not used for analysis were cryopreserved and maintained in liquid nitrogen for approximately 2 years. These cells were then thawed and cultured for 7 days in our basic culture medium, before being re-analyzed for phenotypic and functional characteristics. Figure 8 depicts the process followed (a) and the cell phenotype (b), which is comparable to the cells prior to cryopreservation, with high expression of an extended panel of mature activating receptors. Crucially the NK cells retained their capacity to lyse not only K562 cells (Figure 8C), but also a further four leukemic cell lines of different origins (THP-1 (7D) MOLM13 (7E), RAJI (7F) and KG1a (7G)) at varying effector to target ratios over short (4h) and longer (24h) co-culture times. The killing capacity was accompanied by increased expression of CD107a from a mean of 40.4 ± 1 3% in NK cells prior to co-culture to a high of 74.58 ± 13.6 in co-culture with K562 at low E:T ratios (Figures 8H–M) and secretion of Granzyme A and B, Granulysin and INFγ (Figure 8N). Intracellular staining confirmed increased INFγ and TNFα content in NK cells (Figure 8O). We also showed that these NK cells are capable of antibody dependent cell mediated cytotoxicity (ADCC) with significantly enhanced killing of Raji lymphoma cells in the presence of the anti-CD20 antibody Obinutuzumab (Figure 8P).

Based on the findings of our study, we propose a workflow for the production of NK cell therapeutic products which enables preselection of the highest quality CBUs with early batch testing, thus minimizing wastage and cost of production (Figure 1). This workflow relies on close liaison with the UCB bank or UCB supplier but eliminates the variability and time sensitivity associated with the processing of fresh units as well as the challenges associated with depleting T cells from frozen units. Utilizing the selection criteria which we identified, UCB units can be immediately diverted to a processing lab, where CBMCs can be isolated, CD3 depleted and cryopreserved at different volumes (cell numbers). Small aliquots from each batch can then be used for miniaturized culture and assessment of expansion potential and cytotoxicity within as little as 5–7 days. Thus, starting material qualification can be achieved in small volumes at lower cost before investing in the expansion of a whole unit. Following this derisking, the most highly proliferative, and cytotoxic units can then be chosen for large scale expansion and clinical manufacturing. This constitutes a novel approach to the processing of starting material and allows a qualification step, prior to investing in expansion of a whole UCB unit.