Daudi, SW480, and Phoenix-Ampho cell lines were obtained from ATCC. K562 with HLA-A*24:02-transduced cell line was kindly provided by Fred Falkenburg (Leiden University Medical Centre, Netherlands) and subsequently transduced with luciferase for in vivo imaging purposes. EBV-LCL was kindly provided by Phil Greenberg (Seattle, WA, USA). Phoenix-Ampho and SW480 cells were cultured in DMEM supplemented with 1% Pen/Strep (Invitrogen) and 10% FCS (Bodinco), whereas all other cell lines in RPMI with 1% Pen/Strep and 10% FCS. All cell lines were authenticated by short tandem repeat profiling/karyotyping/isoenzyme analysis and were passaged for a maximum of 2 months, after which new cell line stocks were thawed for experimental use. Furthermore, all cell lines were routinely verified by growth rate, morphology, and/or flow cytometry and tested negative for mycoplasma using MycoAlert Mycoplasma Kit (Lonza, Breda, Netherlands). Peripheral blood mononuclear cells (PBMCs) from healthy donors were isolated by Ficoll-Paque (GE Healthcare, Eindhoven, Netherlands) from buffy coats supplied by Sanquin Blood Bank (Amsterdam, Netherlands).

Clone FE11 was generated as previously described (28). FE11 and LM1 [non-functional γ9δ2TCR with length mutation on the complementary determining region 3 (CDR3) of the δ2-chain (31)] γδTCRs were subcloned to pMP71 retroviral vectors containing both γTCR and δTCR chains, separated by a ribosomal skipping T2A sequence. pU57 constructs containing a ribosomal skipping P2A sequence, followed by full-length human CD8α, were purchased from Baseclear (Leiden, Netherlands). Thereafter, CD8α was subcloned into pMP71 vector using XhoI and HindIII restriction sites downstream of γ115TCR-T2A-δ115_LM1 sequence to generate a TEGLM1_CD8α ( Supplementary Table 2 ) construct that contained NcoI and XhoI restriction sites up- and downstream of LM1 γδTCR chains. NcoI and XhoI restriction sites were then inserted up- and downstream of FE11 γδTCR sequences by site-directed mutagenesis PCR, after which this sequence was ligated to P2A-CD8α sequence in pMP71 vector using the introduced NcoI and XhoI sites, generating a TEG011_CD8α construct ( Supplementary Table 1 ). Where indicated, CD4⁺, CD8⁺, CD4⁺CD8αα⁺, and CD4⁺CD8αβ⁺ TCR-transduced T cells were sorted using a FACSAria II (BD) flow cytometry to >99% purity. Expression levels of CD8α mutants were measured by flow cytometry using anti-CD8α antibody (clones RPA-T8).

IFNγ ELISPOT was performed using antihuman IFNγ mAb1-D1K (I) and mAb7-B6–1 (II) (Mabtech) per the manufacturer’s protocol. Then 15,000 TEG cells (TEG011, TEGLM1, TEG011_CD8α, or TEGLM1_CD8α) were co-incubated with 50,000 target cells (E:T ratio 1:3) for 18–24 h in nitrocellulose-bottomed 96-well plates (Millipore). IFNγ spots were visualized with TMB substrate (Sanquin), and subsequently the number of spots was quantified using ELISPOT Analysis Software (Aelvis). Where indicated, blocking of CD8α was performed using 10 μg/ml anti-CD8α antibody clone OKT8 (eBioscience) and blocking of CD8β with 10 μg/ml anti-CD8β clone 2ST8.5H7 (Abcam).

TEGs were generated as previously described (30). Briefly, Phoenix-Ampho packaging cells were transfected with gag-pol (pHIT60), env (pCOLT-GALV), and pMP71 retroviral constructs containing both γTCR and δTCR chains separated by a ribosomal skipping T2A sequence and followed by CD8α sequence separated by P2A sequence where applicable, using FugeneHD reagent (Promega, Leiden, Netherlands). PBMCs from a healthy donor preactivated with 30 ng/ml anti-CD3 (clone OKT3, Miltenyi Biotec) and 50 IU/ml IL-2 (Proleukin, Novartis, Arnhem, Netherlands) were transduced twice with viral supernatant within 48 h, in the presence of 50 IU/ml IL-2 and 6 µg/ml polybrene (Sigma-Aldrich, Zwijndrecht, Netherlands). TCR-transduced T cells were expanded by stimulation with anti-CD3/CD28 Dynabeads (500,000 beads/10⁶ cells; Thermo Fisher Scientific, Breda, Netherlands) and 50 IU/ml IL-2. Thereafter, transduced T cells were depleted of the non-engineered T cells.

Non-engineered T cells were depleted as previously described (27). In brief, transduced T cells were incubated with a biotin-labeled anti-αβTCR antibody (clone BW242/412; Miltenyi Biotec, Leiden, Netherlands) and then incubated with an anti-biotin antibody coupled to magnetic beads (anti-biotin MicroBeads; Miltenyi Biotec), most recently reported to preferentially bind to the βTCR chain (37). Thereafter, the cell suspension was loaded onto an LD column, and αβTCR⁺ T cells were depleted by MACS cell separation per the manufacturer’s protocol (Miltenyi Biotec). After depletion, TEGs were expanded using a T-cell rapid expansion protocol (REP) (30).

The separation of CD4⁺ TEGs was performed using CD4 Microbeads (Miltenyi Biotech) as per the manufacturer’s instructions. Briefly, TEGs that were previously expanded on REP were incubated with magnetic microbeads cells and loaded into LS column for MACS cell separation. Thereafter, CD4⁺ selected or bulk (with CD4:CD8 ratio 50:50) TEGs were expanded separately on the next REP cycle prior to in vitro functional assay. TEG expression was monitored prior to functional assays or in vivo infusion by flow cytometry using anti-αβTCR-APC (clone IP26, eBioscience), anti-pan-γδTCR-PE (clone IMMU510, Beckman Coulter), anti-CD8-PerCP-Cy5.5 (clone RPA-T8, Biolegend), anti-CD4-PeCy7 (clone TPA-R4, Biolegend), anti-CD4-FITC (clone TPA-R4, Biolegend), and Vδ1-FITC (clone TS8.2, Thermo Fisher Scientific) antibodies.

The NOD.Cg-Prkdc^scidIl2rg^tm1Wjl Tg(HLA-A24)3Dvs/Sz (NSG-A24:02) mice (38) were bred and housed in the breeding unit of the Central Animal Facility of Utrecht University. Experiments were conducted per institutional guidelines after obtaining permission from the local ethical committee, and performed in accordance with the current Dutch laws on animal experimentation. Mice were housed in individually ventilated cage (IVC) system to maintain sterile conditions and fed with sterile food and water. After irradiation, mice were given the antibiotic ciproxin in the sterile water throughout the duration of the experiment. Both male and female mice were randomized with equal distribution among the different groups, based on age and initial weight (measure on Day −1) into 10 mice/group. Adult NSG-A24:02 mice (11–20 weeks old) received sublethal total body irradiation (1,75 Gy) on day −1 followed by intravenous injection of 1×10⁵ K562-HLA-A*24:02 luciferase tumor cells on day 0, and received 2 intravenous injections of TEG011, TEG011_CD8α, or TEGLM1_CD8α cells on days 1 and 6 as previously reported (34). Together with the first TEGs injection, all mice received 0,6 × 10⁶ IU of IL-2 (Proleukin; Novartis) in 100 µl incomplete Freund’s adjuvant (IFA) subcutaneously and subsequently administered every 3 weeks until the end of the experiment. Mice were monitored at least twice a week for any symptoms of disease (sign of paralysis, weakness, and reduced motility), weight loss, and clinical appearance scoring (scoring parameter included hunched appearance, activity, fur texture, and piloerection). The humane endpoint was reached when mice showed the aforementioned symptoms of disease, experienced a 20% weight loss from the initial weight (measured on day −1), developed extramedullary solid tumor masses (if any) reached 2 cm³ in volume, and when clinical appearance score 2 was reached for an individual parameter or a total score of 4.

The following antibodies were used for flow cytometry analysis: huCD45-PB (clone HI30; Sony Biotechnology), pan-γδTCR-PE (clone IMMU510; Beckman-Coulter), mCD45-APC (clone 30-F11, Sony Biotechnology), αβTCR-FITC (clone IP26; Biolegend), CD4-PeCy7 (clone RPA-T4, Biolegend), CD8-PerCPCy5.5 (clone RPA-T8, Biolegend), PD-1-BV711 (clone EH12.2H7, Biolegend), and TIM3-BV650 (clone F38-2E2, Biolegend). To exclude non-viable cells from the analysis, Fixable Viability Dye eFluor506 was used (eBioscience). All samples were analyzed on a BD LSRFortessa using FACSDiva Software (BD Biosciences).

Mouse peripheral blood samples were obtained via cheek vein (max. 50–70 µl/mouse) once a week. Red blood cells were lysed using 1× RBC lysis buffer (Biolegend) and were then stained with a mixture of antibody panels as listed above. The persistence of TEG cells was counted as absolute cell number tumor-reactive TEG cells expressing following cell surface markers huCD45⁺γδTCR⁺CD8⁺ and huCD45⁺γδTCR⁺CD4⁺CD8⁺ populations or non-reactive TEG cells expressing huCD45⁺γδTCR⁺CD4⁺ marker observed in mouse peripheral blood using Flow-count Fluorospheres (Beckman Coulter) and measured by flow cytometry.

At the end of the study period, bone marrow (mixed from tibia and femur) and spleen sections were isolated and processed into single-cell suspension. Femur and tibia from the hind legs were collected; bone marrow cells were collected by centrifugation of the bones at 10,000 rpm for 15 s and resuspension of the cells in phosphate buffer saline (PBS).

A small section of the spleen was minced and filtered through a 70 µm cell strainer (BD); incubated with 1× RBC lysis buffer cells for maximum 4 min, and subsequently cells were washed and resuspended in PBS.

Absolute cell number of TEG cells were quantified using Flow-count Fluorospheres and measured from a total of 10⁶ cells stained for the presence of TEG cells in spleen and bone marrow by flow cytometry analysis (BD LSRFortessa).

Formalin-fixed femur for bone marrow sections were embedded in paraffin and cut into 4 μm sections. Hematoxylin and eosin (H&E) staining was performed for the femur, for bone marrow section. Tissue sections were evaluated to assess for any differences in the presence, distribution, and extension of neoplastic foci indicating tumor tissue. Tissue sections of the femur were evaluated for quantification of tumor tissue by dividing the area covered by the tumor cells by the total area of bone marrow tissue visible in the section using the ImageJ analysis system software (NHI, Bethesda, Maryland, USA) and expressed as a percentage. Images were taken using an Olympus BX45 microscope with the Olympus DP25 camera and analyzed using DP2-BSW (version 2.2) or ImageJ software.

Experimental data were analyzed using GraphPad Prism (GraphPad Software Inc., La Jolla, CA, USA) and shown as mean ± standard deviation (SD) or standard error of mean (SEM) with *P < 0.05; **P < 0.01; ***P < 0.001; and ****P < 0.0001. Statistical significances between groups were assessed using a non-parametric Kruskal-Wallis test, a two-way ANOVA, and a mixed-effects model with repeated measures where indicated.

We previously identified an allo-restricted CD8α-dependent Vγ5Vδ1TCR clone FE11 (28), which showed in vitro antitumor reactivity against HLA-A*24:02-expressing tumor cells (34). We therefore investigated whether introduction of CD8αα or CD8αβ along with Vγ5Vδ1TCR derived from clone FE11 could enhance antitumor reactivity of CD8⁺, and also functionally reprogram CD4⁺ TEG011 cells. Hence, we co-transduced T cells with the FE11 γδTCR, and with either CD8α alone or CD8α together with CD8β ( Figure S1 ). Subsequently, we sorted separate sets of CD4⁺ TEG011 cells that co-expressed either exogenous CD8αα (CD4⁺CD8α⁺) or CD8αβ (CD4⁺CD8αβ⁺) as well as TEG011 cells expressing only endogenous CD4 and CD8 as negative and positive controls for tumor recognition, respectively ( Figure 1A ). Thereafter, TEG cells were co-cultured with SW480 and EBV-LCL target cells or healthy PBMCs as mock control. Both CD4⁺CD8α⁺ and CD4⁺CD8αβ⁺ TEG011 cells secreted significantly higher levels of IFNγ upon exposure to tumor targets than CD4⁺ TEG011 cells. The acquired antitumor reactivity of CD4⁺CD8α⁺ and CD4⁺CD8αβ⁺ TEG011 cells could be blocked by CD8α and CD8β blocking antibodies ( Figure 1B ), confirming the strict dependence of FE11 γδTCR on introduced CD8 molecules. Taken together, we showed that introduction of CD8α alone is sufficient to re-establish antitumor reactivity of CD4⁺ T cells expressing FE11 γδTCR. Introduction of CD8β did not further enhance tumor recognition but was functionally involved in the molecular interaction with its target when present.

For clinical administration, co-expression of both CD8α and the γδTCR in one vector is preferred to allow reproducible and cost-effective production processes (26, 27, 39). Moreover, co-expressing both CD8α and the γδTCR in one vector can also overcome the difference in transduction efficiency when they were transduced separately. Therefore, we generated new retroviral constructs carrying either FE11 γδTCR or a non-functional length mutant clone LM1 γδTCR [ (31); served as mock control] followed by full-length human CD8α receptor sequences (TEG011_CD8α and TEGLM1_CD8α, Figure 1C ). The complete sequence of transgenes for these retroviral constructs is listed in Supplementary Tables 1, 2 , respectively. Subsequently, αβT cells were transduced with either FE11 γδTCR without human CD8α receptor (TEG011), FE11 γδTCR with human CD8α receptor (TEG011_CD8α), or LM1 γδTCR with human CD8α receptor (TEGLM1_CD8α). After TEG expansion, we performed magnetic selection of CD4⁺ T cells for each TEG constructs. To elucidate whether introduction of transgenic CD8α receptor adequately rescues TEG011 reactivity of non-tumor reactive CD4-transduced cells once delivered by the very same vector, we co-cultured tumor target HLA-A*24:02-transduced CML tumor cells (K562), SW480, and EBV-LVL cells with either CD4⁺ TEG011_CD8α, CD4⁺ TEGLM1_CD8α, or CD4⁺ TEG011 (without introduction of the CD8α receptor). Healthy T cells and TEG011 bulk cells (with CD4:CD8 1:1 ratio) were used as the untransformed mock target and positive effector control, respectively ( Figure 1D ). CD4⁺ TEG011_CD8α cells produced a significantly higher IFNγ level compared to CD4⁺ TEG011, which was equivalent to those of TEG011 bulk cells against all tumor targets, without affecting healthy cells. The equivalent IFNγ level between CD4⁺ TEG011_CD8α and TEG011 bulk cells comprised of only 50% CD8⁺ TEG011 implied that reprogrammed CD4⁺ TEG011_CD8α are surprisingly poorer cytokine secretors. Importantly, enhanced tumor recognition was restricted to CD4⁺ TEG011_CD8α cells and not CD4⁺ TEGLM1_CD8α mock cells, highlighting the specific role of CD8α as co-stimulation for the introduced FE11 γδTCR. We concluded that introduction of transgenic CD8α receptor in combination with Vγ5Vδ1TCR derived from clone FE11 allowed reprogramming of CD4⁺ T cells towards HLA-A*24:02-expressing tumor cells in vitro, though activity was lower when compared to CD8⁺ TEG011.

In previous studies, we have shown TEG011 efficacy against HLA-A*24:02-expressing tumor cells in vitro and an extended in vivo safety profile, as well as peripheral persistence of TEG011, where long-term persistence of TEG associated with reduced probability for developing extramedullary solid tumor masses in vivo (34, 40). To assess the consequence of the additional expression of TEG011_CD8α, NSG transgenic mice expressing human HLA-A*24:02 (NSG-A24:02) were irradiated, received luciferase-labeled K562 HLA-A*24:02⁺ cells, and subsequently received two intravenous injections of either mock control TEGLM1_CD8α, TEG011_CD8α, or TEG011 cells. All infused TEG variants showed comparable γδTCR expression, where the transduced αβT cells expressed Vδ1⁺ TCR for TEG011 and TEG011_CD8α ( Figure S2 ). Mice were monitored for tumor burden assessed by bioluminescent imaging, T cell persistence and infiltration, as well as any other signs of discomfort. Mice were sacrificed when the humane endpoints were reached (experimental outline Figure 2A ). TEG011_CD8α-treated mice had a significantly lower tumor burden over time compared to the mock control TEGLM1_CD8α and TEG011-treated groups ( Figure 2B ), indicating superior tumor control in vivo by TEG011_CD8α. All tumor-bearing mice eventually developed tumor, and measurement of individual mouse indicating tumor growth over time for each treatment group is shown in Figures 2C, D . Despite the significant in vivo tumor control, we observed only a trend towards an improved overall survival for TEG011_CD8α-treated mice ( Figure S3 ). This could be due to limited treatment window of this mouse model contributed by aggressive tumor growth of K562 HLA-A*24:02-transduced cells.

As TEG011 cells carry CD8α-dependent Vγ5Vδ1TCR, we focused our in vivo analysis to tumor-reactive CD8-expressing TEG cells (as validated by in vitro functional T cell assay in Figure 1D ) while taking into account the non-tumor reactive CD4⁺ TEG cells. Therefore, we assessed CD8-expressing TEG cell product properties and persistence by measuring viable huCD45⁺γδTCR⁺CD8⁺ single-positive and huCD45⁺γδTCR⁺CD4⁺CD8⁺ double-positive cells (present in mock control TEGLM1_CD8α and TEG011_CD8α only) in mouse peripheral blood using flow cytometry (gating strategy depicted in Figure S4 ). TEG cells persisted up to 4 weeks after infusion in the mouse peripheral blood with biological variations between mice ( Figure 3A ). To address this interindividual variation in T-cell persistence, we analyzed separately the percentage of mice where CD4⁺ and CD8⁺ T cells reached at least 500 cells/ml in the peripheral blood over time, a threshold described previously (41) ( Figure S5A ). We observed a higher percentage of mice with persisting CD4⁺ and CD8⁺ T cells in TEG011_CD8α group when compared to mock TEGLM1_CD8α and TEG011 group. Despite some imbalance in the CD4:CD8 ratio with lower numbers for CD8⁺ TEG011 infused ( Figure S2 ), more CD8⁺ TEG011 persisted over time when compared to CD8⁺ single-positive TEG011_CD8α. Vice versa, endogenous CD4 T cells for TEG011_CD8α were lower before infusion when compared to TEG011 prior to infusion, while CD4⁺CD8⁺ double-positive TEG011_CD8α were higher in numbers over time when compared to both CD4⁺CD8⁺ double-positive TEGLM1_CD8α and CD4⁺ TEG011 cells ( Figure S5B ). As a net effect, we observed more CD8-expressing T cells for TEG011_CD8α cells when compared to TEG011 ( Figure 3B ). Next, we investigated the expression of PD1 and TIM3 on CD8⁺ single-positive cells and CD4⁺ single-positive or CD4⁺CD8⁺ double-positive cells. Higher numbers of T cells expressing PD1 or TIM3 were observed on TEG011_CD8α cells, as compared to mock TEGLM1_CD8α and TEG011 cells ( Figures S6A, B ). CD8⁺ single-positive TEG011 and TEG011_CD8α showed an increased PD1 expression when compared to CD8⁺ single-positive TEG_LM1 ( Figure S6A ). A partial decline of TIM3 expression was most pronounced over time in CD8⁺ single-positive TEG011_CD8α ( Figure S6B ).

Next, we investigated infiltration of TEG cells into spleen and bone marrow on weeks 1 and 2 after infusion. Specifically, we compared the TEG011 and TEG011_CD8α groups to elucidate the contribution of transgenic CD8α co-expression in TEG011 infiltration in vivo, and focused on the total sum of CD8-expressing TEG011 cells. We detected a significantly higher number of CD8-expressing TEG cells infiltrating in the spleen and bone marrow of TEG011_CD8α-treated mice at both time points ( Figure 3B ). Importantly, we did not observe rapid clearance of CD4⁺CD8⁺ double-positive TEG011_CD8α cells in these tissues within these time points, whereas CD8⁺ single-positive TEG011 cells were barely detected. Thus, we conclude that CD8α co-stimulation with TEG011 improves overall in vivo tumor control, T cell persistence, and infiltration of CD8-expressing TEG011 cells.

We previously reported an extensive in vivo safety profile of TEG011 against healthy tissues that express HLA-A*24:02 molecules, in which no significant histological lesions were observed in major organs, including liver, spleen, and intestine (40). For histopathology analysis, we collected a femur bone marrow section from each treatment group at the end of the study period to further evaluate antitumor efficacy of the new TEG011_CD8α cells ( Figure 4A ). Tissue sections were assessed for the presence and extension of the neoplastic foci composed by round, large, undifferentiated tumor cells. The mock control TEGLM1_CD8α-treated group showed evident 19,2% neoplastic infiltration, whereas the TEG011-treated group showed up to 3,4% neoplastic infiltration of a homogeneous population of neoplastic cells in the bone marrow. Interestingly, we did not observe any neoplastic infiltration in the bone marrow of mice in the TEG011_CD8α group, and the appearance of bone marrow cell composition and cellularity was normal ( Figure 4B ). In conclusion, although the number of analyzed bone marrows was limited, our data imply that TEG011_CD8α effectively cleared tumor cells in bone marrow, emphasizing the role of CD8α co-stimulation for better in vivo tumor control of TEG011 cells. Overall, our data indicate that introduction of transgenic CD8α on TEG011 cells effectively improves in vivo tumor control and better T cell infiltration into bone marrow.