γδ T cell-mediated cytotoxicity against patient-derived healthy and cancer cervical organoids

Cervical cancer is a leading cause of death among women globally, primarily driven by high-risk papillomaviruses. However, the effectiveness of chemotherapy is limited, underscoring the potential of personalized immunotherapies. Patient-derived organoids, which possess cellular heterogeneity, proper epithelial architecture and functionality, and long-term propagation capabilities offer a promising platform for developing viable strategies. In addition to αβ T cells and natural killer (NK) cells, γδ T cells represent an immune cell population with significant therapeutic potential against both hematologic and solid tumours. To evaluate the efficacy of γδ T cells in cervical cancer treatment, we generated patient-derived healthy and cancer ectocervical organoids. Furthermore, we examined transformed healthy organoids, expressing HPV16 oncogenes E6 and E7. We analysed the effector function of in vitro expanded γδ T cells upon co-culture with organoids. Our findings demonstrated that healthy cervical organoids were less susceptible to γδ T cell-mediated cytotoxicity compared to HPV-transformed organoids and cancerous organoids. To identify the underlying pathways involved in this observed cytotoxicity, we performed bulk-RNA sequencing on the organoid lines, revealing differences in DNA-damage and cell cycle checkpoint pathways, as well as transcription of potential γδ T cell ligands. We validated these results using immunoblotting and flow cytometry. We also demonstrated the involvement of BTN3A1 and BTN2A1, crucial molecules for γδ T cell activation, as well as differential expression of PDL1/CD274 in cancer, E6/E7+ and healthy organoids. Interestingly, we observed a significant reduction in cytotoxicity upon blocking MSH2, a protein involved in DNA mismatch-repair. In summary, we established a co-culture system of γδ T cells with cervical cancer organoids, providing a novel in vitro model to optimize innovative patient-specific immunotherapies for cervical cancer.


Introduction
According to latest WHO data, cervical cancer is the fourth common cancer among women worldwide (1).The majority of cervical cancers are strongly associated with persistent infection with high-risk oncogenic human papilloma virus (HR-HPV) (2,3).Among HR-HPVs, HPV16 and HPV18 account for about 70% of cervical cancers (4).Yet, additional infectious agents have also been implicated as co-drivers of cervical cancer (5).Considering that only a quarter of cervical cancer patients respond to chemotherapy, the development of more personalized therapies is urgently needed.Recent advances in organoid technology have enabled the culture of human ectocervical organoids that faithfully replicate healthy and cancerous tissues (6).These organoids serve as a promising in vitro model, preserving original cellular heterogeneity and accurately recapitulating epithelial architecture and functionality.Recognizing the absence of in vitro models for HPV, we have earlier reported the transformation of healthy cervical organoids with HPV-derived oncogenes E6 and E7 via lentiviral transfer (5).These transformed organoids exhibit significant deviations from their corresponding healthy controls (5).
A numerically minor yet important subset of T lymphocytes in the peripheral blood endowed with anti-cancer activity is gd T cells (7,8).gd T cells differ from conventional ab T cells in at least two important aspects: (i) gd T cells do not recognize peptides presented by MHC class I or class II molecules but rather recognize phosphorylated intermediates ("phosphoantigens") of the cholesterol synthesis pathway and additional unconventional ligands; and (ii) gd T cells do not require MHC molecules for antigen recognition, which opens the possibility of applying gd T cells across HLA barriers (9,10).gd T cells express CD3-associated T cell receptors (TCR) composed of g and d chains (11,12).Like other cytotoxic effectors, gd T cells directly participate in the elimination of tumour cells, but they also indirectly control the tumour immune response by modulating the activity and functions of other immune cells (7).In healthy adults, approximately 1-5% of circulating T lymphocytes express Vd2 paired with Vg9; the proportion of such Vd2 T cells can rapidly increase during the acute phase of several infectious diseases (11,12).Recent studies have shed light on the mode of recognition of phosphoantigens (pAg) by gd T cells.In this context, an indispensable role of members of the butyrophilin (BTN) family of transmembrane molecules has been discovered.According to the current model, microbe-or tumour-derived pyrophosphates bind to the intracellular B30.2 domain of BTN3A1 which then interacts with BTN2A1 to induce a conformational alteration of the extracellular BTN3A1/2A1 complex which is recognized by the gd TCR (13)(14)(15)(16).
gd T cells have well-established protective roles in cancer (8,17,18).Multiple mouse models have demonstrated their protective role in cancer progression (17,19).However, previous studies so far have only investigated the effect of gd T cells on cervical cancerderived cancer cell lines and tissue slices (17,20) or utilized NK cells (21), while studies comparing their effects on patient-derived organoids have not yet been reported.In this study, we provide a model to assess gd T cell-mediated cytotoxicity by live cell imaging and flow cytometry-based assays.

Isolation of cervical tissue and cervical organoid culture
Human cervical specimens were obtained from volunteers undergoing standard surgical procedures at the Department of Gynaecology, CharitéUniversity Hospital, and August-Viktoria Klinikum, Berlin (Ethics Approval EA1/059/15 from local authorities), informed consent was obtained from all donors.Samples were processed within 2-3 hours after removal.
Organoids were maintained by passaging every 1-2 weeks at a 1:2-1:5 ratio using enzymatic digestion with TrypLE followed by mechanical singularization of cells using a syringe with an 18G needle (BD medical, BDAM303129) as described earlier (22).

2D cervical stem cell lines culture
Passage 0 epithelial cells were cultured in plastic cell culture vessels without an irradiated 3T3-J2 monolayer and seeded in a Collagen I (1:100, Fisher Scientific, A1064401) pre-coated T25 flask.After most epithelial cells attached to the surface, the medium was changed every 2-3 days until confluence reached 80-90%.The cells were then passaged at a ratio of 1:2.For the passage 0, the medium was removed and the cells were washed three times with PBS (Capricorn, PBS-1A) and dissociated with 1 mL TrypLE (for each T25 flask) incubating 10 min at 37°C.The cells were detached from the flask surface and the cell suspension was collected with 10 mL Advanced DMEM/F12.After centrifugation at 400g for 4 min at 4°C, the cell pellet was resuspended with cervical organoid medium and seeded on irradiated 3T3-J2 monolayers.
From passage 1, ectocervical epithelial cells were cultured with an irradiated 3T3-J2 layer and the medium was refreshed every 2-3 days.When it reached 80%-90% confluency, the cells were enzymatically detached twice.The first time, 0.5 mL TrypLE was used for each T25 flask for 1 min to detach the 3T3-J2 feeder cells, and then the cervical epithelial cells were detached with another 1 mL TrypLE and incubated for 10 min.

Whole mount immunofluorescence assays (IFAs) and microscopic analyses
Phase contrast and brightfield images were taken using an IX50 (Olympus) microscope with a 4x or 10x objective.Images were contrast adjusted and scale bars were added using FIJI (ImageJ; RRID : SCR_002285) (23).
Signal was recorded using a ChemiDoc MP Imaging System (Biorad).

Vg9Vd2 T cell expansion
Vg9Vd2 T cells were expanded from peripheral blood mononuclear cells (PBMCs) isolated from leukocyte concentrates of healthy blood donors using a Ficoll-Hypaque (Biochrom, Biocoll L 6113/5) density gradient according to the manufacturer's protocol.Leukocyte concentrates were provided by the Institute of Transfusion Medicine UKSH Campus Kiel and were used in an anonymized fashion and Ethics Approval D 434/22 was provided by local authorities.The cell pellet was resuspended in RPMI medium (Capricorn, RPMI-STA) containing 10% FCS (Sigma, F7524).Before expansion, the initial Vg9Vd2 T population among CD3+ T cells was checked by staining PBMCs with anti-Vd2-FITC and anti-CD3-APC.Samples with a Vg9Vd2 T cell proportion over 2% were deemed acceptable and the sample was further processed.Vg9Vd2 T cells were selectively activated by stimulating PBMCs with 5 mM zoledronic acid (Novartis) and 50 IU/mL IL-2 (Novartis) over the expansion period of 14 days essentially as described (26).IL-2 was supplemented every other day.Purity of Vg9Vd2 T cell lines was checked with anti-Vg9-FITC and anti-CD3-APC.Lines were used for experiments when Vg9Vd2 T cells represented more than 90% of the total cell population.Antibodies and dilutions are listed in Supplementary Table 1.

Vg9Vd2 T cell co-culture with cervical cells
The B-cell lymphoma Daudi cell line (ATCC, CCL-213; RRID : CVCL_0008) with known sensitivity to Vg9Vd2 T cell killing (27) was included as positive control.All lines were washed with PBS and pelleted in 15 mL falcon tubes by centrifuging at 1400 rpm for 5 min.Cells were stained with PKH67 (Sigma, PKH67GL) green dye was prepared according to manufacturer protocol.After incubation at RT for 4 min in the dark, 7 mL RPMI medium containing 10% FCS was added and incubated for further 1 min to stop the staining.Cells were used at a ratio of 10:1 (T cells: epithelial cells).Co-culture was performed for 4 hours with the respective surface blocking mAb, if appropriate, and thereafter cells were washed in PBS, and taken up in sample buffer containing 0.2 mg/mL propidium iodide (PI) for 20 min at 4°C and analysed by flow cytometry.Data was then analysed with FlowJo v10 (RRID : SCR_008520).The absolute number of viable epithelial cells (FITC+PI−) was calculated with the formula derived from Sacchi et al. ( 28): Cytotoxicity = 100 − ½(% of living epithelial cells in co À culture with gd T cells=% of living epithelial cells without gd T) Â 100 2.8 CD107a degranulation assay 50 × 10 3 effector cells were cultured alone or together with 5 × 10 3 tumour cells for 4 h at a 10:1 ratio in the presence of anti-CD107a-PE mAb (clone: H4A3; BD Biosciences) and monensin (3 mM; added after 1 h).Thereafter cells were washed twice in PBS and analyzed by flow cytometry.Where appropriate, the synthetic pAg bromohydrin pyrophosphate (BrHPP; Innate Pharma) was added at 300 nM to activate Vg9Vd2 T cells.Cells were immediately acquired on a BD FACSCanto I flow cytometer.Data was then analysed with FlowJo v10.

Live cell imaging
Prior to co-culture, a flat-bottom 384 well-plate (black with clear bottom, Corning) was coated with Poly-2-hydroxyethyl methacrylate (poly-HEMA; Sigma, P3932) in 100% ethanol as described earlier (29).30 μl poly-HEMA was added in each well and the plate was left to dry overnight under the culture hood.The plate was washed with PBS immediately before use.
The imaging protocol was adapted from previous studies (30, 31).Organoid lines were cultured for around seven days and Vg9Vd2 T cells were expanded for 14 days as described above.Matrigel was removed by incubation with cell-recovery solution for 30 min at 4°C.A part of the suspension of organoids was taken and digested into single cells using TrypLE then the number of single cells was counted to estimate organoid equivalents.The organoids were washed with cold PBS and thereafter organoids were collected and stained with CellTrace CFSE Cell Proliferation Kit (Thermo, C34554) according to the supplier's protocol.Then, Vg9Vd2 T cells were added again at a ratio 10:1 (T cells: epithelial cells).NucRed Dead 647 ReadyProbe Reagent (Thermo, R37113) was added to distinguish dead and live cells.In the respective conditions, gd T cells were activated with 300 nM BrHPP, "Dead cells" were boiled for 10 min at 95°C before the start of the experiment, and 1 mg/mL Puromycin (InVivoGen, ant-pr-1) treated cells served as death control over time.Measurements were performed using the CELLAVISTA 4 automated cell imager in combination with the SYBOT X-1000 with CYTOMAT 2 C-LiN system (all SYNENTEC GmbH).Wells were imaged every hour for a total of 8 hours and afterwards fluorescence data and images were extracted with YT-Software (SYNENTEC GmbH) using the Real Cytoplasm (2F) application.The settings were modified to detect all organoids in the green channel and subsequently, the average intensity of the yellow and red signal within each spheroid area was analysed.The following settings were used for different channels.Exciter: Blue (475/28) -Emissionfilter: Green Filter 530nm (530/43), Exciter: Green (529/24) -Emissionfilter: Amber Filter 580nm (607/70), Exciter: Red (632/22) -Emissionfilter: Far Red Filter 670nm (685/ 40).Scatter plots were assembled in R (v4.1.0)with ggplot2 (v3.4.0).

2.10
Live/dead cytotoxicity assay 1x10 4 cervical organoid cells were seeded into Collagen I precoated 96-well plates without feeder cells.On the next day, cells were co-incubated with 1x10 5 gd T cells (EC : TC = 1:10) for four hours in Opti-MEM (Gibco, 11058021).To quantify dead/live cells, the MultiTox-Glo Multiplex Cytotoxicity Assay kit (Promega, G9272) was used according to manufacturer's instructions.The values were normalized to the respective untreated control, and the ratio of dead/live cells was then normalized to the healthy organoids to obtain fold changes.

Analysis of bulk RNA sequencing
Quality of raw reads was assessed by FastQC.Reads were mapped to the human reference genome (assembly GRCh37) using the splice aware aligner STAR (v2.7) (32) in two-pass mode.Quality GTF files downloaded from Gencode (v28) (33), were used to map reads into genes with FeatureCounts in order to obtain raw counts.From this matrix of raw counts, genes with no expression across samples were filtered out, and differential gene expression was performed using DESeq2 (v1.32.0) (34) in R (35).Volcano plots were generated using ggplot2 (v3.4.0).Genes were ranked based on their log2 fold change compared to healthy controls, and gene set enrichment analysis (GSEA) was performed using the fgsea implementation from the clusterProfiler R package (v4.0.5) (36).Gene sets from Hallmark (H), Curated Gene Sets (C2) and GO Biological Process (C5-BP), retrieved from MsigDB (37) using msigdbr (v7.5.1) were studied.The gene expression heatmaps of the TPMs were generated using the pheatmap package (v1.0.12).Principal Component Analysis (PCA) was performed on the TPM normalized counts and the plots were generated using factoextra (v1.0.7) on the result of the pca function in base R.All the analyses were performed in R (v4.1.0).RNA-seq DGE and GSEA analysis results are included in Table 1.

Statistical analysis and panel composition
Basic calculations were performed using MS EXCEL 2016 (Microsoft).Figures were plotted using 3 Results

Ectocervical organoids accurately represent the phenotype of the parental tissue
To establish a suitable model mimicking ectocervical tissue, we have established patient-derived organoids (PDOs) cervical tissues in our earlier studies (5,6,22).These lines were grown in T25 tissue flasks on irradiated 3T3-J2 mouse fibroblast feeder cells and subsequently embedded in Matrigel to generate 3D organoids (Figure 1A).In order to create paired HPV E6E7-positive lines (referred to as HPV+), the healthy organoid lines were transformed with a construct containing human papilloma virus (HPV) oncogenes E6 and E7 as described earlier (5,22).We were able to passage all lines for approximately 20 passages prior to growth arrest.Visual inspection under brightfield microscopy revealed no discernible phenotypic difference between healthy, HPV+ and cancer lines (Figure 1B).Amplification of integrated HPV16 E6E7 oncogenes was detected by PCR of genomic DNA in the HPV+ and cancer organoids (Figure 1C).Since HPV16 and 18 are the most common causes of cervical cancers, we also assessed the HPV18 status in our organoids.HPV18 was detected in one of the lines (Cancer D3), also reflecting the naturally occurring ratio of HPV16 and HPV18 infections of two to one (38).HPV18 primer sensitivity was verified using HeLa cells as a positive control (Figure 1C).We have summarized the characteristics of the organoid lines in Table 1.
To validate the ectocervical nature of our organoids, we next characterized the organoid lines for marker gene expression and localization by whole mount immunofluorescence assays.All lines were positive for cytokeratin 5 (KRT5), a marker of ectocervix, in the outer layers of the organoids resembling the in vivo stratified squamous epithelium of the parental tissue.As the basal layer is responsible for proliferation in vivo, marker of proliferation KI67 (KI67)-positive cells are expected in this layer.Interestingly, KI67positive cells were indeed detected in the parabasal layers of healthy PDOs, while they were located throughout the whole organoid for HPV+ and cancer-derived lines (Figure 1D).A more detailed characterization of ectocervical organoid cultures has been previously provided in our studies (5,6).Thus, the organoids reflect the in vivo phenotypes of healthy and cancerous ectocervical tissues.

gd T cells exhibit enhanced killing of HPV+ and cancer-derived organoids
gd T cells are an important subset of T lymphocytes involved in anti-tumour immunity.They can kill tumour target cells by the secretory pathway via the release of perforin and granzymes as well as by death receptor pathways (e.g.Fas/Fas-ligand) (11).To first assess whether the tissue identity is also reflected in the immune response of tumour-killing T cell subsets, we co-incubated in vitro expanded Vg9Vd2 T cells with healthy, HPV+ and cancer-derived organoid lines.Before co-incubation, the purity of gd T cell lines was assessed; only cultures with > 90% gd T cells were used.Representative examples of purity determinations as analysed by flow cytometry are presented in Supplementary Figure 1.To determine gd T cell-mediated cytotoxicity, organoids were stained  with Nucred Dead 647 (Yellow: general stain, Red: dead cells) as previously described for breast cancer-PDOs (31).This staining enabled the detection of an increasing number of dead cells within the organoid population by automatized live cell imaging.Live cell imaging revealed distinct responses of gd T cells to HPV+ and cancer-derived organoid lines after four hours of co-incubation (Figures 2A-C).The average accumulation of red stained dead cells in these groups was higher in comparison with healthy organoids, as shown by an increased ratio of Dead/Live signal (Figures 2A-C).
When visualizing individual organoid's Dead/Live (Red/yellow) ratios as scatter plots, a clear shift to the Red/Dead signal was observable in cancer and HPV+ organoids when co-incubated gd T cells, even in the absence of the pAg BrHPP (Figure 2B).We observed a significant increase in the ratio of red/dead to yellow/live signal in cancer and HPV+ organoids in comparison to healthy organoids after 4 hours after analysing single organoids (Figure 2C).Consistently, additional activation of gd T cells by BrHPP resulted in enhanced cytotoxicity in all organoid lines (Figure 2B; Supplementary Figure 2).Unmerged scatters plots for untreated, unstained, puromycin-treated and dead cells are provided in Supplementary Figure 2. To further confirm the findings, organoid-derived cells were co-incubated with gd T cells and cytotoxicity was assessed by an alternative, protease-based cytotoxicity assays after 4h.Higher cytotoxicity was again observed in cancer PDOs after co-incubation with unstimulated gd T cells (Figure 2D, left), while treatment with BrHPP additionally resulted in a significant increase in gd T cell-mediated cytotoxicity against HPV-transformed organoids and cancer lines (Figure 2D, right).
To confirm the differences in gd T-mediated cytotoxicity induced by gd T cells against healthy and HPV+ and cancerderived cervical organoids, we determined degranulation/Lamp-1 mobilization of gd T cells by flow cytometry.As positive control we included the B lymphoblast cancer cell line Daudi.As expected, we could show that the addition of BrHPP significantly enhanced degranulation in all conditions (Figure 2E, right).However, the addition of BrHPP also resulted in a significant increase in degranulation of HPV+ and cancer organoids in comparison with healthy controls (Figure 2E).Confirming our previous results using an alternative cytotoxicity read-out, we demonstrated that unstimulated gd T cells already exhibit significantly increased cytotoxic activity against cancer and HPV+ cells in comparison to healthy controls as measured by propidium iodide (PI) staining to detect dead cells (Figure 2F).Again, the addition of BrHPP increased the overall response and significantly enhanced the killing of HPV+ and cancer organoids as compared to healthy cells (Figure 2F).

Transcriptomic and protein analysis reveals differences in molecules relevant for gd T cell activation
We have observed significant differences in gd T cell-mediated cytotoxicity against distinct organoid lines.To identify putative underlying factors facilitating increased cytotoxicity on cancer and HPV+ organoids, explorative bulk RNA-sequencing (RNAseq) was conducted.Gene expression analysis revealed differences between conditions and also between patient isolates (Figure 3A; Table 1).Cancer lines exhibited comparable expression of genes that were distinguishable from the other conditions, as shown by the first principal component (Figure 3A; Supplementary Figure 3).Isolates transformed with HPV oncogenes E6E7 grouped as intermediates between healthy and cancer organoids (Figure 3A; Supplementary Figures 3, 4).HPV+ isolates generally moved into the cancer direction on the first principal component.Interestingly, the transformation of HPV+ D3 resulted in a clear shift in the second principal component from the other lines.Differential gene expression analysis showed a strong upregulation of cell proliferation-associated genes in the HPV transformed isolates such as CDC7, MCM6/7, and RFC3 (Figure 3B; Supplementary Material Table 1).It is notable that there is a clear bias of upregulation of genes in the HPV+ condition in comparison with healthy organoids (Figure 3B).Cancer organoids showed an upregulation of pathways involved in DNA-damage, cell cycle checkpoint, and DNA-binding/modification pathways and downregulation of pathways involved in differentiation of squamous epithelia (Figure 3B; Supplementary Figure 4; Table 1).No significant elevation of antiviral/IFN-associated pathways was detected.
Having gained insights into the differential expression of transcripts between conditions we further explored the possible underlying mechanisms of enhanced killing of HPV+ and cancer organoids.For this, we investigated the expression of putative gd T cell ligands and molecule involved in gd T cell activation (Figure 3C) as summarized in Kabelitz et al. (11).As expected, we observed pronounced differences in expression patterns between the conditions.Upon transformation with E6E7 gene expression patterns of healthy organoids shifted more towards the cancer direction.Some molecules associated with gd T cell activation such as the Butyrophilin (BTN)-family, UL16 Binding Protein (ULBP)-family, MHC Class I Polypeptide-Related Sequence A (MICA) and B (MICB), A nnexin A 2 (ANXA2), and Apolipoprotein A1 (APOA1 were lower in cancer organoids.In contrast, others such as protein C receptor (PROCR), and ATP Synthase Peripheral Stalk-Membrane Subunit B (ATP5F1/BP) were expressed in a higher amount than healthy controls (Figure 3C).CD1d was only highly expressed in Cancer D3.
In addition to the well characterized genes involved in gd T cell activation, we also investigated the regulation of other genes such as members of the human MutS homologue (MSH)-family, heat shock protein family D (Hsp60) member 1 (HSPD1) and Killer Cell Lectin Like Receptor G2 (KLRG2).These genes have been described to play a role in DNA mismatch repair or are involved in stress signalling of cells (39,40).Interestingly, ectopically expressed MSH2 has been previously identified as a target for human Vg9Vd2 gd T cells (11,39,41,42).We found these genes to be upregulated in HPV+ and cancer organoids (Figure 3C).Another highly upregulated gene was found to be Butyrophilin-Like Protein 9 (BTNL9) which has been implicated to be of prognostic significance in various cancer types (43,44).Its potential involvement in the regulation of gd T cell mediated killing of tumour cells is unknown but deserves further investigation in light of our current findings with HPV-transformed and cancer organoids.Additionally, programmed cell death ligand 1 (PD-L1/CD274) was significantly downregulated in cancer organoids.The PD-L1 receptor PD1 conveys a negative signal on T cells.(45).PD-L1 downregulation might therefore support the enhanced killing of cancer cells by gd T cells.
To confirm whether the regulation of transcripts could be translated into protein expression, we analysed the protein expression of selected differentially expressed genes by western blotting (Figure 3D).We confirmed that KI67 was elevated in cancer-derived organoids, consistent with earlier reports (46) and as was observed in immunofluorescence (Figure 1D).Additionally, expression of MSH2 was upregulated in HPV+ and cancer organoids and PD-L1 was downregulated in cancer organoids in comparison with healthy controls, pointing to possible functional differences in gd T cell-mediated killing.As PD-L1 needs to be expressed on the cell surface to fulfil its function, we also conducted flow cytometry to confirm our western blotting and RNAsequencing results, which, indeed, confirmed decreased surface expression of PD-L1 in cancer organoid isolates (Figure 3E).
Next, we addressed the functional relevance of some of the above molecules for gd T cell-mediated killing of HPV+ and cancer organoids.To this end, we co-incubated organoids with expanded gd T cells with BrHPP in the absence or presence of blocking antibodies against BTN2A1 and BTN3A1 (Figure 3F).While the inhibitory anti-BTN3 antibody used here (clone 103.2) does not discriminate between isoforms BTN3A1/A2/A3 (47), the isoform critically involved in gd T cell activation is BTN3A1 (48).We found that blockade of BTN2A1 or/and BTN3A1 significantly reduced killing in all conditions as was in line with earlier reports (49).However, blocking in cancer organoids seemed to be slightly less affected than in HPV+ organoids (Figure 3F).In view of the results shown in Figures 3C, D, we also addressed the potential role of MSH2 which has been identified by Dai et al. (39) as a gd T-cell ligand when ectopically expressed on the surface of tumour cells.Blocking of MSH2 by antibodies led to an overall decrease of gd T cell-mediated cytotoxicity, most notably against cancer organoids in the presence of BrHPP (Figure 3G).Blocking could decrease gd T cell-mediated cytotoxicity of BrHPP activated gd T cells to the level of unstimulated gd T cells (Figure 3G).While the inhibitory effect of anti-MSH2 antibody blocking was much lower than the effect of anti-BTN2A1/3A1 antibodies, our results suggest that multiple ligand-receptor interactions are involved in the organoid-gd Tcell interaction.
In summary, we have identified multiple possible target molecules and pathways that may contribute to enhanced the killing of HPV-infected cervical tissue by gd T cells.While the precise role of some molecules (including BTNL9) requires further investigation, the currently established organoid-gd T cell coculture system should prove useful for testing additional modulators (e.g.small molecules, antibodies, tumour sensitizers, etc.) with the ultimate goal of improving the efficacy of gd T cell immunotherapy for cervical cancer.

Discussion
The association between HPV infection and the onset and the development of cancer has been extensively studied since almost 50 years (50).However, there has been a lack of suitable models that accurately represent and maintain the characteristics of cervical tissue.Previous studies have relied on immortalized cell lines or tissue slices (20).Here, we present a model to study the interaction of tumour-reactive T lymphocyte populations with healthy, matched HPV16 E6E7 transformed, and cancer-derived primary ectocervical tissues.To achieve this, we utilized an earlier established human ectocervical organoid model including paired HPV16 E6E7 transformed organoids (5,6).We detected infectionspecific differences in proliferation patterns of healthy and HPVinfected organoids by staining with KI67 antibody.Healthy controls exhibited proliferating cells only in the parabasal regions of organoids, cancer organoids showed a wider distribution of proliferation while HPV+ showed an intermediate phenotype.
Similar was observed in HPV+ tissue and cancer biopsies confirming the preservation of tissue origin characteristics (46,51).gd T cells play a distinct role in the immune surveillance against virally infected (including HPV) cells and tumour cells (3,8,11,52).They are believed to be important in the cervical defence against HPV infection (3,53).To establish an in vitro model for in-depth analysis, we co-incubated organoid-derived cells with in vitro expanded Vg9Vd2 gd T cells isolated from healthy donors.In view of the HLA independence of gd T cells, the use of allogeneic gd T cells from healthy donors is a valuable approach also from a translational perspective.In fact, the adoptive transfer of allogeneic gd T cells from healthy donors has already been applied in phase I studies with no major adverse reactions (10).It was thus possible to compare responses of gd T cells to healthy, HPV-transformed as well as tumour organoids.Our results demonstrated elevated responses of unstimulated and phosphoantigen-activated gd T cells to HPVinfected cells in comparison to healthy controls, independently of the type of assay used.Live cell imaging, protease-dependent cytotoxicity assays, and flow cytometry-based cytotoxicity and degranulation assays demonstrated elevated responses to HPV+ and/or cancer organoids.However, the observed differences in the potency of the assays might reflect differential involvement of the secretory (perforin and granzyme-dependent) and death receptor pathways (e.g., TRAIL, Fas/CD95).It is well known that gd T cells can utilize both pathways to kill tumour targets (54,55).The organoid co-culture system established in our current study will allow us to dissect the precise mechanisms of killing of cancer versus HPV-infected organoids by gd T cells in future studies.This will also include the analysis of soluble mediators such as TRAIL, Fas-ligand, granzyme B, IFNg, and others.
To unravel possible mechanisms behind the responses of gd T cells to HPV16-infected organoids, we conducted bulk RNA sequencing.The RNAseq revealed differences between HPV infected vs healthy controls.Importantly, HPV-transformed organoids displayed an intermediate state between healthy and cancer organoids with replicating some, but not all patterns of cancer tissue.Consistent with earlier studies we detected evidence for HPV16-induced DNA damage (5,56).Interestingly, HPVinduced antiviral responses including the type I interferon pathway were negligible in the pathway analysis.Dekkers et al. (31) have earlier reported a high association between cancer cell killing and the preservation of tumour-specific inflammatory features in breast cancer organoids.In their study, sensitivity to cancer metabolome-sensing engineered T cells (TEGs) was highly associated with the expression of antiviral genes such as MX1, IFIT1, OASL and XAF1 (31).
Our pathway analysis demonstrated some overlap in DNAmodification pathways and gd T cell ligands or associated molecules between cancer and HPV+ conditions.We reproduced earlier reports (57, 58) of altered transcription and expression of genes such as TTK protein kinase (TTK), maternal embryonic leucine zipper kinase (MELK), forkhead box M1 (FOXM1), mismatch repair enzyme as mutS homolog 2 (MSH2), matrix metalloproteases (MMPs) and programmed cell death ligand 1 (PD-L1/CD274).While up-regulation of PD-L1 is typically associated with poorer outcomes in chemotherapy treatment (58), it was almost absent in our cancer isolates.While exhaustion the role of Programmed Cell Death 1 (PD-1/CD279), Hepatitis A Virus Cellular Receptor 2 (TIM3/HAVCR2), Lymphocyte Activating 3 (LAG3) and other checkpoint molecules has been less studied in gd T cells compared to conventional ab T cells (59), the low to absent PD-L1 expression on cancer organoids may favour persistent effector activity of gd T cells.Interestingly, the expression of BTN-family members including BTN2A1/3A1, crucial in pAg-mediated gd T cell activation was largely down-regulated in the cancer isolates.However, our blocking studies using inhibitory antibodies against BTN2A1 and BTN3A1 clearly indicated the significance of these BTN molecules in triggering the cytotoxic activity of gd T cells against HPV+ and cancer organoids., in line with previous reports (49).While not directly measured in our study, it can be assumed that HPV+ and cancer organoids produce increased amounts of endogenous pAg isopentenyl pyrophosphate (IPP), leading to BTN2A1/3A1-dependent gd T cell activation and subsequent tumour cell killing (8,11,60).It is noteworthy, however, that blocking of BTN2A1/3A1 pathways by inhibitory antibodies only partially reduced the killing of organoids.Given that such antibody blockade completely wipes out all phosphoantigen-mediated activation of human gd T cells, this strongly suggests that other pathways are involved in organoid-gd T cell interaction, one of which might be MSH2.Additionally, transcripts of BTNL9 and cellular stress molecules HSPD1, KLRG2, and DNA damage associated molecule MSH2 were upregulated in cancer organoids.While the roles of HSPD1 and KLRG2 play a role in regulating gd T cell-mediated cytotoxicity in cancer have been described (61, 62), there is limited information about the involvement of BTNL9 in gd T cell-mediated killing of cancer cells.Given our results, further investigation is warranted.Moreover, the upregulation of MSH2 may be functionally relevant.MSH2 ectopically expressed on the surface of tumour cells has been identified as a ligand for the human Vg9Vd2 TCR (39,41,42) Our study demonstrated that the killing of HPV+ and cancer organoid cells by BrHPP-activated gd T cells was reduced in the presence of anti-MSH2 antibody, supporting the hypothesis that the increased levels of MSH2 contribute to sensitizing of HPV+ and cancer organoid cells to cytotoxicity mediated by gd T cells.Taken together, our findings suggest that BTN2A1/3A1 and MSH2 are involved in triggering the cytotoxic activity in gd T cells towards HPV+ and cancer organoids, but additional pathways might be involved.
In summary, we have established a suitable model to study cellular immune responses to healthy, HPV-transformed and cancerous ectocervical tissue using organoids in combination with short-term expanded gd T cells.Co-incubation of organoid-derived cells with gd T cells resulted in enhanced killing of HPVtransformed and cancer organoids.We demonstrated the involvement of BTN-family members BTN2A1/3A1 in gd T cellmediated cytotoxicity and obtained evidence for the additional involvement of human MutS homologue (MSH)-family member MSH2.The presented in vitro model will be highly valuable for indepth analysis of additional pathways and mechanisms of how to enhance gd T cell effector function with the ultimate goal of improving gd T cell immunotherapy of cervical carcinoma.Such strategies might include (but are not restricted to) sensitizing cancer cells with gd T cell activating drugs like zoledronate (63), bispecific gd T cell engagers (64), or agonistic anti-BTN3A1 antibodies (65).

1
FIGURE 1 Characterization of healthy and cancer-derived ectocervical organoid cultures.(A) Schematic representation of the establishment of organoids derived from ectocervical tissue.(B) Representative brightfield images of three-dimensional (3D) and two-dimensional (2D) cultures of healthy ectocervical organoids, HPV oncogene E6E7-transformed paired organoids (HPV+) and cancer-derived organoids.Scale bars represent 50 µm for 3D organoids and 20 µm for 2D cultures.(C) HPV16 and HPV18 status of the patient-derived organoid cultures.Genomic DNA was amplified by RT-qPCR and DCt values were calculated with GAPDH as housekeeping gene.Agarose gel electrophoresis was performed to determine HPV status.AGS0 and HeLa cells served as negative controls for HPV16; HeLa cells additionally served as a positive control for HPV18 infection.HPV+ and Cancers D1 and D2 showed the presence of integrated HPV16 oncogenes E6E7.Cancer D3 showed integration of HPV18.RT-qPCR experiments show mean (± 95% CI) from ≥ 3 technical replicas of two independent biological replicates.(D) Representative immunofluorescent images of 3D organoids.Nuclei were stained with Hoechst33342 (blue), ectocervical tissue with cytokeratin 5 (KRT5, green) and proliferating cells with marker of proliferation KI67 (KI67, red).Scale bars represent 100 µm for HPV+ organoids and 200 µm for healthy and cancer organoids.

2 gd
FIGURE 2 gd T cells specifically target HPV-positive and cervical cancer organoids.(A) Representative fluorescent images of ectocervical organoids either untreated or co-incubated with gd T cells for four hours.gd T cells were activated with 300 nM BrHPP, "Dead cells" were boiled for 10 min at 95°C before start of the experiment, and Puromycin-treated cells served as death control over time.Cells were stained with NucRed Dead 647 Ready Probe Reagent.Yellow marks viable cells, red marks dead cells.Scale bars represent 200 µm.The experiment was performed at least twice with similar results.(B) Scatter plots of dead/live (red/yellow) cell staining in individual organoids show the increasing shift of yellow to red staining of HPV+ and cancer organoids.Merge of all conditions for organoids incubated with gd cells in the absence of 300 nM BrHPP is seen in the box at the lower right.Merges of the control conditions are given.Unmerged control conditions are depicted in Supplementary Figure 2. (C) Ratio of dead/live (red/yellow) cell staining of individual organoids after four hours.Organoids with a ratio > 10 were excluded from the graph for visual purposes.Data shows ≥ 40 organoids per condition.Statistical significance between conditions was determined using a Kruskal-Wallis test with Dunn's correction for multiple testing.***p < 0.001 ****p < 0.0001 (D) Quantification of cytotoxicity measured by MultiTox-Glo Multiplex Cytotoxicity Assay after four hours of co-incubation of organoids with gd T cells in the absence or presence of 300 nM BrHPP as indicated.Data was normalized to untreated or BrHPP-treated controls and the ratio of dead/live cells was calculated; fold-change to healthy control was assessed.Data is derived of 12-20 replicates per condition from two independent experiments.(E) Quantification of degranulation of gd T cells by CD107a staining with and without additional activation of gd T cells by BrHPP using flow cytometry.The B lymphoma cell line Daudi was included as a positive control.Data shows four independent experiments.(F) Quantification of cytotoxicity induced by gd T cells by propidium iodide (PI) staining with and without activation of gd T cells by BrHPP using flow cytometry.Data shows ≥ three independent experiments run in triplicates.All box plots show Median with Tukey whiskers.Statistical significance in (D-F) was determined using a Two-Way ANOVA with Tukey's correction for multiple testing.ns not significant *p < 0.05 ****p < 0.0001.

3 gd
FIGURE 3 gd T cell mediated killing is partially mediated by BTN family members and MSH2.(A) PCA analysis of bulk RNA sequencing of the organoid lines.(B) Volcano plots of differential gene expression analysis, showing mean Log2 fold-change vs -log10 adjusted p values.HPV16 integration alters the organoid transcriptome.RNA sequencing was performed for 3 independent patient organoid isolates per condition.Downregulated genes are marked in blue, differentially regulated genes with higher p values in green, not significantly regulated genes in grey, and upregulated genes in red.Top10 most significantly up-and downregulated genes are highlighted.(C) Heatmap of the gene expression of gd T cell ligands highlighting the downregulation of BTN family members and the upregulation of DNA damage associated molecules in the cancer isolates.(D) Confirmation of RNAseq results by Western blotting highlighting the differential expression of selected proteins in cancer organoids.(E) Dot plot of PDL1/CD274 (Y axis) and HLA class I (X axis) expression analysed by flow cytometry on organoids as indicated, confirming protein expression results in (D).(F) Quantification of cytotoxicity in organoid lines by propidium iodide staining with and without blocking antibodies for BTN2A1/3 by BrHPP activated gd T cells using flow cytometry.Data shows mean (± 95% CI) of three independent experiments.(G) Representative quantification of a flow cytometry experiment using anti-MSH2 antibody (2-4 replicates per condition).The percentage of propidium iodide-positive cells is shown.Statistical significance in F and G was determined using a Two-Way ANOVA with Tukey's correction for multiple testing.ns not significant *p < 0.05 ***p < 0.001.

TABLE 1
Characteristics of the organoid lines.
HPV+ are derived from the respective healthy lines.