Human PBMC from healthy donors were isolated by ficoll density gradient centrifugation and used for CSP transduction, isolation of B cells and monocytes. The blood collection was done with approval by the institutional review board of Ludwig-Maximilians-University, Munich, Germany. Donors gave informed consent in accordance with the Declaration of Helsinki. PBMC used for B cell isolation and DC generation were from HLA-A2 negative donors such that the B cells or DCs did not provide antigen ligands for the TCR-T58, TCR-D115 or TCR53- tgTCR-T cells in co-culture experiments.

Primary B cells were isolated from PBMC using the CD19 magnetic (negative isolation) Naïve B Cell Isolation Kit II, human (Miltenyi Biotec, Bergisch Gladbach Germany), according to the manufacturer´s instruction. Primary B cells were used freshly prepared for the B cell activation assay. Monocytes were derived from PBMC by plastic adherence and were differentiated to immature dendritic (iDCs) by adding 100 ng/ml GM-CSF and 20 ng/ml IL-4 (both Promokine, PromoCell, Heidelberg, Germany) to VLE RPMI-1640 supplemented with 1% L-glutamine, 1% non-essential amino acids, 1% sodium pyruvate, 1% penicillin/streptomycin and 5% human serum (DC-medium) on day 1 and day 3. iDC were obtained at day 7 and used for co-culture with T cells. Mature DC (mDC) were generated from day7 iDCs by adding Jonuleit maturation cocktail (20 ng/ml IL-4 (PromoCell), 100 ng/ml GM-CSF (PromoCell), 15 ng/ml IL-6 (Sigma-Aldrich), 10 ng/ml IL-1β (PromoCell), 20 ng/ml TNFα (Immunotools) and 1 µg/ml PGE₂ (PromoCell) for 24 h (51).

Monocytes were used to generate ercDC (enriched in renal cell carcinoma) DCs as described (52). Briefly, monocytes from human PBMC were cultivated in DC medium supplemented with 20% RCC-26 conditioned medium on days 1, 3 and 5 for 7 days. RCC-conditioned medium was derived from 2 x 10⁶ renal carcinoma cell line RCC-26 cultivated for 10 days in serum-free AIM-V medium (Gibco/Invitrogen, Carlsbad, CA), then centrifuged and filtered to remove cells (52).

T cells used for retroviral transduction were either primary human T cells from PBMC or human T cells expressing transgenic TCRs (tgTCR)-T cells either with HLA-A2 restricted specificity for the tyrosinase peptide (TCR-T58 or TCR-D115) (53) or with HLA-A2 restricted renal cell carcinoma (RCC) specificity (TCR53) (19, 54). T cells were cultured in RPMI1640 supplemented with 1% L-glutamine, 1% non-essential amino acids, 1% sodium pyruvate, 1% penicillin/streptomycin (RPMI-basic) and 10% human serum (T cell medium, TCM) with human IL-2 (Cancernova GmbH, Reute, Germany) at concentration as indicated.

Cell lines were the human melanoma line SK-Mel23 (ATCC HTB71, from M.C. Panelli, NIH/Bethesda, USA), the human RCC cell lines RCC-53 and RCC-26 (isolated from RCC tissue in our laboratory) (54), the HEK293/Tyr cells transduced in our laboratory to express tyrosinase (HEK/Tyr) (19) and HEK293/Tyr/CD40, which are HEK293/Tyr cells subsequently transduced to express CD40. For HEK293/Tyr and HEK293/Tyr/CD40 single-cell clones were selected for comparable HLA-A2 and tyrosinase expression. HEK293 cells with stable expression of CD40L (HEK293/CD40L) were kindly provided by Kathrin Gärtner, Helmholtz Center Munich. All cell lines were grown in RMPI-basic with 10% FCS at 37°C/6.5% CO2. Mycoplasma testing was performed monthly using VenorGeM Classic (Minerva biolabs, Germany). All cell lines were authenticated by flow cytometry to express the relevant molecules, which were HLA-A2, tyrosinase, CD40, or CD40L, as required.

Three different CD40L:CD28 CSP constructs were designed. Two of them created a type I membrane protein structure in which the C-terminal part corresponding to the soluble CD40L fragment (AA 113-261) was inverted and linked to the transmembrane (TMD) plus cytoplasmic domain (ICD) of CD28 (AA 153-220). For successful intracellular trafficking and surface expression, the PD-1 signal peptide (20 AA) was used. ECD and ICD were connected through a Glycine/Serine linker (10 AA), which provides flexibility of the connected functional domains, and a spacer, either IgG1Fc or Fil3 (third Ig-like repetition corresponding to the Filamin protein), for expression improvement. These CSPs were designated as CD40L:IgGFc:CD28 and CD40L:Fil3:CD28, respectively. The third construct adopted a type II transmembrane protein structure by linking the CD40L ECD and TMD (AA 14-261) with the inverted ICD (AA 180-220) of CD28. This construct was designated as CD40L:CD28i. The CD40L native sequence (provided by Kathrin Gärtner, Helmholtz Center Munich, Germany) and the chimeric sequences (ordered from GeneArt) were cloned in the pMP71 vector for retroviral transduction of primary T cells (CD3/CD28-activated PBMC), or tgTCR-T cells expressing TCR-T58, TCR-D115 or TCR53. Retroviral transduction was performed as described (19, 55). Briefly, human PBMC or tgTCR-T cells were thawed and activated with 5 μg/ml of plate-bound OKT3 (provided by E. Kremmer, Helmholtz Center Munich, Germany) and 1 μg/ml of anti-CD28 (BD Bioscience) for 2 days in TCM with 100 U/ml IL-2. Thereafter, T cells were split into 5 equal parts, each transduced with retrovirus particles encoding either the native CD40L sequence or one of the three chimeric CD40L:CD28 constructs, or no construct (tgTCR/Mock-T cells). After 4 days, CSP-transduced T cells were harvested and cultivated for another 12 days reducing the amount of IL-2 to 50 U/ml. CSP-transduced T cells were frozen on day 12 after transduction.

For re-activation of T cells to induce CD40L:CD28 CSP surface expression, T cells, which were frozen on day 12 after CSP transduction, were thawed and stimulated either with anti-CD3 plus anti-CD28 antibodies or through target cell-expressed peptide/MHC TCR ligands.

For anti-CD3/CD28 stimulation, CD40L:CD28 CSP-transduced T cells were thawed and seeded in 24-well non-tissue culture treated plates that had been coated with anti-CD3 (OKT3, 5 µg/ml) and anti-CD28 antibodies (1 µg/ml) (1 x 10⁶ T cells/ml TCM with 100 U/ml IL-2). After 3 days, T cells were removed from the stimulation plate, diluted 1:4 and further cultured without anti-CD3/CD28 antibodies in TCM with 100 U/ml IL-2 for 3 additional days. On day 6, when endogenous CD40L was downregulated (evidenced by absence on CD4 Mock T cells by flow cytometry) and CSPs were still expressed, T cells were harvested and prepared for flow cytometry or used in B cell and DC activation assay.

For re-activation through peptide/MHC ligands, TCR-T58 or TCR-D115 T cells co-expressing the CD40L:CD28 CSPs or native CD40L, or Mock, were thawed and co-cultured with target cells at a ratio of 1:10. Target cells were HEK293/Tyr cells that had very low endogenous CD40 expression or HEK293/Tyr/CD40 that were transduced to strongly express CD40; or the melanoma cell line SK-Mel23 that endogenously expresses CD40. SK-Mel23 was used either untreated or was pretreated with anti-CD40 antibody (clone HB14 pure functional grade, Miltenyi, at 1:11 concentration) before setting up the co-culture to block the endogenous CD40. After 24 h co-culture, cells were harvested for flow cytometry.

To analyze surface expression of CD40L or CD40L:CD28 CSP, anti-CD40L-PE (89-76, eBioscience) was used in combination with anti-mouse TCRβ-constant region (mTCR)-PB (H57-59, BioLegend) to detect the transgenic TCR expression, anti-CD4-APC-Cy7 (RPA-T4, eBioscience) and anti-CD8-V500 (RPA-T8, BD). 7-AAD (Sigma-Aldrich) was used for live/dead discrimination. HLA-A2 and CD40 expression on HEK293/Tyr cells, HEK393/Tyr/CD40 cells and tumor lines SK-Mel23, RCC-26 and RCC-53 were analyzed using anti-HLA-A2 (ATCC HB54) plus anti-mouse IgG1-A488 (Invitrogen) and anti-CD40-PE (clone H-10, Santa Cruz Biotechnology) together with 7-AAD. Flow cytometry was performed with the LSRII (BD) cytometer and FlowJo v10.7.1 software.

TCRtg-T cells transduced with CD40L:CD28 CSPs were thawed and activated for 6 days using anti-CD3 and anti-CD28 antibody coated plates to upregulate the surface expression of the CD40L:CD28 CSPs (see Re-Activation of T Cells). On day 6, when the endogenous CD40L was downregulated on T cells and CD40L:CD28 CSPs were still expressed, T cells were harvested and washed with PBS, then incubated for 4 h in RPMI-basic without IL-2 to reduce constitutive background p-AKT signals. After 4 h, T cells were harvested, washed with PBS/EDTA (2 mM), and stained with fixable Blue reagent (Thermo Fisher Scientific/Caltag, WalthamMassachusetts USA) for viability (10 min), then washed and co-cultured with HEK293/Tyr/CD40 (1:2 ratio) for 30 min at 37°C and 5% CO₂ to stimulate the T cells through the tgTCR-Tyr/HLA-A2 interaction and to trigger the CSPs through CD40. Unstimulated T cells were kept in culture as a negative control. After 30 min, co-cultures were immediately fixed using Cytofix Fixation Buffer (BD) (15 min, 37°C) followed by permeabilization with ice cold Phosflow Perm Buffer III (BDPhosflow™, 30 min). Antibodies to surface markers, anti-CD45-PE-Cy7 (HI30, BioLegend), anti-CD3-PerCP-Cy5.5 (SK7, eBiosciences), were added together with antibodies for the phosphorylated intracellular proteins p-AKT-PB (S473, M89-61, BD Biosciences), p-mTOR-A647 (S2448, O21-404, BD Biosciences) and p-RPS6-PE (pS235/pS236, (N5-676, eBiosciences). Data were acquired on the LSRII cytometer and analyzed using the FlowJo v10.7.1 software.

T cells transduced with CSPs and re-activated using anti-CD3 plus anti-CD28 antibody-coated 24-well plates (as described in Re-Activation of T Cells) were co-cultured at a 1:1 ratio with naïve human B cells per triplicate in 96-well U-bottom plate overnight at 37°C/5% CO₂. Naïve B cells without stimulation were used as negative control. Positive controls were B cells activated using soluble enhanced trimeric CD40L reagent (Enzo Life Science) (1:10 dilution) and B cells activated using HEK293/CD40L cells. After overnight incubation, cultures were harvested and analyzed by flow cytometry for B cell specific surface activation markers CD83-PE (HB15a, Immunotech), CD86-FITC (2331, BD Biosciences) and Fas-PE-Cy7 (DX2, BioLegend) together with CD19-A700 (HIB19, BD Biosciences) and 7-AAD.

iDCs (0.1 x 10⁶ cells) or ercDCs were harvested and co-cultured with T cells re-activated to express CD40L:CD28 CSPs (see Re-Activation of T Cells) at 1:1 ratios in 96-well U-bottom plates in a final volume of 200 µl TCM. Co-cultures were set in triplicates. iDCs alone, mDC, and iDCs with mock T cells were used as controls. After 24 h, supernatants were harvested for 45Plex bead array and cells were used for flow cytometry using antibodies to CD83-PE (HB15a, Immunotech), CCR7-PB (G043H7, BioLegend), PD-L1-PerCP-Cy5.5 (29E.2A3, BioLegend), HLA-DR-APC-Cy7 (L243, BD Biosciences) and CD80-PE-Cy7 (2D10, BioLegend) together with CD3-A700 (UCHT1, BioLegend) and live/dead fixable blue stain (Thermo Fisher Scientific/Caltag).

TCRtg-T cells (TCR-T58, TCR-D115, TCR53) with or without co-expression of the different CSPs were thawed and co-cultured without re-activation with SK-Mel23, RCC-26 or RCC-53 at 1:2 T cell to target cell ratio in a 96 wells. T cells and target cells cultured alone were used as controls. All target cells expressed CD40 endogenously. Supernatants were collected after 24 h and analyzed for IFN-γ secretion by ELISA.

T cell-mediated target cell killing was performed using a ⁵¹chromium release assay as described (56). Briefly, 1 x 10⁶ target cells were labeled with 50 μCi ⁵¹chromium (Hartmann Analytic, Braunschweig Germany) for 1 h at 37°C. TCRtg-T cells (TCR-T58, TCR-D115, TCR53) with or without the co-expression of the different CSPs were thawed and plated without re-activation at titrated cell numbers reaching T cell to target cell ratios from 10:1 to 1.25:1 in a 96 well plate. Chromium-labelled targets cells were added at a concentration of 2000 cells per well. For spontaneous release, target cells were cultured without T cells. Each culture was set up in duplicates. Cells were co-cultured for 4 h at 37°C. For maximum ⁵¹chromium release, 50 µl of the target cell suspension was pipetted directly to the Luma plate (Canberra Packard, Germany). After the co-culture time, 50 µl aliquot from each co-culture well was transferred to the Luma filter plate, dried and counted using a TopCount machine. Specific cell lysis was calculated by applying the formula:

IFN-γ analysis of T cell/tumor cell co-cultures was done using ELISA kits (BD Bioscience) according to the manufacturer´s instructions. Supernatants of T cell/DC co-cultures were analyzed using a multiplex bead array 45Plex (LKTM014, human XL Cytokine, RnD Systems) according to the manufacturer’s instructions. Data were acquired using the Luminex100 machine with BioPlex Manager 6.1 software (Bio-Rad Laboratories GmbH). Standard curves were fitted using the logistic-5PL regression type.

According to the topological classification of transmembrane proteins, the CD28 protein and the CD40L protein belong to the type I and type II groups, respectively (57, 58). A type I membrane protein is present on the cell surface with its N-terminus oriented towards the extracellular space and the C-terminus located on the cytoplasmic side. A type II membrane protein is anchored with a signal-anchor sequence and, once transported to the cell membrane, is located at the cell surface with its C-terminus oriented into the extracellular space and the N-terminus on the cytoplasmic side. With CD28 and CD40L having different membrane orientation, creating a chimeric CD40L:CD28 protein represented a structural and functional challenge ( Figure 2A ). Three different constructs were designed ( Figures 2B, C ). CD40L physiologically exists also as a soluble protein (AA 113-261) (59). Thus, it was considered that this CD40L ECD sequence should maintain functionality as inverted nucleotide sequence allowing to generate a type I protein structure with the transmembrane (TMD) and intracellular domain (ICD) sequences of CD28. The CD40L ECD and CD28 TM_ICD domains were connected through a Glycine/Serine linker for flexibility and mobility of the connected functional domains (60) and a specific spacer for expression improvement. The CSP variant CD40L:IgGFc:CD28 used the IgG1Fc domain as a spacer to provide the protein with a better membrane stability due to its dimerization property, which was observed during its previous use in the design of CARs for antigen-specific T cell engineering (61). In the second variant, CD40L:Fil3:CD28, the IgG1Fc spacer was exchanged for the third Ig-like repetition corresponding to the Filamin protein. This is a protein sequence that adopts an immunoglobulin-like fold without dimerization properties (62, 63). The rational for this variation was the concern of linking the primarily trimeric CD40L molecule with a dimerizing sequence like the IgG1Fc spacer. The potential oligomerization might result in the formation of protein clusters in the membrane that might interfere with the proper protein function. In the third construct, CD40L:CD28i, the type II transmembrane protein structure of the native CD40L was maintained linking the CD40L ECD plus TMD sequence to the inverted ICD sequence (AA 180-220) of CD28 (CD28i). This design followed that of a published NKG2D/CD3zeta chimeric protein (64). Sequences of the CSPs were cloned into pMP71 retroviral vector and used to stably transduce human T cells, which stably expressed either the HLA-A2 restricted tyrosinase-specific TCR, T58 or D115 (19, 53), or TCR53 (19, 54) specific for RCC.

To assess if the chimeric sequences are transcribed and translated correctly and are able to generate correctly folded proteins, which traffic to the cell surface, CSP-transduced T cells were stained with anti-CD40L antibody at different time points after transduction (3 days, 6 days, 10 days and 13 days) to detect the CSP on the cell surface. T cells transduced with the native sequence of CD40L (nCD40L) were used as control. It was observed that the expression profile varied depending on the CSP design ( Figure 3 and Supplementary Figure 1 ): at 3 days after transduction, the CD40L:IgGFc:CD28 and CD40L:CD28i showed highest expression with 67.7% or 73% CSP positive T cells, while the CD40L:Fil3:CD28 reached only 26% positive T cells. The expression decreased over time, for CD40L:IgGFc:CD28 from 76.7% to 52.6% at day 13, for CD40L:CD28i from 73.% to 14% and to undetectable levels for CD40L:Fil3:CD28. The loss of surface expression was also observed for the nCD40L indicating that it was not a feature of the chimeric proteins. Retroviral vectors should integrate into the genome and result in stable transgene expression. Thus, the loss of CD40L:CD28 CSP expression was surprising and unique to CD40L proteins, since it was not observed for other transgenes (TCRs, PD-1:CD28 CSP) expressed using the same retroviral vector pMP71 ( Supplementary Figure 2 ). Cell culture conditions, as schematically depicted in Figure 4A , were found to relate to cell surface expression of the CD40L proteins ( Figures 4B, C ). Exemplary shown is cell splitting (density reduction) with addition of fresh medium containing IL-2, which was done on day 6 and day 12. Reproducibly, the splitting event turned nearly undetectable CSP expression (day 6, or day 12) on CD8 T cells to fair surface expression on day 10 or day 13 ( Figures 4B, C : CSP positive cells are depicted in blue and are superimposed on CD40L negative cells depicted in red). A similar oscillation of expression was observed for the transgenically expressed native CD40L protein, suggesting that the CD40L extracellular domain, which is shared between CSP and native CD40L, might be involved in the expression kinetic. Indeed, it is well known that the endogenous expression of CD40L is tightly linked to T cell activation showing fast upregulation after stimulation and decay within 16-24 hours after stimulation. The natural kinetic is presumed to minimize bystander activation of CD40 positive cells (65). The kinetic of the endogenous CD40L expression in comparison to the kinetic of the CSPs is best observed in the dot plots of the mock transduced T cells ( Figure 4B , Mock) within the CD4 T cell subset, which are known to express CD40L upon activation (66). In Figure 4C , a line graph depicts the surface expression kinetic of the CD40L:CD28 CSPs and the CD40L native protein upon changes in culture conditions on day 6 and day 12.

Considering that the native CD40L expression is linked to T cell activation, it was evaluated if the CD40L:CD28 CSP expression might follow a similar behavior. To this end, CD40L:CD28 CSP-transduced T cells, which had been frozen on day 12 when CSP expression was lowest, were thawed and objected to an in vitro activation ( Figure 5A ) using anti-CD3 and anti-CD28 antibodies. On day 3, the activation signal was removed and activated T cells were cultured with fresh medium plus IL-2 for another 3 days. CD40L staining was performed to detect surface expression of the CD40L:CD28 CSPs directly after thawing (day 0), on day 3, and day 6. Depicted by histogram display ( Figure 5B ), it is showed that on day 0 CD40L:CD28 CSP expression is not detectable. On day 3 after activation, T cells had upregulated the expression of the CD40L:CD28 CSPs. Once the T cells were transferred to a new plate without antibody stimulation, the expression started to decline again, with variable percentages of expression remaining on day 6 depending on each construct ( Figures 5B, C ). The endogenous CD40L expression kinetic could be followed on the CD4 T cell population by dot plot display ( Figure 5D ) showing the loss of CD40L as evidenced by the absence of CD40L staining on CD4 mock T cells on day 6. Therefore, CD4 mock-transduced T cells were used as a negative control in further assays that assessed the CD40L:CD28 CSP functionality.

After observing that the CD40L:CD28 CSPs displayed similar expression kinetic as the endogenous CD40L protein, and that upregulation is possible with CD3/CD28 activation, it was tested if the physiologic downregulation through CD40 receptor interaction that is well documented for the endogenous CD40L expression (30, 67), also applies to the CD40L:CD28 CSPs. The experimental set-up is depicted in Figures 6A, B . CD40L:CD28-transduced TCR-T58 T cells were co-cultured with the melanoma tumor cell line (SK-Mel23) ( Figure 6A ), which provides the activation stimulus through the TCR-peptide/MHC interaction (HLA-A2/tyrosinase, which is the ligand for TCR-T58) and expresses CD40 endogenously. The CD40L:CD28-transduced T cells were used freshly thawed when their CD40L:CD28 CSP expression level was absent, except for some residual expression of the CD40L:IgGFc:CD28 CSP. CD40L staining to detect CD40L:CD28 CSP surface expression was performed at the start of the co-culture, after 16 h, 24 h and 48 h. As depicted in the line graph of Figure 6A , induced surface expression of CD40L:CD28 CSPs was noticeable for the CD40L:IgGFc:CD28 with a steady incline until 48 h. A delayed and weak surface expression occurred for the CD40L native protein, the CD40L:Fil3:CD28 and CD40L:CD28i CSPs. Endogenous CD40L surface expression on mock T cells was not detected. All CD40L:CD28 CSPs showed strong induction of surface expression when the melanoma cell line was pretreated with CD40L antibody before it was used in the co-culture. These results suggested that CD40L:CD28 CSPs are upregulated by physiologic TCR activation through peptide/MHC interaction and subsequently downregulated by interaction with tumor cell-expressed CD40 receptor. Upregulation was achieved after 16 hours of co-culture, reaching a sustained increase expression until the 48 h time point in the presence of CD40 blockade.

In a second co-culture experiment ( Figure 6B ), HEK293 cells were used that endogenously express HLA-A2 and were transduced to stably express and present the tyrosinase antigen (HEK/Tyr cells). HEK/Tyr cells had very low endogenous CD40 expression and were, thus, additionally transduced to express CD40 (HEK/Tyr/CD40) or left transduced (HEK/Tyr/mock). In co-culture with HEK/Tyr/mock cells, which express only marginal CD40 receptor, strong induction of CD40L:IgGFc:CD28 as well as CD40L native was observed and lower induction of CD40L:Fil3:CD28 and CD40L:CD28i, which was comparable to the CD40L induction on T cells without CD40L:CD28 CSP transduction (mock). In contrast, using the HEK/Tyr/CD40 cells with strong CD40 expression, upregulation was only observed for the CD40L:IgGFc:CD28 CSP. Thus, surface expression of CD40L:CD28 CSPs induced by TCR-peptide/MHC interaction is counteracted by endogenous (SK-Mel23) or transgenic expression (HEK cells) of the CD40 receptor on target cells.

Having determined that the composite proteins can be expressed on the cell surface, assessing if the different domains were functionally incorporated into the chimeric CD40L:CD28 CSPs was the next step. The biologic activity of the CD40L ECD was tested in two systems using a B cell stimulation assay and DC activation. It is well described that B cells express the CD40 receptor. Upon interaction with the CD40 ligand (CD40L), B cells undergo activation, which can be detected by measuring key surface activation markers on the B cells like CD86, Fas and CD83 ( Figure 7A ). Upregulation of the three markers, CD83, CD86 and Fas, are linked to the CD40/CD40L interaction. CD86 and Fas expression might also be influenced by TCR/MHC interaction that occurs between T cells and B cells when the latter serve as APCs. CD83 has been reported to be induced on B cells via CD40 engagement independent of TCR/MHC binding (68). In our system, the CD40L:CD28 CSP-transduced T cells co-expressed the tyrosinase-specific TCR-T58. B cells were isolated from blood of healthy donors. No tyrosinase was present in the system excluding B cell activation through antigen.

T cells without and with CD40L:CD28 CSPs were thawed and re-activated for 6 days using CD3/CD28 antibody-coated 24 well plate ( Figure 7B ) to re-induce CD40L:CD28 CSP expression. T cells were harvested on day 6 when CD40L:CD28 CSP levels were detectable, but the mock T cells were CD40L negative (red histograms), indicating absence of endogenous CD40L that would conceal effects of the CD40L:CD28 CSPs ( Figure 7B ).

To assess B cell activation, T cells were co-cultured with freshly isolated B cells overnight (12 h) at a 1:1 ratio followed by analysis of surface activation markers on the B cell population by flow cytometry ( Figure 7C ). It was observed that CD83, CD86 as well as Fas were induced on B cells that were co-cultured with T cells expressing the CD40L:CD28 CSPs or the native CD40L compared to the co-culture with mock T cells that did not express any CD40L construct or control B cells without T cell co-culture (control). Gradual levels of activation marker expression on B cells were observed depending on the variant of the CD40L:CD28 CSP expressed by the T cells, with highest expression of activation markers achieved with T cells expressing CD40L:IgGFc:CD28, followed by native CD40L, CD40L:CD28i and lowest CD40L:Fil3:CD28. As reference, two different positive controls were included to stimulate the B cell population: one control setting used a commercial soluble CD40L agonist combined with an enhancer to promote CD40L trimerization and, hence, B cell activation. The other control was a co-culture with HEK293/CD40L cells that stably expressed high levels of native CD40L. The sCD40L/enhancer agonist achieved induction of activation markers on B cells similar to the T cell co-culture with CD40L:CD28 CSP-engineered T cells. Highest levels of B cell activation were achieved with the HEK293/CD40L cells.

The observed effects on B cells provide proof of concept that the CD40L ECD within the CSPs is exerting biologic activity with efficiencies matching the surface expression level of each CD40L:CD28 variant.

In the DC maturation assay, immature DCs (iDCs) were generated in vitro following a 7 day protocol (51), and the upregulation of CD83, CCR7, PD-L1, HLA-DR and CD80 was assessed after T cell co-culture ( Figure 8A ). To assess effects on DC maturation, in vitro generated iDCs were co-cultured with TCR-T58 T cells without or with CD40L:CD28 CSP expression ( Figure 8A ) overnight (12 h) in a 1:1 ratio. iDCs without T cells served as a negative control and DCs maturated with Jonuleit cytokine cocktail were used as a positive control. DCs were from HLA-A2 negative donors, thus providing no tgTCR-specific stimulation. In Figure 8B , the expression of DC maturation markers after each culture condition is depicted as the mean fluorescence intensity. For CD83, CCR7 and PD-L1, the expression levels were highest after co-culture with T cells expressing the CD40L:IgGFc:CD28, closely similar to levels observed on mDC. T cells expressing CD40L:Fil3:CD28 or CD40L:CD28i still induced levels above those achieved with Mock T cells, or T cells with native CD40L, or iDCs that were not stimulated at all. For HLA-DR and CD80 markers, no significant difference was seen between the different culture conditions.

In addition to effects on surface markers, CD40L:CD28-expressing T cells also caused changes in secreted cytokines and chemokines in co-cultures with DCs. These included augmented secretion of IL-12p70, IL-10, pro-inflammatory IL-1β, TNF, chemokines MIP-1α (CCL3), MIP-3α (CCL20) and MIP-3β (CCL19) as well as the cytotoxic protein granzyme B (gzmB) ( Figure 8C ). PD-L1 was also induced consistent with the observed upregulation on the cell surface. No changes were seen for Th2 cytokines (IL-4, IL-5, IL-13), IL-17, IL-33, IL-15, eotaxin and fractalkine (CX3CR1), TRAIL, PDGF or EGF (not shown).

In the tumor milieu, myeloid cells adopt TME-associated phenotypes that might be associated with tumor promotion and immune cell inhibition. In RCC, we previously reported a uniquely polarized myeloid subset, which we called ercDC (enriched in renal cell carcinoma DC) (52, 69). ercDC were found to exhibit protumorigenic and immune inhibitory features and a high prevalence in tumor tissue correlated with poor survival. Here we used in vitro generated surrogate ercDCs (52) for co-culture and observed that CD40L:CD28-expressing T cells induced a similar pro-inflammatory and chemotactic secretome as observed for in vitro generated iDCs.

In a next step it was determined whether the CD28 ICD was capable to deliver a proper co-stimulation signal to the T cells. The CD28 co-stimulatory pathway starts with phosphatidylinositol 3′-kinase (PI3K) leading to phosphorylation of the protein Kinase B, also known as AKT, followed by downstream phosphorylation of the mammalian target of rapamycin (mTOR) molecule. Ribosomal protein S6 (RPS6) is then a downstream target of mTOR activation (70, 71).

The functionality of the CD28 ICD within our CD40L:CD28 CSPs was tested by analyzing the phosphorylation of AKT, mTOR and RPS6 in CD40L:CD28 CSP-transduced tyrosinase-specific TCR-T58 T cells after stimulation with HEK293/Tyr/CD40 cells, which can trigger the T58-TCR through HLA-A2/Tyr ligands and the CD40L:CD28 CSPs through CD40 expression. Before setting up the stimulation co-culture, T cells were deprived of IL-2 for 4 hours to reduce the level of constitutive AKT activation. Subsequently, the starved T cells were co-cultured with the HEH293/Tyr/CD40 cells for 30 minutes, which was previously determined as the optimal time for AKT activation to reach its peak (72, 73).

After 30 minutes of stimulation, phosphorylation of AKT protein was increased in CSP-transduced T cells in comparison with the Mock T cells without CSP or T cells carrying the native CD40L that does not contain the CD28 signaling domain, or the unstimulated T cells ( Figure 9A , left graph). Similar results were detected for the phosphorylation of the mTOR and RPS6 proteins ( Figure 9A , middle and right graphs), suggesting that the CD40L:CD28 CSPs activated the AKT pathway.

Induction of phosphorylation was strongest in T cells that expressed the CD40L:Fil3:CD28 or the CD40L:CD28i compared to T cells with the CD40L:IgGFc:CD28. The differences between the three CD40L:CD28 CSPs can be appreciated in the histogram display, showing increased MFI of phosphorylated AKT and mTOR relative to the unstimulated T cells when the T cells expressed the CSPs (colored histogram), while the MFI in Mock and native CD40L-expressing T cells was similar to unstimulated control ( Figure 9B ). The pattern of the phosphorylated RPS6 (p-RPS6) differed from those of AKT and mTOR showing lower MFI in stimulated T cells that expressed no CSP or the native CD40L protein compared with the unstimulated control, while phosphorylation was higher in stimulated T cells expressing the CD40L:Fil3:CD28 or the CD40L:CD28i CSP, reaching levels of the unstimulated T cells.

Finally, the correlation of phosphorylation of the CD28 downstream signaling proteins was depicted by plotting the p-AKT against p-mTOR and p-mTOR against p-RPS6 ( Figure 9C ). The percentages of double-positive populations, corresponding to T cells that had phosphorylated both AKT and mTOR or mTOR plus RPS6, was notably higher in T cells that were transduced with the CD40L:CD28 CSPs compared to Mock T cells without CSP or T cells transduced with the native CD40L that does not contain a CD28 signaling domain.

After the biologic activity of both domains of the CD40L:CD28 CSPs was demonstrated, it was tested if the CD28 signaling provided by the CSPs was able to improve T cell effector function, namely cytokine secretion and cytotoxicity.

Tyrosinase antigen-specific TCR-T58 T cells transduced to express the different CD40L:CD28 CSPs secreted more IFN-γ when stimulated with SK-Mel23 cells (tyrosinase/HLA-A2 and CD40 positive) compared to TCR-T58/mock T cells without CD40L:CD28 CSP ( Figure 10A ). Similar results were observed when CD40L:CD28 CSP-transduced RCC-specific TCR53 T cells were co-cultured with CD40 positive RCC-26 and RCC-53 cell lines ( Figures 10B, C ).

Among the different CSP formats, the CD40L:IgGFc:CD28 (orange) had the lowest effect on cytokine secretion, while the CD40L:CD28i (purple) and the CD40L:Fil3:CD28 enhanced the cytokine secretion more strongly. This is of note, since it contrasts with the expression level of the CSPs on the T cells where the CD40L:IgGFc:CD28 consistently had highest levels and CD40L:Fil3:CD28 was lowest (see for example Figure 3 ).

Target cell killing using tyrosinase-specific TCR-D115 T cells and RCC-specific TCR53 T cells showed that T cells expressing the CD40L:CD28 CSPs executed higher target lysis compared to the T cells without CSPs ( Figure 11 ). As noted before for the cytokine secretion, CD40L:CD28i (purple) and CD40L:Fil3:CD28 (green) CSPs supported T cells more strongly than the CD40:IgGFc : CD28 (orange line).

Of note, anticipating future clinical application, T cells in the functional assays were used freshly thawed without in vitro re-activation, thus the starting expression of the CSPs was low. The results of both the cytokine secretion and the target cell killing suggest that positive effects of the CD40L:CD28 CSPs on two central effector functions can be expected despite low CSP expression.