Extracellular Domains of CD8α and CD8ß Subunits Are Sufficient for HLA Class I Restricted Helper Functions of TCR-Engineered CD4+ T Cells

By gene transfer of HLA-class I restricted T-cell receptors (TCRs) (HLA-I-TCR) into CD8+ as well as CD4+ T-cells, both effector T-cells as well as helper T-cells can be generated. Since most HLA-I-TCRs function best in the presence of the CD8 co-receptor, the CD8αß molecule has to be co-transferred into the CD4+ T-cells to engineer optimal helper T-cells. In this study, we set out to determine the minimal part of CD8αβ needed for optimal co-receptor function in HLA-I-TCR transduced CD4+ T-cells. For this purpose, we transduced human peripheral blood derived CD4+ T-cells with several HLA-class I restricted TCRs either with or without co-transfer of different CD8 subunits. We demonstrate that the co-transduced CD8αβ co-receptor in HLA-I-TCR transduced CD4+ T-cells behaves as an adhesion molecule, since for optimal antigen-specific HLA class I restricted CD4+ T-cell reactivity the extracellular domains of the CD8α and ß subunits are sufficient.


Introduction
Adoptive transfer of T-cells is a strategy used to target both solid tumors [1] and leukemia [2][3][4]. By introducing well-characterised TCRs via gene transfer large numbers of T-cells with defined antigen-specificity can be obtained without long in vitro culture periods. Transfer of HLA-I-TCRs into CD8 + T-cells demonstrated redirected antigen-specificity [5][6][7][8][9][10] and recently the in vivo efficacy of adoptively transferred TCR transduced (td) T-cells was demonstrated in melanoma and synovial cell sarcoma patients [11][12][13]. For optimal maintenance of functional CD8 + immune responses in vivo, however, antigen-specific CD4 + T cells may play an essential role [14,15]. By TCR engineering of CD8 + as well as CD4 + T-cells, both effector T-cells as well as helper T-cells with the same specificity can be generated. However, since most HLA-I-TCRs function best in the presence of the CD8 co-receptor, the CD8 molecule has to be co-transferred into the CD4 + T-cells to engineer optimal helper T-cells [16,17].
The CD8 molecule can be expressed as an aa or an aß dimer, but is on peripheral TCRaß T-cells mostly expressed as an aß dimer [18][19][20][21][22][23]. The a subunit of CD8 binds to the nonpolymorphic residues of the a3 domain of HLA class I, thereby enhancing the avidity of the TCR/MHC complex [24]. The cytoplasmatic tail of the a subunit directly associates with the protein tyrosine kinase Lck [25][26][27][28], promoting signal transduction after T-cell activation. The intracellullar domain of the ß subunit enhances the association of CD8a with lipid raft localized Lck [29,30] and the linker for activation of T-cells (LAT) [31,32]. Although the mechanism is not clear yet, it has been demonstrated that CD8aß heterodimers bind MHC class I molecules more avidly than CD8aa homodimers [32][33][34].
Previously, it was reported that for optimal proliferation, cytokine production and cytotoxicity of HLA-I-TCR td CD4 + T-cells co-expression of CD8ab was needed whereas co-expression of CD8aa only marginally increased functional activity [16,17]. Here, we studied whether the extracellular and/or intracellular part of CD8a and CD8ß were required for this increased functional activity.

Results and Discussion
Extracellular CD8a and ß are required and sufficient to elicit HLA class I restricted IFN-c production To verify that functional activity of high-affinity HLA-I-TCR transduced (td) CD4 + T-cells was improved by the transfer of CD8aa or CD8aß co-receptor, CMV-specific CD4 + T-cells were transduced with the high-affinity HA-2-TCR with either only CD8a or with both the CD8a and CD8ß subunits and purified based on CD8aa or CD8aß expression. T-cells were tested against LCLs pulsed with pp65 peptide stimulating the endogenous CMV-TCR, or with either HA-2 peptide or HA-2 + LCLs stimulating the introduced HA-2-TCR, and antigen-specific IFN-c production was measured ( Figure 1A). As can be observed, HA-2-TCR td CMV-specific CD4 + T-cells co-transferred with CD8aa, CD8aß or negative for CD8 were equally potent in recognizing pp65 peptide pulsed target cells. In addition, all three populations were able to recognize HA-2 peptide loaded target cells. However, only the CD8aß co-transferred T-cells were able to recognize endogenously processed and presented HA-2 ( Figure 1A). These results confirmed previous studies demonstrating that retroviral introduction of CD8ab increased the functional activity mediated via the introduced TCR of the TCR td CD4 + T-cells [16,17].
To more precisely determine the part of the CD8aß coreceptor responsible for increased functional activity of the HLA-I-TCR td CD4 + T-cells, we constructed intracellularly truncated CD8a (DCD8a), Lck mutated CD8a (CD8a Lck), and intracellularly truncated CD8ß (DCD8ß). The HA-2-TCR td CMV-specific CD4 + T-cells were transduced with the different CD8a and CD8ß constructs, purified based on CD8aa or CD8aß expression, and used as effector T-cells in the experiments described here above ( Figure 1B). Results were similar to the experiments with unmodified (wt)CD8a and wtCD8ß co-transferred HA-2-TCR td CMV-specific CD4 + T-cells ( Figure 1A). Only TCR td CD4 + T-cells co-transferred with both CD8a and CD8b produced significant amounts of IFN-c after stimulation with HA-2 + LCLs, irrespective of whether the co-transferred CD8a and ß subunits were intracellularly truncated or whether the Lck binding motif of the CD8a subunit was mutated ( Figure 1B). Mock td CMVspecific CD4 + T cells specifically produced IFN-c only after stimulation with pp65 peptide pulsed LCLs (data not shown). These results demonstrate that for optimal HLA class I restricted IFN-c production of the TCR td CD4 + T-cells, co-transfer of the extracellular domains of CD8a and ß is required but that the intracellular domains can be dismissed.
Polyfunctionality of CD4 + T cells is important for optimal helper function. Therefore, we studied the capacity of HA-2-TCR td CMV-specific CD4 + T-cells co-transferred with different CD8 constructs to produce not only IFN-c, but also TNF-a and IL-2 and upregulate CD40L ( Figure 1C). TCR td CD4 + T-cells cotransferred with either wtCD8aß or DCD8aß produced significantly more cytokines and demonstrated significantly more CD40L upregulation after stimulation with either HA-2 pulsed or HA-2 + LCLs (p,0.05 indicated with asterisks) than CD8 negative or CD8aa expressing TCR td CD4 + T-cells. No significant difference in cytokine production or CD40L upregulation after HA-2 specific stimulation was observed between TCR td CD4 + T-cells co-transferred either with wtCD8aß or DCD8aß. The results in Figure 1C demonstrate that for IFN-c, TNF-a and IL-2 production as well as for CD40L upregulation after stimulation with HA-2 + LCLs co-transfer of CD8a and ß, and most importantly the extracellular domains of these CD8 subunits, is required.
In conclusion, to generate polyfunctional HA-2-TCR td CMVspecific CD4 + T-cells, co-transfer of both CD8a and ß is required, but the intracellular domains of these CD8 subunits can be dismissed.
DCD8a and DCD8ß improve HLA-class I restricted avidity similarly efficient as wtCD8a and wtCD8ß To analyze whether TCR td CD4 + T-cells co-transferred with the different CD8 subunits bind with similar affinity the HLApeptide complex, HA-2-tetramer staining was analyzed for both mock and HA-2-TCR td CD4 + T-cells ( Figure 1D and E). No specific HA-2 tetramer staining could be observed for mock and TCR td CD4 + T-cells without CD8aß co-transfer (data not shown). However, co-transfer of CD8a alone or transfer of CD8aß using two separate retroviral vectors resulted in aspecific staining of every tetramer added (data not shown). Therefore, mock and HA-2-TCR td CMV-specific CD4 + T-cells were co-transferred with multicistronic vectors in which the CD8a and CD8ß molecules were linked with a 2A sequence resulting in equimolar levels of both CD8a and ß molecules ( Figure 1D and E), and analyzed for tetramer staining. As can be seen in Figure 1D, no aspecific staining of HA-2 tetramer is detected on mock transduced CD4 + T-cells co-transferred with the multicistronic CD8aß vectors, whereas TCR td CD4 + T-cells co-transferred with the wtCD8aß or DCD8aß multicistronic vectors demonstrated identical HA-2 tetramer staining, indicating similar avidity for HLA-A2/HA-2-peptide complex ( Figure 1E).
Next, we studied whether co-transfer of wtCD8aß or DCD8aß equally improved the function of CD4 + T-cells transduced with a next generation HA-2-TCR CC , that was codon optimized and cysteine modified to improve TCR cell surface expression. For this purpose, CMV-specific CD4 + T-cells were transduced with the next generation HA-2-TCR CC either without or in combination with different CD8 subunits, purified using flow cytometry based cell sorting and tested for HA-2-specific IFN-c production against target cells loaded with titrated concentrations of the HA-2 peptide as well as against target cells that endogenously process and present the antigen (HA2 + target cells). As demonstrated in Figure 2A, the HA-2-TCR CC td CD4 + T-cells expressing either wtCD8aß or DCD8aß were equally reactive against HA-2 peptide loaded target cells, and approximately a 100 fold more sensitive compared to CD8 negative or CD8aa expressing HA-2-TCR CC td CD4 + T-cells. In addition, whereas CD8 negative or CD8aa expressing HA-2-TCR CC td CD4 + T-cells demonstrated no or low reactivity against HA-2 + target cells, the DCD8aß and wtCD8aß expressing HA-2-TCR CC td CD4 + T-cells were highly reactive against HA2 + target cells. To confirm the significantly increased sensitivity of the wtCD8aß and DCD8aß expressing HA-2-TCR CC td CD4 + T-cells cells compared to the CD8 negative or CD8aa expressing cells, we performed a proliferation assay in which we stimulated PKH-labeled T-cells with HA-2 + and HA-2 2 target cells and analyzed proliferation by measuring PKH dilution at day 5 after stimulation. As can be observed in Figure 2B, wtCD8aß and DCD8aß HA-2-TCR CC td CD4 + T-cells proliferated equally efficient after stimulation with HA-2 + target cells, whereas for the CD8 negative or CD8aa expressing T-cells no antigen-specific proliferation was observed. These data indicate that co-transfer of the extracellular parts of CD8 increase avidity of HA-2-TCR expressing CD4 + T-cells for HLA-A2 + HA-2 + target cells approximately a 100-fold. This increase in avidity is necessary to elicit efficient IFN-c production and proliferation of HA-2-TCR and HA-2-TCR CC td CD4 + T-cells after stimulation with physiologically relevant levels of HA-2.
In general, HLA-I-TCR td CD4 + T-cells require co-transfer of only the extracellular CD8aß domains To confirm the generality of these data, polyclonal peripheral CD4 + T-cells were transduced with next generation high-affinity TCRs specific for HA-1, HA-2 or PRAME and were cotransferred with the different CD8 constructs. Results presented in Figure 3 demonstrate that CD4 + T-cells transduced with either HA-1-TCR CC , HA-2-TCR CC or PRAME-TCR CC displayed IFN-c-, IL-2-and TNF-a production only after stimulation with peptide pulsed target cells (1 mg/ml), and not after stimulation with antigen-positive target cells expressing endogenously processed antigen. Introduction of only the wtCD8a molecule induced some cytokine production against peptide pulsed target cells and antigen-positive target cells. However, when wtCD8aß or DCD8aß were introduced, substantial percentages of transduced T-cells produced IFN-c, IL-2 and TNF-a both after stimulation with peptide pulsed target cells or antigen-positive target cells. These results demonstrate a general trend in requirement and sufficiency of co-transfer of the extracellular domains of CD8a and ß for HLA-class I restricted helper functions.  To study the minimal part of CD8 needed for optimal co-receptor function in HLA-I-TCR td CD4 + T-cells, HA-2-TCR td CMV-specific CD4 + Tcells (A) co-transferred with wtCD8aa or wtCD8aß co-receptor, or (B) co-transferred with either wtCD8a,DCD8a or CD8a Lck in combination with either wtCD8ß or DCD8ß were purified and used in a stimulation assay. Td T-cell populations were tested against HLA-DR1 + LCL-CBH either unpulsed (grey striped bars) or pulsed with pp65 peptide (grey bars), or against HLA-A2 + HA-2 2 LCL-IZA either unpulsed (white bars) or pulsed with HA-2 peptide (black bars), or against HLA-A2 + HA-2 + LCL-JYW (black striped bars). IFN-c production was measured after 18 h of stimulation in duplicate, and a representative experiment out of 3 is depicted. The IFN-c production of the different CD8aß expressing TCR td T-cells was compared to the IFNc production of CD8aa expressing TCR td T-cells within their group using students' t-test. P-values ,0.05 are indicated with an asterisk. (C) To study whether co-transfer of CD8 would also result in polyfunctional helper functions of TCR td CMV-specific CD4 + T-cells, both mock and HA-2-TCR td CMV-specific CD4 + T-cells with or without co-transfer of different CD8 subunits as indicated in the figure were stimulated with HLA-DR1 + LCL-CBH pulsed with pp65 peptide (grey bars; pp65 pep), unpulsed HLA-A2 + HA-2 2 LCL-IZA (white bars; control), HA-2 peptide pulsed HLA-A2 + HA-2 2 LCL-IZA (black bars; HA-2 pep) or HLA-A2 + HA-2 + LCL-JYW (striped bars; HA-2 endogenous). After 5 h of stimulation, T-cells were stained with anti-IFN-c, anti-TNF-a, anti-CD40L and anti-IL-2 mAbs and were analyzed using flow cytometry. The percentage of IFN-c, TNF-a and IL-2 producing or CD40L expressing T-cells after stimulation is depicted. The percentages of cytokine producing and CD40L upregulating CD8aß expressing TCR td T-cells that were significantly higher than CD8 negative and CD8aa expressing TCR td T-cells

Conclusions
We hypothesize that CD8 co-transfer in antigen-experienced CD4 + T-cells potentially poses the risk of overstimulation. To minimize the risk of overstimulation in HLA class I restricted TCR transduced CD4 + T-cells, we analyzed in this study whether cotransfer of a signaling deficient CD8-co-receptor would also result in optimal HLA class I restricted functionality.
We confirmed that for optimal helper functions of HLA-I-TCR td CD4 + T-cells co-expression of CD8aß co-receptors is superior to CD8aa co-receptors, and CD8aß co-expressing T-cells were superior in producing IFN-c, TNF-a, and IL-2, in upregulating CD40L, and in antigen specific proliferation. Expression of the extracellular domains of CD8aß was required both for CD4 + Tcells transduced with unmodified HA-2-TCR, as well as CD4 + Tcells transduced with codon optimized and cysteine modified TCRs. These modifications should result in higher cell surface expression due to improved translation and improved preferential pairing of the TCR-chains, but nevertheless CD8aß co-expression was required for robust HLA-class I restricted helper functions.
Introduction of the CD8aß co-receptor increased the sensitivity of the HLA-I-TCR td CD4 + T cells approximately a 100 fold, leading to efficient recognition of target cells that express the antigen endogenously. Truncation of the intracellular domains of the CD8a and CD8ß subunits did not change the functional properties of the HLA-I-TCR td CD4 + T-cells. Using CD4 + Tcells transduced with several different HLA-I-TCRs we confirmed the generality of the data.
Although it was demonstrated that both for CD8a [35] and CD8ß [36] the intra-as well as the extracellular domain play a role in positive selection of thymocytes, we demonstrate that for the effector function of peripheral T-cells the CD8aß co-receptor functions as an adhesive molecule rather than a signalling molecule. Therefore, to elicit robust helper functions in CD4 + T-cells transduced with high-affinity HLA class I restricted TCRs introduction of the extracellular domains of CD8a and ß subunits Figure 3. In general, co-transfer of the extracellular domains of CD8a and ß is required and sufficient. To confirm the generality of the previous data, total CD4 + T-cells were transduced with codon optimized and cysteine modified HA-1-, HA-2-or PRAME-TCR (transduction efficiency 48%, 48% and 22%, respectively) either with or without co-transfer of different CD8 molecules, as indicated in the figure. One week after transduction, non-purified TCR td CD4 + T-cells were stimulated and tested for cytokine production using flow cytometry. HA-1-or HA-2-TCR td CD4 + T-cells were stimulated either with HA-1 or HA-2 peptide pulsed or unpulsed HLA-A2 + HA-1 -HA-2 2 LCL-IZA, or HLA-A2 + HA-1 + HA-2 + LCL-MRJ, and PRAME-TCR td CD4 + T-cells were stimulated either with PRAME peptide pulsed or unpulsed HLA-A2 + PRAME 2 melanoma cells, or HLA-A2 + PRAME + melanoma cells. is required and sufficient. However, although we demonstrate equal proliferative capacity of HLA-I-TCR td CD4+ T-cells coexpressing either dCD8aß or wtCD8aß, it needs to be studied in vivo whether they also demonstrate equal long-term persistence in vivo.

Construction of retroviral vectors and production of retroviral supernatant
TCRa and TCRß chains of the HA-2-TCR [8], as well as of the codon optimized [37] and cysteine modified [38,39] next generation HA-2-TCR CC , HA-1-TCR CC [40], and PRAME-TCR CC [41] were linked using a self-cleaving T2A sequence [42] and combined with the truncated nerve growth factor (NGF-R) into a retroviral vector. All TCRs used were CD8 dependent and HLA-A*0201restricted. Both single vectors encoding unmodified CD8a (wtCD8a), truncated CD8a (DCD8a), Lck mutated CD8a (CD8a Lck), unmodified CD8ß (wtCD8ß) and truncated CD8ß (DCD8ß) as well as T2A linked wtCD8aß and DCD8aß constructs were engineered. DCD8a consists of amino acids (aa) 1-190, Lck mutated CD8a has mutations at position 201 and 203 (CRA), and DCD8ß consists of aa 1-176. CD8a constructs were combined with the eGFP markergene, whereas the CD8b constructs were combined with the NGF-R markergene. T2A linked wtCD8aß and DCD8aß constructs were engineered without a markergene. Retroviral vectors encoding eGFP or NGF-R alone were used as control vectors (mock). Using the Moloney murine leukemia virusbased retroviral vector LZRS and packaging cells w-NX-A, viral supernatant was generated as previously described in detail [8]. PE-labeled and APC-labeled HA-2 peptide-HLA tetramers were produced as described previously [43]. For data acquisition, a FACSCalibur was used and for data analysis FlowJo software was used.

Donor CD4 + T-cells, stimulation and transduction
After study approval of the Leiden University Medical Center institutional review board and written informed consent according to the Declaration of Helsinki, peripheral blood mononuclear cells (PBMC) were obtained from hereditary hemochromatosis patients. From these PBMC samples, CMV-pp65-specific CD4 + T-cells were isolated as previously described [44]. Briefly , PBMC were stimulated with 2 mg/ml pp65-KYQEFFWDANDIYRI (pp65 peptide; HLA-DRB1*0101-binding) peptide, and after 4 hours (h) of stimulation, IFN-c secreting CD4 + T cells were isolated using the IFN-c secretion assay (Miltenyi Biotec, Bergisch Gladbach, Germany). Subsequently, CMV-specific CD4 + T-cells were further purified by cell sorting based on TCR-BV2-staining. CMV-specific CD4 + T-cell line was stimulated with irradiated (30 Gy) allogeneic PBMCs (1610 6 cells/ml), and 800 ng/ml PHA (Murex Biotec Limited, Dartford, UK), and transduced 2 days after stimulation as described previously [45] with HA-2-TCR or empty vectors combined with NGF-R. After cell sorting based on NGF-R markergene expression, NGF-R markergene positive cells were re-stimulated, transduced with the different CD8 constructs, and sorted on basis of NGF-R expression, CD4 and CD8 cell surface expression. In addition, .94% MACS-enriched (Miltenyi Biotec) CD4 + T-cells derived from total PBMCs were transduced with either next generation HA-1-, HA-2-, or PRAME-TCR or empty vectors combined with NGF-R in combination with the different CD8 co-receptor subunits, and these T-cells were used in experiments without further purification.

Analysis of cytokine production and proliferation
To test the capacity of T-cells to specifically proliferate in response to antigen, a PKH (St. Louis, Missouri, USA) based assay was used [46]. T-cells were labeled with PKH-26 (St. Louis, Missouri, USA) according to manufacturer's instructions, and 1610 4 T-cells were stimulated with 3610 4 target cells. PKH dilution was analyzed using flow cytometry at day 5 after stimulation.

Statistics
Experimental data was evaluated in a paired fashion by use of the students' T-test. Reported P values are 2-sided and were considered statistically different if ,0.05.