MyD88 co-stimulation in donor CD8+ T cells enhances the graft-versus-tumor effect in murine hematopoietic cell transplantation

Abstract

Donor derived lymphocytes from hematopoietic cell transplantation (allo-HCT) or donor lymphocyte infusion can mediate eradication of host tumor cells in a process labeled the Graft-versus-Tumor (GVT) effect. Unfortunately, these treatments have produced limited results in various types of leukemia due to insufficient GVT effect. In this context, molecular engineering of donor lymphocytes to increase the GVT effect may benefit cancer patients. Activating MyD88 signaling in CD8+ T cells via toll-like receptors enhances T cell activation and cytotoxicity. However, systemic administration of toll-like receptor ligands to stimulate MyD88 could induce hyper inflammation or elicit pro-tumor effects. To circumvent this problem, we devised a synthetic molecule consisting of MyD88 linked to the ectopic domain of CD8a (CD8α:MyD88). We used this construct to test the hypothesis that MyD88 co-stimulation in donor CD8+ T cells increases tumor control following allo-HCT in mice by increasing T cell activation, function, and direct tumor cytotoxicity. Indeed, an increase in both in vitro and in vivo tumor control was observed with CD8α:MyD88 T cells. This increase in the GVT response was associated with increased T cell expansion, increased functional capacity, and an increase in direct cytotoxic killing of the tumor cells. However, MyD88 co-stimulation in donor CD8+ T cells was linked to increased yet non-lethal GvHD in mice treated with these engineered CD8+ T cells. Given these observations, synthetic CD8α:MyD88 donor T cells may represent a unique and versatile approach to enhance the GVT response that merits further refinement to improve the effectiveness of allo-HCT.

INTRODUCTION

Allogeneic hematopoietic cell transplantation (allo-HCT) is an established treatment for different blood cancers (1). Over the past several decades, it has become evident that donor-derived lymphocytes increase engraftment and rejection of residual host tumor cells (2, 3). This was initially verified when patients receiving T cell depleted bone marrow grafts experienced an increase in graft failure and tumor relapse (4). The rejection of host tumor cells by donor lymphocytes has been characterized as the unique graft-versus-tumor (GVT) effect (5). Subsequently, donor lymphocyte infusion (DLI) is indicated for high risk patients undergoing Allo-HCT or for the treatment of relapsed disease. This treatment has shown to increase remission and prolong survival in patients with some leukemias, however efficacy remains poor for treating other cancers including lymphoma or acute myelogenous leukemia (6, 7). Further manipulation of donor lymphocyte populations and engineering their function could benefit additional patient populations with aggressive or high-risk disease (8). CD8+ T cells or cytotoxic T cells are a population of lymphocytes that target tumor cells to induce a strong GVT effect through different modalities of killing, including cytokine-, death ligand- and perforin/granzyme-mediated cell death (9). The GVT effect is known to be initiated through T cell activation by residual host dendritic cells and this is crucial for shaping a response to the host tumor cells (10, 11).

MyD88 activation and co-stimulation can be achieved through activation of Toll-like receptors (TLRs), IL-1R, IL-18R, and ST2/IL-33R (12–14). With the exception of TLR3, MyD88 binds to the TIR domain of activated TLRs, initiating the formation of a signaling complex that IRAK-4 binds leading to recruitment and phosphorylation of IRAK-1 (14). Through a series of steps involving TRAF-6 and TAK1, the NF-kB pathway is initiated leading to expression of genes involved in cell survival, proliferation, and inflammation (15). Activation of TLRs and the MyD88 signalosome can also function to recruit factors that activate MAPKs ERK, JNK, and p38. MAPKs recruit transcription factors for growth, proliferation, and survival (14). Activating MyD88 signaling in CD8+ T cells via TLR engagement enhances T cell function and proliferation (12, 13). However, utilizing TLR ligands to stimulate T cells presents major challenges due to ephemeral or low expression of TLRs (16). Additionally, TLR ligand activation of tumor cells has been shown to promote tumor survival and growth (16). To circumvent these challenges, we have developed a construct to produce CD8+ T cell targeted MyD88 co-stimulation. This construct consists of MyD88 directly linked to the ectopic domain of CD8α to induce MyD88 activation with TCR engagement. After transduction with this construct, tumor specific pmel T cells display characteristics of TLR activation and have enhanced efficacy in an experimental melanoma tumor model (17).

The impact of MyD88 in CD8+ T cells in the context of experimental transplantation is still controversial. One study using MyD88 knockout T cells showed that MyD88 was indispensable for effective GVT effect while another report claimed MyD88 was dispensable for GVT effect (18,19). In this study, we sought to define the effect of synthetic MyD88 co-stimulation in donor CD8+ T cells on tumor control in the setting of allo-HCT. Herein we find that synthetic MyD88 co-stimulation in CD8+ T cells significantly improves tumor control in vitro and in vivo. We observed enhanced proliferative capacity and expansion of CD8α:MyD88 transduced donor CD8+ T cells. Furthermore, MyD88 co-stimulation increased the cytotoxic functions of donor CD8+ T cells. The increase in T cell activation and function from MyD88 co-stimulus was linked to an increase in GvHD but not host death. Ultimately, we conclude that MyD88 co-stimulation in donor CD8+ T cells increases the GVT response after allo-HCT.

MATERIALS AND METHODS

Animals and tumor cells

BALB/c, 129×1/SvJ, and C57BL/6 male mice were obtained from The Jackson Laboratory. A20 lymphoma, derived from the BALB/c strain were transduced to express luciferase as described (20, 21). All mice were maintained in specific pathogen-free housing, and all experiments were conducted in accordance with the animal care guidelines from the Office of Animal Welfare Assurance at the University of Maryland School of Medicine Veterinary Resources using protocols approved by the IACUC committee.

Donor cell preparation, plasmid, viral packaging, and transduction

All T cell-deplete bone marrow (TCD-BM) cells were isolated from WT C57BL/6 or 129×1/SvJ mice. T cell depletion was performed with auto-MACS by using anti-CD90.2 microbeads (Miltenyi Biotec). Donor CD8+ T cells (purity ~90–95%) were purified from the spleens of WT C57BL/6 mice by using the CD8+ T cell negative Isolation Kit (STEM CELL). CD8+ T cells were activated to prepare for retroviral transduction by plating 1×10⁶ cells per well in 24 well plates that were coated with 1ug anti-CD3(BD Pharmingen) and supplemented with 2ug/mL soluble anti-CD28(BD Pharmingen) and 100U/mL IL-2 (NovusBio). Plasmids were designed and cloned as described previously using the pMIG retroviral vector (modified MSCV-IRES-GFP) (17). Plasmids for viral packaging were isolated from bacteria using an Endo-free MAXIPREP kit (Qiagen). Retrovirus was prepared by transfecting the Phoenix-ECO (ATCC CRL-3214) packaging cell line with GFP control or CD8α:MyD88 experimental construct plasmids using Lipofectamine 2000 (Invitrogen). Viral supernatant was collected at 48 and 72 hours and used for transduction 48 hours after T cell activation. The activated CD8+ T cells were transduced on Retronectin (Takara Bio). Virus was washed off the next day and cells were given 20U/mL IL-2 and allowed to propagate before T cell infusion or in vitro use.

Antibodies, reagents, and cellular assays

Antibodies including anti-mouse CD11b (M1/70), B220(RA3–6B2), CD3(145–2C11), TCRB(H57–597), TNFα(MP6-XT22), IFN-γ(XMG1.2), GzmB(GB11), CD107a(1D4B), H-2Kb(AF6–88.5), H-2kd(SF1–1.1), Tim-3(B8.2C12), PD-1(29F.1A12) were ordered from Biolegend and used for spectral flow cytometry. The Ki67 antibody (SolA15) from eBioscience was utilized for intracellular flow cytometric staining. Briefly, cells were washed using Flow buffer (PBS with 2% FBS) and FC receptors were blocked with addition of unlabeled anti-CD16/CD32 for 20 minutes. External markers and fixable LIVE/DEAD Aqua or NIR (Invitrogen) were stained together in PBS for 30 minutes and washed 2 times with Flow buffer. Intracellular stains were performed using the Intracellular Fixation & Permeabilization Buffer Set by eBioscience. Briefly, cells were fixed for 30 minutes at room temperature using the IC fixation buffer. Cells were resuspended in 1x permeabilization buffer and incubated for 5 minutes. Samples were split and test antibodies or Isotype control antibodies were added and incubated for 1 hour at room temperature and washed twice in 1x perm buffer before resuspending in flow buffer. Samples were run on the Aurora Spectral flow cytometer (Cytek) in the Center for Innovative Biomedical Resources at the University of Maryland School of Medicine. Unmixed samples were analyzed using Flowjo software (Flowjo LLC). Background staining was assessed with unstained controls, stained isotype controls, and experimental negative controls when possible. “Spike-in dead” and “all dead” controls of mixed cell populations (T cells and tumor) were used as positive controls to test the staining and gating strategy for annexinV/7-AAD staining.

T cell killing assays were setup as modified Mixed lymphocyte cytotoxicity reactions by adding a variable number of transduced C57BL/6 CD8+ T cells (10⁴ - 2×10⁵; 1:1 – 20:1) to a constant 10⁴ tumor cells (BALB/C origin). Assays were performed in T cell media containing an RPMI1640 media base with 10% FBS, 1% non-essential amino acids, 1% glutamine, and 0.05 mM of 2-Mercaptoethanol. Cells were added to a round bottom plate for flow cytometry assays and a black flat bottom plate was utilized for luciferase measurements (round bottoms are not available in black). Mismatched BMDC (BALB/C; H-2d) were added to the wells at a ratio metric 1:10 to the variable T cell number. Assays were quickly stained for surface markers (15 minutes) and washed with calcium buffer followed by the addition of Annexin V/7-AAD purchased as a kit from Biolegend. These samples were placed on ice and ran immediately on the Aurora Spectral flow cytometer (Cytek) mentioned previously. Live tumor control luciferase assays were setup similarly but cultured in black flat bottom plates. On the day of acquisition, these assays were supplemented with 0.15mg/mL luciferin and imaged on the Xenogen IVIS-200 (Perkin-Elmer) and analyzed using the Living Image software (Perkin-Elmer).

GVT and GvHD models

Experimental transplant tumor models utilized male C57BL/6 or 129×1/svj donors and BALB/c host mice which were irradiated with 9.5 Gy before transplant and T cell treatment. One day later, the hosts were inoculated i.v. with A20 cells (doses indicated in figure legends). After injecting tumor, mice received 3.5 × 10⁶ T cell depleted bone marrow (TCD-BM) cells alone or combined with (C57BL/6) wild type naïve CD8+ T cells(nCD8), control transduced CD8+ T cells(pMIG), or CD8α:MyD88 (pCM8) transduced CD8+ T cells (dose indicated in figure legends). In experiment where T cell transfer was delayed, intravenous injection of T cells was performed 7 days after tumor and bone marrow transplant in 200uL of phosphate buffered saline. Bioluminescence imaging was performed to monitor tumor burden as previously described (22). In GVT experiments, tumor burden was expressed as photon flux (photons per second) and disease score and weight were also used to assess the health of the animals through treatment. To assess GvHD we used the groups consisting of TCD-BM alone, TCD-BM plus pMig-GFP transduced control T cells, and TCD-BM plus pCM8 transduced T cells. To separate effects from tumor versus GvHD, these mice were not injected with tumor. T cell dose is indicated in the figure legends. In the GvHD T cell dose escalation experiment, mice were euthanized at the end of the experiment. Liver and large and small intestines were removed, formalin-fixed, sectioned, and stained with H&E. Blinded assessments of histopathological GvHD were made by an experienced pathologist using an established semi-quantitative scoring system (23, 24). Representative pictures were captured at 100x.

Statistics

All data processing was performed in Prism Graphpad V8. Student’s t-tests were used to compare statistical significance between groups of normally distributed samples and asterisks (ns = p> 0.05; *=p<0.05; **=p<0.01;***=p<0.001; ****=p<0.0001) indicate the degree and magnitude of significance between two groups. Kaplan-Meijer plots were created, and survival analysis was performed using the Gehan-Breslow-Wilcoxon test. To compare significance between survival groups, all combinations of groups were separated and paired, and p values were adjusted for multiple comparisons. In assessing statistical differences of tumor burden over time between in vivo groups, a 2-way anova test was utilized up to before mice began dying. To assess significance at individual time points, an octothorpe (#) was used to indicate when the pCM8 group was significantly different (p<0.05) from all other controls when compared using Student’s T Test. Logistic Growth/Inhibition curves were made in GraphPad using the transformed data algorithm in which data is normalized to the control samples (0:1; tumor only) as 100% growth without transforming to log. With this we could derive IC50 values which is defined as the number of T cells to produce 50% growth inhibition of the controls.

RESULTS

MyD88 co-stimulation in donor CD8+ T cells yields enhanced GVT effect.

We previously found that TLR5 agonist CBLB502 enhanced GVT without increasing GvHD (25). This effect was indirect and mediated through stimulation of accessory innate immune cells, increasing CD8+ T cell killing of tumor cells (25). However, other studies have found that CD8+ T cells express other TLRs and utilize MyD88 signaling to amplify activation and effector functions (26). To test the direct effect of the MyD88 pathway on T cells without confounding systemic treatment, we developed a synthetic MyD88 construct (Figure 1A - left) which has since been tested in a solid tumor model and found effective for the treatment of experimental melanoma in mice (17). Our construct contains an ectopic CD8α domain with a MyD88 molecule linked internally, but without the Toll/Il-1 Receptor (TIR) domain. This construct uses the pMIG retroviral vector as the backbone, requiring primary T cell activation for transduction. After transduction, secondary activation through the TCR or anti-CD3 antibodies induces activation of this novel construct. Transduction efficiencies are typically around 60% with a range between 55% and 72% observed (Figure 1A - right). Previous studies show that the effects of this construct are not from ectopic CD8α expression, but from the aggregation and activation of the linked MyD88 in conjunction with TCR stimulus (17).

Figure 1: Synthetic CD8α:MyD88 co-receptor in donor CD8+ T cells increases killing of allogeneic A20 lymphoma cells in vitro.

A) Left: Schematic representing the synthetic CD8α:MyD88 coreceptor. Magnetically sorted CD8 T cells were polyclonally activated with 1ug/mL platebound anti-CD3 and 2ug/mL soluble anti-CD28. On day 3 post activation they were retrovirally transduced to express either CD8α:MyD88(pCM8) or the empty vector(pMIG) with GFP only. Right: Representative for typical transduction efficiency for CD8+ T cells. Efficiency is measured two days post transduction by analyzing GFP expression on CD8+ cells via flow cytometry before injection into corresponding treatment group. B) Schematic describing a cytotoxic MLR with A20-luciferase tumor to test proof of concept of engineered T cells in an allo-HCT tumor model. Transduced T cells (H-2b) were co-cultured with constant A20 tumor (H-2d) and activated bone marrow derived dendritic cells (H-2d) 1:10 to T cells. In order to assess tumor control, luciferin substrate was added at C)24 hours and D) 48 hours after culture, plates were imaged using the Xenogen IVIS system. Total photon flux was plotted to assess tumor signal in each well. Representative results are shown from one of two independent experiments and data are presented as mean ± SD from triplicate samples. Student’s T test was utilized to determine statistical difference between pMig and pCM8 groups at a given timepoint.

As a proof of concept for MyD88-stimulated CD8+ T cell control of tumor in an allogeneic setting, we designed a modified Mixed-Lymphocyte Reaction where H-2b CD8+ T cells were sorted from C57BL/6 mice and transduced to express the control GFP vector (pMig) or the CD8α:MyD88 vector (pCM8). These cells were combined with bone marrow derived dendritic cells (BMDC) from BALB/c (H-2d) mice at a 10:1 ratio. A constant of 10⁴ Luciferase-expressing A20 lymphoma cells were used as live reporter cells to assess tumor control over time (Figure 1B). At 24 and 48 hours, D-Luciferin was added, and the plates were imaged and analyzed. We found that pCM8 transduced CD8+ T cells were better than pMig control T cells in inhibiting tumor growth at high ratios of 20:1 and 10:1 after 24 hours (Figure 1C). After 48 hours of culture, pCM8 transduced CD8+ T cells were better able to control tumor growth with significant differences with the greatest separation at the 1:1 ratio (figure 1D). This experiment provided antecedent evidence that MyD88 co-stimulation in donor CD8+ T cells could increase tumor control in an allogeneic setting.

To determine the impact of MyD88 co-stimulated CD8+ donor T cells in the GVT response, BALB/c (H-2d) host mice were conditioned with total body irradiation (TBI). One day later, mice were first intravenously injected with 0.5×10⁶ A20-luciferase lymphoma cells. This was immediately followed by transplantation with 3.5×10⁶ TCD-BM and treatments including no T cell infusion (TCD-BM only), infusion of 0.5×10⁶ CD8+ naïve T cells, infusion of pMig GFP control CD8+ T cells, or infusion of experimental pCM8 CD8+ T cells (Figure 2A). We performed bioluminescence imaging to measure tumor burden. As expected, there was tumor control in CD8 naïve and pMig control groups compared to TCD-BM only group, yet there was no significant difference between the activated pMig control and the naïve CD8+ T cell control. When the pCM8 T cell group was compared to all controls, there was a significant decrease in tumor burden observed over time (p< 0.0001) (Figure 2B top). Also apparent in each of the pCM8 treated mice were spontaneous but ephemeral regressions that varied in magnitude and time (Figure 2B bottom). To further examine the impact of MyD88 co-stimulation of CD8+ T cells with transplant on the tumor-bearing mice, we monitored survival over 50 days (Figure 2C). Transplant recipients of the naive CD8+ T cells or pMig CD8+ T cells had a median survival (MS) of 25 days compared to the TCD-BM only control which had an MS of 17 days. Recipients of the pCM8 transduced CD8+ T cells had an MS of 47 days which was significantly longer than all other groups [(pMig:pCM8 p=0.0043) (nCD8:pCM8 p=0.0041) (TCD:pCM8 p=0.0039)]. Additionally, mice were scored using a disease metric accounting for fur texture, posture, activity, skin condition, weight loss, and diarrhea (Figure 2D). Disease score in mice treated with pCM8 transduced CD8+ T cells was significantly lower after day 21 and until the end of the study.

Figure 2: Synthetic CD8α:MyD88 co-receptor in donor CD8+ T cells increases graft-versus-tumor (GVT) effect.

A) Schematic describing the experiment in which BALB/c host mice are given total body irradiation, inoculated with 0.5×10⁶ A20-luciferase cells, injected with 3.5×10⁶ T cell depleted bone-marrow (TCD-BM) and 0.5×10⁶ T cells from C57BL/6 or 129×1/SvJ mice for the corresponding T cell adoptive transfer group (n = 5 per group). Using live imaging, tumor measurements were obtained in A20-luciferase tumor bearing mice. B) Total photon flux was plotted over time with average and SEM. Anova was used to analyze mice during the period in which they were all alive and octothorpes (#) indicate significant difference (p<0.05) between pCM8 group and all other (remaining) groups at that time point using a Student’s t test between variables. Additionally, C) survival was measured until the experiment was terminated at 50 days. D) Disease score was assessed on a 0–12 scale by measuring and adding mouse activity (0–2), Fur texture (0–2), Posture (0–2), Diarrhea (0–2), skin (0–2), and weight (0–4) in each animal at time of image acquisition. Deceased mice maintain a disease score of 10 throughout the study. E) A similar experiment was performed but instead used 129×1/SvJ mice as bone marrow and T cell donors into BALB/c hosts and live imaging was performed as in “A”. TCD-BM used an n = 4 and the other groups used n = 5. F) Total photon flux was plotted over time with average and ± SEM. Octothorpes(#) indicate significant of p<0.05 between pCM8 group and all other (remaining) groups at that time point using a Student’s t test between variables. G) survival was measured until the experiment was terminated at 60 days. Anova test was used to ascertain differences between groups over time. Representative results are shown from two independent experiments with similar results for both C57BL/6 and 129×1/SvJ donors.

To determine whether the observed effects were generalizable across transplant models, we performed GVT experiment using 129/SvJ donors to the BALB/c hosts. In the first two weeks, all the CD8+ T cell treatment groups displayed an enhanced control of tumor growth over the ‘TCD-BM only’ control group. After day 18, the pCM8 transduced T cell group began showing a significant decrease in tumor burden compared to other groups throughout the remainder of the experiment (p < 0.0001) (figure 2E). Additionally, the pCM8 group displayed significantly improved survival compared to all other groups [(pMig:pCM8 p=0.0047) (nCD8:pCM8 p=0.0038) (TCD:pCM8 p=0.0047)] (Figure 2F).

In the standard GVT models described above, the bone marrow, tumor cells, and T cells are injected on the same day. This established model (25, 27, 28) for studying GVT in mice, mimics a scenario in which a transplant is performed for treatment, but residual tumor cells remain. However, second line treatments such as donor lymphocyte infusion (DLI) are typically given later after transplant. To test whether the enhanced efficacy of CD8:MyD88 T cells could occur in the setting of DLI, we performed an experiment where the injection of T cells was delayed for 7 days (figure 3A). Starting on day 18, the group treated with the pCM8 transduced T cells displayed significantly reduced tumor growth compared to all other treated and control groups (p<0.0001) (figure 3B). Also apparent in the pCM8 T cell treated group were varying spontaneous tumor regressions (figure 3C), which were also observed in our standard model. The pCM8 transduced T cell group also maintained better survival compared to all other groups [(pMig:pCM8 p=0.0495) (nCD8:pCM8 p=0.0197) (TCD:pCM8 p=0.0039)] (figure 3D). Although the standard model in figure 2 shows greater differences in tumor kinetics and survival, both the standard model and delayed DLI model show a similar pattern of efficacy.

Figure 3. Delayed transfer of CD8α:MyD88 T cells after experimental transplant increases graft-versus-tumor effect.

A) Schematic describing the in vivo experiment in which BALB/c host mice are given total body irradiation, injected with 3.5×10⁶ T cell deplete bone-marrow (TCD) from C57BL/6 mice and inoculated with 0.1×10⁶ A20-luciferase. Seven days later, mice were injected with 0.5×10⁶ T cells from the corresponding T cell adoptive transfer group. Using live imaging, tumor measurements were obtained in A20-luciferase tumor bearing mice. B) Total photon flux was plotted over time with average and SEM. The octothorpes(#) indicate p<0.05 between pCM8 group and all other (remaining) groups at that time point using a Student’s t test between each group. C) Individual values were also plotted to show spontaneous regressions in tumor growth. Additionally, D) survival was measured until the experiment was terminated at 50 days. Representative results are shown as mean ± SEM. Multiple T-tests were utilized to assess significance between groups at a single timepoint while anova test was used to ascertain differences between groups over time.

Together, results from these experiments indicate that synthetic MyD88 co-stimulation in donor CD8+ T cells enhances the GVT effect.

MyD88-stimulated expansion of donor CD8+ T cells induces non-lethal GvHD-related weight loss.

GVT and GvHD are linked in that both are mediated by allogeneic T cells, yet GvHD involves migration and damage to susceptible organs (29). This provided the rationale to examine the effect of this strategy on inducing GVHD. To separate the symptoms caused by the tumor and GvHD, allo-HCT was performed with TCD-BM and T cell infusion without tumor inoculation. In Figure 2, our data indicate there were not significant differences between naïve and activated CD8+ T cell infusion thus naïve CD8+ T cells were excluded from this experiment. Mice were scored for disease over time on the same established metric from Figure 2. We found that mice transplanted with the pCM8 transduced donor CD8+ T cells displayed a significantly higher disease score starting on day 17 through the remaining days of the study (Figure 4A). The weight loss metric was the highest fraction of the score, so mouse weight was assessed separately and found significantly lower than either the pMig activated control T cells or the TCD-BM only control throughout the length of the study (Figure 4B). Despite the observed weight changes, none of the treatment groups, including pCM8 infused mice, experienced mortalities through the length of the study (Figure 4C). To determine whether this effect was correlated with T cell expansion, some mice were bled on day 7 to determine their immune profile. In the pCM8 T cell infused mice, we found that live CD8+ T cells made up a significantly higher proportion of the donor immune cells in the blood indicating increased expansion of these cells (Figure 4D). This expansion was at the cost of donor myeloid cells in the blood which were significantly decreased in mice treated with the CD8+ T cells transduced with pCM8 (Figure 4E). When comparing the myeloid to T cells fractionally, there were on average three times more myeloid cells than T cells in the mice treated with the pMig control group compared to the pCM8 group (Figure 4F). We also examined activation status of these cells by staining for the high affinity IL-2 receptor alpha (CD25). We found that mice treated with CD8+ T cells transduced with the pCM8 construct had donor CD8+ T cells with about 2.5 times more expression of CD25 compared to the pMig GFP control CD8+ T cells (Figure 4G). Additionally, mice treated with the pCM8 transduced CD8+ T cells had significantly lower TIM-3+ PD-1+ double positive donor CD8+ T cells. This was due to a striking decrease in TIM-3 expression in these cells (Figure 4H). Overall, we conclude that although MyD88 co-stimulation in donor CD8+ T cells increases the GVT effect, it also impacts GvHD as correlated with increased T cell expansion and a more activated phenotype.

Figure 4. MyD88 co-stimulated expansion of donor CD8+ T cells correlates with GvHD-related weight loss.

BALB/c host mice were subjected to 9.5Gy TBI, injected with 3.5×10⁶ TCD-BM from C57BL/6 mice and injected with 1×10⁶ of indicated CD8+ T cell group in the absence of tumor (n=5/group). A) Disease score was assessed on a scale by measuring and adding mouse activity (0–2), Fur texture (0–2), Posture (0–2), Diarrhea (0–2), skin (0–2), and weight (0–4) in each animal. B) % weight change was assessed separately and C) survival was tracked during the length of the experiment. Furthermore, live retro-orbital bleeds were used on day 7 to assess D) donor CD8+ T cells, E) donor myeloid cells, and F) their ratios. Donor (H-2k^b) CD8+ T cells were analyzed for G) CD25 expression by measuring mean fluorescence intensity (MFI). H) PD-1 and TIM-3 positivity was also assessed on donor CD8+ T cells. This experiment is representative of two independently replicated experiments with similar results. Representative results are shown as mean ± SEM and significant differences of disease score and weight between groups over time was ascertained using anova test.

We next analyzed the effect of increasing doses of CD8:MyD88 T cells on GvHD coupled with histopathological analysis in order to determine the source of GvHD-related weight loss. Like the previous experiment, both pCM8 transduced T cell doses trended toward higher disease scores than either doses of the pMig control T cells or the TCD-BM only control (Figure 5A). These mice also trended toward lower body weight starting on day 22 through the remaining days of the study (Figure 5B). However, the effects of the pCM8 transduced T cells did not show a dose-dependent increase in disease score or weight loss. Despite the observed weight changes, none of the treatment groups experienced mortalities through the length of the study (Figure 5C). On day 55, the mice were euthanized, the tissues were fixed in formalin and sent for processing into H&E slides. Tissue slides were blind scored for 17 metrics in the intestinal tissues and 19 metrics in the Liver tissues. Minor increases in score were noted in the large intestine of both the 2×10⁶ and 5×10⁶ doses of the pCM8 T cell infusion (Figure 5D). The metrics with the greatest change in the pCM8 groups included increased lamina propria inflammation, increased lymphocyte infiltration, and increased apoptosis (Figure 5E). Together, these results suggest that pCM8 transduced T cells can cause increased but non-lethal GvHD.

Figure 5: MyD88 co-stimulated donor CD8+ T cells induce clinical, histopathological but non-lethal GVHD.

BALB/c host mice were subjected to 9.5Gy TBI, injected with 3.5×10⁶ TCD-BM from C57BL/6 mice and injected with 2×10⁶ or 5×10⁶ of the indicated CD8+ T cell group in the absence of tumor (n=4/group). A) Disease score was assessed as described in Fig 3A. B) % weight change was assessed separately as the strongest indicator of systemic GvHD. Disease score and weight are shown as mean ± SEM and significance between groups over time was ascertained using anova test. C) Survival was tracked until experiment termination on day 55. Furthermore, the mice were euthanized, and liver and large and small intestines were fixed in 10% formalin for processing into H&E stained slides. Histopathological GvHD was scored blind by a trained pathologist and D) summarized. The mean summarized score ± SD is indicated on the graph with score maximums represented by a dashed line and Student’s T Test utilized for statistical comparisons. E) Representative images are displayed for the large intestine. Colored arrows indicate the score indicators of GvHD including Apoptosis(red arrow), Intraepithelial lymphocytic infiltrates (blue arrow), and Lamina propria inflammation (green arrow).

Donor CD8+ T cells have an increased proliferative capacity and survival with MyD88 co-stimulation.

To produce a GVT response, lymphocytes must be activated and expand in response to host antigen presenting cells (30). To investigate whether MyD88 co-stimulated donor CD8+ T cells expand in vivo after allo-HCT, T cell proliferative capacity was assessed on day 4 post-transplant. T cell transduction was assessed before injection (Figure 6A). The spleens of the mice were dissociated and stained for flow cytometric analysis (Figure 6B). Donor CD8+ T cells were analyzed for Ki67, a graded marker indicative of cell proliferation that increases as the cells move towards mitosis (31). Residual host CD8+ cells were also analyzed as an internal control and displayed low Ki67 expression (Figure 6C). The pCM8 transduced donor CD8 T+ cells had significantly higher Ki67 expression compared to either the pMig control or the naïve CD8+ control (Figure 6D). When comparing the transduced GFP+ CD8+ donor T cells between the pMig control T cell and pCM8 T cell groups the differences were more apparent (Figure 6E). When analyzing the frequency of donor CD8+ T cells in the spleens of these mice over time, the pCM8 group remained higher at the earlier timepoints, indicating early expansion of these cells (Figure 6F). We also noted that on day 12 there was a contraction of donor T cells from both groups (Figure 6F). To see if these populations were maintained in the circulation, blood samples from these mice were analyzed. We found that mice treated with pCM8 transduced CD8+ T cells had significantly more donor T cells within the blood at day 12 compared to the pMig control (Figure 6G). In contrast, the inguinal lymph nodes collected from these mice, also displayed a drop off in donor CD8+ T cell populations, mimicking what we saw in the spleen (Figure 6H).

Figure 6: MyD88 co-stimulation enhances proliferation of donor CD8+ T cells in vivo.

(A) CD8+ T cells were sorted and transduced with pMig control or pCM8 experimental constructs and (right) efficiency measured. BALB/c host mice underwent TBI then were injected with 3.5×10⁶ TCD-BM from C57BL/6 mice, 0.5×10⁶ A20-luciferase and 0.5×10⁶ of the indicated CD8+ T cells. On day 4 post injection, mice were euthanized, and their spleens were harvested for cell isolation, staining and analysis. B) Donor and residual host CD8+ T cells were gated as shown. C) Quiescent residual host CD8+ cells were analyzed for Ki67 as a baseline intrinsic control. D) Ki67 was measured in all donor CD8+ T cells and E) between GFP (transduction) positive cells only. F) Donor CD8+ T cell expansion was measured in the spleen of these mice at multiple timepoints. G)Blood and H) Inguinal lymph nodes were taken at day 6 and 12 post-transplant in the presence of A20 tumor. Donor CD8+ populations were assessed in order to determine whether these cells enter circulation or are retained in the secondary lymphatics. This experiment is representative of two independently replicated experiments with similar results shown using mean ± SD. The p values and significance indications were acquired by Student’s t tests.

In order to test survival of donor CD8+ T cells, a modified-MLR was utilized which combined variable ratios of transduced donor CD8+ T cells (H-2b), activated BMDC 1:10 (H-2d), and constant A20 tumor (H-2d). This culture models the allogeneic host cell types in vivo and allows us to analyze how these populations impact the survival of CD8+ T cells with and without MyD88 co-stimulus. At each time point we utilized annexinV/7-AAD to see the fraction of T cells dead (7-AAD+/AnnexinV+) or initiating apoptosis (7-AAD-/Annexin V+) at 24 or 48 hours (Figure 7A). Twenty-four hours after culture we found that the pMig control T cells had a significantly increased frequency of dead cells at every ratio compared to the pCM8 transduced CD8+ T cells (Figure 7B). Additionally, at 24 hours there were significantly more early apoptotic cells in the pMig control cultures at the 10:1 and 1:1 ratios indicating better survival of the pCM8 transduced CD8+ T cells, even when tumor is increased (Figure 7C). At 48 hours the results were similar with increased death of pMig controls compared to the transduced pCM8 CD8+ T cell cultures (Figure 7D). Early apoptosis of the pMig control was also significantly higher than the transduced pCM8 CD8+ T cells after 48 hours (Figure 7E).

Figure 7: MyD88 co-stimulation enhances survival of allogeneic CD8+ T cells.

A) (Top) Transduced T cells (H-2b) were co-cultured with constant A20 tumor (H-2d) and activated bone marrow derived dendritic cells (H-2d) 1:10 to T cells. (Bottom) These cells were stained with phenotypic markers and Annexin V/7-AAD was added to assess death and survival of these cells. Dead cells (AnnexinV+ 7-AAD+) at B) 24 hours and D) 48 hours and Apoptotic cells (AnnexinV+ 7-AAD-) at C) 24 hours and E) 48 hours were analyzed on CD8+ T cells. This experiment is representative from one of two independently replicated experiments with similar results shown using mean ± SD (in triplicate). The p values and significance indications were acquired by Student’s t tests.

Together, these data indicate that the MyD88 co-stimulated donor CD8+ T cells have an increased capacity for proliferation and survival in the allogeneic setting.

MyD88 co-stimulation increases multi-modal functionality in donor CD8+ T cells

To determine the tumor control mechanisms of MyD88 co-stimulated donor CD8+ T cells, CD8+ T cells from mice were characterized 4 days after allo-HCT. Mouse spleens were extracted and stained for cytokines interferon gamma (IFNγ) and tumor necrosis factor alpha (TNFα) (Figure 8A). We were able to detect significantly increased double positive donor CD8+ T cells in mice treated with CD8α:MyD88 donor T cells (Figure 8B). This increase in multi-functional CD8+ T cells was weighted by a significant increase in TNFα positive cells (Figure 8C) and a trending increase in IFNγ (Figure 8D). We also stained the cells for Granzyme B and CD107a to determine their potential for killing with Granzyme B and assess their secretory capacity (Figure 8E). Donor CD8+ T cells transduced with CD8α:MyD88 had three times more Granzyme B positive cells than pMig control or naïve CD8 control T cells (Figure 8F). This group also displayed two times more Granzyme B+ CD107a+ cells than controls (Figure 8G) and had twice the total granzyme B expression of control T cell groups (Figure 8H). Together these data indicate that MyD88 co-stimulated donor CD8+ T cells have enhanced functions for cytotoxicity.

Figure 8: MyD88 co-stimulation enhances multi-modal function of donor CD8+ T cells.

BALB/c host mice were conditioned with TBI (9.5Gy) then injected with 3.5×10⁶ T cell deplete bone-marrow from C57BL/6 mice, 0.5×10⁶ A20-luciferase and 1×10⁶ of the indicated T cell infusion (TCD-BM n=3, pMig n=4, pCM8 n=4). On day 4 post injection, mice were euthanized, and their spleens were harvested for cell isolation, staining and analysis. Donor CD8+ T cells were gated on from live -> singlet -> H-2kb+ populations. A) These cells were analyzed for Interferon gamma and TNFa B) together as double positive cells and individual C) TNFα and D) IFNy single positive populations. E) Additionally, donor CD8+ T cells were examined for expression of Granzyme B and CD107a and assessed for F) granzyme B positivity G) Secretory CD107a+ Granzyme B+ and H) Granzyme B MFI. Results shown are mean ± SD and representative from one of two independently replicated experiments. The p values and significance indications were acquired by Student’s t tests.

MyD88 stimulated donor CD8+ T cells show increased direct cytotoxicity to tumor cells

To determine whether MyD88-stimulated control of tumor is through a direct killing mechanism, we utilized our modified MLR system. Again, CD8+ T cells were transduced with equivalent efficiency (Figure 9A). Transduced CD8+ T cells (H-2b) were co-cultured with 1:10 activated BMDC (H-2d) and a constant 10⁴ A20 lymphoma (H-2d). Different ratios of transduced CD8+ T cells were added to the co-cultures in order to assess killing and tumor control caused by the T cells. At each timepoint, cells were stained for surface markers, Annexin V, and 7-AAD and data was immediately acquired. Tumor cells were analyzed for death and early apoptosis within the population. Additionally, live tumor cells of the total cell population were normalized to the control and used to assess percent tumor growth and obtain an IC50 to measure and compare T cell efficacy. At 24 hours we observed a significant increase in dead tumor cells (Figure 9B) and early apoptotic tumor cells (Figure 9C) in samples treated with CD8α:MyD88 T cells. At 24 hours, significant differences were observed at higher T cell to tumor ratios of 20:1 and 10:1. When assessing how CD8α:MyD88 T cells impacted tumor growth, we found a significant decrease in growth at each ratio (Figure 9D). After 48 hours of co-culture, we found that the samples containing the pCM8 transduced CD8+ T cells had significantly increased dead tumor cells across all ratios (Figure 9E), although the early apoptotic percentages were overall decreased compared to 24 hours (Figure 9F). When analyzing the impact of CD8α:MyD88 transduced CD8+ T cells on % tumor growth after 48 hours, we found the IC50 was still under 1 T cell per tumor cell (0.49) while the pMig IC50 was 2.77 T cells per tumor cells for 50% control of tumor growth (Figure 9G). Tumor growth across all ratios was significantly less in the CD8α:MyD88 T cell group (Figure 9G). Most striking was the 1:1 ratio which was under 50% tumor cell growth in the CD8α:MyD88 T cell group while the GFP control group had tumor growth almost equal to that of the tumor only control (e.g. 100%) at the same ratio (Figure 9G). Together, these data indicate that MyD88 co-stimulated donor CD8+ T cells directly kill tumor cells and do so more efficiently than the transduced GFP control T cells.

Figure 9: Tumor control by MyD88 co-stimulation donor CD8+ T cells is the result of direct killing.

A) (Left) Donor CD8+ T cells were sorted, activated, and transduced with either pMig-GFP control vector (pMIG) or pMIG-CD8α:MyD88 vector (pCM8). Varying T cell concentrations were than co-cultured with a constant number (10⁴) of A20 lymphoma +/−H2kd bone marrow-derived dendritic cells kept at 1:10 ratio (BMDC:T cells). (Right) Tumor cells were analyzed by gating on singlet B220+ cells and further on 7-AAD viability dye by Annexin V apoptosis indicator. Tumor populations were separated by Dead tumor cells at B) 24 hours and E) 48 hours, early apoptotic tumor cells at C) 24 hours and F) 48 hours and Logistic tumor growth curves at C) 24 hours and G) 48 hours. Experiments were performed in triplicate and independently replicated two times with results as mean ± SD. The p values and significance indications were acquired by Student’s t tests. IC50 values were obtained from logistic growth curve analysis in GraphPad Prism software.

DISCUSSION

This study presents a novel strategy of donor T cell engineering and infusion for allo-HCT, where fusion protein CD8α:MyD88 activates MyD88 signaling in donor CD8+ T cells without the need for systemic administration of a ligand or cytokine. This approach resulted in an enhanced GVT effect. To the best of our knowledge, this is the first strategy designed to boost the GVT response through molecular engineering of MyD88 for donor T cell infusion. Donor lymphocyte infusions in some leukemias often fail to clear tumor and patients often succumb to their disease. For example, one study found that patients with adult lymphoid malignancies had an overall survival of only 13% in 3 years after receiving a donor lymphocyte infusion (32). Our results provide proof of concept that synthetic engineering of CD8α:MyD88 may improve the effectiveness of donor lymphocyte infusion in the setting of allo-HCT. However, due to the link between GVT and GvHD it was rational to predict that enhanced MyD88 in donor CD8+ T cells may elicit some GvHD symptoms in these mice. Our results indicated that mice treated with CD8α:MyD88 T cells showed significantly increased weight loss, yet these mice did not display other severe GvHD related side effects and survived throughout the course of the study. When we escalated the dose of administered T cells, we found there was no dose dependent increase in GvHD-related symptoms. Upon histopathological analysis we found that weight loss was associated with minor increases in lamina propria inflammation, intraepithelial lymphocyte infiltration, and apoptosis occurring in the large intestine. However, these mice survived throughout the course of the study, even at the highest dose, indicating MyD88 co-stimulation induces increased but non-lethal GvHD in mice.

From our previous study we can conclude that the CD8α domain of the construct enhances activation of MyD88 in conjunction with activation of TCR (17). Only when these cells underwent antigen-specific activation, MyD88 was found aggregating at or proximal to the immunological synapse, indicating that the CD8α domain of the construct promotes MyD88 aggregation during TCR activation(17). In this study, the TCR activation signal would come from allogeneic MHC stimulation by host dendritic cells and tumor cells. Our CD8α:MyD88 and GFP control T cells are already activated going into the host mouse; however, not all T cells respond to allogeneic stimulation. Host DCs and tumor are likely to play a role in further stimulating and shaping the responding population of T cells. We found that CD8α:MyD88 T cells displayed enhanced capacity to proliferate in the allo-HCT system. Ki67 is a graded marker that increases as cells move towards S phase, G2 phase, and peaks at M phase (31). In addition to showing an increased proliferative capacity, CD8α:MyD88 donor CD8+ T cells expanded early in secondary lymphatics, where they would see residual host APCs. This was followed by a contraction of all donor T cell populations by day 12. Systemically, MyD88 co-stimulated CD8+ T cells maintained a larger presence in the blood through day 12 post-transplant indicating systemic retention of these cells. The contraction of these cells in secondary lymphatics is important to make room for reconstitution of other immune cells while retention in the blood shows these cells could still be active in eliminating tumor cells. Ultimately, the proliferation profile, expansion and notable increase in survival of the CD8α:MyD88 T cells further mimic treatment of T cells with TLR ligand but without systemic off-target effects (33).

TLR2 agonists in CD8+ T cells increase T-bet and its binding to Granzyme B, Perforin, and IFNγ promoters (34). In previous studies using our CD8α:MyD88 construct in pmel-specific CD8+ T cells, we also found a notable increase in IFNγ and other cytokines (17). In the current study, we measured several functional cytokines in the donor CD8+ T cells after isolating them from the spleens of these mice on day 4 post-transplant. We were able to detect increased multiple cytokine producing donor CD8+ T cells. There were also more granzyme B positive donor CD8+ T cells and these cells co-expressed secretory marker CD107a. The increase in granzyme B and CD107a double positive donor T cells in the mice treated with CD8α:MyD88 T cells could indicate an increased expansion of allogeneic responders. We know that this construct can produce a more sensitive MHC and TCR interaction (17), thus it is likely that synthetic CD8α:MyD88 increases the reactivity of donor CD8+ T cells to host antigens and tumor antigens in this allo-HCT system. The observed increase in T cell functions led us to test whether MyD88 co-stimulation in T cells elicits direct killing in target tumor cells. We found that lymphoma cells co-cultured with CD8α:MyD88 T cells had significantly more apoptotic and dead cells down to the 1:1 T cell to Tumor ratio and at each timepoint. Conversely, pMig control T cells could not control tumor growth. This occurred in a T cell-dose dependent manner, indicating the T cells were the dominant parameter modulating tumor death.

The biological role of MyD88 in T cells in the context of experimental transplant has been controversial. Most studies analyzing MyD88 utilize a knockout approach. One study showed that MyD88 is indispensable for effective GVT (18), which complements other studies that found antigen specific CD8+ T cells lacking MyD88 are partially defective and have impaired responses to lymphocytic choriomeningitis virus (35, 36). In contrast, another study concluded that MyD88 ablation reduces GvHD while maintaining GVT effects (37). In both studies, pan T cells including CD4+ T cells were utilized. Our previous studies (38, 39) have shown that although CD4+ T cells contribute to GvT effect, a small number (5×10⁴) of CD4+ T cells could cause hyperacute lethal GvHD between 1–2 weeks after transplant in this model system. Because of this concern, we have devised this CD8α:MyD88 construct to selectively enhance CD8+ T cell function as a strategy to improve the GvT effect without causing severe and lethal GVHD. Although we did see an increase in GvHD symptoms of the gut with our CD8α:MyD88 modified CD8+ T cells, we did not observe any lethality at even 5×10⁶ CD8+ T cell dose. Therefore, our study would support a new strategy of engineering human CD8+ T cells for DLI. Currently DLI in the clinic does not separate CD4+ and CD8+ lymphocytes, so the use of this construct would require a different protocol from the current clinical procedure. Further modifications to this construct (e.g., inducible suicide genes, tumor specific TCRs) may increase safety, specificity, and efficacy in the clinical setting.

To summarize, synthetic engineering of MyD88 co-stimulation into donor CD8+ T cells represents a unique and versatile approach to enhance the GVT response. We found these cells were more activated and primed to kill tumor cells in the setting of allo-HCT. Overall, our approach provides novel proof of concept for engineering donor T cell infusions to improve their efficacy.

KEY POINTS.

MyD88 co-stimulation in donor CD8+ T cells improves the graft-versus-tumor effect.

MyD88 co-stimulated CD8+ T cells are more activated with improved cytotoxic function.