Article Text
PDF
Abstract
Background Chimeric antigen receptor (CAR)-T cells face many obstacles in solid tumor therapy, including heterogeneous antigen expression and inefficient T cell persistence. Guanylyl cyclase C (GUCY2C) has been identified as a suitable tumor antigen for targeted therapy due to its intestinal-restricted expression pattern in normal tissues and steady overexpression in gastrointestinal tumors, especially colorectal cancer. An antigen-sensitive and long-lasting CAR-T cell targeting GUCY2C was investigated in this study.
Methods Using constructed tumor cell lines with various GUCY2C expression densities, we screened out an antigen-sensitive single chain variable fragment (scFv) that enabled CAR-T cells to efficiently eradicate the GUCY2C lowly expressed tumor cells. CAR-T cells with different compositions of the hinge, transmembrane and costimulatory domains were also constructed for selection of the long-lasting CAR-T format with durable antitumor efficacy in vitro and in tumor-bearing mice. The underlying mechanism was further investigated based on mutation of the hinge and transmembrane domains.
Results We found that the composition of the antigen-sensitive scFv, CD8α hinge, CD8α transmembrane, and CD28 costimulatory domains boosted CAR-T cells to rapidly kill tumors, maintain high expansion capacity, and long-term efficacy in various colorectal cancer models. The durable antitumor function was attributed to the optimal CAR tonic signaling that conferred CAR-T cells with autonomous activation, proliferation, survival and cytokine release in the absence of antigen stimulation. The tonic signaling was associated with the length and the cysteine residues in the CD8α hinge and transmembrane domains.
Conclusions This study demonstrated a potent GUCY2C-targeted CAR-T cell for gastrointestinal tumor therapy and highlights the importance of adequate tonic signaling for effective CAR-T cell therapy against solid tumors.
- Colorectal Cancer
- Adoptive cell therapy - ACT
- Chimeric antigen receptor - CAR
Data availability statement
No data are available. All data relevant to the study are included in the article or uploaded as online supplemental information.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Guanylyl cyclase C (GUCY2C) has been identified as a suitable gastrointestinal tumor antigen for development of targeted therapies including bispecific antibody and CAR-T. CAR-T cells face many obstacles in solid tumor therapy, including heterogeneous antigen expression and inefficient T cell persistence. A murine GUCY2C CAR-T has been validated to be capable of eliminating colorectal cancer metastases in syngeneic mouse tumor models. However, it has never been thoroughly studied on the optimal human GUCY2C CAR-T design against the complicated solid tumors and the underlying mechanism, which is important for the clinical application in future.
WHAT THIS STUDY ADDS
This study provides an antigen-sensitive and long-lasting GUCY2C-targeted CAR-T cell that overcame the obstacles of solid tumor therapy including heterogeneous antigen expression and inefficient T cell persistence. The durable antitumor function of our GUCY2C CAR-T was attributed to the optimal CAR tonic signaling that was associated with the length and the cysteine residues in the CD8α hinge and transmembrane domains.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
This study demonstrates a potentially leading GUCY2C CAR-T candidate that could rapidly and durably inhibit GUCY2C-positive tumors, providing the opportunity for further clinical applications in metastatic gastrointestinal cancers. We also illustrate the optimal tonic signaling and basal T cell activation as informative parameters to guide the rational design of next-generation CAR-T cells for cancer therapy.
Introduction
Gastrointestinal (GI) cancer encompasses tumors arising from the proximal esophagus to the distal rectum and represents over a quarter of all cancers. Colorectal cancer (CRC) is the most frequent GI cancer type and led to more than 900,000 deaths worldwide in 2020.1 2 Two-thirds of patients with cancer can adopt the surgical resection after diagnosis, but about 50% of these patients show tumor recurrence afterwards.3 Adjuvant chemotherapy, epidermal growth factor receptor-targeted therapy, antiangiogenic therapy and immune checkpoint-based immunotherapy are main frontline treatments. However, a majority of patients usually develop drug resistance after several rounds of therapies.4 For the third-line therapy of CRC, only three drugs have been approved which showed the overall response rate of less than 5%.5 Therefore, development of new therapies for refractory or relapsed CRC is imperative, especially for the third-line or late-line therapies.
Guanylyl cyclase C (GUCY2C) is a type I transmembrane receptor that produces the second messenger cGMP after activation by its hormone ligands guanylin or uroguanylin and regulates the intestinal homeostasis, tumorigenesis, and obesity.6 7 GUCY2C is selectively expressed in the intestinal epithelial tight junctions and a subset of neurons and is enriched in the apical side of these normal cells.8–10 It is steadily overexpressed during the tumorigenesis of CRC and is also highly expressed in other GI cancers such as stomach, esophageal, and pancreatic cancers.9 Moreover, GUCY2C distributes in both the apical and basolateral membranes of tumor cells due to disruption of the tight junctions,11 which is beneficial for interaction with targeted therapies. The expression pattern of GUCY2C makes it to be a suitable antigen for GI cancers, especially for the metastatic lesions without apical-basolateral polarization. Previous studies have reported a GUCY2C-targeted bispecific antibody and a CAR-T cell that showed potent antitumor efficacy, proving the feasibility of developing GUCY2C-targeted therapies.12–14
In this study, we described the development and preclinical evaluation of a GUCY2C-targeted CAR-T cell with high antigen sensitivity and durable efficacy. The CAR molecule was composed of a single chain variable fragment (scFv) with high binding affinity to GUCY2C lowly expressed cells, CD8α-derived hinge and transmembrane domains, CD28 costimulatory domain, and CD3ζ intracellular domain. The durable antitumor function was attributed to the optimal CAR tonic signaling that conferred CAR-T cells with antigen independent activation, proliferation, survival and cytokine release. Moreover, we elucidated that the tonic signaling was associated with the positively charged residues of the antigen binding domain, the length and the cysteine residues in the CD8α hinge and transmembrane domains. Our study provided a potent GUCY2C CAR-T cell with promising antitumor efficacy and durable persistence and demonstrated the importance of adequate CAR tonic signaling for solid tumor CAR-T therapy.
Methods
Cell lines
HCT116 cells were obtained from the American Type Culture Collection and cultured in Iscove’s Modified Dulbecco’s Medium medium (Gibco, Cat# 31980030) with 10% fetal bovine serum (Bioind, Cat# 04-001-1 C04001-500). The low GUCY2C-expressing cell (HCT-116-hGCC-L) and high GUCY2C-expressing cell (HCT-116-hGCC-H) were generated by transduction of HCT116 cells with human GUCY2C lentivirus and flow cytometry sorting according to the GUCY2C expression intensity. HCT116-hGCC-L(H)-GFP cells were generated by lentiviral transduction of HCT116-hGCC-L(H) cells with firefly luciferase and GFP genes.
Anti-GUCY2C scFv affinity detection
The humanized scFvs YM01 and YM02 against GUCY2C were derived from previously reported GUCY2C-specific antibody clones, respectively.12 13 The scFv sequences were cloned into the pFUSE-hIgG-Fc2 plasmid (Invivogen) that was transfected to 293F cells (Life Technologies) for production of YM01 or YM02 scFv-Fc proteins. The scFv-Fc proteins were purified using protein G-Sepharose column (GE healthcare).
For the sandwich ELISA, the plate was coated with YM01 or YM02 scFv-Fc in phosphate buffered saline (PBS). Recombinant human GUCY2C-6×his protein (Acro Biosystems) at various concentrations were then added, and the horseradish peroxidase conjugated anti-6×His tag (Abcam) at 1:5000 dilution was further incubated for analysis of the binding affinity.
For flow cytometry, YM01 and YM02 CAR-T cells were stained with the LIVE/DEAD Cell Stain Kit (Invitrogen) and labeled with the recombinant human GUCY2C-6×his protein for 0.5 hour at 4°C. After washing, CAR-T cells were stained with anti-6×His-Alexa Fluor 647 conjugate (Biolegend, Cat#326611) and anti-Strep tag II-P-phycoerythrin (PE) conjugate for 0.5 hour at 4°C. Cells were analyzed on a Northern Lights instrument (Cytek Biosciences) after washed twice. Binding of hGUCY2C was determined by mean fluorescence intensity of Alexa Fluor 647 on live CAR-T cells. Non-linear regression analysis (GraphPad Prism V.6) was used to determine the EC50.
CAR-T cell production
Human T cells were isolated and activated by priming peripheral blood mononuclear cells from healthy donors with CD3/CD28 Dynabeads (Thermo Fisher) at a ratio of 1.5:1 (beads: cells). One day after activation, T cells were transduced with various CAR-encoding lentiviral vectors and maintained at 1.5×106 cells/mL in X-VIVO media with 500 U/mL human interleukin-2 (IL-2, SL Pharm, China). Beads were removed 4 days after activation. The CAR-T cell number and viability were tested every 2 days using automated cell counter (Countess 3 FL, Thermo Fisher).
Flow cytometry
All flow cytometry measurements were performed on Northern Lights (Cytek Biosciences). GUCY2C expression on HCT-116-hGCC-L and HCT-116-hGCC-H cell lines was detected using PE-conjugated anti-GUCY2C antibody (R&D system, Cat#: FAB2157P). During the CAR-T cell culture, CD3 expression was detected using CD3 antibody (Biolegend, Cat#300439). Anti-Strep tag II monoclonal antibody (clone 8F8D1) was developed by immunization of mice with the Strep tag II peptide and hybridoma screening. The CAR expression was detected by incubation of cells with PE labeled anti-Strep tag II at the dilution of 1:200 for 20 min at room temperature. For the CAR-T cell differentiation panel, CAR-T cells were stained with anti-human CD45RA antibody (BioLegend, clone HI100, BV510) and anti-human CD62L antibody (BioLegend, clone DREG-56, APC/Fire750). For the T cell exhaustion panel, T cells were stained with anti-human LAG-3 antibody (BioLegend, clone 11C3C65, PE/Cy7), PD-1 antibody (BioLegend, clone EH12.2H7, BV650). The T cell phenotype was evaluated by staining of CD25 (BioLegend, clone M-A251, FITC), CD27 (BioLegend, clone O323, APC), CD69 (BioLegend, clone FN50, APC/Cy7), 4-1BB (BioLegend, clone 4B4-1, PE/Cy7). All FACS plots of CAR-T cell phenotype were analyzed on gated CAR+ cells. For mock T cells, the whole T cell populations were analyzed.
Real-time cytotoxicity assay
HCT-116-hGCC-L cells were seeded into 96-well E-plates (Acea Biosciences, USA) at 2×104 cells per well and monitored overnight using the impedance-based real-time cytotoxicity assay (RTCA) iCELLigence system (Acea Biosciences). Then CAR-T cells were added at an effector to target (E:T) ratio of 1:3. The co-cultured cells were monitored for 2 days using the RTCA system, and the impedance of each well that represented the tumor cell survival was plotted over time. Each experiment was performed in triplicate.
Chronic stimulation assay
For multiple rounds of co-culture with HCT116-hGCC-L-GFP or HCT116-hGCC-H-GFP cells, tumor cells were seeded into 24-well plates (1×105 cells/well) 1 day prior to the addition of mock T or CAR-T cells at certain E:T ratio in at least four duplicates. At the end of co-culture, cells in each well were harvested for quantification of residual tumor cells (GFP+) and CAR-T cells (CD3+ and strep tag II+) using flow cytometry with Count Bright absolute counting beads (Thermo Scientific).
Cell line-derived xenograft model
Each immunodeficient mouse (NCG mouse, 5–6 weeks old, GemPharmatech) received a subcutaneous injection of 1 million HCT-116-hGCC-L cells. GUCY2C CAR-T cells (5 million) were intravenously injected when tumor volumes reached ~200 mm3. Tumor volumes and body weights were monitored twice a week. Tumor volume was calculated using the formula 1/2×length×width2. The level of CAR-T cells in peripheral blood was evaluated by staining of anti-CD3-APC and anti-Strep tag II-PE and flow cytometry analysis (ACEA Biosciences, NovoCyte 2060R).
Organoid cytotoxicity assay
The organoid cytotoxicity assay was performed following the previous report.15 Briefly, a 8 µL of matrigel gelatin was added to each well of an IBAC O2 chip (Daxiang Biotech, OE100811) and premoistened for 48 hours. The gelatin was allowed to solidify consecutively at room temperature and 37°C for 1 hour. Enzymatically dissociated organoids were counted using a hemocytometer, and 1000 cells were seeded into each well of the chip. Organoids were cultured for 48 hours in culture medium supplemented with Y-27632 (Daxiang Biotech, IA100101).
CAR-T cells were stained with CellTracker Deep Red (Invitrogen, C34565), and resuspended in the X-VIVO medium with the addition of a fluorescent caspase-3/7 indicator (C10423, Invitrogen, dilution ratio of 1:1000). The organoid medium was replaced, and 60 µL of cell suspension containing 1500 of CAR-T cells were added to each well. Cells were co-cultured at 37°C for 5 days and were monitored at indicated time points using an ImageXpress Confocal microscope (Molecular Devices). The fluorescence intensity of caspase-3 overlapping with T cells was measured.
Patient-derived xenograft model
Fresh tissues in tissue storage solution (Miltenyi, Cat# 130-100-008) were cut into pieces with a diameter of 2 mm and subcutaneously inoculated into 6-week-old NCG female mice. When tumor volume reached ~250 mm3, mice were intravenously injected with mock T cells or CAR-T cells (5 million) suspended in 100 µL PBS. Tumor volume was measured twice a week. When tumor volume reached over 2000 mm3, mice were sacrificed.
Statistical analysis
Statistical analyses were performed by using GraphPad Prism V.8.0. For comparisons of three or more groups, a one-way or two-way analysis of variance (ANOVA) was used followed by Tukey’s multiple comparisons test. A p<0.05 was considered as significant difference.
Results
GUCY2C expression pattern in CRC and normal tissues
To evaluate the expression level of GUCY2C in CRC tissues and normal tissues, we developed an immunohistochemistry (IHC) antibody that specifically recognized the extracellular domain of GUCY2C to characterize the membrane expression (online supplemental figure S1a). Using this antibody, we performed the IHC staining of 390 CRC tissue samples and a tissue microchip containing 33 different normal tissues. Among 390 CRC samples, 138 cases (35.4%) displayed high expression level (IHC score of +3), and 90 cases (23.1%) showed medium level (IHC score of +2). Low expression level (IHC score of +1) was found in 96 cases (24.6%), and only 66 cases (16.9%) showed negative staining (figure 1). In normal tissues, we observed restricted expression of GUCY2C on the apical side of small intestine and colon, and negative expression in all other tissues (online supplemental figure S1b). These results verified the expression pattern of GUCY2C that was suitable for developing CAR-T therapy.
Supplemental material
GUCY2C expression in human metastatic colorectal cancers and immunohistochemistry. (a) Statistics of different immunointensities of GUCY2C in mCRC. Histological type: 3+, strong expression; 2+, moderate expression; 1+, weak expression; 0, negative expression. (b) Representative tissue sections with different intensities of GUCY2C.
The sensitivity of the antigen-binding domain is critical for the antitumor activity of GUCY2C CAR-T cells
The basic elements of a CAR molecule consist of the antigen binding domain, hinge, transmembrane, costimulatory, and CD3ζ intracellular domains. To generate a GUCY2C-targeted CAR-T cell, we first screened two humanized scFvs (named YM01 and YM02) that specifically recognize the extracellular domain of GUCY2C (online supplemental figure S2a). To evaluate the antigen sensitivity of scFvs, we further detected the binding capacity of YM01 and YM02 to HCT-116-hGCC-L and HCT-116-hGCC-H cells using flow cytometry. We found that both scFvs bound equally well to HCT-116-hGCC-H cells. Moderate binding of YM02 to HCT-116-hGCC-L cells was observed, while YM01 showed sufficient binding to HCT-116-hGCC-L cells, suggesting the higher sensitivity of YM01 to low density of antigens on the cell surface. Meanwhile, neither scFv could bind to GUCY2C-negative HCT116 cells, indicating the GUCY2C-specific recognition of both scFvs (online supplemental figure S2b). Furthermore, we verified that YM01 could not recognize murine GUCY2C (online supplemental figure S2c).
We next constructed two CAR-T cells using YM01 or YM02 in combination with CD28-derived hinge, transmembrane, and endodomain, and the CD3ζ signaling domain (figure 2a). Both CAR-T cells exhibited approximately 40% of CAR transduction efficiency and maintained steady expression over culture time (figure 2b). We examined the binding of recombinant human GUCY2C proteins to above CAR-T cells and found that YM01 CAR-T showed extremely higher avidity to GUCY2C protein compared with YM02 CAR-T (figure 2c). Then we evaluated the cytotoxic ability of CAR-T cells to HCT-116-hGCC-L cells. Using the RTCA, we observed faster and higher cytolytic activity of YM01 CAR-T cells over YM02 CAR-T cells (figure 2d). Accordingly, YM01 CAR-T cells triggered much higher secretion of cytokines including interferon-γ (IFN-γ), tumor necrosis factor-α (TNF-α), and interleukin-2 (IL-2) than YM02 CAR-T cells after exposure to HCT-116-hGCC-L cells (figure 2e). To evaluate the long-term effector function and persistence of CAR-T cells, we examined the killing efficiency and expansion level of both CAR-T cells after multiround chronic stimulation with HCT-116-hGCC-L cells (figure 2f). After repetitive tumor inoculation for three times with HCT-116-hGCC-L cells, significantly lower amount of residual tumor cells (figure 2g) as well as higher number of CAR-T cells (figure 2h) were observed in YM01 CAR-T group, indicating that YM01 CAR-T cells showed more durable antitumor activity along with stronger expansion and persistence.
YM01 CAR-T cells show enhanced avidity to GUCY2C and GUCY2C-specific cytotoxicity at low effector:target (E:T) ratios. (a) Schematic representation of retroviral vectors encoding YM01 CAR-T and YM02 CAR-T. (b) Representative flow cytometry analysis of YM01 CAR and YM02 CAR transduction efficiency in primary T cells at different time points. (c) CARs avidity to GUCY2C was detected by flow cytometric staining with purified hGUCY2CECD-hFC labeled with PE. hGUCY2CECD-hFC binding curves for YM01 CAR-T and YM02 CAR-T cells, gated on live Strep Tag II+CAR T cells. (d) The real-time cytotoxicity assay (RTCA) of YM01 CAR-T and YM02 CAR-T cells in vitro against HCT-116-hGCC-L at E:T ratio of 1:1 using RTCA (n=3). After 24 hours of co-culture, IFN-γ, TNF-α, IL-2 in the supernatant were measured by Cytokine Bead Array (CBA) Kit, respectively (e). (f) Schema of the repetitive chronic stimulation assay. Tumor cells were seeded in 48-well plates 1 day prior to the addition of T cells. At day 0, CAR-T cells were added at T cell to tumor cell ratio of 1:3. Every 3 days, all T cells were collected and transferred into a new well in which 1×105 HCT-116-hGCC-L cells were seeded 1 day before. Residual HCT-116-hGCC-L cells (g) and number of CAR-T cells (h) at the round 3 of co-culture were collected and enumerated by flow cytometry. (i) Tumor size was measured and monitored over time. (j) Number of CAR-T cells (CD3+CAR+) in peripheral blood of mice detected in various groups by flow cytometry over time (n=6). Bars represent means±SD, *p≤0.05, **p≤0.01, ***p≤0.001 by one-way or two-way ANOVA. ANOVA, analysis of variance; PE, P-phycoerythrin.
We further evaluated the antitumor efficacy and persistence of both CAR-T cells in vivo. NCG mice were subcutaneously inoculated with HCT-116-hGCC-L cells. A single intravenous injection of 5 million mock T or CAR-T cells was given to each mouse 10 days post tumor inoculation (online supplemental figure S3a). As shown in figure 2i and online supplemental figure S3b,c, YM02 CAR-T cells slowed down the tumor growth compared with mock T cells. In contrast, YM01 CAR-T cells eradicated most of the tumors after treatment for 20 days, and tumor recurrence was not observed within 1 month. Flow cytometry analysis of the peripheral blood showed the sustained expansion of YM01 CAR-T cells from day 4 to day 12 and persistence until day 20 (figure 2j and online supplemental figure S3d,e). In comparison, YM02 CAR-T cells decreased continuously after injection. Moreover, the expansion of total T cells was observed in all groups, and the YM01 CAR-T group showed about three times higher peak level than the other groups (online supplemental figure S3f). The body weight of mice in YM01 CAR-T group steadily increased, and other groups showed a decrease in body weight at the late stage mostly due to the high tumor burden (online supplemental figure S3g). These results demonstrate that YM01 scFv endowed CAR-T cells with high antitumor activity and expansion potential due to the superior antigen sensitivity.
CD28 costimulatory domain contributes to enhanced antitumor function of YM01-based CAR-T cells against low GUCY2C-expressing cells
To investigate whether CD28 or 4-1BB is conducive to the function of YM01-based CAR-T cells, we prepared two CAR-T cells (named 28z and BBz) using YM01 scFv, CD8α-derived hinge and transmembrane domain, CD3ζ endodomain, together with CD28 or 4-1BB co-stimulatory domain (figure 3a). Both CAR-T cells exhibited evident and gradually increased CAR expression ratio during the culture time (figure 3b,c). The cytolytic activity and expansion capacity of 28z and BBz were first evaluated using HCT-116-hGCC-L target cells. RTCA showed that 28z cells inhibited tumor cell growth and eliminated tumor cells faster than BBz cells (figure 3d). Consistently, 28z cells secreted higher levels of cytokines (TNF and IL-2) than BBz cells after incubation with tumor cells (figure 3e). Moreover, under chronic stimulation of HCT-116-hGCC-L cells for four or five times, 28z cells killed more tumor cells (figure 3f and online supplemental figure S4a) and showed superior persistence than BBz cells (figure 3g and online supplemental figure S4b). Both CAR-T cells showed similar differentiation phenotype (figure 3h), and 28z showed higher PD1 expression than BBz (figure 3i) without stimulation. After chronic stimulation, the 28z CAR-T showed 4.0-fold more naïve and stem cell memory T cells together with 2.6-fold increase in central memory T cells compared with BBz CAR-T (figure 3h). Meanwhile, 28z exhibited higher PD1 expression but lower LAG-3 ratio compared with BBz (figure 3i). Next, we investigated the effects of 28z and BBz CAR-T cells on HCT116-hGCC-H tumor cells. The 28z CAR-T showed higher acute cytotoxicity than BBz CAR-T (online supplemental figure S4c). After chronic stimulation for three times, a significantly lower amount of residual tumor cells as well as a higher number of CAR-T cells was observed in 28z CAR-T group (online supplemental figure S4d,e). Together, these results prove that CD28 co-stimulatory domain is advantageous to the durable efficacy of YM01-based CAR-T cells.
The antitumor activity of YM01 CAR-T cells with either CD28 or 4-1BB costimulation in vitro. (a) Schematic representation of retroviral vectors encoding CD8H/TM.28z CAR-T and CD8H/TM.BBz CAR-T. (b) Representative flow cytometry analysis of CD8H/TM.28z CAR and CD8H/TM.BBz CAR transduction efficiency in primary T cells at different time points. (c) Median fluorescence intensity (MFI) was determined by flow cytometry to evaluate the CAR density on the surface of T cells by detection of the Strep Tag II expression. (d) The real-time cytotoxicity assay (RTCA) of CD8H/TM.28z CAR-T and CD8H/TM.BBz CAR-T cells in vitro against HCT-116-hGCC-L at E:T ratio of 1:3 using RTCA (n=3). CAR-T cells (CD8H/TM.28z CAR-T and CD8H/TM.BBz CAR-T) were co-cultured with against HCT-116-hGCC-L at E:T ratio of 1:3, after 24 hours of co-culture, IFN-γ, TNF-α, IL-2 in the supernatant were measured by Cytokine Bead Array (CBA) Kit, respectively (e). (f–g) Residual HCT-116-hGCC-L cells (f) and number of CAR-T cells (g) at the round 4 of co-culture were collected and enumerated by flow cytometry. (h) The differentiation phenotypes of 28z and BBz CAR-T cells before and after chronic stimulation with tumor cells. The T cell subsets were analyzed using flow cytometry labeling of CD62L and CD45RA. (i) The proportions of PD1+ and LAG3+ cells in 28z and BBz CAR-T cells before and after chronic stimulation with tumor cells. Bars represent means±SD, *p≤0.05, **p≤0.01, ***p≤0.001 by one-way ANOVA. E:T, effector to target; RTCA, real-time cytotoxicity assay; ANOVA, analysis of variance.
The length of hinge domain is crucial for the effector function of YM01-CD28-based CAR-T cells and CD8α transmembrane domain improved the durable efficacy
Since the spacer length and transmembrane domain have been reported to regulate the CAR signaling threshold and quantity,16–18 we further investigated the effects of different hinge/transmembrane domains on the antitumor function of YM01-CD28-based CAR-T cells. We first generated four CAR variants (named CD8HCD8TM, CD8HCD28TM, CD28HCD28TM, and IgG4HCD28TM) using CD8α−, CD28−, or IgG4-derived hinge domains, and transmembrane domains originated from CD8α or CD28 (figure 4a and online supplemental figure S5a). Flow cytometry analysis showed similar expression ratio of these CAR variants on the surface of primary T cells (figure 4b and online supplemental figure S5b). However, the expression density of CAR molecules was much lower in IgG4HCD28TM cells compared with CD8HCD8TM, CD8HCD28TM, and CD28HCD28TM cells (figure 4b and online supplemental figure S5c). Intriguingly, IgG4HCD28TM cells lost the effector functions including cytolytic activity and cytokine release, while the other groups showed high tumor eliminating activities (figure 4c,d). Then we evaluated the durable antitumor functions of CD8HCD8TM, CD8HCD28TM, and CD28HCD28TM cells after chronic stimulation by HCT-116-hGCC-L cells. The minimum amount of residual tumor cells was observed in the CD8HCD8TM group (figure 4e), in accordance with the maximum number of viable CAR-T cells (figure 4f), suggesting that the CD8α-derived transmembrane domain was incentive to the long-term potency of YM01-CD28-based CAR-T cells.
YM01 CAR-T containing the CD8 hinge region and CD8 transmembrane domain demonstrates superior antitumor efficacy in vitro. (a) Schematic representation of retroviral vectors encoding YM01 CAR constructs differed in H and TM domains. (b) CAR expression and binding to GUCY2C of four YM01 CAR-T cells as measured by Strep Tag II expression or PE labeled GUCY2CECD antigen on day 12, respectively. (c) The real-time cytotoxicity assay (RTCA) of four GUCY2C CAR-T cells in vitro against HCT-116-hGCC-L at E:T ratio of 1:3 using RTCA (n=3). (d) CAR-T cells were co-cultured with against HCT-116-GCC-L at E:T ratio of 1:3, after 24 hours of co-culture, IFN-γ, TNF-α, IL-2 in the supernatant were measured by Cytokine Bead Array (CBA) Kit, respectively. Residual HCT-116-hGCC-L cells (e) and number of CAR-T cells (f) at the round 4 of co-culture were collected and enumerated by flow cytometry. (g) Schematics of YM01-IgG4 hinge(H)-based CAR with different length of spacers (CH3, 103aa and CH2CH3, 216aa). CD28 transmembrane (TM) domain was used in all YM01-IgG4 hinge-based CARs (h) The Real-time cytotoxicity assay (RTCA) of four GUCY2C CAR-T cells in vitro against HCT-116-hGCC-L at E:T ratio of 1:3 using RTCA (n=3). (i) Measurement of IFN-γ secretion in co-cultured supernatants with HCT-116-hGCC-L cells at E:T ratio of 1:3 after 24 hours (n=3). (j) Residual HCT-116-hGCC-L cells at the round 3 of co-culture were collected and enumerated by flow cytometry. Bars represent means±SD, *p≤0.05, **p≤0.01, ***p≤0.001 by one-way ANOVA. ANOVA, analysis of variance; E:T, effector to target.
The anergy of IgG4HCD28TM cells is probably attributed to the short spacer length that not only disturbed the interaction with target antigen but also affected the CAR expression level. To further validate the impact of spacer length on the CAR-T function, we constructed two additional CAR variants with modified IgG4-Fc spacer domains, named IgG4HCH3CD28TM and IgG4HCH2CH3CD28TM (figure 4g). Elongation of the IgG4 hinge domain improved the interaction of CAR-T cells with GUCY2C protein (online supplemental figure S5d). Meanwhile, restorations of the cytolytic activity and cytokine release were observed in both IgG4HCH3CD28TM and IgG4HCH2CH3CD28TM CAR-T cells against HCT-116-hGCC-L cells, which were comparable to those of CD8HCD8TM cells (figure 4h,i). Furthermore, the tumor killing efficiency of IgG4HCH3CD28TM and IgG4HCH2CH3CD28TM CAR-T cells was significantly elevated compared with IgG4HCD28TM cells after repetitive exposure to HCT-116-hGCC-L cells, and IgG4HCH2CH3CD28TM cells showed the stronger anti-tumor activity (figure 4j). Consistently, the expansion of IgG4HCH3CD28TM and IgG4HCH2CH3CD28TM CAR-T cells was also much higher than that of IgG4HCD28TM cells (online supplemental figure S5e). These results demonstrated that the length of hinge domain is crucial for the effector function of YM01-CD28-based CAR-T cells.
The composition of CD8α-derived hinge and transmembrane domains contributes to the adequate tonic signaling of YM01-CD28-based CAR-T cells
Since CD8HCD8TM CAR-T cells displayed the highest antitumor activity among above YM01-CD28-based CAR-T cells, we further investigated the underlying mechanism. Previous studies have reported that the extent of CAR-T cell tonic signaling was relevant to the self-activation and exhaustion, and therefore, affected the function of CAR-T cells.19 Thus, we evaluated the tonic signaling of YM01-CD28-based CAR-T cells with different hinge and transmembrane domains (CD8HCD8TM, CD8HCD28TM, CD28HCD28TM, and IgG4HCD28TM) in the absence of antigen and Dynabeads stimulation. Flow cytometry analysis of T cell activation markers showed that the expression levels of CD25, 4-1BB, and CD69 in CAR-T cells without antigen activation were much higher than those in mock T cells, indicating that these YM01-CD28-based CAR-T cells were self-activated (figure 5a). Among the CAR-T cells, IgG4HCD28TM showed the lowest level of activation markers, and CD8HCD8TM exhibited the highest activation (figure 5a). Similar expression of PD-1 and a bit higher expression of Lag-3 was observed in CD8HCD8TM CAR-T cells (figure 5b), suggesting that the self-activation of CD8HCD8TM CAR-T was not redundant to cause cell exhaustion. Analysis of the T differentiation subsets showed that CD8HCD8TM CAR-T contained the highest proportion of effector memory T cells and effector T cell subsets (figure 5c), indicating the strongest tonic signaling of CD8HCD8TM CAR-T cells. Then we evaluated the proliferative ability and cytokine release of these CAR-T cells in the absence of interleukin-2. Both mock T cells and IgG4HCD28TM CAR-T cells could not grow and died after culture for 2 days. In contrast, CD8HCD8TM, CD8HCD28TM, and CD28HCD28TM CAR-T cells could survive and proliferate even after culture for 12 days (figure 5d). Consistently, higher levels of IL-2, IFN-γ, and TNF were observed in CD8HCD8TM, CD8HCD28TM, and CD28HCD28TM CAR-T groups than those in the IgG4HCD28TM CAR-T and mock T groups. Moreover, CD8HCD8TM could secret much higher cytokine levels than the other three CAR-T cells. Notably, only CD8HCD8TM produced sufficient amount of IL-2 while the other three CAR-T cells could not (figure 5e), suggesting the expansion and survival potential of CD8HCD8TM in IL-2 free condition. We next explored whether elongation of the hinge domain of IgG4HCD28TM CAR-T cells could improve the tonic signaling and detected the cell activation and proliferation without antigen stimulation. IgG4HCH3CD28TM showed similar expression of CD25 and 4-1BB compared with IgG4HCD28TM, and could not proliferate either. However, the self-activation and proliferation of IgG4HCH2CH3CD28TM were significantly improved, even to a higher level than those of CD8HCD8TM (figure 5f,g). These results support that the composition of CD8α-derived hinge and transmembrane domains benefited YM01-CD28-based CAR-T cells with efficient tonic signaling, and the proper length of the hinge domain was indispensable for the generation of tonic signaling.
The tonic signaling of YM01 CAR-T cells with different hinge (H) and transmembrane domain (TM) during ex vivo expansion. (a) Activation and exhaustion (b) marker expression of four YM01 CAR-T cells was evaluated 8 days post Dynabeads removal. (c) Memory T cell phenotypes of four YM01 CAR-T cells with different hinge and TM. (d) In vitro proliferation of human YM01 CAR-T cells with different hinge and TM following 4 days post Dynabeads removal. No cytokines were added to culture media at any point during expansion. (e) T cells were collected from culture, washed, and replated at 1×106/mL. Cells were kept in culture for 24 hours at which time supernatant from each culture was collected. Cytokine were detected by Cytokine Bead Array (CBA) Kit. (f) Activation maker of CD25 and 4-1BB in IgG4 based hinge CAR-Ts. (g) In vitro proliferation of IgG4 based hinge CAR-T cells following 4 days post Dynabeads removal. No cytokines were added to culture media at any point during expansion. All experiments in this figure were performed once. Bars represent means±SD, ***p≤0.001 by one-way ANOVA. ANOVA, analysis of variance.
The cysteine residues in the CD8α-derived hinge domain are crucial for the tonic signaling and durable function of YM01-CD28-based CAR-T cells
The CD8α-derived hinge domain contained two cysteine residues which probably contribute to the dimerization of CAR molecules by forming disulfide bonds and furthermore enhanced function of CAR-T cells.18 20 To test this hypothesis, we performed cysteine-to-serine mutations in the CD8 hinge domain of CD8HCD8TM CAR-T and generated the CD8H(mut)CD8TM CAR-T (figure 6a). The transduction efficiency of CD8HCD8TM and CD8H(mut)CD8TM CARs were evaluated using flow cytometry after cells were incubated with a YM01-targeted antibody. Similar transduction efficiency ratio was observed in both CD8HCD8TM and CD8H(mut)-CD8TM CAR-T cells. However, the cysteine residue mutation led to a reduction of CAR expression intensity as well as binding affinity to the GUCY2C protein (figure 6b). We found that both the cytolytic activity and the IFN-γ secretion of CD8H(mut)CD8TM CAR-T cells were significantly lower compared with CD8HCD8TM CAR-T cells (figure 6c,d). Chronic stimulation assay also showed that the residual tumor cells increased by threefold and the remaining number of CAR-T cells was lower in CD8H(mut)CD8TM group compared with CD8HCD8TM group after CAR-T cells were exposed to HCT-116-hGCC-L tumor cells twice (figure 6e). In accordance with the impaired antitumor function, the antigen-independent expression of T cell activation markers including CD25 and 4-1BB was downregulated in CD8H(mut)CD8TM cells (figure 6f). Unlike CD8HCD8TM CAR-T cells with continuous expansion in the absence of IL-2, CD8H(mut)CD8TM CAR-T cells failed to proliferate over time (figure 6g). These results proved that the cysteine residues in the CD8α-derived hinge domain play essential roles in the generation of tonic signaling and durable function of YM01-CD28-based CAR-T cells.
The tonic signaling of YM01-CD8HCD8TM.28z CAR-T cells largely relies on the two cysteine of CD8 hinge. (a) Mutations of two cysteine residues in the hinge of CD8HCD8TM CAR. (b) CAR expression and binding to GUCY2C of CAR-Ts as measured by Strep Tag II expression or PE labeled hGUCY2CECD antigen on day 10, respectively. (c) The real-time cytotoxicity assay (RTCA) of CAR-T cells in vitro against HCT-116-hGCC-L at E:T ratio of 1:3 using RTCA (n=3). (d) CAR-T cells were co-cultured with against HCT-116-hGCC-L at E:T ratio of 1:3, after 24 hours of co-culture, IFN-γ in the supernatant were measured by Cytokine Bead Array (CBA) Kit, respectively. (e) Residual HCT-116-hGCC-L cells and number of CAR-T cells at the round 2 of co-culture were collected and enumerated by flow cytometry. (f) Activation marker expression of CAR-T cells was evaluated 8 days post Dynabeads removal. (g) In vitro proliferation of human CAR-T cells with CD8 hinge and CD8 hinge mut following 4 days post Dynabeads removal. No cytokines were added to culture media at any point during expansion. Bars represent means±SD, ***p≤0.001 by one-way ANOVA. ANOVA, analysis of variance; E:T, effector to target; PE, P-phycoerythrin.
The YM01-CD28-based CAR-T cell with CD8α-derived hinge and transmembrane domains exhibits efficient antitumor activity in CRC organoids and in vivo
After confirming that CD8α-derived hinge and transmembrane domains endowed YM01-CD28-based CAR-T cell (named YM01-CD8HCD8TM.28z) with superior antitumor activity in vitro, we further evaluated its performances in CRC organoids and patient-derived xenograft (PDX) animal model. We chose a CRC organoid with low expression of GUCY2C (figure 7a), treated the tumor organoids with CAR-T or mock T cells and evaluated the intensity of caspase-3/7 in tumor cells and the infiltration of CAR-T cells into the organoids over time. We found that CAR-T cells rapidly surrounded the CRC organoids, and efficiently infiltrated after 48 hours (online supplemental figure S6a–c). Compared with mock T group, the CAR-T group showed stronger fluorescence intensity of caspase 3/7 in the CRC organoids after co-culture for 72–120 hours (figure 7b,c). Meanwhile, we established a CRC PDX model in immunodeficient NCG mice using a CRC tissue expressing GUCY2C (figure 7d and online supplemental figure S6d). Tumor-bearing mice were treated with 5 million mock T cells or CAR-T cells. Rapid tumor regression was observed in mice treated with CAR-T cells, and the antitumor effect persisted during the follow-up period (figure 7e). These results demonstrate that the YM01-CD28-based CAR-T cell with CD8α-derived hinge and transmembrane domains can efficiently eradicate primary CRC tumors even with low GUCY2C expression.
YM01-CD8HCD8TM.CD28z CAR T cells exhibit robust effector functions in vitro and in vivo. (a) Flow cytometric analysis of GUCY2C expression in colorectal cancer organoids. (b) Organoids were collected and counted before assessing the CAR-T treatment by brightfield and immunofluorescence imaging. Representative single tiles of brightfield and caspase-3/7 (green) images of the co-culture treated with YM01-CD8HCD8TM.CD28z CAR T at E:T ratio of 1:3. (c) Normalized caspase-3/7 signal of the respective groups were collected. (d) Timeline for in vivo tumor experiments with a patient-derived xenograft (PDX) tumor model. 20 days after tumor implantation, mice were infused intravenously with 5 million CAR-T cells per mice (n=6/group). (e) Tumor size was measured during the time. *p≤0.05, **p≤0.01, ***p≤0.001 by repeated measure two-way ANOVA for multiple comparisons. ANOVA, analysis of variance; E:T, effector to target.
Discussion
One key limitation for applying CAR-T therapy to epithelial-derived tumors such as CRC is the paucity of tumor-specific antigens.21 In this context, targeting tumor-associated antigens risks the development of “on-target, off-tumor” toxicities and therapy-limiting autoimmunity.22 23 Our study chose GUCY2C as an ideal target antigen to develop CAR-T therapy due to its sequestered expression on the apical surfaces of intestinal epithelia and steady overexpression in various stages of CRC.9 24 25 We also proved its expression feature in multiple normal tissues and CRC tissues by IHC assay (online supplemental figure S1). The intestinal-restricted expression especially the featured localization in normal intestinal epithelia makes GUCY2C as a safe antigen for CAR-T therapy. Thus, it is plausible to choose the tumor-associated antigens for successful development of CAR-T therapy based on the variations in expression ratio, density, and localization between normal tissues and tumor lesions. For example, claudin 18.2 is one of such tumor-associated antigens and has been proven to be expressed in several GI cancer types and in the tight junctions of normal gastric epithelium.26 Our previous clinical study demonstrated that claudin 18.2 CAR-T was effective for gastric cancer therapy and only caused 10.8% of reversible grade 3/4 GI adverse events, proving its feasibility for development of CAR-T therapy.27 28
Prevention of tumor escape arising from heterogeneous antigen expression is another challenge for CAR-T cells in solid tumor therapy, emphasizing a need to tune CAR-T cell function against tumors with low antigen density.29 Recent studies have reported a potent GPC2 CAR-T cell tuned for low antigen density.30 In this regard, we established the HCT-116-hGCC-L cell line and the chronic stimulation model for screening of a functional and durable CAR-T and confined the optimal CAR molecule. We found that the chronic stimulation model was superior to the RTCA for screening the functional domains of CAR-T cells, possibly because it mimicked the tumor relapse and could reflect the persisting effects of CAR-T cells (figure 4). Thus, it is important to establish a suitable in vitro evaluation model for screening out desirable CAR-T cell candidate. Besides the low antigen density, the heterogeneity in antigen expression ratio is another problem for CAR-T therapy against solid tumors, as tumor resistance or relapse is usually associated with antigen escape. Although our GUCY2C CAR-T showed durable antitumor effect, it is noteworthy to further design dual-targeted or multitargeted CAR-T cells to overcome the antigen escape.
Our study showed that YM01 CAR-T cells harboring CD28 instead of 4-1BB costimulatory domain possessed stronger antitumor function and persistence, possibly due to the variations in CAR-T phenotype after stimulation (figure 3). Meanwhile, the superior effect of 28z CAR-T was not associated with the tumor antigen density (online supplemental figure S4). Some studies have demonstrated that CD28 costimulation conferred CAR-T cells with brisk and strong effector function, while 4-1BB signaling resulted in slower but more durable T cell responses.31 It seems that our study is contradictory to these reports. However, compared with hematological tumors, many solid tumor types exhibit higher antigen heterogeneity and have lower chances to encounter CAR-T cells due to the complicated microenvironment.32 In such situation, CD28 costimulation is likely to be more effective for CAR-T activation and eliciting the antitumor effects. Moreover, our functional studies of BBz and 28z CAR-T cells were comparable since both CAR-T cells harbored the same other domains and used the identical manufacture process. We believe that the superiority of CD28 over 4-1BB costimulatory domain is regardless of the change with other hinge or transmembrane domains.
Optimization of CAR structure to improve T cell function has been a central focus of the field. Many recent studies have identified that each component of the CAR structure, including regions that do not directly interact with antigen, have an impact on CAR-T cell function.33–36 In this study, we hypothesized that a more flexible spacer domain might improve YM01-based CAR-T cell signaling. Therefore, we constructed CAR-T cells with different spacers including 45 amino acid CD8α hinge, 39 amino acid CD28 hinge or a short 12 amino acid modified IgG4 hinge region. YM01 CAR-T cells with shorter IgG4 hinge were almost unable to recognize and kill GUCY2C-positive tumor cells. However, YM01 CAR-T cells with CD8α hinge or CD28 hinge killed low GUCY2C-expressing HCT-116 cells equally well (figure 4c and e). Interestingly, the tumor killing effect was restored when we elongated the IgG4 hinge (figure 4h and j). As the GUCY2C CAR-T potency is positively correlated with the spacer length, we confirm that a long and flexible hinge domain benefits YM01-based CAR-T cells targeting to the GUCY2C antigen, possibly at the membrane-distal spatial epitopes.
One important finding in our study lies in that a long and flexible hinge domain such as CD8α hinge is conducive to the YM01-based CAR-T cell tonic signaling and thereby durable antitumor functions. T cell receptor (TCR)-induced tonic signaling is well known to benefit the homeostasis, survival, and differentiation of T cells.37 Cell-autonomous tonic signaling has been also identified in CAR-T cells such as CD28-based CD19 CAR-T and GD2-targeted CAR-T.19 38 Unlike the well-controlled TCR tonic signaling, CAR-induced tonic signaling is complicated. Strong CAR tonic signaling can lead to T cell exhaustion and impaired antitumor function, while insufficiency of tonic signaling is unable to elicit desirable efficacy of CAR-T cells.39 40 Therefore, an adequate level of tonic signaling is beneficial for the CAR-T cell function, and tuning the tonic signaling degree through the appropriate CAR design is an important approach to engineering robust CAR-T cell therapies. In fact, Chen et al have also observed this phenomenon and validated the contribution of tonic signaling to the antitumor efficacy of CAR-T cells.41 42 They introduced positively charged patches (PCPs) into the scFv of the CD19 CAR-T for increase in tonic signaling, and improved antitumor activity was observed. They found that extremely high or low tonic signaling strength were both adverse to the CAR-T fitness, which was also validated in our study. Using the CAR-Toner database,42 we found that the PCP score of YM01 was higher than that of YM02, which was consistent with the higher antitumor activity of YM01 CAR-T (figure 2). We also screened out another GUCY2C CAR-T with higher tonic signaling than YM01 CAR-T, named YM03 CAR-T that showed higher T cell activation (online supplemental figure S7a), effector T cells (online supplemental figure S7b), and exhaustion markers (online supplemental figure S7c). The excessive tonic signaling of YM03 CAR-T significantly impaired the antitumor function, resulting in lower cytokine secretion (online supplemental figure S7d) and inefficient killing of tumor cells (online supplemental figure S7e) compared with YM01 CAR-T. Again, our study proved that adequate tonic signaling was important for the persistence and antitumor efficacy of CAR-T cells.
One interesting experimental phenomenon is the discrepancy between the ELISA analysis of antigen-binding affinity (online supplemental figure S2a) and the antitumor effects of YM01 CAR-T and YM02 CAR-T. This is possibly due to the differences in GUCY2C-binding motif between YM01 and YM02, and the limitation of the ELISA assay using the recombinant GUCY2C protein since we observed higher binding capacity of YM01 than YM02 to GUCY2C-expressing cells (online supplemental figure S2b). Consistent with the cellular binding data, YM01 CAR-T secreted relatively higher levels of cytokines after tumor cell stimulation (figure 2e), especially higher IL-2, aiding in the proliferation and survival of CAR-T cells. Interestingly, the superiority of YM01 CAR-T function was amplified in the chronic stimulation assay and the animal study, likely owing to the variations in frequency and duration of antigen stimulation. The great potency of YM01 CAR-T is associated with the higher binding capacity to target cells and higher PCP score compared with YM02 CAR-T (figure 2g–j).
In summary, we demonstrated a potentially leading GUCY2C CAR-T candidate that could rapidly and durably eliminate GUCY2C-expressing tumors, providing the opportunity for further clinical applications in metastatic GI cancers. We also illustrate tonic signaling and basal T-cell activation as informative parameters to guide the rational design of next-generation CAR-T cells for cancer therapy.
Data availability statement
No data are available. All data relevant to the study are included in the article or uploaded as online supplemental information.
Ethics statements
Patient consent for publication
Ethics approval
All experimental procedures were carried out in accordance with Animal Care Ethics and the animal studies were approved by the Peking University Institutional Animal Care and Use Committee (2023KT01).
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
CQ, DL and CL contributed equally.
Contributors Conceptualization, YD and LS; Formal analysis, CQ, DL and CL; Funding acquisition,
CQ, LS; Investigation, CQ, DL, CL; Methodology, MM, MT, CZ and XWei; Resources, XL and TH; Supervision, YD, JL and LS; Writing–original draft, CQ, DL and FD; Writing–review and editing, YD and LS.
Funding This study was supported by Beijing Natural Science Foundation (L232080), "Open Competition to Select the Best Candidates" Key Technology Program for Cell Therapies of NCTIB (NCTIB2023XB01008), National Key R&D Program of China, Stem Cell Research and Organ Repair (2022YFA1106500), Beijing Hospitals Authority Youth Program (QMS20201101), Science Foundation of Peking University Cancer Hospital (JC202303, JC202406), Clinical Medicine Plus X - Young Scholars Project of Peking University, Peking University Clinical Scientist Training Program, the Fundamental Research Funds for the Central Universities, and Beijing Engineering Research Center of Research and Development of New Antitumor Drugs and New Technologies.
Competing interests The YM01-CD28z CAR construct has been patented, and DL, XL, TH and YD are listed among the inventors. There are no conflicts to declare.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.