Article Text
PDF
Abstract
Background Chimeric antigen receptor natural killer (CAR-NK) therapy holds great promise for treating hematologic tumors, but its efficacy in solid tumors is limited owing to the lack of suitable targets and poor infiltration of engineered NK cells. Here, we explore whether immunogenic cell death (ICD) marker ERp57 translocated from endoplasmic reticulum to cell surface after drug treatment could be used as a target for CAR-NK therapy.
Methods To target ERp57, a VHH phage display library was used for screening ERp57-targeted nanobodies (Nbs). A candidate Nb with high binding affinity to both human and mouse ERp57 was used for constructing CAR-NK cells. Various in vitro and in vivo studies were performed to assess the antitumor efficacy of the constructed CAR-NK cells.
Results We demonstrate that the translocation of ERp57 can not only be induced by low-dose oxaliplatin (OXP) treatment but also is spontaneously expressed on the surface of various types of tumor cell lines. Our results show that G6-CAR-NK92 cells can effectively kill various tumor cell lines in vitro on which ERp57 is induced or intrinsically expressed, and also exhibit potent antitumor effects in cancer cell-derived xenograft and patient-derived xenograft mouse models. Additionally, the antitumor activity of G6-CAR-NK92 cells is synergistically enhanced by the low-dose ICD-inducible drug OXP.
Conclusion Collectively, our findings suggest that ERp57 can be leveraged as a new tumor antigen for CAR-NK targeting, and the resultant CAR-NK cells have the potential to be applied as a broad-spectrum immune cell therapy for various cancers by combining with ICD inducer drugs.
- Adoptive cell therapy - ACT
- Chimeric antigen receptor - CAR
- Natural killer - NK
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary 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
During immunogenic cell death, a key hallmark is the translocation of some endoplasmic reticulum proteins to the cell surface. Whether these proteins translocated to the cell surface can be used as therapeutic targets has not been investigated, so it is significant to explore the possibility of these proteins as chimeric antigen receptor natural killer (CAR-NK) targets.
WHAT THIS STUDY ADDS
We found that ERp57 is spontaneously expressed on the surface of various types of tumor cell lines, and low-dose oxaliplatin treatment can also induce its translocation. To target ERp57, a nanobody Nb G6 was identified from a naïve VHH phage library with high binding affinity to both human and mouse ERp57 for constructing CAR-NK cells. ERp57-targeting CAR-NK92 cells can effectively kill various tumor cell lines in vitro and in vivo, and its antitumor activity is enhanced by low-dose immunogenic cell death-inducible drug oxaliplatin.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Our study highlights the potential of ERp57 as a pan-cancer therapeutic target. The specific, high-affinity Nb to ERp57 screened in this study could also be used for other forms of ERp57-targeted therapies, such as CAR-T, CAR-macrophages, and so on. This study provides an alternative for cancer types without a suitable target or where the target disappears during treatment.
Introduction
Immunotherapy strategies such as immune checkpoint blockades and adoptive immune cell transfer (ACT) that aim to fight cancers by enhancing the functions of patients’ immune systems or leveraging the killing capability of infused engineered immune cells against cancer cells, have achieved remarkable progress and become standard therapeutic modalities for treating certain hematological and solid tumors. One of the most successful ACT is chimeric antigen receptor T cells (CAR-T cells) therapy, which has been developed and applied in clinical settings for various cancer types.1 2 Despite its success in the clinical treatment of various cancers, CAR-T therapy also confronts challenges due to several limitations such as graft-versus-host disease, cytokine release syndrome (CRS), immune effector cell-associated neurotoxicity syndrome (ICANS), and high production costs which make it impractical to most patients.3–6
Unlike T-cell therapy, natural killer (NK) cell-based ACT has not been associated with CRS or ICANS, regardless of whether autologous or allogeneic NK cells were used.7 8 CAR engineered in NK cells offers several advantages over T cells, such as unrequired pre-sensitization, the ability to kill tumor cells via CD16-mediated antibody-dependent cell-mediated cytotoxicity, and a shorter lifespan, which renders NK cells much safer.7 Furthermore, the availability of various sources of NK cells, including induced pluripotent stem cells, umbilical cord blood, peripheral blood, and various NK cell lines, greatly facilitates the development and clinical use of CAR-NK.7
Despite its promising results in hematologic tumors, CAR-NK therapy shows limited efficacy for solid tumors. One potential explanation for the suboptimal efficacy of CAR-NK therapy in solid tumors is inadequate infiltration of CAR-NK cells. Moreover, the current CAR-NK therapy targeting tumor-associated antigens that are also expressed in normal tissues to a certain extent, such as GD2, EGFR, and PSCA, results in potential on-target off-tumor toxic side effects.9–11 More importantly, the expression of tumor-associated antigens is highly heterogeneous, not only in different individuals but also in different lesions or tumor cells.12 Hence, identifying a tumor-specific and relatively universal target is of great significance for CAR-NK construction.
Some chemotherapeutic drugs, in addition to exerting direct cytotoxicity, can also induce immunogenic cell death (ICD) in tumors when administered at low doses, resulting in the translocation or release of damage-associated molecular patterns and thereby enhancing the antitumor immunity.13 14 The central event of ICD is the translocation of calreticulin (CALR) from the inner lumen of the endoplasmic reticulum (ER) to the surface of the cell membrane. This process is also accompanied by the co-translocation of protein disulfide isomerase-associated 3 (PDIA3, also known as ERp57).15 The translocation of both proteins occurs simultaneously as a complex, and knocking down either one compromises the translocation of the other protein.16 Therefore, the translocation of these two proteins (CALR and ERp57) to the cell surface may create an opportunity for CAR-NK cells to recognize cancer cells.
Herein, we demonstrate that the translocation of ERp57 and CALR from ER to cell surface could be induced by oxaliplatin (OXP) treatment, and unexpectedly, we also found that many tumor cell lines spontaneously expressed ERp57 on their cell surface. Using a naive nanobody phage library, one of the nanobodies (Nb G6) that can recognize both human and mouse ERp57 was used for CAR-NK construction. Robust antitumor effects were observed with ERp57-targeting CAR-NK cells in both in vitro cell models and in vivo tumor models including cancer cell-derived xenograft (CDX) and patient-derived xenograft (PDX). Expectedly, ICD inducer OXP could synergistically reinforce the antitumor performance of ERp57-targeting CAR-NK owing to the induction of ERp57. Our study demonstrates that ERp57 translocation on the cell surface could serve as a target for CAR-NK cells, and ICD inducers improving the translocation of ERp57 could augment the tumor infiltration and antitumor activity of the CAR-NK cells.
Materials and methods
Antibodies and reagents
Anti-ERp57 antibody (Cat# ab10287, Abcam), Anti-Calreticulin antibody (Cat# ab92516, Abcam), Anti-HMGB1 antibody (Cat# ab18256, Abcam) and Alexa Fluor 647-IgG H&L (Cat# ab150075, Abcam) were purchased from Abcam (Cambridge, Massachusetts, USA). Fetal bovine serum (Cat. # 10270106, Gibco) and horse serum (Cat. # 26050088, Gibco) were purchased from Gibco BRL (Grand Island, New York, USA). MonoRab Rabbit Anti-Camelid VHH Cocktail-iFluor 647 (Cat# A02019, Kingsray) was obtained from Kingsray Biotechnology, China. CellMask Orange Plasma Membrane stain (Cat# C10045, Thermo) was purchased from Thermo Fisher Scientific (Waltham, Massachusetts, USA). Oxaliplatin (Cat# S122412, Selleck) was obtained from Selleck Chemicals (Houston, Texas, USA). HRP-coupled anti-HA antibody (Cat. # 100028-MM10-H, Sino Biological) was purchased from Sino Biological, China. The Bright-Lite Luciferase Assay System (Cat. # DD1204-01, Vazyme Biotech) was from Vazyme Biotech, China. The 3,3’,5,5’-tetramethylbenzidine (TMB) solution (Cat. # PR1200, Solarbio Life Sciences) and Luria-Bertani (LB) broth (Cat. # L8291, Solarbio Life Sciences) were purchased from Solarbio Life Sciences, China. Liver Dissociation Kit Mouse (Cat. # 130-105-807, Miltenyi Biotec), Lung Dissociation Kit Mouse (Cat. # 130-095-927, Miltenyi Biotec), Dissociation of the adult mouse heart using the Multi Tissue Dissociation Kit 2 (Cat. # 130-110-203, Miltenyi Biotec), Dissociation of mouse kidney using the Multi Tissue Dissociation Kit 2 (Cat. # 130-110-203, Miltenyi Biotec) were purchased from Miltenyi Biotec GmbH (Bergisch Gladbach, Germany). Inositol (Cat. # I7508-50G, Sigma-Aldrich) and folic acid (Cat. # F8758-5G, Sigma-Aldrich) were purchased from Sigma-Aldrich (St. Louis, Missouri, USA). Lenti-X Concentrator (Cat. # 631231, Clontech) was from Clontech (Tokyo, Japan). For immunofluorescence assays, the concentrations of primary were used at 1:100 dilution, and secondary antibodies were used at 1:1,000 dilution, unless otherwise specified.
Cell lines and cell culture
A549, HCT116, 22RV1 and ASPC1 cell lines were obtained from the American Type Culture Collection (Manassas, Virginia, USA). EC109, MKN45, IM95, NK92, MOLM13-luc and 293T/17 were purchased from Procell Life Science & Technology (Wuhan, China). A549-luc and HCT116-luc cells were constructed and maintained by our own laboratory. MKN45, HCT116, EC109, 22RV1, ASPC1 and MOLM13-luc cells were cultured in RPMI 1640 supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine and 1% penicillin/streptomycin. 293T/17, A549 and IM95 were maintained in high-glucose Dulbecco’s Modified Eagle’s Medium supplemented with 10% FBS, 2 mM L-glutamine, and 1% penicillin/streptomycin. NK92 cells were cultured in alpha minimum essential medium supplemented with 0.2 mM inositol, 0.1 mM 2-mercaptoethanol, 0.02 mM folic acid and 100 U/mL recombinant interleukin-2, adjusting to a final concentration of 12.5% horse serum and 12.5% FBS. Human peripheral blood mononuclear cells (PBMCs, Cat. # PB025c) from healthy donors were obtained from Milestone Biotechnologies (Shanghai, China). All the cell lines were maintained at 37°C in a humidified 5% CO2 incubator.
ERp57 gene profile from public patient with cancer databases
Using the GEPIA2 (http://gepia.cancer-pku.cn/) website, a comparative analysis of ERp57 gene expression was performed between cancerous and non-tumor tissues. GEPIA2 is a website that provides data visualization tools for analyzing RNA sequencing expression data obtained from The Cancer Genome Atlas (TCGA) an Genotype-Tissue Expression (GTEx) projects. The data screening conditions are: Normal data match TCGA normal and GTEx data; data type choose messenger RNA; differential methods choose Linear Models for Microarray Analysis (LIMMA) method. |Log2FC| cut-off and q value cut-off choose 1 and 0.01, respectively.
Nanobody library screening against ERp57
A VHH phage display library (Shenzhen KT Health Life Technology) comprising~2×109 independent clones was used for screening ERp57-targeted Nbs. Briefly, panning was performed in immunotubes coated with the ERp57 protein. After three rounds of panning, 192 individual clones were selected for phage ELISA using horseradish peroxidase (HRP)-coupled anti-M13 secondary antibody. The TMB solution was added to react with the HRP conjugates, and the absorbance was measured at 450 nm within 30 min after adding the stop solution. After two rounds of phage ELISA confirmation, clones that were positive in both rounds were sequenced.
Expression and purification of nanobodies
Nb coding sequences were cloned into pColdII vector with N-terminal His and C-terminal HA tag, and then transformed into BL21 (DE3) cells. Bacteria were cultured in LB broth to reach an OD600 of 0.6, followed by IPTG (0.5 mM) induction overnight at 15°C. Bacteria were then harvested and high pressure crushed in lysis buffer (50 mM NaH2PO4, 300 mM NaCl, 10 mM imidazole with protease inhibitor). After lysis, centrifuge lysate at 17,000×g for 30 min at 4°C to pellet the cellular debris, followed by Ni-NTA purification and gel filtration with Superdex 75 Increase 10/300 GL (GE Healthcare Life Sciences, Cat# 29148721). Protein was concentrated using an Amico Ultra-15 centrifugal filter (Millipore, Cat# UFC901024) with a 10 kDa cut-off.
ELISA
To confirm Nb candidates, ERp57 protein or the control protein bovine serum albumin (BSA) in phosphate-buffered saline (PBS) were immobilized on ELISA plates at a concentration of 10 µg/mL overnight at 4°C. Next day, 3% BSA was added to the plate to block the non-specific binding sites. Serial dilutions of HA-tagged Nbs in 0.1% phosphate-buffered saline with Tween 20 (PBST) were incubated with the immobilized antigen, followed by incubation with HRP-coupled anti-HA antibody. Simultaneously, an irrelevant nanobody control (Nb C9) was used as a negative control. Following three times washing, TMB solution was added to react with the HRP conjugates for 5 min with gentle shaking. Once the color became deep blue, the reaction was stopped with a stop solution. The absorbance was measured at 450 nm within 30 min after adding the stop solution. Curve fitting was performed using nonlinear regression by GraphPad Prism V.8.
Surface plasmon resonance
Surface plasmon resonance (SPR) experiments were performed using a Biacore S200 instrument (GE Healthcare) to investigate the binding kinetics of Nbs to human and mouse ERp57. The SPR experiments were performed at 25°C in 10 mM HEPES pH 7.4, 150 mM NaCl, 3 mM EDTA and 0.05% Tween-20. The human and mouse ERp57 proteins were immobilized on a Series S-CM5 sensor chip (Cytiva) by amine coupling (NHS/EDC) to flow cells. A mock coupled surface was used for background subtraction. twofold serial dilutions of Nbs from 3.9 nM to 62.5 nM were passed over the CM5 Series S sensor chip surface at 30 µL/min for 120 s followed by 180 s of dissociation flow. After each cycle, the surface was regenerated with 10 mM glycine, pH 3 for 120 s. Binding data were fitted by the Langmuir binding equation for a 1:1 interaction model using Biacore S200 Evaluation Software.
Immunofluorescence staining
For surface detection of ERp57 and CALR, the cells were washed twice with PBS and fixed with 0.25% paraformaldehyde in PBS for 5 min, followed by incubation with 4% donkey serum solution for 1 hour at room temperature. Then, the primary antibody diluted in cold-blocking buffer was incubated for 1 hour at room temperature. For intracellular HMGB1 staining, cells were fixed with 4% paraformaldehyde for 20 min and permeabilized with 0.1% Triton X-100 for 10 min. All staining was examined and photographed with a Leica SP8 laser scanning confocal microscope.
Plasmid construction and lentivirus production
The second-generation CARs with CD8 signal peptide, VHH Nb, CD8 hinge, CD28 transmembrane domain, CD28 co-stimulatory domain and CD3ζ domain were synthesized and inserted into pCD513B-CopGFP lentiviral vector by GENEWIZ (Jiangsu, China). The bicistronic vector was linked to the transduction marker CopGFP by the “self-cleaving” T2A peptide. To produce lentivirus, HEK293T/17 cells were co-transfected with pMD2.G, psPAX2, and the plasmids of interest (G6-CAR, C9-CAR, pLVX-luciferase-IRES-ZsGreen). The viral supernatants were collected from 48-hour and 72-hour culturing media and concentrated by Lenti-X Concentrator. The detailed concentration steps were carried out according to the manufacturer’s instructions.
Generation of CAR-NK cells
NK92 cells were transduced with G6-CAR or C9-CAR lentivirus in the presence of 4 µg/mL polybrene. After 48 hours, the cell culture medium was supplemented with the selected antibiotic puromycin (2.5 µg/mL) and selected with puromycin for 10 days. To confirm the expression of the CAR on the cell surface of NK92 cells, cells were stained with anti-VHH-AF647 antibody and analyzed by flow cytometry. CAR-expressing NK92 cells were sorted by FACS using SH800ZBP sorter. After sorting, the collected positive cells were rinsed with a prewarmed medium and then cultured for subsequent use.
Flow cytometry analysis
CAR-modified NK92 cells were stained with anti-VHH-AF647 antibody in FACS staining buffer (PBS with 2% FBS) on ice for 30 min. Then, cells were washed with PBS and analyzed on a DxFLEX flow cytometer (Beckman Coulter, Miami, Florida, USA). For detection of CAR-NK92 cells infiltration in the bone marrow of acute myeloid leukemia (AML) mice. The mice were sacrificed by cervical dislocation and the femurs were surgically removed. Then all the bone marrow cells were washed out with PBS, and small debris and muscle tissue were removed with a 70 µm nylon cell strainers. Single-cell suspension preparation for organs (heart, liver, spleen, lung and kidney) was performed according to the Miltenyi Biotec manufacturer’s instructions. The filtrate was spun down at 300 g for 5 min, the supernatant was discarded and 2 mL ammonium chloride erythrocyte lysate (1×) was added and incubated at room temperature for 1 min. 10 mL PBS was added to neutralize the lysate, then 300 g for 5 min to discard the supernatant. Then the cells were resuspended in FACS staining buffer (PBS with 2% FBS) and analyzed for CopGFP signal using a flow cytometer. For cell surface ERp57 detection, organ single-cell suspension was stained with anti-ERp57 antibody on ice for 30 min. Then, cells were washed with PBS and stained with anti-rabbit Alexa Fluor 647-IgG H&L on ice for 30 min. After washed with PBS, samples were analyzed on a DxFLEX flow cytometer. Flow data were then analyzed using FlowJo V.10 (TreeStar Ashland, Oregon, USA).
In vitro cytotoxicity assays
To evaluate the cytotoxic activity of CAR-NK cells, a luciferase-based cytotoxicity assay was used. Briefly, 2×105 target cells (A549-luc, HCT116-luc and MOLM13-luc) were first treated with or without OXP for 4 hours, and then co-incubated with CAR-NK cells for another 4 hours at five different effector:target (E:T) ratio. Luciferase activity was calculated to detect relative light units using the Bright-Lite Luciferase Assay System according to the manufacturer’s protocol. Percentages of cytotoxicity were normalized with wells containing cancer cells only as 0% cytotoxicity and with parental cancer cells untransfected with luciferase construct as 100% cytotoxicity. The lactate dehydrogenase release assay was assessed following the manufacturer’s instructions. The percentage of cytotoxicity was calculated using a standard formula: 100 × (CAR-NK treatment release − spontaneous release)/(positive lysate release − spontaneous release) %.
Immunohistochemistry
To analyze and evaluate the expression level of ERp57 in PDX tumor tissues and clinical samples, immunohistochemistry of ERp57 was performed. The PDX tissue was derived from a patient with stage III C gastric adenocarcinoma and was kindly gifted by Dr Yuanqiao He from Nanchang University. Informed consent was obtained from the patient, and the procedures involving human samples were approved by the medical ethical committee of the Shenzhen People’s Hospital and Nanchang University. The colorectal carcinoma, gastric carcinoma and rectal carcinoma tissue sections treated with or without OXP chemotherapy were obtained from a biobank (Outdo Biotech, Shanghai, China). The sections were deparaffinized and antigen retrieval was performed in Tris-EDTA buffer (pH 9.0) at 100°C for 15 min. After blocking (5% BSA and 0.05% Tween-20 in PBS) for 1 hour at room temperature, and the sections were incubated with primary mouse anti-ERp57 antibody overnight at 4°C. ERp57 detection was performed by biotin-conjugated goat anti-rabbit IgG secondary antibody and ABC kit followed by colorimetric detection using diaminobenzidine. The images were obtained using a 3DHISTECH scanner (Sysmex, UK). Staining scoring of samples was automated using Image J’s IHC Profiler plugin.
Animal studies
All animal procedures were approved by the Institutional Animal Care and Use Committee of Shenzhen People’s Hospital (AUP-220101-LZJ-0600–01) and were performed in accordance with the Guide for the Care and Use of Laboratory Animals. Six weeks old male NCG mice (Prkdc-Il2rg-) were purchased from GemPharmatech (Guangdong, China) and were maintained at standard room temperature (~25°C) with a 12 hours light-dark cycle and humidity ranging 40–60% in the animal facility at the Shenzhen People’s Hospital. For the MOLM-13-luc AML model, the mice were randomly divided into three groups (four mice per group). NCG mice were injected with MOLM13-luc luciferase-expressing cells (2×105) via the tail vein. On day 3, 1×107 G6-CAR-NK92 or C9-CAR-NK92 cells were given once every 4 days through the tail vein, for a total of four injections. For all the in vivo experiments, CAR-NK92 cells were irradiated with 10 Gy using a γ-ray irradiator (Schering, USA).17 18 IVIS Spectrum imaging system (PerkinElmer, USA) was performed to measure the in vivo tumor burden every 4 days. To establish the HCT116 xenograft model, NCG mice were injected subcutaneously with 2×106 HCT116 cells in 100 µL PBS. When the tumor volume reached around 50 mm3, mice were randomly divided into six groups (five mice per group). On day 6, OXP was dissolved in PBS and administered via intraperitoneal (i.p.) injection at a dose of 2.5 mg/kg, which was injected every 2 days, for a total of nine injections. On day 8, G6-CAR-NK92 cells or C9-CAR-NK92 cells (1×107 cells) were given once every 4 days through the tail vein, for a total of four injections. For the PDX model, the PDX tissue was rapidly cut into 3×3×3 mm fragments on ice, and subsequently implanted subcutaneously in the right forelimbs of NCG mice. When tumor volume reached around 50 mm3, mice were randomly divided into three groups (five to six mice per group). OXP was administered via i.p. injection at a low dose of 2.5 mg/kg, which was injected every 2 days, for a total of nine injections. On day 29, G6-CAR-NK92 cells or C9-CAR-NK92 (1×107 cells) were given once every 4 days through the tail vein, for a total of four injections. Animal weight and tumor volume were collected every 3 days. The tumor size (mm3)=length×width2/2. When the tumor burden reached 2000 mm3 or the animal’s weight reduced ≥15%, mice were euthanized with carbon dioxide narcosis. The survival data for the animals were recorded at the same time.
Statistical analysis
All the data were analyzed with GraphPad Prism V.8 (GraphPad Software, San Diego, California, USA). The data were expressed as mean±SE (SEM). Differences among groups were evaluated by two-way analysis of variance with Tukey’s multiple comparison test or unpaired t-test analysis as indicated. Differences were considered significant at *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001.
Results
ERp57 translocates to the cell surface in multiple cancer cell lines and can be enhanced by OXP treatment
Some chemotherapeutic agents such as OXP, actinomycin D and mitoxantrone can induce ICD in tumor cells.14 The process accompanies with co-translocation of CALR and ERp57 complex from ER to cell surface.19 To verify whether ERp57 can translocate to the cell surface during the ICD process, we first performed ICD induction on A549 and HCT116 cells treated with OXP, followed by immunofluorescence staining to observe ERp57 translocation. As expected, ERp57 was detected on the cell membrane surface of A549 after ICD induction, while it was not detected in the untreated control cells (figure 1A). Surprisingly, ERp57 was also detected on the cell membrane surface of HCT116 cells without OXP treatment, while ERp57 translocation was intensified after ICD induction (figure 1B, online supplemental figure S1). Furthermore, we measured the total ERp57 expression levels in cells before and after ICD induction. The results indicated that OXP treatment did not significantly affect the total ERp57 expression levels in either HCT116 or A549 cells (online supplemental figure S2), implying that cell surface expression of ERp57 mainly results from the translocation from ER to the cell membrane. Consistent with previous studies,13 14 CALR translocation to the cell surface was also detected in HCT116 cells after ICD induction, but not in the untreated control cells (figure 1C). HMGB1 release, another important marker for ICD was also detected in the cytoplasm after ICD induction compared with untreated control cells of which HMGB1 was mainly stained in the nuclei (figure 1D). To determine whether ERp57 can also translocate to the cell surface on the other types of cancers, esophageal carcinoma (ESCA-EC109), stomach adenocarcinoma (STAD-IM95 and MKN45), prostate adenocarcinoma (PRAD-22RV1) and pancreatic adenocarcinoma (PAAD-ASPC1) cell lines were tested. Interestingly, all these cells intrinsically expressed ERp57 on the cell surface without ICD induction, indicating that ERp57 might exert other functions warranting further investigation (figure 1E). Furthermore, to demonstrate the safety of ERp57 targeting, we performed in vivo ICD induction according to previous study,20 followed by cell surface detection of ERp57 with an antibody recognizing both human and mouse ERp57 proteins. Flow cytometry assays showed that the cell surface of ERp57 was not detected in all vital organs (heart, liver, spleen, lung and kidney) with or without ICD induction (figure 1F–G), indicating that ERp57 is a safe target. Finally, we also analyzed the expression level of ERp57 between normal tissues (N) and tumor samples (T) across multiple cancer types by the GEPIA2 website. The results showed significant upregulation of ERp57 expression among multiple types of tumor tissues (online supplemental figure S3).
Supplemental material
The determination of ERp57 exposure on the cell surface after ICD induction. (A, B) ERp57 exposure on the cell surface of A549 cells and HCT116 cells was assessed by immunofluorescence staining after OXP treatment (ICD) or without treatment (mock). Cellmask (a plasma membrane dye) was used to indicate the cell membrane. (C) CALR exposure on the cell surface of HCT116 cells was examined after OXP treatment (ICD) or without treatment (mock). (D) Immunofluorescence staining of HMGB1 release from the nuclei to the cytoplasm in HCT116 cells. After the indicated treatments, cells were permeabilized and stained with an anti-HMGB1 antibody. Nuclei were stained with DAPI. (E) Immunofluorescence staining of ERp57 exposure on the cell surface of ESCA (EC109), STAD (IM95 and MKN45), PRAD (22RV1) and PAAD (ASPC1) cell lines. (F) Flow cytometry analysis of ERp57 translocation on the cell surface of organs isolated from OXP-treated mice. Mice were treated with OXP via intraperitoneal injection at a low dose of 2.5 mg/kg every 2 days for a total of four injections. On day 8, the mice were anesthetized and the organs (heart, liver, spleen, lung and kidney) were collected for ERp57 analysis. (G) Heart, liver, spleen, lung and kidney were prepared as single-cell suspensions, blocked with anti-CD16/32 for 20 min and incubated with the anti-ERp57 antibody on ice for 30 min. Samples were analyzed by flow cytometry to detect ERp57 surface expression and dead cells were removed with propidium iodide. CALR, calreticulin; ESCA, esophageal carcinoma; ICD, immunogenic cell death; OXP, oxaliplatin; PAAD, pancreatic adenocarcinoma; PBS, phosphate-buffered saline; PRAD, prostate adenocarcinoma; STAD, stomach adenocarcinoma.
Identification and characterization of ERp57-binding nanobodies
To identify ligand targeting ERp57 to develop CAR NK therapy, in vitro phage display screening of a naïve alpaca VHH library was first performed to isolate high-affinity binding Nbs to ERp57 (figure 2A–C). The purified recombinant human and mouse ERp57 were examined by SDS-PAGE with an approximate molecular weight of 58 kDa and were confirmed by western blotting with an anti-His tag antibody (figure 2B). After three rounds of panning, seven Nbs that can potentially bind to the human ERp57 protein were identified through phage ELISA (online supplemental figure S4a,b). All seven Nbs were cloned and purified with an approximate molecular weight of 15 kDa (figure 2D). Then the Nbs were verified with the antibody of anti-HA tag which was integrated at the C terminals of all the Nbs (figure 2D). Further ELISA assay revealed that four Nbs (NbA11, B9, C10 and G6) obtained with phage ELISA showed positive binding activity to the human ERp57 protein (figure 2E). The CDR3 sequences of these Nbs are presented in online supplemental table S1. The binding kinetics of the four selected Nbs to human and mouse ERp57 protein were determined by SPR, which demonstrated that NbA11, B9 and G6 showed binding affinity to human ERp57 protein with KD values of 418.8 nM, 989.9 nM and 1.59 nM, respectively, whereas Nb C10 indicated no evident binding activity (figure 2F). Subsequently, these four Nbs were further assessed the binding ability to mouse ERp57 owing to the high homology between human and mouse ERp57. Interestingly, Nb G6 showed high binding affinity to mouse ERp57 protein with a KD value of 2.85 nM whereas the other three Nbs did not bind to mouse ERp57 protein (online supplemental figure S4c). In addition, Nb G6 was applied for the immunofluorescence staining to detect the ERp57 in HCT116 cells with/without ICD induction; the results indicated that Nb G6 could recognize the ERp57 translocated from ER to cell membrane, and more importantly, it was able to block the binding of the commercial ERp57 antibody to ERp57 (online supplemental figure S5). Given the best binding affinity of Nb G6 to human and mouse ERp57, it was further applied for subsequent CAR-NK construction and examination.
Screening and characterization of ERp57-targeting Nbs. (A) The molecular structure of ERp57. ERp57 has 505 amino acids and consists of 4 domains, a-b-b’-a’, together with an N-terminal signal sequence and QDEL C-terminal endoplasmic reticulum retention/retrieval motif. Both a and a’ domains hold a thioredoxinlike active site and each contains a redox-active CGHC catalytic sequence. b and b’ domains contain binding sites for calreticulin and calnexin. (B) The purified recombinant proteins of human and mouse ERp57 proteins were examined by SDS-PAGE, and confirmed by immunoblotting anti-His tag antibody. (C) Schematic diagram of ERp57 nanobody phage display library screening. (D) The seven purified nanobodies were identified from phage display screening and phage ELISA. They were examined by SDS-PAGE and stained with Coomassie Blue after purification. Their expression was verified with an anti-HA tag antibody. (E) ELISA assessment for ERp57 nanobodies. (F) Multicycle kinetics analysis for Nbs binding to human ERp57 was evaluated by surface plasmon resonance. The flow cell temperature was 25°C. ERp57 protein was immobilized to flow cells on a Series S Sensor Chip CM5 by amine coupling (NHS/EDC). A mock coupled surface was used for background subtraction during analysis. Nb solutions with two fold serial dilutions from 3.9 nM to 62.5 nM were injected on the sensor chip surface at 30 µL/min for 120 s, followed by 180 s of dissociation flow. After each cycle, the surfaces were regenerated with 10 mM glycine (pH 3.0) for 120 s. C9 is an irrelevant nanobody control. The equilibrium dissociation constant (KD), the association constant (Ka), and the dissociation constant (Kd) are shown in the figure. Nbs, nanobodies; RU, response units.
Generation and in vitro assessment of CAR-NK cells
To target cell surface ERp57-positive cells, ERp57-targeting CAR-NK cells were constructed using the ERp57-targeting nanobody (Nb G6). An irrelevant nanobody (Nb C9) was used to construct negative control CAR NK cells (C9-CAR-NK92). The CAR backbone consisted of a CD8 hinge, a transmembrane region (CD28), followed by the intracellular domains of co-stimulatory CD28, and the intracellular domain of CD3ζ activation domain. The CARs were cloned into the upstream of a T2A sequence and CopGFP in a lentiviral vector (figure 3A–B). After transduction, C9-CAR-NK92 cells and G6-CAR-NK92 cells were selected in vitro for 10 days in a medium containing puromycin (2.5 µg/mL) to obtain stable C9-CAR and G6-CAR molecule expression. After sorting by flow cytometry with VHH antibody, the percentage of C9-CAR and G6-CAR positive NK92 cells were 94.3% and 95.2%, respectively (figure 3C). The high level of CAR expression was also confirmed by immunofluorescence analysis with VHH antibody (figure 3D). Meanwhile, the ERp57 level on the cell surface of G6-CAR-NK92 cells was analyzed by immunofluorescence staining, and no obvious fluorescence signal was detected, indicating no potential fratricide effect among G6-CAR-NK92 cells (online supplemental figure S6).
Generation and characterization of ERp57 targeted CAR-NK92 cells. (A) Schematic diagram of CAR-NK92 with an anti-ERp57 nanobody (Nb G6) or an irrelevant nanobody (Nb C9), a transmembrane region (CD28), followed by the intracellular domains of co-stimulatory CD28, and the intracellular domain of CD3ζ. (B) Structure schematic of the CAR-NK92 cells based on G6 or C9 nanobody. (C) Flow cytometry analysis of CAR expression on the surface of C9-CAR-NK92 and G6-CAR- NK92. Data showed the representative results from three independent experiments. (D) Immunofluorescence staining for CAR expression on the cell surface in C9-CAR-NK92 and G6-CAR-NK92. (E) Luciferase-based cytotoxicity assay of G6-CAR-NK92 cells or control C9-CAR-NK92 cells against HCT116-luc cells at five different E:T ratios (n=3). HCT116-luc cells were treated with or without 150 µM OXP for 4 hours and then co-incubated with CAR-NK cells for another 4 hours. (F) Cytotoxicity assay in ERp57-negative A549-luc cells with G6-CAR-NK92 cells or control C9-CAR-NK92 cells. A549-luc cells were treated with or without 150 µM OXP for 4 hours and then co-incubated with CAR-NK cells for another 4 hours. Data are presented as mean±SEM. Two-way analysis of variance test was employed. (G–I) ELISA was performed to measure the levels of granzyme B (G) IFN-γ (H) and TNF-α (I) released after co-culturing C9-CAR-NK92 cells or G6-CAR-NK92 cells with HCT116-luc cells (n=3). Data are presented as mean±SEM. Unpaired Student’s t-test was employed. ns, no significance; **p<0.01; ***p<0.001; ****p<0.0001. CAR, chimeric antigen receptor; E:T, effector:target; ICD, immunogenic cell death; IFN, interferon; NK, natural killer; OXP, oxaliplatin; TNF-α, tumor necrosis factor-alpha.
We next evaluated the capacity of ERp57-targeted CAR-NK92 cells to eradicate ERp57 positive (HCT116-luc) cell lines treated with or without OXP at five different E:T (effector-to-target) ratios. In the ERp57-positive HCT116-luc cells, G6-CAR-NK92 cells could robustly inhibit HCT116-luc cells at both E:T cell ratios of 5:1 and 10:1, regardless of OXP treatment (figure 3E). More importantly, the killing ability of G6-CAR-NK92 cells was significantly enhanced in the OXP-treated group (ICD) compared with the MOCK group at E:T cell ratio of both 5:1 and 10:1 while control C9-CAR-NK92 showed no significant killing activity, even under the OXP-treated condition (figure 3E). It is noteworthy that OXP treatment alone did not significantly kill A549-luc cells (online supplemental figure S7a) and HCT116-luc cells (online supplemental figure S7b). In addition, both CAR-NK92 cells did not show a significant killing effect on ERp57-negative A549 cancer cells, with no statistical difference between the groups of C9 and G6 CAR-NK cells. Whereas, OXP treatment could enhance the sensitivity of G6-CAR-NK92 cells to A549 cells due to the ERp57 exposure on ICD induction (figure 3F). This suggests that G6-CAR-NK92 cells have a high specificity for targeting ERp57 (figure 3F). Furthermore, we genetically expressed ERp57 on the surface of A549 cells (online supplemental figure S8). The results demonstrated that A549-ERp57 cells were more susceptible to killing by G6-CAR-NK92 cells but did not show increased sensitivity to C9-CAR-NK92 cell-mediated killing (online supplemental figure S8c-d). Moreover, we found that the G6-CAR-NK92 cells released more granzyme B, interferon-γ, and tumor necrosis factor-alpha (TNF-α) than C9-CAR-NK92 cells when co-cultured with HCT116-luc cells (figure 3G–I). We have also demonstrated the function of CAR in primary NK cells and primary T cells. The results showed that both primary CAR-NK cells and primary CAR-T cells targeting ERp57 could effectively eliminate ERp57-positive HCT116 cells (online supplemental figure S9a-d). Additionally, we also evaluated whether primary CAR-NK cells and primary CAR-T cells expressed ERp57 on their cell surface at rest or activation by co-incubation with ERp57-positive HCT116 cells. The results showed that ERp57 expression was not detected on the cell surface of both primary CAR-NK cells and primary CAR-T cells at rest or activation (online supplemental figure S9e-f). Collectively, these results demonstrate that ERp57-targeting CAR-NK cells were successfully established and exhibited prominent in vitro antitumor performance.
G6-CAR-NK92 cells show potent anti-acute myeloid leukemia activity in vivo
CAR-NK cell therapy has demonstrated remarkable antitumor effects in hematological malignancies.21 Therefore, we aimed to first investigate the antitumor effect of CAR-NK cells in a mouse model with hematological malignancy expressing ERp57. We found a substantial expression of ERp57 on the cell surface of the MOLM13-luc cells which are a human AML cell line (figure 4A). We next evaluated the capacity of ERp57-targeting CAR-NK92 cells to eradicate MOLM13-luc cells at three different E:T ratios. Compared with C9-CAR-NK92 cells, G6-CAR-NK92 cells significantly inhibited MOLM13-luc cells at E:T cell ratios of 5:1 and 10:1 (figure 4B); Meanwhile, a clear killing process could be observed in the supplemented video (online supplemental video 1). Therefore, we evaluated the antitumor effect of G6-CAR-NK92 cells in an AML model induced by MOLM13-luc cells. NCG mice were intravenously injected with 2×105 MOLM13-luc cells on day 0; on day 3, the mice were injected with PBS, C9-CAR-NK92 cells or G6-CAR-NK92 cells (1×107 per mouse in 100 µL PBS) via the lateral tail vein every 4 days (figure 4C). The MOLM13-luc tumors was monitored by bioluminescent image following D-luciferin substrate administration. The results showed that both CAR-NK92 cells produced a significant reduction in the tumor burden (efflux signal) on day 15 compared with the PBS groups, whereas G6-CAR-NK92 cells yielded more potent anti-AML activity compared with the control C9-CAR-NK92 cells due to the G6 Nb targeting against ERp57 (figure 4D–E). This was consistent with previous reports, which showed that NK92 cells exhibit moderate cytotoxicity against various leukemia cell lines22 23 and primary NK cells also exhibited the ability to kill MOLM13 cells.24 Unsurprisingly, G6-CAR-NK92 cells treatment led to a significant increase in overall survival (OS) in comparison to PBS or C9-CAR-NK92 cells (figure 4F). The infiltration analysis of CAR-NK92 cells in the bone marrow further disclosed that the number of cells expressing Nb (detected by anti-VHH antibody) was greater in G6-CAR-NK92 cell-treated group than in C9-CAR-NK92 cell-treated group (figure 4G–H). In addition, we also performed quantification of MOLM13-luc cells in the bone marrow from mouse femurs. The results revealed that G6-CAR-NK92 cells significantly reduced the number of MOLM13-luc cells in the bone marrow compared with the C9-CAR-NK92 cells group and the PBS group (figure 4I). This finding is consistent with the in vivo imaging results (figure 4D–E). However, we recognize the limitations in measuring the NK cell population in the bone marrow using CAR molecule staining alone. The use of additional markers in combination with CAR staining would enhance the accuracy in identifying NK cell population. Besides, the observed increase in CAR-NK cell percentage may result from a reduction in MOLM13 cells rather than an actual rise in CAR-NK cells. Further experimental validation is required to confirm this. Together, these results demonstrate that ERp57-targeting CAR-NK92 cells can efficiently inhibit the progression of hematological malignancies with high ERp57 expression on the cell surface.
CAR-NK92 cells regress the progression of acute myeloid leukemia in a xenograft mouse model. (A) Immunofluorescence evaluation of ERp57 expression on the cell surface of MOLM13-luc cells. (B) Luciferase-based cytotoxicity assay of G6-CAR-NK92 cells or C9-CAR-NK92 cells against MOLM13-luc cells at three different effector:target ratios (n=3). (C) Schematic schedule of treatments with CAR NK cells. NCG mice were injected with luciferase-expressing MOLM13-luc cells (2×105) via the tail vein. On day 3, 1×107 G6-CAR-NK92 or C9-CAR-NK92 cells were given once every 4 days through the tail vein, for a total of four injections. The control group without CAR-NK cell treatment was given an equivalent volume of PBS (Mock). IVIS imaging was performed every 4 days. (D) MOLM13 leukemia progression was monitored by serial bioluminescent (BL) imaging using an IVIS Lumina imaging system after administration of D-luciferin substrate (300 mg/kg, intraperitoneal). AML burden was assessed by quantification of BL radiance obtained as photon/s/cm2/sr in the target zone encompassing the entire body of each mouse. (E) Quantification of flux (photons/s). Data represent mean±SEM. *p<0.05 by two-way analysis of variance (n=4). (F) Kaplan-Meier survival curves with different treatments. The survival of AML mice was significantly improved with G6-CAR-NK92 treatment. (G) Therapy regimen schedule diagram for assessing the infiltration of CAR-NK92 cells in the bone marrow of AML mice. NCG mice were injected with MOLM13-luc luciferase-expressing cells (2×105) via the tail vein. On day 3, 1×107 G6-CAR-NK92 or C9-CAR-NK92 cells were given once every 4 days through the tail vein, for a total of four injections. The control group without CAR-NK cells treatment was given an equivalent volume of PBS (Mock). (H) Percentage of CAR-NK92 cells in the bone marrow from mouse femurs receiving G6-CAR-NK92 or C9-CAR-NK92 cells treatments by flow cytometry. (I) Luminescent signal quantification of MOLM13-luc cells in the bone marrow from mouse femurs. The bone marrow cells from mouse femurs were flushed out with PBS, followed by centrifugation and resuspension in 2 mL PBS. After incubation with luciferase substrate for 5 min, the luminescent signal was detected using the Tecan Spark 10M. Data represent mean±SEM. Differences were considered significant at *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. AML, acute myeloid leukemia; CAR, chimeric antigen receptor; NK, natural killer; PBS, phosphate-buffered saline.
ICD induction enhances the antitumor effect of ERp57-targeting CAR-NK cells in a solid tumor model
Inspired by the antitumor activity of ERp57-targeting CAR NK cells in the hematological malignant model, we next assessed whether G6-CAR-NK92 cells could also exhibit excellent tumor inhibitory effect in a solid xenograft model and whether ICD induction could potentiate the efficacy of G6-CAR-NK92 cells. NCG mice were subcutaneously injected with HCT116 cells (day 0). On day 6, OXP was administered via i.p. injection at a low dose of 2.5 mg/kg every 2 days for a total of nine injections. On day 8, CAR-NK92 cells (1×107 cells) were given via tail vein every 5 days for a total of four injections (figure 5A). The low-dose OXP alone did not obviously reduce tumor size when compared with PBS control (figure 5B, online supplemental figure s10). However, G6-CAR-NK92 cells monotherapy significantly repressed tumor burden in mice when compared with untargeted C9-CAR-NK92, PBS or OXP alone, whereas untargeted C9-CAR-NK92 alone did not produce notable tumor inhibition, indicating that ERp57 expressing in HCT116 cancer cells is a key factor for the function of G6-CAR-NK92 cells (figure 5B). The addition of ICD inducer OXP seemed to increase the antitumor effects for both of G6-CAR-NK92 cells and C9-CAR-NK92 cells, but more potent antitumor efficacy was observed in G6-CAR-NK92 plus ICD group (figure 5B). This suggests that OXP treatment may trigger other antitumor mechanisms in CAR-NK92 cells, possibly promoting CAR-NK92 cells infiltration or triggering cytotoxic effects through NKp46 recognition in NK92 cells against OXP-induced CALR translocated in tumor cells.25 When compared with G6-CAR-NK92 monotherapy, ICD plus G6-CAR-NK92 combination showed reduced tumor growth (figure 5B). These data illustrate that ERp57 targeting G6-CAR-NK92 cells can suppress the growth of solid tumors which express intrinsically ERp57 on the cell surface; meanwhile, ICD induction appears to enhance the antitumor efficacy of ERp57-targeting CAR-NK92 cells due to surface ERp57 induction during ICD process. No significant toxicity was observed after CAR-NK92 cells treatment based on body weight loss and histological analysis for important organs (figure 5c, online supplemental figure S11), indicating that ERp57 is a safe target. The growth curves of tumor burden from each group are shown in figure 5D–I. Finally, the infiltration of CAR-NK92 cells in the tumor was analyzed with an antibody against CopGFP which was stably expressed in engineered CAR-NK cells. As shown in figure 5J–K, the number of cells expressing CopGFP was elevated in tumor tissues receiving ERp57-targeting G6-CAR-NK92 cells regardless of ICD induction, and importantly, OXP treatment significantly increased the infiltration of NK cells in G6-CAR-NK92-treated group. Taken together, these results indicate that ERp57-targeting G6-CAR-NK92 cells can efficiently suppress the progression of solid tumors which inherently express ERp57 on the cancer cell surface, and ICD inducers can potentially augment the antitumor effect of G6-CAR-NK92 cells possibly through the elevation of surface ERp57 expression and enhancement of infiltration of CAR-NK92 cells in the tumors.
ERp57-targeting G6-CAR-NK92 treatment suppresses the progression of HCT116 tumor in vivo and ICD inducer OXP synergistically augments its antitumor function. (A) Diagram of treatment schedule in HCT116 xenograft model. NCG mice were subcutaneously injected with 2×106 HCT116 cells (day 0). On day 6, OXP was administered via intraperitoneal injection at a dose of 2.5 mg/kg, which was given every 2 days for a total of nine injections. On day 8, G6-CAR-NK92 cells or C9-CAR-NK92 (1×107 cells) were injected once every 4 days through the tail vein, for a total of four injections (n=5 tumors per group). (B) Tumor growth curves with various treatments. Data are presented as mean±SEM. The difference for each group was analyzed by two-way analysis of variance analysis (n=5 tumors per group). (C) Body weight of each group at the indicated time points. (D–I) Individual tumor growth curves from each group (n=5 tumors per group). (J) Infiltration of G6-CAR-NK92 and C9-CAR-NK92 cells in the tumors 3 days after the last CAR-NK92 treatment in figure 5A. CopGFP positive cells were determined by immunofluorescence staining with an anti-CopGFP antibody. (K) Quantification of CopGFP-positive area ratio relative to C9-CAR-NK92 group in the tumors. Data are presented as mean±SEM. ns, no significance; *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. CAR, chimeric antigen receptor; ICD, immunogenic cell death; NK, natural killer; OXP, oxaliplatin; PBS, phosphate-buffered saline.
ERp57-targeting CAR-NK cells significantly repress tumor progression in a PDX model
CDX models have been extensively employed to assess the antitumor effects of various therapeutics, but their limitations are that they cannot accurately mimic biological processes and the heterogeneity of human tumors during tumor development. Therefore, PDX models are paying considerable attention to tumor studies including drug assessment due to the high similarity to human tumors in components and heterogeneity.26 Given that ERp57-targeting G6-CAR-NK92 therapy showed excellent antitumor activity in the CDX model, we next investigated whether G6-CAR-NK92 therapy could obtain similar results in a PDX model. A PDX model from gastric cancer tissue (fourth passage) was used to evaluate the therapeutic effects. The level of ERp57 expression on tumor tissues from this PDX model after OXP treatment was first determined. When PDX tumor burden reached~50 mm3, OXP was injected intraperitoneally at a dose of 2.5 mg/kg every 2 days for three injections. ERp57 protein expression in STAD PDX samples treated with or without OXP was then assessed by immunohistochemistry staining. The results showed that the OXP treatment samples showed strong positive staining for ERp57 expression compared with PBS control samples (figure 6A). Since a low dose of OXP showed a similar effect on tumor growth to PBS vehicle (figure 5B, online supplemental figure S10), we directly assess the antitumor performance of CAR NK cells under the condition of ICD induction in order to obtain the most prominent antitumor effect and prove the concept. In vivo antitumor effects of CAR-NK92 in the PDX model were then conducted following the experimental schedule shown in figure 6B. NCG mice were subcutaneously implanted with STAD PDX samples (day 0). On day 26, mice were randomly grouped when tumor volume reached~50 mm3. OXP was injected intraperitoneally at a dose of 2.5 mg/kg every 2 days for nine injections. On day 29, CAR-NK92 cells (1×107 cells) were given once every 4 days through the tail vein for a total of four injections. The results showed that ERp57-targeting G6-CAR-NK92 cells under the condition of ICD induction significantly retarded tumor growth and prolonged mouse survival compared with the C9-CAR-NK92 cells plus OXP group or OXP alone group; although C9-CAR-NK92 cells plus OXP treatment also inhibited the tumor progression, mice survival was not significantly prolonged due to the rapid development after treatment cessation (figure 6C,D). In addition, no obvious weight loss was observed for all the mice during the whole treatment process (figure 6E). The tumor growth curves in each group are shown in figure 6F–H. To determine whether these findings are clinically relevant, we further examined whether ERp57 is upregulated in clinical samples with/without OXP chemotherapy. Different types of tumor tissues (colorectal carcinoma n=3, gastric carcinoma n=3 and rectal carcinoma n=3) treated with or without OXP were stained for ERp57 expression. The results showed that the OXP-treated samples displayed strong positive staining for ERp57 compared with untreated clinical samples (figure 6I–L), suggesting that OXP treatment could boost the antitumor effect of ERp57-targeting CAR-NK cells. Collectively, these results illustrate that ERp57-targeting CAR-NK cells have great potential to be applied to patients with cancer whose tumor has surface expression of ERp57 induced by clinically approved ICD inducers, which warrants further investigation for potential clinical translation.
ERp57-targeting G6-CAR-NK92 cells synergize with OXP to control gastric cancer progression in a PDX mouse model. (A) Examination of ERp57 expression after OXP treatment in a PDX model. NCG mice were implanted subcutaneously with gastric cancer PDX samples. When the tumor burden reached~50 mm3, OXP was injected intraperitoneally at a dose of 2.5 mg/kg every 2 days for a total of nine injections. The expression of ERp57 protein in gastric cancer PDX samples treated with or without OXP was then assessed by IHC. The staining scoring of samples was automatically analyzed by Image J’s IHC Profiler plugin. (B) Diagram of experimental schedule in the gastric cancer PDX model. NCG mice were subcutaneously implanted with gastric cancer PDX samples (day 0). When tumor volume reached~50 mm3, mice were randomly grouped. OXP was administered via intraperitoneal injection at a dose of 2.5 mg/kg, every 2 days for a total of nine injections. Three days later, G6-CAR-NK92 cells and C9-CAR-NK92 cells (1×107 cells) were given once every 4 days through the tail vein, for a total of four injections. (C) Tumor growth curves in PDX mice. All the data are presented as mean±SEM. The differences for each group were analyzed by two-way analysis of variance. (D) Kaplan-Meier survival curves of tumor-bearing mice after treatment with CAR-NK92 cells or OXP. The p value was analyzed by log-rank (Mantel-Cox) test. (E) The body weight of each group was measured at the indicated time points. (F–H) Individual tumor growth curves from each group with different treatments (n=5 or 6 tumors per group). (I–K) The expression of ERp57 protein in colorectal carcinoma (I) gastric carcinoma (J) and rectal carcinoma (K) clinical samples treated with or without OXP chemotherapy (n=3). (L) Staining scoring of samples was automatically recorded by Image J’s IHC Profiler plugin. Data are presented as mean±SEM. ns, no significance; **p<0.01; ***p<0.001; ****p<0.0001. Scale bars, 20 µm. CAR, chimeric antigen receptor; ICD, immunogenic cell death; IHC, immunohistochemistry; NK, natural killer; OXP, oxaliplatin; PBS, phosphate-buffered saline; PDX, patient-derived xenograft.
Discussion
Some chemotherapeutic drugs have been demonstrated for the induction of ICD in tumors, including Adriamycin, anthracyclines, epirubicin, mitoxantrone, and OXP. These drugs are widely used in clinical practice for treatments against various malignancies.27 Conventional antitumor drugs are administered at high doses to obtain maximal killing effects on tumor cells, which also results in cytotoxicity to normal cells, including immune cells. Thus, an increasing number of studies have begun exploring the effectiveness of these drugs at lower doses in order to reduce the side effects and provoke possible antitumor immunity. Voorwerk et al treated 67 patients with metastatic triple-negative breast cancer (TNBC) using low-dose ICD inducer Adriamycin and the non-ICD-inducer cisplatin (CDDP) in combination with programmed cell death protein-1 (PD-1) antibodies. They observed a clinical response rate of 35% for Adriamycin and 23% for CDDP when compared with 20% for the overall response rate; favorable tumor microenvironment induced by Adriamycin may be the main mediator to enhance the anti-TNBC efficacy of PD-1 blockade, reminiscent of ICD occurrence caused by Adriamycin.28 Consistently with previous studies, we found that the low dose of OXP treatment can trigger the translocation of CALR and ERp57 from the ER to the cell surface. Interestingly, we also found that a variety type of tumor cell lines overexpresses ERp57 without ICD treatment on the cell surface. This finding is consistent with previous studies that exposure to CALR was detected on the plasma membrane of CD33 malignant blasts in patients with AML regardless of chemotherapy,29 30 and CALR exposure is accompanied by ERp57 exposure.16 The underlying mechanism may be associated with the fact that cancer cells are exposed to high levels of ER stress regardless of treatment, which leads to spontaneous exposure of CALR in the majority of cancer cells in patients.30 In fact, overexpression and translocation of GRP78, the master unfolded protein response regulator and ER chaperone, to the cell surface is induced in response to elevated ER stress in a wide range of solid tumors and hematological malignancies, but not in normal tissues.31 We also found higher levels of ERp57 transcripts in many tumors compared with normal tissues, and translocation of ERp57 to the cell surface was detected in many cancer cells, suggesting that overexpression and translocation of ERp57 might be also associated with elevated ER stress in cancer cells; however, further investigations on this are warranted. In terms of normal tissues, elevated ER stress can trigger apoptosis, which results in the rapid elimination of these cells.32 The highly selective expression and translocation of GRP78 to the cell surface in response to elevated ER stress in a variety of tumors make it an ideal target for CAR T-cell therapy, which has demonstrated robust anti-AML activity.33 This suggests that translocated proteins CALR and ERp57 possibly induced by ER stress could be promising therapeutic targets. Therefore, CAR-NK cells targeting CALR and ERp57 with high-affinity Nbs may be a potential therapeutic strategy for pan-cancer therapy; ICD inducer drugs could boost the therapeutic efficacy of strategies targeting translocated CALR and ERp57 through first priming the tumor cells for ICD induction, as we demonstrated in the current study.
Currently, single chain variable fragment (scFv) is widely used for CAR construction due to its small size, high affinity, and good specificity.34 However, scFv also shows some unavoidable drawbacks to CAR construction. First, the deletion of the constant region may cause auto-aggregation with high hydrophobicity and loss of targeting.35 CAR aggregation may also lead to excessive cytotoxic signaling independent of tumor antigens, ultimately resulting in premature T-cell anergy.36 37 Moreover, for the construction of bispecific CARs, the cross-pairing could be potentially induced between two independent scFv molecules which might cause the loss of binding affinity to either target.38 Therefore, other antibody formats are warranted for CAR-NK or CAR-T construction.39 Nbs are unique antibody fragments engineered from the heavy-chain antibody discovered in camelids and Chondrichthyes that consist of only two heavy chains. They represent the smallest functional antibody units identified to date, with a relative molecular mass of approximately 15 kDa.40 This distinct structure allows Nbs to retain the benefits of conventional antibodies, while also possessing additional functions. First, Nbs have a longer CDR3 sequence compared with conventional antibodies, which provides flexibility in conformational expansion and enables them to reach epitopes that are typically inaccessible to conventional antibodies.41 42 Second, the Nb gene sequence exhibits high homology with the human VH gene family III sequence, resulting in low immunogenicity and strong compatibility with humans.43 Finally, Nbs rarely show auto-aggregation due to high hydrophobicity, which is a common phenomenon to scFv.44 Therefore, Nbs were chosen for CAR construction in this study. We screened for ERp57 targeting Nbs using a VHH phage display library comprising~2 × 109 independent clones and found that Nb G6 had a high binding affinity to both human and mouse ERp57, with KD values of 1.59 nM and 2.85 nM, respectively. This affinity is suitable for CAR-NK construction since the affinity of NK cell receptors to ligands typically ranges from KD values of 1.4 to 51 nM.45–47
A crucial issue of CAR-NK cell therapy is to obtain stable sources for the construction and maintenance of CAR-NK cells that are of high purity and capable of rapid amplification. However, the clinical-scale enrichment of primary NK cells is difficult and limited to specialized laboratories. Additionally, genetic modification of NK cells is still challenging, and there is significant variability in NK cell yield and cytotoxic activity among donors. Given that our main objective is to evaluate the potential of ERp57 as a novel target for cell therapy, we selected the readily accessible and easily manipulable NK92 cell line for CAR cell construction. Additionally, NK92 cells could provide more consistent results in terms of yield and cytotoxic activity compared with primary NK cells. Preclinical studies have also shown that CD123-CAR-NK92 cell lines demonstrate superior cytotoxic activity compared with CD123-CAR-NK cells derived from primary human donors.48 Several clinical trials have disclosed the appreciable antitumor activity of NK92/CAR cells against different hematological and solid tumors.49 Our results showed that G6-CAR-NK92 cells had a robust ability to lyse ERp57-positive HCT116-luc cells at both E:T ratios of 5:1 and 10:1, and that their killing ability was significantly improved by OXP treatment. The mechanism behind this may be due to the translocation of ERp57 induced by OXP. Previous studies have shown that CAR-T cells have a threshold tumor antigen density below which cancer cells are unable to be killed,50 and higher-avidity binding affinity may increase T-cell responses to tumors with low antigen density.51 Therefore, further studies are warranted to improve the translocation levels of ERp57 and determine the antigen density threshold for better ERp57-targeting CAR-NK cells performance.
Identification of specific therapeutic targets for AML has been a challenge because of the heterogeneity of AML and the overlap in antigen expression between healthy tissues such as hematopoietic progenitor cells (HPCs) and AML blasts.52 For example, CD33, CLL1, and CD123 highly expressed in AML blasts are also present on normal HPCs or mature neutrophils, which makes it challenging to selectively target AML cells without affecting healthy cells.52 Our study discovered that the AML cell line MOLM13-luc spontaneously expresses ERp57 on its cell surface. This finding is in agreement with previous research that similar levels of CALR exposure in untreated patients or AML cell lines (Kasumi-1 and MOLM-13) compared with idarubicin or daunorubicin treatment.30 Further investigation is needed to elucidate the underlying mechanism for ERp57 translocation. Our study has demonstrated that treatment with ERp57-targeting G6-CAR-NK92 cells can significantly increase OS compared with both PBS and C9-CAR-NK92 cells in MOLM-13 AML mouse model, suggesting a promising new option for AML treatment.
In the current study, we also evaluated the effect of G6-CAR-NK92 cells in a solid tumor model induced by HCT116 colorectal cancer cells. We observed that the therapeutic effects of G6-CAR-NK92 cells can be improved by OXP treatment. The increased killing ability of G6-CAR-NK92 cells by OXP may involve multiple mechanisms. Selective upregulation of ERp57 translocation by OXP and enhanced infiltration of CAR-NK92 cells in tumors may be the main reasons. Consistent with previous studies, OXP treatment improved the infiltration of CAR-T cells into tumor sites.20 To make our findings more clinically relevant, we also assessed the antitumor effects of G6-CAR-NK92 cells on a PDX model. Consistent with the HCT116 model, G6-CAR-NK92 cells significantly reduced PDX tumor burden and prolonged mouse survival compared with the C9-CAR-NK92 cells and OXP alone groups, suggesting that ERp57 is a good candidate for CAR-NK or potential target after ICD induction. More importantly, we further demonstrated that clinical tumor samples with OXP treatment contained more surface expression of ERp57 than those without OXP treatment, clearly illustrating that ICD inducers combined with agents targeting ICD-translocated markers might be a promising strategy for pan-cancer therapy.
Taken together, G6-CAR-NK92 cells that specifically target ERp57, which is spontaneously expressed on the surface of multiple tumor cell lines and can be further induced by ICD, showed excellent antitumor activity against tumors with inherent or induced ERp57 expression. The tumor-killing capacity of G6-CAR-NK92 cells can be significantly improved by low-dose OXP treatment both in vitro and in vivo. The encouraging results from our study indicate that ERp57-targeting G6-CAR-NK92 cells have great potential as a novel off-the-shelf therapy for treating multiple types of cancer. However, variations in the exposure levels for ERp57 on the cell surface of different tumor types may potentially limit the recognition capabilities and therapeutic efficacy of CAR-NK cells. Besides ICD induction, other strategies to improve the surface expression of ERp57 may also expedite to enhance the efficacy of ERp57-targeting CAR NK therapy. Therefore, additional preclinical and clinical investigations are required to evaluate their feasibility and safety in clinical settings.
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
Ethics statements
Patient consent for publication
Ethics approval
Not applicable.
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
LZ, HW and JZ contributed equally.
Contributors LZ, HW, and JZ contributed equally to this work, performed the study and wrote the manuscript. GS and JM conducted protein purification and surface plasmon resonance analysis. YJ, ZD, JL, Y-QH, and DW were responsible for data collection. JCS, CCX, ZJL and JGW as the guarantors of this study contributed to conception and design of the study. All authors read and approved the final manuscript.
Funding This research was funded by the National Natural Science Foundation of China (32101219, 82373775), Shenzhen Science and Technology Innovation Committee (RCBS20210706092213007), Shenzhen Peoples’s Hospital (SYWGSCGZH202304), the Guangdong Zhujiang Program (0920220233), the Scientific and Technological Innovation Project of China Academy of Chinese Medical Sciences (CI2023D003, CI2021B014, CI2023D008), the Innovation Team and Talents Cultivation Program of National Administration of Traditional Chinese Medicine (No: ZYYCXTD-C-202002), the CACMS Innovation Fund (CI2023E002, CI2021A05101 and CI2021A05104), the Science and Technology Foundation of Shenzhen (JCYJ20210324115800001; JCYJ20210324113608023; JCYJ20200109120205924), the National Key Research and Development Program of China (2022YFC2303600 and 2020YFA0908000), the Science and Technology Foundation of Shenzhen (Shenzhen Clinical Medical Research Center for Geriatric Diseases), the Shenzhen Medical Research Fund (B2302051), the Distinguished Expert Project of Sichuan Province Tianfu Scholar (CW202002), the Shenzhen Science and Technology Innovation Committee (SZSTI) (RCYX20221008092950121), the National Natural Science Foundation of China (82373775; 82070517; 82170842; 8230130506), the HUILAN Public Welfare, the Fundamental Research Funds for the Central public welfare research institutes (ZZ16-ND-10-23, ZZ15-ND-10, ZZ14-ND-010, ZZ14-FL-002, ZZ14-YQ-050, ZZ14-YQ-051), the Shenzhen Science and Technology Innovation Commission (JCYJ20200109120205924), GuangDong Basic and Applied Basic Research Foundation (2021A1515012164; 2022A1515110745; 2021A1515111188), International Science and Technology Cooperation for Shenzhen Technology Innovation Plan (GJHZ20200731095411034).
Competing interests No, there are no competing interests.
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.