Schematic representation of cloning vectors pDRIVE T–MAGE-A3 (A) and pDRIVE T–CPD–MAGE-A3 (B). Primers were designed for MAGE-A3 and CPD–MAGE-A3 without start or stop codons and with 5′ SphI and 3′ BglII ends. C, Polymerase chain reaction was performed with 30 amplification cycles (92°C for 30 seconds, 58°C for 30 seconds, 72°C for 1 minute), and products were purified from agarose gel. This experiment was repeated 3 times, with a representative gel shown. bp indicates base pairs.
Schematic representation of constructed expression vectors pQE-70–MAGE-A3 (A) and pQE-70–CPD–MAGE-A3 (B). bp indicates base pairs. C and D, We transformed BL21 Star (DE3-pLysS) cells with expression vectors. Three clones were selected, induced with 1mM isopropyl-d-thiogalactosidase at 37°C, and analyzed on 4% to 20% sodium dodecyl sulfate–polyacrylamide gel electrophoresis. This experiment was repeated 3 times, with a representative gel shown. The amount of total protein loaded was approximately 5 μg. The sizes of the recombinant proteins were approximately 35 kDa for MAGE-A3 (C) and 38 kDa for CPD–MAGE-A3 (D).
A, Culture of BL21 harboring CPD–MAGE-A3 was conducted in various bacterial growth media: (1) Superior Broth, (2) Turbo Broth, (3) 2XYT Broth, (4) Power Broth, (5) Hyper Broth, and (6) Luria-Bertani Broth. B, Protein expression analyzed 6 hours after IPTG induction at indicated concentrations. Both experiments were repeated 3 times, with representative gels shown.
Control, induced, and purified recombinant proteins of both MAGE-A3 and CPD–MAGE-A3 were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (A) and subjected to Western blot analysis with MAGE-A3 (B) and CPD (C) antibodies as depicted. All experiments were repeated 3 times, with representative gel (A) and Western blots (B and C) shown.
A, Dendritic cells were pulsed for indicated times and then stained with MAGE-A3 antibody followed by visualization with fluorescence microscopy at lower magnification (original magnification ×10). B and C, Higher-magnification images show the presence of CPD–MAGE-A3 in dendritic cell cytosol (original magnification ×40 [B] and ×100 [C]). All experiments were performed in triplicate, with representative fluorescent images shown.
Dendritic cells were pulsed with 3 µmol of CPD–MAGE-A3 protein, followed by intracellular staining with indicated antibody and analysis by flow cytometry. The axes show fluorescent intensity. This experiment was repeated 3 times, with representative flow diagrams shown. EGFP indicates enhanced green fluorescent protein; FITC, fluorescein isothiocyanate; and PE, phycoerythrin.
Batchu RB, Gruzdyn O, Potti RB, Weaver DW, Gruber SA. MAGE-A3 With Cell-Penetrating Domain as an Efficient Therapeutic Cancer Vaccine. JAMA Surg. 2014;149(5):451-457. doi:10.1001/jamasurg.2013.4113
Copyright 2014 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
In conjunction with chemotherapy, immunotherapy with dendritic cells (DCs) may eliminate minimal disease burden by generating cytotoxic T lymphocytes. Enhanced cytosolic bioavailability of tumor-specific antigens improves access to human leukocyte antigen (HLA) class I molecules for more efficient cytotoxic T lymphocyte generation. Various cell-penetrating domains (CPDs) are known to ferry covalently linked heterologous antigens to the intracellular compartment by traversing the plasma membrane.
To determine whether generating melanoma antigen family A, 3 (MAGE-A3), a tumor-specific cancer-testis antigen, as a fusion protein with CPD will enhance the cytosolic bioavailability of MAGE-A3.
MAGE-A3 was amplified by polymerase chain reaction using complementary DNA from renal tissue and cloned in frame with a CPD (YARKARRQARR) at the amino-terminal end and hexahistidine at the carboxy-terminal end to generate CPD–MAGE-A3 in a pQE-70 expression vector. Cultures were grown in Escherichia coli BL21 Star (DE3-pLysS) cells followed by nickel–nitrilotriacetic acid affinity purification of recombinant proteins.
Main Outcomes and Measures
Measurement of DC membrane penetration of CPD–MAGE-A3 vs MAGE-A3 and determination of the effect of CPD–MAGE-A3 pulsing on DC phenotypic expression of cell-surface antigens.
Media composition and isopropyl-d-thiogalactosidase induction were optimized to achieve high levels of protein expression followed by purification. Western blot analysis with MAGE-A3 antibodies recognized both MAGE-A3 and CPD–MAGE-A3 proteins, while CPD antibodies recognized only CPD–MAGE-A3. Purified CPD–MAGE-A3 exhibited more efficient DC membrane penetration than did MAGE-A3 alone as confirmed by immunofluorescence analysis. High-level expression of several unique DC markers (CD80, CD83, CD86, and HLA-DR) by flow cytometry was consistent with a mature DC phenotype, indicating that pulsing with CPD–MAGE-A3 did not alter specific cell-surface antigens required for T-cell activation.
Conclusions and Relevance
We have demonstrated for the first time, to our knowledge, that cloning and purification of MAGE-A3 with CPD enhances its cytosolic bioavailability in DCs without altering cell-surface antigens, potentially making it a more potent therapeutic cancer vaccine compared with existing MAGE-A3 protein and peptide vaccines.
The immune system has the ability to detect and eliminate tumor cells through the generation of cytotoxic T lymphocytes (CTLs), a process known as immune surveillance.1 However, cancer cells may evade this process, and immunotherapy with dendritic cells (DCs) is a relatively nontoxic alternative that may overcome elements of host immune incompetence.
Dendritic cells are antigen (Ag)–presenting cells that endocytose exogenous Ags and play a central role in both initiating and modulating the immune response leading to stimulation of naive T cells.2 Although tumor cells are potential immunogens because they express tumor-specific Ags (TSAs), they are generally not capable of initiating therapeutically useful immune responses owing to their immune inhibitor properties.3 In contrast, DCs can capture TSAs and process them into peptides that bind to human leukocyte antigen (HLA) class I and II molecules. The DCs then migrate to lymph nodes where they interact with both CD8+ and CD4+ T lymphocytes to generate CTLs and helper T lymphocytes, respectively, to shape the adaptive immune response, making them ideally suited for cancer immunotherapy.2
First-generation vaccines based on exogenous pulsing of DCs with tumor lysates, RNA, or peptides have all demonstrated limited success.4 This is because the exogenously pulsed TSA enters the cytoplasm to access the HLA class I pathway by the incompletely understood and inefficient process of cross-priming unique to DCs, where endocytosed TSAs are leaked into the cytosol.5 The process of cross-priming requires high TSA concentrations, providing the rationale for development of better methods for generating CTLs, and accounts for the relative ineffectiveness of prior immunotherapy attempts.6
A nonimmunogenic, 11–amino acid motif on the HIV-1 transactivator of transcription protein, known as the protein transduction domain or cell-penetrating domain (CPD) (YGRKKRRQRRR), that can ferry large, covalently linked heterologous proteins in and out of cells is responsible for this phenomenon.7 Various synthetic CPDs are capable of delivering Ags through the plasma membrane into the cellular compartment.8
Cancer-testis Ags are a family of TSAs whose expression is restricted to immune-privileged gonadal germ cells, thereby making them ideal targets for tumor immunotherapy. Melanoma antigen family A, 3 (MAGE-A3) is a cancer-testis Ag that has attracted particular attention as a candidate for cancer immunotherapy because it is expressed by a wide variety of human cancer types, with several HLA class I- and II-restricted epitopes. These characteristics suggest that MAGE-A3–based vaccines would be immunogenic and that the resulting immune responses are unlikely to target healthy tissues.
Clones of MAGE-A3 CTLs have been generated and shown to lyse cancer cell lines.9 In one large clinical study, MAGE-A3 protein–based tumor vaccine administered in the adjuvant setting prevented disease relapse of non–small cell lung cancer.10 Unlike peptide vaccines, recombinant protein vaccines have the potential to induce a broad array of immune responses because they possess a number of HLA class I and II peptides.4
We sought to address the problem of inadequate cytoplasmic TSA expression by using synthetic CPDs to create fusion proteins that penetrate through the plasma membrane. We hypothesize that MAGE-A3 Ags generated as fusion proteins in-frame with CPDs will gain access to the HLA class I pathway in the cytosol, eventually leading to robust CTL responses. Adequate CTL responses may increase tumor cell killing and achieve durable tumor regression.
Protein expression vectors for MAGE-A3 and CPD–MAGE-A3 were constructed as follows. We performed polymerase chain reaction to amplify MAGE-A3 from kidney tissue complementary DNA and both subcloned it into a pDRIVE T vector (Qiagen) and further cloned it in-frame with YARKARRQARR sequence in a CPD vector. Both MAGE-A3 and CPD–MAGE-A3 were excised and cloned into a pQE-70 protein expression vector (Qiagen) in-frame with the downstream hexahistidine sequence to generate pQE-70–MAGE-A3 and pQE-70–CPD–MAGE-A3.
BL21 Star (DE3-pLysS)–competent cells (Invitrogen) designed for protein expression were transformed with pQE-70–CPD–MAGE-A3 and pQE-70–MAGE-A3 and cultivated in 250 mL of 1 of 5 different commercial media obtained from the pEX Protein Expression Media Optimization Kit (US Biological): (1) Turbo Broth, (2) Superior Broth, (3) Power Broth, (4) Hyper Broth, and (5) Luria-Bertani Broth (Miller), each containing 100-μg/mL ampicillin. In addition, 2XYT Broth (AMRESCO) was used. Induction of protein expression was performed when the optical density reached 1.0 at 600 nm by adding increasing concentrations of isopropyl-d-thiogalactosidase (IPTG) from 0.1mM to 2.0mM. Cultures were further incubated at 30°C and 37°C for various times ranging from 5 to 12 hours. Bacterial cells were centrifuged at 5000g for 10 minutes, and the pellet was suspended in 5 volumes of Bacterial PE (G-Biosciences). The suspension was then vortexed for 1 minute and incubated on ice for 5 minutes. Five microliters of lysozyme (G-Biosciences) containing DNase and RNase was added, and the suspension was incubated at 37°C for 30 minutes followed by centrifugation at 20 000g, 4°C, for 30 minutes.
Cell lysates were passed through a 0.45-μm filter, equilibrated with 50mM sodium phosphate (pH 7.0), 0.5M sodium chloride, 10mM imidazole, and 1% Triton X-100, and left at 4°C for 1 hour with occasional mixing. The resin was set in a small hexahistidine-tag purification column by nickel–nitrilotriacetic acid (Qiagen) (0.5 × 2.5 cm) and washed with 50mM sodium phosphate (pH 7.0) and 40mM imidazole. The adsorbed recombinant proteins were eluted with 0.5M sodium chloride containing 150mM imidazole buffer (pH 7.0) and collected in 0.5-mL fractions.
Total protein was resolved on 4% to 12% polyacrylamide gel and transferred onto a polyvinylidene fluoride membrane by Western blotting. The detection of recombinant proteins using our custom-made CPD sequence (YARKARRQARR) (Biocarta) was performed at 1:10 000 dilution. We used MAGE-A3 monoclonal antibody to identify both MAGE-A3 and CPD–MAGE-A3 protein bands by chemiluminescence assay.
Peripheral blood mononuclear cells were collected from healthy individuals after approval from the Wayne State University Institutional Review Board and written informed consent were obtained. Monocyte-derived DCs were generated as previously described.11 Cytospin slides were incubated with MAGE-A3 mouse antihuman antibody at room temperature for 1 hour followed by 2 washes with phosphate-buffered saline (pH 7.2) containing 0.05% octylphenoxypolyethoxyethanol (IGEPAL CA-630; Sigma). The cells were further incubated with goat antimouse Alexa 488 (Molecular Probes), washed, fixed, and analyzed using a confocal microscope.
Cells were collected during the logarithmic phase, digested with resuspended medium, and added to a 6-well plate (105 cells/well). We added CPD–MAGE-A3 or MAGE-A3 at a concentration of 1nM. The cells were incubated at 37°C for 24 hours according to our previous protocol with minor modification.11 The cells were then stained with fluorescein isothiocyanate to detect surface antigens, including HLA-DR, CD83, CD80, and CD86 (R&D Systems). For intracellular detection of MAGE-A3 or CPD–MAGE-A3, DCs were stained with either the custom-made CPD-specific antibody (recognizing YARKARRQARR) or anti–MAGE-A3 antibody. Isotype-specific negative controls designed for both extracellular and intracellular staining were used in all experiments.
We used polymerase chain reaction to amplify MAGE-A3 from kidney tissue complementary DNA, subcloned it into a pDRIVE T vector (Figure 1A) (Qiagen), and further cloned it in-frame with CPD sequence YARKARRQARR in a CPD vector (Figure 1B). Both MAGE-A3 and CPD–MAGE-A3 were excised for further cloning into expression vectors (Figure 1C).
Both MAGE-A3 and CPD–MAGE-A3 were digested from the pDRIVE T vector and ligated into the pQE-70 protein expression vector in-frame with downstream hexahistidine (Figure 2A and B). Identity of the vectors was confirmed by restriction digestion analysis and DNA sequencing (data not shown). The pQE-70 expression vectors are designed to express a target protein under the regulation of promoter T5/lac in E coli BL21 Star (DE3-pLysS) cells that are specifically designed for efficient lysis and induction of recombinant protein. Three clones of each recombinant protein were selected, and protein induction was attempted at various times and temperatures in Luria-Bertani Broth. The temperature shift from 37°C to 33°C after a 5-hour IPTG induction improved recombinant protein expression, with the highest expression observed in clone 3 of MAGE-A3 (Figure 2C) and clone 1 of CPD–MAGE-A3 (Figure 2D).
We investigated the effect of different media composition on the growth of E coli carrying CPD–MAGE-A3 and the induction of protein expression (Figure 3A). While Luria-Bertani Broth and Power Broth showed the slowest biomass production as well as low rates of induction of recombinant protein, we observed a 3-fold increase in biomass production in 2XYT Broth and a 2-fold increase in Superior Broth after a 5-hour induction. Protein expression was analyzed at various concentrations of IPTG in 2XYT Broth, with maximal induction achieved at 1.0mM (Figure 3B).
Recombinant proteins were purified from the selected high-expression clones using nickel-charged resin affinity chromatography. The MAGE-A3 and CPD–MAGE-A3 proteins were effectively eluted at imidazole concentrations of 150mM and 200mM, respectively, although there was minor contamination in both cases. The purified MAGE-A3 and CPD–MAGE-A3 proteins each showed as a single band at approximately 35 kDa and 38 kDa, respectively, which were similar to their calculated molecular mass (Figure 4A). Although more than 90% of the expressed protein was produced in the insoluble fraction as inclusion bodies, approximately 0.1 mg of both recombinant proteins was purified from the soluble fraction of the cell lysate. Western blot analysis with MAGE-A3 antibodies recognized both MAGE-A3 and CPD–MAGE-A3 proteins (Figure 4B), while CPD antibodies recognized only CPD–MAGE-A3 (Figure 4C).
We compared the ability of CPD–MAGE-A3 vs MAGE-A3 to access the cytoplasm of DCs by pulsing the proteins on day 6 of culture with 3mM MAGE-A3 control and 3mM CPD–MAGE-A3. We observed very little fluorescent staining in MAGE-A3–pulsed DCs after 2 hours (Figure 5A). In contrast, CPD–MAGE-A3 penetrated the DCs within 5 minutes after pulsing. This clearly demonstrates efficient DC penetration and a rapid way to introduce CPD–MAGE-A3 into the cytoplasm. Studies using deconvolution and confocal microscopy confirmed that CPD–MAGE-A3 was localized to the DC cytosol (Figure 5B and C).
To understand whether the CPD of CPD–MAGE-A3 changed the surface antigen expression of DCs, we analyzed several unique DC markers by flow cytometry. The persistent, high-level expression of CD80, CD83, and CD86 in DCs pulsed with CPD–MAGE-A3 is consistent with a mature phenotype, indicating that this process did not interfere with the expression of critical monocyte-derived DC surface molecules (Figure 6).
Although chemotherapy remains one of the mainstays of cancer treatment, cancer vaccines are being increasingly recognized as an important component of combinatorial therapy.12,13 Some of the key advantages of DC-based immunotherapy are its excellent safety profile and its negligible adverse effects, which may positively affect a patient’s quality of life.
Various methods have been used to formulate DC vaccines to generate tumor-specific CTLs. In ex vivo pulsing, the tumor protein, lysate, and peptide preparation is pulsed with DCs that are able to internalize, process, and present Ags.14 First-generation vaccines with exogenous pulsing showed limited success. Even when these ex vivo strategies were complemented with addition of either cytokines4 or Toll-like receptors15 that enhance immune responses and induce DC maturation, the rate of tumor regression did not exceed that observed with standard chemotherapy. Along these lines, several immune-targeted clinical trials have correlated marginally improved overall survival with the presence of CTLs, but only when concomitant chemotherapy was administered.16
The CPD method described here to pulse DCs may represent a significant improvement over prevailing techniques to generate therapeutic vaccines. Dendritic cells transduced with CPD-ovalbumin induced both CD4+ and CD8+ T-lymphocyte responses and regressed ovalbumin-expressing tumors in mice.17 Vaccination of DCs with the transactivator of transcription–CPD of carcinoembryonic antigen has been shown to efficiently inhibit tumor growth when compared with DCs expressing carcinoembryonic antigen alone.18 These results demonstrate that the use of CPD-recombinant TSA is more effective in generating tumor cell lysis than TSA alone. The 16–amino acid peptide pentratin from the DNA-binding domain of the Drosophila antennapedia gene more efficiently presented the immunodominant ovalbumin epitope to generate CTLs when compared with a control construct and also generated an interferon γ response in C57BL/6 mice.19 Growth of melanoma in mice was prevented when challenged with DCs transduced with transactivator of transcription–CPD–Trp2 protein, but Trp2 protein–pulsed DCs alone were shown to be ineffective.20
MAGE-A3 has been shown to be immunogenic and its peptides are presented by HLA class I and II molecules, generating both helper T lymphocytes and CTLs.21 These attributes make it an ideal candidate for immunotherapy. Indeed, it has been successfully used as an immunotherapy target in melanoma and non–small cell lung cancer.22,23 More recently, MAGE-A3 protein has been incorporated into the first US Food and Drug Administration–approved DC vaccine, APC 8015 (sipuleucel-T), to be used in combination with chemotherapy for the treatment of prostate cancer.24 We have demonstrated that the addition of CPD to MAGE-A3 increases its cytosolic bioavailability and thereby may further facilitate antitumor CTL responses compared with prior studies.
Expression of proteins within tumor cells demonstrates both interindividual and intraindividual variation as the cancer progresses. Although most patients with cancer express high levels of MAGE-A3 protein, some have low-level expression, which is due to hypermethylation of the promoter and will impair recognition by DC vaccine–generated CTLs.25 The DNA hypomethylating agent 5-aza-2′-deoxycytidine has been shown to increase the expression of MAGE-A3 in Mel 313 melanoma cells expressing low levels of the protein, but not in Mel 275 melanoma cells already expressing high levels of MAGE-A3.25 These observations indicate a possible role for 5-aza-2′-deoxycytidine as an adjunct to immunotherapy to upregulate tumor-specific MAGE-A3 expression in patients who have low expression levels, thereby potentially increasing the pool of patients eligible for treatment.
We have demonstrated for the first time, to our knowledge, that cloning and purifying MAGE-A3 with CPD enhances its cytosolic bioavailability in DCs without altering specific CD antigens required for T-cell activation, potentially making it a more potent therapeutic cancer vaccine compared with existing MAGE-A3 protein and peptide vaccines. Further, we show that bacterial-recombinant proteins can easily be engineered to purify large amounts of CPD–MAGE-A3. Use of full-length proteins circumvents the need to define HLA class I allele binding before vaccination and increases the number of epitopes recognized by CTLs when compared with peptide-pulsed DCs. Finally, the use of proteins rather than plasmids or viral vectors for in vitro DC vaccine preparation avoids the practical and theoretical safety concerns regarding genomic modification.26
Corresponding Author: Scott A. Gruber, MD, PhD, MBA, John D. Dingell VA Medical Center, 4646 John R St, Detroit, MI 48201 (firstname.lastname@example.org).
Accepted for Publication: June 29, 2013.
Published Online: March 26, 2014. doi:10.1001/jamasurg.2013.4113.
Author Contributions: Dr Batchu had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Batchu, Gruzdyn, Weaver, Gruber.
Acquisition of data: Batchu, Gruzdyn.
Analysis and interpretation of data: Batchu, Gruzdyn, Potti, Gruber.
Drafting of the manuscript: Batchu, Gruber.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Batchu, Gruzdyn.
Obtained funding: Batchu, Weaver.
Administrative, technical, and material support: Batchu, Gruzdyn, Potti, Gruber.
Study supervision: Batchu, Weaver, Gruber.
Conflict of Interest Disclosures: None reported.
Previous Presentation: This paper was presented at the 2013 Annual Meeting of the Association of VA Surgeons; April 23, 2013; Milwaukee, Wisconsin.