Extracellular vesicle-mediated in vitro transcribed mRNA delivery for treatment of HER2+ breast cancer xenografts in mice by prodrug CB1954 without general toxicity

Abstract

Prodrugs are harmless until activated by a bacterial or viral gene product; they constitute the basis of gene delivered prodrug therapies called GDEPTs, which can kill tumors without major side effects. Previously, we utilized the prodrug CNOB (not clinically tested) and enzyme HChrR6 in GDEPT to generate the drug MCHB in tumors. Extracellular vesicles (EVs) were used for directed gene delivery and HChrR6 mRNA as gene. Here, the clinical transfer of this approach is enhanced by: 1. Use of CB1954 (tretazicar) for which safe human dose is established; HChrR6 can activate this prodrug. 2. EVs delivered in vitro transcribed (IVT) HChrR6 mRNA, eliminating the potentially harmful plasmid transfection of EV-producer cells we utilized previously; this has not been done before. IVT mRNA loading of EVs required several steps. Naked mRNA being unstable, we ensured its prodrug activating functionality at each step. This was not possible using tretazicar itself; we relied instead on HChrR6’s ability to convert CNOB into MCHB, whose fluorescence is easily visualizable. HChrR6 mRNA-translated product’s ability to generate fluorescence from CNOB vicariously indicated its competence for tretazicar activation. 3. Systemic IVT mRNA loaded EVs displaying an anti-HER2 scFv (“IVT EXO-DEPTs”) and tretazicar caused growth arrest of human HER2⁺ breast cancer xenografts in athymic mice. As this occurred without injury to other tissues, absence of off-target mRNA delivery is strongly indicated. Many cancer sites are not amenable for direct gene injection, but current GDEPTs require this. In circumventing this need, a major advance in GDEPT applicability has been accomplished.

Introduction

We recently described our initial attempts (1) at developing targeted gene-delivered enzyme prodrug therapy (GDEPT) to treat HER2+ve breast cancer (BC), which remains a serious disease. While antibody-based therapeutics and kinase inhibitors targeting the HER2 receptor are effective (2–4), 10-year follow up studies show drug-resistant relapse in ≥25% in early-stage treated patients (5). Prodrugs are harmless in their native state but are converted into toxic drugs when activated by an enzyme of bacterial or viral origin (6–8). Confining the enzyme-encoding gene to the tumor will localize toxicity to the cancer abating the side effects of conventional chemotherapy. Current GDEPTs are not targeted and require gene injection in cancer site, limiting their usefulness: many cancers, especially multiple sites of metastatic cancer, are not amenable to direct gene injection. Our goal is a regimen for gene delivery specifically to cancer upon systemic administration.

Our previous approach (1) had several innovative features: A. Use of extracellular vesicles (EVs; also called exosomes) (9), generated by Human embryo kidney 293 (HEK293) cells, for gene delivery; these EVs are non-toxic in mice (1,10). EV-mediated foreign mRNA delivery had not been accomplished before (Discussion). EVs are 30 – 100 nm lipid bilayers generated by body cells, performing intercellular exchange of biomolecules believed to be physiologically important (11). Being natural antigen carriers, they may be less immunogenic than viruses and nanoparticles. That they can avoid the lysosomal-endosomal pathway and phagocytosis effectively enough for targeted delivery of si- and miRNAs is well established [(12,13); Discussion].

B. We used mRNA instead of DNA for gene delivery. mRNA is translated directly upon cytosol entry, resulting in efficient gene expression; in contrast, DNA transcription has to precede translation, requiring nucleus entry, which is highly inefficient (<0.10% of cytosolic DNA enters the nucleus) (14). mRNA does not pose the danger of insertional mutagenesis. And mRNA, unlike DNA, is translated in dormant cells (14), which are present in tumors (15). mRNA-based gene delivery has had limited application, but the available comparisons with DNA indicate its superiority (16).

C. Our regimen consisted of a new prodrug CNOB (C₁₆H₇CIN₂O₄) that we discovered, and our bacterial enzyme, HChrR6, which we have improved and humanized. HChrR6 reductively activates CNOB (17) into the toxic drug MCHB (C₁₆H₉CIN₂O₂), which can kill several cancer cell types by apoptosis. MCHB is strongly fluorescent and can be quantitatively imaged in living mice; this greatly facilitated the development of HChrR6/CNOB therapy (17,18).

D. Loading of exogenous functional mRNA into the EVs has been problematic; we accomplished this by constructing a novel plasmid (1).

E. We designed a novel chimeric construct, called the “extracellular-vesicle-HER2-binding (EVHB)” protein; it contains a high affinity anti-HER2 scFv [(K(d) = 10⁻⁹ mol/L] and attaches to the EV surface by its lactadherin C1-C2 domains, targeting them to the HER2 receptor. These HChrR6 mRNA-loaded and targeted EVs are termed “P EXO-DEPTs”; “P” indicates that the mRNA loading involved a plasmid. For ease of reference, schematics of the EVHB protein and the P EXO-DEPTs are reproduced as Figure S1 (1).

In vitro, the P EXO-DEPTs delivered functional HChrR6 mRNA to HER2⁺ but not to HER2^- cells, and when administered systemically along with CNOB, they arrested the growth of HER2⁺ human BC xenografts in athymic mice (1), showing that they delivered the HChrR6 mRNA to the xenografts. But how specific was this delivery to the xenografts was not investigated.

This aspect is explored here. Also, we have made changes to our previous regimen to enhance its clinical transfer prospects: use of the prodrug, CB1954 (tretazicar) which, unlike CNOB, has been in clinical trial with its safe human dose published (6,7); and delivery by targeted EVs of in vitro transcribed (IVT) HChrR6 mRNA instead of mRNA inserted using a plasmid; plasmids can have harmful effects.

IVT mRNA loading of EVs required several steps (Figure 1). Naked mRNA being unstable, we ensured its prodrug activating functionality during these steps. This was not possible with CB1954 (see below); we relied instead on HChrR6’s ability to convert CNOB into MCHB, whose fluorescence is easily visualizable. HChrR6 mRNA-translated product’s ability to generate fluorescence from CNOB (the “MCHB test”; see below) vicariously indicated its competence for CB1954 activation.

Figure 1.

Schematic of IVT EXO-DEPT preparation and use, involving several steps. (1) Generation of in vitro transcribed (IVT) HChrR6 mRNA. (2) Its transfection into HEK293 cells in the presence of Polyethylenimine. (3) Confirmation that extracellular vesicles (EVs) generated by loaded HEK293 producer cells contain HChrR6 mRNA. (4) Conversion of loaded EVs into IVT EXO-DEPTs by incubation with purified EVHB protein. (5) IVT EXO-DEPT-mediated delivery of HChrR6 mRNA to BT474 cells. (6) Addition of the prodrug CNOB or CB1954/Tretazicar. (7) Prodrug conversion within the cells into the drug. (8) Cell death.

We show that systemic IVT mRNA loaded EVs displaying ML39 scFv (“IVT EXO-DEPTs”) and tretazicar caused near-complete growth arrest of human HER2⁺ breast cancer xenografts in athymic mice at a lower EV dose than before (1). As this occurred without injury to other tissues, absence of off-target mRNA delivery is strongly indicated. Many cancer sites are not amenable for direct gene injection, but current GDEPTs require this. In circumventing this necessity, a major advance in GDEPT applicability has been accomplished.

Material and Methods

Cell lines, and viability.

HEK293 and the HER2⁺ human breast cancer BT474 cell lines have been described (1). They were purchased from ATCC between 2005 to 2009 and authenticated (December 2017; Genetic Resources Core Facility, Johns Hopkins). Experiments were done within one-year of the Mycoplasma test (LookOut Mycoplasma PCR Detection; Sigma-Aldrich; Saint-Louis, MO; and PCR based MycoDetect™ kits; Greiner Bio-One North America, NC) showing negativity. Cells were cultured (moisturized incubator, 37°C, 5%CO₂) in DMEM medium (Thermo Fisher Scientific, Waltham, MA)-10% fetal bovine serum (FBS) and used within 20 passages; MTT assay (Roche, Basel Switzerland) determined viability.

EV preparation.

As before (1,19), 5 x 10⁶ HEK293 “producer” cells [in 10-mL DMEM, plus 10% EV-depleted FBS (“DMEM-EDFBS”)] were plated in a 100-mm dish (4-day incubation). Cells and apoptotic bodies were removed (centrifugation, 600 and 2000 x g, respectively, 30-minutes). The supernatant was centrifuged (100,000 x g, 75-minutes); pelleted EVs were suspended in PBS. Bradford Assay quantified protein. The resulting EVs conformed the MISEV standard (9): uniform peak (NanoSight) and average size of ca. 100 nm [NanoSight and transmission electron microscopy (TEM)]; positive for EV-specific proteins: CD63, CD81 and lactadherin (1).

Construction of the EVHB protein.

The EVHB protein binds EVs by its C1-C2 domains, making them target HER2⁺ cells because of its anti-HER2 scFv (Figure S1). As before, the anti-HER2 ML39 scFv DNA sequence (from pACgp67B-HER2m; Addgene; Watertown, MA) was inserted into p6mLSC1C2 (1) to construct pEVC1C2HER. The producer HEK293 cells, transfected with this plasmid, generated EVs displaying EVHB, which was purified as before (1).

Loading EVs with HChrR6 mRNA and construction of IVT EXO-DEPTs.

Loading functional exogenous mRNA into the EVs proved difficult (see below). We previously accomplished this by constructing a new plasmid termed, pXPort/HChrR6 mRNA, using the System Biosciences XPort plasmid and the zipcode sequences (20) inserted at the 3’-UTR of the HChrR6 gene (MSCV promoter). pXPort/HChrR6 mRNA-transfected HEK293 cells generated HChrR6 mRNA-containing EVs (1).

Here, we succeeded, to our knowledge for the first time, in loading EVs with functional IVT mRNA (Results). These were converted into IVT EXO-DEPTs, as before (Figure S1): 2 x 10⁷ EVs were incubated with pure EVHB (1 µg, room temperature, 15 minutes). NanoSight (NS300; Melvin Instruments), and TEM (JEOL JEM1400) analyses showed that they fully resembled P EXO-DEPTs [Supplementary Figure S2; (1)], conforming to the MISEV standard (9).

MCHB test.

This test was used to vicariously determine the capability of translated HChrR6 mRNA to activate tretazicar (which was not practical using tretazicar itself; Results). HChrR6 enzyme (previously named ChrR21) can also reductively activate tretazicar: when tretazicar (15 µM) was incubated with HChrR6 (10 μgml⁻¹, 37°C), it killed >80% breast cancer cells (24-hour incubation, p>0.01) (21). In the MCHB test, in vitro translated HChrR6 mRNA was tested for generating MCHB fluorescence upon CNOB addition in a standard reaction mixture; this contained translated HChrR6 protein + CNOB (15µM) + NADPH (1mM) (17,22). MCHB fluorescence was measured in SpectraMax plate reader (Molecular Devices; excitation, 570 nm; emission, 620 nm). Generation of fluorescence from CNOB vicariously indicated the enzyme’s competence for tretazicar activation.

HChrR6 mRNA transcription, translation and enzyme assay in vitro.

To generate IVT mRNA, the HChrR6 gene was extracted from puC57-HChrR6, using KpnI (1,17,22) and was cloned into pcDNA™6/myc-HisA (ThermoFisher); in-frame insertion was confirmed (Stanford University Protein and Nucleic Acid Facility). pcDNA™6/myc-HisA-SHChrR6 was used to transcribe HChrR6 mRNA in vitro (HiScribe™ T7 ARCA mRNA Kit with tailing; New England Biolabs, Ipswich, MA). mRNA was quantified (NanoDrop 1000 Spectrophotometer; Thermo Fisher; Wilmington, DE) and used to synthesize cDNA (M-MuLV reverse transcriptase, New England Biolabs); RNaseH treatment (New England Biolabs) removed any remaining mRNA. Quantitative reverse transcription-PCR (qRT-PCR) employed Maxima SYBRGreen/ROX qPCR Master Mix Kit (Thermo Fisher Scientific) and 7500/7500 Fast Real-Time PCR System (Applied Biosystems; Foster City, CA); GAPDH mRNA served as endogenous control.

The primers employed were for HChrR6 and GAPDH have been specified before (1). mRNA amount estimation employed a standard curve relating Ct value with mRNA quantity. EV mRNA copy number calculation was by the formula: [X (ng)×6.0221×10²³ (molecules/mole)]/[N×330 (g/mole)×10⁹ (ng/g)], where X is the amount of EV mRNA, N is its length, and 330 is the average molecular weight of individual nucleotides (Rhode Island Genomics and Sequencing Center).

IVT mRNA was translated in vitro (PURExpress® kit, New England Biolabs); the resulting HChrR6 protein was MCHB test positive. E. coli dihydrofolate reductase (DHFR Control Template) supplied in the kit was used to ensure correct translation; it was MCHB test negative, serving as negative control for the translated HChrR6 mRNA.

Loading of EVs with IVT HChrR6 mRNA.

The EV-depleted fresh medium (DMEM-EDFBS) was prepared as before [ultracentrifugation; 110,000g; 4°C; (1)]. HEK293 cells (7×10⁵) seeded in 6-well plates were transfected with HChrR6 IVT mRNA (2 µg/well) using Polyethyleneimine (PEI, 2.5 × mRNA), unbound PEI was removed, the medium replaced (24- hours later) with fresh EV-depleted medium, incubated (37°C, 4-days) and the EVs were isolated. Total RNA extraction from the producer cells and the EVs employed RNeasy Mini Kit (QIAGEN, Hilden, Germany); HChrR6 mRNA was quantified as above.

Flow cytometry.

IVT EXO-DEPTs (Figure 3 legend provides numbers) were stained with PKH26 dye [4 µL in 1-mL Diluent C; Sigma-Aldrich; (1)], washed twice to remove unbound dye and quantified for protein. They were incubated with BT474 cells (3 wells of a 6-well plate), washed (PBS), dislodged (trypsin), suspended in DMEM/10% FBS, harvested (900×g; 4°C; 5- minutes), washed with 1 mL of acid buffer (0.5 mol/L NaCl; 0.2 mol/L acetic acid, pH 3.0), to remove non-internalized EVs), and treated with Flow Cytometry Fixation Buffer (R&D Systems; MN; 4°C; overnight). The cells were pelleted (900×g; 4°C; 5-minutes), resuspended in 1 mL FACS buffer (PBS/1% BSA/0.1% NaN₃) and analyzed at the Scanford FACS analyzer (excitation, 488 nm; emission, 590/20 nm).

Figure 3. A. IVT EXO-DEPTs interaction with HER2⁺ BT474 cells and HChrR6 mRNA delivery.

Flow cytometry results of 1.62 × 10⁶ BT474 cells incubated for 6-hours (upper panels), or 24-hours (lower panels) with PKH26-labeled (red) or unlabeled (yellow) 5 × 10¹⁰ IVT EXO-DEPTs. At 6-hours, 0.6% cells bounded the EVs (red; upper right panel); at 24 hours this number increases to 80% (lower right panel) (n=3). B. HChrR6 mRNA copy number in BT474 cells after 6-hour incubation with IVT EXO-DEPTs (red bar). No HChrR6 mRNA was detected in BT474 cells incubated for the same duration with directed EVs not containing the IVT mRNA (* p<0.05; n=3); consequently, no blue bar (representing unloaded EVs) is seen in the figure.

In vitro assays.

To test functionality of the translated HChrR6 mRNA, different amounts of it (Results) were subjected to the MCHB test. HChrR6 mRNA in IVT- and P EXO-DEPTs was quantified as above. For comparison of their efficacy in donating HChrR6 mRNA, BT474 cells (see Figure 4 legend for numbers) were incubated (DMEM-EDFBS; 100 µL in a 96-well plate; 37°C; 5% C0₂) with IVT- or P EXO-DEPTs [or with directed unloaded “control” EVs (1)]. Incubation medium was replaced with fresh DMEM-EDFBS containing 15 µM CNOB. After incubation (Results), the exposed cells were subjected to the MCHB test by measuring MCHB fluorescence.

Figure 4. Comparative efficacy of IVT- and P EXO-DEPTs.

Comparison of the kinetics of HChrR6 gene expression by BT474 cells (10⁴) receiving the HChrR6 mRNA by IVT EXO-DEPTs (red bars) or by P EXO-DEPTs (blue bars), as determined by the MCHB test. The number of the two kinds of the EVs was adjusted to deliver 10⁴ mRNA copies. This required 3×10⁵ IVT- and 5×10⁷ P EXO-DEPTs (* p < 0.05, ** p < 0.01 and *** p < 0.001; n=3). The IVT EXO-DEPTs generate superior results.

In vivo assay of IVT EXO-DEPT functionality.

Animal experiments were approved by Stanford University Institutional Animal Care and Use Committee (IACUC). As before (1), the number of mice was determined using the G*Power calculator (Universitat Dusseldorf) for F-tests of one-way ANOVA by setting Type I error at 5% (alpha= 0.05); power was kept at 0.8, and the number of treatment groups, 4. The effect size, f, was calculated as 0.707 and indicated total sample size of 28 with 7 mice per treatment group. Five to six-week old female BALB/C athymic nude mice (Charles River Laboratories, Wilmington, MA) were implanted subcutaneously with 17β-estradiol pellets (Innovative Research of America, Sarasota, Florida) to support growth of xenografts of BT474 (estrogen positive) cells. 10⁷ cells were injected subcutaneously. For full details of the procedure, see (1). Caliper was used to measure tumor size at 2-day intervals. Tumor volume and growth rate were calculated using the formula: tumor width²×its length/2, and from slopes of linear regression, respectively. When the tumors reached a volume of ca. 180 mm³, the mice were assigned to four groups and treatments started (Results).

Toxicology and histopathology.

Following mice euthanasia, serum and whole blood were collected. Organs (spleen, kidney, liver, heart, lung and brain) were harvested and fixed in 10% neutral buffered formalin to perform toxicology and histopathology analyses; these were conducted at SRI Biosciences, Menlo Park, CA. The tissues were processed by routine methods, sectioned, mounted on slides, and stained with hematoxylin and eosin.

Data and statistical analysis.

Excel2010 software was used. Statistics were determined using Student’s t-test; p values <than 0.05 were considered significant.

Results

Rationale of the MCHB test.

Making IVT EXO-DEPTs and their use involved multiple steps, e.g., IVT mRNA synthesis, its loading into HEK293 cells, isolating mRNA-loaded EVs, and others illustrated in Figure 1. As naked mRNA can be highly unstable [(23,24); Discussion], it was necessary to ascertain that the translated HChrR6 mRNA remained tretazicar activation-competent during these steps. The drug generated by tretazicar reduction is (5-(aziridine-1-yl)-4N-acetoxy-2-nitrobenzamide (referred to from here on as “nitrobenzamide”). Reliable detection of nitrobenzamide requires complex HPLC-based methods (25), which was impractical, given the multiplicity of steps (Figure 1). However, the fact that HChrR6 can activate both CNOB and tretazicar (21) suggested a facile alternative, namely the MCHB test. This was based on the hypothesis that if the translated HChrR6 mRNA could generate MCHB fluorescence upon CNOB addition, it would be capable of activating tretazicar as well; MCHB is strongly fluorescent and thus easily observable. This hypothesis proved correct.

IVT mRNA in EVs.

IVT HChrR6 mRNA was synthesized; the MCHB test-based confirmation of its functionality is shown in Figure 2A: increasing amounts of the translated mRNA generated increasing fluorescence upon CNOB addition. HEK293 cells were transfected with this IVT mRNA; qRT-PCR revealed that the EVs generated by these cells contained ca. 3.3×10⁻² copy of the mRNA/EV. Two copies of the 25-nucleotide “zipcode” sequence promotes mRNA-loading into the EVs (20). However, addition of these to the IVT mRNA did not enhance the loading; so, IVT mRNA without the zipcode was used. As the previously made P EXO-DEPTs (1) contained 2×10⁻⁴ copy/EV, the IVT method proved an advance, decreasing the number of EVs required to donate one mRNA copy from 5,000 to <50 (Figure 2B).

Figure 2.

In vitro transcribed (IVT) HChrR6 mRNA functionality and its amount in IVT EXO-DEPTs. A. Test of the functionality of IVT HChrR6 mRNA. Its translated product converts CNOB into MCHB as measured from MCHB fluorescence. Increasing amounts of the translated HChrR6 mRNA protein (quantified by Bradford assay) generated increasing fluorescence (n=3). Numbers in the abscissa denote: 1. PBS; 2. CNOB 15µM; 3. NADPH 1mM; 4. full reaction mix (Rx) without HChrR6 protein. (CNOB was 15µM, NADPH, 1mM); 5. Rx + 2.5 ng protein; 6. Rx + 5 ng protein; 7. Rx + 10 ng protein; 8. Rx + 20 ng protein; 9. Rx + 40 ng protein; 10. Rx + 80 ng protein; 11. Rx + 160 ng protein. B. The IVT EXO-DEPTs contain more HChrR6 mRNA compared to P EXO-DEPTs: <50 of the former (red bar), but 5,000 of the latter (blue bar) are needed to donate one mRNA copy (n=3).

Was the HChrR6 mRNA in the EVs adhered to their surface? To investigate this, the loaded EVs were incubated with or without RNaseA, followed by mRNA extraction and quantification. No difference resulted from RNase treatment (Figure S3); thus, the mRNA was mainly inside the EVs. These IVT EVs were converted into EXO-DEPTs capable to target HER2⁺ cells by incubation with the EVHB [Figure S1, (1); Materials and methods].

Binding of IVT EXO-DEPTs to cells and their mRNA delivery efficacy.

FACS analysis of (PKH26-labeled) IVT EXO-DEPT binding to BT474 cells showed 0.6% attachment at 6-hour incubation, which increased to 80% at 24 hours (Figure 3A). 0.6% attachment roughly corresponds to the exposure of each BT474 cell to some 10 HChrR6 mRNA copies. Quantification of the mRNA delivered by the EVs into the cells can be only approximate due to its degradation, etc. in the recipient cells. To get a rough estimation, total RNA was extracted from the cells at the 6-hour incubation point, and the HChrR6 mRNA copies were measured (qRT-PCR); some 5 copies of HChrR6 mRNA were detected per recipient cell, suggesting a high transfection ratio; there was no HChrR6 mRNA in cells treated with directed EVs devoid of mRNA (Figure 3B). We suspect that the efficient mRNA delivery by the EVs to the cells is related to their display of the high affinity [(K(d) = 10 ⁻⁹ mol/L] anti-HER2-scFv that promoted efficient attachment and content release.

EXO-DEPT efficacy in vitro.

We compared the efficacy of IVT EXO-DEPTs with P EXO-DEPTs in conferring prodrug activating capability on the recipient BT474 cells. Separate batches of the cells were incubated with each kind; numbers were adjusted to deliver the same mRNA copies per recipient cell (see legend to Figure 4); gene expression was determined by the MCHB test. It increased with time and, except for the first 24 hours, the IVT EXO-DEPT-treated cells showed greater expression during the 120-hour experiment. While the P EXO-DEPT-mediated expression peaked at 72 hours, it continued to increase during the entire experiment with the IVT EXO-DEPTs. Thus, translation of the mRNA donated by the IVT EXO-DEPTs was more efficient and more durable. Cells incubated with control EVs (directed but without HChrR6 mRNA) were MCHB test negative.

For EVs to deliver mRNA to BT474 cells, they needed to be directed. This was shown in our previous study: mRNA loaded but non-directed EVs failed to confer prodrug activating capability on BT474 cells; loaded and directed ones did [Reference (1); Figures 3A and 4C].

Treatment of implanted human HER2⁺ orthotopic BC xenografts in athymic mice by IVT EXO-DEPTs + tretazicar.

As HChrR6 enzyme effectively activates both CNOB and tretazicar (21), we followed our previous protocol, which achieved near-complete growth arrest of xenografts in mice (1): the EVs were administered in two doses and tretazicar in six (Figure 5A). As before, we administered a total of 2.8×10⁷ mRNA copies/mouse. Since the IVT EXO-DEPTs had more HChrR6 mRNA, fewer of these were needed, 6.8 x 10⁸ instead of 1.5 x 10¹¹ needed for the P EXO-DEPTs (1). Each test group consisted of seven mice. Treatments began after the mice developed measurable (180 mm³) tumors. In mice receiving no treatment, there was vigorous tumor growth, reaching a volume of 400–450 mm³ and growth rate of 5 mm³/day (Figures 5B and C). But in mice receiving both the IVT EXO-DEPTs and tretazicar (CB1954) (“full treatment”), the xenograft growth was minimal and stopped by day eight. The last EV and tretazicar doses were given on days four and seven, respectively (Figure 5A), showing that the tumor growth arrest continued for 21 more days without further treatment. Control mice receiving CB1954 alone exhibited tumor growth resembling the untreated mice.

Figure 5.

In vivo effectiveness of the IVT EXO-DEPTs. A. Administration schedule; the EXO-DEPT numbers and the tretazicar amount are for each dose per mouse. B. Volume of orthotopically implanted BT474 HER2⁺ xenografts in mice, as measured by a caliper; each data point is mean value for individual treatment groups (p < 0.001). The four groups of mice used are identified in the Figure. C. Box-and-Whisker plot of the growth rate of the tumors (*** p < 0.001; n=7). Mice receiving full treatment (IVT EXO-DEPTs + tretazicar) show near-complete arrest of xenograft growth.

Tumor targeting specificity can result from enhanced permeability retention (EPR) effect resulting from blood vessel fenestration in tumors and their impaired lymphatic drainage. Whether this applied to EVs and if so, did the scFv-based targeting strategy we used conferred any advantage was investigated in our previous study (1). This was done by comparing the effect of administering the prodrug with mRNA loaded non-directed or directed EVs in separate groups of mice. There was significant arrest of xenograft growth with the former (loaded, non-directed), but the growth arrest was 50% more pronounced when the directed EVs were used [Reference (1); Figure 4F: compare green and blue curves]. The result bolstered the suitability both of EVs as targeting vehicles and our scFv-based targeting strategy.

However, whether the anti-Her2 scFv displayed by our EVs by itself retarded tumor growth was not investigated. Current HER2⁺ BC therapies (e.g., trastuzumab) rely on interfering with HER2-mediated signaling by interacting with the HER2 receptor, and the anti-HER2 scFv of our EVs could have had a similar effect. This was investigated here by injecting mice only with directed EVs not containing HChrR6 mRNA (no prodrug was administered). Their xenograft growth resembled the untreated group (Figures 5B and C), indicating that the ML39 scFv played no role in arresting tumor growth and the near-complete growth arrest seen in mice receiving the full treatment was due entirely to prodrug activation. A pilot experiment in which only two groups of mice were used--untreated and those receiving full treatment (n=5)--gave very similar results. Experiments were terminated (day 31), as the tumors in the control groups had begun to exceed the volume allowed by the animal protocol.

Because mRNA delivered by the IVT EXO-DEPTs was expressed for a longer duration than the P EXO-DEPT-delivered mRNA (Figure 4), we wondered if fewer IVT EXO-DEPTs would also prove effective. The in vivo experiment was repeated using either half or one log fewer EVs (1.7×10⁸ or 3.4×10⁷ EVs, respectively); “full dose” (Figure 5) was used as control. Treatment schedule and the number of mice used for each group were as in Figure 5. The lower doses, however, had no significant effect on the xenograft growth (Supplementary Figure S4). The reason for this is not known but may be related to the short half-life of EVs in circulation (26), which, at the lower doses, may have decreased the EV number below that required for effective attachment to the xenografts.

Specificity of tretazicar activation in HER2⁺ xenografts.

Since xenograft growth was arrested (Figure 5), it is clear that in the xenografts tretazicar was converted into its cytotoxic product, nitrobenzamide, showing successful delivery of functional HChrR6 mRNA. EVs temporarily colonize pancreas, spleen and liver (27); furthermore, the ML39 scFv of our EVHB protein might also have recognized mouse ErbbB2-expressing cells. Thus, it was possible that the IVT EXO-DEPTs delivered HChrR6 mRNA also to normal tissues causing them to activate tretazicar, triggering general injury. To investigate this, we performed toxicology and histopathology analyses of mice that received no, or the full treatment -- the latter utilized the therapeutic regimen of Figure 5 (that resulted in near-complete growth-arrest of the xenografts).

The animals were sacrificed at the end of the experiment, and whole blood and serum were collected along with tissues (liver, spleen, kidney, lung, heart, and the brain). Figure 6 compares results in treated and untreated mice for hematology (6A) and serum chemistry (6B). (Abbreviations in the abscissa are standard; see Supplementary Table S1 for full names.) The treatment did not significantly affect serum chemistry and whole blood hematology panels. Changes in platelet counts (PLT), mean platelet volume (MPV) and total neutrophil counts (ANS) were significant (p<0.05). Since these changes were small and affected only three hematological parameters, no serious bone marrow suppression is indicated.

Figure 6.

Hematology and serum chemistry. A. Hematology of treated (red) and non-treated (blue) mice. B. Serum chemistry of treated and non-treated mice. Abscissa abbreviations for the constituents are standard; they are spelled out in full in Supplementary Table S1, which also provides numerical values. Mice receiving full treatment show essentially no off-target injury.

Four out of seven mice of the treatment group showed minor elevation of liver enzymes AST and ALT (Figure 6B); however, the overall group averages were not statistically significant, and this outcome is known to be a transient effect of tretazicar treatment (28). Histopathology evaluations of the tissues, including the liver, showed no difference between treated and untreated mice. Thus, the IVT EXO-DEPT + tretazicar treatment is effective without causing notable off-target injury. We suspect that the high specificity of the anti-HER2 ML39 scFv for human HER2 receptor (29,30) and the much higher HER2 expression of the BC xenografts compared to the normal tissues account for this result.

Discussion

Prodrug chemotherapy offers the possibility of confining toxic drug largely to tumors and abating the side effects of most current chemotherapy. Clinical trials of prodrugs have been disappointing (6–8) and the possible reasons for this have been noted. Examples are, problems with gene-delivery vehicles, and dearth of: facile methods for detecting prodrug activation (17,18); of potent activating enzymes (21,22,31–33); and of quantification of tumor transfection with the GDEPT gene for generating sufficient bystander effect to kill the whole tumor (17,18,34,35).

Our 2018 paper (1) addressed some of these issues. First, use of EVs instead of viruses and nanoparticles for gene delivery. That EVs might be superior is widely recognized, as shown in numerous studies involving si- and miRNA delivery. Examples are, let 7ainordertodeliverlet-7a (let7a) to treat breast cancer in mice (36); plasma siRNAs to blood cells for selective gene silencing (37); siRNA to the brain to treat opiate addiction (38); miR-26 to treat liver cancer (39); miR-16–based mimics to treat malignant pleural mesothelioma (40) and use of DNA minicircle for prodrug treatment (41).

Second, since a major problem revealed in prodrug clinical trials is the dearth of sufficient gene transfer and its short-lived expression in the tumor (6–8), we used mRNA instead of DNA for gene delivery, the former being capable of more efficient translation. We believe to be the first to load EVs with functional exogenous mRNA; this was accomplished by constructing a novel plasmid, pXPort/HChrR6 mRNA (1). Around the same time, Kojima et al (42) may also have succeeded in this. They used plasmids to bind the catalase mRNA to the EV CD63 protein and showed that their producer cell EVs transferred catalase activity to recipient cells. But they did not report catalase activity of the producer cells, nor whether catalase expression in the recipient cells was insensitive to the transcription inhibitor, actinomycin D. Thus, the species transported may have been the catalase-encoding plasmid and/or the catalase protein itself, rather than the catalase-encoding mRNA. Earlier attempts at delivering functional mRNA via EVs had uniformly failed. Electroporation did not work (1); in other attempts, we obtained EVs containing a reporter mRNA generated by plasmid-transfected cells, but this mRNA fragmented during its EV-mediated delivery to recipient cells (19); a similar phenomenon was seen by others (43).

The work reported here advances our therapeutic regimen for clinical transfer in significant ways. First, our systemically administered EXO-DEPTs delivered HChrR6 mRNA to xenografts without off-target injury, showing a high degree of xenograft targeting specificity. Specific delivery of the gene to cancer upon systemic administration is critical for GDEPT effectiveness: not all cancers --and especially cancer metastases-- are accessible for direct gene injection upon which the current GDEPTs depend. We note that since the anti-HER2 scFv in the EVHB protein can be replaced by moieties able to target other receptors, our approach can specifically treat any disease significantly over-expressing a marker. Among cancers alone, the over-expressed receptors include bombastin, folate and the ligand PSMA as well as others (44–46).

The second element facilitating clinical transfer is the demonstration that our approach can arrest xenograft growth when tretazicar is used as prodrug; this prodrug has been in clinical trials with its safe dose in humans established (6,7), unlike our previously used prodrug, CNOB (1,17,18).

A third seminal element, also to our knowledge accomplished here for the first time, is the loading of the EVs with IVT mRNA avoiding the use of plasmids, which could be harmful to patients. This was a somewhat daunting task since naked mRNA is prone to fragmentation. For example, when introduced into dendritic cells, it was not expressed until transfected in a nanoparticle format; and modifications like the use of RNA patch were needed for effective expression (23,24,47,48). Thus, at each step in this work (Figure 1), we ensured the functionality of our mRNA for prodrug activation. As mentioned, this was impractical using tretazicar itself. But the MCHB test, relying on the HChrR6 enzyme capability to activate both CNOB and tretazicar, and MCHB fluorescence, enabled us to accomplish this.

We constructed IVT EXO-DEPTs for safety reasons, but they afforded additional benefits. They contained greater HChrR6 mRNA copies and were able to arrest xenograft growth with 220-fold fewer EVs than the P EXO-DEPTs used previously (1). The reasons for greater mRNA in the IVT EXO-DEPTs likely are: i) the IVT mRNA is available directly for transfer into EVs upon entry into the producer cell cytosol; with the pXPort/HChrR6 plasmid (1) in contrast, the mRNA would become available only after transcription, requiring nucleus entry; ii) a given amount of genetic material of naked mRNA has an order of magnitude more copies than an equivalent amount of the plasmid encoding it; and, iii) smaller size of mRNA than its encoding plasmid.

The second beneficial IVT EXO-DEPT characteristic is that the mRNA they delivered resulted in more stable gene expression. While the reason for this remains to be determined, it is a salutary outcome in that short-lived gene expression has constrained GDEPT effectiveness (6–8).

Further improvement of our GDEPT is, however, needed. We have so far attained xenograft growth arrest, but successful therapy requires tumor regression and elimination. We posit that recruiting immunity will help achieve this, since HER2⁺ tumor ablation evokes a strong anti-HER2 immune response (49). We are testing this in immune-competent FVB/NJ mice, which spontaneously develop HER2⁺ breast cancer driven by an oncogenic form of human HER2 (HER2Δ16) (50). As stated, HEK 293 EVs we have been using are, by and large, non-immunogenic, but to further minimize immune rejection, we are using autologous EVs generated by dendritic cells of these mice. Immunogenicity of the human components of the EVHB protein (C1-C2 domains and the anti-HER2 scFv) will, however, remain. While possibly deleterious, it could conceivably be beneficial instead. The lack of general toxicity we report here suggests that the IVT EXO-DEPTs are rapidly attracted to --and may be arrested by-- the HER2⁺ tumor. Thus, the immune rejection may target the tumor and bolster its killing.

Another critical improvement that is needed is to make the IVT EXO-DEPTs more efficient to minimize the need for repeated dosing. Increasing the mRNA content of the EVs and extending its expression in recipient cells can help achieve this. If the IVT mRNA loading of the producer cells can be increased, it is likely that the EVs they generate will also have enhanced content. Available measures for increased mRNA loading of cells include complexation of mRNA with Lipofectin (51) and use of a nonlipid cationic reagent, e.g., Transmessenger. Stability of mRNA and therefore the likelihood of prolonged expression can be achieved by, for example, the use of β-globin mRNA sequences (52), chemical modifications like replacement of uridine or cytidine by pseudouridine or 5-methylcytidine (53), and reversible addition-fragmentation chain transfer (RAFT) polymerization (54). Moreover, incorporating the TEV start site in the HChrR6 mRNA can strongly enhance its translation (55,56). Although with some of these measures care would need to be exercised regarding potential toxicity, there is high probability that suitable methods will be found, given the multiple possibilities noted above and the continuing progress in the field. To conclude: we have taken further important steps for clinical transfer of our regimen, and promising avenues are available to enhance this potential.