Synthetic Anticancer Gene Medicine Exploits Intrinsic Antitumor Activity of Cationic Vector to Cure Established Tumors

The clinical realisation of the potential benefits of gene therapy, in particular for cancer, remains limited, partly because of the challenges of systemic delivery (1, 2). To kill tumor cells at remote sites, a viral or synthetic vector has to travel in the vascular compartment to the tumor blood vessels, across the vessel wall, and through the interstitium to the cell (3). Nonviral delivery systems based on synthetic molecules such as cationic lipids or polymers are relatively simple compared with virus, but have been seen as being limited by low transfection efficiency, in particular in the tumor, after systemic administration.

Here we show that systemic administration of such systems as highly potent gene medicines is feasible if the intrinsic antitumor activity of the dendrimer and its capability for efficient systemic delivery can be exploited. In particular, the PPI_G3 dendrimer acts as a complexation and transfection agent as well as an active drug with intrinsic antitumor activity, thus creating a gene medicine with an efficacy comparable to or better than standard therapy (4).

Nanoparticles. The plasmid constructs were based on previously published sequences (5) and commercially available expression plasmids [pORF9-mTNFα and pORF-hTNFα (Invivogen, San Diego, CA); pCMV SPORTβgal (Invitrogen, Renfrewshire, United Kingdom)] and were purified using an Endotoxin-free Giga Plasmid Kit (Qiagen, Hilden, Germany). Nanoparticles of plasmid DNA (50 μg) and fractured PAMAM dendrimer (SuperFect, Qiagen, Hilden, Germany), linear polyethylenimine (ExGen 500, MW 22 kDa, Helena Biosciences, Sunderland, United Kingdom), or polypropylenimine dendrimer generation 3 (PPI_G3; DAB-Am16) was prepared in 5% dextrose according to the instructions of the manufacturer (100 μL SuperFect, 9 μL polyethylenimine) or as previously described (6).

Animals. Female mice (CD1-nu; initial mean weight, 20 g) housed in groups of five at 19°C to 23°C with a 12-hour light-dark cycle were fed a conventional diet (Rat and Mouse Standard Expanded, B&K Universal, Grimston, United Kingdom) with mains water ad libitum. Experimental work was carried out in accordance with U.K. Home Office regulations and approved by the local ethics committee.

Tumor cell suspensions [LS174T human colorectal adenocarcinoma (ATCC CCL-188), A431 epidermoid carcinoma (ATCC CRL-1555), and C33a human cervix carcinoma (ATCC HTB31)] in exponential growth were injected s.c. (10⁶ cells per flank). Tumors typically reached diameters of 5 mm or more within 7 (LS174T) to 10 days (A431 and C33a).

β-Galactosidase expression was evaluated 48 hours after i.v. administration of lacZ plasmid nanoparticles; the whole tumor was incubated overnight in 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) and histologic sections (paraffin, 20 μm) fixed in 10% formalin and counterstained with eosin.

Gene therapy treatment was administered by i.v. tail-vein injection (50 μg of DNA in 200 μL) once daily on only one occasion (day 0) or on an alternative schedule of five occasions (treatment days 0, 2, 4, 6, and 8). Animals (n = 5) were weighed and volume was determined from daily caliper measurements (volume = d³ × π / 6).

Results are expressed as means ± SE and statistical significance was determined by one-way ANOVA followed by the Bonferroni post test (GraphPad Prism software). Results were expressed as relative tumor volume (rel.Vol_tx = Vol_tx / Vol_t0) and responses classified analogous to Response Evaluation Criteria in Solid Tumors (RECIST; ref. 7). Progressive disease is defined as an increase in relative tumor volume >1.2-fold, stable disease as a relative volume between 0.7 and 1.2 of starting volume, partial response as measurable tumor with a reduction of more than 30% (0-0.7), and complete response as the complete absence of any measurable tumor. Differences were considered as significant when P < 0.05.

Expression from specific promoters. For the determination of relative expression levels, xenografts were excised 48 hours after injection, snap-frozen, and ground to powder in liquid nitrogen. Frozen tissue was homogenized with Hybaid Ribolyser (Hybaid, Middlesex, United Kingdom) and lysing matrix D tubes (Qbiogene, Livingstone, United Kingdom). RNA was extracted using Trizol (Invitrogen) and was treated with DNase I before cDNA synthesis using GeneAmp reverse transcription-PCR reagents (Applied Biosystems, Cheshire, United Kingdom). For DNA extraction, powdered tissue was incubated in tissue lysis buffer [100 mmol/L Tris-HCl (pH 8.5), 5 mmol/L EDTA, 0.2% SDS, 200 mmol/L NaCl, 100 μg proteinase K/sample] treated with RNase A, and DNA was collected by phenol/chloroform extraction and ethanol precipitation.

Quantitative PCR (GRI Opticon using SYBR Green for detection) with the primers 5′-CCCAGACCCTCACACTCAGA-3′ and 5′-GCAGAGAGGAGGTTGACTTT-3′ was used for detection of tumor necrosis factor α (TNFα) cDNA. The primers 5′-GATACCTGTCCGCCTTTCTCC-3′ and 5′-AGCAGAGCGCAGATACCAAATAC-3′ were used for detection of the vector backbone sequence in genomic DNA samples. The primers 5′-TTCCGCAAGTTCACCTACC-3′ and 5′-CGGGCCGGCCATGCTTTACG-3′ were used for detection of ribosomal S15 genomic DNA and cDNA. Primer dimers were excluded from the quantification by performing the optical read step at 82°C. Levels of TNFα cDNA or vector DNA detected in each sample were normalized to S15 cDNA or genomic DNA, respectively, and the overall TNFα expression level in each tumor was adjusted for the normalized levels of vector. The reactions were done in triplicate and repeated twice and the results presented as pooled data.

The concentration of soluble p75TNF receptor 2 (TNFα-R2) in serum was also measured 48 hours after injection (n = 5; Quantikine M kit, R&D Systems, Inc., Abingdon, United Kingdom).

In contrast to other polymers used in synthetic vectors such as linear polyethylenimine (MW 22 kDa; ref. 8) or fractured PANMM dendrimers (9), the lower-generation polypropylenimine dendrimers are small molecules (e.g., DAB-Am16, MW 1.67 kDa) that form nanoparticulate complexes with DNA, which efficiently transport plasmid DNA (10) and oligonucleotides (11) into cells. In particular, the lower-generation polypropylenimine dendrimers and derivatives were found to strike a favorable balance between the requirement of DNA binding and toxicity (10, 12).

We show here that nanoparticles made by the simple mixing of PPI_G3 and a expression plasmid for lacZ are indeed able to reach established s.c. A431 epidermoid carcinoma tumor xenografts and mediate transgene expression after a single tail-vein injection in CD1 nude mice (Fig. 1B). The lacZ transgene expression in these well-developed xenografts is predominantly found in the viable periphery of the tumor.

To evaluate the therapeutic potential of the system, established A431 tumors were treated with nanoparticles of PPI_G3 (250 μg) and a TNFα expression plasmid (pORF9-mTNFα, 50 μg) and compared with the effect induced by plasmid alone, PPI_G3 dendrimer polymer alone, and an uncomplexed combination of DNA and PPI_G3 dendrimer given as two separate injections (Fig. 1). The treatment of a total of five injections (200 μL, in 5% glucose) administered via the tail vein every other day (treatment days 0, 2, 4, 6, and 8) was given once tumors were established (>5 mm). Unchecked tumor growth required euthanasia of all untreated animals within 3 to 4 weeks, but all treated tumors showed some initial growth delay (Fig. 1A). TNFα plasmid complexed with dendrimer induced a measurable tumor volume reduction within 24 hours. Over the next 2 to 3 weeks, this led to a marked regression and long-term survival in 100% of tumors [50% complete response, 50% partial response (RECIST); Fig. 1C and D]. Histologic sections of tumor taken 48 hours after treatment show widespread central necrosis (Fig. 1B,a-c), a common phenomenon for this tumor, and also a known mechanism for the antitumor effects of TNFα (13). Interestingly, the administration of the dendrimer alone also leads to expression of TNFα (see below). The TNFα plasmid given as naked DNA gave rise to an initial growth delay (Fig. 1A) and delayed progression (Fig. 1C), but did not lead to an overall improvement of response and outcome (Fig. 1D). Administration of plasmid DNA and polymer given as two separate injections led to a more pronounced initial growth delay but also did not improve the overall outcome dramatically (Fig. 1). When the PPI_G3 dendrimer was complexed with a promoterless version of the expression plasmid, the growth delay was more pronounced and no progression was observed for over 5 weeks. About 50% of tumors did not progress over the duration of the experiment (12 weeks) and showed complete and partial tumor response. The injection of free PPI_G3 dendrimer led to a similar potent growth retardation with comparatively longer delay of progression-free survival. Around 80% of tumors did not progress over the duration of the experiment, showing stable disease (1 of 9), partial response (2 of 9), as well as complete response (3 of 9; Fig. 1D).

TNFα is a potent but potentially highly toxic pleiotropic cytokine which is capable of inducing necrosis in tumors (13). Because of its potential systemic toxicity, its therapeutic use is limited to local administration (e.g., in isolated limb perfusion procedures; ref. 14). In our experiments, the systemic use of the TNFα gene therapy was well tolerated and no signs of acute toxicity or significant weight loss were observed (cf. Supplementary Fig. 1B and D). Whereas treated and untreated animals developed at the same overall speed, the animals in the TNFα + DAB control group, for reasons not entirely clear, showed a 20% body weight increase in the observation period.

Taken together, these data suggest that the antitumor activity observed with the systemic nanoparticle gene therapy (PPI_G3 + TNFα plasmid + nonspecific EF1α/HTLV promoter) is a result of the therapeutic effects of both plasmid and polymer. TNFα plasmid DNA and free dendrimer give rise to a significant growth delay but their coadministration does not show an additive effect. It is only when both compounds act together in the form of nanoparticles that the full therapeutic effect is observed. First, the ability of the dendrimer to deliver the complexed DNA to the tumor allows efficient expression of TNFα in the tumor. Second, the intrinsic antitumor activity of dendrimer acts together with the therapeutic DNA to provide an overall effect stronger than that observed after administration of either component alone or after coadministration.

This not only is the first report of a pronounced antitumor effect of cationic polymers, such as PPI_G3, but it also shows that the exploitation of both carrier and pharmacologic functionality of the constituent parts of a gene delivery system may complement the effects of the therapeutic gene and thus lead to improved genetic therapies.

The therapeutic effects of the PPI_G3 dendrimer nanoparticles are not unique to the A431 epidermoid carcinoma model but were also observed in other established xenografts models, such as the LS174T colorectal carcinoma or the C33a cervix carcinoma (Fig. 3, top). Whereas the C33a xenograft response to treatment is quite similar to the observations in the A431 model, the growth delay in LS174T is clear but less pronounced, suggesting that the specific nature of the tumor may play an important role.

The effects induced by the murine TNFα expression plasmid (pORF9-TNFα) and of the human form (pORF-hTNFα) are of a very similar magnitude and kinetics (Supplementary Fig. 1A). A single i.v. injection of either plasmid as naked DNA does not have any appreciable effect, but when the plasmids are complexed into a nanoparticle, both forms of TNFα lead to clear tumor regression.

It is possible to further increase the specificity of gene medicines through transcriptional targeting (15). We have tested the promoter fragments for human telomerase, the telomerase RNA (hTR) and the telomerase reverse transcriptase (hTERT), to express TNFα in a tumor-specific fashion (Fig. 2A-D; refs. 5, 16). The growth delay obtained with the gene medicine containing the specific promoters and a nonspecific composite promoter was comparable [elongation factor 1 α (EF1α) and human T-cell leukemia virus (HTLV)] and no statistically significant difference was found between the specific and nonspecific promoters over 3 weeks (Supplementary Table 2). Determination of the relative expression levels by quantitative PCR (Fig. 2C) and measurement of the sTNFα-R2 in serum (17) 48 hours after a single treatment confirm that TNFα is being expressed from all promoters and that it is biologically active. This single time point, however, cannot capture potential differences in activity over the duration of the experiment. The long-term response data (more than 16 weeks; Fig. 2B) suggest that the hTR promoter fragment may have advantages over the nonspecific EF1α/HTLV promoter, an observation which is consistent with recent in vivo imaging data (18). Interestingly, the dendrimer alone also seems to induce expression of endogenous TNFα and a biological response in the form of sTNFα-R2 shedding (Fig. 2C and D; ref. 17).

We compared the relative efficacy of polypropylenimine dendrimers as a transfection agent and as single intrinsically active agents with that of other commonly used cationic polymers (linear polyethylenimine, ExGen; fractured PAMAM dendrimer, SuperFect; Fig. 3C and D). Nanoparticles of mTNFα expression plasmid complexed with PPI_G3, ExGen, or SuperFect all delayed tumor growth relative to the untreated control, but only the polypropylenimine dendrimer leads to a significant regression of the tumors. Interestingly, most of the activity of the ExGen and SuperFect complexes may in fact be explained by the intrinsic antitumor activity of the polymers themselves, which all efficiently suppress tumor growth (Fig. 3D). The kinetics of the transgene expression in our experience leads to an initial regression of tumors within the first week whereas the initial effect of the polymers is predominantly static with potential tumor regression generally occurring only after 2 to 3 weeks.

Our data thus show for the first time that telomerase promoter–based transcriptional targeting with systemically injected synthetic gene medicines is a promising therapeutic alternative to local administration and targeting strategies (19, 20). This is also the first report of therapeutically relevant antitumor activity not only from the polypropylenimine dendrimers but also from other structurally different cationic polymers which are commonly used as transfection reagents. These effects, together with the lack of apparent toxicity, potentially make the low molecular weight PPI_G3 dendrimer an interesting anticancer agent, either as a free drug or as part of a gene medicine. On the other hand, it will be important to understand the nature of the intrinsic polymer activity as it may well influence gene expression when these agents are being used as molecular biology tools or used in a therapeutic context.

Taken together, our findings show that genetic medicines based on a combination of efficient systemic delivery, transcriptional targeting, and intrinsic polymer activity are highly potent anticancer agents and may help to deliver some of the promise of gene therapy.

Grant support: Scottish Enterprise (“Proof of Concept”), the Biotechnology and Biological Sciences Research Council, Cancer Research UK, and the University of Glasgow.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.