In vivo silencing of a molecular target by short interfering RNA electroporation: tumor vascularization correlates to delivery efficiency Free

Screening for a molecular target for cancer therapy requires multiple steps, of which an important one is evaluation of the knockdown effect of the target molecule on pregrown xenograft tumors. However, methods currently used for local administration of knockdown reagents, such as short interfering RNA (siRNA), are not satisfactory as to simplicity and efficiency. We established an electroporation method involving a constant voltage and “plate and fork” type electrodes and used it for in vivo delivery of siRNA. The delivery efficiency correlated to the electric current. The electric current correlated to the microvascular density and vascular endothelial growth factor (VEGF) expression and exhibited a threshold that guaranteed efficient delivery. Consequently, we showed that the vascularization and VEGF expression in tumors determined the efficiency of delivery of siRNA by electroporation. VEGF was chosen as a model target. VEGF siRNA electroporation suppressed the growth of tumors exhibiting high VEGF expression to less than 10% of the control level, but it had no effect on low VEGF-expressing tumors. Notably, a long interval (20 days) of electroporation was enough to obtain a satisfactory effect. Systemically injected siRNA could also be delivered into tumors by this method. Our data will provide the technical basis for in vivo electroporation, and this simple and efficient siRNA delivery method is applicable to in vivo comprehensive screening for a molecular target. [Mol Cancer Ther 2008;7(1):211–21]

Molecular target-based therapy is nowadays the major strategy for treating cancer (1). However, the selection of molecular targets is a rate-limiting step for the development of anticancer drugs, and comprehensive screening of differentially expressed molecules in cancers for a molecular target is necessary for drug development (2, 3). In vitro tests can narrow candidate molecules down, but in vivo screening is needed for further narrowing down the candidates for the following drug design (3). In this context, it is important to develop a convenient system for in vivo comprehensive screening.

RNA interference technologies are currently the most widely used in functional genomic studies (4–7). There is an intense research effort aimed at developing short interfering RNA (siRNA) as therapeutics against various diseases, such as viral infections, neurodegenerative disorders, and cancers (8–10). Therefore, the feasibility of using siRNA for in vivo screening of candidate molecules is high. However, a drug delivery system for siRNA has not been fully developed for in vivo use. We reported previously the use of atelocollagen as a local drug delivery system for siRNA and observed a powerful anticancer effect of siRNA targeting human vascular endothelial growth factor (VEGF; ref. 11). However, a problem with atelocollagen is that careful handling is required because it is soluble at low temperature (below 10°C) but solidifies to refibrillation at more than 10°C (12–14). Another concern is that injection of siRNA with atelocollagen into a tumor has to be done repeatedly at short intervals (four injections every 10 days; ref. 11).

The goal of the present study was to develop an in vivo drug delivery system for siRNA that is convenient enough for in vivo comprehensive screening of molecular targets for cancer. We modified an electroporation apparatus that includes “plate and fork” type electrodes (15) for treating s.c. xenografted tumors. Electroporation was introduced in the 1960s and comprises the application of controlled electric fields to facilitate cell permeabilization (16, 17). Although a few papers have reported the potential of electroporation for in vivo siRNA delivery (18–21), few studies have been conducted to develop in vivo electroporation of siRNA for practical use. In the present study, we first evaluated our “plate and fork” type electrodes as to electric resistance and current in tumors. Second, we determined the conditions allowing the efficient in vivo delivery of siRNA, which identified the tumors this method could be applied to. Finally, we chose VEGF as a model target and evaluated the efficacy of siRNA electroporation as to tumor growth. We found through this study that the vascularization and VEGF expression in tumors determined the efficiency of delivery of siRNA by electroporation.

A siRNA targeting human VEGF was successfully established in our previous work (11): VEGF siRNA (target bases 189-207 in an open reading frame) 5′-GGAGUACCCUGAUGAGAUCdTdT-3′ (sense) and 5′-GAUCUCAUCAGGGUACUCCdTdT-3′ (antisense). We used a scrambled control of the siRNA: VEGF siRNA-SCR 5′-ACGCGUAACGCGGGAAUUUdTdT-3′ (sense) and 5′-AAAUUCCCGCGUUACGCGUdTdT-3′ (antisense). All siRNAs were synthesized by Dharmacon. In this study, we mainly used siSTABLE siRNA to acquire a nuclease-resistant siRNA without changing any nucleotide sequences. We also used Cy3-labeled siSTABLE VEGF siRNA-SCR to monitor the efficiency of delivery of siRNA via electroporation on injection into tumors grown in nude mice. Each freeze-dried siRNA was reconstituted in RNase-free water to prepare an 80 μmol/L stock solution. VEGF siRNA did not affect the expression of VEGF-related genes, such as those of VEGF-B, VEGF-C, VEGF-D, and placenta growth factor, as reported previously (11).

All cell lines, except for MKN-1, were obtained from the American Type Culture Collection and then cultured according to their instructions. PC-3 cells derived from human prostate adenocarcinomas were cultured in Ham's F-12 medium modified by Kaighn (F-12K) with 10% heat-inactivated fetal bovine serum (FBS). A549 cells derived from human lung carcinomas were cultured under the same conditions in DMEM with 10% FBS. ACHN cells from human renal adenocarcinomas were cultured in MEM with 10% FBS. MKN-1 cells derived from human gastric adenosquamous carcinomas were obtained from the Japanese Collection of Research Bioresources Cell Bank and cultured in RPMI 1640 with 10% FBS.

PC-3 cells were transfected with siSTABLE or unmodified siRNA in serum-free medium using LipofectAMINE PLUS (Invitrogen) as reported previously (11, 22). Finally, the medium was replaced with fresh F-12K with heparin (20 μg/mL). Each conditioned medium was collected for ELISA. Total RNA was also extracted for Northern blot analysis.

Secretion of VEGF into the cell culture supernatants and the tumor contents of VEGF in the xenografts was determined using a Quantikine Human VEGF Immunoassay kit (R&D Systems). Tumors were homogenized in ice-cold CelLytic-MT Mammalian Tissue Lysis/Extraction Reagent with a protease inhibitor mixture (Sigma) and then centrifuged (11). Each supernatant was examined by ELISA.

Total RNA was extracted from PC-3 cells transfected with various siRNAs using a RNeasy Mini kit (Qiagen). The cDNA probe used for Northern hybridization was a 600-bp EcoRI-HindIII cDNA fragment derived from human full-length VEGF₁₆₅ (a kind gift from Dr. Masabumi Shibuya, Institute of Medical Science, University of Tokyo; ref. 23). A cDNA fragment (1,200-bp) for human glyceraldehyde-3-phosphate dehydrogenase was used as an internal control for the amount of RNA in each sample (24). Total RNA (15 μg) was electrophoresed in formaldehyde/1% agarose gels and then transferred to Nylon membranes (Hybond-N⁺ from GE Healthcare Bio-science). All cDNA probes (25 ng) were labeled with [α-³²P]dCTP using a Rediprime II random prime labeling system (GE Healthcare Bio-science). The membranes were hybridized with ³²P-labeled cDNA probes in Amersham Rapid-hyb buffer (GE Healthcare Bio-science) for 6 h at 65°C. The hybridized membranes were washed under high-stringency conditions (2× SSPE containing 0.1% SDS at room temperature for 10 min, 1× SSPE containing 0.1% SDS at 65°C for 20 min, and 0.1× SSPE containing 0.1% SDS at 65°C for 20 min, respectively). Autoradiography was done at −80°C for 2 days. The exposed imaging plates (Fujifilm) were scanned with a Typhoon 9400 (GE Healthcare Bio-science), and the relative VEGF expression values were normalized to the glyceraldehyde-3-phosphate dehydrogenase level in various lanes.

A total of 3.0 × 10⁶ PC-3, A549, ACHN, or MKN-1 cells in 0.3 mL of each serum-free medium were s.c. inoculated through a 24-gauge needle into the lower flanks of 8-week-old athymic nude mice obtained from SLC as described previously (11, 22, 25–27).

A pair of electrodes comprising a plate-type one (anode made of platinum; 3.87 mm wide and 6.67 mm long) and a fork-type one (cathode made of stainless steel; 3 tines of 6.67 mm in length and 1.30 mm apart) was used to conduct the electric current into xenografted tumors (Fig. 1A). The distance between the electrodes was always fixed at 0.77 cm using a screw as a stopper to maintain the same distance. The fork-type electrode was inserted into the tumor, scooping from the bottom of the tumor, and the plate-type electrode was put in contact with the surface of the tumor as shown in Fig. 1B.

Nude mice with s.c. xenografted tumors were deeply anesthetized with pentobarbiturate (50 mg/kg i.p.). A pair of “plate and fork” electrodes was inserted into a tumor, and then the electric resistance in the tumor was measured. When we applied a quite weak alternating current (AC, 30 kHz; 1 mA and 20 ms duration) to tumors through the electrodes, we determined the voltage value. Then, we determined the impedance value (Supplementary Fig. S1).⁵

According to the above conditions, we can regard impedance as resistance, because the inductive reactance and capacitive reactance can be almost ignored when an AC current of a high frequency (30 kHz) is applied (28).⁶ So, we determined the resistance value in tumors. siRNA (10 μL ×2 injections, total 20 μL) at various concentrations in PBS was injected through a 29-gauge needle (0.33 × 13 mm; Terumo) into tumors. Immediately, electric pulses were delivered to each tumor with a Square Electro Porator (CUY21; Nepagene). We used a square pulse of constant voltage throughout electroporation. Three pulses of a preselected voltage followed by three pulses of the same voltage but opposite polarity were applied at pulse duration of 50 ms with a 100-ms gap between pulses (Fig. 1B). During the first pulsing, we measured a difference of electric potential (ΔE, as shown in Supplementary Fig. S1)⁵ between the internal resistance (quite small resistance, 0.1 Ω) preset in the electroporator, and then calculated the current (DC). Therefore, this current value means the one that is actually passed in the tumors.

Tumor-bearing nude mice were treated with siSTABLE VEGF siRNA via electroporation after 3 or 4 weeks when the tumors had reached an average volume of ∼60 to 80 mm³. A dilute siSTABLE VEGF siRNA solution in sterile PBS (20 μL) was directly injected into each xenograft, which was then electronically pulsed with a “plate and fork” electrode. As a control, PBS alone or siSTABLE VEGF siRNA-SCR was injected followed by electronic pulsing. We also examined a therapy group without any injection or electroporation. We repeated the therapy twice every 20 days and monitored the tumor size up to day 40. Tumor diameters were measured at regular intervals with digital calipers, and the tumor volume in mm³ was calculated with the following formula: volume = (width)² × length / 2 (11). All of the animal experiments were done in compliance with the guidelines of the Institute for Laboratory Animal Research, Nagoya University Graduate School of Medicine. In particular, during electroporation for nude mice with the xenografts (that is, insertion of the electrodes), to minimize pain, we did the treatment under deep anesthetized condition with pentobarbiturate. For the therapy procedure, we created an endpoint, when the tumor volume does not exceed 5% of the body weight.

siSTABLE VEGF siRNA or siSTABLE VEGF siRNA-SCR (10 nmol/mouse, 133 μg/mouse body, 200 μL/injection) was i.v. injected in nude mice having s.c. PC-3 xenografts (∼60-80 mm³) under the anesthetized condition. Then, electroporation was immediately done (70 V, 91 V/cm). As a control, PBS alone was injected followed by electroporation. We monitored the tumor size up to day 20.

Immunohistochemical staining for CD31 and microvessel counting of CD31-positive vessels were done as described previously (11, 22, 25).

Cy3-labeled siSTABLE VEGF siRNA-SCR (500 pmol/tumor, 6.7 μg/tumor: 20 μL injection volume) was directly injected into PC-3 tumors, and then the tumors were electrically pulsed (0, 35, 50, 60, 70, 90, and 100 V). After 24 h, the tumors were excised and snap frozen in liquid nitrogen, and then 2-μm-thick sections were cut with a cryostat and washed thrice with PBS to remove the undelivered Cy3-labeled siSTABLE siRNA outside the cells and finally were stained the nuclei with SYTOX Green (Invitrogen). All of the sections were observed under a confocal microscope system (LSM5Pascal version 3.5; Carl Zeiss) to obtain fluorescence (red, Cy3-labeled siSTABLE siRNA; green, nucleus) and differential interference contrast images. To calculate the efficiency (%) of delivery of the Cy3-labeled siSTABLE siRNA, first we determined the total number of cells (SYTOX Green-positive cells plus the cells that were recognized by differential interference contrast). Second, we determined the number of cells that exhibited red fluorescence from Cy3-labeled siSTABLE siRNA. Then, we divided the number of red cells by the total number of cells.

siSTABLE VEGF siRNA-SCR and unmodified VEGF siRNA-SCR were each labeled at the 5′-end with T4 polynucleotide kinase (Takara Bio) and [γ-³³P]ATP (GE Healthcare Bio-science) as reported previously (11). Analysis of ³³P-labeled siRNA by PAGE followed by autoradiography revealed a single band.

³³P-labeled siSTABLE or unmodified VEGF siRNA-SCR (500 pmol/tumor, 6.7 μg/tumor: 20 μL injection volume) was microinjected into PC-3 xenografts, and then the xenografts were pulsed with the optimal voltage determined. On days 5, 10, 15, and 20 after electroporation, the tumor was excised and homogenized as reported previously (11). The supernatant was analyzed by PAGE (15% gel) followed by autoradiography.

Unmodified or siSTABLE VEGF siRNA (5.4 μg/tube) was reacted with 5, 25, or 50% FBS at 37°C. Aliquots of the reaction mixtures were removed at different times. The nuclease reactions were stopped by adding RNase inhibitor (1 units/μL; Roche). All samples were analyzed by PAGE (15% gel) followed by staining with a SYBR Gold Nucleic Acid Gel Stain kit (Molecular Probes). All reagents used for PAGE were of nuclease-free grade.

For calculation of the half-life of each siRNA in serum, all bands were analyzed with an imaging densitometer (GelDoc 1000 system, Bio-Rad) as reported previously (26).

Data were analyzed using the Mann-Whitney U test, and probability values of less than 0.05 were considered to indicate significant differences.

We compared two types of electrodes for in vivo electroporation (that is, “plate and fork” and “plate and plate” electrodes; Fig. 1A). With the “plate and plate” type electrodes, the tumor was placed between the two plates. The tumors for which electroporation was applied had been grown to 60 to 80 mm³ on mouse flanks by s.c. grafting PC-3 cells. The “plate and fork” electrodes conducted a significantly higher electric current than the “plate and plate” ones (P < 0.001) when the same voltage (70 V, 91 V/cm) was applied to the PC-3 tumors (Fig. 1C). Based on the results, we used the “plate and fork” electrodes for further study.

To determine the optimal voltage for delivery of siRNAs into tumors, Cy3-labeled siSTABLE VEGF siRNA-SCR (red in Fig. 2A) was injected into the PC-3 tumors, and then the tumors were pulsed. After 24 h, the tumors were excised and frozen sections were observed under a confocal microscope system. Figure 2A shows that voltages ranging from 50 to 70 V were highly successful for delivering the siRNAs into the PC-3 tumors. Among the conditions, 70 V (91 V/cm) was the best to deliver the siRNAs into tumors (Fig. 2A, bar graph). We observed that many nuclei were missing at voltages between 90 and 100 V, indicating severe damage to tissues. Therefore, we determined 70 V (91 V/cm) to be a safe and functional voltage for delivering the siRNAs into the xenografts. siRNAs were hardly delivered without electroporation (0 V).

When current and resistance were compared in tumors derived from four cancer cell lines, we found that the tumors could be divided into two groups: one conducted a high current (PC-3, A549, and ACHN) and the other conducted a low current (MKN-1; Fig. 2B). It is of note that MKN-1 tumors conducted a lower current than PC-3 ones (Fig. 2A). Consequently, the siRNA delivery efficiency was much poorer for MKN-1 tumors (Fig. 2A). These results collectively suggested that the delivery efficiency correlated to the electric current.

The high current group showed strong staining with CD31 and a red appearance of tumors, which indicate a high microvascular density, whereas the low current-conducting tumors, MKN-1, exhibited weak CD31 staining and a white appearance, which are consistent with a poor microvasculature (Fig. 2C). Furthermore, the high current group cells and tumors showed high expression of VEGF, whereas MKN-1 cells and tumors only expressed small amounts of VEGF (Fig. 2C).

The VEGF contents in tumors significantly correlated to microvascular density and current (P < 0.001) and inversely correlated to resistance (Fig. 3A). The plots in circles in Fig. 3A were obtained for MKN-1 xenografts. Cutoff values differentiating the high current and low current groups could be set: 40 pg/mg protein (VEGF contents in xenografts), 30 (microvascular density), 1,200 Ω (electric resistance), and 0.15A (electric current; Fig. 3A). The electric currents observed in the high current group tumors were 2.4-fold higher than those expected from the data for MKN-1 tumors (Fig. 3B).

A functional siRNA sequence for down-regulating human VEGF in PC-3 cells was determined in our previous study (11). We compared unmodified siRNA and siSTABLE siRNA with the same functional sequence. On transfection into PC-3 cells via lipofection, siSTABLE VEGF siRNA continuously suppressed VEGF secretion to below 5% of the level in scrambled siRNA controls (SCR), at least up to 168 h (7 d) after transfection. On the other hand, unmodified VEGF siRNA gradually lost the RNA interference effect (Fig. 4A). We found that siSTABLE VEGF siRNA had decreased the mRNA level of VEGF in PC-3 cells at 168 h after transfection by Northern blot analysis (Fig. 4B). siSTABLE siRNA did not show any cytotoxicity toward PC-3 cells at 1,000 nmol/L at least up to 96 h after transfection (data not shown).

Figure 4C and D reveals the high stability of siSTABLE VEGF siRNA even with 50% FBS. In contrast, unmodified VEGF siRNA was dramatically degraded in a FBS dose-dependent manner. The half-life of each siRNA determined by densitometric analysis is shown (Fig. 4C). Thus, siSTABLE VEGF siRNA seemed to be considerably more stable in an environment where nucleases were ubiquitous. To confirm this, ³³P-labeled siSTABLE VEGF siRNA-SCR was injected into PC-3 xenografts, and then the xenografts were pulsed at 70 V (91 V/cm). ³³P-labeled siSTABLE VEGF siRNA-SCR remained intact in tumors for at least 20 days after electroporation, whereas unmodified siRNA had almost completely disappeared by around 5 days after electroporation (Fig. 4E). Similar results were obtained using Cy3-labeled siSTABLE VEGF siRNA-SCR (data not shown). Therefore, we decided to use siSTABLE siRNA for further study. These results also indicated a condition for electroporation (that is, the interval for tumor therapy was determined to be 20 days).

The conditions for in vivo siRNA electroporation were established based on the results so far obtained. We then chose VEGF as a model target to evaluate the efficacy of our method as to tumor growth, because we knew that VEGF siRNA was effective for the suppression of PC-3 tumors (11). siSTABLE VEGF siRNA was delivered into pregrown PC-3 tumors via electroporation (Fig. 5A). This treatment markedly suppressed tumor growth compared with siSTABLE VEGF siRNA-SCR or PBS (P < 0.001; Fig. 5A). These growth inhibitory effects were dependent on the dose of the siRNA (Fig. 5A). It is of note that the siSTABLE VEGF siRNA at 250 pmol/tumor (3.3 μg/tumor) sufficiently inhibited tumor growth on injection twice with a 20-day interval, whereas the previously reported atelocollagen-mediated delivery method needed unmodified siRNA at 500 pmol/tumor (6.7 μg/tumor) and four injections at 10-day intervals (11). No gross adverse effects, for example, loss of body weight, were observed during the experimental period (data not shown).

To examine the therapeutic effects, we stained intratumor vessels with CD31/PECAM-1-specific antibodies. Dramatically low microvascular densities were observed in tumors treated with siSTABLE VEGF siRNA on 10, 20, and 30 days after the first therapy, whereas the tumors treated with siSTABLE VEGF siRNA-SCR (a control siRNA) showed higher microvascular densities (Fig. 5B). On day 40, the tumor weight was significantly lower for tumors with siSTABLE VEGF siRNA than those for other controls (Fig. 5C). Furthermore, the decrease in the VEGF content in tumors treated with siSTABLE VEGF siRNA correlated to the injected dose of the siRNA and was much lower than those for other controls (Fig. 5C), this being consistent with the results in Fig. 5A and B.

Xenografted tumors derived from the various cells examined in Fig. 2B and C were also examined to further evaluate our electroporation system, because these tumors exhibited various microvascular densities and VEGF contents. Tumors of PC-3, A549, and ACHN cells, which had a rich microvasculature and high VEGF contents (Fig. 2C), responded very well to the in vivo electroporation with siSTABLE VEGF siRNA (Fig. 6A). However, the poor microvasculature tumors of MKN-1 cells did not respond at all (Fig. 6A).

PC-3 tumors were subjected to electroporation immediately after i.v. injection (via a tail vein) of siSTABLE VEGF siRNA (10 nmol/mouse body, 133 μg/mouse body, 200 μL/injection). As shown in Fig. 6B, siSTABLE VEGF siRNA administered via a tail vein had an antitumor effect, although a 40-fold amount of the siRNA was needed compared with local injection of the siRNA.

Drug target discovery, which involves the identification and early validation of disease-modifying targets, is an essential first step of the drug development pipeline (3). We attempted to apply the RNA interference strategy to in vivo screening for molecular targets for cancer therapy. Our strategy of siRNA delivery via electroporation showed sufficient delivery efficiency and resulted in significant suppression of tumor growth when done with VEGF siRNA. The electroporation method revealed an antitumor effect almost equivalent to the previously reported one with atelocollagen (11). However, we could reduce the total dose of siRNA to 25% and the treatment interval to 50% with the proposed method compared with the atelocollagen-mediated method. Thus, 6.6 μg siSTABLE VEGF siRNA (twice, 20-day interval) was needed to treat the PC-3 tumor via the electroporation method during the period of therapy (40 days), whereas 26.4 μg unmodified VEGF siRNA (four times, 10-day intervals) was needed via the atelocollagen-mediated method (11). Therefore, this electroporation method is convenient enough for in vivo comprehensive screening for molecular targets.

Electroporation, as a gene delivery strategy, initially comprised the combination of naked DNA injection and pulsed electric field treatment (29, 30). The success of in vitro delivery by electroporation has led to the development of in vivo applications (31–33). This nonviral gene delivery approach allows remarkably efficient transgene expression, and the expression can be long lived in the case of muscle (34). However, the electroporation apparatuses used in these previous studies involved “plate and plate” tweezers-type, caliper-type, or needle-type electrodes (35), which did not appear to be applicable to xenografted tumors for highly efficient delivery. The “plate and fork” electrodes, which consisted of one plate and some tines of fork, were originally devised by Maruyama et al. (15). These electrodes were used for the transfer of a plasmid vector for erythropoietin expression into rat skin and were compared with needle-type and disc-type electrodes (15). We modified the electrodes to be suitable for treating cancer xenografts on the lower flanks of nude mice. Our electrodes comprising one plate and three tines of fork decreased the electrical resistance in tumors, which enabled us to obtain not only a higher current with a lower voltage but also a current stream within a broader area in tumors (Supplementary Fig. S2).⁵ Harada et al. used 10 pulses of a higher voltage (150 V/cm) and duration (50 ms) to deliver murine interleukin-12 plasmid injected into hepatocellular carcinomas using the same electroporator (CUY 21, Nepa Gene Co., Ltd.) but different electrodes consisting of a pair of needles (36). They applied severer electroporation conditions than ours because of the quite high resistance (1,620 ± 90 Ω) in hepatocellular carcinomas with their needle-type electrodes (37). Moreover, such severe electroporation conditions seem to be needed to deliver plasmid DNA into the tumors probably because they are larger molecules than siRNAs. Our electroporation conditions (70 V, 91 V/cm) also delivered enhanced green fluorescent protein–expressing plasmid DNA into the tumors (efficiency of delivery ∼49%).⁷

Therefore, the electroporation conditions for significant efficacy vary with the molecule to be delivered. In general, there are differences in effective variables between a drug and a gene for delivery by electroporation. High field strength and a short pulse length gave good results, at least with some of the drugs investigated (e.g., bleomycin), whereas electroporation for genes benefits from a combination of a low electric field and a long pulse length (38). Heller et al. reported the electroporation conditions (8 pulses of 1,300 V/cm, 0.099 ms duration, and 1 pulse/s) to effectively deliver bleomycin into tumors (39).

Our study showed that the delivery efficiency depended on the current and that a high microvascular density was required for a high current. Thus, this study revealed the conditions for the application of this electroporation method to target tumors: the electric resistance should be below 1,200 Ω and the current should be over 0.15 A to obtain sufficient siRNA delivery. Interestingly, the current obtained in tumors exhibiting high microvascular density was more than 2-fold that expected from the data for those exhibiting low density. This indicates the importance of the microvascular bed for the electric current and suggests that there is a vascular bed volume threshold that guarantees a high current and efficient delivery. In addition, the finding that a high microvascular density was required for a high current led to another finding, that is, the VEGF content correlated to vascular density and current. Thus, this report showed for the first time the close relationships among microvascular density, electric current, resistance, VEGF content, and efficiency of delivery of siRNA via electroporation.

We chose VEGF as a model target to evaluate our electroporation system. There is compelling evidence that uncontrolled angiogenesis is a major contributing factor in both tumor growth and metastasis (40–42). Several growth factors have been identified as potential positive regulators of angiogenesis. Among them, VEGF is the only growth factor most consistently found in a wide variety of conditions associated with angiogenesis (43). Inhibition of VEGF activity or disabling of the function of its receptors has been shown to inhibit both tumor growth and metastasis in a variety of animal tumor models (44–47). Consistent with this, the present study showed that VEGF siRNA electroporation significantly suppressed the growth of tumors exhibiting high VEGF expression but not that of those exhibiting low expression. As siRNA for VEGF was effective for tumors of all tested cell lines highly expressing VEGF, our data strongly support the idea that VEGF is a general target for cancer therapy. The use of siRNA, even modified to increase stability but eventually eliminated from the body, might not be the best choice. To obtain the RNA interference effect targeting VEGF, there is another way involving a short hairpin RNA–expressing vector, which propagates as an integrated gene on delivery. This strategy might solve the dilemma that VEGF inhibition in tumors leads to not only less vascularization but also less delivery efficiency via electroporation at the same time. We obtained the efficiency (49%) of delivering plasmid DNA via electroporation, indicating that our electroporation conditions (70 V, 91 V/cm) are acceptable for delivering them into tumors. However, one should also take into consideration that the integration of a foreign plasmid DNA into a host gene is dangerous, especially when used in humans.

Finally, this study showed that a combination of systemically administered VEGF siRNA and local electroporation significantly suppressed the growth of PC-3 tumors. This method is far from clinical application but provides an intriguing possibility for future cancer therapy. For example, microelectrodes, transvascularly set near multiple cancerous lesions in an organ, will facilitate the delivery of siRNA, which is administered via a main artery dominating the organ. Because electroporation is superior to other delivery methods, as revealed in this study, the development of such technologies as microelectrodes will be beneficial for cancer therapy.

We thank Yuriko Fujitani for the excellent technical assistance; Drs. Kosho Suzuki, Takahiro Yamauchi, and Kazuo Kita for the helpful suggestions regarding the experiments; Dr. Masabumi Shibuya for generously providing the cDNA probe to human VEGF used for Northern hybridization; and Dr. Toshiyuki Moriizumi, who designed the electroporator (CUY21), for providing a schema of the wiring diagram as shown in Supplementary Fig. S1.⁵