The role of electrophoresis played in electric field-mediated gene delivery in vivo
remains a subject of considerable debate. Some studies suggest that electrophoresis plays a critical role in gene transfer following electropermeabilization (18
), and others suggest that electrophoresis is not even involved (36
). Adding to this ambiguity, few studies have distinguished between electrophoresis through porous matrix compartments (e.g.
, interstitium and cytosol) and across membrane barriers (e.g.
, cell membrane and nuclear envelope). In a recent study, the magnitude of in vivo
electrophoresis in tumor interstitium was quantified directly for the first time (2
). High-energy electrophoretic pulses (400 V/cm, 50 ms) were able to induce interstitial transport but the total distance after application of 10 pulses was less than 2.5 µm, or approximately one fourth the diameter of a typical tumor cell, in either 4T1 or B16.F10 tumor. The short distance could be caused by collagen fibers since previous studies have demonstrated that collagen matrix is the major barrier to interstitial macromolecular transport (19
). Therefore, alterations in the collagen component of tumor ECM may improve electric field-induced interstitial pDNA transport, thereby improving efficiency of pDNA delivery. To investigate this possibility, effects of rh-Rlx on tumor collagen contents were measured in this study. It was observed that the collagen contents were reduced in 4T1 and B16.F10 tumors grown in the hind leg of rh-Rlx treated mice. However, the collagen contents of 4T1 and B16.F10 tumors were increased in most tumors transplanted in DSCs (see ). This finding was consistent with another report published in the literature, in which rh-Rlx treatment did not reduce the overall collagen levels in DSC tumor models (34
). The dependency of rh-Rlx-induced changes in collagen content on tumor type and location of tumor transplantation may be attributed to the following mechanisms.
Firstly, the developmental stage of the tumors varied between the hind leg and DSC tumor models. Both tumor models were seeded by a subcutaneous injection of tumor cell suspensions. However, the hind leg tumors were ~5 mm in diameter whereas the DSC tumors were only ~2 mm in diameter at the initiation of rh-Rlx treatment. This difference in initial tumor age and size might have translated into the difference in intrinsic rates of collagen metabolism between the two tumor models, and therefore different tumor responses to rh-Rlx treatment. Secondly, the intrinsic level of collagen in 4T1 tumors grown on the hind leg was much higher than that in the DSC although there was < 2-fold difference in the collagen contents for B16.F10 tumors grown at different places (see ). This discrepancy could be caused by various factors in tissue microenvironment, which are dependent on tumor type and site of transplantation (25
). This is because tumor ECM is largely produced by host stromal cells (38
). Due to their different anatomical locations, DSC and hind leg tumors may have developed different ECM. Thirdly, the anatomical location of the allograft may have influenced the extent of tumor vascularization. While the systemic rh-Rlx dosage was equal for both tumor models, the local concentration of rh-Rlx depended on tumor vasculature. Hind leg tumors were implanted on the highly vascularized quadricep muscle, and therefore might have had the opportunity to recruit significantly more neovascularization during the course of the study than DSC tumors, which were established in a less vascularized DSC environment. A highly vascularized hind leg tumor would have received a greater rh-Rlx dose than the less vascularized DSC tumor, which might have accounted for the greater effect of rh-Rlx treatment on the collagen content of the hind leg tumors. Finally, the net change in the collagen content depend on the interplay between collagen synthesis and collagen degradation (27
). Both can be upregulated by relaxin. As a result, the relaxin treatment might have reduced the collagen fiber length but not necessarily the total content (see ) (34
The collagen contents in tumors were correlated with the efficiency of gene transfer. In the hind leg tumors, where the collagen content was reduced significantly, rh-Rlx treatment increased the mean transgene expression in both 4T1 and B16.F10 tumors (see ). The collagen content, and the content reduction, were greater in 4T1 tumors that in B16.F10 tumors, which might attribute to the level of significance in the improvement of transgene expression since the improvement was statistically significant only in 4T1 tumors. It was likely that the extent of improvement in B16.F10 tumors was less than inter-tumoral data variation, caused presumably by tumor heterogeneity (see ), so that the improvement was statistically insignificant. In light of the dependence of gene transfer on pDNA interstitial transport, which in turn depended on tissue collagen content, it was likely that the increased transgene expression in rh-Rlx treated hind leg tumors was the result of improved electromobility of pDNA. Relaxin treatment alone, without electric field treatment, did not result in any increase in transgene expression, indicating that rh-Rlx treatment did not directly influence gene transfer.
Without exposure to external electric fields, transgene expression was significantly greater in 4T1 tumors than B16.F10 tumors, in both untreated and rh-Rlx treated mice (P < 0.05). This difference might be attributed partly to endocytosis or macropinocytosis of pDNA by various cells in these tumors since they were the only possible mechanisms known for gene transfer via naked pDNA (36
). The amount of endocytosed or macropinocytosed pDNA depends on the number of cells that are close to pDNA infused into tumors. Therefore, parameters for intratumoral infusion of pDNA might play an important role in determining the transgene expressions shown in (42
). The infusion conditions used in this study had previously been optimized for maximizing the spatial distribution of gene vectors in 4T1 hind leg tumors (46
). As a result, it was likely that pDNA was available to more cells in 4T1 tumors than in B16.F10 tumors. Following the exposure to external electric field, the difference in transgene expression between the two tumors was no longer significant (P > 0.05), showing the electric field to be effective in reducing some heterogeneity observed across tumor types.
In the DSC tumor model, where the collagen content was slightly increased, rh-Rlx treatment did not increase the electromobility of pDNA for any of the four electric fields investigated (see ). This observation was consistent with previous studies correlating interstitial pDNA electromobility with the collagen content (19
), but differed from that in another study where rh-Rlx treatment enhanced diffusion of IgG and dextran-2M in DSC tumors (34
). The enhancement was attributed to the reduction in collagen fiber length instead of total collagen content, which loosened fiber matrix structures and increased pore size in extracellular matrix. As a result, the resistance to diffusion of macromolecules was reduced (34
). In this study, it was possible that the pore size was not adequately increased for pDNA transport after relaxin treatment since pDNA is larger than IgG and dextran. More importantly, pDNA is a highly negatively charged molecule and its transport can be significantly hindered by charge-charge interactions with collagen fibers. These interactions were likely to be a main mechanism for explaining the lack of improvement in pDNA electromobility in relaxin treated tumors grown in DSC.
To mimic the charge-charge interactions, passive transport of pDNA in collagen gel was investigated and compared to that in agarose gel. In agarose gel, where only steric interactions between Rho-pDNA and agarose fibers were expected, Rho-pDNA transport followed the expected pattern of free diffusion through a porous matrix. However, Rho-pDNA diffusion through the collagen gel indicated that binding of Rho-pDNA to collagen had occurred because the diameter of Rho-pDNA distribution volume quickly reached a plateau after injection into the gel (see ). When unlabeled pDNA was added, Rho-pDNA was able to diffuse a greater distance from the site of injection. This behavior was characteristic of the unlabeled pDNA competing with the Rho-pDNA for available binding sites on collagen fibers. By blocking these binding sites, the unlabeled pDNA allowed the Rho-pDNA to distribute to a greater volume around the injection site. pDNA-collagen binding has been suggested previously in studies when investigating the use of implantable pDNA-embedded collagen matrices for sustained release (47
). Cohen-Sacks et al.
showed that 60–80% of embedded pDNA was released from the collagen matrix within 2 h when submerged in TE buffer. The remaining pDNA was not released from the collagen gel until SDS detergent was added to the buffer. The authors proposed that SDS was required to break an electrostatic interaction between pDNA and the free lysine groups in the collagen (47
). Binding of pDNA to collagen fibers might explain the apparent discrepancy between the findings reported here, and those reported previously by Brown et al.
), because IgG and dextran-2M could not bind to collagen strongly. The binding phenomenon might also explain why rh-Rlx treatment resulted in no increase in pDNA mobility in 4T1 tumors in the DSC model (see ) but an enhancement in transgene expression in the hind leg model (see ), suggesting that it would be the collagen content, regardless of its matrix structures, that determined the hindrance to pDNA transport.
In summary, results from this study demonstrated that effects of rh-Rlx on the collagen content depended on microenvironment in tumor tissues and that rh-Rlx treatment would enhance electric field-mediated gene delivery only if it could effectively reduce the collagen content in collagen-rich tumors. The results also suggested that the mechanism of enhancement was due to improvement in interstitial transport of pDNA. In future studies, the correlation between increased gene transfer and improved interstitial transport needs to be verified directly in certain DSC tumor models in which the collagen content can be significantly reduced by either infecting tumors with adenoviral vector for relaxin (35
) or co-injecting collagenase with pDNA into tumors (48