Stromelysin generates a fibronectin fragment that inhibits Schwann cell proliferation.

Our previous report (Muir, D., S. Varon, and M. Manthorpe. 1990. J. Cell Biol. 109:2663-2672) described the isolation and partial characterization of a 55-kD antiproliferative protein found in Schwann cell (SC) and schwannoma cell line-conditioned media and we concluded that SC proliferation is under negative autocrine control. In the present study the 55-kD protein was found to possess metalloprotease activity and stromelysin immunoreactivity. The SC-derived metalloprotease shares many properties with stromelysin isolated from other sources including the ability to cleave fibronectin (FN). Furthermore, limited proteolysis of FN by the SC-derived protease generated a FN fragment which itself expresses a potent antiproliferative activity for SCs. The active FN fragment corresponds to the 29-kD amino-terminal region of the FN molecule which was also identified as an active component in SC CM. Additional evidence that a proteolytic fragment of FN can possess antiproliferative activity for SCs was provided by the finding that plasmin can generate an amino-terminal FN fragment which mimicked the activity of the SC metalloprotease-generated antiproliferative FN fragment. Both the 55-kD SC metalloprotease and the 29-kD FN fragment could completely and reversibly inhibit proliferation of SCs treated with various mitogens and both were largely ineffective at inhibiting proliferation by immortalized or transformed SC lines. Normal and transformed SC types do secrete the proform of stromelysin, however, transformed cultures do not produce activated stromelysin and thus cannot generate the antiproliferative fragment of FN. These results suggest that, once activated, a SC-derived protease similar to stromelysin cleaves FN and generates an antiproliferative activity which can maintain normal SC quiescence in vitro.

Schwannoma cell-derived inhibitor of the neurite-promoting activity of laminin.

During the purification of laminin-proteoglycan complexes from rat RN22 Schwannoma cell-conditioned medium, a laminin-rich fraction was obtained which lacked neurite-promoting activity. Since laminin from several sources is known to have potent neurite-promoting activity, this result suggested that either this laminin was inactive or its activity was somehow masked by associated molecule(s). The latter possibility was supported by the demonstration that the inactive laminin-containing fraction inhibited active laminin-containing fractions. This inhibitory activity was partially purified by using ion exchange chromatography and isopycnic centrifugation. The purified material contained proteoglycan based on its high affinity for cationic resin, high buoyant density, large heterodisperse appearance on electrophoretic gels, ability to label with inorganic sulfate, sensitivity to trypsin and glycosaminoglycan lyases, and heat stability. A quantitative in vitro bioassay was used to monitor the inhibitor after treatments aimed at defining its activity. The isolated Schwannoma-derived inhibitor (a) inhibits the neurite-promoting activity of purified rat, mouse, and human laminin; (b) is active whether presented to laminin in solution or after either the inhibitor or laminin is first bound to the culture substratum; (c) does not act by displacing laminin from the substratum; (d) can be prevented from binding to neurite-promoting laminin substrates by polyclonal and monoclonal anti-laminin or polyclonal anti-entactin antibodies; and (e) is abolished by proteases or glycosaminoglycan lyases but not by heat. The above results suggest that the neurite-promoting activity of laminin is subject to regulation through association with a proteoglycan and entactin.

Schwann cell proliferation in vitro is under negative autocrine control.

In healthy adult peripheral nerve, Schwann cells are believed to be generally quiescent. Similarly, cultures of isolated rat sciatic nerve Schwann cells hardly proliferate in serum-supplemented medium. The possibility that Schwann cells negatively regulate their own proliferation was supported by the demonstration that conditioned media from Schwann cell cultures inhibited the proliferation of mitogen-stimulated test cultures. The inhibition could be complete, was dose dependent, and was exhibited when the test Schwann cells were under the influence of different types of mitogens such as cholera toxin, laminin, and living neurons. The inhibition of proliferation was completely reversible and a rapid doubling of cell number resulted when treatment with conditioned medium was withdrawn from mitogen-stimulated Schwann cells. Conditioned medium from cholera toxin-stimulated and immortalized Schwann cell cultures contained less antiproliferative activity than that found in medium from quiescent Schwann cell cultures. However, media conditioned by two actively proliferating rat Schwannoma cell lines were rich sources of antiproliferative activity for Schwann cells. Unlike the mitogen-stimulated Schwann cells, whose proliferation could be inhibited completely, the immortalized and transformed Schwann cell types were nearly unresponsive to the antiproliferative activity. The antiproliferative activity in Schwann and Schwannoma cell conditioned media was submitted to gel filtration and SDS-PAGE. The activity exists in at least two distinct forms: (a) a high molecular weight complex with an apparent molecular mass greater than 1,000 kD, and (b) a lower molecular weight form having a molecular mass of 55 kD. The active 55-kD form could be derived from the high molecular weight form by gel filtration performed under dissociating conditions. The 55-kD form was further purified to electrophoretic homogeneity. These results suggest that Schwann cells produce an autocrine factor, which we designate as a "neural antiproliferative protein," which completely inhibits the in vitro proliferation of Schwann cells but not that of immortalized Schwann cells or Schwannoma lines.

Mapping of domains in human laminin using monoclonal antibodies: localization of the neurite-promoting site.

Monoclonal antibodies were made against a truncated form of human laminin isolated from placenta. 12 antibodies were isolated and characterized. All antibodies stained basement membranes in placenta and immunoprecipitated laminin from media of cultured choriocarcinoma cells. Three antibodies, 3E5, 4C7, and 4E10, partially blocked the neurite-promoting activity of laminin. Addition of a second antibody, goat anti-mouse IgG, caused more complete blocking of the activity. Two of the blocking antibodies, 4C7 and 4E10, reacted with epitopes within the globular domain at the end of the long arm of laminin, and the third one, 3E5, reacted at the end of the rod-like portion of the long arm adjacent to the globular domain, as shown by electron microscopy after rotary shadowing. Five nonblocking antibodies used in the same test reacted with epitopes in other domains of the molecule. Blocking antibodies 3E5 and 4E10 could be used in immunoblotting and both antibodies reacted with the same polypeptides in pepsin fragments of human laminin, the predominant polypeptides being approximately 400 kD. When a crude extract of human amnion was used as a source of intact laminin, the 4E10 antibody detected a single polypeptide of approximately 400 kD. A nonblocking antibody, 2E8, which reacted at the center of the laminin cross, reacted predominantly with a 200-kD polypeptide in human laminin fragments and exclusively with a 200-kD polypeptide in amnion extract and in rat laminin. Our results with human laminin match the results by Edgar, D., R. Timpl, and H. Thoenen, 1984, EMBO (Eur. Mol. Biol. Organ.) J., 3:1463-1468, in which the neurite-promoting activity of mouse laminin resides at the end of the long arm, which is also the site for heparin binding. However, since the active fragments of human laminin did not bind to heparin, the neurite-promoting site should be different from the heparin-binding site. Our results further suggest that the neurite-promoting site may be contained in or close to the 400-kD component of laminin.

Laminin promotes neuritic regeneration from cultured peripheral and central neurons.

The ability of axons to grow through tissue in vivo during development or regeneration may be regulated by the availability of specific neurite-promoting macromolecules located within the extracellular matrix. We have used tissue culture methods to examine the relative ability of various extracellular matrix components to elicit neurite outgrowth from dissociated chick embryo parasympathetic (ciliary ganglion) neurons in serum-free monolayer culture. Purified laminin from both mouse and rat sources, as well as a partially purified polyornithine-binding neurite promoting factor (PNPF-1) from rat Schwannoma cells all stimulate neurite production from these neurons. Laminin and PNPF-1 are also potent stimulators of neurite growth from cultured neurons obtained from other peripheral as well as central neural tissues, specifically avian sympathetic and sensory ganglia and spinal cord, optic tectum, neural retina, and telencephalon, as well as from sensory ganglia of the neonatal mouse and hippocampal, septal, and striatal tissues of the fetal rat. A quantitative in vitro bioassay method using ciliary neurons was used to (a) measure and compare the specific neurite-promoting activities of these agents, (b) confirm that during the purification of laminin, the neurite-promoting activity co-purifies with the laminin protein, and (c) compare the influences of antilaminin antibodies on the neurite-promoting activity of laminin and PNPF-1. We conclude that laminin and PNPF-1 are distinct macromolecules capable of expressing their neurite-promoting activities even when presented in nanogram amounts. This neurite-promoting bioassay currently represents the most sensitive test for the biological activity of laminin.

Cholinergic neuronotrophic factors: intraocular distribution of trophic activity for ciliary neurons.

Chick ciliary ganglionic neurons require an interaction with their peripheral targets for survival during a critical period of their embryonic development in vivo. It has recently been shown that survival of these neurons in dissociated cell cultures is supported by extract from whole chick embryo. In this study, an assay system based on microwell cultures of ciliary ganglionic neurons was used to demonstrate that a very rich source of trophic factor for them is the intraocular target tissues they innervate. Out of 8000 trophic units present in a 12-day embryo, 2500 were contained in the eye. A subdissection of the eye showed its activity to be localized in a fraction containing the ciliary body and choroid coat, with a specific activity almost 20-fold higher than that of the whole embryo. This selective intraocular distribution at a time when survival or death of ciliary ganglionic neurons is decided in vivo suggests that this soluble factor may be involved in the normal development of the ciliary ganglion.

Human amnion membrane serves as a substratum for growing axons in vitro and in vivo.

The epithelial cell layer of human amnion membrane can be removed while the basement membrane and stromal surfaces remain morphologically intact. Such a preparation has been used as a substratum for the in vitro culture of dissociated neurons. Embryonic motor neurons from chick ciliary ganglion attached to both surfaces but grew extensive neurites only on the basement membrane. On cross sections of rolled amnion membranes, regenerating axons of cultured neurons were guided along pathways of basement membrane that were immunoreactive with an antibody to laminin. In addition, when rolled amnion membranes were implanted into a lesion cavity between the rat septum and hippocampus, cholinergic neurons extended axons through the longitudinally oriented implant into the hippocampus. Thus, this amnion preparation can serve as a bridge to promote axonal regeneration in vivo in damaged adult brain.

Brain injury causes a time-dependent increase in neuronotrophic activity at the lesion site.

A cavity was made in the brain (entorhinal cortex) of developing or adult rats, and a small piece of Gelfoam was emplaced to collect fluid secreted into the wound. The neuronotrophic activity of the fluid was assayed with sympathetic and parasympathetic neurons in culture. The results show that wounds in the brain of developing or adult rats stimulate the accumulation of neuronotrophic factors and that the activity of these factors increases over the first few days after infliction of the damage.

Laminin-like antigen in rat CNS neurons: distribution and changes upon brain injury and nerve growth factor treatment.

Using several antibodies against rat or human laminin and an avidin-biotin immunocytochemical protocol, laminin-like immunoreactivity was detectable in the rat nervous system in expected locations, i.e., associated with blood vessels and reactive astrocytes. However, laminin staining was also abundantly present within neuronal cell bodies in most parts of the developing and adult rat CNS. Medial septum neuronal immunoreactivity was lost after septo-hippocampal disconnection, but could be preserved or even restored by intraventricular administration of nerve growth factor. Thus, at least for medial septum neurons, this laminin-like molecule can be accumulated or produced independent of direct hippocampal (target) contact. It remains to be determined whether CNS neuronal "laminin" processes activities similar to those found for laminin in vitro.

Cyclic nucleotide phosphodiesterases associated with bovine retinal outer-segment fragments.

ATP-dependent cyclic GMP phosphodiesterase activity (EC 3.1.4.16) associated with bovine retinal outer-segment fragment preparations was stimulated an order of magnitude by light, confirming the results of Miki et al. (1973) Proc. Natl. Acad. Sci. U.S. 70, 3820-3824 at Yale for the frog system. In contrast to the results of the Yale group, however, light stimulation was not observed for cyclic AMP as substrate. A direct relationship of bovine rhodopsin bleaching to phosphodiesterase activation differs from a previous report by the Yale group that full activation of the frog enzyme was achieved by bleaching of a maximum of 2% rhodopsin. Phosphodiesterase activity could be qualitatively removed from the fresh outer-segment preparations with isotonic sucrose which apparently did not disrupt the plasmalemma or discs. Activity recovered from the washing was not light sensitive. Two Km values were determined for cyclic AMP, 5 and 0.05 mM; for cyclic GMP a Km of 0.22 mM was found. All Km values were determined in the presence of 1 mM ATP in the dark. Sonication of fresh outer segments or storing at -20 degrees C abolished the light response. However, storage at -76 degrees C fully preserved it.

Converting an alcohol to an amine in a cationic lipid dramatically alters the co-lipid requirement, cellular transfection activity and the ultrastructure of DNA-cytofectin complexes.

Cytofectins are positively charged lipophilic molecules that readily form complexes with DNA and other anionic polynucleotides. Normally, cytofectins are combined with an activity-augmenting phospholipid such as dioleoylphosphatidylethanolamine (DOPE), and a film of dried, mixed lipid is prepared and hydrated to form cationic liposomes. The liposome solution is then mixed with a plasmid DNA solution to afford cytofectin-DNA complexes which, when presented to living cells, are internalized and the transgene is expressed. One of the most potent cytofectins, dimyristoyl Rosenthal inhibitor ether (DMRIE), is presently being used to deliver transcriptionally active DNA into human tumor tissues. Here we report the remarkable consequences of replacing the alcohol moiety of DMRIE with a primary amine. The resulting cytofectin, called beta-aminoethyl-DMRIE (betaAE-DMRIE), promoted high level transfection over a broad range of DNA and cationic lipid concentrations. A comparison of in vitro transfection activity between DMRIE and betaAE-DMRIE in 10 cell types revealed that betaAE-DMRIE was more active than DMRIE, and that betaAE-DMRIE, unlike DMRIE, was maximally effective in the absence of colipid. The consequences of the alcohol-to-amine conversion on the structure of the cytofectin/DNA complex was also examined by Atomic Force Microscopy. Strikingly dissimilar images were found for plasmid DNA alone and for the plasmid complexes of betaAE-DMRIE and DMRIE/DOPE.

Enhancer and promoter chimeras in plasmids designed for intramuscular injection: a comparative in vivo and in vitro study.

Enhancers and promoters from various muscle-specific genes were substituted for or combined with the enhancer/promoter of the human cytomegalovirus (CMV) IE gene in a luciferase reporter gene plasmid in an effort to identify new promoter chimeras with increased expression activity after direct intramuscular injection. The regulatory sequence substitutions or additions varied in content, location, and orientation relative to the CMV regulatory sequences. The expression activities of the derivative and parent plasmids were compared quantitatively in vivo using a standard mouse intramuscular injection assay, and in vitro by transfection of differentiated C2C12 mouse myoblasts and BHK hamster kidney cells, to test whether cultured cell transfection could substitute for at least some animal experimentation. In vivo, 1 of 19 of the enhancer/promoter chimeras increased expression levels. In vitro, some chimeras showed significant expression augmentation in C2C12 cells, but not in BHK cells. We conclude that because of differences in plasmid expression profiles, these cell culture systems cannot readily substitute for in vivo testing of new plasmid constructs.

Gene therapy by intramuscular injection of plasmid DNA: studies on firefly luciferase gene expression in mice.

Direct injection of nonviral, covalently closed circular plasmid DNA into muscle results in expression of the DNA in myofiber cells. We have examined the expression of firefly luciferase DNA constructs injected into adult murine skeletal muscle. Considerable variation in luciferase enzyme expression was noted among constructs with different regulatory elements, among different batches of the same DNA construct, and among similar transfection experiments performed at different times. This variation was minimized by using single batches of plasmid DNA and by performing comparable sets of experiments concurrently. A quantitative experimental protocol was defined for comparing various aspects of the transfection process. We report that a luciferase construct containing the human cytomegalovirus immediate-early gene promoter plus intron A (a construct termed "p-CMVint-lux") showed the highest expression among several constructs tested. Dose-response and time course analyses of p-CMVint-lux DNA injections showed that maximal luciferase expression was achieved with 25 micrograms of DNA at 7-14 days post-injection. Selected manipulations of the transfection process were examined for their influence on luciferase expression. Variations in the rate of DNA injection, needle size, injection volume, and vehicle temperature had no significant effect on luciferase expression. The presence of endotoxin, cationic peptide, muscle stimulants or relaxants, vasoconstrictors, metal chelators, or lysosomal lytic reagents had no significant effect on expression. However, linearization of the DNA, injection of the DNA in water rather than saline, or inclusion of a DNA intercalating agent nearly abolished luciferase expression. And finally, increasing the injection dose by giving multiple injections over a 10-day period increased expression proportionally to the number of injections.

Cancer gene therapy using plasmid DNA: safety evaluation in rodents and non-human primates.

To evaluate the safety of a plasmid DNA-lipid complex, a series of good laboratory practice (GLP) safety studies were conducted with VCL-1005, a plasmid DNA expression vector containing both the human class I MHC HLA-B7 heavy-chain and the beta 2-microglobulin (beta 2m) light-chain genes formulated with the cationic lipid, DMRIE/DOPE. In mice, the repeated intravenous injection of VCL-1005 at plasmid DNA doses of 0.1, 1.0, or 10 micrograms for 14 days had only incidental effects on clinical chemistry and hematology, and did not result in any organ pathology. Repeated intrahepatic injections of VCL-1005 in mice did not result in significant liver histopathology or significant alterations in liver enzymes. In cynomolgus monkeys, the repeated intravenous administration of VCL-1005 at a cumulative dose of 720 micrograms of DNA had no effects on clinical chemistry, hematology, or organ pathology. Thus, systemic administration of a plasmid DNA expression vector containing the coding sequence for a foreign MHC class I molecule did not result in significant toxicity or a pathological immune response in animals. These results suggest that the direct transfer of VCL-1005, a plasmid DNA-lipid complex, could be used for the safe in vivo delivery of recombinant DNA for a cancer gene therapy trial.

An improved plasmid DNA expression vector for direct injection into skeletal muscle.

In previous work, the direct injection of 50 micrograms of a plasmid DNA vector encoding firefly luciferase (VR1205) into murine quadriceps muscle produced an average of 6.5 ng of luciferase per muscle at 7 days postinjection. In this report, various elements of the VR1205 vector were modified to increase gene expression levels or to eliminate undesired viral sequences. Expression of the modified vectors was then compared to VR1205 using the intramuscular injection assay. In general, modifications to promoter, enhancer, and intronic sequences either decreased luciferase expression levels or had no effect. However, modifications to the polyadenylation and transcriptional termination sequences, plasmid backbone elements, and the luciferase gene itself each increased luciferase expression levels. The best-expressing vector, designated VR1255, contained a combination of these incrementally beneficial changes. A single intramuscular injection of 50 micrograms of VR1255 produced 300 ng of luciferase at 7 days postinjection, an expression level 46-fold higher than the VR1205 vector (or 22-fold higher, excluding modifications to the luciferase gene) and 154-fold higher than a commercially available luciferase expression vector. Thus, VR1255 represents an improved plasmid DNA vector that may be useful for gene therapy applications.

Sequence and structural organization of the human gene encoding ciliary neurotrophic factor.

Ciliary neurotrophic factor (CNTF) is a potent polypeptide hormone whose actions appear to be restricted to the nervous system where it promotes survival, neurotransmitter synthesis and neurite outgrowth in certain neuronal populations. We have cloned the gene encoding human CNTF (hCNTF) and have characterized its structure and organization. The hCNTF gene appears to be a unique-copy gene with a simple genetic organization, since only a single intron interrupts the coding domain. The hCNTF gene is located on chromosome 11, as determined using human-hamster somatic cell hybrids. The CNTF protein is highly conserved in evolution. The amino acid (aa) sequences of rat and rabbit CNTF translated from cDNAs display approx. 85% homology with the deduced aa sequence encoding hCNTF.

Nerve growth factor receptor immunoreactivity in neurons of the normal adult rat spinal cord and its modulation after peripheral nerve lesions.

Motoneurons of the rat spinal cord express low-affinity nerve growth factor receptor (LNGFR) and corresponding mRNA during development, and re-express it after their axotomy by peripheral nerve injury. The present study establishes the anatomical and quantitative baseline of LNGFR immunoreactive (LNGFR-IR) neurons of the entire normal adult female rat and then investigates the temporal course for the re-expression of LNGFR-IR in lumbar motoneurons after either a crush lesion (which is followed by regeneration and reconnection to the muscle) or a cut lesion with removal of the distal stump (where a neuroma but no reconnection is formed). In the normal adult spinal cord, two types of LNGFR-IR neurons were recognized: (1) small populations of large motoneurons located in the ventral horn mainly in correspondence to the regions of the phrenic, cremasteric and dorsolateral nuclei, and (2) a more numerous and more dorsally located population of small neurons. With a sciatic cut lesion, the number of LNGFR-IR motoneurons at spinal levels L4-L6 rapidly and dramatically increased to a maximum between post-lesion days 1 and 7, apparently involving most axotomized motoneurons of the region, and returned to the baseline level by day 30. With a crush lesion, similar numbers and virtually the same time-course of LNGFR-IR appearance were seen, but the onset of progressive disappearance of LNGFR-IR neurons was delayed by one week, so that at 30 days, the most caudal motoneurons (which are last to reach their target) were still LNGFR-IR. Comparison of these two time courses gives clues to the kind of signals that may be involved in initiating and/or maintaining the LNGFR response.

The localization of nerve growth factor-like immunoreactivity in the adult rat basal forebrain and hippocampal formation.

The role of nerve growth factor (NGF) as a target derived neurotrophic agent for specific cell populations in the peripheral nervous system has been well documented and much evidence suggests that NGF may serve a similar neurotrophic role in the CNS supporting the cholinergic neurons of the basal forebrain. Previous attempts to localize NGF by immunocytochemical methods, however, have not yielded evidence confirming the regional distribution expected based upon reported levels of extractable NGF. In the present study, affinity purified polyclonal antibodies to beta-NGF and a modified immunohistochemical protocol were used to demonstrate specific NGF-like immunoreactivity in the adult rat hippocampal formation and basal forebrain. In the hippocampal formation, NGF-like immunoreactivity was localized primarily within the hilus of the dentate gyrus and within stratum lucidum of the CA3 and CA2 hippocampal subfields. Staining appeared to be associated with cell processes and was similar to the reported distribution of mossy fibers suggesting that granule cells may either serve as a primary source of hippocampal NGF or that mossy fibers selectively accumulate NGF produced by other cell populations. In the basal forebrain, NGF-like immunoreactivity was localized within neuronal cell bodies of the medial septum, diagonal band, and nucleus basalis of Meynert and was further demonstrated to colocalize exclusively with LNGF-R positive neurons. These findings demonstrate the presence of an NGF-like antigen in association with cholinergic neurons of the basal forebrain and strongly support the hypothesis that NGF may serve as an endogenous trophic factor for this adult neuronal population.

Immunotherapy of established tumors in mice by intratumoral injection of interleukin-2 plasmid DNA: induction of CD8+ T-cell immunity.

Intratumoral (i.t.) injection of a plasmid DNA vector encoding the murine interleukin-2 (IL-2) gene was used to treat established renal cell carcinoma (Renca) tumors in BALB/c mice. Tumor regression was observed in 60-90% of mice that were injected i.t. for 4 days with IL-2 plasmid DNA complexed with the cationic lipid DMRIE/DOPE ((+/-)-N-(2-hydroxyethyl)-N,N-dimethyl-2,3-bis(tetradecyloxy)-1-propa naminium bromide/dioleoylphosphatidylethanolamine). The mice remained tumor-free until the conclusion of the study, which was 4 months after tumor challenge. In a rechallenge experiment, mice that were rendered tumor-free for 6 months by IL-2 plasmid DNA treatment rejected a subsequent challenge of Renca cells but could not reject a challenge with the unrelated, syngeneic CT-26 tumor. Spleen cells from cured mice contained Renca-specific cytotoxic T lymphocytes, and adoptive transfer of mixed lymphocyte cultures into naive mice at 2 days after challenge with Renca cells prevented tumor growth. In vivo depletion of T-cell subsets at the time of i.t. injection with IL-2 plasmid DNA demonstrated that CD8+ T cells, but not CD4+ T cells, were the primary effectors of the antitumor response.

Plasmid DNA gene therapy: studies with the human interleukin-2 gene in tumor cells in vitro and in the murine B16 melanoma model in vivo.

The plasmid DNA vector pVCL-1102 containing the coding sequence for the human IL-2 gene was evaluated for expression in tumor cells in vitro and in vivo. In vitro transfection of murine B16 tumor cells with pVCL-1102 resulted in the expression of 36,000 IU (5.7 micrograms) of biologically active IL-2/10(6) cells/48 h. In vitro transfection of human tumor lines and primary cultures from human biopsies with pVCL-1102 resulted in the expression of 1,289 to 9345 IU of IL-2/10(6) cells/48 h and 30 to 794 IU of IL-2/10(6) cells/48 h, respectively. In vivo, direct intratumor injection of pVCL-1102 resulted in retention of intact plasmid DNA in the tumor tissue and IL-2 secretion by cell cultures derived from the injected tumors. Formulation of pVCL-1102 with the cationic lipid DMRIE/DOPE inhibited DNA degradation and enhanced in vivo transfection efficiency over plasmid DNA alone. Antitumor activity of the pVCL-1102/DMRIE/DOPE complex was evaluated in a B16 melanoma model in mice. An IL-2-specific effect could not be demonstrated in a subcutaneous model because the intratumor injection of plasmid DNA lacking the IL-2 coding sequence also resulted in a significant reduction in tumor volume. However, an IL-2-specific effect was observed when B16 cells were transfected in vitro prior to implantation into the mouse. Transient transfection of B16 cells with pVCL-1102 rendered the cells less tumorigenic in vivo and produced a significant reduction in tumor volume. These data demonstrate that a plasmid DNA expression vector can be used to deliver the IL-2 gene to tumor cells in vitro and in vivo, resulting in the expression of significant levels of IL-2 protein. These data also illustrate the need for the use of appropriate controls when evaluating the in vivo biological activity of plasmid DNA in murine tumor models.