Nonviral Gene Delivery to the Lateral Ventricles in Rat Brain - Cell Press

doi:10.1006/mthe.2001.0272, available online at http://www.idealibrary.com on IDEAL

ARTICLE

Nonviral Gene Delivery to the Lateral Ventricles in Rat Brain: Initial Evidence for Widespread Distribution and Expression in the Central Nervous System James G. Hecker,*,1 Leon L. Hall,* and Van R. Irion* *Department of Anesthesiology, University of California–Davis, Davis, California 95616 Received for publication July 26, 2000; accepted in revised form January 16, 2001.

The use of DNA for nonviral gene expression depends on several factors. These include (i) delivery and accessibility to the targeted tissue, (ii) protection from extracellular degradation, (iii) sufficient uptake by cells of interest, and (iv) protection from intracellular degradation to allow translation of adequate levels of intracellular or secreted proteins. As an initial step in demonstrating the feasibility of nonviral, cationic lipid-mediated gene therapy, we present evidence for the successful delivery and expression of heat shock protein Hsp70 and reporter gene enzymes in the central nervous system (CNS) of the rat after injection into the lateral ventricle. Gene delivery is accomplished using optimized formulations of plasmid DNA, which have been complexed with the cationic lipid MLRI. Results from DNA vectors encoding for green fluorescent protein (GFP), luciferase, and Hsp70 are reported. Standard immunofluorescent methods were used to demonstrate widespread expression of the reporter proteins and of Hsp70. Stereology analysis has been completed on three coronal sections, which illustrates the distribution of expression along the longitudinal axis. These initial findings support the further development of nonviral, lipid-mediated gene delivery technology for transient expression of protective, intracellular proteins and represent an important step leading to in vivo studies to identify potential clinical benefits. Key Words: DNA; gene transfer; GFP; luciferase; Hsp70; transfection; formulation; cationic lipid; in vivo; neurons; CNS.

INTRODUCTION As sequencing of the complete human genome nears completion, delivery and characterization of the expression of exogenous nucleic acid sequences assumes increasing importance. For gene therapy of those diseases which require chronic and high levels of protein expression, inherited enzyme deficiencies for example, viral vectors may offer advantages. However, the risks and duration of gene delivery must be matched to the risks of a disease or to the risks of prophylactic use before a surgical procedure. For short-term expression the delivery of DNA by means of a nonviral, cationic lipid may provide a more favorable risk/benefit analysis. Preoperative expression of neuroprotective gene sequences in the central nervous system may be one such clinical application. DNA can be delivered by a variety of nonviral means 1 To whom correspondence and reprint requests should be addressed at School of Medicine Neurosciences Building, 1515 Newton Court, Room 411, University of California–Davis, Davis, CA 95616. Fax: (530) 757-8827. E-mail: [email protected].

MOLECULAR THERAPY Vol. 3, No. 3, March 2001 Copyright © The American Society of Gene Therapy 1525-0016/01 $35.00

(1). Of potential delivery techniques, cationic lipid-mediated methods show promise for delivery of either mRNA or DNA into central nervous system (CNS) tissues (2–5), while avoiding several potential problems with the use of viral DNA vectors (6, 7). Cationic lipids are used to protect DNA from degradation in the extracellular environment (1). Cationic lipids are commonly comprised of a polar head group and nonpolar symmetric or dyssymmetric carbon based tail. Negatively charged nucleic acids condense and self-assemble into heterogeneous complexes of lipids and nucleic acids when mixed with cationic lipids (5). Ordering and structure of the self-assembled complexes affects transfection and varies with temperature, concentration, charge ratio, buffer, time, and lipid composition. We refer to our optimized, controlled incubation of DNA and cationic lipid which leads to the selfassembly of these nucleic acid/cationic lipid complexes as “formulation.” In vitro transfections were optimized using novel cationic lipids developed by Nantz et al. (8 –10). DNA dose and cationic lipid to DNA ratio was systematically varied

375

ARTICLE

FIG. 1. Immunofluorescence micrographs of rat CNS 24 h after cationic lipid mediated transfection of green fluorescent protein (GFP) DNA into the lateral ventricle. A and B show representative CNS sections stained with a neuron-specific NeuN antibody and photographed at 20⫻ and 40⫻ magnification, respectively. NeuN staining illustrates the number of neurons detectable within the section. C, D, and E show adjacent sections to those shown in A and B stained with an antibody specific for GFP and photographed at magnifications of 10⫻, 20⫻, and 40⫻, respectively. Cytoplasmic staining can be identified in a distribution similar to that observed for NeuN staining. The size and morphology of the GFP-stained cells suggest that many, but not all, of these cells are neurons. F represents a negative control with no primary antibody, conducted on an adjacent serial section. The control image has been acquired and photographed with enhanced settings to confirm the absence of cellular staining.

376

MOLECULAR THERAPY Vol. 3, No. 3, March 2001 Copyright © The American Society of Gene Therapy

ARTICLE

FIG. 2. Immunofluorescence micrographs of 30-␮m sections of rat CNS 24 h after cationic lipid mediated transfection of luciferase DNA into the lateral ventricle. A and B show representative CNS sections stained with a neuron-specific NeuN antibody and photographed at 20⫻ and 40⫻ magnification, respectively. NeuN staining illustrates the number of neurons detectable within the section. C, D, and E show adjacent sections to those shown in A and B stained with an antibody specific for luciferase and photographed at magnifications of 10⫻, 20⫻, and 40⫻, respectively. Cytoplasmic staining can be identified in a distribution similar to that observed for NeuN staining. The size and morphology of the luciferase stained cells suggest that many, but not all, of these cells are neurons. F represents a negative control in which the primary antibody was omitted, conducted on an adjacent, serial section. The control image has been acquired and photographed with identical settings to those used in E and shows the complete lack of fluorescence intensity which was observed.


377

ARTICLE in a series of transfections to arrive at our optimized formulation. Carrier RNA was also added in increasing ratio. The addition of carrier RNA has been shown previously to enhance expression after transfection of both mRNA and DNA (11). Initial analyses of DNA vectors encoding ␤-galactosidase (␤-gal) and luciferase reporter genes were performed using cell lysis and chemiluminescent assay (Galacto-Light, Tropix, Bedford, MA; or enhanced luciferase assay kit, Analytical Luminescence Laboratories, Ann Arbor, MI) on a Monolight 2010 luminometer (Analytical Luminescence). Inducible Hsp70 (12) is the most protective of the heat shock proteins against subsequent hypoxia or ischemia. Upregulation by heat shock of the inducible heat shock proteins (Hsp) is protective against a subsequent, nearlethal stressor (13–17) such as ischemia. Delivery of Hsp70 using the herpes simplex virus (HSV) vector protects neurons in vitro (18). However, there is currently no practical way to induce endogenous Hsp70 in humans. Nonviral transfection of vulnerable cells would provide one approach for transient expression, while avoiding viral vectors. Transient prophylactic expression of protective intracellular proteins could reduce tissue damage after a subsequent, severe stressor. We have initiated a series of studies that are designed to demonstrate and characterize the nonviral DNA delivery and expression of the reporter enzymes luciferase and green fluorescent protein (GFP), and of inducible Hsp70. In this study we use novel cationic lipids which were developed by Nantz et al. (9, 10). Expression of Hsp70 following lipid-mediated DNA delivery illustrates one potential method of protecting the CNS.

MATERIALS

AND

METHODS

Our first goal in designing these experiments was to demonstrate delivery and expression using several carefully designed vectors. DNA vectors encoding reporter gene sequences for ␤-galactosidase (␤-gal) and firefly luciferase (Photinus pyralis) were developed and tested. From these we selected the optimized luciferase DNA vector pNDlux.2, described in detail below, for further experiments. Next, sequences encoding for GFP and Hsp70 were subcloned into pNDlux.2, replacing the luciferase sequence. GFP and Hsp70 vectors are therefore identical to the pND luciferase vector, except for the coding region. Because the coding sequence is small relative to the size of the entire vector, we assume that luciferase, GFP, and Hsp70 vectors are taken up by similar mechanisms when complexed in identical optimized formulations. Immunohistochemistry techniques using both fluorescent and enzymatic detection methods were then used to further characterize GFP, luciferase, and Hsp70 expression in vivo.

Vectors Luciferase DNA Vector (pNDlux.2) The pND luciferase expression vector (gift of Gary Rhodes, Ph.D., University of California–Davis) contains the human CMV immediate early promoter (HCMV IE1) and CMV IE1 intron, a polylinker cloning site, and the RNA terminator/polyadenylation site derived from bovine growth hormone (BGH) (19). These elements are contained in a pUC19 replicon. cDNA encoding the luciferase gene from the firefly Photinus pyralis (pGL2I⫹, Promega, Madison, WI) was cloned into expression vector pND to give the plasmid pND2-lux.

378

GFP DNA Vector (pNDeGFP) Full-length enhanced GFP (eGFP, a variant of jellyfish Aequorea victoria GFP) cDNA (pEGFP, Clontech, Palo Alto, CA) was subcloned into the pND vector described above to give the DNA vector, pNDeGFP (also a gift of Gary Rhodes).

Hsp70-1 DNA Vector (pNDHsp70) Full-length Hsp70-1 cDNA (SPC-130, StressGen, Victoria, B.C.) was subcloned using SalI and XbaI into the pND vector described above to give the DNA vector, pNDHsp70.

Lipid Formulation Cationic lipids were described previously by Balasubramaniam et al. (9) and by Bennett et al. (8, 10). Chloroform was added to dry MLRI (myristoyl lauroyl Rosenthal inhibitor, previously referred to as LMHME). MLRI is the dissymmetric myristoyl (14:0) and Lauroyl (12:0) substituted compound formed from the tetraalkylammonium glycerol-based prototypic cationic lipid DORI (N-(1-(2,3-dioleoyloxy)propyl)-N-(1-(2-hydroxy)ethyl)-N,N-dimethyl ammonium iodide). MLRI was mixed 50:50 with dioleoylphosphatidyl-ethanolamine (DOPE) in chloroform). The solution was vortexed and aliquoted into glass vials. Aliquots were then dried under vacuum and stored at 0°C. Prior to each transfection nuclease-free water was added to dry, thin film MLRI:DOPE to give 1 mM:1 mM final concentration. The solution was vortexed, sonicated for 1 min, heated to 65°C for 1 min, vortexed, and sonicated again.

Transfection Formulation and Delivery Under an Approved Animal Care Protocol, anesthetized animal subjects were mounted in a stereotaxic small animal surgery frame (Stoelting, Wood Dale, IL). Using sterile technique, the transfection site was prepared for delivery using coordinates of 0.9 –1.0 mm posterior and 1.5 mm lateral of midline relative to Bregma, at a depth of approximately 3–3.5 mm. After aspiration of CSF to verify intra-ventricular cannula placement, the transfection formulation was prepared. In a glass vial, 6.5 ␮l of cationic lipid, MLRI per 1 ␮g of DNA is added to 10 ␮g of DNA, vortexed 5 s, and incubated at 37°C for 18 min. Carrier RNA at a concentration of 20 ␮g/␮l (Torula yeast total RNA, 5⬘-3⬘, Boulder, CO) is added at a ratio of 100 ␮g per 1 ␮g of DNA, and again vortexed for 5 seconds. The formulated lipid/DNA complex is rapidly loaded into a syringe and infused over 30 min using a syringe infusion pump (Model 101, Stoelting). Animals were closely monitored for signs of discomfort, toxicity, or neurologic injury, and none was observed.

Tissue Preparation for Reporter Enzyme Localization and Immunohistochemistry At either 24 or 44 h after transfection the animal subjects were deeply anesthetized and perfused through the ascending aorta with saline, followed by 4% paraformaldehyde in 0.1 M, pH 7.4, sodium phosphatebuffered saline (PBS). The brain is dissected and removed, usually leaving the olfactory bulb in the skull. Brains are postfixed in the same paraformaldehyde fixative overnight at 4°C, and are then blocked and placed in PBS containing 20% glycerol at 4°C. Brains not sectioned within 2–3 days of perfusion are frozen in 2-methylbutane at ⫺30°C in a flask containing ethanol and dry ice, and stored at ⫺70°C until sectioning. Brains are sectioned using a microtome in dry ice, following standard techniques. Brains are sectioned in the coronal plane, beginning approximately 6 –7 mm anterior relative to Bregma. Ten series of 30-␮m serial sections from 250- to 300-g Sprague–Dawley rats are collected for each brain. Infusions are performed at approximately 0.9 –1.0 mm posterior and 1.5 mm lateral relative to Bregma. This plane of injection is usually found at or very close to section number 16 in a complete series of 24 sections through the brain in this size animal. These sections correspond to Plate #20 or 21 from Paxinos (20). Sections are stored at ⫺20°C in 24-well plates containing PBS with 20% glycerol, with one free-floating section per well. Each 24-well plate therefore contains a complete series of serial sections across the entire brain, with every tenth section of brain per well in each series. MOLECULAR THERAPY Vol. 3, No. 3, March 2001 Copyright © The American Society of Gene Therapy

ARTICLE Distance along the longitudinal axis of the brain from the plane of infusion was estimated by counting the number of 30-␮m sections from the plane of infusion. Adjacent sections in 24-well plate are 10 sections apart: sections from neighboring wells in each series are therefore 300 ␮m apart. For example, if the injection track is found in coronal section #18, then section #10 is 8 ⫻ 300 ␮m or roughly 2400 ␮m distant from the plane of infusion.

DAB Staining Sections were initially stained as free-floating sections in 24-well netwell plates, using the avidin– biotin–peroxidase (ABC) method (Vectastain kit, Vector Labs, Burlingame, CA). H2O2 was used routinely to eliminate the staining due to endogenous peroxidase activity of infiltrating red and white blood cells in the immediate neighborhood of the infusion cannula. Free-floating sections in 24-well plates are washed in 0.02 M K–PBS and incubated in blocking buffer [0.5% TX-100, 5% normal goat serum (NGS), in 0.02 M K–PBS] for 4 h at room temperature. Sections were pretreated with 1% H2O2 for 15 min before incubation with the primary antibody. Primary antibodies are diluted in 0.1 M sodium phosphate buffer, pH 7.4, containing 0.1% bovine serum albumin, 0.3% Triton X-100, and 2% normal goat serum. Tissue sections were incubated in the primary antibody at 4°C and the bound antibody was visualized with the ABC method (Vectastain Kit, Vector Labs) using 3,3⬘-diaminobenzidine (DAB) as the peroxidase substrate. The primary antibody (mouse monoclonal antibody for the inducible form of Hsp70, SPA-810, StressGen) was diluted 1:300 into 0.3% X-100, 2% normal goat serum in K–PBS, and incubated at 4°C for 24 h. Sections were then washed and incubated with biotinylated secondary antibody (BA-9200, biotinylated goat anti-mouse, Vector Labs) diluted at 1:225 in K–PBS with 0.3% X-100, 2% normal goat serum for 1 h, washed and incubated in avidin solution (Biomeda, Foster, CA) for 45 min. Cells are then incubated in DAB solution (0.05% DAB and 0.04% H2O2 in 0.02 M K–PBS) for 10 min. Primary rabbit anti-luciferase antibody, (CR2029R, Cortex Biochem, San Leandro, CA), diluted 1:3000, was used for luciferase transfected animals. A biotinylated goat anti-rabbit secondary antibody (BA-1000, Vector) was then used. Otherwise we use the same protocol as described above.

Indirect Immunofluorescence More recent experiments have been processed using sections mounted on polylysine-coated microscope slides (Columbia Diagnostics, Inc., Springfield, VA) and air dried for a minimum of 2 h. To reduce autofluorescence sections were incubated in 1% (w/v) sodium borohydride for 8 min at RT. Sections were then washed in PBS and blocked in PBS containing 3% BSA, 20% normal serum of the secondary antibody host species, and 0.1% (v/v) Tween 20 for 20 min at room temperature. The blocking solution is drained off and the sections were covered with the primary antibody appropriately diluted in blocking solution and incubated overnight at 4°C in a humid chamber. Following incubation in the primary antibody, the sections were washed three times in PBS and incubated for 45 min at room temperature with the secondary FITC conjugated antibody diluted in PBS. The sections were washed as above, with a final wash for 2 min in double-distilled water, and mounted in an antifade mounting medium. The following antibodies were found to be optimal for indirect immunofluorescence. Primary antibodies. All primary antibodies are monoclonal, except where noted: Mouse anti-GFP (8362-1 Clontech, 1:250), mouse anti-Hsp70 (SPA810, StressGen, Vancouver, BC, 1:50), mouse anti-NeuN (MAB377, Chemicon, 1:50), and rabbit polyclonal anti-luciferase (CR2029R, Cortex, 1:50). Secondary antibodies. Goat anti-mouse FITC (#31543, Pierce, 1:50) and goat anti-rabbit/FITC (#31573, Pierce, 1:50). A Nikon 600 microscope with camera mount was used for all for photographic documentation of results. Negative controls were included on serial sections and processed identically and simultaneously alongside the positive section, except for omission of the primary antibody. Film negatives or slides were scanned into Photoshop 5.0 using a Photoshop plug-in and a Polaroid SprintScan slide scanner at a resolution of 2700 dpi. Unless specified, photographs were printed without adjustment using Photoshop MOLECULAR THERAPY Vol. 3, No. 3, March 2001 Copyright © The American Society of Gene Therapy

5.0 (Adobe, Seattle, WA) on a Fuji Pictrography 3000 (Fuji Photo Film, Elsmford, NY) at 320 dpi.

Stereology Initial stereological analysis of the distribution, uptake, and expression of GFP after delivery to the lateral ventricle of the rat has been carried out in representative sections using StereoInvestigator (MicroBrightfield), beginning with sections from rostral, caudal, and mid brain.

Positive Control for Hsp70-1 Expression Positive controls include a heat-shocked animal. Following an amended Animal Resources protocol, a heating blanket was used to raise the rectal temperature of a single rat to 42°C for 10 min. This is a sufficient heat shock to elicit a robust Hsp70 response (21). Under deep anesthesia, the brain was perfused 24 h later, and processed using our standard methods.

Control for Endogenous Hsp70 Expression Under deep anesthesia, animals were injected with PBS. Animals were processed as previously described and analyzed for Hsp70 expression.

Negative Controls for Immunohistochemistry Controls which omit the primary antibody are described under Results. Animals injected with luciferase vectors have been stained using GFP and Hsp70 primary antibodies, Hsp70 injected animals have been stained using GFP and luciferase primary antibodies, and GFP-injected animals have been stained using luciferase and Hsp70 primary antibodies.

RESULTS In Vivo Delivery of GFP, Luciferase, and Hsp70-1 to Rat CNS Results which show uptake and expression of GFP, luciferase, and Hsp70-1 24, 47, or 26 h after delivery of DNA vectors in vivo in rat brain are shown in the following figures. Positive controls included a heat shocked animal, while negative controls included Hsp70 immunohistochemistry after PBS and luciferase vector infusion, and GFP immunohistochemistry after Hsp70 and luciferase vector infusion.

GFP Immunofluorescence Figure 1 is representative of indirect immunofluorescence 24 h after delivery of our GFP DNA vector. Transfected sections were processed with either the neuronspecific monoclonal mouse antibody NeuN, or with a monoclonal antibody for GFP. Figures 1A and 1B show representative sections of a GFP-transfected animal stained with a monoclonal antibody for the neuron specific NeuN, at magnifications of 20⫻ and 40⫻, respectively. Figures 1C through 1E show representative examples of GFP transfection in the same animal at increasing magnifications. Figure 1C is a 10⫻ magnification, Fig. 1D is 20⫻, and 1 E is at a magnification of 40⫻, from the same section. The lower magnification conveys the extent of transfection and expression seen across the entire section. At the two higher magnifications cytoplasmic staining is clearly noted. This includes but is not limited to cells that by size and morphology appear to be neurons.

379

ARTICLE

FIG. 3. Immunofluorescence micrographs of rat CNS in Hsp70-transfected and heat-shocked-positive control animals. A and B show detection of heat shock protein 70 (Hsp70), 24 h after cationic lipid-mediated transfection of Hsp70 DNA into the lateral ventricle, at 20⫻ and 40⫻ magnification, respectively. C and D show detection of Hsp70 in a heat-shocked control animal at 20⫻ and 40⫻ magnification, respectively. In both animals the representative figures show widespread distribution of cells stained positive for Hsp70. The size and morphology of the cells suggest that they include neurons. E and F represent negative controls for transfected and heat-shocked experiments, respectively, in which no primary antibody was used. No cellular staining is observed.

380


ARTICLE Figure 1F is a negative control using an adjacent serial section in which the primary antibody was omitted, which was processed identically and simultaneously. The photograph was obtained after a longer exposure time and brightness and contrast were enhanced in the image to allow the negative control images to be visualized. This photograph accurately reflects the lack of cellular staining in any region of these negative control sections. Contrast this with the negative control described in Fig. 2F below.

Luciferase Immunofluorescence Figure 2 is an example of a luciferase DNA vector transfected animal. Brain sections underwent indirect immunofluorescent detection of neurons, 47 h after transfection, using the neuron-specific monoclonal antibody NeuN. Luciferase expression was detected using a specific polyclonal antibody against luciferase. Figures 2A and 2B are from a section stained for neurons, at magnifications of 20⫻ and 40⫻, respectively. Figure 2C shows an adjacent section which has been stained using a polyclonal luciferase antibody, at a magnification of 10⫻. As in the GFP infused animals, expression is noted in Fig. 2C to be widely distributed across the section at lower magnification, but is not uniformly homogeneous. Figures 2D and 2E are at magnifications of 20⫻ and 40⫻, and cells that by size and morphology appear to include neurons are clearly identifiable. Figure 2F is a negative control in which the primary antibody was omitted, which was processed identically and simultaneously. Microscope, camera, and exposure setting were fixed and unchanged between photographs Figs. 2E and 2F, so that the complete lack of fluorescence in the negative control would be illustrated. No changes of any kind were made, including brightness, contrast, or levels, before printing these images, in order to show the contrast between positive and negative.

Hsp70 Immunofluorescence DNA Hsp70 vector-transfected rat brain sections and sections from a heat-shocked positive control animal were processed for indirect immunofluorescent detection of Hsp70. Figures 3A and 3B illustrate Hsp70 immunofluorescence of a perfused animal 26 h after a DNA Hsp70 infusion. Figure 3A is at 20⫻, while Fig. 3B is a 40⫻ magnification. Figure 3E is a negative control for Figs. 3A and 3B in which the primary Hsp70 antibody was omitted. Figures 3C and 3D are similar magnifications from a heat-shocked positive control animal, illustrating a similar pattern of expression. Figure 3F is a negative control for Figs. 3C and 3D in which the primary Hsp70 antibody was omitted, in a serial section from the same heatshocked animal. Photographic exposure time and brightness and contrast of the scanned image have once again been adjusted to display the lack of cellular staining in the negative controls. MOLECULAR THERAPY Vol. 3, No. 3, March 2001 Copyright © The American Society of Gene Therapy

Stereology Initial stereological analysis of the distribution, uptake and expression of GFP after delivery to the lateral ventricle of the rat has been completed on sections #4, 12, and 18 out of one series of 30-␮m sections. Results of positive cell counts are shown in Fig. 4. Each of the fine points shown is a cell that was immunofluorescent. These sections correspond approximately to Plate numbers 9, 17, and 25 from Paxinos and Watson (20). These images clearly demonstrate the degree of staining observed across each coronal section. Some imperfections in section staining and mounting are evident. The immunofluorescent results presented in Fig. 1, above, are from this same animal. As one of several controls for endogenous Hsp70 expression, 17 ␮l of PBS was injected into hippocampus, and stained using our standard methods for Hsp70. In animals perfused and sectioned 24 h after injection, and stained for Hsp70 expression using the ABC/DAB method (Vector), a low level of homogeneous, endogenous expression of Hsp70 was noted near the injection site, presumably due to injury. Cellular expression of endogenous Hsp70 is noted only within a very short distance, corresponding to several cell bodies, from the injection track, with decreasing expression as distance from the injection site increases. Similar patterns of very limited endogenous expression are noted when animals injected with luciferase or GFP vectors are stained for Hsp70.

DISCUSSION Delivery and expression in the CNS of DNA vectors using cationic lipids has been reported previously. However, the extent of distribution, uptake and expression of nonviral gene delivery in previous reports has been limited (4, 22–25). Here we report results of indirect immunofluorescence after in vivo nonviral transfection of reporter genes and the stress response gene Hsp70. We have confirmed expression of GFP, luciferase, and Hsp70 vectors using several independent detection techniques, including chemiluminescent assay, immunohistochemistry, direct fluorescence (GFP), and indirect immunofluorescence. We have shown widespread distribution in a fairly uniform pattern across entire sections, and to sections distant within the brain from the plane of infusion into the lateral ventricle. We have performed an extensive series of negative controls to confirm expression, including injection of PBS alone to detect endogenous Hsp70 expression. Positive controls for GFP expression included direct fluorescence, and positive controls for Hsp70 included a heat shocked animal. We have completed an initial analysis on three isolated coronal sections from the brain of an animal infused with our DNA GFP vector, using modern techniques of stereology (26, 27). These studies demonstrate expression and quantification of reporter gene products and Hsp70 in the CNS of rat brain using nonviral delivery, and support the feasibility of CNS therapy for preoperative expression of pro-

381

ARTICLE

FIG. 4. Stereology was conducted on sections 4, 12, and 18 of a series of 24- to 30-␮m sections through the rat CNS. These sections correspond to plates 9, 17, and 25 in Paxinos and Watson (20). The stereology was performed following immunofluorescent detection of green fluorescent protein (GFP) conducted 24 h after cationic lipid mediated transfection of GFP DNA into the lateral ventricle of the CNS. Each dot represents a single cell identified as positive for GFP expression. Positive cell counts were 26,468 for section #4, 32,911 for #12, and 38,674 for #18. The stereology image characterizes the extensive GFP expression cataloged throughout the CNS, including a fairly uniform distribution across each coronal section.

tective intracellular proteins. We have achieved reporter gene expression levels which appear to be much more widespread than those reported with other nonviral DNA transfections after 24 to 48 h (4, 23, 25). Our in vivo formulations are based on our in vitro experiments (data not shown), in which we have optimized and compared a variety of commercially available cationic lipids in a variety of cell lines. Although the extrapolation of in vitro results to in vivo application has been reported to be problematic, we have found in vitro optimization of formulations to be valuable for determining initial in vivo formulation. Now that we have demonstrated widespread in vivo expression, we will continue to characterize and quantify the time course of in vivo delivery, uptake, and expression. Note that we are comparing our results to other reported nonviral expression results: some viral DNA vectors can achieve higher in vitro infection efficiencies and levels of expression. Figures 1 and 2, sections A and B, are stained for NeuN to demonstrate the density of neurons in this region of the brain. These may be compared to adjacent sections,

382

which were stained for GFP or luciferase, shown in Figs. 1 and 2, sections C–E. Immunostaining and comparison to free-floating sections suggest that antibody penetration in slide mounted sections may not be complete. Although we have not yet confirmed this with confocal microscopy, this is not critical so long as comparisons are made between sections that have been stained in identical manner. Results of initial stereological analysis (Fig. 4) demonstrate abundant immunofluorescent staining within the three sections from distant regions of the brain that we have analyzed so far. Of the representative series of brains we have examined it is worthy to note that we have not yet found a section absent of expression. This suggests that the distribution of DNA/cationic lipid complexes after injection into the CSF in the lateral ventricle of the rat brain is more widespread than previously reported. We postulate that bulk transport occurs due in part to cardiac oscillations along perivascular capillaries, as suggested by Greitz and Hannerz (28) in humans. They demonstrated that cardiac oscillations resulted in considerable bulk mixing between supratentorial and infratentorial MOLECULAR THERAPY Vol. 3, No. 3, March 2001 Copyright © The American Society of Gene Therapy

ARTICLE CSF compartments, and proposed an updated model of CSF circulation and outflow, with the main absorption of CSF by the blood. The results of Greitz and Hannerz provide support for the work of Rennels et al. (29), who described a rapid paravascular flux from CSF to extracellular fluid (ECF) which accounted for extensive perivascular circulation and distribution from the CSF into the brain parenchyma. They described a rapid microcirculation of CSF/ECF “throughout the brain ‘behind’ the blood– brain barrier by fluid movements through the pericapillary basal laminae,” which could be largely abolished by the absence of cardiac pulsation. These reports from the neuroradiology literature are supported by in vivo experiments as well. Recent reports using liposomes or viral delivery vectors demonstrate uptake which is noted to be primarily ependymal, but these same reports describe neurophysiological effects which are not localized to the ependyma. Yang (30), using an adenoviral vector in a murine model, noted that “the ependyma does not form a barrier to the movement of substances between CSF and brain.” Lalwani et al. (31) infused an adenoviral vector for ␤-galactosidase into the cochlea of a guinea pig. They demonstrated ␤-gal staining on the contralateral side as well, and discussed CSF versus intravascular transport of the vector. Yaida and Nowak (32) described the distribution by in situ hybridization of oligonucleotides after hippocampal and intraventricular injection in rat brain. Phosphorothioated oligonucleotides showed significant penetration into and accumulation within brain, and both modified and unmodified oligonucleotides were detectable for at least 2 days. This supports our observation that once the blood– brain barrier between intravascular space to CNS is crossed, with a cannula or spinal needle, for example, then the circulation, distribution, uptake, and expression of a variety of agents via the CSF may be greater than previously realized. Furthermore, comparison of immunofluorescent staining for NeuN versus reporter or Hsp70 staining suggests that the transfected cells include a significant fraction of the neurons available for immunodetection. In the CNS, viral vectors have most often been used for gene delivery (6, 22, 33– 43). Gene therapy has focused on permanent correction of severe genetic defects, cancer, or HIV, and has therefore been directed toward chronic expression of high levels of a gene product. Viral vectors can achieve transduction of 100% of cells in vitro, with extremely high levels of protein expression. However, the extraordinary levels of expression and infectivity which are possible with viral DNA vectors in vitro may be more than is necessary for physiologic significance in vivo, even if they could be routinely achieved. Viral vectors may in fact overwhelm the cellular protein producing apparatus of the cell in vivo (44). Brief therapeutic expression of carefully selected gene sequences may be preferable or continued expression may be deleterious. Viral vectors may not be advantageous for all clinical situations. Indeed, Anderson stated that “. . . two factors suggest that nonviral delivery systems will be the preferred choice in the future: safety, and ease of manufacturing” (35). Any MOLECULAR THERAPY Vol. 3, No. 3, March 2001 Copyright © The American Society of Gene Therapy

report of the efficacy of transfection should therefore include consideration of what constitutes a physiologically relevant and appropriate level of expression when targeted to a particular biological question or clinical situation. Genetic therapeutics involving nonviral delivery, and the lower levels of expression of nonviral when compared to viral DNA delivery may be an advantage in some clinical applications. Several problems are associated with viral vectors. These include immune responses, problems associated with transcriptional and translational control, and difficulties in translating in vitro results to significant in vivo expression (6, 7, 44 –54). Delivery by cationic lipids avoids the problems and safety issues in the production of viral vectors and helper viruses and avoids the complexities in the regulation of viral vectors (1). “Intra-CSF administered recombinant adenovirus causes an immune response-mediated toxicity” (55), and we expect advantages from a nonviral approach. We have not observed an inflammatory response after nonviral vector delivery in routine hematoxylin and eosin staining, and we are continuing in vitro and in vivo analysis of the important issue of cytotoxicity. Nonviral delivery appears to avoid many of the inflammatory response of adenoviruses, and enables the transfection of dividing as well as nondividing cells. We have proposed that expression of Hsp70 in the CNS would confer protection against subsequent ischemia. Intracellular expression of a non-secreted protein such as Hsp70, which exerts its effects on protein folding, is a logical approach since the protein must be expressed at the site of action. Expression of Hsp70 by preoperative gene transfer in a surgical population must be safe and should preferably be transient. Delivery by a nonviral vector may therefore be more appropriate for preoperative surgical patients. Our results demonstrate that nonviral transfections can produce significant protein expression in what appears to be a large fraction of the cells within a 30-␮m section of perfused brain. This is demonstrated by comparison of NeuN and reporter GFP or luciferase expression, and by comparison of Hsp70 staining in Hsp70 transfected versus heat-shocked animals. Levels of immunofluorescent cellular staining appear to be similar in preliminary examinations. Cells in which expression is seen include, but are not limited to, neurons, and we do not suggest that our techniques preferentially transfect neurons. Characterization by cell type using double staining is in progress. Any protein for which the cDNA sequence is known can be encoded in a similar optimized and stabilized vector construct. This technology enables new applications within the field of gene transfer, and the short duration of expression allows gene therapy techniques to be applied in clinical situations which were previously impractical. In future experiments, cell protection after near-lethal stress using our Hsp70-1 vectors will be determined. Tissue protection in vivo will be examined in a functional biological assay, using the four-vessel transient, global ischemia model. This work could potentially lead to improved outcomes in anesthesia delivery and in clinical therapeutics.

383

ARTICLE ACKNOWLEDGMENTS We thank Professor Gary Rhodes for his generous gifts of pND vectors and advice and Professor Michael Nantz for his generous gifts of the cationic lipid and for discussions regarding formulations. We also gratefully acknowledge the valuable assistance of Professor David Amaral. J.G.H. thanks his colleagues in the UC– Davis Department of Anesthesiology for providing him with the necessary time out of the operating rooms to pursue this research. Sources of support include Foundation for Anesthesia Education and Research Starter and New Investigator Awards, NIH K08 NS01960-01A2, American Heart Association Western States Affiliate Beginning Grant-in-Aid 9960061Y, and University of California–Davis Department of Anesthesiology and School of Medicine funds.

REFERENCES 1 Luo, D., and Saltzman, W. M. (2000). Synthetic DNA delivery systems. Nat. Biotechnol. 18: 33–37. 2 Fisher, L. J., and Ray, J. (1994). In vivo and ex vivo gene transfer to the brain. Curr. Opin. Neurobiol. 4: 735–741. 3 Dwarki, V. J., Malone, R. W., and Verma, I. M. (1993). Cationic liposome-mediated RNA transfection. Methods Enzymol. 217: 644 – 654. 4 Zou, L. L., et al. (1999). Liposome-mediated NGF gene transfection following neuronal injury: Potential therapeutic applications. Gene Ther. 6: 994 –1005. 5 Felgner, P. L., et al. (1995). Improved cationic lipid formulations for in vivo gene therapy. Ann. N.Y. Acad. Sci. 772: 1126 –1139. 6 Lawrence, M. S., et al. (1999). Inflammatory responses and their impact on ␤-galactosidase transgene expression following adenovirus vector delivery to the primate caudate nucleus. Gene Ther. 6: 1368 –1379. 7 Dewey, R. A., et al. (1999). Chronic brain inflammation and persistent herpes simplex virus 1 thymidine kinase expression in survivors of syngeneic glioma treated by adenovirusmediated gene therapy: Implications for clinical trials. Nat. Med. 5: 1256 –1263. 8 Bennett, M. J., Malone, R. W., and Nantz, M. H. (1995). A flexible approach to synthetic lipid ammonium salts for polynucleotide transfection. Tetrahedron Lett. 36: 2207–2210. 9 Balasubramaniam, R. P., et al. (1996). Structural and functional analysis of cationic transfection lipids: The hydrophobic domain. Gene Ther. 3: 163–172. 10 Bennett, M. J., et al. (1996). Considerations for the design of improved cationic amphiphile-based transfection reagents. J. Liposome Res. 6: 545–565. 11 Hecker, J. G., and Irion, V. R. (1998) Advances in self-limited gene expression of protective intracellular proteins in vivo in rat brain using mRNA/cationic lipid complexes. Anesthesia Analgesia 86: S346. 12 Feige, U., and Polla, B. S. (1994). Heat shock proteins: The hsp 70 family. Experientia 50: 979 –986. 13 Koroshetz, W. J., and Bonventre, J. V. (1994). Heat shock response in the central nervous system. Experientia 50: 979 –1118. 14 Liu, Y., Kato, H., Nakata, N., and Kogure, K. (1992). Protection of rat hippocampus against ischemic neuronal damage by pretreatment with sublethal ischemia. Brain Res. 586: 121–124. 15 Chen, J., Graham, S. H., Zhu, R. L., and Simon, R. P. (1966). Stress proteins and tolerance to focal cerebral ischemia. J. Cerebral Blood Flow Metab. 16: 566 –577. 16 Narasimhan, P., Swanson, R. A., Sagar, S. M., and Sharp, F. R. (1996). Astrocyte survival and HSP70 heat shock protein induction following heat shock and acidosis. Glia 17: 147–159. 17 Sharp, F. S., Massa, S. M., and Swanson, R. A. (1999). Heat-shock protein protection. TINS 22: 96 –99. 18 Fink, S. L., Chang, L. K., Ho, D. Y., and Sapolsky, R. M. (1997). Defective herpes simplex virus vectors expressing the rat brain stress-inducible heat shock protein 72 protect cultured neurons from severe heat shock. J. Neurochem. 68: 961–969. 19 Chapman, B., Thayer, R., Vincent, K., and Haigwood, N. (1991). Effect of intron A from human cytomegalovirus (Towne) immediate early gene on heterologous expression in mammalian cells. Nucleic Acids Res. 19: 3979 –3986. 20 Paxinos, G., and Watson, C. (1986). The Rat Brain in Stereotaxic Coordinates. Academic Press, San Diego. 21 Planas, A. M., et al. (1997). The heat shock response after brain lesions: induction of 72 kDa heat shock protein (cell types involved, axonal transport, transcriptional regulation) and protein synthesis inhibition. Prog. Neurobiol. 51: 607– 636. 22 Meuli-Simmen, C., et al. (1999). Gene expression along the cerebral-spinal axis after regional gene delivery. Hum. Gene Ther. 10: 2689 –2700. 23 Schwartz, B., et al. (1996) Gene transfer by naked DNA into adult mouse brain. Gene Ther. 3: 405– 411. 24 Brooks, A. I., and Federoff, H. J. (1998). Tfx-20 reagent and gene delivery into mouse CNS. Neural Notes 4: 20 –23.

384

25 Abdallah, B., et al. (1996) A powerful nonviral vector for in vivo gene transfer into the adult mammalian brain: Polyethylenimine. Hum. Gene Ther. 7: 1947–1954. 26 Amaral, D. G., Dolorfo, C., and Alvarez-Royo, P. (1991). Organization of CA1 projections to the subiculum: A PHA-L analysis in the rat. Hippocampus 1: 415– 436. 27 Ishizuka, N., Cowan, W. M., and Amaral, D. G. (1995). A quantitative analysis of the dendritic organization of pyramidal cells in the rat hippocampus. J. Comp. Neurol. 362: 17– 45. 28 Greitz, D., and Hannerz, J. (1996). A proposed model of cerebrospinal fluid circulation: Observations with radionuclide cisternography. Am. J. Neuroradiol. 17: 431– 438. 29 Rennels, M. L., Gregory, T. F., Blaumanis, O. R., Fujimoto, K., and Grady, P. A. (1985). Evidence for a ‘perivascular’ fluid circulation in the mammalian central nervous system, provided by the rapid distribution of tracer proteins throughout the brain from the subarachnoid space. Brain Res. 326: 47– 63. 30 Yang, G.-Y, Zhao, Y.-J., Davidson, B. L., and Betz, A. L. (1997). Overexpression of interleukin-1 receptor antagonist in the mouse brain reduces ischemic brain injury. Brain Res. 751: 181–188. 31 Lalwani, A. K., Walsh, B. J., Reilly, P. G., Muzyczka, N., and Mhatre, A. N. (1996). Development of in vivo gene therapy for hearing disorders: Introduction of adeno-associated virus into the cochlea of the guinea pig. Gene Ther. 3: 588 –592. 32 Yaida, Y., and Nowak, T. S. (1995). Distribution of phosphodiester and phosphorothioate oligonucleotides in rat brain after intraventricular and intrahippocampal administration determined by in situ hybridization. Regul. Pept. 59: 193–199. 33 Costantini, L. C., Bakowska, J. C., Breakefiekd, X. O., and Isacson, O. (2000). Gene therapy in the CNS. Gene Ther. 7: 93–109. 34 Morris, J. C., and Wildner, O. (2000). Therapy of head and neck squamous cell carcinoma with an oncolytic adenovirus expressing HSV-tk. Mol. Ther. 1: 56 – 62. 35 Anderson, W. F. (1998). Human gene therapy. Nature 392: 25–30. 36 Clinical Trials Database (1999). Gene transfer and clinical trials-update. Part 2: Gene delivery systems used and genes transferred. J. Gene Med. 1: 151–153. 37 Trono, D. (2000). Lentiviral vectors: Turning a deadly foe into a therapeutic agent. Gene Ther. 7: 20 –23. 38 Monahan, P. E., and Samulski, R. J. (2000). AAV vectors: Is clinical success on the horizon? Gene Ther. 7: 24 –30. 39 Driesse, M. J., et al. (1999). Distribution of recombinant adenovirus in the cerebrospinal fluid of nonhuman primates. Hum. Gene Ther. 10: 2347–2354. 40 Costantini, L. C., et al. (1999). Gene transfer of the nigrostriatal system by hybrid herpes simplex virus/adeno-associated virus amplicon vectors. Hum. Gene Ther. 10: 2481– 2494. 41 Wang, X., Zhang, G., Yang, T., Zhang, W., and Geller, A. I. (2000). Fifty-one kilobase HSV-1 plasmid vector can be packaged using a helper virus-free system and supports expression in the rat brain. BioTechniques 28: 102–107. 42 Ellinger, S. S., Ellinger, C. A., Aguilar, C. P., Raju, N. R., and Watson, G. L. (1999). Elimination of lysosomal storage in brains of MPS VII mice treated by intrathecal administration of an adeno-associated virus vector. Gene Ther. 6: 1175–1178. 43 Karpati, G., Lochmuler, H., Nalbantoglu, J., and Durham, H. (1996). The principles of gene therapy for the nervous system. TINS 19: 49 –54. 44 Peltekian, E., Parrish, E., Bouchard, C., Peschanski, M., and Lisovoski, F. (1997). Adenovirus-mediated gene transfer to the brain: Methodological assessment. J. Neurosci. Methods 71: 77– 84. 45 Byrnes, A. P., Rusby, J. E., Wood, M. J. A., and Charlton, H. M. (1995). Adenovirus gene transfer causes inflammation in the brain. Neuroscience 66: 1015–1024. 46 Simon, R. H., et al. (1993). Adenovirus-mediated transfer of the CFTR gene to lung of nonhuman primates: Toxicity study. Hum. Gene Ther. 4: 771–780. 47 Cornetta, K., Morgan, R. A., and Anderson, W. F. (1991). Safety issues related to retroviral-mediated gene transfer in humans. Hum. Gene Ther. 2: 5–14. 48 Chan, S. Y., et al. (1999). Tissue-specific consequences of the anti-adenoviral immune response: Implications for cardiac transplants. Nat. Med. 5: 1143–1149. 49 Paillard, F. (1999). Commentary: Gene transfer to neurons with herpes simplex virus/ adeno-associated virus hybrid vectors. Hum. Gene Ther. 10: 2441–2443. 50 Ponder, K. P., Dunbar, R. P., Wilson, D. R., Darlington, G. J., and Woo, S. L. C. (1991). Evaluation of relative promoter strength in primary hepatocytes using optimized lipofection. Hum. Gene Ther. 2: 41–52. 51 Park, F., Ohashi, K., Chiu, W., Naldini, L., and Kay, M. A. (2000). Efficient lentiviral transduction of liver requires cell cycling in vivo. Nat. Genet. 24: 49 –52. 52 Cichon, G., et al. (1999). Intravenous administration of recombinant adenoviruses causes thrombocytopenia, anemia and erythroblastosis in rabbits. J. Gene Med. 1: 360 –371. 53 Snyder, R. O. (1999). Adeno-associated virus-mediated gene delivery. J. Gene Med. 1: 166 –175. 54 Chirmule, N., et al. (1999). Immune responses to adenovirus and adeno-associated virus in humans. Gene Ther. 6: 1574 –1583. 55 Driesse, M., et al. (2000). Intra-CSF administered recombinant adenovirus causes an immune response-mediated toxicity. Gene Ther. 7: 1401–1409.
