Repeated Intrathecal Administration of Plasmid DNA Complexed

Home , PEGylation

Gene Therapy (2003) 10, 1179–1188 & 2003 Nature Publishing Group All rights reserved 0969-7128/03 $25.00 www.nature.com/gt RESEARCH ARTICLE Repeated intrathecal administration of plasmid DNA complexed with polyethylene glycol-grafted polyethylenimine led to prolonged transgene expression in the spinal cord

L Shi1, GP Tang1, SJ Gao1,YXMa1, BH Liu1,YLi1, JM Zeng1,YKNg2, KW Leong1,3 and S Wang1 1Molecular and Biomaterials Laboratory, Institute of Materials Research & Engineering, National University of Singapore, Singapore Singapore; 2Department of Anatomy, National University of Singapore, Singapore; and 3Department of Biomedical Engineering, Johns Hopkins University, Baltimore, USA

Gene delivery into the spinal cord provides a potential attenuation was associated with apoptotic cell death and approach to the treatment of spinal cord traumatic injury, detected even using complexes containing a noncoding DNA amyotrophic lateral sclerosis, and spinal muscular atrophy. that did not mediate any gene expression. When each These disorders progress over long periods of time, component of the complexes, PEI polymer or naked DNA necessitating a stable expression of functional genes at alone, were tested in the ﬁrst dosing, no reduction was found. therapeutic levels for months or years. We investigated in Using polyethylene glycol (PEG)-grafted PEI for DNA this study the feasibility of achieving prolonged transgene complexes, no attenuation of gene expression was detected expression in the rat spinal cord through repeated intrathecal after repeated intrathecal injections, even in those rats administration of plasmid DNA complexed with 25 kDa receiving three doses, administered 2 weeks apart. Lumbar polyethylenimine (PEI) into the lumbar subarachnoid space. puncture is a routine and relatively nontraumatic clinical With a single injection, DNA/PEI complexes could provide procedure. Repeated administration of DNA complexed with transgene expression in the spinal cord 40-fold higher than PEG-grafted PEI through this less invasive route may naked plasmid DNA. The transgene expression at the initial prolong the time span of transgene expression when needed, level persisted for about 5 days, with a low-level expression providing a viable strategy for the gene therapy of spinal cord being detectable for at least 8 weeks. When repeated dosing disorders. was tested, a 70% attenuation of gene expression was Gene Therapy (2003) 10, 1179–1188. doi:10.1038/ observed following reinjection at a 2-week interval. This sj.gt.3301970

Keywords: gene transfer; polyethylenimine; PEGylation; spinal cord; repeated administration

Introduction complications may limit the widespread use of these viral vectors, particularly in repeated administration. Spinal cord injury and other degenerative neurological Nonviral gene delivery systems based upon plasmid illnesses like amyotrophic lateral sclerosis and spinal DNA/chemical complexes are quickly gaining recogni- muscular atrophy come with devastating effects on the tion as an alternative to viral gene vectors for their individual, and are associated with high societal costs potential in avoiding problems inherent in viral systems. because of chronic care and lost productivity. Gene Among the nonviral vectors in use, the polycationic therapy through gene transfer into the central nervous polymer polyethylenimine (PEI) has shown high trans- system (CNS) holds promise in treating these disorders.1 fection efficiency both in vitro and in vivo.4 Once in the Since most spinal cord disorders evolve over years, nervous system, PEI may mediate DNA transfection in chronic therapy may necessitate a long-term transgene terminally differentiated nondividing neurons.4–6 After expression at therapeutic levels. Several viral gene direct brain injection, PEI/DNA complexes can provide vectors, including adenoviruses and herpes simplex transgene expression levels higher than those obtained viruses, have been efficient in attaining persistent gene with the HIV-derived vector and within the same range expression in the spinal cord.2,3 However, potential side as that achieved with adenoviral vectors.5 effects of induction of immunological and pathogenic Owing to the transient expression nature of plasmid DNA, nonviral gene therapy protocols would require repeated administrations in order to achieve efficient long-term expression of therapeutic genes. Several Correspondence: Dr S Wang, Molecular and Bio-Materials Laboratory, Institute of Materials Research & Engineering, 3 Research Link, Singapore laboratories have addressed this issue in peripheral 117602, Singapore organs using repeated intravenous or intratracheal Received 21 June 2002; accepted 10 October 2002 administration.7–9 Unique attributes of the CNS, includ- PEG–PEI and gene expression in the spinal cord L Shi et al 1180 ing limited access because of the skull and the blood– brain barrier and high vulnerability to injury, pose severe obstacles to safe repetitive gene delivery. The commonly used gene delivery techniques, like direct injection into the brain and spinal cord tissues, are highly invasive and not practical in a repeat injection scheme. One way to achieve repeated CNS administration is to implant a cannula connected to a subcutaneous reservoir that provides continuous infusion of a therapeutic gene.10,11 For gene delivery into the spinal cord, a less invasive method, thus potentially allowing multiple administrations, is lumbar intrathecal injection into the cerebrosp- inal fluid (CSF). Lumbar puncture poses little danger to the spinal cord as the relative wide subarachnoid space at the site allows a certain degree of mobility of nerves within the CSF, thus avoiding damage caused by needle puncture. This method leads to relatively widespread gene expression in the CNS, and is suitable in treating degenerative neurological and neurogenetic disorders, spinal cord trauma, pain as well as fatal leptomeningeal metastases.10,12–14 We showed in this study the feasibility of achieving gene transfer in the spinal cord by intrathecal administration of PEI/DNA complexes in rats. We showed further that repeated administration of these complexes at biweekly intervals could achieve unattenuated transgene expressions if the PEI was modified with a polyethylene glycol (PEG), a group of synthetic polymers that are frequently used to improve solubility and reduce antigenicity, immunogenicity and toxicity of the mole- cules to which they are attached.15

Results Transgene expression in the spinal cord after intrathecal delivery of PEI/DNA complexes We first evaluated transgene expression induced by intrathecal injection of naked plasmid DNA. Various amounts, from 2 to 40 mg, of luciferase-encoded plasmid DNA driven by the CMV IE enhancer chicken b-actin promoter (pCAGLuc) were injected at lumbar levels. Transgene expression in the spinal cord was dose- dependent, reaching a maximum level at 20 mg DNA (Figure 1a). When the plasmid DNA was complexed with PEI, transgene expression improved significantly. For example, the expression level induced by the PEI complexes containing 4 mg of DNA could be 40-fold of that induced by the same amount of the naked plasmid Figure 1 Luciferase activities in the spinal cord. (a) Dose response of naked plasmid DNA pCAG-Luc 2 days after lumbar intrathecal injection. DNA. At their maximum levels, the gene expression (b) Dose response of naked plasmid DNA pCAG-Luc complexed with PEI mediated by PEI/DNA complexes (4 mg of DNA at N/P 2 days after lumbar intrathecal injection. The PEI/DNA ratio used was ratio 15) is about five-fold that mediated by 20 mgof fixed at 15. (c) Effects of PEI/DNA ratios 2 days after injection. A volume naked DNA (Figure 1b). The efficiency of PEI/DNA of 4 mg of DNA was used for each rat. (d) Time course of luciferase complexes in mediating transgene expression after activities in the spinal cord. A volume of 4 mg of DNA at a PEI/DNA ratio of 15 was used for each rat. Five to 10 rats were used per group. Values are intrathecal injection was related to charge ratios. A 7 maximum efficiency was observed at a charge ratio of 15 presented as means s.e.m. equivalents of PEI nitrogen per DNA phosphate, with a luciferase activity of 2.5 Â 105 RLU/spinal cord at 4 mgof DNA (Figure 1c). Luciferase activity decreased when the charge ratio was increased to 20 equivalents. Luciferase spinal cord as early as 1 day after injection (Figure 1d). activity was detected along the entire spinal cord, with This level of transgene expression retained for at least peak levels in the thoracic, lumbar and sacral regions 5 days. After that it rapidly declined by an order of that are close to the injection site (Figure 3a). In a time magnitude at day 7. However, this lower level of course study of PEI/DNA complexes, the intrathecal expression was still detectable even after 8 weeks (Figure injection yielded a significant level of luciferase in the 1d).

Gene Therapy PEG–PEI and gene expression in the spinal cord L Shi et al 1181

Figure 2 Localization of luciferase expression in the spinal cord. Immunostaining with an antibody against luciferase 2 days after lumbar intrathecal injection of PEI/pCAG-Luc complexes at N/P¼15. A strong staining in meninges is shown in (a). Positively stained cells in the spinal parenchyma exhibit different morphologies. Possible cells in (b), (c) and (d) are astrocytes, neurons and microglia, respectively.

To determine histologically the regions and cell types parenchyma, with typical morphologies of astroglial cells in the spinal cord that expressed luciferase, cryostat and neurons, respectively (Figure 2). sections of the lumbar and thoracic parts of the spinal cord were stained using antibodies against luciferase. A strong staining was seen in spinal meninges in animals Repetitive injection: attenuation of gene expression receiving PEI/DNA complexes (Figure 2a). The lucifer- upon readministration ase activity in this region accounted for 50% of the total To test whether long-term gene expression can be activity in the whole spinal cord (Figure 3b). Many achieved with repeated injections of PEI/DNA com- positively stained cells were observed also in spinal plexes, rats were given a second injection at different

Gene Therapy PEG–PEI and gene expression in the spinal cord L Shi et al 1182

Figure 3 Distribution of luciferase activity in various parts of the spinal cord. (a) Luciferase activities in the lumbar/sacral, thoracic and cervical segments of the spinal cord, as well as the brainstem. (b) Comparison of Figure 4 Repetitive administration of PEI/DNA complexes at different luciferase activities in meninges and parenchyma of the lumbar/sacral and time intervals. A volume of 4 mg of DNA was complexed with PEI at N/ thoracic spinal cord. A volume of 4 mg of DNA at a PEI/DNA ratio of 15 P¼15 and injected intrathecally. The control rats received an injection of was used for each rat. Five rats were used per group. 5% glucose vehicle. After 1, 2, 4 and 8 weeks, all rats were given a second injection of 4 mg of DNA complexed with PEI at N/P¼15. Rats were killed 48 h later. Five to 15 rats per group were used. Values are presented as RLU per tissue in (a) and percentage of controls (b). ***Po0.0005 when compared to the control animals. times after the first administration, and luciferase activities were assayed 48 h later. Control animals were injected with 5% glucose, a solution used for the preparation of PEI/DNA complexes, at the first time, followed by the same treatment done to their counterparts. As shown in Figure 4, a repeat injection of PEI/ DNA 2 weeks after the initial one yielded a transgene expression that was 70% below control value. However, as the time interval between the two injections increased up to 4 weeks, the reduction of gene expression was not observed. To identify the cause for the attenuation, we tested each component of the PEI/DNA complexes separately. Injection of naked DNA alone, up to a dose of 20 mg, or PEI polymer alone, at a dose similar to that used for DNA complexes preparation, caused no attenuation of gene expression upon re-injection of PEI/DNA complexes 2 weeks later (Figure 5). On the contrary, injection of complexes containing a noncoding plasmid with a nonfunctional promoter did induce the reduction of gene expression (Figure 5). The degree of attenuation wor- Figure 5 Identification of agents responsible for the inhibitory effects in sened with an N/P ratio increasing from 15 to 50. These the repetitive injection. The first injection used 5% glucose vehicle, PEI results indicate that PEI/DNA complexes were the cause polymer alone, naked DNA (20 mg), or noncoding plasmid DNA (pLuc-s, 4 mg) complexed with PEI at different N/P ratios of 15, 30, 50, respectively. for the inhibition observed, not the components alone or After 2 weeks, all rats were given a second injection of 4 mg of DNA transgene expression upon the first dosing. complexed with PEI at N/P¼15. Rats were killed 48 h later. A total of 10 Knowing that PEI exhibits some cytotoxicity,16 we next rats per group were used. examined the histology of the spinal cord that received intrathecal injection of PEI/DNA complexes 2 weeks earlier. H&E staining revealed no histopathological and indication of nuclei fragmentation and apoptotic cell inflammatory alterations (Figure 6a and b). However, death. TUNEL-positive cells were found mainly in the TUNEL staining of the sections from the same animal meninges of the lumbar and thoracic spinal cord (Figure revealed abnormally bright nuclei in some cells, an 6d). Some TUNEL-positive cells were also found in the

Gene Therapy PEG–PEI and gene expression in the spinal cord L Shi et al 1183

Figure 7 Agarose gel electrophoresis of 0.5 mg of pcDNAlacZ plasmid DNA and its complexes with 25 kDa PEI or PEG–PEI at N/P ratios from 0to4.

Table 1 DNA–polymer complex sizea Figure 6 Micrographs of the spinal cord and NT2/D1 cells. H&E stained tissue sections from a control animal (a) and an animal intrathecally Particles Aggregate injected with 4 mg of DNA complexed with PEI at N/P¼15 (b). The samples were collected 2 weeks after injection. TUNEL stained tissue Mean (nm) s.d. (nm) % Mean (nm) s.d. (nm) % sections from a control animal (c) and an animal intrathecally injected with 4 mg of DNA complexed with PEI at N/P¼15 (d). The arrow indicates TUNEL-positive cells. The samples were collected 2 weeks after PEI (n=7) 136 25 44 754 112 28 injection. Note TUNEL-positive cells in meninges in (d). TUNEL stained PEI–PEG (n=5) 93 18 57 826 46 7 NT2 cells after incubation with 0.5 mM PEG-PEI (e) or 0.5 mM PEI (f) for 24 h. The original photographs were taken at Â 200 magniﬁcation. aComplexes were prepared at a DNA concentration of 0.2 mg/ml, N/P ratio=15 for PEI and N/P=20 for PEI–PEG complexes. A volume of 1 ml was used for the measurement. Sizes were measured with N4 Plus Submicron Particles Sizer (COULTER, USA) at room white matter, but none in the gray matter of the spinal temperature. Scattering light was detected at 901, running for 200 s cord parenchyma. for each sample, and analyzed in the Unimodal Analysis mode. For data analysis, the refractive index medium (1.332) of 5% glucose at room temperature was used. PEI PEGylation eliminates attenuation of gene expression in repetitive administration Several studies have shown that PEI PEGylation may as used for our in vivo experiments. Two peaks were reduce its cytotoxicity.17–19 We synthesized a conjugate observed in the size measurement of PEI/DNA com- consisting of PEI covalently modiﬁed with PEG. The plexes at N/P ratio 15. The dominant one, which ratio of PEG versus PEI was 2.5, as determined from accounted for 44% of complexes, showed an average 1H-NMR spectra using integral values obtained for the diameter of 136 nm and another (28%) was 754 nm. This

–CH2CH2O– protons of PEG and –CH2CH2NH– protons second portion was probably because of aggregation of of PEI. PEGylation of PEI at this ratio did not complexes. For PEGylated PEI, we observed a small signiﬁcantly affect the amount of polymer required to complex diameter of 93 nm. This is consistent with a retard the migration of DNA (Figure 7). We then studied recent report20 demonstrating that attaching one or two the inﬂuence of PEGylation on the size of DNA PEG to PEI could enhance DNA condensation forming complexes using light scattering (Table 1). Complexes compact complexes smaller than those complexed by PEI were prepared using DNA and polymer concentrations alone. Furthermore, the percentage of the second portion

Gene Therapy PEG–PEI and gene expression in the spinal cord L Shi et al 1184 of PEG–PEI complexes with a much larger diameter (826 nm) decreased in comparison to PEI complexes, accounting for only 7%. PEGylated PEI was significantly less toxic in two neuronal precursor cells lines, PC12 and NT2/D1 cells (Figure 8). As a matter of fact, PEG–PEI stimulated cell proliferation in both cell lines at lower concentrations (Figure 8). In another experiment, cell toxicity was tested in NT2/D1 cells using DNA/polymer complexes instead of polymer alone to mimic the in vivo use of these polymers. DNA/polymer complexes were more toxic than polymers that did not interact with DNA (Figure 9). PEGylation reduced PEI cell toxicity also when the polymer formed complexes with DNA (Figure 9). As well, apoptosis decreased significantly in NT2/D1 cells after PEI PEGlyation, as demonstrated by TUNEL staining. For example, 0.5 mM of PEI induced typical apoptotic nuclear morphology, with chromatin fragmentation into bright patches (Figure 6f). The same concentration of PEG–PEI did not convert any cells into TUNEL positive (Figure 6e). We then tested DNA complexed with PEG–PEI in our intrathecal gene delivery model. Transgene expression in the spinal cord was dose-dependent, reaching a plateau at 4 mg DNA (Figure 10a). The efficiency of PEG–PEI/ DNA complexes in mediating transgene expression increased when charge ratios were increased, with the highest luciferase activity of 4.8 Â 105 RLU/spinal cord at N/P ratio 40 (Figure 10b). PEI PEGylation did not change tissue and cellular distribution patterns of transgene expression (data not shown). In a repetitive administration experiment, animals were given a second injection of

Figure 9 Cell viability assay with DNA/polymer complexes in NT2/D1 cells. The cells were treated with various concentrations of DNA/polymer complexes or polymers, expressed as the amount of nitrogen or phosphate, for 3 h (a) or 24 h (b) in a serum-containing medium. Cell viability was determined by an MTT assay and expressed as a percentage of control, that is, the untreated sister cultures. Each point represents the mean7s.e.m. of four cultures.

polymer DNA complexes 2 weeks after the ﬁrst administration of PEG–PEI/DNA complexes or 5% glucose vehicle solution and gene expression was analyzed 48 h later. We observed no difference in terms of gene expression levels between animals receiving PEG–PEI/ DNA complexes and control animals (Figure 11). In another experiment, each animal received three doses, administered 2 weeks apart. There was no evidence of inhibitory effects of previous administrations on gene expression imposed on the next administration (Figure 11).

Discussion The blood–brain barrier precludes a vascular route of transgene delivery to the CNS. Direct stereotactic Figure 8 Cell viability assay in NT2/D1 (a) and PC12 (b) cells. The cells injection into speciﬁc location therefore becomes the were treated with various concentrations of PEI or PEG–PEI for 24 h in a serum-containing medium. Cell viability was determined by an MTTassay most widely used approach for gene delivery to the CNS. and expressed as a percentage of control, that is, the untreated sister This procedure is highly invasive and the resulting cultures. Each point represents the mean7s.e.m. of four cultures. The data transgene expression can usually be detected only in the shown here are representative of three independent experiments. immediate vicinity of the injection site. On the other

Gene Therapy PEG–PEI and gene expression in the spinal cord L Shi et al 1185 nontraumatic clinical procedure, allowing a safe repeated administration for gene delivery. The distribution of gene expression in the CNS after lumbar injection is, however, relatively limited, probably because of CSF flow. In this study, the transgene expression was still localized in the sacral, lumbar and thoracic regions, although lower but detectable gene expression could be observed in the cervical regions and the brain. This distribution was not improved by using PEG-grafted PEI. Cellular distribution of transgene expression mediated by PEI observed in this study was consistent with the report by Abdallah et al5 and somehow different from the studies using cationic liposome13 or adenoviral vectors,21 which showed a restricted gene expression in meninges. Transient transgene expression can be one disadvan- tage associated with nonviral gene delivery systems, dependent on the applications. One way to overcome this problem is repetitive administrations. In the case of cationic lipidic vectors, repetitive intravenous administrations are not effective at frequent intervals.7,8,22 Apoptosis and inactivation of gene expression triggered by interferon-g and tumor necrosis factor-a, which may be induced by direct immunostimulatory effects of unmethylated CpG sequences in plasmid DNA, have been identified as two causes, among others, for this 7,22 Figure 10 Gene expression in the spinal cord mediated by PEG–PEI. (a) ineffectiveness. The same mechanism may be in force Dose response of plasmid DNA pCAG-Luc complexed with PEG–PEI, 2 when PEI is used as a delivery system for plasmid DNA days after lumbar intrathecal injection. A PEI/DNA ratio of 20 was used. and be responsible for observed apoptotic cell death in (b) Effects of PEI/DNA ratios, 2 days after injection. A volume of 4 mgof the spinal cord after intrathecal injection of PEI/DNA DNA was injected into each rat. Eight rats per group were used. Values are 7 complexes. The CNS is considered to be an immune presented as means s.e.m. privileged site and inflammatory and immune cells do not function in the CNS as efficiently as in other organs. However, CNS ventricles and parenchyma are immuno- logically separate. Administration of antigens and viruses into the CSF elicited immune responses, probably by antigenic presentation in the deep cervical lymph nodes.23,24 Overexpression of gene products may cause cell death and apoptosis, 25 but seems unlikely to act as a main cause for what was observed in this study, in view of the observation that noncoding plasmid DNA complexed with PEI inhibited subsequent gene expression too. Direct toxic effects of PEI/DNA complexes, especially their aggregates or precipitates, may also be responsible for cell death. It is of note that apoptosis was observed almost exclusively in meninges. The pia mater of spinal meninges forms the boundary of the CSF and spinal parenchyma, acting as a barrier in transport- ing DNA/polymer complexes from the CSF to the spinal Figure 11 Repetitive injection with DNA/PEG–PEI complexes. In the cord tissues. Consistent with a previous study using experiment with two injections, rats were first given 5% glucose (vehicle intracerebro-ventricular injection of PEI/DNA com- control group) or 4 mg of DNA complexed with PEG–PEI at N/P ratio 20 plexes,5 we have observed significant transgene expres- (PEG–PEI group) and then a second injection of 4 mg of DNA complexes for all animals 2 weeks later. In the experiment with three injections, rats sion in CNS parenchyma after CSF gene delivery, were given two doses of 5% glucose (vehicle control group) or 4 mg of DNA indicating the PEI/DNA complexes may penetrate complexed with PEG–PEI (PEG–PEI group) followed by a third injection through the pia mater. No obvious cell death in the gray of 4 mg of DNA complexes for all animals at biweekly intervals between the and white matters of the spinal cord indicates further three administrations. Rats were killed 48 h after the last injection for that PEI/DNA complexes entering the spinal parenchy- luciferase activity assay. Eight rats per group were used. ma may not be toxic to glial cells and neurons at these low doses. This is consistent with our previous report that PEI/DNA complexes delivered through a retrograde axonal transport caused no toxicity in brainstem neurons hand, intraventricular or intrathecal injection into CSF at concentrations for efficient DNA transfection.26 How- may reduce direct damage to CNS tissues and provide ever, PEI/DNA complexes do not disperse well, espe- means for widespread expression of transgenes in the cially at higher N/P ratios. Aggregates of the complexes whole organ. Lumbar puncture is a routine and relatively could precipitate on the surface of the pia/dura mater,

Gene Therapy PEG–PEI and gene expression in the spinal cord L Shi et al 1186 causing the death of leptomeningeal cells. Further plasmid DNA once PEI/DNA complexes were formed, investigation is required to clarify the mechanism of thus affecting cell susceptibility and gene expression apoptosis observed in the current study. profiles in repetitive administration experiments. The apoptotic cell death observed in this study was In summary, we have confirmed the feasibility of using associated with attenuation of transgene expression PEGylated PEI for repetitive administration of DNA/ following a second injection. As shown in this study, polymer complexes into the spinal cord through lumbar the attenuation is not because of the PEI polymer itself intrathecal injection. A prolonged transgene expression because injection of the polymer alone had no effects. at a significant level was demonstrated for up to 6 weeks Neither is the attenuation caused by the gene products after three repeated dosing. Advantages of this method because injection of the complexes containing noncoding in spinal cord gene therapy include minimal damage to plasmid DNA was as inhibitory as those containing neural tissues, flexibility in controlling the therapeutic functional plasmid DNA. It appears that PEI/DNA period, and possibility of achieving widespread gene complexes, but not the separate component alone, are expression. the cause of the attenuation. PEI polymer alone is probably capable of inducing apoptosis at higher concentrations, as shown in the in vitro experiment of this study. In an in vivo study injecting PEI alone into the Materials and methods CSF, such as the one reported here, opsonization of the highly positively charged polymer might reduce its Materials cytotoxic effects. Indeed, PEI polymer alone was less PEI (MW 25 000) and PEG (MW 2000) were obtained toxic than PEI/DNA complexes in in vitro cell viability from Sigma-Aldrich (St Louis, MO, USA). NHS-PEG- assays using serum-containing medium (Figure 9). MAL (MW 2000) was from Shearwater Polymers Relatively inefficient cellular uptake of the branched (Huntsville, AL, USA). pGL-3-luc basic vector was from 25 kDa PEI compared to the PEI/DNA nanocomplexes Promega. Immuno-staining kit, Qiagen kit, Luciferase may also have accounted for this diminished toxicity. kit, and TUNEL kit were obtained from Dako, Qiagen, Naked plasmid DNA alone is not stable in the body fluid Promega, and Roche, respectively. (3-(4,5-Dimethyl-thia- and may degrade rapidly. DNA degradation may de- zol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) was crease CpG contents,7,27 thus eliminating CpG-induced obtained from Sigma. Cell culture media were from Life immune responses. Technologies. PEI PEGylation in this study eliminates apoptosis and attenuation of gene expression upon a second administration of DNA/polymer complexes. The exact under- Synthesis of PEG–PEI lying mechanism is unclear. One drawback of using A volume of 1 ml of PEI (0.1 M) was mixed with 60 mlof polycations, including PEI, is poor solubility of their NHS-PEG-MAL solution (5 mM). The mixture was complexes with DNA,18 especially at a higher N/P ratio. stirred at room temperature for 45 min, after which Precipitates or aggregates of these DNA/polymer com- 50 mlof1M of glycine was added to quench the reaction. plexes could be more toxic than their nonaggregated The compound was purified by PD-10 column eluting counterparts and polymers alone, as indicated in our with PBS solution (PBS containing 5 mM EDTA), cen- finding that the higher the N/P ratio of the complexes, trifugal filter and dialysis against water. The ratio of PEG the stronger the attenuating effect on subsequent gene and the PEI in the copolymer was determined from 1H- expression. PEGylation often can improve the solubility NMR spectra using integral values obtained for the of macromolecules, minimize aggregation of particulates –CH2CH2O– protons of PEG and the –CH2CH2NH– and reduce their interaction with proteins in the protons of PEI. physiological fluid. Previous studies have shown that PEGylation of PEI reduced the surface charge of PEI/ DNA particles, increased their dispersion ability at high Plasmid concentrations, decreased plasma protein binding and The plasmid used was pCAGLuc (kindly donated by erythrocyte aggregation, prolonged blood circulation Yoshiharu Matsuura, National Institute of Infectious and reduced systemic toxicity after tail vein injection.17– Diseases, Tokyo, Japan), encoding firefly luciferase 19 Some of these effects may have contributed to the driven by a composite promoter CAG consisting of the improvement in gene expression upon repeated dosing CMV IE enhancer, chicken b-actin promoter and rabbit b- observed in the present study. It has indeed been globin polyadenylation signal. All plasmid DNAs were reported that PEGylation of polycations provides im- amplified in E. coli and purified according to the proved transfection efficiency.20,28–31 supplier’s protocol (Qiagen, Hilden, Germany). The Proteins and peptides modified by PEG usually quantity and quality of the purified plasmid DNA was become less immunogenic and antigenic, because of assessed by optical density at 260 and 280 nm and by reduced immune recognition caused by PEG-induced electrophoresis in 1% agarose gel. The purified plasmid steric hindrance.32–34 Covalent attachment of PEG to the DNA was resuspended in TE buffer and kept in aliquots surface of the adenovirus35,36 reduces both the humoral at a concentration of 1 mg/ml. To construct the noncod- and cellular immune responses against viral proteins, ing plasmid pLuc-s, the SalI/HindIII luciferase fragment allowing substantial gene expression of the vector upon of pGL-3 basic vector (Promega, Madison, WI, USA) was re-administration without compromising the immune excised and replaced by SalI/HindIII CAGLuc cassette system of immunocompetent animals.37 There exists the from pCAGLuc. pLuc-s did not show any luciferase possibility that PEGylation of PEI in the present study expression after in vitro cell transfection and in vivo might alter immune stimulatory effects of CpG motifs in intrathecal injection.

Gene Therapy PEG–PEI and gene expression in the spinal cord L Shi et al 1187 Preparation and characterization of DNA/polymer PC12 cells were cultured in rat tail collagen-coated complexes dishes in RPMI 1640 medium supplemented with 5% The PEI or PEG–PEI/DNA complexes were prepared FCS and 10% heat-inactivated horse serum. For cell according to Boussif et al.4 Plasmid DNA was diluted in viability assay, the cells (10 000 cells/well) were seeded 5% glucose to the chosen concentration (usually 0.5– into 96-well microtiter plates (Nunc, Wiesbaden, Ger- 2 mg/ml). After vortexing, the appropriate amount of many). After 24 h, culture media were replaced with 0.1 M PEI or PEG–PEI was added dropwise into DNA serum-supplemented tissue culture media containing solutions and the solutions revortexed. The required serial dilutions of polymer or polymer/DNA complexes amount of PEI, according to DNA concentration and solutions and the cells were incubated for 24 h. In some number of equivalents needed, was calculated by taking experiments, the cells were treated for 3 h before into account that 1 mg DNA contains 3 nmol of phos- changing back to normal culture media and the cell phate and that 1 ml of 0.1 M PEI holds 100 nmol of amine viability assay was done 21 h later. For cell viability nitrogen. For agarose gel electrophoresis, polymer DNA assay, 20 ml sterile filtered MTT (5 mg/m) stock solution complexes mixed with a loading buffer were loaded onto in PBS was added to each well, reaching a final an ethidium bromide containing 1% agarose gel. Gel concentration of 0.5 mg MTT/ml. After 4 h, unreacted electrophoresis was run at room temperature in TEB dye was removed by aspiration. The formazan crystals buffer at 80 V for 60 min. DNA bands were visualized by were dissolved in 100 ml/well DMSO (BDH Laboratory a UV (254 nm) illuminator. For complex size measure- Supplies, England) and measured spectrophotometri- ments, an N4 Plus Submicron Particle Sizer (COULTER, cally in an ELISA plate reader (Model 550, Bio-Rad) at a USA) was used. Scattering light was detected at 901, wavelength of 595 nm. The spectrophotometer was running of 200 s at room temperature. For data analysis, calibrated to 0 absorbance using culture medium without the refractive index medium (1.332) of 5% glucose was cells. Quintuplicate determinations for each treatment used. The data obtained were analyzed in the Unimodal were performed. The relative cell growth (%) related to Analysis mode. control cells cultured in media without polymer or polymer/DNA complexes was calculated by [A] test/[A] Â Animals and injection procedures control 100. Adult male Wistar rats (8 weeks old, 180–200 g) were Histological examination and immunostaining used throughout the study. Before each injection, the At 2 or 14 days after PEI/DNA intrathecal injection, animals were anesthetized by an i.p. injection of sodium animals were anesthetized with sodium pentobarbital pentobarbital (60 mg/kg). The lumbar injection was and perfused transcardially with saline solution (200 ml) given using a 10 ml micro-syringe connected with a 26- followed by 150 ml of ice-cold fixative (4% paraformal- gauge needle, after the skin around L4–L5 was exposed. dehyde, 0.05% glutaraldehyde in PBS). Tissues in lumbar The slight movement of the tail indicated the proper and thoracic portions were removed, respectively, from injection into the subarachnoid space. For each rat, 20 ml the whole spinal cord and kept at À801C until sectioning. of 5% glucose solution containing 4 mg of plasmid DNA For H&E staining, paraffin-embedded sections were complexed with PEI or PEG–PEI were given by two prepared. For immunostaining, the tissues were cut in injections. After a slow injection of 10 ml of polymer DNA short length and fixed on the plate with tissue freezing complexes over 2–5 min, the syringe was left in place for medium. After setting up on an ultramicrotome at À201C 5 min to limit diffusion of the DNA from the injection site (OT) and À151C (CT), respectively, sections of thickness owing to the backflow pressure. The skin was closed 35 mm were cut. The sections were transferred to a with surgical clips after the injection. The animals were gelatin-coated slide and fixed with acetone. A goat kept warm until they recovered. antiluciferase (1/200, in PBS 1% BSA, Dako) and an FITC-conjugated anti-goat polyclonal antibody were Luciferase activity assay used as the primary and secondary antibodies, respec- Animals were intraperitoneally anesthetized with so- tively. After antibody incubation and washing, the dium pentobarbital and then perfused transcardially sections were mounted in FluorSave (Calbiochem). with 0.1 M phosphate-buffered saline (PBS, 200 ml/rat). Photographs were taken with digital camera attached After the perfusion, the whole spinal cord was exposed. to the microscope. The spinal cord was separated into cervical, thoracic and lumbar/sacral parts before homogenization. For some TUNEL staining experiments, meninges were separated from the corre- For in vitro TUNEL staining, NT2 cells were seeded in a sponding parts of the spinal cord. The tissue samples 24-well plate with glass slides in each well, grown and were homogenized with 400 ml of lysis buffer (Promega), treated as above. The slides were then fixed with a and then subjected to three freeze–thaw cycles. The freshly prepared paraformaldehyde solution (4% in PBS, lysate was centrifuged at 14 000 g for 5 min at 41C, and pH 7.4). After 1 h of fixation at RT, the slides were rinsed 100 ml of supernatant was used for the luciferase activity with PBS and incubated with blocking solution (3% assay at room temperature employing an assay kit from H2O2) for 10 min at RT. The slides were then incubated in Promega. Measurements were made in a single-well a permeabilization solution containing 0.1% Tritons X- luminometer (Berthold Lumat LB 9507, Germany) for 100 in 0.1% sodium citrate for 2 min at 41C. After rinsing 10 s. the slides twice with PBS, 50 ml of TUNEL reaction solution from the TUNEL staining kit (Roche) was added Cell viability assay to each sample. After 60 min incubation at 371C and NT2 cells were cultured in DMEM supplemented with washing, samples were analyzed under a fluorescence

10% FCS at 371C, 5% CO2, and 95% relative humidity. microscope. For the negative control, samples were

Gene Therapy PEG–PEI and gene expression in the spinal cord L Shi et al 1188 incubated in 50 ml of a label solution without terminal blood components, extended circulation in blood and potential transferase. The same procedure was used for the tissue for systemic gene delivery. Gene Therapy 1999; 6: 595–605. sectioning samples from in vivo studies. 18 Nguyen HK et al. Evaluation of polyether–polyethlenimine graft copolymers as gene transfer agents. Gene Therapy 2000; 7: 126–138. Acknowledgements 19 Kichler A et al. Intranasal gene delivery with a polyethylenimine–PEG conjugate. J Controlled Release 2002; 81: 379–388. This work was funded by the Agency for Science, 20 Petersen H et al. Polyethylenimine–graft–poly(ethylene glycol) Technology and Research (A* STAR). copolymers: influence of copolymer block structure on DNA complexation and biological activities as gene delivery dystem. Bioconjug Chem 2002; 13: 845–854. 21 Mannes AJ, Caudle RM, O’Connell BC, Iadarola MJ. Adenoviral References gene transfer to spinal cord neurons: intrathecal vs. intrapar- enchymal administration. Brain Res 1998; 793: 1–6. 1 Costantini LC, Bakowska JC, Breakefield XO, Isacson O. Gene 22 Tan Y, Li S, Pitt BR, Huang L. The inhibitory role of CpG therapy in the CNS. Gene Therapy 2000; 7: 93–109. immunostimulatory motifs in cationic lipid vector-mediated 2 O’Leary MT, Charlton HM. A model for long-term transgene transgene expression in vivo. Hum Gene Ther 1999; 10: 2153–2161. Gene Therapy expression in spinal cord regeneration studies. 23 Fabry Z, Raine CS, Hart MN. Nervous tissue as an immune 1999; 6: 1351–1359. compartment: the dialect of the immune response in the CNS. et al 3 Yamamura J . Long-term gene expression in the anterior Immunol Today 1994; 15: 218–224. horn motor neurons after intramuscular inoculation of a live 24 Stevenson PG, Hawke S, Sloan DJ, Bangham CR. The immuno- Gene Therapy 7 herpes simplex virus vector. 2000; : 934–941. genicity of intracerebral virus infection depends on anatomical 4 Boussif O et al. A versatile vector for gene and oligonucleotide site. J Virol 1997; 71: 145–151. in vivo Proc transfer into cells in culture and : polyethylenimine. 25 Detrait ER et al. Reporter gene transfer induces apoptosis in Natl Acad Sci USA 92 1995; : 7297–7302. primary cortical neurons. Mol Ther 2002; 5: 723–730. et al in vivo 5 Abdallah B . A powerful nonviral vector for gene 26 Wang S et al. Transgene expression in the brain stem effected by transfer into the adult mammalian brain: polyethylenimine. intramuscular injection of polyethylenimine/DNA complexes. Hum Gene Ther 7 1996; : 1947–1954. Mol Ther 2001; 3: 658–664. et al 6 Goula D . Size, diffusibility and transfection performance of 27 Klinman DM, Yamshchikov G, Ishigatsubo Y. Contribution of linear PEI/DNA complexes in the mouse central nervous CpG motifs to the immunogenicity of DNA vaccines. J. Immunol system. Gene Therapy 1998; 5: 712–717. 1997; 158: 3635–3639. 7LiSet al. Effect of immune response on gene transfer to the lung 28 Choia YH et al. Polyethylene glycol-grafted poly–L-lysine as Am J via systemic administration of cationic lipidic vectors. polymeric gene carrier. J Control Release 1998; 54: 39–48. Physiol 1999; 276(Part 1): L796–L804. 29 Lee M et al. Repression of GAD autoantigen expression in 8 Meyer O, Schughart K, Pavirani A, Kolbe HV. Multiple systemic pancreas-b-cells by delivery of antisense plasmid/PEG–g-PLL expression of human interferon-beta in mice can be achieved complex. Mol Ther 2001; 4: 339–346. upon repeated administration of optimized pcTG90-lipoplex. 30 Sagara K, Kim SW. A new synthesis of galactose–poly–L-lysine Gene Therapy 2000; 7: 1606–1611. pancreas-b-cells for gene delivery to hepatocytes. J Control et al 9 Jenkins RG . An integrin-targeted non-viral vector for Release 2002; 79: 271–281. Gene Therapy 7 pulmonary gene therapy. 2000; : 393–400. 31 Ahn CH, Char SY, Bae YH, Kim SW. Biodegradable poly(- 10 Imaoka T, Date I, Ohmoto T, Nagatsu T. Significant behavioral ethylenimine) for plasmid DNA delivery. J Control Release 2002; recovery in Parkinson’s disease model by direct intracerebral 80: 273–282. gene transfer using continuous injection of a plasmid DNA– 32 Francis GE et al. PEGylation of cytokines and other therapeutic Hum Gene Ther 9 liposome complex. 1998; : 1093–1102. proteins and peptides: the importance of biological optimisation 11 Jackson CA et al. Repetitive intrathecal injections of poliovirus of coupling techniques. Int J Hematol 1998; 68: 1–18. replicons result in gene expression in neurons of the central 33 Harris JM, Martin NE, Modi M. Pegylation: a novel process Hum Gene Therapy nervous system without pathogenesis. 2001; for modifying pharmacokinetics. Clin Pharmacokinet 2001; 40: 12 : 1827–1841. 539–551. 12 Ram Z et al. Intrathecal gene therapy for malignant leptome- 34 Hinds KD, Kim SW. Effects of PEG conjugation on insulin Cancer Res 54 ningeal neoplasia. 1994; : 2141–2145. properties. Adv Drug Deliv Rev 2002; 54: 505–530. et al 13 Meuli-Simmen C . Gene expression along the cerebral-spinal 35 Chillon M, Lee JH, Fasbender A, Welsh MJ. Adenovirus Hum Gene Ther 10 axis after regional gene delivery. 1999; : complexed with polyethylene glycol and cationic lipid is 2689–2700. shielded from neutralizing antibodies in vitro. Gene Therapy 14 Martino G, Furlan R, Comi G, Adorini L. The ependymal route 1998; 5: 995–1002. Trends to the CNS: an emerging gene therapy approach for MS. 36 O’Riordan CR et al. PEGylation of adenovirus with retention of Immunol 2001; 22: 483–490. infectivity and protection from neutralizing antibody in vitro and Poly(ethyleneglycol): Chemistry and 15 Harris JM, Zalipsky S (eds). in vivo. Hum Gene Ther 1999; 10: 1349–1358. Biological Applications. ACS Book: Washington, DC, 1997. 37 Croyle MA, Chirmule N, Zhang Y, Wilson JM. ‘Stealth’ 16 Godbey WT, Wu KK, Mikos AG. Poly(ethylenimine)-mediated adenoviruses blunt cell-mediated and humoral immune re- gene delivery affects endothelial cell function and viability. sponses against the virus and allow for significant gene Biomaterials 22 2001; : 471–480. expression upon readministration in the lung. J Virol 2001; 75: 17 Ogris M, Brunner S, Schuller S, Kircheis R, Wagner E. PEGylated 4792–4801. DNA/transferrin–PEI complexes: reduced interaction with

Gene Therapy