- Email: [email protected]
CpG-Depleted Plasmid DNA Vectors with Enhanced Safety and Long-Term Gene Expression in Vivo
doi:10.1006/mthe.2002.0598, available online at http://www.idealibrary.com on IDEAL
ARTICLE
CpG-Depleted Plasmid DNA Vectors with Enhanced Safety and Long-Term Gene Expression in Vivo Nelson S. Yew,* Hongmei Zhao, Malgorzata Przybylska, I-Huan Wu, Jennifer D. Tousignant, Ronald K. Scheule, and Seng H. Cheng Genzyme Corporation, 31 New York Avenue, Framingham, Massachusetts 01701-9322, USA *To whom correspondence and reprint requests should be addressed. Fax: (508) 872-4091. E-mail: [email protected].
Systemic delivery of cationic lipid–plasmid DNA (pDNA) complexes induces an acute inflammatory response with adverse hematologic changes and liver damage. Immunostimulatory CpG motifs in the pDNA are known to contribute substantially to this response. Here we constructed a pDNA vector (pGZB) that has been depleted of 80% of the CpG motifs present in the original vector. Compared with the unmodified vector, systemic administration of pGZB induced considerably fewer changes in blood parameters, reduced levels of inflammatory cytokines, and decreased liver damage. Despite the extensive sequence modifications within pGZB, there were still robust levels of transgene expression. Furthermore, in contrast to the transient expression observed from the unmodified vector, we observed sustained or increasing expression for up to 49 days from pGZB in the lung and liver of immunocompetent BALB/c mice. Studies adding CpG motifs in trans or in cis indicate that the reduced CpG content of pGZB was the major determinant for its persistent expression. This combination of decreased toxicity and sustained expression suggests that CpG-depleted pDNA vectors can greatly improve the safety and efficacy of synthetic gene delivery systems. Key Words: plasmids, CpG motifs, gene expression, inflammation, gene therapy
INTRODUCTION A potentially efficacious and relatively simple mode of gene therapy involves systemic delivery of complexes composed of cationic lipids or polymers and plasmid DNA [1]. Two problems associated with this approach are the toxicity of the complex and the lack of sustained transgene expression from the plasmid. Mice injected intravenously (i.v.) with cationic lipid–pDNA complexes show an acute inflammatory response, which includes complement activation and induction of the cytokines interleukin-12 (IL12), IL-6, interferon-␥ (IFN-␥,) and tumor necrosis factor␣ (TNF-␣) [2–5]. There are also marked hematologic and serologic changes, such as leukopenia, thrombocytopenia, and elevated levels of serum transaminases indicative of liver damage [5,6]. Though transient and dose-dependent, these harmful effects are unacceptably serious at the doses required to achieve therapeutic levels of expression. The second problem is that expression from current pDNA vectors is short-lived, peaking at day 1 after systemic delivery, then declining precipitously over the next 2 weeks to basal levels [7]. This is most often observed with vectors that contain the immediate-early gene promoter from cytomegalovirus (CMV), although other promoters are also
MOLECULAR THERAPY Vol. 5, No. 6, June 2002 Copyright © The American Society of Gene Therapy 1525-0016/02 $35.00
prone to inactivation over time [8]. The levels of transgene expression that can be attained with current synthetic vectors are generally quite low, and the lack of sustained expression further limits the ability to provide therapeutic levels of a given protein for any extended period of time. We hypothesized that the observed toxicity and transient expression may be partly due to the immunostimulatory properties of pDNA. Bacterial DNA and bacterially derived pDNA have the expected mathematical frequency of CpG dinucleotides, and the DNA is predominantly unmethylated, whereas in mammalian DNA the frequency of CpG dinucleotides is suppressed and largely methylated [9]. Recognition of these differences by the host leads to a pleiotropic inflammatory response that includes the activation of B cells, monocytes, macrophages, dendritic cells, and natural killer cells [10–14]. We and others have shown previously that immunostimulatory CpG motifs within the pDNA vector contribute substantially to the induction of proinflammatory cytokines by cationic lipid–pDNA complexes instilled into the lung [3,15,16]. The cytokine response induced by the complex can be partially suppressed with dexamethasone, by administering antibodies to the cytokines TNF-␣ and IFN-␥, or by inhibiting
731
ARTICLE
doi:10.1006/mthe.2002.0598, available online at http://www.idealibrary.com on IDEAL
FIG. 1. Diagram of CpG-depleted plasmid DNA vectors and transgenes. CMV, cytomegalovirus immediate-early gene enhancer/promoter; HI, hybrid intron; CAT, chloramphenicol acetyltransferase cDNA; pA, bovine growth hormone polyadenylation signal; ORI, replication origin region; KAN, kanamycin resistance gene; sHFIX, synthetic human factor IX cDNA. Each symbol represents five CpG motifs.
neutrophil influx using anti-Mac-1␣ and anti-LFA-1 antibodies [3,16,17]. These treatments were also shown to cause a marked increase in transgene expression levels [3,16,17]. Another strategy has been to decrease the overall number of CpG motifs within the pDNA, and we showed previously that reducing the CpG content of the vector reduced considerably the induction of several proinflammatory cytokines [18]. Here we have further reduced the CpG content of several pDNA vectors by replacing large portions of the vectors with synthetic, non-CpG sequence. We evaluated the acute toxicity induced by these vectors after systemic administration, measuring cytokine levels, cell numbers in the blood, and liver enzymes. We also determined the level and duration of expression in the lung and liver. Finally, we carried out studies to determine the mechanism for the greatly improved performance of these CpGdepleted vectors.
RESULTS Reduced Toxicity of CpG-Depleted pDNA Vectors after Systemic Delivery into BALB/c Mice We previously constructed a pDNA vector, pGZA-sCAT, that contained reduced numbers of CpG motifs (256 versus 526 CpGs in the unmodified vector, counting both strands of the pDNA [18]). This was achieved by minimizing the origin region required for replication of the plasmid, inserting a synthetic, non-CpG kanamycin resistance gene, and replacing the chloramphenicol acetyltransferase (CAT) reporter gene with a synthetic, non-CpG and codon-optimized counterpart (sCAT). To construct pGZB-sCAT, we eliminated an additional 154 CpGs by replacing the CMV immediate-early gene enhancer and promoter region with a synthetic, non-CpG version, and
732
by replacing the hybrid intron and polyadenylation signal with their non-CpG counterparts (Fig. 1). The only nonsynthetic portion remaining is the minimal region required for plasmid replication. pGZB-sCAT contains 102 CpGs, 19% of the CpG content of the original pCF1-CAT vector. To assess the acute toxicity of cationic lipid–pDNA complexes containing this CpG-depleted vector, we injected BALB/c mice with cationic lipid GL-62 complexed to pCF1-CAT, pGZA-sCAT, or pGZB-sCAT (GL-62:pDNA ratio, 0.5:0.5 mM; 16.5 g of pDNA). Blood and lungs were collected 24 hours after injection for analysis. As reported earlier [5], systemic administration of complexes containing pCF1-CAT resulted in a loss of leukocytes and platelets from the serum (Fig. 2). Similar losses were observed in mice that received complexes containing the partially CpG-reduced pGZA-sCAT. In contrast, the loss of leukocytes and platelets was less in mice that received complexes containing the most CpG-reduced vector, pGZBsCAT. Levels of the serum transaminases ALT and AST were highly elevated in mice that received pCF1-CAT or pGZA-sCAT complexes, indicative of liver damage and necrosis. In contrast, the levels of ALT and AST in mice that received pGZB-sCAT were only marginally elevated. In addition, the levels of the proinflammatory cytokines IL-12 and IFN-␥ were considerably lower in mice that received pGZB-sCAT compared with those that received pCF1-CAT, with cytokine levels closely correlating with the CpG content of the pDNA vector. Taken together, these results indicate that the CpG content of the pDNA affects several indicators of toxicity in addition to cytokines, and that the CpG-depleted pGZB-sCAT vector induced significantly less toxicity compared with both the unmodified pCF1-CAT vector and the partially CpGreduced pGZA-sCAT.
MOLECULAR THERAPY Vol. 5, No. 6, June 2002 Copyright © The American Society of Gene Therapy
doi:10.1006/mthe.2002.0598, available online at http://www.idealibrary.com on IDEAL
ARTICLE
FIG. 2. Comparison of the toxicity induced by unmodified and CpG-depleted pDNA vectors after systemic delivery. BALB/c mice were injected i.v. with 100 l of cationic lipid GL-62 complexed with pGZB-sCAT, pGZA-sCAT, or pCF1CAT (0.5:0.5 mM ratio of GL-62:pDNA; 16.5 g of pDNA). Blood was collected and lungs were harvested 24 hours after injection. Cell differentials (leukocytes, platelets) and liver enzymes (AST, ALT) were measured from the blood of five of eight mice per group. Cytokines (IL-12 and IFN-␥) were assayed from the remaining three mice per group. The levels of CAT in the lung were measured from all mice (n = 8 mice per plasmid). Water was used as the vehicle control. Data are expressed as mean ± SEM.
We also measured the levels of CAT in the lung 1 day after injection of complex. Higher levels of CAT were observed from both pGZA-sCAT and pGZB-sCAT compared with pCF1-CAT (Fig. 2). Lower levels of CAT were observed from pGZB-sCAT compared with pGZA-sCAT, suggesting that the synthetic CMV promoter was less active than the wild-type promoter. However, the levels were still substantially above that of the original pCF1-CAT vector, indicating that much of the original promoter activity was retained despite the extensive sequence alterations. Mice were also injected with complex at a dose of 1:1 mM (GL-62:pDNA ratio; 33 g of pDNA). At this higher dose there were no longer measurable differences between the unmodified and CpG-depleted vectors with regard to the loss of leukocytes and platelets, but liver enzyme levels and cytokines were reduced with the CpG-depleted vectors (Fig. 3). These results indicate that complexes containing the pGZA and pGZB vectors were still inflammatory, but that CpG reduction continued to be beneficial even at this higher dose. Persistent CAT Expression from pGZB-sCAT in the Lung Besides toxicity, another limiting issue with synthetic vectors is the lack of sustained transgene expression. We evaluated the persistence of expression from pGZB-sCAT in the lung after systemic delivery. Complexes of cationic lipid GL-62 and pCF1-CAT, pGZA-sCAT, or pGZB-sCAT were injected i.v. into BALB/c mice, and expression was measured for 35 days. Expression from pCF1-CAT declined rapidly within the first 2 weeks, which is characteristic for unmodified vectors that contain the CMV promoter (Fig. 4A) [8]. Expression from pGZA-sCAT was initially higher than that from pCF1-CAT, declined over the first 3 weeks, and then increased between day 21 and day 35. Expression from pGZB-sCAT also declined within the first week but then increased, with levels of CAT at day 35 higher than day 1 levels. These results indicate that the CpG-reduced vectors pGZA-sCAT and pGZB-sCAT displayed dramatically different profiles of expression over time compared with the unmodified pCF1-CAT vector, with the most CpG-reduced pGZB vector having the highest expression over time. Interestingly, we did not observe the generation of anti-CAT antibodies in any group (data not shown), so differences in these expression profiles are not a result of differential antibody response to the transgene.
MOLECULAR THERAPY Vol. 5, No. 6, June 2002 Copyright © The American Society of Gene Therapy
Cumulative Expression after Repeat Administration of pGZB-sCAT The profile observed after a single injection of pGZB-sCAT indicated that repeat administration might be particularly effective in generating a durable increase in CAT levels over that which could be achieved with a single dose. To test this possibility, complexes of cationic lipid GL-62 and
733
ARTICLE
doi:10.1006/mthe.2002.0598, available online at http://www.idealibrary.com on IDEAL
FIG. 3. Comparison of the toxicity induced by unmodified and CpG-depleted pDNA vectors at a higher dose of complex. BALB/c mice were injected i.v. with 100 l of cationic lipid GL-62 complexed with pGZB-sCAT, pGZA-sCAT, or pCF1-CAT (1.0:1.0 mM ratio of GL-62:pDNA; 33.0 g of pDNA). Blood was collected and lungs were harvested 24 hours after injection. Cell differentials (leukocytes, platelets) and liver enzymes (AST, ALT) were measured from the blood of five of eight mice per group. Cytokines (IL-12 and IFN-␥) were assayed from the remaining three mice per group. The levels of CAT in the lung were measured from all mice (n = 8 mice per plasmid). Water was used as the vehicle control. Data are expressed as mean ± SEM.
from a single dose. These data indicate that redosing with pGZB-sCAT was effective in providing cumulatively higher levels of CAT than could be achieved with a single administration of complex.
either pCF1-CAT or pGZB-sCAT were injected i.v. into BALB/c mice. As noted earlier, CAT expression from pGZBsCAT increased over the first 28 days, and then reached a plateau between day 28 and day 49 (Fig. 4B). The levels of CAT in this study were unusually high, possibly because of the variable transfection efficiency of different lots of cationic lipid. Some groups of mice were given a second dose of either GL-62:pCF1-CAT or GL-62:pGZB-sCAT 14 days after the first dose, and expression in the lung was followed for an additional 35 days. Expression from mice that received a second dose of pCF1-CAT increased transiently but then declined to undetectable levels. In contrast, expression from mice given a second dose of pGZBsCAT increased at day 1 after redosing and continued to increase through day 49 (Fig. 4B). The levels at day 49 were 270% higher than the maximum levels observed
734
Improved Persistence of CAT and Factor IX Expression from pGZB Vectors in the Liver A second major target organ for both viral and synthetic gene delivery vectors is the liver. To determine the level and persistence of expression from pGZB-sCAT in liver, pGZBsCAT was delivered using a hydrodynamics-based protocol that efficiently transduces hepatocytes [19,20]. pGZB-sCAT or pCF1-CAT pDNA was injected through the tail vein into BALB/c mice, and the levels of CAT in the liver were measured at 1, 14, and 28 days after injection (Fig. 5A). As observed in the lung, expression from pCF1-CAT declined sharply within the first 2 weeks. Expression from pGZB-sCAT also declined, but at a significantly slower rate compared to pCF1-CAT. These results indicate that pGZB-sCAT could also provide increased persistence of expression in liver. We also evaluated a pGZB vector that expresses a secreted protein, factor IX. The vector pGZB-sHFIX, which has a synthetic, non-CpG human factor IX cDNA, contains 96 CpGs, compared with the vector pCFA-HFIX, which has the wild-type factor IX sequence and contains 488 CpGs (Fig. 1). pGZB-sHFIX or pCFA-HFIX pDNA was injected using the hydrodynamics-based protocol into BALB/c mice, and the levels of factor IX in the plasma were measured at 14, 28, and 42 days after injection (Fig. 5B). The levels of factor IX from pCFA-HFIX declined to essentially background levels by day 14 and remained low. In contrast, the levels of factor IX from pGZB-sHFIX declined ~ 10-fold by day 14, but then remained steady at ~ 1 g/ml to day 42, the last time point. Hence, considerable improvements in the longevity of expression of both a secreted (HFIX) and nonsecreted transgene (CAT) were observed in liver using the pGZB vector. CpG Content as a Major Determinant of Sustained Expression The basis for the improvement in the longevity of expression was not obvious from the above studies. One possibility was that the sequence changes made within the CMV enhancer/promoter somehow rendered the promoter insensitive to inactivation over time. A second possibility was that the reduced CpG content was mainly responsible
MOLECULAR THERAPY Vol. 5, No. 6, June 2002 Copyright © The American Society of Gene Therapy
ARTICLE
doi:10.1006/mthe.2002.0598, available online at http://www.idealibrary.com on IDEAL
A
B
FIG. 4. Expression kinetics and redosing of pGZB-sCAT. (A) Persistence of CAT expression from pGZB-sCAT in the lung after i.v. delivery into BALB/c mice. BALB/c mice were injected i.v. with 100 l of cationic lipid GL-62 complexed with pGZB-sCAT, pGZA-sCAT, or pCF1-CAT (0.5:0.5 mM ratio of GL-62:pDNA; 16.5 g of pDNA). Lungs were harvested at different days after injection, and CAT assays were done (n = 4 mice per group). (B) Increased expression upon repeat i.v. delivery of pGZB-sCAT. BALB/c mice were injected i.v. with 100 l of cationic lipid GL-62 complexed with either pGZB-sCAT or pCF1-CAT (0.5:0.5 mM ratio of GL-62:pDNA). Some groups of mice were injected with a second dose of either GL-62:pGZBsCAT or GL-62:pCF1-CAT at day 14 after the initial injection (arrow). Lungs were harvested at different days after injection, and CAT assays were done (n = 4 mice per group). Data are expressed as mean ± SEM.
day 14, while expression from pGZAsCAT and psCFA-CAT recovered slightly with low but detectable levels of CAT at day 35. These results show that the synthetic CMV promoter in the context of a CpG-rich backbone is not sufficient to confer the substantial increase in expression over time seen with pGZB-sCAT. The data imply that the CpG content of the vector was the most important variable determining the expression profile.
DISCUSSION
for the sustained promoter activity. We conducted two studies in an attempt to distinguish between these possibilities. First, BALB/c mice were injected with cationic lipid–pDNA complexes containing either pGZB-sCAT alone, or pGZB-sCAT plus pOri-Kan, a plasmid that contains an unmodified kanamycin resistance gene and bacterial replication origin (Fig. 1). The pOri-Kan plasmid contains 288 CpGs, adding in trans a large number of CpGs to the dose given to the mice. The levels of CAT in the lung were measured for 35 days. Expression from pGZB-sCAT increased over time (Fig. 6A). In contrast, expression from mice that received both pGZB-sCAT plus pOri-Kan declined to undetectable levels by day 14, then slowly recovered at the later time points. These results show that the addition of a CpG-rich plasmid in trans had a strong negative effect on the persistence of expression. In a second study we constructed the plasmid psCFACAT, which is identical to pCF1-CAT with the exception that the wild-type CMV enhancer/promoter was replaced with the synthetic, non-CpG CMV enhancer/promoter from pGZB-sCAT (Fig. 1). psCFA-CAT contains 448 CpGs. BALB/c mice were injected i.v. with cationic lipid complexes containing pCF1-CAT, pGZA-sCAT, pGZB-sCAT, or psCFA-CAT, and the levels of CAT in the lung were measured for 35 days. Expression from pCF1-CAT, pGZA-sCAT, and psCFA-CAT all declined over the first 14 days (Fig. 6B). Expression from pCF1-CAT continued to decline after
MOLECULAR THERAPY Vol. 5, No. 6, June 2002 Copyright © The American Society of Gene Therapy
In this study we have shown that a pDNA vector that has been depleted in CpG content could overcome to a large extent two important limitations of current synthetic vectors, namely, the acute toxicity and transient expression. Compared with the unmodified vector, systemic delivery of cationic lipid complexes containing the CpG-depleted pGZB vector induced less inflammation, fewer hematologic changes, and decreased liver toxicity. The pGZB vector also displayed the striking property of sustained and even increasing expression over time. These improvements in safety and persistence of expression are vital steps toward making synthetic vectors truly effective gene delivery vehicles. Earlier studies have demonstrated the substantial role of CpG motifs in the acute inflammatory response to cationic lipid–pDNA complexes. Systemic administration of complex induces several proinflammatory cytokines, hematologic changes, and elevation of liver enzymes and acute-phase response proteins [3,5,6,21]. These toxicities are not observed after intravenous injection of either cationic lipid alone or pDNA alone, implicating cationic lipid complexed to pDNA as the causative agent [3,6,22]. Methylating the pDNA in the complex substantially decreases cytokine induction, indicating that the mammalian innate immune system is recognizing the unmethylated CpGs in the pDNA and is likely triggering the cascade of responses [23]. Therefore, one could predict that removal of CpGs from the pDNA would lessen the recognition of the pDNA as foreign, and consequently reduce the subsequent deleterious effects.
735
ARTICLE
doi:10.1006/mthe.2002.0598, available online at http://www.idealibrary.com on IDEAL
A
B
FIG. 5. Persistence of transgene expression from pGZB vectors in the liver. (A) BALB/c mice were injected rapidly through the tail vein with 5 g of pGZB-sCAT or pCF1-CAT in 2 ml of Mirus delivery solution. Livers were harvested at 1, 14, and 28 days after injection, and CAT assays were done (n = 4 mice per time point). (B) BALB/c mice were injected rapidly through the tail vein with 10 g of pGZB-sHFIX or pCFA-HFIX in 2 ml of Mirus delivery solution. Plasma was taken at 1, 14, 28, and 42 days after injection, and the levels of HFIX were measured (n = 5 mice per time point). Data are expressed as mean ± SEM.
The results of these studies agree with this hypothesis, but substantial numbers of CpGs remain even in our most CpG-depleted vector, and the remaining toxicity is not negligible. Preliminary experiments have been done with higher doses of complex than those reported here. Mice given a dose of 2:2 mM (GL-62:pDNA ratio; 66 g of pDNA) using the unmodified pDNA vectors exhibited considerable mortality, whereas mice injected at the same dose using the pGZB vectors typically survived (data not shown). However, there were outward signs of ill effects at this dose (for example, ruffled coat, inactivity), indicating that the range of tolerable doses, though expanded,
A
B
is still quite narrow with the CpG-reduced vectors. It is possible to create a nearly non-CpG vector by PCR amplification of only the expression cassette portion of the pDNA vector, that is, the promoter, intron, cDNA, and polyadenylation signal. Linear PCR fragments have been shown to elicit a reduced inflammatory response when delivered systemically as a lipid–protamine–DNA complex [24]. It remains to be determined whether the toxicity can be completely eliminated or that factors other than CpG motifs contribute to the observed host response. An additional and somewhat surprising consequence of CpG reduction was that the pGZB vector retained substantial transcriptional activity. The pGZB vector contains 31 nucleotide changes within the CMV enhancer/promoter region, several of which reside within transcription factor–binding sites, notably within the sites for Sp1, ATF/CREB, and NFB. However, factor recognition and binding require only a subset of base contacts [25]. The CMV enhancer region is also highly redundant, with multiple binding sites for each factor, and deletion of portions of the enhancer led to only a partial loss of activity [26]. The activity of the modified CMV enhancer/promoter was reduced relative to the wild-type sequence, but it may be possible to restore some of the activity by inserting additional binding sites or other enhancer sequences that contain fewer CpG motifs.
FIG. 6. Effect of CpGs in trans or in cis on the persistence of transgene expression. (A) BALB/c mice were injected i.v. with 100 l of cationic lipid GL-62 complexed with either pGZB-sCAT alone or a 1:1 (wt/wt) mixture of pGZB-sCAT + pOri-Kan (0.5:0.5 mM ratio of GL-62:pDNA; 16.5 g of pDNA). Lungs were harvested at 1, 7, 14, 21, and 35 days after injection, and CAT assays were done (n = 5 mice per time point). (B) BALB/c mice were injected i.v. with 100 l of cationic lipid GL-62 complexed with pCF1-CAT, pGZA-sCAT, pGZB-sCAT, or psCFA-CAT (0.5:0.5 mM ratio of GL-62:pDNA). Lungs were harvested at 1, 7, 14, 21, and 35 days after injection, and CAT assays were done (n = 4 mice per time point). Data are expressed as mean ± SEM.
736
MOLECULAR THERAPY Vol. 5, No. 6, June 2002 Copyright © The American Society of Gene Therapy
doi:10.1006/mthe.2002.0598, available online at http://www.idealibrary.com on IDEAL
The partial loss of transcriptional activity was more than compensated for by the sustained character of pGZB expression, the mechanism of which remains to be determined. Studies with the vector containing the synthetic CMV enhancer/promoter in the unmodified backbone (psCFACAT) indicate that the synthetic CMV promoter alone is not sufficient for the increased persistence. Rather, the overall reduced CpG content of pGZB appears to be a strong determinant. The decreased toxicity of pGZB could result in the survival of more plasmid-bearing cells, with consequently more expression over time. Alternatively, the decreased toxicity may permit the gradual formation of more active transcription factor complexes on the pGZB plasmids, increasing expression. Other promoters that appear to have greater longevity, such as the ubiquitin promoter or liver-specific promoters, may be more resistant to any possible inflammation-induced inactivating factors [27–29]. These data suggest that potentially therapeutic levels of transgene expression over an extended period may now be achievable with synthetic vectors. The decreased toxicity may allow the use of more concentrated lipid formulations, resulting in greater levels of expression. In addition, the ability to redose effectively may enable the gradual accumulation of expression into the therapeutic range. CpG depletion alone will not solve all the limiting issues of synthetic vectors, and targeting complexes away from cells involved in immune recognition will also likely be necessary [30]. However, CpG-depleted vectors do provide a platform for further improvements in the safety and utility of synthetic gene delivery.
MATERIALS
AND
METHODS
Plasmid vector construction. To construct pGZB-sCAT, a 1.1-kb DNA SphI fragment was synthesized by Entelechon (Regensburg, Germany); this fragment encompasses the enhancer/promoter from the CMV immediateearly gene (–522 to +16 relative to the transcription start site), hybrid intron, and bovine growth hormone polyadenylation signal. In this sequence all the CG dinucleotides were changed to TG. The plasmid pOriK-syn [18], which contains the minimal origin of replication and synthetic kanamycin resistance gene, was linearized with SphI, and the 1.1-kb synthetic fragment was inserted to create pGZB. The synthetic CAT cDNA was excised from pGZA-sCAT [18] by digestion with NotI, blunted with the Klenow fragment of DNA polymerase I, and ligated into the blunted EcoRI site of pGZB to create pGZB-sCAT. To construct pCFA-HFIX, DNA encoding human factor IX (gene symbol F9) was obtained by amplification of human liver cDNA (QuickClone cDNA, Clontech, Palo Alto, CA) using PCR with the Advantage polymerase (Clontech). The primers used for amplification were as follows: 5⬘-CCGGATCCTATGCAGCGCGTGAACATGATCATGGC-3⬘ and 5⬘-GGCTCGAGTCCATCTTTCATTAAGTGAGCTTTGTTT-3⬘ (bold letters refer to added BamHI and XhoI sites). The resulting PCR fragment was cloned into pcDNA3.0 (Invitrogen, Carlsbad, CA). The BamHI–XhoI fragment was excised from pcDNA3.0, blunted using the Klenow fragment of DNA polymerase I, and then ligated into the NotI site of pGZA to create pGZA-HFIX. An SphI fragment of pCFA containing the kanamycin resistance gene and replication origin was isolated and inserted in place of the corresponding SphI fragment from pGZA-HFIX to create pCFA-HFIX. To construct pGZB-sHFIX, 700-bp and 745-bp fragments were synthesized by Entelechon and ligated together to create a 1.4-kb fragment encoding human factor IX. The sequence was designed to incorporate codons that
MOLECULAR THERAPY Vol. 5, No. 6, June 2002 Copyright © The American Society of Gene Therapy
ARTICLE
are present in highly expressed human genes, to eliminate rare codons, and to remove all CG dinucleotides. EcoRI and SfiI sites were incorporated into the 5⬘ and 3⬘ ends of the 1.4-kb fragment, respectively. The 1.4-kb fragment was ligated into the EcoRI and SfiI sites of pGZB to create pGZB-sHFIX. To construct psCFA-CAT, pCFA-CAT [16] was digested with MfeI and XbaI and the backbone vector fragment was isolated. pGZB-sCAT was digested with MfeI and XbaI and the synthetic CMV promoter was isolated, and then ligated into pCFA-CAT to create psCFA-CAT. Cationic lipids and plasmid DNA. Cationic lipid GL-62 (N1-spermine cholesterylcarbamate) was formulated as described [27,31]. Plasmid DNA was purified by alkali lysis, ultrafiltration, and column chromatography [31]. The purified plasmid DNA contained < 10 g Escherichia coli chromosomal DNA per mg pDNA and < 10 endotoxin units per mg pDNA as determined by the chromogenic Limulus amebocyte lysate assay (BioWhittaker, Walkersville, MD). Administration of pDNA and cationic lipid–pDNA complexes into mice. Mice were injected through the tail vein with 100 l of cationic lipid GL62:pDNA as described [5]. Using the average nucleotide mass of 330 Daltons, the amount of pDNA delivered at a dose of 0.5:0.5 mM (GL62:pDNA ratio) is 16.5 g. For delivery to the liver, injections were done by the Mirus Corporation (Madison, WI) using the rapid, high-volume protocol described by Zhang et al. [19]. Briefly, 5–10 g of pDNA was mixed with 10 l of TransIT In Vivo Polymer Solution (Mirus Corporation) and then diluted into ~ 2 ml of Mirus Delivery Solution. The solution was injected rapidly into the tail vein over a period of 6–8 seconds. The pDNA has been shown to be taken up primarily by the liver hepatocytes using this procedure [20]. Cytokine and toxicity assays. Mice were anesthetized 24 hours after injection of the cationic lipid–pDNA complexes, and blood was collected by retro-orbital bleed using heparinized microcapillary tubes. Whole-blood and serum samples were analyzed by IDEXX Veterinary Services (West Sacramento, CA). Levels of the cytokines IL-12 and IFN-␥ were measured by ELISA kits using protocols supplied by the manufacturer (R&D Systems, Minneapolis, MN). Assays for transgene expression. CAT enzyme activity was measured from lung homogenates as described [31]. Levels of factor IX were measured using an ELISA kit (Asserachrom IX:Ag) following the protocol supplied by the manufacturer (Diagnostica Stango, Asnieres-Sur-Seine, France). Purified human factor IX protein (Sigma, St. Louis, MO) was used as a standard. Plasma samples were diluted 1:50 into phosphate buffer (1⫻ Reagent 4 from the kit) before conducting the assay.
ACKNOWLEDGMENTS We thank our colleagues at Genzyme Corporation: Richard Gregory for critical reading of the manuscript; Sirkka Kyöstiö-Moore for the factor IX cDNA; Brad Hodges for the factor IX assay; Seann O’Connell, Caroline DiCesare, and Nick Wan for purified plasmid DNA; and the Animal Care and Technical Services staff for assistance with the in vivo studies. RECEIVED FOR PUBLICATION DECEMBER 7, 2001; ACCEPTED MARCH 12, 2002.
REFERENCES 1. Nishikawa, M., and Huang, L. (2001). Nonviral vectors in the new millennium: delivery barriers in gene transfer. Hum. Gene Ther. 12: 861–870. 2. Plank, C., Mechtler, K., Szoka, F. C., Jr., and Wagner, E. (1996). Activation of the complement system by synthetic DNA complexes: a potential barrier for intravenous gene delivery. Hum. Gene Ther. 7: 1437–1446. 3. Li, S., et al. (1999). Effect of immune response on gene transfer to the lung via systemic administration of cationic lipidic vectors. Am. J. Physiol. 276: L796–804. 4. Dow, S. W., et al. (1999). Lipid–DNA complexes induce potent activation of innate immune responses and antitumor activity when administered intravenously. J. Immunol. 163: 1552–1561. 5. Tousignant, J. D., et al. (2000). Comprehensive analysis of the acute toxicities induced by systemic administration of cationic lipid:plasmid DNA complexes in mice. Hum. Gene Ther. 11: 2493–2513. 6. Loisel, S., Le Gall, C., Doucet, L., Ferec, C., and Floch, V. (2001). Contribution of plasmid DNA to hepatotoxicity after systemic administration of lipoplexes. Hum. Gene Ther. 12: 685–696.
737
ARTICLE
doi:10.1006/mthe.2002.0598, available online at http://www.idealibrary.com on IDEAL
7. Song, Y. K., Liu, F., Chu, S., and Liu, D. (1997). Characterization of cationic liposomemediated gene transfer in vivo by intravenous administration. Hum. Gene Ther. 8: 1585–1594. 8. Loser, P., Jennings, G. S., Strauss, M., and Sandig, V. (1998). Reactivation of the previously silenced cytomegalovirus major immediate-early promoter in the mouse liver: involvement of NFB. J. Virol. 72: 180–190. 9. Krieg, A. M. (1999). Direct immunologic activities of CpG DNA and implications for gene therapy. J. Gene Med. 1: 56–63. 10. Krieg, A. M., et al. (1995). CpG motifs in bacterial DNA trigger direct B-cell activation. Nature 374: 546–549. 11. Klinman, D. M., Yi, A. K., Beaucage, S. L., Conover, J., and Krieg, A. M. (1996). CpG motifs present in bacteria DNA rapidly induce lymphocytes to secrete interleukin 6, interleukin 12, and interferon-␥. Proc. Natl. Acad. Sci. USA 93: 2879–2883. 12. Ballas, Z. K., Rasmussen, W. L., and Krieg, A. M. (1996). Induction of NK activity in murine and human cells by CpG motifs in oligodeoxynucleotides and bacterial DNA. J. Immunol. 157: 1840–1845. 13. Sparwasser, T., et al. (1997). Macrophages sense pathogens via DNA motifs: induction of tumor necrosis factor-␣-mediated shock. Eur. J. Immunol. 27: 1671–1679. 14. Sparwasser, T., et al. (1998). Bacterial DNA and immunostimulatory CpG oligonucleotides trigger maturation and activation of murine dendritic cells. Eur. J. Immunol. 28: 2045–2054. 15. Freimark, B. D., et al. (1998). Cationic lipids enhance cytokine and cell influx levels in the lung following administration of plasmid: cationic lipid complexes. J. Immunol. 160: 4580–4586. 16. Yew, N. S., et al. (1999). Contribution of plasmid DNA to inflammation in the lung after administration of cationic lipid:pDNA complexes. Hum. Gene Ther. 10: 223–234. 17. Tan, Y., Li, S., Pitt, B. R., and Huang, L. (1999). The inhibitory role of CpG immunostimulatory motifs in cationic lipid vector-mediated transgene expression in vivo. Hum. Gene Ther. 10: 2153–2161. 18. Yew, N. S., et al. (2000). Reduced inflammatory response to plasmid DNA vectors by elimination and inhibition of immunostimulatory CpG motifs. Mol. Ther. 1: 255–262.
738
19. Zhang, G., Budker, V., and Wolff, J. A. (1999). High levels of foreign gene expression in hepatocytes after tail vein injections of naked plasmid DNA. Hum. Gene Ther. 10: 1735–1737. 20. Liu, F., Song, Y., and Liu, D. (1999). Hydrodynamics-based transfection in animals by systemic administration of plasmid DNA. Gene Ther. 6: 1258–1266. 21. Norman, J., et al. (2000). Liposome-mediated, nonviral gene transfer induces a systemic inflammatory response which can exacerbate pre-existing inflammation. Gene Ther. 7: 1425–1430. 22. Scheule, R. K., et al. (1997). Basis of pulmonary toxicity associated with cationic lipidmediated gene transfer to the mammalian lung. Hum. Gene Ther. 8: 689–707. 23. Hemmi, H., et al. (2000). A Toll-like receptor recognizes bacterial DNA. Nature 408: 740–745. 24. Hofman, C. R., Dileo, J. P., Li, Z., Li, S., and Huang, L. (2001). Efficient in vivo gene transfer by PCR-amplified fragment with reduced inflammatory activity. Gene Ther. 8: 71–74. 25. Pabo, C. O., and Sauer, R. T. (1992). Transcription factors: structural families and principles of DNA recognition. Annu. Rev. Biochem. 61: 1053–1095. 26. Boshart, M., et al. (1985). A very strong enhancer is located upstream of an immediate early gene of human cytomegalovirus. Cell 41: 521–530. 27. Yew, N. S., Przybylska, M., Ziegler, R. J., Liu, D., and Cheng, S. H. (2001). High and sustained transgene expression in vivo from plasmid vectors containing a hybrid ubiquitin promoter. Mol. Ther. 4: 75–82. 28. Herweijer, H., et al. (2001). Time course of gene expression after plasmid DNA gene transfer to the liver. J. Gene Med. 3: 280–291. 29. Qin, L., et al. (1997). Promoter attenuation in gene therapy: interferon-␥ and tumor necrosis factor-␣ inhibit transgene expression. Hum. Gene Ther. 8: 2019–2029. 30. Hwang, S. H., Hayashi, K., Takayama, K., and Maitani, Y. (2001). Liver-targeted gene transfer into a human hepatoblastoma cell line and in vivo by sterylglucoside-containing cationic liposomes. Gene Ther. 8: 1276–1280. 31. Lee, E. R., et al. (1996). Detailed analysis of structures and formulations of cationic lipids for efficient gene transfer to the lung. Hum. Gene Ther. 7: 1701–1717.
MOLECULAR THERAPY Vol. 5, No. 6, June 2002 Copyright © The American Society of Gene Therapy
ARTICLE
CpG-Depleted Plasmid DNA Vectors with Enhanced Safety and Long-Term Gene Expression in Vivo Nelson S. Yew,* Hongmei Zhao, Malgorzata Przybylska, I-Huan Wu, Jennifer D. Tousignant, Ronald K. Scheule, and Seng H. Cheng Genzyme Corporation, 31 New York Avenue, Framingham, Massachusetts 01701-9322, USA *To whom correspondence and reprint requests should be addressed. Fax: (508) 872-4091. E-mail: [email protected].
Systemic delivery of cationic lipid–plasmid DNA (pDNA) complexes induces an acute inflammatory response with adverse hematologic changes and liver damage. Immunostimulatory CpG motifs in the pDNA are known to contribute substantially to this response. Here we constructed a pDNA vector (pGZB) that has been depleted of 80% of the CpG motifs present in the original vector. Compared with the unmodified vector, systemic administration of pGZB induced considerably fewer changes in blood parameters, reduced levels of inflammatory cytokines, and decreased liver damage. Despite the extensive sequence modifications within pGZB, there were still robust levels of transgene expression. Furthermore, in contrast to the transient expression observed from the unmodified vector, we observed sustained or increasing expression for up to 49 days from pGZB in the lung and liver of immunocompetent BALB/c mice. Studies adding CpG motifs in trans or in cis indicate that the reduced CpG content of pGZB was the major determinant for its persistent expression. This combination of decreased toxicity and sustained expression suggests that CpG-depleted pDNA vectors can greatly improve the safety and efficacy of synthetic gene delivery systems. Key Words: plasmids, CpG motifs, gene expression, inflammation, gene therapy
INTRODUCTION A potentially efficacious and relatively simple mode of gene therapy involves systemic delivery of complexes composed of cationic lipids or polymers and plasmid DNA [1]. Two problems associated with this approach are the toxicity of the complex and the lack of sustained transgene expression from the plasmid. Mice injected intravenously (i.v.) with cationic lipid–pDNA complexes show an acute inflammatory response, which includes complement activation and induction of the cytokines interleukin-12 (IL12), IL-6, interferon-␥ (IFN-␥,) and tumor necrosis factor␣ (TNF-␣) [2–5]. There are also marked hematologic and serologic changes, such as leukopenia, thrombocytopenia, and elevated levels of serum transaminases indicative of liver damage [5,6]. Though transient and dose-dependent, these harmful effects are unacceptably serious at the doses required to achieve therapeutic levels of expression. The second problem is that expression from current pDNA vectors is short-lived, peaking at day 1 after systemic delivery, then declining precipitously over the next 2 weeks to basal levels [7]. This is most often observed with vectors that contain the immediate-early gene promoter from cytomegalovirus (CMV), although other promoters are also
MOLECULAR THERAPY Vol. 5, No. 6, June 2002 Copyright © The American Society of Gene Therapy 1525-0016/02 $35.00
prone to inactivation over time [8]. The levels of transgene expression that can be attained with current synthetic vectors are generally quite low, and the lack of sustained expression further limits the ability to provide therapeutic levels of a given protein for any extended period of time. We hypothesized that the observed toxicity and transient expression may be partly due to the immunostimulatory properties of pDNA. Bacterial DNA and bacterially derived pDNA have the expected mathematical frequency of CpG dinucleotides, and the DNA is predominantly unmethylated, whereas in mammalian DNA the frequency of CpG dinucleotides is suppressed and largely methylated [9]. Recognition of these differences by the host leads to a pleiotropic inflammatory response that includes the activation of B cells, monocytes, macrophages, dendritic cells, and natural killer cells [10–14]. We and others have shown previously that immunostimulatory CpG motifs within the pDNA vector contribute substantially to the induction of proinflammatory cytokines by cationic lipid–pDNA complexes instilled into the lung [3,15,16]. The cytokine response induced by the complex can be partially suppressed with dexamethasone, by administering antibodies to the cytokines TNF-␣ and IFN-␥, or by inhibiting
731
ARTICLE
doi:10.1006/mthe.2002.0598, available online at http://www.idealibrary.com on IDEAL
FIG. 1. Diagram of CpG-depleted plasmid DNA vectors and transgenes. CMV, cytomegalovirus immediate-early gene enhancer/promoter; HI, hybrid intron; CAT, chloramphenicol acetyltransferase cDNA; pA, bovine growth hormone polyadenylation signal; ORI, replication origin region; KAN, kanamycin resistance gene; sHFIX, synthetic human factor IX cDNA. Each symbol represents five CpG motifs.
neutrophil influx using anti-Mac-1␣ and anti-LFA-1 antibodies [3,16,17]. These treatments were also shown to cause a marked increase in transgene expression levels [3,16,17]. Another strategy has been to decrease the overall number of CpG motifs within the pDNA, and we showed previously that reducing the CpG content of the vector reduced considerably the induction of several proinflammatory cytokines [18]. Here we have further reduced the CpG content of several pDNA vectors by replacing large portions of the vectors with synthetic, non-CpG sequence. We evaluated the acute toxicity induced by these vectors after systemic administration, measuring cytokine levels, cell numbers in the blood, and liver enzymes. We also determined the level and duration of expression in the lung and liver. Finally, we carried out studies to determine the mechanism for the greatly improved performance of these CpGdepleted vectors.
RESULTS Reduced Toxicity of CpG-Depleted pDNA Vectors after Systemic Delivery into BALB/c Mice We previously constructed a pDNA vector, pGZA-sCAT, that contained reduced numbers of CpG motifs (256 versus 526 CpGs in the unmodified vector, counting both strands of the pDNA [18]). This was achieved by minimizing the origin region required for replication of the plasmid, inserting a synthetic, non-CpG kanamycin resistance gene, and replacing the chloramphenicol acetyltransferase (CAT) reporter gene with a synthetic, non-CpG and codon-optimized counterpart (sCAT). To construct pGZB-sCAT, we eliminated an additional 154 CpGs by replacing the CMV immediate-early gene enhancer and promoter region with a synthetic, non-CpG version, and
732
by replacing the hybrid intron and polyadenylation signal with their non-CpG counterparts (Fig. 1). The only nonsynthetic portion remaining is the minimal region required for plasmid replication. pGZB-sCAT contains 102 CpGs, 19% of the CpG content of the original pCF1-CAT vector. To assess the acute toxicity of cationic lipid–pDNA complexes containing this CpG-depleted vector, we injected BALB/c mice with cationic lipid GL-62 complexed to pCF1-CAT, pGZA-sCAT, or pGZB-sCAT (GL-62:pDNA ratio, 0.5:0.5 mM; 16.5 g of pDNA). Blood and lungs were collected 24 hours after injection for analysis. As reported earlier [5], systemic administration of complexes containing pCF1-CAT resulted in a loss of leukocytes and platelets from the serum (Fig. 2). Similar losses were observed in mice that received complexes containing the partially CpG-reduced pGZA-sCAT. In contrast, the loss of leukocytes and platelets was less in mice that received complexes containing the most CpG-reduced vector, pGZBsCAT. Levels of the serum transaminases ALT and AST were highly elevated in mice that received pCF1-CAT or pGZA-sCAT complexes, indicative of liver damage and necrosis. In contrast, the levels of ALT and AST in mice that received pGZB-sCAT were only marginally elevated. In addition, the levels of the proinflammatory cytokines IL-12 and IFN-␥ were considerably lower in mice that received pGZB-sCAT compared with those that received pCF1-CAT, with cytokine levels closely correlating with the CpG content of the pDNA vector. Taken together, these results indicate that the CpG content of the pDNA affects several indicators of toxicity in addition to cytokines, and that the CpG-depleted pGZB-sCAT vector induced significantly less toxicity compared with both the unmodified pCF1-CAT vector and the partially CpGreduced pGZA-sCAT.
MOLECULAR THERAPY Vol. 5, No. 6, June 2002 Copyright © The American Society of Gene Therapy
doi:10.1006/mthe.2002.0598, available online at http://www.idealibrary.com on IDEAL
ARTICLE
FIG. 2. Comparison of the toxicity induced by unmodified and CpG-depleted pDNA vectors after systemic delivery. BALB/c mice were injected i.v. with 100 l of cationic lipid GL-62 complexed with pGZB-sCAT, pGZA-sCAT, or pCF1CAT (0.5:0.5 mM ratio of GL-62:pDNA; 16.5 g of pDNA). Blood was collected and lungs were harvested 24 hours after injection. Cell differentials (leukocytes, platelets) and liver enzymes (AST, ALT) were measured from the blood of five of eight mice per group. Cytokines (IL-12 and IFN-␥) were assayed from the remaining three mice per group. The levels of CAT in the lung were measured from all mice (n = 8 mice per plasmid). Water was used as the vehicle control. Data are expressed as mean ± SEM.
We also measured the levels of CAT in the lung 1 day after injection of complex. Higher levels of CAT were observed from both pGZA-sCAT and pGZB-sCAT compared with pCF1-CAT (Fig. 2). Lower levels of CAT were observed from pGZB-sCAT compared with pGZA-sCAT, suggesting that the synthetic CMV promoter was less active than the wild-type promoter. However, the levels were still substantially above that of the original pCF1-CAT vector, indicating that much of the original promoter activity was retained despite the extensive sequence alterations. Mice were also injected with complex at a dose of 1:1 mM (GL-62:pDNA ratio; 33 g of pDNA). At this higher dose there were no longer measurable differences between the unmodified and CpG-depleted vectors with regard to the loss of leukocytes and platelets, but liver enzyme levels and cytokines were reduced with the CpG-depleted vectors (Fig. 3). These results indicate that complexes containing the pGZA and pGZB vectors were still inflammatory, but that CpG reduction continued to be beneficial even at this higher dose. Persistent CAT Expression from pGZB-sCAT in the Lung Besides toxicity, another limiting issue with synthetic vectors is the lack of sustained transgene expression. We evaluated the persistence of expression from pGZB-sCAT in the lung after systemic delivery. Complexes of cationic lipid GL-62 and pCF1-CAT, pGZA-sCAT, or pGZB-sCAT were injected i.v. into BALB/c mice, and expression was measured for 35 days. Expression from pCF1-CAT declined rapidly within the first 2 weeks, which is characteristic for unmodified vectors that contain the CMV promoter (Fig. 4A) [8]. Expression from pGZA-sCAT was initially higher than that from pCF1-CAT, declined over the first 3 weeks, and then increased between day 21 and day 35. Expression from pGZB-sCAT also declined within the first week but then increased, with levels of CAT at day 35 higher than day 1 levels. These results indicate that the CpG-reduced vectors pGZA-sCAT and pGZB-sCAT displayed dramatically different profiles of expression over time compared with the unmodified pCF1-CAT vector, with the most CpG-reduced pGZB vector having the highest expression over time. Interestingly, we did not observe the generation of anti-CAT antibodies in any group (data not shown), so differences in these expression profiles are not a result of differential antibody response to the transgene.
MOLECULAR THERAPY Vol. 5, No. 6, June 2002 Copyright © The American Society of Gene Therapy
Cumulative Expression after Repeat Administration of pGZB-sCAT The profile observed after a single injection of pGZB-sCAT indicated that repeat administration might be particularly effective in generating a durable increase in CAT levels over that which could be achieved with a single dose. To test this possibility, complexes of cationic lipid GL-62 and
733
ARTICLE
doi:10.1006/mthe.2002.0598, available online at http://www.idealibrary.com on IDEAL
FIG. 3. Comparison of the toxicity induced by unmodified and CpG-depleted pDNA vectors at a higher dose of complex. BALB/c mice were injected i.v. with 100 l of cationic lipid GL-62 complexed with pGZB-sCAT, pGZA-sCAT, or pCF1-CAT (1.0:1.0 mM ratio of GL-62:pDNA; 33.0 g of pDNA). Blood was collected and lungs were harvested 24 hours after injection. Cell differentials (leukocytes, platelets) and liver enzymes (AST, ALT) were measured from the blood of five of eight mice per group. Cytokines (IL-12 and IFN-␥) were assayed from the remaining three mice per group. The levels of CAT in the lung were measured from all mice (n = 8 mice per plasmid). Water was used as the vehicle control. Data are expressed as mean ± SEM.
from a single dose. These data indicate that redosing with pGZB-sCAT was effective in providing cumulatively higher levels of CAT than could be achieved with a single administration of complex.
either pCF1-CAT or pGZB-sCAT were injected i.v. into BALB/c mice. As noted earlier, CAT expression from pGZBsCAT increased over the first 28 days, and then reached a plateau between day 28 and day 49 (Fig. 4B). The levels of CAT in this study were unusually high, possibly because of the variable transfection efficiency of different lots of cationic lipid. Some groups of mice were given a second dose of either GL-62:pCF1-CAT or GL-62:pGZB-sCAT 14 days after the first dose, and expression in the lung was followed for an additional 35 days. Expression from mice that received a second dose of pCF1-CAT increased transiently but then declined to undetectable levels. In contrast, expression from mice given a second dose of pGZBsCAT increased at day 1 after redosing and continued to increase through day 49 (Fig. 4B). The levels at day 49 were 270% higher than the maximum levels observed
734
Improved Persistence of CAT and Factor IX Expression from pGZB Vectors in the Liver A second major target organ for both viral and synthetic gene delivery vectors is the liver. To determine the level and persistence of expression from pGZB-sCAT in liver, pGZBsCAT was delivered using a hydrodynamics-based protocol that efficiently transduces hepatocytes [19,20]. pGZB-sCAT or pCF1-CAT pDNA was injected through the tail vein into BALB/c mice, and the levels of CAT in the liver were measured at 1, 14, and 28 days after injection (Fig. 5A). As observed in the lung, expression from pCF1-CAT declined sharply within the first 2 weeks. Expression from pGZB-sCAT also declined, but at a significantly slower rate compared to pCF1-CAT. These results indicate that pGZB-sCAT could also provide increased persistence of expression in liver. We also evaluated a pGZB vector that expresses a secreted protein, factor IX. The vector pGZB-sHFIX, which has a synthetic, non-CpG human factor IX cDNA, contains 96 CpGs, compared with the vector pCFA-HFIX, which has the wild-type factor IX sequence and contains 488 CpGs (Fig. 1). pGZB-sHFIX or pCFA-HFIX pDNA was injected using the hydrodynamics-based protocol into BALB/c mice, and the levels of factor IX in the plasma were measured at 14, 28, and 42 days after injection (Fig. 5B). The levels of factor IX from pCFA-HFIX declined to essentially background levels by day 14 and remained low. In contrast, the levels of factor IX from pGZB-sHFIX declined ~ 10-fold by day 14, but then remained steady at ~ 1 g/ml to day 42, the last time point. Hence, considerable improvements in the longevity of expression of both a secreted (HFIX) and nonsecreted transgene (CAT) were observed in liver using the pGZB vector. CpG Content as a Major Determinant of Sustained Expression The basis for the improvement in the longevity of expression was not obvious from the above studies. One possibility was that the sequence changes made within the CMV enhancer/promoter somehow rendered the promoter insensitive to inactivation over time. A second possibility was that the reduced CpG content was mainly responsible
MOLECULAR THERAPY Vol. 5, No. 6, June 2002 Copyright © The American Society of Gene Therapy
ARTICLE
doi:10.1006/mthe.2002.0598, available online at http://www.idealibrary.com on IDEAL
A
B
FIG. 4. Expression kinetics and redosing of pGZB-sCAT. (A) Persistence of CAT expression from pGZB-sCAT in the lung after i.v. delivery into BALB/c mice. BALB/c mice were injected i.v. with 100 l of cationic lipid GL-62 complexed with pGZB-sCAT, pGZA-sCAT, or pCF1-CAT (0.5:0.5 mM ratio of GL-62:pDNA; 16.5 g of pDNA). Lungs were harvested at different days after injection, and CAT assays were done (n = 4 mice per group). (B) Increased expression upon repeat i.v. delivery of pGZB-sCAT. BALB/c mice were injected i.v. with 100 l of cationic lipid GL-62 complexed with either pGZB-sCAT or pCF1-CAT (0.5:0.5 mM ratio of GL-62:pDNA). Some groups of mice were injected with a second dose of either GL-62:pGZBsCAT or GL-62:pCF1-CAT at day 14 after the initial injection (arrow). Lungs were harvested at different days after injection, and CAT assays were done (n = 4 mice per group). Data are expressed as mean ± SEM.
day 14, while expression from pGZAsCAT and psCFA-CAT recovered slightly with low but detectable levels of CAT at day 35. These results show that the synthetic CMV promoter in the context of a CpG-rich backbone is not sufficient to confer the substantial increase in expression over time seen with pGZB-sCAT. The data imply that the CpG content of the vector was the most important variable determining the expression profile.
DISCUSSION
for the sustained promoter activity. We conducted two studies in an attempt to distinguish between these possibilities. First, BALB/c mice were injected with cationic lipid–pDNA complexes containing either pGZB-sCAT alone, or pGZB-sCAT plus pOri-Kan, a plasmid that contains an unmodified kanamycin resistance gene and bacterial replication origin (Fig. 1). The pOri-Kan plasmid contains 288 CpGs, adding in trans a large number of CpGs to the dose given to the mice. The levels of CAT in the lung were measured for 35 days. Expression from pGZB-sCAT increased over time (Fig. 6A). In contrast, expression from mice that received both pGZB-sCAT plus pOri-Kan declined to undetectable levels by day 14, then slowly recovered at the later time points. These results show that the addition of a CpG-rich plasmid in trans had a strong negative effect on the persistence of expression. In a second study we constructed the plasmid psCFACAT, which is identical to pCF1-CAT with the exception that the wild-type CMV enhancer/promoter was replaced with the synthetic, non-CpG CMV enhancer/promoter from pGZB-sCAT (Fig. 1). psCFA-CAT contains 448 CpGs. BALB/c mice were injected i.v. with cationic lipid complexes containing pCF1-CAT, pGZA-sCAT, pGZB-sCAT, or psCFA-CAT, and the levels of CAT in the lung were measured for 35 days. Expression from pCF1-CAT, pGZA-sCAT, and psCFA-CAT all declined over the first 14 days (Fig. 6B). Expression from pCF1-CAT continued to decline after
MOLECULAR THERAPY Vol. 5, No. 6, June 2002 Copyright © The American Society of Gene Therapy
In this study we have shown that a pDNA vector that has been depleted in CpG content could overcome to a large extent two important limitations of current synthetic vectors, namely, the acute toxicity and transient expression. Compared with the unmodified vector, systemic delivery of cationic lipid complexes containing the CpG-depleted pGZB vector induced less inflammation, fewer hematologic changes, and decreased liver toxicity. The pGZB vector also displayed the striking property of sustained and even increasing expression over time. These improvements in safety and persistence of expression are vital steps toward making synthetic vectors truly effective gene delivery vehicles. Earlier studies have demonstrated the substantial role of CpG motifs in the acute inflammatory response to cationic lipid–pDNA complexes. Systemic administration of complex induces several proinflammatory cytokines, hematologic changes, and elevation of liver enzymes and acute-phase response proteins [3,5,6,21]. These toxicities are not observed after intravenous injection of either cationic lipid alone or pDNA alone, implicating cationic lipid complexed to pDNA as the causative agent [3,6,22]. Methylating the pDNA in the complex substantially decreases cytokine induction, indicating that the mammalian innate immune system is recognizing the unmethylated CpGs in the pDNA and is likely triggering the cascade of responses [23]. Therefore, one could predict that removal of CpGs from the pDNA would lessen the recognition of the pDNA as foreign, and consequently reduce the subsequent deleterious effects.
735
ARTICLE
doi:10.1006/mthe.2002.0598, available online at http://www.idealibrary.com on IDEAL
A
B
FIG. 5. Persistence of transgene expression from pGZB vectors in the liver. (A) BALB/c mice were injected rapidly through the tail vein with 5 g of pGZB-sCAT or pCF1-CAT in 2 ml of Mirus delivery solution. Livers were harvested at 1, 14, and 28 days after injection, and CAT assays were done (n = 4 mice per time point). (B) BALB/c mice were injected rapidly through the tail vein with 10 g of pGZB-sHFIX or pCFA-HFIX in 2 ml of Mirus delivery solution. Plasma was taken at 1, 14, 28, and 42 days after injection, and the levels of HFIX were measured (n = 5 mice per time point). Data are expressed as mean ± SEM.
The results of these studies agree with this hypothesis, but substantial numbers of CpGs remain even in our most CpG-depleted vector, and the remaining toxicity is not negligible. Preliminary experiments have been done with higher doses of complex than those reported here. Mice given a dose of 2:2 mM (GL-62:pDNA ratio; 66 g of pDNA) using the unmodified pDNA vectors exhibited considerable mortality, whereas mice injected at the same dose using the pGZB vectors typically survived (data not shown). However, there were outward signs of ill effects at this dose (for example, ruffled coat, inactivity), indicating that the range of tolerable doses, though expanded,
A
B
is still quite narrow with the CpG-reduced vectors. It is possible to create a nearly non-CpG vector by PCR amplification of only the expression cassette portion of the pDNA vector, that is, the promoter, intron, cDNA, and polyadenylation signal. Linear PCR fragments have been shown to elicit a reduced inflammatory response when delivered systemically as a lipid–protamine–DNA complex [24]. It remains to be determined whether the toxicity can be completely eliminated or that factors other than CpG motifs contribute to the observed host response. An additional and somewhat surprising consequence of CpG reduction was that the pGZB vector retained substantial transcriptional activity. The pGZB vector contains 31 nucleotide changes within the CMV enhancer/promoter region, several of which reside within transcription factor–binding sites, notably within the sites for Sp1, ATF/CREB, and NFB. However, factor recognition and binding require only a subset of base contacts [25]. The CMV enhancer region is also highly redundant, with multiple binding sites for each factor, and deletion of portions of the enhancer led to only a partial loss of activity [26]. The activity of the modified CMV enhancer/promoter was reduced relative to the wild-type sequence, but it may be possible to restore some of the activity by inserting additional binding sites or other enhancer sequences that contain fewer CpG motifs.
FIG. 6. Effect of CpGs in trans or in cis on the persistence of transgene expression. (A) BALB/c mice were injected i.v. with 100 l of cationic lipid GL-62 complexed with either pGZB-sCAT alone or a 1:1 (wt/wt) mixture of pGZB-sCAT + pOri-Kan (0.5:0.5 mM ratio of GL-62:pDNA; 16.5 g of pDNA). Lungs were harvested at 1, 7, 14, 21, and 35 days after injection, and CAT assays were done (n = 5 mice per time point). (B) BALB/c mice were injected i.v. with 100 l of cationic lipid GL-62 complexed with pCF1-CAT, pGZA-sCAT, pGZB-sCAT, or psCFA-CAT (0.5:0.5 mM ratio of GL-62:pDNA). Lungs were harvested at 1, 7, 14, 21, and 35 days after injection, and CAT assays were done (n = 4 mice per time point). Data are expressed as mean ± SEM.
736
MOLECULAR THERAPY Vol. 5, No. 6, June 2002 Copyright © The American Society of Gene Therapy
doi:10.1006/mthe.2002.0598, available online at http://www.idealibrary.com on IDEAL
The partial loss of transcriptional activity was more than compensated for by the sustained character of pGZB expression, the mechanism of which remains to be determined. Studies with the vector containing the synthetic CMV enhancer/promoter in the unmodified backbone (psCFACAT) indicate that the synthetic CMV promoter alone is not sufficient for the increased persistence. Rather, the overall reduced CpG content of pGZB appears to be a strong determinant. The decreased toxicity of pGZB could result in the survival of more plasmid-bearing cells, with consequently more expression over time. Alternatively, the decreased toxicity may permit the gradual formation of more active transcription factor complexes on the pGZB plasmids, increasing expression. Other promoters that appear to have greater longevity, such as the ubiquitin promoter or liver-specific promoters, may be more resistant to any possible inflammation-induced inactivating factors [27–29]. These data suggest that potentially therapeutic levels of transgene expression over an extended period may now be achievable with synthetic vectors. The decreased toxicity may allow the use of more concentrated lipid formulations, resulting in greater levels of expression. In addition, the ability to redose effectively may enable the gradual accumulation of expression into the therapeutic range. CpG depletion alone will not solve all the limiting issues of synthetic vectors, and targeting complexes away from cells involved in immune recognition will also likely be necessary [30]. However, CpG-depleted vectors do provide a platform for further improvements in the safety and utility of synthetic gene delivery.
MATERIALS
AND
METHODS
Plasmid vector construction. To construct pGZB-sCAT, a 1.1-kb DNA SphI fragment was synthesized by Entelechon (Regensburg, Germany); this fragment encompasses the enhancer/promoter from the CMV immediateearly gene (–522 to +16 relative to the transcription start site), hybrid intron, and bovine growth hormone polyadenylation signal. In this sequence all the CG dinucleotides were changed to TG. The plasmid pOriK-syn [18], which contains the minimal origin of replication and synthetic kanamycin resistance gene, was linearized with SphI, and the 1.1-kb synthetic fragment was inserted to create pGZB. The synthetic CAT cDNA was excised from pGZA-sCAT [18] by digestion with NotI, blunted with the Klenow fragment of DNA polymerase I, and ligated into the blunted EcoRI site of pGZB to create pGZB-sCAT. To construct pCFA-HFIX, DNA encoding human factor IX (gene symbol F9) was obtained by amplification of human liver cDNA (QuickClone cDNA, Clontech, Palo Alto, CA) using PCR with the Advantage polymerase (Clontech). The primers used for amplification were as follows: 5⬘-CCGGATCCTATGCAGCGCGTGAACATGATCATGGC-3⬘ and 5⬘-GGCTCGAGTCCATCTTTCATTAAGTGAGCTTTGTTT-3⬘ (bold letters refer to added BamHI and XhoI sites). The resulting PCR fragment was cloned into pcDNA3.0 (Invitrogen, Carlsbad, CA). The BamHI–XhoI fragment was excised from pcDNA3.0, blunted using the Klenow fragment of DNA polymerase I, and then ligated into the NotI site of pGZA to create pGZA-HFIX. An SphI fragment of pCFA containing the kanamycin resistance gene and replication origin was isolated and inserted in place of the corresponding SphI fragment from pGZA-HFIX to create pCFA-HFIX. To construct pGZB-sHFIX, 700-bp and 745-bp fragments were synthesized by Entelechon and ligated together to create a 1.4-kb fragment encoding human factor IX. The sequence was designed to incorporate codons that
MOLECULAR THERAPY Vol. 5, No. 6, June 2002 Copyright © The American Society of Gene Therapy
ARTICLE
are present in highly expressed human genes, to eliminate rare codons, and to remove all CG dinucleotides. EcoRI and SfiI sites were incorporated into the 5⬘ and 3⬘ ends of the 1.4-kb fragment, respectively. The 1.4-kb fragment was ligated into the EcoRI and SfiI sites of pGZB to create pGZB-sHFIX. To construct psCFA-CAT, pCFA-CAT [16] was digested with MfeI and XbaI and the backbone vector fragment was isolated. pGZB-sCAT was digested with MfeI and XbaI and the synthetic CMV promoter was isolated, and then ligated into pCFA-CAT to create psCFA-CAT. Cationic lipids and plasmid DNA. Cationic lipid GL-62 (N1-spermine cholesterylcarbamate) was formulated as described [27,31]. Plasmid DNA was purified by alkali lysis, ultrafiltration, and column chromatography [31]. The purified plasmid DNA contained < 10 g Escherichia coli chromosomal DNA per mg pDNA and < 10 endotoxin units per mg pDNA as determined by the chromogenic Limulus amebocyte lysate assay (BioWhittaker, Walkersville, MD). Administration of pDNA and cationic lipid–pDNA complexes into mice. Mice were injected through the tail vein with 100 l of cationic lipid GL62:pDNA as described [5]. Using the average nucleotide mass of 330 Daltons, the amount of pDNA delivered at a dose of 0.5:0.5 mM (GL62:pDNA ratio) is 16.5 g. For delivery to the liver, injections were done by the Mirus Corporation (Madison, WI) using the rapid, high-volume protocol described by Zhang et al. [19]. Briefly, 5–10 g of pDNA was mixed with 10 l of TransIT In Vivo Polymer Solution (Mirus Corporation) and then diluted into ~ 2 ml of Mirus Delivery Solution. The solution was injected rapidly into the tail vein over a period of 6–8 seconds. The pDNA has been shown to be taken up primarily by the liver hepatocytes using this procedure [20]. Cytokine and toxicity assays. Mice were anesthetized 24 hours after injection of the cationic lipid–pDNA complexes, and blood was collected by retro-orbital bleed using heparinized microcapillary tubes. Whole-blood and serum samples were analyzed by IDEXX Veterinary Services (West Sacramento, CA). Levels of the cytokines IL-12 and IFN-␥ were measured by ELISA kits using protocols supplied by the manufacturer (R&D Systems, Minneapolis, MN). Assays for transgene expression. CAT enzyme activity was measured from lung homogenates as described [31]. Levels of factor IX were measured using an ELISA kit (Asserachrom IX:Ag) following the protocol supplied by the manufacturer (Diagnostica Stango, Asnieres-Sur-Seine, France). Purified human factor IX protein (Sigma, St. Louis, MO) was used as a standard. Plasma samples were diluted 1:50 into phosphate buffer (1⫻ Reagent 4 from the kit) before conducting the assay.
ACKNOWLEDGMENTS We thank our colleagues at Genzyme Corporation: Richard Gregory for critical reading of the manuscript; Sirkka Kyöstiö-Moore for the factor IX cDNA; Brad Hodges for the factor IX assay; Seann O’Connell, Caroline DiCesare, and Nick Wan for purified plasmid DNA; and the Animal Care and Technical Services staff for assistance with the in vivo studies. RECEIVED FOR PUBLICATION DECEMBER 7, 2001; ACCEPTED MARCH 12, 2002.
REFERENCES 1. Nishikawa, M., and Huang, L. (2001). Nonviral vectors in the new millennium: delivery barriers in gene transfer. Hum. Gene Ther. 12: 861–870. 2. Plank, C., Mechtler, K., Szoka, F. C., Jr., and Wagner, E. (1996). Activation of the complement system by synthetic DNA complexes: a potential barrier for intravenous gene delivery. Hum. Gene Ther. 7: 1437–1446. 3. Li, S., et al. (1999). Effect of immune response on gene transfer to the lung via systemic administration of cationic lipidic vectors. Am. J. Physiol. 276: L796–804. 4. Dow, S. W., et al. (1999). Lipid–DNA complexes induce potent activation of innate immune responses and antitumor activity when administered intravenously. J. Immunol. 163: 1552–1561. 5. Tousignant, J. D., et al. (2000). Comprehensive analysis of the acute toxicities induced by systemic administration of cationic lipid:plasmid DNA complexes in mice. Hum. Gene Ther. 11: 2493–2513. 6. Loisel, S., Le Gall, C., Doucet, L., Ferec, C., and Floch, V. (2001). Contribution of plasmid DNA to hepatotoxicity after systemic administration of lipoplexes. Hum. Gene Ther. 12: 685–696.
737
ARTICLE
doi:10.1006/mthe.2002.0598, available online at http://www.idealibrary.com on IDEAL
7. Song, Y. K., Liu, F., Chu, S., and Liu, D. (1997). Characterization of cationic liposomemediated gene transfer in vivo by intravenous administration. Hum. Gene Ther. 8: 1585–1594. 8. Loser, P., Jennings, G. S., Strauss, M., and Sandig, V. (1998). Reactivation of the previously silenced cytomegalovirus major immediate-early promoter in the mouse liver: involvement of NFB. J. Virol. 72: 180–190. 9. Krieg, A. M. (1999). Direct immunologic activities of CpG DNA and implications for gene therapy. J. Gene Med. 1: 56–63. 10. Krieg, A. M., et al. (1995). CpG motifs in bacterial DNA trigger direct B-cell activation. Nature 374: 546–549. 11. Klinman, D. M., Yi, A. K., Beaucage, S. L., Conover, J., and Krieg, A. M. (1996). CpG motifs present in bacteria DNA rapidly induce lymphocytes to secrete interleukin 6, interleukin 12, and interferon-␥. Proc. Natl. Acad. Sci. USA 93: 2879–2883. 12. Ballas, Z. K., Rasmussen, W. L., and Krieg, A. M. (1996). Induction of NK activity in murine and human cells by CpG motifs in oligodeoxynucleotides and bacterial DNA. J. Immunol. 157: 1840–1845. 13. Sparwasser, T., et al. (1997). Macrophages sense pathogens via DNA motifs: induction of tumor necrosis factor-␣-mediated shock. Eur. J. Immunol. 27: 1671–1679. 14. Sparwasser, T., et al. (1998). Bacterial DNA and immunostimulatory CpG oligonucleotides trigger maturation and activation of murine dendritic cells. Eur. J. Immunol. 28: 2045–2054. 15. Freimark, B. D., et al. (1998). Cationic lipids enhance cytokine and cell influx levels in the lung following administration of plasmid: cationic lipid complexes. J. Immunol. 160: 4580–4586. 16. Yew, N. S., et al. (1999). Contribution of plasmid DNA to inflammation in the lung after administration of cationic lipid:pDNA complexes. Hum. Gene Ther. 10: 223–234. 17. Tan, Y., Li, S., Pitt, B. R., and Huang, L. (1999). The inhibitory role of CpG immunostimulatory motifs in cationic lipid vector-mediated transgene expression in vivo. Hum. Gene Ther. 10: 2153–2161. 18. Yew, N. S., et al. (2000). Reduced inflammatory response to plasmid DNA vectors by elimination and inhibition of immunostimulatory CpG motifs. Mol. Ther. 1: 255–262.
738
19. Zhang, G., Budker, V., and Wolff, J. A. (1999). High levels of foreign gene expression in hepatocytes after tail vein injections of naked plasmid DNA. Hum. Gene Ther. 10: 1735–1737. 20. Liu, F., Song, Y., and Liu, D. (1999). Hydrodynamics-based transfection in animals by systemic administration of plasmid DNA. Gene Ther. 6: 1258–1266. 21. Norman, J., et al. (2000). Liposome-mediated, nonviral gene transfer induces a systemic inflammatory response which can exacerbate pre-existing inflammation. Gene Ther. 7: 1425–1430. 22. Scheule, R. K., et al. (1997). Basis of pulmonary toxicity associated with cationic lipidmediated gene transfer to the mammalian lung. Hum. Gene Ther. 8: 689–707. 23. Hemmi, H., et al. (2000). A Toll-like receptor recognizes bacterial DNA. Nature 408: 740–745. 24. Hofman, C. R., Dileo, J. P., Li, Z., Li, S., and Huang, L. (2001). Efficient in vivo gene transfer by PCR-amplified fragment with reduced inflammatory activity. Gene Ther. 8: 71–74. 25. Pabo, C. O., and Sauer, R. T. (1992). Transcription factors: structural families and principles of DNA recognition. Annu. Rev. Biochem. 61: 1053–1095. 26. Boshart, M., et al. (1985). A very strong enhancer is located upstream of an immediate early gene of human cytomegalovirus. Cell 41: 521–530. 27. Yew, N. S., Przybylska, M., Ziegler, R. J., Liu, D., and Cheng, S. H. (2001). High and sustained transgene expression in vivo from plasmid vectors containing a hybrid ubiquitin promoter. Mol. Ther. 4: 75–82. 28. Herweijer, H., et al. (2001). Time course of gene expression after plasmid DNA gene transfer to the liver. J. Gene Med. 3: 280–291. 29. Qin, L., et al. (1997). Promoter attenuation in gene therapy: interferon-␥ and tumor necrosis factor-␣ inhibit transgene expression. Hum. Gene Ther. 8: 2019–2029. 30. Hwang, S. H., Hayashi, K., Takayama, K., and Maitani, Y. (2001). Liver-targeted gene transfer into a human hepatoblastoma cell line and in vivo by sterylglucoside-containing cationic liposomes. Gene Ther. 8: 1276–1280. 31. Lee, E. R., et al. (1996). Detailed analysis of structures and formulations of cationic lipids for efficient gene transfer to the lung. Hum. Gene Ther. 7: 1701–1717.
MOLECULAR THERAPY Vol. 5, No. 6, June 2002 Copyright © The American Society of Gene Therapy