Introduction

For many genetic diseases, a gene therapeutic approach giving prolonged and relatively high serum levels of the relevant protein would have significant therapeutic potential. Naked plasmid DNA (pDNA) has been shown to transduce hepatocytes when introduced into rodents using so-called "hydrodynamic" delivery conditions, i.e., relatively large volumes delivered at relatively high rates^1,2. Expression resulting from hydrodynamic delivery of naked pDNA has been shown to be relatively high, e.g., in the range of the viral vector systems, and to be largely a result of hepatocyte expression. We have recently demonstrated that such a hydrodynamic delivery of pDNA could be extended to a larger animal model, namely the New Zealand White rabbit, by using balloon catheters together with a delivery approach that uses the venous vasculature of the liver³. Thus, this vector and approach have the potential of a gene therapeutic that can be administered repeatedly without the complications of host immune responses to the vector.

Although hydrodynamic delivery of pDNA does not suffer from immune responses against the vector, we and others^4,5 have noted that this delivery method can result in the generation of antibodies against the protein produced. Since the treatment of patients with genetic disease will no doubt often involve treating "knockouts" for the particular gene in question, the generation of antibodies against the transgene product would represent a significant barrier to treatment by the hydrodynamic route.

We and others⁶ have also noted that the hydrodynamic approach, although well tolerated in rodents and rabbits, is associated with some acute toxicities. In particular, delivery-associated rapid (<24 h) and transient elevations in the serum transaminases indicate acute hepatocellular damage in both mice (data herein) and rabbits³. In the rabbit model, transaminase levels appear to correlate with expression. These hepatocellular effects may be related directly to the mechanism(s) by which the hydrodynamic delivery process results in hepatocyte transduction⁷.

Given the coincidence of apparent liver cellular cytotoxicity and initially high expression resulting from hydrodynamic delivery of pDNA, we hypothesized that the combination of these two factors might have contributed to the observed humoral responses. As a corollary, we reasoned that separating these two factors in time, e.g., by eliminating the initially high transgene expression, might lead to a reduction in subsequent immune responses, possibly improving long-term transgene expression. To effect a separation of expression and any delivery-associated inflammation, we have used small interfering RNA (siRNA)^8,9,10,11,12 to "switch off" temporarily the initial expression from the transgene. We have found as a consequence of this strategy not only that humoral responses against the transgene product are significantly attenuated, but also that the host immune system is better "tolerized" with respect to the expressed protein. The improvement in tolerization achieved by the siRNA approach was particularly profound in a knockout model, implying that such a strategy might have practical therapeutic consequences in genetic disease patients.

Results and discussion

Pilot experiments (see Supplementary material) demonstrated: (i) that hydrodynamic delivery of an -galactosidase A (gal) plasmid can elicit a humoral response against the gene product, (ii) that this response is more pronounced in the knockout (Fabry) model, (iii) that it may correlate with initial expression levels, and (iv) that the siRNA-mediated suppression of liver expression was specific and relatively long-lasting for both the nonsecreted CAT reporter and the gal.

-Galactosidase a expression recovers after siRNA-mediated attenuation

To explore the question of duration of the siRNA effect in more detail for gal, we coadministered a plasmid bearing gal (pCMV) and an gal-specific siRNA to BALB/c mice. Fig. 1A demonstrates that in the absence of the siRNA, serum gal levels declined from an initial (day 1) level of 10 g/ml to 1 g/ml over a 112-day period. When we codelivered this same amount of pCMV with 10 g of an gal-specific siRNA, initial (day 1) serum gal levels were depressed 100-fold. Over the next 2–3 weeks, serum gal levels increased 10-fold, reaching levels seen in the absence of siRNA at the 3-week time point. The degree of this transient attenuation of serum gal levels was dependent on the relative siRNA/pDNA ratio, i.e., lower relative ratios led to less profound decreases in serum gal, without changing the overall kinetic profile (data not shown).

Figure 1.

Coadministration of an gal-siRNA specifically and transiently suppresses initial gal expression from a CMV-driven vector and attenuates antibodies to gal in BALB/c mice. (A) A CMV-driven pDNA bearing human gal (pCMV; 10 g) was hydrodynamically injected either alone or together with 10 g of an siRNA against CAT or against gal. Serum gal expression was assayed over time by ELISA as described under Materials and Methods. Symbols indicate means SD; N = 14 animals/group. (B) A CMV-driven pDNA bearing pCMV-sSEAP (pSEAP; 10 g) was hydrodynamically injected either alone or together with 10 g of an siRNA against gal. Serum SEAP expression was assayed over time as described under Materials and Methods. Symbols indicate means SD; N = 5 animals per group. Serum antibodies against gal were determined at day 112 post-administration of a pDNA containing gal (pCMV) either alone or with an siRNA against CAT or gal. The number of animals with a given antibody titer are shown both (C) in tabular form and (D) as a bar graph in terms of percentages of the total number of animals in the group. Titers <200 are considered negative. The antibody distributions for the pCMV alone and pCMV + CAT-siRNA groups were not statistically different. However, the distributions of these two groups (in which 30–40% of the animals were negative) were statistically different (P < 0.0001) from that of the pCMV + gal-siRNA group (in which 72% of the animals were negative).

Full figure and legend (177K)

The specificity of this transient siRNA effect on gal expression was demonstrated in two ways. Fig. 1A shows that coadministering the same dose (10 g) of an irrelevant siRNA, namely CAT-siRNA (see Supplementary material), had essentially no effect on the serum levels of gal. Specificity was also demonstrated by coadministering the siRNA specific for gal with a noncognate pDNA, namely the secreted alkaline phosphatase (SEAP) reporter plasmid pGZB-sSEAP. Fig. 1B shows that coadministering gal-siRNA together with pGZB-sSEAP failed to alter significantly either the magnitude or the kinetic profile of serum SEAP expression over a 28-day period. Taken together, these results demonstrate that the effects of an siRNA on serum expression from its cognate pDNA (here, gal) are an immediate, specific, and dose-dependent attenuation of expression that recovers over time (in this case 3 weeks) to a level of expression that would have been seen from that plasmid at this time point in the absence of any siRNA.

It should be emphasized that the declining serum gal concentrations observed here (and throughout these studies) are likely due to a combination of factors in addition to antibody-mediated clearance, e.g., promoter shut-off. For example, the rapid initial loss of expression (few days) after hydrodynamic delivery is not likely due to antibodies. However, the declining serum gal levels are not likely to be the result of cytotoxic T lymphocyte responses, since we have observed prolonged liver expression of gal in the face of declining serum levels of the expressed transgene (see Supplementary material). Perhaps a reflection of the fact that multiple mechanisms are operating that affect serum gal levels, an examination of expression levels and antibody titers for individual animals revealed no direct correlation (data not shown). It should therefore be stressed that we are using the average strength of the humoral response only as a readout to test our hypothesis that the concurrence of initially high expression and inflammation is related to the strength of the subsequent immune response.

siRNA-mediated attenuation of Gal expression suppresses the antibody response to Gal

We have noted previously that hydrodynamic delivery elicits an inflammatory response that can be characterized by serum levels of IL-12 and the serum transaminases, e.g. alanine aminotransferase, ALT³. In the current study, 18 h after injecting pCMV into BALB/c mice, ALT and IL-12 levels had increased from baseline (see Supplementary material). Delivering an siRNA together with pDNA did not significantly alter these levels (see Supplementary material). These inflammatory parameters had returned to baseline levels by 2 weeks²⁶. Given our hypothesis, namely, that the coincidence of initially high expression from the liver together with liver inflammation contributed to the (humoral) immune response against the transgene product, we asked whether the observed siRNA-mediated attenuation of initial expression might also decrease the subsequent antibody response to gal. Figs. 1C and D demonstrate that at day 112 after hydrodynamic delivery of 10 g of pCMV the majority of the mice (22/36) had detectable antibodies against gal; antibody titers ranged up to 3200. Codelivery of 10 g of an irrelevant siRNA, namely CAT-siRNA, did not change this distribution of antibody responses significantly. In contrast, codelivery of the cognate siRNA, namely gal-siRNA, decreased the antibody response dramatically, and 70% (33/46) of the mice had undetectable (<200) anti-gal antibody titers. Taken together, the data shown in Fig. 1 suggest that by using siRNA to decrease the initial expression of gal, the resulting antibody response was significantly attenuated in the BALB/c mouse.

siRNA-mediated suppression of expression is independent of promoter

To ask whether siRNA could also affect the expression of gal driven by a less ubiquitously expressed promoter, we performed an analogous set of experiments using a hepatocyte-restricted promoter (HRP), namely, one based on the albumin promoter (pHRP; see Supplementary material). Fig. 2A demonstrates that in BALB/c mice, hydrodynamically delivered pHRP resulted in initial serum gal levels of 30 g/ml, similar to those produced by pCMV (Fig. 1A and Supplementary material). Thereafter, serum levels decreased by approximately 1 log/month over the next 2 months and then stabilized at 800 ng/ml, a kinetic profile not unlike that seen with pCMV (Fig. 1A). Coadministering an irrelevant siRNA (CAT-siRNA) had essentially no effect on this profile (Fig. 2A). In contrast, Fig. 2A shows that coadministering the cognate gal-siRNA decreased initial expression by 2 logs, an effect essentially identical to that seen with pCMV (Fig. 1A). Thus, the effects of a coadministered gal-siRNA on serum gal expression levels were both qualitatively and quantitatively identical with those of plasmids driven by either a ubiquitously expressed (pCMV) or a hepatocyte-restricted promoter (pHRP).

Figure 2.

Coadministration of an gal-siRNA specifically and transiently suppresses initial gal expression from a hepatocyte-restricted promoter-driven vector in BALB/c mice. (A) A hepatocyte-restricted promoter-driven pDNA bearing gal (pHRP; 10 g) was hydrodynamically injected either alone or together with 10 g of an siRNA against CAT or against gal. Serum gal expression was assayed over time by ELISA as described under Materials and Methods. Symbols indicate means SD; N = 5 animals/group. (B) Serum anti-gal antibody titers were determined at day 112 for the three groups shown in (A); the number of animals with a particular titer are presented as a percentage of the total number of animals as in Fig. 1D. Titers <200 are considered negative. Antibody distributions for the three groups are not statistically different.

Full figure and legend (88K)

Using the pCMV construct harboring the CMV promoter, we showed (above) that hydrodynamic delivery resulted in significant antibody responses against gal, which were attenuated when the cognate siRNA was coadministered (Figs. 1C and D). Fig. 2B shows that using the more tissue-specific pHRP construct, from which the expression was more restricted to hepatocytes, resulted in a dramatic reduction (compared to pCMV, see Figs. 1C and D) of the humoral response. Even in the absence of the cognate siRNA, no significant antibodies were generated against gal in the BALB/c mouse. This finding is consistent with the expectation that pHRP should restrict expression of gal mostly to hepatocytes¹⁷ and, unlike pCMV, should not result in gal expression in antigen-presenting cells such as the Kupffer cells of the liver.

siRNA suppresses initial Gal expression and anti-gal antibodies in fabry mice

It might be expected that the expression of human gal should generate a stronger anti-gal antibody response in a Fabry mouse (which is a knockout for gal²⁰) than in a BALB/c mouse, which expresses the mouse gal homolog. This expectation is consistent with data (see Supplementary material) showing a more rapid onset of anti-gal antibodies (by day 42) and much more rapidly declining serum human gal levels in the Fabry mouse. To ask whether siRNA-mediated reductions in the initial gal expression levels would also translate into a reduced antibody response and perhaps more persistent expression in the Fabry mouse, we codelivered pCMV and its cognate siRNA using hydrodynamic injection; we monitored serum gal and anti-gal antibody levels over 112 days. Fig. 3A shows that, as in the BALB/c model, coadministering gal-siRNA with pCMV reduced the initial (day 1) serum levels of gal by approximately 2 logs. An irrelevant siRNA, namely CAT-siRNA, had essentially no effect on serum gal levels. Over time, animals that had received both pCMV and gal-siRNA demonstrated higher (5-fold) serum levels of gal than those that had received pCMV alone or pCMV together with CAT-siRNA. Compared to the BALB/c, these long-term gal levels in the Fabry mouse were 20-fold lower.

Figure 3.

Coadministration of an gal-siRNA specifically and transiently suppresses initial gal expression from a CMV promoter-driven vector in Fabry mice. (A) A CMV-driven pDNA bearing gal (pCMV; 10 g) was hydrodynamically injected either alone or together with 10 g of an siRNA against CAT or against gal. Serum gal levels were assayed over time by ELISA as described under Materials and Methods. Symbols indicate means SD; N = 11–14 animals/group. (B) Serum anti-gal antibody titers were determined at day 112 for the three groups shown in (A); the number of animals with a particular titer are presented as a percentage of the total number of animals as in Fig. 1D. Titers <200 are considered negative. Antibody distributions for the three groups are not statistically different.

Full figure and legend (117K)

Compared with results in the BALB/c strain (Figs. 1C and D), Fig. 3B shows that CMV-driven expression of gal in the Fabry mouse led to a significantly higher distribution of anti-gal antibody titers. This result was not unexpected in this gal knockout, especially since expression was driven by a CMV promoter, with the possibility of expression in antigen-presenting cells. This robust anti-gal response is consistent with the 20-fold reduction in long-term expression noted above (Fig. 3A). Coadministration of an gal-siRNA with pCMV in this model had no significant effect on anti-gal antibody titers. These data thus imply that this model represents a higher hurdle (than the BALB/c) in terms of immune responses against the transgene product and that expression of gal from a CMV promoter leads to significant humoral responses that cannot be diminished by attenuating the initial expression.

We next asked whether substituting a plasmid bearing a hepatocyte-restricted promoter, pHRP, could lead to a reduced humoral response in the Fabry model. Fig. 4A depicts the serum gal expression profiles over time when expression was driven by the HRP. Initial expression levels were essentially identical (10 g/ml) to those seen with the CMV promoter in BALB/c or Fabry and the hepatocyte-restricted promoter in the BALB/c model. Thus, these two promoters generated equivalent initial serum levels in both models. Unlike the CMV promoter (Fig. 3A), Fig. 4A shows that the hepatocyte-restricted promoter resulted in significantly higher long-term serum levels (100 ng/ml) in the Fabry model; coadministration of CAT-siRNA had no effect on this expression profile. Coadministration of the gal-siRNA, however, resulted in long-term (3 months) serum levels that approached 1 g/ml, similar to those obtained in the BALB/c mouse with both promoters.

Figure 4.

Coadministration of an gal siRNA attenuates anti-gal antibodies and specifically and transiently suppresses initial gal expression from an HRP-driven vector in Fabry mice. (A) An HRP-driven pDNA bearing gal (pHRP; 10 g) was hydrodynamically injected either alone or together with 10 g of an siRNA against CAT or against gal. Serum gal levels were assayed over time by ELISA as described under Materials and Methods. Symbols indicate means SD; N = 15–27 animals/group. (B) Serum anti-gal antibody titers were determined at day 112 for the three groups shown in (A); the number of animals with a particular titer are presented as a percentage of the total number of animals as in Fig. 1D. Titers <200 are considered negative. The antibody distributions of the pHRP and pHRP + CAT-siRNA groups are not statistically different (P = 0.20), but each is statistically different from the pHRP + gal-siRNA group (P = 0.04 and 0.004, respectively).

Full figure and legend (104K)

Fig. 4B demonstrates that compared to a CMV-driven construct (Fig. 3B), the use of an expression plasmid driven by an HRP in the Fabry model led to a significantly attenuated antibody response, viz. 70% of the animals treated with pHRP had essentially no (titer <200) anti-gal titers, while only 14% of the animals treated with the CMV plasmid were in this category (Fig. 3B). Including an irrelevant (CAT) siRNA had essentially no effect on this antibody distribution. Importantly, Fig. 4B also shows that coadministering an gal-siRNA together with the HRP-driven plasmid resulted in a significant further attenuation of the anti-gal response, such that virtually all animals had undetectable anti-gal antibody titers. The absence of a significant humoral response under these conditions is consistent with the improved long-term serum levels in this group of animals (Fig. 4A). Thus, the maneuver of coadministering a cognate siRNA together with an expression plasmid driven by an HRP conferred long-term serum expression levels and an absence of anti-gal antibodies in the Fabry knockout model that were essentially equivalent to those produced (by either promoter) in the BALB/c model (Figs. 1 and 2).

Hydrodynamic delivery of plasmids expressing Gal are tolerizing, and coadministering an Gal-siRNA together with plasmids bearing a hepatocyte-restricted promoter can further enhance this state of tolerance

There is some evidence to suggest that expressing a secreted foreign protein from the liver leads to immune tolerance to that protein, i.e., it fails to provoke a humoral response^17,27. In the absence of gal-siRNA, we have shown (above) that both BALB/c and Fabry mice respond to the expression of gal from the liver by generating (to varying degrees) a humoral response to gal; by coadministering an gal-siRNA we showed that this response was significantly blunted. To ask whether this siRNA-mediated attenuation of the humoral response also led to a state of "tolerance," we challenged mice that had been expressing gal for a prolonged period (112 days) with purified gal protein in complete Freund's adjuvant (CFA). Animals that had developed tolerance would be expected to have little or no increase in antibody titer in response to this challenge, while animals that were not tolerized would be predicted to develop significant increases in anti-gal titers.

Figs. 5A and B demonstrate that 3 weeks after challenge with gal in CFA, naive BALB/c mice and naive Fabry mice responded with robust anti-gal humoral responses, with titers increasing 2–3 logs. Hydrodynamic delivery of pCMV to both mouse strains resulted in significant anti-gal titers at week 16 (day 112) as described above, with the knockout Fabry strain exhibiting the highest prechallenge titers, as expected. Interestingly, these titers were essentially unchanged after challenge in both mouse strains, indicating that significant tolerance had been developed to the transgene product, namely gal. These results were essentially unchanged in those animals that had received an siRNA (CAT-siRNA or gal-siRNA) together with the original pCMV.

Figure 5.

Hydrodynamic delivery of gal plasmids is tolerizing in mice. CMV- and HRP-driven gal plasmids (10 g of pCMV and pHRP, respectively) were hydrodynamically delivered either alone or together with 10 g of an siRNA against either CAT or gal in either (A) BALB/c or (B) Fabry mice. Serum anti-gal antibody titers were determined at 16 weeks postadministration (W16). The animals were then challenged with human gal protein in CFA (see Materials and Methods) and their serum anti-gal antibody titers redetermined after an additional 3 weeks (W19). Titers of individual animals are shown before (W16) and after (W19) the gal protein challenge. Naive groups of BALB/c and Fabry mice, i.e., not given pDNA or siRNA, were challenged with the gal protein in CFA and their titers are shown in the leftmost graph in the upper row. The number of animals/group is shown for each individual graph.

Full figure and legend (206K)

Given the apparent higher stringency of the Fabry model as deduced from its greater humoral responses to gal seen herein, we also asked whether a state of tolerance could be induced in this model by hydrodynamic delivery of the hepatocyte-restricted pHRP plasmid. Results with the pHRP construct were qualitatively similar to those obtained with the pCMV construct, but with some important differences. Fig. 5B shows that the distribution of titers at week 16 postadministration with the HRP was significantly lower than that of the analogous CMV group. Challenge with gal protein again indicated that a significant degree of immune tolerance to the transgene product had been developed. Thus, the determining factor in generating a state of tolerance appeared to be simply the hydrodynamic delivery and subsequent expression of gal from the liver.

Perhaps the most important immune-related findings are the tolerance results obtained in the Fabry model. Using pCMV, we have shown in the Fabry model (Fig. 3) that a significant anti-gal antibody response correlates with relatively low long-term serum levels of gal, but that this situation improves when an HRP is used to drive expression (Fig. 4). Further improvements in long-term expression and attenuation of humoral responses were seen by including an gal-siRNA together with the pHRP (Fig. 4). Fig. 5B shows that the decreased humoral response seen under these conditions also correlated with an improved state of tolerance to the transgene product. Thus, this combination of HRP together with cognate siRNA resulted in essentially no long-term antibody response (Figs. 4B and 5) and essentially no response to a protein challenge (Fig. 5B), i.e., a state of immune tolerance to gal was apparently achieved. We also note that this same state of tolerization could be achieved 6 weeks after plasmid administration (data not shown).

In general, decreasing initial serum levels of gal by coadministering its cognate siRNA was seen to reduce the antibody response (except for CMV/Fabry) and increase long-term gal serum levels (relative to pDNA alone). Importantly, for the Fabry mouse, the combination of HRP and cognate siRNA was shown essentially to eliminate the anti-gal antibody response, and long-term serum gal levels approached those reached in the lowest immune response condition, i.e., 1000 ng/ml, HRP/BALB/c (Fig. 2). These relatively high serum gal levels are at least a log higher than those attained with the CMV promoter in this model (Fig. 1) and are predicted to be therapeutic in the Fabry mouse.

Thus, strategies either using an HRP or separating delivery and expression in time using a cognate siRNA both served to attenuate anti-gal antibody responses. Taken together, these data are consistent with our working hypothesis, namely, that the humoral immune response against gal may be a consequence of initial, delivery-related liver inflammation together with high liver-derived expression of gal, some of which could occur in antigen-presenting cells such as Kupffer cells or liver sinusoidal endothelial cells. The humoral response could also result from gal expression in non-antigen-presenting cells (APCs), such as hepatocytes, that later release gal as a consequence of delivery-associated hepatocellular damage; this released neoantigen could then be taken up by APCs to initiate a humoral response. Based on these results, which demonstrate that an immune response can be induced even when an HRP is used, provided that local (liver) inflammation is also present, we speculate that even the relatively noninflammatory AAV vector bearing an HRP (which has been shown not to generate an antibody response when expressing gal¹⁷) could illicit an immune response if its expression kinetics sufficiently overlapped (in time) an inflammatory liver environment.

An important finding of these studies is that long-term, plasmid DNA-mediated expression of human gal from the murine liver appears to be tolerizing (Fig. 5). Tolerization is particularly efficient in the BALB/c mouse using a hepatocyte-restricted promoter and can be improved in this mouse model by coadministering the cognate siRNA. Perhaps most importantly, for gene therapy applications that seek to use the liver as a depot, tolerization could be significantly improved even in the knockout model using the siRNA coadministration strategy. As a therapeutic strategy, an siRNA coadministered with a plasmid vector could potentially tolerize a patient to subsequent administrations of a therapeutic protein, thereby making the protein therapeutic more efficacious. Finally, by extension, these data may imply that viral gene therapy approaches targeting the liver might benefit as well in terms of reduced immune responses to the transgene product by separating any delivery-associated inflammation from expression, e.g., using analogous siRNA or "gene-switch" strategies.

Materials and methods

Plasmid DNAs and siRNAs

Ubiquitous (CMV) and hepatocyte-restricted pDNAs

The two expression cassettes used in these studies are shown schematically in the Supplementary material. Briefly, the ubiquitously expressed cassette contains CpG-free synthetic elements including a CMV promoter, hybrid intron, bovine growth hormone polyadenylation signal, and kanamycin resistance gene, together with a CpG-reduced, minimal bacterial origin of replication. This CpG-reduced backbone has been shown to result in reduced toxicity and enhanced long-term gene expression^13,14,15. The synthetic human -galactosidase A (1.3 kb) and synthetic secreted alkaline phosphatase (1.5 kb) transgenes in this ubiquitously expressed cassette, namely pGZB-sHAGA and pGZB-sSEAP, respectively, have been described previously¹³. The sHAGA and sSEAP cDNAs are devoid of CpG dinucleotides. For simplicity, and to emphasize its relatively unrestricted expression, pGZB-sHAGA is referred to as "pCMV." The CMV-driven pCF1-CAT vector was constructed as described previously¹⁶. The hepatocyte-restricted DC190 expression cassette (see Supplementary material) contains a human serum albumin promoter (nucleotides -486 to +20) to which are appended two copies of the human prothrombin enhancer (nucleotides -940 to -860)¹⁷. This cassette was incorporated into the CpG-reduced backbone and the synthetic human -galactosidase A cDNA inserted to obtain pGZDC190-sHAGA as described¹⁸. For simplicity, and to emphasize its restricted expression in hepatocytes, this vector is abbreviated as "pHRP." For comparison, the mouse gal open reading frame codes for a 419-amino-acid protein which has 78% identity with the human analog¹⁹.

SiRNAs

The gal and CAT-siRNAs were synthesized by Xeragon (Huntsville, AL, USA). The 21-nucleotide (nt) gal siRNA with a 2-nt 3' overhang was designed to target nt 307–327 (nt 231–251 starting from the AUG) of the synthetic human -galactosidase A mRNA¹³. The sequence of gal-siRNA was 5'-GUCUGAAGGUUGGAAGGAUGC for the sense and 5'-AUCCUUCCAACCUUCAGACAC for the antisense strand. The 23-nt sequence of CAT-siRNA was that published previously¹⁰, viz., 5'-GGAGUGAAUACCACGACGAUUUC for the sense and 5'-AAUCGUCGUGGUAUUCACUCCGA for the antisense strand. Control siRNAs included: (i) CAT-siRNA with an inverted nuclear sequence for 21 nt at the 5' (CAT-inv-siRNA), having sequences 5'-UUAGCAGCACCAUAAGUGAGGCU for the sense and 5'-CCUCACUUAUGGUGCUGCUAAAG for the antisense strand, (ii) the single-stranded sense oligonucleotide from CAT-siRNA (CAT-S-oligo), and (iii) the single-stranded antisense oligonucleotide from CAT-siRNA (CAT-AS-oligo). Stock solutions of siRNAs and oligonucleotides were made at 20 M in RNase-free water and stored at -20°C.

Animal experiments

Six- to seven-week-old female BALB/c mice were purchased from Taconic (Germantown, NY, USA) and were acclimated for 1 week prior to being placed on study. Female Fabry mice (gal-/-²⁰) were placed on study at 10 to 11 weeks of age. Fabry mice are on an SV129 and C57BL/6 genetic background²⁰. Despite their slightly different age ranges, both strains were considered to have mature immune systems^21,22,23. All mice were housed at an AAALAC-accredited facility in microisolator cages and were tested routinely for viral (Sendai virus, pneumonia virus of mice, mouse hepatitis virus, Theiler's murine encephalomyelitis virus, reovirus, epizootic diarrhea of infant mice virus, mouse parvovirus) and bacterial (Mycoplasma pulmonis, Helicobacter, Pasteurella pneumotropica) pathogens as well as parasites (Aspiculuris tetraptera, Syphacia muris, Syphacia obvelata) using a sentinel-based strategy; all mice remained negative for these pathogens for the duration of the studies. All experiments were conducted under an approved IACUC protocol. All mice depicted in a given figure survived to the end of the study shown.

Plasmid DNA and siRNA were delivered by a hydrodynamic procedure described previously¹. Briefly, nucleic acid (DNA and/or RNA) in 2 ml of saline was injected into a tail vein over a period of 4–5 s. The doses of pDNA and siRNA are indicated in each figure. Using this procedure, pDNA has been shown to be delivered primarily to the hepatocytes².

Blood was collected from the orbital venous plexus of mice under anesthesia (2–3% isoflurane) using heparinized microhematocrit capillary tubes into serum-separating tubes. Serum was separated by centrifuging samples at 10,000g for 10 min and then stored at -80°C until assayed.

Immune tolerance to gal was evaluated in mice at 16 weeks post-administration of pDNA by intraperitoneally injecting 50 g of purified recombinant human gal (Genzyme Corp., Boston, MA, USA) in 100 l of PBS emulsified with 100 l of CFA (Sigma–Aldrich Corp., St. Louis, MO, USA)¹⁷. Serum was collected 3 weeks after this challenge and assayed for anti-gal antibody titers to assess tolerance to the gal protein.

Quantitative analyses

CAT enzymatic activity was determined in liver homogenates²⁴ and the levels of SEAP were quantified in mouse serum³ as described previously. Levels of gal in mouse serum were determined by an enzyme-linked immunosorbent assay (ELISA) using a polyclonal antibody that recognizes human, but not mouse, gal as described¹⁶. The sensitivity of this assay was 10 pg/ml¹⁸. Serum samples were also assayed for anti-gal antibody titers by ELISA using highly purified recombinant human gal to coat the 96-well plate as described previously²⁵. Goat anti-mouse immunoglobulin G (IgG), IgM, and IgA were used as the secondary antibodies in the assay²⁵. Titers are expressed as the reciprocal of the highest serum dilution giving an OD₄₉₀ 0.1. Because antibodies could be detected in both strains by day 84 after hydrodynamic delivery, and because titers had reached a plateau by day 112 (data not shown), this later time point was used to compare the immune responses between the two mouse strains. Although a formal possibility, cellular immune responses were not investigated in this study because our results with a similar gal construct demonstrated continued liver expression over several months while serum levels declined and anti-gal antibodies increased (see Supplementary material).

To ask whether the measured serum gal levels were affected by the presence of anti-gal antibodies, gal protein was spiked into serum containing known anti-gal antibody titers. At titers <12,000, serum gal quantitation was not affected. Since virtually all titers in the present study were less than 12,000, the serum gal levels as determined by ELISA should be relatively accurate. To ask the converse, that is, whether anti-gal antibody titers were affected by serum gal, serum with known antibody titers was spiked with known amounts of gal protein. Anti-gal titers were found to be unaffected at an gal concentration of 1 g/ml. Since all long-term serum gal levels are 1 g/ml, it is unlikely that long-term determinations of either antibody titers or gal protein levels are affected significantly.

Inflammation resulting from hydrodynamic delivery was evaluated by quantifying serum levels of the cytokine IL-12 (Mouse IL-12 ELISA kit; R&D Systems, Minneapolis, MN, USA) and the serum transaminase ALT (IDEXX Veterinary Services, Inc., West Sacramento, CA, USA) approximately 18 h after injection.

Statistics and error analysis

Results comparing two groups were analyzed using the two-tailed unpaired Student t test; those comparing multiple groups were analyzed by ANOVA. Statistical analyses of antibody distributions were performed by multinomial regression. Values shown represent means, and error bars represent standard deviations. Results were considered significantly different if P < 0.05. Additional details are given in the legends to individual figures.

References

1. 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. | Article | PubMed | ISI | ChemPort |
2. Liu, F., Song, Y. and Liu, D. (1999). Hydrodynamics-based transfection in animals by systemic administration of plasmid DNA. Gene Ther. 6: 1258–1266. | Article | PubMed | ISI | ChemPort |
3. Eastman, S. J., et al. (2002). Development of catheter-based procedures for transducing the isolated rabbit liver with plasmid DNA. Hum. Gene Ther. 13: 2065–2077. | Article | PubMed | ChemPort |
4. Ye, P., et al. (2004). Naked DNA transfer of factor VIII induced transgene-specific, species-independent immune response in hemophilia A mice. Mol. Ther. 10: 117–126. | Article | PubMed | ChemPort |
5. Dagnaes-Hansen, F., et al. (2002). Physiological effects of human growth hormone produced after hydrodynamic gene transfer of a plasmid vector containing the human ubiquitin promotor. J. Mol. Med. 80: 665–670. | Article | PubMed | ChemPort |
6. Kobayashi, N., Nishikawa, M., Hirata, K. and Takakura, Y. (2004). Hydrodynamics-based procedure involves transient hyperpermeability in the hepatic cellular membrane: implication of a nonspecific process in efficient intracellular gene delivery. J. Gene Med. 6: 584–592. | Article | PubMed | ChemPort |
7. Hodges, B. L. and Scheule, R. K. (2003). Hydrodynamic delivery of DNA. Expert Opin. Biol. Ther. 3: 911–918. | Article | PubMed | ChemPort |
8. Hamilton, A. J. and Baulcombe, D. C. (1999). A species of small antisense RNA in posttranscriptional gene silencing in plants. Science 286: 950–952. | Article | PubMed | ISI | ChemPort |
9. Elbashir, S. M., Harborth, J., Lendeckel, W., Yalcin, A., Weber, K. and Tuschl, T. (2001). Duplexes of 21-nucleotide RNAs mediate RNA interference in cultured mammalian cells. Nature 411: 494–498. | Article | PubMed | ISI | ChemPort |
10. Caplen, N. J., Parrish, S., Imani, F., Fire, A. and Morgan, R. A. (2001). Specific inhibition of gene expression by small double-stranded RNAs in invertebrate and vertebrate systems. Proc. Natl. Acad. Sci. USA 98: 9742–9747. | Article | PubMed | ChemPort |
11. McCaffrey, A. P., Meuse, L., Pham, T. T., Conklin, D. S., Hannon, G. J. and Kay, M. A. (2002). RNA interference in adult mice. Nature 418: 38–39. | Article | PubMed | ISI | ChemPort |
12. Lewis, D. L., Hagstrom, J. E., Loomis, A. G., Wolff, J. A. and Herweijer, H. (2002). Efficient delivery of siRNA for inhibition of gene expression in postnatal mice. Nat. Genet. 32: 107–108. | Article | PubMed | ISI | ChemPort |
13. Przybylska, M., et al. (2004). Partial correction of the alpha-galactosidase A deficiency and reduction of glycolipid storage in Fabry mice using synthetic vectors. J. Gene Med. 6: 85–92. | Article | PubMed | ChemPort |
14. Zhao, H., Hemmi, H., Akira, S., Cheng, S. H., Scheule, R. K. and Yew, N. S. (2004). Contribution of Toll-like receptor 9 signaling to the acute inflammatory response to nonviral vectors. Mol. Ther. 9: 241–248. | Article | PubMed | ChemPort |
15. Yew, N. S., et al. (2002). CpG-depleted plasmid DNA vectors with enhanced safety and long-term gene expression in vivo. Mol. Ther. 5: 731–738. | Article | PubMed | ChemPort |
16. Yew, N. S., et al. (1999). Increased duration of transgene expression in the lung with plasmid DNA vectors harboring adenovirus E4 open reading frame 3. Hum. Gene Ther. 10: 1833–1843. | Article | PubMed | ChemPort |
17. Ziegler, R. J., et al. (2004). AAV2 vector harboring a liver-restricted promoter facilitates sustained expression of therapeutic levels of alpha-galactosidase A and the induction of immune tolerance in Fabry mice. Mol. Ther. 9: 231–240. | Article | PubMed | ChemPort |
18. Hodges, B. L., Taylor, K. M., Joseph, M. F., Bourgeois, S. A. and Scheule, R. K. (2004). Long-term transgene expression from plasmid DNA gene therapy vectors is negatively affected by CpG dinucleotides. Mol. Ther. 10: 269–278. | Article | PubMed | ChemPort |
19. Gotlib, R. W., et al. (1996). The entire genomic sequence and cDNA expression of mouse alpha-galactosidase A. Biochem. Mol. Med. 57: 139–148. | Article | PubMed | ChemPort |
20. Ioannou, Y. A., Zeidner, K. M., Gordon, R. E. and Desnick, R. J. (2001). Fabry disease: preclinical studies demonstrate the effectiveness of alpha-galactosidase A replacement in enzyme-deficient mice. Am. J. Hum. Genet. 68: 14–25. | Article | PubMed | ISI | ChemPort |
21. Brazolot Millan, C. L., Weeratna, R., Krieg, A. M., Siegrist, C. A. and Davis, H. L. (1998). CpG DNA can induce strong Th1 humoral and cell-mediated immune responses against hepatitis B surface antigen in young mice. Proc. Natl. Acad. Sci. USA 95: 15553–15558. | Article | PubMed | ChemPort |
22. Anspach, J., Poulsen, G., Kaattari, I., Pollock, R. and Zwollo, P. (2001). Reduction in DNA binding activity of the transcription factor Pax-5a in B lymphocytes of aged mice. J. Immunol. 166: 2617–2626. | PubMed | ChemPort |
23. Johnson, K. M., Owen, K. and Witte, P. L. (2002). Aging and developmental transitions in the B cell lineage. Int. Immunol. 14: 1313–1323. | Article | PubMed | ISI | ChemPort |
24. 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. | PubMed | ISI | ChemPort |
25. Ziegler, R. J., et al. (1999). Correction of enzymatic and lysosomal storage defects in Fabry mice by adenovirus-mediated gene transfer. Hum. Gene Ther. 10: 1667–1682. | Article | PubMed | ISI | ChemPort |
26. 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. | Article | PubMed | ISI | ChemPort |
27. Mingozzi, F., et al. (2003). Induction of immune tolerance to coagulation factor IX antigen by in vivo hepatic gene transfer. J. Clin. Invest. 111: 1347–1356. | Article | PubMed | ISI | ChemPort |

Appendices

Appendix A

Supplementary data

Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ymthe.2005.04.007

Acknowledgements

We thank Bradley Hodges for constructing the pHRP vector and Fei Wang for statistical analysis of antibody distributions. We also gratefully acknowledge the Comparative Medicine and Vector Production Groups at Genzyme for their expert assistance.