Efficacy of Gene Therapy for a Prototypical Lysosomal Storage Disease (GSD-II) Is Critically Dependent on Vector Dose, Transgene Promoter, and the Tissues Targeted for Vector Transduction

ARTICLE

doi:10.1006/mthe.2002.0563, available online at http://www.idealibrary.com on IDEAL

Efficacy of Gene Therapy for a Prototypical Lysosomal Storage Disease (GSD-II) Is Critically Dependent on Vector Dose, Transgene Promoter, and the Tissues Targeted for Vector Transduction Enyu Ding,1 Huimin Hu,1 Bradley L. Hodges,1 Felicia Migone,1 Delila Serra,1 Fang Xu,1 Yuan-Tsong Chen,1,2 and Andrea Amalfitano1,2,* 1

Division of Medical Genetics, Department of Pediatrics, and 2Department of Genetics, Duke University Medical Center, Durham, North Carolina 27710, USA *To whom correspondence and reprint requests should be addressed. Fax: (919) 684-2362. E-mail: [email protected].

Lysosomal storage diseases are an intriguing target for gene therapy approaches, as transduction of a “depot” organ with a transgene encoding a lysosomal enzyme can be followed by secretion, systemic distribution, downstream uptake, and lysosomal targeting of the enzyme into non-transduced tissues. These benefits are of utmost importance when considering gene therapy approaches for glycogen storage disease type-II (GSD-II). GSD-II is a prototypical lysosomal storage disorder caused by lack of intralysosomal acid -glucosidase (GAA) activity. Lack of GAA can result in a proximal limb myopathy and respiratory and cardiac failure, each due to abnormal glycogen accumulation in the skeletal muscles or cardiac tissues, respectively. After converting the liver into a “depot” organ, we found that intravenous injection of the [E1-,polymerase-]AdGAA vector allowed for hepatic secretion of GAA over an at least 20-fold dosage range. We noted that very low plasma GAA levels (derived from hepatic secretion of GAA) can allow for GAA uptake by muscle tissues (skeletal or cardiac), but significantly higher plasma GAA levels are required before glycogen “cross-correction” can occur in these same tissues. We also demonstrated that liver-specific enhancer/promoters prolonged GAA transgene expression from persistent [E1-,polymerase-] adenovirus based vector genomes for at least 180 days, and significantly diminished the amounts of neutralizing anti-GAA antibodies elicited in this animal model. Finally, we demonstrated that skeletal muscles can also serve as a “depot” organ for GAA secretion, allowing for secretion of GAA and its uptake by noninfected distal tissues, although glycogen reductions in non-injected muscles were not achieved by the latter approach. Key words: Adenovirus, acid--glucosidase, Pompe, myopathy, dystrophy

INTRODUCTION Glycogen storage disorder type-II (GSD-II, or Pompe disease) is a lethal, autosomal recessive metabolic myopathy caused by a lack of acid -glucosidase (GAA) activity in the cardiac and skeletal muscles. Lack of sufficient GAA activity results in a massive accumulation of glycogen in intramuscular lysosomes, disturbing normal muscle cell functions [1]. The clinical spectrum of GSD-II can vary as to age of onset and organ involvement, with the relative severity generally correlating with the amounts of residual GAA activity present in the muscle tissues derived from affected patients [1]. At present, only supportive measures are available for patients with GSD-II, as there is no effective treatment.

436

Recently, however, our group and others have shown that enzyme replacement therapy for the infantile onset form of GSD-II (Pompe disease) can demonstrate promising clinical results [2,3]. Enzyme replacement strategies are based on the lysosomal targeting characteristics of exogenously presented lysosomal enzymes such as human GAA [4]. However, enzyme replacement therapies may be limited due to several factors, including large-scale production issues and/or the need for repeated, frequent, and life-long intravenous enzyme infusions. Gene therapy approaches offer the potential for long-term efficacy in these diseases, as well as a potential for greater efficacy than bolus infusions of lysosomal enzymes.

MOLECULAR THERAPY Vol. 5, No. 4, April 2002 Copyright © The American Society of Gene Therapy 1525-0016/02 $35.00

doi:10.1006/mthe.2002.0563, available online at http://www.idealibrary.com on IDEAL

ARTICLE

evaluated before any gene therapy approach can be demonstrated to show efficacy, not only in animal models of GSD-II, but also when considering future human clinical trials with GAA-encoding vectors.

RESULTS

FIG.1. Dose–response curve after intravenous injection of [E1,polymerase-] AdCMV-GAA vector in vivo. Gaa–/– mice were intravenously injected with a modified [E1,polymerase-] AdCMVGAA vector at the indicated particle doses per mouse. Each treatment group consisted of three age-matched mice. All plasma samples were obtained from the mice 3 days postinjection. Error bars in all figures reflect the standard deviations of the averaged values.

For example, we previously demonstrated that a single intravenous administration of an [E1-,polymerase-] adenovirus (Ad) based vector encoding GAA can allow for the potential correction of all affected muscles in a mouse model of GSDII [5]. This method of adenovirus vector delivery directed both high-level hepatic expression and secretion of GAA into the plasma space, which was followed by “cross-correction” of non-transduced muscle cells via receptor-mediated uptake of hepatically derived GAA, a result subsequently repeated by other groups [6]. Recently, we demonstrated that the improved [E1-,polymerase-] class of Ad vector was capable of persisting in the livers of immune-competent Gaa–/– mice for at least 6 months [7]. Furthermore, hepatically derived GAA persisted in muscle cells for at least 6 months. The latter result correlated with long-term (6 months) correction of pathologic intramuscular glycogen accumulations in the same muscles [7]. Here, we have set out to delineate the key variables that need to be overcome before the efficacy of most gene therapy approaches can be demonstrated in the current mouse model of GSD-II, the Gaa–/– mouse. For example, we have evaluated how the dosage of intravenously injected adenovirus vectors encoding GAA influences hepatic secretion and tissue targeting of GAA in the immune-competent Gaa–/– mouse model. We also used viral (cytomegalovirus, CMV) and nonviral enhancer/promoter elements, such as a modular liver-specific promoter (LSP), a murine-derived albumin enhancer/promoter (ALB), and the human elongation factor-1-a enhancer/promoters (EF) to drive GAA transcription, and compared their in vivo capabilities within mouse liver tissues. We also have investigated whether skeletal muscle can serve as a depot organ for GAA secretion in vivo, as previous reports have suggested that muscle tissues cannot allow for significant GAA secretion [8,9]. The results of our studies suggest that each of these variables must be critically considered and

MOLECULAR THERAPY Vol. 5, No. 4, April 2002 Copyright © The American Society of Gene Therapy

Detection of GAA in Plasma of Gaa–/– Mice Injected with Increasing Doses of the [E1-, polymerase] AdCMVGAA Vector Our previous studies demonstrated that the intravenous injection of high doses of Ad vectors encoding GAA can not only allow for hepatic expression and secretion of GAA into the blood, but also result in muscle uptake of GAA and cross correction of abnormal glycogen accumulation in multiple muscle tissues of Gaa–/– mice [5,7]. Because Ad vectors can allow for extremely high levels of transgene expression in liver tissues, we wished to evaluate the efficacy of these vectors at lower dosages. Gaa–/– liver tissues transduced with the [E1-,polymerase-] Ad CMV-GAA vector (GAA expression driven by a CMV enhancer/promoter) at varying doses had plasma samples collected at 3 days postinjection, a time when maximal levels of gene expression are obtained after hepatic transduction with this vector [5] (Fig. 1). The results demonstrated that [E1-,polymerase-]Ad CMV-GAA vector doses > 1.25  1010 particles/mouse resulted in significantly increased plasma GAA activities compared with the background GAA activities detected in mock-infected Gaa–/– mice. As the [E1-,polymerase-]Ad CMV-GAA vector dose increased above 1.25  1010 particles/mouse, the GAA plasma activity levels increased nearly linearly. We also carried out a GAA specific immunoblot analysis of the same plasma samples. As we have previously demonstrated, intravenous injection of the GAA encoding [E1-,polymerase-]Ad CMV-GAA vector resulted in detection of the ~ 110-kDa precursor protein isoform of GAA

FIG. 2. Immunodetection of GAA isoforms in plasma of Gaa–/– mice injected with either the [E1-polymerase-]Ad CMV-GAA or [E1-polymerase-]Ad LSP-GAA vector. Gaa–/– mice were injected with the indicated vectors at the following vector particles/mouse: A, 5  1010; B, 2.5  1010; C, 1.25  1010; D, 0. 625  109; E, 0.25  109. Each lane was loaded with 4 l of plasma derived from an individually injected mouse at 3 days postinjection. The results depict the patterns delineated in all injected mice assessed in a similar manner (data not shown). KO, mock-injected Gaa–/– mouse plasma; WT, mock-injected wild-type mouse plasma.

437

ARTICLE

doi:10.1006/mthe.2002.0563, available online at http://www.idealibrary.com on IDEAL

A

A

B

B

FIG. 3. Heart or quadriceps GAA activities in Gaa–/– mice treated with varying doses of the modified [E1-polymerase-]Ad CMV–GAA vector. GAA activity levels within heart (A) or quadriceps (B) tissues isolated 18 dpi from three vector-injected Gaa–/– mice were quantified and averaged. Heart and quadriceps protein samples derived from mock-injected Gaa–/– mice or wild-type mice were used as controls. The age of mice from all the treated groups is the same as that of the control group. Error bars reflect standard deviations of the sample groups.

in plasma. In contrast to the GAA activity analysis (Fig. 1), the GAA precursor isoform was detected in the plasma of all mice, at all [E1-,polymerase-]Ad CMV-GAA vector dosages (Fig. 2). We also noted that the ~ 76-kDa isoform of GAA was detected in the plasma at the highest vector dosages injected, which could either represent secretion of the 76-kDa protein directly from hepatocytes, or proteolytic cleavage of the 110kDa GAA isoform after secretion into the plasma space. The analysis of plasma GAA (by activity levels or by GAA immunoblot detection) illustrated that even though dosages of the [E1-polymerase-] AdCMV–GAA vector below 1.25  1010 did not demonstrate plasma GAA activity levels above the background levels noted in plasma derived from uninfected Gaa–/– mice (Fig. 1), the GAA-specific immunoblot assay was capable of detecting low level hepatic GAA secretion into the plasma space even at the much lower dosages. We have previously noted that mouse plasma contains high background GAA activity levels, likely due to the presence of neutral glucosidases present in mouse plasma. Our current results suggest that these high background levels can prevent the detection of very low activity levels of GAA protein present in mouse blood derived from any hepatically targeted vector; however, these levels can be detected by GAA specific immunoblot analysis.

438

FIG. 4. Immunoblot analysis of GAA isoforms in heart and quadriceps muscles from mice treated with the different doses of the modified [E1-polymerase]AdCMVGAA vector. Heart (A) and quadriceps (B) muscles were harvested at about 18 dpi. Each lane represents 100 g of protein extracted from the respective tissues of individual mice treated with the indicated dose of the modified [E1-polymerase-]Ad CMV–GAA vector. The patterns presented are representative of the pattern noted in all mice injected with the indicated vector dosages analyzed in a similar manner (data not shown). Gaa–/– mice were injected with the indicated vectors at the following vector particles/mouse: A, 5  1010; B, 2.5  1010; C, 1.25  1010; D, 6.25  109; E, 2.5  109. The tissue extracts from untreated Gaa–/– mouse and wild-type mouse were used as controls, and represent the typical patterns observed in this assay when control animals are examined. Recombinant CHO cell derived rGAA (10 ng) was used as a molecular weight standard. Note that the quadriceps image was exposed longer to the autoradiographic film than the image of the heart samples; longer exposures of the heart immunoblot similarly displayed presence of intralysosomal isoforms of GAA (~ 76 and 67 kDa) at the lowest vector dosages injected (data not shown).

We have previously demonstrated that high-level, hepatic secretion of GAA into the plasma space of Gaa–/– mice can allow for enough uptake of GAA by non-transduced muscle tissues to allow for “cross-correction” of the abnormal glycogen contents in these same tissues [5,7]. We therefore evaluated how decreasing amounts of hepatic secretion of GAA (after infection with decreasing amounts of the GAA encoding Ad vector) influenced either GAA uptake or glycogen levels in non-transduced muscle tissues by 18 days postinjection (dpi), a time when maximal “cross-correction” can be demonstrated [5,7]. The results indicated that as the vector dosage increased (and as plasma GAA activity levels increased), the GAA activity levels measured in heart tissues increased proportionately (Fig. 3). Immunoblot detection of GAA protein isoforms within heart muscle further confirmed the trends (Fig. 4). The immunoblot assay also indicated that the tissue GAA activity results were not sensitive enough to detect the

MOLECULAR THERAPY Vol. 5, No. 4, April 2002 Copyright © The American Society of Gene Therapy

doi:10.1006/mthe.2002.0563, available online at http://www.idealibrary.com on IDEAL

very low levels of GAA that had been taken up and intralysosomally targeted within heart muscle tissues at the lowest vector dosages given (Fig. 4). Finally, quantitative measurement of the heart glycogen contents were also analyzed in relation to the vector dosages administered (Fig. 5). Although by GAA-specific immunoblot assay GAA protein was detectable in heart muscle at all vector dosages tested, the glycogen contents of heart muscles in Gaa–/– mice only begin to significantly decrease at doses of the [E1-,polymerase-]Ad CMV-GAA vector that were at least 1.25  1010 particles (compared with untreated, agematched Gaa–/– mice). Substantial amounts of glycogen reduction can only be achieved in heart tissues of Gaa–/– mice (18 days after GAA encoding vector injection) when plasma GAA activity levels approached ~ 11,800 nmol/hour/ml within 3 days of vector injection, and heart tissue GAA activities exceeded ~ 5.5 nmol/hour/mg protein at ~ 18 dpi. Similarly, we analyzed Gaa–/– quadriceps muscle responsiveness to decreasing [E1-,polymerase-]Ad CMVGAA vector dosages. GAA activities in quadriceps muscles also paralleled the decreasing plasma GAA levels (Fig. 3), although the more sensitive immunoblot assay again demonstrated that intralysosomal GAA protein isoforms are detected in this muscle at all vector dosages (Fig. 4). Finally, quantitative measurement of the quadriceps muscle glycogen contents was also analyzed in relation to the [E1-,polymerase-]Ad CMV-GAA vector dosages administered (Fig. 5). Doses of [E1-,polymerase-]Ad CMV-GAA exceeding 1.25  1010 particles allowed for the glycogen contents of the quadriceps muscles of Gaa–/– mice to begin to decrease, compared with the levels noted in the quadriceps muscles of untreated, age-matched Gaa–/– mice. Despite the higher plasma GAA activity levels achieved at these higher vector dosages (~ 31,500 nmol/hour/ml), the quadriceps GAA activity levels were only 3.4 nmol/hour/mg. The results suggested that quadriceps muscles seem to be less responsive to similar plasma GAA activity levels (as assessed by overall glycogen reduction levels) compared with heart muscle. This phenomenon is likely secondary to a decreased efficiency of GAA uptake by skeletal muscle. Use of Nonviral Enhancer/Promoter Elements to Drive GAA Transcription from [E1-,polymerase-]Ad Vectors As lower levels of GAA secretion from liver tissues demonstrated significant evidence of efficacy in heart tissues, we next evaluated whether alternative (nonviral) enhancer/promoter elements might also be capable of driving GAA expression to the efficacious levels noted with use of the CMV-derived enhancer element. Use of these nonviral elements might offer several potential advantages relative to the use of the CMV-based enhancer element. These advantages include tissue-specific limiting of the expression of GAA (relative to the ubiquitously expressed

MOLECULAR THERAPY Vol. 5, No. 4, April 2002 Copyright © The American Society of Gene Therapy

ARTICLE

A

B

FIG. 5. Glycogen content of Gaa–/– mice treated with varying doses of the [E1-, polymerase-]AdCMV-GAA vector. Muscle tissues (heart (A) and quadriceps (B)) from three vector-treated Gaa–/– mice were analyzed for glycogen content and compared with the glycogen levels measured in the same muscles of age-matched, mock-injected Gaa–/– mice. Cohorts of [E1-,polymerase-]AdCMVGAA treated mice that had significantly decreased amounts of glycogen as compared with the mock-injected Gaa–/– mice are indicated as follows: *P < 0.05; #P < 0.01. P values were determined by Student’s t-test. All error bars represent the standard deviations within each data set.

CMV enhancer/promoter element) as well as prolonged transcriptional activity, as these tissue-specific enhancer/promoter elements are not subject to the downregulation events associated with use of CMV-based enhancer elements in liver tissues [10,11]. We therefore constructed a series of [E1-,polymerase-]Ad vectors encoding identical GAA gene sequences, the expression of which was placed under the control of the LSP (modular liver-specific enhancer/promoter [12]), ALB (murine albumin enhancer/promoter [13]), or EF (elongation factor 1- enhancer/promoter [11]) derived elements. These elements allow for both high-level and sustained expression of transgenes in liver tissues, in the context of Ad or adenoassociated virus (AAV) based vectoring systems [12,13]. We generated high-titer preparations of each of the vectors and analyzed their ability to mediate hepatic secretion of GAA after intravenous administration. At equivalent intravenous Ad vector particle number injections, none of the alternative enhancer elements allowed for plasma GAA activity levels to exceed the background GAA activity levels measured in plasma of uninfected Gaa–/– mice, at all time points analyzed (data not shown). Unfortunately, repeated attempts with the ALB or EF enhancer driven GAA encoding [E1-polymerase-]Ad vec-

439

ARTICLE

doi:10.1006/mthe.2002.0563, available online at http://www.idealibrary.com on IDEAL

tors failed to demonstrate adequate secretion of GAA into mouse plasma at maximal dosages, as assessed by both GAA activity levels and GAA-specific immunoblot assays. In all instances, these two versions of [E1-,polymerase-] GAA vectors also did not allow for evidence of glycogen cross-correction in heart or quadriceps tissues derived from these animals, a result predicted by our studies of decreasing dosages of the [E1-polymerase-]CMVGAA vector. Based on the lack of efficacy, the EF GAA-encoding vector constructs were not evaluated further. Use of the more sensitive GAA-specific immunoblot assay demonstrated that the vector using the LSP enhancer to drive GAA transcription allowed for detectable (albeit lower) levels of GAA hepatic secretion into the plasma (Fig. 2). Although GAA protein is detected in the plasma over a range of vector dosages after intravenous injection of Gaa–/– mice with the [E1-,polymerase-]LSP-GAA encoding vector (as was shown for the [E1-,polymerase-] CMV/GAA vector), the levels never approach the maximal levels achieved with similar particle dosages of the [E1-,polymerase-] CMV/GAA vector (Fig. 2). We demonstrated some evidence of low-level uptake of GAA in the hearts and quadriceps muscles of Gaa–/– mice infected with the [E1-,pol-]Ad LSP–GAA vector by 18 dpi (by GAA-specific immunoblot; data not shown), but these very low levels were insufficient to allow for significant reduction of the glycogen content of these same tissues, at all vector dosages administered (data not shown) This confirmed our observations noted with lower dosages of the [E1-,pol-]Ad CMV–GAA vector, namely that if plasma GAA activity levels do not exceed 11,800 ng/hour/ml by 3 dpi, no evidence of glycogen “cross-correction” will be evidenced in heart or skeletal muscle tissues of intravenously injected Gaa–/– mice by ~ 18 dpi. Lack of Persistent Detection of GAA in the Plasma of Gaa–/– Mice: Role of Anti-GAA Antibodies versus GAA Gene Expression from Ad-Based Vectors We have previously demonstrated that use of the [E1-,polymerase-]Ad CMV–GAA vector resulted in highlevel, transient detection of GAA in the plasma of intravenously injected mice, with complete lack of GAA detection being noted in the plasma at time points > 12 dpi, even at the highest doses of Ad vector injected [5,7]. The lack of plasma GAA detection was not due to loss of vector genomes, as the modified vector was detected for at least 190 days in the liver of the injected animals [7]. This lack of plasma GAA detection positively correlated with the rapid onset of anti-GAA antibodies in the same animals. One caveat to those studies, however, was that the CMV enhancer element was also subject to a very rapid downregulation of transcription within days of vector injection. Because of this rapid shutdown, one could argue that the loss of detection of GAA protein in the plasma of [E1-,polymerase-]Ad CMV-GAA injected mice may have also been due to lack of adequate GAA expression/secre-

440

A

B

FIG. 6. Quantitative assessment of human GAA DNA and RNA sequences present in liver tissues of mice injected with the [E1-,polymerase-]Ad CMV-GAA vector, the [E1-,polymerase-]Ad LSP-GAA vector, or the [E1-,polymerase-]Ad mALB-GAA vector. PCR and RT-PCR detection of GAA or G3PDH sequences were carried out on liver-derived DNA (A) and RNA (B) samples harvested from 2–3 of the respective mice at the indicated time points, averaged, and compared. All GAA DNA or RNA quantitations were normalized to mouse G3PDH DNA or RNA sequences quantitated in the same samples. *Note that all liver samples derived from [E1-,polymerase-]Ad LSP-GAA or [E1-,polymerase-]Adm ALB-GAA vector treated mice demonstrated GAA RNA transcription at all time points, this was not true for [E1-,polymerase-]Ad CMV-GAA treated mice, as only one of two mice at 56 dpi, and only one of three mice at 190 dpi, demonstrated detectable GAA gene transcription. The data values as presented, however, depict the average value for all mice at these time points (the two cohorts are indicated with an asterisk).

tion from the liver at later time points, rather than solely attributable to the onset of rising anti-GAA antibody titers. We hypothesized that use of the [E1-polymerase-] LSPGAA vector would clarify the impact anti-GAA antibody responses may be having in plasma clearance of GAA in Gaa–/– mice, as the LSP enhancer should not be subject to the rapid downregulation of the CMV enhancer previously noted in liver tissues. We therefore compared the vector genome persistence levels, GAA mRNA expression levels, and plasma GAA secretion patterns at several time points after injection of similar doses of the [E1-,polymerase-] CMV-GAA vector or the [E1-,polymerase-] LSP-GAA vector, to clarify the impact anti-GAA antibody responses have in this model system of GSD-II. We confirmed that vectorspecific DNA sequences and GAA-specific RNA transcripts are detectable for extended time periods in the liver tissues

MOLECULAR THERAPY Vol. 5, No. 4, April 2002 Copyright © The American Society of Gene Therapy

doi:10.1006/mthe.2002.0563, available online at http://www.idealibrary.com on IDEAL

ARTICLE

FIG. 7. Immunoblot detection of GAA in plasma from [E1-,polymerase-]Ad LSP-GAA injected Gaa–/– mice. [E1-,polymerase-]Ad LSP-GAA injected Gaa–/– mice had 4 l of plasma retrieved from the mice at the indicated time points postinjection. The samples were electrophoretically separated, transferred to nylon membrane, probed with an anti-GAA specific antibody, and GAA protein was visualized. Results from three individual mice (A, B, or C), each injected with equivalent amounts of the [E1-,polymerase-]Ad LSP-GAA vector (particles = 5  1010), are presented. The patterns presented are representative of the patterns observed from several experiments with this vector (data not shown). No GAA was detected in plasma samples at later time points as well (data not shown). We used 10 g of rGAA protein as the rGAA positive control.

of Gaa–/– mice injected with either of the Ad vectors, using quantitative PCR and RT-PCR based methods (Fig. 6). The results demonstrate that a [E1-,polymerase-] CMV-GAA genome generated at least 100-fold more GAA mRNA expression than what was transcribed from the genome of a [E1-,polymerase-] LSPGAA vector at 3 dpi, explaining why similar doses of the latter failed to allow for high-level GAA secretion from hepatocytes (Fig. 2). However, the lower levels of GAA transcription generated from the [E1,polymerase-] LSP–GAA vector were sustained at roughly equivalent levels during the first few months of the experiment, and were not subject to the 100-fold drop of GAA expression noted within days of injection of the [E1-,polymerase-] CMV–GAA vector. Despite the steady-state GAA expression levels derived from the persistent [E1-,polymerase-]Ad LSP–GAA vector, immunoblot analysis of plasma samples derived from [E1-,polymerase-] LSP–GAA vector injected Gaa–/– mice again demonstrated loss of

detection of GAA in plasma at 17 dpi, relative to the levels detected at 3 dpi (Fig. 7). Gaa–/– mice injected with the [E1-,polymerase-]Ad LSPGAA vector had their anti-GAA antibody levels quantitated at several time points post-injection (Fig. 8). This vector elicited significantly elevated anti-GAA antibody responses by 17 dpi when compared with noninfected control mice, titers which increased further at later time points (Fig. 8A). As the RT-PCR data confirmed that the [E1-,polymerase-] LSP-GAA vector was capable of stable levels of GAA mRNA transcription throughout this time period (Fig. 6), the anti-GAA antibody data suggested that loss of detection of plasma GAA mice 17 days after transduction (and beyond) in Gaa–/– mice injected with the [E1,polymerase-]Ad LSP-GAA vector is due to the elicitation of anti-GAA antibodies. The use of the tissue-specific LSP enhancer element significantly diminished the relative amounts of the anti-GAA antibodies detected, compared

A

B

FIG. 8. Anti-GAA antibody levels in [E1-,polymerase-]Ad LSP-GAA treated mice. (A) Intravenous injection of Gaa–/– mice with 5.0  1010 vector particles of the [E1-,polymerase-]Ad LSP-GAA yielded significantly elevated anti-GAA antibody levels, compared with mock-injected Gaa–/– mice (n = 11 for [E1-,polymerase-]Ad LSPGAA treated Gaa–/– mice; n = 7 for mock infected Gaa–/– mice). (B) The impact that use of alternative enhancers to drive GAA expression has on anti-GAA antibody responses. Gaa–/– mice (n = 3 for each dosage) were intravenously injected with the respective vectors at the indicated dosages, and the amounts of anti-GAA antibody elicited were compared with each other and mock-injected Gaa–/– mice at 17-19 dpi (n = 11). At all dosages tested, the [E1-,polymerase-]Ad CMV-GAA vector elicited greater amounts of anti-GAA antibodies than that elicited after similar particle number injections with the [E1-,polymerase-]Ad LSP-GAA vector. This result was confirmed even when comparing cohorts that had significantly different vector particle numbers injected (that is, compare antibody titers of Gaa–/– mice injected with 6.25  109 vector particles of [E1-,polymerase-]Ad CMV-GAA and Gaa–/– mice injected with 5.0  1010 vector particles of [E1-,polymerase-]Ad LSP-GAA). Note that these latter particle numbers had yielded nearly identical amounts of GAA hepatic secretion, based on GAA-specific immunoblot assay (Fig. 2). Finally, similar results were obtained when later time points were similarly assessed (data not shown).

MOLECULAR THERAPY Vol. 5, No. 4, April 2002 Copyright © The American Society of Gene Therapy

441

ARTICLE

doi:10.1006/mthe.2002.0563, available online at http://www.idealibrary.com on IDEAL

FIG. 9. Immunoblot detection of GAA in plasma of SCID mice intramuscularly injected with [E1-,polymerase-]Ad CMV-GAA. SCID mice were intramuscularly injected with 3  109 particles of [E1-,polymerase-]Ad CMV-GAA. Plasma (4 l) from two mice (A and B) was serially collected at the indicated dpi, electrophoretically separated, transferred to nylon membrane, and probed with an anti-GAA polyclonal antibody. Note that extended autoradiographic exposure times were required to detect the low GAA plasma levels (secreted from muscle) with this method. As a result, the GAA polyclonal antibody is noted to cross-react with an as yet unidentified ~ 110 kDa mouse plasma protein, which is visualized in the mock-injected SCID plasma or mock-injected wildtype mouse derived plasma, a result we have noted previously [5]. Despite the nonspecific binding, one can discern the vector-derived GAA precursor protein isoforms present in the plasma, a conclusion further substantiated by the presence of the 76 kDa mature form of GAA at days 10–84 in these samples. rGAA denotes 10 ng of rGAA derived from CHO cells.

with studies in mice injected with an [E1-,polymerase-]Ad CMV-GAA vector hepatically secreting the same amounts of GAA, a pattern that was also repeatable over all dosage ranges of either vector used (Fig. 8B). The use of the tissue-specific LSP enhancer in the [E1-,polymerase-]Ad vector backbone did not completely prevent the elicitation of significant amounts of anti-GAA specific antibodies, although the absolute amounts elicited were always less than those produced after similar injections with an identical vector using the CMV enhancer to drive GAA expression. This result is consistent with some reports suggesting that use of tissue-specific promoters in Ad based vectors may allow for avoidance of anti-transgene antibody responses [14]. Human GAA Secreted from [E1-,polymerase-]Ad CMV-GAA Injected Skeletal Muscle Our evaluation of escalating intravenous dosing regimens, as well as the use of alternative enhancer/promoter elements to modulate GAA expression from [E1-,polymerase-]Adbased vectors, was based on using liver tissue as a depot organ for GAA secretion. We next determined whether alternative, more accessible tissues might also serve as a potential “depot” organ for GAA secretion. Skeletal muscle allows for high-level secretion of several proteins after transduction with a number of vectors, including Ad-based vectors [15,16]. It has also been demonstrated that skeletal muscle can be transduced by [E1-]Ad vectors encoding GAA, however, only localized GAA expression was

442

achieved (that is, secretion of GAA and uptake of GAA by other noninfected tissues were not demonstrated) [8,9]. We determined whether use of [E1-,polymerase-]Ad vectors might allow for an improved ability of muscle tissues to secrete GAA, as this class of vector has been demonstrated to be capable of improved in vivo performance characteristics compared with [E1-]Ad-based vectors [17]. We therefore injected the gastrocnemius muscles of SCID mice with high-titer preparations of the [E1-,polymerase-] CMV–GAA vector. Although at all time points post-injection plasma GAA activity levels were not significantly different from the background GAA activity levels of mock-injected SCID mice (data not shown), secreted GAA protein is detectable by the more sensitive GAA-specific immunoblot method in the [E1-,polymerase-] CMVGAA vector intramuscularly injected mice (Fig. 9). Expression in this immune-deficient animal model was sustained, as GAA protein isoforms were detected in the plasma for several months after injection. We have previously confirmed that SCID deficient mice do not mount anti-GAA antibody responses, nor do they allow for as rapid a shutdown of the CMV enhancer, both factors that likely accounted for the sustained plasma detection of GAA in these mice [7]. Identical injections into the gastrocnemius muscles of immune-competent Gaa–/– mice also did not result in elevated plasma GAA activity levels compared with mockinjected Gaa–/– mice (data not shown). However, high levels of GAA protein were not only detected by GAA-specific immunoblot in the injected muscle, but also in uninfected liver and muscle tissues (Fig. 10). Within days of vector injection the GAA protein levels were highest in liver, with barely detectable amounts of GAA protein present in cardiac muscle or the non-injected contralateral gastrocnemius muscles. The absolute amounts of the GAA protein did seem to decrease in these tissues over time (Fig. 10). As a critical control, we confirmed that the large amounts of the GAA isoforms detected in liver tissues were not due to inadvertent Ad vector transduction of the liver (which if present may have also allowed for hepatic secretion of GAA and subsequent uptake by non-injected muscle tissues), via a sensitive PCR-based method specific for [E1-,polymerase-]Ad CMV-GAA vector genomes (Fig. 11). The PCR assay is unable to detect significant levels of [E1-,polymerase-]Ad CMV-GAA vector genome in liver DNA after the intramuscular injections, whereas the [E1-,polymerase-]Ad CMV-GAA vector genome is readily detected in the injected muscle. The results indicate that the large amounts of GAA protein detected in the liver at these same time points (Fig. 10) are most likely exogenously derived, and not due to inadvertent transduction of the liver tissues. Although substantial GAA secretion and distal uptake of muscle secreted GAA by several organs were demonstrated, this level of expression by the muscle “depot” was not sufficient to allow for “cross correction” of abnormal

MOLECULAR THERAPY Vol. 5, No. 4, April 2002 Copyright © The American Society of Gene Therapy

doi:10.1006/mthe.2002.0563, available online at http://www.idealibrary.com on IDEAL

ARTICLE

other gene therapy based strategies for patients affected by GSD-II. For example, several different doses of the [E1-,polymerase-] Ad CMVGAA vector (spanning a 20-fold range: 2.5  109 to 5  1010 particles/mouse) were found to be capable of directing high-level hepatic secretion of GAA. We also demonstrated that even at the lowest doses of vector injected, uptake of the hepatically secreted GAA into several different muscle tissues could be demonstrated. FIG. 10. Immunodetection of GAA in tissues of Gaa–/– mice intramuscularly injected with [E1-,polymerase-]Ad CMV- However, it was only when GAA. Adult Gaa–/– mice were injected into the right gastrocnemius muscle with 3  109 [E1-,polymerase-]Ad CMV- [E1-,polymerase-] CMV–GAA vecGAA particles. Individual Gaa–/– mice are labeled A–D. At the time points indicated (dpi), the mice were sacri- tor dosages exceeded 1.25  1010 ficed and the indicated tissues had 100 g of their protein content electrophoretically separated, transferred to particles/mouse that enough GAA a nylon membrane, and probed with an anti-GAA antibody. High levels of the 110-kDa GAA precursor protein (#) were present in the injected gastrocnemius muscle and livers of Gaa–/– mice at early time points. Within the was secreted from the liver into non-injected tissues, the ~ 76 kD (“+”) and ~ 67 kD (“*”) processed GAA isoforms are present in the liver tis- the plasma (plasma GAA activity sues as well as in the injected gastrocnemius muscles, when compared with similarly analyzed proteins derived level > 11,800 nmol/hour/ml) to from the same tissues of mock-injected, age-matched Gaa–/– mice. Furthermore, the arrows indicate the pres- allow for the cross-correction of ence of the ~ 67 kDa GAA isoform, present only in cardiac muscles, or the contralateral, uninjected gastrocnethe abnormal glycogen content of mius muscles of the vector injected Gaa–/– mice; these GAA protein isoforms are notably absent from samples –/– derived from the same muscles of mock-injected Gaa mice. Because of the low levels of GAA present, extended non-injected muscles (such as the autoradiographic exposure times were required to produce the image. Cross-reactivity of anti-GAA antibody to heart tissues) by ~ 18 dpi. unidentified mouse-derived proteins has been minimized, but cannot be eliminated in the tissues derived from Although lower doses of the vector Gaa–/– mice at these exposure times. Detection of the ~ 76 kDa GAA isoforms in the experimental samples was resulted in detectable hepatic GAA not able to be confidently confirmed, due to the presence of background nonspecific binding of the probe in secretion and uptake by non–/– the same noninjected Gaa mouse tissues. transduced muscle tissues (based on immunoblot analysis), these glycogen accumulation in non-injected muscle tissues lower levels were insufficient to allow for cross-correction such as the heart or the contralateral gastrocnemius mus- of non-injected muscle tissues. Furthermore, skeletal muscle, as predicted by our previous hepatic targeting/dosage cles (that is, quadriceps muscles) required higher plasma studies using high dosages of the [E1-,polymerase-]Ad GAA activity levels (compared with cardiac tissues) before evidence of glycogen “cross-correction” could be demonCMV-GAA or [E1-,polymerase-]Ad LSP-GAA vectors (data not shown). Glycogen levels were also not significantly strated. It is interesting to contemplate why skeletal muscles decreased in the directly injected muscle tissues at 3 or 14 dpi, despite GAA activity levels of ~ 26.7 nmol/hour/mg seem to be relatively less responsive than cardiac muscles to equivalent amounts of exogenously provided GAA. One protein being measured at 3 dpi in these tissues. possibility is that the amounts of the putative lysosomal targeting receptor for GAA may be at a lower density in DISCUSSION skeletal muscle membranes, relative to cardiac muscle, as has been demonstrated for the IGFII/mannose-6-phosWe previously demonstrated that intravenous injection of phate receptor [18]. Another possibility is that glycogen [E1-,polymerase-]Ad vectors encoding GAA resulted in the high-level hepatic secretion of GAA into the vascular sys- stored in skeletal muscle is intrinsically less accessible to exogenously provided GAA, relative to cardiac muscle. tem of several strains of mice, including Gaa–/– mice, a model of GSD-II. Despite both the elicitation of anti-GAA Differing metabolic rates of glycogen flux within the two antibodies and a rapid shutdown of the CMV promoter types of muscle likely also vary. The vascularity of the tis(used to drive GAA expression from the Ad vector used in sues may also vary, further influencing relative amounts of GAA taken up within them. Future studies will be those studies), the glycogen content of all Gaa–/– mouse required to discern among these possibilities. muscles analyzed after a single treatment was significantly We also investigated the potential of alternative reduced for extended periods of time (up to 190 days in enhancer/promoter elements to drive expression of a GAA cardiac muscle) [5,7]. transgene within the context of an [E1-,polymerase-]Ad vecHere, we evaluated several key parameters that could tor, relative to the CMV enhancer used to drive GAA transubstantially influence the efficacy of future gene therapy scription in our original [E1-,polymerase-]Ad vector approaches using GAA-encoding Ad vectors, as well as most

MOLECULAR THERAPY Vol. 5, No. 4, April 2002 Copyright © The American Society of Gene Therapy

443

ARTICLE

doi:10.1006/mthe.2002.0563, available online at http://www.idealibrary.com on IDEAL

constructs. Of several nonvirally derived enhancer/promoter elements analyzed (all previously shown to allow for high-level hepatic expression of vector transgenes), only a modular, liver-specific enhancer/promoter (LSP) allowed for detectable secretion of GAA from liver tissues in vivo. Despite this, the absolute amounts of GAA mRNA transcribed from the [E1-,polymerase-]Ad LSP-GAA vector genome were significantly less than the amounts measured from the genome of the [E1-,polymerase-]Ad vector containing –/– a CMV-based enhancer/promoter element, FIG. 11. Detection of [E1-,pol-]Ad GAA vector DNA in tissues of Gaa mice 3 days after intramuscular injection. Primers used in this PCR assay are specific for the CMV-GAA transgene, yielding an ~ 750at least at early time points after injection. bp product if the sequence is present in the sample. One microgram of DNA extracted from the injected At the highest injected dosages of the gastrocnemius demonstrated high copy number presence of vector sequences. Note the lack of [E1-,polymerase-] LSP-GAA vector, a slight detectable vector GAA sequences in 1 g of DNA derived from the liver tissues of these same animals. reduction in glycogen levels in the heart The results strongly suggest that the high amounts of GAA protein found in the livers of the same mice tissues was noted, but the levels of GAA (Fig. 10) were likely exogenously derived, and not due to direct transduction of liver tissues with the [E1-,polymerase-]Ad CMV-GAA vector. As a control for the sensitivity of the PCR analysis, 1 g of liver secreted did not allow for any evidence of DNA from a mock-infected mouse was spiked with decreasing amounts of the [E1-,polymerase-]Ad CMVcross-correction of glycogen in skeletal GAA vector DNA containing plasmid, at the indicated copy numbers/per hepatocyte. The PCR can detect at least one vector genome per 250 hepatocytes. muscles, such as the quadriceps. Our results also demonstrated that the [E1-,polymerase-] LSP-GAA and [E1-,polymerase-] ALB–GAA vector persistently transcribed [E1-,polymerase-]Ad CMVGAA vectors, and subsequently GAA mRNA in the livers of all injected Gaa–/– mice for at express and secrete GAA. In contrast to previous studies least 6 months, but our results also suggested that the elic- using [E1-]Ad vectors encoding GAA, we not only itation of anti-GAA antibodies within several weeks of gene observed GAA expression in the injected muscle, but we also demonstrated that the level of expression was adetransfer may have limited the time that muscle tissues were exposed to hepatically secreted GAA. This result sug- quate to allow for secretion of GAA into the plasma, as detected by a GAA-specific immunoblot assay. We also gests that in the Gaa–/– mouse model of GSD-II, a rapid, high-level expression of GAA is required before cross-cor- confirmed that the muscle-derived GAA detected in rection will be demonstrated with any putative GAA gene plasma was secreted by the injected muscle (and not from delivery vector, as rapid onset of anti-GAA specific anti- inadvertent transduction of liver tissues after intramusbodies will abrogate any efficacy at later time points after cular injection), by a sensitive, PCR-based ([E1-,polymerase-]Ad CMV-GAA genome specific) detection techvector injection [12]. nique. Unfortunately, despite the uptake of Our results demonstrated that the relative titers of antimuscle-secreted GAA by multiple muscle groups, none of GAA antibodies elicited in Gaa–/– were lowered (but not completely avoided) by the use of a tissue-specific these muscles in the Gaa–/– mice demonstrated any evienhancer/promoter to drive GAA transcription from an dence of glycogen “cross-correction.” This result was in [E1-,polymerase-]Ad vector. This is somewhat similar to fact consistent with our hepatic dosing studies, which other reports demonstrating that tissue-specific demonstrated that high plasma levels of GAA are required enhancer/promoter elements can completely prevent the before glycogen “cross-correction” can be demonstrated onset of AAT antibodies after transduction of human AAT in Gaa–/– mice by 18 dpi (before onset of anti-GAA antiinto C3WheJ mice [14]. Our lack of similar results may be bodies). Our results also demonstrated another limitation due to the fact that mice express several isoforms of mouse of muscle targeting approaches, namely that the liver tisAAT, each highly homologous to human AAT, a situation sue avidly sequestered most of the GAA secreted from the that is not replicated for mouse Gaa [19,20]. Injection of injected gastrocnemius muscle, which likely prevented C3WheJ mice with an Ad vector using a tissue-specific uptake of GAA by other muscle tissues. Possibly, injection enhancer to drive AAT transcription, as well as the expres- of greater numbers of muscle fibers within an individual sion of several AAT protein isoforms by the mice, may animal with muscle-optimized versions of the [E1-,polyhave both contributed to the complete lack of elicitation merase-]Ad GAA encoding vector and/or use of alternative of anti-AAT antibodies in those studies. vectors (that is, AAV-based vectors) may allow for skeleIn an attempt to further broaden potential clinical tal muscle to serve as an efficacious source of GAA secretreatment options for GSD-II patients, we also analyzed tion. Studies are underway in our laboratories to address the ability of skeletal muscle to be transduced by these possibilities.

444

MOLECULAR THERAPY Vol. 5, No. 4, April 2002 Copyright © The American Society of Gene Therapy

doi:10.1006/mthe.2002.0563, available online at http://www.idealibrary.com on IDEAL

Finally, we noted that glycogen reductions were also not noted in the directly injected muscles, although significant amounts of GAA activity were noted in this tissue. This dichotomy was likely due to the Ad vector only infecting a small percentage of the muscle fibers within the entire gastrocnemius muscle. Ad vectors have a poor ability to transduce adult muscle fibers due to downregulation of the CAR molecule, which likely caused the lack of complete infection [21]. Therefore, although the overall levels measured seemed elevated, in fact the GAA protein was likely being expressed in only a small fraction of muscle cells, preventing significant drops in glycogen levels from being measured when the entire muscle glycogen content was assessed. Higher vector dosage levels, an increased number of intramuscular injections, or use of either tropism-modified Ad vectors or alternative serotype AAV vectors may overcome this hurdle.

MATERIALS

AND

METHODS

Construction of [E1,polymerase-] adenovirus vectors encoding GAA. The construction of the modified [E1-,polymerase-] adenovirus encoding human GAA (transcription driven by the CMV based enhancer/promoter element) has been described [7]. Plasmid pAAV-LSP-cFIX (contains LSP) was described previously [12,22]. The BamHI subfragment containing the modular LSP enhancer/promoter element (786 bp) from pAAV.LSP-cFIX was subcloned into the BamHI site of pcDNA3 (Invitrogen, Carlsbad, CA) generating pcDNA3+LSP. The KpnI and EcoRV subfragment containing the LSP enhancer from pcDNA3+LSP was subcloned into KpnI-NotI digested pShuttle pA (NotI site blunt-end filled) yielding pShuttle+LSPpA. The HindIII and XhoI subfragment of pcDNA-GAA containing the GAA-cDNA was subcloned into the HindIII and SalI sites of pShuttleLSPpA, generating pShuttle LSPGAApA [4]. The HindIII-XhoI subfragment containing the GAA gene of pcDNA3-GAA was subcloned into the HindIII and SalI site of pShuttlepA yielding pShuttleGAApA. pShuttleGAApA was digested with XhoI and partially digested with NotI; the NotI-XhoI subfragment of murine pBSKK+Albumin (containing the murine ALB promoter [13]) was subcloned into pShuttle-GAApA, yielding pShuttle ALB GAApA. The XbaI-HindIII subfragment of pAdEF-AAT containing the human elongation factor 1a (EF) promoter was subcloned into the XbaI-HindIII site of pShuttlepA, yielding pShuttleEFpA [11]. The HindIII-XhoI subfragment of pcDNA3GAA (containing the GAA-cDNA) was subcloned into the HindIII-SalI site of pShuttleEFpA, yielding pShuttleEF1aGAApA. PShuttleLSPGAApA, pShuutleALBGAApA, and pShuttleEFGAApA were digested with PmeI, and homologously recombined with pAdDpol, yielding p[E1-,polymerase-]LSP GAA, p[E1-,polymerase-]ALB GAA, and p[E1,polymerase-]EFGAA, respectively. The bacterial plasmids were digested with PacI and CaPO4 transfected into C7 cells, and the infectious vectors were subsequently isolated, amplified, concentrated, titered, and genomic structures confirmed as described [23,24]. Particle numbers are assessed essentially as previously, except virus was disrupted in 0.5% SDS before optical density measurement [7]. In vivo administration of GAA encoding vectors. Different doses of the [E1-,polymerase-] GAA-encoding vectors were intravenously administered (via the retro-orbital sinus) into 3-month-old Gaa–/– mice [25]. Plasma or tissue samples were obtained and processed at the indicated time points postinjection. In one set of experiments, the [E1-,polymerase-]Ad CMVGAA vector was directly injected into skeletal muscle at the particle dosages indicated in the figure legend. All animal procedures were done in accordance with Duke University Institutional Animal Care and Use Committee approved guidelines. Detection of cell/plasma GAA activities and tissue glycogen content assays. For plasma GAA activity detection, blood samples were collected

MOLECULAR THERAPY Vol. 5, No. 4, April 2002 Copyright © The American Society of Gene Therapy

ARTICLE

by retro-orbital sampling into heparinized capillary tubes, and plasma isolated via centrifugation. For tissue GAA activity measurement, the respective tissues were obtained from treated or control mice, snap-frozen in liquid nitrogen, homogenized, sonicated in water, and insoluble proteins removed by centrifugation. The protein concentrations of the clarified suspensions were quantified via the Bradford assay. GAA activity in the plasma or tissues was then determined as described [5,7]. Glycogen content of tissues derived from mouse tissues was measured using the Aspergillus niger assay system, as described [26]. A two-tailed homoscedastic Student’s t-test was used to determine significant differences in glycogen content between control and vector exposed tissues. Western blot analysis of GAA. For direct detection of GAA in plasma, 4 l of each plasma sample was electrophoresed through a 6% SDS-polyacrylamide gel, and electrotransferred onto a nitrocellulose membrane. The membrane was blocked with nonfat milk at 4C overnight, incubated with a 1:2000 dilution of a rabbit anti-human GAA polyclonal antibody, washed, probed with a 1:5000 dilution of an anti-rabbit, IgG- peroxidase linked antibody derived from donkey, and visualized via the ECL detection system (Amershan Pharmacia, Piscataway, NJ). For immunodetection of GAA in tissues, the respective tissues were frozen, homogenized, sonicated, and centrifuged to remove insoluble proteins. The protein content of the supernatants was measured by the Bradford assay. Samples (100 g protein) were electrophoretically separated and transferred to a nitrocellulose membrane. The blots were blocked with nonfat milk, incubated with the same primary and secondary antibodies as described above, and visualized as described above. ELISA detection of plasma anti-GAA antibody from treated mice. Recombinant GAA (5 g) in carbonate buffer was coated onto each well of a 96-well plate at 4C overnight. After washing with phosphate buffered saline (PBS) containing 0.05% Tween 20, 1:10 dilutions of the plasma (all samples yielded absorbance values that were within the linear range of the assay at this dilution) were added to each well, and incubated for 1 hour at room temperature. The wells were washed with 0.05% Tween 20 + PBS, incubated with a 1:2500 dilution of alkaline phosphase-conjugated sheep antimouse IgG (H+L) at room temperature for 1 hour, washed, and alkaline phospatase substrates (p-Nitrophenyl phosphate) added. The absorbance values of the plates were read at 405 nm with a Bio-Rad microplate reader. PCR detection of vector derived DNA and RT-PCR detection of GAA mRNA. For PCR quantitation of vector DNA in Gaa–/– tissues, tissue DNA was extracted as described [17]. Liver tissue DNA (1 g) derived from each respective mouse was subjected to PCR amplification. For [E1-,polymerase] vector DNA detection, primers were those that spanned nucleotides 70–450 of the GAA DNA (forward, 5-GCACTCCGACTACATCCGGAGAAGA-3, and reverse, 5-CCTTCTGGTGTTATTTGTCGACCTC-3; PCR product of 380 bp). As an internal control for ensuring adequate DNA isolation and amplification, a second PCR was repeated on all samples, specific for sequences present in the mouse glyceraldehyde-3-phosphate dehydrogenase (G3PDH) gene (forward, 5-ACCACAGTCCATGCCATCAC-3, and reverse, 5-TCCACCACCCTGTTGCTGTA-3; PCR product ~ 452 bp). All PCRs were subjected to the following procedures: 94C for 3 minutes; followed by 16 or 22 cycles of 94C for 30 seconds, 55C for 20 seconds, and an extension at 72C for 1 minute, and a final extension cycle at 72C for 10 minutes. All samples were positive for amplification derived from the mouse G3PDH gene, confirming the integrity of the DNA samples evaluated with the GAA minigene-specific primers described above. With these parameters, the amount of the respective PCR products generated was quantitated, as compared with control reactions of decreasing amounts of liver DNA isolated from a mock-injected Gaa–/– mouse (for detection of murine G3PDH sequences), or constant amounts of Gaa–/– derived DNA spiked with increasing quantities of the p[E1-,polymerase-]Ad CMV-GAA plasmid. Human GAA or murine G3PDH specific PCR products were visualized in 0.7% agarose gels, as described [7]. Quantitation of [E1-,polymerase-]Ad vector genomes or of the murine G3PDH gene detected in each sample was analyzed by densitometric analysis of images of the gels with use of the SCION imaging software package as described [23]. The vector genome copy number of all samples was normalized relative to the G3PDH DNA content of the respective samples.

445

ARTICLE

doi:10.1006/mthe.2002.0563, available online at http://www.idealibrary.com on IDEAL

For detection of GAA-specific mRNA sequences, total RNA was isolated from the livers or gastrocnemius muscles of the respective mice using the Promega total RNA isolation kit as per the manufacturer’s recommendations (Madison, WI). Tissue (5 g) RNA was digested with DNase I for 15 minutes at room temperature and the reaction was stopped by the addition of EDTA to 25 mM, and heating for 10 minutes at 65C. Complementary DNA was generated from equivalent amounts of the DNase I-treated RNA samples by the addition of 20 ng of random primers (Gibco BRL) and SuperScript reverse transcriptase (RT; Gibco BRL) in vendor supplied RT buffer (Gibco BRL) at 42C for 1 hour. Equivalent volumes of cDNA derived from the reactions were subjected to PCR amplification, using primers that spanned nucleotides 70–450 of the GAA cDNA and thermocycling parameters as described for vector genome DNA detection (above). As a control for ensuring adequate RNA isolation and amplification, a second RT-PCR reaction was repeated with all cDNA samples, which was specific for sequences present in the mouse G3PDH transcript (reaction parameters were also as described above). The amounts of GAA RNA detected in each sample were normalized to the amounts of G3PDH RNA detected in each sample. As a quantitative assessment of GAA mRNA expression levels, cDNAs derived from the RT-PCR of mock-infected Gaa–/– liver tissues were spiked with decreasing amounts of the pAd[E1-,polymerase-]CMVGAA plasmid as indicated. The reaction products of these control reactions were visualized and compared with the relative intensities of the same PCR products present in the experimental tissue samples, to ensure that the cycling parameters were conducted within the linear range of detection of the GAA-specific PCR quantitative assay. The amounts of [E1-,polymerase-]Ad vector transcripts or G3PDH transcripts detected were compared by densitometric analysis of the gels containing the PCR products and use of the SCION imaging software package as described [23]. The relative amounts of GAA transcripts per sample was then normalized relative to the G3PDH RNA transcripts detected in the same samples; this amount was also compared with the amounts of vector genome detected in the samples.

ACKNOWLEDGMENTS We thank Jude Samulski (Chapel Hill, NC) for the LSP enhancer containing plasmid; Randy Eisensmith (New York, NY) for the EF1- enhancer containing plasmid; and Dwight Koeberl (Durham, NC) for the murine albumin enhancer containing plasmid. We acknowledge support from the Muscular Dystrophy Association (USA), NIH grant RO1-DK52925, SynPac Pharmaceuticals, and the Genzyme Corporation. RECEIVED FOR PUBLICATION NOVEMBER 13, 2001; ACCEPTED FEBRUARY 12, 2002.

REFERENCES 1. Hirschhorn, R., and Reuser, A. J. (2001). Glycogen storage disease type II: acid -glucosidase (acid maltase) deficiency. In The Metabolic and Molecular Basis of Inherited Disease (C. R. Scriver et al., Eds.), pp. 3389–3420. McGraw-Hill, New York. 2. Amalfitano, A., et al. (2001). Recombinant human acid -glucosidase enzyme therapy for infantile glycogen storage disease type II: results of a phase I/II clinical trial. Genet. Med. 3: 132–138. 3. Van den Hout, H., et al. (2000). Recombinant human -glucosidase from rabbit milk in Pompe patients. Lancet 356: 397–398. 4. Vanhove, J. L. K., Yang, H. W., Wu, J. Y., Brady, R. O., and Chen, Y. T. (1996). Highlevel production of recombinant human lysosomal acid -glucosidase in Chinese ham-

446

ster ovary cells which targets to heart muscle and corrects glycogen accumulation in fibroblasts from patients with Pompe disease. Proc. Natl. Acad. Sci. USA 93: 65–70. 5. Amalfitano, A., et al. (1999). Systemic correction of the muscle disorder glycogen storage disease type II after hepatic targeting of a modified adenovirus vector encoding human acid--glucosidase. Proc. Natl. Acad. Sci. USA 96: 8861–8866. 6. Pauly, D. F., et al. (2001). Intercellular transfer of the virally derived precursor form of acid -glucosidase corrects the enzyme deficiency in inherited cardioskeletal myopathy Pompe disease. Hum. Gene Ther. 12: 527–538. 7. Ding, E. Y., et al. (2001). Long term efficacy after [E1-,polymerase-] adenovirus mediated transfer of the human acid--glucosidase gene into GSD-II knockout mice. Hum. Gene Ther. 12: 955–965. 8. Pauly, D. F., et al. (1998). Complete correction of acid -glucosidase deficiency in Pompedisease fibroblasts in vitro, and lysosomally targeted expression in neonatal rat cardiac and skeletal muscle. Gene Ther. 5: 473–480. 9. Tsujino, S., et al. (1998). Adenovirus-mediated transfer of human acid maltase gene reduces glycogen accumulation in skeletal muscle of Japanese quail with acid maltase deficiency. Hum. Gene Ther. 9: 1609–1616. 10. 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 NF--B. J. Virol. 72: 180–190. 11. Guo, Z. S., Wang, L. H., Eisensmith, R. C., and Woo, S. L. (1996). Evaluation of promoter strength for hepatic gene expression in vivo following adenovirus-mediated gene transfer. Gene Ther. 3: 802–810. 12. Wang, L. L., Takabe, K., Bidlingmaier, S. M., Ill, C. R., and Verma, I. M. (1999). Sustained correction of bleeding disorder in hemophilia B mice by gene therapy. Proc. Natl. Acad. Sci. USA 96: 3906–3910. 13. Pinkert, C. A., Ornitz, D. M., Brinster, R. L., and Palmiter, R. D. (1987). An albumin enhancer located 10 kb upstream functions along with its promoter to direct efficient, liver-specific expression in transgenic mice. Genes Dev. 1: 268–276. 14. Pastore, L., et al. (1999). Use of a liver-specific promoter reduces immune response to the transgene in adenoviral vectors. Hum. Gene Ther. 10: 1773–1781. 15. Svensson, E. C., et al. (1997). Long-term erythropoietin expression in rodents and nonhuman primates following intramuscular injection of a replication-defective adenoviral vector. Hum. Gene Ther. 8: 1797–1806. 16. Monahan, P. E., et al. (1998). Direct intramuscular injection with recombinant AAV vectors results in sustained expression in a dog model of hemophilia. Gene Ther. 5: 40–49. 17. Hu, H. M., Serra, D., and Amalfitano, A. (1999). Persistence of an [E1-, polymerase-] adenovirus vector despite transduction of a neoantigen into immune-competent mice. Hum. Gene Ther. 10: 355–364. 18. Liu, J. L., et al. (1998). Insulin-like growth factor-I affects perinatal lethality and postnatal development in a gene dosage-dependent manner—manipulation using the cre/loxp system in transgenic mice. Mol. Endocrinol. 12: 1452–1462. 19. Borriello, F., and Krauter, K. S. (1991). Multiple murine  1-protease inhibitor genes show unusual evolutionary divergence. Proc. Natl. Acad. Sci. USA 88: 9417–9421. 20. Tardiff, J., and Krauter, K. S. (1998). Divergent expression of 1-protease inhibitor genes in mouse and human. Nucleic Acids Res. 26: 3794–3799. 21. Huard, J., et al. (1995). Differential short-term transduction efficiency of adult versus newborn mouse tissues by adenoviral recombinants. Exp. Mol. Pathol. 62: 131–143. 22. Somia, N. V., Kafri, T., and Verma, I. M. (1999). Piecing together more efficient gene expression. Nat. Biotechnol. 17: 224–225. 23. Hodges, B. L., et al. (2000). Multiply deleted [E1-,polymerase-, and pTP-] adenovirus vector persists despite deletion of the preterminal protein. J. Gene Med. 2: 250–259. 24. Amalfitano, A., et al. (1998). Production and characterization of improved adenovirus vectors with the E1, E2b, and E3 genes deleted. J. Virol. 72: 926–933. 25. Raben, N., et al. (1998). Targeted disruption of the acid -glucosidase gene in mice causes an illness with critical features of both infantile and adult human glycogen storage disease type II. J. Biol. Chem. 273: 19086–19092. 26. Kikuchi, T., et al. (1998). Clinical and metabolic correcton of Pompe disease by enzyme therapy in acid maltase-deficient quail. J. Clin. Invest. 101: 827–833.

MOLECULAR THERAPY Vol. 5, No. 4, April 2002 Copyright © The American Society of Gene Therapy