Gene Ther Mol Biol Vol 4, 45-58. December 1999.

Efficient expression of ribozyme and reduction of stromelysin mRNA in cultured cells and tissue from rabbit knee via Adeno-associated Virus (AAV)

Research Article

Elisabeth Roberts*, Piruz Nahreini*, Kristi Jensen, Ira von Carlowitz, Karyn Bouhana, Stephen Hunt III**, Thale Jarvis, Larry Couture***, and Dennis Macejak

Ribozyme Pharmaceuticals Inc., 2950 Wilderness Place, Boulder, Colorado 80301, USA

*these authors contributed equally; ** Parke-Davis Pharmaceutical Research, Division of Warner-Lambert Company, Ann Arbor, MI 48105; *** Present address: City of Hope National Medical Center, 1500 E. Duarte Rd., Duarte, CA 91010; A preliminary account of this study was presented at Cold Spring Harbor symposium on Gene therapy on September 25th, 1996.

______________________________________________________________________________________

Correspondence: Dennis Macejak, Ribozyme Pharmaceuticals Inc., 2950 Wilderness Place, Boulder, Colorado 80301; Tel: (303) 546-8153; Fax: (303)-449-6995; E-mail: macejad@rpi.com .

Key Words: hammerhead ribozymes, Adeno-associated virus, gene delivery, osteoarthritis, stromelysin, nerve growth factor receptor

Received: 26 October 1999; accepted 3 November 1999

Summary

The potential use of Adeno-associated virus (AAV) as an efficient gene delivery vector is increasingly being recognized in human gene therapy. We have investigated the utility of recombinant AAV (rAAV) vectors for the delivery and expression of hammerhead ribozymes targeted against a cellular mRNA encoding a matrix metalloproteinase, stromelysin. Stromelysin expression has been linked to the pathogenesis of human osteoarthritis. We have constructed several rAAV backbone plasmids containing single or multiple hammerhead ribozymes expression cassettes under the control of either the tRNA, U1, or U6 promoter, and have used these plasmids to generate rAAV. These rAAV vectors also contain a selectable marker, the truncated nerve growth factor receptor (NGFR) driven by the cytomegalovirus immediate early gene promoter. rAAV expressing stromelysin-specific ribozyme transduced ex vivo cultured rabbit synovial fibroblasts (RSFs) with a greater than 95% efficiency. Stable ribozyme expression can readily be detected throughout the life span of RSFs in culture. Furthermore, ribozyme mediated knockdown of stromelysin mRNA was detected in RSFs infected by a rAAV containing the tRNA-based transcription unit.

I. Introduction

Human arthritic disorders are the major cause of chronic disability among adults (Vincenti, Clark, and Brinckerhoff, 1994). Several etiological factors have been reported to play roles in the initiation and progression of degenerative cartilage disorders. For example, a group of proteolytic enzymes, including several members of the matrix metalloproteinase (MMP) family, are strongly implicated in the pathology of arthritis (Vincenti, Clark, and Brinckerhoff, 1994) (Cawston, 1996). MMPs degrade extracellular matrix components such as collagens, gelatins, proteoglycan, and fibronectin (Woessner, 1991).

MMPs normally play an important role in embryogenesis, wound healing, and tissue remodeling (Werb, Alexander, and Adler, 1992); however, abnormal expression has been implicated in a wide range of human diseases such as atherosclerosis, arthritis (Murphy and Hembry, 1992), glomerulonephritis, corneal ulceration, periodontitis, encephalomyelitis, and tumor metastasis (Stetler-Stevenson, Aznavoorian, and Liotta, 1993).

Most MMPs (except for MMP-11 and MT-MMP) are secreted into the extracellular matrix as proenzymes. Stromelysin (MMP-3) may be a key mediator in arthritic diseases. It degrades proteoglycans and a wide range of other matrix components (Woessner, 1991) and activates the proenzyme forms of collagenase (Suzuki et al., 1990), gelatinases (Miyazaki et al., 1992) (Ogata, Enghild, and Nagase,1992), and other MMPs, leading to initiation of a proteolytic cascade. Synovial fibroblasts derived from osteoarthritic or rheumatoid synovium produce high levels of stromelysin upon stimulation (Brinckerhoff and Auble, 1990). In addition, there is a significant up-regulation of stromelysin and other MMPs in articular tissues from patients with osteo- or rheumatoid arthritis (Hembry et al., 1995) (Okada et al., 1992). Thus targeted inhibition of one or more of these proteolytic activities may be a valid therapeutic approach for arthritis.

Ribozymes are RNA-based enzymes that have the ability to cleave RNA molecules in a sequence-specific manner. Sequence specificity comes from the base-pairing of ribozyme sequences with nucleotides spanning the cleavage site of the target RNA. Cleaved RNA is rapidly degraded in the cell, resulting in a decrease in expression of the encoded protein. Because of their sequence-specificity, ribozymes can be utilized as a therapeutic agent to down-regulate a specific RNA in the background of other cellular RNAs (Rossi, 1999). We have previously reported that a chemically stabilized synthetic hammerhead ribozyme targeting nucleotide position 1049 in the human stromelysin mRNA significantly reduced the target RNA levels upon intra-articular injection (Flory et al., 1996). In this study, we have used recombinant AAV (rAAV) for the expression of hammerhead ribozymes targeting stromelysin mRNA at the same nucleotide position.

Because ribozymes can function as RNA molecules they can be synthesized in a variety of transcription units besides RNA polymerase II mRNAs. However, the success of ribozyme efficacy is dependent upon at least three parameters: ribozyme intracellular localization, ribozyme cleavage activity, and ribozyme RNA levels. Ribozyme localization is somewhat dependent upon the inherent characteristics of the particular transcription unit used and of the RNA elements contained within the transcript (Bertrand et al., 1997; Thompson et al., 1995b). Ribozyme cleavage activity is effected by flanking sequences contained within the transcript (Chowrira, Pavco, and McSwiggen, 1994; Thompson et al., 1995a) and ribozyme RNA levels are limited by promoter strength and ribozyme RNA stability (Thompson et al., 1995a) (Thompson et al., 1995b) (Rossi, 1999). Thus, we have designed and constructed a variety of ribozyme transcription units that maintain catalytic activity and have the potential to accumulate to high levels in the nucleus or to utilize endogenous splicing machinery to promote ribozyme/target RNA hybridization. We have previously reported the construction of a modified tRNA promoter for the expression of an HIV specific hammerhead ribozyme (Thompson et al., 1995a). In this report, we compare the efficacy of ribozymes expressed from modified tRNA, U6 snRNA, and U1 snRNA transcription units.

Ribozyme expression via retroviral and adenoviral vectors has been reported in several studies (Macejak et al., 1999) (Thompson et al., 1995a), however the utility of AAV for ribozyme expression has not been explored extensively. Retroviruses and adeno-associated virus (AAV), by virtue of integrating into chromosomal DNA, are attractive gene delivery vehicles for the treatment of human disorders in which a long-term therapy is essential for an effective treatment. The attractive features of AAV as a vector are non-pathogenicity, low immunogenicity, stable and efficient expression of transgenes from the integrated or episomal form, infection of non-dividing cells, broad host range, generation of high titer (8 x 10⁸ IU/ml), and physically stable virions. Recently, AAV-mediated expression of ribozymes targeting a mutated rhodopsin mRNA was shown to slow the rate of photoreceptor degeneration in a transgenic rat model (Lewin et al., 1998).

AAV is a small non-pathogenic human parvovirus whose genome (4681 bases) is a single-stranded DNA of either polarity, flanked by inverted terminal repeats (ITRs). AAV-ITRs are 145 bases in length and function as the sole cis-acting elements essential for chromosomal excision (rescue), integration, replication, and encapsidation of nascent viral DNA (Muzyczka, 1992). AAV infects a variety of mammalian cells with a broad host range; however, some human megakaryocytic cell lines are refractory to AAV infection, presumably because they lack the putative AAV receptor (Ponnazhagan et al., 1996). AAV can establish a lytic or latent infection in mammalian cells in the presence or absence of a helper virus, respectively. Productive AAV infection is dependent on a helper DNA virus, which is usually adenovirus; however, herpes virus and vaccinia virus can substitute for the helper functions of adenovirus (Schlehofer, Ehrbar, and zur Hausen, 1986) (Thomson et al., 1994). In the absence of a helper virus, the AAV genome preferentially integrates into a defined region of human chromosome 19 (q13.3-qter), and establishes a stable latent infection (Kotin et al., 1990) (Samulski et al., 1991). In this study, we demonstrate the utility of rAAV as a vector for ribozyme expression in primary synovial cells of the rabbit knee, and report the feasibility of rAAV-mediated gene therapy for arthritic disorders.

II.Results

A. rAAV-LacZ transduction of synoviocytes and chondrocytes.

Although AAV is known to infect a variety of mammalian cell lines, some cells, such as megakaryocytic MB-O2 and MO7e cell lines, are refractory to AAV infection (Ponnazhagan et al., 1996). Therefore, we first tested whether AAV can infect primary cells from the intra-articular lining of rabbit knees. Primary synovial fibroblasts from rabbit knee tissue (RSFs) were cultured ex vivo and were infected with rAAV containing a b-galactosidase gene driven by the cytomegalovirus immediate early promoter. rAAV-mediated LacZ expression was readily detected in RSF cells (Figure 1). The LacZ expression was detectable 30 hrs post-infection and remained stable during the entire life span of primary RSF cells in culture (5 – 7 passages, data not shown).

Rabbit cartilage explants infected with rAAV also demonstrated LacZ expression (Figure 1). It is not clear, however, whether the infected cells in the cartilage explants were chondrocytes or another cell type(s). In addition, rAAV-mediated LacZ expression can readily be detected in the chondrocyte sarcoma cell line SW1353 (Figure 1). Together, these observations indicate that rAAV-LacZ can infect the ex vivo cultured RSFs and cells of rabbit cartilage explants.

B. rAAV vectors and ribozyme transcription units

The backbone plasmid, in which all ribozyme transcription units were inserted, contains a modified nerve growth factor receptor (NGFR) gene, driven by the cytomegalovirus early promoter, as a selection marker. Because of a deletion in the c-terminus of the cDNA, NGFR is biologically inactive when expressed from rAAV-infected cells. However, rAAV-mediated expression of this altered NGFR can be readily detected on the membrane of transduced cells by Fluorescence Activated Cell Scanning (FACS; see Figure 5A).

Several ribozyme transcription units were cloned into the backbone rAAV plasmid, and the resulting plasmids were used to generate rAAV particles. The rAAVs used in this study are shown in Figure 2 and transcription units are shown in Figure 3A. The modified tRNA transcription unit was originally developed to increase the copy number and maintain catalytic activity of ribozyme expressed inside cells (Thompson et al., 1995a). To generate ribozyme with minimal flanking sequence we chose the U6 snRNA promoter since the U6 promoter is extragenic except for the "G" at +1. We have developed a U6+1-Rz (called "U6C", see Figure 3A) that contains a 5'/3' stem interaction for stability, analogous to that in the improved "TRZ-motif" (Thompson et al., 1995a). Certain splicosomal RNAs interact with pre-mRNA molecules during normal pre-mRNA processing and the sequences which mediate these interactions in the context of a snRNP are known. The first snRNP to contact a pre-mRNA contains U1 RNA which hybridizes to the 5' splice site. We reasoned that replacing the U1 RNA sequences that hybridize to the 5' splice site with a ribozyme (Figure 3A) could promote hybridization of a U1-ribozyme snRNP to a target cleavage site in the nucleus.

To confirm that cleavage activity was maintained by each ribozyme within the U1-, U6-, or tRNA-derived RNA transcripts, ribozyme RNAs were synthesized in vitro to contain the flanking sequences predicted for each transcription unit. The chimeric ribozyme RNAs were then tested for cleavage activity against a RNA containing the target site. All ribozyme transcripts contained comparable cleavage activity in vitro that was partially reduced compared to a chemically synthesized ribozyme with no extraneous sequence (Figure 3B). To further characterize these ribozyme transcripts within cells, we analyzed their stability in 293 cells following Actinomycin D treatment (Figure 3C). The U6- and tRNA-derived ribozyme RNAs had a similar stability with a half-life of 1-1.5 h. On the other hand, no reduction in the level of U1-derived ribozyme RNA was detected over 4h. In addition, we observed that the U1-ribozyme was immunoprecipitable with anti-trimethyl-G or anti-SM antibodies (our unpublished results), indicating that the U1-ribozyme transcript goes through a maturation/modification process analogous to authentic U1 RNA.

C. Efficient transduction and expression of a stromelysin-specific hammerhead ribozyme in RSFs

Since rAAV could infect RSFs and the ribozyme chimeric transcripts retained cleavage activity, a series of rAAV containing one or more expression cassettes encoding ribozymes targeting stromelysin mRNA were prepared (Figure 2). Although ribozyme expression was detected in 293 cells, the promoter activity of the ribozyme transcription units in primary RSFs was uncertain. Thus, we investigated ribozyme expression in rAAV-infected RSFs. RSF cells were infected (moi = 20) with rAAV containing either single or multiple transcription units (see Figure 2). Northern analyses demonstrate that each transcript was expressed in RSFs (data from initial experiments not shown, but see Figure 5B). We then investigated the duration of ribozyme expression. RSF cells were infected with a subset of rAAV (containing either U6 or both U1 and U6 transcription units). Cells were harvested and total RNA purified for Northern analysis at different cell passages post-infection. Cells were infected at passage 3, and analyzed for RNA 48 hours after infection, prior to passaging (Figure 4, P3 lane). Ribozyme expression from U6 and U1 transcription units is readily detectable during the entire life span of RSF cells in culture (6 passages; Figure 4). Ribozyme expression from the U1 and U6 promoters significantly increased upon passaging. Transduction efficiencies as determined by NGFR expression using FACS analysis were similar (>90%) with each of the rAAVs tested.

The level of stromelysin RNA was induced upon rAAV infection alone 3-4 fold over that of Il-1 treatment alone (data not shown). Thus, in terms of ribozyme efficacy, cells infected with the empty vector, vAT22, is the appropriate control for comparison. The level of stromelysin target RNA was slightly reduced in RSFs expressing U1- or U6-derived ribozymes compared to RSFs infected with the control vector (Figure 5C, vAT23 or vAT36 versus vAT22; p < 0.05). The greatest reduction (>60%) in stromelysin RNA was observed in RSF expressing tRNA-ribozyme (vAT42 versus vAT22; p < 0.05). Interestingly, the U1- and U6-derived ribozyme appeared to have an additive effect in vAT43 infected cells, but this effect was lost in cells infected with virus containing the trimeric ribozyme cassette (vAT44).

E. Reduction in target RNA is due to a ribozyme mechanism.

To confirm that the reduction in stromelysin RNA observed was due to ribozyme cleavage activity and not due to over-expression of the tRNA motif itself, we compared target RNA levels in RSFs infected with rAAV encoding either a ribozyme targeting an irrelevant sequence (vAT30), the ribozyme targeting site 1049 of stromelysin (vAT42), or an attenuated version of the stromelysin ribozyme (vAT46); all genes were driven by the tRNA promoter within an otherwise identical rAAV vector. The attenuated control contains two base substitutions in the ribozyme catalytic core domain. This attenuated analog is still capable of binding the target site, but has reduced cleavage activity. Previous work demonstrated the greatly reduced catalytic effects of base substitutions in the ribozyme core on ribozyme activity in vitro (Ruffner, Stormo, and Uhlenbeck, 1990) as well as in cell culture and in vivo (Jarvis et al., 1996) (Flory et al., 1996).

The active and attenuated versions of the stromelysin-specific ribozyme and the irrelevant HIV-specific ribozyme were each readily detectable in these cells by Northern analysis (Figure 6A). The probe used in this Northern blot was designed to hybridize to tRNA sequences, such that the same probe will detect all three ribozyme transcripts as well as endogenous tRNA. After normalization to endogenous tRNA, it appears that the irrelevant (HIV-specific) ribozyme (vAT30) level is the greatest, yet no reduction in the level of stromelysin RNA is observed with rAAV encoding irrelevant ribozyme (Figure 6B). As observed previously, RSF infected with rAAV containing active ribozyme (vAT42) significantly reduced stromelysin mRNA levels as compared to either control virus (Figure 6B; p < 0.05). This reduction was not dependent upon IL-1 treatment since vAT42-infected RSFs without IL-1 also displayed reduced target RNA levels compared to RSFs infected with either vector control vAT22 or vAT30 (p < 0.05). The attenuated version of the stromelysin-specific ribozyme (vAT46) did not show a statistically significant decrease in stromelysin mRNA levels relative to the control vector vAT22 in the presence or absence of IL-1. Statistical analysis was done with Kruskal-Wallis One Way Analysis of Variance on Ranks to test for differences between groups. When significant, post-hoc analysis was done using the Dunnett's test and p values < 0.05 were considered significant. These results confirm that the decrease in stromelysin mRNA observed in vAT42 infected cells is largely due to a ribozyme cleavage-dependent mechanism.

III. Discussion

Arthritis is a chronic debilitating disease affecting 16-40 million people in the USA (Lawrence et al., 1989). There is a strong correlation between the levels of matrix metalloproteinase (MMP) expression and secretion from the intra-articular cell lining of the joint, and the initiation and progression of the extracellular matrix degeneration seen in human arthritic disorders, such as osteoarthritis and rheumatoid arthritis. In addition, MMPs are directly implicated in tissue remodeling and tumor metastasis in the angiogenic phase of tumor growth. For example, MMP2 is shown to trigger the migration of breast epithelial cells via specific cleavage of laminin-5 (Ln-5) (Brooks et al., 1996) (Giannelli et al., 1997). The cleaved form of Ln-5 is commonly discovered in tumors and in tissues undergoing remodeling, but not in quiescent tissues. Therefore, MMPs are very attractive therapeutic targets whose abnormal levels of expression can potentially be controlled via genetic and non-genetic interventions. Besides T cells and macrophages, synovial fibroblasts and chondrocytes are directly involved in initiating the arthritic disease, primarily by overexpression and secretion of proteases into the intra-articular region of the joint. Cytokines secreted by T cells and macrophages are known to induce synovial fibroblasts to express and secret augmented levels of MMPs into the joint cavity (Burger et al., 1998). Natural inhibitors of MMPs, such as tissue inhibitor of metalloproteases (TIMPs), normally function to keep the activities of these proteases within physiological homeostasis during the course of wound healing, tissue remodeling, and embryogenesis (Nagase, 1996) (Brown, 1997). However, the perturbation of this balance in favor of increased MMPs expression and secretion would eventually lead to extracellular matrix destruction associated with arthritis and tumor metastasis. Because arthritic disorders, for the most part, are chronic in nature, a long-term therapeutic genetic intervention may be essential to halt the disease process.

AAV vectors are attractive in this regard because they are nonpathogenic parvoviruses of human origin, which integrate into chromosomal DNA of a host cell and stably express the therapeutic gene during the entire life span of transduced cells. Second, AAV vectors are less dependent on the proliferative nature of the target cells for efficient transduction. This is supported by many reports indicating that these vectors efficiently infect dividing and non-dividing cells (Kaplitt et al., 1994) (Ponnazhagan et al., 1997) (Koeberl et al., 1997). This is critical for gene delivery to synovial fibroblasts and chondrocytes, which are quiescent cells in vivo.

In this report, rAAV-mediated expression of b-galactosidase was readily detectable in ex vivo infected cultures of RSFs and chondrocytes. More importantly, stromelysin-specific ribozymes driven by U1-, U6,- and tRNA-based promoters were efficiently expressed in RSFs. Such expression cannot be assumed as we have previously observed that a similar U1-ribozyme transcription unit was not expressed from a recombinant adenovirus in primary human or rat smooth muscle cells (Macejak et al., 1999). Because the ribozyme transcripts are small RNA molecules (relative to most mRNAs) we were able to express more than one transcription unit from a single rAAV vector. Although the three ribozyme transcription units used here are thought to direct nuclear localization (Bertrand et al., 1997) (Thompson et al., 1995a), their subnuclear localization may be unique. The U1-ribozyme RNA was extremely stable (Figure 3C), contained a trimethyl-G “cap” and was bound by SM proteins, consistent with this ribozyme transcript being part of a snRNP. Expression of multiple ribozyme transcription units from one vector may enable the ribozyme transcripts to attack the target RNA at different points along its maturation/transport pathway. Consistent with this hypothesis we observed an additive effect with the U1- and U6-derived ribozymes expressed in combination compared to their sole expression (Figure 5C). The situation does not appear to be that simple, however, since expression of all three ribozyme transcription units in combination did not generate the most reduction in target RNA. This may be due in part to the lower level of expression of all three ribozyme transcripts compared to their individual expression levels (Figure 5B).

Interestingly, the levels of U1- and U6-derived ribozyme increased in rAAV-infected RSFs upon passaging (Figure 4). There are several possible explanations for this observation. The first is related to the fact that the conversion of ss-DNA of rAAV genome to a double stranded form is essential for the expression of a transgene. This rate-limiting step is reported to be more efficient in dividing cells as compared to growth-arrested cells. Second, genomic integration of rAAV may enhance transgene expression and the formation of double-stranded AAV geno