Foamy Virus as a Gene Transfer Vector to the Central Nervous System

Abstract

Engineered Foamy virus vectors (FV) have been lauded for their superior safety profiles and stable integration patterns compared to their gammaretroviral counterparts. The drawback, however, has been that FV incorporation is allegedly cell cycle dependent, which would limit its utility in post-mitotic tissues such as the central nervous system. In this brief communication, we challenged this notion and tested Foamy virus in vivo. We injected equal titers of FV and Lentivirus (LV) into the adult rat brain and found that at 1 week, FV transduced a significantly greater volume of BrdU negative brain parenchyma than did LV. By 8 weeks, however, the volume of transduced tissue was greatly reduced—comparable to LV—and restricted to BrdU+. Taken together, these data implicate a role for FV in short-term gene delivery strategies to the CNS.

Gene therapy is a promising therapeutic technique for disorders of the central nervous system (CNS). However, effectiveness in clinical trials has been limited due to physiologic barriers unique to this organ system as well as the post-mitotic state of many of the cellular targets in the brain and spinal cord. Currently the HIV-derived lentivirus (LV) and adeno-associated virus (AAV) represent the most widely used viral transfer vectors for CNS gene delivery. The former is limited by relatively low titers, gene capacity and safety concerns; the latter is limited by short-term expression of transgenes. A recently described potential alternative to these vectors is Foamy Virus (FV) ¹. FV constitute a subfamily (Spumavirinae) of Retroviridae that are endemic in some mammalian species including chimpanezee and cats, but not in human populations^2,3. To date, exposure and subsequent infection of humans with FV has not been associated with any pathology and human-to-human transmission has yet to be demonstrated. Moreover, long-term studies of infected animal care workers have failed to demonstrate negative consequences⁴. In fact, FV does not appear to be pathogenic even in its natural mammalian hosts and infection is not associated with an increased risk of developing malignancies⁵.

In addition to its safety benefits, FV possesses several novel properties and characteristics essential to effective gene delivery. The FV life cycle is unique in that single-stranded viral RNA is converted to double-stranded DNA (DS DNA) within virion-producing cells⁶ unlike LV and Moloney Leukemia Virus (MLV) where reverse transcription to DS DNA occurs in target cells. This is significant in that it affords FV improved stability relative to LV and MLV since it is able to transcribe DNA independent of target cell nucleotide pools and can then use DNA as the functional genome. This aspect also allows the FV genome to be sustained as a pre-integration complex that can enter the nucleus in the absence of mitosis. Once cell division commences, it can integrate into the genome and the transgene can be expressed⁷.

In general, the ability of a virus to integrate into genomic DNA offers the possibility of long-term transgene expression, but with the associated risk of malignant transformation resulting from insertional mutagenesis and oncogene activation. The potential to produce cell transformation by this mechanism was unfortunately realized in a gene therapy clinical trial in which several patients developed leukemia after retroviral integration near the LMO2 protooncogene⁸. This negative potential is highly dependent upon the location of viral integration into the host genome. Specifically, it has been determined that MLV preferentially integrates into active promoter regions and LV prefers integration into actively transcribed DNA^{9, 10}. However, FV has a unique integration pattern that favors non-transcribed areas of chromosomal DNA, which increases its appeal as a safe gene transfer vehicle¹¹.

In addition to its attractive integration pattern and safety gains, FV vectors have demonstrated a broad host range, large packaging capacity and the ability to concentrate stable virions via ultracentrifugation¹. For example, the FV system has been shown to effectively transduce human fibroblasts¹¹, human hematopoeitic stem cells¹² and rat hippocampal neurons in vitro¹³ as well as cells derived from other various mammalian species. The FV system also appears to transduce multiple different cell types with equal efficiency suggesting that a receptor mediating infection may be broadly expressed in many tissue types^{14, 15}.

Although FV appears to have many advantages for gene transfer, controversy still exists regarding the cell-cycle dependence of FV genomic integration, which might undermine its potential as an effective vehicle for delivery to post-mitotic tissues such as the CNS. Trobridge and Russell (2004) reported that FV-mediated transgene expression is dependent upon cell mitosis but not DNA synthesis. They also suggested that FV vectors may only work in quiescent stem cells that later divide and thereby permit integration¹⁶. By implication this would suggest that FV vectors may not be suited for the CNS in which the majority of the cells are post-mitotic. In this study we produced replication incompetent lentivirus and foamy virus and applied both, at equal titers, to cells in vitro and then directly in the central nervous system of adult rats to compare the transduction efficiency of the two vectors in vivo.

To assess and compare the efficacy of cell transduction in vitro, three distinct cell types were tested: 293T cells, rat-derived primary CNS cortical cultures, and pan-purified A2B5+ oligodendrocyte progenitor cells (Figure 1a). Cells were cultured with virus for 24 hours and the percentage of EGFP+ cells was calculated 72 hours later. FV readily transduced dividing cells and primary cell cultures in vitro with a similar efficiency to lentivirus (Figure 1b). However, FV failed to transduce A2B5+ progenitor cells in vitro, in contrast to LV, which readily transduced these cells albeit with lesser efficiency when compared to primary cultures and cell lines. These differences are likely related to the rate at which different cultures actively divide and the apparent inability of FV to incorporate into cells in G₀ or diminished efficiency in cells that divide slowly.

Figure 1. Lentivirus transduces primary purified cell cultures more efficiently than foamy virus.

Replication incompetent FV and LV were produced according to published protocols used to generate high titer LV¹⁵. Viral titers were determined on 293T cells and FV was subsequently diluted 10-fold in all experiments to allow for equivalent titers to be compared. Four plasmids¹, (kind gift of Dr. Russell) one of which was modified to include a MND promoter sequence driving EGFP expression (A), were used to transduce 293T cells in twenty 75 cm² flasks. Virus was concentrated by centrifugation at 50,000g and resuspended in 100 μl of serum free DMEM. A 5 μl aliquot of virus was thawed and mixed with 1 ul of protamine sulfate (4μg/ml) and incubated for 20 minutes at room temperature (RT). Virus was then added to fresh culture media and applied to cells. Forty-eight hours later media was changed and the percentage of green cells/visual field was assessed using an inverted fluorescent microscope. Although FV efficiently transduced both 293T cells and primary cortical-derived rat brain cultures, it failed to transduce A2B5+ cells in culture and overall was less efficient in primary cell cultures than LV. Transduction studies were conducted in triplicate and mean values are represented here with ± one standard deviation.

To assess and compare the efficacy of cell transduction in vivo, FV and LV were injected into the corpus callosum of adult rats in accordance with the NIH Animal Protection Guidelines which were approved by Institutional Animal Care and Utilization Committee of Case Western Reserve University School of Medicine. At one week post injection, FV transduced a greater volume of brain parenchyma relative to LV, as measured by EGFP+ cell volume (Figure 2, top 2 panels). The volume transduced by FV injection at one week was 20.5 mm³ and by LV was 1.17 mm³. However, at 8 weeks post-injection the volume of brain transduced by FV was 0.521 mm³ and by LV was 0.367 mm³ (Figure 2, bottom 2 panels). The extensive loss in EGFP+ tissue volume over time in FV transduced rat brain suggests a transient phase of virus entry and gene expression in the absence of permanent integration into all cells. Although LV transduced tissue volume was also reduced at 8 weeks, LV integrated into the genome more effectively, maintaining 31.6% of the initially transduced volume compared to FV which only maintained 2.5%. These data bring to light an important caveat regarding long-term gene delivery using FV: despite the presence of stable pre-integration complexes within transduced cell nuclei, long-term gene expression is feasible only in tissues that are actively dividing.

Figure 2. Compared with lentivirus, foamy virus is able to transiently transduce large volumes of brain parenchyma, however, with less efficient integration and subsequent long-term transgene expression.

Access to brain parenchyma was accomplished by placement of bilateral burr holes through the skull, 0.0 mm anterior and −2.0 mm and +2.0 lateral to bregma with injection occurring at a depth of 3.4 mm. 5 ul of virus was injected into the left corpus callosum at a rate of 0.25 μl/min followed by removal of the needle at 1mm/min. Immediately following removal, injection was carried out into the contralateral corpus callosum in an identical manner. Rats were sacrificed at one week and eight weeks and perfused with approximately 150 ml of 0.9% saline over 20 minutes. Immediately following saline treatment, animals were perfused with approximately 150 ml of 4% PFA for 20 minutes. The entire brain was removed and post-fix immersed in 4% PFA overnight at 4o C and then cryoprotected in sucrose for 24 hours. Finally, the brain was embedded in O.C.T and sectioned. Every thin section with EGFP+ cells was photographed, the gross anatomy outlined (including the EGFP-positive region), and the infected tissue was reconstructed using the FDA approved 3D-Doctor software (Able Software Corp., Lexington, MA). In addition to being able to visualize the transduced (EGFP+) region in three dimensions, the software also provided quantitative transduction volumes. The calculated brain volume transduced by FV at one week post injection was 20.5 mm³ compared to LV which transduced 1.17 mm³. At eight weeks post injection only 0.521 mm³ was transduced by FV and 0.367 mm³ by LV (n=4).

In order to investigate the relationship between FV-mediated transduction and the cell cycle, BrdU was injected (100mg/kg) intraperitoneally daily for 5 days after injection of virus into brain parenchyma. Rats were again sacrificed at 1 and 8 weeks, the EGFP+ volume was calculated and immunohistochemical analysis conducted using anti-BrdU antibody (Roche). At 1 week, EGFP+ signal was identified in both BrdU+ and BrdU-cells in the brain after both FV injection and LV injection (Figure 3, top 2 panels). However, at 8 weeks post-injection of FV, the overlap of BrdU+ cells with EGFP+ cells was distinctly different (Figure 3, bottom 2 panels). Namely, the EGFP+ signal was confined to BrdU+ cells, suggesting a cell cycle dependence of FV for long-term gene expression. A restriction to proliferating cells is consistent with the extensive loss of FV-infected volume at 8 weeks and supports a transient phase of transduction and transgene expression that is highly efficient but temporally limited. In contrast, at 8 weeks LV facilitated long-term transgene expression in cells that were both BrdU positive and negative. This supports the well documented ability of LV to maintain expression in post-mitotic cells for a sustained period of time.

Figure 3. Integration and long-term transgene expression by foamy virus is cell cycle-dependent.

Rats underwent virus injected as described in figure legend 2 after which BrdU was injected (100mg/kg) intraperitoneally daily for 5 days. In order to visualize dividing cells, immunohistochemistry was performed on brain cryosections. Tissue was incubated for 48 hours in mouse anti-BrdU antibody (Roche Inc.), followed by a 1 hour incubation in goat anti-mouse Alexa 594 (Invitrogen). At one week post-injection, both LV and FV demonstrated transgene expression in both dividing and non-dividing cells. However, by eight weeks FV-mediated transgene expression was limited to dividing cells whereas LV was able integrate and express the transgene in both dividing and non dividing cells.

Multiple reports have demonstrated the ability of FV to efficiently transduce CNS-derived cells in vitro ^{19, 13}, but to our knowledge this is the first report demonstrating in vivo applications of FV to the CNS. In order to fully exploit the advantageous characteristics of FV, however, it is necessary to first define the pattern of transduction, gene expression and integration of FV into the CNS as we have done in this brief communication. Herein we have shown that high-titer FV efficiently transduces brain parenchyma at one week post-injection. This transduction and gene expression is cell cycle independent as indicated by discordant BrdU+ signal and EGFP signal. A high level of FV transduction suggests that it might be a very useful mechanism to facilitate widespread gene expression over a short time period. At 8 weeks the differences in transduced brain volume between FV and LV are not statistically significant. This indicates that despite its initial gains, over time, integration and transgene expression by FV is at best equal to LV. Nevertheless, this characteristic holds considerable promise for selectively examining clonal relationships between neural cells and targeting transduction to particular cell populations through timing and localization of virus administration. Ultimately we would propose using this knowledge in order to generate a Lenti-Foamy hybrid virus which incorporates the highly efficient transient transduction ability and chromosomal integration pattern of FV with the ability of LV to efficiently and permanently integrate into post-mitotic tissue.