**The Influence of Envelope C-Terminus Amino Acid Composition on the Ratio of Cell-Free to Cell-Cell Transmission for Bovine Foamy Virus**

#### **Suzhen Zhang †, Xiaojuan Liu †, Zhibin Liang, Tiejun Bing, Wentao Qiao and Juan Tan \***

Key Laboratory of Molecular Microbiology and Technology, Ministry of Education, College of Life Sciences, Nankai University, Tianjin 300071, China; zhangsuzhen819819@163.com (S.Z.); 1395312330@qq.com (X.L.); liangzhibin@mail.nankai.edu.cn (Z.L.); btj1987@126.com (T.B.); wentaoqiao@nankai.edu.cn (W.Q.)

**\*** Correspondence: juantan@nankai.edu.cn; Tel.: +86-22-2350-4547; Fax: 86-22-2350-0950

† These authors contributed equally to this work.

Received: 12 November 2018; Accepted: 29 January 2019; Published: 31 January 2019

**Abstract:** Foamy viruses (FVs) have extensive cell tropism in vitro, special replication features, and no clinical pathogenicity in naturally or experimentally infected animals, which distinguish them from orthoretroviruses. Among FVs, bovine foamy virus (BFV) has undetectable or extremely low levels of cell-free transmission in the supernatants of infected cells and mainly spreads by cell-to-cell transmission, which deters its use as a gene transfer vector. Here, using an in vitro virus evolution system, we successfully isolated high-titer cell-free BFV strains from the original cell-to-cell transmissible BFV3026 strain and further constructed an infectious cell-free BFV clone called pBS-BFV-Z1. Following sequence alignment with a cell-associated clone pBS-BFV-B, we identified a number of changes in the genome of pBS-BFV-Z1. Extensive mutagenesis analysis revealed that the C-terminus of envelope protein, especially the K898 residue, controls BFV cell-free transmission by enhancing cell-free virus entry but not the virus release capacity. Taken together, our data show the genetic determinants that regulate cell-to-cell and cell-free transmission of BFV.

**Keywords:** bovine foamy virus; infectious clone; particle release; cell-free transmission

#### **1. Introduction**

Foamy viruses (FVs), also known as spumaviruses, are a group of *Retroviridae* with unique features that differentiate them from orthoretroviruses. FVs infect humans [1,2] and other mammals, including bovines [3], simians [4], felines [5], and equines [6]. However, FV infection does not cause any clinical symptoms in its natural hosts, despite the significant cytopathic effect it causes in fibroblasts or fibroblast-derived cell lines as well as in epithelial cells, such as baby hamster kidney (BHK) cells [7,8].

Viruses have two major transmission strategies: cell-free transmission, involving the release of virus particles into the extracellular space, and cell-to-cell transmission [9–11]. Retroviruses exhibit different degrees of cell-free and cell-to-cell transmission. Unlike most other retroviruses, such as the human immunodeficiency virus (HIV) [12–16], murine leukemia virus (MLV), feline foamy virus (FFV), prototype foamy virus (PFV), and simian foamy virus (SFV), which transmit through both cell-to-cell and cell-free pathways, bovine foamy virus (BFV) infection is tightly cell-associated [17,18].

In contrast to other retroviruses, the envelope (Env) protein of PFV plays an important function in the budding and release of PFV particles [19]. In particular, the leader peptide (LP) in the N-terminal region of PFV Env is essential for virus budding. In LP, the three lysine residues (K14, K15, and K18) undergo ubiquitination, which regulates PFV release [20].

The Env protein determines FV's wide host range [1–6]. The cellular receptor of FVs has not been determined; however, it was reported that heparin sulfate might act as an attachment factor

facilitating PFV and SFV entry [21,22]. Different from orthoretroviruses, the assembly and budding of FV particles require direct and specific interaction between the N-terminus of Gag and the Env leader protein Elp [23,24]. FV Gag, lacking the myristoylation membrane targeting signal, cannot produce cell-free Gag-only virus-like particles [18,24,25]. Instead, co-expression of FV Gag and Env leads to the generation of Env-dependent sub-viral particles (SVPs), which sets FVs apart from orthoretroviruses [23,24,26–28].

Bao and colleagues selected high-titer (HT) cell-free BFV-Riems isolates using the in vitro evolution procedure [18]. Yet, they did not generate infectious viral DNA clones and did not explore the molecular mechanisms that have enabled BFV cell-free transmission. Using the BFV strain BFV3026, which we isolated in 1996, we generated an infectious clone called pBS-BFV-B [29]. BFV-B is deficient in cell-free transmission, which does not allow for the development of a BFV vector. We have now screened for BFV variants with enhanced cell-free transmission in BICL cells (derived from BHK-21 cells) by serial virus passaging and successfully created a BFV infectious clone—called pBS-BFV-Z1—with cell-free transmission ability. Through sequence alignment and mutagenesis, we determined the C-terminal region of Env as one determinant for BFV cell-free transmission, and thus uncovered the molecular mechanism by which BFV spreads via cell-free transmission.

#### **2. Materials and Methods**

#### *2.1. Cell Lines and Viruses*

BHK-21, Cf2Th, HEK293T, BFVL (BHK21-derived indicator cells containing a *luciferase* gene under the control of the BFV LTR) [30], and BICL (BHK21-derived indicator cells containing an enhanced green fluorescent protein under the control of the BFV LTR) cells [31,32] were maintained in Dulbecco's modified Eagle's medium (Thermo Fisher, Waltham, MA, USA) containing 10% fetal bovine serum (GE Healthcare, Cincinnati, OH, USA) and 1% penicillin-streptomycin (Thermo Fisher, Waltham, MA, USA) at 37 ◦C in a 5% CO2 atmosphere. BFV3026 was stored in our lab and cultured with Cf2Th and BICL cells. No mycoplasma and viruses contamination were detected in any cells we used.

#### *2.2. Plasmids and Transfection*

BFV3026 full-length genomic DNA clone pBS-BFV-B was generated by amplifying viral DNA extracted from BFV3026-infected Cf2Th cells. The BFV infectious clone pBS-BFV-Z1 was constructed using the same methods of pBS-BFV-B as previously reported [29]. Chimeric BFV clones between clone B and Z1 were generated by shared different restriction sites. Mutations were generated using site-direct PCR mutagenesis (Toyobo, Osaka, Japan), and all mutations were verified by DNA sequencing (Genewiz, Beijing, China). The plasmids expressing Env and Gag were constructed by inserting the coding sequences of BFV Env and Gag into the indicated vectors, including pCMV-3HA and pCE-puro-3Flag. HEK293T and BHK-21 cells were transfected using polyethylenimine (PEI) (Polysciences, Warrington, PA, USA) according to the manufacturer's protocol [33]. BHK-21 cells (2 × <sup>10</sup>5) were seeded in six-well plates. Twenty-four hours later, 2 <sup>μ</sup>g empty vector pBS, pBS-BFV-B, pBS-BFV-Z1, or chimeric infectious clones were transfected into BHK-21 cells. Eight μL PEI (1 mg/mL) was added at DNA:PEI (μg:μg) ratios of 1:4 and incubated for 10 min at room temperature. Cells were harvested 48 h after transfection. Cf2Th cells were transfected using Lipofectamine 2000 (Thermo Fisher, Waltham, MA, USA).

#### *2.3. Titration of Cell-Free BFV3026*

BFV3026 was isolated from lymphocytes of peripheral blood of cattle by our lab in 1996 and cultured with Cf2Th cells. The parental cell-free BFV3026 virus was obtained by freezing-thawing highly infected Cf2Th cells. The later serial passage screening was carried in BICL cell lines. BFV was adapted to cell-free transmission in BICL cells by serial passaging using cell-free culture supernatants as previously reported [18]. In the case of viral infection, BICL cells expressed GFP and green signals were visible under a fluorescent microscope.

BFV cell-free titers were determined by infecting BICL indicator cells with gradient diluted BFV-containing supernatants. BICL cells were seeded (2 × <sup>10</sup><sup>4</sup> cells/well) in 48-well plates. Twenty-four hours later, BFV-containing supernatants were serially diluted 1:5 (triplicate for each dilution) and added to BICL cells. Six days post-infection, BFV titers were determined by fluorescence microscopy for GFP signals in BICL cells. If at least two of the three wells were positive for GFP, that dilution was considered positive for infection.

#### *2.4. Cell-Free Infection*

BHK-21 cells were transfected with indicated infectious clones for 2, 4, and 6 days. Then, cell culture supernatants were collected and filtered with a 0.45 μm membrane. BFVL cells in 12-well plates (2.5 × 105 cells/well) were then infected with equal volumes of these filtered supernatants for 48 h and subjected to luciferase assay. The infectivity of the cell-free virus was determined by luciferase activity.

#### *2.5. Cell Co-Culture Assay*

BHK-21 cells were transfected with indicated infectious clones for 2, 4, and 6 days. Then, the 5% of cells were harvested and co-cultured with 1.5 × <sup>10</sup><sup>5</sup> BFVL cells in 12-well plates. Forty-eight hours post co-culture, luciferase activity was measured to quantitate virus titer.

#### *2.6. Luciferase Reporter Assay*

Forty-eight hours after infection or co-culture, the BFVL cells were harvested in lysis buffer. Then, luciferase assays were performed using the luciferase reporter assay system (Promega, Madison, WI, USA). In addition to normalizing the luciferase activities, we also detected the GAPDH expression of cell lysates by western blotting. All of the results from the luciferase reporter assay were the averages of three independent experiments.

#### *2.7. Hirt DNA Extraction*

BICL cells infected with cell-free BFV p43 (passage 43) were harvested, washed with 1.5 mL 1× PBS, and lysed with 250 μL buffer K (20 mM HEPES, 140 mM KCl, 5 mM MgCl2, 1 mM Dithiothreitol) and 7.5 μL 0.5% Triton X-100 for 10 min at room temperature. Following centrifugation (10 min at 1500× *g*), the pellet was dissolved in 400 μL TE and 90 μL 5 M NaCl at 4 ◦C overnight. The supernatant post centrifugation was extracted with phenol chloroform: isoamylalcohol (24:1), then precipitated for DNA in ethanol with 0.3 M NaAc at −20 ◦C for 1 h. The purified DNA was then washed with 70% ethanol, dissolved in 20 μL TE, and stored at −20 ◦C.

#### *2.8. Enrichment of Wt and Sub-Viral BFV Particles*

Six milliliters of cell culture supernatants containing BFV particles or SVPs (including Env-only and Gag-Env SVPs) were filtered through 0.45 μm membranes and layered on a 1 mL cushion of 20% sucrose in PBS (*w*/*v*). After ultracentrifugation (Optima LE-80K, Beckman Coulter) at 4 ◦C for 1.5 h at 210,053× *g*, the non-visible pellet was re-suspended in 30 μL loading buffer containing 2% SDS and subjected to western blotting.

#### *2.9. Co-Immunoprecipitation*

HEK293T cells transfected with pCMV-3HA-Gag and pCMV-3HA-Env were lysed in lysis buffer (50 mM Tris, pH 7.4; 150 mM NaCl; 2 mM EDTA; 3% Glycerol; 1% NP-40; complete, EDTA-free protease inhibitor cocktail tablets). The cell lysates were sonicated and centrifuged at 13,000× *g* for 10 min at 4 ◦C. Following centrifugation, supernatants were incubated with mouse anti-Gag antibody for 2 h at 4 ◦C and rotated with Protein A-agarose (Merck Millipore, Darmstadt, Germany) for 3 h or overnight at 4 ◦C. After six washes with lysis buffer, the immunoprecipitated materials were boiled in 40 μL of 2× SDS loading buffer and subjected to western blotting using rabbit anti-HA antibody.

#### *2.10. Immunofluorescent Assay*

BHK-21 cells seeded in 12-well plates containing coverslips were infected with infectious clones. After 48 h, BHK-21 cells were fixed with 500 μL of 4% formaldehyde in PBS for 10 min on a shaker. Cell membranes were permeabilized with 500 μL of penetration solution (0.1% Triton X-100 in 4% formaldehyde) for 10 min. Then, cells were blocked in 500 μL of blocking solution (50% serum, 6% non-fat milk power, 15% BSA, 5% NaN3, 20% Triton in PBS) for 2 h at room temperature or 4 ◦C overnight. Cells were then incubated in diluted primary antibody for 2 h at room temperature. The coverslips were washed in the 12-well plates with PBST and incubated in diluted fluorochrome-conjugated secondary antibody for 40 min in the dark. After 5 washes, cells were incubated in 500 μL of DAPI for 10 min in the dark. The coverslips were mounted onto glass slides and allowed to air dry for 1 h. For long-term storage, slides were stored flat at 4 ◦C protected from light.

#### *2.11. Western Blotting*

Proteins were separated by SDS-PAGE (polyacrylamide gel electrophoresis). Then, the separated proteins were transferred onto a polyvinylidene difluoride (PVDF) membrane (GE Healthcare, Cincinnati, OH, USA) by electroblotting for 1 h at 100 V in 4 ◦C. Following incubation in PBS containing 5% nonfat milk for 45 min at room temperature, the PVDF membranes were subsequently incubated with primary antibody for 90 min. Then, the membranes were incubated with secondary antibodies—goat anti-mouse or goat anti-rabbit IgG conjugated to horseradish peroxidase, for an additional 45 min. Protein bands were detected by chemiluminescence (Merck Millipore, Darmstadt, Germany).

#### *2.12. Statistical Analysis*

Data were expressed as the mean ± SD of the results of three independent experiments in which each assay was performed in triplicate. Data were compared using the unpaired two-tailed *t* test. A *p* value of <0.05 was considered significant (\* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001).

#### **3. Results**

#### *3.1. Construction of a Cell-Free BFV Infectious Clone pBS-BFV-Z1*

To isolate a high-titer cell-free BFV stain, we performed in vitro evolution experiments in BICL cells, which express GFP upon BFV infection. The original virus strain used for selection was Cf2Th-associated BFV3026, which was isolated in China. After 53 passages, the cell-free infectivity of BFV reached a plateau of 105 IU/mL (Figure 1A). The full-length viral genomic DNA was amplified from the hirt DNA extracted from BICL cells infected with cell-free BFV (passage 43) and then inserted into the pBluescript SK-(pBS) vector. The constructed infectious viral DNA clone was named pBS-BFV-Z1.

We next characterized the pBS-BFV-Z1 clone for its ability to produce infectious cell-free particles by harvesting viruses in the supernatants of BHK-21 cells that were transfected with BFV infectious clones. As shown in Figure 1B, the BFV particle release was observed in Z1-transfected cells but not in cells transfected with the cell-associated clone, pBS-BFV-B. At the same time, we also analyzed the intracellular Gag expression levels and observed that Gag expression in BFV-B was less than BFV-Z1 (Figure 1B). Furthermore, we observed that the replication capacity of the Z1 clone was 40 times greater than that of the B clone, as measured by co-culture assay (Figure 1C). Notably, the cell-free Z1 virus particles were infectious (Figure 1C) and spread in a long-term infection (Figure 1D). Together, these

data demonstrate that the pBS-BFV-Z1 clone is highly infectious and produces infectious cell-free virus particles.

**Figure 1.** The cell-free and cell-to-cell transmission activity of BFV-Z1. (**A**) Screen of the high-titer cell-free bovine foamy virus (BFV) strain in BICL cells with cleared supernatants. Around 53 passages, BFV cell-free titers were determined by gradient dilution. BICL cells were seeded in 48-well plates (2 <sup>×</sup> 104 cells/well), then BFV-containing supernatants were serially diluted 1:5 in the plate (triplication for each dilution). Four to six days post-infection, the cell-free BFV titers were calculate and analyzed, as determined by fluorescence microscopy for GFP signals in BICL cells. (**B**–**C**) BHK-21 (baby hamster kidney) cells (2 <sup>×</sup> 105) were seeded in 6-well plates. After 24 h, 2 <sup>μ</sup>g empty vector pBS, pBS-BFV-B, or pBS-BFV-Z1 were transfected into BHK-21 cells. Four days later, 5% cells co-cultured with 1.5 <sup>×</sup> 105 BFVL cells and part of supernatants infected 2.5 <sup>×</sup> <sup>10</sup><sup>5</sup> BFVL cells, respectively, for cell-to-cell and cell-free transmission activity using a luciferase activity assay after 48 h, and the results were the averages of three independent experiments and data were analyzed using GraphPad Prism software (compared with BFV-B, \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001). The remainder of the supernatants and cells transfected for 2 days were harvested for western blot with the indicated antibodies (B). (**D**) Part of the supernatants were collected to infect fresh BICL cells (marked P1) prior to ultracentrifugation, and the expression of GFP was observed. If more than 80% BICL cells were positive for green fluorescence, then the supernatants were collected and cleared to infect fresh BICL cells (marked P2).

#### *3.2. The C Terminus of Env Determines the Ability of BFV to Generate Infectious Cell-Free Particles*

To identify which regions in the sequence of pBS-BFV-Z1 changed and enabled the production of infectious cell-free virus particles, we aligned the sequences of the two BFV clones, B and Z1. Changes were identified at multiple nucleotide positions (Tables S1 and S2). We then swapped the sequences between the B and Z1 clones and generated six chimeric BFV clones (*Eco*R I within *pol* gene, and *Nde* I within *env*) (Figure 2A). As shown in Figure 2, clone S1S2 and clone Z1 produced similar levels of intracellular Gag (Figure 2B) as well as comparable cell-to-cell transmission activity (Figure 2D). Unfortunately, none of these chimeric clones produced infectious cell-free BFV particles (Figure 2C). Notably, the expression of Gag in BFV-B could be detected for two days (Figure 1B) but not for four days that was passaged one time with a low ratio of transfected cells (Figure 2B). One possibility is that the activity of clone B is much weaker than clone Z1 (Figure 1C).

**Figure 2.** Activities of chimeric BFV infectious clones between pBS-BFV-B and pBS-BFV-Z1. (**A**) The structure model of BFV proviral genome is shown on the top, followed with a schematic overview of the chimeric virus between BFV-B and BFV-Z1. The gray box represents the BFV3026 proviral genomic DNA of pBS-BFV-B, and the black box represents pBS-BFV-Z1. The full-length viral DNA was divided into three segments based on the single restriction enzyme sites of *Eco*R I and *Nde* I, and the new chimeric clones were generated by exchanging the fragments between pBS-BFV-B and pBS-BFV-Z1 plasmids. (**B**–**D**) BHK-21 (2 <sup>×</sup> <sup>10</sup>5) cells were transfected with 2 <sup>μ</sup>g empty vector pBS, pBS-BFV-B, pBS-BFV-Z1, or chimeric infectious clones as indicated for 2, 4, or 8 days. Cells transfected for 4 days were subjected to western blotting with indicated antibodies (**B**). The activity of cell-free and cell-to-cell transmission was measured by a luciferase activity assay at the indicated times (**C**,**D**), and the data were the averages of three independent experiments. Compared with BFV-B, \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001.

Previous studies have shown that Env protein is essential for PFV release and entry [26,34,35]. Clone S1S2 contains a chimeric *env* gene, including the N terminus of EnvZ1 and the C terminus of EnvB, but lacks cell-free transmission ability. To test the role of Env in the ability of the Z1 clone to produce infectious cell-free virus particles, we engineered another two chimeric clones, Z1(1–9554) and Z1(EnvcB). Clone Z1(1–9554) contained an entire *env* gene from clone Z1, whereas clone Z1(EnvcB) had the clone Z1 sequence except for the C-terminal region of Env, which was derived from clone B (Figure 3A). The intracellular Gag expression from Z1(1–9554) and Z1(EnvcB) was similar to that from clones Z1 and S1S2 (Figure 3B). Excitingly, the Z1(1-9554) clone produced high levels of infectious cell-free virus particles, although still two to three-fold lower than the parental clone Z (Figure 3C). In contrast, the cell-to-cell transmission of Z1(EnvcB) was two-fold more efficient than that of clone Z1, as measured by co-culture assay (Figure 3D). However, compared with clone Z1, chimera Z1(EnvcB) was unable to transmit by cell-free particles, which suggests that replacement of the C-terminal

Env with that from clone B abrogated the cell-free infectivity of clone Z1 (Figure 3C). Furthermore, Z1(1–9554) exhibited cell-free transmission activity, although to a lesser degree than Z1 (Figure 3C). Overall, these results indicate that the C-terminus of Env is crucial for BFV cell-free infectivity.

**Figure 3.** The C terminus of Env determines the ability of BFV to generate infectious cell-free particles. (**A**) We constructed clone BFV-Z1(1-9554) by the single restriction enzyme sites of *Nhe* I, which cuts downstream of Env in the BFV proviral genome, and clone BFV-Z1(EnvcB) was constructed by *Nde* I and *Nhe* I. The gray box represents the BFV3026 proviral genomic DNA of pBS-BFV-B, and the black box represents pBS-BFV-Z1. (**B**–**D**) BHK-21 cells (2 <sup>×</sup> 105) were transfected with 2 <sup>μ</sup>g empty vector pBS, pBS-BFV-B, pBS-BFV-Z1, pBS-BFV-Z1(1-9554), pBS-BFV-S1S2, or pBS-BFV-Z1(EnvcB) for 4 days, and cells were subjected to western blotting with indicated antibodies (**B**). The cell-free and cell-to-cell transmission activity was measured with the indicated assays (**C**,**D**), and the data were the averages of three independent experiments.

To identify the specific sites that affect the cell-free transmission ability of clone Z1, we compared the amino acid sequence of the C-terminal region of Env between clone Z1 and clone B and found differences at five sites: H816Y, P823S, E898K, R976K, and L978S (Figure 4A), which are all located in the C-terminus of transmembrane (TM) protein gp48 (aa 572–990). Using chimera S1S2 that bears the C-terminal region of the clone B Env, we mutated the amino acids at the above five positions to the counterparts in clone Z1. Our results showed that two single point mutations—H816Y and E898K, especially E898K—rendered clone S1S2 to produce infectious cell-free virus particles. Furthermore, the cell-free infectivity of clone S1S2-898/976/978, which had the amino acids at position 898, 976, and 978 of Z1 Env, was similar to that of Z1(1–9554) (Figure 4B). We also mutated the above five positions in clone Z1 Env to the respective amino residues in clone B Env. The K898E mutation markedly impaired the cell-free infectivity of Z1, and mutations Y816H and K898E together abrogated the cell-free infectivity of Z1 (Figure 4D). These results indicate that amino acid K898 together with Y816 in the Env protein are major determinants of BFV cell-free transmission.

**Figure 4.** K898 in the *env* gene of BFV-Z1 is crucial for BFV cell-free transmission. (**A**) Amino acid site mutations, marked in red, in the C terminus of Env by sequence alignment analyses are shown. (**B**,**C**) Construction of infectious clones with point mutations on the basis of pBS-BFV-S1S2 from B to Z1 and their transfection to BHK-21 cells. The transmission activity was analyzed by measuring luciferase activity, and the data were the averages of three independent experiments. Compared with BFV-S1S2, \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001. (**D**,**E**) Construction of clones with point mutations based on pBS-BFV-Z1 from Z1 to B. Then, these were transfected into BHK-21 cells and the transmission activity was analyzed. Compared with BFV-Z1, \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001, \*\*\*\* *p* < 0.0001. "•" stands for omission.

#### *3.3. The 14-AA Deletion in Gag Gene Increases BFV Cell-Free Transmission Activity*

Gag plays an indispensable role in the assembly and release of PFV particles. We found that Gag of the Z1 clone is 14 aa shorter than Gag of the B clone (Figure 5A). We tested whether this 14-aa deletion had an effect on BFV cell-free transmission by constructing two chimeric clones, Z1(Gag60) and Z1(1-9554; Gag60), which had the 14 aa inserted back in clones Z1 and Z1(1-9554). BICL cells were transfected with the above BFV clones and observed similar levels of Gag expression from all viral DNA clones (Figure 5B). However, the cell-free infectivity of Z1(Gag60) was four- to five-fold lower than that of Z1 (Figure 5C), although the cell-to-cell transmission activities of these two clones were comparable (Figure 5D). We further observed that Gag-B, Gag-Z1, and Gag60 were similarly distributed in the cytoplasm, suggesting that this 14-aa sequence does not affect the cellular localization of Gag (Figure 5E).

**Figure 5.** The 14 aa deletion in *gag* gene increases BFV cell-free transmission activity. (**A**) Partial results—sequence alignment of amino acids in Gag-B and Gag-Z1 is shown in the top left corner. Starting with the 176th amino acid, 14 consecutive amino acids are missing in Gag-Z1. (**B**) BHK-21 cells were transfected with empty vector pBS, pBS-BFV-B, pBS-BFV-Z1, pBS-BFV-Z1(1-9554), pBS-BFV-Z1(Gag60), or pBS-BFV-Z1(1-9554; Gag60), and partial lysates of cells after 4 d of transfection were tested by western blotting analysis. (**C**,**D**) The cell-free and cell-to-cell transmission activity was detected separately at the indicated transfection time, and the data were the averages of three independent experiments. (**E**) BHK-21 cells were transfected with indicated infectious clones, then subjected to immunofluorescent assays. Confocal microscopy was adopted for observation and image capture. "-" stands for missing.

Unlike orthoretroviruses, the release of FV virions requires both Gag and Env proteins [36]. FV can form Gag-Env SVPs or Env-only SVPs [26,34,37]. We therefore examined the ability of Env-B and Env-Z1 to mediate SVPs release. Different *env* and *gag* genes were cloned into the pCMV-3HA or the pCE-puro-3Flag vector. Gag plasmids alone or together with different Env plasmids were transfected into HEK293T cells. Two days post-transfection, virus particles in the culture supernatants were harvested through ultracentrifugation and analyzed by western blotting. As shown in Figure 6A, SVPs were detected when Gag was expressed together with Env but not when Gag was expressed alone. Similar levels of SVPs were produced by Env and Gag from both clone B and clone Z1. However, Env-Z1, but not Env-B, was completely processed in virus particles.

**Figure 6.** The interaction between Gag and Env. (**A**) HEK293T (4 <sup>×</sup> <sup>10</sup>6) cells were transfected with different eukaryotic expression plasmids, 3HA-Gag and 3HA-Env, for 2 days. Cells and supernatants were harvested separately, and the cell culture supernatants were cleared by ultracentrifugation following filtration with a 0.45 μm membrane. Levels of Gag and Env in cells and supernatants were measured by western blotting. (**B**) Immunoprecipitation (with anti-HA) and immunoblot (with anti-HA and anti-Flag) of HEK293T (4 <sup>×</sup> <sup>10</sup>6) cells co-transfected with the same source of eukaryotic expression plasmids encoding 3HA-Gag and 3Flag-Env.

Next, co-immunoprecipitation (Co-IP) assays were performed to measure the interaction between Env and Gag. 3Flag-Env and 3HA-Gag were co-transfected into HEK293T cells followed by IP with anti-HA antibody. The results of western blots showed similar levels of interaction between Env and Gag of both the B and Z1 clones (Figure 6B). Taken together, the 14-amino acid sequence that is missing in the Z1 Gag contributes to the increased cell-free transmission of clone Z1, but this activity is not a result of greater ability of assembling SVPs nor interacting with Env.

#### *3.4. The C-Terminal Region of Env Modulates the Entry of Cell-Free BFV Particles But Does Not Affect Virus Release*

We showed that chimeric clone Z1(EnvcB), which contains the C-terminus of Env-B, is defective in cell-free transmission (Figure 3C). To further understand the underlying causes, we examined the ability of EnvcB in mediating the assembly and release of Gag-Env particles. The results showed similar levels of SVPs that were produced with either EnvcB, Env-Z1 or Env-B together with Gag (Figure 7A). In addition, all three Env proteins formed Env-only SVPs (Figure 7A), which is consistent with the previous reports [26,34,37]. We also observed that there were more uncleaved gp130 in Env-B-containing SVPs that were produced with Gag-Env and Env alone compared to Env-Z1 SVPs and EnvcB SVPs, which suggests that this C-terminal region of Env affects Env protein processing.

**Figure 7.** The C-terminal region of Env modulates the entry of cell-free BFV particles but does not affect virus release. (**A**) Immunoblot analysis of Gag and Env in HEK293T cells co-transfected with the indicated plasmids for 48 h and viral particles in supernatants. (**B**) Immunoblot analysis of HEK293T cells transfected with different HA-Env and NL4-3.luc (env- ). Luciferase reporter assays to analyze the infectivity of viral particles in supernatant, and the data were the averages of three independent experiments. (**C**) Immunoblot analysis of HEK293T cells transfected for 48 h with plasmids encoding different Env and BTas, and luciferase reporter assays to analyze the membrane fusion ability mediated by different Env. The data were the averages of three independent experiments.

To determine whether the defective processing of BFV Env diminishes Env entry function, we used BFV Env proteins to pseudotype HIV-1(env-) viruses and measured virus infectivity. BFV Env proteins were co-expressed with HIV-1 DNA clone NL4-3.Luc.env-, which harbors a luciferase reporter gene and does not express HIV-1 Env. The culture supernatants containing BFV Env-pseudotyped HIV-1 particles were used to infect Cf2Th cells. Luciferase expression in Cf2Th cells reflects virus entry mediated by BFV Env proteins. The results showed that both Env-B and EnvcB were able to mediate HIV-1 entry, but the efficiency was 85-fold lower than Env-Z1. Furthermore, the K898E mutation reduced the entry efficiency of Env-Z1, which is in agreement with the increased entry of EnvcB

harboring the E898K mutation (Figure 7B). We also tested the efficiency of these Env proteins in mediating cell-cell fusion. HEK293T cells were co-transfected with BTas DNA and different Env DNA and then co-cultured with the BFVL indicator cells. Env-mediated fusion of 293T and BFVL cells allows BTas to activate the expression luciferase reporter gene in BFVL cells. These results showed that EnvcB and EnvB led to much higher cell fusion than EnvZ1 (Figure 7C). The E898K mutation impaired the ability of EnvcB to mediate cell fusion (Figure 7C). Together, these data suggest that the C-terminal region of Env affects Env processing and therefore has a key role in the function of Env to mediate virus-cell or cell-cell fusion.

#### **4. Discussion**

In this study, we generated high-titer cell-free BFV variants by serial virus passaging in vitro and further constructed a cell-free infectious clone, pBS-BFV-Z1. We also identified the viral genes and key amino acids that regulate BFV cell-free transmission. In particular, we found that the C-terminal region of Env, especially the K898 residue, is essential for BFV cell-free transmission activity.

It is known that BFV is highly cell-associated and spreads mainly through cell-to-cell transmission [17,18]. Although FFV, SFV, and PFV can transmit by both cell-to-cell and cell-free pathways, the sequence similarity between BFV and other FVs (FFV, SFV, and PFV) is low, and it is thus difficult to identify the viral genetic determinants through sequence alignment. We thus performed in vitro evolution to generate high-titer cell-free BFV variants. There are two factors that influence cell-free transmission, the number and the infectivity of the released virions [9,10]. Using the two distinct isolates, BFV-B and BFV-Z1, we detected Env-dependent production of BFV Gag-Env SVPs and demonstrated that Env-B was as efficient as Env-Z1 and EnvcB in mediating SVPs release (Figure 7A). However, Env-B was not completely processed in virus particles when compared with Env-Z1. It is reported that all known FV Env proteins contain a conserved optimal cleavage site RX(K/R)R between the SU and TM subunits and processing of SU/TM does not affect the egress of viral particles but is essential for the infectivity of the released virus particles [26,38,39]. Indeed, Env-Z1 was 85-fold more efficient than EnvcB and EnvB in mediating cell-free virus entry (Figure 7B). Our study demonstrates that the C-terminal sequence of Env regulates BFV cell-free entry efficiency rather than virion release capacity.

BFV Env is essential for virus budding, release, entry, and membrane fusion [18,19]. Compared with BFV-B, there are a few mutations in BFV-Z1 Env (eight changes among 990 residues), including three mutations in the N-terminus and five in the C-terminus. Previous reports state that the ubiquitination of three lysine sites in the N-terminal cytoplasmic tail region of PFV Env protein increases virus infectivity and decreases the production of SVPs [20]. In our study, the four N-terminal mutations (D42N, V55I, and P134H) did not change cell-free virus infectivity (Figure 3C). In contrast, the C-terminal mutations H816Y, P823S, E898K, R976K, and L978S, especially the E898K mutation, enhance cell-free transmission (Figure 4B,D). It is well-known that lysine (K) can undergo methylation, acetylation, succinylation, ubiquitination, and other modifications, which play an important role in regulating the protein activity and structure adjustment. Interestingly, we found that the equivalent position of the 898 residue in four currently described BFV isolates from the United States (GenBank accession number NC001831.1) [40], China (accession number AY134750.1) [41], Poland (accession number JX307861) [42,43], and Germany (accession number JX307862) [43], which spread only through cell-to-cell, is not a lysine (K). Nevertheless, the equivalent position is occupied by a lysine in other high titer cell-free strains, such as SFV and PFV. These suggest that the K898 in Env has an important role in FV cell-free transmission. The underlying mechanism warrants further study.

The Gag protein of BFV-Z1 lost a 14-amino acid sequence compared to BFV-B. This 14-residue deletion led to a smaller Gag-Z1, as shown by the results of western blotting (Figure 5B). Interestingly, Gag-Z1 enhanced cell-free infectivity by four- to five-fold (Figure 5C). Similar deletions have also been reported in other high titer cell-free BFV strains, although there is differing in the number of missing amino acids and their locations (Professor Martin Löchelt's unpublished data). It is known that the

Gag-Env interaction is very important for the budding and release of BFV virions. Yet, in our study, the interaction of Gag and Env in BFV-B and BFV-Z1 was almost the same, which suggests that the contribution of Gag-Z1 to enhanced cell-free transmission is not through promoting interaction with Env. The other changes of the BFV-Z1 genome contributed little to BFV cell-free transmission.

In summary, we demonstrated that the C-terminus of Env, especially the K898 residue, is critical for BFV cell-free transmission. This function of Env C-terminal sequence results from promoting BFV cell-free entry efficiency rather than viral particle release capacity. Our study thus suggests the possibility of generating high-titer BFV vectors through engineering viral Env—in particular, the C-terminal sequence.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1999-4915/11/2/130/s1, Table S1: The mutation amino acid (aa) sites in coding region of Cell-free BFV; Table S2: The mutation sites in noncoding region of Cell-free BFV.

**Author Contributions:** Conceptualization, S.Z., X.L., J.T.; Methodology, S.Z., X.L.; Formal Analysis, S.Z., X.L., J.T.; Resources, Z.L., T.B.; Writing-Original Draft Preparation, S.Z., X.L., Z.L.; Writing-Review & Editing, W.Q., J.T.; Supervision, W.Q., J.T.; Project Administration, W.Q., J.T.; Funding Acquisition, J.T.

**Funding:** This work was supported by grants from the National Natural Science Foundation of China (31670151).

**Acknowledgments:** We thank Martin Löchelt (German Cancer Research Center) for his generous sharing of screening basis for high-titer cell-free BFV strain. We also thank Chen Liang (McGill University, Canada) for the critical reading of the manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **In-Vivo Gene Therapy with Foamy Virus Vectors**

#### **Yogendra Singh Rajawat 1, Olivier Humbert <sup>1</sup> and Hans-Peter Kiem 1,2,3,\***


**\*** Correspondence: hkiem@fredhutch.org; Tel.: +1-206-667-4425

Received: 29 September 2019; Accepted: 20 November 2019; Published: 23 November 2019

**Abstract:** Foamy viruses (FVs) are nonpathogenic retroviruses that infect various animals including bovines, felines, nonhuman primates (NHPs), and can be transmitted to humans through zoonotic infection. Due to their non-pathogenic nature, broad tissue tropism and relatively safe integration profile, FVs have been engineered as novel vectors (foamy virus vector, FVV) for stable gene transfer into different cells and tissues. FVVs have emerged as an alternative platform to contemporary viral vectors (e.g., adeno associated and lentiviral vectors) for experimental and therapeutic gene therapy of a variety of monogenetic diseases. Some of the important features of FVVs include the ability to efficiently transduce hematopoietic stem and progenitor cells (HSPCs) from humans, NHPs, canines and rodents. We have successfully used FVV for proof of concept studies to demonstrate safety and efficacy following in-vivo delivery in large animal models. In this review, we will comprehensively discuss FVV based in-vivo gene therapy approaches established in the X-linked severe combined immunodeficiency (SCID-X1) canine model.

**Keywords:** gene therapy; in-vivo gene therapy; hematopoietic stem and progenitor cells; foamy virus vector; pre-clinical canine model; SCID-X1

#### **1. Introduction**

Therapies based on gene transfer to hematopoietic stem and progenitor cells (HSPCs) have achieved tremendous curative outcomes over the past decade and due to revolutionary success in some of these gene therapy clinical trials, these outcomes will redefine the clinical management of patients [1–3]. Pioneering gene therapy trials have shown that the genetic engineering of HSPCs can be a potentially superior alternative to allogeneic transplantation in the treatment of hematological monogenetic disorders including primary immunodeficiencies [4–6]. Transfer of therapeutic genes into long-term repopulating HSPCs can potentially cure blood disorders such as hemoglobinopathies and primary immunodeficiencies. Specifically, with regards to X-linked severe combined immunodeficiency (SCID-X1), recent data showed that this approach could be curative in animal models [7,8] together with very promising clinical results using gene therapy in SCID-X1 patients [1,9]. Despite these advances, gene therapy continues to face a number of challenges which, if not resolved, could be detrimental to the clinical translation of these approaches [4]. These challenges range from the translation of research findings to clinical practice, covering issues with regards to the need for a conditioning regimen, vector-related genotoxicity, specific vector design, and the requirement of sophisticated manufacturing facilities, all of which present various obstacles towards efficacious and practically feasible gene therapy. Manipulation of HSPCs ex-vivo in clinical trials have several drawbacks including a cumbersome and expensive process of extracting and purifying HSPCs from the patient and returning the genetically modified cells to the patient, which also causes a delay in treatment. Further, relative refractoriness of human hematopoietic stem cells to clinically available vector systems remains a significant obstacle for

most applications. To circumvent the current challenges and achieve practical HSPCs gene therapy, much work has focused on the development of newer delivery vehicles (viral vector based, or non-viral vector based using nanoparticles), alternative conditioning regimens (non-genotoxic), or simpler methods to target HSPCs in-situ by in vivo administration of the therapeutic transgene which bypasses the need for ex-vivo manipulation of these cells.

Priority should be given to the development of new viral vectors that overcome the known obstacles to stem cell transduction, such as the ability to transduce nondividing cells and utilization of virus that target receptors specific to primitive repopulating cells. With keeping in mind application and clinical translatability, various vectors such as γ-retroviral vector (γ-RVV, [9]), lentiviral (LVV, [1,10–12]), adenoviral [13–16] and adeno-associated viral [17,18] vectors have been designed to target refined cell populations with varying clinical and preclinical success. Although, significant progress has been made in the design of viral vectors, there are several limitations that still need considerable attention, particularly in the large-scale production of viral vectors under good manufacturing practice (GMP), which constitutes a major bottleneck. In recent times, AAV vectors have achieved clinical success; however, pre-existing immunity, relatively low transduction efficacy to non-dividing cells and relatively low transgene/cDNA (~4.4 kb) delivery capacity restricted their use [19,20]. Adenoviral vectors offer several advantages, including large cloning capacity and transduction of a number of tissues based on serotype, but their use has been restricted mainly due to their potent induction of acute immune response that could be fatal [21,22]. γ-RVV and LVV have shown very encouraging clinical outcomes specifically in monogenetic disorders of hematopoietic origin. However, the most commonly used envelope (VSV-G) to pseudotype clinical γ-RVV and LVV vectors targets HSPCs inefficiently and these vectors have significant batch to batch variability in large scale manufacturing and in their subsequent transduction of target cells [23,24]. Further, VSV-G pseudotyped lentiviral vectors are immunogenic, which leads to the elimination of corrected HSPCs in-vivo due to potent humoral and cellular immune responses, resulting in lower engraftment in patients [25,26]. To circumvent limitations of the currently available viral vector platform, there is an unmet need to either improve the current viral vector platform or to identify novel viral vectors applicable to HSPCs gene therapy.

More than two decades ago, a new vector system based on foamy viruses from the spumavirus family was described that can be used for gene transfer into murine hematopoietic stem cells (HSCs) [27,28]. Foamy viruses are non-pathogenic retroviruses with a wide tissue tropism that are commonly found in mammalian species. David Russell and his group developed replication-defective foamy virus vectors (FVVs) and demonstrated that these vector particles efficiently transfer a marker gene into repopulating mouse HSCs and into human CD34+ cells ex-vivo [29,30]. Since then, our laboratory has optimized parameters for efficient transduction of human and canine CD34+ HSPCs by FVVs [31–33]. In the last decade, our group has demonstrated efficacy of FVV in correction of monogenetic diseases of hematopoietic origin in large animal models (non-human primates and canine) and has established the safety of FVV in-vivo [7,8,34,35]. In these pre-clinical studies, FVV-mediated transgene delivery maintained high and persistent levels of marker-gene expression possibly due to improved transduction of quiescent cells and to the novel envelope/receptor system used for stem cell entry [7,8]. In the following sections, we will discuss advantages of FVV for therapeutic gene therapy applications and will specifically focus on our studies using in-vivo FVV delivery for the treatment of canine model of SCID-X1 since general characteristics and different aspects of FV biology have already been discussed elsewhere in this special issue.

#### **2. Limitations of Ex-Vivo Gene Therapy**

Hematopoietic stem and progenitor cells ex-vivo gene therapy utilizing viral vectors have been used in multiple clinical trials to circumvent the complications associated with allogeneic bone marrow transplantation. Despite the undeniable therapeutic benefits offered by HSPCs gene therapy treatment for various monogenetic disorders such as hemoglobinopathies [36,37], primary immunodeficiencies [1,9,38,39] and inborn metabolic disorders [40,41], this approach poses several

limitations which include requirement of cytotoxic conditioning regimen to promote engraftment of donor cells, lifelong administration of immunoglobulins in cases of immunodeficiencies, safety concerns due to vector genotoxicity (insertional mutagenesis), invasive procedure to procure HSPCs, and requirement of costly GMP facility for production of the cell product. Further, ex-vivo transduction protocols used in SCID-X1 clinical trials using SIN lentiviral vectors [1,42] required manipulation of HSCs by culture in multiple cytokines for approximately 3–4 days that may compromise HSCs pluripotency due to entry into pathways of differentiation and limit their engraftment potential and long-term repopulating capacity. Similar findings were reported in SCID-X1 dogs transplanted with bone marrow CD34+ cells from normal healthy donors [43]. Therefore, innovation is needed for long-term efficacy of ex-vivo gene therapies [44]. Alternatively, to circumvent these problems, an attractive approach is to transduce HSCs in-vivo within their natural environment by direct intravenous injection (in-vivo gene therapy). For comprehensive information of ex-vivo studies using FVV in HSPCs, we refer the reader to the excellent review by Vassilopoulos et al., in the same edition. In the current review, we will focus on FVV as an in-vivo gene therapy platform established in the canine model of SCID-X1. We will also discuss progress made in vector design and regimen used to mobilize HSCs out of the bone marrow.

#### **3. Viral and Non-Viral Vectors Platforms for In-Vivo Gene Therapy**

In lieu of various shortcomings of ex-vivo gene therapy, an alternative strategy that overcomes current limitations and still utilizes the benefits of gene therapy would provide a great advancement toward clinical translation. We propose that a possible answer lies in the use of in-vivo gene therapy. In-vivo gene transfer strategies administer the therapeutic vector either directly to the target organ or via the vascular system into blood vessels feeding that organ. In-vivo gene transfer offers several advantages over ex-vivo strategies including ease of administration, no need for HSPCs collection, manipulation and culture outside the body and thus no requirement for costly cell processing GMP facilities, and increased safety due to absence of myeloablative conditioning and transplantation procedures. Finally, this novel platform is portable and easy to disseminate worldwide particularly in under-developed countries with a large patient population needing treatment.

Multiple approaches and various delivery vehicles have been utilized for in-vivo transfer of therapeutic genes. These platforms include the use of non-viral [45] and viral vectors (integrating and non-integrating) [46] based approaches. Non-viral gene delivery systems have gained considerable attention as a promising alternative to viral delivery to treat various diseases [47,48]. However, despite extensive research, little is known about the parameters that underline the safe and effective in-vivo use of the nanoparticle-based delivery. So far, nanoparticles have shown promise in targeting and delivering cargo to various tumors very effectively. This success has been attributed to enhanced permeability and retention (the EPR effect) that can permit passive accumulation into tumor interstitial tissue. However, sub-optimal delivery is achieved with most nanoparticles because of heterogeneity of vascular permeability, which limits nanoparticle penetration. With regards to in- vivo gene therapy using nanoparticles, we cannot rely on passive accumulation and need to target tissue specific delivery of therapeutic cargo. Moreover, slow drug release limits bioavailability, which also restricts the use of nanoparticles for delivery of therapeutic cDNA. Overall, the modest efficacy, limited stability of nanoparticle conjugated to delivery cargo and lack of specificity of non-viral delivery are the central issues that need to be addressed. Thus, although nanoparticle-based approaches remain an attractive potential choice for in-vivo gene therapy, many questions still need to be answered for their effective clinical translation.

In early attempts with in-vivo gene therapy with viral vectors, VSV-G pseudotyped lentiviral vectors were used for direct intravenous injections in female rat brains using a stereotactic approach and showed effective transduction in multiple cell types including terminally differentiated neurons [49]. These early studies showed that LVVs can successfully be administered intravenously, transgene expression could be sustained for several months without detectable pathology, and they provided proof

of concept which eventually led to multiple follow-up studies with various vector systems. Intravenous administration of early adenoviral vector platforms in ornithine transcarbamylase deficiency (OTCD) clinical trials [50] resulted in fatal systemic inflammatory response syndrome in one patient [22]. Further, the use of first generation γ-RVVs (non-SIN γ-RVV) derived from Moloney murine leukemia virus (MMLV) with duplicated viral enhancer sequences within the long terminal repeats (LTRs) led to leukemogenesis in early SCID-X1 clinical trials [51–53]. These early setbacks paved the path for the design of novel and safe viral vectors. In the past two decades, not only have there been inventions of a variety of safer viral vector such as "gutless adenoviral vectors" [54–56] and self-inactivating (SIN) γ-RVVs [9,57] and LVVs [12,58], but these vectors have also become the backbone of ex-vivo gene therapy clinical trials [24,59–61].

With regards to use of integrating viral vectors (γ-RVVs and LVV's) in-vivo, intravenous administration of the retroviral replicating vector, Toca 511, recently demonstrated efficacy in orthotopic immune-competent mouse glioma model [62]. Further, a phase 1/2 study of a non-primate lentiviral vector based upon the equine infectious anemia virus (EIAV) expressing three genes involved in dopamine metabolism demonstrated safety of local lentiviral gene delivery into the central nervous system with evidence of clinical benefit [63]. In-vivo gene delivery using a lentiviral vector has also been applied clinically to the eye [64] and demonstrated that EIAV vectors provide a safe platform with robust and sustained transgene expression for ocular gene therapy. Cantore et al. reported the efficacy and safety of liver-directed in-vivo gene therapy in large and small animal models using lentiviral vectors. These vectors targeting the expression of a canine factor IX transgene in hepatocytes were well tolerated and provided a stable long-term production of coagulation factor IX in dogs with hemophilia B [65]. Even though these studies represent evidence of tremendous improvement in γ-RVV's and LVV's design for use as in-vivo gene delivery vehicles, major hurdles such as safety and host immune response against the vector and envelope used for pseudotyping restricts their use in clinical applications. Therefore, various strategies have been proposed to improve existing platforms to be utilized for in-vivo gene delivery [66–68].

Existing viral vectors have shown varying degrees of therapeutic efficacy in ex-vivo gene therapy clinical trials; nonetheless, little progress has been made for in-vivo clinical use with the exception of AAV vectors. In recent times, AAV vector-based in-vivo gene therapy has seen tremendous success for monogenetic disorders, which is evident with the recent approval of alipogene tiparvovec (Glybera, EMA, Amsterdam, Netherlands; year 2012) for the treatment of a rare inherited disorder, lipoprotein lipase deficiency, voretigene neparvovec (Luxturna, USFDA, Silver spring, Maryland, USA; year 2017) for the treatment of Leber's congenital amaurosis and for the treatment of pediatric spinal muscular atrophy (SMA) with bi-allelic mutations in the survival motor neuron 1 (SMN1), geneonasemnogene abeparvovec-xioi (Zolgensma, USFDA, year 2019). Although AAV based approaches have seen clinical success, these vectors have several drawbacks including limited scope with regard to target tissues and cell types that do not divide rapidly. AAV is largely maintained episomally, with very limited vector getting integrated in genome that will limit long-term efficacy. Although AAV vectors have little or no acute toxicity, there are reports of development of hepatocellular carcinoma in the mice model [69], ocular toxicity in mice [70], and the use of AAV vectors resulted into severe toxicity in non-human primates and pigs [71]. Altogether, current viral vector-based approaches for in-vivo gene therapy need to be further improved by leveraging recent discoveries in viral biology, progress in vector design and transduction, or exploring the use of novel viral vectors. In the following section, we discuss our promising data using in-vivo administration of FVV's for the treatment of SCID-X1 in the dogs.

#### **4. In-Vivo Gene Therapy for Canine SCID-X1 with FVVs**

FVs are unique retroviruses which belong to Spumaretrovirus and are nonpathogenic to their natural host [72]. FVs are prevalent in many mammals including nonhuman primates but they are not endemic in human populations [73]. Cell membrane associated heparan sulfate is a receptor for the prototype foamy virus in many species including humans [74]. As heparan sulfate is expressed in a variety of cell types, FVs are able to infect many tissues. Prototype FVVs were developed owing to several unique properties including lack of pathogenicity [75], broad tropism (can transduce many therapeutic targets), large transgene capacity, unique replication strategy which provides the ability to persist in quiescent cells, safer integration profile [31,76,77] and resistance to serum complement inhibition [27,78] which is a determining factor for in-vivo gene therapy. FVVs system have evolved from early replication competent vectors to third generation non replicating viral vectors which are efficient gene delivery vehicles that have shown great promise for gene therapy in various preclinical animal models including our SCID-X1 dogs [79].

The SCID-X1 dog model provides a fantastic opportunity to delineate various therapeutic strategies that are very much translatable to human SCID-X1 patients. Our collaborators, Felsburg and colleagues, have established a SCID-X1 dog model in basset hounds breed in which immunodeficiency is caused by a naturally occurring genetic mutation in the common gamma chain (γC) [80,81]. The mutation is a four base pair frameshift deletion in the signal peptide region that results in a pre-mature termination codon in exon1 [82]. Unlike genetically engineered γC deficient mice, canine SCID-X1 has a clinical and immunologic phenotype representative of human SCID-X1, thus making it an ideal pre-clinical model to improve gene therapy strategies for human SCID-X1.

In our very first study, we evaluated the efficacy of FVV gene therapy in treating SCID-X1 [7]. Five neonatal SCID-X1 dogs were treated by in-vivo administration of the FVV, containing Green Fluorescent Protein (GFP) and the coding sequence for the human common gamma chain (γC) linked by a 2A element, and expressed under control of the elongation factor 1 promoter (EF1α) (EF1α-EGFP-2A-γC). All five animals were intravenously injected with 4.0–8.4 <sup>×</sup> 108 infectious unit of FVV (age at injection varied from one day old to 13 days old). The injection of FVV was well tolerated by all five pups with no adverse effect. γC+ lymphocytes were detected in peripheral blood within 14 days post-treatment and, by 84 days, γC+ cells comprised 30%–58% of the total lymphocytes. Four out of five dogs showed a parallel trend for GFP+ lymphocytes. While promising, these results were limited by the relatively slow rate of lymphocyte reconstitution in these animals. The dogs surviving long-term eventually recovered normal lymphocyte counts at 112 days post-treatment (R2202 and 2203, Figure 1 inset, blue lines, Figure 1 includes selected results of two dogs from first study [7]). GFP+ (i.e., gene corrected) lymphocytes eventually accounted for 73% to 91% of circulating lymphocytes and expression of γC was sufficient for the development of CD3+ T cells, comprising 7% to 43% of total lymphocytes in peripheral blood by six weeks after administration of FVV. As expected, the majority of CD3+ cells expressed GFP that originated from the gene therapy vector and stained positive for CD4 and CD8. Most of the CD3+ cells also stained positively for CD45RA, a marker for naïve T cells, indicating recent thymic emigration.

We further assessed the T-cell receptor (TCR) diversity in each treated animal by spectratyping that analyzes genetic rearrangement of the 17 families of TCR Vβ segments. The longest surviving dog, R2202, initially expressed polyclonal TCR at early timepoints but eventually lost TCR diversity by 322 days post-treatment. These results demonstrated that delivery of the γC gene via FVV in-vivo in SCID-X1 dogs enabled thymocyte development and maturation as demonstrated by robust TCR rearrangement. Normal T cells counts as well as functionality of the γC-dependent signaling pathway were also restored, as demonstrated by tyrosine phosphorylation of the downstream STAT5 effector via activation of the γC pathway by IL-2 stimulation in peripheral blood mononuclear cells (PBMCs). Moreover, these γC+ lymphocytes were able to proliferate and re-enter into the cell cycle upon mitogen (phytohemagglutinin, PHA) stimulation as assessed by BrdU incorporation. Overall, FVV injection restored T-cell-mediated immunity with normal number and functionality of T cells generated. Specific antibody responses and immunoglobulin class switching was also evaluated in treated animals after immunization with the T cell-dependent neoantigen bacteriophage, ΦX174. This neoantigen is routinely being used in human patients with SCID-X1 to evaluate success of treatment with bone marrow transplantation or gene therapy. We found that treated animals showed a primary and secondary antibody response that is very similar to that seen in healthy human and canine subjects, indicating

that our treatment restored both the B and T cell cytokine signaling required for class switching and memory responses to this neoantigen.

**Figure 1.** Immune reconstitution in foamy virus vector (FVV) treated X-linked severe combined immunodeficiency (SCID-X1) dogs: Bottom left graph shows % gene corrected lymphocytes in peripheral blood of various animals. Top left inset highlights the early kinetics of gene marking in treated animals (blue vs. green vs. red lines). The mobilized dog H867 had a stable level of gene marking almost 1260 days post treatment. Bottom right graph represents absolute number of CD3+ T lymphocytes in peripheral blood of treated dogs. Top right inset emphasizes the days required to attain normal numbers of absolute T lymphocytes (blue vs. green vs. red lines; dashed black lines shows counts in healthy dogs). H867 maintained normal levels of CD3+ T cell counts for over three years. Data in this figure was reproduced from previous studies [7,8] and contains extended data on H864 and H867. R2202 and R2203 were part of the cohort of five dogs from our first study [7] and R2258, R2260, H864 and H867 were part of our second study [8], additional details are included in the text. EF1α-FVV: elongation factor 1 α promoter (EF1α-GFP-2A-γC) carrying foamy virus vector (FVV); PGK-FVV: human phosphoglycerokinase (PGK-mCherry-γC) promoter carrying FVV.

To assess the safety and potential genotoxicity of FVV, retroviral integration site (RIS) analysis was performed longitudinally on peripheral blood of treated animals. Based on the identification of only 20 unique RISs across all samples, we concluded that all dogs displayed a polyclonal hematopoietic contribution in gene-modified cells over time. To determine if our in-vivo FVV treatment resulted in significant off-target activity (intended target cell population was bone marrow or blood derived HSCs or HSPCs), we assessed the biodistribution of the DNA provirus from various tissues by RIS. The majority of the identified integrants originated from the perfused blood into the tissues except for one event detected in the gut. We also found two integration events in the ovaries of R2202 but no integration was observed in the testis of R2203. Taken together, intravenous FVV gene therapy resulted in a very low frequency of off-target transduction events and are thus not likely to be passed on in the germline. Although clonal diversity and TCR repertoire were relatively low in these animals, these results provided proof of concept that FVV can safely be used in a pre-clinical model for in-vivo gene therapy. In conclusion, this first study demonstrated feasibility and safety of FVV in-vivo gene therapy in SCID-X1 dogs. Further, this study proved that in-vivo gene therapy using FVVs could achieve immune reconstitution in a clinically relevant large animal model of SCID-X1.

#### **5. Optimization of FVV In-Vivo Gene Delivery**

Our preliminary study using FVV, EF1α-EGFP-2A-γC was equally efficacious (in terms of T lymphocyte reconstitution) to earlier studies using in-vivo γ-retroviral vectors (γ-RVV) to treat canine SCID-X1 [83] or to ex-vivo γ-RVV clinical trial results reported in human patients [9]. However, this study demonstrated limited gene marking in the B and myeloid cell lineages as was reported in γ-RVV studies. In particular, treated animals still developed opportunistic infections (Table 1) and produced low immunoglobulin (Ig)G levels, and marking levels in granulocytes and monocytes were very low (0.6%), indicating that more efficient transduction of multipotent HSCs is required to achieve long-term phenotypic correction.

**ID Age at Injection (Days Old.) Foamy Viral Vector Dose of Vector (Infectious Units) Mobilization Survival of Dogs (Days Post Treatment) Health Status or Infectious Complications** H867 16 PGK.mCherry.2A.γ<sup>C</sup> 4.0 <sup>×</sup> 108 G-CSF/AMD3100 1260 Healthy and Alive H864 16 PGK.mCherry.2A.γ<sup>C</sup> 4.0 <sup>×</sup> 108 G-CSF/AMD3100 ~486 *Bordetella bronchiseptica* R2258 <sup>18</sup> EF1α.EGFP.2A.γ<sup>C</sup> 4.0 <sup>×</sup> <sup>108</sup> NO ~820 Papillomavirus PGK.mCherry.2A.γ<sup>C</sup> 4.0 <sup>×</sup> <sup>108</sup> R2260 <sup>18</sup> EF1α.mCherry.2A.γ<sup>C</sup> 4.0 <sup>×</sup> <sup>108</sup> NO ~820 Papillomavirus PGK.EGFP.2A.γ<sup>C</sup> 4.0 <sup>×</sup> <sup>108</sup> R2202 1 EF1α-EGFP-2A-γ<sup>C</sup> 4.2 <sup>×</sup> <sup>108</sup> NO ~334 Coccidiosis;

R2203 1 EF1α-EGFP-2A-γ<sup>C</sup> 4.2 <sup>×</sup> 108 NO ~120 Canine Parainfluenza virus

Canine Distemper virus

**Table 1.** Description of SCID-X1 dogs treated by intravenous injection of FVV in various in-vivo gene therapy studies [7,8]. Two out of five dogs from the EF1α-EGFP-2A-γC study [7] were selected for inclusion in the table.

The suboptimal immune reconstitution observed in the preliminary dog study prompted us to evaluate several strategies to further optimize our FVV in-vivo gene delivery protocol. The kinetics of immune reconstitution may be enhanced by modifying FVVs design, for example, by using a stronger promoter in place of the short form of the human EF1α promoter to drive expression of γC. In addition, targeting HSCs more efficiently may increase gene marking in other cell lineages that do not have a selective advantage like T lymphocytes. This could, in principle, be achieved by using mobilizing agents to increase the number of circulating HSCs in peripheral blood at the time of vector administration.

In our next study using FVV for in-vivo gene therapy [8], we hypothesized that utilizing an alternative promoter to EF1α promoter could result in more robust γC expression in cells of hematologic origin. For this purpose, we redesigned our FVV with a human phosphoglycerokinase (hPGK) promoter to drive expression of the codon optimized human γC cDNA. Performance of each vector was compared side by side in a competitive repopulation assay by intravenous injection of equal doses of the FVVs, EF1α-EGFP -γC and PGK-mCherry -γC, in two newborn SCID-X1 animals. Competitive injection of EF1α-EGFP-γC and PGK-mCherry-γC in the same SCID-X1 dog bestowed an ideal opportunity to study the efficacy of each promoter under similar physiological conditions in the same animal. The absolute number of circulating lymphocytes steadily increased in both treated dogs during the first six months post-treatment, plateaued around 6–8 months, and remained within the normal range during the course of 2.5 years post-treatment (dogs R2258 and R2260, Figure 1, Green Lines). Strikingly, the majority of gene marking (70% to 90%) in peripheral blood came from the PGK-mCherry-γC vector in both animals, whereas marking from the EF1α-EGFP-γC vector comprised only a small fraction (5% to 10%). Interestingly, the early kinetics of gene marking in peripheral blood lymphocytes in these two animals was substantially improved as compared with animals treated with the EF1α-EGFP-γC vector from our previous study (R2202 and R2203, Figure 1). The fraction of gene-corrected peripheral lymphocytes reached 40% in both R2258 and R2260 at six weeks post-injection, as compared with 5% for the EF1α-EGFP-γC alone treated animal, demonstrating superior therapeutic performance of the PGK-mCherry-γC vector as compared to EF1α-EGFP-γC. Nevertheless, this new vector did not result

in improved targeting of the most primitive HSCs as showcased by limited gene marking in non-T cell lineages (B and myeloid cells).

In an attempt to target HSCs more effectively, we chose to treat a new animal cohort with a mobilization agent to increase the frequency of HSCs in peripheral blood at the time of FVV administration. Stem cell mobilization is defined as a process in which certain drugs are used to cause the trafficking of stem cells from the bone marrow into the blood and is commonly used to collect and store stem cells that may be used later as for bone marrow replacement therapy during a stem cell transplant. Granulocyte colony-stimulating factor (G-CSF) mobilized HSPCs from peripheral blood is the most widely used source of HSPCs for clinical transplantation [84]. As an alternative, plerixafor (AMD3100) was shown to not only efficiently mobilize the HSPCs in various species including humans [85,86], but also augment the mobilization and yield of CD34+ HSPCs when used in combination with G-CSF for clinical transplantation [87,88]. Furthermore, plerixafor was found to be very effective at mobilizing CD34+ HSPCs in dogs (3–10 fold increase in circulating CD34+ HSPCs count) [89]. Therefore, we hypothesized that mobilization prior to injection of FVV may enhance HSPCs transduction efficiency in-vivo. In the next cohort of animals, we treated two SCID-X1 dogs with both plerixafor (4 mg/kg, subcutaneously, single dose) and G-CSF (5 μg/kg, subcutaneously, twice a day for five days), which resulted in a 6.4–7.2-fold increase in circulating CD34+ cells [8]. Plerixafor treatment significantly increased the kinetics of lymphocyte expansion and gene marking as compared to non-mobilized PGK-mCherry-γC-treated animals. The fraction of gene-corrected lymphocytes in peripheral blood of mobilized animals reached 80% at six weeks post-treatment, whereas it took >20 weeks in non-mobilized animals to reach similar levels (green lines vs. red lines comparison in Figure 1). Accordingly, the time required to reach normal lymphocyte counts was markedly reduced in the mobilized animals (red lines, Figure 1). The two non-mobilized FV vector-treated animals (R2258 and R2260) initially showed normal frequency (90%) of CD3+CD45RA+ T cells in peripheral blood, but their frequency subsequently declined to 50% at one-year post-treatment. In comparison, levels of CD3+CD45RA+ cells have remained stable for both mobilized dogs, up to 18 months in H864 and for over 36 months post-treatment in H867, which continues to be monitored.

Thymic output was assessed by analysis of T cell receptor excision circles (TRECs) originating from TCR genes rearrangement that occurs during T-lymphocyte maturation. In the non-mobilized animals, TRECs were initially 10-fold lower in treated dogs (1000–1200 TREC/million peripheral blood mononuclear cells, PBMCs), as compared with a normal littermate control (12,000–14,000 TREC/million PBMCs), and then gradually declined over time. In contrast, TRECs in the mobilized animals reached normal levels as early as three months post-treatment and remained similar to the littermate control for up to three years post-treatment. In summary, we found that mobilization before FVV injection of SCID-X1 canines improved kinetic of T-lymphocyte reconstitution and increased thymic output to levels comparable to those in a healthy control.

The majority of expanded CD3+ lymphocytes were mature and expressed the coreceptor CD4 or CD8, with a small fraction of cells being CD4/CD8 double positive or double negative. Both mobilized animals H864 and H867 showed normal CD4:CD8 cell ratios, averaging two. The majority of circulating T lymphocytes in non-mobilized (R2258/R2260) and mobilized (H864/H867) also stained positive for TCR α/β starting at two months post-treatment, consistent with observations from healthy canines and humans. When assessing TCR diversity in each treated animal by TCR β spectratyping, we found that the two animals mobilized with G-CSF/AMD3100 showed robust spectratype profiles comparable to that of an aged-matched normal littermate, characterized by Gaussian distribution of fragments sized across 17 families of TCR vector β segments, and stable for up to three years post-treatment.

Similar to what we described in our previous study, we also verified functionality of the γC-dependent signaling pathways in all FVV-treated animals as well as effective stimulation in response to T-cell mitogen phytohemagglutinin. Primary and secondary antibody responses, and immunoglobulin class switching after immunization with the T cell–dependent neoantigen bacteriophage ΦX174 was also documented in these animals. In addition, polyclonal IgM, IgG, and IgA concentrations were measured from serum of mobilized dogs at multiple timepoints post-treatment and showed IgG (1850–3152 mg/dL) and IgM levels (250–382 mg/dL) in the treated SCID animals that were comparable to a healthy littermate control (IgG: 670–1650 mg/dL: IgM: 100–400 mg/dL), indicating partial restoration of B-lymphocyte function. Although, antibody levels were within normal range for the mobilized FVV treated dogs (H864 and H867), the gene marking levels in B lymphocytes were low throughout the study. In fact, despite substantially improving T-lymphocyte reconstitution, HSPC mobilization did not increase gene marking in myeloid cells (0–1.5%) or B lymphocytes (0–4%) in mobilized dogs, which is consistent with the levels of gene marking in myeloid and B lymphocytes in non-mobilized dogs. Thus, FVV in-vivo gene therapy can result in low levels of correction of HSCs or myeloid progenitors in addition to circulating common lymphoid or T cell progenitors that experience a selective growth advantage after gene correction.

To assess the safety and potential genotoxicity of FVVs, tissues from non-mobilized animals (R2258 and R2260) were collected and analyzed by RIS for biodistribution assessment of the foamy provirus. The vast majority of RISs (90%) detected in tissues were also found in peripheral blood at the same time point, suggesting that they originated from contaminating blood cells present in perfused tissues. Interestingly, ovaries and testes showed the smallest number of integration events (37 and 56, respectively, as compared with 766 and 469 in blood), and none of the RISs found exclusively in the gonadal tissues appeared at notable frequencies except for one integration site in the ovaries (chromosome 38; 34,522; 4.28%). No unique RIS at a noteworthy frequency was detected in semen from mobilized male H867. Taken together, these results suggested that off-target transduction events by in-vivo FVV treatment are rare and thus unlikely to be passed on in the germline, a finding also supported by the study of progeny issued from FVV-treated male R2260. RIS analysis from peripheral white blood cell DNA showed a marked increase in integration events in mobilized dogs H864 and H867 as compared with non-mobilized dogs R2258 and R2260, despite the use of an equal dose of FVV PGK-γC, consistent with the greater therapeutic activity of this vector. No clonal dominance was observed in any animal, but some persisting clones contributing to 0.1% of total gene marking were found in the non-mobilized animals, albeit with no indication of expansion. Taken together, our studies indicated that the use of G-CSF/AMD3100 mobilization before intravenous FVV delivery increases both the kinetics of lymphocyte recovery and diversity of immune reconstitution in SCID-X1 canines.

Altogether, we have so far treated nine dogs (five with EF1α-EGFP-2A-γC; two with both EF1α-EGFP-2A-γC and PGK-mCherry-2A-γC and two dogs with PGK-mCherry-2A-γC and G-CSF/AMD3100 mobilization) [5,6] and all treated dogs demonstrated efficacy and safety of FVV in-vivo gene therapy in canine model. Out of nine treated SCID-X1 dogs, five lived more than a year and three lived for over 2.5 years. Most importantly, the kinetics of lymphocyte reconstitution in our study using PGK-mCherry-2A-γC and G-CSF/AMD3100 mobilization is comparable that of SCID-X1 patients treated using ex-vivo gene therapy [9]. The long-term treated dogs demonstrated that correction of cellular and humoral immune compartment is sustained for over three years. Even though we have seen therapeutic correction of SCID phenotype in the dogs, particularly in T cell immune reconstitution, there is still room for improvement in myeloid and B cell gene marking. This could be achieved by treatment with more effective HSC mobilization regimen, by directly targeting primitive HSCs in their niche through intra-osseous delivery [90] of the viral vector, or with the use of a selection strategy for gene-modified HSCs [34,91]. Our findings from FVV-treated dogs are directly translatable to human SCID-X1 patients and validate FVVs as an effective vehicle for in-vivo delivery of the therapeutic transgene to correct SCID-X1 and potentially for other hematologic disorder.

#### **6. Summary and Future Perspective**

Ex-vivo HSPCs gene therapy clinical trials using non-SIN γ-RVV, SIN γ-RVV and LVV for SCID-X1 patients have demonstrated tremendous clinical success and will change the current practice of patient care. However, these therapies still require, in most cases, high doses of conditioning with chemotherapy, and thus patients are myelosuppressed for prolonged periods requiring hospitalization. In addition, conditioning can lead to genotoxicity and secondary malignancies. All ex-vivo approaches require invasive procedures to procure HSPCs and appropriate facilities, with very high cost. Due to these limitations, it will be challenging to apply current ex-vivo gene therapy strategies using integrating viral vectors on a broad scale. Therefore, there is an unfulfilled and ongoing quest for a safer and affordable treatment option for SCID-X1 patients. In-vivo gene therapy offers several advantages and could be a potential alternative to mitigate some of the challenges seen with ex-vivo HSPCs gene therapy. In terms of in-vivo gene therapy, γ-RVV and LVV have been used pre-clinically for disease models (as discussed briefly in this review) other than SCID-X1, with varying degree of therapeutic efficacy. However, challenges such as safety and immunogenicity remain a major hurdle for clinical translation of these vectors involving in-vivo delivery.

In-vivo gene therapy with FVVs has provided encouraging long term safety and efficacy results in the pre-clinical SCID-X1 dog model. These findings demonstrate comparable efficacy in terms of immune reconstitution and T cell functionality to human SCID-X1 clinical trials with the use of γ-RVV and LVV in ex-vivo gene therapy. Interestingly, in our SCID-X1 model, we have seen production of immunoglobulins that show normal B cell function with no prior conditioning, whereas, in human clinical trials normal function of B cells and production of immunoglobulin's (IgG and IgA) was attributed to use of conditioning regimen. Therefore, FVV's could provide an alternative platform for in-vivo gene therapy to mitigate some of the challenges possessed by γ-RVV and LVV. Our current pre-clinical FVV in-vivo gene therapy offers a path forward as an effective, safe, and accessible platform that may provide prompt treatment of newborn SCID-X1 patients without the complications associated with conditioning or manipulation of HSPCs. Moreover, this treatment scheme could be applied in other hematological disorders with monogenetic mutations, particularly in those disorders where manipulation of HSPCs ex-vivo is near impossible and where transfer of the therapeutic gene in few stem cells could be curative such as Fanconi anemia.

The excellent therapeutic benefits reported in patients and regulatory approval of viral vector-based gene therapy products such as Glybera, Luxturna, and Zolzensma are providing enough impetus to continue the exploration of novel in-vivo gene transfer approaches. However, the choices of viral vector for in-vivo gene therapy will depend on specific disease, target tissue in hand, size of transgene delivered, and packaging capacity of vector. Specific challenges that need to overcome for in-vivo gene transfer strategies include the induction of immunity by the viral vector, access of the gene therapy vector to the targeted cells/organ, efficient targeting of the vector to the cell and translocation of the genetic material to the nucleus, and any toxicity induced by expression of virus and/or transgene. Further, ideal in-vivo gene therapy that should be affordable as Glybera (now withdrawn from market), Luxturna, and Zolzensma are very expensive and could be a concern for widespread usage and commercial interest. Among the integrating viral vectors, FVV provides a suitable platform due to several advantages offered as compared to γ-RVV and LVV as discussed in this study. With our recent pre-clinical data in the canine model of SCID-X1, FVV have so far successfully shown clear long-term safety and therapeutic efficacy. This portable in-vivo gene delivery platform circumvents some of the challenges imposed by some clinically used viral vectors. Importantly, with the advent of novel gene-editing approaches, FVV, in its engineered integration-deficient form, provides an attractive option for the in-vivo delivery of editing reagents (Cas9, gRNA and donor template) due to its large packaging capacity. Altogether, FVV could be an answer to some of the challenges faced today in clinical translation via the in-vivo delivery of gene therapeutics.

**Author Contributions:** Conceptualization—H.-P.K. and Y.S.R.; writing—original draft preparation, Y.S.R.; writing—review and editing, Y.S.R., O.H., H.-P.K.; supervision, H.-P.K.; funding acquisition, H.-P.K.

**Funding:** This research was funded by National Institutes of Health, National Institute of Allergy and Infectious Diseases grant P01: AI097100, National Heart, Lung, and Blood Institute grant P01: HL122173 and Fred Hutchinson Cancer Research Center support grant P30: CA15704.

**Acknowledgments:** We thank Helen Crawford for her assistance with manuscript and figure preparation.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **FV Vectors as Alternative Gene Vehicles for Gene Transfer in HSCs**

#### **Emmanouil Simantirakis 1, Ioannis Tsironis <sup>1</sup> and George Vassilopoulos 1,2,\***


Received: 8 January 2020; Accepted: 15 March 2020; Published: 19 March 2020

**Abstract:** Hematopoietic Stem Cells (HSCs) are a unique population of cells, capable of reconstituting the blood system of an organism through orchestrated self-renewal and differentiation. They play a pivotal role in stem cell therapies, both autologous and allogeneic. In the field of gene and cell therapy, HSCs, genetically modified or otherwise, are used to alleviate or correct a genetic defect. In this concise review, we discuss the use of SFVpsc\_huHSRV.13, formerly known as Prototype Foamy Viral (PFV or FV) vectors, as vehicles for gene delivery in HSCs. We present the properties of the FV vectors that make them ideal for HSC delivery vehicles, we review their record in HSC gene marking studies and their potential as therapeutic vectors for monogenic disorders in preclinical animal models. FVs are a safe and efficient tool for delivering genes in HSCs compared to other retroviral gene delivery systems. Novel technological advancements in their production and purification in closed systems, have allowed their production under cGMP compliant conditions. It may only be a matter of time before they find their way into the clinic.

**Keywords:** foamy virus; gene therapy; HSC; gene marking; FV gene transfer to HSCs; gene therapy alternatives

#### **1. Introduction**

Hematopoietic Stem Cells (HSCs) are rare cells residing in the Bone Marrow (BM) that can support blood cell production for the lifetime of an individual. This is achieved through HSCs potential to differentiate and self-renew [1]. The self-renewal process is the central dogma in stem cell biology and is a character that is lost when HSCs are extensively manipulated ex vivo; this observation argues for the substantial contribution of the BM milieu in supporting HSCs. Practically, both HSC nature and BM nurture, in a coordinated interplay, allow HSCs to express their unique characters [2]. What has been elusive is the HSC per se; HSC cannot be identified by morphology but only by surface markers and functional assays.

In humans, HSCs are considered as CD34+ cells residing in the mononuclear cell population of the BM. Not all HSCs are CD34+ but a significant fraction of CD34+ cells are HSCs; this circular argument can be rephrased to convey that CD34+ cells are enriched in HSC and CD34 marking is used to assay whether a certain cell population has enough HSC to serve as a donor for a transplantation experiment [3]. The latter is the sole surrogate marker to assay whether a population of cells has the potential to support hematopoiesis in a myeloablated host. In addition, Stem Cell Transplantation (SCT), is the ultimate marker to support the potential of a viral vector to correct a genetic defect in a HSC; the marked or corrected cells of the donor will be present in the blood cells of the recipient host and will provide phenotypic correction for a lifetime.

Allogeneic SCT (ASCT) has been used since around the '70s for the treatment of acute leukemias. ASCT for genetic defects such as thalassemia was introduced in the 90s and achieved high cure rates for those who had a suitable donor [4]. In contrast, transplantation with autologous gene-corrected HSCs entered the therapeutic armamentarium around the year 2000 and was only recently endorsed as a therapeutic option by the FDA [5]. Gene correction for HSC requires the permanent integration of the vector in the host-HSC genome; if episomal, the vector is destined to dilute out at some point in the subsequent cell divisions since episomes do not faithfully follow the cell's DNA mitotic cycles [6]. Retroviruses are viruses with potential of permanent integration, which is a prerequisite for viral gene expression and a productive life cycle.

There are two subfamilies of Retroviruses: the first is the *Orthoretrovirinae* that includes the Genera *Lentivirus* and *Gammaretrovirus* (GV), and the second is *Spumaretrovirinae* that includes the genera *Simiispumavirus* and *Felispumavirus* [7]. The human foamy virus isolates are actually chimpanzee foamy viruses. There are three to date: human isolate HSRV clone 13 SFVpsc\_huHSRV.13, human isolate BAD327 SFVptr\_huBAD327, and human isolate AG15 SFVptr\_huAG15 [7]. For simplicity and familiarity reasons, in this article, wherever we refer to SFVpsc\_huHSRV.13 we shall use the name PFV, HSRV, or FV vectors. The first vectors to be developed for gene transfer purposes were those with the gamma-retroviral backbones and were used for genetic correction of immunodeficiency syndromes after extended preclinical validations [8,9]. However, along with the successes, novel problems emerged; The not-so-random insertion of the retroviruses close to oncogenes, resulted in the development of leukemias [10]. The development of high throughput technologies allowed the mapping of Retroviral Integration Sites (RIS) and shed light on the gammaretroviral vector potential to cause leukemias, in relation to their preference for specific genomic sites. Gammaretroviral vectors as well as Lentiviral vectors demonstrate a preference for the integration of their retroviral cDNA into transcriptionally active sites in the host genome [11]. In addition, the Gammaretroviral vectors showed an increased preference for landing near oncogenes, a potentially dangerous condition [12]. Further data showed that different genera of retroviruses have distinct preferences for genomic integrations: Gammaretroviral vectors and Foamy Viral vectors prefer to integrate in proximity to transcription start sites and regulatory elements like CpG islands, while Lentiviral vectors tend to integrate within coding sequences [13–15]. Beyond safety, a problem that retroviruses had in relation to the need for long-term expression in the context of HSCs, was the frequent observed silencing of the transferred transgene. Although silencing was related to the methylation of the retroviral DNA [16,17], a problem that could be solved with the use of insulator elements [18] or partially reversed with hypomethylating agents [19], the quest for a more efficient gene transfer vehicle did not seize.

In a chronological order, the two other kinds of retroviruses that were tested for their HSC gene transfer potential were the Lenti- and the Foamy- or Spuma- derived vectors. In a seminal paper on Lenti-vectors, the authors reported long-term gene marking in CD34+ cells transplanted in NOD/SCID mice [20]. Reports on the potential of Spuma- (or Foamy Virus, FV) derived vectors to transduce efficiently both murine and human HSCs and to express the transferred gene in a long-term manner, came soon after [21]. Although, currently, Lenti has gained tremendous popularity as a gene transfer tool and has obtained commercial approval for human use, FV-derived vectors are a safe and efficient alternative for gene transfer into HSCs [22,23]. Beyond HSC gene transfer, FV vectors of feline origin have been developed as gene transfer vehicles and have been tested so far in cell lines [24]; an interesting application of these vectors is their use as vehicles for the transfer of genes that induce immune responses to feline viruses [25]. In this review, we will present all the relative data that confirm the potential of FV-derived vectors as a non-inferior vehicle for HSC gene transfer and therapy. The main features of each retroviral gene delivery system are summarized in Table 1.


**Table 1.** Features of Retroviral gene delivery systems.

#### **2. Features of FV Vectors for HSC Gene Delivery**

A number of features are essential for any vector to be suitable for HSC gene transfer. Target cell tropism is the principal condition, and FV vectors have shown that they can transduce a number of cell lines and murine hematopoietic progenitors and stem cells early on during the vector development history [21,33,34]. The broad spectrum of permissive cells for FV transduction, suggested that an abundant molecule on the cell surface was facilitating FV attachment and/or entry. This molecule was later identified as heparan sulfate [35,36]. The expression of heparan sulfate in a wide variety of cells and its use by FVs as a non-specific receptor could explain the wide cell tropism that this vector demonstrates.

Another key principal for any gene transfer vehicle is its ability to deliver its cargo into non-dividing cells such as the HSCs. This principle was assessed through experiments that tested FV vector transgene expression in vivo after transplantation of transduced HSCs. Given that reporter gene expression can be traced in transplant recipients, it can be inferred that true, non-dividing HSCs were transduced by FV vectors [21,37]. However, gene transfer in HSCs occurs after the 5FU treatment of the HSC donor animals and the ex vivo manipulation of cells under strong cytokine stimulation; in such conditions, HSCs are practically forced to divide. Thus, conclusions on FV gene transfer into non-dividing cells from such experiments could not be confidently reached. It was later shown in a different cellular system that although FV DNA lacks nuclear localization signals, the FV DNA can survive long enough in the cytoplasm in anticipation of a subsequent cell division that will result in nuclear membrane break down [38,39]. Under such conditions, FV DNA can enter the nucleus and establish a productive infection or transduction. However, it should be clarified that FVs are complex retroviruses. Their replication demands an RNA intermediate from which viral cDNA synthesis occurs by the activity of viral reverse transcriptase. Furthermore, in in vitro systems, it has been observed that the reverse transcription of viral RNA is completed before virus budding [13].

Following cellular entry and nuclear penetration, gene transfer vehicles must also have long-term expression in daughter cells. This is a prerequisite for therapeutic procedures whose results are expected to span the lifetime of an individual. Vector gene silencing occurs through DNA methylation/histone acetylation and has been a central problem in gene transfer with retroviral vectors [16,17]. This does not seem to be a problem with FV vectors, since (i) transgene silencing has not been reported in any of the in vivo studies published so far and (ii) it has been shown that FV Long Terminal Repeats (LTRs) have the potential to insulate the vector genome when integrated in the host cell DNA [31,40]. The FV insulation can be seen as a double-edged sword; it protects the transgene from external effects but also protects the genomic environment from the effects of the FV vectors. This prompts the issue of activation of neighboring oncogenes at the integration site. Compared to LVs and GVs in in vitro immortalization assays, FV vectors had the lowest rate of read-through transcripts. When the LMO2

site was targeted for insertion, LMO2 mRNA increments were 280x for GV, 200x for LV, and 45x for FV vectors, normalized for genomic integration site, indicating insulator properties of the LTR. This low read-through transcription of an integrated FV vector occurs due to a 36 bp long CTCF binding motif inside its LTR. CTCF is the major chromatin insulator protein in vertebrates and binds the CCCTC sequence via various combinations of 11 zinc fingers [41].

After cell tropism and long-term expression is the issue of safety. From data on zookeepers that are chronic FV carriers, we have reassuring evidence that wtFV, causes no harm to its hosts [42,43]. However, beyond the wtFV non-pathogenic characteristics, foamy viral vectors possess a number of features making them attractive for use in gene therapy. Current FV designs consist of a split four plasmid system: the transfer vector and three accessory or helper packaging plasmids [44,45]. The packaging plasmids encode the Gag, Pol, and Env proteins; the Tas (or Bel1) and Bet proteins required for the infection and replication of the wtFV are dispensable for vector production and can be omitted. Furthermore, the vectors have a self-inactivating (SIN) design; deletion of the 3'LTR U3 region in the transfer plasmid is copied in in the 5'LTR sequence during packaging, rendering the deleted FVs safe for gene therapy applications. Bet deletion also renders the host cell susceptible to superinfection (multiple rounds of infection), a desirable feature for difficult to transduce cell targets [15,31,46]. PFV vectors were developed independently in the States and in Europe by Russel's [44] and Rethwilm's [47] groups, respectively. Here, we shall elaborate on the vectors developed by the Russel group. The FV vectors originated from the wtFV strain SFVpsc\_huHSRV.13 [48]. The wt provirus map is presented in Figure 1. The viral cDNA contains three overlapping open reading frames (ORFs). The viral cDNA also encodes the genes *gag*, *pol,* and *env*. The latter three are typical of all retroviruses. The aforementioned overlapping ORF encodes the genes *bel1*/*tas*, *bel2*, and *bel3*. Tas is a transactivator protein that binds into the internal promoter present in *env* and 5 LTR promoters and enhances the transcription of the wt integrated FV viral genome. Additionally, Bet results from the translation of spliced *bel1* and *bel2* ORF mRNA. Bel1-3 and Bet are not required for viral replication in vitro. Thus, they are omitted from vector designs. The current vectors comprise of four plasmids. A transfer vector with deleted viral LTRs. Some important cis-acting elements exist between the 5 LTR and *gag*, a part of the 5 *gag* (CAS I) sequence, a part of the 3 *pol* (CAS II) sequence, and a part of the 5 *env* sequence. The 3 LTR bears a deletion in the U3 region. Additionally, the 5 LTR is fused with the CMV promoter to render the vector Tas independent. In order to avoid the generation of the replication competent virus in the packaging cell lines, the Gag, Pol, and Env are expressed by separate plasmids. The coding sequences bear minimal overlap [21,28,44,49].

In the recent years, more light has been shed on the integration site profile of FVs on the host cell genome. On studies performed on human HSCs transduced with FV vectors, the vectors seemed to prefer integrating upstream of transcription start sites specifically within CpG islands, with only 4.4% of the integration sites to be 50kb proximal to proto-oncogenes [13,31,41]. Although Integration Sites (IS) proximity to the genes remains largely random, the preference that the FV vectors display makes them relatively safer than GV and LV vectors, as they prefer constitutively lamina associated regions (cLAD) and less often CpGs [15] to integrate.

An issue of interest in gene therapy applications is the transgene payload that a vector can carry. Transgene payload is of paramount importance because regulated gene expression requires non-coding sequences of significant length. As a rule of thumb, one should not exceed the length of the wt genome or a significant drop in titer is inevitable. Foamy viruses have the largest genome among retroviruses and as a result, FV vectors have enough space to accommodate a little over 9 kb of exogenous DNA [44].

**Figure 1.** Third generation Foamy Virus (FV) vector system. At the top wtPFV provirus genome is depicted. The LTR contains the entire U3 region of the Human Foamy Virus (HFV.). Red arrows indicated the promoters driving the expression of FV genes, as well as the internal promoter present inside the *env* sequence driving the expression of *tas*. *tas* is expressed at basal levels when the provirus integrates into host cell genome. Upon translation, Tas binds to the LTR and internal promoter and enhances the transcription of *gag*, *pol*, and *env.* Additionally, Tas enhances the transcription of *bel1*, *bel2*, and *bel3. bel1* and *bel2* transcripts splice into a single mRNA whose translation generates Bet. In third -generation systems, the FV genome is split into a four-plasmid system comprised of three helper plasmids encoding *gag*, *pol,* and *env* under the control of a CMV promoter and a pA (polyadenylation signal) to allow the packaging cell lines to express the proteins at high levels. Transfer vector is comprised of deleted LTRs. Viral RNA expression is driven by a CMV promoter and an important cis-acting element sequence between the 5 LTR and *gag*, a part of the 5 *gag* (CAS I) sequence, a part of the 3 *pol* (CAS II) sequence and a part of the 5 *env* sequence. These sequences are necessary for efficient virion assembly. The CMV 5 LTR fusion renders the viral vector production in packaging cell lines (HEK293T) Tas independent. Moreover, there is a deletion of the U3 region of LTR. Following integration into the transduced cell genome, the Tas dependent LTR promoter is regenerated. This fact renders the vector SIN and shuts the expression driven by 5 LTR off. Transgene expression is driven by internal promoters.

Finally, and aiming at practical applications, comes the issue of titers and storage. FV vectors can be concentrated by ultracentrifugation [44,45]. The codon optimization and expression design optimization of packaging and transfer vector plasmid sequences have allowed for the generation of high titer FV vector supernatants. These modifications rendered FV vectors as efficient as lentiviral vectors producing <sup>10</sup> <sup>×</sup> 107 TU/mL of crude supernatant [44]. Finally, FV supernatants can be concentrated and purified in closed systems using affinity chromatography with POROS-Heparin columns, followed by Tangential Flow Filtration and ultracentrifugation [50] or size exclusion chromatography using CaptoCore columns followed by sepharose-heparin affinity chromatography and ultracentrifugation [51]. Both purification approaches are performed in closed systems compliant with cGMP and yield significantly pure preparations. Lastly, in regard to storage, although not reported in the literature per se, we have developed a freezing medium that allows repeated freeze/thaw cycles with 80–90% yields in vector titer.

#### **3. Gene Marking Studies in Small Animals**

Replication incompetent FV vectors were shown to transduce nearly every cell line including hematopoietic cell lines [33,34]. As the next leap forward, FV vectors were tested for their gene marking potential with murine HSCs after transplantation into lethally irradiated hosts [21]. The data showed that FV vectors could transduce murine HSC with a marking efficiency of about 50% and after transplantation there was sustained expression of the transgene for over 6 months. Similarly, when human HSCs (CD34+) cells were transduced and transplanted in ablated NOD/SCID animals, high levels of gene marking were observed across all lineages, indicating the transduction of a true human HSC [37]. In a direct head-to-head comparison between GV, LV and FV vectors with identical constructs, FV vectors performed as efficiently as the LV vectors, indicating FV vectors' potential for clinical applications [52].

Another notable mention is that in close to 500 animals that we have transplanted, we never encountered an adverse outcome such as leukemia or lymphomaThese early observations on murine and human HSCs were the first indications that FV derived vectors were a relatively safe vector system that did not cause any harm in the HSC genomes and could thus be further tested in preclinical animal models as a therapeutic gene transfer vehicle.

#### **4. Therapeutic Gene Transfer in Murine Preclinical Models**

The testing ground for any gene therapy vector has traditionally been the genetic correction of β-thalassemia. Two such FV vectors have been developed and tested side-by-side. The expression cassette had the complete human β-globin gene under the control of the short native β-globin gene promoter with either an α-globin HS40 sequence, or a mini-LCR with the core sequences of the HS2 and HS3 regulatory elements from the β-globin LCR. Both vector viral stocks were used to transduce erythroleukemia lines, murine Lin- HSCs from normal and thalassemic mice and human CD34+ cells from β-thalassemia patients. The vectors had comparable efficiency in all settings, although the HS40 was marginally superior and more stable in vivo. In the thalassemic mouse model Hbbth3/+, the transplantation of FV-transduced HSCs with the HS40 vector resulted in 43% of peripheral blood expressing human β globin at 6 months post transplantation. This level of expression is adequate to establish a thalassemia carrier phenotype and a therapeutic effect [53].

Another monogenic recessive disorder amenable to treatment with gene therapy is chronic granulomatous disease (CGD) [54]; the X-linked form of the disease results from mutations in the CYBB gene that encodes the gp91phox, the larger subunit of the oxidase flavocytochrome b558. Patients with CGD lack production of microbicidal superoxide, resulting in recurrent infections and early deaths in childhood. An FV vector carrying the *gp91phox* gene under the control of a PGK or a MSCV-LTR promoter was tested for its ability to restore superoxide levels in transplanted animals. With an average of 41.5% chimerism, the transplanted animals displayed phagoburst activity that reached 40% of the wt animals, clearly indicating a robust therapeutic effect in the preclinical model [55,56]. Overall, superoxide production reached 70% of normal with low vector copy numbers per cell (<2), a finding that argues for copy number dependent expression and minimal cytotoxic effects. These levels of superoxide production are similar to what has been reported for lenti-based vectors [57].

FV vectors have also been used for the genetic correction of the Wiskott-Aldrich syndrome (WAS) mouse model [58]. WAS is an X-linked disorder characterized by eczema, immunodeficiency, and micro-thrombocytopenia resulting in bleeding tendency [59]. Uchiyama et al., used a WAS cDNA under the control of two different promoters (an endogenous and an A2UOCE-derived 631 bp fragment) and demonstrated the correction of the WAS phenotypic disorders. T-cell receptor-mediated responses, B-cell migration, platelet adhesion, and podosome formation in dendritic cells (DCs) were restored at levels that can translate into a therapeutic effect. In addition, they showed improvement in gene transfer rates with repeated transduction cycles and confirmed earlier integration site analyses. The IS distribution showed FV vectors landing within transcriptional units at a frequency of 22%–44%, versus 46%–72% for a similar WAS protein-expressing Lenti vector and a low tendency for integration near

oncogenes (<5%). As the authors state, the combination of complete phenotypic restoration with low copy numbers (<2) and a relatively safe IS profile, make the WAS protein-expressing FV vectors attractive for clinical applications.

One of the first attempts of gene therapy in humans targeted the SCID-X1 disorder with retroviral vectors. The attempt was considered a success [8] but it was also an alarm on the side effects that should be addressed before the wide application of gene therapy [10]. The disease was targeted with an FV vector to test whether FVs could be an alternative vector system and to test whether the integration sites in T cells were relatively "safer" when FVs were used to transfer the γc gene [60]. In the ED40515 (γc-) T cell line, the FV vector integration sites were located in close proximity to the transcriptional start sites in 13% of integration events relative to 25% with the retroviral vectors. In addition, in the 100 IS analyzed, there were none located within oncogenes, as opposed to three, present in the retrovirally-transduced cells. Finally, animals transplanted with FV-corrected HSC cells, recovered their T and B cell counts and their serum levels of IgM, IgG, and IgA.

A relative problem with HSC gene transfer applications, specifically where in vivo selective advantage of transduced cells does not apply, has been the low gene transfer rates. A potential solution to this problem could be the use of pharmacologic agents that could selectively eliminate non-transduced cells. In an attempt to test whether resting HSC can be transduced and selected in vivo, investigators used the MGMT(P140K) DNA repair protein in an FV backbone. Transplanted mice were treated at 4 weeks post Bone Marrow Transplantation (BMT), with sub-myeloablative conditioning with O6-BG (O-6-Benzylguanine) and BCNU(bis-chloroethylnitrosourea) at different doses and analyzed 6 months later, the mock-transduced animals had undetectable levels of MGMT expression as compared to 55% of positive animals in the FV transduced group [61]. These results confirm the potential of FV vectors to transduce relatively resting HSC cells at low Multiplicity Of Infection (MOIs) and to selectively enhance their presence in the peripheral blood with pharmacologic manipulations. This approach was also used in the design of FV vectors that could block HIV replication. A cassette, that carried three anti-HIV targets in the form of short hairpins (*tat*/*rev* at site I and site II and to human CCR5), was able to confer 4 log reductions in HIV replication assays [49]. FV vectors are particularly useful for anti-HIV gene therapy, since using Lenti-based vectors for RNAi delivery could be problematic since the anti-HIV sequences can jeopardize the vector packaging process [62]. Beyond therapeutic gene transfer targeting single gene disorders, FV vectors designed to deliver shRNA through PolIII promoters after HSC transfer and transplantation provided a significant long-term downregulation of target genes [63].

Overall, the potential of FV vectors to correct the genetic defect in monogenic disorders amenable to HSC gene addition has been well established in various disease settings. In addition, FV vectors perform as good as their LV counterpart and have not been linked to any adverse effects.

#### **5. Large Animal Preclinical Models**

Large animal models offer the potential to simulate conditions that are much closer to the clinical setting as compared to the murine preclinical models. FV vectors have been used to treat two such conditions, the leukocyte adhesion deficiency (LAD) and the pyruvate kinase deficiency (PKD).

LAD is a stem cells disorder that affects white blood cell migration in response to chemotactic signals and as a result, affects immune responses. Patients suffer from bacterial and fungal infections that most commonly occur on the skin and mucous membranes. The defects affect the leukocyte integrin ITGB2 gene (CD18) that prevents the expression of the CD11/CD18 adhesion complex on the cell surface [64]. The canine form of the disease (CLAD) recapitulates the severe deficiency phenotype of LAD-1 in children [65].

Canine HSCs were transduced with an FV vector expressing normal CD18 cDNA and were transplanted to affected dogs [66]. After a year, the dogs had their WBC counts restored to normal and survived without antibiotics, both signs of a functional cure of their disease. The observed long-term (24 months) lymphocyte marking rates of 5–10% are compatible with normal life expectancy. This result was achieved with a single overnight exposure to the vector and a nonmyeloablative conditioning regimen. Furthermore, the copy number per cell was low and in the order of 0.83-1.25 provirus copies per diploid cell.

In the canine setting, FV vectors have also been used to treat a Basenji pyruvate kinase deficient dog model. Pyruvate kinase deficiency causes severe hemolytic anemia, which is potentially lethal [67]. Since PKD does not provide survival advantage to the successfully transduced cells, the vector was enhanced by co-expressing the mutant MGMTP140K that provides resistance to O6-benzylguanine and BCNU, potent inhibitors of the wt MGMTP protein that is expressed in all normal tissues. At 100 days post HSC transplantation with transduced cells, the gene marking rates were 3.5% for granulocytes and 0.4% for lymphocytes. After three rounds of treatment with O6BG and BCNU, gene marking raised to 33% in granulocytes and 5.5% in lymphocytes [68]. This also translated to the correction of the phenotypic disorder, as evidenced by the normalization of LDH (an indicator of hemolysis) and the achievement of transfusion independence.

Finally, it has to be mentioned that FV vectors are resistant to lysis by human serum, a property that has sparked interest for direct in vivo delivery of the vectors, avoiding all ex vivo manipulations of HSCs. This has been tried with the X-SCID canine model [69]. A total of five newborn SCID-X1 dogs received i.v. infusions of FV vector preparations with doses ranging from 4.0-8.4 x10ˆ8 particles. The functional outcomes showed marginal immune reconstitution. The dogs did not survive past one year and succumbed to infections. In regard to off target viral integrations, two such sites were recorded: one in the gut, another (potential) at the virus infusion site and none in the germ cells. An improved protocol was later implemented from the same group that included stem cell mobilization with G-CSF/AMD3100 prior to FV vector delivery and the substitution of the EF1a promoter with that of the *pgk* gene in the FV vector [59]. The data argue for the faster recovery of T cell numbers and a broad TcR repertoire, and are clinically relevant when considering the overall survival that climbed to 2.5 years as compared to 330 days in the previous study. On the issue of safety, a major concern when a vector with broad cell tropism is tested, two conclusions emerged from this study: (i) the vast majority of integration sites were shared between tissues and blood indicating blood contamination and (ii) the gonads had the smallest number of integration events (37 and 56 for ovaries and testes, respectively, as compared to 766 and 469 in blood) with none of them being unique except for one integration site in the ovaries that may have been derived from non-germ cells.

#### **6. Conclusions**

Foamy virus vectors have been extensively tested in marking and in gene therapy studies with small and big preclinical animal models. The gene therapy trials featured in this review are summarized in Table 2. Their non-pathogenic nature is an attractive feature for clinical applications. This is further supported by all the positive outcomes that the gene therapy community has communicated from two decades of testing.It is therefore strange that these vectors have not found their way into the clinic. It seems from the relative literature that most manufacturing issues have been overcome and what is missing is the interest of pharmaceutical companies to develop it as a product. In the meantime, the gene therapy world has been dominated by LV vectors that are offered as pharmaceutical products often at extraordinary prices [70]. FVs with the relative ease of production, concentration, and purification could become a poor man's Ferrari for nations and insurance systems that cannot afford million-dollar price tags for a relatively simple and one-off treatment.



**Author Contributions:** G.V. set the conceptual frame for the manuscript; E.S. authored the manuscript; I.T. was instrumental in assigning the references. All authors have read and agreed to the published version of the manuscript.

**Funding:** There are no financial resources to be declared.

**Acknowledgments:** The authors would like to thank Christian Yanes for English text editing.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Viruses* Editorial Office E-mail: viruses@mdpi.com www.mdpi.com/journal/viruses

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18