Molecular Vision 2007; 13:2001-2011 <http://www.molvis.org/molvis/v13/a225/> Received 3 September 2007 | Accepted 12 October 2007 | Published 18 October 2007

Download Reprint

Targeted expression of two proteins in neural retina using self-inactivating, insulated lentiviral vectors carrying two internal independent promoters

Susan L. Semple-Rowland, Kristofer S. Eccles, Elizabeth J. Humberstone
 
 

Department of Neuroscience, University of Florida McKnight Brain Institute, Gainesville, FL 32611

Correspondence to: Susan L. Semple-Rowland, Department of Neuroscience, University of Florida McKnight Brain Institute, 100 Newell Drive, Bldg 59, Rm L1-100, Box 100244, Gainesville, FL 32610-0244; Phone: (352) 392-3598; FAX: (352) 392-8347; email: rowland@mbi.ufl.edu

Abstract

Purpose: There is increasing interest in developing viral vectors capable of reliably delivering multiple therapeutic genes to targeted cell populations. Currently, bicistronic vectors carrying two transgenes linked by an internal ribosomal entry site (IRES) are the most commonly employed vectors to accomplish this goal. We and others have found that the protein encoded downstream of the IRES in these vectors is not reliably expressed. The purpose of this study was to determine if replacement of the IRES in our self-inactivating, insulated, lentiviral vectors with a second, independent, cell-specific promoter would produce a vector that reliably expressed two proteins in targeted retinal cells in vivo.

Methods: Five dual promoter lentiviral vectors were constructed using our self-inactivating (SIN), insulated, lentiviral backbone. Each vector carried two independent transgenes encoding a fluorescent protein (GFP or tdTomato) whose expression was driven by three photoreceptor promoters (interphotoreceptor retinoid binding protein-IRPB1783; guanylate cyclase activating protein 1-GCAP292; rhodopsin-mOP500) and one ubiquitously expressed promoter (elongation factor 1α-EF1α). Constructs were packaged and injected into the optic vesicles of developing chicken embryos. The day before hatching, the retinas were removed and examined as whole mount tissues and as frozen sections using fluorescent microscopy.

Results: In our first experiment, we characterized the expression of the three photoreceptor promoters in chicken retina. The activities of GCAP292 and IRBP1783 were restricted to cone cells. GCAP292 was also active in a small sub-group of inner nuclear cells. The activity of mOP500 was restricted to rod cells. In our second experiment, we characterized the activity of three dual promoter vectors: GCAP292-GFP-IRBP1783-tdTomato, IRBP-tdTomato-GCAP292-GFP, and IRBP1783-tdTomato-mOP500-GFP. All three vectors produced easily detectable levels of GFP and tdTomato in transduced retinas, a result that prompted further analyses of the expression characteristics of these vectors. In retinas treated with either of the GCAP292/IRBP1783 dual promoter vectors, GFP and tdTomato were only detected in cone cells. No GFP was detected in the inner retina. In retinas treated with IRBP1783-tdTomato-mOP500-GFP, tdTomato was detected only in cone cells and GFP was detected only in rod cells, a result indicating that these promoters retained their intrinsic expression specificities in this dual promoter vector. In our final experiment, the ubiquitously expressed EF1α promoter was paired with either GCAP292 or mOP500 creating EF1α-tdTomato-GCAP292-GFP and EF1α-tdTomato-mOP-GFP. In retinas treated with EF1α-tdTomato-GCAP292-GFP, GFP was only detected in cone cells. In retinas treated with EF1α-tdTomato-mOP500-GFP, GFP was detected in rod cells and in several cells within the inner retina.

Conclusions: The results of this study show that it is possible to construct dual promoter lentiviral vectors that reliably express two proteins in a cell-specific manner. Among the dual promoter vectors created for this study, we have identified two vectors that specifically target expression of both transgenes to cone cells and one vector that specifically targets expression of one transgene to cone cells and the other transgene to rod cells. The ability to create one lentiviral vector that is capable of targeting expression of multiple genes to single or multiple cells in vivo should prove very useful in the development and delivery of complex, combination therapies to diseased tissues.

Introduction

Over the last decade, our understanding of the genetic causes of many retinal diseases has burgeoned as evidenced by the ever increasing numbers of mapped disease loci and identified genes. This information has led to the development and testing of several gene-based treatment strategies for these diseases, the most successful of which have been viral-mediated corrective gene treatments [1-4]. The results of these studies have been encouraging, but emerging evidence suggests that successful treatment of many photoreceptor diseases may require combination therapies in which corrective gene treatments are combined with anti-apoptotic and/or neurotrophic treatments that promote photoreceptor survival. To date, only a few studies have been done to investigate the potential benefits of combination therapies in the treatment of photoreceptor disease [5-7]. In all of these studies, expression of one of the two therapeutic genes was driven by a cell specific promoter while the expression of the other was driven by a ubiquitously expressed promoter. The results obtained from these studies while encouraging, could potentially be improved if the expression of all of the therapeutic genes is targeted to appropriate cells. In order to fully realize the therapeutic potential of combination therapies we will need to be able to deliver multiple genes to the neural retina and target the expression of these genes to either a single retinal cell type or to specific multiple cell types.

The most commonly employed method to deliver multiple genes to a single cell type is to use vectors that carry two transgenes that are linked by an internal ribosomal entry site (IRES). We and others who routinely use lentiviral vectors have found that the gene located downstream of the IRES in these bicistronic vectors is either not expressed, or is expressed at levels too low to be effective for many applications [8,9]. One possible solution for this problem was first described by Yu et al [9]. To co-express two genes in a single cell, these investigators constructed lentiviral vectors in which they replaced the IRES with a second independent promoter; thereby creating the first dual promoter lentiviral vectors. These investigators were able to show that high, sustained expression of two genes in transduced stem cells in culture was possible using vectors containing genes driven by either two constitutive promoters or a constitutive promoter and a cell-specific promoter.

The purpose of this study was to determine if we could achieve targeted expression of two proteins to specific retinal cell types in vivo by replacing the IRES in our SIN lentiviral vectors with a second, independent, cell-specific promoter. Five dual promoter lentiviral vectors were constructed using three photoreceptor promoters and one ubiquitously expressed promoter. The expression characteristics of these vectors were determined by examining the distribution of two fluorescent reporter proteins in the retinas of late-stage chicken embryos that had been injected with the vectors.

Methods

Construction of lentiviral vectors

The insulated pFIN, self-inactivating lentiviral vector used in this study was derived from our pTYF transducing vector that includes a cPPT-DNA flap sequence immediately upstream of the multiple cloning site [10]. To create the pFIN vector, the 2x250 bp chicken β-globin HS4 core sequence was removed from pNI-CD (kindly provided by Dr. Adam West) using KpnI, blunted, and ligated into a PmeI site that was introduced into the 3'-LTR of the pTYF vector using a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA; Figure 1A). Placement of the insulator sequence in the 3' LTR produces an integrated transgene that is flanked by insulator sequences (Figure 1B), a configuration that has been shown to minimize the potential for transcriptional interference from genomic DNA flanking the integrated transgene [11]. All subsequent vectors were built using the pFIN transducing vector as the starting backbone.

Single promoter vectors

GCAP292-GFP: The EF1α promoter in the pTV-EFGFP vector [12] was removed using NotI and PmeI and the GCAP292 promoter extending from -292 to +302 of the gene encoding chicken GCAP1 (transcription start point=+1; GenBank AF172707), which was amplified from a chicken GCAP cosmid clone (ccos24) [13] using Pfu DNA polymerase (Stratagene) and sense (5'-ACC CGT GTG CTT TTC-3') and antisense (5'-GCT CCA GTC ACT CT-3') primers containing a NotI and PmeI site, respectively, was ligated into the pFIN vector [14]. The GCAP292-GFP fragment was amplified from the pTV-GCAP292-GFP vector using Pfu and sense (5'-TTC ATC AGT CGA CCC GCA CCC GTG TG-3') and antisense (5'-GCT TTA CGT CGA CTC ACT TGT ACQA GCT CGT CCA-3') primers containing SalI sites. The resulting product was ligated into the SalI site of the pFIN vector creating pFIN-GCAP292-GFP.

mOP500-GFP: mOP500-GFP was amplified from pTR-MOPS500-GFP (UF5-mops; kindly provided by William Hauswirth, University of Florida) using Pfu polymerase and sense (5'-AGC AGG CGT CGA CGG TTC CTA GAT CTG AAT TCG GT-3') and antisense (5'-AAC TCG CGT CGA CGG TTT GTC CAA ACT CAT CAA TG-3') primers that introduced SalI sites that were used to insert the DNA fragment into the pFIN vector creating pFIN-mOP500-GFP.

IRBP1783-tdTomato

The murine interphotoreceptor retinoid-binding protein (IRBP) promoter BP1783-EGFP (kindly provided by Jeffrey Boatright, Emory University) using Pfu polymerase and sense (5'-GCG GCC GCC AGT GTG ATG-3' GAT A) and antisense (5'-GTT TAA ACT GGC GAC CGG TGG ATC CCA G-3') containing NotI and PmeI sites, respectively. This promoter was ligated into the NotI and PmeI sites of the pFIN vector. The cDNA encoding tdTomato was amplified from pRSETB-tdTomato (kindly provided by Roger Tsien, University of California San Diego) using Pful and sense (5'-AAT GAT GTT TAA ACC TGT ACG ACG ATG ACG ATA AG-3') and antisense (5'-TTA CAT GTC GAC TCC TTT CGG GCT TTG TTA-3') primers containing PmeI and SalI sites, respectively. The cDNA encoding tdTomato was ligated downstream of the IRBP1783 promoter using PmeI and SalI creating pFIN-IRBP1783-tdTomato.

Dual promoter vectors

Each dual promoter vector was constructed by inserting two transgenes into the multiple cloning site of the pFIN transducing vector backbone (Figure 1A). The transgenes were arranged head-to-tail so that transcription for both proceeded in the same direction and the transgenes shared the same polyadenlyation site located in dl.R region of the 3' LTR (Figure 1B).

GCAP292-GFP-IRBP1783-tdTomato: The IRBP1783-tdTomato DNA fragment was removed from the pFIN-IRBP1783-tdTomato vector using KpnI and was ligated into the KpnI site of the pFIN-GCAP292-GFP vector.

IRBP1783-tdTomato-GCAP292-GFP: The GCAP292-GFP DNA fragment was removed from the pFIN-GCAP292-GFP vector using SalI and was ligated into the SalI site of pFIN-IRBP1783-tdTomato.

IRBP1783-tdTomato-mOP500-GFP: The mOP500-GFP DNA fragment was removed from the pFIN-mOP500-GFP vector using SalI and was ligated into the SalI site of pFIN-IRBP1783-tdTomato.

EF1α-tdTomato-GCAP292-GFP: The IRBP1783 promoter was removed from pFIN-IRBP1783 using NotI and PmeI and was replaced with the EF1α promoter obtained from pTYF-EF1α-PLAP vector [10]. GCAP292-GFP was removed from pFIN-GCAP292-GFP using SalI and ligated into the SalI site of the resulting pFIN-EF1α-tdTomato vector.

EF1α-tdTomato-mOP500-GFP: GCAP292-GFP was removed from pFIN-EF1α-tdTomato-GCAP292-GFP using SalI and replaced with a SalI fragment containing mOP500-GFP that was obtained from pFIN-mOP500-GFP.

Packaging lentiviral vectors

Each of the pFIN self-inactivating transducing vectors described above was packaged into vesicular stomatitis virus G (VSV-G) glycoprotein-pseudotyped lentivirus using a three plasmid packaging system [10] with the following modifications. The 293FT cells (Invitrogen, Carlsbad, CA) were plated at a density of 3x10⁷ cells in T225 flasks and were transiently transfected the following day (80-90% cell confluence) with the three plasmid vectors, pNHP, pHEF-VSVG and the pFIN transducing vector using Superfect (Qiagen, Valencia, CA). Media containing virus was collected at 24 and 40 h post-transfection. The media collected 24 h post-transfection was centrifuged at 783xg for 5 min at 4 °C to remove cell debris and then stored at 4 °C until the next day. The media collected 40 h post-transfection was similarly centrifuged, mixed with the 24 h sample, and filtered through a 0.45 μM low-protein-binding Durapore filter device (Millipore, Billerica, MA). All but 26 ml of the filtrate, which was stored at 4 °C until further use, was concentrated using a Centricon-70 ultrafiltration column (Millipore) that was centrifuged at 2500xg for 1 h at 4 °C. The retentate recovered from the column was mixed with the unconcentrated media (26 ml), loaded into a 30 ml conical polyallomer tube (Beckman, Fullerton, CA), and centrifuged at 103,680xg in a SW28 rotor (Beckman) for 1.5-2.0 h at 4 °C. In most cases, the viruses prepared for this study were prepared using three T225 flasks. The viral pellets produced from 3 flasks were overlayed with 25 μl sterile DMEM or PBS and were stored at 4 °C overnight. The next day the viral solution was gently mixed on ice for 2 h using a tilting shaker, divided into aliquots, and stored at -70 °C until use. Particle titers were determined using a p24 ELISA kit obtained from SAIC (Frederick, MD), according to the protocols provided. Infectious titers of virus carrying tissue-specific promoters were estimated by multiplying the particle titers of the viruses by the ratio of infectious titer to particle titer obtained from our previous analyses of an EF1-nlacZ lentivirus [14]. The estimated titers of all virus preparations used in this study were about 10¹⁰ transducing units/ml. In all but one case, each virus was packaged one time and that preparation was used throughout the study. The criterion that was used to determine if a vector had been successfully packaged was that the virus was able to efficiently transduce embryonic retina. In these experiments, the presence of one or both of the fluorescent reporter proteins in the fully-developed retinas of treated embryos served as an indicator of packaging success. The one virus that failed to meet this requirement was associated with an unusually high death rate in the treated embryos that was attributed to contamination of the virus preparation. This problem was resolved by repackaging the virus.

Embryo injections

All protocols were approved by IACUC and adhered to the standard conventions on animal use for scientific studies. All eggs used in this study were obtained from our wild-type Rhode Island Red breeding colony that we maintain at the University of Florida. We have noted that the shells of the eggs produced by our colony are thicker and sturdier than the shells of eggs that we have purchased from commercial vendors. Use of thicker shelled eggs significantly reduces unexpected egg breakage. Use of thin shelled eggs in this procedure requires additional care to avoid breaking the shells while opening the egg and injecting the embryo. Lentivirus (about 1x10⁷ transducing units) was injected into the neural tubes of embryonic day 2 (E2) chicken embryos as previously described [14] with the following modifications. Eggs were incubated vertically with the blunt ends uppermost without rotation for the first two days prior to injection. On E2, embryos were injected through a hole made in the eggshell that was placed in the center of the blunt end of the egg rather than in the equatorial plane of the egg, an approach that has been shown to increase the viability of the injected embryos [15]. Following injection, eggs were sealed and incubated as previously described. For these experiments 10-15 eggs were injected with each virus preparation. Approximately 50% of the injected embryos survived to E20, and of those, approximately 50-80% had sufficient transduction of the retina to allow analyses of the expression characteristics of the virus. A minimum of 4 retinas was analyzed for each vector.

Preparation of retinal whole mounts and immunohistochemistry

On E19-20, the embryos were sacrificed and the eyes removed. The anterior segments of the eyes were removed using a razor blade and the eye cups of containing the retina/pigment epithelium were incubated in DMEM containing 0.05 U/ml dispase II neutral protease, grade II solution (Roche Applied Science, Indianapolis, IN) and 20 μM HEPES buffer solution (Invitrogen) for 30 min at 37 °C to aid in the removal of the pigment epithelium from the neural retina. Following the 30 min incubation, the retina was gently teased away from the underlying pigment epithelium using fire polished glass rods. Isolated retinas were fixed in 4% paraformaldehyde prepared from electron microscopic grade 20% paraformaldehyde solution (Electron Microscopy Sciences. Hatfield, PA) for 30-60 min at room temperature, rinsed in PBS, sandwiched between cover glasses, and photographed using a Zeiss Axiovert fluorescent microscope. The Zeiss fluorescent filter sets used to visualize GFP and tdTomato were filter set 38HE and filter set 00, respectively. tdTomato was imaged before GFP to minimize photobleaching of tdTomato. The retinas were then equilibrated in 30% sucrose in PBS overnight, cryosectioned at 10 μm, and either stained with DAPI alone or in combination with a rod transducin polyclonal antibody (1:1000; G-t1: K20 Santa Cruz Biotechnology, Inc., Santa Cruz, CA) that was visualized using either Alexa Fluor 488 or Alexa Fluor 594 secondary antibodies (Invitrogen). DAPI was visualized using the Zeiss filter set 02 and Alexa Fluor 488 and 594 were visualized using filter sets 38HE and set 00, respectively.

Results

In our first experiments, we examined the individual expression characteristics of the GCAP292, mOP500 and IRBP1783 promoters. Specifically, we were interested in characterizing the photoreceptor cell-specificity of these promoters in fully-developed chicken retina, information that would allow us to identify any changes in the expression characteristics of these promoters when placed in the context of dual promoter vectors. Examination of the retinas of embryos treated with pFIN vectors carrying GCAP292-GFP (Figure 2A-D), mOP500-GFP (Figure 2E-H), or IRBP1783-tdTomato (Figure 2I-L) revealed that the levels of fluorescent protein produced by these vectors were sufficient high to be easily detected in retinal whole mounts. In retinas of embryos treated with the GCAP292-GFP lentivector, GFP was detected in the photoreceptors in the outer nuclear layer and in a small sub-population of neurons in the external region of the inner nuclear layer (Figure 2B). Examination of sections of these treated retinas that had been immunostained with rod transducin antibody (Figure 2C) revealed that the GFP detected in the photoreceptor cell layer was limited to cone cells (Figure 2D). No GFP was detected in photoreceptors stained with the rod transducin antibody. In retinas of embryos treated with the mOP500-GFP lentivector, GFP was localized to the outer nuclear layer (Figure 2F) and more specifically, to rod photoreceptors (Figure 2G,H). Finally, in retinas of embryos treated with the IRBP1783-tdTomato lentivector, tdTomato was detected in the outer nuclear layer (Figure 2J) and was specifically limited to cone photoreceptors (Figure 2K,L).

In our second series of experiments, we constructed several dual promoter lentiviral vectors using the GCAP292, mOP500 and IRBP1783 promoters, as well as the ubiquitously expressed EF1α promoter to determine if cell-specific expression of two proteins could be achieved using two independent promoters. The fluorescent reporter proteins, GFP and tdTomato, were used to monitor expression of the paired promoters.

Two vectors were built utilizing the IRBP1783 and GCAP292 promoters, GCAP292-GFP-IRBP1783-tdTomato (Figure 3A-E) and IRBP1783-tdTomato-GCAP292-GFP (Figure 3F-J). Examination of whole mounts of retinas treated with either of these vectors revealed that the relative level of expression of the two fluorescent proteins across populations of transduced cells was highly variable. The levels of the two proteins in some cell populations were nearly equal while in others one protein was expressed at much higher levels than the other (Figure 3A,F). Examination of sections of these whole mounts showed that GFP and tdTomato were localized to the photoreceptor cell layers, an expression pattern that was maintained regardless of the order of the transgenes in the vector (Figure 3B-E; Figure 3G-J). The photoreceptor cell types in which tdTomato and GFP were expressed was determined by immunostaining sections of the whole mounts with rod transducin antibody. No tdTomato (Figure 3E) or GFP (Figure 3J) fluorescence was detected in the stained rod cells, suggesting that expression of both IRBP1783 and GCAP292 was limited to cone cells in context of these dual promoter vectors. The absence of GFP in the inner nuclear layer of retinas treated with these vectors suggests that pairing the GCAP292 promoter with the IRBP1783 promoter increased the specificity of the GCAP292 promoter.

We next paired IRBP1783 with mOP500, promoters whose activities when used in single promoter vectors are limited to cone and rod cells, respectively (Figure 2). In retinas treated with IRBP1783-tdTomato-mOP500-GFP, no retinal cells were identified that contained both tdTomato and GFP (Figure 3K). Examination of sections of these whole mounts revealed that the cells containing tdTomato were cone cells (Figure 3L) and those containing GFP were rod cells (Figure 3M-O). These data show that these promoters retained their individual expression characteristics in the context of the dual promoter vector.

In our final two vectors, we paired the ubiquitously expressed EF1α promoter with either GCAP292 or mOP500, placing EF1α upstream of the photoreceptor promoters. Examination of whole mounts of retinas treated with EF1α-tdTomato-GCAP292-GFP revealed that GFP was localized to one cell layer within the retina, whereas tdTomato was distributed throughout the thickness of the retina. A small number of GFP-positive cells also contained tdTomato (Figure 3P). Examination of sections of these retinas revealed that GFP was only present in the photoreceptor layer. The shapes and positions of the cell bodies of the GFP-positive cells within this layer suggested that these cells were cone cells (Figure 3Q,R). tdTomato, on the other hand, was detected in all cell layers (Figure 3Q,S,T), a pattern consistent with the ubiquitous expression characteristics of the EF1α promoter. Immunostaining of sections with rod transducin antibody revealed that while tdTomato was present in cone and rod cells, the majority of tdTomato-positive cells were cone cells (Figure 3U).

Examination of whole mounts of retinas treated with EF1α-tdTomato-mOP500-GFP revealed a fluorescent pattern that differed from that observed in retinas treated with EF1α-tdTomato-GCAP292-GFP. GFP in retinas treated with EF1α-tdTomato-GCAP292-GFP was restricted to the photoreceptor cell layer while GFP in retinas treated with EF1α-tdTomato-mOP500-GFP was detected throughout the thickness of the whole mount (Figure 3V), an expression pattern that was confirmed by examination of sections of these retinas. Analyses of sections of retinas treated with EF1α-tdTomato-mOP500-GFP showed that GFP was present not only in the photoreceptor layer, but also in several cell types within the inner retina (Figure 3W). Interestingly, within the photoreceptor layer, GFP was detected only in rod cells (Figure 3Z, AA). Occasionally, GFP was observed in the portions of the Müller cells that extend into the outer nuclear layer and their processes that form the outer limiting membrane (Figure 3A, arrow). tdTomato was detected in all cell layers in these retinas, a distribution similar to that observed in retinas treated with the EF1α-tdTomato-GCAP292-GFP vector (Figure 3X). Examination of sections stained with rod transducin antibody suggested that the majority of photoreceptors expressing tdTomato were cone cells (Figure 3B).

Discussion

The results of these experiments show that it is possible to target expression of two proteins to specific retinal cell types using lentiviral vectors carrying two independent internal promoters. We have designed three dual promoter vectors that exhibit photoreceptor-specific expression, two of which limit expression of both proteins to cone cells and one that expresses one protein in cone cells and the other in rod cells (Figure 4). To our knowledge, this is the first report of successful delivery and expression of multiple transgenes to specific cells in vivo using an insulated, dual promoter, SIN lentiviral vector.

The expression of the transgenes carried by dual promoter lentiviral vectors is defined by several factors including the intrinsic activities of promoters used to drive transgene expression, promoter-promoter interactions in multiple promoter vectors, transcriptional interference from surrounding cellular chromatin, and the stability of the transgene transcripts [16]. The intrinsic activities of individual promoters play a lead role in defining the expression characteristics of dual promoter vector. One of the goals of this study was to create a dual promoter vector carrying transgenes whose expression was limited to photoreceptor cells. To accomplish this, we selected three well-characterized photoreceptor promoters for use in these vectors. Our analyses of the activities of each of these promoters in chicken retina showed that GCAP292, a proximal promoter fragment derived from the chicken GCAP1 gene [13], is active in cone photoreceptor cells and in a sub-population of neurons in the inner nuclear layer, a pattern consistent with our previous analyses of the activity of this promoter [14]. mOP500, a promoter derived from the mouse rod opsin gene that is expressed in rod and cone cells in rat and mouse retina [17], exhibits rod-specific activity in chicken retina. Finally, IRBP1783, a fragment of the mouse IRBP gene whose expression has been shown to be restricted to photoreceptor cells [18,19], exhibits cone-specific activity in chicken retina.

In selecting the promoters that we used in our vectors, we did not limit our selection to promoters derived from chicken genes. In many of our previous studies, we have successfully used promoters derived from human, rat, mouse, and chicken genes to drive expression of various proteins in chicken retina [4,14,20]. We have found that the activities and specificities of the heterologous promoters in chicken retina have, in general, closely matched those observed in the species of origin. This is not particularly surprising since many of these promoters contain specific, highly conserved, cis-acting elements that are activated by trans-acting factors that are also highly conserved across species. The real benefit to being able use heterologous promoters is that the size of the pool of promoter candidates that may be used in developing viral vectors increases.

Keeping in mind the intrinsic expression characteristics of the GCAP292, mOP500, and IRBP1783 promoters, we then paired them and re-examined the expression characteristics of the resulting dual promoter vector. Pairing GCAP292 with IRBP1783 increased the specificity of the GCAP292 promoter, limiting its activity to cone cells. Pairing IRBP1783 with mOP500, on the other hand, produced a vector exhibiting two distinct cellular specificities: the expression of the IRBP1783 transgene was limited to cone cells and the expression of the mOP500 transgene was limited to rod cells. The results of these analyses show that the dual promoter configuration used in our vector does not necessarily alter the expression characteristics of the promoters used to construct them and that these promoters, in turn, are major determinants of the expression characteristics of the final vector. We were able to construct vectors that limited expression of both transgenes either to cones alone or to cones or rods. It should be possible to create dual promoter vectors that target expression of both transgenes to both cones and rods. Promoter candidates for these vectors include, but are not limited to, a 110-bp human rhodopsin kinase (GRK1) promoter [21,22], a 1.8-kb human retinal guanylate cyclase -1 promoter [20], and a 257-bp mouse IRBP promoter (IRBP156) [19].

In addition to the intrinsic activities of the individual promoters, interactions between promoters can also shape the expression characteristics of dual promoter vectors. Promoter-promoter interactions can influence the cellular specificity of the individual promoters and can also affect transgene expression levels. Unless specific steps are taken to minimize these interactions, the overall expression of these vectors can be influenced by interactions between promoters located outside of the proviral DNA integration site, as well as within the dual promoter transgene. We have reduced the possibility of promoter interactions originating from outside the integration site by inserting an HS4 insulator into the deleted U3 (dl.U3) region of the 3' LTR of our pFIN vector. By inserting the insulator into the dl.U3, we have taken advantage of the unique retroviral RNA reverse transcription process that will, upon completion, produce proviral DNA that is flanked by insulators. HS4 insulators have been shown to protect against repression or inappropriate activation of vector promoters by promoters surrounding the integration site [23-26]. In the presence of the HS4 insulators, we still found evidence suggesting possible promoter-promoter interactions. Pairing GCAP292-GFP with either IRPB1783 or EF1α limited expression of the GFP to cone cells, while pairing mOP500-GFP with EF1α expanded expression of GFP so that it was fount not only in rod cells but also in inner retinal cells. The specific mechanisms underlying the changes observed in the expression of the GCAP292-GFP and mOP500-GFP transgenes in these vectors are not known. In the case of the EF1α-tdTomato-mOP500-GFP transgene, the presence of GFP in the cells of the inner retina could reflect non-specific activation of the mOP500 promoter by EF1α promoter enhancer elements, an interaction that could be facilitated by the absence of the trans-acting factors that normally would activate the mOP500 promoter in rod cells. It was of interest that within the photoreceptor layers of retinas treated with EF1α-tdTomato-mOP500-GFP, GFP was not detected in cone cells. This observation was surprising since one might expect the same loss of specificity of the moP500 promoter that was observed in the inner retina in the cone cells. Perhaps cone cells contain trans-acting factors that interfered with activation of the mOP500 promoter. The order in which promoters appear in the vector can also affect promoter-promoter interactions. In this study, we did not examine all possible arrangements of the various promoters. In the one instance that we did build and test mirror image vectors (GCAP292-GFP-IRBP1783-tdTomato versus IRBP1783-tdTomato-GCAP292-GFP) we did not observe any differences in their retinal expression profiles. If promoter-promoter interactions are a problem, it may be possible to reduce these interactions by reversing the order of the promoters in the vector. For example, it may be possible to reduce the non-specific expression of GFP in the inner retina of animals treated with EF1α-tdTomato-mOP500-GFP by reversing the order of the transgenes in this vector.

As indicated above, the expression of dual promoter vector transgenes may also be affected by the stability of the transgene transcripts. We would be prompted to explore this possibility if a well-characterized promoter did not appear to be active when placed in the context of a dual promoter vector. If experimental analyses of transcript levels in transduced retinas suggest that transcript stability is a primary cause of low expression of a transgene, then it may be possible to resolve the problem by replacing the transgene promoter with one that exhibits higher activity. We are currently adding the woodchuck hepatitis virus posttranscriptional regulatory element (WPRE) to our pFIN vector backbone, a modification that has been shown to significantly enhance transgene expression [27]. It is likely that many of the applications in which these vectors are used will not require equal expression levels of both transgenes, especially in cases where one transgene encodes a reporter protein. Nonetheless, it should be possible to optimize expression of both transgenes by carefully selecting promoter pairs.

A final point that deserves mention is our finding that the relative levels of GFP and tdTomato expressed in transduced cells were not equal among all cells expressing the transgenes. This variability was particularly apparent in retinas treated with either GCAP292-GFP-IRBP1783-tdTomato or IRBP1783-tdTomato-GCAP292-GFP, vectors that limited expression of GFP and tdTomato to cone cells. In our experimental model, we deliver the viral vectors to the developing optic vesicles that contain the retinal progenitor cells. These progenitor cells divide and pass their integrated proviral DNA to their daughter cells, which then go on to differentiate into the mature cells of the neural retina. Thus, we would predict that daughter cells of transduced progenitor cells would exhibit similar protein expression levels. This prediction was born out by our observation that in many retinas photoreceptors with similar protein levels were clustered together (Figure 3A), a distribution pattern consistent with the reported distribution of the daughter cells of progenitor cells in developing chicken retina [28]. There were also several instances in which the intensities of the fluorescent GFP and tdTomato signals were very different in adjacent cells (Figure 3A,D,I). Factors that could lead to variations in protein expression levels between adjacent cells include differences in integrated transgene copy number, chromosomal position effects at the site of proviral DNA integration, and cell-specific differences in the activities of the transgene promoter. Differences in transgene copy number and chromosomal integration sites could only account for the variation in protein expression in adjacent cells if the cells originated from two different progenitor cells. This is possible, but it is unlikely that these factors underlie all of the variation that we observed. In addition, the potential for chromosomal position effects to influence transgene expression in this study was minimized by the presence of the HS4 insulators that flank the integrated transgene. The variation in transgene expression observed between adjacent cells could reflect cell-specific differences in the relative activity levels of the transgene promoters in the various types of cone cells present in chicken retina. The cone cell population in chicken retina is heterogeneous; it consists of double cones and four classes of single cones, all of which are morphologically and biochemically distinguishable [29-31]. This heterogeneity, which is not often considered when targeting cone cells, could contribute to the protein expression differences that were observed in adjacent cone cells.

At the outset of this study, our goal was to develop a lentiviral vector that would reliably co-express two proteins in a single targeted cell. We were able to achieve this goal using dual promoter lentiviral vectors, something we were unable to achieve using IRES elements. In constructing our vectors, we used promoters derived from chicken and mouse genes. The ability of these promoters to work together in our dual vectors suggests that it will be possible to use these vectors in species other than chicken. Dual-promoter vectors can be used to target expression of two genes to a single cell type or to two different cell types. The flexibility of these vectors should prove useful in several types of applications including those designed to treat autosomal dominant diseases that require co-delivery of genes encoding post-transcriptional gene silencing molecules and normal copies of the affected gene, and in those designed to delivery multiple therapeutic genes to diseased tissues.

Acknowledgements

We would like to thank Dr. Roger Tsien (University of California San Diego) for providing the cDNA encoding tdTomato, Dr. Jeffrey Boatright (Emory University) for providing the IRBP1783 promoter, and Dr. William Hauswirth (University of Florida) for providing the mOP500 promoter. This study was supported by grants from the National Eye Institute EY11388, a grant from the University of Florida McKnight Brain Institute, and Core Grant EY08571. Preliminary results from this study were presented at the 2007 Association for Research in Vision and Ophthalmology meeting.

References

1. Acland GM, Aguirre GD, Ray J, Zhang Q, Aleman TS, Cideciyan AV, Pearce-Kelling SE, Anand V, Zeng Y, Maguire AM, Jacobson SG, Hauswirth WW, Bennett J. Gene therapy restores vision in a canine model of childhood blindness. Nat Genet 2001; 28:92-5.

2. Acland GM, Aguirre GD, Bennett J, Aleman TS, Cideciyan AV, Bennicelli J, Dejneka NS, Pearce-Kelling SE, Maguire AM, Palczewski K, Hauswirth WW, Jacobson SG. Long-term restoration of rod and cone vision by single dose rAAV-mediated gene transfer to the retina in a canine model of childhood blindness. Mol Ther 2005; 12:1072-82.

3. Ali RR, Sarra GM, Stephens C, Alwis MD, Bainbridge JW, Munro PM, Fauser S, Reichel MB, Kinnon C, Hunt DM, Bhattacharya SS, Thrasher AJ. Restoration of photoreceptor ultrastructure and function in retinal degeneration slow mice by gene therapy. Nat Genet 2000; 25:306-10.

4. Williams ML, Coleman JE, Haire SE, Aleman TS, Cideciyan AV, Sokal I, Palczewski K, Jacobson SG, Semple-Rowland SL. Lentiviral expression of retinal guanylate cyclase-1 (RetGC1) restores vision in an avian model of childhood blindness. PLoS Med 2006; 3:e201.

5. Bennett J, Zeng Y, Bajwa R, Klatt L, Li Y, Maguire AM. Adenovirus-mediated delivery of rhodopsin-promoted bcl-2 results in a delay in photoreceptor cell death in the rd/rd mouse. Gene Ther 1998; 5:1156-64.

6. Buch PK, Maclaren RE, Duran Y, Balaggan KS, MacNeil A, Schlichtenbrede FC, et al. In Contrast to AAV-Mediated Cntf Expression, AAV-Mediated Gdnf Expression Enhances Gene Replacement Therapy in Rodent Models of Retinal Degeneration. Mol Ther 2006 Jul 25.

7. Schlichtenbrede FC, MacNeil A, Bainbridge JW, Tschernutter M, Thrasher AJ, Smith AJ, Ali RR. Intraocular gene delivery of ciliary neurotrophic factor results in significant loss of retinal function in normal mice and in the Prph2Rd2/Rd2 model of retinal degeneration. Gene Ther 2003; 10:523-7.

8. Mizuguchi H, Xu Z, Ishii-Watabe A, Uchida E, Hayakawa T. IRES-dependent second gene expression is significantly lower than cap-dependent first gene expression in a bicistronic vector. Mol Ther 2000; 1:376-82.

9. Yu X, Zhan X, D'Costa J, Tanavde VM, Ye Z, Peng T, Malehorn MT, Yang X, Civin CI, Cheng L. Lentiviral vectors with two independent internal promoters transfer high-level expression of multiple transgenes to human hematopoietic stem-progenitor cells. Mol Ther 2003; 7:827-38.

10. Coleman JE, Huentelman MJ, Kasparov S, Metcalfe BL, Paton JF, Katovich MJ, Semple-Rowland SL, Raizada MK. Efficient large-scale production and concentration of HIV-1-based lentiviral vectors for use in vivo. Physiol Genomics 2003; 12:221-8.

11. Ramezani A, Hawley TS, Hawley RG. Performance-and safety-enhanced lentiviral vectors containing the human interferon-beta scaffold attachment region and the chicken beta-globin insulator. Blood 2003; 101:4717-24.

12. Chang LJ, Urlacher V, Iwakuma T, Cui Y, Zucali J. Efficacy and safety analyses of a recombinant human immunodeficiency virus type 1 derived vector system. Gene Ther 1999; 6:715-28.

13. Semple-Rowland SL, Larkin P, Bronson JD, Nykamp K, Streit WJ, Baehr W. Characterization of the chicken GCAP gene array and analyses of GCAP1, GCAP2, and GC1 gene expression in normal and rd chicken pineal. Mol Vis 1999; 5:14 <http://www.molvis.org/molvis/v5/a14/>.

14. Coleman JE, Fuchs GE, Semple-Rowland SL. Analyses of the guanylate cyclase activating protein-1 gene promoter in the developing retina. Invest Ophthalmol Vis Sci 2002; 43:1335-43.

15. Bednarczyk M, Lakota P, Siwek M. Improvement of hatchability of chicken eggs injected by blastoderm cells. Poult Sci 2000; 79:1823-8.

16. Sinn PL, Sauter SL, McCray PB Jr. Gene therapy progress and prospects: development of improved lentiviral and retroviral vectors--design, biosafety, and production. Gene Ther 2005; 12:1089-98.

17. Glushakova LG, Timmers AM, Issa TM, Cortez NG, Pang J, Teusner JT, Hauswirth WW. Does recombinant adeno-associated virus-vectored proximal region of mouse rhodopsin promoter support only rod-type specific expression in vivo? Mol Vis 2006; 12:298-309 <http://www.molvis.org/molvis/v12/a32/>.

18. Boatright JH, Buono R, Bruno J, Lang RK, Si JS, Shinohara T, Peoples JW, Nickerson JM. The 5' flanking regions of IRBP and arrestin have promoter activity in primary embryonic chicken retina cell cultures. Exp Eye Res 1997; 64:269-77.

19. Boatright JH, Knox BE, Jones KM, Stodulkova E, Nguyen HT, Padove SA, Borst DE, Nickerson JM. Evidence of a tissue-restricting DNA regulatory element in the mouse IRBP promoter. FEBS Lett 2001; 504:27-30.

20. Coleman JE, Wu K, Fulle HJ, Semple-Rowland SL. The 5' flanking sequence of the human retGC1 gene acquires a photoreceptor cell restricted activity pattern over the course of retinal development. Mol Vis 2004; 10:720-7 <http://www.molvis.org/molvis/v10/a87/>.

21. Young JE, Vogt T, Gross KW, Khani SC. A short, highly active photoreceptor-specific enhancer/promoter region upstream of the human rhodopsin kinase gene. Invest Ophthalmol Vis Sci 2003; 44:4076-85.

22. Young JE, Gross KW, Khani SC. Conserved structure and spatiotemporal function of the compact rhodopsin kinase (GRK1) enhancer/promoter. Mol Vis 2005; 11:1041-51 <http://www.molvis.org/molvis/v11/a123/>.

23. Burgess-Beusse B, Farrell C, Gaszner M, Litt M, Mutskov V, Recillas-Targa F, Simpson M, West A, Felsenfeld G. The insulation of genes from external enhancers and silencing chromatin. Proc Natl Acad Sci U S A 2002; 99 Suppl 4:16433-7.

24. Chung JH, Bell AC, Felsenfeld G. Characterization of the chicken beta-globin insulator. Proc Natl Acad Sci U S A 1997; 94:575-80.

25. Recillas-Targa F, Pikaart MJ, Burgess-Beusse B, Bell AC, Litt MD, West AG, Gaszner M, Felsenfeld G. Position-effect protection and enhancer blocking by the chicken beta-globin insulator are separable activities. Proc Natl Acad Sci U S A 2002; 99:6883-8.

26. Aker M, Tubb J, Groth AC, Bukovsky AA, Bell AC, Felsenfeld G, Kiem HP, Stamatoyannopoulos G, Emery DW. Extended core sequences from the cHS4 insulator are necessary for protecting retroviral vectors from silencing position effects. Hum Gene Ther 2007; 18:333-43.

27. Zufferey R, Donello JE, Trono D, Hope TJ. Woodchuck hepatitis virus posttranscriptional regulatory element enhances expression of transgenes delivered by retroviral vectors. J Virol 1999; 73:2886-92.

28. Bowmaker JK, Knowles A. The visual pigments and oil droplets of the chicken retina. Vision Res 1977; 17:755-64.

29. Bowmaker JK, Heath LA, Wilkie SE, Hunt DM. Visual pigments and oil droplets from six classes of photoreceptor in the retinas of birds. Vision Res 1997; 37:2183-94.

30. Morris VB, Shorey CD. An electron microscope study of types of receptor in the chick retina. J Comp Neurol 1967; 129:313-40.

31. Iwakuma T, Cui Y, Chang LJ. Self-inactivating lentiviral vectors with U3 and U5 modifications. Virology 1999; 261:120-32.

32. Fekete DM, Perez-Miguelsanz J, Ryder EF, Cepko CL. Clonal analysis in the chicken retina reveals tangential dispersion of clonally related cells. Dev Biol 1994; 166:666-82.