Molecular Imaging of Biological Gene Delivery Vehicles for Targeted Cancer Therapy: Beyond Viral Vectors

Abstract

Cancer persists as one of the most devastating diseases in the world. Problems including metastasis and tumor resistance to chemotherapy and radiotherapy have seriously limited the therapeutic effects of present clinical treatments. To overcome these limitations, cancer gene therapy has been developed over the last two decades for a broad spectrum of applications, from gene replacement and knockdown to vaccination, each with different requirements for gene delivery. So far, a number of genes and delivery vectors have been investigated, and significant progress has been made with several gene therapy modalities in clinical trials. Viral vectors and synthetic liposomes have emerged as the vehicles of choice for many applications. However, both have limitations and risks that restrict gene therapy applications, including the complexity of production, limited packaging capacity, and unfavorable immunological features. While continuing to improve these vectors, it is important to investigate other options, particularly nonviral biological agents such as bacteria, bacteriophages, and bacteria-like particles. Recently, many molecular imaging techniques for safe, repeated, and high-resolution in vivo imaging of gene expression have been employed to assess vector-mediated gene expression in living subjects. In this review, molecular imaging techniques for monitoring biological gene delivery vehicles are described, and the specific use of these methods at different steps is illustrated. Linking molecular imaging to gene therapy will eventually help to develop novel gene delivery vehicles for preclinical study and support the development of future human applications.

Introduction

Nucleic acids and their analogs have many therapeutic applications in cancer that have been exploited to deliver genes as DNA plasmids, to mediate gene knockdown via RNA interference (RNAi) mechanisms, or to alter pre-mRNA splicing to ameliorate disease-causing mutations [1]. Naked therapeutic genetic molecules are generally difficult to deliver primarily due to rapid clearance [2] via nucleases, which limit the serum half-life of unmodified small interfering RNA to 5–60 min [3] and DNA to 10 min [4], a lack of organ-specificity, and a low efficiency of cellular uptake following systemic delivery. Although nucleic acid modifications, such as incorporation of targeting ligands and engineering using tissue-specific promoters, can overcome some of these limitations, specialized gene delivery vehicles (GDV) that improve delivery efficiency and cell-specificity are preferred.

Viral vectors and cationic liposomes are at the forefront of GDV technology with a large number already in clinical trials. Despite their potential, significant limitations remain, including immune recognition [5] of most viral GDVs, mutagenic integration [6] for some viruses, and inflammatory toxicity and rapid clearance of liposomes [7]. For example, immune activation can require the concomitant use of immunosuppressive strategies to overcome readministration problems with current GDVs [8]. Furthermore, most current GDVs lack sufficient specificity for cancer tissue, which limits their gene delivery capability. This is particularly important given that optimizing the specific affinity of GDVs would improve the efficiency of gene/protein delivery and reduce unwanted transfection of noncancerous tissues [9].

Molecular imaging of reporters for specific genes could play a crucial role in optimizing gene delivery and expression by enabling quantitative monitoring of reporter gene expression and the therapeutic effect of transgenes in vivo [10]. Despite sustained efforts to improve the efficacy and safety of liposome formulations and viral vectors using imaging technologies, research into alternative GDVs may provide the least complex solution for novel therapeutic and diagnostic applications. This review will introduce imaging applications for biological delivery vector systems that have been studied so far with an emphasis on bacteria-mediated cancer gene therapy. Additionally, exemplary examples of combined imaging and therapy will be briefly reviewed.

Molecular Imaging Modalities and Imaging Reporter Genes

Several molecular imaging technologies have been used to try to monitor transgene expression for gene therapy, including MRI, optical imaging, and radionuclide imaging techniques such as PET and SPECT [10–14]. Reporter genes with optical signatures (e.g., fluorescence and bioluminescence) are low-cost alternatives for real-time analysis of gene expression in small-animal models. Fluorescence imaging uses a fluorescent protein, such as green fluorescent protein (GFP), that is excited with external illumination, and the emission is subsequently detected [15]. GFP encoding cDNA can easily be included in the myriad of therapeutic vectors and serve as a monitoring tool for gene therapy. Nevertheless, excitation and emission wavelengths in the range of 500 nm (e.g., GFP) have limited penetration in mammalian tissues (1–5 mm). Since mammalian tissues absorb light that is used to excite these fluors, the tissues also fluoresce when excited at these wavelengths. The combination of absorption of specific signal and autofluorescence of tissues can result in a poor signal-to-noise ratio [16]. Recently, red-shifted mutants of GFP (RFP) have been shown to have an advantage over GFP in that red light penetrates tissues more efficiently than green [17]. A newer approach to fluorescence imaging of deeper structures uses fluorescence-mediated tomography [18]. The subject is exposed to continuous wave or pulsed light from different sources in an imaging chamber and detectors arranged in a spatially defined order capture the emitted light. Mathematical processing of this information results in a reconstructed tomographic image. Fluorescence mediated tomography is still in its infancy and requires extensive mathematical validation prior to routine implementation.

Bioluminescent photoproteins, such as luciferase, have been used as reporter proteins in living animals [10, 19, 20]. The most common bioluminescence reporter genes are firefly luciferase (fluc) and renilla luciferase (rluc). (Note, in general a lowercase abbreviation refers to the gene while an uppercase abbreviation refers to the protein.) The fluc (550 aa, 62 kDa) from the North American firefly, Photinus pyralis, is one of the best studied luciferases due to its high quantum yield (> 88%). Light is produced by catalyzing the oxidation of its small molecule substrate, beetle D-luciferin (benzothiazole), in an ATP-dependent process [21]. Another well-studied luciferase, rluc, is from the sea pansy, Renilla reniformis (311 aa, 34 kDa). RLuc utilizes coelenterazine as a substrate and emits light with a peak at 480 nm in a process that does not require ATP. RLuc has a low enzymatic turnover and quantum yield (6%) [22]. Gaussia luciferase (gluc) is from the marine copepod Gaussia princeps. GLuc (185 aa, 19.9 kDa) is the smallest luciferase known and is naturally secreted [23]. This luciferase emits light at a peak of 480 nm with a broad emission spectrum extending to 600 nm. RLuc and GLuc induce flash kinetics within the first 10 s that rapidly decay with time, whereas FLuc has glow kinetics [19, 23]. The bacterial luciferase gene (lux) operon encodes five gene clusters, lux C, D, A, B, and E. Lux A and B control luciferase enzyme expression, while lux C, D, and E control fatty aldehyde enzyme complex production, which synthesizes the substrates. Because the lux operon encodes all of the proteins necessary for light-emitting systems to function in bioluminescent bacteria, including luciferase, substrate, and substrate-regenerating enzymes, bacteria that express the lux operon only require oxygen and do not require an exogenous substrate to produce bioluminescence [9, 24]. The advantage of bioluminescence is its minimal background noise, since luciferase is not a natural constituent of mammalian organisms. Bioluminescence-based approaches currently lack detailed tomographic information and are limited to relatively small animals [9, 25, 26].

The advantage of MR for imaging gene expression is its excellent three-dimensional spatial resolution (tens of millimeter range). Owing to the indirect nature of the enhancement produced by MR contrast agents, much higher concentrations of injected material (10 ∼ 100 µM concentrations and higher) are generally necessary to produce sufficient image contrast [10, 13]. The low sensitivity often entails long imaging times, and consequently data acquisition is slow [13]. While magnetic resonance spectroscopy (MRS) does not usually produce three-dimensional images, this technique does provide accurate measurements of gene expression over short time frames and may eventually be harnessed to produce true spatial images, although at a much poorer spatial resolution than MRI. Radionuclide imaging with PET and SPECT has been used to characterize enzyme activity, receptor/transporter status, and biodistribution of various radiolabeled substrates (tracers) [27]. For these reasons, the most significant progress has been made using these techniques to image gene therapy, allowing the monitoring of gene delivery and the identification of therapeutic and/or reporter gene expression in living subjects. While the sensitivity of PET imaging is high (as little as 10⁻¹¹ ∼ 10⁻¹² M of tracer can be detected) and the speed of imaging is relatively rapid (minutes), these techniques lack micrometer spatial resolution (1–2 mm with micro-PET) [10, 12]. An alternative approach to PET is SPECT imaging. While the sensitivity of the single-photon system is intrinsically about one to two orders of magnitude less than the PET systems, the required radiopharmaceuticals and imaging systems are more readily available. Further details on the instrumentation available and the relative advantages of the various types of imaging instrumentation may be found elsewhere [10, 11, 28].

Biological Gene Delivery

Many forms of microorganisms have evolved to infect cells effectively and stably while evading host immune responses. Furthermore, many such microorganisms are tolerated by the immune system, including commensal bacteria in the gut and transfusion-transmitted viruses in the liver. The “ideal” biological GDV should have the appropriate packaging size for cargo, the ability to evade immune recognition, target cell-specificity, and efficient cargo delivery. The requirements at each of these steps differ between applications, and thus different GDVs are likely to be required. Although the most-developed GDVs in this class, such as viruses, show great promise for a wide range of disease applications, they have the aforementioned limitations. On the other hand, unconventional biological GDVs with unique properties might fill therapeutic niches poorly served by mainstream viral and liposomal GDVs (Table 1).

Table 1.

Comparison of selected parameters of the main vector types

	Viral vectors	Bacterial vectors	Bacteriophages	Minicells
Natural cancer targeting	±	+++	±	±
Safety	+	+	+++	+
Efficiency	+++	+	+	+
Low production costs	+	+++	+++	+
Simple production	+	+++	+++	+
Simple delivery	++	+++	+	+++
Amount of cargo molecule	++	+++	+	+++

Any bacterium suitable for use as a GDV must possess several features. It must (1) be genetically tractable, (2) be nonpathogenic so it is well-tolerated by the host, (3) exhibit preferential replication and accumulation within tumors, (4) be motile and nonimmunogenic, and (5) be susceptible to antibiotics so it can be cleared from the body following gene delivery [29]. As bacterial vectors satisfy the above criteria, two strategies are emerging depending on whether the bacteria is used as a protein or gene delivery tool (Fig. 1a, b) [30].

Fig. 1a–g.

Production of biological gene delivery vehicles labeled with molecular imaging probes. a Strains of bacteria with desirable properties are transformed with the plasmid cargo. Imaging reporter genes such as the lux operon are transduced into the bacteria. b The bacteria are cotransformed with the plasmid encoding the therapeutic gene and the imaging reporter gene. c Docking of bacteria with functionalized multiple-sized nanoparticles through biotinylated antibodies and surface-antigen interactions (microbot). Streptavidin-coated nanoparticles can carry biotinylated cargo. d Minicells are derived from a minCDE-chromosomal deletion mutant of Salmonella enterica serovar Typhimurium (S. typhimurium). Target genes are incubated with minicells overnight. Bispecific antibody (BsAb) is used to target recombinant minicells to tumor cells. One arm of these antibodies recognizes the O-polysaccharide component of the minicell surface lipopolysaccharide and the other a tumor-preferential cell surface-receptor, such as the epidermal growth factor receptor (EGFR), which is overexpressed in several cancers. e The coat proteins of the bacteriophages can be engineered to incorporate targeting ligands. Phage nanoparticles with multiple peptides engineered for different functions can then be produced with phagemid technology to enhance gene delivery efficiency. f Retrovirus. g Adenovirus

The basic idea of gene delivery is that transformed bacterial strains deliver genes encoded on a eukaryotic plasmid into cells where they are expressed as therapeutic gene products. The delivery process might involve intracellular localization of the bacteria (Fig. 2b). In another approach, bacteria are not used for gene transfer, but rather persist in the target tissues and produce the therapeutic proteins in situ. The transgene can be expressed in the cytoplasm after the bacteria penetrate the cells by prokaryotic transcription and translation machinery. Bacteria that do not enter host cells express the transgene in the intercellular space. In this case, extracellularly expressed protein may need to be fused to a cell permeable peptide (CPP) in order to penetrate the cell membrane (Fig. 2a).

Fig. 2a–g.

Intracellular delivery of cargo by delivery vehicles. a Extracellular secretion of protein by bacteria, such as E. coli or S. typhimurium, which are defective in ppGpp synthesis. Proteins made by bacteria must be engineered with a leader secretory signal sequence (S) for extracellular (bacteria) secretion. For intracellular transport, bacterial proteins need to be fused to cell permeable peptide (CPP) sequences. b Intracellular delivery of genes/proteins by bacteria such as S. typhimurium, S. typhi or E. coli expressing the invasion of Y. pseudotuberculosis. The bacteria invade host cells and remain in the vacuole. There they die due to metabolic attenuation and release their expression plasmid or protein. By an unknown mechanism, the plasmids cross the vesicular membrane and reach the cell nucleus of the host cells where they are expressed. c Bacteria enter cells via induced phagocytosis. Bacterial toxin causes endosomal compartments to disintegrate. The therapeutic cargo then separates from the bacterium and is delivered to the nucleus. d Minicells carrying the target gene, such as siRNA, bind biomarkers on the surface of cancer cells via BsAb and enter the cell by endocytosis. After endocytosis, the minicells traverse the well-established early and late endosomal pathways, terminating in acidified organelles, the lysosomes, where they are degraded and release their cargo. e Bacteriophage nanoparticles carrying target molecules bind biomarkers on the surface of cancer cells via a targeting peptide and enter the cell by endocytosis. After endocytosis, the bacteriophages traverse the well-established early and late endosomal pathways, terminating in acidified organelles, the lysosomes, where they are degraded and release their cargo. f Retroviral gene transfection. g Adenoviral gene transfection

Bacteria

That bacteria colonize tumors and that there is some beneficial effect of bacterial infection in cancer cells was first recognized by Dr. William B. Coley over 100 years ago [31]. Since then, oncolytic strategies using various bacterial strains, such as Salmonella typhimurium (S. typhimurium) [32–36], E. coli [24, 37], Clostridium [38], Bifidobacteria [39], and Listeria [40-44], have been widely explored as alternatives to traditional cancer therapies such as radiotherapy or chemotherapy. These bacteria reportedly have a propensity to naturally accumulate and replicate in a wide variety of solid tumors [45]. For cancer therapy, tumor-targeting bacteria have been applied alone [36] or in combination with conventional therapeutics [46] and have also been manipulated to deliver therapeutic molecules [47]. In particular, S. typhimurium has been bioengineered to generate bioluminescence [9, 48] or fluorescence [49] signals and has been employed to monitor bacterial migration to tumors in living small-animal models. These data may facilitate the prediction of the therapeutic efficacy of bacterio-therapy through visualization of bacterial accumulation and replication in specific organs.

Hoffman and colleagues have exploited the nutritional needs of S. typhimurium to target and kill tumor cells [32, 35, 36] by selecting strains of S. typhimurium that are dependent on the presence of the amino acids leucine and arginine for growth. They also reisolated S. typhimurium from the tumor after infection of human colon cancer cells in nude mice and demonstrated that the reisolated bacteria had increased tumor-targeting capability in vivo as well as in vitro [32]. These S. typhimurium strains were used to treat many types of metastatic cancers in mouse models, and tumor regression was observed after intravenous (IV) administration. In these experiments, GFP-expressing, attenuated S. typhimurium strains were successfully monitored after IV injection into various mice tumor models [35]. This approach had several advantages for imaging living animals [50]: (1) it produced a strong fluorescent signal that could be used to image unrestrained animals, and (2) the use of different types of fluorescent probes enabled multi-color imaging. However, because excitation is required, the applicability of this technique depends on the number and location of the bacteria in the animal [50]. This limitation, however, can be overcome using skin-flap techniques. For example, the spatiotemporal dynamics of bacterial infection following oral administration have been reported in great detail using this approach [49].

In an interesting application, GFP-labeled bacteria have also been used to screen for tumor-expressing promoters from a huge library. A population of Salmonella containing a random library of 14,028 genomic DNA fragments cloned upstream of a promoterless gene for GFP was IV injected into tumor-free nude mice and into human PC3 prostate s.c. tumors in nude mice to screen for candidate promoters preferentially induced in bacteria growing in tumors [51]. This strategy led to the identification of the pflE and ansB promoters, which are known to be induced in hypoxic conditions, as well as many other candidate promoters not related to hypoxia.

Our laboratory developed bioluminescent Salmonella and E. coli using the bacterial lux operon to visualize bacterial location in infected mice. The lux operon was integrated into Salmonella’s chromosome using Phage22 transduction. The bioluminescence emitted from the bacteria was detected in various tumor models after IV bacterial injection using a cooled-CCD camera. The lux operon can also be cloned into a plasmid, and transformed bacteria will generate light [9, 24]. However, using this strategy, E. coli fails to maintain the pLux expression plasmid in the absence of selection, particularly in infected animals [48]. Therefore, we employed a balanced-lethal host-vector system that coupled the expression of the Lux operon to the gene for aspartate β-semialdehyde dehydrogenase (asd), a key enzyme in the biosynthetic pathway of diaminopimelic acid (DAP), which is an essential component of the peptidoglycan of all Gram-negative and certain Gram-positive bacteria [52]. E. coli asd mutants require DAP and undergo lysis in its absence. In order to select bacteria that retained pLux in vivo, the chromosomal copy of asd was mutated in the host strain, resulting in viability only of strains that retained the Asd⁺pLux construct. The procedure for constructing the asd mutant of wild-type MG1655 was based on a method previously reported by Datsenko and Wanner [53]. In addition to strategies for maintaining exogenous genes, strides have also been made in increasing expression of these light emitting proteins and thus enhancing their detectability. For example, Riedel et al. have improved a system for luciferase tagging Listeria monocytogenes using a highly active, constitutive promoter that results in 100-fold higher luciferase activity in broth than any native promoter tested. This allowed for imaging of lux-tagged L. monocytogenes in food products, during murine infection, and in tumor-targeting studies [54].

In addition to the optical reporter genes, a reporter gene for radionuclide imaging was also explored. The HSV1-tk gene has been utilized as an effective reporter gene for nuclear imaging using PET or a gamma camera. Bermudes and colleagues showed that VNP20009 Salmonella constitutively expressing the HSV1-tk gene could be detected noninvasively by [¹²⁴I]FIAU PET in xenograft tumor models. These data demonstrated the efficiency and duration of vector targeting as well as indicated the location of tumors. Moreover, the ability to detect Salmonella vectors noninvasively by PET imaging has potential for use in clinical settings and could aid in development of these vectors by demonstrating the efficiency and duration of targeting as well as indicating the locations of tumors [55, 56].

Recently, MRI has also been utilized to image bacteria in living small animals. Benoit et al. reported that the magnetotactic bacteria Magnetospirillum magneticum AMB-1 could produce positive MRI contrast and colonize mouse tumor xenografts, providing a potential tool for MRI visualization of bacteria in preclinical and translational studies to track cancer despite its lower sensitivity. Following IV injection of ⁶⁴Cu-labeled AMB-1, PET imaging also revealed increased bacterial colonization of tumors, but decreased colonization of other organs, such as liver and spleen, after 4 h [57].

For therapeutic applications, tumor-targeting bacteria have been transformed with plasmids encoding therapeutic genes [58, 59]. A major problem hindering full exploitation of therapeutic gene-expressing Salmonella in cancer treatment is its toxicity in nontumoral reticuloendothelial organs, mainly the liver and spleen. Studies have shown that when injected through the tail vein, the majority of Salmonella initially localizes in those organs (Fig. 3a) [24, 48]. Therefore, it is necessary to establish an inducible system that allows complete, deliberate control over the expression of a gene of interest in the Salmonella transformants after host infection. Loessner et al. employed the inducible promoter P_BAD from the arabinose operon of E. coli to control expression of reporters such as GFP, FLuc, or Lux in vivo in the S. typhimurium strain SL7207 (hisG, ΔaroA), which can be activated by the sugar L-arabinose [60]. We also confirmed that RLuc8 expression was restricted in subcutaneously implanted tumors using Salmonella transformed with pBAD-rluc8 (Fig. 3b). To advance this system, we engineered attenuated Salmonella typhimurium that was defective in ppGpp synthesis (strain ΔppGpp) to carry cytotoxic proteins (cytolysin A) and express reporter genes [61]. In this system, we transformed Salmonella with a plasmid encoding a cytotoxic protein (cytolysin A) under the control of P_BAD. Subsequently, the bacterial luciferase (lux) operon from S. typhimurium-Xen26 (Xenogen-Caliper) was transduced in the transformed Salmonella. Engineered anti-tumoral bacteria labeled with imaging molecules allowed determination of bacterial fate in a simple, noninvasive manner that was amenable to repeated observation.

Fig. 3a, b.

Specific gene expression in tumors by bacterial vectors. a A bacterial expression plasmid was constructed in which a Renilla luciferase variant (RLuc8) was placed under the control of a constitutive promoter (pLac-RLuc8). S. typhimurium defective in ppGpp synthesis (strain ΔppGpp) was transformed with the pLac-RLuc8 plasmid. The transformed bacteria were IV administered into immunocompetent BALB/c mice bearing CT-26. Due to early distribution of bacteria in the liver and spleen, bacterial bioluminescence was produced not only in the tumor, but also in the liver and spleen. b To verify that remote control of the bacterial gene expression is possible, a bacterial expression plasmid was constructed in which a Renilla luciferase variant (RLuc8) was placed under the control of the P_BAD promoter (pBAD-RLuc8). The mutant S. typhimurium ΔppGpp strain was electrotransformed with the pBAD-RLuc8 vector. The transformed bacteria were IV administered into immunocompetent BALB/c mice bearing CT-26. RLuc8 expression was induced with L-arabinose (60 mg) at 4 dpi. Bioluminescence was detected in the tumors after L-arabinose administration, but not in its absence. Bioluminescence was observed only in the tumor and not in the liver or spleen because the bacteria were cleared from these organs (liver, spleen) at 4 dpi

Nanoparticles and bacteria have also been used independently to deliver genes and proteins into mammalian cells to monitor or alter gene expression and protein production [62]. In addition, Akin et al. reported the simultaneous use of nanoparticles and bacteria to deliver DNA-based model drug molecules in vivo and in vitro. In this approach, cargo (in this case, a fluorescent or a bioluminescent gene) is loaded onto the nanoparticles, which are carried on the bacterial surface (Listeria monocytogenes) (Fig. 1c). When incubated with cells, the cargo-carrying bacteria (“microbots”) were internalized, and the genes released from the nanoparticles were expressed (Fig. 2c). Mice injected with microbots also successfully expressed the genes, as seen by luminescence in different organs. This new approach may be used to deliver different types of cargo into live animals and a variety of cells in culture without the need for complicated genetic manipulation.

Bacteria-like Particles

Minicells, or bacteria-like particles, have been produced by derepressing cryptic polar sites of cell fission through inactivating genes controlling normal bacterial cell division [63]. These minicells can be packaged with therapeutically relevant concentrations of a range of therapeutics and selectively targeted to cancer cells via bispecific antibodies (BsAbs) (Fig. 1d) [64, 65]. A two-wave cancer treatment has been explored in which minicells are specifically and sequentially delivered to tumor xenografts. First, siRNA- or shRNA-encoding plasmids are delivered to compromise drug resistance by knocking down a multidrug resistance protein. Subsequently, targeted minicells containing cytotoxic drugs are administered, which then eliminate the formerly drug-resistant tumors (Fig. 2d) [66]. Monoclonal BsAbs or anti-O-polysaccharide was labeled with Alexa Fluor-488 or Fluor-647 to visualize minicells invasion into cells in vitro. In vivo visualization may also be possible by conjugating the minicells with a near-infrared fluorophore.

Bacteriophages

Bacteriophages are natural viruses that exclusively infect bacteria and are ingested without serious side effects through the consumption of fermented food, such as yoghurt. Development of phagemid vectors, plasmids with a bacteriophage origin of replication that are packaged into phage particles upon the addition of helper phages, has made the genetic manipulation of bacteriophages as easy as manipulating plasmids used in the production of viral vectors [67]. The phagemid system allows for precise control over the relative composition of wild-type and modified coat proteins [68] while eliminating potentially immunogenic genes encoding bacteriophage proteins. The bacteriophage coat proteins can be engineered to incorporate targeting ligands without significantly affecting structure. Phage nanoparticles with multiple peptides engineered for different functions can then be produced with phagemid technology to enhance gene delivery efficiency [47]. Based on these technologies, disease-specific phage library-driven fluorescent probes could be developed for use in the early detection of cancers by optical imaging systems (Figs. 1e and 2e). For example, Newton et al. selected bacteriophages in vivo labeled with a near-infrared fluorophore (AlexaFluor 680) that displayed affinity to prostate cancer [69]. A tumor-specific bacterial phage was also generated by displaying up to five copies of melanocyte-stimulating hormone (-MSH) peptide analogs on the phage surface. To visualize these phages, streptavidin was conjugated to diethylenetriaminepenta acetic acid, which enabled radiolabeling with ¹¹¹In [70] and hence radionuclide imaging.

Viruses

Over the last century, many viral types, including adenovirus, herpes simplex virus, Newcastle disease virus, myxoma virus, vaccinia virus, and vesicular stomatitis virus, have been studied intensively as novel agents for cancer treatment [71]. Targeted gene expression can be obtained via different strategies, such as exploiting tissue/tumor-specific promoters that only activate the gene in certain tissues or conditions [72, 73], the use of antibody against tumor-specific antigen to modify tropism and redirect the virus to cancer cells [74-77], and natural or artificial modification of conditionally replicating viruses [78–81]. These techniques are largely thought to limit undesired spreading of the virus to normal organs. The tropism of viral vectors can be visualized by encoding imaging reporter genes in the viral genome. Recombinant vaccinia virus-mediated expression of the human norepinephrine transporter (hNET) allowed deep tissue imaging by radiolabeled metaiodobenzylguanidine (MIBG) [82]. In addition to hNET, the Na/I symporter (hNIS) has also been commonly used in adenovirus [83] or in measles virus Edmonston [84] as an imaging reporter gene to monitor viral gene therapy.

Tumor-specific promoters have been actively investigated as a means to confine gene expression to tumors and limit undesired gene expression in other normal organs. A combination of transcriptional targeting and noninvasive imaging could improve the safety of viral-based therapeutic approaches in cancer. For example, prostate-specific adenoviral vectors encoding firefly luciferase produced a robust signal in prostate cancers. Repetitive imaging over a 3-week period after IV injection into tumor-bearing mice revealed that the virus could locate and illuminate metastases in the lung and spine [85]. To enhance the expression of a prostate-specific promoter, Rubinchik et al. placed the tetracycline transactivator gene under the control of a prostate-specific ARR2PB promoter, and a mouse Tnfsf6 (encoding FASL)-GFP fusion gene under the control of the tetracycline responsive promoter. A single construct with both expression cassettes carried by adenovirus showed that GFP expression was restricted to prostate cancer cells [86]. Sato et al. developed a two-step transcriptional amplification (TSTA) method using an adenoviral vector by utilizing the prostate specific promoter to drive a potent GAL4VP16 transcriptional activator, which in turn binds to tandem repeats of the GAL4 binding site that activates the secondary reporter or therapeutic gene. Analysis of real-time gene expression by both optical imaging and the combined modality of PET/CT showed that this approach enhanced the weak native PSA promoter up to 1,000-fold and was even stronger than the robust expression driven by the viral cytomegalovirus promoter [87].

Conclusions

The biological GDVs discussed above, although less well established than viral vectors and liposomal delivery agents, present tantalizing opportunities for niche therapeutic applications and provide a rich opportunity for further research and development. Besides their intrinsic advantages, biological GDVs can also benefit tremendously from the development of tangential technologies. For example, bacteria and bacteriophage production can benefit from the research and industrial experience of the vaccine production and food probiotic industries. Given the state-of-art molecular imaging techniques that are available, the barrier between bench and bedside applications is likely not as high as is perceived; hence the development of biological GDVs through molecular imaging techniques probably deserves greater attention from the gene therapy community.