Evaluation of adenovirus capsid labeling versus transgene expression

Adenoviral (Ad) vectors have been utilized for a variety of gene therapy applications. Their utilities are attributed to the unparalleled efficiency of gene transfer in both in vitro and in vivo contexts [1, 2]. Our group, along with others, have incorporated imaging reporters of either bioluminescent [3] or fluorescent nature [4‐8], as well as suicide genes within the adenovirus genome as a transgene for analytical and/or therapeutic purposes. These molecules have also been incorporated as capsid components [3, 8]. Determining the best locale for imaging modalities and/or therapeutic genes could influence the design of Ad and conditionally replicative adenoviruses (CRAds) for monitoring of viral replication, gene transfer, and biodistribution thus improving these vectors for clinical applications.

Recognizing reporter transgene incorporation or capsid incorporation yields potential advantages and disadvantages; our report compares and evaluates the benefits of imaging via transgene incorporation versus imaging via capsid incorporation. In this regard, when interpreting CRAd imaging end point data the results are often based on detection of virus-encoded expression opposed to direct viral particle analyses [9]. Imaging of viral infection via transgene expression from the early region 3 (E3) of replication-competent Ad is dependent on cells producing viral progeny due to activation of transgene by the E3 promoter [10‐13]. Therefore, one speculated disadvantage of imaging through transgene expression is that transgene imaging is thought to be less accurate with respect to CRAd biodistribution, progeny production, and virus accumulation in tumors [9]. On the other hand, one potential advantage of capsid-incorporated reporter imaging is that capsid based reporter imaging is thought to be more accurate with respect to direct particle localization as well as imaging capacity seen in combination with gene expression [9]. One potential advantage of reporter imaging within Ad E1 or E3 is that transgene expression allows the incorporation of complex imaging reporters, whereas in some cases the capsid loci (i.e. protein IX [pIX]) may not be compatible. Capsid incorporation of the reporter gene must be compatible with the pIX loci and subsequent CRAd capsid assembly. In the event, that ligand incorporation is not compatible with pIX, the resulting CRAds might have defective pIX particles and/or capsids. These resulting virus progeny could have reduced virus stability in addition to being temperature sensitive and/or non-infectious [14, 15].

In order to evaluate capsid incorporated imaging versus transgene imaging, we have genetically incorporated firefly luciferase (Luc) as a transgene within the deleted E3 region of wild type Ad, or at the 3'-end of minor capsid pIX gene, respectively. In order to determine the benefits of moiety expression at pIX versus that of E3, the Luc protein was expressed under the control of the native promoters at either locale. We examined viral production, virus replication, Luc expression, and activity with these constructs in vitro and in vivo.

Ad vectors have been used for a variety of therapeutic applications. This study compares in vitro and vivo imaging of luciferase protein following Luc incorporation on the capsid protein IX or as a transgene within the deleted E3 region of Ad. We demonstrated that when tested in vitro or in vivo, both viruses express functional luciferase protein at either the pIX or E3 locale. In addition, we showed that throughout the in vivo imaging study the Luc expressed under the control of the E3 promoter yields higher reporter gene activity compared to that of Luc expressed in the pIX loci. In the in vivo imaging study the signal magnitude difference between the Luc activity from Luc incorporated within the E3 locale or the pIX locale was markedly higher on imaging days one through four.

Adenovirus has been exploited for cancer gene therapy by means of viral particle monitoring by incorporating imaging modalities within the Ad genome [20‐25]. Traditionally, these molecules have been incorporated into the genome as transgenes, typically within the E1 region [26‐29]. Also along these same lines, to establish the safety of oncolytic viruses imaging modalities have been incorporated within the wild type or CRAd genomes. For example, Ono et. al., incorporated the gene, which encodes for enhanced green fluorescent protein (EGFP) in the deleted E3 region of a wild type Ad. This report demonstrated that strong EGFP fluorescence was detected in these viral-infected cells in a replication-dependent manner. Through a series of analyses, this report conveyed that EGFP, controlled by the Ad major late promoter, provides a valuable tool whereby noninvasive imaging can be accomplished to monitor Ad replication for preclinical uses and ultimately human applications. CRAds were envisioned and proposed for cancer gene therapy as an alternative for surgery, radiation and chemotherapy [30‐32], however; to date the use of CRAds or conventional therapies as single agents to combat cancer have showed limited efficacy for cancer therapy [33‐35]. In this regard, researchers have employed a series of combination therapies which utilize CRAd agents in combination with conventional therapies (i.e., surgery, radiation, chemotherapy, and cell therapies) to yield improved pre-clinical and clinical cancer therapy [36‐41]. With respect to the clinical use of CRAds, there is speculation as to whether transgene expression could provide endpoint data related to viral replication, spread, tropism specificity, viral persistence, and virus-host cell interaction. In this regard, researchers had begun to attempt to improve on Ad monitoring systems, thereby labeling capsid particles with imaging modalities. Many groups have incorporated imaging modalities in capsid locales such as pIX or pV [3, 6, 42, 43]. For instance, our group as well as Meulenbroek and colleagues have demonstrated the feasibility of incorporating the fluorescent moiety EGFP within the adenovirus capsid pIX. These studies illustrated that labeled particles allow qualitative assessments of viral particle localization within cells in vitro as well as in vivo[4, 5].

Our more recent studies demonstrated that we could incorporate herpes simplex virus type 1 thymidine kinase (HSV-tk) at the pIX locale whereby it could metabolize conversion of substrate permitting an imaging signal. This capacity allowed assessments of CRAd parameters in vivo related to viral persistence. Based on these findings, we sought to explore the full potential of the capsid incorporation approach for utility in CRAd imaging analysis. Along those same lines, we incorporated a fusion of HSV-tk-Luc within the Ad pIX. This study was perform in a non-replicative Ad, however, we were able to demonstrate functional HSV-tk and luciferase activity in an in vitro and in vivo context [3]. This study illustrated dynamic imagining in the context of our capsid-incorporated platform, and will be transitioned to a CRAd context.

Fluorescent and other imaging modalities have been tested at the capid or transgene loci, respectively, but very little information has been acquired to compare an identical modality at multiple sites within the Ad genome [44]. Therefore, we sought to compare transgene expression of Luc versus that of capsid-incorporated Luc under the control of the Ad native promoters. Our data indicates that in vitro DNA replication rates and in vitro DNA replication levels of Ad-wt-E3-Luc and Ad-wt-pIX-Luc were comparable to the Adwt vector (Figure 1). In addition, the in vitro DNA replication rates and in vitro DNA replication levels of Ad-wt-E3-Luc and Ad-wt-pIX-Luc were comparable to one another (Figure 1). This is an important finding in that, this particular capsid modifications or transgene modification does not appear to dramatically impair virus replication.

Our Western blot analyses indicate that the Luc protein is expressed from the Ad-wt-E3-Luc virus at its expected molecular mass (Figure 2B). Western blot analyses indicate that the capsid associated Luc is effected by capsid incorporation into the pIX locale. Our pIX-modified Ad only expresses the pIX fusion protein since the native pIX gene has been replaced with the modified pIX gene. Our protein analysis of pIX-Luc indicates that a truncated version of pIX-Luc is being produced after infection into A549 cells (Figure 2). Results from our laboratory also demonstrate a similar finding with respect to proteolytic pIX cleavage products observed after protein analysis of various viruses (Ad5-wt-IX-EGFP, Ad5-wt-IX-mRFP1, and Ad5-wt-IX-mRFP1-E3-V-EGFP) containing pIX conjugated fluorescent tags [13]. The Ad genome encodes a gene for cysteine protease that recognizes consensus sequence motifs (M,I,L)XGG/X and (M,I,L)XGX/G contained in precursor proteins, where X is any amino acid. This protease cleaves the residue at the site of "/"[18, 19]. The adenovirus protease plays a role in protein maturation of adenoviral proteins by cleaving precursors of IIIa, VI, VII, μ and terminal proteins [18, 19, 45]. We found that the protein sequence for Luc contains a few putative cleavage sites for Ad protease. Being that Luc is a universally utilized maker for in vitro and in vivo applications; its utility in a capsid-incorporated context is informative. To further confirm that, Ad protease is involved in the cleavage of pIX-Luc we analyzed protein from a stable cell line expressing pIX-Luc. These results confirmed that the expression of full-length pIX-Luc (Figure 3), therefore the proteolytic cleavage seen from cells infected with Ad-wt-pIX-Luc is likely a result of Ad protease (Figure 2A-C). The Luc expressed from the E3 region is not affected by Ad proteases (Figure 2B), Luc expressed from the E3 region is localized in the cytoplasm, so this soluble form is not cleaved by Ad protease [46]. However, pIX-Luc is relocalized from the cytoplasm to the nucleus during viral assembly. Ad cysteine protease localization is nuclear; the protease activity is observed in the nucleus fraction.

We would note that Ad-wt-pIX-Luc yields direct functional activity of incorporated Luc protein as expected (Figure 4). It is important to note that the truncated version of pIX-Luc is capable of being assembled within the viral capsid (data not shown) and able to generate direct in vitro Luc enzymatic activity in the presence of Luc substrate and ATP (Figure 4). We speculate that the truncated pIX-Luc must contain the enzyme active site allowing for Luc activity. In contrast, Ad-wt-E3-Luc needs to be infected within cells to generate functional Luc protein and activity (Figure 5). At 48 and 72 h.p.i., there appears to be a significant difference between relative light units observed after infection with Ad-wt-E3-Luc as compared to that of Ad-wt-pIX-Luc (Figure 5).

Due to the fact that the Luc protein is expressed on the pIX capsid and is constitutively active on the viral capsid in the presence of substrate and ATP, we expected that imaging observed on day 0 would yield substantially higher luciferase activity in tumors injected with Ad-wt-pIX-Luc as compared to that of Ad-wt-E3-Luc (Figure 6). On day 0, in vivo Luc signal generated from tumors injected with Ad-wt-pIX-Luc was similar to that of Luc signal generated from tumors injected with Ad-wt-E3-Luc. This finding may be due to the lack of signal intensity. In this instance, it might be possible to distinguish pIX-associated signal with a stronger imaging molecule such as HSV-tk [8]. It is also plausible that the number of pIX-Luc molecules incorporated on the Ad viral particles may not be substantial enough to generate a higher signal in vivo as compared to the Ad-wt-E3-Luc virus, even though in our in vitro study we could observe substantial Luc activity in the direct in vitro context after analysis of direct Ad-wt-pIX-Luc particles (Figure 4). It is likely that pIX-Luc incorporation can vary from batch to batch with each preparation of Ad-wt-pIX-Luc (data not shown). Therefore, it is likely that the pIX-Luc integrity can be improved through additional virological methods or other molecular methods such as the addition of linkers.

In vivo, we observed that tumors injected with the Ad-wt-E3-Luc virus yielded higher Luc counts/second compared to tumors injected with Ad-wt-pIX-Luc. Although, the difference between the signal magnitudes between groups was not statistically significant, we observed a trend of a higher signal magnitude when tumors were injected with Ad-wt-E3-Luc versus that of Ad-wt-pIX-Luc (Figure 6). The in vivo result was different from what was observed in A549 cells under in vitro conditions (Figure 5). The differential outcome between the in vitro and in vivo results confirms that often times these two systems does not actually mimic one another, due to the complexities of in vivo model systems (i.e. viral lateralization of virus in a monolayer cell system versus that of viral lateralization in a tumor model system).

Although, experimental conditions were maximally optimized significant differences may have been difficult to observe due to a variety of factors such as injection techniques, natural tumor heterogeneity, virus lateralization, and/or mouse sample size. In addition, we must comment that differences seen between the viruses are likely to be attributed to relative promoter activity. Differences seen with respect to luciferase activity might be affected because at the E3 locale, Luc is expressed most like its native form whereas Luc protein expressed at the pIX locale is conjugated to the pIX protein, possibly yielding a slightly diminished signal. When designing vectors to express imaging or therapeutic genes, limits of the gene its self and the locale its self are of the utmost importance.

We did not observe temporal differences between tumors injected with either virus, however temporal patterns observed in this study herein were similar to that seen with Ad-wt-pIX-monomeric red fluorescent protein 1 [16]. In our 2006 report, maximum signal seen with respect to tumors injected with Ad-wt-pIX-monomeric red fluorescent protein 1 was observed within 2 and 6 days and diminished around day 10 [16]. These results were also consistent with those seen in a clinical setting, generally there is a peak in circulating viral DNA detected which is typical observed over several days and diminishes over time to baseline, indicating viral clearance [47‐49]. We speculate that Luc signals would have returned to baseline levels on or about day 20, however due to excess tumor burden in a few the mice we concluded the experiment on day 15.

In this report, we demonstrate that both Ad-wt-pIX-Luc and Ad-wt-E3-Luc replicate comparable to one another and similar to other vectors in our laboratory (Figure 1). We demonstrate that Ad-wt-pIX-Luc and Ad-wt-E3-Luc express luciferase protein at either locale (Figure 2) and that functional luciferase activity is retained in vitro (Figure 4 and 5) as well as in vivo (Figure 6), whereby the virus expressing Luc within the E3 locale yields a higher result with respect to reporter readout as compared to the Ad-wt-pIX-Luc virus. Our group has attempted to optimize the Ad genome incorporation of therapeutic genes and reporter genes for improved Ad and CRAd virus readout and therapeutic efficacies. This study provides a road map forward for optimization of CRAd design. Although the luciferase signals generated from Ad-wt-pIX-Luc and Ad-wt-E3-Luc were not statistically different throughout the duration of the in vivo experiment (Figure 6), there was a consistent trend whereby Ad-wt-E3-Luc yielded a higher signal throughout the duration of the in vivo experiments. This trend observed whereby the E3 imaging is superior to capsid-incorporated imaging is important. In a clinical setting, the achievement of maximal signal threshold is necessary for sensitive orthotopic in vivo applications. This would be a clear advantage for expressing imaging motifs within the E3 locale. As it relates to the capsid-incorporated imaging strategy, one disadvantage associated with capsid-incorporated imaging, might be proteolysis associated with Ad precursor proteins/capsid-incorporation. Truncated protein observed in our study was not detrimental to our study, however this possibility must be considered thoroughly when designing vectors. An advantage to expressing imaging motifs within a capsid-incorporated locale is that direct virus particle locale (i.e. virus biodistribution) can be visualized. Our study herein, examined intratumoral imaging, thus this paradigm was not directly observed. However, we demonstrated that capsid-incorporated imaging was comparable to that of E3 imaging. Therefore, in order to achieve maximal imaging threshold or therapeutic efficacy, one likely approach may be the combination of incorporation of dual imaging modalities or therapeutic gene incorporation at multiple genome locales (i.e. E3 and pIX). In summary, if multiple parameters are desired, such as imaging readout and therapeutic efficacy, placement of the most critical modality within the E3 region is likely the best option.