Adenoviral vector construction

Replication incompetent E1- and E3-deleted Ad5CMV-EGFP and Ad5-ROBO4-EGFP vectors were created using a two-plasmid rescue method. The plasmids encoded expression cassettes comprised of the human cytomegalovirus (CMV) major immediate-early promoter/enhancer or the ROBO4 enhancer/promoter elements coupled to the enhanced green fluorescent protein EGFP gene, followed by the bovine growth hormone polyadenylation signal. These expression cassettes were cloned into a shuttle plasmid (pShuttle, Qbiogene, Carlsbad, CA) and confirmed using restriction enzyme mapping and partial sequence analysis. The shuttle plasmids were linearized with Pme I enzyme and integrated into the Ad5 genome by homologous recombination with pAdEasy-1 plasmid in E. coli strain BJ5183. Recombinant viral genomes were transfected into HEK293 cells using SuperFect Transfection Reagent (QIAGEN, Chatsworth, CA), where they were packaged into virus particles. Ad5CMV-EGFP and Ad5ROBO4-EGFP were propagated in HEK293 cells, purified twice by CsCl gradient centrifugation and dialyzed against 10 mM HEPES, 1 mM MgCl₂, pH 7.8 with 10% glycerol. The viral particle (vp) concentration was determined by absorbance of dissociated virus at A₂₆₀ nm using a conversion factor of 1.1×10¹² vp per absorbance unit. Viral titer also was measured by a 50% tissue culture infectious dose (TCID₅₀) assay. Briefly, HEK293 cells were plated into 96-well tissue culture plates at 5×10³ cells per well, and then serial dilutions of viral stock were added directly to the cells. Cells were incubated for 14 days, and relative cell density was determined using a crystal violet staining assay. Cell culture medium was removed and surviving cells were then fixed and stained with 2% (w/v) crystal violet (Sigma-Aldrich) in 70% ethanol for 3 hours at room temperature. The plates were extensively washed, air-dried, and optical density was measured at 570 nm using a V Max plate reader (Molecular Devices Corporation, Sunnyvale, CA). The number of wells with observable cytopathic effect per each row was determined. The viral titer was calculated by the Karber equation: T  =  10^1+D(S-0.5) × V⁻¹, where T is infectious titer in TCID₅₀ ml⁻¹, D is the log₁₀ of the dilution, S is the log₁₀ for the initial dilution plus the sum of ratios, and V is the volume in ml of the diluted virus used for infection. Adenoviral vectors with the viral particle concentrations 1.1−2.9×10¹² vp ml⁻¹, 1.3−2.0×10¹⁰ TCID₅₀ ml⁻¹ and 1∶84-1∶92 TCID₅₀ to vp ratios were used in presented study.

Ethics statement

The Animal Studies Committee of Washington University in St. Louis approved all procedures under protocol numbers 20120029 and 20110035. During surgery for orthotopic tumor implantation, animal pain and suffering was minimized by using inhalational anesthesia and postoperative analgesia.

Generation of composite mice

Rag2^−/− (Rag2^tm1.1Cgn) mice [13], in a mixed genetic background, were bred in house. Transgenic human coxsackie virus and adenovirus receptor (hCAR) mice on a mixed genetic background, likely C57Bl6/J and DBA [14], were obtained from Sven Pettersson. ROSAR26R (Gt(ROSA)26Sor^tm1Sor) knock-in mice were obtained in house. Rag2^−/− mice were serially intercrossed with R26R and hCAR transgenic mice to generate the composite mouse line, hCAR/wt:R26R/R26R:Rag-2^−/−, subsequently denoted as hCAR:Rag2^−/− mice. The warfarin liver detargeting experiments were performed using wt/wt:R26R/R26R:Rag2^−/− littermates. The R26R conditional LacZ alleles were designed for Cre recombination experiments but were not used in this study.

Creation of orthotopic and subcutaneous heterotopic tumors

Mice from 6 to 12 weeks of age were used in the present work. The 786-O human kidney cancer cell line was freshly obtained from ATCC and cultured in RPMI with 10% FBS with penicillin/streptomycin/amphotericin B. Xenograft tumors were established by injection of 5×10⁶ cells in 50 µl of RPMI media using aseptic technique. Kidney orthotopic tumors were established by left kidney subcapsular injection of 4×10⁶ 786-O cells in 40 µL of RPMI media. Carprofen, 5 mg/kg sc X 3 days, (Pfizer Animal Health, NY, NY) was used for postop analgesia. A total of 36 mice were both injected with 786-O cells and subsequently with Ad vectors (see below). As preliminary experiments revealed a 100% kidney orthograft “take” rate, bioluminescence imaging was not preformed in this study. Mice were sacrificed 4–6 weeks post 786-O injection when the subcutaneous xenograft tumors reached a diameter of 4 mm.

Ad vector injections, host organ, and tumor harvest

Mice harboring established subcutaneous and kidney tumors were tail vein injected with 5.0×10¹⁰, 1.0×10¹¹, or 1.5×10¹¹ viral particles of Ad5ROBO4-EGFP or Ad5CMV-EGFP in 200 µl of saline. For warfarin experiments, mice were administrated warfarin (5 mg/kg) dissolved in peanut oil subcutaneously on day −3 and day −1 prior to vector injection. Organs and tissue were harvested from mice anesthetized with 2.5% 2, 2, 2-tribromoethanol (Avertin, Sigma-Aldrich, St. Louis, MO) 72 hours post vector injection for all experiments.

Tissue harvest and immunofluorescent localization of reporter gene expression

Mice under Avertin anesthesia were perfused via the left ventricle with phosphate-buffered saline (PBS, pH 7.4), followed by 4% paraformaldehyde/PBS for whole body fixation. Mouse organs and tumors were collected, post-fixed in 4% paraformaldehyde for 2 hours at room temperature, cryopreserved in 30% sucrose for 16 hours at 4°C, and cryo-embedded in NEG50 (Thermo Fisher Scientific, Waltham, MA) over 2-methylbutane/liquid nitrogen. Sixteen-micrometer frozen sections were air-dried, washed in PBS, blocked with protein block (1% donkey serum in PBS containing 0.1% Triton X-100), and incubated with primary antibodies including: rat anti-endomucin, 1∶1,000, (#14-5851-81 eBioscience, San Diego, CA), Armenian hamster anti-CD31, 1∶1,000, (#MAB1398Z EMD-Millipore, Billerica, MA), and rabbit anti-GFP, 1∶400, (#A11122 Life Technologies, Carlsbad, CA). After PBS washes, the slides were incubated with corresponding Alexa Fluor 488 and Alexa Fluor 594, 1∶400, (Jackson ImmunoResearch Laboratories, West Grove, PA) conjugated secondary antibodies and counterstained for nuclei with SlowFade Gold Antifade mounting reagent with 4′,6-diamidino-2-phenylindole (DAPI) (Life Technologies). Fluorescence microscope images were collected using an FVII digital camera with Extended Focal Imaging (EFI) function (Olympus America, Center Valley, PA). To quantify the tissue section GFP fluorescence, the areas of GFP (+) cells and dual CD31/endomucin (+) blood vessels were measured and normalized by total tissue area per field. Imaging experiments were repeated 2–4 times on independent sets of vector-injected mice. Quantitative imaging experiments were based on the ratio of Ad5ROBO4 vector expression (green fluorescence emission from fluorophore tagged the anti-EGFP secondary antibody) divided by total vascular endothelial area (red fluorescence emission from the fluorophore tagged anti-CD31/endomucin secondary antibodies) using image analysis software (MicroSuite Biological Suite Version 5, Olympus), from 4–5 mice in each group.

Cultured cell, tissue, and whole organ protein expression analysis by immunoblotting

Anesthetized mice were perfused via the left ventricle with cold phosphate-buffered saline (PBS, pH 7.4) containing 1 mM PMSF (Sigma-Aldrich), organ tissues and tumors were snap frozen in liquid nitrogen, and stored in the liquid nitrogen vapor phase. Frozen tissues were pulverized using a liquid nitrogen-chilled mortar and pestle (Cell Crusher, Thermo-Fisher), and the crushed powder lysed on ice in radioimmunoprecipitation assay buffer (RIPA; 20 mM Tris-HCl (pH 7.6), 0.15 M NaCl, 1% sodium deoxycholate, 1% NP40, 1 mM EDTA, 1 mM EGTA) supplemented with Protease Inhibitor Cocktail, 1∶10, Sigma-Aldrich) for 30 minutes. The late log-phase 786-O cells grown on a 10-centimeter tissue culture dish was washed once with cold PBS, and lysed on ice with 500 µl cold RIPA buffer supplemented with protease inhibitors (see above). Protein lysates were separated on polyacrylamide gels and transferred to polyvinylidene difluoride (PVDF) membranes. Protein loading in individual lanes was normalized first to β-tubulin and then VE-Cadherin. Membranes were blocked in Tris-buffered saline (TBS, pH 7.6) containing 0.5% Tween 20 (TBST) and 5% nonfat dry milk and incubated in 5% BSA in TBST, containing the following antibodies: rabbit polyclonal anti-ROBO4 (generous gift from Dean Li, University of Utah), chicken anti-EGFP, 1∶1,000, (#A10262 Life Technologies), goat anti-VE-Cadherin, 1∶400, (#A1002 R&D Systems, Minneapolis, MN), and polyclonal anti-β-tubulin, 1∶20,000, (Novus Biologicals, Littleton, CO) overnight. Membranes were washed three times with TBST and incubated in TBST containing 5% milk with the corresponding IgG-horseradish peroxidase conjugate, 1∶5,000, (Santa Cruz Biotechnology, Santa Cruz, CA) for 1 hour. After three TBST washes, peroxidase activity was revealed by enhanced chemiluminescence using ECL2 or SuperSignal West Femto Western Blotting Substrate (both from Thermo Scientific) and imaged using a Chemidoc XRS imaging system (Bio-Rad Laboratories, Hercules, CA). The immunoblotting was quantified by densitometry with Quantity One one-dimensional analysis software (Bio-Rad Laboratories). Immunoblotting was repeated 2–4 times using extracts from independent experiments from separate sets of mice. Representative blots and densitometry analysis were presented for each group of immunoblots in each experiment.

Statistical analysis

Significance between groups in the differential fluorescent area experiments was determined using one-way ANOVA with Tukey's correction for multiple group comparisons (GraphPad Prism, San Diego, CA).

Endogenous ROBO4 in renal cancer xenografts and orthotopic tumors

While investigators agree that ROBO4 expression is restricted to ECs and bone marrow hematopoietic stem cells [15]–[17], ROBO4 expression level in normal organs and particularly in tumors is controversial [18]–[22]. The 786-O human kidney cancer cell line was chosen for our proof of concept tumor endothelial cell (EC) expression engineering because these cells form hypervascular tumors in immunodeficient hosts due to constitutive, high level expression of hypoxia-inducible factor-2, and its downstream target vascular endothelial growth factor [23], [24]. To determine differential expression of ROBO4 in this kidney cancer model, and to provide a backdrop for our investigation of the human ROBO4 enhancer/promoter fragment used in our Ad5ROBO4-EGFP vector, we tested ROBO4 protein levels in 786-O orthotopic (KO) and subcutaneous (SC) tumors compared to liver using a validated ROBO4 antibody (Figure 1) [25], [26]. Prior to immunoblotting, we noticed that the 786-O tumor vascular density appeared to be less than that of normal organs despite its hypervascularity in tissue sections stained with a CD31/endomucin cocktail (Figures 1A and 2). Indeed, image analysis revealed a 1.6- and a 1.7-fold lower vascular area density in KO and SC 786-O tumors, respectively, compared to liver (Figure 1B). Based on this vascular density analysis, we immunoblotted for the EC specific marker, VE-cadherin in our ROBO4 immunoblots and all subsequent Westerns to normalize for this differential tumor/host organ vascular density. Consistent with the tissue immunofluorescence, densitometric analysis of VE-cadherin immunoblotting revealed a 2.3- and 1.6-fold decrease in KO and SC tumor versus liver expression (Figures 1C and 1D). Liver was selected as our control normal organ because of its avid Ad vector uptake [27], [28]. To further test for ROBO4 endothelial specificity, 786-O cells grown in culture were lysed and run on the same immunoblots as livers and tumors obtained from the mice. Compared to prior reports suggesting robust ROBO4 upregulation in tumor angiogenesis [12], [17], [18], endogenous ROBO4 was very modestly upregulated 2.0-fold in KO, and 1.4-fold in SC 786-O tumors, compared to liver (Figure 1E).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Endogenous ROBO4 upregulation despite lower vascular density in orthotopic and xenograft tumors.

A. Immunofluorescence of the vascular endothelium in liver (upper panel) and 786-O human renal cell carcinoma (RCC) subcutaneous xenograft tumor (lower panel). B. Vascular area analysis of liver (Li), kidney orthotopic (KO) tumors, and subcutaneous (SC) xenograft tumors (n = 6 mice analyzed). C. Immunoblot of endogenous ROBO4 and the endothelial cell specific VE-cadherin from liver, kidney orthotopic and subcutaneous xenograft 786-O RCC tumors, and from the derivative 786-O cells grown in culture. D. Densitometry analysis of VE-cadherin/tubulin ratio from C mirrors the vascular area determination in B. E. Densitometry analysis of endogenous ROBO4 normalized to VE-cadherin expression reveals a 1.4- to 2-fold increase in SC and KO tumors compared to liver. C–E: Immunoblot and densitometry was repeated twice with two independent sets of protein extracts from two different tumor-bearing mice with essentially the same results. A. Magnification: 100X, Red: endomucin/CD31 antibody cocktail, Blue: DAPI. B. *p<0.05, one way ANOVA with Tukey's correction, mean ± SD.

https://doi.org/10.1371/journal.pone.0083933.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Adenoviral Type 5 (Ad5) vector expression in a host organ panel in tumor bearing immunodeficient Rag2^−/−, and hCAR:Rag2^−/− composite mice.

Host organ EGFP reporter expression following intravenous injection of 1.0×10¹¹ viral particles (vp) of either (A, B) Ad5ROBO4-EGFP (ROBO4) or (C, D) Ad5CMV-EGFP (CMV) vectors. A. Ad5ROBO4 vector expression in Rag2^−/− mice is widespread but focal in liver endothelial cells (ECs), and rarely detectable in single splenic microvessels. All other organs are negative for vector reporter expression. B. Multiorgan, EC restricted vascular expression is evident in Ad5ROBO4-injected hCAR:Rag2^−/− composite mice. C. Ad5CMV vector expression in Rag2^−/− mice is detectable in liver hepatocytes, sporadic splenic (inflammatory) and adrenal cells. D. Ad5CMV vector expression in hCAR:Rag2^−/− mice is complex, hepatocyte localized, but decreased in frequency compared to C in liver, mixed inflammatory and endothelial cell localized in spleen, and EC localized in all other organs including brain and skin. A and B: n = 5–6 mice, combined from 3–6 independent experiments, C and D: n = 3–4 mice combined from 3–4 independent experiments Magnification: 100X, Red: endomucin/CD31, Green: EGFP immunofluorescence, Blue: DAPI.

https://doi.org/10.1371/journal.pone.0083933.g002

Endothelial specificity of the Ad5ROBO4 vector following systemic administration

While the 3 kb enhancer promoter fragment of human ROBO4 enhancer/promoter had been previously validated for endothelial expression in single copy and endogenous locus transgenic knock-in mice [12], [29], there was no guarantee of its similar specificity when used in the context of an Ad vector. Therefore to test expression localization, we compared our Ad5ROBO4 versus Ad5CMV vectors in a nine organ panel harvested from tumor bearing immunodeficient Rag2^−/− mice (negative littermate mice derived from hCAR:Rag2^−/− × Rag2^−/− matings, Materials and Methods) (Figures 2A and 2C). Ad5ROBO4-mediated reporter gene expression was exclusively EC localized predominantly in the liver, with sporadically detectable rare EC expression in spleen and heart (Figure 2A). In stark contrast, Ad5CMV-mediated EGFP expression was predominantly detectable in liver hepatocytes, and occasional reticuloendothelial system (RES) cells in spleen and adrenal gland (Figure 2C). Large format images highlighted the potent cell type specific transcriptional retargeting from hepatocytes to liver ECs mediated by the 3 kb ROBO4 enhancer/promoter (Figure S1).

Enhanced but promiscuous endothelial Ad vector expression mediated by ubiquitous hCAR transgene expression

As undetectable host organ Ad5ROBO4 vector expression was likely due to extremely low levels of EC CAR protein, we intercrossed Rag2^−/− with hCAR transgenic mice to test the maximal EC expression capacity of our Ad5ROBO4 vector. We performed comparative experiments by intravenously injecting 1.0×10¹¹ vp of either Ad5ROBO4-EGFP or Ad5CMV-EGFP vectors into hCAR:Rag2^−/− tumor bearing composite mice, but for this analysis focused on host organs (Figures 2B and 2D). Endothelial expression was markedly increased in all organs except for liver and skin in Ad5ROBO4 injected mice (Figure 2B). The Ad5CMV vector expression pattern in composite mice was surprisingly complex (Figure 2D). Frequent EC expression was detected in kidney glomeruli and peritubular vessels, muscle, myocardium, adrenal and skin, with focal expression in brain vessels (Figure 2D). In sharp contrast, liver Ad5CMV-mediated EGFP expression remained restricted to hepatocytes, while splenic expression was a composite of vessel ECs and interfollicular RES cells (Figure S1 and Figure 2D). Of interest, and consistent with an intravenous injection “first pass” phenomenon, appreciable hepatocyte detargeting was produced by ubiquitous hCAR transgene expression (Figure S1).

Ad5ROBO4 transcriptionally targets tumor endothelial cells

After elucidation of EC expression patterns of the Ad5ROBO4 and Ad5CMV vectors in host organs either without (Figures 2A and 2C) or with (Figures 2B and 2D) the hCAR transgene, we next tested for differential vector expression in 786-O tumors (Figure 3). In contrast to our nine-organ host panel wherein Ad5ROBO4 vector expression was essentially restricted to liver ECs, both KO and SC tumors evidenced easily detectable, though sporadic, Ad5ROBO4 EC EGFP immunofluorescence in Rag2^−/− mice (Figure 3A). The addition of the hCAR transgene produced markedly increased intratumoral Ad5ROBO4 vessel expression (Figure 3B). To test for tumoral Ad5ROBO4 mediated EC cell type specificity, we co-stained tumor sections with antibodies cognate for pericyte, desmin, PDGFRβ, and NG2, or inflammatory cell, CD45, antigens (Figure S2). As a subset of intratumoral vascular pericytes remained intimately associated with capillary ECs, a fraction of these cells labeled with the red-emitting fluorochrome appeared to partially co-localize with the green EGFP immunofluorescent signal. However, closer examination revealed distinct segmentation of vector transgene expression to ECs rather than overlying pericytes (Figure S2). These data supported the EC specificity of the Ad5ROBO4 vector in the intratumoral microenvironment. In contrast to Ad5ROBO4, Ad5CMV-mediated EGFP expression was undetectable in KO and SC tumors grown in Rag2^−/− mice (Figure 3C). The presence of the hCAR transgene only elevated intratumoral Ad5CMV-mediated EGFP expression to a level approaching, but still lower than, that of Ad5ROBO4 in hCAR negative mice (Figure 3D versus 3B). Moreover, only half of the sporadically dispersed intratumoral EGFP positive cells in Ad5CMV vector injected hCAR:Rag2^−/− mice were co-localized within ECs, while the other half were perivascular stromal cells (Figure 3D, third and sixth columns, 400X magnification).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Vascular restricted ROBO4-directed reporter expression in kidney orthotopic and subcutaneous xenograft tumors.

The tumors were harvested from the same Ad vector injected mice whose host organs were depicted in Figure 2. A. Ad5ROBO4 vector expression in Rag2^−/− mice is sporadic yet easily detectable in vascular ECs within tumors in either orthotopic or subcutaneous microenvironments. B. The hCAR transgene markedly increases Ad5ROBO4 vector expression throughout both the orthotopic and subcutaneous tumors following injection into hCAR:Rag2^−/− mice. C. Ad5CMV expression in Rag2^−/− mice is undetectable in either orthotopic or subcutaneous tumors. D. Isolated intratumoral endothelial cell reporter expression in hCAR:Rag2^−/− mice. White line: kidney-tumor boundary; Arrows: glomerular tufts. Magnifications: 40X, first and fourth columns, and 200X, second and fifth columns, 400X, third and sixth columns. Red: endomucin/CD31, Green: EGFP immunofluorescence, Blue: DAPI.

https://doi.org/10.1371/journal.pone.0083933.g003

To semiquantitatively test for differential Ad vector-mediated host organ versus tumor expression, we immunoblotted extracts from both KO and SC tumors as well as liver, in hCAR:Rag2^−/− mice injected with 5.0×10¹⁰ vp of either the Ad5ROBO4 (n = 4 mice) or Ad5CMV (n = 4 mice) vectors, and probed for EGFP protein expression normalized to either VE-cadherin or β-tubulin (Figures 4A–D). Four independent experiments yielded consistent results, and one representative data set of the four is presented in Figure 4A. The near total absence of Ad5CMV-regulated tumor immunofluorescent expression (Figure 3D) was validated as neither the orthotopic nor xenograft immunoblots contained detectable EGFP protein (Figures 4A and 4D). In contrast, KO and SC tumor extracts from Ad5ROBO4 injected mice contained easily detectable EGFP expression that was 2.1- to 2.3-fold elevated compared to liver when normalized to either β-tubulin or VE-cadherin respectively (Figures 4A, 4B, and 4C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Semiquantitative immunoblotting reveals differential Ad5ROBO4 reporter expression in tumor compared to liver.

A. Immunoblot of EGFP, VE-cadherin, and β-tubulin loading controls in tissue protein extracts from hCAR:Rag2^−/− mice injected with either Ad5ROBO4, left three lanes, or Ad5CMV, right three lanes. B. and C. Densitometry analysis of Ad5ROBO4 vector EGFP expression normalized to either VE-cadherin or β-tubulin. D. Densitometry of AdCMV vector EGFP expression. As AdCMV expression was hepatocyte specific, this blot was only normalized to β-tubulin. A–D: Representative immunoblots from n = 4 mice injected with either Ad5ROBO 4 or Ad5CMV vectors. Li: liver, KO: kidney orthotopic tumor, SC: subcutaneous tumor.

https://doi.org/10.1371/journal.pone.0083933.g004

Ad5ROBO4 vector transcriptionally targets metastatic tumor endothelial cells

During tissue immunofluorescence analysis we serendipitously detected intra-ovarian and peritoneal metastases in an Ad5ROBO4 injected (1.5×10¹¹ vp) hCAR:Rag2^−/− mouse bearing an orthotopic tumor (Figure S3A–D). Nearly all of the microvessels within the intra-ovarian and peritoneal metastases expressed EGFP. In contrast, there was almost no expression within stromal ECs within the metastasis-bearing ovary except for perifollicular microvessels (Figure S3A–3C). Nor was expression detectable in ECs within the fallopian tube abutting the peritoneal metastasis (Figure S3D).

Pharmacological liver detargeting increases the tumor EC expression bias of the Ad5ROBO4 vector

As real world clinical applications demand sufficient target cell vector payload expression in the context of low hCAR expressing ECs, we manipulated Rag2^−/− host mice attempting to decrease liver vector uptake and conversely increase tumor viral particle delivery without the necessity of the hCAR transgene. For simplicity and across the board comparative analyses we chose warfarin administration, because the predominant mechanism for Ad vector liver sequestration is mediated by coagulation Factor X-viral capsid hexon binding [30], [31]. However, it is important to keep in mind that levels of vector liver detargeting similar to warfarin can be achieved using vector capsid mutation(s) [31]–[33]. First we tested liver detargeting efficiency in our Rag2^−/− mice. Warfarin, 5 mg/kg, pretreatment on day −3 and −1 before injection of 1.0×10¹¹ vp AdCMV-EGFP, revealed a marked diminution of hepatocyte reporter expression (Figures 5C, 5D and S4). Compared to the multiorgan EC expression mediated by the hCAR transgene (Figures 2B and 2D), warfarin failed to achieve appreciable multiorgan EC expression in Ad5CMV injected mice (Figures 5C and 5D) except for clusters of EGFP positive RES cells in the spleen, sporadic EC expression in the lung and adrenal gland, and only 1–4 individual positive ECs in muscle and skin. Brain lacked Ad5CMV EC vector expression. Encouragingly, warfarin treatment prior to Ad5ROBO4 vector injection produced a similar paucity of host organ EC expression other than in liver and spleen, and very focal areas in the lung (Figures 5A, 5B and S5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Warfarin pretreatment “detargets” the liver without producing multiorgan Ad5ROBO4 vector expression.

A. Ad5ROBO4 injection into vehicle-treated Rag2^−/− mice produced vascular EC expression predominantly in liver and spleen. B. Warfarin pretreatment, 5 mg/kg, on day −3 and −1 prior to Ad5ROBO4 injection vector injection increased the frequency of splenic and lung EC expression and produced sporadic, infrequent expression in kidney and heart. Liver expression was present but diminished. C. AdCMV injection into vehicle treated mice predominantly produced hepatocyte expression with focal RES cell splenic and scattered lung expression. D. Warfarin pretreatment prior to AdCMV injection markedly decreased the frequency of hepatocyte EGFP expression, while increasing sporadic splenic, and lung expression, inducing focal adrenal cellular and rare muscle and skin EC expression. A, B and S5: Representative images from n = 5 mice from 2 independent experiments, C and D: n = 3 mice from 3 independent experiments. 1×10¹¹ vp were injected in each group. Magnification 100X; Red: endomucin/CD31, Green: EGFP immunofluorescence, Blue: DAPI.

https://doi.org/10.1371/journal.pone.0083933.g005

Ad5ROBO4 EC targeting is maintained and differentially enhanced in both orthotopic and xenograft tumors compared to host organs following warfarin-liver detargeting

As warfarin-mediated liver detargeting effectively eliminated Ad5ROBO4-mediated EC expression in most host organs examined (Figure 5B and S5), our next question was whether EC expression in the tumors of mice whose host organs were examined in Figure 5, would be similarly diminished (Figure 6). Similar to data reported in Figure 2, scattered EC vector expression was detectable in both KO and SC tumors of vehicle (peanut oil) treated, Ad5ROBO4-injected mice (Figure 6A). Warfarin pretreatment produced a heterogeneous, but high level, regional increase in intratumoral EC vector expression (Figure 6B). In contrast, Ad5CMV expression was undetectable in intratumoral vessels either without or with warfarin pretreatment (Figure 6C and 6D). To quantify warfarin's effects on Ad5ROBO4-mediated EC expression in KO and SC tumors versus host organs, we measured the area of EC co-localized reporter versus total EC immunofluorescent areas in tissue sections from Ad5ROBO4-injected vehicle (n = 4) or warfarin-injected (n = 5) mice (Figure S5). Warfarin produced an eight-fold increase in EGFP-EC co-localization in orthografts and a six-fold increase in xenografts (p<0.05 warfarin compared to vehicle, Figure S5). Five of seven host organs evidenced minimal areas positive for Ad5ROBO4 expression, albeit with single outlier mice in each organ. Warfarin produced a 1.7-fold decrease in EGFP-EC co-localization in liver and a 2.6-fold increase in co-localization in the spleen (p<0.05 compared to vehicle, Figure S5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Warfarin liver detargeting enhances tumor neovascular endothelial cell reporter expression of the Ad5ROBO4 vector.

A. Ad5ROBO4 produced easily detectable scattered tumor endothelial cell EGFP immunofluorescence in both kidney orthotopic and subcutaneous 786-O tumors in vehicle-treated Rag2^−/− mice. B. Warfarin pretreatment markedly enhanced the multiplicity of tumor endothelial cell reporter gene expression within both orthotopic and subcutaneous tumors in Ad5ROBO-injected mice. C. Ad5CMV injection failed to produce detectable tumor EC expression in vehicle-treated, or D. warfarin-treated mice. A–D: tumors from the same mice as in Figure 5. Magnifications: 40X first and fourth columns, 100X second and fifth columns, and 400X third and sixth columns. Red: endomucin/CD31, Green: EGFP immunofluorescence, Blue: DAPI.

https://doi.org/10.1371/journal.pone.0083933.g006

To quantitatively test for warfarin mediated shifts of host organ versus tumor vector expression we immunoblotted tissue extracts, probing for EGFP, VE-cadherin, and the β-tubulin loading control (Figures 7A–7C). As the ultimate question was the warfarin-mediated differential tumor EC versus host organ expression, we focused our experiments exclusively on mice injected with the Ad5ROBO4 vector. Host organ whole tissue extract immunoblotting of vehicle treated mice substantiated the markedly predominant liver vector expression (Figure 7A, left seven lanes and 7B, white bars). Warfarin pretreatment produced a 2.5-fold increase in spleen and a 3-fold decrease in liver vector expression (Figure 7A, right seven lanes and 7B, black bars). Following this determination of host organ expression distribution, the effect of warfarin on tumor versus liver expression was tested in independent immunoblotting of KO, SC, and liver whole tissue extracts (Figure 7C). Warfarin pretreatment produced a 2.1-fold increase in KO, a 2.0-fold increase in SC, and a 4.7-fold decrease in liver vector expression normalized to VE-cadherin compared to vehicle (Figure 7C, left three gel lanes vehicle, right three lanes warfarin). The fold-increase in tumor Ad5ROBO4 expression mediated by warfarin-liver detargeting was closely comparable to that produced by ubiquitous hCAR transgene expression, and reinforces the conclusion that the Ad5ROBO4 vector is preferentially, though not exclusively, expressed in tumor versus host organ endothelium.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. Warfarin pretreatment enhances tumor Ad5ROBO4 vector expression.

A. Multiorgan immunoblot of vehicle (left lanes) or warfarin (right lanes) pretreated Rag2^−/− mice injected with 1.0×10¹¹ vp of Ad5ROBO4. B. Densitometry of A revealed that vehicle pretreatment was associated with robust liver, detectable splenic, and trace to undetectable expression in all other sampled organs. Warfarin pretreatment produced a 2.5-fold increased splenic and a 3-fold decreased liver expression while all other organs still evidenced trace to undetectable expression. C. Immunoblot and densitometry of liver and tumor EGFP, VE-cadherin, and β-tubulin expression in vehicle (left lanes) or warfarin (right lanes) from the same pretreated, Ad5ROBO4-injected mice as in A and B. EGFP densitometry normalized to VE-cadherin, revealed a 4.7-fold decrease in liver and 2-fold increase in increase KO and SC tumor expression produced by warfarin pretreatment. A–C: representative immunoblots from n = 2 mice from 2 independent experiments.

https://doi.org/10.1371/journal.pone.0083933.g007