Effective in vivo and ex vivogene transfer to intestinal mucosa by VSV-G-pseudotyped lentiviral vectors

The pRRLsin-hCMV-β-Gal vector was constructed by insertion of the β-Gal reporter gene from plasmid pSV-β-Gal (Promega, Madison, WI, USA), and the pRRLsin-hCMV-fLuc vector was constructed by insertion of the fLuc gene [32], respectively, into the multiple cloning site (MCS) of pRRLsin-hCMV-MCS-pre, a third-generation, self-inactivating LV construct kindly provided by Dr. Luigi Naldini (University of Milan, Italy) [33] (Figure 1A).

Human embryonic kidney cell line 293T, and human colon cancer cell lines Caco-2, LoVo, and WiDr, (ATCC, Manassas, VA, USA), were cultured in Dulbecco's modified Eagle's medium (DMEM), Ham's F12K, or RPMI-1640 medium, respectively, with 10% fetal bovine serum and 1% penicillin-streptomycin. LV virus was produced in 293T cells using a third-generation packaging system as previously described [34], using either pRRLsin-hCMV-β-Gal or pRRLsin-hCMV-fLuc, to produce LV-β-Gal or LV-fLuc, respectively. Virus titers were determined by HIV-1 p24 ELISA (Perkin Elmer, Waltham, MA, USA) and expressed as p24 equivalent units (ng/ml). Vector transductions were performed on 1 × 10⁵ target cells with 8 μg/mL polybrene (Sigma, St. Louis, MO, USA) at 37°C. Polybrene is a small, positively charged molecule that binds to cell surfaces and neutralizes surface charge and has been shown to enhance cell transduction by retroviruses [35].

After replacement of the medium and further incubation for 24 hr, cells were washed twice in phosphate-buffered saline (PBS), fixed in 2% formaldehyde/0.2% glutaraldehyde (Sigma, St. Louis, MO) for 10 min at room temperature, and stained with 20 mg/ml X-Gal solution (5-bromo-4-chloro-3-indolyl beta-D-galactoside; Sigma, St. Louis, MO, USA) at 37°C for 2 hr.

All in vivo studies were performed according to institutional guidelines under protocols approved by the UCLA Animal Research Committee. Briefly, 6 to 8-week-old female BALB/c mice (Charles River Laboratories Inc., Wilmington, MA, USA) were divided into a non-treated control (NC) group, and two groups that received intrarectal instillation of either LV or vehicle solution (DMEM). Prior to instillation, anesthetized mice were given an intrarectal enema of 50% ethanol (vol/vol in distilled H₂O). Pretreatment with ethanol enemas has been shown to increase intestinal transduction with other vectors [36]. Two hours after the enema, 100 μL of vehicle or viral vector solution containing a total of approximately 1000 ng p24 equivalent units was instilled intrarectally. The mice were inverted for 30 sec after administration of intrarectal products to prevent leakage.

Health status and body weight was monitored throughout the study. On days 2, 7, and 21 after vector or vehicle instillation, cohorts of mice were sacrificed, and the entire colon was removed. Colon length from cecum to anus, weight, and a macroscopic colonic damage score were recorded The macroscopic colonic damage score was based on the degree of tissue adhesion, mucosal ulceration, and intestinal wall thickness (Table 1) [37].

Ex vivo bioluminescence imaging was performed 2 days after rectal instillation of LV-fLuc using the Xenogen IVIS system (Caliper Life Sciences, Alameda, CA, USA). Bioluminescent signal intensity was expressed as photons per second per cm² per steridian (p/s/cm²/sr), and color images were processed with Living Image and IGOR-PRO analysis software (Wave Metrics, Portland, OR, USA).

For routine histology, colon and other tissues (spleen, liver, lung, kidney) were fixed in 4% paraformaldehyde overnight, placed in 30% sucrose/PBS for 2 hr, and embedded in OCT compound. Serial 5-μm cryosections were stained with hematoxylin/eosin (H&E) for evaluation of histopathology score. Mucosal inflammation was scored using a semi quantitative technique (Table 2) [38].

For X-Gal histochemistry, tissues were fixed in 2% glutaraldehyde for 2 hr, embedded in OCT compound, and 10-μm cryosections were stained in 1 mg/ml X-Gal solution at 37°C for 24sr, rehydrated, and counterstained with 0.1% Nuclear Fast Red (Sigma, St. Louis, MO, USA). The number of positive staining cells was counted in five independent fields in random areas on two non-consecutive slides at 200× magnification.

Quantification of vector copy number was performed at each time point by TaqMan qPCR assay to detect the HIV-1 packaging signal sequence, using 300 ng genomic DNA (equivalent to 5 × 10⁴ genomes) isolated from murine colon and other tissues, including stomach, small intestine, liver, kidney, spleen, lung, heart, brain, and bone marrow [39]. A reference curve was prepared by amplifying serial dilutions of LV-encoding plasmid in a background of genomic DNA isolated from untransduced murine colon. Genomic DNA from a PC3 cell line previously confirmed by flow cytometry to be 100% transduced by LV-GFP vector was used as a positive assay control.

All human endoscopic biopsies were collected from healthy HIV-negative volunteers at UCLA Medical Center, Los Angeles, USA. The protocol for the use of human endoscopic biopsies was approved by Institutional Review Board of the David Geffen School of Medicine at UCLA (#02-05-001-13).

Explant cultures were established as previously described [31]. Briefly, explants were placed on presoaked Gelfoam^® (Pharmacia and Upjohn Company, Kalamazoo, MI, USA) rafts with the epithelium uppermost. Tissues were transduced by pipetting LV solution and polybrene (8 μg/mL) onto the top of each Gelfoam^®-supported explant. After 2 hr incubation, tissues were washed three times, placed on fresh Gelfoam^® rafts, and incubated at 37°C for a total of 24 hr. Explants were then either fixed in glutaraldehyde, embedded in OCT, and 10-μm cryosections prepared for X-Gal as above, or were fixed in formaldehyde, embedded in OCT, and 7-μm sections prepared for immunofluorescence staining.

The identity of transgene-expressing cells in harvested murine colon tissues or human colorectal explant tissues was examined by immunofluorescence double-staining with antibodies against β-Gal and cell-specific phenotypic markers. Briefly, serial 7-μm cryosections were stained for E. coli β-Gal (Promega, Madison, WI, USA). Additional staining included antibodies directed against human and murine cytokeratin AE1/AE3 and CD45 (Dako North America Inc., Carpinteria, CA, USA); and CD4, CD8, and CD11c (BD Pharmingen, Franklin Lakes, NJ, USA). β-Gal staining was then visualized with Alexa Fluor^® 594. Cytokeratin and CD45 were visualized with Alexa Fluor^® 488 (Invitrogen Corporation, Carlsbad, CA, USA). The CD4, CD8, and CD11c antibodies were already conjugated with Alexa Fluor 488. The Vector mouse-on-mouse (M.O.M™) Immunodetection Kit (Vector Laboratories, Burlingame, CA, USA) was used to avoid the high background staining caused by antibody binding to endogenous murine immunoglobulin when secondary antibodies were used to amplify primary murine antibodies. Slides were mounted and nuclei counterstained using VECTASHIELD^® and DAPI (4',6-diamidino-2-phenylindole; Vector Laboratories, Burlingame, CA, USA). In order to quantify LV-mediated β-Gal transduction efficiency, the total number of cells showing co-localization of β-Gal-positive and CD45 or CD4 positive signals in 3 randomly selected fields per section at 200 × magnification were identified, and expressed as percentages of the total number of β-Gal-positive cells or the total number of CD45 +/CD4+ cells.

All values were expressed as means ± SD. Comparisons between groups were made using the Student t-test and the Mann-Whitney U test, and p values <0.05 were considered statistically significant.

LV-mediated transduction efficiency was first assessed in vitro using human colorectal adenocarcinoma cell lines Caco-2, WiDr, and LoVo. In each cell line, a dose-dependent increase in the number of cells showing positive signals upon X-Gal staining was observed with increasing LV concentration (Figure 2A, B). The highest dose tested was 10 ng p24 equivalent units of LV, which corresponds to a biological infectious titer of 10 ⁶ transducing units (TU) on a standardized cell line such as 293T [40]. At this dose, transduction rates per 1 × 10⁵ target cells were 40.6% for WiDr, 13.7% for Caco-2, and 13.3% for LoVo, indicating that although these colon adenocarcinoma cell lines are less permissive for LV compared to 293T embryonic kidney cells, significant gene transfer can be achieved with higher multiplicities of infection. Seppen et al. have previously reported higher rates of transduction in Caco-2 cells but used a transwell system with GFP expressing first and third generation lentiviral vectors [41].

The safety of gene transfer procedures using LV expressing either fLuc or β-Gal was assessed by both macroscopic and histopathologic criteria in three experimental groups; mice receiving the LV, mice receiving a mock installation (the viral plasmid solution in DMEM), and a control group (N = 8 per condition) Intrarectal administration of 1000 ng p24 equivalent units of VSV-G-pseudotyped LV, corresponding to a 293T cell-based standardized biological titer of 10 ⁸ TU, did not affect body weight or induce any gross abnormalities upon routine observation. The ratio of colon weight-to-length was 0.33 ± 0.04 in the non-treated control (NC) group, 0.34 ± 0.5 in the mock instillation control group, and 0.27 ± 0.4 in the LV instillation group. Thus, there were no significant differences among these three groups in macroscopic damage assessment or histopathological evaluation scores (Additional File 1). More specifically, the use of ethanol enemas did not appear to induce significant mucosal damage. No pathological findings could be observed in kidney, spleen, lung, and liver harvested from control and LV-treated mice at each time point (data not shown).

Strong positive signals were observed in the distal colon adjacent to the rectum in 8/8 (100%) of mice exposed to VSV-G-pseudotyped LV encoding the fLuc reporter gene (Figure 3A). Quantitative measurement of bioluminescent photon emission from LV-transduced distal colon showed a significantly higher level of transduction than that of mock-treated colon (57560 ± 28960 p/s/cm²/sr for the LV instillation group vs. 6260 ± 813.3 p/s/cm²/sr for the mock instillation group; p < 0.001) (Figure 3B). Additional studies are needed to determine whether the physical distribution of the ethanol enemas and/or the LV influenced the location of the bioluminescence signal.

Real-time qPCR was employed to quantify the vector copy number of LV integrated into genomic DNA from murine colon tissues after in vivo administration. Using spiked samples to obtain a reference curve, this method was determined to be sensitive enough to detect 50 copies of LV per 5 × 10⁴ cellular genomes. Genomic DNA from both positive control cells (data not shown) and transduced rectosigmoid colon showed amplification of proviral LV sequences. The average copy number per 300 ng genomic DNA from LV-transduced colon was 231.3 ± 183.7 (Figure 3C). As expected, no detectable qPCR signals were observed in genomic DNA from colon tissues of non-treated or mock-treated controls. Importantly, even in LV-treated animals at 2 days post-vector instillation, no detectable qPCR signals were found in any extra-intestinal tissues examined, including stomach, small intestine, liver, kidney, spleen, lung, heart, brain, and bone marrow (Figure 3C). As the LV was not treated with DNAse prior to injection, we cannot exclude the possibility that the data in Fig 3C in part reflect plasmid contamination.

Histochemistry was performed using optimized X-Gal concentrations and pH conditions to minimize background staining due to endogenous mammalian β-galactosidase in the GI tract. As expected, under these conditions, no staining was observed in mouse colon tissues from non-treated and mock-treated control groups, nor in the non-intestinal tissues (kidney, spleen, lung, and liver) from any animals including the LV-treated group.

In contrast, positive-staining cells were observed in colon tissues from animals treated with LV expressing the E. coli β-Gal reporter gene. The initial number of positive-staining cells in the LV-treated group was 62.3 ± 18.5 per microscopic field at 200× magnification on Day 2 post-vector instillation (Figure 4). However, the number of positive-staining cells was then observed to decrease, to 19.1 ± 5.3 per 200 × field on Day 7, and 15.7 ± 5.1 per 200 × field on Day 21. These positive-staining cells were found predominantly toward the luminal surface of the colon, but were identified in both mucosal epithelium and, importantly, the lamina propria (LP) (Figure 5A).

To further characterize the identity of the transduced cells, immunofluorescence double-staining was performed using an E. coli β-Gal-specific antibody in combination with various cellular immunophenotypic antibodies. Quantification of double-positive staining in colon tissues harvested on Day 2 post-LV instillation demonstrated that 27 ± 5% of the β-Gal-positive cells were also positive for cytokeratin. Consistent with their histological origin, β-Gal/cytokeratin double-positive cells were observed solely in the epithelial layers of colon tissues from transduced animals. Cells exhibiting double-positive staining for both β-Gal and the common leukocyte antigen CD45 were found in both mucosal epithelium and lamina propria (LP) regions. These β-Gal/CD45 double-positive cells represented 70 ± 11% of the total population of β-Gal-positive cells.

A subset of LV-transduced white blood cells was further identified to consist of T lymphocytes, dendritic cells, or macrophages, as 27 ± 2% of the total population of β-Gal-positive cells were also positive for CD4. In fact, of the total population of CD4-positive cells, 48 ± 13% were β-Gal-positive, indicating a significant transduction of half of this cell population. These β-Gal/CD4 double-positive cells were observed only within the LP of transduced murine colon. Notably, there were no cells showing co-expression of both β-Gal and CD8 or CD11c (Figure 5B-F).

Human colon tissue specimens were incubated with 1000 ng p24 units of LV-β-Gal, and examined within 24 hr of explant culture. The number of X-Gal-positive cells after transduction was 12.4 ± 1.5 per low power (×200) field. In contrast to the murine studies, the transduced cells in human samples were primarily found within the LP (Figure 6A). These data might reflect the availability of both apical and basal aspects of the explant to LV transduction, and/or the use of polybrene. By double immunofluorescence staining using human cellular immunophenotypic antibodies, 84 ± 16% of β-Gal-positive cells were also found to be positive for CD45. Of these, 41 ± 12% of β-Gal-positive cells were found to be CD4-positive, and 35 ± 8% were CD11c-positive. Notably, no β-Gal-positive cells expressed CD8 or cytokeratin (Figure 6B-F). Non-specific background staining was not observed using this technique (Additional File 2). This identifies transduced cells as mucosal lamina propria CD4+ lymphocytes, macrophages, dendritic cells, and other leukocytes. Consistent with the results obtained after in vivo LV instillation in mice, 54 ± 15% of the total population of CD4-positive cells in ex vivo LV-transduced human explant tissues were also positive for β-Gal. We did not perform immunohistochemical examination of mesenteric lymph node tissue and so we do not know whether LV transduced cells could be identified at this site.