AAV2 replicates significantly higher in PT3 cells

AAV has been isolated from SSE of the anogenitals and autonomous parvoviruses preferentially replicate in malignant cells. Thus, to test the hypothesis that AAV preferentially replicates in cervical cancer cells we compared three primary cervical cancer isolates and two archival cervical cancer cell lines to normal primary human foreskin keratinocytes (NK) for the ability to allow AAV autonomous replication and virion production within the organotypic epithelial raft culture system. All of these cells, except the normal keratinocytes, contain human papillomavirus type 16 (HPV-16) DNA. The NK cells represent a mixed culture of cells isolated from multiple individuals. The six types of cells were infected with AAV, transferred into the raft culture system to form a stratified squamous epithelium, harvested on day 6, DNA extracted, and analyzed by Southern blot. Two types of analyses were done as depicted in Figure 1A. First, AAV DNA replication was analyzed in the various squamous cell lines as SSE rafts, as a "first plate" analysis. Second, AAV virion production was measured by generating putative AAV virus stocks from equivalent "first plate" rafts and then a portion was used to infect a "second plate" of adenovirus-infected HEK293 cells. Any AAV DNA replication in the 293 cells would be due to AAV virions produced in the first plate rafts.

The Southern blot analysis of AAV replication directly in the first plate rafts is shown in Figure 1B. As can be seen, of the six cell types one isolate showed an unusually high level of AAV replication compared to other isolates. PT3 allowed for approximately a 10 fold higher level of AAV DNA replication compared to all other cervical cancer cell lines by densitometric analysis. All the other cervical cancer lines, and normal keratinocytes, also demonstrated AAV replication, but at a much low level. A quantification of the DNA replication levels is shown in Figure 1C. These results are comparable to a similar first plate raft experiment of AAV DNA replication shown in Figure 1D. However, coupled with this experiment is a second plate analysis of AAV virion production as shown Figure 1E. Note that PT3 was, in addition to higher AAV DNA levels, also demonstrated higher levels of virion production as well. Thus, PT3 is super permissive for complete AAV's full life cycle.

Gene expression analysis with normalization to ACTB, GAPDH, or HG-U133A housekeeping genes

As PT3 allowed much higher levels of AAV replication we expected these cells to over express cellular components PCNA, POLD1, RFC, RPA1, and RPA [41, 42]. Thus the transcriptome of PT3, representing the high AAV replication scenerio, was compared to low/normal AAV replication cell types PT1 and NK by DNA microarray analysis. Total RNA prepared from PT3, PT1 and NK was examined for the expression levels of Affymetrix HG-U133A (14,500 human genes). The RNA samples were isolated in-house and sent to the University of Iowa DNA Core for analysis. Three different methods for data normalization using ACTB, GAPDH, and Affymetrix U-133A housekeeping expression, respectively were utilized.

Normalization of all probe sets (5789 probe sets) to expression of GAPDH as a control gene revealed 1440 probe sets that were up-regulated, and 429 probe sets that were down-regulated, in PT3 compared to PT1 and NK cell lines, for a total of 1869 genes of all differently expressed genes. Yet again the same seven AAV-critical genes were up-regulated in PT3 compared to PT1 and NK, (Table 1), this time when normalized to GAPDH. These data provide evidence that the cellular components reported to be involved in AAV in vitro DNA replication may also be involved in vivo AAV DNA replication as well. Furthermore these data suggest a mechanistic explanation as to why PT3 allows high AAV DNA replication.

Comparison of Normalization Techniques

Based on the number of transcripts identified as differentially expressed, the three techniques used to normalize the array data could be ordered by the number of genes identified as differentially expressed as follows: GAPDH (1869 probe sets) > ACTB (1781 probe sets) > U-133A (1478 probe sets). Although the three array normalization methodologies differed in the number of genes defined as down- or up-regulated in expression in PT3 compared to PT1 and NK cell lines, all identified the same 7 up-regulated genes (PCNA, POLD1, RFC3, RFC4, RFC5, RPA1, and RPA2) except RPA1 in normalization using HG-U133A housekeeping genes (Table 1). This finding suggested that these seven genes were clearly differentially over-expressed in PT3 versus PT1 and NK cell lines.

Verification of microarray results by real-time quantitative PCR

As we did the microarray analysis using a single mRNA isolation/cDNA probe analysis, we needed to verify the transcriptional over-expression of these seven genes by real-time quantitative PCR. To determine the optimum amount of cDNA template in initial experiments, we performed undiluted, 1:10 diluted, and 1:100 diluted cDNA template in parallel. Gene expression of NK, PT1 and PT3 cDNA templates were normalized to the Ct value of reference genes: GAPDH or ACTB, respectively, to calculate the Ct values of PCNA, POLD1, RFC3, RFC4, RFC5, RPA1, and RPA2. We used NK as calibrator (Figure 2A and 2B). The RT-qPCR results confirmed the microarray results, that PCNA, POLD1, RFC3, RFC4, RFC5, RPA1, and RPA2 were over-expressed in PT3 (at least a 1.8 fold difference between two groups [PT3 vs Non-PT3]). The relative quantitative expression of the 7 genes between PT3 and Non-PT3 samples was set at a significance level of 0.05. To see the comparative gene expression levels of PCNA, POLD1, RFC3, RFC4, RFC5, RPA1, and RPA2, comparing the microarray and qPCR results, we used non-PT3 (NK and PT1) cells as the calibrator (Figure 3A and 3B).

Super-permissiveness does not correlate with cytotoxicity

It has been reported in CNN in 2005, in work done by Craig Meyers, that AAV preferentially kills cancer cells http://www.cnn.com/2005/HEALTH/06/22/cancer.virus/. This reported cancer cell killing may be related to parvovirus replication as certain parvoviruses have been reported to preferentially replicate in malignant cells [44]. Thus we tested the high and low AAV-permissive cells for their sensitivity to killing by AAV infection. The results are shown in Figure 4 and demonstrate that PT3 was not preferentially sensitive to killing by AAV2 infection compared to other squamous cells.

Cell lines

Primary human foreskin keratinocytes (NK) were purchased from Clonetics Inc.(San Diego, CA). PT1, PT2, and PT3 primary cell lines were isolated from three cervical cancer patients as described previously [48]. These cervical cancer isolates were at approximately passage 10–15 when used in these experiments. CaSki and SiHa cervical cancer cell lines were purchased from American Type Culture Collection (Rockville, MD). All the cells were cultured in keratinocyte serum-free medium (Invitrogene, Carlsbad, CA) in 37°C under 5% CO₂ prior to raft formation.

AAV replication in squamous cells using the organotypic epithelial raft cultures

On day 1, 10⁶ normal primary keratinocytes, three primary cervical cancer cells and CaSki, SiHa cells were infected with 10⁸ infectious units of wild type AAV-2 virus (multiplicity of infection [moi] = 100). On day 2 the cells were trypsinized, plated onto J2-containing collagen rafts as described previously [34–37]. On day 3 these organotypic skin rafts were raised to the air interface and allowed to form an SSE over a period of 3 days (day 6 overall) using E medium. Southern blot analysis was done to detect AAV DNA replication. Rafts were harvested on day 6. After washing with phosphate-buffered saline (PBS, 137 mM NaCl, 10 mM Phosphate, 2.7 mM KCl, pH 7.4.) twice, all the rafts were minced and lysed. Total DNA was extracted and 10 μg of total cellular DNA were analyzed for AAV DNA replication levels by agarose gel electrophoresis, Southern blotting, and probing with ³²P-AAV Cap DNA probe to pick up only the wt AAV genome. Finally, a quantification of the Southern blot was done by densitometric analysis using an Alpha Imager 2000 (Alpha Innotech Corporation, San Leandro, CA). The densitometric data was quantified using AlphaImager™ 2000 software. Densitometric data was analyzed by the unpaired t-test and presented as mean ± standard error (SE).

"Second plate" analysis of AAV virion production

Instead of harvesting the keratinocyte rafts for the analysis of AAV DNA replication on day 6, in certain experiments the SSE rafts were analyzed for AAV virion production by the infection of a "second plate" of adenovirus infected HEK293 cells. Putative AAV virus stocks were generated by freezing day 6 rafts and grinding the rafts with mortar and pestle. The remains of the raft were placed in one ml of DMEM medium, vortexed for 1 minute and centrifuged at 8,000 g for 15 minutes to remove debris, and the supernatant was filtered through a 20 um filter. One third of the putative virus stock was used to infect a 6 cm plate on 80% confluent monolayer HEK293 cells. These cells were also infected with Ad helper virus at an moi of 5. Any AAV infectious units produced in the original raft would be amplified in the Ad-infected 293 cells. After 36 hours of infection total DNA was extracted and 10 μg of total cellular DNA were analyzed for AAV DNA replication levels by agarose gel electrophoresis, Southern blotting, and probing with ³²P-AAV cap DNA probe.

AAV2 cytotoxicity in cervical cancer cell isolates

AAV2 virus stock was serially diluted with Dulbecco's medium (supplemented with 10% FBS and 100 U/ml penicillin). Normal keratinocytes and three primary cancer cell lines (PT1, PT2 and PT3) were seeded (4 × 10⁵/dish) one day prior to infection with serially diluted wild type AAV 2 in 1 ml culture media at a multiplicity of infection (moi) of 100, 1,000, 10,000 AAV particles. Culture media was replaced with E medium after overnight incubation at 37°C and were incubated for additional 6 days with fresh media at one day interval. At day 7 the cells were washed with PBS, fixed in formaldehyde and stained with methylene blue. The experiment was done three times.

Total RNA extraction and cDNA synthesis

For real-time quantitative PCR (qPCR), total RNA samples from 1 × 10⁶ cultured cells was extracted from NK, PT1 and PT3 cell lines using Total RNA Purification System Kit (Invitrogen, USA) according to the manufacturer's protocol. Concentration of mRNA was quantified using NanoDrop^® ND-1000 Spectrophotometer (NanoDrop technology, USA). One μg was transcribed into first strand cDNA using Transcriptor First Strand cDNA Synthesis Kit (Roche, Indianapolis, IN) according to the manufacturer's protocol. For Affymetrix microarray analysis, total RNA was isolated from NK, PT1 and PT3 cell lines using Trizol (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. After treatment with 5 U/μg of RNase-free DNase I at 37°C for 1 hour, all the samples were frozen in and sent to University of Iowa DNA facility for microarray analysis. After cDNA synthesis, samples were applied to a Human Genome GeneChip HG-U133A (Affymetrix Inc. Santa Clara, CA).

Array filtering and significant expressed gene identification

Microarray data in the form of CEL files were imported into BRB ArrayTools developed by Dr. Richard Simon and Amy Peng Lam http://linus.nci.nih.gov/BRB-ArrayTools.html. HG-U133A microarray raw expression intensities of NK, PT1, and PT3 data were scaled to a target intensity of 100 units, normalized independently, using the robust multichip average (RMA) algorithm for the quantification of the expression level of target genes, and passed by the filtering and subletting criteria with any one absent (A) or marginal call (M). Genes that had more than 50% missing data across all observations were excluded from the analysis. Also, we selected those genes with an expression level of ≥ 20 in ≥ 25% of samples. Fold change has been transformed based on log₂ (PT1/NK), log₂(PT3/NK), log₂ (PT3/PT1), log₂ (PT3/non-PT3), respectively. Fold change above 2.0 was defined as differentially expressed genes between two cell lines, where it is meet fold >2 SD (above 97% confidence).

Real-time quantitative PCR

All PCR reactions were carried out in a final volume of 50 μl containing 1× of FastStart Universal SYBR Green Master [ROX] (Roche Applied Science, USA) with a reporter dye at the 5' end (FAM) and a quencher at the 3' end, 300 nM of each gene specific primers (Table 1), 50 ng cDNA, and then added sterile deionized water. The standard cycling condition was 50°C for 2 min, 90°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min.

To quantify the relative expression of each gene, real-time qPCR data were first reported as (1) NK: PT1 and PT3 as well as (2) non-PT3 (NK and PT1):PT3 ratios. The comparative threshold cycle (C_t) values for NK, PT1, and PT3 samples were normalized for reference genes (ΔC_t = C_{t target} - C_{t ACTB or GAPDH}) and compared with a calibrator using the ΔΔCt method [49]. As calibrator, the average Ct value of each gene in all samples grouped together was taken. All reported real-time quantitative PCR reactions were performed and analyzed using an ABI 7500 System SDS Software Ver1.3 (Applied Biosystems, USA). Fold units were calculated dividing the expression fold changes of the candidate genes by the expression fold changes of the reference gene (ACTB or GAPDH).

Statistical analysis

Comparison of the relative quantitative expression of the 7 genes between PT3 and Non-PT3 samples was done with an unpaired t-test comparing two groups, with a significance level of 0.05 using Microsoft Excel 2003 program and presented as mean ± standard error (SE). All real time quantitative PCR were performed in triplicate to ensure quantitative accuracy.