A potential approach for assessing the quality of human and nonhuman adenoviral vector preparations

Abstract

Various types of human and nonhuman adenoviral (AdV) vectors are being used as gene delivery vectors in preclinical and clinical investigations. The objective of this study was to determine the ratio between the 2 best assays that would effectively address the variability in the titration of various AdV vectors in different cell lines and help obtain consistent results in preclinical and clinical studies using different AdV vectors. Here, we compared plaque-forming units, tissue culture infectious dose 50, focus-forming units (FFU), virus particle (VP) count, and genome copy number (GCN) of purified preparations of human AdV type C5, bovine AdV type 3, and porcine AdV type 3 to determine a correlation between infectious and noninfectious virus particles. Our results suggest that a VP:FFU or a VP:GCN ratio could accurately reflect the quality of an AdV preparation and could serve as an indicator to control batch-to-batch variability.

Adenoviruses (AdVs) are non-enveloped viruses with icosahedral capsids of 90 to 100 nm in size containing a double stranded DNA genome ranging from 26 to 46 kb (1). Several AdVs belonging to human, simian, bovine, porcine, canine, ovine, avian, and other species have been developed as gene delivery vectors for recombinant vaccines and gene therapy applications, including the treatment of metabolic disorders and cancers (2–8). Several AdV vectors are currently being evaluated as potential gene delivery vehicles in clinical trials (9–11). There are various types of these human and nonhuman AdV vectors, such as replication-competent (e.g., E3-deleted), replication-defective (e.g., E1-; E4-; E1 and E3-; E1 and E4-; or E1, E3 and E4-deleted), conditional replication-competent (e.g., expressing E1A under an inducible promoter), and helper-dependent/gutless (e.g., containing at least the AdV inverted terminal repeats and the packaging sequences) vectors (12). Adenovirus vectors can efficiently infect several cell types, leading to high levels of transgene expression (13,14). There can also be high levels of immunogen-specific humoral and cellular immune responses due to AdV vector-mediated Toll-like receptor (TLR)-dependent as well as TLR-independent pathways (15,16).

Adenovirus titration by plaque assay to determine plaque-forming units (PFU) is still considered a widely accepted method to titrate infectious AdV particles. For AdVs that do not easily develop plaques in infected cell monolayers, tissue culture infectious dose 50 (TCID₅₀) by TCID₅₀ assay is one of the preferred tests for AdV titration. However, when conditional replication-competent and helper-dependent vectors were developed, AdV vector titration (either by plaque assay or TCID₅₀ assay) lost its applicability. Subsequently, virus particle (VP) count assay was developed to quantitate various AdV particles in purified preparations (17–19). Now, the VP assay is based on a spectrophotometric estimation of the amount of protein and nucleic acid in a purified AdV vector preparation. The number of VP is calculated using the formula:

$$\text{VP\hspace{0.17em}count}=\text{absorbance\hspace{0.17em}at\hspace{0.17em}}260\hspace{0.17em}\text{nm}\times \text{dilution\hspace{0.17em}factor}\times 3.5\times {10}^9$$(19,20).

A VP:PFU ratio is preferred for AdV vector titration, whenever it is applicable. Since the VP count is affected by the total protein amount in the virus preparation, this assay will include empty VP (i.e., without AdV genome sequences), mature VP (i.e., containing AdV genome sequences), and ruptured particles. The ratio of empty and ruptured VP:infectious VP can be in the range of 11 to 2300 (21–23), suggesting that most of the VP in an AdV preparation represents non-infectious virus particles. For the majority of AdV vectors, the number of non-infectious VP in vector preparations is critical to the AdV-mediated adjuvant effect. The VP:PFU ratio for human AdV vectors varies in different preparations from 12:1 to 100:1 (24,25). This ratio is being used as a guide to reduce batch-to-batch variability for clinical applications. AdV adjuvant effect and vector toxicity are mainly due to viral capsid proteins; therefore, vector titration needs to be efficient for all types of vectors. The VP:PFU ratio is not a good measure for many other AdV vectors. For example, in one study, VP:PFU ratios for human AdV type C5 (HAdV-C5) expressing green fluorescent protein (GFP) (HAd-GFP), bovine AdV type 3 (BAdV-3) expressing GFP (BAd-GFP), and porcine AdV type 3 (PAdV-3) expressing GFP (PAd-GFP) were 72:1, 608:1, and 100:1, respectively (23). In another study, the VP:PFU ratios for HAdV-C5 expressing luciferase and metabolically biotinylated fiber-mosaic AdV were 9:1 and 530:1, respectively (26).

In an effort to determine a ratio between the 2 best assays, which could effectively address the variability in titration of various AdV vectors in different cell lines, we selected HAdV-C5, BAdV-3, and PAdV-3 and propagated them in human embryonic kidney (HEK) 293 (27), Madin-Darby bovine kidney (MDBK), and porcine kidney (PK)-15 cell lines, respectively, by infecting each cell line with a multiplicity of infection of 5 PFU per cell. All 3 AdVs were of mammalian origin; however, for an elaborative study, we could have used mammalian, avian, reptile, and amphibian AdVs. We chose HAdV-C5, BAdV-3, and PAdV-3 for our study since these AdVs use different receptors for virus internalization (28–31). The rationale of using only wildtype AdVs was to avoid any adverse impact on vector growth characteristics due to a transgene. Each AdV was purified by cesium chloride density gradient centrifugation (32). The purified AdV preparations were titrated for PFU, TCID₅₀, focus-forming units (FFU), VP, and genome copy number (GCN). Titration of HAdV-5, BAdV-3, and PAdV-3 for PFU, TCID₅₀, and FFU were conducted using HEK 293, MDBK, and PK-15 cell lines, respectively.

For PFU, serially log-diluted HAdV-C5, BAdV-3, or PAdV-3 was used to infect HEK 293, MDBK, or PK-15 cell lines and the plaques were counted on day 7, 10, or 12 post-infection, respectively. For TCID₅₀, serially log-diluted HAdV-C5-, BAdV-3-, or PAdV-3-infected cells were examined for cytopathic effect on day 5, 7, or 10 post-infection, respectively, and the TCID₅₀ titers were determined by the Reed and Muench formula (33):

$$\text{proportionate\hspace{0.17em}distance}=\frac{50\%-\text{dilution\hspace{0.17em}just\hspace{0.17em}below\hspace{0.17em}}50\%\hspace{0.17em}\text{CPE}}{\text{dilution\hspace{0.17em}just\hspace{0.17em}above\hspace{0.17em}}50\%\hspace{0.17em}\text{CPE}-\text{dilution\hspace{0.17em}just\hspace{0.17em}below\hspace{0.17em}}50\%\hspace{0.17em}\text{CPE}}$$

For FFU, serially log-diluted HAdV-C5, BAdV-3, or PAdV-3 was used to infect HEK 293, MDBK, or PK-15 cell lines, respectively. Three to 5 days post-infection, infected cell monolayers were fixed with cold methanol. Virus-infected individual cells or cell clusters were incubated at room temperature for 1 h with 1:1000 diluted rabbit hyperimmune serum against HAdV-C5, BAdV-3, or PAdV-3 (34). The plates were washed carefully 4× with phosphate-buffered saline (PBS), incubated at room temperature for 1 h with 1:2000 diluted goat anti-rabbit horseradish peroxidase-conjugated antibody and washed 4× with PBS. The plates were incubated for 30 min with a solution containing 0.015% hydrogen peroxide and 0.05% diaminobenzidine and washed twice with PBS. The clusters of virus-infected cells and individually infected cells were stained dark brown with a light brown background. The total number of virus-infected cells was counted under a light microscope and the virus titer in FFU/mL was determined (35) (Figure 1). Numbers of VP in purified preparations of HAdV-C5, BAdV-3, and PAdV-3 were determined spectrophotometrically by measuring optical density (OD) at 260 nm. For the determination of GCN, 1 mL of purified preparations of HAdV-C5, BAdV-3, or PAdV-3 was used to extract viral DNA (36) and the amount of DNA was determined spectrophotometrically by measuring the absorbance (i.e., OD) at 260 nm. Finally, GCN was determined using the genome size in nucleotides with the following formula (37):

Figure 1.

Focus-forming units (FFU) assay for human adenovirus type C5 (HAdV-C5), bovine adenovirus type 3 (BAdV-3), and porcine adenovirus type 3 (PAdV-3). Human embryonic kidney (HEK) 293, Madin-Darby bovine kidney (MDBK), or porcine kidney (PK)-15 cells were infected with serially log-diluted purified preparations of HAdV-C5, BAdV-3, or PAdV-3, respectively.

$$\text{number\hspace{0.17em}of\hspace{0.17em}virus\hspace{0.17em}genome\hspace{0.17em}copies}=\frac{\text{DNA\hspace{0.17em}amount\hspace{0.17em}}(\text{ng})\times 6.022\times {10}^{23}}{\text{DNA\hspace{0.17em}length\hspace{0.17em}}(\text{bp})\times {10}^9\times 650}$$

Each assay was done in triplicate and 3 independent preparations of each virus were used with similar results.

Plaque-forming units, TCID₅₀, FFU, VP, and GCN titers (log mean ± SD) with the purified preparation of HAdV-C5 were 11.19 ± 0.04, 11.26 ± 0.08, 11.5 ± 0.08, 13.18 ± 0.01, and 12.43 ± 0.02, respectively (Figure 2A). Similarly, PFU, TCID₅₀, FFU, VP, and GCN titers (log mean ± SD) with the purified preparation of BAdV-3 were 10.27 ± 0.03, 10.26 ± 0.10, 10.4 ± 0.10, 12.56 ± 0.03, and 12.29 ± 0.02, respectively (Figure 2B). Whereas, PFU, TCID₅₀, FFU, VP, and GCN titers (log mean ± SD) with the purified preparation of PAdV-3 were 9.53 ± 0.32, 10.39 ± 0.21, 10.42 ± 0.20, 12.57 ± 0.03, and 12.52 ± 0.03, respectively (Figure 2C).

Figure 2.

Titration of human adenovirus type C5 (HAdV-C5), bovine adenovirus type 3 (BAdV-3), and porcine adenovirus type 3 (PAdV-3) by plaque-forming units (PFU), tissue culture infectious dose 50 (TCID₅₀), focus-forming units (FFU), virus particle (VP), and genome copy number (GCN). Titers of A — HAdV-C5, B — BAdV-3, and C — PAdV-3 by various assays and VP ratios with other titration assays for D — HAdV-C5, E — BAdV-3, and F — PAdV-3 are shown. The statistical analysis was done by unpaired t-test with Welch’s correction. *P < 0.05, ***P < 0.001, ****P < 0.0001. ns — not significant at P > 0.05.

Since plaque formation by PAdV-3 in PK-15 cells was not as efficient as that of HAdV-C5 in HEK 293 cells or BAdV-3 in MDBK cells, PAdV-3 titer in PFU was significantly lower than PAdV-3 titers in TCID₅₀ or FFU. In this study, PAdV-3 served as an AdV with less efficient plaque-forming ability, which may be due to the less efficient cell-to-cell virus transmission and the overgrowth of PK-15 cells with time, suggesting that the plaque assay is not the best choice for titration of every AdV or AdV vector. Either TCID₅₀ or FFU could be an excellent alternative to plaque assay for titration of any AdV or AdV vector that replicates in a cell line. However, our results with HAdV-C5, BAdV-3, and PAdV-3 suggest that FFU could provide more consistent results for AdVs or AdV vectors compared to PFU or TCID₅₀. Unlike PFU or TCID₅₀, FFU seems to work even in situations in which cell-to-cell virus transmission is limited. Focus-forming unit titration could easily be adapted for foreign gene expression in place of viral gene expression using the transgene-specific antibody for any type of AdV vector. Focus-forming units titration is more tedious compared to titration for PFU or TCID₅₀, but the results are available earlier than PFU or TCID₅₀ since FFU can be performed a few days post infection.

The other 2 assays included in this study, VP and GCN, are not dependent on virus replication. The protocol for VP is straightforward and quick because VP can be estimated directly by measuring the absorbance at 260 nm of a purified preparation of the AdV or AdV vector. For the GCN assay, however, the viral genomic DNA needs to be extracted from a purified preparation before estimating GCN from the absorbance at 260 nm of a purified preparation of AdV genome. Since the VP assay is affected by the total protein in the preparation, the VP count includes empty, ruptured, and mature AdV particles. The GCN assay is based on the total DNA estimation, so the GCN count consists predominantly of mature AdV particles (both containing the full or partial genomes). The VP titer, therefore, will always be higher than the GCN titer of the same AdV preparation. For titration of various AdV vectors, VP titers are widely used in preclinical and clinical studies (38,39).

Since AdV vectors differ in type of vector, type of virus, and species of origin, and the cell line used for vector propagation, it is difficult to anticipate that a single vector titration method will be applicable for all AdV vectors. To address this issue, a ratio between 2 titration assays such as VP:PFU is often used for AdV vectors (23,40). VP:PFU ratios for purified preparations of HAdV-C5, BAdV-3, and PAdV-3 were 93.33, 206.91, and 1047.50, respectively; VP:TCID₅₀ ratios for HAdV-C5, BAdV-3, and PAdV-3 were 68.29, 215.15, and 134.94, respectively; VP:FFU ratios for HAdV-C5, BAdV-3, and PAdV-3 were 35.47, 139.67, and 134.08, respectively; and VP:GCN ratios for HAdV-C5, BAdV-3, and PAdV-3 were 4.14, 1.96, and 1.08, respectively (Figures 2D to F).

Various other combinations of ratios were examined, but the ratios with either VP or GCN provided similar conclusions. Since PAdV-3 was not efficient in producing plaques in PK-15 cells, a higher VP:PFU ratio of 1047.50 compared to 93.33 and 206.91 for HAdV-C5 and BAdV-3, respectively, clearly indicated the poor plaque-forming ability of PAdV-3. The VP:TCID₅₀ and VP:FFU ratios provided somewhat similar results, but FFU could be applied to all types of vectors, including conditional replication-competent and helper-dependent vectors, suggesting that the VP:FFU ratio would give the best outcome compared to VP:PFU or VP:TCID₅₀ ratios. The measurement of transgene expression is important for vaccine or gene therapy applications, but for conditional replication-competent vectors, especially for cancer therapy, there is no need to assess transgene expression. Since transgene expression can be measured by several other techniques, there is no need to combine it with vector titration. Thus, it seems that the VP:FFU or VP:GCN ratio could be applied to all types of AdV vectors.

Normalizing the vector amount for the same type of vector by using either the VP:FFU or VP:GCN ratio will be important for the reproducibility of preclinical and clinical studies with AdV vectors. This process will also reduce the impact of batch-to-batch variability and variability due to vector type. For the benefit of all researchers, it will be critical to include the VP:FFU or VP:GCN ratio for each vector preparation in every manuscript. It will certainly help everyone compare results with different AdV vectors and avoid vector dose-associated toxicity.

In summary, this study examined 5 methods of AdV titration — PFU, TCID₅₀, FFU, VP, and GCN — in an attempt to determine the best ratio to apply to various AdV preparations for assessing vector quality. Our findings indicate that a VP:FFU or VP:GCN ratio could serve as a guideline to adjust batch-to-batch variability in vector quality or to adjust the vector dose for another type of AdV vector for preclinical or clinical evaluations. The inclusion of the VP:FFU or VP:GCN ratio for each AdV vector in every manuscript could assist in interpreting the results from various laboratories using different AdV vectors. Additional studies with different types of AdV vectors in cell culture, as well as in animal models, are required to fully explore the advantages and challenges of the best titration ratio described in this manuscript.