Genome concentration, characterization, and integrity analysis of recombinant adeno-associated viral vectors using droplet digital PCR

Abstract

Precise, reproducible characterization of AAV is critical for comparing preclinical results between laboratories and determining a safe and effective clinical dose for gene therapy applications. In this study, we systematically evaluated numerous parameters to produce a simple and robust ddPCR protocol for AAV characterization. The protocol uses a low ionic strength buffer containing Pluronic-F68 and polyadenylic acid to dilute the AAV into the ddPCR concentration range and a 10-minute thermal capsid lysis prior to assembling ddPCR reactions containing MspI. A critical finding is that the buffer composition affected the ITR concentration of AAV but not the ITR concentration of a double stranded plasmid, which has implications when using a theoretical, stoichiometric conversion factor to obtain the titer based on the ITR concentration. Using this protocol, a more comprehensive analysis of an AAV vector formulation was demonstrated with multiple ddPCR assays distributed throughout the AAV vector genome. These assays amplify the ITR, regulatory elements, and eGFP transgene to provide a more confident estimate of the vector genome concentration and a high-resolution characterization of the vector genome identity. Additionally, we compared two methods of genome integrity analysis for three control sample types at eight different concentrations for each sample. The genome integrity was independent of sample concentration and the expected values were obtained when integrity was determined based on the excess number of positive droplets relative to the number of double positive droplets expected by chance co-encapsulation of two DNA targets. The genome integrity was highly variable and produced unexpected values when the double positive droplet percentage was used to calculate the genome integrity. A protocol using a one-minute thermal capsid lysis prior to assembling ddPCR reactions lacking a restriction enzyme used the non-ITR assays in a duplex ddPCR milepost experiment to determine the genome integrity using linkage analysis.

Fig 1. Enzyme effects on the ITR2 concentration.

Prior to droplet formation, ddPCR reactions were prepared using a single-stranded vector genome plasmid, pssAAV2, and the indicated restriction nuclease or nucleases. The concentration of ITR and eGFP for each enzyme condition is indicated by the black and gray bars respectively. The ITR2 to eGFP concentration ratio is indicated. The reported numeric errors and error bars represent the 95% confidence interval.

Fig 2. Effect of buffer additives on DNase I activity and eGFP concentration.

DNase I reactions were prepared that comprised RPP30 gBlock and AAV2 with different additives. Samples were incubated at 37°C for 30 min and then serially diluted into the ddPCR concentration range with polyA buffer. Capsids were lysed at 95°C for 10 min prior to assembling ddPCR reactions. (A) RPP30 concentration without and with DNase I. The estimated RPP30 concentration for samples containing DNase I is indicated with numerical values because the bars are not visible. (B) eGFP concentration with DNase I digestion. Error bars represent the 95% confidence interval.

Fig 3. Buffer effects on the viral genome ITR2 concentration.

Prior to droplet formation, ddPCR reactions were prepared using a single-stranded viral vector, AAV2, that was diluted into the ddPCR range with either polyA buffer or PCR buffer and lysed for 10 min at 95°C before being used as the template for ddPCR. Two separate duplex reactions with 5 U MspI were prepared: ITR2/eGFP and SV40/eGFP. (A) The concentration from the ITR (black bar) and eGFP (black striped bar) duplex reaction is shown. The concentration from the SV40 and eGFP duplex reaction is indicated by the gray and gray striped bars respectively. (B) The concentration ratio of ITR2 to eGFP (black bars) and SV40 to eGFP (gray bars) is indicated. The error bars represent the 95% confidence interval.

Fig 4. Effect of buffer composition on the viral genome concentration.

Seven different dilution buffer formulations were used to dilute AAV2 after DNase I digestion. Buffer identities are listed in S1 Text. Buffers 1‒3 are based on DNA suspension buffer and buffers 4‒7 are based on PCR buffer. For reference, buffer 1 is polyA buffer and buffer 7 is PCR buffer. Duplex reactions with ITR2 FAM and eGFP HEX were analyzed. (A) The concentration of the ITR (black bars) and eGFP (gray bars). (B) The concentration ratio of ITR2 to eGFP. The error bars represent the 95% confidence interval.

Fig 5. Effect of F68 on the AAV2 viral genome concentration.

PolyA buffer and PBS with 100 ng/μL polyadenylic acid (PBS + pA) were prepared with Pluronic F-68 concentrations that varied from 0–0.1% and used to dilute AAV2 after DNase I digestion. Duplex reactions with ITR2 FAM and eGFP HEX were analyzed. (A) The concentration of the ITR (black bars) and eGFP (gray bars). (B) The concentration ratio of ITR2 to eGFP. The error bars represent the 95% confidence interval.

Fig 6. Comparison of pre-droplet and in-droplet capsid lysis.

After DNase I digestion, viruses were serially diluted using polyA+ buffer into the ddPCR concentration range. The viral samples were either lysed at 95°C for 10 min (pre-droplet) or added directly (in-droplet) to ddPCR reactions and lysed using the polymerase activation step of the PCR thermal cycle. The (A) eGFP concentration and (B) ITR2/eGFP concentration ratio. The error bars represent the 95% confidence interval.

Fig 7. Effect of proteinase K on AAV5.

After DNase I digestion, AAV5 was incubated either without (no PK) or with (PK) proteinase K for 2 hours at 50°C and then further processed as described in the Materials and Methods. The two samples were serially diluted in parallel with polyA+ buffer and either added directly to the ddPCR supermix (no capsid lysis step) or had the capsid thermally lysed (with capsid lysis step) prior to adding to the ddPCR supermix. The ITR2 concentration (black bars) and eGFP concentration (gray bars) is shown. The values indicate the ITR2/eGFP concentration ratio. The error bars and numerical errors represent the 95% confidence interval.

Fig 8. High-resolution AAV vector genome characterization.

Viruses were prepared using the optimized workflow and the genomes were characterized using singleplex ddPCR assays against the indicated targets for (A) rAAV2 RSS and (B) AAV2. Error bars represent the 95% confidence interval.

Fig 9. Comparison of genome integrity methods.

Female DNA with HaeIII was used as a template for duplex ddPCR reactions using RPP30 FAM and SOD1 HEX and the data was analyzed as (A) the percentage of double positive droplets or (B) linkage percentage. (C and D) Plasmid pAV-CMV-GFP with either HindIII or MspI using CMV-Enh FAM and eGFP HEX assays and (E and F) synthetic DNA with HaeIII using RPP30 FAM and SOD1 HEX assays were analyzed correspondingly. The means and standard deviations are indicated. The standard deviations are smaller than the marker when not visible. RS: gBlock with RPP30 and SOD1 target sequences that has no HaeIII site between the assays. R/S: gBlock with RPP30 and SOD1 target sequences on the same molecule with a single HaeIII site between the assays.

Fig 10. Milepost experiment using pAV-CMV-GFP.

Plasmid pAV-CMV-GFP was used as a template for duplex ddPCR reactions using CMV-Enh FAM and the indicated HEX assay. (A) Schematic showing approximate distances from the end of the CMV-Enh amplicon (blue) and the beginning of the HEX amplicons (green). Reactions were prepared with 5 U of (B) HindIII or (C) MspI. The concentration of CMV-Enh (black bar), the HEX assay (gray bar), and linkage concentration (white bar) is shown with the corresponding standard deviations. The calculated linkage percentage between CMV-Enh FAM and the corresponding duplexed HEX assay is indicated.

Fig 11. Milepost experiment using AAV2.

An AAV2 sample was DNase I treated and lysed for 10 min at 95°C prior to droplet formation (pre-droplet lysis) and then used as a template for duplex ddPCR reactions using CMV-Enh FAM and the indicated HEX assay with either no enzyme, MspI, MseI, or a double digest with MspI and MseI. The (A) calculated linkage percentage and (B) concentration ratio of CMV-Enh FAM to the HEX assay with the corresponding standard deviations are shown.

Fig 12. Effect of lysis time on linkage.

(A) AAV2 and (B) AAV5 samples were either lysed in droplets (in drop) or prior to droplet formation for 1, 2.5, 5, 10, or 15 minutes at 95°C and used as a template for a milepost analysis using CMV-Enh FAM and the indicated HEX assay. The error bars represent the standard deviations.

Fig 13. Linkage percentage of AAV2.

The viral vector AAV2 was lysed prior to droplet formation for 1 minute at 95°C and then used as a template for a milepost experiment using CMV-Enh_s-FAM as the anchor assay and either CMV-Pro, eGFP, GFP-L or SV40 as the HEX assay. The linkage percentage was calculated using equations based on an average concentration for the FAM and HEX assays (average) or equations that compensated for differences in assay concentration between the FAM and HEX assay (compensated). The error bars represent the standard deviation and are smaller than the marker when not visible.