Origins of truncated supplementary capsid proteins in rAAV8 vectors produced with the baculovirus system

Abstract

The ability to produce large quantities of recombinant Adeno-Associated Virus (rAAV) vectors is an important factor for the development of gene therapy-based medicine. The baculovirus/insect cell expression system is one of the major systems for large scale rAAV production. So far, most technological developments concerned the optimization of the AAV rep and cap genes in order to be expressed correctly in a heterologous system. However, the effect of the baculovirus infection on the production of rAAV has not been examined in detail. In this study we show that the baculoviral cathepsin (v-CATH) protease is active on several (but not all) rAAV serotypes, leading to a partial degradation of VP1/VP2 proteins. Subsequently, we identified the principal v-CATH cleavage site in the rAAV8 capsid proteins and demonstrated that the cleavage is highly specific. The proteolytic degradation of VP1/VP2 AAV capsid proteins reduces the infectivity of rAAV vectors but can be prevented by the use of a baculovirus vector with a deletion of the chiA/v-cath locus or by addition of the E64 protease inhibitor during production. Moreover, the codon optimization of AAV cap performed for several serotypes and originally aimed at the removal of potential alternative initiation codons, resulted in incorporation of additional forms of truncated VP1 into the rAAV capsids.

Fig 1. The baculovirus v-CATH protease induces rAAV8 capsid degradation.

rAAV8 vectors encoding the γSGC transgene were produced in the baculovirus system with the v-cath and chiA genes deleted (ΔchiA/v-cath) or with the unmodified bacmid system (WT) and purified using immuno-affinity chromatography. The protease inhibitor E64 was added to the WT production runs at various times post infection (0, 24, 72 or 96h p.i.). The cells were harvested and lysed at 96h p.i. In the final sample E64 was added following harvest, cell lysis and clarification of the cell culture. Samples of 3 x 10⁹ vg of rAAV8 γSGC were loaded on the SDS-PAGE followed by Coomassie staining (A) or by Western blot using Anti VP B1 antibody (B). The red dash (-) pinpoints to the major 61 kDa degradation band originating from VP1/VP2 unique domains. The asterisk (*) indicates a minor degradation band originating from the VP1/VP2 unique domain. The hash (#) indicates minor degradation band originating from VP1 unique domain.

Fig 2. rAAV8 capsid degradation by v-CATH studied with anti-VP polyclonal antibody.

rAAV8 vectors encoding the γSGC transgene were produced in an unmodified bacmid system in the presence or the absence of E64 protease inhibitor and purified using immuno-affinity chromatography. Purified vectors were loaded on SDS-PAGE stained with Coomassie blue (left panel). The corresponding Western blot is shown on the right panel.

Fig 3. The baculovirus v-CATH protease does not degrade empty rAAV8 capsids.

Empty rAAV8 capsids were produced in unmodified bacmid system by infecting Sf9 cells only with Bac-rep/cap baculovirus at MOI 0.05, purified using immuno-affinity chromatography, analyzed on SDS-PAGE and revealed by Coomassie staining.

Fig 4. Amino acid sequence alignment of VP proteins of AAV serotype 1 to rh10 at the major cleavage site induced by v-CATH.

Amino acid sequence of AAV VP1 proteins of serotype 1 to 9 and rh10 were aligned using multalin software [35]. The display includes amino acids comprised from aa 151 to 200. Amino acids conserved among all studied rAAV serotypes are displayed in red color. Conserved amino acids among several serotypes are displayed in blue color. Gaps introduced into the alignment are represented by dash police. rAAV serotypes (1, 6, 8) cleaved by the baculovirus v-CATH are displayed in bold characters at the major cleavage position in the rAAV VP amino acid sequences. Minor cleavage site happens in rAAV8 following LN amino acids in bold italic police in the figure. Both major and minor cleavage sites by v-CATH in rAAV8 are indicated by vertical red arrows. The amino-acid sequences of rAAV serotypes (2, 9, rh10) not cleaved by v-CATH but corresponding to the amino acid region cleaved for susceptible serotypes are underlined.

Fig 5. Effect of v-cathepsin protease on various rAAV serotypes.

rAAV vectors of serotypes 1, 2, 6, 8, 9, rh10 encoding the γSGC transgene were produced in an unmodified bacmid system in the presence or in the absence of E64 protease inhibitor and purified using immuno-affinity chromatography. VP proteins were revealed by Western blot using Progen B1 antibody. Supplementary bands originating from codon optimization performed in the cap genes are marked by an asterisk (*). Supplementary bands originating from v-CATH degradation are marked with a dash (-).

Fig 6. Mutation of V-CATH major cleavage site in AAV8 VP1/2 proteins confirms high specificity of the protease.

Based on sequence analysis of several AAV VP serotype sequences, key amino acids Leu₁₈₉ was identified as critical for susceptibility to V-CATH cleavage, while Glu₁₉₁ was suspected to be involved in v-CATH susceptibility. Mutated rAAV8 (on cap8 codon optimized sequence) encoding the γSGC transgene were produced in an unmodified bacmid system, purified and studied by Western blot using anti-VP B1 antibody. From left to right, rAAV8 samples with amino acid sequence Leu₁₈₉-Gly₁₉₀-Glu₁₉₁ (as found in WT cap8 sequence); Ile₁₈₉-Gly₁₉₀-Gln₁₉₁ (double mutant); Ile₁₈₉-Gly₁₉₀-Glu₁₉₁ (single mutant); Leu₁₈₉-Gly₁₉₀-Gln₁₉₁ (single mutant). Major degradation product of the VP1/2 unique domain by V-CATH is signaled by a dash. Secondary degradation product of VP1/2 unique domain by V-CATH is signaled by *. AAV8 was produced in absence (A) or presence (B) of E64 protease inhibitor, purified and then analyzed.

Fig 7. Supplementary truncated VP1 band is a product of translation and not a degradation product.

Origin of supplementary truncated VP1 protein band (signaled on the figure by *) was studied by Western blot using anti-VP B1 antibody. (1) AAV2 WT. 5 x 10⁹ vg of AAV2 WT produced by transfection of HEK 293 cells. (2) Empty AAV2. AAV2 empty capsids produced with baculovirus deleted of chiA/v-cath genes, with the cap2 gene codon optimized. (3) Empty AAV2 with silenced VP1 expression. AAV2 empty capsids produced with baculovirus deleted of chiA/v-cath genes, with cap gene codon optimized, including silencing of VP1 alternative start codon (ACG → ACT).

Fig 8. WT cap8 gene does not lead to additional truncated VP product in rAAV8 capsid.

rAAV8 was produced in a bacmid system with the v-cath and chiA genes deleted. WT cap8 gene sequence was used without any of the codon optimization. Western blot against VP proteins was used on bulk samples of Sf9 cells infected with baculovirus deleted of chiA/v-cath genes encoding WT cap8 gene.

Fig 9. In vivo evaluation of rAAV vectors.

rAAV8-mSeAP produced either using unmodified baculovirus (WT) or baculovirus with v-cath and chiA and p10 genes inactivated (ΔCCΔp10). 10⁹ vg of purified AAV8 were injected intramuscularly into mouse TA muscle (n = 4 per type of vector) and 2 mice were injected with PBS (as negative control). (A) Time course expression of seric mSeAP was measured on days 0, 2, 7, 14, 21, 28, and 35 post injection and is expressed as counts per second (* p<0.05). (B) The number of rAAV-mSeAP genome present in the injected muscles was measured and normalized to the level of the titin gene (cellular reference gene) (** p<0.01).