Further Improved AAV Vectors Produced in Insect Cells

Abstract

The present invention relates to the production of adeno-associated viral vectors in insect cells. The insect cells therefore comprise a first nucleotide sequence encoding the adeno-associated virus (AAV) capsid proteins, whereby the initiation codon for translation of the AAV VP1 capsid protein is a non-ATG, suboptimal initiation codon and wherein the coding sequence for one or more amino acid residues have been inserted between the suboptimal translation initiation codon and the codon encoding the amino acid residue that corresponds to the amino acid residue at position 2 of the wild type capsid amino acid sequence of which the first amino acid residue is alanine, glycine valine, aspartic acid or glutamic acid. The insect cell further comprises a second nucleotide sequence comprising at least one AAV inverted terminal repeat (ITR) nucleotide sequence; a third nucleotide sequence comprising a Rep52 or a Rep40 coding sequence operably linked to expression control sequences for expression in an insect cell; and, a fourth nucleotide sequence comprising a Rep78 or a Rep68 coding sequence operably linked to expression control sequences for expression in an insect cell. The invention further relates to adeno-associated viral vectors with an altered ratio of the viral capsid proteins.

Claims

1.-16. (canceled)

17. A nucleic acid molecule having a first nucleotide sequence comprising an open reading frame (ORF), wherein the ORF in 5′ to 3′order comprises: (i) a first codon, which is a suboptimal translation initiation codon selected from the group consisting of CTG, ACG, TIG and GTG; (ii) a second codon encoding an amino acid residue selected from the group consisting of alanine, glycine, valine, aspartic acid and glutamic acid; (iii) optionally, one or more codons encoding additional amino acid residues following the second codon; and, (iv) a sequence encoding adeno-associated virus (AAV) capsid proteins including VP1, which sequence lacks only the translation initiation codon of VP1.

18. The nucleic acid molecule according to claim 17, wherein the AAV capsid proteins are AAV serotype 5, AAV serotype 8, or AAV serotype 9 capsid proteins,

19. The nucleic acid molecule according to claim 18, wherein the capsid proteins have an amino acid sequence selected from the group consisting of SEQ ID NO:22, SEQ ID NO:28, SEQ ID NO:30, SEQ ID NO:71 and SEQ ID NO:73.

20. The nucleic acid molecule according to claim 1, wherein the second codon encodes alanine.

21. The nucleic acid molecule according to claim 17, wherein the second codon is selected from the group consisting of GCT, GCC, GCA, GCG and GGU.

22. The nucleic acid molecule according to claim 21, wherein the second codon is GCT

23. A first nucleic acid construct comprising the nucleic acid molecule according to claim 17, wherein the first nucleotide sequence encoding the AAV capsid proteins is operably linked to an expression control sequence for expression in an insect cell.

24. The nucleic acid construct according to claim 23, wherein the nucleotide sequence of the ORF is operably linked to a promoter selected from the group consisting of polyhedron promoter, p10 promoter, 4×Hsp27 EcRE+minimal Hsp70 promoter, deltaE1 promoter and E1 promoter.

25. The nucleic acid construct according to claim 23, wherein the construct is an insect-compatible vector.

26. The nucleic acid construct according to claim 25, wherein the vector is a baculoviral vector.

27. The nucleic acid construct according to claim 23, wherein the nucleic acid molecule comprises an ORF selected from the group consisting of: SEQ ID NO:51, SEQ ID NO:69, SEQ ID NO:42, SEQ ID NO:47, SEQ ID NO:48 and SEQ ID NO:50.

28. The nucleic acid construct according to claim 27, wherein the ORF is SEQ ID NO:51.

29. An insect cell comprising the first nucleic acid construct according to claim 23.

30. The insect cell according to claim 29, wherein the insect cell further comprises: (a) a second nucleotide sequence comprising at least one AAV inverted terminal repeat (ITR) nucleotide sequence; (b) a third nucleotide sequence comprising a Rep78 or a Rep68 coding sequence operably linked to an expression control sequence for expression in an insect cell; (c) optionally, a fourth nucleotide sequence comprising a Rep52 or a Rep40 coding sequence operably linked to an expression control sequence for expression in an insect cell.

31. The insect cell according to claim 30, comprising: (a) the first nucleic acid construct that further comprises the third and fourth nucleotide sequences as defined in claim 30(b) and (c); and, (b) a second nucleic acid construct comprising the second nucleotide sequence of claim 30(a), which is an insect cell-compatible vector.

32. The insect cell according to claim 31, wherein the second nucleic construct of (b) is a baculoviral vector.

33. The insect cell according to claim 30, wherein the second nucleotide sequence further comprises at least one nucleotide sequence that encodes a gene product of interest for expression in a mammalian cell that becomes incorporated into the genome of an AAV serotype 5 produced in the insect cell

34. The insect cell according to claim 33, wherein the second nucleotide sequence comprising said at least one nucleotide sequence encoding said gene product of interest comprises two AAV ITR nucleotide sequences between which the sequence encoding the gene product of interest is located.

35. The insect cell according to claim 30, wherein the first nucleotide sequence, the second nucleotide sequence, the third nucleotide sequence and, optionally, the fourth nucleotide sequence are stably integrated in the genome of the insect cell.

36. An AAV virion comprising a VP1 capsid protein and comprising in its genome at least one nucleotide sequence encoding a gene product of interest, wherein: (a) the at least one nucleotide sequence encoding the gene product of interest is not a native AAV nucleotide sequence, (b) the AAV VP1 capsid protein comprises, from N terminus to C terminus: (i) a first amino acid residue encoded by a translation initiation codon; (ii) a second amino acid residue selected from the group consisting of alanine, glycine, valine, aspartic acid and glutamic acid; (iii) optionally, one or more additional amino acid residues following the second amino acid residue; and (c) encoded by the native VP1 translation initiation codon.

37. The AAV virion of claim 36, wherein the translation initiation codon encoding the first amino acid residue is a suboptimal initiation codon selected from the group consisting of CTG, ACG, TTG and GTG.

38. The AAV virion according to claim 36, wherein the gene product of interest encodes a Factor IX or a Factor VIII protein.

39. The AAV virion according to claim 38, wherein the Factor IX or Factor VIII protein is a human Factor IX or Factor VIII protein.

40. A method for producing AAV virions in an insect cell, comprising the steps of: (a) culturing the insect cell of claim 29 under conditions such that AAV virions are produced; and (b) optionally recovering the AAV virions.

Description

DESCRIPTION OF THE FIGURES

[0101] FIG. 1: Various mutant capsids harbouring reporter transgene SEAP were purified and resolved on an NuPage gel. Three capsid proteins, VP1 (87 kDa), VP2 (72 kDa) and VP3 (62 kDa) are shown.

[0102] FIG. 2: In vitro potency assay with various AAV5 capsid mutants carrying seap expression cassette in Hela cells. The activity of the reporter gene is measured indirectly as emission of light and is expressed in RLU (relative light units). NTC=negative control.

[0103] FIG. 3: In vitro potency assay with various AAV5 capsid mutants carrying seap expression cassette in Huh7 cells. The activity of the reporter gene is measured indirectly as emission of light and is expressed in RLU (relative light units). NTC=negative control.

[0104] FIG. 4: In vivo potency assay of various capsid mutants carrying seap expression cassette in C57BL/6 mice. The activity of the reporter gene is measured indirectly as emission of light and is expressed in RLU (relative light units).

[0105] FIG. 5: In vivo potency assay of various AAV5 capsid mutants carrying FIX expression cassette in C57BL/6 mice. FIX expression was monitored in mice upon administration of two different vectors i.e. capsid variant 160 and 765. Both capsids carry FIX expression cassette. FIX is measured in plasma at week 1, 2 and 4 post injections by means of specific ELISA. IU/ml represents international units of FIX protein found in 1 ml of plasma. PBS—phosphate buffered saline.

[0106] FIG. 6: Mutant capsids harbouring reporter transgene SEAP were purified and resolved on an NuPage gel to show the three capsid proteins VP1, VP2 and VP3. Three clones of construct 43 are shown.

[0107] FIG. 7: In vitro potency assay with various AAV5 capsid mutants carrying seap expression cassette in HeLa cells (A) and in Huh7 cells (B). The activity of the reporter gene is measured indirectly as emission of light and is expressed in RLU (relative light units) (a.u.: arbitrary units).

EXAMPLES

1. Introduction

[0108] The initial baculovirus system for production of rAAV was described by Urabe et al (Urabe et al. [2002] Human Gene Therapy 13(16):1935-1943) and consists of three baculoviruses, namely Bac-Rep, Bac-cap and Bac-vec, co-infection of which into insect cells e.g. SF9 resulted in generation of rAAV. The properties of such produced rAAV, i.e. physical and molecular characteristic including potency, did not differ significantly from the rAAV generated in mammalian cells (Urabe [2002] supra). In order to accomplish efficient generation of rAAV vectors in insect cells the AAV proteins needed for the process had to be expressed at appropriate levels. This required a number of adaptations of operons encoding for Rep and Cap proteins. Wild type AAV expresses large Rep78 to small Rep52 from two distinct promoters p5 and p19 respectively and splicing of the two messengers results in generation of Rep68 and Rep52 variants. This operon organization results in limited expression of Rep78 and relatively higher expression of Rep52. In order to mimic the low 78 to 52 ratio Urabe and colleagues constructed a DNA cassette in which expression of Rep78 was driven by the partially deleted promoter for the immediate-early 1 gene (ΔIE-1) whereas Rep52 expression was controlled by a strong polyhedrin promoter (polh). The spliced variants of large and small Reps were not observed in insect cells which likely relates to the difference in splicing processes between mammalian and insect cells. Another technical challenge to be overcome was related to the expression of the three major viral proteins (VP's). Wild type AAV expresses VP1, 2 and 3 from p40 promoter. Arising messenger RNA is spliced into two species: one responsible for VP1 expression whereas the second expresses both VP2 and VP3 via a “leaky ribosomal scanning mechanism” where the protein is initiated from non-canonical start i.e. ACG, is occasionally missed by the ribosome complex which than proceeds further until it finds the canonical start of VP3. Due to the differences in splicing machinery between vertebrate and insect cells the above described mechanism did not result in generation of proper capsids in insect cells. Urabe et al., decided to introduce a modification of translational start of VP1 which was similar to these found in the VP2 in such a way that the translational start of VP1 was changed to ACG and the initiation context, which consists of 9 nucleotides preceeding VP1, was changed to those preceeding VP2. These genetic alterations resulted in expression of the three VPs in the correct stoichiometry that could properly assemble into capsids from a single polycistronic mRNA. The transgene cassette on the other hand was similar to what was previously described for mammalian based systems, flanked by ITRs as the only in trans required elements for replication and packaging.

[0109] With the growing number of newly discovered AAV serotypes that hold different desired properties, there is a need for generation of these capsids in the BEV system. Although a successful production of AAV2 in the insect cells has been shown, not all serotypes perform equally well in the system adapted for AAV2. it seems that adapting a new serotype for optimum production and potency is not a trivial task and will require a tailor made approach. Previous attempts to adapt the rAAV5 sequence for production by BEVS in insect cells met a limited success, resulting in low incorporation of VP1 to the capsid (Kohlbrenner et al. (2005) Molecular Therapy 12 (6):1217-1225, Urabe et al. (2006) Journal of Virology 80(4):1874-1885). To circumvent this problem, Urabe et al. generated a chimeric type 2/5 virus which contains the N-terminal 136 amino acid residues from AAV type 2 and the remainder sequence from AAV serotype 5. Such virus was reported to produce well and to display similar potency to that of the wild type AAV5 (Urabe et al. (2006) supra). However, the resulting virion was a chimera and it does not represent the “true” rAAV5 serotype.

[0110] In order to generate genuine rAAV5 in insect cells with improved infectivity and/or potency, we designed several capsid protein 5 mutants. It seems important for the infectivity that the stoichiometry of the three viral proteins is balanced. For example, as previously reported we noticed that the lack of VP1 synthesis drastically influences the potency of the vector. Furthermore, we observed that the potency of the vectors was negatively correlated with the high incorporation of VP3 as compared to VP1 and VP2. Viral preparations with an excessive amount of VP3 were poor in transducing cells in vitro and in vitro. Finally we have constructed a genuine (or “true”) rAAV5 capsid which displays superior potency to the chimeric rAAV5 generated by Urabe et al (2006, supra). This new capsid was found to have balanced VP stoichiometry, and similar or superior potency as compared to the chimeric AAV2/5.

2. Methods

[0111] 2.1. Generation of rAAV5 Vectors

[0112] rAAV5 batches were generated by co-infecting expresSF+® insect cell line (Protein Sciences Corporation) with three different baculoviruses, which comprised expression cassettes for the capsid (rAAV5 variant library), replicase and transgene (Seap or Factor IX) under the control of a CMV and LP 1 promoter, respectively. Capsid expression cassettes were under the control of a polyhedron promoter. Rep expression cassettes were as described in WO 2009/14445 (BACVD183) and under control of a deltaE1 and polyhedron promoter driving expression of Rep78 and Rep52, respectively. ExpresSF+® cells were infected at a 5:1:1 (Rep:Cap:Transgene) volumetric ratio using freshly amplified baculovirus stocks. After a 72 hour incubation at 28° C., cells were lysed with 10× lysis buffer (1.5M NaCl, 0.5M Tris-HCl, 1 mM MgCl.sub.2, 1% Triton X-100, pH=8.5) for 1 hour at 28° C. Genomic DNA was digested by Benzonase treatment for 1 hour at 37′C. Cell debris was removed by centrifugation for 15 minutes at 1900×g after which the supernatant containing the rAAV5 particles was stored at 4° C. Vector titers were determined in this so-called crude cell lysate with a specific Q-PCR directed against the promoter region of the transgene. Briefly, affinity purified vectors were analysed by Q-PCR AAVs were treated with DNAse at 37° C. to degrade extrageneous DNA. AAV DNA was then released from the particles by 1M NaOH treatment. Following a short heat treatment (30 minutes at 37-C) the alkaline environment was neutralized with an equal volume of 1M HCl. The neutralized samples contained the AAV DNA that was used in the Taqman Q-PCR. Q-PCR was performed according to standard procedures using primers and probes listed in Table 1 below.

2.2. Purification of rAAV5 Vectors

[0113] rAAV5 particles were purified from crude lysates by a batch binding protocol using AVB sepharose (affinity resin, GE healthcare). rAAV5 crude cell lysates were added to washed (with 0.2M HPO.sub.4 pH=7.5 buffer) resin. Subsequently, samples were incubated for 2 hours at room temperature under gentle mixing. Following the incubation the resin was washed in 0.2M HPO.sub.4 pH=7.5 buffer and bound vectors were eluted by the addition of 0.2M Glycine pH=2.5. The pH of the eluted vectors was immediately neutralized by the addition of 0.5M Tris-HCl pH=8.5. Purified rAAV5 batches were stored at −20° C. Purified vectors were titered by a specific Q-PCR.

[0114] In order to generate higher vector amounts for in vivo study a modified purification protocol was used. Briefly, following the harvest, the clarified lysate was passed over a 0.22 μm filter (Millipak 60, 0.22 μm). Next, vector particles were affinity purified by means of a 8 ml AVB sepharose column (GE Healthcare) on a AKTA explorer (FPLC chromatography system, GE healthcare). Bound rAAV5 particles were eluted from the column with 0.2M Glycine pH=2.5. The eluate was immediately neutralized by 60 mM Tris HCl pH=7.5. The buffer of the neutralized eluates was exchanged to PBS 5% Sucrose with the help of 100 KDa ultrafiltration (Millipore) filter. The final product was then filtered on a 0.22 μm filter (Millex GP), aliquoted and stored at −20° C. until further use. Following the purification virus titers were determined with a specific Q-PCR

TABLE-US-00001 TABLE 1 TAQMAN Q-PCR primers SEQ ID Description NO: primers used for detection of Seap transgene pr59 AATGGGCGGTAGGCGTGTA CMV promotor fwd 55 pr60 AGGCGATCTGACGGTTCACTAA CMV promotor rev 56 pb12 TGGGAGGTCTATATAAGCAG CMV promotor probe Fam-MGB 57 primers used for detection of Factor IX transgene pr1103 CAAGTATGGCATCTACACCAAAGTCT FIX fwd 58 pr1104 GCAATAGCATCACAAATTTCACAAA FIX rev 59 pb25 TGTGAACTGGATCAAGGAGAAGACCAAGC FIX probe Fam-Tamra 60
2.3. VP Protein Composition of rAAV5 Variants

[0115] VP protein composition of purified rAAV5 variants was determined on Bis-tris polyacrylamide gels (Nupage, Life technologies) stained with Sypro Ruby. Briefly, 15 μl of purified rAAV5 was mixed with 5 μl 4×LDS loading buffer (Life technologies) and loaded on a Bis-Tris polyacrylamide gel. The samples were electrophoretically separated for 2 hours at 100 Volts. Following electrophoresis the proteins were fixed for 30 minutes with 10% NaAC/7% EtOH and stained with Sypro Ruby (Life technologies) for 2 hours VP proteins were then visualized under UV light on an ImageQuant system (GE Healthcare).

2.4. In Vitro Potency

[0116] To investigate in vitro potency of the different serotype 5 capsid variants, two continuous cell lines were used. Here, 1×10.sup.5 Hela and Huh7 were infected with rAAV5 variants at various multiplicity of infection. The experiments were performed in a 24-well plate with approximately 80% confluency at 1e5 cells/well. In both experiments wild type adenovirus was used at a multiplicity of infection of 30. This addition of wild type adenovirus is only applied in in vitro potency tests, in order to accelerate the process of second strand synthesis to within about 24 hours, thereby allowing the assay to be performed in a relatively shorter period of time and avoiding the need of cell passages. 48 hours after the start of the infection Seap expression was measured in the supernatant using the Seap reporter assay kit (Roche). Luminescence was measured on a Spectramax L luminometer (Molecular devices) at 470 nm with an integration time of 1 second.

2.5. In Vivo Potency

[0117] To investigate in vivo potency of the different serotype 5 capsid variants, two different experiments were performed. Briefly, the potency of rAAV5 vectors constructs 159-164 harbouring Seap reporter gene was investigated in C57BL/6 mice. Different vectors were injected intramuscularly in mice at a dose of 5×10.sup.12 gc/kg. Groups consisted of 5 mice each, 7 groups in total including a PBS group. Mice plasma was obtained 2, 4 and 6 weeks after the injection after which the mice were sacrificed. Seap activity was measured in the plasma using the Seap reporter assay kit from Roche. Luminescence was measured on a Spectramax L luminometer (Molecular devices) at 470 nm with an integration time of 1 second.

[0118] Next, the in vivo potency of variant AAV5(765) was compared to that of AAV5(160) and AAV5(92). AAV5(92) was a kind gift received from laboratory of dr. Kotin (Urabe et al, 2006) C57BL/6 mice were injected intravenously at doses of 2×10.sup.12 gc/kg and 2×10.sup.13 gc/kg with 765 or 160 both harbouring FIX as a reporter gene. In total seven groups of five mice each were injected including a PBS group Plasma was collected 1, 2 and 4 weeks following injection after which the mice were sacrificed. Factor IX protein present in the plasma was measured with a factor IX specific ELISA (VisuLize FIX antigen kit, Kordia). Optical density was measured at 450 nm on a Versamax ELISA plate reader (Molecular devices).

3. Results

[0119] 3.1. Generation of rAAV5 in RBEVS

[0120] AAV is a mammalian virus that uses its host's machinery to express its genes, among which a cap gene. The mechanism by which a correct stoichiometry of VP1:VP2:VP3 is achieved in a mammalian host are not present or are not optimal in insect cells. Therefore, Urabe et al., developed a strategy of genetic adjustments to organization of cap polycistronic mRNA which resulted in production of three VP's of AAV2 in insect cells at the correct stoichiometry (Urabe et al. (2002) supra). The attempts to establish similar methods to produce rAAV5 in BEVS proved to be unsuccessful to achieve sufficient infectious particles. Without wishing to be bound by any theory, this seems to be caused by a low incorporation of VP1 into the capsids (Urabe et al. (2006) supra). Thereby, Urabe et al., building on the previous success with the type 2 serotype, replaced the N-terminal portion of the type 5 VP1 with that of the type 2, to produce infectious AAV5 particles (Urabe et al. (2006) supra). Although successful, the chimeric AAV2/5 chimeric capsid does not comprise bonafide type 5 particles and as such may have altered properties as compared to AAV5, which could represent the combination of the two capsids rather than those from the type 5.

[0121] In order to allow for AAV5 virion production in insect cells with an improved infectivity and potency, in the present invention a series of genetic alteration to cap5 expression cassette of AAV5 were made (Table 2). As previously noted (Urabe et al. (2006) supra) the wild type cap5 gene (here clone number 763) did not support generation of rAAV. Lack of recognition of native AAV splicing signals in insect cells most likely resulted in low expression of separate VP's and lack of vector production. Due to the fact that eukaryotic ribosomes read mRNA unidirectional from 5′ to 3′, the first translation initiation start (here VP1) of polycistronic cap5 mRNA is detrimental for expression of all three proteins. The wild type initiation start is composed of ATG, a so-called strong translation initiation codon, that does not allow for ribosomal read through and thereby blocks the expression of other two VPs, which leads to lack of rAAV production. Due to the fact that wild type AAV uses ribosomal read through to express VP2 (non-canonical translation initiation start, ACG) and VP3 (ATG), lead us to investigate the translational start of VP1 and its immediate surroundings to alter the expression and/or assembly of three VP's.

[0122] It has been reported before that the nucleotide context of the translational start have an influence on the strength of the translational initiation (Kozak (1987) Nucleic Acid Research 15(20):8125-8148; WO2007/046703). The preferred nucleotides seem to be A at the position (−3) and G at the position (+4) with AUG counting +1, +2 and +13 respectively (Kozak supra; WO2007/046703). Table 2 details the specific changes that were introduced to the translational initiation start, its upstream and downstream context to tune the expression of three VPs. We have investigated the upstream initiation context that originally surrounds VP2 translational start; various non-canonical start codons (ACG, CTG, TTG, GTG), various mutagenic changes to the +2 wild type triplet and insertion between the +1 initiation triplet and the +2 wild type triplet. The expression cassettes encompassing combination of these features were used for generation of rAAV.

TABLE-US-00002 TABLE 2 Description of AAV5 capsid variants, A number of different mutations surrounding the  translational start of VP1 were generated to improve the stoichiometry of three VPs expressed  in insect cells. Nucleotides and amino residues changed as compared to the wild type serotype capsid sequence are indicated in bold. VP2 initiator SEQ context- Start Amino acid ID Bac. VD No. upstream codon additions(s) 5′ part of capsid sequence NO: AAV5 wild type — ATG — TCT TTT GTT GAT CAC CCT CCA GAT TGG T ... 39  S   F   V   D   H   P   P   D   W Changes surrounding the CP1 translation start 159 CCTGTTAAG ACG — TCT TTT GTT GAT CAC CCA CCC GAT TGG T ... 41  S   F   V   D   H   P   P   D   W 160 CCTGTTAAG ACG GCTA TCT TTT GTT GAT CAC CCA CCC GAT TGG T ... 42  S   F   V   D   H   P   P   D   W 161 CCTGTTAAG ACG — GCT TTT GTT GAT CAC CCA CCC GAT TGG T ... 43  A   F   V   D   H   P   P   D   W 162 CCTGTTAAG CTG — ACT TTT GTT GAT CAC CCA CCC GAT TGG T ... 44  T   F   V   D   H   P   P   D   W 163 CCTGTTAAG CTG ACTT AGC TTT GTT GAT CAC CCA CCC GAT TGG T ... 45  S   F   V   D   H   P   P   D   W 164 CCTGTTAAG CTG — AGT TTT GTT GAT CAC CCA CCC GAT TGG T ... 46  S   F   V   D   H   P   P   D   W 761 CCTGTTAAG ACG GCTA TCT TTT GTT GAT CAC CCA CCC GAT TGG T ... 47  S   F   V   D   H   P   P   D   W 762  — ACG GCTA TCT TTT GTT GAT CAC CCA CCC GAT TGG T ... 48  S   F   V   D   H   P   P   D   W 763 (wild type — ATG — TCT TTT GTT GAT CAC CCT CCA GAT TGG T ... 49 AAV5)  S   F   V   D   H   P   P   D   W 764 — TTG GCTA TCT TTT GTT GAT CAC CCA CCC GAT TGG T ... 50  S   F   V   D   H   P   P   D   W 765 — CTG GCTA TCT TTT GTT GAT CAC CCA CCC GAT TGG T ... 51  S   F   V   D   H   P   P   D   W 766 — GTG GCTA TCT TTT GTT GAT CAC CCA CCC GAT TGG T ... 52  S   F   V   D   H   P   P   D   W  43 CCTGTTAAG CTG GCTA TCT TTT GTT GAT CAC CCA CCC GAT TGG T ... 69  S   F   V   D   H   P   P   D   W Bac. VD No's 159-164 and 43 are operably linked to a polH promoter (SEQ ID NO: 53) Bac. VD No's 761-766 are operably linked to a short polH promoter (SEQ ID NO: 54)
3.2. Small Nucleotide Changes Surrounding the Translation Initiation Start of VP1 have Profound Effects on the Potency of the Vector

[0123] Baculovirus constructs harbouring all variants of cap5 expression cassettes listed in table 2 were successfully generated. Subsequently, these baculovirus constructs in combination with baculoviruses harbouring Rep(s) and transgene (reporter gene e.g. SEAP or FIX) were used for generation of rAAV. Some of the tested constructs irrespectively of multiple attempts did not support generation of rAAV production. This included wild type AAV5 (construct 763) and some of the constructs harbouring non-canonical starts, TTG (construct 764), GTG (construct 766). All the other constructs listed in table 2 resulted in successful generation of rAAV.

[0124] The three viral proteins (VPs) of successfully produced rAAV type 5 variants were isolated. The stoichiometry of the three VPs was investigated by electrophoretic separation (SDS-PAGE) of purified vectors (FIGS. 1 and 6). It appears that the small modifications introduced to the expression cassette of cap5 gene have a profound influence on the expression and/or assembly of the three VP proteins which is reflected in the composition of the capsids. We have noted that the adaptation of serotype 5 capsid to the insect cells by introducing non-canonical start codon (ACG) and the nine nucleotide upstream context CCTGTTAAG, which was reported by Urabe et al., as a modification allowing for insect cell production of serotype 2, resulted in low incorporation of VP1 (low VP1/VP2 ratio) and incorporation of excessive levels of VP3 into the capsid (high VP3/VP1 ratio) resulting in aberrant stoichiometry of the three VPs (FIG. 1, construct 159). Similarly, modification of nucleotide +4 to constitute G and to resemble closer canonical Kozak sequence, which resulted in exchange of serine at position +2 for alanine (construct 161), resulted in low incorporation of VP1 and high incorporation of VP3 (low VP1/VP2 high VP3/VP1). Use of different non-canonical codon CTG in combination with upstream CCTGTTAAG and downstream modification, i.e. change +4 nucleotide to A (construct 162), or +4-5 to AG (construct 164) or insertion of ACT as a second triplet with the modification of original +2 triplet to AGC (construct 163) did not improve VP1 incorporation to the capsid resulting in low VP1/VP2 high VP3/VP1. One of the constructs that showed a VP1/VP2 ratio close to 1 was construct 160 which encompasses direct upstream insertion of CCTGTTAAG, non-canonical ACG and insertion of an additional alanine in position +2 encoded by GCT as compared to the wild-type sequence, although the incorporation of VP3 was still in excess (equal VP1/VP2 high VP2/VP1). Subsequently, the promoter sequence in the construct 160 was mutated such that it resembles more precisely the wild type polyhedrin promoter. This generated the mutant 761. The VP2 initiation context was removed creating mutant 762. In both cases (761 and 762) there was a slight negative influence on the stoichiometry of the virus (lower VP1 incorporation) as compared to construct 160 (FIG. 1) Next, translation initiation start site of VP1 in construct 160 (to preserve the beneficial GCT directly downstream from the translation start codon) was altered to wild type ATG (mutant 763), TTG (mutant 764), CTG (mutant 765), GTG (mutant 766). All but the 765 mutant resulted in lack of detectable production of rAAV. Interestingly, combination of CTG as a non-canonical VP1 initiation start and addition of GCT triplet (encoding extra alanine) immediately following the translational start (765) resulted in higher incorporation of VP1 than VP2 and strong attenuation of VP3 ultimately resulting in balanced wild type AAV like VP stoichiometry (high VP1/VP2 moderate VP3/VP1) Finally, construct 43, which is like construct 160 with CTG as VP1 initiation codon instead of ACG, resulted in VP1 production with an almost native VP ratio (FIG. 6).

3.3. Superfluous Expression of P3 is Responsible for a Low Potency of True Type 5 AAV Mutants in BEVS.

[0125] In order to study the potency of the library of serotype 5 capsids, i.e., the ability of the vector to drive the expression of its genetic material, that have different VP stoichiometry in vitro and in vivo study where performed. Two different continuous cell lines were used i.e. Hela (FIG. 2 and FIG. 7A) and Huh7 (FIG. 3 and FIG. 7B). In both cases the set of mutants which showed incorporation of VP1 below that of VP2 and excessive incorporation of VP3 (constructs 159, 161-164) showed very reduced potency (FIG. 2-3). The potency of the vector was much improved by balancing VP1 and VP2 incorporation (construct 160). Shortening of the promoter (construct 761) and removal of the initiator constructs (construct 762) had a negative effect on the vector potency. The most potent vector, construct 765 (FIG. 2-3) showed VP1 to VP2 ratio in favour of the former and significantly decreased VP3 incorporation. Finally, the polH promoter (not shortened) in combination with the initiator construct, the CTG initiation codon and additional GCT triplet (encoding extra alanine) (construct 43) showed a good potency, albeit somewhat less than the potency of construct 765 (FIGS. 7A and B).

[0126] A subset of mutants (constructs 159-164) was tested in vivo (C57BL/6 mice) for potency. The vectors carried a reporter gene SEAP. Mice were injected with capsid 5 variants at a dose 5e12 gc/kg and monitored in time. In line with in vitro observation, variant that showed the best potency out of the tested set (160) also had VP1/VP2 in equimolar amounts (FIG. 4).

3.4. Insect Cell Produced Gemuine AAV5 (765) Performs Superior to the Chimeric Type 2/5 Mutant In Vivo

[0127] In order to investigate the potency of the AAV5 (765) in vivo three vector batches were prepared. These included the chimeric type 2/5 (92) (Urabe et al. (2006) supra), the genuine type AAV5 that contains excessive amounts of VP3 (160) and the best in vitro performing genuine type 5 AAV with wild type stoichiometry of VP's (765). All batches were produced under the same conditions using baculovirus constructs harbouring Rep proteins and FIX expression cassette (as described in WO 2006/36502). In order to compare the potency of the three vector preparations black 6 mice were injected with two different doses of the vectors, i.e. low dose 2e12 gc/kg and a high dose 2e13 gc/kg. In total seven groups including the vehicle group consisting of 5 animals each, were included in the experiment. Following the start of the experiment, blood was collected at week 1, 2 and 4. The expression of FIX was monitored in the blood by means of specific ELISA. The results corroborated the previous in vitro findings were newly generated 765 mutant displayed significant improved potency over 160 construct. Interestingly, 765 construct was also significantly better that the type 2/5 chimera (construct 92) published by Urabe et al. (2006) (supra) (FIG. 5). Unpaired t test was used to investigate the differences between 765 vs. 160 and 765 vs. 92. In all cases i.e. week 1, 2 and 4 there was a statistical significant difference with a p value <0.05.

4. Discussion

[0128] Generation of rAAV in insect cells requires a number of adjustments in the genetic organization of the cap gene. In mammalian cells AAV expresses its VP proteins from a single open reading frame by utilizing alternative splicing and the poorly utilized ACG initiator start for VP2. This results in a VP1:VP2:VP3 stoichiometry of 1:1:10. In insect cell these mechanisms failed to produce AAV vectors with a correct VP stoichiometry (Urabe et al. (2002) supra). This is a known problem which has previously been circumvented by Urabe et al., to generate rAAV2 serotype by changing the VP1 initiator triplet to ACG and by mutating the 9 nucleotides upstream from the translation initiation start site. These changes resulted in production of all three rAAV2 VP's in a correct stoichiometry. Similar genetic alteration in rAAV5 expression cassette resulted in low VP1 production and low potency of produced virus Building upon the success of the genetic adaptation to rAAV2, Urabe et al. decided to: make a series of six domain swap mutants where, rAAV5 received various length of N-terminal portion of VP1 from AAV2 (ranging from 7 amino acids up until 136 amino acids). This approach resulted in the production of a chimeric rAAV5 that showed a correct stoichiometry of VP's. Moreover, the domain swap mutants, resulted in a potency that was similar or superior to that of rAAV5 produced in 293T cells (Urabe et al. (2006) supra). Although, Urabe et al., demonstrated that chimeric rAAV5 can be generated in insect cells the obtained vector does not comprise bona fide AAV5 particles and as such may differ in various aspects such as susceptibility to pre-existing neutralizing antibodies, intracellular trafficking, bio-distribution and/or targeting from the true AAV5 serotype. At the same time the Urabe et al., reported that the attempts to produce infectious genuine rAAV5 failed due to low synthesis of VP1 polypeptide (Urabe et al. (2006) supra).

[0129] Here we have constructed a library of cap5 mutants objected at understanding the determinants underlying low potency of genuine rAAV5 produced in insect cells First, we have examined a mutant (159) that incorporated a number of adaptations which were previously used for successful generation of rAAV2 in insect cells (Urabe et al. (2002) supra). This mutant contains 9 nucleotide upstream VP2 initiator context placed upstream of VP1 translational start and non-canonical translation initiation start ACG. These 9 nucleotides were previously used by Urabe et al., to express serotype 2 gene in insect cells (Urabe et al. (2002) supra). This particular sequence naturally flanks non-canonical start codon (ACG) of VP2. Next, the wild type ATG was change to either ACG or CTG and in order to provide optimal downstream context from the start codon various mutations were introduced. Most of the mutants showed aberrant VPs stoichiometry with low incorporation of VP1 and excessive presence of VP3 (low VP1/VP2 and high VP3/VP1 ratio). The ratio VP1/VP2 was much improved in the genetic design 160, which still however showed excessive incorporation of VP3 into the vector particles. Finally, one of the genetic designs i.e. 765 showed high incorporation of VP1 (high VP1/VP2 ratio) and reduced incorporation of VP3 as compared to other tested variants (balanced VP3/VP2 ratio).

[0130] The low ratio of VP1/VP2 proteins has been postulated before to be responsible for the low vector potency (Hermonat et al. (1984) Journal of Virology 51(2):329-339; Tratschin et al (1984) Journal of Virology 51(3):611-619). Unique VP1 part of AAV is buried inside the capsid and becomes exposed during intracellular trafficking of virus to the nucleus. It first becomes exposed as a response to lowering pH in the lumen of endosome. Free N-terminal part of VP1 contains phospholipase domain which upon exposure to the outside of capsid becomes available to hydrolase specifically the 2-acyl ester (sn-2) bond of phospholipid substrates, resulting in release of lysophospholipids and free fatty acid allowing, in turn, endosomal escape of AAV. Unique portion of VP1 contains nucleus localization signals (clusters of basic amino acids) and was implicated in nucleus targeting of AAV. Finally, some authors suggest that unique portion of VP1 may play a role in virus uncoating in the nucleus. Low VP1/VP2 ratio and excessive incorporation of VP3 into viral particles (high VP3/VP1 ratio) may result in either 1) decreased incorporation of VP1 into the assembled particles on average or 2) generation of two particle populations A) correctly assembled particles (having close to wild type stoichiometry 1:1:10, i.e. 5 VP1 molecules per vector particle) B) VP3/VP2 only particles. In both situations (1 and 2) such vector preparation may have altered potency. The excessive amounts of VP3 proteins (as compared to VP1 or VP2) present in the vector preparation likely results in impaired trafficking of the vector to the nucleus due to disturbed endosomal escape. In order to test the hypothesis that the VP stoichiometry is detrimental for vector potency and to generate more potent vector the library of mutants of serotype 5 capsid was tested in vitro and in vivo.

[0131] It appeared that the VP's stoichiometry correlated well with the potency of the vector. As shown before (Hermonat et al. (1984) supra; Tratschin et al. (1984) supra; WO2007046703A2) low VP1N/VP2 ratio has strong influence on the potency of the virus. The mutants 159, 161-164 all have shown low VP1/VP2 ration and drastically reduced potency. Improved ratio between VP1/VP2 had significant impact on the potency of the vector (160). Interestingly further improvement in the VP1/VP2 ratio