Arch Virol (2008) 153:2233–2243 DOI 10.1007/s00705-008-0252-1

ORIGINAL ARTICLE

An improved self-deleting retroviral vector derived from avian leukemia and sarcoma virus Caroline Torne-Celer Æ Karen Moreau Æ Claudine Faure Æ Ge´rard Verdier Æ Corinne Ronfort

Received: 22 July 2008 / Accepted: 30 October 2008 / Published online: 19 November 2008 Ó Springer-Verlag 2008

Abstract We have previously developed a self-deleting avian leukosis and sarcoma virus (ALSV)- based retroviral vector carrying an additional attachment (att) sequence. Resulting proviruses underwent deletion of viral sequences and were flanked either by two LTRs (LTRs proviruses) or by the additional att sequence and the 30 LTR (att proviruses). Herein, we have tried to increase (1) the selfdeleting properties of this vector, either by raising the selection pressure applied on target cells or by optimizing the size of the internal att sequence, (2) the titer of the vector by deleting or inverting some viral sequences. Moreover, a new type of provirus flanked by att sequences at each end was isolated. Finally, under specific conditions, 100% of proviruses had internal sequences deleted, and as many as 92–100% of proviruses were no longer

C. Torne-Celer
K. Moreau
C. Faure
G. Verdier
C. Ronfort (&) Universite´ de Lyon, Lyon, France e-mail: [email protected] K. Moreau
G. Verdier
C. Ronfort INRA, UMR754 Re´trovirus et Pathologie Compare´e, IFR128 Biosciences Gerland-Lyon Sud, Lyon, France C. Torne-Celer
K. Moreau
C. Faure
G. Verdier
C. Ronfort Universite´ Lyon 1, 69007 Lyon, France K. Moreau
G. Verdier
C. Ronfort Ecole Nationale Ve´te´rinaire de Lyon, 69000 Marcy L’e´toile, France K. Moreau
G. Verdier
C. Ronfort Ecole Pratique des Hautes Etudes, Lyon, France C. Torne-Celer
C. Faure CNRS, Centre de Ge´ne´tique Mole´culaire et Cellulaire, 69622 Villeurbanne, France

mobilizable by a replication-competent virus. The inactivation procedure achieved here might improve the biosafety of retroviral vectors.

Introduction Retroviral vectors are powerful tools for transferring new genetic information into the genome of animal cells. Retroviral vectors offer several advantages, including the efficient delivery and stable integration of a foreign gene into the host genome. These vectors have been widely used for gene transfer into animals [1–3] and are popular vehicles for delivering genetic material to humans for gene therapy [4–6]. The retroviral vectors currently used are replicationdefective and lack the virus coding regions; therefore, the viral proteins are supplied in trans, either in packaging cell lines engineered to generate infectious virus or by transient vector production [7–11]. Popular retroviral vectors are based on murine leukemia virus (MuLV) and on lentiviruses such as human immunodeficiency virus (HIV), which infect and replicate in terminally differentiated and nondividing cells. Vectors derived from avian leukemia and sarcoma virus (ALSV) and spleen necrosis virus (SNV) have also been widely developed (reviewed in [12, 13]). The use of retroviral vectors, especially those derived from lentiviruses, raises concerns about their safety, considering the pathogenicity of the parental virus. One goal is to prevent the emergence of replication-competent recombinants (RCRs). Continuous improvements in packaging cell lines for producing virus stocks have minimized the risk of generating RCRs [7]. Moreover, once integrated into the target cell, a replication-defective vector

123

2234

C. Torne-Celer et al.

theoretically cannot replicate because it lacks the genetic information for viral proteins. However, infectious viruses can be generated if the missing proteins are supplied in trans, which can occur either in the event of superinfection of these cells by exogenous retroviruses or by transcomplementation with endogenous retrovirus proteins, as has been reported recently [14]. To overcome these problems, several self-inactivating or self-deleting vectors have been developed: 1.

2.

3.

Self-inactivating (SIN) vectors are the most widely used vectors. They are obtained by deleting enhancerpromoter sequences from the U3 region of the LTRs. The transcriptional inactivation of the LTR in the SIN provirus and the expression of the transgene from an internal promoter prevent mobilization by a replication-competent virus and minimize any risk of vector spread [7, 15]. However, HIV-derived SIN vector may still generate a small number of full-length transcripts in transduced cells, since a cryptic transcriptional activator resides in the HIV leader region [16]. In Cre-loxP vectors, the strategy employs the duplication of LTR and the ability of the P1 phage sitespecific recombinase (Cre) to excise any sequences positioned between two loxP target sequences. LTRmediated duplication places vector sequences, including Cre, between loxP sequences in the integrated proviruses. This enables Cre to excise from the provirus most of the unrelated viral and non-viral sequences, allowing the expression of the transgene from an internal promoter positioned in the remaining U3 sequence [17, 18]. However, rearrangements in the retroviral vectors might sometimes reduce the efficiency of excision [17, 19]. In E- and psi-vectors [20–22], the packaging sequence and selectable gene are flanked by two direct repeats, leading to their deletion during the reverse transcription step via a template misalignment mechanism. Such vectors delete either a GFP gene in approximately 22–29% of proviruses formed or both the selectable marker and psi sequence at an efficiency greater than 99% in a single replication cycle. However, the resulting provirus still maintains two functional LTRs, and there is a low but detectable frequency (5–15%) of deletions and rearrangements during transfection or propagation of producer cells [20].

Finally, vectors with artificial primer binding sites that require cotransfection of a recombinant tRNA for complementation in the producer cell further decrease the potential for mobilization because the artificial sequence is missing in the target cell [23, 24].

123

Retroviral integration is carried out by the viral enzyme IN and involves the extremities of the LTRs which, when joined together, form the attachment (att) site [25]. We have recently reported the effects on the replication life cycle of introducing an additional att sequence internally into a vector backbone [26]. The vector was shown to lose sequences during the replication cycle, giving rise to two main kinds of proviruses: (1) proviruses flanked by LTRs (LTR proviruses), from most of which internal sequences were deleted, probably during the reverse transcription step, (2) proviruses flanked, at the 50 side by the additional att sequence and at the 30 side by an LTR (att proviruses) from which the whole 50 part of the genome (including the 50 LTR) was deleted. Among the clones analyzed in this previous study, 61% were flanked by LTRs and had internal sequences deleted, and 31% were att proviruses. In total, 77% of the proviruses had lost psi and were not mobilizable by a replication-competent virus. Furthermore, 92% of the proviruses had lost the selectable gene, which was no longer necessary in the target cell. The att vector could therefore be considered an efficient self-deleting vector. However, such a vector still had several features that needed to be improved. In the present paper, the retroviral vector was subjected to the analysis of certain parameters to improve both the percentage of deleted proviruses and the titer of the vector.

Materials and methods Reagents Primers were purchased from Sigma-Proligo (Saint Quentin Fallavier, France). Antibiotics (G418 and puromycin) were purchased from Cayla (Toulouse, France). Taq enzyme (GoTaq DNA polymerase) and T4 DNA ligase were from Promega (Charbonnie`res-les-Bains, France). Cell lines The QT6 cell line is derived from quail cells [27]. The Isolde cell line [28] is an ALSV-based packaging cell line derived from QT6 cells. These quail cell lines do not harbor endogenous retroviruses [29]. Both were maintained in F10 medium as described previously [28]. Vectors NP3Catt (Fig. 1a) and NP3C (Fig. 3b) vectors have been described previously [26]. Briefly, they are avian erythroblastosis virus (AEV)-based vectors containing the bacterial neo and puro selectable marker genes. In both vectors, the neo gene is driven by the AEV LTR, while the

A self-deleting retroviral vector

2235 > 2.0 kb

H

A P neo probe

P H

G

LTR

pro

puro

p A

N129

R

U5 U3

N859

N130

invR

O4 O6O7

pro

puro

-

-

-

pro

puro

-

+

-

pro

puro

-

+

-

pro

puro

+

+

-

puro

-

-

+

B Clone structure

Classes

PCR amplifications invR/O6

invR

B

P

SA

SD R

puro probe

E

J

neo

NP3Catt

P

att

R/O4 or R/O7

P

N129/O4

B

LTR

1.9 kb

U3 I

II

att proviruses LTR proviruses

att

att +

III

fulllength proviruses

IV

rearranged proviruses

att J

neo

?

U3

P

pro LTR

C

U5

U3

TTGGTGTGCACCTGGGTAGATGGACAGACCGTTGAGTCCCTAACGATTGCGAACACCTGAATGAAGCGGAAGGCTTCATTGATGTAGTCTTATACAATAGCCTTACACAATAATGCTATGTAACGATGAAACAGCAGTATTCCATATAAGGAAAGTA

att 122

att 157

att (att 91) att 40

att 75

Fig. 1 Structure of the NP3Catt vector, expected structure of puroresistant NP3Catt proviruses and description of att sequences of various sizes. a Map of the retroviral construct NP3Catt. The large boxes represent LTRs of AEV origin, marker genes (neo and puro selectable genes), SV40 sequences (promoter (pro) and polyadenylation signal (pA)). The small boxes represent part of the gag and env genes (dG and dE), and the lines represent viral non-coding sequences (leader, J region from AEV and 30 non-coding region, W: psi packaging signal). SD and SA splice donor and acceptor sites. The internal att sequence (depicted by a hashed box) is constituted by 45 bp of the U5 30 end and 46 bp of the U3 50 end (the sequence is shown in Fig. 1c). Single-headed arrows represent the direction and approximate location of the primers used for PCR amplifications in Figs. 1b and 2. The positions of HindIII (H), BstXI (B) and PstI (P) restriction sites and the puro and neo probes (black lines) are also indicated, as well as the size of fragments obtained after digestion with either HindIII or both HindIII and BstXI and hybridization with the puro probe. b Structure of puroR NP3Catt proviruses expected

from NP3Catt integration. According to previous results [26], puroR proviruses can be classified into 4 classes (att proviruses, LTR proviruses, full-length proviruses (identical to the parental vector), and rearranged proviruses) after PCR amplification with N129/O4, R/ O4 or R/O7 and invR/O6 pairs of primers. ? PCR amplification, - no PCR amplification. For proviruses classified in class I (which are negative for the three PCR tests), the structure is further confirmed by amplifying the junction fragments (by inverse PCR) and sequencing of the resulting products (as described in [26]). In class II, the presence/absence of the psi sequence is monitored by sequencing the amplified products (R/O4 or R/O7 products) or by a PCR amplification with primers W/O7 or W/N859. DU3 indicates a deletion of the U3 sequence found in the internal att sequence of some proviruses. c Sequences of the att sites used in this study are indicated. U5 sequences are in bold while U3 sequences are not. The canonical CA and TG used by IN for integration are underlined. Sequences are from Ref. [42]

puro gene is driven by an internal simian virus 40 (SV40) promoter. NP3Catt also contains a 91-bp att sequence [46 bp from the U5 right end and 45 bp from the U3 left end (Fig. 1c)] inserted between the neo and puro transcription units, in normal orientation relative to the LTR

sequences. NP3Catt derivatives (NP3Catt40, NP3Catt75, NP3Catt122 and NP3Catt157) (Table 2) contain an att sequence of 40 bp (25 bp from U5 plus 15 bp from U3), 75 bp (40 ? 35), 122 bp (62 ? 60) and 157 bp (80 ? 77) (Fig. 1c), respectively. In NP3Catt(DpA) and NP3C(DpA)

123

2236

C. Torne-Celer et al.

was checked using an immunocytochemistry test as previously described [30]. Southern blots and PCR amplifications were performed as described previously [26]. Primers W (CTTAGGAGG GCAGAAGCTGA), N859 (GCCGATTGTCTGTTGTG CCC), invR (GTGGTGGAAGGTAAAATGGC) and pbs (TAACGATCACGTCGGGGTC) hybridize the psi packaging sequence, the neo gene, the LTRs and the pbs, respectively. Primers are represented in Fig. 1a.

vectors (Fig. 3a, b), the SV40 polyA signal was removed. In NP(-)3Catt(DpA) and NP(-)3C(DpA) vectors, the puro transcription unit has been inserted in reverse orientation and the SV40 polyA sequence was also removed. In NP3Catt(inv) (Fig 3a), att sequences have been inserted in reverse orientation. Virus assays Plasmid DNAs were transfected into the Isolde packaging cell line as described previously [26]; transfected cells were selected with G418 (200 lg/ml). Viruses were harvested from a pool of more than 100 G418-resistant clones and used to infect fresh QT6 cells. Titers of infectious particles transducing neo and puro resistance were measured following infection of QT6 cells with these supernatants and selection with either G418 (200 lg/ml) or puromycin drugs (0.2, 0.5 or 2 lg/ml). Puro-resistant (puroR) clones att1 to att13 [26], A1 to A15, and B1 to B13 (this paper) were isolated following selection with 0.2, 0.5 and 2 lg/ml of puromycin, respectively. Release of RCRs

A

O1

M1 M2

v /in

77

- +

2 N1

9/N

13

Inverse PCR One lg of genomic DNA was digested with PstI (Promega, France) and ligated; the left and right ends of proviruses were then amplified by PCR and sequenced as described previously [26]. The right end of provirus NP3Catt157 number 6 (provirus 157-6) (Fig. 2) was amplified with oligonucleotides N2 (GGGGGATCGATCCTCTAGC) and N3 (CTCCACAGTTGAATGCACAG), located in the J region of AEV downstream of the PstI site. Thirty-five cycles of PCR were performed (94°C for 45 s, 50°C for

0 /p K2

- +

- +

bs

/N K2

13

0

- +

1.3 kb

1.4 kb

0.4 kb

0.42 kb

B 1.4 kb 1.3 kb

420 bp

400 bp N1

att

inv77 K2

O1 O7

pro

puro

R

pbs

N129

N130

G

p E A

neo LTR

att

N2

J

host DNA

N3 N4

...GCATATTCTAGATAATAGAACAAATGTtgtagtcttatacaatagcc...

Fig. 2 Structure of the puroR NP3Catt157 clone 6 classified in class V. a Specific PCR amplification of provirus with oligonucleotides O1-inv77, N129-N130, K2-pbs and K2-N130. A 1/10 fraction of PCR reactions were loaded onto the gel. The sizes of the amplification products are indicated on the right side of the figure. M1 lambdaHindIII marker, M2 100-bp ladder marker. - Negative control (QT6 cell DNA). b Structure of the NP3Catt157-6 provirus as deduced from PCR experiments in a and inverse PCR analysis. The legend is the

123

...tgaatgcggaaggcttcaCAAATGTCAAACCAGTACAAGATAT...

same as in Fig. 1. Single-headed arrows represent the direction and approximate location of the primers used for PCR amplification in a and their resulting amplified sizes. Large arrows indicate regions that have been amplified by inverse PCR (as described in Ref. [26]) and sequenced. The resulting sequence of the viral-cellular DNA junctions is given below (host DNA is in capital letters, att sequences are in small letters). The duplication of host DNA sequences is underlined

A self-deleting retroviral vector

2237

B

A LTR

att

G

neo

NP3Catt

pro

neo

LTR

puro

neo

LTR

LTR

G

NP(-)3C( pA) SD

pro

neo SD

E

pro

att

G

G

NP3C( pA)

SA

SD

LTR

LTR

LTR

puro

p A

E

pro

puro

pro

puro

p A

E

LTR

SA

E

LTR

E

LTR

SA

SD

neo

NP3Catt(inv)

E

puro

att

G

G

neo

SA

SD

NP(-)3Catt( pA)

LTR NP3C

SD

att

G

NP3Catt( pA)

p A

puro

LTR

SA

SD

LTR

pro

E

neo

puro

pro

SA

LTR

SA

Fig. 3 Retroviral constructs used in this study. All vectors are derivatives of NP3Catt (a) and NP3C (b) constructs. Vectors depicted in a carry an internal att sequence, while vectors in b do not. DpA indicates the deletion of the SV40 polyA signal. The puro transcription unit has been inserted in the reverse orientation in NP(-)3Catt

(DpA) and NP(-)3C(DpA) vectors, which is indicated by an arrow. Att sequence has been inserted in reverse orientation in the NP3Catt(inv) vector. All other abbreviations are the same as in Fig. 1

45 s, and 72°C for 3 min), followed by an extension at 72°C for 10 min. A 1/100 fraction of this reaction was used for a second (nested) PCR, using oligonucleotides N1 (ACCAAAGCGGCCATCGTGCC) and N4 (TCCTGAA GGTGTAAGGAATG), located within the first set of primers. Conditions of the nested PCR were the same as above. A 1/5 fraction of the nested PCR was electrophoresed on a 2% agarose gel, and the remaining PCR products were used directly for DNA sequencing (GATC-Biotech, Marseille, France).

NP3C vector [26]. These results suggest that insertion of the att site has a slight negative effect on the transduction of the neo gene and a positive effect on the transduction of the puro gene. Proviruses were isolated following infection of QT6 cells with helper-free stocks of NP3Catt vector, selection with puromycin and picking of puro-resistant (puroR) clones. From data reported previously [26], the structure of NP3Catt-resulting proviruses can be determined according to the following protocol: First, the overall structure of the provirus is checked by Southern blot: (1) the 30 viral-host DNA junction fragment is detected using the HindIII enzyme and hybridization with a puro probe; this analysis should reveal one band per integration, the size of which should be greater than 2 kb (Fig. 1a); (2) to ensure that the integrated proviruses did not rearrange, the structure in the 30 region of proviruses is monitored using HindIII and BstXI enzymes and hybridization with a puro probe (Fig. 1a), which should reveal a fragment of 1.9 kb in size; (3) blots are then hybridized with a neo probe to check for the presence of the neo gene. Secondly, the structure of puroR proviruses is more accurately determined by PCR: (1) the presence of a full-length provirus (carrying neo) is monitored by using the N129/O4 pair of oligonucleotides (hybridizing the neo gene and the SV40 promoter, respectively) (Fig. 1a); (2) the presence of the 50 LTR is monitored by using the R/O7 and R/O4 pairs of oligonucleotides (R is specific for the LTR while O4 and O7 hybridize the SV40 promoter, Fig. 1a); (3) some clones displaying a rearranged structure are detected with invR (which is specific for the R region of the LTR) and O4 or O6 oligonucleotides. Finally, puroR proviruses can be divided into four classes according to these PCR analyses (Fig. 1b) [26]:

Results and discussion Structure of the NP3Catt vector and protocol for determining the structures of proviruses The NP3Catt and NP3C avian erythroblastosis virus (AEV)-based vectors carry two selectable genes, neo and puro, whose expression is driven by the viral LTR and an internal SV40 promoter, respectively. The NP3Catt vector contains a 91-bp-long internal att sequence inserted between the two selectable genes (Figs. 1a, 3), while NP3C (see Fig. 3b) lacks this sequence [26]. The vectors NP3C and NP3Catt were produced at titers of 1.1 and 0.3 9 104 neo-resistance focus-forming units (RFFU) per ml, respectively. The NP3C and NP3Catt puro titers were 2.1 and 4.8 9 103 puro-RFFU/ml, respectively [26]. It is interesting to note that the neo titers of the NP3Catt vector were repeatedly lower (2.5–5 times) than the corresponding titers obtained with the NP3C vector. Conversely, the puro titers of the NP3Catt vector were repeatedly higher (1.2–5 times) than the corresponding titers obtained with the

123

2238

C. Torne-Celer et al.

Table 1 Characterization of puroR proviruses isolated at 3 puromycin concentrations Puromycin concentration (lg/ml)

Total number of puroR clones isolated

Number of Number of att proviruses neoR (class I)e clonesd

Number of LTR proviruses (class II)f

Number of fulllength proviruses (class III)g

Number of rearranged proviruses (class IV)h

Number of DW clonesi

0.2

13a

1

4

7

1

1

10

0.5

b

14

0

8

5

0

1

14

2

12c

0

7

3

0

2

11

a

These clones have been described in detail in Ref. [26]

b

Clones of series A. Fifteen clones were isolated. One of them was not further studied because it carried two proviruses

c

Clones of series B. Thirteen clones were isolated. One of them was not further studied because it carried two proviruses

d

PuroR clones were tested individually for resistance to G418. Clones resistant to G418 selection were classified as neoR. Presence/absence of the neo gene was further checked by PCR with primers N129-N130

e

PuroR clones yielding no amplification product with N129/O4, R/O4 or R/O7 as well as with invR/O6 pairs of primers were classified as att proviruses according to the classification in Fig. 1b. The extremities were further checked by sequencing host-virus junctions after inverse PCR

f

NeoS PuroR clones yielding an amplification product with R/O4 or R/O7 primers were classified as LTR proviruses

g

PuroR proviruses resulting from integration of the full-length vector (parental vector)

h

PuroR clones displaying a rearranged structure were detected by the invR-O4 or invR-O6 pairs of primers and assumed to result from autointegration

i DW indicates the deletion of the packaging sequence. The presence of the W sequence was determined in LTR proviruses, either by sequencing the R/O4 or R/O7 PCR amplifications (clones isolated at 0.2 lg/ml of puromycin) or by PCR amplification with primers W/O7 or W/N859 (clones selected with 0.5 and 2 lg/ml of puromycin)

1.

2.

3.

4.

Clones flanked by the additional att sequence in 50 and the LTR in 30 (class I) which have the entire 50 region deleted, are neither hybridized by the neo probe nor amplified by the three oligonucleotides pairs used for screening (Fig. 1b). For these clones, junctions between viral and host genomic DNA is checked by sequencing the PCR products amplified by an inverse PCR approach [26]. Clones flanked by the LTRs (class II) are detected by the R-O4 or R-O7 pairs of primers. Among them, the presence of the W sequence is determined either by sequencing the resulting product or by a W-N859 PCR amplification [W and N859 are specific for the psi and neo gene, respectively (Fig. 1a)]. Clones displaying the structure of the full-length vector (class III) are hybridized by the neo probe and further detected by the N129-O4 and R-O4 primers. Clones displaying a rearranged structure (class IV) are detected using the invR-O6 or invR-O4 pairs of primers.

In the first set of experiments [26], thirteen clones were studied and were classified as follows: four clones belonged to class I, seven to class II (among which six had the psi sequence deleted), one to class III and one to class IV (data shown in Table 1, lane 1, clones obtained after selection with 0.2 lg/ml puromycin [26]). Proviruses of class II were assumed to be the result of the integration of a small-size cDNA formed during reverse transcription [26]. By contrast, proviruses of class I were assumed to be the

123

result of the integration of a small-size cDNA formed either during reverse transcription or at the time of integration [26]. In the present report, we tried to improve both the percentage of self-deleted proviruses and the titer of the vector. Attempts to increase the proportion of att proviruses by increasing the selection pressure Our first aim was to increase the proportion of NP3Catt proviruses flanked by the att sequence instead of the 50 LTR (i.e., class I versus class II proviruses). Several LTR clones, having lost the neo gene (class II, Fig. 1b), should carry the puro gene in an unfavorable context for expression. Indeed, vectors carrying internal promoters frequently display promoter interaction leading to the reduced expression of one or the other promoter [31–33]. Therefore, we have tested the possibility of eliminating these 50 LTR proviruses by increasing selection pressure on the puro gene. NP3Catt helper-free stocks were used to infect fresh QT6 cells, which were then selected with 0.2 lg/ml of puromycin (the same concentration as in previous work [26]), or with 0.5 or 2 lg/ml of the drug puromycin. The NP3Catt vector gave titers of 4.5 9 103 (average of five experiments of transfection, infection and selection), 5 9 103 (average of eight experiments) and 4 9 103 (average of two experiments) puro-RFFU/ml following selection with 0.2, 0.5 and 2 lg/ml of puromycin,

A self-deleting retroviral vector

2239

Table 2 Titers of NP3C and NP3Catt vectors carrying internal att sequences of various lengths and number of puroR proviruses generated by these vectors ranged in the different classes Vector

Titersa

Structuresb

Neo titers

Puro titer

Puro/neo ? puro

Class I

Class II

Class III

Class IV

Class V

7.6 9 103

2.0 9 103

0.21

–

–

–

–

–

NP3Catt40

7.0 9 10

3

3.9 9 103

0.36

4

9

3

0

0

16

NP3Catt75 NP3Cattc

3.6 9 103 2.8 9 103

3.6 9 103 3.2 9 103

0.50 0.53

ND 8d

ND 5d

ND 0d

ND 1d

ND 0d

14

NP3Catt122

1.9 9 103

3.6 9 103

0.65

ND

ND

ND

ND

ND

NP3Catt157

2.6 9 103

3.9 9 103

0.60

2

7

0

0

1e

NP3C

Total number of clones

10

a

Each vector was transfected into the packaging cell line. The supernatants were used to infect fresh QT6 cells. Infected cells were then selected with G418 or puromycin (0.5 lg/ml), and resistant colonies were scored 10–15 days later. Neo and puro titers are expressed in neo- or puroRFFU/ml (resistance focus-forming units per ml). The results given in this table are an average of three independent experiments of transfection, infection and titration

b

The structure was determined according to the protocol described in Fig. 1b and in the text (including PCR and inverse PCR approaches)

c

NP3Catt vector carries a 91-bp-long att sequence (Fig. 1a, c) [26]

d

Proviruses of series A (this article, Table 1)

e

One clone with a distinctive structure, which is depicted in Fig. 2

respectively. Therefore, increasing selection pressure on the puro gene did not affect transduction of the puro gene, as revealed by the puro titer. Fifteen (called A1–A15) and 13 (called B1–B13) puroR clones were isolated after infection and selection with 0.5 and 2 lg/ml of puromycin, respectively. Resistance to the drug neomycin was tested for each clone by applying 200 lg/ml of the drug G418. All clones were found to be neo-sensitive (neoS) (Table 1). Each of these clones was then subjected to molecular analysis (by Southern blot and PCR) using the protocol described above. The results of the PCR experiments are summarized in Table 1, and proviruses were distributed into the four classes described in Fig. 1b). (1) Four of the 13 proviruses (31%) obtained with 0.2 lg/ml puromycin (clones from Ref. [26], Table 1, lane 1), 8 of the 14 clones (57%) obtained with 0.5 lg/ml puromycin (series A, lane 2) and 7 out of the 12 clones (58%) obtained with 2 lg/ml puromycin (series B, lane 3) displayed an att in the 50 junction and an LTR in the 30 junction (proviruses classified in class I) (Table 1). (2) Conversely, the proportion of LTR proviruses (class II) decreased from 54% (7/13) at 0.2 lg/ml puromycin to 36% (5/14) and 25% (3/12) at 0.5 and 2 lg/ml puromycin, respectively. (3) Proviruses carrying the full-length vector (class III) were only found at a puromycin concentration of 0.2 lg/ml, but not beyond a concentration of 0.5 lg/ml. It is possible that the configuration of these proviruses (full-length vector) do not allow a high level of expression of the puro gene. (4) Class IV proviruses (rearranged) were not eliminated by increasing the selection pressure (7% (1/14) and 17% (2/12) at 0.5 and 2 lg/ml of puromycin, respectively). According to the Fisher’s exact test, the distribution in the four

classes according the puromycin drug concentration was significantly different (P value = 0.001). Therefore, the distribution of proviruses in the four classes is dependent on the puromycin drug concentration applied to infected cells. With respect to the selectable gene (neo), it is noteworthy that 92% of proviruses at 0.2 and 2 lg/ml puromycin (11/12 and 12/13, respectively) and 100% of proviruses (14/14) at 0.5 lg/ml puromycin had the entire neo sequence deleted. With respect to the psi sequence, 100% (14/14) of proviruses at 0.5 lg/ml and 92% (11/12) of proviruses at 2 lg/ml had this sequence deleted and were not rescuable, whereas this number was 77% (10/13 clones) following a selection with 0.2 lg/ml of puromycin in the first set of experiments [26]. Although these data are not significantly different (in the Fisher’s exact test), the trend suggests that selecting target cells beyond 0.5 lg/ml of puromycin might be a means to increase the proportion of self-deleted proviruses. This concentration was used thereafter. Finally, these data confirm that the NP3Catt vector could be considered an efficient vector, since as many as 92–100% of the proviruses were not rescuable, and the same proportion of the proviruses had the neoselectable gene deleted, which is no longer desirable once the vector is integrated into the target cells. Attempts to increase the proportion of att proviruses by varying the size of the additional att sequence and identification of a less-represented new class of proviruses To determine the effects of the size of the internal att sequence on the proportion of provirus classes, four other

123

2240

NP3Catt-derivative vectors, NP3Catt40, NP3Catt75, NP3Catt122 and NP3Catt157, were constructed carrying internal att sequences of 40, 75, 122 and 157 bp, respectively (Fig. 1c). The results of three infections with viruses collected from three different pools of producer cells are reported in Table 2. It is noteworthy that: (1) according to the neo titers of the vectors, each vector replicated normally when passaged in recipient cells (approximately 1.9 9 103 to 7.0 9 103 neo-RFFU/ml); (2) the puro titer was almost the same regardless of the att size (3.2–3.9 9 103 puro-RFFU/ml). Moreover, puro titers of the series of NP3Catt vectors were always higher than the corresponding titer of the NP3C vector lacking the internal att sequence; this observation has been noted previously in repetitive experiments [26]. According to the Fisher’s exact test, the distribution of neo and puro titers varies according to the plasmid used, i.e., to the att sequence size (P value \ 2.2 9 10-6). Therefore, the puro/(puro ? neo) ratio (which may be a measure of the transduction efficiency of the puro gene in relation to the neo transduction efficiency) was calculated. The puro/ (puro ? neo) ratio increased with the size of the att sequence. The best efficiency of puro transduction was found with the 122- and 157-bp-long att sequences. Next, 16 and 10 clones obtained following infection with NP3Catt40 and NP3Catt157, respectively, and selection with 0.5 lg/ml of puromycin were picked. Proviruses were analyzed according to the protocol described above. Of the 16 NP3Catt40 puroR clones, four (i.e., 25%) carried a class I provirus, nine (56%) a class II provirus and three (19%) a class III provirus. Of the ten NP3Catt157 puroR clones, two (20%) were put in class I and seven (70%) in class II (Table 2). The last NP3Catt157 clone (clone number 157-6) displayed a distinctive structure and could not be assigned to one of the provirus classes. This clone was analyzed by further PCR amplifications (Fig. 2a). We made the following observations: (i) PCR amplifications, either with O1 and inv77 (surrounding the puro gene) or with the N129–N130 pair of oligonucleotides (amplifying the neo gene), indicated the presence of both the neo and puro genes in the DNA of this clone (Fig. 2a). (2) PCR with either R-O7 or N129-O4 pairs of primers carried out on the DNA of this clone were all negative (data not shown), suggesting that the LTR and neo gene were not linked to the SV40 promoter. We hypothesized that the neo gene would be linked on the right-hand side of the puro gene. (3) This assumption was confirmed by PCR with K2 (hybridizing the 30 non-coding region) and either the pbs oligonucleotide (hybridizing the leader region) or the N130 oligonucleotide (Fig. 2a). Finally, the left and right junctions as well as sequences surrounding the LTR were characterized by inverse PCR and sequencing of the resulting product. These analyses demonstrated that the

123

C. Torne-Celer et al.

provirus was flanked at both extremities by the att sequence (U3 on the left side, U5 on the right side) and that integration resulted in a small-size duplication (7 bp long) of host DNA (Fig. 2b). This provirus resulted from an integration process, since (1) cleavage involves IN recognition att sequences and (2) provirus is flanked by a short duplication of target DNA, which is a hallmark of integration [25]. Until now, att–att integration, as displayed by clone 157-6, has not been observed since the additional att sequence was always found at one provirus extremity and an LTR was found at the other end. It is possible that only such a long att sequence of 157 bp is able to generate such proviruses and that internal att sequence of smaller size are not. This clone represents the first demonstration that an internal ALSV att sequence can be recognized and cleaved by IN during the integration process, although at a low frequency. A 49-bp-long SNV additional att sequence was shown to be a substrate for integration, generating several proviruses like the above-described att–att provirus [34, 35]. This property to integrate at low frequency by way of an internal att sequence seems to be restricted to avian viruses, since such MuLV vectors were not found to work [36, 37]. Finally, the data showing the proportion of each class of proviruses generated by these vectors are summarized in Table 2. For the NP3Catt vector, we have taken into account the previously obtained ratio (series A obtained at 0.5 lg/ml of puromycin) (Table 1). Each vector generated class I (att) and class II (LTR) proviruses. Only the NP3Catt40 vector formed proviruses of class III. Moreover, only the NP3Catt157 generated the class V provirus (att–att). Class IV proviruses (rearranged) were found only with the NP3Catt vector. According to the Fisher’s exact test, these data are not significantly different (P value [ 0.05). Nevertheless, we can observe some trends in the distribution of clones. Indeed, the highest proportion of class I att proviruses (8/14 = 57%) was found with the NP3Catt vector. The vector NP3Catt157 generated more class II than class I proviruses. It is possible that the 157-bp-long att sequence is more efficient than the 91-bp-long att sequence at facilitating strand transfer during reverse transcription, bringing together the LTR and the internal att sequence (which is how class II proviruses may be generated; see model of Fig. 7A in Ref. [26]). This increase in recombination during reverse transcription would lead to an increase in the proportion of class II proviruses. The vector NP3Catt40 generated more class II than class I proviruses and was the only one to generate class III proviruses. It is possible that the 40-bplong att sequence is a less efficient substrate altogether for the abovementioned strand transfer during reverse transcription and/or for the integration process (which is how class I proviruses may be generated (see models in Fig. 7B

A self-deleting retroviral vector

and C in Ref. [26]). Hence, proviruses of class I would be produced at low frequency. Altogether, these data indicate that, at a selection pressure of 0.5 lg/ml of puromycin, (1) increasing the size of the internal att sequence beyond 40 bp seems to disfavor generation of proviruses of class III (full-length), and (2) increasing the size of the internal att sequence to 157 bp does not favor production of class I proviruses over class II proviruses. Finally, the optimal att size would be between 91 and 122 bp. Attempts to increase the NP3Catt vector titer According to the previous experiments, the vectors NP3C and NP3Catt are produced at titers of 7.6–11 9 103 and 2.8–3 9 103 neo-RFFU/ml, respectively (Table 2 and previous work [26]). The viral titers achieved with these vectors are still not high enough for many gene transfer applications. This difficulty could easily be circumvented, first by using a higher-producer cell line than the one used here, which usually produces vector at a titer of 105 neo-RFFU/ ml [28]. Titers could also be increased by pseudotyping vectors with the vesicular stomatitis virus protein G, since titers as high as 109 might be obtained after concentration of the viral particles [11]. Second, it is noteworthy that a similar vector lacking the puro transcription unit (N3C) is produced at a titer of 105 neo-RFFU/ml (data not shown). The reduction in titers of vectors carrying the puro transcription unit in relation to the abovementioned N3C might be due to the insertion of a polyadenylation sequence along with the puro gene. Indeed, the RNAs transcribed from the 50 LTR to this heterologous polyA signal should not be reverse transcribed because they lack the 30 LTR required to complete the reverse transcription process. Only the full-length RNAs transcribed from the 50 to the 30 LTRs may be reverse transcribed efficiently and give rise to infectious particles. Alternatively, the reduction of titer might also be due to transcriptional interference between the LTR and the internal SV40 promoters, as described previously by others [31, 32, 38]. To test if the deletion of the heterologous polyA signal could increase NP3C and NP3Catt viral titers, we generated mutants in which the SV40 polyadenylation region was removed. The resulting NP3C(DpA) and NP3Catt(DpA) vectors (Fig. 3a, b) gave an approximately seven- to tenfold increase in neo and puro titers in comparison with the NP3C and NP3Catt vectors, respectively (Table 3). Therefore, the deletion of the SV40 polyadenylation signal significantly increased virus titers of both the self-deleting and the control vectors.

2241 Table 3 Titers of vector NP3C and NP3Catt derivatives Vector

Neo titer

Puro titer

NP3Catt

3 9 103

4 9 103

9 9 10

3

2 9 103

NP3Catt(DpA)

2 9 10

4

4 9 104

NP3C(DpA)

8 9 104

2 9 104

2 9 10

4

2 9 104

5 9 10

4

2 9 104

1 9 10

3

6 9 102

NP3C

NP(-)3Catt(DpA) NP(-)3C(DpA) NP3Catt(inv)

Each vector was transfected into the packaging cell line Isolde. The supernatants were used to infect fresh QT6 cells. Infected cells were then selected with G418 or puromycin (0.5 lg/ml), and resistant colonies were scored 10–15 days later. Neo and puro titers are expressed in neo- or puro-RFFU/ml (resistance focus-forming units per ml). Results are the mean of at least two independent transfection–infection–titration experiments

To reduce promoter interference, the distance between LTR and SV40 promoters was increased by inverting the orientation of the puro transcription unit (Fig. 3). At the same time, the SV40 polyA signal was removed as well because the SV40 DNA fragment contains a bidirectional poly(A) signal [39]. This resulted in the generation of vectors NP(-)3C(DpA) and NP(-)3Catt(DpA) (Fig. 3). Neo and puro titers were 5–10 times higher, respectively, when comparing the NP(-)3C(DpA) and NP()3Catt(DpA) vectors with NP3C and NP3Catt vectors (Table 3). In these vectors, the deletion of the SV40 polyA signal implies that RNAs transcribed from the SV40 promoter are stopped at cellular polyadenylation sites. The strategies to maximize expression of the inverted puro transcription units should include the provision of an efficient unidirectional poly(A) signal. Insertion of a unidirectional polyA signal in this transcription unit would probably have no effect on neo gene transduction but may promote the expression of the puro gene and, subsequently, may further increase transduction of the puro gene. Inverting the orientation of the internal att sequence had less effect on transduction of the puro gene To determine if an additional ALSV att sequence, in opposite orientation, could be used as well for self-deletion, the internal att sequence in the NP3Catt vector was inserted in the U3/U5 orientation, in the opposite orientation relative to the LTRs, which gives rise to vector NP3Catt(inv) (Fig. 3a). A similar neo titer was observed in comparison with the NP3Catt vector (Table 3), but a sevenfold decrease in puro titer was observed, suggesting that the att sequence was not functional in this context.

123

2242

Att vector as a model of safe retroviral vectors for gene transfer experiments The present study exploits features of the retroviral life cycle, namely the att substrate for integration, to construct a new generation of vectors that self-delete at a high frequency during one round of replication. The precedent study [26] demonstrated that the insertion of an att sequence in a vector genome confers efficient self-deletion without lowering the vector titer or impairing the expression of the vector. The NP3Catt vector mainly generated two kinds of proviruses: (i) those flanked by an att sequence and an LTR (class I), and (ii) those flanked by 2 LTRs (class II). From previous data [26] and those presented here, we have shown that (i) depending on size of the att and on selection pressure, as many as 57% of proviruses belong to class I and have lost the 50 LTR (Table 2), which can reduce promoter interference and increase the expression of the puro gene, (2) as many as 92–100% of proviruses, according to experiments, have the psi sequence deleted, reducing the likelihood of RCR viruses, (3) under some selection conditions, as many as 92–100% of proviruses have the selectable neo gene deleted (Table 1). A drug resistance gene may be necessary for the selection of virus-producing helper cells, but is no longer desirable once the vector is integrated into target cells. Deletion of the neo-selectable marker may also decrease the probability of eliciting an immune response against therapeutically infected cells or inducing alteration of cellular gene expression [40, 41]. Finally, the most efficient att vector (1) would have an att sequence whose size is around 90–122 bp long to generate mostly att proviruses and to eliminate full-length proviruses; size should be higher to produce att–att proviruses (at least 157 bp long). (2) It should carry the transcription unit in reverse orientation, with a unidirectional polyadenylation signal, and (3) it should be produced in an cell line generating a high titer of viral particles or should be pseudotyped with the VSV-G protein. Conclusion This vector design offers significant biosafety features. Indeed, the self-deleting retroviral vector described in this report consistently self-deletes in recipient cells, mostly during the reverse-transcription step, but also at the time of integration into target DNA [26]. The deletion of the packaging sequence in almost all of the resulting proviruses (92–100% according to the experiment) should prevent mobilization by replication-competent viruses and should minimize any risk of spread of vector viruses. Deletion of neo sequences in 100% of the clones may minimize

123

C. Torne-Celer et al.

interactions involving metabolism or immune responses against this bacterial enzyme. We have also succeeded in infecting human cells (expressing the avian ALSV receptor (tva)) with the NP3Catt vector. In these cells, the vector also displayed self-deleting properties at high frequencies (Torne-Celer et al., in preparation). Therefore, the att vector offers a reasonable alternative to the conventional vectors currently used for gene transfer experiments. These results have general implications for the design and development of more sophisticated and improved retroviral vectors for gene transfer experiments as well as for gene therapy. Acknowledgments This work was supported by research grants from the Institut National de la Recherche Agronomique (INRA), the Centre National de la Recherche Scientifique (CNRS), and the Association pour la Recherche sur le Cancer (ARC). We thank Transgene SA (Strasbourg), the INRA and the Ligue Nationale Contre le Cancer for fellowships (CTC) and the Ministe`re de l’Education Nationale et de la Recherche (MENRT) for fellowships (KM). Special thanks to Dr. Antoine Drynda and Dr. Yahia Chebloune for their helpful contributions to this work as well as to Dr. Christophe Terzian for statistical analyses. Thanks are also due to Dr. M. Mehtali (Transgene) and collaborators for fruitful discussions and to the INRA translation service for correction of the English.

References 1. Clark J, Whitelaw B (2003) A future for transgenic livestock. Nat Rev Genet 4:825–833 2. Hofmann A, Kessler B, Ewerling S, Weppert M, Vogg B, Ludwig H, Stojkovic M, Boelhauve M, Brem G, Wolf E, Pfeifer A (2003) Efficient transgenesis in farm animals by lentiviral vectors. EMBO Rep 4:1054–1060 3. Pfeifer A (2004) Lentiviral transgenesis. Transgenic Res 13:513– 522 4. Ailles LE, Naldini L (2002) HIV-1-derived lentiviral vectors. Curr Top Microbiol Immunol 261:31–52 5. Aiuti A, Slavin S, Aker M, Ficara F, Deola S, Mortellaro A, Morecki S, Andolfi G, Tabucchi A, Carlucci F, Marinello E, Cattaneo F, Vai S, Servida P, Miniero R, Roncarolo MG, Bordignon C (2002) Correction of ADA-SCID by stem cell gene therapy combined with nonmyeloablative conditioning. Science 296:2410–2413 6. Cavazzana-Calvo M, Thrasher A, Mavilio F (2004) The future of gene therapy. Nature 427:779–781 7. Sinn PL, Sauter SL, McCray PB Jr (2005) Gene therapy progress and prospects: development of improved lentiviral and retroviral vectors-design, biosafety, and production. Gene Ther 12:1089– 1098 8. Hu WS, Pathak VK (2000) Design of retroviral vectors and helper cells for gene therapy. Pharmacol Rev 52:493–511 9. Xu K, Ma H, McCown TJ, Verma IM, Kafri T (2001) Generation of a stable cell line producing high-titer self-inactivating lentiviral vectors. Mol Ther 3:97–104 10. Cockrell AS, Ma H, Fu K, McCown TJ, Kafri T (2006) A translentiviral packaging cell line for high-titer conditional self-inactivating HIV-1 vectors. Mol Ther 14:276–284 11. Loewen N, Poeschla EM (2005) Lentiviral vectors. Adv Biochem Eng Biotechnol 99:169–191 12. Hughes SH (2004) The RCAS vector system. Folia Biol (Praha) 50:107–119

A self-deleting retroviral vector 13. Ronfort C, Legras C, Verdier G (1997) The use of retroviral vectors for gene transfer into bird embryo. In: Transgenic animals: generation and use. Harwood Academic Publishers GMBH, Switzerland, pp 83–94 14. Dong B, Silverman RH, Kandel ES (2008) A natural human retrovirus efficiently complements vectors based on murine leukemia virus. PLoS ONE 3:e3144 15. Cockrell AS, Kafri T (2007) Gene delivery by lentivirus vectors. Mol Biotechnol 36:184–204 16. Logan AC, Haas DL, Kafri T, Kohn DB (2004) Integrated selfinactivating lentiviral vectors produce full-length genomic transcripts competent for encapsidation and integration. J Virol 78:8421–8436 17. Choulika A, Guyot V, Nicolas JF (1996) Transfer of single genecontaining long terminal repeats into the genome of mammalian cells by a retroviral vector carrying the cre gene and the loxP site. J Virol 70:1792–1798 18. Russ AP, Friedel C, Grez M, von Melchner H (1996) Selfdeleting retrovirus vectors for gene therapy. J Virol 70:4927– 4932 19. Fernex C, Dubreuil P, Mannoni P, Bagnis C (1997) Cre/loxPmediated excision of a neomycin resistance expression unit from an integrated retroviral vector increases long terminal repeatdriven transcription in human hematopoietic cells. J Virol 71:7533–7540 20. Julias JG, Hash D, Pathak VK (1995) E-vectors: development of novel self-inactivating and self-activating retroviral vectors for safer gene therapy. J Virol 69:6839–6846 21. Delviks KA, Hu WS, Pathak VK (1997) Psi- vectors: murine leukemia virus-based self-inactivating and self-activating retroviral vectors. J Virol 71:6218–6224 22. Delviks KA, Pathak VK (1999) Development of murine leukemia virus-based self-activating vectors that efficiently delete the selectable drug resistance gene during reverse transcription. J Virol 73:8837–8842 23. Grunwald T, Pedersen FS, Wagner R, Uberla K (2004) Reducing mobilization of simian immunodeficiency virus based vectors by primer complementation. J Gene Med 6:147–154 24. Lund AH, Duch M, Lovmand J, Jorgensen P, Pedersen FS (1997) Complementation of a primer binding site-impaired murine leukemia virus-derived retroviral vector by a genetically engineered tRNA-like primer. J Virol 71:1191–1195 25. Brown PO (1997) Integration in retroviruses. In: Coffin JM, Hugues SH, Varmus HE (eds) Cold Spring Harbor Laboratory Press, Cold spring Harbor, pp 161–204 26. Torne-Celer C, Moreau K, Faure C, Chebloune Y, Verdier G, Ronfort C (2008) A novel self-deleting vector carrying an additional sequence recognized by the viral Integrase (IN). Virus Res 135:72–82 27. Moscovici C, Moscovici MG, Jimenez H, Lai MM, Hayman MJ, Vogt PK (1977) Continuous tissue culture cell lines derived from chemically induced tumors of Japanese quail. Cell 11:95–103

2243 28. Cosset FL, Legras C, Chebloune Y, Savatier P, Thoraval P, Thomas JL, Samarut J, Nigon VM, Verdier G (1990) A new avian leukosis virus-based packaging cell line that uses two separate transcomplementing helper genomes. J Virol 64:1070–1078 29. Frisby DP, Weiss RA, Roussel M, Stehelin D (1979) The distribution of endogenous chicken retrovirus sequences in the DNA of galliform birds does not coincide with avian phylogenetic relationships. Cell 17:623–634 30. Ronfort C, Chebloune Y, Cosset FL, Faure C, Nigon VM, Verdier G (1995) Structure and expression of endogenous retroviral sequences in the permanent LMH chicken cell line. Poultry Sci 74:127–135 31. Emerman M, Temin HM (1984) Genes with promoters in retrovirus vectors can be independently suppressed by an epigenetic mechanism. Cell 39:449–467 32. Emerman M, Temin HM (1986) Quantitative analysis of gene suppression in integrated retrovirus vectors. Mol Cell Biol 6:792–800 33. Emerman M, Temin HM (1986) Comparison of promoter suppression in avian and murine retrovirus vectors. Nucleic Acids Res 14:9381–9396 34. Panganiban AT, Talbot KJ (1993) Efficient insertion from an internal long terminal repeat (LTR)-LTR sequence on a reticuloendotheliosis virus vector is imprecise and cell specific. J Virol 67:1564–1571 35. Panganiban AT, Temin HM (1984) Circles with two tandem LTRs are precursors to integrated retrovirus DNA. Cell 36:673– 679 36. Ellis J, Bernstein A (1989) Retrovirus vectors containing an internal attachment site: evidence that circles are not intermediates to murine retrovirus integration. J Virol 63:2844–2846 37. Lobel LI, Murphy JE, Goff SP (1989) The palindromic LTR– LTR junction of Moloney murine leukemia virus is not an efficient substrate for proviral integration. J Virol 63:2629–2637 38. Soriano P, Friedrich G, Lawinger P (1991) Promoter interactions in retrovirus vectors introduced into fibroblasts and embryonic stem cells. J Virol 65:2314–2319 39. Bushman AR, Burnett L, Berg P (1981) The SV40 nucleotide sequence. Appendix A. In: Tooze J (ed) DNA tumor viruses, 2nd edn. Cold Spring Harbor Laboratory, New York, pp 799–841 40. Valera A, Perales JC, Hatzoglou M, Bosch F (1994) Expression of the neomycin-resistance (neo) gene induces alterations in gene expression and metabolism. Hum Gene Ther 5:449–456 41. von Melchner H, Housman DE (1988) The expression of neomycin phosphotransferase in human promyelocytic leukemia cells (HL60) delays their differentiation. Oncogene 2:137–140 42. Thoraval P, Savatier P, Xiao JH, Mallet F, Samarut J, Verdier G, Nigon V (1987) Partial nucleotide sequence of the avian erythroblastosis virus (AEV ES4). Nucleic Acids Res 15:9612

123