Gene Therapy (2012) 19, 800 -- 809
& 2012 Macmillan Publishers Limited All rights reserved 0969-7128/12
www.nature.com/gt
ORIGINAL ARTICLE
Novel random peptide libraries displayed on AAV serotype 9 for
selection of endothelial cell-directed gene transfer vectors
K Varadi1,2, S Michelfelder3, T Korff 4, M Hecker4, M Trepel3, HA Katus1, JA Kleinschmidt2 and OJ Müller1
We have demonstrated the potential of random peptide libraries displayed on adeno-associated virus (AAV)2 to select for AAV2
vectors with improved efficiency for cell type-directed gene transfer. AAV9, however, may have advantages over AAV2 because
of a lower prevalence of neutralizing antibodies in humans and more efficient gene transfer in vivo. Here we provide evidence
that random peptide libraries can be displayed on AAV9 and can be utilized to select for AAV9 capsids redirected to the cell
type of interest. We generated an AAV9 peptide display library, which ensures that the displayed peptides correspond to the
packaged genomes and performed four consecutive selection rounds on human coronary artery endothelial cells in vitro. This
screening yielded AAV9 library capsids with distinct peptide motifs enabling up to 40-fold improved transduction efficiencies
compared with wild-type (wt) AAV9 vectors. Incorporating sequences selected from AAV9 libraries into AAV2 capsids could not
increase transduction as efficiently as in the AAV9 context. To analyze the potential on endothelial cells in the intact natural
vascular context, human umbilical veins were incubated with the selected AAV in situ and endothelial cells were isolated.
Fluorescence-activated cell sorting analysis revealed a 200-fold improved transduction efficiency compared with wt AAV9
vectors. Furthermore, AAV9 vectors with targeting sequences selected from AAV9 libraries revealed an increased transduction
efficiency in the presence of human intravenous immunoglobulins, suggesting a reduced immunogenicity. We conclude that
our novel AAV9 peptide library is functional and can be used to select for vectors for future preclinical and clinical gene
transfer applications.
Gene Therapy (2012) 19, 800--809; doi:10.1038/gt.2011.143; published online 29 September 2011
Keywords: adeno-associated virus; random peptide display library; endothelial cell; cardiovascular system; vector targeting
INTRODUCTION
Endothelial cells have a central role in vascular diseases such as
hypertension and atherosclerosis, as well as allograft vasculopathy
and stenosis after balloon dilatation or bypass vein grafting. Gene
transfer approaches into the endothelium with adenoviral vectors
were able to prevent hypertension in spontaneously hypertensive
rats.1,2 Although adenoviral vectors revealed only a transient
expression and strong inflammatory response after endothelial
gene transfer in a rabbit model, adeno-associated virus (AAV)
vectors enabled a more sustained transduction of endothelial
cells.3 However, transduction efficiency of endothelial cells was
found to be very low in vivo3,4 and in vitro.3,5,6
Transduction efficiency of endothelial cells could be increased
by introduction of an endothelial-targeting peptide identified by
phage display within the AAV capsids or selection of random AAV
display peptide libraries on endothelial cells.7--13 Such retargeted
vectors resulted in an increase in transduction rates by about
40-fold. Selected peptides allowed transduction of endothelium
to a certain extent in vivo.7--9,11--13 However, these previous approaches used AAV2 serotype vectors that might not be ideal for
future clinical approaches because there is a high prevalence of
immunoglobulin G against AAV2 in human serum.14,15 In contrast,
AAV9 serotype vectors are less affected by neutralizing factors in
human serum peptide14 and allow a highly efficient gene transfer
after intravenous vector administration.16--19 Thus, we reasoned
that AAV9 vectors might be a more suitable basis for developing
an endothelial gene transfer vector using endothelial-targeting
1
peptides. Because peptides identified by phage display might
result in conformational changes of the peptide when incorporated directly into the viral surface or might not function efficiently
for vector-targeting purposes, combinatorial approaches displaying a library of randomized peptides on the AAV surface8,20,21 or
capsid modification by shuffled or error-prone PCR22--26 may
represent an alternative. Therefore, the aims of our study were (i)
to identify a suitable insertion site for targeting peptides within
the AAV9 capsid surface, (ii) to generate a random AAV9 peptide
display library for selection of an AAV9-based targeting vector on
endothelial cells and (iii) to compare vectors selected from AAV9
libraries with those from our previous AAV2 libraries. We could
confirm that site A589 in the AAV9 capsid is a suitable region for
insertion of a targeting peptide as predicted from a structure
comparison of AAV9 to AAV2. We were able to successfully insert a
highly diverse oligopeptide library into this region and finally
selected vectors with reduced immunogenicity enabling highly
efficient transduction of endothelial cells in vitro as well as in
human umbilical veins in situ.
RESULTS
Endothelium-targeted heptapeptides displayed at AAV9 capsid
amino-acid residue A589 modulate the transduction efficiency
of AAV9 vectors on human endothelial cells
Previously, we have shown that selection of a random AAV2
peptide library displayed in close vicinity to arginine residue R588
Department of Internal Medicine III, University Hospital Heidelberg, Heidelberg, Germany; 2Applied Tumor Virology, German Cancer Research Center, Heidelberg, Germany;
Department of Oncology and Hematology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany and 4Department of Physiology, University of Heidelberg,
Heidelberg, Germany. Correspondence: Dr OJ Müller, Department of Internal Medicine III, University Hospital Heidelberg, Im Neuenheimer Feld 410, 69120 Heidelberg, Germany.
E-mail: [email protected]
Received 17 January 2011; revised 15 June 2011; accepted 20 June 2011; published online 29 September 2011
3
Endothelial vectors selected from AAV9 libraries
K Varadi et al
801
AAV2 wild-type sequence
GGC
AAC
AGA
CAA
GCA
GCT
G586
N587
R588
Q589
A590
A591
AAV9 wild-type sequence
GCC
CAA
GCA
CAG
GCG
CAG
A587
Q588
A589
Q590
A591
Q592
GCC
CAG
GCG
GCC
A
Q
A
A
G
AAV9 modified sequence
SfiI
SfiI
GGC
CAA
GCA
GGC
CAAG
G587
Q588
A589
G590
frame-shift!
Oligonucleotide design (double-strand)
T
A
GGC
21-mer
GCC
CAG
CGT
CCG
21-mer
CGG
G
A589
G590
7-mer
A598
Q599
AAV9 modified sequence with oligonucleotide insertion
GGC CAA GCA GGC AAT GAT GTT AGG GCG GTG AGT GCC CAG GCG GCC
G587 Q588 A589 G590 N591 D592 V593 R594 A595 V596 S597 A598 Q599 A600 A601
Figure 1. Design of the site for insertion of an oligonucleotide library into the AAV9 cap-gene. (a) Because of an additional amino acid within
the AAV9 VP, residue A589 corresponds to R588 from AAV2 (green). The wtAAV9 cap open reading frame (ORF) was modified to allow
engineering of two SfiI recognition sequences (frames and arrows) required for oligonucleotide insertion, resulting in a total of four amino
acid modifications (red). Two adenine nucleotides separating the SfiI recognition sites induce a frameshift in the cap ORF (indicated by opaque
residues) and prevent the creation of AAV without insert. Upon insertion of a suitable oligonucleotide (dashed frame), the frameshift is
restored and AAV displaying a peptide (blue) are produced. (b) Transduction efficiencies of wt and mutant AAV2 or AAV9 vectors on HCAECs.
Wt AAV9-CMV-EGFP vectors were compared with mutant vectors displaying heptapeptide NDVRAVS behind residue 588 (AAV2) or 589 (AAV9).
After an incubation of 72 h at MOI 1 104 g.c./cell, transduction efficiencies were monitored by flow cytometry. Efficiencies are given in %
EGFP-positive cells as means þ s.d. (n ¼ 4, two independent vector productions).
involved in heparan sulfate proteoglycan (HSPG) binding, yielded
peptides with an improved transduction efficiency of AAV2
vectors for human endothelial cells.8 Unavailable data of the
capsid structure and yet unrevealed binding sites for receptors
involved in the transduction process, however, limited the
possibilities by which plausible peptide insertion sites could be
predicted for AAV9. An alignment of AAV2 and AAV9 capsid
sequences revealed a high variability in the amino acid composition surrounding R588 (not shown), suggesting that this region
could also be involved in AAV9 tropism. Because of an additional
amino acid within the AAV9 capsid sequence the corresponding
site to R588 is A589 (Figure 1a, green letters). We engineered a
plasmid construct comparable to our previous AAV2 plasmid
library by introducing two incompatible SfiI restriction sites in the
AAV9-cap open reading frame, which enables cloning of oligonucleotides in an oriented fashion. A frameshift in the capsid open
reading frame prevents capsid formation from plasmids without
oligonucleotide insertion. The architecture of our construct leads to
the production of capsid mutants displaying a heptapeptide at
capsid site A589 (VP1 numbering) with modification of the wildtype (wt) sequence at only four residues (Figure 1a, red letters).
We created wt and mutant AAV9 vectors displaying NDVRAVS,
an endothelium-targeting peptide obtained by in vitro-selection of
an AAV2 library8 and compared their transduction efficiency to
wtAAV2 and AAV2-NDVRAVS on human coronary artery endothelial cells (HCAECs). Although wtAAV2 and wtAAV9 showed
transduction efficiencies of 19.2±10.7% and 12.4±3.9%, respectively, AAV2- and AAV9-NDVRAVS improved efficiencies to
71.7±15.7% and 74.2±8.9% transduced cells, respectively
(Figure 1b), confirming our assumption that capsid site A589 has
a role in the transduction process of AAV9 vectors and is suitable
for insertion of targeting peptides.
random insertions. Oligonucleotides encoding randomized heptapeptides were produced following a NNK-triplet design (N, A, C,
G or T; K, G or T) that circumvents the expression of ocher and
opal stop codons, but does not restrict the variety of amino acids
within the peptide. The random library was cloned and produced
at the plasmid level and virus particle level. Both, plasmid and
virus library were characterized in terms of complexity and amino
acid composition. Counting bacterial colonies after transformation
with the plasmid library yielded a complexity of 6.02 108 clones
for the entire library. Sequence analysis of 79 randomly assigned
plasmid clones revealed an amino acid distribution comparable to
the theoretical expectation (Supplementary Figure S1a). The
functional complexity of the plasmid library was calculated
3.9 108 because 30 out of 79 analyzed clones possessed one
or two stop codons within the randomized heptamer.
We used the plasmid library to create the intermediate-step
transfer shuttle library that resulted in a total of 3.2 1012 genome
copies (g.c.) and 5 107 infectious units (IU), covering the complexity
of the plasmid library (Supplementary Table S1a). Human embryonic
kidney cells (HEK)293T were infected with a multiplicity of infection
(MOI) of 0.5 IU/cell using the entire transfer shuttle library production,
thus functionally maintaining the complexity of the plasmid library in
the final virus library. The genomic titer of the final virus library was
4.7 1011 g.c. in total, and sequence analysis of 70 random clones
from the virus library revealed a decreased average presence of
cysteines (C), aspartic (D) and glutamic acid (E) and threonine (T)
within the heptamere, whereas the presence of asparagine (N) and
serine (S) was slightly increased (Supplementary Figure S1a).
Altogether, the analysis of the AAV9 plasmid and virus library
suggests a display of random peptides with sufficient diversity and
broad amino acid usage for the selection process. Furthermore, these
data show that insertion of peptides at capsid site A589 does not
hamper genome packaging or capsid assembly.
The AAV9 peptide library displays random heptameres with a
virtually unbiased amino acid distribution
After identification of a suitable AAV9 capsid site for the insertion
of targeting ligands, we set out to generate a capsid library with
Selection of the AAV9 library on HCAECs results in enrichment of
distinct peptides
We applied three different MOIs of 10, 100 or 1000 g.c./cell and
selected the AAV9 library four consecutive times on HCAECs. After
& 2012 Macmillan Publishers Limited
Gene Therapy (2012) 800 -- 809
Endothelial vectors selected from AAV9 libraries
K Varadi et al
802
round three and four we analyzed the heptamere composition of
several clones recovered from each applied MOI. Independent of
selection round or MOI, peptide RGDLRVS represented the most
prominent peptide (Figure 2a). Second most common peptides
were NFTRLSA and SLRSPPS with a more restricted distribution
than RGDLRVS. Notably, some peptides shared certain residues or
had similar sequences (RGDLRVS/FRVG, S/LT/IRSPPS) whereas others
differed in amino acid composition (DAKWDYR, GAIPDLR). These
results show that three to four rounds of selection of the AAV9
library on HCAECs yielded a consistent enrichment of certain
peptides independent of the MOI applied.
Enriched peptides improve transduction efficiency of AAV9
vectors in a serotype-dependent manner, but do not confer
specificity for HCAECs
To assess the potential of enriched peptides to transduce HCAECs,
an enhanced green fluorescent protein (EGFP) reporter gene
under control of the cytomegalovirus (CMV) promoter was
packaged in AAV9 capsids displaying selected heptapeptides
(Figure 2b). The vector production yielded AAV9 mutant vector
titers below those of wtAAV9 or wtAAV2 vectors but generally
higher than those of comparable mutant AAV2 vectors (Supplementary Table S1). Transduction efficiencies were compared with
those of wtAAV9, AAV9-NDVRAVS, as well as to AAV9-TEWDQPF,
which displays a random control peptide. Almost all mutant AAV9
vectors displaying any of the selected peptides significantly
outperformed wtAAV9, TEWDQPF and endothelium-targeting
NDVRAVS at MOIs 1 103 (Po0.05) and 1 104 (Po0.001)
g.c./cell, except NFTRLSA and NNVRGFV, which showed no
significant differences to NDVRAVS at MOI 1 103 g.c./cell
(Figure 2b). Peptides SL/IRSPPS and NLHSPPA enabled by trend
the highest transduction efficiencies at MOI 1 104 g.c./cell
compared with the other selected peptides.
The improvement of AAV9 transduction by display of endothelium-targeted peptides raised the question whether also specificity is increased. We therefore compared transduction efficiencies
of wt and mutant AAV9 vectors displaying SLRSPPS and RGDLRVS
on different cell types. At a MOI of 2.5 103 g.c./cell, wtAAV9
MOI 10
MOI 100
MOI 1000
R3 6/8
1/8
1/8
3/7
1/7
1/7
1/7
1/7
4/10
2/10
1/10
1/10
1/10
1/10
R4 8/15
4/15
2/15
1/15
3/14
3/14
2/14
2/14
2/14
1/14
1/14
7/14
3/14
2/14
1/14
1/14
vectors showed efficiencies of below 10% on primary cells
(HCAECs, human coronary artery smooth muscle cells) or cell
lines (HEK293T, HeLa, 911 and HepG2). Both, SLRSPPS and
RGDLRVS increased transduction efficiencies of other cell types
similarly to HCAECs, although to a different degree with the two
targeting peptide insertions (Figure 3a). Remarkably, transduction
patterns of SLRSPPS and RGDLRVS correlated with the origin of
the investigated cell types. SLRSPPS significantly (Po0.01) outperformed RGDLRVS on primary cells (HCAECs, human coronary
artery smooth muscle cells) whereas RGDLRVS showed higher
transduction efficiencies on cell lines (Po0.01 for 293T, HeLa, and
HepG2; Po0.05 for 911).
Both NDVRAVS and SLRSPPS are peptides that are endotheliumtargeted (HCAECs) but were selected from peptide libraries
displayed on different serotype capsids. Although selected
originally from an AAV2 random peptide library, NDVRAVS
efficiently enhances transduction of AAV9 to 58.4±5.1% in
contrast to 2.3±1.6% with wtAAV9 (Figure 2b). We therefore
wanted to know whether peptide sequences enhance transduction efficiency in a serotype-independent manner and incubated
HCAECs at a MOI of 5 103 g.c./cell with either wt or mutant AAV2
or AAV9 displaying peptides selected on HCAECs in context of
an AAV9 library (SLRSPPS, RGDLRVS, NLHSPPA) or in context of an
AAV2 library (NDVRAVS, NSSRDLG, NSVSSAS). Both AAV2 and
AAV9 wild-type vectors showed only basal transduction efficiencies of below 10% (Figure 3b). Overall, analyzed AAV9 peptides
displayed on AAV9 capsids yielded the highest transduction
efficiencies of HCAECs when compared with AAV2 peptides
displayed on AAV2 capsids. Cross-display of AAV9 peptides on
AAV2 capsids strongly reduced transduction. Similarly, except for
peptide NDVRAVS, efficiency of peptides selected from AAV2
libraries was low when displayed on AAV9 vectors indicating that
the peptides are not solely responsible for the modulation of
transduction efficiency but highly depend on the capsid context
of the serotype they are displayed on.
So far, we could show that our selected peptides improve
transduction efficiency over wt and mutant AAV9-NDVRAVS
vectors in a serotype-dependent manner, but do not confer
endothelium specificity to AAV9.
All sequences
Figure 2. Enriched peptides and their transduction efficiencies of HCAECs. (a) Individually assigned clones obtained from selection rounds
three (R3) and four (R4) of the AAV9 library on HCAECs at MOIs 10, 100 and 1000 g.c./cell were sequenced and translated into peptide chains
(one letter code). Amino-acid groups are indicated by different colors. The abundance of enriched peptides is indicated by their occurrence
out of total clones analyzed for each MOI. Column ‘all sequences’ gives an overview over all identified peptides from highest to lowest
abundance, top to bottom. (b) Wt and vectors-harboring enriched peptides were incubated at MOIs 1 103 (white bars) or 1 104 (black bars)
g.c./cell and transduction efficiencies were determined 48 h later by flow cytometry. Beside peptide NDVRAVS (positive control) from selection
of an AAV2 library and a non-selected random peptide TEWDQPF (negative control), seven other peptides were chosen based on their
frequency of occurrence. Efficiencies are given in % EGFP-positive cells as means þ s.d. (n ¼ 4, two independent vector productions).
Gene Therapy (2012) 800 -- 809
& 2012 Macmillan Publishers Limited
Endothelial vectors selected from AAV9 libraries
K Varadi et al
Figure 3. Cell and AAV serotype specificity of selected peptides.
(a) AAV9 wt and mutant vectors displaying SLRSPPS and RGDLRVS
were incubated with primary (HCAECs, human coronary artery
smooth muscle cells (HCASMC)) and immortalized cells (HEK293T,
HeLa, 911, HepG2) for 48 h at MOI 2.5 103 g.c./cell and transduction efficiencies were determined by flow cytometry. Efficiencies are
given in % EGFP-positive cells as means þ s.d. (n ¼ 4, two independent vector productions). (b) Transduction efficiencies of wt
and mutant AAV-CMV-EGFP vectors displaying endothelium-targeted peptides from AAV2 and AAV9 library selections on HCAECs.
Peptides SLRSPPS, RGDLRVS and NLHSPPA selected on HCAECs in
this study, as well as NDVRAVS, NSSRDLG and NSVSSAS selected on
HCAECs in an older study8 were cross-displayed on AAV2 or
AAV9 vector capsids, respectively. After 48 h incubation at MOI
5 103 g.c./cell, EGFP-positive HCAECs were analyzed by flow
cytometry. Efficiencies are given in % EGFP-positive cells as
means þ s.d. (n ¼ 3).
AAV9 vectors displaying peptides SLRSPPS and RGDLRVS bind a
common cell surface receptor
The differences in transduction patterns between AAV9 displaying
SLRSPPS and RGDLRVS on different cell types (Figure 3a) suggest
potential targeting of different transduction mechanisms or
pathways. To further investigate this possibility, we performed
competition assays with AAV9 vectors displaying these peptidesharboring fluorescent EGFP or non-fluorescent EGFPDGly67
expression cassettes. We reasoned that in case SLRSPPS and
RGDLRVS would target the same pathway, a co-incubation of a
‘fluorescent’ SLRSPPS vector with a ‘non-fluorescent’ RGDLRVS
vector supplied in excess should reduce the transduction signal
& 2012 Macmillan Publishers Limited
obtained from the ‘fluorescent’ SLRSPPS vector (and vice versa).
Preliminary co-incubations with fluorescent and non-fluorescent
vector variants displaying either peptide revealed a saturation
effect upon applying 25--100 g.c. excess of non-fluorescent
vector (Supplementary Figure S1b). We compared transduction
efficiencies of fluorescent vectors displaying SLRSPPS or RGDLRVS
with co-incubations of non-fluorescent vectors displaying
SLRSPPS, RGDLRVS as well as their scrambled counterparts
PSLPSRS and SDLRRGV at 75 g.c. excess and found significant
reductions of transduction in both groups (Figure 4a).
Co-incubation of fluorescent/non-fluorescent SLRSPPS vectors
resulted in a decrease of EGFP-positive cells from 100% (SLRSPPS
alone) to 58.2±3.8% (Po0.01) and of RGDLRVS from 100%
(RGDLRVS alone) to 56.7±5.0% (Po0.01). In contrast, the controls
using non-fluorescent vectors PSLPSRS and SDLRRGV did not alter
the transduction efficiency significantly. Importantly, co-incubations of fluorescent SLRSPPS/RGDLRVS vectors with non-fluorescent RGDLRVS/SLRSPPS vectors lowered the transduction
efficiencies to 60.8±2.8 (Po0.01) and 51.7±4.2 (Po0.05),
respectively, suggesting the partial use of a common transduction
mechanism.
The cell surface receptors responsible for AAV9 transduction still
remain to be unraveled. However, for AAV2, a cluster of several
basic arginines, R588 among them, represent binding sites for
HSPG and mutants lacking one or several of these arginine
residues exhibit a reduced infectivity.27,28 As both, SLRSPPS and
RGDLRVS deliver arginine residue(s) to the otherwise argininenegative AAV9 capsid at this site (Figure 1a) we asked whether our
enriched peptides could have re-targeted AAV9 to the HSPG
receptor. We incubated HCAECs and HepG2 cells with wt or
mutant AAV9 displaying SLRSPPS or RGDLRVS in the presence or
absence of solute heparin. Because HepG2 is expressing HSPG,
wtAAV2 control--vector efficiently transduced HepG2 in the
absence of heparin (94.8±1.5%, Figure 4b) but failed to do so
in its presence (Po0.001). Furthermore, only low transduction
efficiencies (o10%) were posed by wt AAV2 (HCAECs) or AAV9
(HCAECs and HepG2). Heparin did not alter the transduction
patterns of SLRSPPS on any cell type but did for RGDLRVS to a
small, but significant extent (Po0.05 for HCAECs and Po0.01 for
HepG2). In summary, we have shown that peptides SLRSPPS and
RGDLRVS partially share a common transduction mechanism and
that heparin treatment only affects peptide RGDLRVS but not
SLRSPPS.
Peptides SLRSPPS and RGDLRVS significantly evade neutralization
by antibodies in vitro
AAV vectors capable of escaping neutralization by pre-existing
antibodies are of high interest in gene therapy. We therefore
analyzed the feasibility of our leading mutants SLRSPPS and
RGDLRVS to avoid neutralization by capsid antibodies present in
human serum in vitro. Co-incubations of wt or mutant vectors with
serial dilutions of pooled intravenous immunoglobulin (IVIG) or
ADK9, a capsid-binding AAV9 antibody (JA Kleinschmidt, unpublished data), were performed on HEK293T cells using a MOI of
1.5 104 g.c./cell. We considered a vector sample as neutralized
when transduction efficiency dropped below 50% of the value
obtained by vector incubations without serum or antibody.
Control wtAAV2 and AAV2-NDVRAVS vectors, which displayed in
the absence of IVIG transduction efficiencies of 498%, were
neutralized throughout the entire range of dilutions (Figure 5a).
Similarly, wtAAV9 vectors were neutralized by both, IVIG and ADK9
up to dilutions of 1:1280 (translating into 3.125 pg ADK9 antibody
per 1.5 108 g.c.). However, mutants SLRSPPS and RGDLRVS
revealed improved in vitro antibody evasion over wtAAV9, wtAAV2
or AAV2-NDVRAVS about four and eight times, respectively,
(B1:320 and B1:160, Figure 5a) and were only affected by IVIG
(Figures 5a and b). Furthermore, the result correlates with our
Gene Therapy (2012) 800 -- 809
803
Endothelial vectors selected from AAV9 libraries
K Varadi et al
804
displaying peptide NDVRAVS, and therefore are more suitable for
targeted peptide display than AAV2-based vectors.
Peptide SLRSPPS strongly enhances transduction of human
umbilical vein endothelial cells (HUVECs) in human
umbilical veins
To reproduce our in vitro results in human tissue in situ, we used
postnatal human umbilical cords and infiltrated veins with wt or
mutant AAV9 vectors displaying SLRSPPS. To facilitate evaluation
of transduction efficiency, we isolated HUVECs from veins
incubated with vector and expanded them in vitro for 4 days to
obtain a sufficient cell amount. Flow cytometry analysis confirmed
the presence of endothelial cells, as 99.8±0.06% cells (n ¼ 3) were
CD31 positive. Although HUVECs showed a virtually non-existent
transduction by wtAAV9, AAV9 displaying peptide SLRSPPS
allowed a transduction efficiency of 80.3±11.9% (Po0.001,
Figure 6). In summary, in vitro selection of the AAV9 library yields
endothelium-targeted peptides that improve transduction efficiency of AAV9 not only in cell culture but also in intact human
tissue in situ.
Figure 4. Modulation of transduction efficiency in the presence of
competing capsid mutants or soluble heparin. (a) HCAECs were
exposed to AAV9-CMV-EGFP vector mutants displaying SLRSPPS or
RGDLRVS at MOI 5 103 g.c./cell exclusively (no competitor) or
as a co-incubation with ‘non-fluorescent’ AAV9-CMV-EGFPDGly67
mutants (competitors) displaying SLRSPPS, RGDLRVS or their
scrambled counterparts PSLPSRS and SDLRRGV supplied in 75 excess (MOI 3.75 105 g.c./cell). Efficiencies determined by flow cytometry are shown in % EGFP-positive cells as means þ s.d. (n ¼ 3) and
were calculated as relative percentage to single incubations with
mutants displaying SLRSPPS or RGDLRVS (set to 100%). (b) Cell-typedependent transduction efficiencies of wt and mutant AAV9-CMVEGFP vectors in the presence or absence of heparin: HCAECs (EC) or
HepG2 (Hep) cells were incubated with wt or mutant AAV9 vectors
displaying SLRSPPS or RGDLRVS at MOI 5 103 g.c./cell in the presence (w/) or absence (w/o) of 425 IU/ml heparin. WtAAV2 was used as
a positive control for the assay. Efficiencies determined by flow cytometry are shown in % EGFP-positive cells as means þ s.d. (n ¼ 3).
previous finding that ADK9 recognizes wtAAV9 vectors but not
wtAAV2, mutant AAV2 or mutant AAV9 vectors in the context of a
capsid enzyme-linked immunosorbent assay (not shown). Our
results substantiate that endothelium-targeted AAV9 vectors
displaying SLRSPPS or RGDLRVS improve immune escape from
human serum antibodies in vitro over wtAAV9, wtAAV2 and AAV2
Gene Therapy (2012) 800 -- 809
DISCUSSION
Efficiency of endothelial gene transfer using AAV vectors is low.
Insertion of peptides identified by phage display7,9,11--13 or
selection of random AAV display peptide libraries8,10 significantly
increased transduction of endothelial cells with AAV2 vectors. The
lower prevalence of neutralizing antibodies14 and a more efficient
gene transfer in animal models17--20 might render human AAV9
more suitable for endothelium targeting than AAV2.
Elucidation of the AAV2 capsid structure29 together with
identification of membrane-attached HSPG as a primary receptor
for AAV2 binding and infection30 as well as isolation of capsid
regions involved in HSPG binding27,28 enabled a more precise
design of AAV2 capsid mutants with altered tissue tropism and
enhanced transduction efficiencies. In contrast, only preliminary
X-ray crystallographic data of the AAV9 capsid structure were
published recently and structure determination is still under
progress.31 Because insertion of heptapeptides behind arginineresidue 588 (R588), part of several basic amino-acid residues
clustered within the threefold spike of the capsid surface involved
in HSPG binding,32 altered AAV2 tropism, we reasoned
that peptide sequence alignment of AAV2 and AAV9 capsids
would indicate the position of similar infection-relevant regions
within the AAV9 capsid. AAV2 and AAV9 share about 83% of their
capsid amino-acid sequences33 and sequence comparison of VP3
subunits comprising the HSPG-binding domain revealed highly
conserved regions (R484/R485 and R488/R489, (AAV2/AAV9)) as
well as variable regions (R585/S586 and R588/A589), which might
contribute to the different tissue tropism of these two serotypes.
Comparison of vectors displaying endothelium-targeted control
peptide NDVRAVS behind R588 of AAV2 or A589 of AAV9
confirmed our assumption as both mutants were able to
transduce HCAECs with significant higher efficiency than both
wt AAV2 and AAV9 (Figure 1b).
A total of four amino acid modifications were necessary to
introduce the SfiI site allowing oligopeptide insertion behind
residue A589 and preventing capsid formation without peptide
insertion (Figure 1a, red letters), leaving the residual wtAAV9
capsid sequence unaffected. In contrast, other studies intended to
create vectors with stronger capsid modifications either by errorprone PCR amplification or by in vitro recombination of parental
cap-genes from several AAV serotypes (DNA-family shuffling).20,22--26
These attempts shared beside improvement of transduction
efficiency another common advantage: removal of antigenic
epitopes from surface-exposed capsid domains. Neutralizing
antibodies in general bind to capsid epitopes, thus preventing
& 2012 Macmillan Publishers Limited
Endothelial vectors selected from AAV9 libraries
K Varadi et al
805
Figure 5. Neutralized transduction of AAV9 vectors in the presence of IVIG or ADK9. HEK293T cells were exposed for 48 h to wtAAV2 or
wtAAV9, as well as AAV2-NDVRAVS, AAV9-SLRSPPS or AAV9-RGDLRVS at an MOI 1.5 104 g.c./cell after preincubation with serial dilutions of
IVIG (a) or ADK9 capsid antibody (b). Neutralization of transduction was assumed when efficiency dropped below 50% of values obtained with
vectors in the absence of serum or antibody (not shown). Efficiencies determined by flow cytometry (% EGFP-positive cells) were calculated as
means þ s.d. (n ¼ 3) and are shown here as reciprocal % neutralization of transduction. The dilution of IVIG or ADK9 at which immune escape
has occurred can be extrapolated at the intersection of continuous and dashed lines (50% threshold).
Figure 6. In situ transduction of HUVEC with AAV9 displaying a selected peptide. Transduction efficiencies of wt and mutant AAV9
SLRSPPS vectors harboring a scCMV-EGFP genome were assessed by
in situ incubation of HUVEC. Umbilical veins were exposed for 2 h to
the vector and cells were collected by dispase treatment. Transduction efficiencies were analyzed after 96 h in vitro propagation of
cells by flow cytometry and are given in % EGFP-positive cells as
means þ s.d. (n ¼ 3 umbilical cords for each group).
interaction of AAV with cellular surface receptor(s).34 For AAV2,
several epitopes have been mapped within the loops of the
threefold spike, the most prominent domain of the capsid.35
Huttner et al.36 reported two AAV2 mutants carrying peptide
insertions I-534 and I-573 with a reduced affinity for AAV
antibodies in the majority of the analyzed serum samples. These
two insertions, especially I-573, are located in close proximity to
R588, probably pointing to a relationship between capsid domains
involved in tropism and domains serving as epitopes. We
therefore compared the ability of our AAV9 mutants SLRSPPS
and RGDLRVS to wt AAV2 and AAV9 vectors as well as to AAV2NDVRAVS to escape neutralization by IVIG, a pool of human
antisera applicable for in vitro neutralization studies of AAV20,32
(Figure 5a) and ADK9, a monoclonal antibody raised in mice
against wtAAV9 capsids (Figure 5b). AAV9 mutants RGDLRVS and
SLRSPPS improved in vitro escape from IVIG neutralization over
either wt or AAV2-NDVRAVS, underlining the suitability of AAV9
rather than AAV2 as a backbone for targeted peptide display. Still,
IVIG dilutions below 1:160 completely neutralized transduction of
& 2012 Macmillan Publishers Limited
the AAV9 mutants. However, the improved evasion lowers the
vector dose required for gene transfer attempts thereby reducing
the dose-dependent immune response to vectors in treated
subjects.37
We performed in parallel four selection rounds with MOIs 10,
100 and 1000 and analyzed enriched peptides after round three
and four (Figure 2a). Noticeably, the amount of different peptides
isolated from the MOI 10 approach was lower and more specific in
comparison with other approaches with higher MOIs, although a
comparable number of clones was analyzed. Still, RGDLRVS
represented the most prominent peptide. SLRSPPS, which confers
the best transduction efficiencies to AAV9 vectors (Figure 2b) by
trend, was also present after round three. Therefore, raising
the stringency of the selection process by lowering the MOI
(and therefore the complexity) of the library may lead to
propagation of ‘positive’ mutants (that is, mutants truly enhancing
transduction efficiency) at the same time depleting ‘false-positive’
mutants that only bind to the cell surface but are neither
internalized nor replicated and represent contaminating carryovers in the pool of positive virus mutants.
In addition, not only the peptide but also the serotype
backbone it is embedded in seems to influence the transduction
efficiency. Cross-display of clones selected from AAV9 libraries
on AAV2 capsids and vice versa showed that efficiency of AAV9derived peptide sequences is generally lower in the AAV2 context.
Also most sequences selected on AAV2 resulted in decreased
transduction in the AAV9 context, underlining the need of a
selection process in the context of the capsid on which the
peptide is considered to be expressed (Figure 3b). These findings
confirm our assumption that peptide sequences selected from an
library displayed on one serotype not necessarly work in the
context of another serotype. Thus, it cannot be generalized from
the perfomance of a peptide sequence from one serotype to
another, justifying generation of an AAV9 library. Interestingly,
NDVRAVS was detected only once (1/25) during the first of three
selection rounds on HCAECs8 but had a transduction efficiency
similar to other more frequently enriched peptides. A similar
observation was made with SLRSPPS in this study, which was also
under-represented after four selection rounds but outperformed
the most abundant peptide RGDLRVS. Obviously, there is no
correlation between abundance of peptides and their efficiency
and therefore criteria for the choice of peptides for further
analyzes should be rather their sequence than their abundance.
Gene Therapy (2012) 800 -- 809
Endothelial vectors selected from AAV9 libraries
K Varadi et al
806
Another important finding is the reasonable transduction
efficiency SLRSPPS conferred to AAV9 at the low MOI of
1.000 g.c./cell posing an improvement over the 10-fold higher
MOI we previously applied to enable efficient transduction of
HCAECs.8 Such high efficiency with a low vector dose could also
be advantageous for a future clinical use because a lower vector
exposure might provoke a reduced immune response37 that
complements the ability of this vector to escape neutralizing
antibodies.
Some peptides share certain fragments (RGD, SPPS) and nearly
all peptides show biased incorporations of single-amino acids
(mostly L, N, R and S). The elevated occurrence of these amino
acids in peptides from AAV2 libraries selected on different tissues
has been reported several times in literature.8,21,24 Our attention
was drawn toward arginine residues present in our most efficient
peptides. The binding of AAV2 to HSPG receptors depends mainly
on a subset of basic arginines (particularly R585/R588) that are
absent in the corresponding region of AAV9. According to our
results, wtAAV2 vectors efficiently transduced HSPG-positive
HepG2 cells in a heparin-dependent manner (Figure 4b). Because
augmentation of missing arginines can be achieved by displaying
enriched heptamers such as SLRSPPS and especially RGDLRVS,
we analyzed the effect of heparin on transduction efficiencies. For
RGDLRVS, relevant (HCAECs) and notable (HepG2) transduction
efficiencies were attained, which were slightly modulated by
heparin. Because of the minimal inhibitory effect heparin carries
on RGDLRVS here and the improved transduction RGDLRVS
exercises at the same time on HSPG-deficient HCAECs, a
transduction that has been redirected to HSPG receptors can be
excluded. The trimer RGD is known to promote binding to a
multitude of receptors from the integrin family (reviewed by
Ruoslahti et al.38) and it has been demonstrated that integrins
have an essential role as cellular entry receptors for AAV239 and
AAV9.40 Perabo et al.21 showed that the expression of RGD at
position 587 enables AAV2 to efficiently infect cells via an integrin
subtype and, importantly, those RGD-expressing vectors lacked
cellular specificity, which the authors ascribed to the wide-ranged
expression profile of integrins on different tissues. Our competition experiments suggest that both peptides are capable
of competing each other but not if the competing peptide is
displayed in a scrambled manner indicating that (i) the peptides
specifically modulate transduction patterns and (ii) partly use a
common transduction pathway. Because both vectors cannot be
completely inhibited by each others, additional alternative
receptors/postentry mechanisms might exist, which also could
explain different transduction patterns of cell lines and primary
cells (Figure 3a).
An important finding in this study is the suitability of our library
to select peptides in vitro, which also efficiently transduce human
(umbilical vein) endothelium ex vivo. According to our in vitro data,
at least 24 h of incubation upon vector exposition is necessary
to detect EGFP expression. Because endothelial integrity was
severely affected after 24 h, we were not able to detect EGFP
expression in endothelium in cryo-sections of human umbilical
veins after incubation with AAV9-SLRSPPS. To overcome these
technical limitations, we established a readout using flowcytometry. By thoroughly rinsing veins before harvesting endothelial cells, carry-over of vector contaminations into the cell
culture was prevented (see Materials and Methods).
In summary, we have shown the suitability of our AAV9 random
peptide library displayed behind residue A589 for selection of
peptides with an improved transduction efficiency of HCAECs
in vitro. Selected peptides outperformed transduction efficiency of
wtAAV9 and AAV9 displaying the known AAV2-library-derived
endothelium-targeted peptide NDVRAVS on HCAECs. A more
detailed analysis of two enriched peptides, SLRSPPS and RGDLRVS,
indicates the partial use of a common transduction mechanism
other than HSPG as well as a reduced immunogenicity.
Gene Therapy (2012) 800 -- 809
Furthermore, peptide SLRSPPS showed also a high in situtransduction efficiency of human umbilical vein endothelium,
rendering this targeting peptide a candidate for future preclinical
and clinical approaches.
MATERIALS AND METHODS
Cell culture
All primary cell types used in this study were provided cryo-preserved by
Promocell (Heidelberg, Germany) unless otherwise noted. HCAECs and
human coronary artery smooth muscle cells were maintained in a medium
supplied by the cell provider (C-22120 and C-22162, respectively).
Passaging of all primary cells was carried out using the Detach Kit
(Promocell). Cell lines HEK293T, HeLa, 911 and HepG2 were maintained in
Dulbecco’s modified Eagle’s medium (Sigma-Aldrich, Hamburg, Germany)
supplemented with 10% heat-inactivated fetal bovine serum (PAA, Coelbe,
Germany) and 2 mM L-glutamine (Gibco, Karlsruhe, Germany). Passaging
was performed with 0.05% Trypsin-EDTA (Gibco). All media were supplied
with 1 Pen/Strep/Fungizone solution (Promocell).
AAV vector production and purification
For production of self-complementary (sc) AAV-CMV-EGFP-vectors, rep/
cap-plasmids lacking inverted terminal repeats (ITRs), plasmid pDGDVPcontaining adenovirus helper functions41 and ITR-positive plasmid
dsAAV-CMV-EGFP containing a CMV-EGFP expression cassette42 were used.
Vectors bearing a non-fluorescent CMV-EGFP expression cassette packaged
a scCMV-EGFPDGly67 instead of a scCMV-EGFP genome. scCMVEGFPDGly67 was generated by in vitro-mutagenesis with forward 50 -GACC
ACCCTGACCTACGTGCAGTGCTTCAGCGCG-30 and reverse 50 -GCGGCTGAAG
CACTGCACGTAGGTCAGGGTGGTC-30 primers using the Quick Change II
Site-Directed Mutagenesis Kit (Stratagene, Waldbronn, Germany). For
generation of AAV2-based vectors, rep/cap-plasmids pDP2rs (wt)43 or
pMT187XX2 plus insert (capsid mutants)44 was used. AAV9 vectors were
produced with rep/cap-plasmids p5E18-VD2/9 (wt)45 or p5E18-VD-2/9SfiI1759 plus insert (capsid mutants). A total of 1 108 HEK293T cells were
transfected with EGFP-, rep/cap- and pDGDVP plasmids at a ratio of
1:1.2:3.5 (200, 240 and 700 mg) or 1:3.5 (200 and 700 mg) in case of EGFP and
pDP2rs (AAV2 vectors). For transfection, the protocol of Reed et al.46 was
slightly modified: 0.323 g/l PEI (Polysciences, Heidelberg, Germany) was
solved in water and freeze-thawed three times. A 80-ml transfection
solution containing 150 mM NaCl, plasmid DNA and 9.12 ml PEI-stock (N:Pratio 20) was prepared, incubated 12 min at room temperature to allow
DNA--PEI complex formation and was mixed to the medium. After 48--72 h
cells were harvested by Trypsin-EDTA and lysed by four freeze-thaw cycles
in cell lysis buffer (50 mM Tris Cl pH 8.5, 150 mM NaCl, 5 mM MgCl2). Cell
lysates were sonified with a Sonorex TK device (Bandelin, Berlin, Germany)
for 1 min at 48 W and treated with 100 U Benzonase (Sigma-Aldrich) per ml
of lysate for 30 min at 37 1C. AAV particles were then purified by iodixanol
density gradient ultracentrifugation as described.47
Generation of random libraries on the plasmid level
A 975-bp fragment of the wtAAV9-cap open reading frame was
synthesized (Geneart, Regensburg, Germany, plasmid pGA4) harboring
two incompatible SfiI restriction sites separated by two adenines at
nucleotide positions 1759 through 1786 of cap9 (Figure 1a). The modified
975-bp cap9 fragment (cap9mut) was excised with BsiWI/XcmI from pGA4
and was inserted into p5E18-VD-2/9, resulting in p5E18-VD-2/9-SfiI1759.
Plasmid p5E18-VD-2/9-SfiI1759 then was digested with HindIII and EcoRV,
releasing a 2774-bp fragment containing the entire cap9mut and a part of
rep2. This fragment was cloned into pMT-187-XX2. To enable the HindIII/
EcoRV digestion of pMT-187-XX2, the EcoRV site had to be generated by in
vitro mutagenesis with the Quick Change II Site-Directed Mutagenesis Kit
using forward 50 --CAATTACAGATTACGAGTCAGATATCGTGCCAATGGGGC
GAG--30 and reverse 50 --CTCGCCCCATTGGCACGATATCTGACTCGTAATCTG
TAATTG--30 primers according to the manufacturer’s recommendations.
The resulting plasmid was named pMT-187-XX2-rep2/cap9mut. Finally, the
entire rep2/cap9mut sequence was excised from pMT-187-XX2-rep2/
& 2012 Macmillan Publishers Limited
Endothelial vectors selected from AAV9 libraries
K Varadi et al
cap9mut with XbaI and was ligated into ITR-positive plasmid pSSV9(ref. 48)
resulting into the library plasmid backbone pKV-AAV9Lib/BB. To create
the oligonucleotide library, a degenerated single-stranded oligonucleotide 50 --CAGTCGGCCAAGCAGGC(NNK)7GCCCAGGCGGCTGACGAG--30 was
synthesized encoding a random seven-residue peptide insert flanked by
two incompatible BglI restriction sites (University of Freiburg, Oligonucleotide Synthesis Core Facility). This architecture allowed for directional
in-frame cloning into SfiI-digested pKV-AAV9Lib/BB plasmid (or p5E18-VD2/9-SfiI1759 for production of vector-harboring peptides). Second strand
synthesis was carried out with the Sequenase Kit (Amersham, Freiburg,
Germany) and second strand primer 50 --CTCGTCAGCCGCCTGG--30 . The
double-stranded insert was purified with the QIAquick Nucleotide Removal
Kit (Qiagen, Hilden, Germany). Digestion with SfiI cleaved a 12-bp stuffer
fragment within pKV-AAV9Lib/BB and the linearized plasmid backbone
was purified with the QIAquick PCR Purification Kit (Qiagen). Backbone and
insert were ligated at a 1:30 molar ratio and transformed into electrocompetent DH5a bacteria using the Gene Pulser (Biorad, München,
Germany). Transformed bacteria were grown over night at 30 1C to an
OD600 of 0.2. The plasmid library termed pKV-AAV9Lib was purified using
the Qiagen’s Plasmid Maxi Kit. Before purification, an aliquot was taken
from the culture and serial dilutions were plated on agar plates containing
100 mg/ml ampicillin. Library plasmid complexity was extrapolated for
a total of 10 mg transformed library plasmid by determining the number
of colonies.
Generation of AAV transfer shuttle library and the random
AAV display peptide library
A modified two-step protocol was used to create the AAV9 library.8 After
construction of a plasmid library we produced in the first step chimeric
wt and mutant AAV capsids, the transfer shuttle library particles by
PEI-transfecting 1 108 HEK293T cells with the ITR-positive pKV-AAV9Lib
plasmid along with ITR-deficient pRSV-VP3co-expressing a codon-optimized wtAAV2 VP3 capsid protein and pDGDVP at a ratio of 1:1:15 (20, 20
and 300 mg). Cells were harvested 72 h posttransfection and transfer
shuttle particles were purified by an iodixanol step gradient resulting in a
total volume of 2 ml. After determining replicative (3.2 1012 g.c.) and
infectious titers (5 107 IU), the final virus library was generated by
incubating 1 108 HEK293T with the transfer shuttle library at a MOI of 0.5
replicative particles per cell for 4 h in a medium containing 5% fetal bovine
serum. For superinfection, transfer shuttle library-containing medium was
removed, cells were washed once with 1 phosphate-buffered saline and
fresh medium containing wtAd5 used at a MOI of 10 infectious particles
(ip) per cell was supplied. Harvesting and titration of virus library
(4.7 1011 g.c./ml) occurred 48 h later upon appearance of 50% cytopathic
effect induced by Ad.
In vitro biopanning of AAV library on HCAECs
A total of 1 106 HCAECs were infected with AAV9 virus library at MOIs of
10, 100 or 1000 g.c./cell. After an incubation of 4 h, AAV9 library-containing
medium was removed, cells were washed with 1 phosphate-buffered
saline and were supplied with fresh medium containing wtAd5 at a MOI
of 20 ip/cell. Once an B50% cytopathic effect occurred after 48--72 h,
we harvested and titrated library particles as described. These particles
were re-applied in the next selection round using the same MOIs. The
protocol was repeated three more times resulting in a total of four
consecutive selection rounds, whereas individual clones from the third and
fourth selection round from each applied MOI were sequenced to identify
enriched peptides.
PCR amplification, sequencing and titrating of AAV particles
DNA from AAV particles was purified with the QIAamp MinElute Virus Spin
Kit (Qiagen) and subjected to standard PCR reaction with AAV9 cap-specific
forward 50 -GGAGGATCCGCAGGTACAGGTGTGT-30 and reverse primer
50 -GCTTGATGAATTCTGGACCTGCTATGGC-30 spanning the entire oligonucleotide insertion site. PCR products were cloned into pCR-2.1-TOPO (TOPO
TA Cloning Kit, Invitrogen, Darmstadt, Germany). Randomly assigned
& 2012 Macmillan Publishers Limited
clones were sequenced using the primer 50 -CAAACAAGGAACTGGAAGA
G-30 . Genomic titers were determined by quantitative real-time PCR
(TaqMan, Applied Biosystems, Darmstadt, Germany) using CMV-specific or
rep-specific primers and probe for vectors and viruses, respectively, as
described previously.49 Replicative titers of AAV virus particles were
determined by dot blot analysis.50 Detection of AAV genomes was
performed with a 32P-dCTP labeled (Random Primed DNA Labeling Kit,
Roche, Mannheim, Germany) SalI fragment from plasmid p5E18-VD2/9.
In vitro and in situ gene transfer studies with wt and selected
AAV vectors
Studies involving human material have been approved by the authors’
Institutional Review Board. For in vitro analysis of AAV transduction
efficiencies, 5--15 104 cells/well were seeded in 12-well plates (Corning,
München, Germany) in 1 ml corresponding medium and were incubated
the next day with AAV vectors at MOIs ranging from 1 103 to 1 104 in
500 ml medium (maximum 5% fetal bovine serum) for 48--72 h. For in situ
gene transfer human umbilical cords between 15 and 25 cm in length were
maintained at 37 1C but never 46 h. Residual blood was removed, veins
were rinsed with 1 HANKS buffer without Mg2 þ and Ca2 þ (PAA) and
were perfused with HUVEC medium devoid of hydro-cortisone (C-22110)
supplied with or without 1.75 109 g.c./ml AAV9-CMV-EGFP wt vector or
vector displaying peptide SLRSPPS. Depending on vein diameter, about
0.25 ml vector solution per cm of vessel was applied. After 2 h of
incubation at 37 1C, vector solution was removed and veins as well as parts
of the cord potentially contaminated by vector solution were rinsed
thoroughly with 1 HANKS buffer. HUVECs were then isolated by infusing
veins with 3.125 mg/ml dispase (Invitrogen) and incubating 30 min at
37 1C. Detached cells were collected by flushing veins with 1 HANKS
buffer. Cells were resuspended in the medium and maintained in flasks
pre-coated with 2% gelatin-type B (Sigma-Aldrich) with a medium
exchange occurring every other day. Expression of endothelial marker
CD31 was confirmed by detaching cells from flasks using non-enzymatic
cell dissociation solution (Sigma, Seelze, Germany) and incubation with 1:2
phosphate-buffered saline-diluted fluorescein isothiocyanate-labelled
platelet/endothelial cell adhesion molecule-1 antibody (LS-C43861, Lifespan Biosciences, Eching, Germany) for 45 min at 4 1C. For analysis of
transduction efficiency or immunostaining, cells were harvested and
resuspended in 1 Hank’s balanced salt solution lacking Mg2 þ and Ca2 þ
(Gibco) with 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid,
10 mM NaN3, 2% newborn calf serum (Sigma) and 1 mg/ml propidium
iodide (Sigma). Percentage of EGFP-positive cells was monitored by flow
cytometry on a fluorescence-activated cell sorting Calibur device (BecktonDickinson, Heidelberg, Germany). Transduction efficiencies were evaluated
with Flow Jo software (v.7.6.1, Tree Star, Inc., Olten, Germany) and
statistical calculations were done with Sigma Plot (v.10.0, Systat Software,
Inc., Erkrath, Germany). Efficiencies are given in % EGFP-positive cells and
values are means± s.d. An independent Student’s t-test was performed to
detect significant (Po0.05, Po0.02) or highly significant (Po0.01 and
Po0.001) differences.
In vitro neutralization assay with wt AAV and single mutants
In a round-bottom multi-well plate (96 well), 1:2 serial dilutions of 10%
pooled IVIG (Gamunex R, Talecrid Biotherapeutics, Frankfurt, Germany) or
AAV9 capsid antibody ADK9 (JA Kleinschmidt, unpublished data) were
made in a total of 30 ml medium lacking fetal bovine serum. The final
concentration of ADK9 antibody in the lowest dilution (1:10) was 0.4 ng.
Another 30 ml serum-free medium containing AAV-CMV-EGFP vectors
(wtAAV2, wtAAV9, mutant AAV2-NDVRAVS, AAV9-RGDLRVS or AAV9SLRSPPS) were added to the serial dilutions and incubated 45 min at
room temperature. Forty out of 60 ml neutralized vector solution resulting
in MOIs of 1.5 104 g.c./cell were added to 1 104 293T cells seeded the
day before in a multi-well plate (96 well) in 100 ml medium. Cells were
incubated for further 48 h. For analysis of transduction efficiency cells were
processed as described above. Neutralization was assumed when
transduction efficiency of samples treated with serum was reduced to
50% of mock-treated cells.
Gene Therapy (2012) 800 -- 809
807
Endothelial vectors selected from AAV9 libraries
K Varadi et al
808
Heparin and capsid competition assays with wtAAV and
single mutants
A total of 5 104 HCAECs or 2.5 105 HepG2 cells were seeded the day
before and were incubated with wt or selected AAV9-CMV-EGFP vectors
displaying peptides RGDLRVS and SLRSPPS in the presence or absence of
420 IU/ml heparin (Neolab, Heidelberg, Germany) at MOI 5 103 g.c./cell.
Wt AAV2-CMV-EGFP vector was used as a positive control. After 4 h,
medium containing vectors and/or heparin was replaced by fresh medium,
and cells were incubated for a total of 48 h. The capsid competition assay
was performed by co-incubation of vectors-harboring fluorescent EGFP
expression cassettes and competing vectors containing non-fluorescent
EGFPDGly67. A total of 7.5 105 HCAECs were seeded and incubated the
next day with AAV9-CMV-EGFP mutants displaying SLRSPPS or RGDLRVS
alone (efficiencies were set to 100%) or co-incubated with mutant AAV9CMV-EGFPDGly67 vectors displaying peptides RGDLRVS and SLRSPPS, as
well as mutants displaying the scrambled counterparts SDLRRGV and
PSLPSRS. Before use, each CMV-EGFPDGly67 vector was tested on HCAECs
at MOI 5 103 g.c./cell to ensure the absence of fluorescence signals other
than background. CMV-EGFP vectors were applied at MOI 5 103 g.c./cell
and CMV-EGFPDGly67 vectors in excess at MOIs 1 105 (20 ), 2.5 105
(50 ), 3.75 105 (75 ) and 5 105 (100 ) g.c./cell. After 4 h
co-incubation, vector-containing medium was replaced by fresh medium
and was further incubated to a total of 48 h. Transduction efficiencies were
evaluated by flow-cytometry as described.
CONFLICT OF INTEREST
The authors declare no conflict of interest.
ACKNOWLEDGEMENTS
We thank Barbara Leuchs (VP and DU, the German Cancer Research Center,
Heidelberg, Germany) and her team for excellent assistance in vector production and
titration and Ender Serbest (Department of Physiology, University of Heidelberg,
Heidelberg, Germany) for outstanding technical assistance related to processing of
human umbilical cords. This work was supported by the German Research
Foundation (1654/3-2 to JAK and OJM).
REFERENCES
1 Miller WH, Brosnan MJ, Graham D, Nicol CG, Morecroft I, Channon KM et al.
Targeting endothelial cells with adenovirus expressing nitric oxide synthase
prevents elevation of blood pressure in stroke-prone spontaneously hypertensive
rats. Mol Ther 2005; 12: 321--327.
2 Xu Y, Gong B, Yang Y, Awasthi YC, Boor PJ. Adenovirus-mediated overexpression
of glutathione-s-transferase mitigates transplant arteriosclerosis in rabbit carotid
allografts. Transplantation 2010; 89: 409--416.
3 Gruchala M, Bhardwaj S, Pajusola K, Roy H, Rissanen TT, Kokina I et al. Gene
transfer into rabbit arteries with adeno-associated virus and adenovirus vectors.
J Gene Med 2004; 6: 545--554.
4 Sen S, Conroy S, Hynes SO, McMahon J, O’Doherty A, Bartlett JS et al. Gene
delivery to the vasculature mediated by low-titre adeno-associated virus
serotypes 1 and 5. J Gene Med 2008; 10: 143--151.
5 Denby L, Nicklin SA, Baker AH. Adeno-associated virus (AAV)-7 and -8 poorly
transduce vascular endothelial cells and are sensitive to proteasomal degradation.
Gene Therapy 2005; 12: 1534--1538.
6 Pajusola K, Gruchala M, Joch H, Luscher TF, Yla-Herttuala S, Bueler H. Celltype-specific characteristics modulate the transduction efficiency of adenoassociated virus type 2 and restrain infection of endothelial cells. J Virol 2002; 76:
11530--11540.
7 Chen YH, Chang M, Davidson BL. Molecular signatures of disease brain endothelia
provide new sites for CNS-directed enzyme therapy. Nat Med 2009; 15:
1215--1218.
8 Müller OJ, Kaul F, Weitzman MD, Pasqualini R, Arap W, Kleinschmidt JA et al.
Random peptide libraries displayed on adeno-associated virus to select for
targeted gene therapy vectors. Nat Biotechnol 2003; 21: 1040--1046.
9 Nicklin SA, Buening H, Dishart KL, de Alwis M, Girod A, Hacker U et al. Efficient and
selective AAV2-mediated gene transfer directed to human vascular endothelial
cells. Mol Ther 2001; 4: 174--181.
Gene Therapy (2012) 800 -- 809
10 Waterkamp DA, Müller OJ, Ying Y, Trepel M, Kleinschmidt JA. Isolation of
targeted AAV2 vectors from novel virus display libraries. J Gene Med 2006; 8:
1307--1319.
11 White K, Büning H, Kritz A, Janicki H, McVey J, Perabo L et al. Engineering adenoassociated virus 2 vectors for targeted gene delivery to atherosclerotic lesions.
Gene Therapy 2008; 15: 443--451.
12 Work LM, Büning H, Hunt E, Nicklin SA, Denby L, Britton N et al. Vascular bedtargeted in vivo gene delivery using tropism-modified adeno-associated viruses.
Mol Ther 2006; 13: 683--693.
13 White SJ, Nicklin SA, Büning H, Brosnan MJ, Leike K, Papadakis ED et al. Targeted
gene delivery to vascular tissue in vivo by tropism-modified adeno-associated
virus vectors. Circulation 2004; 109: 513--519.
14 Boutin S, Monteilhet V, Veron P, Leborgne C, Benveniste O, Montus MF et al.
Prevalence of serum IgG and neutralizing factors against adeno-associated virus
(AAV) types 1, 2, 5, 6, 8, and 9 in the healthy population: implications for gene
therapy using AAV vectors. Hum Gene Ther 2010; 21: 704--712.
15 Calcedo R, Vandenberghe LH, Gao G, Lin J, Wilson JM. Worldwide epidemiology
of neutralizing antibodies to adeno-associated viruses. J Infect Dis 2009; 199:
381--390.
16 Goehringer C, Rutschow D, Bauer R, Schinkel S, Weichenhan D, Bekeredjian R et al.
Prevention of cardiomyopathy in delta-sarcoglycan knockout mice after
systemic transfer of targeted adeno-associated viral vectors. Cardiovasc Res
2009; 82: 404--410.
17 Inagaki K, Fuess S, Storm TA, Gibson GA, McTiernan CF, Kay MA et al. Robust
systemic transduction with AAV9 vectors in mice: efficient global cardiac gene
transfer superior to that of AAV8. Mol Ther 2006; 14: 45--53.
18 Pacak CA, Mah CS, Thattaliyath BD, Conlon TJ, Lewis MA, Cloutier DE et al.
Recombinant adeno-associated virus serotype 9 leads to preferential cardiac
transduction in vivo. Circ Res 2006; 99: e3--e9.
19 Zincarelli C, Soltys S, Rengo G, Rabinowitz JE. Analysis of AAV serotypes 1-9
mediated gene expression and tropism in mice after systemic injection. Mol Ther
2008; 16: 1073--1080.
20 Grimm D, Lee JS, Wang L, Desai T, Akache B, Storm TA et al. In vitro and in vivo
gene therapy vector evolution via multispecies interbreeding and retargeting of
adeno-associated viruses. J Virol 2008; 82: 5887--5911.
21 Perabo L, Büning H, Kofler DM, Ried MU, Girod A, Wendtner CM et al. In vitro
selection of viral vectors with modified tropism: the adeno-associated virus
display. Mol Ther 2003; 8: 151--157.
22 Koerber JT, Maheshri N, Kaspar BK, Schaffer DV. Construction of diverse adenoassociated viral libraries for directed evolution of enhanced gene delivery
vehicles. Nat Protoc 2006; 1: 701--706.
23 Maheshri N, Koerber JT, Kaspar BK, Schaffer DV. Directed evolution of adenoassociated virus yields enhanced gene delivery vectors. Nat Biotechnol 2006; 24:
198--204.
24 Perabo L, Endell J, King S, Lux K, Goldnau D, Hallek M et al. Combinatorial
engineering of a gene therapy vector: directed evolution of adeno-associated
virus. J Gene Med 2006; 8: 155--162.
25 Li W, Asokan A, Wu Z, Van Dyke T, DiPrimio N, Johnson JS et al. Engineering and
selection of shuffled AAV genomes: a new strategy for producing targeted
biological nanoparticles. Mol Ther 2008; 16: 1252--1260.
26 Yang L, Jiang J, Drouin LM, Agbandje-McKenna M, Chen C, Qiao C et al.
A myocardium tropic adeno-associated virus (AAV) evolved by DNA shuffling and
in vivo selection. Proc Natl Acad Sci USA 2009; 106: 3946--3951.
27 Kern A, Schmidt K, Leder C, Müller OJ, Wobus CE, Bettinger K et al. Identification of
a heparin-binding peptide on adeno-associated virus type 2 capsids. J Virol 2003;
77: 11072--11081.
28 Opie SR, Warrington Jr KH, Agbandje-McKenna M, Zolotukhin S, Muzyczka N.
Identification of amino acid residues in the capsid proteins of adeno-associated
virus type 2 that contribute to heparan sulfate proteoglycan binding. J Virol 2003;
77: 6995--7006.
29 Xie Q, Bu W, Bhatia S, Hare J, Somasundaram T, Azzi A et al. The atomic structure
of adeno-associated virus (AAV-2), a vector for human gene therapy. Proc Natl
Acad Sci USA 2002; 99: 10405--10410.
30 Summerford C, Samulski RJ. Membrane-associated heparan sulfate proteoglycan
is a receptor for adeno-associated virus type 2 virions. J Virol 1998; 72: 1438--1445.
31 Mitchell M, Nam HJ, Carter A, McCall A, Rence C, Bennett A et al. Production,
purification and preliminary X-ray crystallographic studies of adeno-associated
virus serotype 9. Acta Crystallogr Sect F Struct Biol Cryst Commun 2009; 65:
715--718.
32 Lochrie MA, Tatsuno GP, Christie B, McDonnell JW, Zhou S, Surosky R et al.
Mutations on the external surfaces of adeno-associated virus type 2 capsids that
affect transduction and neutralization. J Virol 2006; 80: 821--834.
33 Gao G, Vandenberghe LH, Alvira MR, Lu Y, Calcedo R, Zhou X et al. Clades of
Adeno-associated viruses are widely disseminated in human tissues. J Virol 2004;
78: 6381--6388.
& 2012 Macmillan Publishers Limited
Endothelial vectors selected from AAV9 libraries
K Varadi et al
809
34 Moskalenko M, Chen L, van Roey M, Donahue BA, Snyder RO, McArthur JG et al.
Epitope mapping of human anti-adeno-associated virus type 2 neutralizing
antibodies: implications for gene therapy and virus structure. J Virol 2000; 74:
1761--1766.
35 Wobus CE, Hugle-Dorr B, Girod A, Petersen G, Hallek M, Kleinschmidt JA.
Monoclonal antibodies against the adeno-associated virus type 2 (AAV-2) capsid:
epitope mapping and identification of capsid domains involved in AAV-2-cell
interaction and neutralization of AAV-2 infection. J Virol 2000; 74: 9281--9293.
36 Huttner NA, Girod A, Perabo L, Edbauer D, Kleinschmidt JA, Büning H et al.
Genetic modifications of the adeno-associated virus type 2 capsid reduce the
affinity and the neutralizing effects of human serum antibodies. Gene Therapy
2003; 10: 2139--2147.
37 Bessis N, Garcia-Cozar FJ, Boissier MC. Immune responses to gene therapy vectors:
influence on vector function and effector mechanisms. Gene Therapy 2004;
11(Suppl 1): S10--S17.
38 Ruoslahti E. RGD and other recognition sequences for integrins. Annu Rev Cell Dev
Biol 1996; 12: 697--715.
39 Asokan A, Hamra JB, Govindasamy L, Agbandje-McKenna M, Samulski RJ. Adenoassociated virus type 2 contains an integrin alpha5beta1 binding domain
essential for viral cell entry. J Virol 2006; 80: 8961--8969.
40 Akache B, Grimm D, Pandey K, Yant SR, Xu H, Kay MA. The 37/67-kilodalton
laminin receptor is a receptor for adeno-associated virus serotypes 8, 2, 3, and 9.
J Virol 2006; 80: 9831--9836.
41 Dubielzig R, King JA, Weger S, Kern A, Kleinschmidt JA. Adeno-associated
virus type 2 protein interactions: formation of pre-encapsidation complexes.
J Virol 1999; 73: 8989--8998.
42 Wang Z, Ma HI, Li J, Sun L, Zhang J, Xiao X. Rapid and highly efficient transduction
by double-stranded adeno-associated virus vectors in vitro and in vivo.
Gene Therapy 2003; 10: 2105--2111.
43 Grimm D, Kay MA, Kleinschmidt JA. Helper virus-free, optically controllable, and
two-plasmid-based production of adeno-associated virus vectors of serotypes 1
to 6. Mol Ther 2003; 7: 839--850.
44 Michelfelder S, Lee MK, deLima-Hahn E, Wilmes T, Kaul F, Müller O et al. Vectors
selected from adeno-associated viral display peptide libraries for leukemia
cell-targeted cytotoxic gene therapy. Exp Hematol 2007; 35: 1766--1776.
45 Gao GP, Alvira MR, Wang L, Calcedo R, Johnston J, Wilson JM. Novel adenoassociated viruses from rhesus monkeys as vectors for human gene therapy. Proc
Natl Acad Sci USA 2002; 99: 11854--11859.
46 Reed SE, Staley EM, Mayginnes JP, Pintel DJ, Tullis GE. Transfection of mammalian
cells using linear polyethylenimine is a simple and effective means of producing
recombinant adeno-associated virus vectors. J Virol Methods 2006; 138: 85--98.
47 Hauswirth WW, Lewin AS, Zolotukhin S, Muzyczka N. Production and purification
of recombinant adeno-associated virus. Methods Enzymol 2000; 316: 743--761.
48 Samulski RJ, Chang LS, Shenk T. A recombinant plasmid from which an infectious
adeno-associated virus genome can be excised in vitro and its use to study viral
replication. J Virol 1987; 61: 3096--3101.
49 Veldwijk MR, Topaly J, Laufs S, Hengge UR, Wenz F, Zeller WJ et al. Development
and optimization of a real-time quantitative PCR-based method for the titration of
AAV-2 vector stocks. Mol Ther 2002; 6: 272--278.
50 Grimm D, Kern A, Rittner K, Kleinschmidt JA. Novel tools for production and
purification of recombinant adenoassociated virus vectors. Hum Gene Ther 1998;
9: 2745--2760.
Supplementary Information accompanies the paper on Gene Therapy website (http://www.nature.com/gt)
& 2012 Macmillan Publishers Limited
Gene Therapy (2012) 800 -- 809