Preferential Labeling of Inhibitory and Excitatory Cortical Neurons by Endogenous Tropism of Adeno-Associated Virus and Lentivirus Vectors

PCA

JEa

ib

Jc

vd

e

Anftboosdb(ttEcnsrVrdVstb

K

Rva

*8EA�ommscpc

Neuroscience 161 (2009) 441–450

0d

REFERENTIAL LABELING OF INHIBITORY AND EXCITATORYORTICAL NEURONS BY ENDOGENOUS TROPISM OF
DENO-ASSOCIATED VIRUS AND LENTIVIRUS VECTORS
l(nchtFotma

mpg“TuemtaoavneW2Vec2s

ppiwrfh

bttcp

. L. NATHANSON,a,b Y. YANAGAWA,c K. OBATAd AND. M. CALLAWAYa*

Systems Neurobiology Laboratories, Salk Institute for Biological Stud-es, 10010 North Torrey Pines Road, La Jolla, CA 92037, USA

Department of Bioengineering, University of California, San Diego, Laolla, CA, USA

Department of Genetic and Behavioural Neuroscience, Gunma Uni-ersity Graduate School of Medicine, Maebashi, Gunma, Japan

Neuronal Network Mechanisms Research Group, RIKEN Brain Sci-nce Institute, Saitama, Japan

bstract—Despite increasingly widespread use of recombi-ant adeno-associated virus (AAV) and lentiviral (LV) vectorsor transduction of neurons in a wide range of brain struc-ures and species, the diversity of cell types within a givenrain structure is rarely considered. For example, the abilityf a vector to transduce neurons within a brain structure isften assumed to indicate that all neuron types within thetructure are transduced. We have characterized the trans-uction of mouse somatosensory cortical neuron typesy recombinant AAV pseudotyped with serotype 1 capsid

rAAV2/1) and by recombinant lentivirus pseudotyped withhe vesicular stomatitis virus (VSV) glycoprotein. Both vec-ors used human synapsin (hSyn) promoter driving DsRed-xpress. We demonstrate that high titer rAAV2/1-hSyn effi-iently transduces both cortical excitatory and inhibitoryeuronal populations, but use of lower titers exposes atrong preference for transduction of cortical inhibitory neu-ons and layer 5 pyramidal neurons. In contrast, we find thatSV-G-LV-hSyn principally labels excitatory cortical neu-ons at the highest viral titer generated. These findingsemonstrate that endogenous tropism of rAAV2/1 andSV-G-LV can be used to obtain preferential gene expres-ion in mouse somatosensory cortical inhibitory and excita-ory neuron populations, respectively. © 2009 IBRO. Publishedy Elsevier Ltd. All rights reserved.

ey words: AAV, LV, tropism, neurons, cortex, mouse.

ecombinant adeno-associated viruses (AAVs) and lenti-iruses (LVs) hold promise as gene therapy vectors andre valuable experimental tools due to their apparent

Corresponding author. Tel: �1-858-453-4100�1158; fax: �1-858-546-526.-mail address: [email protected] (E. M. Callaway).bbreviations: AAV, adeno-associated virus; CAG, hybrid CMV/chicken-actin promoter; GAD67, glutamate decarboxylase 67; GFP, green flu-rescent protein; hSyn, human synapsin I promoter; LV, lentivirus;CAMK, mouse �-calcium/calmodulin-dependent protein kinase II pro-oter; NeuN, neuronal nuclei antibody; rAAV2/1, recombinant AAV with

erotype 2 backbone packaged with serotype 1 capsid; RFP, red fluores-ent protein; VSV, vesicular stomatitis virus; VSV-G-LV, lentivirus

eseudotyped with vesicular stomatitis virus glycoprotein; WPRE, wood-huck hepatitis virus posttranscriptional regulatory element.

306-4522/09 $ - see front matter © 2009 IBRO. Published by Elsevier Ltd. All rightoi:10.1016/j.neuroscience.2009.03.032

441

ow toxicities and stable long-term transgene expressionMcCown, 2005). These vectors are particularly useful in theervous system due to their ability to infect non-dividingells (Naldini et al., 1996; Miao et al., 2000). Such vectorsave opened up an extensive range of possibilities due toheir ability to cause expression of virtually any gene.urthermore, one of the chief advantages of genetic meth-ds is the ability to target gene expression to particular cellypes, for example within complex neuropil, which containsany distinct cell types with their axonal and dendriticrbors intimately intertwined.

Cell type specific gene expression can be achieved byany different approaches. The most successful ap-roaches to date have involved the generation of trans-enic mouse lines using bacterial artificial chromosome orknock-in” technologies (Hanks et al., 1995; Heintz, 2001).hese strategies take advantage of large stretches of reg-latory genomic DNA or endogenous genetic regulatorylements to generate expression of a transgene whichimics expression of an endogenous gene. Although

hese methods are extremely useful, transgenic methodsre not practical in humans or in most mammalian speciesther than rodents. Thus, it is desirable to also have thebility to generate cell type specific expression from viralectors. Using viral vectors, selectivity can be achieved byatural or engineered tropism (Bowles et al., 2003; Mullert al., 2003; Perabo et al., 2003; Rabinowitz et al., 2004;arrington et al., 2004; Choi et al., 2005; Maheshri et al.,

006; Perabo et al., 2006; Wu et al., 2006a; Li et al., 2008;an Vliet et al., 2008), or insertion of gene regulatorylements into the viral genome (Chen et al., 1999; Cuc-hiarini et al., 2003; Dittgen et al., 2004; Zheng and Baum,005; Hioki et al., 2007). However, these approaches aretill in their infancy and not well understood.

As viral vector technologies become increasingly so-histicated and as they are combined with other ap-roaches, such as cell type specific promoters, there is an

ncreasing level of complication involved in understandinghy a particular approach is or is not successful. As a

esult it is important not only to understand the individualactors that influence cell type specific expression, but alsoow they interact.

Despite the potential for variable tropism observedetween viral serotypes and the likely dependence on viraliter, there have been few careful studies of the cell typeshat are transduced within a given brain area. And studiesarefully examining the relationships between viral tro-ism, titer, and cell type specific regulatory elements are
ven more rare or non-existent. Some studies have de-
s reserved.

mailto:[email protected]

stgeeeiceAp

V

TtMc(62otst

V

PCS(p((vpurfrmIglMg(wcpca

V

VAwGviwwb

Cctttpmppteciapncsohtitbmvarca

V

GmaAtsgcakietGGGtiGnncacmar0dapfii

J. L. Nathanson et al. / Neuroscience 161 (2009) 441–450442

cribed the ability to selectively transduce a particular cellype when using a putative cell type specific promoter in aiven vector. However, without direct comparisons of genexpression patterns observed between specific and gen-ral promoters, it is not possible to determine whetherxpression in the targeted cell type resulted from selectiv-

ty conferred by the promoter versus viral tropism, or aombination of both. Here we describe the transductionfficiencies of rAAV2/1 (AAV2 backbone packaged withAV1 capsid) and vesicular stomatitis virus (VSV)–G-seudotyped LV in the adult mouse somatosensory cortex.

EXPERIMENTAL PROCEDURES

irus promoters

he human synapsin I promoter (hSyn) (Kugler et al., 2003a), washe 469 bp human sequence chrX: 47,364,154–47,364,622 (UCSCarch 2006 assembly) (Kent et al., 2002). The mouse �-calcium/

almodulin-dependent protein kinase II promoter (mCAMK) (CKa13)Dittgen et al., 2004) was the 1289 bp mouse sequence chr18:1,084,084–61,085,372 (UCSC July 2007 assembly) (Kent et al.,002) cloned from pLenti-CaMKIIa-hChR2-EYFP-WPRE, courtesyf K. Deisseroth (Stanford, CA, USA). The hybrid CMV/chicken �-ac-

in promoter (CAG) promoter (Niwa et al., 1991), was the �1700 bpequence cloned from pCAG–green fluorescent protein (GFP), cour-esy of D. D. O’Leary (La Jolla, CA, USA).

irus production

romoters were cloned into either AAV or LV transfer vectors.omponents of AAV include ITR-AAV2 inverted terminal repeat,D/SA-splice donor/accepter sequence (human beta globin)

Kaspar et al., 2002), red fluorescent protein (RFP) DsRed-Ex-ress (Mikkelsen et al., 2003), and bovine growth hormone polyA) signal. Components of LV (modified version of pCSC-SP-PWMarr et al., 2004) include long terminal repeats, Psi-element foriral genome packaging, Rev response element, cPPT-centralolypurine tract, woodchuck hepatitis virus posttranscriptional reg-latory element (WPRE), upstream border of the 3= long terminalepeat polypurine tract. Plasmids were purified using endotoxin-ree maxiprep kits (QIAGEN, Valencia, CA, USA). RecombinantAAV2/1 serotype was produced by BBS/CaCl2 or polyethylenei-ine (PEI) mediated co-transfection of 293T cells with the AAV2

TR flanked transfer vector, pXR1 (AAV type 1) and pXX6-80 (Ad5enome) (Rabinowitz et al., 2002). Cells were harvested and

ysed, treated with benzonase (Sigma-Aldrich D9542, St. Louis,O, USA), and virus was concentrated and purified by iodixanolradient centrifugation and harvested in the 40% iodixanol bandZolotukhin et al., 1999). Virus dilutions were in 40% iodixanol. LVas produced by BBS/CaCl2 mediated co-transfection of 293Tells with the hSyn transfer vector and plasmids pMDL, pRev, andVSVG (Naldini et al., 1996). Medium was collected and virusoncentrated through multiple centrifugation steps (Tiscornia etl., 2006).

irus titration

iral genomes were quantified using a qPCR cycler (ABI 7900HT,pplied Biosystems, Foster City, CA, USA) and SYBR Greenith DsRed-Express specific primers 5=-AGGACGTCATCAAG-AGTTC and 5=-TCTGGGTGCCCTCGTAG and serially dilutedirus (see Table 1). AAV genomic DNA was isolated by lysing AAVn 2 M NaOH for 30 min at 56 °C, and neutralizing with HCl. LVas incubated with DNase I (NEB, Ipswich, MA, USA), and RNAas isolated (QIAamp MinElute virus Spin kit, QIAGEN), followed
y reverse transcription (Superscript III kit, Invitrogen, Carlsbad, v
A, USA). Viral DNA was diluted and compared to a standardurve created from a known quantity of transfer vector. Infectiousiters were determined by infecting confluent 293T cells with sixen-fold dilutions of virus, followed 60 h later by two PBS washes,rypsinization, centrifugation to isolate cells, DNA isolation byroteinase K digestion in SNET lysis buffer (20 mM Tris (pH 8), 1M EDTA, 1% SDS, 0.4 M NaCl) at 55 °C for 12 h, followed byhenol/chloroform extraction and alcohol precipitation. qPCR waserformed in the same manner as used to determine the genomiciter. Since isolation of DNA from the infected cells was not 100%fficient and differed between samples, infectious titers were cal-ulated by normalizing the qPCR titer by the efficiency of DNAsolation. Since nearly all isolated nucleic acid was cellular in originnd not viral, and assuming cell numbers were equal between sam-les, DNA isolation efficiency was estimated by the concentration ofucleic acid as determined by spectrophotometry. The isolation effi-iency of cellular nucleic acid and viral DNA was assumed to be theame. Individual normalization factors were calculated as the ratiosf the concentrations of precipitated nucleic acid divided by theighest concentration value of all precipitated nucleic acid prepara-ions. Thus, this normalization method presumably underestimatesnfectious titer by the amount of our most efficient nucleic acid isola-ion procedure. Average normalization ratios for rAAV2/1-hSyn-RFPatch 1, rAAV2/1-hSyn-RFP batch 2, rAAV2/1-CAG-RFP, rAAV2/1-CAMK-RFP, and lentivirus pseudotyped with vesicular stomatitis

irus glycoprotein (VSV-G-LV)–hSyn-RFP were 2.2, 1.5, 1.7, 1.4nd 2.0, respectively. Even though LV integrates and AAV largelyemains episomal, both forms of DNA can be isolated via this pro-edure using SNET buffer lysis, phenol/chloroform extraction andlcohol precipitation.

irus injections

lutamate decarboxylase 67 (GAD67)–GFP (�neo) mice (Tama-aki et al., 2003) were injected with viruses following procedurespproved by the Salk Institute Animal Care and Use Committee.ll experiments conformed to named international guidelines on

he ethical use of animals. The number of mice used and theiruffering were minimized to our abilities. In this study these trans-enic mice are referred to as GAD67-GFP knock-in mice. Be-ause we could not obtain robust and consistent GAD or GABAntibody staining in mice, we used GFP expression in GAD67nock-in mice, produced by Tamamaki et al. (2003), as an inhib-tory neuron marker for co-label studies. Produced by insertingnhanced GFP cDNA into the ATG translation initiation codon ofhe GAD67 locus, GAD67-GFP knock-in mice express enhancedFP in GABAergic neurons under the control of the endogenousAD67 gene promoter. Tamamaki et al. found 80%–90% of theFP-labeled cells were positive for GAD67 or GABA immunoreac-

ivity in the somata, and almost all labeled cells were immunoreactivef the neuropil was also included. Conversely, basically all of theAD67-immunoreactive cells labeled with GFP in the perikarya oreuropil. These animals were reported to exhibit normal growth,ormal behavior and no abnormality at the macroscopic level. GABAontent was significantly lower than in wild-type mice at birth, but nott the 6–7 week-old stage. GAD67-GFP knock-in mice produced byrossing GAD67-GFP knock-in mice with wild-type C57BL/6 or ICRice were injected between 10 and 30 weeks old. Animals werenesthetized using a cocktail of ketamine and xylazine, and/or isoflu-ane. Virus was stereotaxically injected into somatosensory cortex,.5 mm posterior to bregma, 3–3.5 mm lateral to the midline, atepths of 0.75 and 0.45 �m from the surface. Virus was delivered viaglass micropipette (�30–50 �m tip diameter) using air pressure

ulses applied via a Picospritzer II (General Valve Corporation, Fair-eld, NJ, USA) at 20 psi, with one pulse per second. Virus wasnjected for 5 min per depth, which usually corresponded to 1–2 �l of
irus. The pulse duration was adjusted to modulate the flow rate.

H

TPBimbTMbsI51twd(

I

Tliiahttevtsdita(cdytt1mha

dTwp

Itpsw

T(vaenDfifsjimppbwii2wsrstlbcfic

dis(t

T

V

rrrrrrV

a

b

c s/ml).

J. L. Nathanson et al. / Neuroscience 161 (2009) 441–450 443

istology and immunohistochemistry

wo to 3 weeks following injection, animals were perfused withBS followed by 4% paraformaldehyde in phosphate buffer.rains were post-fixed in 4% paraformaldehyde for 16 h and sunk

n 30% sucrose in PBS; 25–35�m thick sections cut on a freezingicrotome were stained with one or more of the following anti-odies: chicken anti-GFP (Aves Laboratories GFP-1020 1:500,igard, OR, USA), rabbit anti-DsRed (Clontech 632496 1:200,ountain View, CA, USA), and mouse anti–neuronal nuclei anti-ody (NeuN) (Millipore MAB377 1:100, Billerica, MA, USA). Asecondary antibodies we used one or more of Cy2 (Jacksonmmuno 703-225-155 1:100, West Grove, PA, USA), AlexaFluor68 (Invitrogen A11036 1:100), and Cy5 (Jackson Immuno 715-75-151 1:100) fluorophore-conjugated antibodies raised againsthe appropriate species. Mounted sections were counterstainedith 10 �M DAPI (Sigma-Aldrich D9542) in PBS. Sections wereehydrated through xylenes and coverslipped using KrystalonEMD Chemicals, Gibbstown, NJ, USA).

maging and cell-counting

he data described in the Results are based on observations ofabel from at least three, and typically four to five viral vectornjections, for each vector (see Table 2). In most cases a singlenjection was analyzed from each animal, however, in some cases

second injection was made and analyzed from the oppositeemisphere. In all cases, reporter gene expression was restrictedo the injected hemisphere, with no evidence of retrograde infec-ion via axon terminals. For the purpose of statistical analyses,ach injection was considered as an independent sample, as theariability between injections is presumed to be far greater thanhe variability between animals. For each injection analyzed, tis-ue sections were scanned to identify the region with the highestensity of labeling and this was considered the center of the

njection site. Neurons expressing reporter from the injected vec-or were counted in this section and up to eight adjacent sectionsnd scored for co-label with GFP (inhibitory neurons) and RFPvirus label) and/or NeuN. The total numbers of labeled cellsounted and scored varied between sections and layers due toifferences in labeling density for different vectors. Statistical anal-sis was performed using two-tailed, unequal variance Welch’s-tests. Typically, about 700 neurons were counted for each injec-ion (range: average of �1300 neurons per injection for AAV-hSyn:5 to �350 neurons per injection for AAV-CAG) and a total ofore than 1400 neurons for each vector. An exception was AAV-Syn diluted 125-fold, for which relatively few cells were labelednd only 543 cells were scored from a single injection analyzed.

To count and score labeled cells, fluorophores were indepen-ently imaged at 10� or 20� magnification using either a NikonE300 with a Bio-Rad radiance 2100 system or an Olympus BX51ith a Bio-Rad Radiance 2100MP system at three confocal

able 1. Genomic and infectious titers

irus Genomic titera Re

AAV2/1-hSyn-RFP (batch 1) 1:1 8.4�1012 10AAV2/1-hSyn-RFP (batch 1) 1:5 1.7�1012 2AAV2/1-hSyn-RFP (batch 1) 1:25 3.4�1011

AAV2/1-hSyn-RFP (batch 2) 1.2�1011

AAV2/1-CAG-RFP 5.8�1011

AAV2/1-mCAMK-RFP 2.6�1011

SV-G-LV-hSyn-RFP 3.5�1011

Determined using qPCR from lysed virus (genome copies/ml).Normalized to rAAV2/1-hSyn-RFP (batch 1) 1:1.Determined using qPCR from infected 293T cells (infectious particle

lanes. Images were projected and merged into RGB color space. w

ndependent RGB channels were linearly adjusted in Adobe Pho-oshop to optimize image brightness. Quantification of overlap oferikarya and/or neuropil cell labeling was done using the RGBeparated confocal images in Adobe Photoshop. Cortical layersere determined by DAPI or NeuN staining.

RESULTS

he studies described here focus on the ability of rAAV2/1AAV) and VSV-G-LV (LV) vectors to transduce inhibitoryersus excitatory mouse somatosensory cortical neurons,nd of the mCAMK promoter to preferentially drive genexpression in excitatory mouse somatosensory corticaleurons. All viral vectors used expression of the RFP,sRed-Express (Clontech) (Mikkelsen et al., 2003) (re-

erred to as RFP) as a reporter (Table 1). To facilitatedentification of GABAergic inhibitory neurons without con-ounds related to GAD or GABA antibody specificity andtaining reliability, all vectors were assessed following in-

ections into the somatosensory cortex of transgenic micen which GFP was inserted into the GAD67 locus (Tama-

aki et al., 2003). In these mice, almost all of the GFP-ositive cells showed GAD67 immunoreactivity in theerikarya or neuropil (Tamamaki et al., 2003). Conversely,asically all of the GAD67-immunoreactive cells labeledith GFP in the perikarya or neuropil. In the typical exper-

ment (see Experimental Procedures for details), corticalnjections of the viral vector were followed by a period of–3 weeks to allow for reporter gene expression. Animalsere then perfused and the brains sectioned and double-tained for anti-GFP and anti-RFP with appropriate fluo-escent secondary antibodies to amplify the green and redignals, respectively (Table 2). Sections were also coun-erstained with DAPI to allow for identification of corticalayers. In selected cases, anti-NeuN, a neuron specific anti-ody (Peterson et al., 1996), was used to allow for quantifi-ation of neurons versus glia. Labeled neurons were quanti-ed and scored as single, double, or triple stained usingonfocal microscopy (e.g. Fig. 1).

Transduction of cortical neurons was tested with threeifferent promoters driving RFP expression in AAV. These

ncluded the hybrid CAG (Niwa et al., 1991), the humanynapsin (hSyn) (Kugler et al., 2003a), and the mCAMKDittgen et al., 2004) promoters. LV was tested using onlyhe hSyn promoter. Genomic and infectious titers (Table 1)

nomic titerb Infectious titerc Relative infectious titerb

8.1�1010 100.0%1.6�1010 20.0%3.2�109 4.0%1.1�1010 13.2%4.2�109 5.2%3.3�109 4.1%2.9�109 3.5%

lative ge

0.0%0.0%4.0%1.5%6.8%3.0%4.1%

ere determined using qPCR from lysed virus and infected

2f1t

A

TaoaaovtlnNedsNNdpllrsspi2nt3ltde(

tinttr

Pd

TvchrveAle

2.V

irus

expr

essi

onco

unts

inG

AD

67-G

FP

mic

e

us#

Of

anim

als

#O

fin

ject

-ions

#O

fto

tal

sect

ions

#C

ount

edce

llsS

umm

ary

L2/3

exc.

L2/3

GA

D67

�L4 ex

c.L4 G

AD

67�

L5 exc.

L5 GA

D67

�L6 ex

c.L6 G

AD

67�

Tot

al#

ofce

llsA

vgce

lls/in

jM

axce

lls/in

jM

ince

lls/in

j

V-h

Syn

1:1

34

1013

9828

679

814

193

523

049

589

4372

1093

1807

384

V-h

Syn

1:5

25

1514

6492

424

421

618

0368

770

035

363

9112

7826

0840

5V

-hS

yn1:

254

525

146

557

2137

243

291

199

247

2785

557

725

446

V-h

Syn

1:12

51

19

614

94

8798

184

213

543

543

543

543

V-h

Syn

(bat

ch2)

44

2833

286

413

046

598

179

096

150

3808

952

1413

481

V-C

AG

47

1818

850

032

249

831

431

107

150

2488

355

580

229

V-m

CA

MK

33

1748

612

063

3153

310

796

2114

5748

675

710

8-h

Syn

55

2780

766

889

3783

671

562

4133

0966

272

756

2

henu

mbe

rof

anim

als,

inje

ctio

ns,s

ectio

ns,a

ndce

llsco

unte

dar

esh

own

for

som

atos

enso

ryco

rtic

alla

yers

2/3–

6.T

heav

erag

e,m

axim

uman

dm

inim

umnu

mbe

rsof

cells

coun

ted

are

show

nfo

rh

viru

s.In

hibi

tory

neur

ons

are

defin

edby

GA

D67

-GF

P�

expr

essi

on.

Exc

itato

ryne

uron

coun

tsar

ein

dica

ted

by“e

xc.”


93T cells, respectively (see Experimental Procedures forurther details). AAV-CAG, AAV-mCAMK, LV-hSyn and:25 diluted AAV-hSyn viruses were of similar infectiousiter.

AV biased expression towards inhibitory neurons

he CAG promoter in AAV drove expression in excitatorynd inhibitory neurons, and glia (Fig. 1d–f). Quantificationf neuron expression required staining for NeuN to excludebundant populations of glial cells (Fig. 2). The percent-ges of transduced neurons which were either excitatoryr inhibitory were quantified and compared to expectedalues as follows. The percentage of transduced neuronshat were inhibitory was expressed as the number of triple-abeled (GFP, RFP and NeuN) neurons divided by the totalumber of transduced neurons (double-labeled RFP andeuN). The remaining neurons were assumed to bexcitatory. Expected values (assuming random trans-uction) were estimated by quantifying NeuN expres-ion in GAD67-GFP mice. Cells double-labeled (GFP andeuN) were counted as inhibitory, and the remainingeuN cells as excitatory. Because neuronal labeling variedepending on cortical layers, labeled neurons and ex-ected values were quantified for each distinct cortical

ayer. The calculated expected inhibitory percentages forayers 2/3, 4, 5 and 6 were 15%, 10%, 21% and 12%,espectively. Quantification of expression from AAV-CAGhowed that in cortical layers 2/3, 4 and 6, reporter expres-ion was strongly biased towards inhibitory neurons com-ared to excitatory neurons (Fig. 3). The percentages of

nhibitory neurons that expressed RFP were 75% in layer/3, 88% in layer 4, and 60% in layer 6, which were five- toinefold higher than expected in these layers. In contrast,he percentage of inhibitory neurons in layer 5 was only6%, which is closer to the expected value of 21% in that

ayer. (Layer 1 neurons are only inhibitory and data areherefore not shown.) Welch’s t-test analysis indicated theistributions of AAV-CAG are significantly different than thexpected values in all layers, including layer 5 (P�0.004)Fig. 3).

These observations indicate that under the conditionsested (titer and promoter), rAAV2/1 can selectively labelnhibitory relative to excitatory neurons with a bias up toinefold relative to expected values. We hypothesized thathe higher proportions of inhibitory neurons observed withhe AAV-CAG vector described above were related to theelatively low titer of that virus (see Table 1).

roportion of excitatory and inhibitory neuron labelepended on AAV titer

o systematically and quantitatively test the role of AAViral titer, we used the hSyn promoter in order to avoidomplications related to the transduction of glial cells. TheSyn promoter has been shown to drive near perfect neu-on restrictive transgene expression in rat brain in adeno-irus (Ralph et al., 2000; Glover et al., 2002), AAV (Kuglert al., 2003b) and LV (Hioki et al., 2007). Our use ofAV-hSyn in mice also successfully eliminated glia ex-

pression (data not shown). As hypothesized, high titerTab

Vir

AA

AA

AA

AA

AA

AA

AA

LV

Tea

c

Ance

tncti

blwntpAAn

FiAeri(a n, o.


AV-hSyn (see titer in Table 1) labeled vast quantities ofeurons, including excitatory and inhibitory neurons in allortical layers at proportions indistinguishable from theirxpected distributions (Figs. 1j–l, 3 and 4).

Of the dilutions of AAV-hSyn tested, only the highestiter showed labeling that was not biased towards inhibitoryeurons. Twenty-five-fold dilution of this virus resulted in alear shift towards greater percentages of labeled inhibi-ory neurons compared to excitatory neurons in all layers,

ig. 1. Virus injections in GAD67-GFP mice. GAD67-GFP mice injecten somatosensory cortex corresponding to the AAV-CAG-RFP injectAV-CAG-RFP (d–f), AAV-hSyn-RFP diluted 1:25 (g–i), AAV-hSyn-xpression patterns through all cortical layers. Close-up images centeespectively. GAD67-GFP label is shown in green, and virus expressnhibitory neurons are shown by upward pointing arrows, excitatory neD) and medial (M) directions are indicated by the orientation grid. Oriapplies to a, d, g, j, m. Scale bar in b applies to b, c, e, f, h, i, k, l,

ncluding layer 5 (Figs. 1g–i, 3 and 4). At 1:25 dilution, the i

ias towards inhibitory neurons was most pronounced inayers 2/3 and 4, where 81% and 95% of labeled neuronsere inhibitory, respectively. The percentage of labeledeurons that were inhibitory was lowest in layer 5 (69%), but

his still represents a threefold higher percentage than ex-ected by chance and twice the percentage observed withAV-CAG. The expression difference between 1:25 dilutedAV-hSyn and AAV-CAG, which are similar in titer, is pro-ounced in layer 5 (P�0.001), suggesting the possibility of an

AV and LV. The left column (a–c), illustrates GAD67-GFP expressionn in the next column. Subsequent columns show virus expression:(j–l), and LV-hSyn-RFP (m–o). The first row (a, d, g, j, m) showsyers 2/3 and 5 are shown in rows 2 (b, e, h, k, n) and 3 (c, f, i, l, o),. Co-labeled cells are of varied yellow hue. Examples of co-labeleddownward pointing arrows, and nonneuronal cells by boxes. Dorsal

f rows 2 and 3 is as in b. Scale bars�50 �m in a and b. Scale bar in

d with Aion showRFP 1:1red on laion in redurons by

entation o

nfluence of the CAG or hSyn promoters in this layer.

iep(dVotnitnt

L

Rr

tnGnlhsil

wtlnhgtH

F(w l (M) dire

FiebP(L


Lower titer virus resulted in higher proportions of inhib-tory versus excitatory neuron label. A systematic differ-nce in expression was seen across all layers when com-aring between dilutions of 1:1, 1:5 and 1:25 of AAV-hSynFig. 4). Percentages of inhibitory neurons labeled at 1:5ilution were intermediate to the values at 1:1 and 1:25.ery few cells were labeled with the 1:125 dilution, andnly one animal was injected; but nevertheless a clear biasowards inhibitory neuron labeling was still apparent. Fi-ally, a different batch of the AAV-hSyn (batch 2) with an

nfectious titer equivalent to a 1:8 dilution of the highestiter AAV, resulted in a labeling bias towards inhibitoryeurons that was intermediate to the 1:5 and 1:25 dilu-ions.

V predominantly labeled excitatory neurons

esults with VSV-G-LV contrasted sharply with those fromAAV2/1. Using the same hSyn promoter, LV at the highest

ig. 2. NeuN labeling distinguishes neurons from glia in rAAV2/1-CAGc) NeuN labels neurons. All three panels are images of the same sectiith arrows. Layers 2/3, 4, and 5 are indicated. Dorsal (D) and media

ig. 3. Differences in promoter, titer and virus type affect excitatory ann dark grey and inhibitory neuron label in light grey. Distributions areach viral vector tested. The thick horizontal lines indicate percentagears indicate the standard deviation. Welch’s t-test statistics compar-values. Some error bars for expected values are smaller than the th

rAAV2/1 with CAG promoter), n�6. AAV-hSyn 1:1 (undiluted rAAV2/1 with humV-hSyn (VSV-G-pseudotyped LV with human synapsin promoter), n�5. AAV-

iter produced, predominantly labeled excitatory corticaleurons (Figs. 1m–o and 3). LV-hSyn only co-labeled withFP-labeled inhibitory neurons in 7%, 4%, 8% and 6% ofeurons in layers 2/3, 4, 5 and 6, respectively. In all cortical

ayers, the percentages of labeled neurons that were in-ibitory were lower than expected, by about half. t-Testtatistics indicate a difference from expected with P�0.04

n all layers, and lower in layers 2/3, 5 and 6 (P�0.013 forayer 2/3; P�0.011 for layer 5; P�0.013 for layer 6).

The observation that LV-hSyn expression is biased to-ards excitatory neurons further supports the conclusion

hat preferential transduction of inhibitory neurons withow titer AAV-hSyn is due to preferential transduction andot the result of a bias in transcription introduced by theSyn promoter. The contrast between vectors further sug-ests that VSV-G-LV might preferentially transduce exci-atory versus inhibitory cortical neurons at the titer tested.owever, an alternative possibility is suggested by the ob-

tions. (a) GAD67-GFP expression. (b) rAAV2/1-CAG-RFP expression.set of inhibitory neurons is marked with boxes and excitatory neuronsctions are indicated. Scale bar�50 �m.

ry neuron labeling. Percentages of excitatory neuron label are shownparately for mouse somatosensory cortical layers 2/3, 4, 5 and 6, for

bitory/excitatory neurons expected from unbiased transduction. Errorto unbiased transduction values are indicated by the correspondingf the lines. Sample sizes are the number of viral injections. AAV-CAG

-RFP secon. A sub

d inhibitoshown ses of inhi

ing virusickness o

an synapsin promoter), n�4. AAV-hSyn 1:25 (diluted 25-fold), n�5.mCAMK (rAAV2/1 with mouse �-CaMKII promoter), n�2–3.

socidinio

saLstottruiehce3dmptcwAf

DscvcaicT

Fvedo

FDcGrsdcD


ervation that in the case of some LV-hSyn injections, webserved fewer and less bright GFP expression at theenter of LV injections (Fig. 5). This suggested that the LVnjections could have resulted in selective inhibitory cellamage or changes in gene expression. However, dimin-

shed NeuN antibody label in this area suggested pan-euronal damage. Cells were not counted at the centers of

njection where GFP or NeuN labeling was abnormal. Thesebservations are discussed further below.

ig. 4. Dilution of rAAV2/1 vector shifts labeling towards inhibitory neuiral titer. Distributions of rAAV2/1-hSyn expression at 1:1 (diamond, nquivalent batch 2 version (x, n�4) are shown for somatosensoryistributions from unbiased neuron transduction. Standard deviations anly one viral vector injection.

ig. 5. Labeling is impaired at the center of some LV injections.ecreased GFP and NeuN labeling indicates neuron toxicity or de-reased gene expression at the center of some LV injections. GAD67-FP� cells are shown in green (a), LV-hSyn-RFP� cells are shown in

ed (b), NeuN� neurons are shown in blue (c), and DAPI� cells arehown in blue (d). The impaired region delineated in white, defined byecreased GFP or NeuN label, is not included in the double labeling

tounts. The center of the injection is marked by the arrow shown in d.orsal (D) and lateral (L) directions are indicated. Scale bar�50 �m.

A previous study by Dittgen et al. (2004) describedelective expression of reporter in excitatory neurons using1.3 kb mouse �-CaMKII (mCAMK) promoter in VSV-G-

V. Our observations with LV-hSyn suggest that the ob-erved bias might be due, at least in part, to selectiveransduction with LV (or reduction in the number or labelingf inhibitory neurons) rather than entirely due to transcrip-ional regulation by the mCAMK promoter. To further testhe ability of the mCAMK promoter to bias transcription andeporter gene expression towards excitatory neurons, wesed the same 1.3 kb mCAMK promoter in AAV. The

nfectious titer used was similar to the titer that biasedxpression towards inhibitory neurons when using theSyn and CAG promoters (see Table 1). Under theseonditions, the mCAMK promoter consistently drove genexpression in both inhibitory and excitatory neurons (Figs.and 6), indicating that in the AAV context this promoter

oes not restrict expression only to excitatory neurons inouse somatosensory cortex. Nevertheless, the mCAMKromoter appears to introduce a strong transcriptional biasowards excitatory neurons (Fig. 3). In particular, the per-entages of RFP expressing neurons that were inhibitoryere far lower with the AAV-mCAMK vector than with theAV-CAG or AAV-hSyn vectors at similar titers (P�0.05

or both in all layers).

DISCUSSION

etermining whether cell type specificity of gene expres-ion following injections of viral vectors into the brain isonferred by transcriptional regulation by the promoterersus viral tropism, or a combination of both, requires aareful assessment of each factor. Specificity arising frompromoter can only be determined in viruses capable of

nfecting a broad range of cell types. Likewise, viral tropisman only be determined using a general purpose promoter.hese two ideal situations are difficult to obtain, and in reality

citatory and inhibitory neuron expression is quantified as a function of(square, n�5), 1:25 (triangle, n�5), and 1:125 (circle, n�1), and 1:83, 4, 5 and 6. The thick horizontal bars indicate expected neuronted, except for the 1:125 dilution which consists of cells counted from

rons. Ex�4), 1:5layers 2/re indica

hese two variables are at best minimally confounded.

tsemeopLnYetid(pitseccattr

stidicrgtfpGicgwo

iaiHdRioieltngbwi

etdvhdiAomsst

ciiAietmfdso

Fcd


In the studies described here, we attempted to assesshe tropism of rAAV2/1 and VSV-G-LV in mouse somato-ensory cortex by using putative general neuronal promot-rs, CAG and hSyn. The results indicate that these pro-oters are capable of driving gene expression in bothxcitatory and inhibitory neurons, but that the proportionsf inhibitory/excitatory neurons expressing reporter de-end on the titer of AAV and also differ between AAV andV. High titer rAAV2/1-hSyn labels excitatory and inhibitoryeurons in proportions consistent with their constituency.et, a dilution of the virus results in a shift away fromxcitatory neurons. In contrast, LV-hSyn at the highest titerhat we could produce, fails to label large numbers ofnhibitory neurons. Because these different vectors labelifferent cell types even when using the same promoterhSyn), we attribute these differences to differential tro-ism of viral vectors (AAV vs. LV). The most conservative

nterpretation of these results is that rAAV2/1 preferentiallyransduces inhibitory relative to excitatory mouse somato-ensory cortical neurons, while VSV-G-pseudotyped LVfficiently transduces excitatory neurons, and that hSynonveys no bias. Nevertheless, other extreme possibilitiesannot be entirely ruled out. For example, if hSyn conveysbias towards excitatory neurons, then there might poten-

ially be no bias conveyed by LV tropism. But if this werehe case, then the tropism of AAV towards inhibitory neu-ons must be even stronger than was observed.

Importantly, following LV injections we sometimes ob-erved fewer and less brightly labeled GAD67-GFP posi-ive cells near injection centers. This suggested that the LVnjections could have resulted in selective inhibitory cellamage or changes in gene expression. However, dimin-

shed NeuN antibody label suggests that these injectionsaused cell damage to both inhibitory and excitatory neu-ons. DAPI staining density appeared normal, suggestinglial cell encroachment to damaged regions. Enhanced byhe activity of WPRE in LV, cell death might have resultedrom exceedingly high levels of DsRed-Express (RFP) ex-ression. More likely however, since many low-expressingFP cells resided in regions without any RFP expression,

s that impurities in the LV preparation were the cause ofell death. In contrast to AAV purification with an iodixanolradient, LV was prepared by pelleting centrifugation,hich can concentrate impurities along with the virus. An-

ig. 6. mCAMK promoter drives expression in inhibitory neurons in rAells are shown in red (b), and co-labeled cells are shown in yellowirections are indicated. Scale bar�50 �m.

ther caveat is that impurities in the LV preparation could t

nfluence LV tropism. Regions with obvious neuron dam-ge or diminished gene expression, as assessed by dimin-

shed GFP or NeuN label, were not counted in the results.owever, because we assessed inhibitory and excitatoryistributions from GFP expression, the observed rates ofFP and GFP double-labeling could underestimate inhib-

tory neuron transduction in peripheral areas with lessbvious impairment. While the effect is apparent in inhib-

tory neurons, we could not directly assess LV-mediatedffects on gene expression in the excitatory neuron popu-

ation. If the effect is uniformly applicable to all neuronypes, our results indicating LV bias towards excitatoryeurons would be supported. On balance, the results sug-est that tropism of LV towards excitatory neurons is likely,ut these confounds prevent a definitive conclusion. Eitheray, caution is warranted when using LV to target cortical

nhibitory neurons.In addition to the differences in proportions of cell types

xpressing reporter with different vectors under transcrip-ional control from the same promoter, we also observedifferences when using different promoters with the sameector (low titer AAV). Results were similar for CAG andSyn promoters, except in cortical layer 5, where the 1:25iluted AAV-hSyn resulted in a far higher proportion of

nhibitory neurons expressing reporter (69%) than withAV-CAG (36%). Informal observations of the morphologyf the layer 5 excitatory neurons suggest that they areostly tall-tufted pyramids (Larsen and Callaway, 2006),

uggesting that the CAG promoter may generate a tran-criptional bias towards expression in this particular cellype.

The mCAMK promoter was able to generate a far morelear, although incomplete, bias towards gene expression

n excitatory cortical neurons, relative to results with hSynn AAV of similar titer (see Fig. 3, AAV-hSyn 1:25 vs.AV-mCAMK). Nevertheless, it is clear that this promoter

n AAV did not restrict gene expression exclusively toxcitatory neurons. The use of an AAV titer that biasesransduction towards inhibitory neurons makes this fact farore apparent than it would be with LV, which appears to

avor transduction (or survival) of excitatory neurons. In-eed, we used the same 1.3 kb mCAMK promoter de-cribed in LV by Dittgen et al. (2004); Dittgen et al. (2004)bserved that the mCAMK promoter restricted expression

AD67-GFP� cells are shown in green (a), rAAV2/1-mCAMK-RFP�s indicate co-labeled inhibitory neurons. Dorsal (D) and medial (M)

AV2/1. G(c). Arrow

o cortical pyramidal neurons, identified by a characteristic

aielbcrtteb

itaIvcmrovsr(iae2ep

tcmtubhcsmn

deivgratttdooAt

rmviceit

cmrmdtlrsbt

AiNta

B

C

C

C

D

G

G

H

H

H


pical dendrite and dendritic branching in layer 1, suggest-ng that the promoter was capable of generating a nearlyxclusive bias towards excitatory neurons. Given the like-

ihood that both AAV and LV have strong and oppositeiases towards transduction of inhibitory versus excitatoryortical neurons, the present results do not allow an accu-ate quantification of the extent of the transcriptional biashat is likely introduced by the mCAMK promoter. Never-heless, our results suggest that some of the bias towardsxcitatory neurons observed by Dittgen et al. (2004) coulde due to LV tropism.

Tropism of rAAV2/1 can be due to both extra- andntra-cellular factors including receptor-mediated viral en-ry, intracellular trafficking to the nucleus, virion uncoating,nd conversion from single- to double-stranded DNA.dentifying any of these specific factors responsible forirus tropism may be difficult to determine due to theomplexities involved in transduction, but surface receptorolecules can provide specific entry pathways. Sialic acid

esidues are abundant on the surface of cells, and variantsf these residues are primary receptor sites for manyiruses (Olofsson and Bergstrom, 2005). �-2,3 And �-2,6ialic acids present on N-linked glycoproteins are primaryeceptors for both AAV1 and AAV6 cellular transductionWu et al., 2006b). However, additional receptors mediat-ng cell-entry and/or differences in post-entry mechanismsre probable, since AAV6 has been shown to be morefficient than AAV1 for liver transduction (Grimm et al.,003). The abundances of �-2,3 and �-2,6 sialic acids onxcitatory and inhibitory neurons are not known, and mayrovide an explanation for the bias we observe.

Another explanation for differences in AAV and LVropism is the compositions of the AAV and LV transgeneonstructs. The AAV ITR and LV LTR have inherent pro-oter activity, and may themselves contribute to biased

ransgene expression. In addition, our AAV constructssed a splice donor/accepter sequence from human �-glo-in (SD/SA) (Kaspar et al., 2002) and a bovine growthormone poly (A) signal, elements not found in the LVonstruct. Likewise, our LV construct contains a WPREequence not present in the AAV constructs. These ele-ents might unequally affect transgene expression amongeuron types.

Most viral tropism studies do not characterize trans-uction beyond tissue regions (brain, heart, etc.) and gen-ral cell type classifications (neurons, glia, etc.). The brain

s a diverse cellular environment, and characterization of airus or promoter must take into account cell type hetero-eneities. Our results demonstrate tropism between neu-ons in mouse somatosensory cortex, but other brain areasnd other species may not show the same differentialropism. We are not aware of previous reports on theransduction efficiencies of rAAVs in cortical neuron sub-ypes, but Haberman et al. (2002) have illustrated howiffering transgene expression in inhibitory and primaryutput neurons of rat inferior collicular cortex can alter theutcome of receptor-based gene therapy. When usingAV to deliver antisense cDNA for the NMDA receptor 1,

hey found that reduction in seizure sensitivity differed as a

esult of the specificity of the promoter used. The Haber-an et al. (2002) results illustrate the need to characterize

iral activity at more detailed levels to determine the abil-ties of viruses and promoters to function in the necessaryell types. In other cases too, greater cell type specificxpression of therapeutic agents will minimize potential

nteractions posed by undesired expression in other cellypes.

An imbalance of excitatory and inhibitory activity inortex is likely a cause of many epileptic events (Haber-an et al., 2003; Richichi et al., 2004). Since lower titer

AAV2/1 biases expression towards inhibitory neurons inouse somatosensory cortex, targeted expression pre-ominantly to these inhibitory neurons may not require cellype specific promoters. Also, since rAAV2/1 efficientlyabels inhibitory neurons at low titers, we reason thatAAV2/1 may be a suitable vector for testing promoters forubtypes of cortical inhibitory neurons. In addition, LV maye well suited for determining if an inhibitory neuron cellype promoter is also active in excitatory neurons.

cknowledgments—We thank Timothy Liu for his help with virusnjections and tissue processing. This work was supported by theIH Neuroplasticity of Aging Training Grant AG000216 (awarded

o UCSD), Chapman Charitable Trust, Aginsky Research Schol-rship, and NIH grants DA011828 and MH063912 to E.M.C.

REFERENCES

owles DE, Rabinowitz JE, Samulski RJ (2003) Marker rescue ofadeno-associated virus (AAV) capsid mutants: a novel approachfor chimeric AAV production. J Virol 77:423–432.

hen H, McCarty DM, Bruce AT, Suzuki K (1999) Oligodendrocyte-specific gene expression in mouse brain: use of a myelin-formingcell type-specific promoter in an adeno-associated virus. J Neuro-sci Res 55:504–513.

hoi VW, McCarty DM, Samulski RJ (2005) AAV hybrid serotypes:improved vectors for gene delivery. Curr Gene Ther 5:299–310.

ucchiarini M, Ren XL, Perides G, Terwilliger EF (2003) Selectivegene expression in brain microglia mediated via adeno-associatedvirus type 2 and type 5 vectors. Gene Ther 10:657–667.

ittgen T, Nimmerjahn A, Komai S, Licznerski P, Waters J, MargrieTW, Helmchen F, Denk W, Brecht M, Osten P (2004) Lentivirus-based genetic manipulations of cortical neurons and their opticaland electrophysiological monitoring in vivo. Proc Natl Acad SciU S A 101:18206–18211.

lover CP, Bienemann AS, Heywood DJ, Cosgrave AS, Uney JB(2002) Adenoviral-mediated, high-level, cell-specific transgene ex-pression: a SYN1-WPRE cassette mediates increased transgeneexpression with no loss of neuron specificity. Mol Ther 5:509–516.

rimm D, Zhou S, Nakai H, Thomas CE, Storm TA, Fuess S, Matsus-hita T, Allen J, Surosky R, Lochrie M, Meuse L, McClelland A,Colosi P, Kay MA (2003) Preclinical in vivo evaluation ofpseudotyped adeno-associated virus vectors for liver gene ther-apy. Blood 102:2412–2419.

aberman R, Criswell H, Snowdy S, Ming Z, Breese G, Samulski R,McCown T (2002) Therapeutic liabilities of in vivo viral vectortropism: adeno-associated virus vectors, NMDAR1 antisense, andfocal seizure sensitivity. Mol Ther 6:495–500.

aberman RP, Samulski RJ, McCown TJ (2003) Attenuation of sei-zures and neuronal death by adeno-associated virus vector gala-nin expression and secretion. Nat Med 9:1076–1080.

anks M, Wurst W, Anson-Cartwright L, Auerbach AB, Joyner AL(1995) Rescue of the En-1 mutant phenotype by replacement of
En-1 with En-2. Science 269:679–682.

H

H

K

K

K

K

L

L

M

M

M

M

M

M

N

N

O

P

P

P

R

R

R

R

T

T

V

W

W

W

Z

Z


eintz N (2001) BAC to the future: the use of bac transgenic mice forneuroscience research. Nat Rev Neurosci 2:861–870.

ioki H, Kameda H, Nakamura H, Okunomiya T, Ohira K, NakamuraK, Kuroda M, Furuta T, Kaneko T (2007) Efficient gene transduc-tion of neurons by lentivirus with enhanced neuron-specific pro-moters. Gene Ther 14:872–882.

aspar BK, Vissel B, Bengoechea T, Crone S, Randolph-Moore L,Muller R, Brandon EP, Schaffer D, Verma IM, Lee KF, HeinemannSF, Gage FH (2002) Adeno-associated virus effectively mediatesconditional gene modification in the brain. Proc Natl Acad Sci U S A99:2320–2325.

ent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, Zahler AM,Haussler D (2002) The human genome browser at UCSC. Ge-nome Res 12:996–1006.

ugler S, Kilic E, Bahr M (2003a) Human synapsin 1 gene promoterconfers highly neuron-specific long-term transgene expressionfrom an adenoviral vector in the adult rat brain depending on thetransduced area. Gene Ther 10:337–347.

ugler S, Lingor P, Scholl U, Zolotukhin S, Bahr M (2003b) Differentialtransgene expression in brain cells in vivo and in vitro from AAV-2vectors with small transcriptional control units. Virology 311:89–95.

arsen DD, Callaway EM (2006) Development of layer-specific axonalarborizations in mouse primary somatosensory cortex. J CompNeurol 494:398–414.

i W, Asokan A, Wu Z, Van Dyke T, Diprimio N, Johnson SJ, Govin-daswamy L, Agbandje-McKenna M, Leichtle S, Eugene RedmondD Jr, McCown TJ, Petermann KB, Sharpless NE, Samulski RJ(2008) Engineering and selection of shuffled AAV genomes: a newstrategy for producing targeted biological nanoparticles. Mol Ther16:1252–1260.

aheshri N, Koerber JT, Kaspar BK, Schaffer DV (2006) Directedevolution of adeno-associated virus yields enhanced gene deliveryvectors. Nat Biotechnol 24:198–204.

arr RA, Guan H, Rockenstein E, Kindy M, Gage FH, Verma I,Masliah E, Hersh LB (2004) Neprilysin regulates amyloid betapeptide levels. J Mol Neurosci 22:5–11.

cCown TJ (2005) Adeno-associated virus (AAV) vectors in the CNS.Curr Gene Ther 5:333–338.

iao CH, Nakai H, Thompson AR, Storm TA, Chiu W, Snyder RO, KayMA (2000) Nonrandom transduction of recombinant adeno-asso-ciated virus vectors in mouse hepatocytes in vivo: cell cycling doesnot influence hepatocyte transduction. J Virol 74:3793–3803.

ikkelsen L, Sarrocco S, Lubeck M, Jensen DF (2003) Expression ofthe red fluorescent protein DsRed-express in filamentous ascomy-cete fungi. FEMS Microbiol Lett 223:135–139.

uller OJ, Kaul F, Weitzman MD, Pasqualini R, Arap W, KleinschmidtJA, Trepel M (2003) Random peptide libraries displayed on adeno-associated virus to select for targeted gene therapy vectors. NatBiotechnol 21:1040–1046.

aldini L, Blomer U, Gallay P, Ory D, Mulligan R, Gage FH, Verma IM,Trono D (1996) In vivo gene delivery and stable transduction ofnondividing cells by a lentiviral vector. Science 272:263–267.

iwa H, Yamamura K, Miyazaki J (1991) Efficient selection for high-expression transfectants with a novel eukaryotic vector. Gene
108:193–199.
lofsson S, Bergstrom T (2005) Glycoconjugate glycans as viral re-ceptors. Ann Med 37:154–172.

erabo L, Buning H, Kofler DM, Ried MU, Girod A, Wendtner CM,Enssle J, Hallek M (2003) In vitro selection of viral vectors withmodified tropism: the adeno-associated virus display. Mol Ther8:151–157.

erabo L, Endell J, King S, Lux K, Goldnau D, Hallek M, Buning H(2006) Combinatorial engineering of a gene therapy vector: di-rected evolution of adeno-associated virus. J Gene Med 8:155–162.

eterson DA, Lucidi-Phillipi CA, Murphy DP, Ray J, Gage FH (1996)Fibroblast growth factor-2 protects entorhinal layer II glutamatergicneurons from axotomy-induced death. J Neurosci 16:886–898.

abinowitz JE, Bowles DE, Faust SM, Ledford JG, Cunningham SE,Samulski RJ (2004) Cross-dressing the virion: the transcapsidationof adeno-associated virus serotypes functionally defines sub-groups. J Virol 78:4421–4432.

abinowitz JE, Rolling F, Li C, Conrath H, Xiao W, Xiao X, SamulskiRJ (2002) Cross-packaging of a single adeno-associated virus(AAV) type 2 vector genome into multiple AAV serotypes enablestransduction with broad specificity. J Virol 76:791–801.

alph GS, Bienemann A, Harding TC, Hopton M, Henley J, Uney JB(2000) Targeting of tetracycline-regulatable transgene expressionspecifically to neuronal and glial cell populations using adenoviralvectors. Neuroreport 11:2051–2055.

ichichi C, Lin EJ, Stefanin D, Colella D, Ravizza T, Grignaschi G,Veglianese P, Sperk G, During MJ, Vezzani A (2004) Anticonvul-sant and antiepileptogenic effects mediated by adeno-associatedvirus vector neuropeptide Y expression in the rat hippocampus.J Neurosci 24:3051–3059.

amamaki N, Yanagawa Y, Tomioka R, Miyazaki J, Obata K, KanekoT (2003) Green fluorescent protein expression and colocalizationwith calretinin, parvalbumin, and somatostatin in the GAD67-GFPknock-in mouse. J Comp Neurol 467:60–79.

iscornia G, Singer O, Verma IM (2006) Production and purification oflentiviral vectors. Nat Protoc 1:241–245.

an Vliet KM, Blouin V, Brument N, Agbandje-McKenna M, Snyder RO(2008) The role of the adeno-associated virus capsid in genetransfer. Methods Mol Biol 437:51–91.

arrington KH Jr, Gorbatyuk OS, Harrison JK, Opie SR, Zolotukhin S,Muzyczka N (2004) Adeno-associated virus type 2 VP2 capsidprotein is nonessential and can tolerate large peptide insertions atits N terminus. J Virol 78:6595–6609.

u Z, Asokan A, Samulski RJ (2006a) Adeno-associated virus sero-types: vector toolkit for human gene therapy. Mol Ther 14:316–327.

u Z, Miller E, Agbandje-McKenna M, Samulski RJ (2006b) Alpha2,3and alpha2,6 N-linked sialic acids facilitate efficient binding andtransduction by adeno-associated virus types 1 and 6. J Virol80:9093–9103.

heng C, Baum BJ (2005) Evaluation of viral and mammalian promot-ers for use in gene delivery to salivary glands. Mol Ther 12:528–536.

olotukhin S, Byrne BJ, Mason E, Zolotukhin I, Potter M, Chesnut K,Summerford C, Samulski RJ, Muzyczka N (1999) Recombinantadeno-associated virus purification using novel methods improves
infectious titer and yield. Gene Ther 6:973–985.
(Accepted 13 March 2009)(Available online 24 March 2009)

Documents

Preferential Labeling of Inhibitory and Excitatory Cortical Neurons by Endogenous Tropism of Adeno-Associated Virus and Lentivirus Vectors