Cell Host & Microbe

Article

Nature and Duration of Growth Factor Signalingthrough Receptor Tyrosine KinasesRegulates HSV-1 Latency in NeuronsVladimir Camarena,1,4 Mariko Kobayashi,2 Ju Youn Kim,2 Pamela Roehm,3 Rosalia Perez,1 James Gardner,3

Angus C. Wilson,2 Ian Mohr,2 and Moses V. Chao1,4,5,6,7,*1Molecular Neurobiology Program, Skirball Institute for Biomolecular Medicine2Department of Microbiology3Department of Otolaryngology4Department of Cell Biology5Department of Physiology and Neuroscience6Department of Psychiatry7Center for Neural Science

New York University School of Medicine, 540 First Avenue, New York, NY 10016, USA

*Correspondence: chao@saturn.med.nyu.edu

DOI 10.1016/j.chom.2010.09.007

SUMMARY

Herpes simplex virus-1 (HSV-1) establishes life-longlatency in peripheral neurons where productive repli-cation is suppressed. While periodic reactivationresults in virus production, the molecular basis ofneuronal latency remains incompletely understood.Using a primary neuronal culture model of HSV-1latency and reactivation, we show that continuoussignaling through the phosphatidylinositol 3-kinase(PI3-K) pathway triggered by nerve growth factor(NGF)-binding to the TrkA receptor tyrosine kinase(RTK) is instrumental in maintaining latent HSV-1.The PI3-K p110a catalytic subunit, but not the b ord isoforms, is specifically required to activate 3-phos-phoinositide-dependent protein kinase-1 (PDK1) andsustain latency.Disrupting thispathway leads tovirusreactivation. EGF and GDNF, two other growthfactors capable of activating PI3-K and PDK1 butthat differ fromNGF in their ability to persistently acti-vateAkt, donot fully supportHSV-1 latency. Thus, thenature of RTK signaling is a critical host parameterthat regulates the HSV-1 latent-lytic switch.

INTRODUCTION

The ability of herpes simplex virus to establish and maintain

a life-long latent infection in peripheral neurons is fundamental

to its survival and function as a human pathogen. Classically,

the latent state is defined as the absence of infectious virus

production despite the presence of episomal viral genomes in

neuronal nuclei. Expression of the more than 80 ORFs encoded

by HSV-1 is highly restricted in latently infected neurons (Knipe

and Cliffe, 2008). The exception is a latency-associated RNA

transcript (LAT) that accumulates to high levels in the neuronal

nucleus. Several functions have been proposed for LAT,

320 Cell Host & Microbe 8, 320–330, October 21, 2010 ª2010 Elsevie

including the ability to modulate the chromatin state of the viral

episome, inhibit apoptosis, and produce microRNAs that

suppress lytic gene expression (Bloom et al., 2010).

Periodically, the virus changes its relationship with the

neuronal host and reactivation from latency ensues, resulting in

the coordinate expression of lytic genes and production of infec-

tious virus that spreads back to the epithelium. A variety of

conditions can promote reactivation, including exposure to UV

light, stress, fever, anxiety, and nerve trauma (Cushing, 1905;

Glaser and Kiecolt-Glaser, 2005; Warren et al., 1940; Wheeler,

1975). While herpes reactivation following surgery on the

trigeminal ganglion was first reported over a century ago, the

mechanisms underlying latency and reactivation remain largely

unknown.

Experiments using animal model systems have been instru-

mental in understanding latency (Wagner and Bloom, 1997). In

addition to defining viral genes required for reactivation, these

systems have revealed important roles for components of both

innate and acquired immunity in modulating viral reactivation

(Knickelbein et al., 2008; Leib et al., 1989a, 1989b; Thompson

et al., 2009). At its core, however, latency involves a precisely

tuned interaction between the virus and host neuron. Conse-

quently, the intricate details of this relationship are difficult to

tease out in animal models due to the confounding influence of

nonneuronal cell types and the actions of immune defenses.

Instead, a detailed molecular understanding of HSV-1 latency

in neurons requires a cell-culture model that utilizes a homoge-

nous neuronal population that faithfully recapitulates the hall-

marks of latency and reactivation.

Sympathetic neurons can be cultured as a pure population of

cells that depend upon trophic support from nerve growth factor

(NGF) or glial-derived neurotrophic factor (GNDF) (Glebova and

Ginty, 2005). Indeed, latency can be established in primary

sympathetic neurons cultured in the presence of NGF (Wilcox

and Johnson, 1987, 1988; Wilcox et al., 1990). This agrees with

studies in latently infected rabbits showing that NGF withdrawal

can induce HSV-1 reactivation in sensory and sympathetic

neurons in vitro or after anti-NGF treatment in vivo (Hill et al.,

1997). Importantly, NGF stimulates a range of physiological

r Inc.

Cell Host & Microbe

PI3-Kinase Signaling Regulates HSV-1 Latency

responses in neurons including but not limited to differentiation,

survival, inflammation, regeneration, cell-cycle arrest, and cell

death by interacting with multiple cell surface receptors and trig-

gering at least five independent signaling pathways. Surprisingly,

since publication of the initial reports describing NGF-dependent

latency, the specific NGF-responsive receptors and signal-

transduction pathways required to maintain latency and prevent

reactivation have not been deciphered.

Here we have developed a simple, real-time readout for reac-

tivation in living neurons and employed small-molecule chemical

inhibitors along with gene-silencing techniques to determine the

signaling components that control HSV-1 latency. Significantly,

we find that a continuous neuronal signaling program mediated

by NGF through the TrkA receptor, PI3-kinase (PI3-K) p110a iso-

form, PDK1, and Akt is required to suppress HSV productive

(lytic) growth and maintain latency. Disrupting this signaling

pathway, even transiently, using selective small molecule inhib-

itors or shRNA-mediated gene silencing resulted in efficient

reactivation. Moreover, these studies reveal that the duration

of growth factor signaling to Akt is a critical parameter regulating

latency in neurons. Specific growth factors therefore have

different abilities to support latency and suppress lytic HSV-1

replication.

RESULTS

To define the cellular requirements to sustain HSV-1 latency in

neurons, we modified a primary neuronal cell culture model for

establishing HSV-1 latency in vitro (Wilcox and Johnson, 1987,

1988), such that reactivation can be monitored in real time.

Dissociated superior cervical ganglia (SCG) neurons from E21

rat embryos were cultured with 50 ng/ml NGF in the presence

of 5-fluorouracil and aphidicolin to remove nonneuronal cells.

SCG neurons isolated in this manner resulted in sufficiently

pure populations of neurons to enable a study of virus-neuron

interactions without interference from other cell types.

Once established, these neuronal cultures were subsequently

infected with HSV-1 (Figure 1A). An otherwise wild-type HSV-1

strain expressing GFP fused to the Us11 true-late (g2) protein

served as a reporter to follow the lytic phase of the viral life cycle

and allowed reactivation to be detected in living neurons

(Benboudjema et al., 2003). Replicate wells of virus-infected

neurons were treated with acyclovir (ACV) for up to 6 days to

suppress lytic HSV-1 replication. At this point, ACV can be

removed and the infected cultures maintained for weeks without

the production of infectious virus as detected by plaque assay

(Figure 1B). Likewise, there was no detectable expression of

mRNAencoding ICP27 (Figure 1C, lanes 2 and3), a critical imme-

diate-early (IE) regulator essential for productive replication,

indicating that the virus had entered a nonreplicating state. This

was reinforced by the accumulation of LAT transcripts, which

were readily detected by RT-PCR in SCG neurons (Figure 1C),

and reproducibly found in 20% of the neuronal nuclei (three

independent experiments, n = 1500) by in situ hybridization after

ACV removal (Figure 1D). Finally, accumulation of GFP-Us11,

a reporter gene expressed late in the productive growth cycle,

was also not detected (Figure 1E, top panels). The absence of

(1) detectable infectious virus production, (2) detectable produc-

tive lytic cycle gene expression, and (3) the concurrent accumu-

Cell Hos

lation of nuclear LATs are accepted hallmarks of latency in

neurons (Knipe and Cliffe, 2008; Bloom et al., 2010).

Depletion of NGF using an anti-NGF antibody resulted in

productive viral replication (reactivation), evident from the

production of infectious virus measured 6 days after adding

anti-NGF (Figure 1B), the selective accumulation of ICP27

mRNA in GFP-positive cultures (Figure 1C), and late GFP-Us11

reporter expression, which was readily detected after 1–2 days,

and steadily increased up until day 6 (Figures 1E and 1F). LATs

were detected in all cultures even during productive viral growth

(Figure 1C), consistent with studies showing that LAT expression

is not limited to latently infected cells (Devi-Rao et al., 1991) (data

not shown). Importantly, GFP-US11 accumulation was routinely

observed in approximately 10%–20% of wells in each experi-

ment, representing a baseline level of spontaneous reactivation.

Taken together, these results indicate that NGF depletion repro-

ducibly activated expression of viral productive cycle genes in

latently infected neurons and thereby verified the reported

requirement for NGF to suppress productive replication and

maintain latency in cultured sensory neurons (Smith et al.,

1994). Activation of productive cycle lytic genes in latently in-

fected neurons, culminating in the release of infectious virus, is

the hallmark of HSV-1 reactivation from latency.

NGF withdrawal results in apoptosis of SCG neurons (Deck-

werth and Johnson, 1993), and it is conceivable that HSV-1

reactivation occurs through activation of a cell-death pathway.

To address this, a pan-caspase inhibitor, Z-VAD-fmk, was

added to the cultures prior to NGFwithdrawal. While the inhibitor

effectively prevented cell death in response to NGF depletion

under these conditions (Figure 2A), latently infected SCGs

reactivated to equivalent levels (Figure 2B). In the absence of

Z-VAD-fmk, GFP-positive cells induced by NGF withdrawal

displayed intact nuclei by Hoechst staining (data not shown).

Thus, caspase-dependent apoptosis per se was not required

for viral reactivation induced by NGFdeprivation.

Latency Requires NGF-TrkA SignalingNext we began to explore the mechanism(s) by which NGF sup-

pressed lytic replication and maintained latency. NGF interacts

with two receptors, the TrkA receptor tyrosine kinase (RTK) and

the p75 neurotrophin receptor (Figure 2C). The earlier in vitro

studies were carried out prior to the identification of TrkA as an

NGF receptor (Kaplan et al., 1991; Wilcox and Johnson, 1988;

Wilcox et al., 1990) and before the multiple NGF signaling path-

ways were defined; consequently, little information is available

on the role of individual NGF receptors in controlling HSV-1

latency. A large body of work has established that NGF signaling

through the Trk and p75 receptors is remarkably complex and

capable of triggering at least five major signaling pathways that

orchestrate diverse physiological responses (Chao, 2003).

To address the receptor requirements for NGF-dependent

latency, infected SCG cultures were treated with the pharmaco-

logical agent K252a (Figure 2D), at a concentration (200 nM) that

selectively blocks Trk receptors (Figure S1), but not other RTKs

(Berg et al., 1992). Addition of K252a resulted in reactivation

levels and kinetics similar to those observed upon NGF depletion

using anti-NGF antibody. To test whether the p75 receptor

participates in HSV-1 reactivation, cells were treated with the

anti-p75 antibody (9651), which blocks NGF binding to the

t & Microbe 8, 320–330, October 21, 2010 ª2010 Elsevier Inc. 321

A

PhaseE F

Days post-treatment

6 days

Dissect & plate SCG neurons + NGF

Infect with GFP-HSV1 + ACV

ACV removal & GFP reporter screen

6 days

anti-NGF

NGF anti-NGF

control antibodies

Us11- GFP

D

LAT Hoechst Merge NF200

20

80

40

60

1 2 3 4 5 60

6 days

10

30

50

70

% G

FP+

wel

ls

C

GFP+

LAT

ICP27

GAPDH

HSV + + + +_

1 2 3 4 5

B

anti-NGF

PFU

/Wel

l

controlantibody

1

101

210

310

410

510

GFP-

Figure 1. Establishment of a Nonreplicating

HSV-1 Infection in Cultured SCG Neurons

(A) Protocol for establishment of HSV-1 infections

in SCG cultures. Dissociated SCG were seeded

in 96-well plates in media supplemented with

50 ng/ml NGF. After 6 days, cells were treated

with 50 mM acyclovir (ACV), and 1 day later in-

fected with HSV-1 EGFP-Us11 (MOI = 1 based

upon titer on Vero cells). Infected cultures were

maintained with ACV for an additional week to

allow the virus to establish a nonreplicating

infection.

(B) Plaque assays were performed after 6 days of

treatment with blocking antibodies against NGF

or BDNF (control). Twenty individual wells were

assayed for each condition. Bar shows the average

number of plaque forming units (PFU) per well.

(C) Detection of HSV-1 transcripts by RT-PCR.

After establishing a nonreplicating HSV-1 infection

in the presence of NGF and ACV, the ACV was

removed and GFP fluorescence monitored over

a 7 day period. RNA was collected from two

GFP-negative samples (lanes 2 and 3) represent-

ing latently infected cultures and two GFP-positive

samples (lanes 4 and 5) representing the small

number of cultures that spontaneously reactivated

in the absence of any treatment. Semiquantitative

RT-PCR was used to determine the expression of

LAT, ICP27, and cellular GAPDH RNA transcripts.

Lane 1, uninfected SCG culture.

(D) Accumulation of LAT in nuclei of latently in-

fected neurons. HSV-1-infections were estab-

lished as described in (A). After ACV removal,

neurons were fixed and probedwith a combination

of fluorescent in situ hybridization (FISH) and indi-

rect immunofluorescence. Viral LAT RNA tran-

scripts were detected by RNA FISH [green], and

nuclear DNA was visualized using Hoechst stain

(red). The merge of the RNA FISH and Hoechst

signals highlights the latently infected neurons

(yellow). Neurofilament protein NF200, a neuronal

cytoplasmic marker, was detected by indirect

immunofluorescence (white). Bar = 20 mm.

(E) Detection of reactivating virus by fluorescent microscopy. Phase contrast and fluorescent images of representative clusters of latently infected neurons.

Reactivation was induced by NGF deprivation, resulting in the expression of the HSV-1 EGFP-Us11 fusion protein as a true late (g2) gene product. The image

was captured on day 3 following NGF depletion. Bar = 20 mm.

(F) Reactivation assay showing the percentage of wells containing GFP-positive cells detected by fluorescent microscopy of living neurons over a 6 day period

after the removal of ACV and NGF. NGF was depleted using anti-NGF blocking antibodies and compared with anti-BDNF (control) antibodies. The error bars

indicate the standard error of the mean.

Cell Host & Microbe

PI3-Kinase Signaling Regulates HSV-1 Latency

receptor and consequently ablates downstream signaling

(Huber and Chao, 1995; Skoff and Adler, 2006). Reactivation

was not detected in latently infected SCGs treated with 9651

(Figure 2E). Taken together, these results indicate that reactiva-

tion of latent HSV-1 upon NGF depletion specifically involved

TrkA, but not p75. Moreover, the results suggest that signals

emanating from the NGF-bound TrkA receptor are required to

suppress lytic replication and maintain latency.

Requirement for PI3-K to Suppress Reactivationand Maintain LatencyBinding of NGF to the TrkA receptor can activate the mitogen-

activated protein (MAP) kinase pathway, phospholipase Cg,

and PI3-K (Figure 3A). To determine which of these pathways

were required to maintain latency, we first treated sympathetic

322 Cell Host & Microbe 8, 320–330, October 21, 2010 ª2010 Elsevie

cultures with a panel of well-characterized chemical inhibitors

that have been used previously to examine TrkA signaling in

sympathetic neurons (MacInnis and Campenot, 2002; Ye et al.,

2003). While PD98059 inhibited MAP kinase kinase (MEK), and

consequently blocked ERK activation in these neuronal cultures

(Figure S2A), reactivation was not detected compared to

cultures treated with the TrkA inhibitor K252a (Figure 3B). Impor-

tantly, inhibition of lytic replication by PD98059 was not

observed in acutely infected SCG neuron cultures, indicating

that ERK activity was not required for the productive cycle of

HSV-1 replication (Figures S2B–S2C). To determine whether

PI3-K signaling contributes to the maintenance of latency in

neurons, cultures were treated with the broad specificity PI3-K

inhibitor LY294002. Remarkably, while inhibiting ERK activation

did not induce reactivation, the PI3-K inhibitor LY294002

r Inc.

D

DMSO

K252aanti-NGF

E

anti-NGF

IgG controlanti-p75

1 2 3 4 5 60 1 3 4 5 60

% re

activ

ated

wel

ls

% re

activ

ated

wel

ls

Days post-treatment Days post-treatment

20

80

40

60

10

30

50

70

20

80

40

60

10

30

50

70

anti-

NG

F

NG

F

anti-

NG

F+

ZVAD

0

200

50

150

100

BA

Dea

d ce

lls

anti-NGF

IgG controlIgG + ZVAD

anti-NGF + ZVAD

20

80

40

60

10

1 2 3 4 5 60%

reac

tivat

ed w

ells

Days post-treatment

30

50

70

p75

NGF

TrkA

cell cycle arrest

neurite outgrowth

C

2

γplasticity

differentiationsurvival

survival growth

κinflamation

survival

Figure 2. Inhibition of TrkA Induces HSV-1

Reactivation

(A) Z-VAD-fmkwas effective in blocking apoptosis.

Cultured neurons were treated with anti-NGF anti-

body in the presence or absence of Z-VAD-fmk

(20 mM). After 48 hr, neurons were stained with

Hoechst 33342 dye and scored for nuclear frag-

mentation indicative of apoptosis. Bar indicates

mean (n = 200).

(B) Pan-caspase inhibitor Z-VAD-fmk does not

prevent the reactivation after NGF withdrawal.

GFP-HSV-1-infected SCG neuron cultures were

treated with anti-NGF or control antibodies in the

presence or absence of 20 mM Z-VAD-fmk. Plot

shows the number of GFP+ wells, indicative of

reactivated virus, scored over a 6 day period.

(C) Schematic showing signaling pathways acti-

vated by NGF. SCG neurons express two NGF

receptors, p75 or TrkA, and activate multiple path-

ways that give rise to different biological effects

(Chao, 2003).

(D) Reactivation assay comparing treatments with

the Trk inhibitor K252a (200 nM in DMSO), anti-

NGF antibody (positive control) or DMSO alone.

See also Figure S1.

(E) Reactivation assay comparing anti-p75 (9651)

blocking antibody with anti-NGF and IgG control

antibodies. The error bars indicate the standard

error of the mean.

Cell Host & Microbe

PI3-Kinase Signaling Regulates HSV-1 Latency

resulted in robust reactivation, with a greater fraction of wells

showing reactivation than with the TrkA inhibitor K252a

(Figure 3C). The effect of LY294002 was specific because

LY303511, a close structural analog of LY294002 that does not

inhibit PI3-K, did not result in detectable HSV-1 reactivation

(Figure 3C). The ability of LY294002 to block PI3-K signaling

was readily demonstrated by monitoring phosphorylation of

a downstream target (Figure S2D).

Lytic Gene Transcription Occurs within Hoursof Inhibiting the PI3-K PathwayAlthough Us11-GFP fluorescent protein provides a convenient

real-time marker for HSV-1 reactivation, it relies on the accumu-

lation of sufficient protein quantities for detection by live fluores-

cent imaging. This likely contributes to the gradual increase in

positive wells in the time courses. As an alternative, we prepared

RNA from infected cultures (in duplicate) collected 20 hr after

exposure to LY294002 and performed RT-PCR to detect repre-

sentative IE (ICP27), early (UL30 and UL5) lytic transcripts

(Figure 3D). As expected, LAT RNA was readily detected before

Cell Host & Microbe 8, 320–330,

and after LY294002 treatment, whereas

the lytic genes were only detected after

addition of the inducer.

To evaluate the number of neurons

undergoing independent reactivation

events we pretreated cultures with WAY-

150138, a compound that specifically

blocks viral spread by preventing

encapsidation of the viral DNA genome

(Newcomb and Brown, 2002; Pesola

et al., 2005; van Zeijl et al., 2000). Infected

sympathetic neuron cultures were treatedwithWAY-150138 and

reactivation induced with LY294002. Small but significant

numbers of GFP-positive neurons could be detected in 70% of

wells indicating that a number of independent reactivation

events occur per individual culture (Figure 3E). Presumably

some or all of these reactivation events give rise to infectious

virus that spreads to neighboring cells. The effectiveness of the

compound in preventing the spread of virus in cultured SCG

neurons was addressed by performing a lytic infection at

a MOI of 0.1 and by visualizing the infected neurons by fluores-

cence microscopy (Figure 3F). After 72 hr, the majority of

neurons expressed GFP, but in the presence of WAY-150138

only the cluster of neurons that were initially infected were GFP

positive. This provides a basis for scoring the number of

GFP-positive wells rather than individual cells.

Subunit Specific PI3-K Signaling SuppressesHSV-1 ReactivationThe PI3-K holoenzyme comprises an 85-KDa regulatory subunit

partnered with one of three catalytic subunits (p110 a, b, and d),

October 21, 2010 ª2010 Elsevier Inc. 323

C

DMSO

K252a

LY294002

DMSO

K252a

PD98059

B

LY294002

DMSO

PIK-75

TGX115

IC87114

D

LY303511

LY303511

1 2 3 4 5 60

1 2 3 4 5 60

1 2 3 4 5 60

sllew detavitcaer

%

sllew detavitcaer

%

sllew detavitcaer

%

Days post-treatment

Days post-treatment

Days post-treatment

20

80

40

60

10

30

50

70

20

80

40

60

10

30

50

70

20

80

40

60

10

30

50

70

NGFTrkA

MEK1/2

ERK1/2

PLCγ PDK1

PI3-K

plasticitydifferentiation

survival

Latent HSV-1 genome

Cytoplasm

Nucleus

A

survival growth

-

LAT

ICP27

GAPDH

- + +LY294002:

UL30

UL5

DMSO LY294002(20 uM)

sllec + PFG fo .o

N

50

+ WAY-150138

100

150

200

E

DMSO

WAY-150138

>

phase GFPF G

••••••••••• •••• •••

Figure 3. PI3-K Is Required in Latency

(A) Schematic of divergent TrkA-mediated signaling pathways. Molecules that were tested and found to be either required (black), or not required (gray), for

maintenance of latency are highlighted. PI3-K, phosphatidylinositol 3-kinase; PDK1, phosphoinositide-dependent kinase 1; MEK, mitogen-activated protein

kinase (MAPK)/ERK kinase; PLCg, phospholipase Cg; Trk, tropomyosin-related kinase receptor, NGF, nerve growth factor.

(B) Reactivation assay comparing the response of GFP-HSV-1 infected SCG neurons treated with MEK inhibitor PD98059 (50 mM) to Trk inhibitor K252a (200 nM)

or vehicle control DMSO. See also Figures S2A–S2C.

(C) Reactivation of HSV-1 after 6 days of treatment with the PI3-K inhibitor LY294002 (10 mM), LY303511 (an inactive analog of LY294002; 10 mM), the Trk inhibitor

K252a (200 nM), or DMSO. See also Figure S2D.

(D) Detection of HSV-1 transcripts by RT-PCR. After establishing a nonreplicating HSV-1 infection in the presence of NGF and ACV, the ACVwas removed and the

cultures were treated with LY294002 or DMSO. The RNA was collected 20 hr later. Semiquantitative RT-PCR was used to determine the expression of LAT,

ICP27, UL30, UL5, and cellular GAPDH RNA transcripts.

(E) Number of reactivated GFP+ neurons after 48 hr of treatment with LY294002 or vehicle DMSO in the presence of WAY-150138 (20 mg/ml). Ten wells were

scored for each condition and data points indicate the number of individual GFP+ cells in a well containing 103 SCG neurons.

(F) WAY-150138 prevents HSV-1 spread in SCG neurons. Cultures were infected with HSV-1 Us11-GFP at aMOI of 0.1 in the presence ofWAY-150138 or DMSO

and the spread of virus visualized by GFP fluorescence after 72 hr.

Cell Host & Microbe

PI3-Kinase Signaling Regulates HSV-1 Latency

324 Cell Host & Microbe 8, 320–330, October 21, 2010 ª2010 Elsevier Inc.

Cell Host & Microbe

PI3-Kinase Signaling Regulates HSV-1 Latency

each of which is expressed in sympathetic neurons (Bartlett

et al., 1999). LY294002 is a broad-spectrum inhibitor capable

of antagonizing all PI3-K p110 isoforms, but small molecule

inhibitors selective for each isoform have also been character-

ized (Feldman et al., 2005; Knight et al., 2006). Latently infected

cultures were treated with three of these inhibitors: TGX115,

a selective inhibitor of p110b and p110d; IC87114, selective for

p110d; and PIK75, an inhibitor of p110a. Surprisingly, treatment

with p110a-selective inhibitor PIK75 resulted in substantial reac-

tivation that was nearly as efficient as LY294002 (Figure 3G). In

contrast, treatment with the p110b and p110d inhibitors

TGX115 and IC87114 did not result in reactivation (Figure 3G).

Thus, the catalytic activity of the PI3-K p110a subunit is most

critical for maintaining latent HSV-1 in cultured sympathetic

neurons.

Depletion of PDK1 with shRNAs Resultsin HSV-1 ReactivationActivation of PI3-K stimulates phosphatidylinositol phosphoryla-

tion and leads to the recruitment of 3-phosphoinositide-depen-

dent protein kinase-1 (PDK1) to the plasma membrane. We

examined the involvement of PDK1 in maintaining latency, using

BX-795, a pyrimidine derivative that inhibits PDK1 by competing

for the ATP-binding pocket of the catalytic site (Feldman et al.,

2005). BX-795 treatment resulted in levels of reactivation similar

to those induced by LY294002 (Figure 4A). Again, inhibition

could be readily demonstrated by monitoring phosphorylation

of a downstream substrate (Figure S3).

Next, the requirement for PDK1 was confirmed using RNA

interference, an independent approach that does not rely

upon chemical inhibitors. PDK1 was depleted using shRNAs

expressed from a pLVTHM lentiviral vector (Figure 4B) that

had been modified to express mCherry, thereby allowing lentivi-

ral infection and HSV-1 reactivation to be monitored simulta-

neously in live cells. Infection with two different PDK1 shRNA

lentiviruses successfully depleted endogenous PDK1 protein

levels and significantly resulted in reactivation at levels compa-

rable to LY294002 (Figure 4C). Parallel infections with a control

lentivirus did not induce reactivation unless neurons were

treated with LY294002, confirming that coinfection with a lenti-

virus does not have a detectable effect on HSV-1 latency or

reactivation.

We also tested a lentivirus expressing shRNA to phospholi-

pase Cg (PLCg), an independent arm of TrkA signaling (Fig-

ure 3A). While PLCg levels were reduced significantly by the

shRNA (Figure 4D), no increase in HSV-1 reactivation was

detected (Figure 4E). Cultures treated with PLCg shRNAs were

still capable of reactivation in response to LY294002 (Figure 4E),

demonstrating that PLCg was not required for productive repli-

cation. Thus, loss of the PLCg from NGF-TrkA signaling is not

sufficient to reactivate latent HSV-1. This result also strengthens

the observations made with the PDK1 shRNAs by showing that

the methodology does not necessarily give rise to reactivation.

Taken together, these findings show that specifically interrupting

the PI3-K signaling pathway either by inhibiting PDK1 activity or

(G) Selective inhibition of the PI3-K catalytic subunit p110a leads to reactivation

PIK75 (PI3-K p110a inhibitor, 0.116 mM), TGX115 (PI3-K p110bd inhibitor, 2.6 mM

the standard error of the mean.

Cell Hos

by selectively depleting PDK1 protein using shRNA resulted in

efficient reactivation. Moreover, these experiments clearly

demonstrate that shRNAs can provide an effective tool to study

HSV-1 latency.

Differential Ability of Growth Factorsto Support HSV-1 LatencyNGF is not alone in its ability to bind its receptor and trigger

PI3-K-mediated signaling. Indeed, it is surprising that a relatively

ubiquitous RTK-linked signal pathway component such as PI3-K

would be involved in suppressing HSV-1 lytic replication and

maintaining latency. This raises the intriguing possibility that

other growth factors that act through the PI3-K pathway and

are expressed in SCG neurons, such as EGF and GDNF, might

also regulate HSV-1 latency.

To address this, SCG neuron cultures were established and

maintained in media containing either NGF and EGF, or NGF

and GDNF (each at 50 ng/ml). Latent HSV-1 infections were

then established in each culture and assayed for reactivation

using blocking antibodies to individual growth factors. Removal

of NGF resulted in reactivation regardless of the presence

or absence of EGF (Figure 5A). In contrast, inclusion of GDNF

resulted in smaller numbers of GFP+ wells suggesting that

GDNF has some capacity to maintain latency after NGF deple-

tion (Figure 5B). Removal of both NGF and GDNF was required

to achieve maximal reactivation in cultures established and

maintained in the presence of both factors. The differential ability

of EGF and GDNF to maintain HSV-1 latency was not due to lack

of RTK activity, since both factors stimulated their respective

receptors, EGFR and c-RET (Figures 5C and 5D). Thus, despite

their ability to bind ligand and stimulate RTK signaling via a PI3K-

dependent pathway, NGF, EGF, and GDNF differed in their

ability to suppress lytic replication and maintain HSV-1 latency

in neurons.

Duration of Akt Activation Is Criticalto Maintain Latency in NeuronsThe serine/threonine kinase Akt represents a key component of

the PI3-K pathway and regulates fundamental cellular processes

such as apoptosis and protein synthesis. Because Akt is a prom-

inent substrate for PDK1-mediated phosphorylation, we treated

latently infected neurons with AKT inhibitor VIII, a cell permeable

allosteric inhibitor of Akt (Calleja et al., 2009), in the presence of

NGF (Figure 6B). Treatment with the inhibitor resulted in robust

activation with 60% of wells scoring positive for GFP in 2 days.

The ability of this compound to prevent activation of Akt as

measured by phosphorylation at serine 473 was confirmed by

immunoblotting (Figure S4). This result demonstrates that activa-

tion of Akt is necessary to maintain latent HSV-1 in sympathetic

neuron cultures.

The differential ability of NGF, EGF, and GDNF to maintain

latency cannot be explained by a simple lack of receptor expres-

sion or PI3-K activity and suggests that the duration of signaling

might be more important. Therefore, the kinetics of growth-

factor signaling in sympathetic neurons was examined. We

. Infected cultures were treated with the PI3-K inhibitor LY294002, LY303511,

), IC87114 (PI3-K p110d inhibitor, 2.6 mM), or DMSO. The error bars indicate

t & Microbe 8, 320–330, October 21, 2010 ª2010 Elsevier Inc. 325

DMSO

LY294002 BX-795A

C

Control lentivirus

ShRNA PDK1(3)

Control + LY294002

ShRNA PDK1(2)

B

Con

trol

ShR

NA

PDK1

(3)

ShR

NA

PDK1

(2)

PDK1

GAPDH

% re

activ

ated

wel

ls

% re

activ

ated

wel

ls

6 days post-treatment

6 days post-ACV removal

20

80

40

60

10

30

50

70

20

80

40

60

10

30

50

70

D

ShR

NA

PLC

-gam

ma

cont

rol

PLC-γ

GAPDH

EshRNA

PLC-γ + LY294002

Control% re

activ

ated

wel

ls

5 days post ACV removal

20

80

40

60

10

30

50

70

90100

shRNA PLC-γ

Figure 4. PDK-1 Is Required in Latency

(A) Inhibition of PDK1 induces reactivation. Reactivation in the presence of BX-

795 (PDK1 inhibitor, 1 mM), LY294002, or vehicle control DMSO. See also

Figure S3.

(B) Depletion of PDK1 was verified by immunoblot analysis of lysates prepared

after 6 days of infection with each lentivirus. GAPDH served as a loading

control.

(C) Depletion of PDK1 using shRNAs promotes reactivation. Following ACV

removal, GFP-HSV-1-infected cultures were infected with two different lentivi-

ruses expressing an mCherry fluorescent marker and shRNAs against rat

PDK1. Cultures infected with a control lentivirus in the presence or absence

of LY294002 acted as positive and negative controls respectively. Number

of wells expressing GFP was scored after 6 days.

(D) Depletion of PLCgwas verified by immunoblot analysis of lysates prepared

after 5 days of infection with each lentivirus. GAPDH served as a loading

control.

(E) Depletion of PLCg using shRNAs does not induce or prevent reactivation.

Following ACV removal, cultures harboring latent GFP-HSV-1 were infected

with a control lentivirus or with a lentivirus expressing an shRNA against rat

PLCg. Addition of LY294002 acted as a reactivation control. Number of wells

expressing GFP was scored after 5 days. The error bars indicate the standard

error of the mean.

Cell Host & Microbe

PI3-Kinase Signaling Regulates HSV-1 Latency

focused on two key phosphorylation sites on Akt: threonine 308

(T308), a major PDK1 substrate, and serine 473 (S473), a target

for phosphorylation by mTORC2, both of which are accepted

326 Cell Host & Microbe 8, 320–330, October 21, 2010 ª2010 Elsevie

indicators of Akt activation. Uninfected cultures of SCG neurons

were treated with each growth factor and lysates were prepared

after different time intervals and analyzed by immunoblotting. As

shown in Figures 6C and 6D, each growth factor produced

a strikingly different profile. In the presence of NGF, Akt was

rapidly phosphorylated on T308 and remained phosphorylated

at S473 over the 18 hr time period, whereas EGF gave only

a short-lived increase in phosphorylation at S473 and no detect-

able phosphorylation at T308, even at the shortest time point.

These responses indicated that NGF and EGF can both activate

Akt, but do so with very different kinetics as measured by phos-

phorylation on T308 and S473.

Treatment with GDNF showed an intermediate profile, with

a very similar profile to NGF at 2 hr but differed at 18 hr when

the phospho-S473 signal had returned to background levels.

To address this further, we performed a second time course

analysis choosing additional time points at which to compare

phosphorylation at S473 in the presence of NGF or GDNF

(Figure 6E). As before, both growth factors gave a similar profile

at early times but differed significantly at 18 hr and 36 hr. The

inability of GDNF to activate Akt (phospho-S473) for long periods

is consistent with its reduced ability to support HSV-1 latency in

neuron cultures. Taken together, these results argue that differ-

ential ability of individual growth factors to maintain latency and

suppress HSV-1 reactivation is directly related to their differing

abilities to provide sustained signaling through PI3-K and Akt.

DISCUSSION

The remarkable ability of HSV-1 to stably colonize and periodi-

cally reactivate from peripheral neurons is well accepted, but

the cellular andmolecular mechanisms responsible for maintain-

ing life-long latency punctuated by episodic reactivation remain

enigmatic. The underlying disparity in our understanding of

latency compared to the productive replication cycle largely

reflects the absence of a tractable experimental system to ask

mechanistic questions about fundamental interactions between

the virus and host neuron.

Here we describe a modified primary neuron cell-culture

system capable of supporting a stable, nonproductive HSV-1

infection that exhibits key hallmarks of latency, including nuclear

LAT accumulation and the absence of detectable lytic gene

expression. Lytic reactivation in live neurons can be scored in

real time using a GFP-reporter virus, and the cultures are

amenable to chemical or biological manipulations, permitting

mechanistic studies. Significantly, we have found that contin-

uous signaling through the canonical PI3-K pathway triggered

by NGF binding to the TrkA receptor was instrumental in main-

taining HSV-1 latency in primary neurons. PI3-K p110a catalytic

subunit activity, but not the alternative b or d isoforms, was

specifically required to suppress lytic replication and sustain

latency. Surprisingly, not all growth factors capable of stimu-

lating PI3-K signaling were equally effective at supporting

HSV-1 latency, and the ability to activate Akt in a sustained

manner appears to be a critical parameter (Figure 6A).

The importance of continuous PI3-K signaling in maintaining

latency highlights the role of the host neuron and cell-type-

specific signal pathways. While this does not diminish the contri-

bution of the host innate and acquired immune responses to

r Inc.

A

+ NGF+ EGF

- NGF- EGF

- NGF+ EGF

5 days post-treatment

B

+ NGF+ GDNF

- NGF+ GDNF

+ NGF- GDNF

- NGF- GDNF

pEGFR (Tyr 1069)

total EGFR

0 m5 h2

EGFC D

0 m5 h2

h81

GDNF

pRet (Tyr 905)

total Ret

5 days post-treatment

20

80

40

60

10

30

50

70

20

80

40

60

10

30

50

70

% re

activ

ated

wel

ls

% re

activ

ated

wel

ls

h81

Figure 5. Differential Effects of Growth

Factors on HSV-1 Reactivation

(A) EGF does not substitute for NGF in maintaining

latency. GFP-HSV-1 latently infected SCG neuron

cultures were established in the presence of NGF

andEGF (50ng/ml each),washed, and then cultured

for 5 days in media containing NGF and EGF (filled

bar), anti-EGF and anti-NGF antibody (open bar)

and EGF plus anti-NGF antibody only (shaded bar).

(B) GDNF can partially substitute for NGF. As in

(A), except that neurons were cultured in media

containing NGF and GDNF (50 ng/ml). Neurotro-

phins were removed by addition of either anti-

NGF or anti-GDNF receptor antibodies. The error

bars indicate the standard error of the mean.

(C and D) Cultured SCG neurons express func-

tional receptors for EGF and GDNF. Neurons

cultures were maintained for 6 hr in media lacking

growth factors and treated with either EGF or

GDNF (50 ng/ml each) and lysates prepared after

5 min, 2 hr, or 18 hr. Levels of activated receptor

were detected by immunoblotting using anti-

bodies to phospho-EGFR (C) and phospho-RET,

the GDNF receptor (D). Downregulation of the

receptor (shown here at 18 hr) after stimulation

has been reported previously (Beguinot et al.,

1984; Pierchala et al., 2006; Richardson et al.,

2006; Stoscheck and Carpenter, 1984).

Cell Host & Microbe

PI3-Kinase Signaling Regulates HSV-1 Latency

suppress reactivation in disease pathogenesis (Knickelbein

et al., 2008), or the potential for LATs to suppress lytic IE gene

expression (Garber et al., 1997), it directly demonstrates that

fundamental features of latency can be reconstituted by infect-

ing pure neuronal cultures with HSV-1 and illustrates that

a pivotal neuron-specific signal-transduction pathway is a critical

regulator of the virus. Importantly, these findings suggest that

neuronal targets of PI3-K/Akt signaling are the likely cellular

effectors responsible for maintaining latency. Alterations to

these cellular targets may transmit the initial reactivation signal

(s) to the repressed viral genome. Prolonged signaling through

the PI3-K/Akt axis could conceivably maintain critical aspects

of the latent state, including nuclear LAT accumulation, viral

microRNA production, cytoplasmic HCF-1 localization, and

maintenance of the viral genome in repressive chromatin state

(Umbach et al., 2008; Kristie et al., 1999; Bloom et al., 2010).

Alternatively, other cellular functions known to be regulated by

PI3-K/Akt, such as cap-dependent translation, may emerge as

important regulators. The cell-type-dependent expression of

receptors such as TrkA that display the appropriate PI3-K/Akt

activation profile are likely to be a critical determinant that limits

latency to peripheral neurons. Future studies using this neuronal

culture system will determine which parameters are most rele-

vant to latency.

Signaling through the PI3-K pathway is emerging as part of

a general mechanism to control the replication of a number of

important viruses (Buchkovich et al., 2008). For example, activa-

tion of the PI3-K pathway by the Epstein-Barr virus latent

membrane protein 2A (LMP2A) promotes the survival of the

host lymphocyte and prevents EBV reactivation (Morrison

et al., 2003; Swart et al., 2000). Along similar lines, recent work

with the Kaposi’s sarcoma-associated herpesvirus has shown

that inhibition of PI3-K signaling facilitates reactivation from

Cell Hos

latency (Peng et al., 2010). Our study shows for that differences

in the duration of PI3-K-mediated Akt activation by RTK signaling

directly correlate with the ability to maintain HSV-1 latency.

Differential outcomes resulting from NGF compared to EGF

signaling have also been reported in uninfected cultured cells

including PC12 cells (Chao, 1992). Furthermore, related strate-

gies relying on differential signal strength and duration dictate

decisions in BDNF-induced neuronal branching and plasticity,

lineage commitment in the immune system, differentiation, and

development (Calao et al., 2008; Hogquist, 2001; Ji et al.,

2010; Murphy and Blenis, 2006; Schmierer and Hill, 2007).

While much has been learned from studying latency in small

animal models, each model presents challenges when it comes

to identifying the molecular processes that regulate latency.

There is compelling evidence, for example, that both the innate

and acquired immune responses contribute to the control of

latency at the whole animal level (Khanna et al., 2003). From an

investigative standpoint, this imposes a substantial layer of

complexity that can confound studies of the intimate virus-

neuron relationship. Pharmacological inhibitors that target path-

ways within the neuron may alter the behavior of immune cells,

and regulatory factors that control the virus within a neuron

may be essential for immune function or even the viability of

the experimental animal. By utilizing a pure population of

neurons that can be readily manipulated, it becomes easier to

interpret studies of the discrete interactions between the virus

and its host cell. By the same logic, defining the spectrum of

virus-neuron interactions will ultimately highlight the unique

contributions of the immune response.

While the mechanistic connections between natural reactiva-

tion cues such as exposure to UV radiation, physical trauma

and psychological stress, and the PI3-K signaling pathway

have yet to be established, there are clear indications that

t & Microbe 8, 320–330, October 21, 2010 ª2010 Elsevier Inc. 327

D

0 m 5 h 2h 81 m 5 h 2h 81

NGF EGFC

pAKT (Thr 308) pAKT (Ser 473)

total AKT

pERK 1/2

total ERK 1/2

GAPDH

pTRKA pERK 1/2

total AKT

pAKT (Ser 473)

pAKT(Thr 308)

total PLC-γ

Hsp90

0 m 5 h 2h 81 m 5 h 2h 81

NGF GDNF

pPLC-γ (Tyr 783)

A NGFTrkA

Latent HSV-1 genome

Cytoplasm

Nucleus

PDK1

PI3-K

AKT

GDNFRet EGF

EGFR

S473 T308p p

p110-α

B

1 2 3 4 5 60

% re

activ

ated

wel

ls

Days post-treatment

20

80

40

60

10

30

50

70

DMSO

AKT VIII inhibitor LY294002

pAKT (Ser 473)

Hsp90

_ GD

NF

NG

F

GD

NF

NG

F

GD

NF

NG

F

GD

NF

NG

F

GD

NF

NG

F

10m 1h 5h 18h 36h

Total AKT

E

Figure 6. NGF, EGF, and GDNF Activate

Equivalent Downstream Pathways in

Cultured Sympathetic Neurons but with

Different Kinetics

(A) Sustained RTK-mediated activation of Akt is

required to maintain HSV-1 latency. GDNF and

EGF signal through the Ret and EGFR tyrosine

kinase receptor, respectively, and like NGF-TrkA,

they activate PI3-K-PDK1-AKT, albeit with

different kinetics. The ability of different growth

factors to maintain latency is directly proportional

to the duration of PI3-K signaling. NGF induces

prolonged PI3-K signaling to maintain latency.

GDNF has an intermediate ability to sustain the

PI3-K pathway and latency. Finally, EGF pro-

vides a transient signal and is unable to maintain

latency. In black are elements of the pathway

tested that maintain latency. PI3-K, phosphatidyli-

nositol 3-kinase; PDK1, phosphoinositide-depen-

dent kinase 1; Trk, tropomyosin-related kinase

receptor; NGF, nerve growth factor; Ret, Ret tyro-

sine kinase receptor; EGFR, epidermal growth

factor receptor.

(B) Reactivation of latent HSV-1 after treatment

with AKT inhibitor VIII (5 mM for 16 hr), PI3-K

inhibitor LY294002 (10 mM) or DMSO. See also

Figure S4. The error bars indicate the standard

error of the mean.

(C and D) Kinetics of growth factor signaling in

SCG neurons. Cultures (4 3 105 cells/lane) were

maintained for 6 hr in media lacking growth factors

and then supplemented with NGF, EGF, or GDNF

(50 ng/ml each). Lysates were prepared after

5 min, 2 hr, or 18 hr and analyzed by immunoblot-

ting to detect phosphorylated forms of TrkA, AKT,

ERK and PLCg. NGF treatment was compared to

EGF (A) or GDNF (B). GAPDH and Hsp90 serve

as loading controls.

(E) Profile of AKT phosphorylation by NGF and

GDNF. Cultures with supplemented with NGF or

GDNF and harvested after 10 min, 1 hr, 5 hr,

18 hr, and 36 hr. Lysates were analyzed by immu-

noblotting with antibodies against AKT phospho-

Ser-473 and total AKT. In the absence of trophic

support, neurons begin to die at around 36 hr as

indicated by decreased levels of Hsp90. Note

that the NGF 5 hr sample is slightly underloaded.

Cell Host & Microbe

PI3-Kinase Signaling Regulates HSV-1 Latency

ablation of NGF signaling can be a potent reactivation stimulus

in vivo. First, anti-NGF treatment in humans and latently infected

rabbits stimulates reactivation and is associated with aggra-

vated herpetic keratitis, the leading cause of infectious blindness

(Lambiase et al., 2008; Mauro et al., 2007). Second, surgical

axotomy (retrogasserian rhizotomy) for the treatment of trigem-

inal neuralgia is a known inducer of HSV-1 reactivation in

humans and can be reproduced in experimental animals (Carton

and Kilbourne, 1952; Cushing, 1905; Walz et al., 1974). NGF is

normally taken up at nerve terminals and transported in a retro-

grade manner to the cell body located in the ganglia (Ginty and

Segal, 2002). Severing of nerve axons blocks transport of

NGF-receptor complexes from the periphery to the cell body

and is approximated in our system by addition of anti-NGF anti-

bodies to the culture media.

328 Cell Host & Microbe 8, 320–330, October 21, 2010 ª2010 Elsevie

In addition to using selective chemical inhibitors to target

specific pathways, we have shown that host gene involvement

can be queried by shRNA-mediated silencing. Future application

of genome-wide shRNA screening techniques could potentially

define neuronal genes required to maintain latency or transition

to productive replication. Conversely, a battery of stimuli or small

molecules can be tested for their ability to provoke reactivation in

the presence of NGF. Other pathways capable of controlling re-

activation independent of PI3K signaling may thus be revealed.

The extent to which other classical reactivation stimuli identified

in humans and animals act on a neuron-autonomous level, or via

influencing secondary systems, can also be addressed. Basic

questions in HSV biology such as the role of LAT RNAs and

the temporal pattern of viral gene expression in reactivating

neurons can also now be explored in detail.

r Inc.

Cell Host & Microbe

PI3-Kinase Signaling Regulates HSV-1 Latency

EXPERIMENTAL PROCEDURES

Cell Culture and HSV-1 Infection

Superior cervical ganglia (SCG) neurons from E21 rat embryos were dissoci-

ated in trypsin (0.1%) at 37�C for 30 min. Approximately 5000 neurons per

well were plated in a 96-well plate coated with rat tail collagen (0.66 mg/ml,

08-115, Millipore). SCG neurons isolated in this manner provide a relatively

pure population of neurons expressing the TrkA receptor (Glebova and Ginty,

2005) andcontain fewnonneuronal cells. The cellsweremaintainedwith neuro-

basal media, glucose (0.4%), B27 supplement, NGF (50 ng/ml), and glutamine

(2 mM) and treated with 5-fluorouracil and aphidicolin to eliminate any dividing

cells that contaminate the cultures. After 6 days, the cells were pretreated with

acyclovir, (ACV, 50mM) for 20 hr, and subsequently infectedwithHSV-1 (Patton

strain GFP-Us11; multiplicity of infection (MOI) = 1 based upon titer on Vero

cells) for 2 hr in the presence of ACV to block productive HSV-1 replication.

Neurons were maintained in ACV for at least 6 days. After ACV removal,

infected neuronal cultures were exposed to different reactivation stimuli. In

an experiment, 22 independently infected wells were analyzed per individual

stimulus. Graphs summarize a minimum of three separate experiments and

error bars indicate the standard error of the mean.

RT-PCR

RNA was isolated from approximately 30,000 latently infected neurons and

analyzed by standard methodologies. The primer sequences are posted in

the supplementary section.

Combined Fluorescent In Situ Hybridization

and Indirect Immunofluorescence

Cells were cultured and infected with HSV-1 (GFP-Us11) as described above

but plated onto 8-well chamber slides at a density of �2 3 104 neurons/

chamber. In situ hybridization was performed by adding a mix containing

four LAT probes (200 nM each) for 5 hr at 42�C. LAT-specific oligonucleotides

were designed against the�2 kb intron region of HSV-1 strain 17 (Farrell et al.,

1991), and were synthesized with a fluorescein tag on the 50 end. All subse-quent incubations for immunofluorescence (IF) were done at RT. Additional

details can be found in the supplement.

shRNA Lentivirus Infection

Lentiviruses expressing shRNAs against rat PDK1 and rat PLCg were gener-

ated using a pLVTHM vector that included an mCherry expression cassette.

SCG cultures were infected with lentivirus (MOI = 5) for 12 hr prior infection

with HSV-1. The efficiency of lentiviral infection as judged by mCherry expres-

sion was approximately 90%. The shRNA sequences are posted in the supple-

mentary section.

Plaque Assay

Cell lysates were prepared from neuron cultures by freeze thawing, and the

amount of infectious virus determined by plaque assay using Vero cells and

serial dilutions of each lysate.

Viability Assay by Hoechst Staining

One week after plating, SCG cultures were treated with anti-NGF blocking

antibodies in the presence or absence of 20 mM ZVAD-fmk (627610, Calbio-

chem). After 48 hr, cells were stained with Hoechst 33342 (Molecular Probes)

and visualized by light microscopy. The number of apoptotic nuclei was deter-

mined by counting 200 cells.

Additional materials and methods can be found in the Supplemental

Information.

SUPPLEMENTAL INFORMATION

Supplemental Information includes four figures, Supplemental Experimental

Procedures, and references and can be found with this article online at

doi:10.1016/j.chom.2010.09.007.

Cell Hos

ACKNOWLEDGMENTS

We thank Kevan Shokat for the specific PI3K inhibitors, Josh Bloom (Wyeth)

for WAY-150138 and Eugene Johnson for advice. This work was supported by

grants to MVC (NS21072, HD23315), ACW (GM61139, S10RR017970) and IM

(AI073898, GM056927) from the NIH.

Received: March 9, 2010

Revised: July 10, 2010

Accepted: August 20, 2010

Published: October 20, 2010

REFERENCES

Bartlett, S.E., Reynolds, A.J., Tan, T., Heydon, K., and Hendry, I.A. (1999).

Differential mRNA expression and subcellular locations of PI3-kinase isoforms

in sympathetic and sensory neurons. J. Neurosci. Res. 56, 44–53.

Beguinot, L., Lyall, R.M., Willingham, M.C., and Pastan, I. (1984). Down-regu-

lation of the epidermal growth factor receptor in KB cells is due to receptor

internalization and subsequent degradation in lysosomes. Proc. Natl. Acad.

Sci. USA 81, 2384–2388.

Benboudjema, L., Mulvey, M., Gao, Y., Pimplikar, S.W., and Mohr, I. (2003).

Association of the herpes simplex virus type 1 Us11 gene product with the

cellular kinesin light-chain-related protein PAT1 results in the redistribution

of both polypeptides. J. Virol. 77, 9192–9203.

Berg, M.M., Sternberg, D.W., Parada, L.F., and Chao, M.V. (1992). K-252a

inhibits nerve growth factor-induced trk proto-oncogene tyrosine phosphory-

lation and kinase activity. J. Biol. Chem. 267, 13–16.

Bloom, D.C., Giordani, N.V., and Kwiatkowski, D.L. (2010). Epigenetic regula-

tion of latent HSV-1 gene expression. Biochim. Biophys. Acta 1799, 246–256.

Buchkovich, N.J., Yu, Y., Zampieri, C.A., and Alwine, J.C. (2008). The TORrid

affairs of viruses: effects of mammalian DNA viruses on the PI3K-Akt-mTOR

signalling pathway. Nat. Rev. Microbiol. 6, 266–275.

Calao, M., Burny, A., Quivy, V., Dekoninck, A., and Van Lint, C. (2008). A perva-

sive role of histone acetyltransferases and deacetylases in an NF-kappaB-

signaling code. Trends Biochem. Sci. 33, 339–349.

Calleja, V., Laguerre, M., Parker, P.J., and Larijani, B. (2009). Role of a novel

PH-kinase domain interface in PKB/Akt regulation: structural mechanism for

allosteric inhibition. PLoS Biol. 7, e17.

Carton, C.A., and Kilbourne, E.D. (1952). Activation of latent herpes simplex by

trigeminal sensory-root section. N. Engl. J. Med. 246, 172–176.

Chao, M.V. (1992). Growth factor signaling: where is the specificity? Cell 68,

995–997.

Chao, M.V. (2003). Neurotrophins and their receptors: a convergence point for

many signalling pathways. Nat. Rev. Neurosci. 4, 299–309.

Cushing, H. (1905). The surgical aspects of major neuralgia of the trigeminal

nerve. J. Am. Med. Assoc. XLIV, 773–779.

Deckwerth, T.L., and Johnson, E.M., Jr. (1993). Temporal analysis of events

associated with programmed cell death (apoptosis) of sympathetic neurons

deprived of nerve growth factor. J. Cell Biol. 123, 1207–1222.

Devi-Rao, G.B., Goodart, S.A., Hecht, L.M., Rochford, R., Rice, M.K., and

Wagner, E.K. (1991). Relationship between polyadenylated and nonpolyade-

nylated herpes simplex virus type 1 latency-associated transcripts. J. Virol.

65, 2179–2190.

Farrell, M.J., Dobson, A.T., and Feldman, L.T. (1991). Herpes simplex virus

latency-associated transcript is a stable intron. Proc. Natl. Acad. Sci. USA

88, 790–794.

Feldman, R.I., Wu, J.M., Polokoff, M.A., Kochanny, M.J., Dinter, H., Zhu, D.,

Biroc, S.L., Alicke, B., Bryant, J., Yuan, S., et al. (2005). Novel small molecule

inhibitors of 3-phosphoinositide-dependent kinase-1. J. Biol. Chem. 280,

19867–19874.

Garber, D.A., Schaffer, P.A., and Knipe, D.M. (1997). A LAT-associated

function reduces productive-cycle gene expression during acute infection

t & Microbe 8, 320–330, October 21, 2010 ª2010 Elsevier Inc. 329

Cell Host & Microbe

PI3-Kinase Signaling Regulates HSV-1 Latency

of murine sensory neurons with herpes simplex virus type 1. J. Virol. 71, 5885–

5893.

Ginty, D.D., and Segal, R.A. (2002). Retrograde neurotrophin signaling: Trk-ing

along the axon. Curr. Opin. Neurobiol. 12, 268–274.

Glaser, R., and Kiecolt-Glaser, J.K. (2005). Stress-induced immune dysfunc-

tion: implications for health. Nat. Rev. Immunol. 5, 243–251.

Glebova, N.O., and Ginty, D.D. (2005). Growth and survival signals controlling

sympathetic nervous system development. Annu. Rev. Neurosci. 28, 191–222.

Hill, J.M., Garza, H.H., Jr., Helmy, M.F., Cook, S.D., Osborne, P.A., Johnson,

E.M., Jr., Thompson, H.W., Green, L.C., O’Callaghan, R.J., and Gebhardt,

B.M. (1997). Nerve growth factor antibody stimulates reactivation of ocular

herpes simplex virus type 1 in latently infected rabbits. J. Neurovirol. 3,

206–211.

Hogquist, K.A. (2001). Signal strength in thymic selection and lineage commit-

ment. Curr. Opin. Immunol. 13, 225–231.

Huber, L.J., and Chao, M.V. (1995). A potential interaction of p75 and trkA NGF

receptors revealed by affinity crosslinking and immunoprecipitation. J. Neuro-

sci. Res. 40, 557–563.

Ji, Y., Lu, Y., Yang, F., Shen, W., Tang, T.T., Feng, L., Duan, S., and Lu, B.

(2010). Acute and gradual increases in BDNF concentration elicit distinct

signaling and functions in neurons. Nat. Neurosci. 13, 302–309.

Kaplan, D.R., Hempstead, B.L., Martin-Zanca, D., Chao, M.V., and Parada,

L.F. (1991). The trk proto-oncogene product: a signal transducing receptor

for nerve growth factor. Science 252, 554–558.

Khanna, K.M., Bonneau, R.H., Kinchington, P.R., and Hendricks, R.L. (2003).

Herpes simplex virus-specific memory CD8+ T cells are selectively activated

and retained in latently infected sensory ganglia. Immunity 18, 593–603.

Knickelbein, J.E., Khanna, K.M., Yee, M.B., Baty, C.J., Kinchington, P.R., and

Hendricks, R.L. (2008). Noncytotoxic lytic granule-mediated CD8+ T cell inhi-

bition of HSV-1 reactivation from neuronal latency. Science 322, 268–271.

Knight, Z.A., Gonzalez, B., Feldman, M.E., Zunder, E.R., Goldenberg, D.D.,

Williams, O., Loewith, R., Stokoe, D., Balla, A., Toth, B., et al. (2006). A phar-

macological map of the PI3-K family defines a role for p110alpha in insulin

signaling. Cell 125, 733–747.

Knipe, D.M., and Cliffe, A. (2008). Chromatin control of herpes simplex virus

lytic and latent infection. Nat. Rev. Microbiol. 6, 211–221.

Kristie, T.M., Vogel, J.L., and Sears, A.E. (1999). Nuclear localization of the C1

factor (host cell factor) in sensory neurons correlates with reactivation of

herpes simplex virus from latency. Proc. Natl. Acad. Sci. USA 96, 1229–1233.

Lambiase, A., Coassin, M., Costa, N., Lauretti, P., Micera, A., Ghinelli, E., Aloe,

L., Rama, P., and Bonini, S. (2008). Topical treatment with nerve growth factor

in an animal model of herpetic keratitis. Graefes Arch. Clin. Exp. Ophthalmol.

246, 121–127.

Leib, D.A., Bogard, C.L., Kosz-Vnenchak, M., Hicks, K.A., Coen, D.M., Knipe,

D.M., and Schaffer, P.A. (1989a). A deletion mutant of the latency-associated

transcript of herpes simplex virus type 1 reactivates from the latent state with

reduced frequency. J. Virol. 63, 2893–2900.

Leib, D.A., Coen, D.M., Bogard, C.L., Hicks, K.A., Yager, D.R., Knipe, D.M.,

Tyler, K.L., and Schaffer, P.A. (1989b). Immediate-early regulatory gene

mutants define different stages in the establishment and reactivation of herpes

simplex virus latency. J. Virol. 63, 759–768.

MacInnis, B.L., and Campenot, R.B. (2002). Retrograde support of neuronal

survival without retrograde transport of nerve growth factor. Science 295,

1536–1539.

Mauro, C., Pietro, L., and Emilio, C.C. (2007). The use of nerve growth factor in

herpetic keratitis: a case report. J. Med. Case Reports 1, 124.

Morrison, J.A., Klingelhutz, A.J., and Raab-Traub, N. (2003). Epstein-Barr virus

latent membrane protein 2A activates beta-catenin signaling in epithelial cells.

J. Virol. 77, 12276–12284.

Murphy, L.O., and Blenis, J. (2006). MAPK signal specificity: the right place at

the right time. Trends Biochem. Sci. 31, 268–275.

330 Cell Host & Microbe 8, 320–330, October 21, 2010 ª2010 Elsevie

Newcomb, W.W., and Brown, J.C. (2002). Inhibition of herpes simplex virus

replication by WAY-150138: assembly of capsids depleted of the portal and

terminase proteins involved in DNA encapsidation. J. Virol. 76, 10084–10088.

Peng, L., Wu, T.T., Tchieu, J.H., Feng, J., Brown, H.J., Feng, J., Li, X., Qi, J.,

Deng, H., Vivanco, I., et al. (2010). Inhibition of the phosphatidylinositol

3-kinase-Akt pathway enhances gamma-2 herpesvirus lytic replication and

facilitates reactivation from latency. J. Gen. Virol. 91, 463–469.

Pesola, J.M., Zhu, J., Knipe, D.M., and Coen, D.M. (2005). Herpes simplex

virus 1 immediate-early and early gene expression during reactivation from

latency under conditions that prevent infectious virus production. J. Virol.

79, 14516–14525.

Pierchala, B.A., Milbrandt, J., and Johnson, E.M., Jr. (2006). Glial cell line-

derived neurotrophic factor-dependent recruitment of Ret into lipid rafts

enhances signaling by partitioning Ret from proteasome-dependent degrada-

tion. J. Neurosci. 26, 2777–2787.

Richardson, D.S., Lai, A.Z., and Mulligan, L.M. (2006). RET ligand-induced

internalization and its consequences for downstream signaling. Oncogene

25, 3206–3211.

Schmierer, B., andHill, C.S. (2007). TGFbeta-SMADsignal transduction:molec-

ular specificity and functional flexibility. Nat. Rev. Mol. Cell Biol. 8, 970–982.

Skoff, A.M., and Adler, J.E. (2006). Nerve growth factor regulates substance P

in adult sensory neurons through both TrkA and p75 receptors. Exp. Neurol.

197, 430–436.

Smith, R.L., Escudero, J.M., and Wilcox, C.L. (1994). Regulation of the herpes

simplex virus latency-associated transcripts during establishment of latency in

sensory neurons in vitro. Virology 202, 49–60.

Stoscheck, C.M., and Carpenter, G. (1984). Down regulation of epidermal

growth factor receptors: direct demonstration of receptor degradation in

human fibroblasts. J. Cell Biol. 98, 1048–1053.

Swart, R., Ruf, I.K., Sample, J., and Longnecker, R. (2000). Latent membrane

protein 2A-mediated effects on the phosphatidylinositol 3-Kinase/Akt

pathway. J. Virol. 74, 10838–10845.

Thompson, R.L., Preston, C.M., and Sawtell, N.M. (2009). De novo synthesis of

VP16 coordinates the exit fromHSV latency in vivo. PLoSPathog. 5, e1000352.

Umbach, J.L., Kramer, M.F., Jurak, I., Karnowski, H.W., Coen, D.M., and

Cullen, B.R. (2008). MicroRNAs expressed by herpes simplex virus 1 during

latent infection regulate viral mRNAs. Nature 454, 780–783.

van Zeijl, M., Fairhurst, J., Jones, T.R., Vernon, S.K., Morin, J., LaRocque, J.,

Feld, B., O’Hara, B., Bloom, J.D., and Johann, S.V. (2000). Novel class of thio-

urea compounds that inhibit herpes simplex virus type 1 DNA cleavage and

encapsidation: resistance maps to the UL6 gene. J. Virol. 74, 9054–9061.

Wagner, E.K., and Bloom, D.C. (1997). Experimental investigation of herpes

simplex virus latency. Clin. Microbiol. Rev. 10, 419–443.

Walz, M.A., Price, R.W., and Notkins, A.L. (1974). Latent ganglionic infection

with herpes simplex virus types 1 and 2: viral reactivation in vivo after neurec-

tomy. Science 184, 1185–1187.

Warren, S.L., Carpenter, C.M., and Boak, R.A. (1940). Symptomatic herpes,

a sequela of artificially induced fever: incidence and C aspects; recovery of

a virus from herpetic vesicles, and comparison with a K strain of herpes virus.

J. Exp. Med. 71, 155–168.

Wheeler, C.E., Jr. (1975). Pathogenesis of recurrent herpes simplex infections.

J. Invest. Dermatol. 65, 341–346.

Wilcox, C.L.J., and Johnson, E.M., Jr. (1987). Nerve growth factor deprivation

results in the reactivation of latent herpes simplex virus in vitro. J. Virol. 61,

2311–2315.

Wilcox, C.L., and Johnson, E.M., Jr. (1988). Characterization of nerve growth

factor-dependent herpes simplex virus latency in neurons in vitro. J. Virol.

62, 393–399.

Wilcox, C.L., Smith, R.L., Freed, C.R., and Johnson, E.M., Jr. (1990). Nerve

growth factor-dependence of herpes simplex virus latency in peripheral

sympathetic and sensory neurons in vitro. J. Neurosci. 10, 1268–1275.

Ye, H., Kuruvilla, R., Zweifel, L.S., and Ginty, D.D. (2003). Evidence in support

of signalling endosome-based retrograde survival of sympathetic neurons.

Neuron 39, 57–68.

r Inc.