Construction and properties of a herpes simplex virus 1 ... · their properties. Attenuated herpes...

6
Construction and properties of a herpes simplex virus 1 designed to enter cells solely via the IL-132 receptor Guoying Zhou and Bernard Roizman* The Marjorie B. Kovler Viral Oncology Laboratories, University of Chicago, 910 East 58th Street, Chicago, IL 60637 Contributed by Bernard Roizman, February 14, 2006 Current design of genetically engineered viruses for selective destruction of cancer cells is based on the observation that atten- uated viruses replicate better in tumor cells than in normal cells. The ideal virus, however, is one that can infect only cancer cells by virtue of altered host range. Such a virus can be made more robust than the highly attenuated viruses used in clinical trials. Earlier, we reported the construction of a recombinant herpes simplex virus 1 (R5111) in which the capacity to bind heparan sulfate was disabled and which contained a chimeric IL-13-glycoprotein D that enabled the virus to infect cells expressing the IL-132 receptor (IL-13R2) commonly found on the surface of malignant glioblastomas or high-grade astrocytomas. In the earlier report, we showed that the recombinant R5111 was able to enter and infect cells via the interaction of the chimeric glycoprotein D with IL-13R2 but that the virus retained the capacity to bind and replicate in cells expressing the natural viral receptors HveA or nectin-1. Here, we report the construction of a recombinant virus (R5141) that can only enter and replicate in cells that express the IL-13R2. The recombinant R5141 does not depend on endocytosis to infect cells. It does not infect cells expressing HveA or nectin-1 receptors or cells expressing IL-13R2 that had been exposed to soluble IL-13 before infection. The studies described here show that the host range of herpes simplex viruses can be altered by genetic manipulation to specifically target cancer cells. glycoprotein D HveA malignant glioblastoma nectin-1 T he ideal oncolytic drug is one that fully discriminates be- tween normal and tumor cells. Preeminent among such agents are viruses. A property of most viruses is that they have a host range that defines the species and type of cell that can become infected or replicate the virus. An additional advantage of viruses is that they can be genetically manipulated to alter their properties. Attenuated herpes simplex virus 1 (HSV-1), among other viruses, has been tested as a potential oncolytic agent with promising results (1–9). Thus viruses attenuated by deletion of the 1 34.5 genes or 1 34.5 and the gene encoding the major subunit of the viral ribonucleotide reductase (U L 39) were found to be safe by inoculation into the central nervous system of patients with malignant glioblastoma (4, 7). The few long-term survivors after such treatment attested to the potential useful- ness of these viruses but also exposed their limitation. Because the principal characteristic of viruses tested in clinical trials that enables them to discriminate between normal and tumor cells is that they cannot overcome the host defenses of the former, their ability to replicate and spread among tumor cells is severely restricted and depends in large part on the genotype of the tumor. An ideal virus would retain at least some of the properties of wild-type viruses to replicate and disseminate efficiently from cell to cell but would discriminate between normal and tumor cells by targeting surface proteins that are either unique to or grossly enriched on tumor cells. In the case of HSV-1, the main mechanism of entry into cells consists of two steps (reviewed in refs. 10–12). In the first, the viral glycoproteins B (gB) and C (gC) contained in the virion envelope bind heparan sulfate proteoglycans on cell surfaces. In the second step, the confor- mation of the viral glycoprotein D (gD) becomes altered as a consequence of its interaction with one of two cell surface receptors, HveA, a member of the TNF- receptor family of proteins, or nectin-1, a member of a family of proteins that forms the intercellular networks that link cells together (10, 13, 14). The change in the conformation of gD is thought to enable a trio of glycoproteins, gH, gL, and gB, to interact with gD and enable the fusion of the envelope with the plasma membrane (15, 16). Subsequent to the fusion event, the capsid is transported to a nuclear pore where it releases the viral DNA directly into the nucleus (17). The goal of the studies described in this article was to genetically engineer HSV-1 so that it does not bind heparan sulfate proteoglycans and its entry into cells requires recognition of cell surface markers other than HveA or nectin-1. In principle, viruses can be retargeted in two different ways. The first is to tag virion surface proteins with a single-chain antibody that recognizes a novel receptor. The extent to which such viruses interact solely with the intended receptor hinges on several factors, but primarily on the nature of the epitope recognized by the antibody (18). The second approach is to incorporate into the virion surface natural ligands for proteins unique or enriched on the tumor cell surface. We have chosen the second approach. In the first phase of these studies we constructed a virus (R5111) that was capable of entering cells via the IL-132 receptor (IL-13R2) but which retained the capacity to enter cells via HveA or nectin-1 (19). The IL-13R2 is monomeric and present in malignant glioblastomas and high- grade astrocytomas; it has also been reported to be present in normal testes (20, 21). We have used the same design to construct a recombinant virus capable of entering cells via the urokinase plasminogen activator receptor while still retaining the capacity to enter cells via HveA or nectin-1 receptors (22). In this article we describe the further modification of a virus that enters cells solely via the IL-13R2 protein. This study estab- lishes the proof of principle that viruses can be targeted to specific single proteins on cell surfaces. Results Construction of R5141 Recombinant Virus. The construction of a mutant dependent on IL-13R2 for entry into cells took place in two stages. The first reported earlier involved the construction of a recombinant virus, R5111, that was able to enter cells via the IL-13R2 but retained the ability to enter cells via the HveA or nectin-1 receptors. The structure of this recombinant is shown in Fig. 1, lane 1. Briefly, in the R5111 mutant virus the polylysine tract in gB was removed and the first 132 residues of gC were Conflict of interest statement: No conflicts declared. Abbreviations: HSV-1, herpes simplex virus 1; gB, glycoprotein B; gC, glycoprotein C; gD, glycoprotein D; IL-13R2, IL-132 receptor; pfu, plaque-forming units; ICP, infected cell protein. *To whom correspondence should be addressed. E-mail: [email protected]. © 2006 by The National Academy of Sciences of the USA 5508 –5513 PNAS April 4, 2006 vol. 103 no. 14 www.pnas.orgcgidoi10.1073pnas.0601258103 Downloaded by guest on February 2, 2021

Transcript of Construction and properties of a herpes simplex virus 1 ... · their properties. Attenuated herpes...

Page 1: Construction and properties of a herpes simplex virus 1 ... · their properties. Attenuated herpes simplex virus 1 (HSV-1), among other viruses, has been tested as a potential oncolytic

Construction and properties of a herpes simplexvirus 1 designed to enter cells solelyvia the IL-13�2 receptorGuoying Zhou and Bernard Roizman*

The Marjorie B. Kovler Viral Oncology Laboratories, University of Chicago, 910 East 58th Street, Chicago, IL 60637

Contributed by Bernard Roizman, February 14, 2006

Current design of genetically engineered viruses for selectivedestruction of cancer cells is based on the observation that atten-uated viruses replicate better in tumor cells than in normal cells.The ideal virus, however, is one that can infect only cancer cells byvirtue of altered host range. Such a virus can be made more robustthan the highly attenuated viruses used in clinical trials. Earlier, wereported the construction of a recombinant herpes simplex virus 1(R5111) in which the capacity to bind heparan sulfate was disabledand which contained a chimeric IL-13-glycoprotein D that enabledthe virus to infect cells expressing the IL-13�2 receptor (IL-13R�2)commonly found on the surface of malignant glioblastomas orhigh-grade astrocytomas. In the earlier report, we showed that therecombinant R5111 was able to enter and infect cells via theinteraction of the chimeric glycoprotein D with IL-13R�2 but thatthe virus retained the capacity to bind and replicate in cellsexpressing the natural viral receptors HveA or nectin-1. Here, wereport the construction of a recombinant virus (R5141) that canonly enter and replicate in cells that express the IL-13R�2. Therecombinant R5141 does not depend on endocytosis to infect cells.It does not infect cells expressing HveA or nectin-1 receptors or cellsexpressing IL-13R�2 that had been exposed to soluble IL-13 beforeinfection. The studies described here show that the host range ofherpes simplex viruses can be altered by genetic manipulation tospecifically target cancer cells.

glycoprotein D � HveA � malignant glioblastoma � nectin-1

The ideal oncolytic drug is one that fully discriminates be-tween normal and tumor cells. Preeminent among such

agents are viruses. A property of most viruses is that they havea host range that defines the species and type of cell that canbecome infected or replicate the virus. An additional advantageof viruses is that they can be genetically manipulated to altertheir properties. Attenuated herpes simplex virus 1 (HSV-1),among other viruses, has been tested as a potential oncolyticagent with promising results (1–9). Thus viruses attenuated bydeletion of the �134.5 genes or �134.5 and the gene encoding themajor subunit of the viral ribonucleotide reductase (UL39) werefound to be safe by inoculation into the central nervous systemof patients with malignant glioblastoma (4, 7). The few long-termsurvivors after such treatment attested to the potential useful-ness of these viruses but also exposed their limitation. Becausethe principal characteristic of viruses tested in clinical trials thatenables them to discriminate between normal and tumor cells isthat they cannot overcome the host defenses of the former, theirability to replicate and spread among tumor cells is severelyrestricted and depends in large part on the genotype of thetumor. An ideal virus would retain at least some of the propertiesof wild-type viruses to replicate and disseminate efficiently fromcell to cell but would discriminate between normal and tumorcells by targeting surface proteins that are either unique to orgrossly enriched on tumor cells. In the case of HSV-1, the mainmechanism of entry into cells consists of two steps (reviewed inrefs. 10–12). In the first, the viral glycoproteins B (gB) and C(gC) contained in the virion envelope bind heparan sulfate

proteoglycans on cell surfaces. In the second step, the confor-mation of the viral glycoprotein D (gD) becomes altered as aconsequence of its interaction with one of two cell surfacereceptors, HveA, a member of the TNF-� receptor family ofproteins, or nectin-1, a member of a family of proteins that formsthe intercellular networks that link cells together (10, 13, 14).The change in the conformation of gD is thought to enable a trioof glycoproteins, gH, gL, and gB, to interact with gD and enablethe fusion of the envelope with the plasma membrane (15, 16).Subsequent to the fusion event, the capsid is transported to anuclear pore where it releases the viral DNA directly into thenucleus (17). The goal of the studies described in this article wasto genetically engineer HSV-1 so that it does not bind heparansulfate proteoglycans and its entry into cells requires recognitionof cell surface markers other than HveA or nectin-1.

In principle, viruses can be retargeted in two different ways.The first is to tag virion surface proteins with a single-chainantibody that recognizes a novel receptor. The extent to whichsuch viruses interact solely with the intended receptor hinges onseveral factors, but primarily on the nature of the epitoperecognized by the antibody (18). The second approach is toincorporate into the virion surface natural ligands for proteinsunique or enriched on the tumor cell surface. We have chosenthe second approach. In the first phase of these studies weconstructed a virus (R5111) that was capable of entering cells viathe IL-13�2 receptor (IL-13R�2) but which retained the capacityto enter cells via HveA or nectin-1 (19). The IL-13R�2 ismonomeric and present in malignant glioblastomas and high-grade astrocytomas; it has also been reported to be present innormal testes (20, 21). We have used the same design toconstruct a recombinant virus capable of entering cells via theurokinase plasminogen activator receptor while still retainingthe capacity to enter cells via HveA or nectin-1 receptors (22).In this article we describe the further modification of a virus thatenters cells solely via the IL-13R�2 protein. This study estab-lishes the proof of principle that viruses can be targeted tospecific single proteins on cell surfaces.

ResultsConstruction of R5141 Recombinant Virus. The construction of amutant dependent on IL-13R�2 for entry into cells took place intwo stages. The first reported earlier involved the constructionof a recombinant virus, R5111, that was able to enter cells via theIL-13R�2 but retained the ability to enter cells via the HveA ornectin-1 receptors. The structure of this recombinant is shown inFig. 1, lane 1. Briefly, in the R5111 mutant virus the polylysinetract in gB was removed and the first 132 residues of gC were

Conflict of interest statement: No conflicts declared.

Abbreviations: HSV-1, herpes simplex virus 1; gB, glycoprotein B; gC, glycoprotein C; gD,glycoprotein D; IL-13R�2, IL-13�2 receptor; pfu, plaque-forming units; ICP, infected cellprotein.

*To whom correspondence should be addressed. E-mail: [email protected].

© 2006 by The National Academy of Sciences of the USA

5508–5513 � PNAS � April 4, 2006 � vol. 103 � no. 14 www.pnas.org�cgi�doi�10.1073�pnas.0601258103

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

1

Page 2: Construction and properties of a herpes simplex virus 1 ... · their properties. Attenuated herpes simplex virus 1 (HSV-1), among other viruses, has been tested as a potential oncolytic

replaced with the amino acid sequences encoding IL-3. Theobjectives were 2-fold, to remove the sites that enabled gB andgC to bind to heparin sulfate proteoglycans as described byLaquerre et al. (23) and at the same time to increase the affinityof the virus particle to the surface of cells expressing theIL-13R�2. To specifically target the IL-13R�2, the codingsequence of IL-13 was inserted between amino acids 24 and 25of gD. As previously reported, the entry and initiation ofinfection by R5111 recombinant virus of cells expressing theIL-13R�2 protein on their surface totally depended on theinteraction of IL-13 present on the surface of the virus withthe cognate receptor inasmuch as soluble IL-13 competed withthe virus for the receptor (24).

The objective of the second stage reported here was toconstruct a recombinant virus that was unable to bind nectin-1or HveA receptors. These studies were done in three phases. Inthe first phase we made linker insertions in the chimeric gD genecontained in the R5111 recombinant virus. The basis for theconstruction of this series of recombinants stemmed from re-ports that suggested that mutations in gD distant from the knownsites of interaction of gD with HveA abolished the ability of themutant gD to interact with nectin-1. Thus gD-carrying substi-tutions in residues 11, 27, 28, 29, 30, 40, and 43 were reported tobind nectin-1 but not HveA (25). In another study (26) substi-tutions of residue 38 disabled gD in fusion assays. Still anotherstudy (27) reported that double or triple amino acid substitutionsat positions 215, 222, and 223 of gD precluded attachment tonectin-1 and corresponding inability to function in fusion assays.These results suggested that the binding sites in gD for HveA andnectin-1 did not overlap. As shown in Table 1, a common andunexpected characteristic of the R5200 series of mutants is thatthey retained the capacity to infect and replicate in J-nectin-1cells but exhibited a reduced capacity to replicate in J-13R orJ-HveA cell lines relative to those of the wild-type virus or theparent R5111 virus.

The construction of the recombinants in phase 2 of this studywas based on the hypothesis that the functional interactionsbetween gD and HveA or nectin-1 overlap and occur at theamino terminus of gD and that modifications of gD at sitesdistant from the amino terminus interfere with the interactionsof the altered protein with HveA or nectin-1. The key mutationsdesigned to test this hypothesis are illustrated in Fig. 1 and Table1. Specifically, the key mutations are as follows.

(i) In recombinant R5121 and R5113, the chimeric gD retainedits signal peptide sequence but residues 1–24 or 1–32 werereplaced with the coding sequence of IL-13. These recom-binants failed to replicate or replicated poorly in any of thecell lines tested.

(ii) In recombinant R5123, the signal peptide sequence of gDwas replaced with that of IL-13. In addition, the codingsequence of IL-13 replaced the residues 1–32 of gD. Weselected three clones for analysis. The clone shown in Table1 replicated better in J-13R cells than in J-nectin-1 cells andfailed to replicate in J-HveA cells.

In the third phase of these studies we introduced mutations inthe sequences immediately upstream from residue 33. In thechimeric gD genes constructed to make additional recombinantviruses we substituted Val-34 with Ser, Arg-35 with Asn, Arg-36with Asp, Val-37 with Ser, or Tyr-38 with Ser, Asn, Asp, or His.The only constructs that yielded viable viruses were the substi-tutions of Val-34 with Ser (R5141) and Val-37 with Ser (R5144).Of the two recombinants only R5141 replicated in J-13R cells butnot in J-nectin or J-HveA cells (Fig. 2 and Table 1). Therecombinant was also replicated and produced plaques in theVero-13R cell line but not in the parental Vero cells (data notshown). The plaques were smaller than those made by wild-typeparent virus; they could be readily counted and indicated that thevirus spread from infected to adjacent, noninfected cells. How-ever, the yields of R5141 in IL-13R�2-expressing cells were�100-fold lower than those obtained from wild-type virus incells expressing HveA or nectin-1 receptor (Fig. 2 and Table 1).

Fig. 1. Schematic representation of the sequence arrangements of thewild-type parent virus HSV-1(F) and R5111 and the chimeric gD contained inrecombinant viruses R5121, R5113, R5123, R5141, and R5144. In all recombi-nant viruses, residues 68–78 of gB were deleted and IL-13 was inserted in placeof residues 1–140 of gC. In R5111, R5121, and R5113, the signal sequence andresidues 1–24 of gD were retained. In R5123, the residues 1–33 were deletedand only the signal peptide sequence of gD was retained. In R5141 and R5144,the signal peptide sequence of gD was replaced by that of IL-13. In addition,these two chimeric gDs contain additional amino acid substitutions as shown.The IL-13 inserted in R5121 and R5113 recombinant viruses carried the E13Ymutation, in which the glutamic acid was substituted by tyrosine. E13Ysubstitution restricts IL-13 binding to the IL-13R�2 (52). R5121, R5141, andR5144 carried a wild-type IL-13 copy in gD.

Table 1. Replication of genetically engineered viruses in celllines expressing specific receptors for viral entry

RecombinantsInsertion (�)Deletion (�)

Yields

J-nectin J-HveA J-13R

HSV-1(F) WT 6 � 108 5 � 108 2 � 101

R5208 �34 GKIFL 5 � 105 6 � 102 6 � 102

R5209 �43 EDLP 4 � 107 6 � 102 9 � 102

R5212 �77 GKIFP 4 � 107 6 � 102 2 � 101

R5213 �77 EDLP 2 � 107 6 � 102 7 � 104

R5216 �83 GRSS 7 � 106 6 � 102 7 � 104

R5217 �84 GKIFP 7 � 106 8 � 101 5 � 105

R5220 �125 GRSS 6 � 106 7 � 101 6 � 102

R5221 �126 GKIFP 6 � 106 7 � 101 6 � 103

R5225 �151 WKIFL 4 � 107 6 � 102 6 � 102

R5231 �187 GRSS 5 � 105 3 � 101 8 � 103

R5237 �243 GRSS 6 � 107 5 � 104 3 � 103

R5239 �246 GKIFP 7 � 106 8 � 101 4 � 101

R5240 �246 EDLP 5 � 105 5 � 101 6 � 102

R5242 �277 GKIFP 5 � 105 3 � 104 4 � 101

R5121 � 1–24 7 � 105 4 � 101 6 � 101

R5113 � 1–32 7 � 103 4 � 101 6 � 103

R5123 � SP, � 1–32 8 � 104 3 � 101 4 � 105

R5141 � SP, � 1–32, V34S 8 � 101 6 � 101 6 � 106

R5144 � SP, � 1–32, V37S 7 � 101 5 � 102 7 � 102

J-nectin, J-HveA, or J-13R cells grown in 25-cm2 flasks were exposed to 0.1pfu of wild-type or recombinant virus per cell. The cultures were harvested24 h after infection. Progeny virus was titrated on Vero-13R cells.

Zhou and Roizman PNAS � April 4, 2006 � vol. 103 � no. 14 � 5509

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

1

Page 3: Construction and properties of a herpes simplex virus 1 ... · their properties. Attenuated herpes simplex virus 1 (HSV-1), among other viruses, has been tested as a potential oncolytic

The experiments described below were designed to characterizethe R5141 recombinant virus with respect to entry into cells andgrowth properties.

The Recombinant R5141 Depends on the Interaction of the IL-13Component of Its Envelope with the IL-13 Receptor for Virus Entry intoCells. Although we earlier demonstrated that the recombinantR5111 enters cells via its interaction with IL-l3R�2, it wasessential to determine whether R5141 recombinant enters J-13Rcells by the same pathway and whether it also depends onIL-13R�2 for entry into the Vero-13R cell line. In this series ofexperiments J-13R or Vero-13R cells were mock-treated orexposed to 1 or 10 ng of IL-13, IL-2, or IL-4 for 1 h. The cellswere then exposed to 0.1 plaque-forming units (pfu) of virus percell. The cells were harvested 24 h after infection, and the viruswas tittered in Vero-ILR cells. The results are shown in Fig. 3.The experimental design was based on the report that IL-13R�2is internalized after binding IL-13 and therefore the receptorwould not be available for viral entry (28–30). As predicted,exposure of J-13R or Vero-13R cells to IL-13 significantlydecreased viral replication, whereas IL-2 or IL-4 had no effect.

R5141 Recombinant Virus Does Not Depend on Endocytosis for Entryinto J-13R or Vero-13R Cells. HSV-1 enters cells by one of twomechanisms. The first involves the fusion of the envelope withthe plasma membrane followed by the release of capsid-tegument structures into the cytoplasm. Uptake of virions byendocytosis constitutes an alternative mode of entry, which doesnot require the interaction between gD and its cell surfacereceptor. However, subsequent events depend on the presence ofat least a portion of gD in the environment of the infected cell.In the absence of gD, the endocytosed virion is degraded.Because R5141 recombinant virus carries a chimeric gD con-taining all but 33 residues of the native protein, it might beexpected that endocytosed virions could infect cells. The ques-tion arose, therefore, whether R5141 enters cells by endocytosis.The experimental design of these studies was based on the resultsreported earlier that the inhibitor PD98095 blocked the repli-cation of a gD mutant in a cell line expressing gD and exposedto gD recombinant virus (24). Consistent with the experimentaldesign reported earlier (24), replicate cultures of Vero, J-13R, orVero-13R cells were mock-treated or incubated in mediumcontaining PD98095 (30 �M) for 1 h and then exposed to 0.1 pfu

of R5141 recombinant virus or the wild-type HSV-1(F) virus percell. As expected, wild-type virus replicated in Vero and Vero-13R but not in J-13R cells. The inhibitor PD980959 had no effecton the replication of HV-1(F) or R5141 virus in cells susceptibleto infection (Fig. 4). We conclude that the R5141 recombinantvirus did not depend on endocytosis for entry and replication inJ-13R or Veo-13R cells.

Fig. 2. Replication of R5141, R5144 virus, and HSV-1(F) in J-nectin, J-HveA,and J-13R cells. Cells grown in 25-cm2 flasks were exposed to 0.1 pfu of therecombinant virus or wild-type HSV per cell and harvested 24 h after infection.Progeny virus was titrated on Vero-13R cells.

Fig. 3. Preexposure of J-13R or Vero-13R cells to IL-13 precludes the infectionand replication of R5141 mutant virus in both cell lines. Replicate cultures ofJ-13R cells (A) or Vero-13R cells (B) were mock-treated or exposed for 1 h toIL-13, IL-2, or IL-4 (1 ng��l or 10 ng��l). Then the cells were infected with 0.1pfu of HSV-1(F) or R5141 per cell and harvested 24 h after infection. Progenyvirus was titrated on Vero-13R cells.

Fig. 4. R5141 recombinant virus does not depend on endocytosis for entryand replication in J-13R or Vero-IL-13R cell lines. Vero, Vero-13R, and J-13Rcells were exposed to a 30 �M concentration of PD98059 inhibitor for 1 h. Thecells were infected with 0.1 pfu of HSV-1(F) or R5141 per cell and thenharvested 24 h after infection. Progeny virus was titrated on Vero-13R cells.

5510 � www.pnas.org�cgi�doi�10.1073�pnas.0601258103 Zhou and Roizman

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

1

Page 4: Construction and properties of a herpes simplex virus 1 ... · their properties. Attenuated herpes simplex virus 1 (HSV-1), among other viruses, has been tested as a potential oncolytic

Comparison of Viral Gene Expression of R5141 Recombinant Virus inVero and Vero-J-13 Cells. In this series of experiments Vero orVero-13R cells grown in 25-cm2 flasks were exposed to 1.0 pfuof HSV-1 or R5141 virus per cell. The cells were harvested 24 hafter infection and processed as described in Materials andMethods, and 100 mg of protein was loaded on a denaturing geland subjected to electrophoresis, then electrically transferredonto a nitrocellulose sheet and reacted with a monoclonalantibody against infected cell protein (ICP) 0, gC, gM, gD, gG,UL38, or US11. The results shown in Fig. 5 were as follows:

(i) As expected, viral proteins were not detected in Vero cellsexposed to R5141 recombinant virus because the recombinantdoes not infect these cells.

(ii) Except for the chimeric gD, R5141 recombinant virusproduced in Vero-13R cells amounts of viral proteins tested inthese assays that were nearly equivalent to those made by thewild-type parent virus in Vero or Vero-13R cells. Neither theantibody H170 to the amino-terminal domain deleted in R5141recombinant nor the ZC25 antibody to the carboxyl-terminaldomain detected significant amounts of the chimeic gD proteinin these assays. IL-13 gD chimeric protein was readily detectedin immunoprecipitates from R5141 virus-infected cells labeledwith [35S]methionine (data not shown).

DiscussionEver since humans embarked on a search for therapeutic agents,the implicit idealized objectives were to find drugs that would bemaximally effective and minimally harmful. In the case of thecurrent treatment of cancer, this objective has not been met.Aggressive treatment that produces long-term cures or signifi-cantly delays relapses is frequently associated with severe sideeffects. There is, in fact, no magic bullet for the simple reasonthat treatment directed primarily to tumor cells also affectsnormal cells that share properties with tumor cells. Becauseviruses target specific cells, they have long been viewed aspotential oncolytic agents. The advent of tools designed to

genetically engineer viruses suitable for human administrationled to design and clinical trials of candidate therapeutic viruses.The requirements for viral therapeutic agents, that is safety andefficacy, are not different from those demanded of other ther-apeutic agents. Here in fact is where problems arose. To convertpathogenic agents that can cause lethal disease so as to specif-ically infect and destroy cancer cell, it was necessary to renderthem replication-defective or delete genes that enabled virusesto replicate in nondividing, resting cells, or delete genes thatwould enable the virus to overcome host defenses to infection.HSVs are particularly suitable for genetic engineering of suchmutants because the technology for genetic engineering of suchviruses has been available for the past 20� years (31), the keyfunction of many of the viral genes is known, and the genomecontains a wide repertoire of genes suitable for genetic engi-neering in all of the categories listed above (32). Thus replica-tion-incompetent viruses that remain cytotoxic, albeit unable tocomplete the replicative cycle, result from deletion of genesencoding the regulatory ICPs 0, 4, 22, 27, and 4 (cited in ref. 32).Replication-competent viruses restricted to reproduce solely individing cells result from deletion of genes encoding viralthymidine kinase or ribonucleotide reductase (33). The premierexample of a gene essential to overcome host defenses andtherefore a suitable target for deletion is �134.5 described earlierin the text (34). The inherent problem in using recombinantviruses from which one or more of these genes have been deletedis that either they retain an unacceptable level of pathogenicityor they do not fully discriminate between normal and tumor cells.Although they do not replicate or spread in normal cells, thelimited viral gene expression is nevertheless cytotoxic to theinfected normal cells. The defect that prevents viral replicationin the normal cell is also manifest to only a slightly lesser degreein tumor cells. In the end, effectiveness of a virus hinges on thegenotype of the tumor cell. For example, tumor cells in whichprotein kinase R is activated after infection are likely to be farmore resistant to the replication of ��134.5 mutants than tumorsin which the kinase is not activated (35). Attempts to overcomethis problem by insertion of genes that encode nonviral cytotoxicagents (e.g., cyclophosphamide, etc.) (36, 37), insertion of cel-lular genes that bolster immunity to tumors such as IL-4 or IL-12(38–40), or combined use of ionizing radiation and virus (41) arealternatives still being tested. The pathway we have chosen is notto restrict the capacity of the virus to replicate by gene deletionbut rather to retarget the virus to novel receptors present inabundant amounts on tumor cells but not on normal cells. Wehave selected for this purpose the IL-13R�2 present in humanmalignant gliomas, high-grade astrocytomas, and less frequentlyin other tissues. In normal tissues it has been reported to bepresent in human testes, but not in other tissues. In this articlewe described the construction and initial characterization of arecombinant virus that targets the IL-13R�2. The issues relevantto this study are as follows:

The R5141 recombinant virus enters cells by its interactionwith the IL-13R�2 inasmuch as depletion of the receptor fromthe cell surface by exposure of the cells to soluble IL-13significantly reduces virus yields. We have also shown that theR5141 recombinant does not depend on endocytosis for entryinto cells. The results indicate that entry into cells is thereforereceptor-mediated at the surface of the cells exposed to thisvirus. The key feature that differentiates this virus from allprevious attempts to construct targeted viruses (e.g., R5111,R5200 series, R5123) is that it is unable to enter cells via aninteraction with the natural viral receptors HveA or nectin-1.Furthermore, the deletion of the poly(K) sequence in gB and thereplacement of the amino-terminal domain of gC with IL-13should render the R5141 recombinant incapable of binding tothe heparan sulfate moiety on cell surfaces.

Fig. 5. Comparison of the accumulation of selected R5141 and wild-typeparent proteins in Vero cells (Left) and Vero-13R cells (Right). Vero andVero-13R cells grown in 25-cm2 flasks were exposed to 1.0 pfu of HSV-1 orR5141 virus per cell. The cells were harvested 24 h after infection, solubilized,subjected to electrophoresis in 10% denaturing polyacrylamide gels, electri-cally transferred onto a nitrocellulose sheet, and reacted with antibodyagainst ICP0, gC, gM, gD (H170 and ZC25), gG, UL38, or US11. The trace amountof gD reacting with anti-gD(H170) represents spillover from an adjacent laneand is not reproducible.

Zhou and Roizman PNAS � April 4, 2006 � vol. 103 � no. 14 � 5511

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

1

Page 5: Construction and properties of a herpes simplex virus 1 ... · their properties. Attenuated herpes simplex virus 1 (HSV-1), among other viruses, has been tested as a potential oncolytic

Elsewhere, we have reported the construction of a virus(R5181) capable of entering and replicating in cells via theurokinase plasminogen activator receptor (22). A key propertyof the target receptor is that it is bound to the cells surface bya glycosylphosphatidylinositol anchor. Replication of R5181recombinant virus is abolished by digestion of cells expressingthis receptor with phosphatidyl inositol-specific phospholipaseC. These results suggest that the construction of viruses targetingspecific cellular surface proteins is not restricted to a specificclass of receptors (22).

Finally, the R5141 presents two puzzles that remain to beresolved. The first concerns the accumulation of low, virtuallyundetectable levels of the chimeric gD. The mechanism under-lying this gross decrease in the accumulation of this protein isunclear. It is conceivable that IL-13 fused to the amino-terminaldomain of gD imparts on the chimeric protein a shorter half-lifeor that it interacts with the receptor in cellular membranes andis degraded. An earlier report concerning a viral mutant that hada similar property concluded that the amounts of gD required forviral infection need not be very large (42). It remains to be seenwhether the lower yields of R5141 relative to those of wild-typeparent virus are related to the low levels of the chimeric gD.

The second problem relates to the actual mechanism thatenables the entry of HSV into cells. Biochemical data (15)supported by structural studies (16) unambiguously indicate thatthe interaction of gD with its receptor results in a conformationchange that most likely enables the interaction of gD withfusogenic viral glycoprotein and ultimately the fusion of theenvelope of the virus with the plasma membrane. Because theinteraction of the chimeric gD with its novel receptor occursthrough an extension of gD that presumably does not impinge onthe structure of the unmodified gD residues, the question arisesas to how the altered interaction enables the conformationalchange in gD that permits the fusion of the envelope with theplasma membrane. Elucidation of this problem may enable thedesign and construction of more efficient chimeric gD structures.

The studies reported here pave the way for the constructionof viruses that target other receptors suitable for both destruc-tion of cancer cells or delivery of life-saving genes to cells thatneed them.

Materials and MethodsCells. Vero cells were obtained from the American Type CultureCollection and maintained in DMEM supplemented with 10%FBS. J1.1, a derivative of a baby hamster kidney (BHK) thymi-dine kinase-negative cell line, lacks both HveA and nectin-1HSV-1 receptors (43). J-HveA and J-nectin-1 cell lines, express-

ing HveA and nectin-1, respectively, were derived from the J1.1cell line and were kind gifts from G. Campadelli-Fiume (Uni-versity of Bologna, Bologna, Italy). The cell lines J-13R andVero-IL-13 derived by transfection of J1–1 and Vero cell lineswith a plasmid encoding the IL-13R�2 have been described (24).

Viruses. The gD��� mutant lacks both the gD gene and the gDprotein in its envelope. gD��� virus enters cells by endocytosis,and at high multiplicities in the absence of complementing gD itcauses cells to undergo apoptosis (44). The recombinant virusR5111 has been described (19).

The R5200 series of recombinant viruses (R5208, R5209,R5212, R5213, R5216, R5217, R5220, R5221, R5225, R5231,R5237, R5239, R5239, R5240, and R5242) listed in Table 1express mutagenized linker-insertion mutant gD originally ob-tained from R. J. Eisenberg and G. H. Cohen (University ofPennsylvania, Philadelphia). They were further modified byinsertion of IL-13 between residues 24 and 25 as in the recom-binant virus R5111. The procedures for construction of therecombinant viruses carrying the gD linker-insertion mutantshave been described (45). In the recombinant R5121 virus IL-13was inserted after the signal peptide of gD at residue 25 of gD.The sequences encoding residues 1–24 of gD were deleted. InR5113 recombinant, IL-13 was inserted between the signalpeptide and residue 33 of gD, replacing residues 1–32.

Reagents. The endocytosis inhibitor PD98059 was purchasedfrom Cell Signaling Technology (Beverly, MA). IL-13, IL-4, andIL-2 were purchased from Research Diagnostics (Flanders, NJ).

Virus Titrations. Replicate cultures of J-13R, J-HveA, J-nectin,and Vero-13R were exposed to 0.1 pfu of recombinant viruses orthe wild-type parental HSV-1(F) virus per cell. After 24 h ofincubation, the cells were harvested and disrupted by gentlesonication. Viral progeny were titrated on Vero-13R cells.

Antibodies. Monoclonal antibodies against the amino-terminalsequence of gD clone H170 (46) or ICP0 (clone H1083) (47)were purchased from the Goodwin Institute (Plantation, FL).Monoclonal antibodies against gC (19), gM (48), gG (49), UL38(50), and US11 protein (51) have been described. The ZC25antibody to the carboxyl-terminal domain of gD was the kind giftof G. H. Cohen and R. J. Eisenberg.

We thank Dr. Gabriella Campadelli-Fiume for the J-1, J-nectin, andJ-Hve cell lines and for advice. These studies were aided by NationalCancer Institute Grants CA115662, CA83939, CA71933, CA78766, andCA88860.

1. Pyles, R. B., Warnick, R. E., Chalk, C. L., Szanti, B. E. & Parysek, L. M. (1997)Hum. Gene Ther. 8, 533–544.

2. Rampling, R., Cruickshank, G., Papanastassiou, V., Nicoll, J., Hadley, D.,Brennan, D., Petty, R., MacLean, A., Harland, J., McKie, E., et al. (2000) GeneTher. 7, 859–866.

3. McKie, E. A., Brown, S. M., MacLean, A. R. & Graham, D. I. (1998)Neuropathol. Appl. Neurobiol. 24, 367–372.

4. Markert, J. M., Medlock, M. D., Rabkin, S. D., Gillespie, G. Y., Todo, T.,Hunter, W. D., Palmer, C. A., Feigenbaum, F., Tornatore, C., Tufaro, F., et al.(2000) Gene Ther. 7, 867–874.

5. Harrow, S., Papanastassiou, V., Harland, J., Mabbs, R., Petty, R., Fraser, M.,Hadley, D., Patterson, J., Brown, S. M. & Rampling, R. (2004) Gene Ther. 11,1648–1658.

6. Detta, A., Harland, J., Hanif, I., Brown, S. M. & Cruickshank, G. (2003) J. GeneMed. 5, 681–689.

7. Papanastassiou, V., Rampling, R., Fraser, M., Petty, R., Hadley, D., Nicoll, J.,Harland, J., Mabbs, R. & Brown, M. (2002) Gene Ther. 9, 398–406.

8. Shah, A. C., Benos, D., Gillespie, G. Y. & Markert, J. M. (2003) J. Neurooncol.65, 203–226.

9. Varghese, S. & Rabkin, S. D. (2002) Cancer Gene Ther. 9, 967–978.10. Campadelli-Fiume, G., Cocchi, F., Menotti, L. & Lopez, M. (2000) Rev. Med.

Virol. 10, 305–319.

11. Spear, P. G. (2004) Cell Microbiol. 6, 401–410.12. Spear, P. G. & Longnecker, R. (2003) J. Virol. 77, 10179–10185.13. Montgomery, R. I., Warner, M. S., Lum, B. J. & Spear, P. G. (1996) Cell 87,

427–436.14. Spear, P. G., Eisenberg, R. J. & Cohen, G. H. (2000) Virology 275, 1–9.15. Fusco, D., Forghieri, C. & Campadelli-Fiume, G. (2005) Proc. Natl. Acad. Sci.

USA 102, 9323–9328.16. Krummenacher, C., Supekar, V. M., Whitneck, J. C., Connolly, S. A., Eisenberg,

R. J., Cohen, G., Wiley, D. C. & Carfi, A. (2005) EMBO J. 24, 4144–4153.17. Batterson, W., Furlong, D. & Roizman, B. (1983) J. Virol. 45, 397–407.18. Jiang, A., Chu, T. H., Nocken, F., Cichutek, K. & Dornburg, R. (1998) J. Virol.

72, 10148–10156.19. Zhou, G., Ye, G. J., Debinski, W. & Roizman, B. (2002) Proc. Natl. Acad. Sci.

USA 99, 15124–15129.20. Debinski, W., Gibo, D. M., Hulet, S. W., Connor, J. R. & Gillespie, G. Y. (1999)

Cancer Res. 5, 985–990.21. Mintz, A., Gibo, D. M., Slagle-Webb, B., Christensen, N. D. & Debinski, W.

(2002) Neoplasia 4, 388–399.22. Kamiyama, H., Zhou, G. & Roizman, B. (November 17, 2005) Gene Ther.,

10.1038�sj.gt.3302685.23. Laquerre, S., Argnani, R., Anderson, D. B., Zucchini, S., Manservigi, R. &

Glorioso, J. C. (1998) J. Virol. 72, 6119–6130.

5512 � www.pnas.org�cgi�doi�10.1073�pnas.0601258103 Zhou and Roizman

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

1

Page 6: Construction and properties of a herpes simplex virus 1 ... · their properties. Attenuated herpes simplex virus 1 (HSV-1), among other viruses, has been tested as a potential oncolytic

24. Zhou, G. & Roizman, B. (2005) J. Virol. 79, 5272–5277.25. Yoon, M. & Spear, P. G. (2004) Proc. Natl. Acad. Sci. USA 101, 17252–17257.26. Connolly, S. A., Landsburg, D. J., Carfi, A., Whitbeck, J. C., Zuo, Y., Wiley,

D. C., Cohen, G. H. & Eisenberg, R. J. (2005) J. Virol. 79, 1282–1295.27. Manoj, S., Jogger, C. R., Myscofski, D., Yoon, M. & Spear, P. G. (2004) Proc.

Natl. Acad. Sci. USA 101, 12414–12421.28. Kawakami, K., Taguchi, J., Murata, T. & Puri, R. K. (2001) Blood 97,

2673–2679.29. Kawakami, K., Joshi, B. H. & Puri, R. K. (2000) Hum. Gene Ther. 11, 1829–1835.30. Feng, N., Lugli, S. M., Schnyder, B., Gauchat, J. F., Graber, P., Schlagenhauf,

E., Schnarr, B., Wiederkehr-Adam, M., Duschl, A., Heim, M. H., et al. (1998)Lab. Invest 78, 591–602.

31. Post, L. E. & Roizman, B. (1981) Cell 25, 227–232.32. Roizman, B. & Knipe, D. M. (2001) in Fields Virology, eds. Knipe, D. M.,

Howley, P., Griffin, D. E., Lamb, R. A., Martin, M. A., Roizman, B. & Straus,S. E. (Lippincott Williams & Wilkins, New York), 4th Ed., pp. 2399–2459.

33. Mineta, T., Rabkin, S. D. & Martuza, R. L. (1994) Cancer Res. 54, 3963–3966.34. He, B., Chou, J., Brandimarti, R., Mohr, I., Gluzman, Y. & Roizman, B. (1997)

J. Virol. 71, 6049–6054.35. Mezhir, J. J., Advani, S. J., Smith, K. D., Darga, T. E., Poon, A. P. W., Schmidt,

H., Posner, M. C., Roizman, B. & Weichselbaum, R. R. (2005) Cancer Res. 65,9479–9484.

36. Kambara, H., Saeki, Y. & Chiocca, E. A. (2005) Cancer Res. 65, 11255–11258.37. Tyminski, E., Leroy, S., Terada, K., Finkelstein, D. M., Hyatt, J. L., Dansk,

M. K., Poter, P. M., Saeky, Y. & Chiocca, E. A. (2005) Cancer Res. 64,6850–6857.

38. Andreansky, S., He, B., van Cott, J., McGhee, J., Markert, J. M., Gillespie,G. Y., Roizman, B. & Whitley, R. J. (1998) Gene Ther. 5, 121–130.

39. Parker, J. N., Gillespie, G. Y., Love, C. E., Randall, S., Whitley, R. J. &Markert, J. M. (2000) Proc. Natl. Acad. Sci. USA 97, 2208–2213.

40. Markert, J. M., Parker, J. N., Gillespie, G. Y. & Whitley, R. J. (2001) Herpes8, 17–22.

41. Advani, S. J., Sibley, G. S., Song, P. Y., Hallahan, D. E., Kataoka, Y., Roizman,B. & Weichselbaum, R. R. (1998) Gene Ther. 5, 160–165.

42. Huber, M. T., Wisner, T. W., Hegde, N. R., Goldsmith, K. A., Rauch, D. A.,Roller, R. J., Krummenacher, C., Eisenberg, R. J., Cohen, G. H. & Johnson,D. C. (2001) J. Virol. 74, 10309–10318.

43. Cocchi, F., Menotti, L., Mirandola, P. & Campadelli-Fiume, G. (1998) J. Virol.72, 9992–10002.

44. Zhou, G., Galvan, V., Campadelli-Fiume, G. & Roizman, B. (2000) J. Virol. 74,11782–11791.

45. Zhou, G., Avitabile, E., Campadelli-Fiume, G. & Roizman, B. (2003) J. Virol.77, 3759–3767.

46. Arsenakis, M., Tomasi, L. F., Speziali, V., Roizman, B. & Campadelli-Fiume,G. (1986) J. Virol. 58, 367–376.

47. Ackermann, M., Braun, D. K., Pereira, L. & Roizman, B. (1984) J. Virol. 52,108–118.

48. Baines, J. D. & Roizman, B. (1993) J. Virol. 67, 1441–1452.49. Longnecker, R. & Roizman, B. (1987) Science 236, 573–576.50. Ward, P. L., Ogle, W. O. & Roizman, B. (1996) J. Virol. 70, 4623–4631.51. Roller, R. J. & Roizman, B. (1992) J. Virol. 66, 3624–3632.52. Debinski, W. & Thompson, J. P. (1999) Clin. Cancer Res. 5, 3143s–3147s.

Zhou and Roizman PNAS � April 4, 2006 � vol. 103 � no. 14 � 5513

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 2,

202

1