Proc. Natl. Acad. Sci. USAVol. 93, pp. 11407-11413, October 1996Colloquium Paper

This paper was presented at a colloquium entitled "Genetic Engineering of Viruses and of Virus Vectors," organized byBernard Roizman and Peter Palese (Co-chairs), held June 9-11, 1996, at the National Academy of Sciences in Irvine, CA

Cell-surface receptors for retroviruses and implications forgene transferA. DUSTY MILLERFred Hutchinson Cancer Research Center, 1100 Fairview Avenue North, Room C2-023, Seattle, WA 98109

ABSTRACT Retroviruses can utilize a variety of cell-surface proteins for binding and entry into cells, and thecloning of several of these viral receptors has allowed refine-ment of models to explain retrovirus tropism. A single recep-tor appears to be necessary and sufficient for entry of manyretroviruses, but exceptions to this simple model are accu-mulating. For example, HIV requires two proteins for cellentry, neither of which alone is sufficient; lOAl murineleukemia virus can enter cells by using either of two distinctreceptors; two retroviruses can use different receptors in somecells but use the same receptor for entry into other cells; andposttranslational protein modifications and secreted factorscan dramatically influence virus entry. These findings greatlycomplicate the rules governing retrovirus tropism. The mech-anism underlying retrovirus evolution to use many receptorsfor cell entry is not clear, although some evidence supports amutational model for the evolution of new receptor specific-ities. Further study of factors that govern retrovirus entry intocells are important for achieving high-efficiency gene trans-duction to specific cells and for the design of retroviral vectorsto target additional receptors for cell entry.

Many features make retrovirus vectors a good choice for genetransfer into animal cells. Most importantly, these vectorsintegrate efficiently into the target cell genome to promotestable gene transfer, and integration is precise with respect tothe virus genome, resulting in unrearranged transfer of thedesired genes. The only other integrating vector is derivedfrom adeno-associated virus, but integration is inefficient (1)and appears not to be precise with respect to the viral genome(2). In addition, retroviral vectors can transduce both dividingand non-dividing cells, although this is true of vectors derivedfrom HIV (3) and not the commonly used vectors derived frommurine leukemia viruses, which require cell division (4).Furthermore, retrovirus vectors can be designed to eliminateall viral protein coding regions without affecting gene transferrates, and can be made in the absence of replication-competentvirus by using retrovirus packaging cell lines, which supply allof the viral proteins required for vector transmission. Genetransfer and expression mediated by such replication-incompetent vectors is called transduction to differentiate thisprocess from virus infection followed by further virus replica-tion.A key consideration in retroviral vector design is the

source of the viral envelope (Env) protein present on vectorvirions, because this protein binds to specific cell-surfaceproteins and is the primary determinant of the range of cellsthat can be transduced by the vector. The name of the virusor the virus group from which the Env protein was derivedwill be referred to as the pseudotype of the vector. Naturally

occurring retroviruses can use a variety of different proteinsfor cell entry, although in general individual retrovirusesappear to recognize- a single receptor. Utilization of addi-tional cell-surface proteins for vector entry has beenachieved by incorporation of polypeptides into the Envprotein to alter its receptor binding properties or by replace-ment of the retroviral Env protein with surface proteins fromother viruses. These alterations can allow targeting of par-ticular cells that express specific proteins or an expansion ofthe range of cells that can be transduced by targeting broadlyexpressed proteins. In this paper I will review the factors thatgovern retrovirus binding and entry into cells and implica-tions for the design of retroviral vectors.

Virus Interference

Early evidence that retroviruses use multiple receptors for cellentry came from studies of virus interference. Infection of acell by a replication-competent retrovirus results in synthesisof a retroviral Env protein that binds to the receptor used forvirus entry. This effectively blocks entry of the original virusand other retroviruses that target the same receptor, whereasentry of retroviruses that use different receptors is unaffected.Interference between retroviruses has been shown to occur atthe level of virus entry into cells and not at some other step inthe virus life cycle. By interference analysis, retroviruses thatinfect human cells have been assigned to eight groups that usedifferent receptors on human cells (Table 1). The genesencoding these receptors are scattered on different chromo-somes (Table 1), indicating that the receptors are differentproteins.

Cloned Retrovirus Receptors

In 1984 CD4 (previously called T4) was identified as areceptor for HIV-1, and became the first known retrovirusreceptor (12, 13). Since then, six additional retrovirus re-ceptors have been identified and their cDNAs cloned (Table2). All except CD4 appear to be sufficient for entry of thecorresponding retroviruses by the criteria that expression ofthese receptors in nonpermissive cells renders the cellssusceptible to infection. In contrast, CD4 transfer intononpermissive mouse cells does not allow infection by HIV.HIV binds to all cells that express CD4, but another factoris required for HIV entry. Recently, a coreceptor for T-celltropic HIV-1 strains has been found and was named fusin toindicate its presumed role in virus entry following HIV-1binding to CD4 (14). Expression of the human CD4 and fusinproteins in mouse cells renders the cells susceptible to HIV-1infection, whereas either protein alone is insufficient. Even

Abbreviations: MLV, murine leukemia virus; AM-MLV, amphotropicMLV; MoMLV, Moloney MLV; CHO, Chinese hamster ovary;GALV, gibbon ape leukemia virus; FeLV, feline leukemia virus.

11407

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

July

22,

202

1

Proc. Natl. Acad. Sci. USA 93 (1996)

Table 1. Retrovirus interference groups in human cells

Interference Human chromosomegroup Virus Description that encodes receptor

1 RD114 Cat endogenous virus 19SNV Avian spleen necrosis virusBaEV Baboon endogenous virusSRV-1 Simian retrovirusSRV-2 Simian retrovirusSRV-3 (MPMV) Simian retrovirusSRV-4 Simian retrovirusSRV-5 Simian retrovirusPO-1-Lou Spectacled langur retrovirusSMRV Squirrel monkey retrovirus

2 MLV-A Amphotropic murine leukemia virus 83 MLV-X Xenotropic murine leukemia virus4 FeLV-C Feline leukemia virus5 FeLV-B Feline leukemia virus -2

SSAV Simian sarcoma-associated virusGALV Gibbon ape leukemia virus

6 BLV Bovine leukemia virus7 HTLV-1 Human T-cell leukemia virus 17

HTLV-2 Human T-cell leukemia virusChTLV Chimpanzee T-cell leukemia virusSTLV Simian T-cell leukemia virus

8 HIV-1 Human immunodeficiency virus 12HIV-2 Human immunodeficiency virusSIV Simian immunodeficiency virus

Interference data are from Sommerfelt and Weiss (5), and for SNV, from Kewalramani et al. (6).Chromosome localization data are from the following references: group 1 (7), group 2 (8), group 5 (9),group 7 (10), and group 8 (11).

more recently, a second protein related to fusin and previ-ously named CC-CKR-5 has been found to be a coreceptorfor macrophage-tropic HIV-1 strains (15, 16).These results, showing that two proteins are required for

HIV-1 entry, raise the possibility that coreceptors are re-

quired for entry of other retroviruses. However, their de-tection will require the identification of nonpermissive cellsfor which transfer of the known receptors does not render thecells susceptible to infection. Some retroviruses have a verywide host range; thus, if other proteins are required for entryof these viruses, functional homologs of these coreceptorsmust be widely distributed in cells from many species.Two of the cloned retrovirus receptors, Raml and Glvrl, are

closely related at the protein sequence level (21, 22, 24), andboth are sodium-dependent phosphate transporters (23).These proteins are members of a large family of known andpresumptive phosphate transporters from many organisms(Fig. 1). However, Raml and Glvrl are clearly distinct sincethe genes encoding these proteins are located on differentchromosomes in humans and mice (8, 9, 30, 31) and they showvery different patterns of expression in animal tissues (23). In

addition, these proteins serve as receptors for distinct groupsof viruses in human cells (Table 1).

The lOAl Retrovirus Can Use Either ofTwo Receptors for Cell Entry

Studies of cloned retrovirus receptors and most virus inter-ference data suggested that individual retroviruses bind to a

single protein for entry into cells. When different viruses bindto the same receptor, they typically show reciprocal inter-fer-ence; that is, infection of cells by either virus blocks entry bythe other virus. The finding of nonreciprocal interferencebetween some retroviruses complicated this picture. In theexample shown (Table 3), transduction by a vector with an

amphotropic, a lOA1, or an ecotropic pseudotype was mea-sured in NIH 3T3 mouse cells infected with amphotropic MLV(AM-MLV), lOAl MLV, Moloney MLV, or no virus. A typicalpattern of interference for viruses that use different receptorsfor cell entry is shown by the amphotropic and ecotropicviruses, where ecotropic vector transduction is blocked by thepresence of ecotropic MoMLV in the target cells, but isunaffected by the presence of amphotropic virus, and ampho-

Table 2. Cloned retrovirus receptors

Retrovirus Receptor Type* Function Refs.

Human immunodeficiency virus CD4 TM1 Immune recognition 12, 13Fusin, CC-CKR-5 - G protein-coupled

(coreceptors) TM7 chemokine receptors 14-16Simian immunodeficiency virus CD4 TM1 Immune recognition 17Murine ecotropic retrovirus Recl TM14 Basic amino acid transport 18-20Murine amphotropic retrovirus Raml TM10-13 Phosphate transport 21-23Gibbon ape leukemia virus Glvrl TM10-13 Phosphate transport 23, 24Bovine leukemia virus Blvr TM1 ND 25, 26Avian leukosis virus type A Tva TM1 LDL receptor-like protein 27Feline immunodeficiency virus CD9 TM4 Signaling protein? 28, 29

ND, not determined; LDL, low density lipoprotein.*TM followed by a number indicates the number of transmembrane domains in the protein.

11408 Colloquium Paper: Miller

Dow

nloa

ded

by g

uest

on

July

22,

202

1

Proc. Natl. Acad. Sci. USA 93 (1996) 11409

hRamlHomo sapiens

rRamlRattus norvegicus

cRamlCricetulus griseus

hGlvrlHomo sapiens

cGIvrlCricetulus griseus

mGlvrlMus musculus

B0222.2C. elegansB0222.3C. elegans

YGO4H. influenzaeORF.. -.. I

ThermoautotrophicumPho4Neurospora crassa

YB81S. cerevisiae

PitAE. cofi

PitBE. coli

PitHStreptomyces halstediiPitMycobacterium leprae

FIG. 1. Dendrogram of amino acid sequence similarities amongphosphate transporters. Distances between sequences were computedby the Genetics Computer Group program PILEUP. Overall sequenceidentity for the branch point at the far left is about 21%. Sequenceswere obtained from GenBank: hRaml, L20852; rRaml, L19931;cRaml, U13945; hGlvrl, L20859; cGlvrl, U13946; mGlvrl, M73696;B0222.2 and B0222.3, U50312; YGO4, P45268; M. thermoauto trophi-cum ORF, S08522; Pho4, M31364; YB8I, P38361; PitA, P37308; PitB,P43676; PitH, P41132; Pit, U15187.

tropic vector transduction is blocked by the presence ofAM-MLV, but is unaffected by the presence of ecotropicMoMLV. Nonreciprocal interference is displayed by the am-photropic and lOAl viruses, where lOAl-pseudotype vectortransduction is unaffected by the presence of amphotropicvirus in the target cells, but amphotropic vector transductionis blocked by the presence of lOAl virus. These data suggestedthat lOAl virus can enter cells by using a different receptorthan that used by amphotropic virus, and that lOAl virus canalso bind to the amphotropic receptor and block amphotropicvirus entry (Fig. 2).

Given that lOAl virus can bind to Raml, we tested theability of Raml to mediate entry of the lOAl-pseudotypeLAPSN vector. We also tested Glvrl due to its similarity toRaml and thus the possibility that Glvrl was the alternativereceptor for entry of lOAl virus. We found that expression ofhuman Raml, rat Raml, human Glvrl, or mouse Glvrlrendered Chinese hamster ovary (CHO) cells susceptible tolOAl-pseudotype LAPSN vector transduction (Table 4). Thus,lOAl virus can bind and enter cells by using either of twodifferent retrovirus receptors. Amphotropic virus can enterCHO cells expressing human or rat Raml, but not thoseexpressing human or mouse Glvrl (data not shown). Theseresults explain the nonreciprocal interference observed be-tween lOAl and amphotropic retroviruses.

Table 3. Nonreciprocal interference between 1OAl andamphotropic retroviruses

Vector titer, FFU/ml

LAPSN 3T3 + 3T3 + 3T3 +pseudotype 3T3 AM-MLV 1OAl MoMLV

Amphotropic 5 x 106 40 3 4 x 1061OAl 7 x 106 6 x 106 2 x 102 6 x 106Ecotropic 3 x 106 2 x 106 2 x 106 40

The LAPSN vector encodes alkaline phosphatase and neomycinphosphotransferase. LAPSN vector with an amphotropic, 1OA1, orecotropic pseudotype was made by using PA317 retrovirus packagingcells, wild-type 1OAl virus, or PE501 packaging cells, respectively.Transduction was measured by staining cells for alkaline phosphatase2 days after exposure to the vectors. Data are from Miller and Chen(32).

Some Receptors Can Promote Entry of Retroviruses ThatNormally Utilize Independent Receptors

In human cells, gibbon ape leukemia virus (GALV) exclusivelyuses Glvrl for entry and amphotropic retrovirus exclusivelyuses Raml. These facts are reflected in the assignment of theseviruses to separate interference groups for human cells (Table1). However, analysis of the hamster homolog of Raml showsthat it can function as a receptor for GALV or amphotropicretrovirus (34). In addition, certain chimeric receptors madebetween Raml and Glvrl can also function as receptors forboth viruses (Fig. 3). In this example, the hybrid receptor RRGpromotes transduction by GALV or amphotropic pseudotypevectors with efficiencies similar to those found for GALVvector transduction of cells expressing the normal humanGlvrl receptor (GGG) or amphotropic vector transduction ofcells expressing the normal rat Raml (RRR) receptor. Thus,the ability of retroviruses to utilize certain receptor homologsfor entry, and therefore the interference pattern of these

UNINFECTEDCELLS

AMINFECTEDCELLS

1 OAlINFECTEDCELLS

ll1. Rann1

.-~- Glvrl

'KNAM'0 En v ()

EnvOAl Viruses/-\, Env

FIG. 2. Nonreciprocal interference between 1OAl and ampho-tropic retroviruses.

A.N-. p9bFk

1 1

r- N., -% -." /0-*Or-N

1 1

e4n.4.-' fin...,

Colloquium Paper: Miller

1

Dow

nloa

ded

by g

uest

on

July

22,

202

1

Proc. Natl. Acad. Sci. USA 93 (1996)

Table 4. A 1OAl-pseudotype retroviral vector can utilize Raml orGlvrl for cell entry

Vector titer,Receptor Species FFU/ml

Ram-1 Human 1 x 106Rat 3 x 105

Glvr-1 Human 6 x 105Mouse 7 x 105

None <500

Retrovirus receptor cDNAs were expressed by using the retroviralvector LXSN. CHO cells were seeded at 2 x 104 per well (d = 3.5 cm)in multiwell dishes on day 1. On day 2, cells were cotransfected with2.5 jig of 13-galactosidase expression plasmid and 2.5 ,ug of the receptorexpression construct. On day 3, one set of dishes was stained for,B-galactosidase to assess transfection efficiency, whereas the other setwas infected with 2 ,u of the LAPSN vector pseudotyped with the 1OAlretrovirus in the presence of 50% medium conditioned by CHO cells.On day 4, cells were stained for alkaline phosphatase and foci ofstained cells were counted. Transfection efficiencies were similar forall constructs, as measured by ,B-galactosidase staining. Data are fromMiller and Miller (33).

retroviruses, will depend on the specific receptors expressed inthe target cells.

Retrovirus Interactions with Homologous Receptor Proteinsfrom Other Species Are Complex

Limitations to retrovirus entry into cells from different speciesare due to variable expression of the receptor or its homologs,or to amino acid sequence differences or posttranslationalmodifications in the receptor homologs in different speciesthat inhibit virus binding or entry. The former mechanism forvirus resistance is primarily applicable to different cells fromthe same organisms that express variable levels of the receptor.An example of this is provided by HIV, which for entryrequires the CD4 receptor that is found primarily on Tlymphocytes and not on cells from many other tissues. Manyexamples of the latter mechanism of virus resistance have beenfound for cells from different species, in which a receptor

homolog is expressed but is nonfunctional. For example,ecotropic retroviruses infect rodent cells, but do not infecthuman cells, even though human cells express a homolog of themurine ecotropic receptor that is 87% identical to the mouseprotein. In this case, only two amino acid changes are requiredto convert the human protein into a functional receptor, or torender the mouse protein nonfunctional as an ecotropic ret-rovirus receptor (35, 36).Other examples of virus restriction in different species are

provided by viruses that utilize Pit receptor family members forentry. A simple example is the restriction of GALV entry intomouse cells, which, like the restriction to ecotropic virusinfection of human cells, is not due to lack of receptor homologexpression, but to minor changes in mouse Glvrl comparedwith human Glvrl.A more complicated example is provided by 1OAl receptor

usage in different species. As noted above, 1OAl-pseudotypevirus can use human Raml or human Glvrl to enter CHO cells(Table 4) and to enter human cells (32). However, in rat cells1OAl virus infection does not block GALV-pseudotype vectortransduction (Table 5), showing that 1OAl cannot bind to orenter cells by using the rat GALV receptor. Indeed, 1OAl hasthe same interference properties as amphotropic retrovirus inrat cells, and thus uses the amphotropic virus receptor forentry. Thus, 1OAl virus behaves like an amphotropic virus inrat cells, but like a combination of an amphotropic virus andGALV in human cells. Therefore, the interference and recep-tor utilization properties of 1OAl virus depends on the cell typeused for the analysis.

Env Amino Acid Sequence Is Not Predictiveof Receptor Utilization

It is not clear how retroviruses have evolved to utilize such adiverse group of receptors for entry into cells. One possibilityis that mutations in the Env protein promote weak binding andentry through interaction with new receptors, and if the newreceptor specificity is beneficial for virus survival, selectivepressure favors further mutations that promote more efficient

GALV vector,NIH 3T3 cells

Amphotropic vector,CHO cells

(Ram-1) RRR

(Glvr-1) GGG

GRR

GGR

RGG

RRG

GRG

GGrG

I I

I1I I

Vector-transduced cell foci,B-galactosidase-positive cell foci

Vector-transduced cell focip-galactosidase-positive cell foci

FIG. 3. GALV- and amphotropic-pseudotype vector transduction of cells expressing hybrid Ram-1/Glvr-1 receptors. CHO or NIH 3T3 cellswere seeded in 3.5-cm dishes at 2 x 104 cells per dish. On day 2, hybrid constructs cloned in the retroviral vector LXSN were cotransfected (1:1)with a plasmid encoding ,B-galactosidase (2.5 ,ug each). Parallel dishes were stained for ,3-galactosidase, whereas the other set was infected withthe retroviral vector LAPSN pseudotyped by amphotropic Env [LAPSN(PA317)] or a GALV Env [LAPSN(PG13)] at a multiplicity of infectionof about 2. Cells were stained for alkaline phosphatase activity on day 5. Values are the number of vector-transduced (alkaline phosphatase-positive)foci divided by the number of ,B-galactosidase positive foci. Results are averages of duplicate dishes. The experiment was performed three timeswith similar results. Data are from Miller and Miller (33).

Construct

I ----r- I I

11410 Colloquium Paper: Miller

I

10-4 10-3 10-2 -10-1 10-3 10-2 10-1 loo

Dow

nloa

ded

by g

uest

on

July

22,

202

1

Proc. Natl. Acad. Sci. USA 93 (1996) 11411

Table 5. 1OAl-pseudotype LAPSN vector can use theamphotropic receptor, but not the GALV receptor,for entrv into 208F rat cells

LAPSN Vector titer, FFU/mlpseudotype 208F 208F + AM-MLV 208F + GALV

1OAl 1 x 107 500 1 X 107Amphotropic 2 x 106 200 2 x 106GALV 2 x 105 2 x 105 100

LAPSN vector with a 1OA1, amphotropic, orGALV pseudotype wasmade by using wild-type 1OAl virus, PA317 retrovirus packaging cells,or PG13 packaging cells, respectively. Transduction was measured bystaining cells for alkaline phosphatase 2 days after exposure to thevectors. Data are from Miller and Chen (32).

utilization of the new receptor. Another model involves re-placement of portions of the retroviral Env protein withportions of cellular proteins that recognize other cell-surfaceproteins. For example, incorporation of a portion of theerythropoietin protein into an existing Env protein might allowthe erythropoietin receptor to function as a new target forretrovirus binding and entry. This model for acquisition of newreceptor specificities parallels the process of retrovirus acqui-sition of cellular oncogenes. Just as oncogenes can improveretrovirus replication and survival, the ability to utilize newreceptors for cell entry should improve the survival potentialof a retrovirus. In addition, the existence of cellular proteinsthat bind cell-surface molecules with high affinity seems amore likely source for the development of radically alteredretrovirus receptor specificities compared with random Envmutations.

Predictions of the model for alteration of retrovirus receptorutilization by incorporation of cellular genes is that retrovi-ruses that utilize similar receptors should contain more closelyrelated Env proteins than retroviruses that use other receptors,and Env proteins should contain regions of similarity withcellular proteins. A comparison of Env proteins from severalretroviruses shows that retroviruses that use different recep-tors can be more highly related than those that use the samereceptor (Fig. 4). For example, GALV, lOA1, and subgroup Bfeline leukemia virus (FeLV-B) all can use Glvrl as a receptor,but FeLV-B is more closely related to FeLV-A and FAIDS,which use different receptors, than it is to GALV or the lOAlvirus. Likewise, the lOAl virus is more closely related to thepolytropic (MoMCF; Moloney mink cell focus-forming virus),xenotropic (NZB), and ecotropic (FrMLV, MoMLV, andAKV) retroviruses than to GALV or FeLV-B. The sameoverall dendrogram is obtained even if one compares only the200 amino acids at the amino termini of the processed Envproteins that are directly involved in receptor binding (notshown). Comparison of these amino terminal receptor-bindingregions of the Env proteins from viruses in the FeLV, MLV,and GALV groups reveals a similar amino acid frameworksurrounding two variable regions, with no common features inthe variable regions that would predict the receptor utilizationpattern (37). In addition, no similarities have been foundbetween regions of retroviral Env proteins and already se-quenced cellular proteins. Thus, the data to date favor amutational origin for new receptor specificities rather than amodel involving acquisition of cellular proteins that can bindnew cell-surface receptors. The mutational model can explainthe diversity in Env sequences among viruses that recognizethe same receptor as the result of convergent evolution ofdifferent parental retroviruses.Another argument in favor of a mutational model for

acquisition of new receptor specificities is the finding that smallchanges in a virus Env protein can result in a new receptorspecificity. For example, no more than six amino acid changesare required to convert an amphotropic Env, which targetsRaml, to one having the receptor utilization properties of

r FAIDS

FeLV-A

FeLV-B

1 QAlAM-MLV

MoMCF

NZB

FrMLV

MoMLV

AKV

GALV

FIG. 4. Dendrogram of amino acid similarities between differentretroviral Env proteins. Distances between sequences were computedby the Genetics Computer Group program PILEUP. Overall sequenceidentity for the branch point at the far left is about 42%. Sequenceswere obtained from GenBank: FAIDS, feline AIDS virus, M18247;FeLV-A, feline leukemia virus subgroup A, M12500; FeLV-B, felineleukemia virus subgroup B, X00188; 1OA1, 1OAl MLV, M33470;AM-MLV, M33469; MoMCF, Moloney mink cell focus-forming virus,J02254; NZB, NZB MLV, K02730; FrMLV, Friend MLV, Z11128;MoMLV, J02255; AKV, AKV MLV, J01998; and GALV, M26927.

IOAI virus, which targets Raml or Glvrl for entry (32, 38).However, because Raml and Glvrl are related proteins, thischange does not represent a dramatic switch in receptorspecificity, and it will be interesting to see if minor amino acidchanges in Env proteins can result in more dramatic changesin receptor utilization.

Endogenous Synthesis of Env Protein Can BlockRetrovirus Entry

Retrovirus receptors can be rendered nonfunctional due toblockade by Env protein synthesized by a replication-competent retrovirus. This is the basis for the virus interfer-ence discussed above. Interference with receptor function canalso result from synthesis of Env proteins from endogenousretroviruses or fragments of retroviruses that are inherited inanimals. A well-documented example of this phenomenoninvolves the Fv-4 locus in mice (39), the phenotype of which isdue to a truncated endogenous ecotropic retrovirus that ismissing the gag and part of the pol genes, but which containsan intact env gene. Synthesis of this endogenous env geneproduct in mouse tissues blocks infection and leukemia causedby ecotropic retroviruses by blocking the ecotropic retrovirusreceptor. Other examples of this phenomenon have been foundfor avian leukosis viruses in chickens (40) and for MCF virusesin mice (41).

Hamster Cells Secrete a Factor That Blocks RetrovirusInfection and a Similar Factor Is Found in Hamster Serum

CHO cells are resistant to infection by many retroviruses. Inmost cases, this resistance can be abrogated by prior treatmentof the cells with the glycosylation inhibitor tunicamycin. Theresistance to GALV and amphotropic retrovirus infection isdue to a secreted protein factor that blocks infection (42).Thus, addition of CHO cell-conditioned medium to CHO cellsthat have been made susceptible to infection by treatment withtunicamycin blocks infection by GALV and amphotropicviruses. In contrast, addition of the conditioned medium doesnot block infection of tunicamycin-treated CHO cells by anecotropic retrovirus, showing that the medium is not simplytoxic, and that the effect is specific for retroviruses withparticular Env proteins. Interestingly, the CHO cell-

Colloquium Paper: Miller

Dow

nloa

ded

by g

uest

on

July

22,

202

1

Proc. Natl. Acad. Sci. USA 93 (1996)

conditioned medium does not block amphotropic vector trans-duction of human or mouse cells, nor does it block transductionof CHO cells made susceptible to amphotropic vector trans-duction by prior introduction of genes expressing human or ratRaml (21, 22), indicating that the factor can bind to and blockthe hamster receptor but not the human or rat receptors.Hamster serum contains a similar factor that can block

retrovirus infection of tunicamycin-treated CHO cells (Table6). Addition of 5% serum from Chinese hamsters completelyblocked transduction by an amphotropic vector, and 12.5%serum from Syrian hamsters also significantly inhibited trans-duction. In contrast, addition of 25% fetal bovine serum hadno effect on transduction of tunicamycin-treated CHO cells bythe amphotropic vector. Like the CHO cell-conditioned me-dium, the hamster sera had no effect on amphotropic vectorinfection of HeLa human cells (Table 6), indicating a speciesspecificity for the factor.

Thus, hamster serum contains a similar, potentially identi-cal, inhibitor of retrovirus infection to that secreted from CHOcells. Based on the principle of virus interference, the factorcould be a fragment of an Env protein that is secreted fromcells and blocks infection by binding to the virus receptor.Alternatively, it could be a normal cellular protein that nat-urally interacts with the phosphate transporter that serves asa receptor for GALV and amphotropic viruses resulting in a

block to infection.

Receptor Glycosylation Can Affect Retrovirus Entry

Retroviral interference can by reversed by inhibitors of gly-cosylation that affect Env processing and subsequent bindingto virus receptors (44). In addition, inhibitors of glycosylationcan have direct effects on a retrovirus receptor to modulateinfection. For example, the ecotropic retrovirus receptor ho-molog on Mus dunni cells functions as a receptor for mostecotropic retroviruses with the exception of MoMLV. Tuni-camycin treatment renders the cells susceptible to infection byMoMLV, and alteration of a single amino acid in the Musdunni receptor to prevent glycosylation at that site results in areceptor that promotes MoMLV infection (45). Thus, subtlechanges in receptor glycosylation can have a major effect onthe ability of a retrovirus to utilize the receptor for cell entry.

Vector Pseudotypes Available for GeneTransfer Applications

Given the complicated factors that govern retrovirus entry intocells from different tissues and different species, it is helpfulthat there are a wide range of retrovirus packaging cell linesthat are available for production of retroviral vectors with

Table 6. Hamster sera inhibit amphotropic vector infection oftunicamycin-treated CHO cells but not HeLa cells

Target Vector titer, Inhibition,cells Additional serum CFU/ml %

CHO None 1 x 103Chinese hamster (5%) <10 >99Syrian hamster (12.5%) 35 97Fetal bovine (25%) 2 x 103

HeLa None 3 x 105Chinese hamster (5%) 2 x 105Syrian hamster (12.5%) 3 X 105

The indicated target cells were plated at 105 per 6-cm dish on day1, infected with an amphotropic-pseudotype vector carrying the neogene on day 2 in the presence of culture medium containing 5% FBS(no additional serum), 5% FBS plus 5% Chinese hamster serum, 5%

FBS plus 12.5% Syrian hamster serum, or 5% FBS plus 25% additionalFBS, and G418-resistant colony formation was measured. Inhibition isreported only when >50%. Data are from Miller and Miller (43).

different pseudotypes. Approximate host ranges of packagingcells derived using mammalian retroviruses are shown in Table7. A listing of specific packaging cell lines can be found in ref.46. The best vector pseudotype for a given application will befurther influenced by the specific target tissue and the expres-sion of suitable levels of receptors with proper posttransla-tional modifications to allow efficient virus entry. For example,Glvrl is overexpressed compared with Raml in hematopoieticcells (23), and vectors with a GALV pseudotype have beenfound to transduce hematopoietic cells more efficiently thanthe same vectors with an amphotropic pseudotype (47, 48).

Conclusions

Retroviruses utilize a diverse set of proteins for cell entry.Single proteins are apparently required for binding and entryof most retroviruses, although two proteins are required forHIV. Although virus entry is dependent on the level ofreceptor expression in particular cells, there are many otherfactors that govern utilization of a receptor or its homologs indifferent species. Subtle alterations in the amino acid sequenceof receptor homologs in different species can dramaticallyaffect virus entry, either as a direct result of changes in theprimary amino acid sequence or as an indirect result of alteredprotein modifications such as glycosylation. Indeed, restrictedvirus host range is not generally due to a lack of expression ofhomologous receptor proteins, but is more often related tominor alterations in these proteins. In addition, soluble pro-teins secreted by some cells and present in some animals, andretroviral Env proteins synthesized from replication-competent viruses or from endogenous virus sequences, canblock receptor utilization. These are all important consider-ations in the design of retroviral vectors for gene transfer incultured cells and in animals.

Recently it has become clear that certain retroviruses canuse more than one receptor for entry into some cell types, andsome receptors can promote entry of retroviruses that nor-mally utilize different receptors in other cells. These resultsseriously complicate attempts to classify retroviruses intogroups based on receptor utilization, as determined by inter-ference analysis, because these groupings depend on theparticular receptors expressed on the cell type used for theanalysis. In fact, this problem was appreciated long before themolecular basis for this phenomenon was determined (49).

Further development of retroviral vectors for gene transferapplications has involved the incorporation of Env proteins

Table 7. Host range of selected retrovirus packaging cells

Targetcells Ecotropic Amphotropic GALV RD114 1OAl

Mouse + + - - +Rat + + + - +Hamster - +/_ + - +Rabbit - + +Mink - + + +Cow - +/_ +Cat - + + + +Dog - + + + +Monkey - + + + +Human - + + + +Chicken - +/- +

The ability of vectors from the indicated packaging cells to transducetarget cells from the indicated species is shown as + if the cells can betransduced, - if the cells cannot be transduced, and as +/- if thereis poor transduction or if there is variable transduction of differentcells from the indicated species. These evaluations are only intendedas a general guide because there are many factors that can influencetransduction rates, including the particular animals from within eachspecies that are the source of the target cells.

11412 Colloquium Paper: Miller

Dow

nloa

ded

by g

uest

on

July

22,

202

1

Proc. Natl. Acad. Sci. USA 93 (1996) 11413

from other virus families, such as the vesicular stomatitis virusG protein (3, 50) and efforts to alter the receptor specificity ofexisting retroviral Env proteins by the incorporation of peptideor antibody domains that can bind to other cell-surface pro-teins (51, 52). An understanding of the principles governingcell entry by naturally occurring retroviruses will help in thedesign and application of these strategies.A fascinating aspect of retroviruses is their utilization of

diverse proteins for cell entry. The analysis presented herefavors a mutational basis for retrovirus evolution to utilize newreceptors, rather than acquisition and expression of cellularproteins that naturally bind to cell-surface receptors, but moreinformation is needed to resolve this issue. Perhaps analysis ofadditional naturally occurring retroviruses and their receptorswill reveal a clear example of acquisition of a cellular gene thatenables utilization of a new cell-surface receptor for entry.Answers to these questions have important implications for thedesign of retroviral vectors with novel receptor specificities,and for the evolution of retroviruses, which are importantagents of disease in humans and in animals.

I thank Michael Emerman and Greg Wolgamot for comments onthis manuscript. This work was supported by grants from the NationalHeart, Lung, and Blood Institute and the National Institute ofDiabetes and Digestive and Kidney Diseases of the National Institutesof Health.

1. Halbert, C. L., Alexander, I. E., Wolgamot, G. M. & Miller, A. D.(1995) J. Virol. 69, 1473-1479.

2. Kotin, R. M. & Berns, K. I. (1989) Virology 170, 460-467.3. Naldini, L., Blomer, U., Gallay, P., Ory, D., Mulligan, R., Gage,

F. H., Verma, I. M. & Trono, D. (1996) Science 272, 263-267.4. Miller, D. G., Adam, M. A. & Miller, A. D. (1990) Mol. Cell. Bio.

10, 4239-4242.5. Sommerfelt, M. A. & Weiss, R. A. (1990) Virology 176, 58-69.6. Kewalramani, V. N., Panganiban, A. T. & Emerman, M. (1992)

J. Virol. 66, 3026-3031.7. Sommerfelt, M. A., Williams, B. P., McKnight, A., Goodfellow,

P. N. & Weiss, R. A. (1990) J. Virol. 64, 6214-6220.8. Garcia, J. V., Jones, C. & Miller, A. D. (1991) J. Virol. 65,

6316-6319.9. Kaelbling, M., Eddy, R., Shows, T. B., Copeland, N. G., Gilbert,

D. J., Jenkins, N. A., Klinger, H. P. & O'Hara, B. (1991) J. Virol.65, 1743-1747.

10. Sommerfelt, M. A., Williams, B. P., Clapham, P. R., Solomon, E.,Goodfellow, P. N. & Weiss, R. A. (1988) Science 242, 1557-1559.

11. Isobe, M., Huebner, K., Maddon, P. J., Littman, D. R., Axel, R.& Croce, C. M. (1986) Proc. Natl. Acad. Sci. USA 83,4399-4402.

12. Dalgleish, A. G., Beverley, P. C., Clapham, P. R., Crawford,D. H., Greaves, M. F. & Weiss, R. A. (1984) Nature (London)312, 763-767.

13. Klatzmann, D., Champagne, E., Chamaret, S., Gruest, J., Gue-tard, D., Hercend, T., Gluckman, J. C. & Montagnier, L. (1984)Nature (London) 312, 767-768.

14. Feng, Y., Broder, C. C., Kennedy, P. E. & Berger, E. A. (1996)Science 272, 872-877.

15. Deng, H., Liu, R., Ellmeier, W., Choe, S., Unutmaz, D., Burkhart,M., Di Marzio, P., Marmon, S., Sutton, R. E., Hill, C. M., Davis,C. B., Peiper, S. C., Schall, T. J., Littman, D. R. & Landau, N. R.(1996) Nature (London) 381, 661-666.

16. Dragic, T., Litwin, V., Allaway, G. P., Martin, S. R., Huang, Y.,Nagashima, K. A., Cayanan, C., Maddon, P. J., Koup, R. A.,Moore, J. P. & Paxton, W. A. (1996) Nature (London) 381,667-673.

17. Sattentau, Q. J., Clapham, P. R., Weiss, R. A., Beverley, P. C.,Montagnier, L., Alhalabi, M. F., Gluckman, J. C. & Klatzmann,D. (1988) AIDS 2, 101-105.

18. Albritton, L. M., Tseng, L., Scadden, D. & Cunningham, J. M.(1989) Cell 57, 659-666.

19. Kim, J. W., Closs, E. I., Albritton, L. M. & Cunningham, J. M.(1991) Nature (London) 352, 725-728.

20. Wang, H., Kavanaugh, M. P., North, R. A. & Kabat, D. (1991)Nature (London) 352, 729-731.

21. Miller, D. G., Edwards, R. H. & Miller, A. D. (1994) Proc. Natl.Acad. Sci. USA 91, 78-82.

22. van Zeijl, M., Johann, S. V., Closs, E., Cunningham, J., Eddy, R.,Shows, T. B. & O'Hara, B. (1994) Proc. Natl. Acad. Sci. USA 91,1168-1172.

23. Kavanaugh, M. P., Miller, D. G., Zhang, W., Law, W., Kozak,S. L., Kabat, D. & Miller, A. D. (1994) Proc. Natl. Acad. Sci. USA91, 7071-7075.

24. O'Hara, B., Johann, S. V., Klinger, H. P., Blair, D. G., Rubinson,H., Dunn, K. J., Sass, P., Vitek, S. M. & Robins, T. (1990) CellGrowth Differ. 1, 119-127.

25. Ban, J., Portetelle, D., Altaner, C., Horion, B., Milan, D.,Krchnak, V., Burny, A. & Kettmann, R. (1993) J. Virol. 67,1050-1057.

26. Ban, J., Truong,; A. T., Horion, B., Altaner, C., Burny, A.,Portetelle, D. & Kettmann, R. (1994) Arch. Virol. 138, 379-383.

27. Bates, P., Young, J. A. & Varmus, H. E. (1993) Cell 74, 1043-1051.

28. Hosie, M. J., Willett, B. J., Dunsford, T. H., Jarrett, 0. & Neil,J. C. (1993) J. Virol. 67, 1667-1671.

29. Willett, B. J., Hosie, M. J., Jarrett, 0. & Neil, J. C. (1994)Immunology 81, 228-233.

30. Gazdar, A. F., Oie, H., Lalley, P., Moss, W. W. & Minna, J. D.(1977) Cell 11, 949-956.

31. Adamson, M. C., Silver, J. & Kozak, C. A. (1991) Virology 183,778-781.

32. Miller, A. D. & Chen, F. C. (1996) J. Virol. 70, 5564-5571.33. Miller, D. G. & Miller, A. D. (1994) J. Virol. 68, 8270-8276.34. Wilson, C. A., Farrell, K. B. & Eiden, M. V. (1994) J. Virol. 68,

7697-7703.35. Albritton, L. M., Kim, J. W., Tseng, L. & Cunningham, J. M.

(1993) J. Virol. 67, 2091-2096.36. Yoshimoto, T., Yoshimoto, E. & Meruelo, D. (1993) J. Virol. 67,

1310-1314.37. Battini, J. L., Heard, J. M. & Danos, 0. (1992) J. Virol. 66,

1468-1475.38. Ott, D. & Rein, A. (1992) J. Virol. 66, 4632-4638.39. Gardner, M. B., Kozak, C. A. & O'Brien, S. J. (1991) Trends

Genet. 7, 22-27.40. Robinson, H. L., Astrin, S. M., Senior, A. M. & Salazar, F. H.

(1981) J. Virol. 40, 745-751.41. Bassin, R. H., Ruscetti, S., Ali, I., Haapala, D. K. & Rein, A.

(1982) Virology 123, 139-151.42. Miller, D. G. & Miller, A. D. (1992) J. Virol. 66, 78-84.43. Miller, D. G. & Miller, A. D. (1993) J. Virol. 67, 5346-5352.44. Rein, A., Schultz, A. M., Bader, J. P. & Bassin, R. H. (1982)

Virology 119, 185-192.45. Eiden, M. V., Farrell, K. & Wilson, C. A. (1994) J. Virol. 68,

626-631.46. Miller, A. D., Miller, D. G., Garcia, J. V. & Lynch, C. M. (1993)

Methods Enzymol. 217, 581-599.47. von Kalle, C., Kiem, H. P., Goehle, S., Darovsky, B., Heimfeld,

S., Torok-Storb, B., Storb, R. & Schuening, F. G. (1994) Blood 84,2890-2897.

48. Bauer, T. R., Miller, A. D. & Hicksein, D. D. (1995) Blood 86,2379-2387.

49. Chesebro, B. & Wehrly, K. (1985) Virology 141, 119-129.50. Burns, J. C., Friedmann, T., Driever, W., Burrascano, M. & Yee,

J. K. (1993) Proc. Natl. Acad. Sci. USA 90, 8033-8037.51. Kasahara, N., Dozy, A. M. & Kan, Y. W. (1994) Science 266,

1373-1376.52. Somia, N. V., Zoppe, M. & Verma, I. M. (1995) Proc. Natl. Acad.

Sci. USA 92, 7570-7574.

Colloquium Paper: Miller

Dow

nloa

ded

by g

uest

on

July

22,

202

1