Envelope Targeting: Hemagglutinin Attachment Specificity Rather than Fusion Protein Cleavage-Activation Restricts Tupaia Paramyxovirus Tropism

To engineer a targeting envelope for gene and oncolytic vectordelivery, we characterized and modified the envelope proteinsof Tupaia paramyxovirus (TPMV), a relative of the morbilli-and henipaviruses that neither infects humans nor has cross-reactiverelatives that infect humans. We completed the TPMV genomicsequence and noted that the predicted fusion (F) protein cleavage-activationsite is not preceded by a canonical furin cleavage sequence.Coexpression of the TPMV F and hemagglutinin (H) proteins inducedfusion of Tupaia baby fibroblasts but not of human cells, afinding consistent with the restricted TPMV host range. To identifythe factors restricting fusion of non-Tupaia cells, we initiallyanalyzed F protein cleavage. Even without an oligo- or monobasicprotease cleavage sequence, TPMV F was cleaved in F1 and F2subunits in human cells. Edman degradation of the F1 subunityielded the sequence IFWGAIIA, placing the conserved phenylalaninein position 2, a novelty for paramyxoviruses but not the causeof fusion restriction. We then verified whether the lack ofa TPMV H receptor limits fusion. Toward this end, we displayeda single-chain antibody (scFv) specific for the designated receptorhuman carcinoembryonic antigen on the TPMV H ectodomain. TheH-scFv hybrid protein coexpressed with TPMV F mediated fusionof cells expressing the designated receptor, proving that thelack of a receptor limits fusion and that TPMV H can be retargeted.Targeting competence and the absence of antibodies in humansdefine the TPMV envelope as a module to be adapted for ferryingribonucleocapsids of oncolytic viruses and gene delivery vectors.

INTRODUCTION

Top

Introduction

Several viruses, including the paramyxoviruses Measles virus(MV) (47), Mumps virus (MuV) (1), and Newcastle disease virus(NDV) (32), are being developed as novel agents for the treatmentof human cancer (3, 39). Neutralizing antibodies, even if theydo not abolish efficacy of human viruses in some xenograft modelsof oncolysis (14) or viral replication after challenge of hostswith preexisting immunity (36), may reduce their efficacy incurrently open clinical trials (9, 16, 55, 58). To circumventneutralizing antibodies prevalent in the general populationthrough vaccination or natural disease, we envisage enclosingthe replicative units of human viruses in envelopes of animalviruses that do not have cross-reactive relatives in humans.

We have begun exploring the use of the Tupaia paramyxovirus(TPMV) envelope glycoproteins as a vector delivery module. Tupaiasor tree shrews are small mammals widely present in southeastAsia. After having been classified as either primates or insectivores,they were finally assigned to a separate order, Scandentia.Tupaias have yielded a rich crop of viruses, including a herpesvirus(2), adenovirus (43), and rhabdovirus (46) that have recentlybeen characterized at the genomic level. TPMV was isolated fromspontaneously degenerating primary kidney cells of an apparentlyhealthy tree shrew (Tupaia belangeri) (53, 54). The virus isgrowing in a trypsin-independent manner in Tupaia cells butnot in any other cell line tested thus far. The cytopathic effectincludes multinucleated syncytia followed by cell lysis. Thenucleotide sequence of the N and P genes revealed that the virusis most closely related to the morbilli- and henipaviruses (54),but the glycoproteins have not been characterized.

The glycoproteins are important determinants of paramyxovirustropism: the attachment protein because it targets one or morereceptors and the fusion (F) protein because it can be activatedby tissue-specific proteases. The attachment protein is a typeII transmembrane protein named hemagglutinin (H), hemagglutinin-neuraminidase(HN) or, when the protein does not have these functions, glycoprotein(G). Viruses relying on oligosaccharides as receptors have neuraminidaseactivity to eliminate these from the cell surface at late infectionsstages and allow efficient particle release. Viruses with peptidylicepitopes of specific proteins as receptors do not need neuraminidaseactivity.

The F protein of paramyxoviruses is a trimeric type I transmembraneprotein that executes membrane fusion. All F proteins characterizedthus far are cleaved in an N-terminal F2 and a C-terminal F1fragment; a hydrophobic fusion peptide is amino terminal tothe F1 fragment. The cleavage site is preceded by either oneor several arginines and lysines, i.e., it is oligo- or monobasic(18). Proteins with oligobasic cleavage sites are cleaved inthe trans-Golgi network by the ubiquitous protease furin orrelated proteases, whereas proteins with monobasic cleavagesites are cleaved by trypsin-like proteases outside the cell.The nature of the protease cleavage site and the localizationof the F protein activating protease can determine viral tropismand pathogenicity. The proteases activating Sendai virus (SeV),tryptase Clara and mini-plasmin, are only present in the respiratorytract of the murine host, thereby limiting the infection tothis organ (27, 51). F protein cleavage is a tropism determinantalso in the avian paramyxovirus NDV: the cleavage sites of pathogenicstrains are oligobasic, whereas those of nonpathogenic strainsare monobasic (28, 33, 45). Paramyxoviruses with monobasic cleavagesites often require trypsin for efficient propagation in culturedcells. As an exception, the recently characterized Hendra virus(HeV) and Nipah virus (NiV) have only one lysine or argininepreceding the fusion peptide, respectively, but neverthelesspropagate efficiently in a variety of cultured cells withouttrypsin (64) and may be cleaved inside the cells by as-yet-unidentifiedproteases (31).

We present here a functional analysis of the TPMV glycoproteinsand discuss their sequences that we deposited in the GenBank(AF079780). We show that F-protein cleavage activation is nota limiting factor in human cells, even if the structure of theF1 subunit amino terminus is unprecedented. We also show thata specificity domain added to the TPMV H ectodomain is sufficientto mediate fusion of target cells, thereby defining receptorrecognition as tropism determinant.

MATERIALS AND METHODS

Top

Materials and Methods

Cells and viruses. TPMV was propagated in Tupaia baby fibroblasts (TBF) (11) asdescribed previously (54). 293T, Vero, and HT1080 were fromthe American Type Culture Collection. MC38cea cells (40) werekindly provided by Jeffrey Schlom, National Institutes of Health.All cells were grown in Dulbecco modified Eagle medium containing10% fetal calf serum with penicillin-streptomycin except forthe Vero cells that were supplemented with 5% fetal calf serum.

Nucleotide sequencing and computer-assisted sequence analysis. To complete the TPMV genomic sequence, degenerated oligonucleotideprimers corresponding to conserved regions in the M gene ofthe morbilliviruses and henipaviruses were constructed, takinginto account the TPMV codon usage in the N and P genes. Theseprimers were combined with primers in the P gene in PCR to amplifyand sequence the beginning of the M gene. The M gene sequencewas then completed by 3'-rapid amplification of cDNA ends (RACE).The same strategy was used for the remaining F, H, and L genes.The TPMV trailer sequence was also determined by 3'-RACE. DNAsequencing and RACE procedures were carried out as describedpreviously (54). Protein similarities were calculated with theCLUSTAL W program (52) on the website of the Pôle BioinformatiqueLyonnais (http://npsa-pbil.ibcp.fr/).

Molecular cloning of the TPMV glycoproteins. Total RNA was isolated from infected TBF cells with the RNeasykit (QIAGEN, Hilden, Germany). The F and H coding sequenceswere PCR amplified with the oligonucleotide primers TPMV-F-S(5'-TTTGGGGGATCCCCAAGGATGGCATCACTGCTA-3'), TPMV-F-AS (5'-TTTGGGTCTAGATTATCCACTTATATCTGTACTG-3'),TPMV-H-S (5'-TTTGGGGGATCCATTATGGATTATCATTCACACACG-3'), and TPMV-H-AS(5'-TTTGGGTCTAGATTACTTAGTATTAGGACATGTAC-3') (restriction sitesused for cloning are underlined) and cloned into the eukaryoticexpression vector pCG, resulting in the plasmids pCG-TPMV-Fand pCG-TPMV-H. The F protein was additionally cloned with aFlag tag (DYKDDDDK, one-letter amino acid code) fused to itscytoplasmic tail (pCG-TPMV-F_Flag). The nucleotide sequence ofthe cloned genes differed slightly from the deposited sequence,resulting in two different amino acids in the F (T485A and I502T)and H (V257A and Y313S) proteins. Since the proteins were functionaland the sequences found in different clones, the plasmids werenot corrected.

Retargeted TPMV H. The expression plasmid pCG-TPMV-HX ${alpha}$ CEA that encodes a fusionprotein of TPMV H and a single-chain antibody (scFv) to humancarcinoembryonic antigen (CEA) was generated by amplificationof the H open reading frame (ORF) with the primers TPMV-H-Sand T-HXL-SphI (5'-TTTGGGGCATGCGCGCGCCCCTTCCCTCGATCTTAGTATTAGGACATGTAC-3'),adding the sequence encoding a factor Xa cleavage site and aBssHII site to the 3' end of the H ORF. The modified ORF wasthen cloned by using BamHI and BssHII into the plasmid pCG-H5-XL(15) that contains the coding sequence for the scFv MFE-23 (8)with a 16-amino-acid linker between the V_H and V_L domains (asdescribed in reference 15).

Expression of the TPMV glycoproteins. Various cell lines, including TBF, 293T, HT1080, Vero, and MC38cea,were transfected in 12-well plates with 2 µg of plasmidDNA by using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) andobserved for syncytium formation or lysed in radioimmunoprecipitationassay buffer (150 mM NaCl, 1.0% Triton X-100, 50 mM Tris-HCl[pH 8.0]) with complete protease inhibitor (Roche Diagnostics,Mannheim, Germany) for Western blot analysis.

Generation of antisera and Western blot analysis. The peptides (C)PSHHRSNGHQNHSFSTDISG, corresponding to the TPMV-Famino acids 534 to 553 with an N-terminal cysteine and MDYHSHTTQTGSNET(C)corresponding to the first 15 TPMV H amino acids with a C-terminalcysteine were synthesized by the Mayo Clinic protein core facility,coupled to keyhole limpet hemocyanin, and used to generate peptideantisera in rabbits (Cocalico Biologicals, Reamston, PA). ForWestern blotting, protein lysates were cleared by centrifugationfor 15 min at 5,000 x g at 4°C and mixed with an equal volumeof urea buffer (200 mM Tris [pH 6.8], 8 M urea, 5% sodium dodecylsulfate [SDS], 0.1 mM EDTA, 0.03% bromophenol blue) containing1.5% dithiothreitol. Some protein samples were digested withPNGase F or Endo H (New England Biolabs, Beverly, MA) accordingto the manufacturer's instructions. After denaturation at 65°Cfor 10 min, the proteins were fractionated on SDS-polyacrylamidegels of various concentrations (Bio-Rad, Hercules, CA) and blottedonto polyvinylidene difluoride membranes (Millipore, Billerica,MA). The antipeptide sera were used in a concentration of 1:10,000.The membranes were subjected to enhanced chemiluminescence detectionwith a horseradish-peroxidase conjugated secondary antibodyand the ECL system (Amersham Biosciences, Piscataway, NJ).

Edman degradation. 293T cells were transfected (Lipofectamine 2000; Invitrogen)with the plasmid pCG-TPMV-F_Flag, lysed in radioimmunoprecipitationassay buffer after 36 h, and immunoprecipitated with ANTI-FLAGM2 affinity gel (Sigma). The immunoprecipitated proteins wereseparated by SDS-polyacrylamide gel electrophoresis (PAGE),blotted to a polyvinylidene difluoride membrane (Sequi-Blot;Bio-Rad), and visualized by Coomassie blue staining. Anothermembrane segment was probed with anti-Flag antibodies to identifythe F protein bands. The F0 and F1 protein fragments were excisedand sequenced by Edman degradation (Mayo Clinic protein corefacility) with an Applied Biosystems 492 Procise cLC sequencer(Applied Biosystems Inc, Foster City, CA).

RESULTS

Top

Results

The TPMV genome and the amino acid sequences of the glycoproteins. The completed TPMV genomic sequence was submitted to GenBankby C. A. Tidona and G. Darai (AF079780). We will restrict ourcomments to the general characteristics of the genome and thepredicted glycoprotein sequences. The 17,904-nucleotide genomeis rule-of-six compliant (20, 37) and larger than those of allknown paramyxovirus except the henipaviruses. TPMV shows thetypical paramyxovirus gene order N-P/C/V-M-F-H-L (Fig. 1C),and all genes are separated by the nontranscribed trinucleotideCUU (antigenome). The trailer region is 590 nucleotides long,much larger than all other trailer regions in the family.

View larger version (32K):
[in this window]
[in a new window]
FIG. 1. TPMV genome organization and glycoproteins structure and sequence. (A) Comparison of the H protein cytoplasmic tails of selected paramyxoviruses. The proteins are aligned based on a charged amino acid postulated to be the first cytoplasmic residues. Amino acids in the transmembrane region are italicized. The number of the last amino acid in each sequence is indicated on the right. (B) Schematic drawing of the TPMV H protein. The cytoplasmic tail (CT), transmembrane region (TM), and three predicted N glycosylation sites ( ${blacktriangledown}$ ) are indicated. (C) Map of the TPMV genome. The black bar represents the 17,904-nucleotide single-stranded TPMV antigenome; the eight open arrows indicate the position of the ORFs of the N, P, C, V, M, F, H, and L proteins; and the small vertical bars indicate the positions of the intergenic trinucleotides. (D) Schematic drawing of the TPMV F protein. SP, signal peptide; FP, fusion peptide; TM, transmembrane region; CT, cytoplasmic tail. The black arrow indicates the cleavage site situated between aa 107 and 108; the black triangles ( ${blacktriangledown}$ ) indicate the N glycosylation sites. (E) Alignment of the F2-F1 junction of selected paramyxoviruses. Identical amino acids are indicated by an asterisk, highly conserved amino acids are indicated by a colon, and well conserved amino acids are indicated by a period. Conserved basic amino acids and the phenylalanine/leucine conserved as position one of the fusion peptide in the majority of paramyxoviruses are in boldface. The number of the last amino acid in each sequence is indicated on the right. Abbreviations not introduced in the text: MoV, Mossman virus; HeV, Hendra virus; RPV, Rinderpest virus; CDV, canine distemper virus; PDV, phocine distemper virus; hPIV1, human parainfluenza virus type 1; hPIV3, human parainfluenza virus type 3; SV5, simian virus 5. The number of the last amino acid in each sequence is indicated on the right.

The F gene codes for a protein of 553 amino acids (aa), includingthe signal peptide. As in other paramyxovirus fusion proteins,three hydrophobic regions exist that were named the signal peptide(SP), the fusion peptide (FP), and the transmembrane domain(TM) (Fig. 1D). The TPMV F protein sequence shows the highestsimilarity to those of the recently identified Mossman virus(25) (34% amino acid identity), the morbilliviruses (33.4% identityto MV) and the henipaviruses (31.8% identity to NiV). The fusionpeptide is predicted to begin with a nearly invariable phenylalanine(aa 109) and is highly conserved, including the three glycinesthat have recently been shown to regulate F protein activation(41). However, in contrast to all other known paramyxovirusF proteins, there is neither a mono- nor an oligobasic cleavagesite preceding F109 (Fig. 1E, top line). This raises the possibilitythat the TPMV F protein is cleaved by an unknown protease thatmay or may not exist in cells of mammals other than tree shrews.Four possible N glycosylation sites (consensus sequence NXS/T)are present in the ectodomain of the TPMV F protein: three onthe F2 fragment (N63, N71, and N90) and one on the F1 fragment(N434) (Fig. 1D, triangles).

The attachment protein gene codes for a 665-aa protein. SinceTPMV virions hemagglutinate Tupaia erythrocytes but have noneuraminidase activity (21), this protein is named hemagglutinin(H). H has a predicted molecular mass of 74 kDa without thethree predicted oligosaccharides (N glycosylation sites at N199,N349, and N509; Fig. 1B, triangles). Sequence identity to theattachment proteins of other paramyxoviruses is low (<20%),but the spacing of cysteines and other structurally importantresidues allows accurate alignment (data not shown). It is noteworthythat the 94 residue H cytoplasmic tail has twice as many residuesas any other paramyxovirus attachment protein cytoplasmic tail(Fig. 1A, bottom line).

Characterization of the TPMV glycoproteins. The ORFs of the TPMV F and H proteins were amplified by reversetranscription-PCR from infected TBF cells and cloned into theeukaryotic expression vector pCG (7). The proteins were transientlyexpressed in different cell lines and analyzed by Western blottingwith rabbit antisera. To raise these antisera, peptides correspondingto aa 534 to 553 of the F protein and aa 1 to 15 of the H proteinwere synthesized, coupled to a carrier protein, and used forrabbit immunization.

Two F protein bands with an estimated molecular masses of 65and 48 kDa were detected after expression in Tupaia cells (Fig.2A, lane F). These estimated sizes correspond to the predictedmolecular mass of the uncleaved F0 protein and the F1 fragment,respectively. Since the antibody is directed against the carboxy-terminalcytoplasmic tail, the amino-terminal F2 fragment cannot be detected.This result suggests that the TPMV F protein is cleaved, despitethe absence of a mono- or oligobasic sequence upstream of theputative fusion peptide. Digestion with PNGase F resulted infaster migration of both fragments, confirming that they areboth glycosylated (Fig. 2A). The shifts in apparent molecularmass of the F0 and F1 bands were consistent with the additionof one oligosaccharide on F1 and several on F0. To determinehow many of the four potential N glycosylation sites in F0 areused, we performed a partial PNGase F digestion. The concentrationof the enzyme was varied so that the number of partially digestedbands could be counted. Figure 2B shows the result of this analysis,detecting four bands in addition to the completely deglycosylatedprotein. Thus, all potential glycosylation sites are used.

View larger version (31K):
[in this window]
[in a new window]
FIG. 2. Western blot analysis of the TPMV glycoproteins. (A) TPMV F protein expressed in TBF cells, fractionated by SDS-10% PAGE, and detected with a peptide antiserum against the cytoplasmic tail. The two bands correspond to the TPMV F0 and F1 fragments. (B) Deglycosylation of the TPMV F0 fragment with increasing concentrations of PNGase F (from left to right) results in five distinct bands, confirming that all four potential N glycosylation sites are used. (C) Cleavage of the TPMV F protein in cells from three species. TPMV F was transiently expressed in different cells lines, fractionated by SDS-10% PAGE, and detected with the anti-F peptide antiserum. The positions of the F0 and F1 fragments are indicated. (D) TPMV F is cleaved membrane proximally, resulting in an F1b fragment. Protein lysates from transfected (central lane) and infected (right lane) TBF cells were fractionated by SDS-15% PAGE, and F proteins were detected using by a peptide antiserum against the cytoplasmic tail. The positions of the F0, F1, and F1b fragments are indicated on the right; the positions of molecular mass markers are indicated on the left in kilodaltons. (E) Expression of the TPMV H protein in 293T cells and glycosylation analysis. Protein extracts of control cells (first lane) or transfected cells treated or not with PNGase F (second and third lanes) were loaded, and the H protein was detected with a peptide antiserum against the cytoplasmic tail.

To determine whether F0 or F1 are transported to the cell surface,we digested total cellular protein extracts with endoglycosidaseH (Endo H). Endo H-resistant glycans are acquired in the Golgicompartment, and proteins containing these glycans are generallytransported rapidly to the cell surface (35). F1 was almostcompletely resistant to Endo H, whereas most F0 was Endo H sensitive(Fig. 2A, right lane). This suggests that the bulk of F1 ison the cell surface, whereas F0 is mainly intracellular.

We then sought to determine whether TPMV F2/F1 cleavage maybe restricted to Tupaia cells and explain tropism. Cleavageoccurred in all cell lines tested, including human (Fig. 2C,293T and HT1080), simian (Fig. 2C, Vero), and murine cells (Fig.2C, MC38cea). Cleavage efficiency varied in the different cellslines and was highest in HT1080 cells that are known to secretematrix-metalloproteinases (MMP) (Fig. 2C, compare the intensityof the F0 and F1 bands). However, the broad-spectrum MMP inhibitorGM 6001 (13) did not inhibit fusion of TBF cells by the TPMVglycoproteins (see below), indicating that proteases other thanMMPs activate TPMV F in these cells (data not shown). Sincecleavage is ubiquitous, it cannot explain tropism.

When protein lysate from transfected or infected TBF cells wasseparated on a more concentrated gel, an additional small bandof ca. 10 kDa was detected (Fig. 2D). We have recently identifiedsimilar F1b fragments generated by a partial membrane-proximalcleavage of the F proteins of the paramyxoviruses MV, Caninedistemper virus (CDV), and MuV, and have shown that partialmembrane-proximal cleavage is important for protein function(60).

We also sought to determine whether the TPMV H protein has theexpected molecular mass and is glycosylated. A Western blotwith H-protein specific antibodies revealed a band of ca. 80kDa (Fig. 2E) corresponding to the predicted molecular mass.Digestion with PNGase F reduced the apparent molecular massof the protein, demonstrating that it is glycosylated (Fig.2E, right lane).

Analysis of the F protein amino terminus by Edman degradation reveals an amino acid upstream of the conserved phenylalanine. Since the TPMV F protein does not have a detectable mono- oroligobasic cleavage site upstream of the fusion peptide, weattempted to determine the exact F2/F1 cleavage site. The F0and F1 proteins were Flag tagged at the common carboxy terminusand purified by using an anti-Flag antibody from human 293Tcells transfected with an expression plasmid. After separationon SDS-PAGE and Western blotting, Coomassie blue-stained bandswere excised and analyzed by Edman degradation. F0 analysisyielded the amino acid sequence EPTPKSQL (one-letter code),corresponding to aa 20 to 27 of the precursor protein. Thisindicates that the signal peptidase cleaves the protein betweenaa 19 and 20 (full sequence of the signal peptide MASLLKTICYIYLITYAKL).The signal peptide was correctly predicted by the computer programSignalP (4). The result for the F1 protein was IFWGAIIA (aa108 to 116), demonstrating that the protein was cleaved betweenaa 107 and 108, leaving an isoleucine upstream of the phenylalaninethat is conserved as the first amino acid of a large majorityof the paramyxovirus F proteins (Fig. 1E).

The TPMV glycoproteins selectively fuse Tupaia cells. TPMV replicates selectively in Tupaia cells and not in cellsof other mammals (54). To verify whether impaired membrane fusionof non-Tupaia cells by the two TPMV glycoproteins may accountfor tropism restriction, TPMV glycoprotein expression plasmidswere cotransfected into Tupaia (TBF) and non-Tupaia (293T, HT1080,Vero, and MC38cea) cell lines. In TBF cells, cotransfectionelicited large, multinucleated syncytia (Fig. 3, bottom panel),a finding comparable to the typical TPMV cytopathic effect inthese cells (not shown). In contrast, there was no syncytiumformation when the proteins were expressed separately (Fig.3, top panel, F protein alone) or in any other cell type tested.Control transfections of primate cells (293T, HT1080, and Vero)with plasmids encoding MV F and H resulted in extensive fusion(data not shown). Cotransfection of TPMV H with MV F and viceversa in TBF and Vero cells did also not lead to syncytium formation,indicating that these proteins cannot functionally interact.Altogether, the experiments described above suggest the lackof a receptor on non-Tupaia cells as a possible reason for thelack of function of the TPMV glycoproteins.

View larger version (138K):
[in this window]
[in a new window]
FIG. 3. The TPMV glycoproteins induce syncytium formation in TBF cells. TBF cells were transfected with a plasmid encoding the F protein alone (upper panel) or with plasmids encoding both TPMV glycoproteins (lower panel) and photographed after 16 h. Coexpression of TPMV F and H results in the formation of multinucleated syncytia.

Receptor-binding conferred by a single-chain antibody displayed on the TPMV H protein restores its fusion support function in non-Tupaia cells. To test whether the lack of a specific receptor accounts forlack of fusion of non-Tupaia cells by the TPMV glycoproteins,we took advantage of the observation that another paramyxoviralattachment protein (MV H) can be targeted to bind designatedreceptors through added specificity domains, including single-chainantibodies (scFv) (15, 42). Analogous to what was done withMV H, we constructed a plasmid coding for a TPMV H protein witha human CEA scFv added to its carboxy terminus (Fig. 4A, shadedbox). H was connected to the scFv by the four-residue linkerIEGR (Fig. 4A, bottom right) that can be cleaved by factor Xaprotease for removal of the displayed specificity domain. Upontransfection of 293T cells and Western blot analysis a proteinof the expected size was detected (Fig. 4B, lane HX ${alpha}$ CEA), andfactor Xa digestion confirmed that this protease can be usedto remove the scFv (Fig. 4B, lane HX ${alpha}$ CEA/FXa).

View larger version (67K):
[in this window]
[in a new window]
FIG. 4. Retargeting of the TPMV H protein. (A) Schematic drawing of the hybrid protein composed of TPMV H and the scFv against CEA. The positions of the transmembrane domain (TM) and the factor Xa cleavage site are indicated by black boxes. In the lower part of the panel the amino acid sequence of the junction is shown. (B) TPMV H and HX ${alpha}$ CEA proteins transiently expressed in 293 cells. The anti-CEA scFv can be cleaved off with factor Xa (right lane). (C) Expression of TPMV F and TPMV HX ${alpha}$ CEA in MC38cea cells results in the formation of syncytia (central panel). The standard MV F protein was coexpressed with a retargeted MV H protein (MV-F/HXL) (15) as a positive control (lower panel). The standard TPMV H protein was unable to fuse murine cells when cotransfected with TPMV F (upper panel).

We then sought to determine whether, in combination with theTPMV F protein, TPMV HX ${alpha}$ CEA can induce fusion of murine MC38ceacells stably expressing human CEA (40). These cells cannot befused when the TPMV glycoproteins are coexpressed (Fig. 4C,upper panel), but when standard H is substituted with the Hprotein displaying the anti-CEA scFv, the cells form syncytia(Fig. 4C, center panel) at least as efficiently as with theretargeted MV envelope protein pair (Fig. 4C, lower panel).The modified H protein coexpressed with the F protein was stillable to induce syncytium formation in TBF cells, albeit withreduced efficiency (data not shown). Thus, we conclude thatreceptor attachment is the factor limiting efficiency of fusionof non-Tupaia cells by the TPMV glycoproteins.

DISCUSSION

Top

Discussion

Paramyxovirus glycoproteins mediate attachment to host cellsand fusion of viral and cellular membranes (reviewed in reference22). We analyzed the glycoproteins of a paramyxovirus that infectsmammals of the order Scandentia and has no cross-reactive relativesthat infect humans. We disclosed a unique F-protein cleavagesequence but also verified that F-protein cleavage is not afactor limiting fusion of human cells. On the other hand, weshowed that the TPMV H protein has a narrow receptor specificityand that fusion of non-Tupaia cells expressing a designatedreceptor occurs only after the display of a cognate specificitydomain on TPMV H.

A ubiquitous protease cleaves TPMV F, producing a unique F1 terminus. The F1 N termini of several paramyxoviruses, including SeV,Simian virus 5 and NDV (38), MV (56), CDV (57), MuV (44), Respiratorysyncytial virus (12), HeV (24), and NiV (26) have been determinedby amino acid sequencing, and this information has been usedto align the cleavage sites and define the preceding mono- andoligobasic signature sequences. Sequencing of the TPMV F1 aminoterminus by Edman degradation indicated that an isoleucine isfound upstream of phenylalanine 109, in spite of the fact thatall paramyxoviral F1 protein sequences examined to date beginwith a phenylalanine or with a leucine in the equivalent position.This observation has important implications for F protein engineering(see below). The sequence preceding the first isoleucine isRDGGT, not a signature protease cleavage site to our knowledge.

The henipaviruses are closely related to TPMV and are also uniquewithin the paramyxoviruses since they have a monobasic cleavagesite that is nevertheless cleaved intracellularly (31). Moreover,the first amino acid of their fusion peptide is a leucine insteadof a phenylalanine (24, 26) (Fig. 1E). Interestingly, it wasrecently shown that all of the amino acids, including a basicone preceding the cleavage site in the NiV F protein, can beindividually mutated with only minor effects on protein function(26). It was also shown that the protease cleaving the nearlyidentical HeV F protein requires calcium as a cofactor (31).Since the NiV and HeV carboxyl-terminal sequences are not homologousto the corresponding TPMV sequences, we suspect that the proteasecleaving TPMV F may be different from that cleaving HeV andNiV F and from furin. Since TPMV F cleavage occurs efficientlyin cells originating from different tissues of four differentspecies, the protease cleaving TPMV F must be ubiquitous.

TPMV envelope as a vector delivery module. These observations not only indicate that F protein cleavageis not a factor limiting TPMV tropism but also suggest thatthe fusion peptide of paramyxoviruses can be extended at itsamino terminus without impairing F protein function. Tropismof another paramyxovirus, MV, was restricted without changingthe F1 amino terminus by introducing two point mutations inthe furin cleavage sequence, making it specific for trypsinor trypsin-like proteases, such as tryptase Clara in lungs (23).The opportunity of leaving one or more additional residues atthe F-protein amino terminus facilitates engineering of paramyxovirusesactivated by certain proteases excreted by tumors (MMPs) thatcleave in the middle of a 6-aa recognition sequence. Indeed,recent experiments with a third paramyxovirus suggest that upto three additional residues can be left upstream of the conservedphenylalanine without compromising fusion efficiency: a recombinantMMP-activated SeV reduces human MMP-secreting xenografts inmice (17). Similarly, TPMV F may be modified for tumor-specificactivation.

The fact that a scFv can be functionally displayed on TPMV Hwithout affecting fusion-support function suggests another levelat which the TPMV envelope can be engineered for the targetingof oncolytic vectors. It was previously shown that the H proteinof MV, a human paramyxovirus, can be targeted by displayingeither small ligands such as EGF and IGF (42) or larger specificitydomains such as scFvs (5, 15, 30, 34), and TPMV H may be similarlytolerant to ligand display. The TPMV envelope has two distinctadvantages over the MV envelope as a vector delivery module.First, it is derived from an animal virus, and since there areno indications that TPMV or TPMV-like viruses infect humans,neutralizing antibodies may not be an issue. Second, TPMV doesnot recognize a receptor on human cells, and therefore "de-targeting"from known and unknown human receptors is not necessary (29,61, 63).

How straightforward is it to enclose a genetic cargo in a TPMVenvelope? We have shown that envelope exchange within the morbillivirusesyields high-titer viruses (62), and other groups have enclosedparamyxoviral ribonucleocapsids in envelopes from differentparamyxovirus subgroups and confirmed efficacy of the chimericviruses as vaccines (6, 48-50, 59). Therefore, we anticipatethat a paramyxoviral chimera with a TPMV envelope may efficientlyspread in vivo. In this perspective, the ribonucleocapsids ofMV and MuV, human paramyxoviruses used in clinical trials ofoncolysis, are premier candidates for enclosure in a TPMV envelope.

Finally, it might be possible to use the TPMV glycoproteinsto pseudotype retroviral and lentiviral vectors. These vectorsare already used to deliver genes in clinical protocols (10,65), but in many cases the target cells have to be transducedex vivo because no efficient targeting is available. Since successfulpseudotyping of retroviruses with the SeV glycoproteins hasbeen reported (19), pseudotyping with the TPMV glycoproteinmay also be possible. The availability of targeting at the receptorattachment level and the perspective of tropism restrictionat the F protein cleavage-activation level make the TPMV envelopea promising module for a wealth of clinical vector applications.

ACKNOWLEDGMENTS

We thank Sompong Vongpunsawad for excellent technical support,Markus Moll for helpful advice on Edman degradation, Ben Maddenfrom the Mayo Clinic Proteomics Research Center for excellentservice, and Jeffrey Schlom for the MC38cea cells.

This study was supported by research grants of the Mayo andSiebens Foundation, the NIH grant R01 CA90636, and a researchscholarship (SP 694/1-1) from the German research foundation(DFG) to C.S.

FOOTNOTES

* Corresponding author. Mailing address: Mayo Clinic Rochester, Molecular Medicine Program, Guggenheim 1838, 200 First St. SW, Rochester, MN 55902. Phone: (507) 284-0171. Fax: (507) 266-2122. E-mail: Cattaneo.Roberto{at}mayo.edu.

Present address: INRS-Institut Armand-Frappier, Universitédu Québec, Laval, Quebec H7V 1B7, Canada.

REFERENCES

Top