Generation of optimized and urokinase-targeted oncolytic Sendai virus vectors applicable for various human malignancies

Article metrics


We previously reported the development of a prototype ‘oncolytic Sendai virus (SeV) vector’ formed by introducing two major genomic modifications to the original SeV, namely deletion of the matrix (M) gene to avoid budding of secondary viral particles and manipulation of the trypsin-dependent cleavage site of the fusion (F) gene to generate protease-specific sequences. As a result, the ‘oncolytic SeV’ that was susceptible to matrix metalloproteinases (MMPs) was shown to selectively kill MMP-expressing tumors through syncytium formation in vitro and in vivo. However, its efficacy has been relatively limited because of the requirement of higher expression of MMPs and smaller populations of MMP-expressing tumors. To overcome these limitations, we have designed an optimized and dramatically powerful oncolytic SeV vector. Truncation of 14-amino acid residues of the cytoplasmic domain of F protein resulted in dramatic enhancement of cell-killing activities of oncolytic SeV, and the combination with replacement of the trypsin cleavage site with the new urokinase type plasminogen activator (uPA)-sensitive sequence (SGRS) led a variety of human tumors, including prostate (PC-3), renal (CAKI-I), pancreatic (BxPC3) and lung (PC14) cancers, to extensive death through massive cell-to-cell spreading without significant dissemination to the surrounding noncancerous tissue in vivo. These results indicate a dramatic improvement of antitumor activity; therefore, extensive utility of the newly designed uPA-targeted oncolytic SeV has significant potential for treating patients bearing urokinase-expressing cancers in clinical settings.


In recent decades, gene therapy has been anticipated as a new therapeutic strategy to treat patients with intractable malignancies. A number of clinical studies of gene therapy for malignancies have been done worldwide, but the therapy's efficacy was relatively limited. Therefore, improvement in some aspects of gene therapy, including the gene transfer efficiency without significant damage to noncancerous tissue, has been much desired. To overcome the limited efficacy of gene transferring by current replication-deficient viruses in vivo, tumor-selective and replication-competent oncolytic viruses have been generated, and clinical studies on the use of such viral vectors have been reported.1, 2 These viruses, including an oncolytic adenovirus and a herpes virus, can replicate specifically in the cancer cells and spread in situ, exhibiting oncolytic activity through a direct cytopathic effect during intracellular viral replication and budding. Clinical studies on the use of oncolytic viruses actually showed reduced tumor size in some patients, suggesting proof of the strategy's concept and its potential efficacy; however, cancer cells infected by these viruses released active virus particles to the adjacent noncancerous tissue as well as to systemic circulation, presenting a potential risk for adverse effects through systemically disseminated virus infection, so-called ‘viremia.’ In fact, some patients who received oncolytic viruses available at present demonstrated flu-like symptoms, including fever and chills, probably through viremia-induced systemic innate immune responses. Thus the development of a new mode of oncolytic viruses that do not produce secondary virus particles causing systemic viremia is highly desirable because of safety concerns.

Sendai virus (SeV) is an enveloped virus with a linear, nonsegmented negative-strand RNA genome of approximately 15.4 kb, which belongs to the large virus family Paramyxoviridae, comprising a variety of mammalian pathogens. SeV displays a narrow spectrum of tissue tropism in susceptible hosts, actually growing in the respiratory tract of mice or in the allantoic cells of embryonated chicken eggs with little appreciable spreading to other tissues in these host organisms, even though its receptor is sialic acid residues that are ubiquitous throughout the body. This restricted tropism primarily depends on the specific tissue proteases required for cleavage activation of viral fusion (F) glycoprotein, and thus the proteases needed for infectivity of progeny (the capacity to penetrate into and initiate infection in the next cell) are available only on the surface of those limited types of tissue.3

Using sophisticated manipulating technology of the genome of SeV, namely ‘Reverse Genetics,’ we recently generated SeV vectors lacking envelope-related genes, including fusion (F) gene-deleted (SeV/ΔF),4 matrix (M) gene-deleted (SeV/ΔM),5 hemagglutinin/neuraminidase (HN) gene-deleted (SeV/ΔHN), both M and F genes-deleted (SeV/ΔMΔF)6 and all of the envelope-related genes-deleted (SeV/ΔMΔFΔHN).7 Among the proteins contained in the viral lipid bilayer, M protein plays a central role in secondary virus assembly and budding from infected cells. Therefore, deletion of the M gene from SeV almost completely abolished virus maturation during formation of infectious particles in infected cells and instead caused cell-to-cell vector spreading through membrane fusion, forming large syncytia and resulting in cell death by the active F protein under supplementation of trypsin.5 This suggested to us that cell-to-cell spread of this vector could be manipulated by the presence of selected F protein-activating proteases.

On the basis of these features of SeV vectors, we previously reported the development of a prototype oncolytic M gene-deleted SeV vector by replacing the trypsin-sensitive amino-acid residues of F protein at the matrix metalloproteinase (MMP)-selective cleavage site (-PLGMTS-).8 As expected, this new class of SeV vectors spread widely from cell to cell through fusion of the plasma membranes, forming syncytia among the MMP-expressing tumor cells in vitro, and led the tumor to extensive death in vivo, without releasing secondary vector particles.8 Our subsequent studies examining the utility of the vector among various human cancer cell lines from different origins, however, showed that the efficacy of this prototype vector was relatively limited because of (1) the requirement for a relatively high expression level of MMPs for efficient cell death and (2) the limited numbers of MMP-expressing tumor cells, less than 20% (unpublished observation).

To overcome these problems, we have been focusing on the manipulation of the cytoplasmic domain of F protein and urokinase-type plasminogen activator (uPA) for the past 2 years. Recent studies showed that truncation of the cytoplasmic domains, not only of retrovirus and herpes virus glycoproteins,9 but also those of Paramyxoviridae (that is, measles virus and Newcastle disease virus) increases their fusion activity.10 Moreover, the plasminogen activator (PA) system consists of two PAs, urokinase (uPA) and tissue type (tPA), which activate plasminogen to the active plasmin. This serine protease can degrade extracellular matrix components directly, for example, fibronectin and proteoglycans, or indirectly by activating other proteases, including MMPs. Importantly, uPA has been shown to be expressed in various types of human malignancies more frequently (over 50%) than MMPs.

We here report the optimized design of protease-specific oncolytic recombinant SeV vectors that have been generated by truncation of the F protein cytoplasmic domain and by the optimization of catalytic activity against MMPs and uPA. These new vectors showed dramatically enhanced antitumor activities in vitro and in vivo.


Optimization of truncation for the cytoplasmic domain of the F gene of SeV to enhance its fusogenic activity

As an initial step, we optimized the truncation of the cytoplasmic tail of wild-type F gene transfected by pCAGGS to LLC-MK2 cells cultivated with trypsin. We examined three deleting mutants (Fct27, Fct14, Fct4), as shown in Figure 1a, and the fusion efficiency was quantified by counting the nuclei per formed syncytium.8 As shown in Figure 1b, syncytia formation was observed only in cells with cotransfection of paired membrane glycoproteins (F and HN). The highest ratio of fusion activity, six times higher than that seen by wild-type F gene cotransfected (P<0.01), was seen in the use of a truncated clone (Fct14); therefore, we concluded that Fct14 should be used in our new design of an oncolytic SeV.

Figure 1

Optimization of fusogenic activity through truncation of the cytoplasmic tail of SeV-F. (a) Design of cytoplasmic domain of F: Three deletion mutants of the cytoplasmic domain of F protein were constructed and inserted in mammalian expression vector (pCAGGS), respectively. (b) LLC-MK2 cells transfected with pCAGGS/F (wild), Fct27, Fct14, Fct4 with or without HN were cultured with trypsin (7.5 μg ml−1). Induction of cell fusion and cell-to-cell viral spreading were assessed by counting nuclear numbers of syncytia formed after hematoxylin staining.

Optimization of the sensitive linker sequence in a protease-specific processing site of the F gene

As a next step, we optimized the sensitive linker sequence in the processing site of F gene against the MT1-MMP and MMP2, urokinase-type (uPA) or tissue-type (tPA) plasminogen activators, that is, proteases that were shown to be expressed by various human cancers. The candidate sequences were based on the information determined previously by phage display.11 Each plasmid subcloned with corresponding mutant F genes shown in Figure 2a was cotransfected with HN gene expression vector to LLC-MK2 cells, and fusion activity was determined as described above. In cases of collagenase and gelatinase, cDNA of MT1-MMP was also cotransfected, because MT1-MMP is required for efficient activation of MMP-2. Expression of active MMP-2 in this experiment was confirmed by zymography (data not shown). Replacement of the trypsin cleavage site (-QSR-) in the collagenase, gelatinase or uPA/tPA cleavable site would add amino acids upstream of the fusion peptide. Active fusion peptides of Paramyxovirus highly conserve the hydrophobic sequence, consisting of phenylalanine at the N terminus of F1. Previously, we revealed that the addition of the sequence –MTS-, but not –LGL- or –LWA-, at the N terminus of activated fusion peptide (F1) of SeV was susceptible to replacement with an efficient membrane fusion; therefore, –MTS- was also examined to determine whether it improved the original gelatinase sequence.

Figure 2

Optimization of linker sequences sensitive to matrix metalloproteinase (MMP) and plasminogen activator (PA). (a) The design of the F cleavage site: F expression plasmid was constructed, and the F cleavage site was redesigned by site-directed mutagenesis. Collagenase (MMP1, MMP3, MT-MMP and so on) and gelatinase substrates (MMP2, MMP9) were selected by the study of the synthesized peptide sequence. The additional MTS sequences were also examined (MMP-G1/MTS, MMP-G2/MTS, MMP-G3/MTS, MMP-G4/MTS) to facilitate the sensitivity to proteases. (b) Fusion assay by cotransfection of modified F- and wild HN-expression plasmids to LLC-MK2 cells that expressed MMP2, MT1-MMP or MT1-MMP/MMP2. Cotransfection of three independent plasmid vectors expressing modified F, HN and EGFP to LLC-MK2 cells was done by FuGene 6 transfection reagent. Two days later, the nuclear numbers of syncytia formed were counted. (c) Fusion assay by cotransfection of modified F- and wild HN-expression plasmids to LLC-MK2 cells in culture media supplemented with either recombinant human uPA or tPA. Two days later, the nuclear numbers of syncytia formed were counted.

Syncytia formation of active MMP2 (MT1-MMP/MMP2)-expressing cells was examined by the addition of each plasmid vector expressing MMP-C1-4, -G1-4, -G1-4/MTS linker. As shown in Figure 2b, MMP-C1, -C2, -C4 linker formed active MMP-2-specific syncytia formation, whereas PA-sensitive linkers did not. MMP-G4/MTS showed the highest efficiency in forming syncytia under cotransfection of both MT1-MMP and MMP2. It has been reported that MMP-G2 (-PQG/LYA-) is an efficient sequence in MMP-expressing HT-1080 cell, in the use of recombinant measles virus,12 but this was not the case with SeV under similar experimental conditions using HT-1080 cells (data not shown).

A similar strategy was adopted for uPA/tPA; in more detail, uPA1 linker (-VGR-) (PA-sub II), a more sensitive linker, uPA2,13 and tPA-specific linker14 were substituted into the trypsin-cleavage site of the F gene (Figure 2a). A syncytia formation assay showed that uPA2 (-SGR/S-) was specific for PA but not for MMPs, and it was the most effective linker to form syncytia in each PA in the culture medium (Figure 2c).

Considering these results, we expected that the combination of Fct14 and F(MMP-G4/MTS) or F(uPA2) might show optimal performance for killing cancer cells. To test this hypothesis, we rescued and evaluated the oncolytic SeV vectors encoding MMP-C1, MMP-G1/MTS, MMP-G2/MTS, MMP-G3/MTS, MMP-G4/MTS, uPA1, uPA2 and tPA linker in the F gene without any modification in their cytoplasmic domain, and truncated Fct14 SeV with the MMP-G4/MTS and uPA2 sequence, respectively.

Optimization of antitumor efficiency of MMP-targeted oncolytic SeV vectors

To evaluate the potency of the syncytia formation and tumor killing activity, we first focused our attention on MMP-targeted oncolytic SeV vectors, directly comparing their effects in vitro and in vivo using HT1080 (high MMP-2 and -9 expression) and SW620 cells (no detectable expression of these), as shown by gelatin zymography (Figure 3a, inset panel). As shown in Figure 3a, conventional SeV vector-deficient F gene (SeV/ΔF) could not lead MMP-expressing HT1080 cells to syncytia formation in vitro, but these cells treated with SeV vector-deficient M gene (SeV/ΔM) having manipulated F to MMP-sensitive sequences (C1 and G1-4/MTS) at a multiplicity of infection (MOI)=0.3 showed various amounts of syncytium. The greatest effect was seen in the use of G4/MTS, findings that were not seen in the case of SW620 cells without MMP expression, and the effect observed with G4/MTS virus was further enhanced when combined with Fct14. The induced syncytia formation was critically sensitive to 50 μg ml−1 of TIMP-2 but not to TIMP-1, endogenous MMP-specific inhibitors. Furthermore, efficacies of syncytia formation corresponded roughly to the cytotoxicity that was assessed by LDH release to culture media (Figure 3b).

Figure 3

Antitumor activity of newly optimized matrix metalloproteinase (MMP)-targeted oncolytic SeV vectors in vitro and in vivo. (a and b) Syncytia formation (a) and LDH-releasing (b) assays assessing cell killing activity in vitro to optimize MMP-sensitive linker sequences of modified F gene of oncolytic SeV. HT1080 cells expressing active MMPs were used, and a control cell line without active expression of typical MMPs, SW620, was also used (a, inset). Each vector was infected at MOI=0.3, and 2 days later, the nuclear numbers of syncytia formed and released LDH were assessed. Note that optimized cytotoxic activity seen with the use of SeV/Fct14(MMP-G4/MTS)ΔM was completely abolished by supplementation with a typical MMP inhibitor, TIMP-2, but not by supplementation with TIMP-1 (50 ng ml−1, respectively). (c and d) Antitumor activity of oncolytic SeVs with various modifications of F gene sensitive to MMPs in vivo. HT1080 (1 × 106 cells) that expressed a high level of active MMPs was subcutaneously burdened to the abdominal wall. Seven days later, each SeV vector was injected intratumorally at once. Note that SeV/Fct14 (MMP-G4/MTS)ΔM-green fluorescent protein (GFP) showed the strongest inhibition of HT1080 tumor growth (c), whereas neither the control vector (SeVΔF) nor SeV/Fct14(MMP-G4/MTS)ΔM-GFP showed a significant effect on the growth of SW620 tumors (d).

Subsequently, established HT1080 and SW620 xenografts on nude mice were intratumorally administered by these vectors once at day 7 after tumor inoculation (Figures 3c and d). As expected from in vitro study, SeV/Fct14(MMP-G4/MTS)ΔM-green fluorescent protein (GFP) showed optimized effect inhibiting the growth of MMP-expressing HT1080 tumors, compared with the effects of other constructs, findings that were not seen in the case of SW620 without expression of MMPs.

Optimization of the antitumor efficiency of uPA-targeted oncolytic SeV vectors

Next, similar assessments were done to optimize uPA-targeted oncolytic SeV vectors. We used two independent human prostate cancer cell lines (PC3, DU145) expressing the uPA and uPAR, determined by western blot analysis (data not shown), and activation of plasminogen in the supernatant of both cell lines was confirmed by casein zymography (Figure 4a, inset). A human colon cancer cell line that does not express PA activity, HT29, was used as a negative control.

Figure 4

Antitumor activity of newly optimized uPA-targeted oncolytic SeV vectors in vitro and in vivo. (a and b) Syncytia formation (a) and LDH releasing (b) assays assessing cell-killing activity in vitro to optimize plasminogen activator (PA)-sensitive linker sequences of modified F gene of oncolytic SeV. PC3 and DU145 cells expressing active uPA were used, and a control cell line without active expression of PA, HT29, was also used (a, inset). Each vector was infected at MOI=0.3, and 2 days later, nuclear numbers of syncytia formed and released LDH were assessed. Note that optimized cytotoxic activity seen with the use of SeV/Fct14(uPA2)ΔM was completely abolished by supplementation with a typical PA inhibitor, PAI-1 (50 ng ml−1, respectively). (c and d) Antitumor activity of oncolytic SeVs with various modifications of the F gene sensitive to PA in vivo. PC3 (1 × 106 cells), which expressed a high level of active uPA, was subcutaneously burdened to the abdominal wall. Seven days later, each SeV vector was injected intratumorally at once. Note that SeV/Fct(uPA2)ΔM-green fluorescent protein (GFP) showed the strongest inhibition of PC3 tumor growth (c), whereas neither the control vector (SeVΔF) nor SeV/Fct(uPA2)ΔM-GFP showed a significant effect on the growth of H29 tumors (d).

As shown in Figure 4a, the oncolytic SeV having uPA2 linker, but not uPA1, showed a significant amount of syncytium formation in uPA-expressing PC3 and DU145, but not in HT29 without PA activity. The optimized effect in vitro was seen in the combination of uPA2 and Fct14 in the M gene-deficient oncolytic SeV, and this effect was almost completely abolished by a PA-specific inhibitor, PAI-I. Cytotoxicity assessed by LDH release also roughly corresponded to the amount of syncytia formation (Figures 4a and b). Such findings obtained in in vitro studies were also represented in vivo; SeV/Fct14(uPA2) ΔM-GFP showed an optimized effect in established PC3 xenografts, and other constructs, including MMP-targeted SeV/Fct14(MMP-G4/MTS)ΔM-GFP, could not suppress the tumor growth. This effect was not seen in the case of H29 without significant expression of PA activity.

New oncolytic SeV vectors selectively induced protease-dependent syncytia formation and inhibited tumor growth through extensive killing of MMP- or uPA-expressing cells

Protease-dependent syncytia formation was examined in vitro, using these optimized MMP-targeted (SeV/Fct14(MMP-G4/MTS)ΔM-GFP) and uPA-targeted vectors (SeV/Fct14(uPA2)ΔM-GFP). Figures 5b and c shows the enzymatic activity of MMP2 and uPA in several human tumor cell lines and normal human cell (human aortic smooth muscle cells). HT1080 showed the highest MMP2 expression, and CAKI-I and BxPC3 expressed a low level, whereas HT29 and PC14 expressed MMP2 at undetectable levels (Figure 5b). Most of the tested tumor cell lines (CAKI-I, BxPC3, PC14, HT1080) expressed both uPA protein and uPAR protein by western blot analysis (data not shown) and exhibited enzymatic activity (Figure 5c), whereas HT29 expressed only uPAR protein but not urokinase activity.

Figure 5

Protease-dependent killing activity of newly developed oncolytic SeVs on various human cancer cells. (a) Cells were infected with MOI=0.3 and 48 h later were photographed under UV light (upper) to activate green fluorescent protein (GFP) fluorescence or visible light (lower). (b) matrix metalloproteinases-2 (MMP2) or (c) uPA expression levels in various human tumor cells. Five hundred microliters of culture medium was added to each tumor cell sample (1 × 106) and the samples were incubated in 6-well culture dish for 2 days. In each case, the conditioned medium was collected and applied to a gelatin gel plate. After gelatin zymography, the expression level of MMP2 was semiquantified by densitometry. The same samples were subjected to a uPA activity kit as described in the Materials and methods section. (d) Cytotoxicity assessed by LDH release against MMP- or uPA-targeted optimized oncolytic SeV vectors. (e) Immunoblot analysis of F protein processing by treatment of the virions with proteases in vitro. Virus particles of the parental SeV/ΔM-GFP (lanes 1), SeV/Fct14(MMP-G4/MTS)ΔM-GFP (lane 2) and SeV/Fct14(uPA2)-GFP (lane 3) were prepared from the culture supernatants of an M protein-expressing helper cell line. The virus particles were incubated with MMP-9 (0.1 mg ml−1) (left) or uPA (0.1 mg ml−1) (right) for 30 min at 37 °C. The viral proteins were then analyzed by western blotting using an anti-F1 antibody.

Infection with the MMP-targeted oncolytic SeV vector induced extensive syncytia in HT1080 and small syncytia in CAKI-I and BxPC3, but not in HT29 and aortic smooth muscle cells (Figure 5a), reflecting their MMP2 expression levels. Infection with a uPA-targeted vector formed large syncytia in all tumor cell lines tested except for HT29 and aortic smooth muscle cells (Figure 5a). The formation of syncytia nearly corresponded to the level of cytotoxicity (Figure 5d), at least in vitro.

To demonstrate the protease-mediated cleavage of the engineered F protein, we collected the MMP-targeted and uPA-targeted vector particles without any protease and digested them by recombinant MMP9 or uPA in vitro (Figure 5e). The virus induced fusion and virus spreading in a highly specific protease-dependent manner strictly paralleled the cleavability of the F0 precursor on the virions into F1 and F2 by the respective proteases; specific cleavage of F0 by MMP9 and uPA for MMP-targeted (SeV/Fct14(MMP-G4/MTS)ΔM-GFP) (Figure 5e, lane 2) and uPA-targeted SeV vector (SeV/Fct14(uPA2)ΔM-GFP) (Figure 5e, lane 3), respectively. Control SeV/ΔM-GFP, which have wild type F (Figure 5e, lane 1), were not processed by both protease.

Subsequently, we assessed whether these findings obtained in vitro would occur in vivo. The growth of HT29 tumors established in the flanks of athymic nude mice, tumors that expressed almost no MMP or uPA, as a negative control, was rarely affected by the bolus intratumor injection of either MMP-targeted vector (SeV/Fct14(MMP-G4/MTS)ΔM-GFP) or uPA-targeted vector (SeV/Fct14(uPA2)ΔM-GFP) on day 6, as shown in Figure 5d. In contrast, the growth of dermal tumors of CAKI-I, BxPC3 and PC14 were significantly suppressed by intratumor injection of uPA-targeted vector (CAKI-I: P=0.039, BxPC3: P=0.048 and PC14: P=0.03) but not by MMP-targeted vector (Figures 6a–c), reflecting the cytotoxic activity seen in vitro (Figure 5d).

Figure 6

Selective inhibition of in vivo growth of tumors highly expressing uPA by bolus injection of uPA-targeted, but not of matrix metalloproteinase (MMP)-targeted, oncolytic SeV vector. CAKI-I (a), BxPC3 (b) and PC14 (c) xenografts were established in Balb/c nude mice. Five days later, the tumors were treated with phosphate-buffered saline, SeV/Fct14(MMP-G4/MTS)ΔM-GFP or SeV/Fct14(uPA2)ΔM-GFP (1 × 107 cell infection units per tumor, respectively). (d) Cytotoxic effects of human tumor cell lines by an infection by MMP- and uPA-targeted oncolytic SeV. Cells were infected with MOI=0.3 and cytotoxicity was measured by LDH assay kit 4 days later.

Finally, we screened the oncolytic activity of these optimized vectors using 33 established human cancer cell lines from various origins. As shown in Figure 6d, extensive oncolytic/cytotoxic activity (>40%) of uPA-type vectors was illustrated on 42.4% of these cancer lines (14/33 cell lines), whereas only 2 cancers (6.1%) were sensitive to the MMP-targeted vector. As a number of publications show that the uPA system appears to be active in the majority of various cancers in situ,15, 16 these results suggest the more extensive utility of uPA-targeted oncolytic SeV than that of MMP-targeting SeV.


In a previous study, we developed a novel prototype of oncolytic recombinant SeV by introducing two genomic modifications to its genome, resulting in cell-to-cell spread through fusion of plasma membranes forming syncytia only among the MMP-expressing tumor cells, and successful inhibition of the tumor growth in vivo, without releasing secondary vector particles.8 On the basis of this concept, in this study, we developed dramatically improved vectors by modifying the F gene sensitive to uPA and MMP associated with truncation of its cytoplasmic tail. These newly optimized vectors showed markedly enhanced cancer killing activity in a protease-dependent manner in vitro and in vivo, and, importantly, modification of amino-acid residues of the F gene for targeting uPA dramatically expanded the cell-killing spectrum of oncolytic SeV. These results suggest the potential utility of SeV/Fct14(PA2)ΔM in clinical gene therapy for cancer patients.

Some recent studies showed that artificial truncation of the cytoplasmic domain of glycoprotein of retrovirus,9 herpes virus17 as well as paramyxovirus10 enhanced their fusogenic activity. Considering this information, we found that the modification of the F gene of SeV (truncation of 14 C-terminal amino acids) resulted in a dramatic increase of fusogenic activity, namely a sixfold increase in the number of syncytia formed compared with that seen by the original F protein (Figure 1). Similar findings were also seen in the case of measles viruses, another member of Paramyxoviridae; modification of the F gene (truncation of 17 C-terminal amino acids) enhanced its fusogenic activity.10 The precise mechanism related to enhancement of fusogenic activity caused by the cytoplasmic tail of Paramyxoviridae, however, has been largely unknown. A possible explanation is the loss of interaction among H, F and M, because the direct interaction of the cytoplasmic tails of H/F and M were shown to downregulate viral fusion activity.10, 18 However, this may not be the case, because the M gene was deleted in the oncolytic SeV vector used in this study. Further studies will be needed to clarify the possible molecular mechanism underlying fusogenic activity and the cytoplasmic domain of F protein.

Subsequently, we further investigated the optimal amino-acid sequence of the cleavage site against MMPs and uPA/tPA to enhance the catalytic potency of oncolytic SeV. We found more sensitive linker sequences (MMP-G4/MTS: -PRAMTS-; uPA2: -SGRS-) than that of the prototype vectors (MMP-G1/MTS: -PLGMTS- and uPA1: -VGR-) (Figure 2); however, the sequence-specific sensitivity may depend on the type of virus. For example, MMP-activated recombinant retroviruses could be generated by the introduction of -PLGLYA- hexamer, and a similar sequence (MMP-G2: -PQGLYA-) could be adapted to the oncolytic measles virus vector, as demonstrated by Springfeld et al.12 Measles virus-F protein containing MMP-G2 (-PQGLYA-) sequence, called ‘MMP-A1’ in their study, could lead HT1080 cells to syncytia formation. In contrast, we found that SeV-F protein with MMP-G2 did not cause syncytia formation in MMP-expressing LLC-MK2 (Figures 2a and b) or HT1080 cells (data not shown). The reason for this distinctive phenomenon is unclear, but we speculate that the addition of sequences, for example, -LYA-, may have interrupted the conformation and function of the fusion peptide in the case of SeV.

Using the optimized MMP-targeted oncolytic SeV (SeV/Fct14(MMP-G4/MTS)), we directly assessed its cytotoxic effects on various types of tumor cell lines. Only two human cancer cell lines (6.1%: HT1080 and MDA-MB-231) were sensitive to this type of vector, showing over 40% cytotoxicity, among 33 screened human cancer cells (Figure 6d). Importantly, 11 cell lines (33.3%) showed almost no response to SeV/Fct14(MMP-G4/MTS). Recently some researchers and we reported on the generation of plasmid DNA, adenovirus vector carrying viral envelope glycoproteins, measles vector and Sendai virus vector activatable by MMPs.8, 12, 19, 20 Their data show that those vectors can target the cytotoxicity of the fusogenic membrane glycoprotein in MMP-overexpressing tumor lines and xenografts, and maintain significant antitumor activity both in vitro and in vivo. MMP-targeted vectors, irrespective of the vectors’ backbone, may be confronted with a similar difficulty in clinical settings.

To overcome such a potential issue, we here further generated a new group of vectors that could be useful against a broad range of human tumors, namely the optimized uPA-targeted oncolytic SeVs. We here assessed uPA-sensitive peptide substrates, F(uPA1) (-VGR-) and F(uPA2) (-SGRS-)13 as well as tPA-sensitive residue F(tPA) (-YGRS-).21 Cell fusion activity was optimal with the use of F(uPA2), and suboptimal activity was seen in the case of F(tPA) (Figure 2c). As western blot analyses revealed significant expression of tPA in a few cancer cell lines among 33 cell lines assessed (as listed in Figure 6d (data not shown)), the uPA2 substrate was utilized for subsequent studies. The result was that greater numbers of human tumor cells (42.4 and 51.5% for >40 and >30% cytotoxicity, respectively) were highly sensitive to new and optimized uPA-targeted oncolytic SeVs. These findings suggest that the targeting uPA should extend the potential utility of oncolytic SeVs in clinical settings.

In addition, there are several theoretical advantages of M gene-deleted paramyxovirus-based vectors over the current oncolytic viruses. First, abolishment of secondary viral particles through deletion of the M gene may reduce the risk of their systemic spread. Second, we should encounter the direct and indirect antitumor effect of the fusogenic membrane glycoprotein of enveloped viral vectors for gene therapy, including measles virus F/H proteins, gibbon ape leukemia virus envelope protein, vesicular stomatitis virus G protein and HN/F proteins of our oncolytic SeVs. Virus-mediated fusogenic membrane glycoprotein expression in target tumor cells has shown direct cytotoxicity, associated with a ‘bystander-like effect,’ against human tumor xenografts in vivo with the use of plasmid DNA, adenoviral or lentiviral vectors.22, 23 In addition to the direct cytotoxic effect, fusogenic membrane glycoprotein delivery can initiate systemic antitumor immunity, thus indirectly suppressing the growth of distant tumors.24 We also observed that treatment of solid tumors by uPA-targeted oncolytic SeVs used in this study resulted in the establishment of tumor-specific antitumor immunity to prevent the recurrence of tumors (Kinoh et al., unpublished data). These multiple mechanisms for antitumor activity of uPA-targeted oncolytic SeVs show promise for the development of treatments for patients with intractable and far advanced malignancies in the clinical setting.

In summary, we successfully developed new and optimized oncolytic recombinant SeVs that target uPA and MMP for extensive killing of cancer cells in vitro and in vivo. We believe that this novel type of oncolytic virus that works without spreading secondary particles may open the way to developing an oncolytic strategy for cancer patients through not only possible improvement of the treatment's efficacy, but also its safety in clinical settings.

Materials and methods

Cell lines and reagents

Tumor cell lines used in this study were obtained from the American Type Culture Collection (ATCC, Rockville, MD, USA) or the European collection of animal cell cultures (Wiltshire, UK) or RIKEN (Tsukuba, Japan). Each cell line was maintained in directed medium supplemented with 10% fetal bovine serum. The following cell lines were used: PC3, DU145, CAKI-I, HT1080, SW620, HT29, LLCMK2, PC14, BxPC3. Normal human aortic smooth muscle cells were purchased from Takara Biotech (Otsu, Japan). Recombinant human PAI-I, human TIMP-1 and human TIMP-2 (Calbiochem, Darmstadt, Germany) at final concentrations of 50 μg ml−1 were used as protease-specific inhibitors.

Plasmid construction of F gene and site-specific mutagenesis

Modification of the F protein cleavage site and deletion of the cytoplasmic domain of the F gene were done after subcloning to the expression vector pCAGGS25 with the Quick-Change Mutagenesis Kit (Stratagene, La Jolla, CA, USA) according to the manufacturer’s instructions. The integrity of mutant constructs was confirmed by direct DNA sequencing.

Construction of MMP2 or MT1-MMP-expressing plasmid vectors

Full-length human MT1-MMP (MMP14) and human MMP2 cDNAs were cloned from total RNA of HT1080 cells by reverse transcription-PCR, and amplified fragments were inserted in a pIRES vector (Takara Biotech, Otsu, Japan). The full sequence of cDNAs were determined by direct sequencing. The enzymatic activity of these genes was checked by gelatin zymography, as will be described later.

Construction and recovery of improved oncolytic SeV vector

A scheme of the designs of SeV construction is shown in Supplementary Figure 1. The parent plasmid pSeV18+/ΔM-GFP, in which the GFP had been substituted for the deleted M gene, was constructed as described before by Inoue et al.5 The parent plasmid was digested with SalI and NheI, and the F gene fragment (9634 bp) was subcloned into LITMUS 38 (New England Biolabs, Beverly, MA, USA). Site-directed mutagenesis was performed using the Quick-Change Mutagenesis Kit (Stratagene), and mutated F gene was returned to the pSeV18+/ΔF-GFP backbone.4 Recovery and amplification of the SeV vector were generated essentially as described before;4, 5 briefly, LLC-MK2 cells were transfected at room temperature with a plasmid mixture containing each plasmid [pSeV+18/F(MMP-C1 MMP-G1/MTS, G2/MTS, G3/MTS, G4/MTS, uPA1, uPA2 and tPA)ΔM-GFP], pSeV+18/Fct14(MMP-G4/MTS, uPA2)ΔM-GFP, pGEM-NP, pGEM-P and pGEM-L in 110 μl of Superfect transfection reagent (Qiagen, Tokyo, Japan). The transfected cells were maintained for 3 h in 3 ml of Opti-MEM (Gibco-BRL) plus 3% fetal calf serum, washed three times with modified Eagles’ medium (MEM) and incubated for 60 h in MEM containing araC (40 μg ml−1). The transfected cells were collected by centrifugation at 1000 g for 5 min, resuspended in Opti-MEM and lysed by three cycles of freezing and thawing. The lysate solution was incubated on the F/M-expressing LLC-MK2 cells in a 24-well plate.5 Twenty-four hours later, cells were washed three times with MEM and incubated for 3–6 days in MEM containing araC and 5 U ml−1 collagenase type IV, Clostridium histolyticum (ICN Biochemical Inc., Costa Mesa, CA, USA) or 7.5 μg ml−1 trypsin plus 10 ng ml−1 urokinase (Cosmobio, Tokyo, Japan). Virus yield is expressed in cell infection units, as described previously.4

Fusion assay and cytotoxicity assay

Cells were plated in 96-well dishes at a density of 5 × 103 cells per well and infected as described above with each SeV vector at an MOI of 0.3. On day 4, cells were fixed with 4% paraformaldehyde. Cells were then stained with hematoxylin, rinsed and covered with 100 μl phosphate-buffered saline. The number (N) of syncytia per well (0.3 cm2) was counted in triplicate wells, and the average nucleus number was calculated. All experiments were repeated at least four times. To evaluate the fusogenic activity against MMP-expressing cells, Sendai envelope expression plasmid (pCAGGS/F and pCAGGS/HN) and MMP expression plasmid were transfected into LLCMK2 in 96-well plates using trans IT reagent. The number of nuclei in the fusion cells was counted as described above. Cytotoxicity was determined by LDH assay using the Cytotoxicity Detection Kit (Roche Diagnostics KK, Tokyo, Japan) according to the manufacturer’s instructions.

Casein and gelatin zymography

Gelatin and casein zymography was performed using the method outlined by Heussen and Dowdle.26 Briefly, serum-free supernatants from each cell culture were analyzed under nonreducing conditions on 7.5% SDS-polyacrylamide gel electrophoresis (SDS-PAGE) (Sigma, St Louis, MO, USA) containing 0.1% casein plus 4 mg of plasminogen or 0.1% gelatin. The gel was then incubated at room temperature in a 2.5% Triton X-100 solution containing 0.05% sodium azide and 0.05 M Tris (pH 7.5; Sigma) followed by incubation at 37 °C in 0.1 M Tris-HCl (pH 8.3)/0.5 M sodium chloride solution for 16 h. The gel was then stained with a 0.25% Coomassie Blue/1.1% acetic acid/45.5% methanol solution for 40 min, followed by destaining in a 25% methanol/10% acetic acid mixture. Enzyme activity was visualized as clear zones on a blue background. Semiquantification of MMP2 was done by the method of Togawa et al.27 Densitometry was done using NIH Image.

Urokinase activity assay

Cells (1 × 106) were seeded in six-well dishes, and 0.5 ml of supernatant was collected after 48 h of incubation at 37 °C. The urokinase activity was measured with a Urokinase Activity Kit (Chemicon International, Temecula, CA, USA).

Western blot analysis

Three Sendai virus vectors (SeV/ΔM-GFP, SeV/Fct14(MMP-G4/MTS)ΔM-GFP, SeV/Fct14(uPA2)ΔM-GFP) were prepared from the culture supernatants of an M protein-expressing helper cell line after infection of an MOI of 3 (cell infection units) and subsequent incubation for 2 days in the absence of any protease. Three SeVs were digested by human recombinant active MMP9 (0.1 μg ml−1) (Calbiochem) or urokinase (Calbiochem) for 30 min at 37 °C. The viral proteins were then analyzed by western blotting using an anti-F1 rabbit antiserum. Immunoblot analysis was carried out using a rabbit peptide antiserum against the F1 (dilution, 1:1000),8 an anti-rabbit-horseradish peroxidase conjugate and the enhanced chemiluminescence detection system (Amersham Biosciences, Piscataway, NJ, USA).

In vivo xenograft model

Female 7- to 8-week-old Balb/c nu/nu mice were obtained from Charles River Co. (Shizuoka, Japan). Human tumor cell lines (HT1080, PC3, PC14, CAKI-I, BxPC3, SW620, HT29) (1 × 107 cells) were propagated and implanted in Matrigel (Beckton Dickinson, Bedford, MA, USA). The cells were injected subcutaneously into the right flank of mice. When the subcutaneous tumors grew to 4–8 mm in diameter (usually after 7 days), the mice were divided into groups (n=5–8 per group), and each SeV vector (1 × 107 cell infection units) in a 100 μl volume was injected intratumorally using a 26-gauge needle. Tumor volume was calculated with the formula Volume=a2b/2, where a is the shortest diameter and b is the longest diameter. Tumor sizes were expressed as mean volumes±s.e.

Statistical analysis

Statistical significance was assessed with the Mann–Whitney U-test as appropriate.


  1. 1

    Aghi M, Martuza RL . Oncolytic viral therapies—the clinical experience. Oncogene 2005; 24: 7802–7816.

  2. 2

    Bell J . Replicating oncolytic virus therapeutics—Third International Meeting. IDrugs 2005; 8: 360–363.

  3. 3

    Tashiro M, McQueen NL, Seto JT . Determinants of organ tropism of sendai virus. Front Biosci 1999; 4: D642–D645.

  4. 4

    Li HO, Zhu YF, Asakawa M, Kuma H, Hirata T, Ueda Y et al. A cytoplasmic RNA vector derived from nontransmissible Sendai virus with efficient gene transfer and expression. J Virol 2000; 74: 6564–6569.

  5. 5

    Inoue M, Tokusumi Y, Ban H, Kanaya T, Shirakura M, Tokusumi T et al. A new Sendai virus vector deficient in the matrix gene does not form virus particles and shows extensive cell-to-cell spreading. J Virol 2003; 77: 6419–6429.

  6. 6

    Inoue M, Tokusumi Y, Ban H, Shirakura M, Kanaya T, Yoshizaki M et al. Recombinant Sendai virus vectors deleted in both the matrix and the fusion genes: efficient gene transfer with preferable properties. J Gene Med 2004; 6: 1069–1081.

  7. 7

    Yoshizaki M, Hironaka T, Iwasaki H, Ban H, Tokusumi Y, Iida A et al. Naked Sendai virus vector lacking all of the envelope-related genes: reduced cytopathogenicity and immunogenicity. J Gene Med 2006; 8: 1151–1159.

  8. 8

    Kinoh H, Inoue M, Washizawa K, Yamamoto T, Fujikawa S, Tokusumi Y et al. Generation of a recombinant Sendai virus that is selectively activated and lyses human tumor cells expressing matrix metalloproteinases. Gene Therapy 2004; 11: 1137–1145.

  9. 9

    Rein A, Mirro J, Haynes JG, Ernst SM, Nagashima K . Function of the cytoplasmic domain of a retroviral transmembrane protein: p15E-p2E cleavage activates the membrane fusion capability of the murine leukemia virus Env protein. J Virol 1994; 68: 1773–1781.

  10. 10

    Cathomen T, Naim HY, Cattaneo R . Measles viruses with altered envelope protein cytoplasmic tails gain cell fusion competence. J Virol 1998; 72: 1224–1234.

  11. 11

    Ke SH, Coombs GS, Tachias K, Corey DR, Madison EL . Optimal subsite occupancy and design of a selective inhibitor of urokinase. J Biol Chem 1997; 272: 20456–20462.

  12. 12

    Springfeld C, von Messling V, Frenzke M, Ungerechts G, Buchholz CJ, Cattaneo R . Oncolytic efficacy and enhanced safety of measles virus activated by tumor-secreted matrix metalloproteinases. Cancer Res 2006; 66: 7694–7700.

  13. 13

    Ke SH, Coombs GS, Tachias K, Corey DR, Madison EL . Optimal subsite occupancy and design of a selective inhibitor of urokinase. J Biol Chem 1997; 272: 20456–20462.

  14. 14

    Ding L, Coombs GS, Strandberg L, Navre M, Corey DR, Madison EL . Origins of the specificity of tissue-type plasminogen activator. Proc Natl Acad Sci USA 1995; 92: 7627–7631.

  15. 15

    Egeblad M, Werb Z . New functions for the matrix metalloproteinases in cancer progression. Nat Rev Cancer 2002; 2: 161–174.

  16. 16

    Dano K, Behrendt N, Hoyer-Hansen G, Johnsen M, Lund LR, Ploug M et al. Plasminogen activation and cancer. Thromb Haemost 2005; 93: 676–681.

  17. 17

    Gage PJ, Levine M, Glorioso JC . Syncytium-inducing mutations localize to two discrete regions within the cytoplasmic domain of herpes simplex virus type 1 glycoprotein B. J Virol 1993; 67: 2191–2201.

  18. 18

    Tahara M, Takeda M, Yanagi Y . Altered interaction of the matrix protein with the cytoplasmic tail of hemagglutinin modulates measles virus growth by affecting virus assembly and cell-cell fusion. J Virol 2007; 81: 6827–6836.

  19. 19

    Johnson KJ, Peng KW, Allen C, Russell SJ, Galanis E . Targeting the cytotoxicity of fusogenic membrane glycoproteins in gliomas through protease-substrate interaction. Gene Therapy 2003; 10: 725–732.

  20. 20

    Allen C, McDonald C, Giannini C, Peng KW, Rosales G, Russell SJ et al. Adenoviral vectors expressing fusogenic membrane glycoproteins activated via matrix metalloproteinase cleavable linkers have significant antitumor potential in the gene therapy of gliomas. J Gene Med 2004; 6: 1216–1227.

  21. 21

    Harris JL, Backes BJ, Leonetti F, Mahrus S, Ellman JA, Craik CS . Rapid and general profiling of protease specificity by using combinatorial fluorogenic substrate libraries. Proc Natl Acad Sci USA 2000; 97: 7754–7759.

  22. 22

    Bateman AR, Harrington KJ, Kottke T, Ahmed A, Melcher AA, Gough MJ et al. Viral fusogenic membrane glycoproteins kill solid tumor cells by nonapoptotic mechanisms that promote cross presentation of tumor antigens by dendritic cells. Cancer Res 2002; 62: 6566–6578.

  23. 23

    Diaz RM, Bateman A, Emiliusen L, Fielding A, Trono D, Russell SJ et al. A lentiviral vector expressing a fusogenic glycoprotein for cancer gene therapy. Gene Therapy 2000; 7: 1656–1663.

  24. 24

    Linardakis E, Bateman A, Phan V, Ahmed A, Gough M, Olivier K et al. Enhancing the efficacy of a weak allogeneic melanoma vaccine by viral fusogenic membrane glycoprotein-mediated tumor cell–tumor cell fusion. Cancer Res 2002; 62: 5495–5504.

  25. 25

    Niwa H, Yamamura K, Miyazaki J . Efficient selection for high-expression transfectants with a novel eukaryotic vector. Gene 1991; 108: 193–199.

  26. 26

    Heussen C, Dowdle EB . Electrophoretic analysis of plasminogen activators in polyacrylamide gels containing sodium dodecyl sulfate and copolymerized substrates. Anal Biochem 1980; 102: 196–202.

  27. 27

    Togawa D, Koshino T, Saito T, Takagi T, Machida J . Highly activated matrix metalloproteinase-2 secreted from clones of metastatic lung nodules of nude mice injected with human fibrosarcoma HT1080. Cancer Lett 1999; 146: 25–33.

Download references


This work was supported in part by Research Grants from the 21st Century Center of Excellence Program, Chiba University Graduate School of Medicine, and by Grants-in-Aid (to YY) from the Japanese Ministry of Education, Culture, Sports, Science and Technology. We thank B Moss, D Kolakofsky, I Saito and H Iba for supplying experimental materials essential for this study; K Ishida, T Kanaya, T Yamamoto, M Yoshizaki, A Tagawa, E Suzuki and N Kohno for their excellent technical assistance; and T Hironaka, T Zhu and A Iida for helpful discussions.

Author information

Correspondence to H Kinoh.

Additional information

Supplementary Information accompanies the paper on Gene Therapy website (

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Kinoh, H., Inoue, M., Komaru, A. et al. Generation of optimized and urokinase-targeted oncolytic Sendai virus vectors applicable for various human malignancies. Gene Ther 16, 392–403 (2009) doi:10.1038/gt.2008.167

Download citation


  • uPA
  • MMP
  • oncolytic Sendai virus vector

Further reading