Sorting signals in the measles virus wild-type glycoproteins differently influence virus spread in polarized epithelia and lymphocytes

Measles virus (MV) infection is a highly contagious disease, still remaining one of the major causes of infant mortality in developing countries (WHO, 2008). In contrast, vaccination with live attenuated MV vaccine strains prevents MV-related fatalities. The differences in virulence between MV vaccine and wild-type strains rely to some extent on cell entry through different receptor molecules. To date, two cellular molecules have been identified to serve as MV entry receptors: CD46 (membrane cofactor protein), which is expressed on all nucleated cells (Dorig et al., 1994; Naniche et al., 1993), and CD150/signalling lymphocyte activation molecule (SLAM), whose expression is restricted to certain cells of haematopoietic origin, including activated B and T lymphocytes (Tatsuo et al., 2000). It is generally accepted that vaccine strains are able to use CD46 and CD150/SLAM for cell entry, whereas wild-type strains can use only CD150/SLAM; this is in agreement with the well-known lymphotropism of MV wild-type infections in vivo (for a review see Yanagi et al., 2006). In addition to CD46 and CD150/SLAM, there is evidence for an additional receptor on epithelial cells (Tahara et al., 2008; Takeda et al., 2007).

In a recent study of intratracheally infected macaques, CD150/SLAM-positive B and T lymphocytes and CD11c-positive myeloid cells were identified as major target cells of MV wild-type infection. Later in infection, epithelia of the mouth and trachea were also found to be infected (de Swart et al., 2007). Infection of epithelia with subsequent shedding of virus and/or infected cells may serve to allow efficient transmission of the virus between humans (Leonard et al., 2008; Takeda, 2008).

As a member of the genus Morbillivirus in the family Paramyxoviridae, MV cell entry and spread is mediated by the two surface glycoproteins inserted into the viral envelope: the receptor-binding haemagglutinin (H) and the fusion protein (F) (for review see Rima & Duprex, 2006). We have shown previously that both glycoproteins of the MV Edmonston vaccine strain (MV_Edm) contain tyrosine-dependent sorting signals in their cytoplasmic tails mediating the transport of both H and F to the basolateral domain of polarized epithelial cells, thereby facilitating virus spread via cell-to-cell fusion within epithelia and also in the respiratory tract of experimentally infected cotton rats (Moll et al., 2004). Furthermore, we demonstrated a crucial role of the transport signals in both MV_Edm glycoproteins for protein localization and virus spread within lymphocytes (Runkler et al., 2008). Effects on cell-to-cell fusion and spread due to changes in the MV glycoprotein surface distribution depend on the availability of cellular receptors on neighbouring cells. Since MV wild-type strains can only bind to CD150/SLAM and not to CD46 receptors, alterations in wild-type H and F surface localization might have different effects on virus spread. To investigate this difference, we generated recombinant MV_Edm containing both parental glycoproteins of the MV wild-type strain WTFb (Johnston et al., 1999), as well as viruses harbouring WTFb-H and WTFb-F proteins with mutations in the potential tyrosine signals in the cytoplasmic tail of one or both glycoproteins (tyrosine mutants). Using these viruses, we demonstrated that only the cytoplasmic tyrosine residue of WTFb-H influences virus spread in lymphocytes, whereas the sorting signals in both WTFb glycoproteins affect virus spread in epithelial cells.

Cell culture.
Madin–Darby canine kidney (MDCK) cells were grown in Eagle's minimal essential medium (MEM) supplemented with 10 % fetal calf serum (FCS), 100 U penicillin ml^–1 and 100 µg streptomycin ml^–1 (all from Gibco). BJAB cells (human B-cell line) were maintained in RPMI 1640 medium (Gibco) supplemented with 10 % FCS, penicillin and streptomycin. B95a (adherent marmoset B-cell line) cells were grown in Dulbecco's modified Eagle's medium (DMEM; Gibco) containing 10 % FCS and antibiotics. Vero-SLAM cells derived from Vero (African green monkey kidney) cells that stably express human CD150/SLAMw (Ono et al., 2001) were grown in DMEM containing 10 % FCS in the additional presence of 0.5 mg G418 ml^–1.

Construction of plasmids encoding full-length MV genomes.
To analyse the role of the tyrosine-dependent signals in the MV wild-type WTFb glycoproteins in the context of an MV infection, we generated recombinant MV_Edm harbouring the MV_Edm N, P, M and L proteins in combination with either the parental WTFb-glycoproteins (rWTFb) or glycoproteins in which aa 549 and 12 in F and H, respectively, were replaced with alanine (rWTFb tyrosine mutants; Fig. 1). For this, genes encoding WTFb-F, -F_549Y/A, -H or -H_12Y/A were cloned into plasmids containing the full-length MV_Edm genome. Plasmid p(+)MV(WTFb-H/F)Ed (Johnston et al., 1999), which encodes the antigenomic MV_Edm (Ed-tag) sequence with the Edmonston F and H genes substituted with the sequences of F and H of the wild-type strain WTFb, was a kind gift from J. Schneider-Schaulies (University of Würzburg, Germany). To generate glycoprotein-encoding expression plasmids, p(+)MV(WTFb-H/F)Ed was digested with EcoRI and NheI and the F-coding region was ligated into EcoRI–XbaI-cleaved pcDNA3 vector (Invitrogen) to yield pcDNA3-WTFb-F. A PacI–SpeI-cleaved fragment of p(+)MV(WTFb-H/F)Ed encoding H was inserted into PacI–SpeI-digested pCG vector (Cathomen et al., 1995) to construct pCG-WTFb-H. Mutant WTFb-F and WTFb-H genes were generated by introduction of site-specific mutations into the pcDNA3 and pCG expression plasmids using the Quikchange site-directed mutagenesis kit (Stratagene). Using complementary primers, aa 549 and 12 in the WTFb-F and WTFb-H genes, respectively, were changed from tyrosine to alanine generating the plasmids pcDNA3-WTFb-F_549Y/A and pCG-WTFb-H_12Y/A. To generate rMV, the coding region of pcDNA3-WTFb-F_549Y/A was excised by NarI and PacI and ligated to p(+)MV(WTFb-H/F)Ed to give the plasmid p(+)MV(WTFb-H/F_549Y/A)Ed. By ligating the PacI–SpeI fragment of pCG-WTFb-H_12Y/A to p(+)MV(WTFb-H/F)Ed the plasmid p(+)MV(WTFb-H_12Y/A/F)Ed was generated. To construct the double mutant p(+)MV(WTFb-(H/F)_Y/A)Ed, pCG-WTFb-H_12Y/A was again digested with PacI and SpeI and ligated to PacI–SpeI cleaved p(+)MV(WTFb-H/F_549Y/A)Ed. The sequence of each construct was confirmed by dideoxy sequencing.

(12K): Fig. 1. Amino acid sequence of the cytoplasmic domains of rWTFb and the rWTFb tyrosine mutants. Bold letters indicate the amino acid at positions 549 and 12 in F and H, respectively. TMD, transmembrane domain; CD, cytoplasmic domain.

Virus rescue.
rMVs were rescued essentially as described by Johnston et al. (1999). Briefly, after infection of HeLa cells with the attenuated vaccinia virus Ankara expressing T7 polymerase (MVA-T7) at an m.o.i. of 0.5, cells were co-transfected using Lipofectamine 2000 (Invitrogen) with one of the full-length plasmids p(+)MV(WTFb-H/F)Ed, p(+)MV(WTFb-H/F_549Y/A)Ed, p(+)MV(WTFb-H_12Y/A/F)Ed or p(+)MV(WTFb-(H/F)_Y/A)Ed and plasmids encoding the MV_Edm-N, -P and -L proteins (pEMC-Na, pEMC-Pa and pEMC-La). One day later, 2x10⁵ BJAB cells were added to the HeLa cells to allow further replication of successfully rescued virus. After 5 days, cell lysates were used to infect BJAB cells. rMV were harvested when 80–90 % of the cells showed cytopathic effects. The identity of the rescued viruses was confirmed by RT-PCR followed by dideoxy sequencing.

Generation of expression plasmids and establishment of stably transfected MDCK cells.
To generate expression plasmids for stable transfection, the plasmids pcDNA3-WTFb-F and -F_549Y/A were cleaved with EcoRI and BsrBI and ligated into EcoRI–SwaI-cleaved pcz-CFG5-fEGN vector (Lindemann et al., 2001) to construct pcz-WTFb-F and pcz-WTFb-F_549Y/A. EcoRI–PmeI fragments of pCG-WTFb-H and -H_12Y/A were ligated into EcoRI–SwaI-cleaved pcz-CFG5-fEGN yielding pcz-WTFb-H and pcz-WTFb-H_12Y/A, respectively. Furthermore, the region spanning the CD150/SLAM gene was amplified by PCR from pcDNA3-ESLAM (a kind gift from J. Schneider-Schaulies). During amplification, a SwaI cleavage site was introduced at the 3'-end of the PCR product. The EcoRI–SwaI-cleaved PCR product was then ligated into the cleaved pcz-CFG5-fEGN vector to obtain pcz-ESLAM.

For stable expression, MDCK cells were transfected with pcz-WTFb-F, pcz-WTFb-F_549Y/A, pcz-WTFb-H, pcz-WTFb-H_12Y/A or pcz-ESLAM using Lipofectamine 2000 according to the manufacturer's protocol. Transfected cells were selected in medium containing 1 mg Zeocin ml^–1 (InvivoGen). Resistant cells were subcloned and assayed for protein expression and polarized growth.

Animal infections and titration.
Inbred cotton rats were obtained from Harlan. For intranasal infection, parental or mutant rWTFb was given in PBS to five 6-week-old, isofluorane-anaesthetized female cotton rats in a volume of not more than 100 µl. At day 4 post-infection (p.i.), animals were euthanized using CO₂ and the lungs were removed and weighed. Lung tissue was minced with scissors and dounced with a glass homogenizer. Lung lavage was performed using a three-way stopcock and 10 ml PBS/1 % EDTA. Serial 10-fold dilutions were assessed for the presence and levels of infectious virus. Infectivity was quantified by the TCID₅₀ method using B95a cells, as described previously (Pfeuffer et al., 2003).

Virus growth analysis.
Virus growth was analysed by infecting 2.5x10⁵ BJAB cells with rWTFb, rWTFb-F_549Y/A, rWTFb-H_12Y/A or rWTFb-FH_Y/A at an m.o.i. of 0.1. After 2 h, cells were washed to remove unbound virus and were cultured in 1 ml RPMI 1640 containing 10 % FCS at 37 °C. Every 12 h, cells were pelleted by low-speed centrifugation and cell-free rMV in the supernatant was titrated by plaque assay. Dilutions of the cell supernatant were adsorbed to Vero-SLAM cells for 2 h, then cells were overlaid with MEM containing 2 % FCS and 0.9 % Bacto Agar (BD). After 4 days, plaques were stained with 0.0125 % neutral red (Merck) and counted.

To analyse syncytia formation in infected cells, B95a or BJAB cells were infected with the different rMVs at an m.o.i. of 0.05. After washing, cells were cultured at 37 °C in medium containing 10 % FCS. Cells were monitored by phase-contrast microscopy and photographed. Nuclei within syncytia were counted after Giemsa staining at day 2 p.i., as described previously (Moll et al., 2002).

Immunostaining.
B95a cells were seeded onto coverslips and directly infected with different rMVs at an m.o.i. of 0.5. At 8 h p.i., cells were washed and cultured in DMEM plus 10 % FCS together with a fusion-inhibitory peptide to prevent disruption of the cells by syncytia formation (Weidmann et al., 2000). At 48 h p.i., cells were incubated with F- or H-specific monoclonal antibodies (mAbs) (K83 and A504, kindly provided by J. Schneider-Schaulies) for 1 h at 4 °C to visualize MV glycoproteins on the cell surfaces. For M staining, cells were treated with methanol/acetone (1 : 1) for 5 min and then labelled with the M-specific mAb 8910 (Chemicon). All primary antibodies were detected by incubation with rhodamine-conjugated goat anti-mouse IgG (Dianova) for 45 min at 4 °C. After immunostaining, cells were mounted in Mowiol (Merck) and 10 % 1,4 diazabicyclo(2.2.2)octane (Sigma) and fluorescence images were recorded using a Zeiss ApoTome/Axiovert 200M microscope.

Domain-selective surface biotinylation and immunoprecipitation.
MDCK cells stably expressing WTFb-F, WTFb-F_549Y/A, WTFb-H or WTFb-H_12Y/A, or MDCK-SLAM cells infected with parental or mutant rWTFb were cultivated on 0.4 µm pore size PET Transwell filters (Greiner Bio-One). To control cell polarity, the transepithelial resistance (TER) was measured using a Millicell-ERS apparatus. Analyses were only performed with cells that gave TER values above 200 Ω cm^–2. At 72 h p.i. and 35 h post-seeding, cells were washed and either the apical or basolateral side of the membranes was incubated twice for 15 min with PBS containing 2 mg S-NHS-biotin ml^–1 (Calbiochem) on ice, while 0.1 M glycine was added to the opposite side of the membrane. After washing with 0.1 M glycine, filter membranes were cut and cells were lysed in 0.5 ml radioimmunoprecipitation assay buffer (1 % Triton X-100, 1 % sodium deoxycholate, 0.1 % SDS, 0.15 M NaCl, 10 mM EDTA, 10 mM iodacetamide, 1 mM PMSF, 5 % aprotinin, 20 mM Tris/HCl, pH 8.5). After clarifying cell lysates by centrifugation at 19 000 g, F and H proteins were immunoprecipitated using F-specific rabbit-antiserum (Fcyt, kindly provided by R. Cattaneo) or H-specific (K83) antibodies and protein A-Sepharose beads (Sigma). Following SDS-PAGE and blotting to nitrocellulose, proteins were detected with streptavidin-biotinylated horseradish peroxidase complex (Amersham) and enhanced chemiluminescence (Pierce). Relative protein amounts on the basolateral and apical domains were quantified using Adobe Photoshop CS3 Extended software.

Fusion assay in polarized cells.
To analyse glycoprotein-induced fusion in polarized MDCK cells, MDCK-SLAM cells were seeded in high density onto coverslips. After 40 h, cells were transfected to co-express parental and mutant F and H proteins in different combinations. At 8 h post-transfection (p.t.), the medium was removed and cells were incubated in DMEM plus 10 % FCS either with or without calcium. To visualize syncytium formation, cells were fixed with methanol/acetone (1 : 1) and incubated with an H-specific mAb (K83). Tight junctions were visualized with a ZO-3-specific antibody (anti-ZO-3). Primary antibodies were detected with rhodamine-conjugated goat anti-mouse immunoglobulins. Afterwards, cells were mounted as described above and analysed using a Zeiss ApoTome/Axiovert 200M microscope.

Role of the cytoplasmic tyrosines for rWTFb release and cell-to-cell fusion in lymphocytes
As in the MV_Edm vaccine strain, the cytoplasmic domains of the glycoproteins of the wild-type strain WTFb contain a single tyrosine residue at positions 549 and 12 in the F and H proteins, respectively. To determine the role of the cytoplasmic tyrosines in the WTFb glycoproteins, we generated recombinant MV_Edm harbouring either the parental WTFb-glycoproteins (rWTFb) or glycoproteins in which the single tyrosine residues in F and H were replaced with alanine (rWTFb tyrosine mutants, Fig. 1) as described in Methods. To analyse the significance of the tyrosine-dependent sorting signals in the WTFb glycoproteins for replication in lymphocytes, BJAB cells (a CD150/SLAM-positive human B cell line) were infected with rWTFb, rWTFb-F_549Y/A, rWTFb-H_12Y/A or rWTFb-FH_Y/A and virus release into the supernatant was quantified by plaque assay. rWTFb-F_549Y/A did not show a prominent difference in virus growth compared with the parental rWTFb virus; newly synthesized particles were released into the supernatant with the same efficiency (Fig. 2). In contrast, virus titres in the supernatant of rWTFb-H_12Y/A- and rWTFb-FH_Y/A-infected cells showed a more than 50-fold reduction (Fig. 2). This shows that mutation of the sorting signal in the WTFb-H only, and not in WTFb-F, interferes with virus release.

(11K): Fig. 2. Growth of rWTFb and the tyrosine mutants. BJAB cells were infected with rWTFb or rWTFb tyrosine mutants at an m.o.i. of 0.1 and incubated at 37 °C. Every 12 h, cell-free viruses were titrated by plaque assay. rWTFb (•), rWTFb-F549Y/A (), rWTFb-H12Y/A () and rWTFb-FHY/A () are shown. Values plotted represent mean results of two independent experiments.

To test the influence on virus spread via cell-to-cell fusion, syncytia formation in BJAB and B95a cells (a marmoset B cell line) was investigated. In Fig. 3, syncytia formation of cells infected with parental or mutant rWTFb at 2 days p.i. is shown. rWTFb-F_549Y/A-infected BJAB and B95a cells did not show a prominent difference in syncytia formation compared to rWTFb infection. In contrast, cell fusion of rWTFb-H_12Y/A- and rWTFb-FH_Y/A-infected cells was obviously enhanced. When we quantified fusion by counting and averaging the number of nuclei per syncytium of 20 randomly chosen syncytia, we found that syncytia in rWTFb- and rWTFb-F_549Y/A-infected cells contained about 130 and 136 nuclei, whereas syncytia caused by infections with rWTFb-H_12Y/A or rWTFb-FH_Y/A produced syncytia with 329 or 306 nuclei on average, respectively. Thus, only the substitution of the cytoplasmic tyrosine in the WTFb-H but not in the WTFb-F protein enhanced fusogenicity and interfered with virus release. Hyperfusogenic properties of rWTFb-H_12Y/A and rWTFb-FH_Y/A not only prevented production of cell-free viruses but also similarly reduced cell-associated virus titres, indicating that extensive cytopathic effects destroy the cells before infectious particles can be assembled intracellularly. Interestingly, this result is in clear contrast with what we have observed for the MV_Edm glycoproteins. Here, tyrosine mutations in both H and F had similar enhancing effects on the fusogenic properties, and virus release was only impaired if both glycoproteins were mutated (Runkler et al., 2008).

(63K): Fig. 3. Fusion activity of rWTFb expressing parental or mutant glycoproteins. B95a and BJAB cells were infected with rWTFb, rWTFb-F549Y/A, rWTFb-H12Y/A and rWTFb-FHY/A and monitored for cytopathic effect by phase-contrast microscopy. Syncytia formation 2 days p.i. is shown. Bar, 400 µm.

Distribution of MV envelope proteins in rWTFb-infected lymphocytes
Cell-to-cell fusion of MV-infected cells can be regulated by interaction of the glycoproteins with the third envelope protein M (Cathomen et al., 1998a, b). To elucidate the impact of the glycoprotein–M interaction on differences in virus spread, surface distribution of all three envelope proteins in infected B95a cells was analysed. To visualize F and H on cell surfaces, cells infected with parental or mutant rWTFb were cooled on ice at 48 h p.i. and labelled with F- or H-specific primary antibodies. To detect the M protein, fixed cells were incubated with an M-specific mAb. Primary antibodies were detected using rhodamine-conjugated anti-mouse IgG. As shown in Fig. 4, F and H accumulated in large aggregates on the cell surface of cells infected with parental rWTFb. In cells infected with rWTFb tyrosine mutants, mutated glycoproteins showed a more punctuate distribution (F in rWTFb-F_549Y/A, H in rWTFb-H_12Y/A), whereas non-mutated glycoproteins were still found in large aggregates. In rWTFb-FH_Y/A-infected cells, neither F nor H accumulated in large clusters at the cell surface. Interestingly, the M protein was found in large aggregates underneath the plasma membrane in all infections, indicating that distribution of M protein does not depend on H or F localization. The finding that only wild-type WTFb glycoproteins but not H or F proteins containing tyrosine mutations colocalized with M on the surface of infected cells (Supplementary Fig. S1, available in JGV Online) clearly indicates that the interaction of M with both glycoproteins is prevented by tyrosine substitution. These observations show that even if only the tyrosine-dependent sorting signal in the WTFb-H protein influences virus spread, the cytoplasmic tyrosines in both glycoproteins affect surface distribution and interaction with M in lymphocytes. Thus, changes in the surface distribution of the glycoproteins are likely not responsible for the enhanced cell-to-cell fusion and the less-efficient virus production in rWTFb-H_12Y/A-infected lymphocytes.

(76K): Fig. 4. Localization of F, H and M in cells infected with rWTFb and rWTFb tyrosine mutants. B95a cells were infected with rWTFb, rWTFb-F549Y/A, rWTFb-H12Y/A or rWTFb-FHY/A. At 48 h p.i., cell surfaces were labelled using either monoclonal anti-F or anti-H antibodies. To visualize the M protein, cells were fixed and permeabilized, and stained with a monoclonal anti-M antibody. Primary antibodies were detected by incubation with rhodamine-conjugated anti-mouse immunoglobulins. Bar, 10 µm.

Role of cytoplasmic tyrosines in WTFb glycoprotein for protein targeting in polarized epithelial cells
To investigate the role of the tyrosine signals in the WTFb wild-type glycoproteins for virus spread in polarized epithelial cells, we generated MDCK cell lines stably expressing the parental WTFb-F or -H proteins or the corresponding tyrosine mutants as described in Methods. To analyse the targeting of the expressed proteins, cells were cultured on permeable filter supports, where they form polarized monolayers. Polarized cells grown on filters were cooled to 4 °C and surface proteins of either the apical or the basolateral domain were labelled by adding the non-membrane-permeating reagent S-NHS-biotin to the respective filter chamber. After cell lysis and immunoprecipitation of F and H by specific antibodies, proteins were separated by SDS-PAGE and transferred to nitrocellulose. Biotinylated proteins were then detected using peroxidase-conjugated streptavidin. As shown in Fig. 5(a), efficient biotinylation of the parental F and H proteins was obtained after labelling the cells from the basolateral side (lanes 2 and 6). This indicates that the WTFb glycoproteins are expressed on the basolateral surface of stably expressing MDCK cells (65 and 76 % of the F and H proteins, respectively). In contrast, both tyrosine mutants were found to be apically expressed (lanes 3 and 7). Only a minor band (less than 5 %) was found after basolateral labelling (lanes 4 and 8). The redirection of the mutant proteins to the apical cell membrane indicates that the tyrosine residues serve as tyrosine-dependent basolateral sorting signals for both WTFb glycoproteins in polarized epithelial cells.

(30K): Fig. 5. Surface distribution of parental or mutant WTFb glycoproteins in polarized MDCK cells. MDCK cells stably expressing WTFb-F, WTFb-F549Y/A, WTFb-H or WTFb-H12Y/A (a) or MDCK cells stably expressing CD150/SLAM infected with rWTFb, rWTFb-F549Y/A, rWTFb-H12Y/A or rWTFb-FHY/A (b) were grown on permeable filter supports. At 40 h post-seeding (a) or 35 h p.i. (b), cells were surface-biotinylated from either the apical (Ap) or the basolateral (Bl) side. After cell lysis, proteins were immunoprecipitated with F- or H-specific antibodies. Precipitated proteins were analysed by SDS-PAGE, transferred to nitrocellulose and probed with peroxidase-conjugated streptavidin.

To analyse targeting of the glycoproteins during MV infection, MDCK cells stably expressing the wild-type receptor CD150/SLAM were generated. MDCK-SLAM cells were infected with parental (rWTFb) and mutant rMV (rWTFb-F_549Y/A, rWTFb-H_12Y/A or rWTFb-FH_Y/A), grown on filters and subjected to a domain-specific surface biotinylation assay. Similar to our observations in stably expressing cells, the WTFb glycoproteins were expressed at the basolateral domain of rWTFb-infected MDCK cells (Fig. 5b, lanes 2 and 4). The non-mutated glycoproteins, the H protein in rWTFb-F_549Y/A-infected cells (lane 6) and the F-protein in rWTFb-H_12Y/A-infected cells (lane 12) were also located at the basolateral domain, whereas the mutated F in rWTFb-F_549Y/A (lane 7) and the mutated H in rWTFb-H_12Y/A (lane 9) were found at the apical surface. As expected, in rWTFb-FH_Y/A-infected cells, both mutated glycoproteins were expressed apically (lanes 13 and 15). This indicates that in the context of a virus infection as well, the tyrosine residues in the cytoplasmic tails of both WTFb-glycoproteins mediate transport to the basolateral domain of polarized epithelial cells.

Functional importance of basolateral WTFb-glycoprotein expression in polarized epithelial cells
To elucidate the importance of the basolateral sorting of WTFb-F and -H for MV spread within epithelia, we analysed whether inactivation of the targeting signal affects the fusion activity of the WTFb glycoproteins in polarized MDCK cells. For this, MDCK-SLAM cells were grown polarized on coverslips and then transfected with plasmids encoding the parental WTFb-glycoproteins or the tyrosine mutants in different combinations. At 6 h p.t., culture medium was replaced with either calcium-depleted (–Ca²⁺) or normal (+Ca²⁺) growth medium. In the absence of calcium, tight junctions cannot be maintained or formed, as confirmed by the lack of ZO-3 staining, a typical tight junction marker (Fig. 6, –Ca²⁺). Therefore, cells cultured in calcium-depleted medium are non-polarized and have no separated apical and basolateral membrane domains. At 18 h p.t., cells were fixed and stained with an MV-H-specific antibody and a rhodamine-conjugated secondary antibody. As expected, all cells grown in medium without calcium showed large syncytia formation (Fig. 6, –Ca²⁺ panels). In contrast, fusion of polarized cells (+Ca²⁺) was only detected when cells co-expressed both parental WTFb-glycoproteins (F+H). In cells expressing either one (F_549Y/A+H and F+H_12Y/A) or both (F_549Y/A+H_12Y/A) tyrosine mutants, fusion was largely prevented. Similar results were obtained with SLAM-negative MDCK cells (Supplementary Fig. S2, available in JGV Online). This shows that fusion of polarized epithelial cells is only possible when both F and H are present at the (baso-)lateral cell surfaces, and demonstrates that only lateral membranes can fuse in polarized cell monolayers. Apically expressed proteins are unable to mediate fusion. Thus, the cytoplasmic tyrosines of both WTFb-glycoproteins are indispensable for glycoprotein-induced cell-to-cell fusion of polarized epithelia. The tyrosine-dependent sorting signals in the WTFb glycoproteins seem, therefore, to be of different importance in the two cell types analysed. Although only the transport signal in the H protein influences virus propagation in lymphocytes, signals in the F and H protein are equally important for virus spread in epithelial cells.

(35K): Fig. 6. Syncytia formation in polarized MDCK cells. Forty hours after seeding on coverslips at high density, MDCK cells stably expressing CD150/SLAM were transfected to co-express parental or mutant F and H proteins in different combinations and were incubated in medium without (–Ca2+) or with (+Ca2+) calcium. Eighteen hours p.t., cells were fixed and tight junctions were visualized with a ZO-3-specific antibody (anti-ZO-3). Transfected cells were stained with the MV H-specific K83 antibody (anti-MV) and rhodamine-conjugated secondary antibodies. Bar, 100 µm.

Replication of rWTFb in cotton rats
To determine whether mutations in the cytoplasmic tyrosines of WTFb-H and WTFb-F also differently influence MV replication in vivo, we used cotton rats, an animal model system that is susceptible to respiratory infection with MV (Niewiesk, 2009). In lungs of cotton rats infected with WTFb, epithelial cells, pneumocytes and macrophages are infected (data not shown). Six-week-old cotton rats were infected intranasally with 10⁵ TCID₅₀ of rWTFb, rWTFb-F_549Y/A or rWTFb-FH_Y/A. At day 4 p.i., the virus titre in lung lavage cells, which consist of leukocytes, and lung tissue was determined (Table 1). In agreement with our findings that the tyrosine mutation in the WTFb-F protein does not influence MV replication in lymphocytes, virus titres in lung lavage cells isolated from rWTFb-F_549Y/A-infected animals were not reduced compared with the titres found in animals infected with parental rWTFb (10^3.5 versus 10^3.3). In contrast, when the H protein was additionally mutated, as in the mutant rWTFb-FH_Y/A, titres were clearly reduced (10^2.3). This result strongly suggests that only a tyrosine mutation in the WTFb-H protein impairs MV replication in leukocytes in vivo, and supports the observations in cultured lymphocytes.

Table 1. Replication of parental and mutant rWTFb in cotton rat lungs Groups of five animals were infected with 105 TCID50 of rWTFb, rWTFb-F549Y/A or rWTFb-FHY/A. On day 4 p.i., the virus titres in the lung lavage cells and lung tissue were assessed on B95a cells. Lung lavage cells were pooled from five animals, virus titres in the lung tissue were determined for each individual animal and the mean was calculated. Titres in lung lavage cells: P value between rWTFb and rWTFb-FHY/A, 0.008; P value between rWTFb-F549Y/A and rWTFb-FHY/A, 0.009. Titres in lung tissue: P value between rWTFb and rWTFb-FHY/A, 0.00076; P value between rWTFb and rWTFb-F549Y/A, 0.00519.

In contrast with the titres in lung lavage cells, virus titres recovered from lung tissue were reduced for both mutants rWTFb-F_549Y/A and rWTFb-FH_Y/A (10^3.14 and 10^3.04 versus 10^4.1 in parental rWTFb). This result suggests that a single F mutation as well as mutations in both WTFb glycoproteins similarly impair virus replication in epithelial cells in vivo, thereby confirming the previous results in cultured MDCK cells. Lymphocytes and respiratory epithelial cells constitute the main target cells during MV infection (de Swart, 2008; Leonard et al., 2008). In this study, we showed that the tyrosine residues in the cytoplasmic tails of the glycoproteins F and H of the WTFb wild-type strain (Y₅₄₉ in the WTFb-F and Y₁₂ in the WTFb-H protein) are of critical importance for virus spread within these cell types. In lymphocytes, infection with recombinant MV expressing H proteins with a mutated cytoplasmic tyrosine resulted in an enhanced capacity to induce cell-to-cell fusion, and was accompanied by a less effective virus production in these cells. In polarized epithelial cells, the cytoplasmic tyrosine residues in both glycoproteins represent sorting signals mediating transport of WTFb-F and -H to the basolateral domain of stably expressing, as well as infected cells. Basolateral expression of both glycoproteins was demonstrated to be crucial for F/H-induced fusion of neighbouring epithelial cells. Our data in cotton rats are consistent with the concept that WTFb-H is required for virus production in lymphocytes and both WTFb-H and -F are required for virus spread between epithelial cells.

As epithelial cells form sheets that line surfaces and often form the interface between the interior and exterior of the organism (Duffield et al., 2008), pathogens like MV which infect via the respiratory route have to overcome an epithelial cell barrier in order to establish a systemic infection. This can be achieved by directly infecting epithelial cells in the respiratory tract and subsequent spread to underlying tissues followed by systemic dissemination. However, a recent study of intratracheally infected macaques indicated that not epithelial cells but cells of the immune system are the first cells infected by MV. MV is likely trapped by dendritic cells (DCs) in the respiratory tract by attachment to the C-type lectin DC-SIGN and is subsequently transported to the tonsils (de Swart et al., 2007; de Witte et al., 2006). During systemic MV replication, lymphocytes were proven to be major MV target cells, whereas replication in respiratory epithelial cells was only found later in infection (de Swart, 2008). Thus, MV infects polarized epithelia late in infection, what is likely required for virus shedding and transmission from host to host. Fitting into this model, it has been shown that MV infects polarized epithelial cells from the basolateral domain and is released apically (Leonard et al., 2008; Maisner et al., 1998; Tahara et al., 2008). The fact that basolateral expression of both WTFb glycoproteins is required for syncytia formation within epithelial cells, suggests that tyrosine-dependent sorting of both F and H is essential for MV spread within the epithelia via cell-to-cell fusion in order to disrupt the epithelial barrier and to be shed efficiently into the airway lumen.

Generation and maintenance of the polarized epithelial cell phenotype is dependent on targeted transport of proteins to either the apical or the basolateral domain. Association of proteins with lipid rafts and cholesterol- and sphingolipid-rich domains plays a crucial role in the segregation of proteins to the apical side (Schuck & Simons, 2004), whereas N- and O-linked glycans are recessive signals for apical transport that only become evident when the dominating targeting signal to the basolateral domain has been removed (Scheiffele et al., 1995; Yeaman et al., 1997). Basolateral targeting depends on specific amino acid motifs in the cytosplasmic domains of transmembrane proteins. The most common motif contains a single tyrosine residue within the consensus motif YXXΦ (where X is any amino acid and Φ is an amino acid with a bulky hydrophobic side chain). In addition, the tetrapeptide NPXY, dileucine-motifs and single leucine residues are able to mediate basolateral transport (Muth & Caplan, 2003; Rodriguez-Boulan et al., 2005). Even if the tyrosine-based motifs in both WTFb-F and -H do not fit perfectly to the known tyrosine-dependent sorting signals YXXΦ and NPXY, they mediate basolateral targeting that can be disrupted by substitution of the critical tyrosine by alanine. The resulting apical sorting of the tyrosine mutants is likely mediated by the N-glycans within both proteins in the absence of the basolateral targeting signal (Alkhatib et al., 1994; Hu et al., 1994). In rWTFb-F_549Y/A- or rWTFb-H_12Y/A-infected cells, only the mutated glycoprotein was retargeted to the apical domain, whereas the second non-mutated glycoprotein was still found at the basolateral side of infected polarized cells. This clearly suggests that surface transport of WTFb-H and -F is not interdependent. In agreement with this, changes in the distribution of one mutant glycoprotein in lymphocytes did not affect surface distribution of the other non-mutated protein.

Binding of the H protein to a cellular receptor is believed to induce a conformational change in the H protein, followed by changes in the complexed F protein. The hydrophobic fusion peptide at the N terminus of the F₁ subunit is then exposed and inserted into the plasma membrane of the target cell. Further structural changes in the F protein pull the virus and cell membranes into close proximity, resulting in fusion into one single membrane (Yanagi et al., 2006; Zhu et al., 2002). MV-induced membrane fusion thus requires the presence of both glycoproteins. Consequently, mutation in only one protein preventing basolateral protein expression of either H or F inhibited syncytia formation in polarized epithelial cell layers. In contrast with polarized epithelial cells, cell-to-cell fusion of infected lymphocytes is only influenced by the mutation of the tyrosine-dependent sorting signal in the WTFb-H protein. This reveals a principal difference in the influence of the tyrosine signals in the wild-type glycoproteins requiring CD150/SLAM as a receptor for cell-to-cell fusion, compared with H and F proteins of vaccine strains that can also use CD46. For the glycoproteins of the MV_Edm vaccine strain, we have observed that intact tyrosine signals are essential for the interaction of F and H with the M protein, resulting in the formation of large glycoprotein clusters on the surface of non-polarized lymphocytes. Tyrosine mutations in only one of the MV_Edm glycoproteins enhanced cell-to-cell fusion in lymphocytes, whereas single mutations in either H or F did not interfere with virus production and release (Runkler et al., 2008). In contrast with this, we demonstrated in this study that mutation of the tyrosine signal in the wild-type F protein did not affect cell-to-cell fusion or virus production, whereas tyrosine mutation in the wild-type WTFb-H protein increased the fusogenicity and significantly interfered with MV propagation in lymphocytes. This finding clearly indicates different requirements for MV wild-type glycoprotein-mediated fusion compared with the fusion caused by H and F proteins of MV vaccine strains. So far, it has been assumed that disruption of the tyrosine-dependent interaction of F or H with M favours cell-to-cell fusion because deficient M–glycoprotein interaction enhances lateral mobility of the glycoproteins in the plasma membrane and facilitates the formation of active H–F fusogenic complexes (Moll et al., 2002; Runkler et al., 2008). However, this study indicates that disrupting the interaction of M with WTFb-F did not increase fusogenicity. Only mutation of WTFb-H resulted in enhanced cell-to-cell fusion in lymphocytes, suggesting that it is not the formation of H–F complexes but the encounter of H with its receptor CD150/SLAM that is the fusion-limiting step. If H mobility is increased as a result of a disrupted M–H interaction in rWTFb-H_12Y/A- or rWTFb-FH_Y/A-infected lymphocytes, receptor finding and binding is probably facilitated, resulting in an increased or more rapid formation of fusion pores. The fact that rWTFb-H_12Y/A and rWTFb-FH_Y/A showed the same increase in fusogenicity emphasizes again that the mobility of wild-type F is not a limiting factor in this scenario. The finding that viruses with an increased fusion activity produced at least 10-fold lower titres in lymphocytes clearly suggests that the tyrosine-dependent signal in the WTFb-H protein is required for efficient virus propagation within these cells, most probably by limiting cell-to-cell fusion and thus preventing syncytia formation and too early cell death.

We thank J. Schneider-Schaulies, Institute of Immunobiology and Virology, University of Würzburg, Germany, and R. Cattaneo, Mayo Clinic College of Medicine, Rochester, USA, for providing MV-specific antibodies and plasmids. This work was supported by grants from the German Research Foundation (DFG) to A. M. (MA 1886/4-4, MA 1886/5-3 and SFB 535/TP A17).

Footnotes

†Present address: National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 1AA, UK.

Two supplementary figures showing colocalization of M with F and H in rWTFb-infected B95a cells, and syncytia formation in SLAM-negative MDCK cells are available with the online version of this paper.