**3. Results**

Endocytosis of HeV and NiV F protein precursors represents a critical step to gain biological activity, and thus, viral infectivity. In order to investigate whether CedV F protein similarly undergoes endocytosis from the cell surface, we performed a qualitative antibody uptake assay as described previously [11]. Therefore, MDCK-2 cells were transfected with plasmid DNA encoding for the CedV F protein. At 24 h p.t. without prior fixation, cell-surface expressed CedV F protein was labeled with CedV F protein-specific antibodies. Next, cells were either kept on ice or shifted to 37 ◦C for 30 min to allow endocytosis to occur. Surface-bound antibodies were then detected with an AF 488-conjugated secondary antibody that was added in excess to saturate surface-bound primary antibodies. Subsequent permeabilization allowed the staining of intracellular F protein-antibody complexes with an AF 568-conjugated antibody. After incubation at 37 ◦C, cells expressing the CedV F protein showed both green surface staining and multiple red fluorescent intracellular vesicles indicating endocytosis of the labeled CedV F protein (Figure 1). In contrast, we observed only green fluorescent signals of F proteins in cells that were kept on ice su ffering a temperature-related inhibition of endocytosis.

**Figure 1.** Endocytosis of CedV F protein in MDCK-2 cells. At 24 h p.t., CedV F-expressing MDCK-2 cells were incubated with a CedV F-specific rabbit antiserum to label surface-expressed F proteins. Then, the cells were shifted to 37 ◦C for 30 min to allow endocytosis to occur. AlexaFluor (AF) 488-conjugated secondary antibodies visualized surface-bound primary antibodies. After fixation and permeabilization, AF 568-conjugated secondary antibodies were used to stain internalized primary antibody-CedV F protein complexes. Magnification, ×60. n = 3.

To evaluate the functional importance of the tyrosine-based motifs within the CedV F protein cytoplasmic tail for endocytosis, we generated seven F protein mutants by tyrosine to alanine substitution (Figure 2). We constructed mutants encoding either single (e1–e5) or multiple (e3, e6) mutations as well as one mutant devoid of all putative endocytosis signals (e7) to investigate potentially additive effects of these mutated motifs (Figure 2). Apart from a classical YXXΦ motif in the membrane-proximal region (Y524XXF, Figure 2), we chose to also mutate the tyrosine residue at amino acid position 522 (Y522XXN, Figure 2) since a similar motif present in the cytoplasmic domain of Measles virus hemagglutinin has been described earlier to affect basolateral sorting and internalization of the protein [45].

**Figure 2.** Schematic overview of the cytoplasmic tail mutants of CedV F protein generated in this study. Tyrosine-based putative endocytosis motifs are underlined. F1: CedV F protein subunit F1; F2: CedV F protein subunit F2; TM: transmembrane domain; CD: cytoplasmic domain; HA: HA-tag at the C-terminus of the F1 subunit; S-S: di-sulfide bond. wt: wild-type.

To assess whether these mutations have an impact on F protein synthesis, we compared expression efficiencies using a pulse-chase analysis. At 24 h p.t., MDCK-2 cells were labeled metabolically with [35] cysteine and -methionine for 15 min. After pulse-labeling, cells were incubated for 2 h followed by immunoprecipitation and SDS-PAGE under reducing conditions. We observed similar amounts of expression for parental and mutant F proteins within 2 h demonstrating that mutations within the cytoplasmic tail do not impair protein synthesis (Figure 3). However, although mutant e7 was expressed at similar levels, proteolytic processing was much less efficient. We generated additional F protein constructs combining selected mutations of particular motifs (e8 = e3 + e4; e9 = e3 + e5). Since these mutants displayed no differences in proteolytic processing to F wt (Supplementary Figure S1), they were not included in subsequent analyses. Together, these findings sugges<sup>t</sup> that proteolytic activation of the CedV F protein is markedly impaired upon the simultaneous disruption of all tyrosine-based motifs.

**Figure 3.** Effect of cytoplasmic tail mutations on CedV F protein expression and cleavage. At 24 h p.t., MDCK-2 cells expressing different CedV F proteins were metabolically labeled for 15 min (pulse) and then incubated for 2 h in serum-free nonradioactive medium (chase). After immunoprecipitation of F proteins from cell lysates and separation on a 12% SDS-gel under reducing conditions, samples were analyzed by autoradiography. n = 2; wt: wild-type.

HeV and NiV F proteins are known to be cleaved within the endosomal compartment. In analogy, blocking clathrin-mediated endocytosis by addition of chlorpromazine also inhibited proteolytic processing of CedV F protein (see Supplementary Figure S2) demonstrating that endocytosis is a prerequisite for CedV F cleavage. Thus, inefficient cleavage of CedV mutant e7 might result from an inability to undergo endocytosis. In order to study the effects of single and multiple cytoplasmic tail mutations on CedV F protein endocytosis, we performed another antibody uptake experiment as described above using all generated mutants. After a period of 30 min at 37 ◦C, we observed intracellular red vesicles for all mutants except mutant e7 with all tyrosine-based motifs disrupted (Figure 4a). Disruption of a single motif such as YXXN (mutant e1) or YXXF (mutant e2), the di-tyrosine motifs (mutants e4, e5) as well as the combination of several mutated tyrosine residues (mutants e3, e6) displayed no difference in comparison to the fluorescent signals of the parental CedV F protein (Figure 1). The qualitative finding of red fluorescent vesicles indicates the internalization of mutants e1 to e6 from the cell surface while endocytosis of mutant e7 was strongly impaired after 30 min of endocytosis.

To further understand the effects of single and multiple mutations on endocytosis, we next aimed to quantify the internalization of the different CedV F proteins and compare their endocytosis rate with the wt CedV F protein in a semi-quantitative biotin internalization assay. Moreover, this assay allowed us to demonstrate that the internalization of F proteins is not induced by antibody cross-linking. Briefly, we performed a surface biotinylation assay using a cleavable NHS-SS biotin derivative. Using the membrane-impermeable reducing agen<sup>t</sup> MESNA, residual biotin that was not internalized from the cell surface after incubation at 37 ◦C for different periods can be cleaved. Cell lysis, immunoprecipitation of F proteins, and Western blot using Streptavidin-HRP were performed with subsequent detection and quantification of intracellular biotinylated F proteins. The amount of internalized protein was quantified by comparison to F-expressing cells that were neither incubated at 37 ◦C nor treated with MESNA, therefore, representing the total amount of biotinylated F proteins (Ctr lane, Figure 4b). Since internalization rates did not appear to be linear over time, endocytosis of biotinylated proteins is displayed additionally as a function of the time in Figure 4c. More than 80% of the parental CedV F protein was internalized after 30 min (Figure 4b,c). Similarly, mutant e1 carrying a substitution of a single tyrosine residue at aa position 522 as well as the di-tyrosine motif mutants e4, e5, and e6 displayed only a marginal decrease in internalization compared to the parental F protein. Interestingly, mutation of the YXX Φ motif (mutant e2 and e3) clearly decreased internalization with weak signals detectable after a period of 15 to 30 min at 37 ◦C (Figure 4b). However, the strongest effects were observed for mutant e7. Here, biotinylated mutant e7 was only detected after 90 min of incubation at 37 ◦C (Figure 4b) leading to a drastically reduced internalization rate over time in contrast to the parental F protein and the other mutants (Figure 4c). An antibody uptake assay of CedV F proteins including a co-staining of endosomal marker protein early endosomal antigen-1 (EEA1) in HeLa cells confirmed the delayed uptake of mutants e2 and e3. After 5 min at 37 ◦C, we observed no internalization of CedV F mutants e2 and e3 and thus, no co-localization with EEA1 (Figure 4d). However, both mutants co-localized with EEA1 after 30 min. In contrast, the parental F protein showed intracellular staining and co-localization with EEA1 as early as 5 min after the shift to 37 ◦C (Figure 4d).

**Figure 4.** *Cont.*

**Figure 4.** Endocytosis of CedV F proteins in MDCK-2 cells. (**a**) Antibody uptake assay of mutant CedV F proteins. MDCK-2 cells were transfected with plasmids encoding the indicated CedV F protein mutants. At 24 h p.t., a CedV F-specific rabbit antiserum was used to label surface-expressed F proteins at 4 ◦C. Then, the cells were shifted to 37 ◦C for 30 min to allow endocytosis to occur. AF 488-conjugated secondary antibodies visualized surface-bound primary antibodies. After fixation and permeabilization, AF 568-conjugated secondary antibodies were used to stain internalized primary antibody-CedV F protein complexes. Magnification, ×60. n = 2. (**b**) At 24 h p.t., CedV F-expressing cells were surface-labeled with cleavable NHS-SS biotin at 4 ◦C followed by a shift to 37 ◦C for the indicated times allowing endocytosis to occur. Residual biotin was then cleaved from the cell surface using MESNA. To quantify the amount of surface-biotinylated proteins that underwent endocytosis in a certain time, samples were compared to the total amount of surface-biotinylated control cells (Ctr) that were neither incubated at 37 ◦C nor treated with MESNA. Following cell lysis, F proteins were immunoprecipitated and samples were separated under non-reducing conditions. After transfer to nitrocellulose, biotinylated proteins were detected with peroxidase-conjugated streptavidin and chemiluminescence. The control lane represents 50% of the total amount of biotinylated F proteins (Ctr 50%). One representative blot is shown for each CedV F protein variant. (n = 3; n = 2 for mutants e1, e3, and e4). wt: wild-type (**c**) CedV F protein internalization in percentage (%) per minute. To quantify the internalization rate, the percentages of internalized F protein amounts measured in the experiment shown in Figure 4b are displayed as a function of the time of incubation at 37 ◦C. Error bars represent the standard error of the mean. (**d**) Intracellular localization of wt and mutant CedV F proteins in MDCK-2

cells after 5 and 30 min at 37 ◦C. An antibody uptake assay of CedV F proteins was performed as described above with slight modifications. After the endocytosis step, surface-bound primary antibodies were blocked using a peroxidase-labeled secondary antibody while internalized primary antibodies were detected with AF 488-conjugated rabbit-specific secondary antibodies after fixation and permeabilization. Likewise, early endosomes were visualized with a primary antibody against the early endosomal antigen-1 (EEA1) and a mouse-specific AF568-conjugated secondary antibody. Scale bars indicate 20 μm. Representative images from two independent experiments are displayed (n = 2). Inserts show magnifications of indicated areas. Magnification, ×63.

Since di fferences in the internalization rate might directly a ffect cell surface expression, we next performed a surface biotinylation assay of F-expressing MDCK-2 cells followed by immunoprecipitation of biotinylated proteins and detection of the HA-tagged proteins in Western blot. In accordance with the antibody uptake assay and the MESNA reduction, we observed that all F proteins reached the cell surface (Figure 5). However, comparing the cell surface expression of all F proteins under non-reducing conditions, the amount of F0 di ffered distinctly between the parental CedV F protein and some mutant F proteins (Figure 5a). In all biotinylation assays performed mutant e1 displayed a cell surface expression similar to the parental CedV F protein, while surface expression of mutants e2–e5 was slightly increased. In contrast, the average level of cell surface expression of mutants, e6 and e7 was substantially higher. For mutant e7, we detected up to 2.5-fold more F0 on the cell surface than for the parental F protein. Under reducing conditions (Figure 5b), both the F0 precursor and the subunit F1 were detected for all F proteins analyzed, indicating that all of them were proteolytically cleaved despite mutations in their cytoplasmic tails. For the parental F protein as well as for mutant e1–e6, quantification of band intensities revealed that more cleavage product F1 than inactive precursor F0 is found at the cell surface. However, mutant e7 rather displayed a reversed cleavage ratio, with less F1 present on the cell surface indicating a reduced amount of cleaved F protein. Although internalization of mutants e2 and e3 was shown to be clearly delayed in the biotin internalization assay, proteolytic activation and surface expression of these mutants still seemed to reach levels comparable to the parental F protein after 24 h p.t.

Finally, to assess whether any of the observed effects influence the biological activity of the F proteins, syncytium formation was analyzed in a fusion assay. At 48 h p.t., co-expression of the parental CedV F and CedV G proteins resulted in the formation of multinucleated syncytia in MDCK2 cells (Figure 6a) and Vero76 cells (Figure 6b). Mutant CedV F e1 and e4 induced syncytium formation comparable to the parental F protein. Surprisingly, despite their reduced endocytosis rate but parental F-like cell surface expression, co-expression of mutant e2 or e3 with CedV G led to a hyperfusogenic phenotype forming syncytia that were markedly increased in number and size (Figure 6a,b). Interestingly, fusogenicity of mutant e5 and e6 displaying an increased surface expression compared to the parental F protein was slightly enhanced while fusion activity of endocytosis-deficient mutant e7 was strongly impaired in both cell lines tested (Figure 6a,b). In addition, we quantified these di fferences using a luciferase reporter gene-based fusion assay in Vero76 cells after 24 h p.t. The hyperfusogenic phenotype of mutant e2 and e3 displayed a 4- and 6.5-fold increase in measurable luciferase reporter activity compared to the parental F protein (Figure 6c). Co-expression of CedV G protein with mutant e5 still resulted in a 2.5-fold increase in reporter activity in Vero76 cells (Figure 6c). A marked decrease in reporter activity was confirmed for mutant e7. Since fusion activity is sensitive to the cell surface expression of both glycoproteins and to exclude that observed e ffects in fusogenicity were due to an altered cell-surface expression of CedV G protein, we performed a surface biotinylation assay in MDCK-2 cells co-transfected with CedV G and wt or mutant CedV F proteins. As depicted in Supplementary Figure S3, both F and G can be detected at the cell surface. Irrespective of the (mutant) F protein combination, the G expression at the cell surface seems to be similar suggesting that observed e ffects in fusion activity are not related to di fferences in CedV G cell surface expression. Taken together, these findings demonstrate that a membrane-proximal YXX Φ, as well as a C-terminal di-tyrosine motif in the CedV F protein cytoplasmic tail, are of functional relevance for endocytosis and biological activity of the protein.

**Figure 5.** Cell surface expression of CedV F proteins. At 24 h p.t., MDCK-2 cells expressing F proteins were surface-labeled with biotin on ice. After cell lysis, biotinylated proteins were immunoprecipitated using NeutrAvidin beads and subjected to SDS-PAGE under non-reducing (**a**) and reducing (**b**) conditions. Precipitated F proteins were visualized using an antibody against the HA-tag (H6908), HRP-labeled secondary antibodies, and chemiluminescence. In (**a**), ConA staining is used as a loading control. Representative blots are shown from (**a**) three or (**b**) four independent experiments. The amount of F0 and F1 protein (in (**a**) % of CedV F0 protein with the parental F protein set to 100% or (**b**) % of F1 protein) is calculated as the mean of three or four independent experiments, respectively. wt: wild-type.

**Figure 6.** Effect of cytoplasmic tail mutations on CedV glycoprotein-mediated fusion activity. Syncytium formation in (**a**) MDCK-2 and (**b**) Vero76 cells co-expressing CedV F and G proteins were visualized by Giemsa staining at 48 h p.t. Magnification, ×20. n = 3; wt: wild-type (**c**) Quantitative reporter gene assay. Vero76 cells were co-transfected with plasmids encoding for the CedV glycoproteins F and G as well as a plasmid containing the luciferase gene under the control of a T7 promoter (pCITE Renilla). At 24 h p.t., Vero76 cells expressing the T7 polymerase were layered on the glycoprotein-expressing cells and incubated for 3 h at 37 ◦C. Then, cells were lysed, and luciferase activity measured using a luminometer. Samples were tested in duplicates in three independent experiments. Reporter activity measured for the parental CedV F protein (wt F) co-transfected with CedV G protein was set to 1 serving as a reference point for fusion activity. Bars represent the fusion activities of the different (mutant) CedV F proteins in relation to the fusion activity of the parental protein and include the standard error of the mean (SEM). Background activity of the luciferase reporter was assessed with cells transfected with pCAGGS CedV G and pCITE Renilla only, layered with T7 polymerase expressing cells. n = 3.
