Protein C was detected in UC purified MEV. Until now it was only reported to be synthesized in Vero cells, but this was the first time it was found in MEV virions . The intensities of the bands containing protein C indicate its low abundance in virus samples, which might explain why C was not detected in HIC purified MEV samples. V and C were until now considered non-structural MEV proteins, and were found not to be necessary for replication of MEV in Vero cells [2, 13, 14]. Therefore, it is very interesting that C is in fact packed into MEV virions. Protein V was not detected in MEV samples, possibly due to even lower abundance than C, or absence from the virions.
Detection of protein H in up to 7 bands in MEV samples implies the presence of various glycoforms as well as the presence of both monomer and dimer on the gel, as previously described .
Occurrence of protein N in multiple bands at MM equal and lower than MMaa is in agreement with previous results which hypothesized that minor bands ranging from 40 to 55 kDa belong to breakdown products of N protein or its truncated forms [23, 32]. It is also interesting to notice that the intensity of the 55–56 kDa band in the samples is typically higher than of that at 59 kDa. This might be the result of changes in transcription or translation, or 59 kDa protein degradation by proteases during purification procedures , resulting in a more abundant 55–56 kDa form present on the gel. It is also interesting to notice that the analysis of peptides present in the PMF spectra of MEV purified by HIC did not indicate truncation of protein with apparent MM of 55 kDa (Additional file 1: Figure S4), however such result needs to be further corroborated.
Mature protein F consists of disulphide linked F1 (MMaa 47 kDa) and F2 (MMaa 13 kDa) fragments generated by F0 cleavage . The uniqueness of MEV F in comparison to F protein of other paramyxoviruses is that all glycosylation sites appear to be on the F2 fragment [24,25,26]. It was proposed that F2 is usually not detected by Coomassie staining because of its diffuse nature due to its carbohydrate content . F was found in multiple bands in UC purified mEV, at MMs equal and higher than MMaa. Bands containing F which were found around 40 kDa probably contain the non-glycosylated F1 fragment. Since MMaa of non-glycosylated precursor F0 would be around 59 kDa, bands found around to 59 kDa may contain F0, as previously reported [25, 29], however since no peptides corresponding to F2 fragment were detected, this cannot be confirmed. Multiple bands containing F1 were found in the 50–55 kDa range, however their origin is unclear for now. It is possible that some of them represent palmitoylated F1  or degradation products of F0. Cross-contamination between bands as a cause of this phenomenon was excluded due to meticulous work in this and all other samples in which this occurs.
Protein M also appeared at several bands at MM equal and higher than its MMaa (40–60 kDa). The doublet of bands around 37–39 kDa was previously reported for MEV M, and some smearing of M was reported even under reducing conditions . Although biologically active form of M seems to be a dimer [51, 52], this does not explain the occurrence of these bands. The origin of M in multiple bands remains to be elucidated.
Protein P was detected at 2 to 3 bands in different MEV samples. The bands detected at approximately 53 kDa correspond well to MMaa of P (54 kDa), thus likely corresponding to the protein without any posttranslational modifications. In the virions P is heavily phosphorylated , thus carrying large negative charge, therefore its migration in the gel should be retarded. This implies that the bands detected at approximately 60 kDa probably present posttranslationally modified P. P which was detected in bands at higher MMs (110 kDa and more) likely represents oligomers of P, since P is known to be a self-associated oligomer . In previous reports P was also readily found in bands ranging from 65 to 70 kDa [23, 25, 29].
Viral protein L (MMaa 248 kDa) was found at its corresponding MM in UC purified MEV, but in HIC purified MEV it remained undetected. This was probably due to its low abundance combined with its co-migration with much more abundant fibronectin, which might result in peptide desorption/ionization suppression.
Results obtained for MUV in this study are similar to those previously published . Although protein V is still often considered to be a non-structural protein and is not necessary for MUV replication [12, 16], it was shown to be present in all the samples analysed in this study, which is in line with previously published studies [32, 55].
Protein HN was detected in 2 to 4 bands at MMs ranging from 70 to 200 kDa, which is higher than its MMaa (64 kDa). This indicates its presence as glycosylated monomer and dimer, as previously reported [32, 56]. The novelty is that in the sample on Fig. 2(c) HN was present in 4 bands, indicating that different glycoforms are likely present.
Protein NP was detected in up to 4 bands, at MM equal or lower than its MMaa, and the comparison of detected peptides in the PMF spectra of UC and HIC purified samples (Additional file 1: Figures S3 and S5, respectively) again indicates C-truncation of proteins present in the bands at MMs lower than 61 kDa, as described in our previous study . Interestingly, in the present study, in HIC purified MUV full-length NP was not detected. When this is considered in parallel with the finding that lower MM forms of N are also more abundant in all MEV samples, it indicates some processes occur, either during virus production in the cells, or during virus purification, which result in more abundant truncated forms.
In this study, for the first time in UC and IAC purified MUV, fragment F2 of protein F was found at MM higher than it MMaa, indicating it is present in its glycosylated form. In HIC purified MUV, the band found around 57 kDa contained only peptides of F1 fragment, indicating it contains glycosylated F1, since its MMaa is 47 kDa. The band around 65 kDa contains peptides from both F1 and F2 fragments, indicating the presence of glycosylated F0 precursor, since its MMaa is 59 kDa [21, 22, 30].
In this study, in UC purified MUV samples protein P was not detected, which is unexpected since it was previously detected in MUV purified by UC . Reports exist showing it is susceptible to protease degradation, which might explain its absence from the gel [23, 54].
In this study, the comparison of two HIC elution fractions, E1 and E2, reveals different protein patterns for both MEV and MUV, which are consistent with other findings such as total and infective particle number as seen in Table 1. It becomes clear that more viral proteins are present in E2 fraction which also contains more infective particles, as previously reported . All these findings indicate that particles in fractions E1 and E2 differ significantly, possibly presenting different virus subpopulations .
Host cell proteins in virus and ECV preparations purified by different purification methods
The presence of ECVs in virus preparations, which was often neglected, complicates determination of HCPs present only in virions and not in ECVs. Here, for the first time, this problem was addressed by comparative analysis of the results obtained for viruses and ECVs purified by different purification methods. Chromatographic techniques such as HIC and IAC result in virus preparations of higher purity in comparison to UC. This is easily observed through total-to-infective particle ratio in such samples, as well as in HCP content when compared to starting material [5, 47, 48]. Comparison of results obtained by different purification methods helps in estimating purification efficiency of these methods, as well as estimation if a method results in enrichment of certain particles (e.g. infective or non-infective virus particles, ECVs) or HCPs in comparison to other available methods.
ECVs were long ago shown to be a major contaminant of virus preparations, as well as a source of HCPs present in such virus preparations [38, 39]. ECVs are similar to MUV and MEV in size – ECVs produced by non-infected Vero cells used in our experiments have a mean diameter of 199 ± 3.8 nm (n = 39, updated data from ), whereas MUV and MEV have mean diameters of 215 ± 1.9 nm and 206 ± 2.5 nm, respectively (n = 67 and n = 68, respectively, updated data from ). Similarity of ECVs to enveloped viruses in size, density and composition makes the preparation of ECV-free virus samples virtually impossible by methods currently available, and in case that production of ECVs is not greatly affected during infection (increased or diminished), up to one third of particles in virus suspensions may be ECVs .
Here the evaluation of which HCPs might be part of the virions was carried out for the first time by comparing HCPs detected in virus preparations purified by different purification methods, and with HCPs present in ECVs produced by non-infected Vero cells. The hypothesis beneath this comparative analysis is that, if an HCP is virion-associated, it will be present in all virus samples, regardless of the purification method used. Otherwise, if the HCP is present in virus preparation obtained only by some purification methods, it is more likely to be a contamination arising from the ECVs present in the virus preparation. To confirm the incorporation of such HCPs into ECVs, a comparison to the proteome of purified ECVs from non-infected Vero cells was carried out under the hypothesis that the composition of ECVs produced by infected and non-infected Vero cells is the same. Although the protein composition of ECVs might change during infection, since ECVs produced by infected cells cannot be distinguished and separated from viral particles in the supernatant of the infected cell culture, the results presented here still give a valuable insight into which HCPs are more likely to be present in viral preparations due to their association with viral particles, and which due to inevitable presence of ECVs in virus preparations.
All of the proteins detected in ECV samples are considered to be exosome markers , except BSA which is likely a contaminant originating from FCS used in cell culture media during production of ECVs .
Fibronectin was found in MUV and MEV, as well as ECVs, and in fraction E1 of HIC purified samples it is present in very high concentrations. Since it is unlikely that any particles (virions or ECVs) would contain such high concentrations of fibronectin as seen in E1, fibronectin is probably co-purified from culture supernatant by HIC under used conditions, with most protein eluting in E1 fraction. Since samples were concentrated by UC prior to SDS-PAGE, free proteins present in the eluates should be removed as the forces during UC are not strong enough to pellet free proteins. However, high salt concentration used in HIC (in this case 1 M (NH4)2SO4) can cause fibronectin aggregation and even precipitation [60, 61]. Therefore, it is presumed that fibronectin has possibly formed large aggregates during HIC purification, which pelleted at 141,000×g used for UC. Fibronectin has previously been reported in MUV samples purified by UC , and it was also detected here in UC purified samples. Its presence in most samples might imply its involvement in particle formation; however, its absence from IAC purified MUV indicates that it might just be a contamination.
Actin, annexins (A1, A2, A4, A5) and cyclophilin A (CypA) are readily found in all samples which, combined with previous reports, strongly supports hypothesis that these proteins are in fact part of the virions. Presence of actin in MEV and MUV was previously reported [20, 21, 26, 31, 62]. It was shown that viruses use cytoskeletal proteins such as actin for transport of viral components inside the cell, as well as in virus budding and maturation [34, 62]. Actin was found to interact with ribonucleocapsid in MEV, and it also seems to interact with ribonucleocapsid, M and glycoproteins of some other paramyxoviruses [62,63,64,65]. It is likely responsible for maintaining the architecture of virions  and ECVs, hence its presence inside the particles is expected. In virions, it might have an additional function, e.g. it was found to be involved in genome transcription in several paramyxoviruses [64, 66].
Annexins are present in the cytoplasm but can also be bound to the plasma membrane surface. Annexin A2 binds cellular actin and is involved in its organization in the proximity of plasma membrane . It is hypothesized that annexins as a part of viral particles aid the attachment of viruses on the host cells and fusion of virus and plasma membrane so it would be logical that they play the same role in the fusion of ECVs and cells. Although contradictory findings about the role of annexin A2 in virus infection have been published [67,68,69,70], it is possible that it might be important for infective virus formation in some cell lines .
Cyclophilin A (CypA) is highly abundant cytosolic protein acting as peptidyl-prolyl isomerase and is therefore often classified as a chaperone. Its presence in the particles might simply arise from its high abundance in the cytoplasm, but it could be incorporated into virions through interaction with viral proteins due to its chaperone function. It has been hypothesized that in some viruses it helps the formation of viral particles or uncoating after the infection, and it was also shown to be necessary for infective HIV-1 production [33, 35].
Integrin β1 and moesin were consistently found in both virus and ECV samples, regardless of the purification method. Their presence in IAC purified MUV supports the hypothesis that they are in fact included in virus particles. Integrin β1 was previously reported in vesicular stomatitis virus , whereas moesin was found in HIV . Since integrins act as membrane receptors, and are involved in connecting extracellular matrix to cytoskeleton, and moesin is involved in the interaction of actin cytoskeleton with the plasma membrane, they are likely present at the virus budding sites and hence included into virions. Whether they have a specific role in the virus lifecycle itself remains unclear, although they seem to be important for the virus uptake to the cells [73, 74].
Virtually all of the detected HCPs have been previously reported as proteins present inside of purified virus particles [33,34,35,36,37, 42, 59, 62, 63, 65, 66, 71, 75,76,77]. However, one should be aware that co-purified ECVs contribute to detected HCPs [38, 39, 41]. Also, protein composition of ECVs might change during virus infection hence yielding ECVs of different composition, and this further underlines that co-purification of ECVs with viruses should not be neglected.