1. Introduction
Most models of influenza virus assembly postulate a direct interaction between the viral matrix protein M1 and the cytoplasmic domains of the envelope glycoproteins hemagglutinin (HA) and/or neuraminidase (NA) as a driving force for virus assembly [
1,
2,
3]. However, such an interaction has never been demonstrated biochemically, for example by co-precipitation approaches. Experiments to demonstrate that an interaction of HA and NA with M1 by indirect methods yielded contradictory results. M1 is not intrinsically targeted to the assembly site, the apical plasma membrane, but rather localizes to the cytosol, the nucleus and a significant fraction to internal membranes, such as the Golgi [
4]. In one report, co-expression of M1 with HA and NA did stimulate the membrane association of the M1 protein significantly [
5], but this effect was not seen in other studies [
6,
7]. Subsequently, detergent extraction at low temperature was used to demonstrate the specific interaction of M1 with HA and NA [
8]. Both HA and NA fractionate into detergent resistant membranes (DRMs), whereas M1 expressed alone was soluble, but became resistant when co-expressed with HA and NA. In order to define the minimal number of elements required for virus budding, it was found that HA, when expressed alone, was released as virus-like particles (VLPs) if sialidase activity is provided. Co-expression of NA and M2 increased VLP production. Expression of M1 alone does not produce VLPs, but M1 is incorporated into VLPs if HA is also expressed. Furthermore, when the cytoplasmic tails (CT) were deleted from NA and especially from HA, tailless glycoproteins were included in the VLPs, but M1 failed to be efficiently incorporated [
9]. The data are consistent with a model in which the glycoproteins control virus budding by sorting to lipid raft nanodomains and recruiting the internal viral core components.
The most pronounced feature of the cytoplasmic tail of HA is the attachment of fatty acids to (mostly) three conserved cysteine residues [
10,
11]. Although this modification is usually described as palmitoylation, we found by mass spectrometry of HA from virus particles that two different fatty acids are attached, palmitate (C 16:0) and stearate (C 18:0). Whereas palmitate is exclusively attached to two cytoplasmic cysteine residues, stearate is bound only to a cysteine at the end of the transmembrane domain (TMD) [
12]. The hydrophobic modification of HA is essential for virus replication, since (depending on the virus strain) either virus mutants with more than one acylation site deleted show drastically impaired growth or could not be created at all by reverse genetics [
13,
14,
15]. Acylation facilitates raft-association of HA [
16] and thus enrichment of the protein in small nanodomains of the plasma membrane [
17,
18], an observation which might explain why palmitoylation affects both assembly of virus particles and the membrane fusion activity of HA [
19].
Given the central role of the cytoplasmic tail of HA for virus budding, it is surprising that recombinant virus with tailless HA could be generated by reverse genetics [
20]. The resulting virus particles had the same morphology and protein composition as wild-type (wt) virus; only the budding efficiency and infectivity of the virions were slightly reduced. A virus where the cytoplasmic tails were deleted from both HA and NA exhibited more severe defects, especially aberrant morphology and altered genome packaging [
21]. The relatively mild phenotype for the virus with tailless HA might be due to the specific constellation of viral genes in these experiments,
i.e., HA from the A/Udorn/72 (H3N2) (Udorn) strain in the background of the A/WSN/33 (H1N1) (WSN) strain. It was subsequently shown that deletion of cytoplasmic acylation sites from HA of the Udorn virus is lethal for virus replication, but recombinant virus could be rescued and grew to only moderately depressed titers by expression of M1 from the WSN strain [
13]. Thus, a defect in virus assembly caused by a lack of acylation could be compensated by a different M1 protein. Although the molecular basis for this effect is mysterious, it is clear that the cytoplasmic tail is important for virus replication. However, it is not known whether (besides the acylation sites) other amino acids of the tail also affect virus assembly and replication as one would assume if HA mediates specific interactions with M1 to recruit it to the assembly site.
In order to characterize the molecular signal for site-specific attachment of stearate we recently analyzed recombinant WSN viruses containing HA with various mutations in its cytoplasmic tail [
22]. Exchange of conserved amino acids in the vicinity of an acylation site had a moderate effect on the stearate content. In contrast, shifting the TMD cysteine to a cytoplasmic location virtually eliminated attachment of stearate, indicating that the location of an acylation site relative to the transmembrane span is the decisive factor for attachment of stearate. More importantly, in this context, is the observation that none of the mutations reduced attachment of fatty acids to HA; each of the cysteine residues were stoichiometrically acylated. Thus, any possible effect of the point mutations on virus replication is not due to reduced acylation of adjacent cysteine residues but an inherent property of the respective amino acids. Therefore, we now analyzed the growth properties, protein and genome composition of these viruses. In addition, we provide a comprehensive sequence comparison of the linker region, transmembrane domain and cytoplasmic tail of all HAs present in the database and report reversions of amino acids we observed during propagation of recombinant viruses.
2. Materials and Methods
2.1. Amino Acid Conservation Analysis
Influenza A virus HA amino acid sequences were extracted from the Influenza Research Database in March 2014 [
23], which contained 33,254 entries of H1 to H17 subtypes with full-length HA sequences. After elimination of redundant sequences, the remaining 17,311 unique sequences were aligned, while 11,543 and 5768 sequences were used for alignment of HAs belonging to group-1 and group-2, respectively. Alignment was done using the MAFFT program, which is based on the FFT-NS2 algorithm [
24,
25]. To represent the sequence conservation graphically, the WebLogo 3 program was applied constructing amino acid sequence logos [
26,
27,
28].
2.2. Cells
Madin-Darby canine kidney (MDCK II, ATCC CRL-2936) and human embryonic kidney (HEK) 293T (ATCC CRL-11268) cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; PAN Biotech, Aidenbach, German) supplemented with 10% FBS (Perbio, Bonn, Germany) at 37 °C, 5% CO2 and 95% humidity, using standard techniques.
2.3. Generation of Recombinant Virus
Recombinant influenza A/WSN/33 (H1N1) virus was generated using an eight-plasmid reverse genetics system [
29], where each plasmid contains the cDNA of a single viral RNA segment, flanked by suitable promoters. In the HA encoding cDNA segment, the codon for I563 (ATA) was replaced by the isoleucine codon TTG, G557 (GGG) was replaced by the alanine codon GCT and the glutamic acid codon GAA, respectively. Q560 (CAG) was replaced by the glutamine codon GAA. G547 (GGG) was replaced by the serine codon ACG. C554 (TGT) and L559 (TTG) were replaced by the serine codon (AGC) and the cysteine codon (TGC), respectively using QuickChange mutagenesis (Stratagene, Waldbronn, Germany), and confirmed by sequencing (GATC Biotech, Konstanz, Germany). For generation of acylation mutants the cysteine codon TGT (Cys) was changed to AGC (Ser) resulting in Ac1, TGC (Cys) was replaced by TCT (Ser) and TGC (Cys) was changed to AGT (Ser) generating Ac2 and Ac3, respectively.
HEK 293T cells in 60 mm dishes were transfected with the eight plasmids encoding WSN cDNA (1 µg each) using TurboFect (Fermentas/Thermo Fisher Scientific, St. Leon-Rot,) in OptiMEM medium (Invitrogen, Karlsruhe, Germany). At 4–6 h after transfection, the medium was replaced by infection medium (DMEM containing 0.2 % bovine serum albumin (BSA), 0.1% foetal bovine serum (FBS), 2 mM glutamine, 100 U/mL penicillin/streptomycin and 1 µg/mL TPCK-treated trypsin, (Sigma-Aldrich, Taufkirchen, Germany) and incubation was continued at 37 °C. At 48 h after transfection, the supernatants were harvested and cleared of debris by low-speed centrifugation (2000× g, 5 min, 4 °C). The supernatant was used to infect MDCK cells for virus propagation. After 1 h of adsorption, the cells were washed with PBS, infection medium was supplied and incubation was continued until a cytopathic effect became evident. The supernatant was then harvested.
2.4. Sequencing of HA from Virus Particles
To check for correctness of the HA sequences in the recombinant viruses, RNA was isolated from cleared cell culture supernatants with Invisorb Spin Virus RNA Mini Kit (Stratec, Birkenfeld, Germany) followed by reverse-transcription and polymerase chain reaction (RT-PCR) using HA specific primers and the OneStep RT-PCR kit (Qiagen, Hilden, Germany). Before sequencing, PCR products were treated with the ExoSAP-IT Cleanup Kit (Affymetrix, Halbergmoos, Germany) to remove excess of primers and desoxinucleotide triphosphates (dNTPs), which might hinder the sequencing reaction. Two µL ExoSAP-IT reagent was added to 5 µL PCR product, and incubated at 37 °C for 15 min and heat-inactivated for 15 min at 80 °C. Sequencing was performed by GATC Biotech (Konstanz, Germany).
2.5. Plaque Assay
Plaque assays were performed on MDCK cells in 6-well plates according to standard procedures. Briefly, cells were infected with serial tenfold dilutions of the virus supernatants in infection medium, incubated for 1 h at 37 °C, washed with PBS and overlaid with 0.9% SeaKem LE agarose (Lonza, Verviers, Belgium) in Eagle’s minimum essential medium (Lonza, Verviers, Belgium) supplemented with 0.2% BSA, 0.1% FBS, 2 mM glutamine and TPCK-treated trypsin (1µg/mL). After 3 days of incubation, the cells were stained using 0.02% neutral red (Biochrom, Berlin, Germany) in PBS and the plaques were counted. Plaque sizes were calculated from digital images using Image J.
2.6. Hemagglutination Assay
HA tests were performed to quantify virus particles by hemagglutination. Freshly drawn chicken blood was washed three times in sterile PBS (centrifugation for 5 min at 2000× g), the working solution was adjusted to 0.5%. Two-fold virus dilution series with a final volume of 50 µL were prepared in 96-V-well microtiter plates. Then, 50 µL red blood cell solutions were added per well, mixed and incubated for 30 min at 4 °C. The titer was determined by the last viable lattice structure found.
2.7. Electron Microscopy
For negative staining, virus supernatants from infected MDCKII cells were harvested and cleared of debris by low-speed centrifugation (2000× g, 5 min, 4 °C). The plaque titer was determined and aliquots containing approximately 107 PfU/mL were mixed with fixative (stock solution of 20% paraformaldehyde (PFA; Roth, No. 0335.3, Karlsruhe, Germany) in 0.5 M HEPES buffer (pH 7.2), which was heated to 60–70 °C for 20 min to shift the equilibrium towards non-polymerized FA) to a final concentration of 2% formaldehyde and incubated for 1 h at room temperature before storing at 4 °C. A drop (10 µL) of the test suspension was placed directly onto a glow-discharged electron microscopy sample support (400 mesh copper grid, covered with a carbon reinforced plastic film). After adsorption for 10 min at room temperature, the grids were washed three times in double distilled water and negatively stained with 1% uranyl acetate. The grids were examined using a JEM-2100 transmission electron microscope (JEOL Corp., Freising, Germany) operated at 200 KV. Micrographs were recorded with a Veleta CCD camera (Olympus Soft Imaging Solutions) at a resolution of 2048 × 2048 pixels.
2.8. Growth Kinetics of Recombinant Virus
To assess growth kinetics, MDCK cells were infected with recombinant WSN at a low m.o.i. (0.01 or 0.001). After 1 h of adsorption and washing with PBS, infection medium (DMEM containing 0.2% bovine serum albumin, 0.1% FBS, 2 mM glutamine, 100 U/mL penicillin/streptomycin and 1 µg/mL TPCK-treated trypsin, Sigma-Aldrich) was added. Aliquots were removed from cell culture supernatant after a defined incubation time, cleared of debris and stored at −80 °C before titer determination by plaque assay.
2.9. Virus Preparation, SDS-PAGE, Western Blotting and Radioactive Labeling
Virus particles were produced by infecting MDCK cells in a 15 cm dish at an m.o.i. of >2. At 24 h after infection, the supernatants were harvested, cell debris was removed (2000× g, 10 min, 4 °C), and the virus was pelleted from the supernatant by ultracentrifugation (Beckman SW-28 rotor, Krefeld, Germany, 28,000 r.p.m., 2 h, 4 °C), resuspended in 100 mL PBS and analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) under non-reducing conditions followed by Western blotting using a polyclonal antiserum against fowl plague virus, which cross-reacts with WSN M1, and horseradish peroxidase (HRP)-coupled secondary antibodies for chemiluminescence detection using ECL Plus substrate (GE Healthcare, Solingen, Germany) and a Fusion SL camera system (Promega, Mannheim, Germany), which detects photons over a wide linear signal-response range. In order to prepare radioactive labelled virus particles the cell culture medium was replaced at 3 h post infectionem (p.i.) by DMEM lacking methionine, cysteine and glutamine, supplemented with 0.2% BSA, 0.1% FBS, 5 mM glutamine, 1 µg/mL TPCK-treated trypsin and 0.3 mCi/mL (11.1 kBq) [35S]-methionine/cysteine (EasyTag™ EXPRE35S35S Protein Labelling Mix; PerkinElmer, Rodgau, Germany). Virus preparations were subjected to SDS-PAGE under non-reducing conditions and fluorography using 1 M salicylate. The dried gel was exposed to X-ray film and bands intensities were analyzed with Bio-1 D software (PeqLab, Erlangen, Germay).
2.10. Quantitative Real-Time RT-PCR
In order to analyze an equivalent number of virions the supernatants from infected MDCK cells were adjusted to an HA titer of 26. RNA was then extracted with the Invisorb Spin Virus RNA Mini Kit (Stratec) and cDNA synthesized with the Maxima H Minus First Strand cDNA Synthesis Kit (Thermo Scientific, Braunschweig, Germany). 4 µL 5× RT-Buffer, 1µL dNTPs (10 mM), 1 µL Oligo(dT)18 Primer (100 µM), 1 µL Maxima H Minus Enzyme mix and 13 µL RNA were used per sample, which were incubated at 50 °C for 30 min. The reaction was terminated by incubation for 5 min at 85 °C.
The reaction product was directly used in a quantitative RT-PCR with the following primers specific for the influenza A virus M gene segment: Forward primer (5′→3′): AGATGAGTCTTCTAACCGAGGTCG, reverse primer (5′→3′): TGCAAAAACATCTTCAAGTCTCTG (Invitrogen). Real-time RT-PCR was performed utilizing a TaqMan probe, OneTaq DNA polymerase (NEB) and the ICYCLER-IQ5 Multicolor Real Time PCR Detection System (BIO-RAD, München, Germany). One assay with a total volume of 25 µL contained 5 µL 5× OneTaq-buffer, 16 µl RNAase free water, 0.5 µL dNTPs (100 µM), 0.25µL forward and reverse primer (100 µM), 0.75 µL probe (10µM, 6FAM-TCAGGCCCCCTCAAAGCCGA-TMR), 0.25 µL OneTaq DNA polymerase and 2 µL cDNA as template. Temperature profile: 3 min, 95 °C (1×); (10 s, 95 °C; 30 s, 55 °C (annealing and elongation)) (40×). Fluorescence values (FAM) were collected during the annealing step. A standard curve was generated by using serial dilutions of an in vitro transcribed, M segment-derived RNA transcript.
4. Discussion
In summary, we have shown here that the cytoplasmic tail of H1 subtype HA is essential for virus replication since WSN virus with tailless HA could not be generated. This is in contrast to a previous study, where viruses containing tailless HA from the Udorn strain (H3 subtype HA) could be rescued in the background of a WSN helper virus and even revealed only little growth defects [
20]. Except for the two cytoplasmic acylation sites, the C-terminal isoleucine residues and the glycine, the tail sequences of HA from Udorn (KGNIRCNICI, stop codon was introduced instead of the K) and from WSN (NGSLQCRICI, stop codon was introduced instead of the G) are different which might explain the different results. In addition, it might be possible that M1 (or any other protein) that was provided from WSN might allow rescue of otherwise non-infectious particles, as described for HA from Udorn having deletions of acylation sites [
13].
Individual amino acids in the tail of HA from WSN had very different effects on virus growth. Deletion of the two cytoplasmic palmitoylation sites and non-conservative substitution of the adjacent isoleucine by glutamine completely prevented rescue of infectious virus (
Table 1). This is in agreement with earlier studies using a less efficient reverse genetics system where rescue of infectious WSN virus was not possible if Cys 560 or Cys 563 were mutated to serine. However, virus particles could be generated if Cys 560 was exchanged by hydrophobic amino acids, such as alanine and especially with tyrosine and phenylalanine [
15]. Also in accordance with our findings are published results using reverse genetics with two other influenza virus strains. Recombinant Udorn (H3 subtype HA) and fowl plague virus (H7 subtype HA) where HA´s two cytoplasmic acylation sites were substituted by serines could be generated, but the resulting virus particles were greatly growth compromised [
13,
14]. Likewise, although the cytoplasmic tail of HA (including two palmitoylated cysteine residues) from Influenza B virus is not absolutely required for virus replication, the resulting virus revealed defects in replication [
37].
In contrast, recombinant viruses with mutations in other conserved amino acids showed no altered morphology of virus particles (
Figure 5), some only subtle defects in growth kinetics (
Figure 3) and M1 and RNP content compared to wild type virus (
Figure 4). The standard methods we used, such as HA-assays and metabolic labeling experiments or Western-blotting, are at best semi-quantitative and have an intrinsic error margin which probably exceeds possible small differences between wild type and mutant viruses. Thus, more precise methods, such as quantitative mass spectrometry are required to reliably demonstrate differences in the protein composition of wild type and mutant virus particles [
38]. The pleomorphic morphology of Influenza virus particles is a further obstacle that prevents rapid progress in the field. Assuming that every (infectious) particle has eight different RNP segments, the number of NP molecules (and also PA, PB1, PB2) is the same in each virion. However, since the membrane’s surface area varies between particles having a different morphology the number of membrane proteins (and thus the ratio of internal to external viral proteins) most likely varies between individual virions. In addition, at least 90% of particles present in a virus preparation are not infectious, but a shift in the ratio of infectious to non-infectious particles is only insufficiently described by calculating the PFU to HA titer ratio. Thus, new and more precise methods need to be developed to understand the complex interplay between viral proteins that leads to virus assembly and bud formation.
Figure 5.
Characterisation of WSN wild type and mutants by negative staining transmission electron microscopy. Representative negative stain images of the indicated virus particles harvested from the supernatant of infected MDCK cells. Scale bars = 200 nm. Differences in particle appearance are not due to morphological changes but to slight modifications of microscope settings and staining conditions of the virus suspensions. Occasionally, filamentous and aberrant formed particles were also observed in each virus preparation, the latter might be an artefact of ultracentrifugation [
39].
Figure 5.
Characterisation of WSN wild type and mutants by negative staining transmission electron microscopy. Representative negative stain images of the indicated virus particles harvested from the supernatant of infected MDCK cells. Scale bars = 200 nm. Differences in particle appearance are not due to morphological changes but to slight modifications of microscope settings and staining conditions of the virus suspensions. Occasionally, filamentous and aberrant formed particles were also observed in each virus preparation, the latter might be an artefact of ultracentrifugation [
39].
However, although all effects are rather small (compared to the essential nature of the cytoplasmic palmitoylation sites and adjacent hydrophobic residues), the repeated occurrence of amino acid reversion suggests that conserved residues in the tail and in the TMD confer a growth advantage (
Figure 2). The most striking example is the repeated and rapid occurrence of an amino acid reversion to serine if a conserved glycine in the TMD was exchanged to isoleucine, a typical constituent of transmembrane domains. Since the other possible single nucleotide exchanges in that codon would specify either a hydrophilic residue (asparagine) or hydrophobic amino acid (isoleucine, leucine, phenylalanine, valine, threonine, methionine), the latter are typical constituents of transmembrane regions, the repeated reversion to serine strongly suggests that a small amino acid is preferred at that position. The transmembrane region of HA is likely to be α-helical [
40] and when the inner part of the TMD (541VL
VVSL
GAI
SFWM
C554) is plotted as a helical wheel, G547 is located together with other small amino acids (V543, S550) on the same side of the helix as the stearoylated cysteine 554. One might speculate that these small residues are located on the outer surface of the trimeric TMD where they might form a groove which binds stearate to align it with the TMD helix as previously speculated [
41]. Alternatively, glycine residues are often located at the interface between oligomeric transmembrane regions where they mediate helix-helix interactions [
42]. In both cases the substitution of the small glycine residue (amino acid volume of 70 Å
3) by the large residue isoleucine (volume of 170 Å
3) might disrupt the assumed interaction and, therefore, a spontaneous reversion to serine (volume of 100 Å
3) confers a growth advantage.
However, our results are difficult to reconcile with the assumption that the cytoplasmic tail of HA alone recruits M1 to the viral assembly site, at least if this is assumed to occur by hydrophilic binding forces, such as salt bridges or hydrogen bonds. The hydrophobic patch ICI (563–565) at the C-terminal end of the tail plus the second palmitoylation site in close proximity (561) likely anchors the tail to the inner side of the plasma membrane (
Figure 6). This assumption is based on the observation that the exchange of the acylated cysteines by serine is lethal whereas their substitution by residues with the propensity to insert into lipid bilayers allows virus rescue [
15]. Likewise, substitution of isoleucine 563 by the hydrophilic residue glutamine prevents virus rescue, while its exchange by a long and hydrophobic leucine residue has only a minor effect. In addition, substitution of leucine 559 by a cysteine, which is stoichiometrically used as acylation site [
22], reduces the virus growth only slightly suggesting that this residue in the cytoplasmic tail also interacts with membranes. In proximity to these acylated or hydrophobic residues is the conserved basic amino acid arginine 562, which has the capacity to interact with the head groups of negatively charged lipids, abundant components at the inner leaflet of the plasma membrane [
43]. Thus, if many amino acid side chains (or attached fatty acids) of the cytoplasmic tail of HA are engaged in interactions with the lipid bilayer, only four, N, G, S and Q, remain to specifically bind to M1 (
Figure 6). However, exchange of two of them, G 557 and Q 560 had very little influence on virus growth. This suggests that they do not bind with high-affinity to M1 since one would assume that replacement of an amino acid making an essential contact with M1 would result in a more drastically impairment of virus growth. However, in our mutagenesis study, we concentrated on amino acids conserved between all HA subtypes, and it might be that even non conserved residues might have a substantial impact on virus assembly and budding.
Figure 6.
Model of the cytoplasmic tail of HA. Amino acids, indicated in single letter code, which were exchanged in this study, are underlined. Essential amino acids, i.e., the non-conservative substitution of which prevented rescue of infectious virus particles, are in red. Fatty acids attached to CT and TMD cysteine residues are depicted by a black (palmitate) or purple (stearate) zigzag line. Hydrophobic or positively charged amino acids that probably interact with the acyl chains or head groups of (negatively charged) lipids are in purple.
Figure 6.
Model of the cytoplasmic tail of HA. Amino acids, indicated in single letter code, which were exchanged in this study, are underlined. Essential amino acids, i.e., the non-conservative substitution of which prevented rescue of infectious virus particles, are in red. Fatty acids attached to CT and TMD cysteine residues are depicted by a black (palmitate) or purple (stearate) zigzag line. Hydrophobic or positively charged amino acids that probably interact with the acyl chains or head groups of (negatively charged) lipids are in purple.
Nevertheless, there is clear evidence that HA of both Influenza A and B virus at least functionally interacts with M1 during budding [
13,
37]. M1, a hydrophobic protein that has the capacity to partially insert into lipid bilayers [
44,
45,
46], might interact with residues located within the inner (and variable) part of the TMD of HA. This assumption might explain why expression of M1 from WSN virus rescues the otherwise lethal deletion of acylation sites from HA of Udorn virus [
13]. The two essential cytoplasmic palmitates might also stick to the hydrophobic surface of M1. Alternatively, HA and M1 might not form protein interactions but promote virus budding indirectly, e.g., by changing the fluidity of the plasma membrane to create a stable assembly site. In addition, the cytoplasmic tail of NA, which is completely conserved through all subtypes, might recruit M1 to the viral assembly site. This is consistent with a recent study using Cryo-EM concluding that expression of M1 and M2 together with either of the viral glycoproteins is the minimal requirement to assemble and release virus-like particles [
47] and also with the observation that the cytoplasmic tails of both HA and NA control the shape of virus particles [
21].