Caveolin-1 influences human influenza A virus (H1N1) multiplication in cell culture
© Sun et al. 2010
Received: 2 March 2010
Accepted: 26 May 2010
Published: 26 May 2010
Skip to main content
© Sun et al. 2010
Received: 2 March 2010
Accepted: 26 May 2010
Published: 26 May 2010
The threat of recurring influenza pandemics caused by new viral strains and the occurrence of escape mutants necessitate the search for potent therapeutic targets. The dependence of viruses on cellular factors provides a weak-spot in the viral multiplication strategy and a means to interfere with viral multiplication.
Using a motif-based search strategy for antiviral targets we identified caveolin-1 (Cav-1) as a putative cellular interaction partner of human influenza A viruses, including the pandemic influenza A virus (H1N1) strains of swine origin circulating from spring 2009 on. The influence of Cav-1 on human influenza A/PR/8/34 (H1N1) virus replication was determined in inhibition and competition experiments. RNAi-mediated Cav-1 knock-down as well as transfection of a dominant-negative Cav-1 mutant results in a decrease in virus titre in infected Madin-Darby canine kidney cells (MDCK), a cell line commonly used in basic influenza research as well as in virus vaccine production. To understand the molecular basis of the phenomenon we focussed on the putative caveolin-1 binding domain (CBD) located in the lumenal, juxtamembranal portion of the M2 matrix protein which has been identified in the motif-based search. Pull-down assays and co-immunoprecipitation experiments showed that caveolin-1 binds to M2. The data suggest, that Cav-1 modulates influenza virus A replication presumably based on M2/Cav-1 interaction.
As Cav-1 is involved in the human influenza A virus life cycle, the multifunctional protein and its interaction with M2 protein of human influenza A viruses represent a promising starting point for the search for antiviral agents.
In the last few years the interaction of viral matrix proteins or precursors with cellular proteins has attracted much attention in the field of medical virology due to the increase in the understanding of their interplay in late viral processes like protein transport, virus assembly and budding. Viral matrix proteins establish the link between outer shell and capsid core of enveloped viruses and bring together these parts in the virus assembly step. Moreover, matrix proteins frequently determine the place where the assembly step occurs. In influenza A viruses two M proteins are located on RNA7 of the negative-stranded, segmented RNA virus. The M1 protein functions as a typical matrix protein, while M2 exerts multiple tasks in the early and late phase of virus infection. M2 tetramers form an ion channel and in the early phase of virus infection M2 serves for the release of viral nucleocapsid by acidification of endosomes. In the late phases, M2 prevents premature activation of newly synthesized HA  and -in concert with M1- contributes to virus budding and morphology. The involvement in virus exit has been assigned to the cytoplasmic tail of the protein [2–4]. Influenza viruses bud from lipid rafts and for this event the components of the viral envelope (haemagglutin HA, neuraminidase NA, M2) and the RNA containing protein complex (vRNP) must come together to form infectious virus [5–7]. Interestingly, the endosomal sorting machinery (ESCRT), which has been involved in late steps of other viruses, does not contribute to influenza virus budding [6, 8]. Accordingly, other routes and gates have been suggested for the transport of influenza proteins and virus assembly/budding .
In several previous investigations caveolin-1 (Cav-1), a multifunctional, raft-resident membrane protein has been linked to the virus replication of retroviruses HIV-1 and amphotropic mouse leukemia virus, rotavirus and respiratory syncytial virus [9–13]. Interestingly, a contribution of Cav-1 to HA transport has been reported for influenza virus infected MDCK cells . In a recent investigation of the enveloped γ-retroviruses budding from lipid rafts we showed that caveolin-1 (Cav-1) interacts specifically with the MLV retroviral matrix protein in the Gag precursor, suggesting that Cav-1 serves in positioning the Gag precursor at lipid rafts . Not surprisingly, Cav-1 is incorporated into MLV virions released from mouse NIH3T3 [13, 15]. Subsequently, competition and inhibition experiments provided evidence that Cav-1 modulates MLV retrovirus production . Taken together, these findings pointed to a general contribution of Cav-1 in virus replication strategy and opened the possibility that other virus families budding from lipid rafts may co-opt the functions of Cav-1. In our search for cellular/viral targets a database screen for Cav-1 binding sites notably revealed that structural proteins like matrix proteins of other viral families, e.g. Orthomyxoviridae with influenza A virus as a representative, exhibit regions of homology with a consensus motif for Cav-1 binding (Cav-1 binding domain, CBD) (Wirth, M, unpublished).
To address the biological relevance of the interplay of Cav-1 with influenza proteins we performed inhibition experiments with a dominant-negative Cav-1 mutant, knock-down by Cav-1 RNAi as well as competition experiments with M2 fusion proteins. We found, that the yield of human influenza virus progeny is affected by the presence/absence of Cav-1. The data suggest that Cav-1 can support the human influenza virus A life cycle. Pull-down and co-immunoprecipitation experiments were performed which showed binding of M2 and Cav-1.
A dominant-negative Cav-1 mutant has been described which functionally inactivates caveolin-1 upon binding . The mutant carries a F92A/V94A double mutation in the scaffolding domain (SD) of canine caveolin-1. Expression in rat adipose and COS-1 cells has been shown to interfere with the interaction of Cav-1 with the insulin receptor and impairs receptor function.
Cav-1 interactions with viral proteins
Type of interaction with protein partner
Type of interaction with Cav-1
Binding to CBD in HIV-1, but not HIV-2 or SIV
Binding to CSD*
(Benferhat et al., 2008; Hovanessian et al., 2004)
Binding to six-helix bundle
Binding to CSD
(Huang et al., 2007)
Cav-1 membrane insertion domain
(Llano et al., 2002)
Matrix, associates with membranes, link between capsid, plasma membrane, and viral membrane proteins
Binding mediated by a CBD in MA, interaction locates MA to lipid rafts domains in PM
Interaction with CSD*†
(Beer and Wirth, 2004; Yu et al., 2006)
Not known, Functioning in Golgi localization?
Binding to several CBDs
Not known, interaction with CSD likely
(Padhan et al., 2007)
Influenza A virus human
Early phase: Ion channel, viroporin
Late Phase: matrix, virus assembly and budding
Binding. Protein regions presumably CBD aa47-55
Binding to CSD*†
Binding to CSD‡
Zou et al. 2009
Influenza A virus human
Colocalization in perinuclear regions
(Scheiffele et al., 1998)
Colocalization with internal viral filaments, colocalization at lipid rafts
Binding not specified, redistribution of Cav-1 after phosphorylation
(Brown et al., 2002; Brown, Rixon, and Sugrue, 2002; McDonald et al., 2004)
Ion channel formation, ER and caveolae localization, important for morphogenesis
Binding aa114-135 (enterotoxic peptide) amphipatic helix at the C-terminus
Binding and colocalization, 2 independent binding sites at the N-terminus (aa2-22)and C-terminus (aa161-178) identified, influence on localization or transport?
(Mir et al., 2007; Parr et al., 2006; Storey et al., 2007)
To confirm data of the pull-down experiments, co-immunoprecipitation experiments were performed using NIH3T3 or MDCK cells after transfection of pEP24c, an expression vector containing M2 PR/8  or a vector harbouring fusion protein of M2 with fluorescent marker EGFP. 24 h later cell lysates were prepared in the presence of octylglucoside, a detergent that disrupt lipid rafts, as described previously . In the first series of experiments, polyclonal anti-Cav-1 antibodies were used to pull-down Cav-1 complexes from lysates. Precipitated complexes were probed for the presence of M2 after Western Blot and immunostaining. In these experiments, the Cav-1 antibodies clearly pulled down a complex that contained a M2 from pEP24c transfected MDCK cells (Fig. 5B, a) or the M2 fusion protein from pM2PR8-EGFP transfected MDCK or NIH3T3 cells (Fig. 5B, b) as well as infected MDCK cells (Fig. 5B, c left panel). In the second experimental setting, vice versa, monoclonal anti-EGFP antibodies were used to precipitate M2 binding partners and a rabbit anti-Cav-1 antibody was used to probe for the presence of caveolin (Fig 5B, c right panel). These types of experimental settings identified M2 complexed with Cav-1 and vice versa in both cell lines, NIH3T3 and MDCK. Thus, the results suggest that M2 has the capability to interact directly or indirectly with caveolin-1 in different cell lines. With respect to the type of interaction, it is notable, that caveolin-1 as well as M2 have been reported to bind cholesterol via cholesterol specific recognition domains [30, 31]. This prompted us to investigate, whether cholesterol is involved in the M2/caveolin-1 interaction. For that purpose methyl-β-cyclodextrin (MβCD) was used to deplete cell lysates from cholesterol before co-immunoprecipatation (Fig. 5B b and 5c). Interestingly, in pM2PR8-EGFP transfected NIH3T3 cells as well as in PR8 virus-infected MDCK cells, signals from co-immunoprecipated proteins decreased to a certain extent, if cholesterol was removed from the lysate before pull-down. These findings imply, that cholesterol seems to support the interaction of M2 with caveolin-1.
Viruses recruit the cellular machinery to support their own multiplication and elicit an early host response to overcome the unwanted viral invaders. In our contribution we investigated the ability of caveolin-1, a multifunctional protein, to interact with components in the influenza A virus life cycle and to interfere with influenza A virus production. Cav-1 represents an organizing element at the plasma membrane and serves on localization and accumulation of proteins in lipid rafts and transmission of signalling events . Furthermore, the protein contributes to intracellular cholesterol transport and has been identified as the main determinant of caveolae, invaginations of the plasma membrane used for entry of molecules and particles into the cell.
Based on previous findings of Cav-1 involvement in the late retroviral life cycle  we investigated the influence of Cav-1 on human influenza A/PR/34 (H1N1) virus multiplication in inhibition experiments. It is crucial for our investigation, that influenza virus entry does not occur via caveolae, but can be mediated by chlatrin-dependent endocytosis or another, not-defined pathway independent of chlathrin-coated pits [32–34]. For example, it has been shown, that a Cav-1 dominant-negative mutant does not affect the entry of influenza virus . The findings are a prerequisite to exclude artifacts that may arise from insufficient entry due to Cav-1 depletion in inhibition experiments. Applying different methods to impair or inhibit Cav-1 function in MDCK, a knock-down procedure, a dominant-negative Cav-1 mutant as well as competition experiments with M2 fusion proteins, we could show that Cav-1 influences human influenza A virus propagation. Inhibition methods have their limitation, e.g., we noticed that Cav-1 RNAi-mediated knock-down resulted in diminution of Cav-1 expression levels in MDCK cells to 25% of Cav-1 wild-type level at the most. Concomitantly, virus yield from these cells decreased 2-3 fold of virus levels observed from wild-type or RNAi-vector treated MCDK cells. Unfortunately, the effect of complete absence of Cav-1 on human influenza A virus production in MDCK cells could not be investigated, as further reduction of Cav-1 levels cannot be achieved with the retroviral RNAi system used . This question may be answered in a Cav-1 (-/-) MDCK cell line, which yet has to be established.
Data from knock-down experiments in MDCK were supplemented by transfection of a dominant-negative Cav-1 mutant as well as Cav-1 over-expression, which decreased viral yields by 38-44%. The results are reminiscent of experiments of Nystrom et al. who observed impairment of the insulin signalling pathway upon expression of both, the dominant-negative Cav-1 mutant and the over-expressed Cav-1 wt cDNA as well . Finally, competition with M2 fusion proteins impaired virus replication, too.
Taken together Cav-1 supports virus multiplication in MDCK, but the cellular pathway directing this Cav-1 property is not known. It is conceivable, that the cellular protein level of Cav-1 is important for the outcome, as it has been suggested for Cav-1 involvement in the insulin pathway .
Hints for the molecular basis of influenza virus/Cav-1 interaction may come from other viruses which co-opt Cav-1. It is evident that individual stages in the various viral life times are affected and different roles are allocated to Cav-1 as well (Table 1). For example, the CBD region in the HIV-1 gp41 transmembrane protein can permeate membranes and is supposed to augment the fusion step upon virus entry. Remarkably, respiratory syncytial virus (RSV), induces Cav-1 phosphorylation, which results in intracellular relocation of proteins during the paramyxovirus life-cycle. In several cases, Cav-1 functions in positioning of viral proteins to intracellular membranes (Rotavirus, SARS) or specialised regions of the plasma membrane like lipid rafts (retrovirus MLV).
To understand the molecular basis of the Cav-1 contribution to influenza A virus propagation we focussed on Cav-1 interactions mediated by the caveolin-scaffolding domain (CSD, aa 81-102) . Database searches and subsequent peptide pull-down assays in combination with co-immunoprecipitation experiments suggested binding of caveolin-1 to M2 presumably to a motif in the M2 protein fitting the CBD consensus . Strikingly, the motif is shared in M2 of nearly all human influenza A viruses. M2 functions within the viral life cycle as a viroporin with proton channel activity that is crucial in the entry phase  and as a maturation cofactor in virus budding. The cytoplasmic tail is implicated in M1 binding and facilitates virus assembly and production [2–4, 35]. Furthermore, Schroeder et al. showed that avian M2 is a cholesterol-binding protein . Most avian influenza A viruses contain two cholesterol recognition motifs (CRAC I, CRAC II) in close vicinity to the transmembrane domain in the cytoplasmic region of M2 [31, 36]. Thus, cholesterol-binding and palmitoylation in combination with a short transmembrane region may direct M2 to the raft periphery in membranes and may promote clustering and merging of rafts which is then followed by the pinching-off of avian viruses . With this model for avian influenza virus in mind it is conceivable that the interaction of Cav-1 with M2 could direct the protein into the vicinity of lipid rafts in human influenza A virus infection. This view may be supported by different observations: Firstly, we observed that the caveolin-1 binding domain is present in M2 of most human influenza A virus strains and overlaps with a CRAC motif for cholesterol binding. Such a high degree of evolutionary conservation generally suggests a constant selective pressure to preserve a specific function in the viral life cycle. Secondly, Cav-1 itself binds cholesterol via a region in the caveolin scaffolding domain . Notably, to some degree Cav-1 binding to M2 is sensitive to the cholesterol depletion (this investigation). Preliminary results of mutagenesis as well as localization experiments indicate a certain role of the M2-CBD in M2 transport and localization (unpublished observations). Taken together our results demonstrate, that Cav-1 exerts an influence on influenza A virus replication and data imply that the binding of Cav-1 to the matrix protein M2 is involved. However, which function or pathway in MDCK cells actually is triggered via Cav-1 interaction with M2, remains to be determined.
The appearance of the aggressive bird influenza (H5N1), the 2009 outbreak of a pandemic influenza (H1N1) of swine influenza origin, and the recent occurrence and rapid dissemination of oseltamivir-resistant human influenza strains are motors that have accelerated the search for new antiviral targets and agents within the last time [37–39]. The investigation of cellular mechanisms involved in 'early' and 'late' viral processes and the identification of cellular actors provides a means to interfere with viral strategies. With this respect, the observed Cav-1/M2 interplay may represent a new, conserved target for e.g. therapeutic intervention with circulating and newly emerging strains of human influenza A virus. Thus, application of high-throughput screening of compound libraries will follow target identification and may result in a new antiviral agent, as exemplified for a cellular target involved in the late retroviral life cycle .
When this manuscript was in preparation Zhou et al. reported binding of a cytoplasmic fragment of M2 from human influenza to Cav-1 in an in vitro assay based on a Cav-1 protein fragment expressed in E. coli and CBD-dependent perinuclear co-localization upon expression in CHO cells . However, no experiments on the functional importance of M2/Cav-1 were performed in this investigation.
MDCK Madin-Darby canine kidney (ATCC CCL-34) and NIH 3T3 (ATCC CRL-1685) were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum and 2 mM L-Glutamine at 37°C in 5% CO2. Influenza A/Puerto RicoR/8/34 (H1N1, Mount Sinai strain) virus was generously provided by Stephan Ludwig (Virology, ZMBE, Muenster, Germany).
BCA protein assay kit (Pierce) Methyl-β-cyclodextrin (MβCD, Sigma), octyl glucoside (Applichem) and other chemicals were of the highest grade commercially available.
pCav-1 wt (myc-tagged canine Cav-1 cDNA in pCIS2) and pCav-1 SD (point mutations F92A V94A in scaffolding domain) are described elsewhere . pM2PR8-EGFP and pM2PR8-dsRED were constructed by PCR-cloning of M2 (A/PR/8/34/(H1/N1) into BamHI/AgeI linearized pEGFP-N1 and pDsRed-Express-N1 (Clontech), respectively. M2 identity was verified by DNA sequencing. pEP24c (M2 cDNA) . pRVH1-Puro-Cav-1 and pRVH1-Puro  are described elsewhere.
Rabbit anti-caveolin 1 polyclonal antibody (pAb), mouse anti-caveolin 1 monoclonal antibody (mAb) mouse anti-EGFP mAb (JL-8) (all BD Transduction Laboratories) mouse anti-Influenza A virus M2 monoclonal antibody (14C2, ABR) were used according to the suggestions of the supplier.
Infections with Influenza A/PR8/34 were performed in the presence of trypsin (1-2 μg/ml) at a multiplicity of infection (m.o.i) of 0.2-10 for 2 h at 37°C. Virus stocks were prepared from supernatants of MDCK cell cultures one day post infection (1 d.p.i.).
Plasmids were transfected into cells via Lipofectamine 2000 (Invitrogen or by calcium phosphate transfection .
Lysates were prepared as described previously .
Influenza A/PR/8/34 titre was determined by plaque assay on MDCK cells. PBS- washed MDCK were inoculated with 500 μl of virus dilution for 1-2h at 37°C. Cells were covered with 2 ml of MEM medium containing 1% purified agar (Oxoid, England) and 1-2 μg/ml trypsin (Sigma). After three days incubation at 37°C, plates were stained with 0.03% neutral red staining to facilitate plaque counting.
20 μM biotinylated peptide encompassing either the conserved CBD within human influenza M2 (Bio-β-Ala-LDRLFFKCIYRFFKHGL-amid) or a mutant where the CBD core motif is exchanged by alanine residues (Bio-β-Ala-LDRLAFKCIYRFAKHGL-amid) were inoculated with 50 μl NIH3T3 cell lysate (2 ml, T75 flask) for 90 min. Complexes were immobilized using 10 μl streptavidin coated paramagnetic microbeads and μ column (Miltenyi). Washed samples were eluted with 1× sample buffer preheated at 95°C for 2 min and 15 μl out of 70 μl eluate were separated by SDS PAGE, blotted to PVDF membrane and probed with rabbit anti-caveolin-1 antibody.
Cell lysates were incubated with rabbit anti-caveolin-1 antibody (1:2000) or mouse anti-EGFP antibody (1:100) at 4°C for 1 h, treated with 20-50 μl protein A- or G microbeads (Miltenyi) at 4°C for 1 h, and processed as described previously . To deplete cholesterol, cell lysates were treated with 10-20 mM MβCD at room temperature for 1 h before co-immunoprecipitation.
Protein concentrations were determined using the BCA kit (Pierce). 5 μg total protein was separated on a vertical 12% separating gel. Subsequently, proteins were transferred to PVDF membranes using a Transblot™ Semi-dry transfer cell (Bio-Rad). After blocking for 1 h (0.2% CA blocking reagent, Applichem) immunostaining was performed with primary antibody followed by 4 washing steps (TBS 0,02% Tween 20) and addition of the secondary antibody at appropriate dilution. The blots were developed with chemoluminescent substrate (Supersignal Femto West, Supersignal Pico West, Pierce). The band intensities were quantified using QuantityOne software (Bio-Rad) and ImageJ.
The recombinant retroviral vectors were produced from 293T triple transfection of pCMV1MLVGP1, encoding MLV gagpol, pVSV-G, pRVH1-Puro-Cav-1 encoding a shRNA for Cav-1 inhibition and a puromycin resistance gene, as described . For knock-down MDCK (60%-80% confluency) were infected with the respective shRNA retroviral vectors in the presence of 4 mg/ml polybrene for 48 hours. Puromycin-resistant clones were pooled and further analysed 10-27 days after infection.
Plasmids pCav-1 wt or pCav-1 SD (Scaffolding domain mutant) were transiently introduced into MDCK, NIH3T3 or MEF 3T3 KO cells using lipofectamine 2000.
Plasmids pM2PR8_EGFP or pM2PR8DsRed were transiently transfected into MDCK cells by lipofection. The cells were infected with influenza A/PR/8 virus 1 day after transfection. Virus titres were determined from supernatants after additional 24 h of incubation at 37°C.
We thank Prof. Su and Dr. J.-X. Bi (NKLBE, Beijing) for enabling the external fellowship (L.S.). L.S. was supported by the Chinese Scholarship Council, the Helmholtz Association and a grant of the Max-Buchner-Forschungsstiftung. We are grateful to Prof. Yoshihiro Kawaoka (Univ. Madison, Wisconsin, U.S.A.) for the kind gift of pEP24c and Prof. Kai Simons (MPI, Dresden, Germany) for supply with pRVH1-Puro-Cav-1 and pRVH1-Puro. We appreciate the helpful suggestions of Prof. Jürgen Bode (HZI) and support by Prof. Wolfgang Garten (Virology, Marburg, Germany).
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.