Entry of influenza viruses into cells is inhibited by a highly specific protein kinase C inhibitor

Influenza viruses are negative-sense RNA viruses with a segmented genome that replicate in the nucleus of infected cells. They are common pathogens of humans and a wide variety of animal species (Cox & Kawaoka, 1998 ), causing significant morbidity and mortality. There are three groups of Influenza viruses, called influenza A, B and C, with influenza viruses A and B being similar in structure and biological properties and influenza virus C having different proteins responsible for binding and entry into target cells (Hay, 1998 ; Ruigrok, 1998 ).

Entry into a host cell is of fundamental importance in infection and pathogenesis for all viruses. Animal viruses generally enter by one of two mechanisms, either via the endocytic pathway of the cell, or by fusion at the plasma membrane. Uptake through endocytosis is an essential component of the route of entry of viruses such as influenza virus and Semliki Forest virus, but other viruses, including herpesviruses and many retroviruses, generally enter by direct fusion at the plasma membrane [reviewed in Marsh & Helenius (1989) and Marsh & Pelchen-Matthews (1994) ]. For most viruses that enter by endocytosis, the virus is believed to be trafficked towards the late endosome and is therefore part of the lysosome-targeted pathway of endocytosis, which results in progressive acidification of the vesicle and the fusion and/or uncoating of the virus particle. The alternative endocytic pathway, whereby endosomes do not reach a low pH and are recycled back to the cell surface, does not seem to be a common route of productive virus entry.

The entry of influenza virus A has been extensively studied with regard to binding, fusion and uncoating (Hernandez et al., 1996 ; Whittaker et al., 1996 ). Following initial interaction of the virus haemagglutinin (HA) with its sialic receptor, the virus is believed to enter the cell by receptor-mediated endocytosis a process dependent on cellular dynamin (Roy et al., 2000 ). Fusion of the virus envelope with the endosomal membrane then occurs in a compartment having an pH equivalent to the late endosome: approximately pH 5·5 (Stegmann et al., 1987 ; Yoshimura & Ohnishi, 1984 ). Exposure of HA to low pH is essential to trigger its fusion activity (Skehel et al., 1995 ; White et al., 1982 ). Once released into the cytoplasm, the virus uncoats following a low-pH-triggered release of the matrix protein, M1, and the genomic ribonucleoproteins (vRNPs) then rapidly enter the nucleus, via nuclear pore complexes (Bui et al., 1996 ; Martin & Helenius, 1991 ).

The acidification occurring as part of endosome maturation is essential for two events in the virus life-cycle. In addition to being a trigger for HA-mediated fusion, the low pH environment is transferred into the interior of the virus, via the M2 ion channel present in the virus envelope (Pinto et al., 1992 ; Sugrue & Hay, 1991 ). It is the M2-mediated acidification of the virus interior that is the target for the anti-influenza drug amantadine. The mechanism of action of amantadine is to block the M2 ion channel, so preventing the interior of the virus from encountering low pH (Bukrinskaya et al., 1982 ; Hay et al., 1985 ; Martin & Helenius, 1991 ). This results in prevention of release of the influenza virus matrix protein (M1) from the vRNPs during virus uncoating (Bui et al., 1996 ), with the end result that the vRNPs do not enter the nucleus, and replication cannot occur. Amantadine is effective only against strains of influenza A virus, and is ineffective against influenza viruses B and C [see Lamb & Pinto (1997) for a review].

Based on the action of certain protein kinase inhibitors, such as H7 and staurosporine, the entry mechanisms of several enveloped viruses, including rhabdoviruses, alphaviruses, poxviruses and herpesviruses, have been proposed to require cellular protein kinase C (PKC) (Cirone et al., 1990 ; Constantinescu et al., 1991 ). These inhibitors have not been reported to have any effect on entry of orthomyxoviruses, such as influenza virus although H7 has been widely reported as a PKC inhibitor that has effects late in infection and affects influenza viral mRNA and vRNP nuclear export, and late protein production (Bui et al., 2000 ; Kurokawa et al., 1990 ; Martin & Helenius, 1991 ; Vogel et al., 1994 ).

PKC is a large superfamily of related proteins which carry out diverse regulatory roles in many key cellular processes (Mellor & Parker, 1998 ; Toker, 1998 ). Current data on the role of PKC in virus entry have relied on the use of the kinase inhibitors H7 and staurosporine, but these are known to be non-selective in their action and inhibit a wide range of different protein kinases (Bradshaw et al., 1993 ; Garland et al., 1987 ; Quick et al., 1992 ). As such, their use in determining a role for PKC must be treated with caution. Recently, a new generation of PKC inhibitors, the bisindolylmaleimides, has been described (Toullec et al., 1991 ). The bisindolymaleimides are thought to block the ATP-binding site on the catalytic domain of PKC and are highly specific, appearing to inhibit all PKC isozymes with similar potency.

In the present study we used the highly specific PKC inhibitor bisindolylmaleimide I.HCl and examined its effect on the entry and replication of influenza virus. We found that virus replication was inhibited in a dose dependent and reversible manner, and that the block in replication occurred very early in infection, apparently during virus endocytosis and uncoating. These results suggest that activity of PKC is crucial for influenza virus entry, and may be a target for future antiviral therapy.

Cells and viruses
Cell culture.
Mink lung epithelial (Mv-1) cells were passaged twice weekly and maintained in Dulbeccos modified Eagles medium (DMEM) containing 10% foetal bovine serum (FBS), 100 U/ml penicillin and 10 µg/ml streptomycin. Mv-1 cells have recently been reported to be susceptible to all strains of influenza virus tested (Schultz-Cherry et al., 1998 ).

Bisindolylmaleimide I.HCl was obtained from Calbiochem. Unless stated otherwise, cells were pre-treated for 10 min with bisindolylmaleimide I.HCl at the concentration stated and maintained in bisindolylmaleimide throughout the infection.

Virus infections. Influenza A virus (strain A/WSN/33) (Stuart-Harris, 1939 ) was obtained from A. Helenius (ETH-Zürich, Switzerland). Stocks of virus were prepared from the supernatant of infected MadinDarby bovine kidney (MDBK) cells and plaque titred on Madin-Darby canine kidney (MDCK) cells (Whittaker et al., 1995 ).

Influenza B virus (strain B/Yamagata/83) was kindly provided by Peter Palese (Mt Sinai School of Medicine, NY, USA). Stocks of virus were grown in MDCK cells at 34 °C and were plaque titred on MDCK cells.

For infection (15 p.f.u. per cell), virus stocks were first diluted in RPMI 1680 medium containing 0·2% BSA and buffered to pH 6·8 with 10 mM HEPES (RPMI-BSA). Virus was adsorbed for 90 min at 0 °C and cells were then washed and maintained in Mv-1 growth medium containing 2% FBS. Infections were carried out in a 5% CO₂ incubator. For infections at high m.o.i. (approximately 150200 p.f.u. per cell), virus stocks were diluted in RPMI-BSA, adsorbed for 90 min at 0 °C and cells washed and maintained in Mv-1 growth medium without sodium bicarbonate, containing 2% FBS, and buffered to pH 7·3 with 20 mM HEPES. Infections were then carried out by quickly shifting the cells to a water-bath set to 37 °C. A protein synthesis inhibitor (1 mM cycloheximide; Sigma) was added to all high m.o.i. experiments to ensure that only signal from input virus was detected, i.e. to exclude any possibility of signal arising from newly synthesized virus protein.

Production of radioactive virus.
MDBK cells were infected with 15 p.f.u. per cell influenza virus, and 167 µCi/ml ³⁵S Pro-mix (Amersham) was added at 6 h p.i. Virus was harvested from the supernatant at 20 h pi, and concentrated on a 3060% sucrose step-gradient. The peak radioactive fractions containing virus (approximately 1x10⁴ c.p.m./10µl) were pooled, and stored at -80 °C.

Virus binding assay.
Approximately 1x10⁵ cells were washed with ice-cold RPMI-BSA and pre-chilled for 15 min on ice, and approximately 1 p.f.u. per cell influenza virus bound; 5x10³ c.p.m. of radioactive virus was mixed with unlabelled virus (in RPMI-BSA) to give the desired m.o.i. Cells were incubated on ice for 90 min and washed extensively with RPMI-BSA to remove unbound virus. Cells were then lysed in SDSPAGE sample buffer, and radioactivity was determined by scintillation counting.

Assay of virus infectivity.
Mv-1 cells were infected with influenza virus (approximately 1 p.f.u. per cell) in RPMI-BSA buffered with 10 mM HEPES and virus was allowed to adsorb at 4 °C for 90 min. Cells were washed and incubated in Mv-1 growth medium containing 10 mM HEPES, pH 7·3, and 2% FBS for 2 h at 37 °C. Cells were then washed with 0·1 M glycine, 0·1 M NaCl, pH 3·0, for 1 min to remove any uninternalized virus from the surface of cells. The cells were then washed and incubated once more in Mv-1 growth medium containing 10 mM HEPES, pH 7·3, and 2% FBS and infection allowed to proceed until 12 h post-infection. Viral supernatants were then collected and stored at -80 °C.

To quantify the amount of virus released, supernatants were serially diluted in RPMI-BSA and adsorbed to a confluent monolayer of MDCK cells in a six-well plate for 60 min at 37 °C. The cells were then rinsed with RPMI and media replaced with DMEM, 0·2% BSA, 2 µg/ml TPCK-trypsin, including 1% agarose, and incubated at 37 °C for 36 h to permit plaque formation.

SDSPAGE and Western blotting.
For analysis by SDSPAGE, cell monolayers were washed with PBS, and scraped into ice-cold PBS with a rubber policeman. Cells were lysed in SDSPAGE sample buffer and subjected to SDSPAGE on 12% acrylamide gels. Samples were transferred to nitrocellulose by semi-dry blotting and membranes blocked in 5% non-fat dry milk. Samples were analysed using the anti-influenza polyclonal antibody IBO (Whittaker et al., 1995 ) (1:500 dilution for 2 h at room temperature). By Western blot analysis, IBO reacts strongly with influenza virus nucleoprotein (NP; 55 kDa) and more weakly with matrix protein (M1; 27 kDa) (G. Whittaker, unpublished observation). As secondary antibodies we used anti-rabbit alkaline phosphatase (Bio-Rad; 1:5000 dilution for 30 min at room temperature). Blots were developed using standard alkaline phosphate reagents. As a control antibody, we used an anti-calnexin polyclonal antibody at a dilution of 1:500 (kindly provided by A. Helenius, ETH Zürich), which reacts with the cellular protein calnexin (90 kDa by SDSPAGE). Blots were scanned and stored using Adobe Photoshop. Adobe Illustrator was used for final layout of the figures.

Indirect immunofluorescence microscopy.
Immunofluorescence microscopy was carried out essentially as described previously (Whittaker et al., 1995 ). Briefly, cells were fixed with 3% paraformaldehyde in PBS for 15 min, quenched with 50 mM NH₄ClPBS and permeabilized for 5 min with 0·1% Triton X-100/PBS. After blocking in 10% goat serum, cells were incubated with primary and secondary antibodies for 30 min each and mounted in Mowiol. Influenza A virus NP was detected using the monoclonal antibody H10, L16-4R5 (ATCC). Influenza virus neuraminidase was detected using the mouse monoclonal antibody H17, L17-5R17 (kindly provided by J. Yewdell, National Institutes of Health, Bethesda, MD, USA). Influenza B virus NP was detected using the mouse monoclonal antibody 23A7 (kindly provided by Wendy Barclay, University of Reading, UK). As secondary antibody we used Oregon Green 514-labelled goat anti-mouse IgG (Molecular Probes). Hoechst 33258 (Molecular Probes) was used at a concentration of 1 µg/ml. Cells were viewed using a Zeiss Axioskop fluorescence microscope fitted with a UV filter and a long pass 520 nm filter and either a 40x or 63x objective lens. Images were photographed using Ektachrome 400 film (Kodak) and scanned into Adobe Photoshop. Adobe Illustrator was used for final layout of the figure.

Bisindolylmaleimide blocks influenza A virus replication
Protein kinase inhibitors such as H7 have been widely used to study influenza virus replication, but their actual target in the infected cells is very poorly defined. As bisindolylmaleimides are highly specific protein kinase C (PKC) inhibitors, we initially tested these compounds to determine their effects on influenza virus replication. We infected Mv-1 cells with influenza virus (WSN) and treated the cells with various amounts of bisindolylmaleimide I. Infected cells were assayed by immunofluorescence microscopy at 8 h post-infection using antibodies specific for the influenza virus late glycoprotein neuraminidase (NA) (Fig. 1). In untreated cells, NA was expressed in all cells and was localized to intracellular membranes and the cell surface. In cells treated with 5 µM bisindolylmaleimide I, most cells still showed normal NA expression, but a minority of cells showed little or no expression of NA, indicating that replication of influenza virus was inhibited. Fig. 1(B) shows that the cells not expressing NA (in the green channel of the immunofluorescence microscope) instead showed a red-orange signal. This signal is due to inherent autofluorescence of bisindolylmaleimide I, which increases in intensity with the amount of drug added to the cells (for example compare Fig. 1B, C and D). When cells were treated with 10 µM bisindolylmaleimide I only a small minority of cells showed NA expression and with 20 µM bisindolylmaleimide I no cells with NA expression could be detected (Fig. 1C and D). These data show that bisindolylmaleimide I prevents expression of an influenza virus glycoprotein and indicate that infection is inhibited in a dose-dependent manner.

(50K): Fig. 1. Effects of bisindolylmaleimide on expression of influenza A virus NA and NP. Mv-1 cells were infected with influenza A virus (WSN) and treated with various amounts of bisindolylmaleimide (520 µM) at the time of infection, or were left untreated (no treatment). Cells were analysed after 8 h by immunofluorescence microscopy using an anti-NA antibody (AD) or an anti-NP antibody (EH). Increasing levels of bisindolylmaleimide inhibited expression of both viral proteins.

To address any possible differences between previous experiments using an alternative kinase inhibitor, H7 (Bui et al., 2000 ), we next examined infected Mv-1 cells using anti-nucleoprotein (NP) antibodies at 8 h post-infection. In untreated cells NP was expressed and in most cells showed a distinct localization to the cytoplasm (Fig. 1E), indicating that vRNP nuclear export had occurred. In the presence of bisindolylmaleimide I, NP expression was markedly reduced, and was completely inhibited at a concentration of 20 µM (Fig. 1H). As with NA expression, the synthesis of NP was inhibited in a dose-dependent manner.

In most cases expression of both NP and NA was completely abolished in the presence of bisindolylmaleimide I, i.e. the inhibitor had an all-or-nothing effect. However, in some cells treated with low levels of bisindolylmaleimide I (5 µM) there was a nuclear signal for NP (Fig. 1F, arrows). This most probably represents a general slowing of virus replication rather than any specific block in virus trafficking. This is in contrast to cells treated with H7, which show NP expression, but with retention of both the NP and vRNPs in the nucleus (Martin & Helenius, 1991 ; Bui et al., 2000 ). Overall, bisindolylmaleimide I seems to block influenza virus infection in a mechanistically quite different manner to the more extensively studied protein kinase inhibitor H7.

To confirm and extend the results of immunofluorescence microscopy, we infected Mv-1 cells with influenza virus (WSN), treated them with various amounts of bisindolylmaleimide I, and analysed cell lysates by SDSPAGE and Western blotting (Fig. 2). Infection with influenza virus was assayed using the anti-influenza polyclonal antibody IBO. As expected, in the absence of bisindolylmaleimide I, lysates of influenza virus-infected cells showed strong reactivity with NP and weaker reactivity with M1, and lysates of mock-infected Mv-1 cells showed no reactivity. Addition of 5 µM bisindolylmaleimide I reduced levels of NP and M1 slightly, and addition of 10 µM bisindolylmaleimide I resulted in further reductions in the expression of NP and M1. Following addition of 20 µM bisindolylmaleimide I, levels of both NP and M1 were suppressed markedly, and both proteins were only barely detectable by Western blot. As a loading control, we assayed a duplicate blot with an anti-calnexin antibody and showed that essentially identical levels of protein were present in each lane. These experiments confirm that influenza virus replication is inhibited in a dose-dependent manner by bisindolylmaleimide I.

(119K): Fig. 2. Expression of influenza A virus proteins is blocked in a dose-dependent manner by bisindolylmaleimide. Mv-1 cells were mock-infected (M), or infected with influenza A virus (WSN) and treated with various amounts of bisindolylmaleimide (020 µM) at the time of infection. Cells were analysed after 8 h by Western blotting, using anti-influenza virus (WSN) sera. Expression of viral protein was inhibited by increasing levels of bisindolylmaleimide. Levels of a control cellular antigen (calnexin; CNX) were essentially identical in all samples.

To examine the overall effects of bisindolylmaleimide I on the production of infectious virus, we infected Mv-1 cells with influenza virus WSN in the presence or absence of drug at various concentrations. Supernatants were collected from these cells and assayed for infectious virus production by plaque assay (Fig. 3). Bisindolylmaleimide I inhibited influenza virus replication in a dose-dependent manner. In cells treated with 510 µM bisindolylmaleimide I, we found that virus infection was inhibited by approximately 1 log unit, at 20 µM by approximately 2 log units, and with 40 µM bisindolylmaleimide virus infectivity was inhibited by more than 3 log units. These data confirm that bisindolylmaleimide has a significant inhibitory effect on influenza virus replication.

(15K): Fig. 3. Production of infectious influenza A virus proteins is blocked in a dose-dependent manner by bisindolylmaleimide. Approximately equal numbers of Mv-1 cells (5x104) were infected with 1 p.f.u. per cell influenza virus (WSN) and treated with various amounts of bisindolylmaleimide, or left untreated. Supernatants were harvested 12 h after infection, serially diluted and analysed by plaque assay on MDCK cells. Data are the mean of duplicate experiments.

Bisindolylmaleimide inhibits influenza B virus replication
We wished to address the effects of bisindolylmaleimide I on the replication of influenza B virus. Mv-1 cells were infected with influenza B virus (B/Yamagata/83) and influenza A virus (WSN), or were mock infected. Cells were treated with bisindolylmaleimide I at the time of infection and, following virus adsorption, were incubated at 34 °C for 12 h, before fixation for immunofluorescence microscopy. Incubations were done at 34 °C (instead of 37 °C) as this allowed most efficient infection with influenza B viruses, and the time was extended from previous experiments to allow for the slower growth at the reduced temperature.

In untreated cells both influenza A and B virus NP was expressed and in most cells showed a predominant localization to the cytoplasm (Fig. 4A and C). In the presence of 10 µM bisindolylmaleimide I influenza virus NP expression was markedly inhibited, with less than 1% of cells expressing NP. These experiments show that influenza B virus is also susceptible to inhibition by bisindolylmaleimide at micromolar concentrations.

(101K): Fig. 4. Bisindolylmaleimide inhibits influenza B virus replication. Mv-1 cells were infected with influenza A WSN virus (A/WSN) or with influenza B Yamagata/83 virus (B/Yama). Cells were treated with bisindolylmaleimide (5 µM) at the time of infection, or were left untreated (no treatment). Cells were incubated at 34 °C and analysed after 12 h by immunofluorescence microscopy using an anti-influenza A virus NP antibody (H10, L16-4R5; panels C and D) or an anti-influenza B virus NP antibody (23A7; panels A and B). Expression of NP for both viruses was inhibited by bisindolylmaleimide.

Bisindolylmaleimide acts early in the infectious cycle
As bisindolylmaleimide I acts differently to other protein kinase inhibitors such as H7, we wished to determine the point in the infectious cycle where the drug is acting. As a preliminary analysis, we carried out experiments where the drug was added at successively later times after the onset of infection, with the aim of determining a point at which the drug no longer inhibited influenza virus replication. Fig. 5 shows the results of an experiment where bisindolylmaleimide I was added 2 h after infection. The expression of NP was unaffected by the addition of bisindolylmaleimide I, up to the maximal level of 20 µM tested, and NP showed essentially the same high levels of cytoplasmic distribution as the untreated control (Fig. 5EH). As little or no replication of viral RNA, or viral protein production or assembly would have occurred within the first 2 h of infection, this indicated that bisindolylmaleimide I acted during the process of virus entry.

(56K): Fig. 5. The effects of bisindolylmaleimide on replication of influenza virus occur early in infection. Mv-1 cells were infected with influenza A virus (WSN) and treated with various amounts of bisindolylmaleimide (520 µM) after 2 h of infection, or were left untreated (no treatment). Cells were analysed after 8 h by immunofluorescence microscopy using an anti-NA antibody (AD) or an anti-NP antibody (EH). The effects of bisindolylmaleimide were generally abrogated when added 2 h after infection.

When we examined the effects of adding bisindolylmaleimide I at 2 h post-infection on expression of the late glycoprotein NA, we found that the inhibitor did show some effect. Although some NA expression was prevented at the higher concentration of bisindolylmaleimide (Fig. 5C and D), the effect was much less pronounced when added at 2 h post-infection. As NP expression was apparently unaffected, inhibition of NA expression by bisindolylmaleimide at this time may be due to non-specific effects on glycoprotein production or trafficking.

Bisindolylmaleimide reversibly blocks virus entry into the nucleus
Because the effects of bisindolylmaleimide on influenza virus infection appeared to occur at early times of infection, we examined the effects of the inhibitor on virus entry. We infected Mv-1 cells with a high m.o.i. of virus and treated infected cells with bisindolylmaleimide I. We then assayed cells 60 min after infection by immunofluorescence microscopy. In these experiments cycloheximide was added to ensure that all signal was due to input virus and not replicating virus. Fig. 6 clearly shows that in the absence of bisindolylmaleimide, vRNPs efficiently enter the nucleus by the 60 min time-point, and cells showed a bright and distinct nuclear signal with little or no fluorescent signal from vRNPs present in the cytoplasm (Fig. 6D). However, on addition of 20 µM bisindolylmaleimide I, the nuclear signal from vRNPs was completely lost, and instead vRNPs were visible in weak punctate areas of the cytoplasm (Fig. 6E). In bisindolylmaleimide-treated cells we also observed low levels of red-orange fluorescence due to the presence of the drug (Fig. 6E), but as the cells were exposed to the inhibitor for much shorter periods of time the fluorescent signal was much weaker than in earlier experiments; compare Fig. 6 to Figs 1, 4 and 5. Overall, these experiments clearly show that bisindolylmaleimide I blocks the entry mechanism of influenza virus within the first 60 min of infection during virus entry.

(77K): Fig. 6. Entry of influenza virus RNPs into the nucleus is reversibly blocked in a dose-dependent manner by bisindolylmaleimide. Mv-1 cells were infected with influenza A virus (WSN) and treated with 20 µM bisindolylmaleimide (+bisindolylmaleimide; panels B and E) at the time of infection, or were left untreated (no treatment; panels A and D). Cells were analysed after 60 min h by immunofluorescence microscopy using an anti-NP antibody (DF), or with Hoechst 33258 (AC). After 60 min those cells treated with bisindolylmaleimide were washed and incubated in medium and analysed after a further 60 min (washout; panels C and F). Entry of vRNPs was blocked in cells treated with bisindolylmaleimide, but resumed after drug washout.

We also wished to determine if the effects of bisindolylmaleimide I were reversible. To address this issue, Mv-1 cells were infected with a high m.o.i. of virus and infection initiated for 60 min in the presence of drug. The drug was then removed from the cells and infection allowed to proceed for a further 60 min (Fig. 6F). Under these conditions we now saw essentially the same nuclear signal of vRNPs as observed in untreated cells. In addition all the red-orange fluorescence due the bisindolylmaleimide was lost from the cells, indicating efficient wash-out of the inhibitor. These experiments show that the bisindolylmaleimide-induced block in influenza virus entry is rapidly and completely reversible.

Effects of bisindolylmaleimide on virus binding
To determine whether the presence of bisindolylmaleimide affected infection due to impaired virus binding, radiolabelled influenza virus (WSN) was mixed with an excess of unlabelled virus and adsorbed at 0 °C onto Mv-1 cells pretreated with 20 µM bisindolylmaleimide I, or left untreated. Virus was allowed to bind for 90 min in the presence or absence of bisindolylmaleimide. Cell-associated counts were then determined. No significant difference in virus binding was detected in the presence or absence of bisindolylmaleimide (Fig. 7). These data show that bisindolylmaleimide has no effect on influenza virus binding to cells.

(23K): Fig. 7. Binding of influenza virus to Mv-1 cells is not affected by bisindolylmaleimide. Radioactive influenza virus (WSN) was prepared and bound to the surface of Mv-1 cells on ice, in the presence of 20 µM bisindolylmaleimide, or without inhibitor (no treatment). The amount of bound virus was determined by scintillation counting. Error bars represent the standard deviation for three independent experiments.

Bisindolylmaleimide is not acting as a weak base
As the effects of bisindolylmaleimide appeared to be acting during influenza virus entry, we wished to determine how the action of the drug compared with non-specific virus entry inhibitors such as NH₄Cl and chloroquine. These two compounds act as weak bases and, when added to cells at millimolar levels, prevent entry of pH-dependent viruses by preventing endosome acidification and virus fusion and uncoating. Importantly, the effects of weak bases such as NH₄Cl and chloroquine are sensitive to the external pH of the medium and they have little or no effect on endosome acidification when the external medium is buffered to pH 7·0. or below (Yoshimura et al., 1982 ; Kielian & Helenius, 1986 ). We therefore used this pH effect as a diagnostic tool to determine if the effects of bisindolylmaleimide, at the concentrations showing antiviral activity, were due to any weak base activity of the compound.

Fig. 8 shows immunofluorescence microscopy of cells 60 min after infection with a high m.o.i. of influenza virus WSN. In the presence of NH₄Cl at 20 mM, a concentration known to prevent endosome acidification and influenza virus infection, no vRNPs were detectable at pH 7·8. However, at pH 6·5 the weak base-effect of NH₄Cl was neutralized by the external medium and vRNP entry into the nucleus was observed. In contrast, bisindolylmaleimide I (20 µM) inhibited virus entry independent of the pH of the external medium, showing that it is not acting non-specifically as a weak base at micromolar concentrations.

(100K): Fig. 8. Bisindolylmaleimide is not acting as a weak base. Mv-1 cells were infected with influenza A virus (WSN) and treated with 20 µM bisindolylmaleimide (+BIM; panels A and B) at the time of infection, or with 20 mM ammonium chloride (+NH4Cl; panels C and D). The external medium was buffered to either pH 7·8 (panels B and D) or pH 6·5 (panels A and C) using HEPES. Cells were analysed after 60 min by immunofluorescence microscopy using an anti-NP antibody. Inhibition of vRNP nuclear entry by bisindolylmaleimide was independent of pH.

We show here that the protein kinase inhibitor bisindolylmaleimide prevents entry of influenza virus into the nucleus of cells early in infection. Bisindolylmaleimide acts in a quite different manner during influenza virus infection to the more thoroughly characterized protein kinase inhibitor H7. Whereas H7 prevents vRNP exit from the nucleus late in infection, our experiments indicate that bisindolylmaleimide has no such function. When virus was allowed to enter cells and bisindolylmaleimide then added, vRNPs appeared to be exported normally at 8 h post-infection (see Fig. 5). Although many studies on the effects of H7 during influenza virus replication have been carried out (Bui et al., 2000 ; Kurokawa et al., 1990 ; Martin & Helenius, 1991 ; Vogel et al., 1994 ), its precise mechanism of action and cellular target remain unclear. We suggest that the target of bisindolylmaleimide in the virus life-cycle is quite different from that of H7, as it does not affect vRNP nuclear export. This would imply that different cellular kinases are responsible for regulating the two phases of the virus life-cycle (entry vs nuclear export) and that bisindolylmaleimide (a highly specific PKC inhibitor) and H7 have different targets. This hypothesis is reinforced by biochemical evidence that the main target of H7 is not PKC (Garland et al., 1987 ; Quick et al., 1992 ), and our own unpublished data which show no effect of H7 at concentraions up to 100 µM on virus entry (M. Bui & G. Whittaker, unpublished results). An effect of bisindolylmaleimide during virus entry is suggested by the finding that inhibition seems to be an all-or-nothing phenomenon. If the inhibitor were affecting a point in the virus life-cycle such as transcription or replication, then a general lowering of protein expression might be anticipated with increased inhibitor concentration. The all-or-nothing effect suggests a specific early block, such that those viruses that escape drug inhibition go on to give normal levels of protein expression.

At present, our data show that bisindolylmaleimide inhibits influenza virus entry within the first 60 min of infection at some point prior to entry of vRNPs into the nucleus. Interestingly, the point of action at which the anti-influenza drug amantadine is observed to act is also at the stage of vRNP nuclear import (Bukrinskaya et al., 1982 ; Kato & Eggers, 1969 ; Martin & Helenius, 1991 ). It is now well-established that at micromolar concentrations amantadine acts by blocking the M2 ion channel of influenza A viruses (Pinto et al., 1992 ; Sugrue & Hay, 1991 ), and in the presence of the drug pH-dependent virus uncoating and nuclear transport do not occur. Thus, amantadine and bisindolylmaleimide appear to inhibit the virus life-cycle at similar points. However, it is well established that amantadine has no effect on replication of influenza B viruses at micromolar concentrations because the function of M2 is replaced by a different protein (NB) in influenza B viruses [see Lamb & Pinto (1997 ) for a review]. Our experiments with influenza B virus clearly show that bisindolylmaleimide inhibits replication of this virus in addition to influenza A virus. Overall, our results demonstrate that amantadine and bisindolylmaleimide appear to have different targets in the entry pathway and that bisindolylmaleimides represents a new class of influenza virus entry inhibitors. Although we cannot exclude an effect on fusion or nuclear import at the present time, we feel that the most likely point at which bisindolylmaleimide has its inhibitory effect is in the process of endocytosis.

Several groups have suggested a role for PKC in endocytosis. Activators of PKC such as phorbol esters stimulate receptor-mediated endocytosis of the HIV receptors CD4 and CXCR4 36-fold (Pelchen-Matthews et al., 1993 ; Signoret et al., 1997 ), and in the case of CD4 appear to divert receptor from the endosomal recycling pathway to a late-endosomal compartment (Pelchen-Matthews et al., 1993 ). Phorbol esters are thought to promote endocytosis either by increasing PKC-dependent phosphorylation to the internalization motif located in the cytoplasmic tail of the receptor (Signoret et al., 1997 ) or by promoting endosomeendosome fusion (Aballay et al., 1999 ). In many cases, the increased levels of endocytosis induced by PMA can be abrogated by addition of PKC inhibitors such as staurosporine or calphostin C. PKC inhibitors have also been shown to block the endocytosis of certain hormone receptors into cells, although the mechanism of action is incompletely understood (Ferrari et al., 1999 ). More specifically, a role for one specific isotype of PKC (PKCλ) has been shown in lysosome-targeted endosomes. PKCλ is targeted to late endosomes, and a dominant negative mutant of PKCλ severely inhibits endocytosis by the lysosome-targeted pathway but not the recycling pathway (Sanchez et al., 1998 ).

It remains to be determined whether bisindolylmaleimide drugs have effects on entry of other viruses in addition to influenza virus. Based on a predicted site of action in the endocytic pathway, it is possible that bisindolylmaleimide I will be inhibitory for other pH-dependent viruses such as vesicular stomatitis virus or Semliki Forest virus. Experiments to address effects of bisindolylmaleimide on the endocytosis of these viruses are currently in progress. One of the most effective ways of preventing virus infection and disease is to prevent the initial entry of virions into their target cells (Dimitrov, 1997 ; Marsh & Pelchen-Matthews, 1994 ). Further studies on the precise mechanism of action of bisindolylmaleimide with respect to virus entry will significantly add not only to our knowledge of virus entry into cells and possible antiviral therapy, but also to our understanding of the cellular mechanisms of endocytosis.

We thank Ruth Collins and Melissa Grabowski for critical reading of this manuscript. We also thank Peter Palese, Wendy Barclay, Ari Helenius and Jon Yewdell for their generous contributions of reagents. C.R. was supported by the Cornell Hughes Undergraduate Research Program and L.M. was supported by the Cornell University Leadership Alliance. These studies were funded in part by a Scientist Development Grant from the American Heart Association (to G.W.).

Footnotes

^b Present address: Department of Microbiology, Duke University Medical Center, Durham, NC 27710, USA.