Antiviral Effects of an Iminosugar Derivative on Flavivirus Infections

Abstract

Endoplasmic reticulum (ER) α-glucosidase inhibitors, which block the trimming step of N-linked glycosylation, have been shown to eliminate the production of several ER-budding viruses. Here we investigated the effects of one such inhibitor, N-nonyl-deoxynojirimycin (NN-DNJ), a 9-carbon alkyl iminosugar derivative, on infection by Japanese encephalitis virus (JEV) and dengue virus serotype 2 (DEN-2). In the presence of NN-DNJ, JEV and DEN-2 infections were suppressed in a dose-dependent manner. This inhibitory effect appeared to influence DEN-2 infection more than JEV infection, since lower concentrations of NN-DNJ substantially blocked DEN-2 replication. Secretion of the flaviviral glycoproteins E and NS1 was greatly reduced, and levels of DEN-2 viral RNA replication measured by fluorogenic reverse transcription-PCR were also decreased, by NN-DNJ. Notably, the viral glycoproteins, prM, E, and NS1 were found to associate transiently with the ER chaperone calnexin, and this interaction was affected by NN-DNJ, suggesting a potential role of calnexin in the folding of flaviviral glycoproteins. Additionally, in a mouse model of lethal challenge by JEV infection, oral delivery of NN-DNJ reduced the mortality rate. These findings show that NN-DNJ has an antiviral effect on flavivirus infection, likely through interference with virus replication at the posttranslational modification level, occurring mainly in the ER.

The Flaviviridae family includes the three genera Hepacivirus (e.g., Hepatitis C virus), Flavivirus, and Pestivirus (e.g., Bovine viral diarrhea virus [BVDV]). The genus Flavivirus comprises more than 70 viruses, many of which are potent human pathogens that can cause severe encephalitic, hemorrhagic, hepatic, or febrile illnesses (35). Of particular importance for public health are the mosquito-borne flaviviruses, such as Yellow fever virus (41, 48), Japanese encephalitis virus (JEV) (45), West Nile virus (25), and dengue viruses (DENs) (16). Severe forms of dengue-related diseases, such as dengue hemorrhagic fever and dengue shock syndrome, have recently become the most serious vector-borne viral diseases in humans. It is estimated that more than 50 million people suffer DEN infection annually, with approximately 2.5 billion people living in at-risk areas (16, 34). Even though these flaviviruses have a major clinical impact, there is still no vaccine for DENs, nor are there any specific antiviral therapeutics available for treatment of infections with JEV or DENs.

Flaviviral virions are composed of a lipid bilayer with two or more envelope proteins surrounding a nucleocapsid, which consists of a single-stranded positive-sense genome RNA associated with multiple copies of capsid proteins. After entering a host cell, flaviviral RNA first translates into a long polyprotein, which is cleaved by cellular and viral proteases into individual structural and nonstructural proteins. RNA replication begins with the synthesis of complementary negative strands, which are then used as templates for reproduction of positive-stranded RNA. Flaviviruses are thought to replicate exclusively in the cytoplasm and to mature on the intracellular membranes of infected cells. Employing the intrinsic secreting pathway of infected cells, flaviviruses bud from the membranes of the endoplasmic reticulum (ER) and Golgi apparatus to release mature virions (40).

Three of the flaviviral proteins carry N-linked glycans: two of these are the precursor of membrane (prM) protein and envelope (E) protein, which are virion components, and the other is the nonstructural protein NS1 (7, 40). During virus secretion through acidic cellular compartments, prM protein associates with E protein. This association prevents an irreversible conformational change of E protein (1). At the late stage of infection, proteolytic cleavage of prM to M protein results in mature virions. NS1 carries 12 invariant cysteine residues, which are highly conserved among the flaviviruses. While no precise function has been ascribed to NS1, it clearly plays some critical role in viral RNA replication (28, 29, 36, 37). An additional NS1-2A-related protein (named NS1′) with a molecular size of 53 kDa is often observed in JEV-infected cells; it is presumably generated by an unknown cellular protease that recognizes an alternative cleavage site within NS2A (7).

During the synthesis of N-linked glycans in mammalian cells, a 14-residue oligosaccharide core unit, (Glc)₃(Man)₉(GlcNAc)₂, is added cotranslationally to the newly synthesized polypeptides in the ER. The chain of glycan molecules is then biochemically modified within the ER and the Golgi apparatus to generate the diversified glycan moieties found in mature glycoproteins. ER α-glucosidases I and II are involved in the trimming of terminal glucose on core oligosaccharides, and the resulting monoglucosylated glycoproteins can bind to ER chaperones, the membrane-bound calnexin (CNX) and/or its soluble homologue calreticulin (CRT) (17, 38, 42). Removal of the last glucose from the glycans by glucosidase II releases the glycoprotein from CNX and/or CRT. The binding between CNX and/or CRT and glycoprotein has been shown to increase the folding efficiency of the given glycoproteins (14, 18, 19). Glycans may be reglucosylated by UDP-Glc:glycoprotein glucosyltransferase (GT) when proper protein folding is not accomplished, and the resulting monoglucosylated glycoproteins can again be recognized and bound by CNX and/or CRT. Such a deglucosylation-reglucosylation cycle continues until proper glycoprotein folding is achieved (20, 39) or until processing of the glycans by ER mannosidase targets the glycoprotein for proteasomal degradation. Many, but not all, glycoproteins are able to interact with CNX and/or CRT. Addition of α-glucosidase I and/or α-glucosidase II inhibitors, such as castanospermine (CST) and deoxynojirimycin (DNJ), prevents the interaction of CNX and/or CRT with folding glycoproteins (39). Enveloped animal viruses often contain one or more viral glycoproteins, and the α-glucosidase inhibitors have been shown to block infection by several viruses (32), such as vesicular stomatitis virus (44), Sindbis virus (43), human immunodeficiency virus (HIV) (15, 50), human cytomegalovirus (47), woodchuck hepatitis virus (4), BVDV (51), and dengue virus serotype 1 (DEN-1) (10).

The iminosugar DNJ and its N-alkylated derivatives have been characterized for their inhibition of α-glucosidase, and N alkylation of DNJ has been shown to increase its inhibitory potency (46). N-Nonyl-DNJ (NN-DNJ), a 9-carbon alkyl derivative of DNJ, was found to be more potent than N-butyl-DNJ (NB-DNJ) in inhibiting hepatitis B virus (HBV) (4) and BVDV (51). In the present study, we observed a potent antiviral effect, in vitro and in vivo, of NN-DNJ on infection by JEV and DEN-2. We found that one of the molecular targets for the action of NN-DNJ against flaviviruses might be the interaction between the ER chaperone CNX and viral glycoproteins. These results may provide a molecular explanation for the ability of α-glucosidase inhibitors to suppress flavivirus replication in infected hosts.

MATERIALS AND METHODS

Cell lines, viruses, and chemicals.

BHK-21 cells were cultured in RPMI 1640 medium containing 5% fetal bovine serum (FBS) and 2 mM l-glutamine. The plaque-purified neurovirulent JEV strain RP-9 (9) was used in in vitro culture and in vivo challenge experiments. The Taiwanese DEN-2 strain PL046 (27) was also used in this study. Virus propagation was carried out in C6/36 cells by using RPMI 1640 medium containing 5% FBS. Virus titers were determined by a plaque-forming assay on BHK-21 cells. NN-DNJ was dissolved in methanol, and its tartrate salt was dissolved in water. For in vitro study, both methanol and water solutions were used; for in vivo study, the NN-DNJ tartrate salt solution was used. N-alkylated DNJ compounds were synthesized by using sodium cyanoborohydride (NaBH₃CN) and the appropriate aldehyde as described previously (33).

Virus infection and titration.

For infection with JEV or DEN-2, monolayers of BHK-21 cells in 6- or 12-well plates were adsorbed with virus for 1 h at 37°C. After adsorption, unbound viruses were removed by gentle washing with serum-free medium, followed by addition of fresh medium containing various amounts of NN-DNJ for further incubation at 37°C. To determine virus titers, culture media were harvested for plaque-forming assays. Various virus dilutions were added to 80% confluent BHK-21 cells and incubated at 37°C for 1 h. After adsorption, cells were washed and overlaid with 1% agarose (SeaPlaque; FMC BioProducts) containing RPMI 1640 with 1% FBS. After incubation for 4 days for JEV and 7 days for DEN-2, cells were fixed with 10% formaldehyde and stained with 0.5% crystal violet.

Indirect immunofluorescence assay.

Cells were fixed in an acetone-methanol (1:1) solution for 3 min and then reacted with a monoclonal antibody (MAb) against JEV NS3 (8) or DEN-2 NS3 (27). After a wash with phosphate-buffered saline (PBS), cells were further stained with a goat anti-mouse fluorescein isothiocyanate (FITC)-conjugated secondary antibody (Jackson ImmunoResearch), and the resulting cells were examined under a Leica fluorescent microscope. Cell nuclei were visualized by 4′,6′-diamidino-2-phenylindole (DAPI) staining in 0.9% sodium chloride at room temperature for 5 min.

XTT assay.

To determine cell viability, a colorimetric XTT-based assay was performed (Cell Proliferation Kit II; Roche). BHK-21 cells in a 96-well plate were incubated with various concentrations of NN-DNJ for 1 or 2 days before the XTT labeling reagent was added to the culture medium. Cells were incubated at 37°C for about 30 min and then read by an enzyme-linked immunosorbent assay (ELISA) reader at 450 nm (Molecular Devices).

LDH assay.

Virus-induced cytopathic effect (CPE) was assessed by the release of the cytoplasmic enzyme lactate dehydrogenase (LDH) using a commercial kit (Cytotoxicity Detection Kit; Roche). Culture supernatants from cell samples were clarified by centrifugation, mixed with the reaction mixture [diaphorase/NADH⁺, 2-(4-iodophenyl)-3-(4-nitrophenyl)-5-phenyltetrazolium chloride(INT)/sodium lactate], incubated at room temperature for about 30 min, and then read by an ELISA reader at 490 nm (Molecular Devices).

RIP.

Cell labeling and immunoprecipitation were performed as described previously (26). Briefly, BHK-21 cells were adsorbed with virus at a multiplicity of infection (MOI) of 5 for 1 h. The virus inoculants were removed; cells were washed once and then replenished with medium plus different concentrations of NN-DNJ. At 12 h postinfection, the medium was removed and replaced with warm methionine (Met)- and cysteine (Cys)-free RPMI 1640 medium containing 100 μCi of ³⁵S-labeled Pro-mix (Amersham)/ml and 2% dialyzed FBS (GIBCO) plus various amounts of NN-DNJ, and the mixture was incubated for 2 h or 20 min at 37°C. Cell lysates were harvested in lysis buffer (1% NP-40, 150 mM NaCl, 50 mM Tris-HCl [pH 7.5], 1 mM EDTA) containing protease inhibitors (Boehringer Mannheim). Mixtures of at least two MAbs specific for JEV (8) and DEN-2 (27) viral proteins were used in the radioimmunoprecipitation (RIP). The resulting immune complexes were washed three times with RIP assay buffer (10 mM Tris-HCl [pH 7.5], 150 mM NaCl, 5 mM EDTA, 0.1% sodium dodecyl sulfate [SDS], 1% Triton X-100, 1% sodium deoxycholate), analyzed by SDS-polyacrylamide gel electrophoresis (PAGE), and fluorographed at −70°C.

Coimmunoprecipitation.

Complex formation between CNX and viral proteins was studied as previously described (6). Cells were prepared and labeled as described above and then lysed under a milder condition using a 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS)-HSE buffer (2% CHAPS in 50 mM HEPES [pH 7.5]-200 mM NaCl-2 mM EDTA). For coimmunoprecipitation experiments, lysates were immunoprecipitated with anti-CNX (StressGen), washed with 0.5% CHAPS-HSE buffer, and then analyzed by SDS-PAGE and fluorographed.

Quantitative detection of DEN-2 using fluorogenic reverse transcription-PCR (RT-PCR).

The viral RNA used in this study was extracted from cells by using a Viral RNA Extraction Kit (Virogen) according to the manufacturer's instructions. Viral RNA was reverse transcribed to cDNA by using a ThermoScript RT kit (GIBCO BRL). To detect positive- and negative-sense viral RNA, a primer annealing to DEN-2 nucleotides 10723 to 10703 (5′-AGAACCTGTTGATTCAACAGC-3′) or nucleotides 9709 to 9729 (5′-CATGAGTTAATCATGAAAGAC-3′), respectively, was used in the RT reaction. Real-time PCR to detect the DEN-2 genome was conducted as previously described (21). ABI PRISM 7700 sequence detection system, version 1.7 (Perkin-Elmer Applied Biosystems), was employed for PCR cycling, real-time data collection, and analysis.

In vivo protective effect of NN-DNJ in a JEV mouse challenge model.

Six- to seven-week-old female ICR mice were obtained from the National Laboratory Animal Breeding and Research Center, Taipei, Taiwan. Throughout the entire experiment period, NN-DNJ (tartrate salt) dissolved in water was orally delivered by feeding tube once per day at 0, 20, or 200 mg of NN-DNJ/kg of body weight/day starting from 1 day before JEV challenge. Each mouse was intraperitoneally (i.p.) challenged with 10⁵ PFU of JEV (approximately 10 50% lethal doses) in 300 μl of PBS and was simultaneously intracranially (i.c.) injected with 30 μl of PBS in the right hemisphere of the brain (i.p. plus i.c. route) (26). To ensure the depth of each i.c. injection, we used 27-gauge one-stop needles for i.c. inoculation. Mouse mortality was monitored daily for 3 weeks. The results of survival curves in different groups were analyzed by using the log rank test.

RESULTS

In vitro anti-JEV effect of NN-DNJ.

We first determined the noncytotoxic concentrations of NN-DNJ for BHK-21 cells. As shown in Fig. 1B, there was no difference in cell viability between untreated cells and cells treated with as much as 100 μM NN-DNJ for 1 or 2 days, and only a slight reduction in cell viability could be observed after treatment with 150 μM NN-DNJ. We then assayed the potential anti-JEV effect of NN-DNJ in a BHK-21 cell culture system. After adsorption with JEV (at an MOI of 0.1 or 5), cells were washed, replenished with medium containing varying amounts of NN-DNJ (0 to 100 μM), and incubated for 2 days. The extent of JEV infection in these cells was then measured by immunostaining with anti-JEV NS3 (8) and an FITC-conjugated secondary antibody. As shown in Fig. 1A, the number of JEV-infected cells was greatly reduced by NN-DNJ in a dose-dependent manner. The reduction was especially evident at an MOI of 0.1; at a high MOI, although the reduction in viral protein expression was not as prominent as at a lower MOI, there was a reduction of approximately 20% in JEV-positive cells when cells were treated with 100 μM NN-DNJ. Titers of infectious virus present in the supernatants were determined by plaque-forming assays and are shown in Fig. 1C. Irrespective of the MOI used, the virus yields from JEV-infected BHK-21 cells were reduced in a dose-dependent manner. By measuring the release of a cytoplasmic enzyme, LDH, into the culture medium, we determined the protective effect of NN-DNJ on the JEV-induced CPE. As shown in Fig. 1D, without any treatment, JEV caused a severe CPE at an MOI of 5, whereas addition of NN-DNJ greatly reduced the JEV-induced CPE. The anti-JEV effect of NN-DNJ was further kinetically studied at different time points after infection using an MOI of 0.1. An NN-DNJ dose-dependent reduction of virus production was observed at each of the three days post-JEV infection (Fig. 2). Together, the decreases in JEV antigen expression, virus production, and virus-induced CPE indicate that NN-DNJ possesses an anti-JEV effect.

FIG. 1.

In vitro anti-JEV effects of NN-DNJ. (A) BHK-21 cells infected with JEV at an MOI of 0.1 or 5 were treated with various doses of NN-DNJ (0 to 100 μM) as indicated on the left. Two days postinfection, cells were fixed and stained with an anti-JEV NS3 MAb and an FITC-conjugated secondary antibody. Cell nuclei were stained by DAPI. Pictures were taken using an inverted fluorescent microscope (Leica) by double exposure of the same fields with filters for FITC and DAPI. (B) Effect of NN-DNJ on BHK-21 cells. The viability of BHK-21 cells incubated with various concentrations of NN-DNJ (0 to 150 μM) for 1 or 2 days was determined by an XTT assay (Roche) according to the manufacturer's instructions. Data are averages and standard errors from two or three wells for each concentration tested. (C) Infectious virus titers in NN-DNJ-treated cell culture supernatants. Virus titers (expressed as PFU per milliliter) produced by BHK-21 cells treated with various doses of NN-DNJ were determined by plaque assays. (D) Levels of CPE induced by JEV infection in the presence of various doses of NN-DNJ were determined by the release of the cytoplasmic enzyme LDH as described in Materials and Methods. Representative data from two independent experiments are shown.

FIG. 2.

In vitro anti-JEV effects of NN-DNJ at different time points postinfection. BHK-21 cells were infected with JEV (at an MOI of 0.1) and treated with various doses of NN-DNJ (0 to 100 μM) for 1 day, 2 days, or 3 days before their culture media were harvested for plaque assays. Representative data from two independent experiments are shown. D.L., detection limit (5 PFU/ml).

In vitro anti-DEN-2 effect of NN-DNJ.

We treated DEN-2-infected BHK-21 cells with NN-DNJ as described above for JEV-infected cells. Figure 3A shows that the spreading ability of DEN-2 at an MOI of 0.1 was greatly suppressed by NN-DNJ treatment, which was consistent with the results shown in Fig. 1. Furthermore, NN-DNJ readily diminished DEN-2 titers to levels much lower than those for JEV at 1, 2, or 3 days postinfection (Fig. 3B). These results clearly demonstrate that JEV and DEN-2 are both susceptible to NN-DNJ treatment.

FIG. 3.

In vitro anti-DEN-2 effects of NN-DNJ. (A) BHK-21 cells infected with DEN-2 (at an MOI of 0.1) were incubated with NN-DNJ (0 to 100 μM) for 2 days. Cells were then stained with an anti-DEN-2 NS3 MAb plus an FITC-conjugated secondary antibody. Pictures were taken using an inverted fluorescent microscope (Leica). (B) DEN-2 titers (in PFU per milliliter) produced from BHK-21 cells infected with DEN-2 (at an MOI of 0.1) for 1, 2, or 3 days in the presence of various amounts of NN-DNJ (0 to 100 μM) were determined by plaque assays. Representative data from two independent experiments are shown. D.L., detection limit (5 PFU/ml).

The antiviral effect of NN-DNJ is mediated at the post-virus adsorption step.

To investigate which steps of the virus life cycle are affected by NN-DNJ treatment, target cells were treated with varying amounts of NN-DNJ before, during, or after virus adsorption as outlined in Fig. 4A. After a 2-day incubation, virus production was determined by plaque assays. As shown in Fig. 4B and C, a significant antiviral effect occurred only when NN-DNJ was added to the target cells after, not before or during, virus adsorption. These data suggest that NN-DNJ acts mainly on the intracellular step(s) of the viral life cycle rather than on viral entry, and the infectivities of JEV and DEN-2 grown in the presence of NN-DNJ were affected.

FIG. 4.

NN-DNJ acted on the postadsorption step of the virus life cycle. (A) Outline of experimental design. NN-DNJ (0, 10, or 100 μM) was added to BHK-21 cells overnight (O/N) before virus adsorption (I), during virus adsorption for 1 h (II), or post-virus adsorption (III) for 2 days. JEV and DEN-2 were infected at an MOI of 0.1 for 2 days; then the culture supernatants were harvested for plaque assays. (B and C) Virus titers for JEV and DEN-2, respectively. Representative data from three independent experiments are shown.

NN-DNJ reduces the synthesis and secretion of viral glycoproteins.

By RIP, we further examined the effects of NN-DNJ on viral proteins. BHK-21 cells infected with JEV or DEN-2 at high MOIs were treated with 10 or 100 μM NN-DNJ for 12 h before being pulse-labeled by using ³⁵S-labeled Pro-mix for 20 min. Newly synthesized glycoproteins E and NS1 were subsequently immunoprecipitated and quantified by SDS-PAGE analysis. NN-DNJ treatment did not significantly affect intracellular levels of JEV glycoproteins E (Fig. 5A, lanes 2, 5, and 8) and NS1 (Fig. 5B, lanes 2, 5, and 8), whereas it greatly decreased intracellular protein levels of DEN-2 E (Fig. 5A, lanes 3, 6, and 9) and NS1 (Fig. 5B, lanes 3, 6, and 9) in a dose-dependent manner. The protein bands with high molecular weights seen in the JEV-infected cell lysates (Fig. 5A, lanes 2, 5, and 8) might represent the intermediate precursor of prM-E, the misfolding and slowly migrating E protein, or some coprecipitated but unidentified cellular proteins. It remains unclear why NN-DNJ exerted a stronger inhibition on viral protein synthesis for DEN-2 than for JEV; nevertheless, this observation seemed to corroborate the obvious reduction in titers of DEN-2 after NN-DNJ treatment as shown in Fig. 1 to 4. In the presence of NN-DNJ, the amounts of glycoproteins E and NS1 secreted into the culture medium were further assessed for both viruses for a 2-h labeling period (Fig. 5C and D). In contrast to the intracellular levels of JEV glycoproteins, which were only slightly affected by NN-DNJ (Fig. 5A and B), the secretion of JEV E and NS1 was greatly hindered by this treatment (Fig. 5C and D, lanes 2, 5, and 8). Likewise, extracellular levels of DEN-2 E and NS1 were also greatly reduced by NN-DNJ treatment compared to those for the untreated groups (Fig. 5C and D, lanes 3, 6, and 9). These results establish that the α-glucosidase inhibitor could affect not only the secretion of viral glycoproteins, as suggested for HBV (5), but also the intracellular levels of some viral glycoproteins, such as DEN-2 E and NS1. It also appears that the secretion of viral glycoproteins is the major step of the virus life cycle inhibited by NN-DNJ treatment.

FIG. 5.

NN-DNJ affected the levels of viral glycoproteins. BHK-21 cells that had been mock infected or infected with JEV or DEN-2 (at an MOI of 5) for 12 h in the absence or presence of NN-DNJ (10 and 100 μM) were labeled with ³⁵S-Pro-mix for 20 min before cell lysates were harvested for immunoprecipitation with JEV or DEN-2 anti-E (A) or anti-NS1 (B). Mock-infected-cell lysates were precipitated by a mixture of the MAbs for JEV and DEN-2. To study the secretion of viral glycoproteins, cells were prepared as described for panels A and B, except that ³⁵S-Pro-mix labeling was done for 2 h; then the culture supernatants were harvested for immunoprecipitation with JEV or DEN-2 anti-E (C) or anti-NS1 (D). Proteins were separated by SDS-12% PAGE and fluorographed. Lanes M, molecular weight standards.

NN-DNJ reduces the replication of DEN-2 RNA.

Detection of DEN-2 RNA and its replication by use of quantitative fluorogenic RT-PCR has recently been developed (21). We used this technique to determine whether NN-DNJ also affects flaviviral RNA replication. We conducted an experiment using a high MOI and a short term of infection to assess the direct effect of NN-DNJ on virus RNA replication during a single round of infection. As shown in Fig. 6A and B, replication of both the positive and negative strands of DEN-2 RNA was inhibited by a high dose (100 μM) of NN-DNJ. The relative Ct values, i.e., the numbers of PCR amplification cycles needed for the PCR products to reach a detection threshold, for positive-strand DEN-2 RNA (Fig. 6A) were increased from 20.10 (without NN-DNJ treatment) to 21.12 (with 10 μM NN-DNJ) and 24.11 (with 100 μM NN-DNJ). This rise in Ct values from 20.10 to 24.11 indicates that it took about 4 more cycles to reach the detection threshold in real-time PCR analysis, which means that the amount of target RNA was approximately 16-fold less after treatment with 100 μM NN-DNJ. For DEN-2 negative-strand RNA (Fig. 6B), the relative Ct values were similarly increased from 21.21 (without NN-DNJ) to 21.99 (with 10 μM NN-DNJ) and 24.85 (with 100 μM NN-DNJ), showing that 100 μM NN-DNJ could also result in a ∼12.5-fold decrease in the amount of negative-strand viral RNA. However, even in the presence of 100 μM NN-DNJ, reduction in viral RNA species levels (∼16-fold) (Fig. 6) could contribute only partly to the virus titer drop (3 log units) (Fig. 3 and 4), confirming that other targets such as the viral glycoproteins (Fig. 5) exist for NN-DNJ. Quantitative fluorogenic RT-PCR results from an infection with a low MOI and a longer incubation period are also included in Fig. 6C and D to demonstrate that NN-DNJ could indeed reduce the copy numbers of DEN-2 viral RNA during multiple rounds of infection.

FIG. 6.

NN-DNJ reduced DEN-2 viral RNA replication. BHK-21 cells infected with a high dose (MOI = 5) of DEN-2 for 13 h (A and B) or with a low dose (MOI = 0.1) of DEN-2 for 44 h (C and D) in the absence or presence of NN-DNJ (10 and 100 μM) were harvested, and their intracellular RNA levels were analyzed. Positive-sense (A and C) or negative-sense (B and D) DEN-2 viral RNA was detected by fluorogenic real-time RT-PCR as described in Materials and Methods. The Ct values shown in each panel indicate the threshold cycle, the calculated fractional cycle number at which the PCR product crosses a threshold of detection, for each reaction.

NN-DNJ affects interactions between viral glycoproteins and CNX.

The ER chaperone CNX is an ER transmembrane protein that associates transiently with newly synthesized glycoproteins which carry monoglucosylated N-glycans. We sought to determine whether flaviviral glycoproteins could interact with CNX and whether such interactions were affected by treatment with NN-DNJ. CNX is a 65-kDa protein with 573 amino acid residues (49), although in SDS-PAGE it migrates anomalously at an apparent molecular mass of about 90 kDa (19). In the absence of NN-DNJ, after a 20-min pulse-label with ³⁵S-Pro-mix, an anti-CNX antibody readily precipitated the proteins comigrating with JEV prM, E, and NS1s (Fig. 7; compare lanes 1 and 2 with lane 4). These coimmunoprecipitations were reduced by treatment with 10 and 100 μM NN-DNJ (Fig. 7, lanes 7 and 10), suggesting that binding between viral glycoproteins and CNX is dependent on glycosylation status. In addition, we observed that the patterns of binding to CNX appeared to be different for JEV prM and JEV E after NN-DNJ treatment (Fig. 7). It remains unclear why JEV E proteins were more sensitive to NN-DNJ treatment than prM in terms of association with CNX. One of the possible explanations is that the newly synthesized E or prM proteins in the ER first interact with CNX individually and that subsequently these two viral proteins form heterodimers only if they are properly folded. This possibility is partly supported by a previous DEN-1 study (10) in which heterodimer complexes were found to be unstable if formed by incompletely folded prM and E proteins. Similarly, even though the interaction was not as prominent as for JEV, protein bands comigrating with DEN prM, E, and NS1 could be detected in cells without NN-DNJ treatment (Fig. 7; compare lane 5 with lanes 12 to 14). Like those of JEV, these coimmunoprecipitated DEN glycoproteins were not detectable after NN-DNJ treatment (Fig. 7, lanes 8 and 11). In addition to CNX, we have also examined if anti-Bip or anti-CRT antibodies could bring down flaviviral glycoproteins; however, we failed to detect any interaction among them (data not shown). Whether other unidentified chaperones may play a role in the folding of DEN glycoproteins remains to be studied further.

FIG. 7.

Interactions between flaviviral glycoproteins and the ER chaperone CNX. BHK-21 cells that were mock infected or infected with JEV or DEN-2 (MOI = 5) for 12 h in the absence or presence of NN-DNJ (10 and 100 μM) were labeled with ³⁵S-Pro-mix for 20 min and then harvested by using CHAPS-HSE buffer. Cell lysates were immunoprecipitated with anti-CNX (lanes 3 to 11) or control MAbs for JEV and DEN-2 viral glycoproteins as indicated at the top of the gel. Proteins were separated by SDS-12% PAGE and then fluorographed. Molecular size standards are included in the marker lanes.

In vivo anti-JEV effect of NN-DNJ in a mouse challenge model.

We further studied the protective ability of NN-DNJ against JEV challenge in a mouse model. Figure 8 shows that daily oral delivery of a high dose of NN-DNJ (200 mg/kg/day), starting from 1 day before JEV challenge, substantially increased the survival rate to 47% (P = 0.0896 for comparison to the untreated group), compared to 7% for the untreated group and 8% for the low-dose (20-mg/kg/day) group. No sign of sublethal illness has been noticed in the surviving mice. This observation suggests that NN-DNJ is also capable of blocking flavivirus infections in vivo.

FIG. 8.

In vivo protective effects of NN-DNJ against a lethal JEV challenge. NN-DNJ (0, 20, or 200 mg/kg/day) was orally administered to 6- to 7-week-old female ICR mice, which were challenged with JEV (RP-9, 10⁵ PFU/mouse [approximately 10 50% lethal doses]) by the i.p.-plus-i.c. route as described in Materials and Methods. Animal survival was monitored daily for 21 days. The numbers of mice in each group were 14, 12, and 30 for 0-, 20-, and 200-mg/kg/day groups, respectively.

DISCUSSION

In the present study, we found that the flaviviral glycoproteins prM, E, and NS1 are transiently associated with the ER chaperone CNX and that this interaction was prevented by the iminosugar derivative NN-DNJ (Fig. 7A), which inhibits ER α-glucosidases I and II. Blocking of their interaction with CNX is likely to affect the proper folding and subsequently the secretion of the glycoproteins (Fig. 5C and D). Immunoprecipitation of newly synthesized DEN-2 glycoproteins after a 20-min pulse showed almost complete inhibition by 100 μM NN-DNJ, while only very low inhibition could be observed for intracellular JEV glycoproteins (Fig. 5A and B). The reason for this profound difference between the sensitivities of DEN-2 and JEV to NN-DNJ treatment is not clear, but this finding is consistent with the results shown in Fig. 1 to 4, demonstrating that NN-DNJ inhibited DEN-2 production to a much greater extent than JEV production. We also found that NN-DNJ blocked DEN-2 RNA replication in both positive- and negative-sense RNA synthesis (Fig. 6). The exact mechanism for such viral RNA suppression remains to be further studied. Conceivably, it could be the consequence of NN-DNJ decreasing the amounts of flaviviral NS1 glycoprotein (Fig. 5), which has been suggested to play a role in flavivirus RNA replication (28, 29, 36, 37). Alternatively, NN-DNJ may have an indirect effect on the microenvironment of the ER, the major site for flavivirus replication (40), making it no longer favorable for the accomplishment of the flavivirus life cycle.

Many inhibitors of α-glucosidase, the first enzyme involved in the trimming of the 14-residue oligosaccharide core unit on N-linked glycans, have been studied for their antiviral effects. Among these inhibitors, CST and DNJ were first demonstrated to possess anti-HIV activity (15). NB-DNJ, an alkylated derivative of DNJ, was shown to have the strongest anti-HIV activity among 47 iminosugars screened (24). In the presence of NB-DNJ, the glycosylation of gp120 was altered (23), which resulted in an impairment of viral entry at the post-CD4 binding level (12). It was further demonstrated that NB-DNJ causes a qualitative defect in gp120 molecules, thereby preventing them from undergoing conformational changes after binding with CD4 molecules (13). Likewise, NB-DNJ also exhibited anti-HBV activity in a tissue culture system (5), probably through blocking N-linked glycosylation on the viral envelope M protein (31). However, problems with achieving therapeutic serum concentrations and dealing with adverse side effects have limited the usefulness of NB-DNJ (22). NN-DNJ, a 9-carbon alkyl derivative of DNJ, was 100 to 200 times more potent than NB-DNJ in inhibiting HBV in a cell-based assay system (4). In an animal model, NN-DNJ appeared to suppress the reproduction of viruses from woodchucks chronically infected with woodchuck hepatitis virus in a dose-dependent manner (4). Furthermore, the antiviral activity of NN-DNJ against infection with BVDV, a member of the Pestivirus genus in the Flaviviridae family, was found to be 46 times more potent than that of NB-DNJ (51). The anti-BVDV capability of NB-DNJ appeared to correlate with how this compound caused misfolding of E2 envelope proteins and impairment of heterodimer formation of E1-E2 envelope glycoproteins (6). The results of the present study add JEV and DEN-2 to the growing list of viruses that are prone to be inhibited by the ER glucosidase inhibitors.

The relatively stronger potency of NN-DNJ among the iminosugars has been attributed to its longer alkyl chain, which makes it more hydrophobic, more susceptible to cellular uptake, better retained in the ER, and/or better for ER targeting (51). In a recent study (30), even at a concentration that no longer influenced the function of ER glucosidase, NN-DNJ still maintained antiviral activity against HBV through an unidentified mechanism. Furthermore, a galactose analogue of NN-DNJ, N-nonyl-deoxygalactojirimycin (NN-DGJ), although it lacks the ability to inhibit α-glucosidase, still has anti-HBV (30) and anti-BVDV (11) activities. These observations suggest that NN-DNJ may possess some other antiviral mechanism in addition to interference with glycosylation processing. Nevertheless, in agreement with the results reported for herpes simplex virus (30), we found that NN-DGJ did not inhibit either JEV or DEN-2 in our in vitro culture system (data not shown). Therefore, the antiviral effects of NN-DNJ on JEV and DEN-2 shown in the present study are more likely mediated primarily by its inhibition of the ER α-glucosidase; however, the possibility of other mechanisms besides α-glucosidase inhibition cannot be rigorously excluded.

Two α-glucosidase inhibitors, CST and DNJ, have been shown to reduce the virus production of DEN-1 by affecting the modification and dimerization of viral envelope glycoproteins prM and E (10). In the present study, we found that NN-DNJ readily blocked the virus production of JEV and DEN-2 (Fig. 1 to 3). In addition to the mechanism involving virion morphogenesis, as suggested by Courageot et al. (10), the antiviral capability of NN-DNJ against JEV and DEN-2 may involve multiple interconnected mechanisms, conceivably exerting their effects in concert. The interactions between the ER chaperone CNX and viral glycoproteins prM, E, and NS1 were inhibited by NN-DNJ (Fig. 7A), which may result in misfolding and blocking of glycoprotein secretion (Fig. 5). As a result, a few misfolded envelope proteins may become dominant enough to disrupt the proper process of viral envelope formation, thereby amplifying the effect of NN-DNJ inhibition on virus assembly and subsequently infectivity. Alternatively, this α-glucosidase inhibitor might act directly to block the biosynthesis of the flaviviral glycoprotein NS1, which has been suggested to play a role in viral RNA replication (28, 29, 36, 37). In the present study, intracellular NS1 levels were greatly reduced by NN-DNJ treatment, especially for DEN-2 (Fig. 5), whereas protein levels of nonglycosylated viral proteins, such as NS3, were not significantly affected during the same period (data not shown). Since all the flaviviral proteins are encoded by a single open reading frame, the specific reduction in viral glycoprotein levels is more likely due to a rapid degradation of misfolded glycoproteins. Accordingly, the lack of a sufficient amount of functional glycoprotein NS1 might contribute to the decline of DEN-2 RNA replication as a result of NN-DNJ treatment (Fig. 6). In addition, this antiviral effect could stem from a scenario whereby NN-DNJ inhibits other, unidentified cellular glycoproteins that directly or indirectly participate in flavivirus replication.

Carbohydrate modifications of proteins and lipids are key factors in modulating their appropriate structures and functions within cells. Utilization of inhibitors to affect glycosylation processing as a treatment for human diseases has been attempted for decades (22) and has recently attracted new attention (2, 3). Our success in protecting mice from a lethal JEV challenge by NN-DNJ (Fig. 8) and the higher in vitro sensitivity of DEN-2 than JEV to NN-DNJ warrant further study of NN-DNJ and its potential derivatives, which may in the future benefit millions of people suffering from dengue-associated diseases worldwide.