|Home | About | Journals | Submit | Contact Us | Français|
Neutralization of flaviviruses requires engagement of the virion by antibodies with a stoichiometry that exceeds a required threshold. Factors that modulate the number of antibodies bound to an individual virion when it contacts target cells impact neutralization potency. However, the contribution of cellular factors to the potency of neutralizing antibodies has not been explored systematically. Here we investigate the relationship between expression level of a viral attachment factor on cells and the neutralizing potency of antibodies. Analysis of the attachment factor DC-SIGNR on cells in neutralization studies failed to identify a correlation between DC-SIGNR expression and antibody-mediated protection. Furthermore, neutralization potency was equivalent on a novel Jurkat cell line induced to express DC-SIGNR at varying levels. Finally, blocking virus-attachment factor interactions had no impact on neutralization activity. Altogether, our studies suggest that cellular attachment factor expression is not a significant contributor to the potency of neutralizing antibodies to flaviviruses.
Flaviviruses are small RNA viruses responsible for considerable morbidity and mortality worldwide for which vaccines and therapeutics are badly needed. Members of this genus that cause severe disease in humans include West Nile virus (WNV), dengue virus (DENV), yellow fever virus, tick-borne encephalitis virus, and Japanese encephalitis virus. WNV is a mosquito-born flavivirus that circulates in an enzootic cycle between mosquitoes and birds; a variety of additional vertebrate species have been shown to represent dead-end hosts for WNV, including humans (Hayes and Gubler, 2006). WNV was first introduced into the Western Hemisphere in 1999 resulting in 59 hospitalizations with severe neurologic disease in the New York city area and seven fatalities (Marfin and Gubler, 2001). WNV has since spread across North America into Canada and Central America (Artsob et al., 2009). It is estimated that more than three million human WNV infections have occurred in the United States since its introduction (Petersen et al., 2012). The majority of human WNV infections do not cause overt disease. Clinically apparent infections range from a self-limiting febrile illness to more severe neurologic manifestations that include encephalitis, meningitis, and a polio-like paralytic syndrome (Sejvar, 2007). At present, treatment of WNV infection is limited to supportive measures. Despite significant progress, there is currently no WNV vaccine licensed for use in humans (reviewed by (Beasley, 2011)).
Antibodies are a critical component of immunologic protection from flavivirus infection (reviewed by (Dowd and Pierson, 2011)). Passive transfer of antibody has been shown to confer protection in several animal models of flavivirus infection. Monoclonal antibodies (mAbs) may also be effective when administered therapeutically (reviewed by (Diamond et al., 2012)). One such mAb, the WNV-specific neutralizing mAb E16, is currently being evaluated in phase I human trials (NCT ID: NCT00927953)(Beigel et al., 2010; Oliphant et al., 2005). Antibodies have the potential to contribute to protection from WNV via several mechanisms, including directly neutralizing virus infectivity (reviewed by (Dowd and Pierson, 2011)). This activity can be augmented in an antibody subclass-dependent manner through interactions between the complement component C1q and the constant region of the antibody heavy chain (Fc region) (Mehlhop et al., 2009). Other Fc-dependent effector functions may also contribute to protection by a variety of mechanisms, including facilitating complement deposition on infected cells and antibody-dependent cellular cytotoxicity (Garcia et al., 2006; Kurane et al., 1984; Laoprasopwattana et al., 2007).
The primary targets of neutralizing antibodies are the envelope (E) proteins incorporated into the virion (reviewed by (Dowd and Pierson, 2011; Roehrig, 2003)). The E protein is an elongated three domain structurethat orchestrates the assembly of virions and their entry into cells (reviewed by (Mukhopadhyay et al., 2005)). Flaviviruses assemble at membranes derived from the endoplasmic reticulum as immature virions that incorporate 60 heterotrimeric spikes of E proteins in complex with the premembrane protein (prM). During egress though the trans-Golgi network, pH-dependent changes in the arrangement of prM and E allow for cleavage of prM by a cellular furin-like protease (Li et al., 2008; Stadler et al., 1997; Yu et al., 2008). Cleavage of prM is required for infectivity of the virion and defines the maturation step in the virus lifecycle (Elshuber et al., 2003). E proteins on the resulting mature virus particle exist as anti-parallel dimers organized into a dense herringbone arrangement (Kuhn et al., 2002; Mukhopadhyay et al., 2003; Mukhopadhyay et al., 2005).
How antibodies neutralize virus infection has been studied extensively. In this regard, one informative perspective is to consider how the number of antibody molecules bound to the virion governs its infectivity; stoichiometric models of neutralization have been presented for several classes of viruses, including flaviviruses (reviewed by (Della-Porta and Westaway, 1978; Parren and Burton, 2001). Neutralization of flaviviruses is a “multiple-hit” phenomenon that requires engagement of the virus particle by antibody with a stoichiometry that exceeds a required threshold (reviewed by (Dowd and Pierson, 2011)). Our estimate of this threshold is ~30 antibody molecules per virion (Pierson et al., 2007). Two factors principally govern the stoichiometry of antibody engagement of the virion at any particular concentration of antibody: antibody affinity and epitope accessibility (reviewed by (Dowd and Pierson, 2011)). Both have been shown to modulate the potency of neutralizing antibodies to flaviviruses (Cherrier et al., 2009; Cockburn et al., 2012; Gromowski and Barrett, 2007; Gromowski et al., 2008; Gromowski et al., 2010; Nelson et al., 2008; Nybakken et al., 2005; Oliphant et al., 2006; Pierson et al., 2007; Stiasny et al., 2006). Because antibodies bind virions quite rapidly relative to the rate of virus attachment to cells, the number of antibodies bound to an individual virion at steady state is likely determined prior to contact with a target cell (Della-Porta and Westaway, 1978; Dowd et al., 2011; Steckbeck et al., 2005). A role for cellular factors in determining the neutralization potency of anti-flavivirus antibodies is not well established, and may not be predicted solely from this stoichiometric perspective. In this study, we explore whether differences in the expression level of a virus attachment factor modulate the neutralization potency of monoclonal and polyclonal antibodies.
The cell biology of flavivirus entry is poorly understood (reviewed by (Fernandez-Garcia et al., 2009)). While several cellular proteins have been identified with a capacity to bind flaviviruses, few have been characterized in detail and shown to be sufficient for virus attachment and entry (Anderson, 2003). The c-type lectins DC-SIGN and DC-SIGNR bind several viral pathogens and markedly increase the efficiency of viral infection (reviewed by (van Kooyk and Geijtenbeek, 2003)). These attachment proteins recognize the carbohydrate moieties of N-linked sugars arrayed on the surface of virions (Davis et al., 2006a; Feinberg et al., 2001); the specificity and affinity of these interactions is governed in part by geometrical orientation of the carbohydrate chains in space (reviewed by (Taylor and Drickamer, 2007)). Both DC-SIGN and DC-SIGNR have been characterized as attachment factors for WNV and DENV (Davis et al., 2006a; Davis et al., 2006b; Dejnirattisai et al., 2011; Navarro-Sanchez et al., 2003; Tassaneetrithep et al., 2003). While a role for DC-SIGN or DC-SIGNR in the lifecycle of WNV in vivo has not been investigated, cells expressing these molecules provide a reductionist system in which to explore how antibody-mediated neutralization of infection is modulated by the efficiency of virus-cell interactions. Furthermore, cells expressing c-type lectins are now commonly used in high-throughput assays of antibody-mediated neutralization (Balsitis et al., 2010; Boonnak et al., 2008; Nelson et al., 2008; Pierson et al., 2007; Wahala et al., 2010).
WNV reporter virus particles (RVPs) are pseudo-infectious virions that allow virus infection to be scored as a function of reporter gene expression, and have been used extensively to study virus entry and its inhibition by antibodies (Dowd et al., 2011; Mehlhop et al., 2009; Nelson et al., 2009; Pierson et al., 2006; Pierson et al., 2007). The introduction of DC-SIGNR into a cell line that is poorly permissive for WNV due to an inability to bind virions (e.g. Raji) markedly increases their permissiveness to infection (Davis et al., 2006b). To quantify the DC-SIGNR expression level required for infection, Raji-DCSIGNR cells were incubated with WNV RVPs and analyzed for virus entry and DC-SIGNR expression two days post-infection. DC-SIGNR surface expression was quantified using a standard curve prepared using Quantum™ Simply Cellular beads with a known number of antibody binding sites (Bangs Laboratories, Inc.). A comparison of the DC-SIGNR expression level of the uninfected Raji cell population to those infected by WNV RVPs (Fig. 1A) revealed that infection was strongly correlated with high expression of DC-SIGNR. In agreement, increased expression of DC-SIGNR correlates with increased susceptibility to WNV infection (Fig. 1B). Roughly 20,000 DC-SIGNR molecules/cell are required to support detectable WNV RVP infection using this system.
To investigate the impact of attachment factor expression on the potency of neutralizing anti-flavivirus antibodies, we pursued three complementary approaches. We first investigated whether differences in the expression of DC-SIGNR on target cells impact the concentration of antibody required to inhibit 50% of WNV infection (EC50). WNV RVPs were incubated with the WNV domain III lateral-ridge (DIII-LR) specific mAb E24 for one hour at 37°C and then added to Raji-DCSIGNR cells (Oliphant et al., 2005; Pierson et al., 2007); this incubation has been shown previously to be sufficient to allow for steady state binding between antibody and the virion (Dowd et al., 2011). Cells were harvested two days post-infection and analyzed for GFP and DC-SIGNR expression using flow cytometry. Analysis of the total cell population for GFP expression revealed the expected sigmoidal neutralization profile for mAb E24 characterized by an EC50 of 0.035 nM (+/- 0.004 nM, n=9)(Fig. 2A)(Pierson et al., 2007). To explore whether DC-SIGNR expression modulated the potency of E24, the data was re-analyzed by gating on cells expressing high or low levels of attachment factor as shown in Fig. 2B. No significant difference in the EC50 was detected between DC-SIGNR high and low expressing cells (n=9, p=0.739) (Fig. 2C and 2D). Similar results were obtained with the mAb E53, which binds the structurally distinct domain II fusion loop (DII-FL) epitopeand neutralizes infection by blocking attachment (data not shown, n=4, p=0.34) (Nybakken et al., 2005), as well as polyclonal antibody present in the sera of eight recipients of a candidate WNV vaccine (Fig. 2E and Fig. S1)(Martin et al., 2007). Furthermore, neither DC-SIGNR nor DC-SIGN expression level significantlymodulated the neutralizationsensitivity of DENV1 RVPs to the type-specific DIII-reactive mAb E105 (p=0.8639, n=5; p=0.4938, n=3, respectively) (Fig. 2F) (Shrestha et al., 2010). Altogether, these data do not reveal a significant impact of DC-SIGNR expression level on the neutralizing potency (EC50) of antibodies to WNV or DENV.
WNV infection of Raji-DCSIGNR cells did not significantly impact DC-SIGNR expression as compared to uninfected cells (when assayed two days post-infection; n=9, p=0.15). However, we could not rule out the possibility that subtle changes in DC-SIGNR expression might occur during the course of the neutralization assay and confound our interpretation of the experiments presented in Fig. 2. Therefore we next created a stable Jurkat cell line that expresses DC-SIGNR under the control of a tetracycline-inducible promoter. Jurkat-DC-SIGNR cells become susceptible to infection only in the presence of tetracycline (Fig. 3A); tetracycline dose-response studies revealed a large range of expression levels was achievable with saturation occurring at approximately 280,000 DCSIGNR molecules per cell. To investigate the impact of differences in DC-SIGNR expression on neutralization potency, cells were induced to express high, medium, or low levels of DC-SIGNR (Fig. 3B). Antibody dose response curves were generated on these populations using mAbs that bind epitopes on each of the three E protein domains (Fig. 3C-E) and polyclonal antibody from a recipient of a WNV DNA vaccine (Fig. 3F). In each case, neutralization activity was not markedly impacted by the level of DC-SIGNR expressed on the target cell.
To complement our analysis of neutralization potency on cells that differ with respect to DC-SIGNR expression, we next compared attachment factor distribution on cells at different points of an antibody neutralization profile. If the neutralizing potency of an antibody is truly independent of attachment factor expression on the target cell, one would expect the distribution of DC-SIGNR expression to be equivalent among populations infected by WNV RVPs with or without antibody present. In contrast, if antibody more effectively prevents infection of cells with low levels of attachment factor expression, the distribution of DC-SIGNR expression on cells infected in the presence of antibody would be skewed towards a higher average number of DC-SIGNR molecules/cell as compared to cells infected without antibody present. This type of comparison can be made quantitatively using previously described chi-squared probability binning analysis described in detail in the materials and methods section (Roederer et al., 2001).
WNV RVPs were incubated with E24 at the EC50 (Fig. 4A, Box 2) or a non-neutralizing antibody concentration (Fig. 4A, Box 1) and added to Raji-DCSIGNR cells. On day two post-infection, cells were harvested and analyzed for GFP and DC-SIGNR expression. Histograms that describe the distribution of DC-SIGNR expression on cells infected (GFP+) in the presence of neutralizing and non-neutralizing concentrations of E24 were compared; no difference was observed (Fig. 4B). As a positive control for this method of analysis, we repeated this experiment using K562-DCSIGNR cells that express an activating Fc-receptor that provides a DC-SIGNR-independent entry pathway in the presence of antibody (Davis et al., 2006b; Pierson et al., 2007). Antibody dependent enhancement (ADE) of infection describes a marked increase in the efficiency of virus infection in the presence of antibody that is mediated by Fc-receptors (Fig. 4C) (reviewed by (Halstead, 2003)). Previous studies using cell lines and primary dendritic cells revealed the magnitude of ADE correlates inversely with the expression of viral attachment factors on target cells (Boonnak et al., 2008; Pierson et al., 2007). Analysis of the DC-SIGNR expression profile of K562-DCSIGNR cells infected in the presence of enhancing concentrations of antibody (Fig. 4C, Box 2) revealed a large shift towards lower DC-SIGNR expression (Fig. 4D), reflecting the susceptibility to infection of cells expressing low levels of DC-SIGNR due to Fc-receptor facilitated ADE.
Since this method was effective in analyzing the role of DC-SIGNR in neutralization, the data from each independent experiment described in Fig. 2 was reanalyzed using this method. At several points on each neutralization curve, the DC-SIGNR expression of infected cells in each sample was compared to the infected cells in the no-antibody control using chi-squared probability binning analysis. No significant differences were observed for WNV RVPs and E24 (n=25), E53 (n=33), or vaccine candidate recipient serum (n=14) on Raji-DCSIGNR cells, or for DENV1 RVPs and E105 (n=28) at any concentration tested in the experiment.
As a final approach, we studied the neutralizing potency of mAb E24 under conditions where the number of DC-SIGNR molecules available for WNV attachment was reduced using three different inhibitors. The mAb 120604 binds the carbohydrate recognition domain region of DC-SIGNR and has been shown to block interactions with WNV (Davis et al., 2006b). Incubation of Raji-DCSIGNR cells with mAb 120604 resulted in a dose-dependent inhibition of WNV RVP infection (EC50=2.3×10-9 M, n=2). Likewise, pre-treatment of Raji-DCSIGNR cells with glucose or mannose blocks WNV RVP infection by competing for binding to the carbohydrate recognition domain (EC50=8.1×10-3 M, n=2 and EC50=3.5×10-3M, n=3, respectively). Raji-DCSIGNR cells were incubated with each of these inhibitors at a concentration that corresponds roughly to the EC50, followed by use in neutralization studies with mAb E24. As a control, cells were also incubated in the presence of each inhibitor at a level shown in our studies not to block WNV RVP entry. A comparison of the potency of E24 in the presence of inhibitory and non-inhibitory concentrations of mAb 120604, glucose, or mannose failed to identify an impact of treatments that functionally reduce the number of DC-SIGNR on Raji cells (Fig. 5).
The efficiency of antibody-mediated neutralization has been shown in several contexts to vary when assayed using different cellular substrates (Kjellén, 1985; Outlaw et al., 1990). For example, studies with a panel of mAbs to La Crosse virus show marked differences in the protective capacity of different antibodies based upon the cell type used in the assay (Grady and Kinch, 1985). In addition, WNV-reactive antibodies have been described that neutralize infection in a cell type-dependent manner via unknown mechanisms (Oliphant et al., 2006). Several studies with influenza A suggest neutralization potency is modulated by the efficiency of virus attachment to cells (Hensley et al., 2009; Yewdell et al., 1986). This raises the intriguing possibility that the neutralization activity of anti-flavivirus antibodies might be impacted by factors that control the efficiency of virus infection of target cells. Evaluating this possibility is challenging in light of an incomplete understanding of factors involved in the attachment and entry of these viruses into cells. To date, only a small number of cellular factors that modulate the efficiency of flavivirus attachment and entry into cells have been characterized; none of these have been shown to be broadly necessary and sufficient for infection among permissive cell types (Anderson, 2003).
DC-SIGN and DC-SIGNR are related c-type lectins which directly bind virus particles and promote more efficient infection (Davis et al., 2006a). In this study, we investigated how changes in expression of these attachment factors (which in turn control the efficiency of virus binding and infection) modulate antibody-mediated neutralization. Using several complementary approaches and two different cell types, we found that changes in the expression of DC-SIGN or DC-SIGNR do not modulate detectable changes in the neutralizing activities of flavivirus-reactive antibodies. These results suggest the fate of the virion is determined by the number of antibodies bound to the virus particle before it contacts the target cell. Thus, changes in the expression of cellular receptors are not likely to be mechanistically responsible for the cell-type dependent neutralization observed for some viruses.
The studies presented within were limited to cell types upon which the cellular attachment structure was defined and could be manipulated; the cellular factors involved in attachment to many cell substrates traditionally used in flavivirus neutralization studies has not yet been established. The functional properties of flavivirus reactive antibodies on cells expressing both DC-SIGNR (or DC-SIGN) and an Fc-receptor reflect an overlay of neutralizing and enhancing activities in a manner which may be quite complex (Boonnak et al., 2008; Pierson et al., 2007). While extending our studies to primary cell types that are targets of virus infection in vivo (e.g. DC-SIGN+ dendritic cells) (Tassaneetrithep et al., 2003; Wu et al., 2000) would have been interesting, these experiments are complicated by the presence of Fc-receptors capable of enhancing viral infection, particularly when low levels of attachment factor are present. As the requirements for flavivirus entry are better understood, the contribution of distinct cellular factors to the effect of antibodies will undoubtedly become more clear.
K562 cells were grown in RPMI-1640 medium (Invitrogen) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (HyClone) and 50 U/ml penicillin-streptomycin (PS)(complete RPMI). Media used for the culture of a K562 line that stably expresses DC-SIGNR was supplemented with 10 μg/ml blasticidin (Invitrogen). Raji and Raji-DCSIGNR cell lines were cultured in complete RPMI. The inducible Jurkat-DCSIGNR cell line was maintained in RPMI supplemented with 7% FBS, 50 U/ml PS, 5 μg/ml blasticidin, and 1000 μg/ml G418. HEK-293T cells were passaged in Dulbecco’s modified Eagle medium (DMEM)(Invitrogen) supplemented with 10%FBS and 50 U/ml PS). All cell lines were maintained at 37°C in the presence of 7% CO2.
Serum from recipients of a candidate WNV DNA vaccine was obtained for use in neutralization studies. A phase I single-site, open-label clinical study (NCT ID: NCT00106769) to evaluate the safety and tolerability of a recombinant nucleic acid vaccine has been described (Martin et al., 2007). The protocol and conduct of the clinical research adhered to the experimental guidelines of the US Department of Health and Human Services and was approved by the NIAID Intramural Institutional Review Board. Sera from eight subjects was collected at 12weeks post-final vaccination and studied for neutralizing activity.
A Jurkat T cell line was constructed to express DC-SIGNR under the control of a tetracycline inducible CMV promoter using the T-REx system (Invitrogen). Briefly, the coding sequence of human DC-SIGNR was cloned into the pT-REx-DEST30 vector by recombination using a previously described pDONR221 entry clone (Davis et al., 2006b). The resulting expression clone, pDC-SIGNR-EXP30, was electroporated into 1.5 × 106 T-REx Jurkat cells in 160 μl of Buffer SF (divided among eight wells) using the Amaxa 96-well shuttle nucleofection system (Lonza) and program CM-150. Transfected cells were selected by repeated passage in complete RPMI containing 5 μg/ml blasticidin and 1000 μg/ml geneticin (Invitrogen). Individual clones were isolated by limiting dilution.
Pseudo-infectious reporter virus particles (RVP) were produced by transfection of HEK-293T cells with DNA plasmids encoding the structural genes and a WNV sub-genomic replicon as described (Pierson et al., 2006; Pierson et al., 2007). Briefly, a mixture (1:3 by mass) of two plasmids encoding a GFP-expressing WNV replicon (pWNIIrepGZ) and the WNVI (NY99) or DENV1 (Western Pacific-74) structural genes were transfected into HEK-293T cells using Lipofectamine LTX (Invitrogen) according to the manufacturer’s specifications. The transfection media was replaced after four hours with a low glucose formulation of DMEM supplemented with 10% FBS and 50 U/ml PS. RVPs were collected at 48 hours, aliquoted, and stored at -80°C until use.
Neutralization of WNV and DENV RVPs was performed as previously described (Dowd et al., 2011; Nelson et al., 2008; Pierson et al., 2007). Briefly, RVPs were incubated with serial dilutions of antibody or human serum for one hour at 37°C and then added to cells (total infection volume of 300ul in each well of a 96-well plate). Infected cells were incubated for 48 hours at 37°C in 7% CO2. GFP-expressing infected cells were fixed with 2% paraformaldehyde (PFA) and then enumerated using a FACSCalibur flow cytometer (BD Biosciences). Neutralization potency (EC50) was estimated by a least squared minimization non-linear regression analysis using Prism software (GraphPad).
The expression of DC-SIGNR was measured by flow cytometry using phycoerythrin-conjugated mAbs (R&D Systems clone 120604). Cells were stained, washed once in cold Dulbecco’s phosphate-buffered saline (PBS), and fixed with 2% PFA in PBS. The number of DC-SIGNR molecules expressed on each cell was estimated by comparison to a standard curve generated using Quantum Simply Cellular beads that have a known number of antibody binding sites (Bangs Laboratories, Inc.) processed in parallel. Use of this standard curve assumes each DC-SIGNR molecule is engaged by a single antibody molecule. Beads and cells were collected using a FACSCaliber cytometer (BD Biosciences).
DC-SIGNR expression profiles of infectedcells (GFP positive cells) were compared between samples where antibody was present or absent during infection using the probability chi-squared binning analysis tool in the FlowJo software package (Tree Star, Inc.). This allows a comparison of two cell populations of different sizes without biasing towards effects at the tails of the distribution. Briefly, the control population (infected cells without antibody present) was divided into approximately 300 gates with roughly equal numbers of cells per gate. The total number of cells in each gate was divided by the total number of cells in the sample, and the resulting metric was recorded as the value for that gate. The same gates used in the control population were then applied to the experimental population (infected in the presence of antibody), and the process was repeated. For each gate, the two population test metrics were compared by traditional chi-squared analysis. The resulting statistic T(χ) is analogous to a t-score, and can be used to find the probability that two populations could be drawn from the same sample.
Neutralization profiles were generated by pre-incubating WNV RVPs with serial dilutions of serum from seven additional vaccine trial volunteers, followed by infection of Raji-DCSIGNR cells. Results are normalized to the infectivity obtained in the absence of serum and are representative of two independent experiments. Error bars show the standard error of the mean of triplicate wells.
We would like to thank Dr. Michael S. Diamond for providing all the antibodies used in this study and for stimulating discussions, and Drs. Barney S. Graham and Heather D. Hickman for critical comments on the manuscript. This study was funded by the intramural research program of the National Institute of Allergy and Infectious Diseases and the NIH Office of AIDS Research. The funding sources had no role in study design; collection, analysis, or interpretation of the data; writing the report; or the decision to publish.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.