Monoclonal IgM antibodies mediate potent complement neutralization by targeting cell-derived epitopes on enveloped viruses

ABSTRACT

When studying virus neutralization by specific antibodies of the adaptive immune response, serum heat treatment is a standard procedure done to eliminate any non-specific effects by the complement system. Although non-specific, innate immune responses such as these may be relevant in determining infection outcome. We observed extensive variation between negative control donors in their ability to neutralize Ebola virus and other enveloped viruses in the presence of complement. The effect was mediated by the classical pathway of complement activation, with serum IgM the main initiator in a manner that depended on the cell line that produced the virus. To identify the epitope, three monoclonal IgM antibodies were isolated after a neutralization screen and probed against arrayed glycans. The antibodies were found to bind ganglioside GM2, and the test virus was rendered refractory to anti-GM2 neutralization when propagated in the presence of an inhibitor of ganglioside synthesis. Gangliosides and other host membrane moieties are known to be carried by viruses and to aid entry to cells. The findings here suggest that under specific circumstances, such moieties incorporated in lipid envelopes of viruses can be recognized by potently neutralizing IgM antibodies present in the serum of immunologically naïve individuals.

IMPORTANCE

Neutralization, that is, the ability to block infection in cell culture, remains the primary functional attribute of therapeutic antibodies and humoral immune responses against viruses. Beyond direct effects—such as preventing virus attachment to the cell—virus inactivation by antibody-mediated complement deposition or lysis is often considered a type of neutralization. Viral envelope glycoproteins are the main targets of neutralizing antibodies. However, because viruses obtain their membranes from cells in which they replicate, the membranes are known to contain cell-derived markers in addition to virally encoded proteins. Having observed marked individual variance in naïve serum donors’ ability to neutralize Ebola virus and other enveloped viruses, we sought to understand the mechanism. Three IgM monoclonals targeting a cell-derived epitope were isolated. The findings expand on the importance of cellular factors exposed on the virion membrane and suggest that they may be targets of potent antibody-mediated immune monitoring.

INTRODUCTION

A major question in the field of infectious diseases is why the outcome of an infection with a certain organism can vary from severe to benign between individuals. In general, dose and route of infection are important factors that can impact disease severity and progression. Beyond the circumstances of a particular transmission event, variability in the microbe or the host arising from genetic or acquired traits can also influence the outcome. Adaptive immunity represents an obvious acquired host trait, and an emerging body of evidence suggests that genetic deficits in innate immunity signaling and lack of virus receptors in some individuals can predispose to severe viral infections or be protective, respectively (1). Regarding the pathogen, it is known that genetic traits can influence virulence, with less emphasis having been placed on non-genetic traits.

Virus neutralization assays are commonly conducted to assess the functionality and potency of immune sera and monoclonal antibodies. A virus of interest is incubated with the test sample, and the resulting reduction in infectivity of susceptible cells is quantified. Antibodies can mediate neutralization by different mechanisms, including induction of conformational changes to the attachment glycoprotein, steric blocking of virus–receptor interactions, inhibition of a post-attachment stage in entry (membrane fusion), or by aggregating virions (2). A reaction that requires the complement (a plasma protein cascade capable of opsonization, membrane permeabilization, and signaling [3]) or other mechanisms giving a similar result in a classic neutralization assay (such as inhibition of cell-to-cell spread, egress, or proteasome targeting) are often considered neutralization, though this depends on definition (2).

We have observed frequent non-specific neutralization against Ebola virus (EBOV) and other enveloped viruses when the standard heat treatment of naïve donor serum is omitted. Interestingly, the potency of this phenomenon varies greatly between individuals, raising the question of whether it could contribute to disease outcomes as a genetic or acquired host trait. To explore the nature of the phenomenon, evaluate therapeutic potential, and to assist in interpreting neutralization results, we sought to (i) isolate the non-specific neutralization trigger; (ii) identify the virion component targeted; (iii) determine whether neutralization by naïve donors is a transient response; and (iv) investigate what determines the neutralization potency differences observed between individual donors. Three IgM mAbs that target a cell-derived epitope and mediate potent neutralization via complement were identified. The results highlight the possible role of host-derived epitopes as acquired traits of a virus, subject to immune recognition.

RESULTS

Potency of non-specific neutralization of enveloped viruses varies substantially between donors but is stable over time

Sporadic neutralization of EBOV expressing the green fluorescent protein (EBOV-GFP) was noted in our earlier work when testing serum samples from presumed naïve donors. To assess the frequency of the phenomenon, samples were sourced from a population assumed to be negative for specific EBOV antibodies (normal donors from the United States). Samples were heated or left unheated and assessed for neutralization potency. None of the heated sera displayed neutralizing activity. In contrast, a minority of samples were able to neutralize >80% of infectivity when heating was omitted, and in some cases, the titers were substantial (up to 1/85) (Fig. 1A). This neutralizing activity extended to other enveloped viruses as well, and titers against Sudan virus (SUDV; another orthoebolavirus) and Lujo virus (LUJV; a mammarenavirus) correlated with titers that were observed against EBOV-GFP (however, the sera appeared more potent against the orthoebolaviruses than against LUJV) (Fig. 1B). To assess whether the non-specific neutralization extended to non-enveloped viruses that the donors were expected to be seronegative against, Wad Medani virus (genus Orbivirus; a virus isolated from ticks collected in India [4]) neutralization was attempted with heated and unheated sera. None of the tested sera had activity against this virus regardless of whether the sera were heat inactivated or not (Fig. 1C). To assess whether the neutralization titers were transient or stable over time, rebleed samples from three donors with activity against EBOV-GFP were tested. In all cases, the titers were found to be stable for at least 5.5–9.5 months (Fig. 1D).

Fig 1.

Identifying the cause of non-specific neutralization. (A) Neutralization of EBOV-GFP with heated or not heated human serum samples (n = 58). Mean and standard deviation from 1 to 3 repeat experiments/serum are shown. (B) Comparing EBOV-GFP neutralization to SUDV-ZSG and LUJV-ZSG neutralization with a subset of the samples in panel A. Means of 2–3 repeat experiments are shown (n = 20). (C) Neutralization of Wad Medani reovirus with the serum samples from panel B. Mean data from two repeat experiments are shown. (D) Stability of neutralization titer in three individuals. Paired serum samples for three donors, with the samples in teal drawn 5.5–9.5 months after the samples in purple. Mean and range of 2–3 repeats. (E) A serum sample without intrinsic neutralization activity can rescue the activity of a heated sample. Three sera (Neut.X, Y, and Z) were used to neutralize EBOV-GFP either without heat treatment, with heat treatment, or mixed with a non-neutralizing serum (Non.x, y, or z) that had no neutralizing activity of its own. Means and standard deviations from three repeats are shown. (F) Dependency of non-specific neutralization on complement cascade factors. Unheated sera were mixed with mAbs (final 20 µg/mL) or AMY-101 acetate (final 5 µM) or vehicle controls prior to adding EBOV-GFP in a neutralization format. Normalized infectivity from three experimental repeats (mean and standard deviation) with three neutralizing sera and one non-neutralizing serum is shown. (G) EBOV-GFP neutralization curves with selected sera after resin depletion of IgG, IgM, or IgA antibodies. (H) Correlation of EBOV-GFP neutralization titers with IgM concentration. Pearson r and two-tailed P-value shown (n = 20). (I) The observed FRNT₅₀ titers grouped by donor IGHM alleles. (J) Serum IgM titers grouped by donor IGHM alleles. FRNT₅₀₍₈₀₎, focus-reduction neutralization titer 50 (80) %; the titer reducing focus counts by either 50% or 80% (for numerical comparisons, the mid-curve value FRNT₅₀ is used). SUDV-ZSG, Sudan virus with ZsGreen reporter gene. LUJV, Lujo virus.

Non-specific neutralization is mediated primarily by IgM antibodies and complement

Next, experiments were carried out to identify the mediators of this broad neutralization of enveloped viruses. As a first step, we sought to understand whether a heat-labile element in the neutralizing sera was the sole neutralizing factor or if multiple factors were involved. Three sera that had been identified as neutralizers were confirmed to lose their activity upon heating (Fig. 1E). Three other sera were confirmed not to have activity, even when left unheated. When heated neutralizers and unheated sera with no activity of their own were mixed, neutralizing activity could be partially restored. The results suggested that a heat-labile and a heat-stable component were both required, that the heat-labile component was generally present in the donors, and that the heat-stable component was the factor determining the ability to neutralize enveloped viruses.

The archetypal serum system that contains heat-labile elements and can mediate virus neutralization is the complement system, and enveloped viruses have long been considered susceptible to complement-mediated inactivation in a manner that can depend on the cells that produced the virus (5). The complement system can be activated by three different pathways: the classical pathway is triggered by the constant portion of antibodies binding antigen; the lectin pathway by soluble pattern-recognition molecules binding carbohydrates; and the alternative pathway by spontaneous hydrolysis of the downstream factor complement component 3 (C3) (3). To assess whether the neutralization was indeed mediated by the complement, and to narrow down the pathway and potential triggers involved, specific inhibitors of classical pathway upstream protease C1s, lectin pathway protease MASP-2, and downstream factors C3 and C5 were tested. All three sera lost their neutralization activity in the presence of antibodies inhibiting C1s or C5, while the antibody inhibiting MASP-2 had no effect. Surprisingly, a peptide inhibitor of C3 (a factor between C1s and C5 in the complement cascade) was able to reverse neutralization only partially, though the effect was statistically significant (paired, two-tailed t-test of neutralizer AMY-101 signals vs. DMSO signals: P = 0.009) (Fig. 1F). In contrast to results from other systems that identified the lectin pathway targeting EBOV (6, 7), the results here were consistent with classical pathway involvement.

In some cases—such as with retroviruses—the classical complement pathway can be activated directly by C1q, the first factor of the pathway, bypassing the need for a binding antibody (8, 9). To probe the need for an antibody and to identify the isotype activating complement in the current scenario, IgG, IgM, and IgA antibodies were independently depleted from the sera. In all cases, IgM depletion resulted in complete loss of neutralizing activity, while IgG depletion reduced the potency of the sera without abrogating activity. IgA depletion had little effect (Fig. 1G). The reciprocal experiment with purified IgG, IgM, and IgA fractions was consistent with IgM having a greater activating potential than IgG and IgA (Fig. S1).

Having identified IgM as the primary initiator of the observed neutralization by donors not exposed to the virus, attempts were made to understand individual differences in the IgM compartment. Neutralization potency was not statistically correlated with IgM concentration in the samples (Fig. 1H), implying that the difference between individual donors was either in IgM specificity or ability to activate the complement. To investigate genetic differences in the ability of IgM antibodies to activate the complement, the immunoglobulin heavy constant mu gene locus (IGHM) was subjected to targeted sequencing. Among 20 donors, 5 were homozygous for IGHM*03 (although in some individuals the two genomic copies displayed minor non-coding variation, both grouping under allele *03), 3 were heterozygous IGHM*03/*06, 11 were homozygous for IGHM*06, and 1 was homozygous for IGHM*04. Between these three alleles, the only non-synonymous difference is at UniProt P01871 position 191 (SDWLG in allele *04, S in *03 and *06). No apparent differences in neutralization titers (or IgM concentrations) were identified based on allelic status (Fig. 1I and J).

Isolating IgM mAbs targeting the viral envelope

In aggregate, the above results suggested that some individuals had IgM antibodies that could engage unidentified epitopes to neutralize enveloped viruses by activating the complement system. As the first step to identify this epitope, monoclonal antibodies were isolated from B cells of the EBOV-naïve donor with the highest identified EBOV-GFP neutralization titer. The screen was designed to target IgM-secreting cells with neutralization read-out (Fig. 2A). The primary screen identified nine wells with neutralizing activity (Fig. 2B and C). Three of the nine hit wells were retained after repeat testing, with no apparent toxicity against the screening cells observed (Fig. 2D). Hit well RNA was sequenced using a targeted approach, and the reads assembled to identify the IGHM, IGK, and IGL (immunoglobulin kappa and lambda) contigs with most reads (Fig. 2E). Expressing the most abundant IGHM sequences of each well with respective IGK and IGL sequences and the J-chain identified in each case a single combination that neutralized the virus in the presence of complement, without apparent toxicity against the cells (Fig. 2F). The sequences were closely related, with CDR (complementarity-determining region) 1, 2, and 3 translations matching in some cases. However, all sequences were unique owing to differences in other regions (Fig. 2E; File S1 and S2). IMGT/V-QUEST analysis found IGHV3-30*02 or IGHV3-30-5*02 as the most likely heavy chain V-genes used (94.1% identity for 9-K21 and 14-H15, 94.4% for 14-E14), IGHJ3*02 as the most-likely J-gene (88% for all three mAbs), and IGHD5-24*01 in reading frame 3 as the hit D-gene (all three mAbs). Regarding the light chain, IGLV10-54*04 was the most likely V-gene (96.1%, 96.5%, and 97.9% identity for 9-K21, 14-E14, and 14-H15, respectively, with 14-E14 also matching IGLV10-54*01 at the same percent). For all three mAbs, the hit lambda J-gene was IGLJ3*02, with 97.3%, 89.2%, and 91.9% identities, respectively. Having found non-synonymous differences in the sequences, all three hit antibodies were expressed at a larger scale and purified by size-exclusion. (For these expressions, the light chain vector was mutated to lambda 3 to match the sequencing results indicating that this gene was used by the hit B cells.) Gel analysis of the purified preparations was consistent with multimeric IgM antibodies (Fig. S2).

Fig 2.

Isolating IgM mAbs that mediate complement-mediated neutralization in an immunologically naïve individual. (A) Schematic outline of the EBOV-GFP neutralization screening strategy. Graph created with BioRender.com. (B) Normalized infectivity data from the screen with the preliminary hit wells (>5 standard deviations below the mean normalized signal) in color. (C) Example images from the screen depicting the strongest hit (9-K21) and a control well. (D) Data from the hit verification round. Raw supernatants of the nine hit wells were tested for neutralization activity at seven dilutions and for toxicity at one dilution. The heat map represents infectivity (viability) normalized against control wells that received no B-cell supernatant. (E) Summary of sequencing results from the hit wells. For each hit well (9-K21, 14-E14, and 14-H15), the CDR sequences of the top contigs (by number of reads) are written out. The neutralizing combinations are depicted by a gray background. The colors of the vertical bars represent the relative read abundancies. (F) Neutralization experiment identifying the functional heavy and light chain combinations. Each kappa and lambda light chain was combined with each heavy chain of a hit well, and the IgM was expressed in small scale. The raw supernatants were tested for neutralization activity against EBOV-GFP at three dilutions and toxicity against Vero-E6 cells.

The mAbs are potent neutralizers in the presence of complement

Experiments with EBOV-GFP showed that complement was strictly required for neutralization by these mAbs (Fig. 3A), matching the findings with the polyclonal serum. Since the polyclonal sera were found active against unrelated enveloped (but not non-enveloped) viruses (Fig. 1B and C), the epitope recognized was hypothesized to be derived from the cell in the form of a post-translational modification or another moiety exposed on the virion membrane. To test this hypothesis, EBOV-GFP was grown on primary-like, terminally differentiated HepaRG cells that also support hepatitis B virus infection (10). After inoculation at a low multiplicity, progressive EBOV infection of the monolayer was observed, and harvest at day 4, when nearly all cells were positive for the fluorescent reporter, resulted in a stock with a Vero-E6 titer of 4 × 10⁵ FFU/mL. This stock was neutralized by the broadly neutralizing antibody ADI-15878 (11) at a similar titer as the stock grown on Huh7 cells. In contrast, neutralization by the isolated IgM mAbs was absolutely dependent on the cell line that produced the virus (Fig. 3B). Similarly, refractory EBOV infectivity was observed in an archived rhesus macaque tissue sample and a liver sample from a mouse infected with EBOV adapted to that species (12) (Fig. S3). Observing neutralization that depends on the source of the virus suggested that the mAbs target the virus particle rather than a virus receptor on the cells, which might yield the same experimental result. When activity was present, the IgM mAbs were highly potent (average 9-K21 IC₅₀ 0.3 ng/mL, 14-E14 and 14-H15 both 1.5 ng/mL, n = 8–9).

Fig 3.

Identifying the epitope mediating non-specific neutralization. (A) Neutralization curves of EBOV-GFP from Huh7 cells in the presence and absence of human complement. (B) Neutralization of Huh7-derived virus vs. HepaRG-derived virus in the presence of complement. (C) Binding of the hit mAbs to glycan epitopes. Glycan array with triplicate antigen spots was probed with the hit antibodies or a negative control antibody. Median fluorescence signals after correction against median signals from the negative control are presented. (D) ELISA against purified ganglioside antigens GM1 and GM2. (E) Influence of ganglioside synthesis inhibition on hit IgM mAb binding to Huh7 cells. Cells were treated with 200 µM miglustat or DMSO as a vehicle control for 4 days. Immunofluorescence analysis was then performed using the hit mAbs or an isotype control as the primary antibodies. (F) Impact of ganglioside synthesis inhibition on virus susceptibility to hit mAb neutralization. After 4 days of pre-treatment with miglustat or vehicle control, EBOV-GFP was passaged on Huh7 cells for 4 days in the presence of the same. Neutralization curves of the resulting virus stocks are presented. (G) Neutralization of virus grown as in F by unheated serum of the B-cell donor. (H) Neutralization of virus grown as in F by the unheated serum panel of Fig. 1A.

The mAbs target ganglioside GM2

Since the epitope of the IgM mAbs appeared to be cell-derived instead of a virus-encoded protein, the mAbs were next tested for their ability to bind glycan antigens. The most prominent hit among 300 arrayed glycans was ganglioside GM2 [GalNAc-β-1,4-(Neu5Ac-α-2,3)-Gal-β-1,4-Glc-β-] (Fig. 3C). Notably, other gangliosides also present on the slide were not reactive. These included the one galactose longer GM1a [Gal-β-1,3-GalNAc-β-1,4-(Neu5Ac-α-2,3)-Gal-β-1,4-Glc-β-], the GalNAc shorter GM3 (Neu5Ac-α-2,3-Gal-β-1,4-Glc-β-), and the one sialic acid longer GD2 [GalNAc-β-1,4-(Neu5Ac-α-2,8-Neu5Ac-α-2,3)-Gal-β-1,4-Glc-β-], as well as GD3 and GT3 that lack the GalNAc but have one or two residue longer sialic acid chains, respectively.

ELISA against purified GM1 and GM2 gangliosides confirmed that the isolated IgM mAbs all bound GM2, while no binding was observed with GM1 antigen (Fig. 3D). To corroborate the role of gangliosides as neutralization epitopes, other antibodies that target gangliosides were tested. A previously reported human IgM mAb against GM2 (13, 14) was found to be non-neutralizing (Fig. S4). However, testing a panel of mouse ganglioside mAbs (15–17) found neutralization by both IgM clones that target GM2, but not by clones that target other gangliosides. These GM2-targeting mAbs reproduced the finding that neutralization depended on the cell line that produced the virus (Fig. S5). Attempts to detect GM2-binding IgM directly in the polyclonal sera were unsuccessful, possibly due to the potent neutralizing antibodies being present at concentrations below the detection limits of the ELISA (Fig. S6).

Next, we tested whether the mAbs could neutralize EBOV-GFP when the envelope was devoid of gangliosides. When Huh7 cells were grown in the presence of miglustat, an inhibitor of ganglioside synthesis (18), immunofluorescence staining by the IgM mAbs was diminished (with 9-K21 showing more residual staining as compared to 14-E14 and 14-H15) (Fig. 3E). Next, EBOV-GFP was passaged with miglustat or vehicle control added to cells as pre-treatment as well as during the virus infection. The resulting stock titers were not markedly influenced by the drug treatment (8.1 × 10⁶ vs. 9.6 × 10⁶ FFU/mL with vehicle control as the average of 3 stocks), nor was there apparent toxicity to the cells associated with the drug treatment. The virus grown in the presence of miglustat was found to be completely resistant to neutralization by the IgM mAbs but remained sensitive to neutralization by the control IgG mAb that binds the viral glycoprotein (Fig. 3F). Finally, we investigated whether the IgM mAbs recapitulated neutralization by the polyclonal serum of the B-cell donor. Like the mAbs, the serum failed to neutralize a virus stock grown in the presence of miglustat (Fig. 3G). Testing the serum panel of Fig. 1A showed that the effect extended to all donors whose unheated serum had shown non-specific neutralization of Huh7-grown virus (Fig. 3H).

DISCUSSION

This work investigated the origin of broad non-specific neutralization of enveloped viruses in vitro, the potency of which greatly varied between donors, and found that it was mediated primarily by IgM antibodies triggering the classical complement pathway. Since the effect was dependent on the cell line used for growing the virus, careful choice of producer cell line and determination of background positivity among the test population are recommended when conducting neutralization experiments in the presence of complement. Indeed, it has been suggested that serologic relatedness of viruses could be misidentified due to apparent cross-reactivity against cell-derived antigenic determinants (5). While the results here pertain to non-virus-specific antibodies, complement has been reported to potentiate neutralization of EBOV by some donor sera when specific antibodies are present (19), although the difference is not consistently observed (20).

Three IgM mAbs mediating complement-dependent neutralization were isolated from the donor with the highest serum titer to investigate which virion component they recognized. Glycan array screening and follow-up experiments identified ganglioside GM2 or related structures as an epitope targeted by the mAbs 9-K21, 14-E14, and 14-H15. Gangliosides (sialic acid-containing glycosphingolipids) are ubiquitously expressed in various tissues, though particularly abundant in the brain. Of note, genetic inactivation of an enzyme in the sialic acid synthesis pathway (21, 22) means that gangliosides of humans and certain other species (23) contain N-acetyl- instead of N-glycolylneuraminic acid, adding a layer of complexity to the recognition of these structures. Finding a ganglioside antigen on the EBOV virion is in itself consistent with previous work establishing that EBOV buds from cells at lipid rafts and that the virions carry ganglioside GM1 derived from the cell membrane on them (24). More than mere incidental carry-over from the cell, these gangliosides are thought to have a functional role in the virus replication cycle. The ganglioside receptor Siglec-1/CD169 mediates EBOV entry into activated dendritic cells, and antibodies targeting Siglec-1/CD169 have been proposed as cross-protective antivirals against filoviruses and beyond (25).

GM2 has been identified as a tumor-associated antigen (26), which may explain its cell-type-specific incorporation in our system. The cell line dependency suggests these mAbs might not neutralize orthoebolaviruses to the same extent in vivo. However, targeting GM2 has been investigated in relation to other viruses. A human IgM antibody against GM2 was found to lyse both HIV-1 particles and infected cells (13) and to limit HIV-1 release by primary cells isolated from infected individuals (14). While we found activity in both mouse IgM mAbs that bound GM2, the above-mentioned human antibody was not active in our system. This may be due to potency differences. The previously published results were obtained with µg/mL concentrations, while the antibodies presented here were active in the ng/mL range. Hypothetically, antibodies targeting non-ubiquitous membrane epitopes might be most relevant against viruses with narrow cell tropism (resulting in a uniform composition of host-derived membrane markers in vivo). Against such viruses, membrane-targeting mAbs might achieve broad-spectrum activity combined with a high resistance barrier. Clearly, investigating this avenue requires representative in vitro models, or better still, ex vivo systems in which infectious virus can be studied directly in clinical material. Beyond direct neutralization, complement activation is known to enhance adaptive immune responses (27), and IgM antibodies targeting cell-derived epitopes on virions could plausibly act as pattern-recognition receptors to bridge innate and adaptive immune responses. It is noteworthy that the mAbs triggered complement-mediated neutralization at concentrations exceeding the potency of the control IgG mAb that binds the viral glycoprotein. Comparing the neutralization and ELISA curves implies that when investigating IgM antibodies such as the ones identified here, functional assays may be more sensitive as the starting point than binding assays. Indeed, we found our ganglioside ELISA inadequately sensitive to directly detect IgM antibodies in the serum of the donor whose B cells yielded the mAbs.

Complement genes are known to be polymorphic in a manner that may influence susceptibility to infectious and non-infectious diseases (28). However, experiments with mixed sera and IgM heavy chain sequencing suggested that in the present setting, individual differences arose from varying antibody specificity or quantity between individuals. So-called natural antibodies (many of which are polyreactive, arise without apparent immune exposure and with minimal somatic hypermutation) are thought to have a role in immunity and homeostasis, for example, by clearing cell debris (29). In contrast to current data, some such antibodies bind pathogens in a specific manner and are also able to direct them to lymphoid tissues (30). In the three individuals tested, neutralization potency was stable for several months, arguing against a cross-reactive response against a recent antigenic exposure. While this might suggest a “natural” origin of the antibodies, their sequences did display divergence from database genomic sequences.

The reason ganglioside antibodies were present in the B-cell donor is unknown. Several neuropathic conditions have been associated with antibodies against gangliosides, although in most cases, the causal role of such antibodies remains to be established (31). One factor that complicates the picture is the existence of ganglioside antibodies, including ones that target GM2, in healthy donors (32, 33). Therefore, though our study was not designed to assess their prevalence, finding the mAbs in a donor with no reported neuropathic condition is consistent with previous results. A limitation of our work is that we only isolated mAbs from a single individual. However, like these mAbs, all tested sera lost their non-specific neutralization activity when the virus was grown in the presence of a ganglioside synthesis inhibitor. This finding suggests that similar antibodies may be present in the other donors as well. Of note, the associations with neuropathies have not prevented the clinical development of ganglioside-targeting antibodies. Due to aberrant glycosylation often present in tumors, many gangliosides are considered tumor-associated carbohydrate antigens, and a number of antibodies have been trialed for cancer therapy (34).

Intriguingly, gangliosides are not the only host membrane components with functional roles when incorporated into the virion membrane. An exposed marker of apoptotic cell debris, phosphatidylserine, can increase the infectivity of a variety of enveloped viruses by binding specific receptors directly or via adaptor proteins (35–39). Conversely, extracellular vesicles with exposed phosphatidylserine are present in many body fluids and inhibit entry of such viruses (40).

In this study, we found that some individuals not exposed to the virus or corresponding vaccine could neutralize EBOV and other enveloped viruses and that the potency of that neutralization was stable over time. mAbs that reproduced the complement-mediated phenomenon were potent in targeting ganglioside GM2. These findings highlight the relevance of host membrane components in virus entry and immune monitoring and raise the possibility that antibodies against similar epitopes may have a role in natural immunity or therapeutic potential against enveloped viruses.

MATERIALS AND METHODS

Samples

De-identified normal donor serum and an anticoagulant citrate dextrose solution (solution A) containing half leukopak were obtained from a commercial source that collected the samples under an IRB-approved protocol. PBMCs were purified from buffy coats using Ficoll-Paque (Cytiva). When indicated, serum samples were used after heating at +56°C for 30 min.

Antibodies and compounds

Mouse immune ascites fluid against the Wad Medani strain EG AR 492 was provided by the University of Texas Medical Branch World Reference Center for Emerging Viruses and Arboviruses. Sutimlimab, narsoplimab, and eculizumab biosimilars were obtained from Proteogenix, AMY-101 acetate from MedChemExpress, and miglustat (N-butyldeoxynojirimycin hydrochloride) from Cayman Chemical. Mouse IgM mAbs against gangliosides (clones GMB16, MK1-16, GMR6, 2A3D2, GMR11, and GMR5) were from Amsbio, and the mouse IgM isotype control from Thermo Fisher Scientific. Human GM2 mAb L55-81 was from Novus Biologicals. ADI-15878 (11) and a non-specific IgG control were expressed in-house. Polyclonal human IgM isotype control (Thermo Fisher Scientific) was used as a negative control in Fig. 3A and IgM lambda isotype control (Southern Biotech) in other experiments. Before using for neutralization, commercial antibodies were dialyzed against PBS using Pur-A-Lyzer Mini Dialysis Kit (Sigma-Aldrich) with a 25 kDa cutoff. Total serum IgM was quantified using ProcartaPlex Human Antibody Isotyping Panel (Invitrogen).

Antibody depletions and purification of polyclonal antibodies

IgG, IgM, and IgA antibodies were depleted by mixing serum with PBS-washed resins (CaptureSelect Fc-XL, POROS CaptureSelect IgM-XL, or CaptureSelect IgA Affinity Matrix; Thermo Fisher Scientific) and incubating for 1 h at room temperature, followed by centrifugation to remove resin. The resulting supernatants were used for experiments and compared to non-depleted serum.

For purifying polyclonal IgG, IgM, and IgA fractions, the same resins were used for IgG and IgA, and POROS CaptureSelect IgM Affinity Matrix (Thermo Fisher Scientific) was used for IgM. PBS-washed resins were mixed with 300 µL of heated serum samples and incubated for 3 h with rotation. After four washes with PBS, the bound antibodies were eluted by two 5-min incubations with 0.1 M glycine buffer, at pH 3.0. The buffer of the pooled eluates was changed to PBS using 30 kDa Amicon columns (Sigma-Aldrich), and the resulting fractions were normalized so that for each isotype, maximal concentration was used in neutralization assays (119 µg/mL IgG, 29 µg/mL IgM, and 8 µg/mL IgA).

Cells and viruses

Huh7 cells (Apath, New York, NY, USA) were cultured in Dulbecco’s modified Eagle medium supplemented with 5% fetal bovine serum (FBS), 1 mM sodium pyruvate, 100 U/mL penicillin, 100 µg/mL streptomycin, and non-essential amino acids (Gibco). Vero-E6 cells were cultured in DMEM with 10% FBS, sodium pyruvate, and penicillin/streptomycin. Expi293F cells (Thermo Fisher Scientific) were cultured in Expi293 Expression Medium (Gibco).

EBOV-GFP (41), SUDV-ZSG (42), LUJV-ZSG (43), and Wad Medani virus Karyana (4) were grown on Huh7 cells unless otherwise indicated. For testing the impact of ganglioside synthesis inhibition on neutralization, cells were pre-treated for 4 days with 200 µM miglustat (or 0.66% DMSO), and the stocks were grown for 4 days in the presence of the same. Archived samples from rhesus macaque liver and spleen were used as such. Mouse-adapted EBOV/ZSG was recovered from CD1 mouse liver by grinding the tissue and passing the clarified supernatant through Detergent Removal Purification Columns (Thermo Fisher Scientific).

HepaRG cells (Gibco) were cultured according to the manufacturer’s recommendations and products. Briefly, cells were thawed using William’s E Medium with Thaw, Plate and General Purpose (TPGP) medium supplement and 2 mM GlutaMax supplement into vessels coated with collagen I from rat tail. After 1 day in culture, the medium was replaced with one containing Maintenance/Metabolism supplement instead of TPGP. On day 4 of culture, the cells were inoculated with EBOV-GFP, and the resulting stocks were recovered on day 4 of infection in the same medium. Comparing the infectivity of day 0 and day 4 samples showed that more than 99.9% of the virus in the HepaRG-grown stocks resulted from replication in these cells. The serum in the Maintenance/Metabolism supplement is from newborn bovine calves.

All work with infectious viruses was done under BSL-4 conditions at the Centers for Disease Control and Prevention, Atlanta, GA.

Neutralization assays

Neutralization assays were performed in 96-well plates on confluent Vero-E6 cell monolayers. Serum or monoclonal antibodies were diluted in Dulbecco’s modified Eagle medium containing 2% heat-inactivated FBS, penicillin/streptomycin, and sodium pyruvate (Gibco). When indicated, IgG/IgM-depleted human complement (PelFreez) was included so that the final dilution with virus was 5%. Dilutions were then mixed with virus stocks diluted in the same medium at 1:1, and the mixes were incubated at 37°C for 60 min. After removing medium from cells, replicate wells were infected with 40 µL inoculum volume for 1 h. The inocula were removed, and an overlay containing 1% carboxy-methylcellulose (Sigma-Aldrich), 4% FBS, sodium pyruvate, and penicillin/streptomycin in minimal essential medium (Gibco) was added. Fluorescent filovirus foci were quantified 5 days later using a BioTek Cytation instrument. Similarly, fluorescent LUJV foci were read on day 3 post-infection. Wad Medani virus was detected 1 day post-infection after fixing the monolayers with formalin, permeabilizing the cells with 0.5% Triton-X100 (Sigma), and detecting viral antigen with the mouse immune ascites fluid listed above. Normalized focus counts were plotted as a non-linear curve fit with four parameters using GraphPad Prism software, and titers reducing focus counts by either 50% or 80% (FRNT) are reported.

IgM neutralization screen

PBMCs were thawed using 10% FBS in RPMI-1640 medium (Gibco) and rested overnight. Cell viabilities were determined using Guava ViaCount reagents and an easyCyte 8HT instrument (Luminex). B cells were selected from 36 million live cells using the EasySep Human Pan-B Cell Enrichment Kit (Stemcell Technologies) and cells expressing an IgG receptor were depleted using anti-IgG microbeads (Miltenyi). The resulting cells were grown in a cell culture flask in ImmunoCult B Cell Expansion Kit medium (StemCell Technologies; with half the recommended supplement concentration). Two days later, cells were washed, and IgM-secreting cells were quantified using the Human IgM ELISpot Basic kit (MabTech). Upon reading the result the following day, the cells were washed and seeded in ImmunoCult B Cell Expansion Kit medium with full supplements so that 384-imaging plates received on average 7 IgM-secreting cells/well. All in all, approximately 28,000 cells were screened. After 5 days in culture, 10 µL of the supernatants (final dilution 1:2) was mixed with EBOV-GFP (with luciferase from transfected cells) and human complement (final 10%) in 2% heat-inactivated FBS, sodium pyruvate, penicillin/streptomycin, and GlutaMax in DMEM not containing phenol red (Gibco). The mixtures were incubated for 1 h at +37°C. After the incubation, 10 µL was added to Vero-E6 monolayers growing in 50 µL of the same medium in 384-imaging wells. Unused B-cell medium and heated serum sample corresponding to the leukopak at final dilution 1:20 were tested in every plate as negative and positive controls, respectively. On day 2 of infections, fluorescent foci were counted from images obtained with a BioTek Cytation reader, and fold differences to wells without sample were determined. Supernatants from hit wells (low observed infectivity) were analyzed for luciferase activity to ensure successful virus dispensing was achieved. Supernatants from corresponding B-cell culture wells were harvested for a repeat experiment. The B cells from hit wells were scraped into 0.2% Triton-X100 + 1 U/µL RNase inhibitor in tris-EDTA buffer and stored at −80°C. Reagents detecting ATP (Promega) were used to assess relative cell viability.

IgM sequencing

Hit well B-cell RNA was isolated using a commercial kit (Zymo) and subjected to reverse transcription and template switch with an oligo-dT primer, blocked template switch oligo /5AmMC6/GCTAATCATTGCAAGCAGTGGTATCAACGCAGAGTACATrGrGrG, SuperScript IV First Strand kit (Thermo), and Template Switching RT Enzyme Mix (New England Biolabs). These reactions were used as templates for PCR with Q5 High-Fidelity 2× Master Mix (NEB), primer GCTAATCATTGCAAGCAGTGGTATCAAC as the forward, and primers TCGTATCCGACGGGGAATTCTCACAG, CTGATGGGTGACTTCGCAGGCGTAG, and TGCGTGACCYGGCARCTGTAG as reverse primers for IGHM, IGK, and IGL genes, respectively. The PCR products were purified, and sequencing libraries were built using the SMRTbell Express Template Prep Kit 2.0 and the SMRTbell Barcoded Adapter Plate 3.0 according to the manufacturer’s instructions (Pacific Biosciences). Reads from a Sequel II system were trimmed for quality at Q40 (BBDuk plugin of the Geneious analysis software) and de novo assembled. Contigs with the most reads (without stop codons) were chosen for expression testing. Analysis of the resulting variable region sequences was performed using IMGT/V-QUEST (44, 45).

For sequencing the IGHM constant region, genomic DNA was isolated from CD19+ depleted (Miltenyi) whole blood using Monarch High-MW DNA Extraction Kit (New England Biolabs). Q5 High-Fidelity 2× Master Mix (NEB) was used first to amplify the approximately 5 kb region of interest with primers gcagtcgaacatgtagctgactcaggtcacCTCAGCCCCGACAGGCAG and tggatcacttgtgcaagcatcacatcgtagGGCTTGCTGGGAAGGGTG, except for 2 samples that were better amplified after replacing the first primer with gcagtcgaacatgtagctgactcaggtcacGGGCAGTCAGGCCTCAGA. Using gel-purified products as templates, Barcoded Universal Primers (Pacific Biosciences) were used in a second round of PCR. Gel-purified, pooled product was used for generating a long-read sequencing library with SMRTbell Express Template Prep Kit 2.0 for Sequel II sequencing.

IgM expression

Synthetic gene fragments corresponding to the most prominent contigs were cloned into pFUSE expression vectors (Invivogen). In the initial phase, miniprep DNA together with an expression vector for the J-chain (Invivogen) was transfected into Expi293F cells so that all IGHM and IGK/IGL combinations for a given well were obtained. After 4 days, the resulting IgM concentrations were estimated by ELISA, normalized, and samples tested for neutralization (and toxicity via ATP as surrogate) as above. The neutralizing heavy and light chain combinations, of which only one was identified for each original hit well, were considered the IgM hit mAbs, and in each case, the light chain was lambda 3. The commercial vector that encoded lambda 2 was mutated to express lambda 3 and to correct a single change in the variable/constant interface. The resulting plasmids were prepped at maxiprep scale and used together with the J-chain expression construct for expression in Expi293F cells with FectoPro reagent (Polyplus). On day 4 of transfection, supernatants were clarified by centrifugation and filtration through 0.2 µm PES membranes. The clarified supernatants (200–300 mL) were concentrated using a 100 kDa cutoff Centricon Plus-70 Ultracell system (MilliPore), and the resulting material was purified by size-exclusion chromatography (Superdex 200 pg columns; Cytiva) against PBS. Appropriate fractions were pooled and the IgM monoclonals stored at −80°C after flash-freezing with liquid nitrogen.

Glycan array and ganglioside ELISA

The hit IgM mAbs and a negative IgM lambda control (Southern Biotech 0158L-01) were tested for glycan binding on Glycan Array 300 (RayBiotech). The mAbs were tested at 5 µg/mL and detected with a goat anti-human IgM antibody with conjugated DyLight650 fluorophore (Thermo Fisher Scientific SA5-10105). The array was read by the manufacturer. Median hit IgM signals of 3 replicate spots were corrected by subtracting the background and the median signal of the negative control antibody.

Bovine-derived gangliosides GM1 and GM2 (Cayman Chemical) were dissolved in 2:1 methanol-chloroform, diluted in ethanol, and coated at 250 ng/96-well onto Immulon 1B plates (Thermo Fisher Scientific) by overnight evaporation. ELISA with purified IgM mAbs was performed after blocking with 1% bovine serum albumin (Sigma-Aldrich) in PBS with test antibodies and secondary anti-human and anti-mouse IgM-HRP secondary antibodies diluted in the same. When testing serum, blocking, sample, and secondary antibody dilutions were done in 1% heat-inactivated FBS in PBS.

Immunofluorescence staining

Huh7 cells pre-treated with 200 µM miglustat or DMSO were fixed using 10% neutral-buffered formalin, permeabilized with 0.5% Triton-X100, and stained with the hit mAbs or isotype control at 1 µg/mL, followed by detection with IrDye650-labeled anti-human IgM antibody and nuclear staining with DAPI (Thermo Fisher Scientific).

ETHICS APPROVAL

Animal experiments were approved by the CDC Institutional Animal Care and Use Committee (#3199), performed in an AAALAC International-approved facility, and conducted in accordance with the Guide for the Care and Use of Laboratory Animals. The CDC is fully accredited by the AAALAC-International.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/mbio.01709-25.

File S1. mbio.01709-25-s0001.docx.

Alignment of IgM heavy chain sequences.

File S2. mbio.01709-25-s0002.docx.

Alignment of IgM light chain sequences.

Supplemental figures. mbio.01709-25-s0003.pdf.

Figures S1-S6.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.