Introduction
Plants are the source of treatments for a wide variety of diseases worldwide [
1‐
4]. This includes viral infections such as human immunodeficiency virus type 1 (HIV-1) [
3,
5]. Many medicinal plants can be found in the family
Cucurbitaceae [
6,
7], including the genus
Momordica. The best-known medicinal plant in
Momordica is the species
Momordica charantia, commonly called "bitter melon." Bitter melon is used to treat a variety of medical conditions, from chronic inflammation [
8] and diabetes [
9‐
11] to cancer [
12,
13]. It has also been reported to have various inhibitory effects on viruses, bacteria, and parasites (see [
14]).
Momordica anti-HIV protein (MAP30), a ribosome-inactivating protein (RIP) isolated primarily from the seeds of
Momordica charantia, has been shown to have anti-HIV activity [
15,
16]. RIPs are N-glycosidases that depurinate ribosomal ribonucleic acid (rRNA) [
17]. This depurination irreversibly inactivates the ribosome, blocking protein synthesis [
18]. Another species within the
Cucurbitaceae is
Momordica balsamina. It also has been reported to produce a RIP with anti-HIV-1 activity [
19,
20]. The anti-HIV-1 activity of RIPs has been attributed to their interaction with viral nucleic acid and interaction with a post-reverse transcription step in replication [
19].
Momordica balsamina is also used to treat other medical conditions, such as gastric ulcers [
21,
22].
Proteins that bind to carbohydrate residues are collectively called carbohydrate-binding agents (CBAs) [
23]. CBAs can block the binding of envelope glycoproteins with their target receptors on cells. The ability of CBAs to bind to glycoproteins and block their interaction with receptors has been proposed as a potential means to inhibit enveloped viruses like HIV-1 [
23‐
26]; however, to date, no agent has become commercially available. A common type of CBA found in plants is lectins, which can bind to glycoproteins and cause red blood cells to agglutinate [
26,
27]. Plant Lectins are reportedly more resistant to heat denaturation than animal proteins [
28]. Indeed, banana and other plant lectins have been proposed to have various anti-microbial functions, including the ability to inhibit HIV-1 and Influenza viruses [
27]. It has been theorized that treatment of HIV-1 with CBAs could select for mutations that leave "holes" in the carbohydrate layer and allow for broadly neutralizing antibodies to be induced [
29]. Therefore, there has been significant interest in finding plant CBAs that could potentially be used to treat HIV/AIDS. Chitinases are another ubiquitous group of plant proteins that catalyze the breakdown of chitin found in the cell walls of fungi and can provide protection against fungal infection [
30]. Some plant chitinases have evolved the ability to hemagglutinate red blood cells and are sometimes called chi-lectins [
31]. Therefore, some of the many chitinases from plants could function as CBAs and potentially be used as antiviral treatments. Hevamine A-like proteins are a type of chitinase that generally act as plant defensins protecting the plant from pathogens with chitin in their cell walls, such as fungi [
32].
This report details the characterization of a 30 kDa protein we have previously reported [
33] from the medicinal plant
Momordica balsamina used by traditional healers in certain African regions. Its N-terminal sequence has homology to Hevamine A-like proteins in other plants. It can bind to HIV-1 gp120 and inhibit the virus. It is a CBA and could represent a potential new type of therapy against HIV that would allow a short-term treatment to induce long-term viral suppression.
Discussion
In this study, we report the anti-HIV-1 activity of water-soluble plant extracts from the medicinal plant
Momordica balsamina. We find that the biological activity is contained in a 30 kDa protein we call MoMo30-plant. Anti-HIV-1 activity has been previously reported in extracts of
Momordica balsamina [
20]. A 30 kDa ribosome-inactivating protein (RIP) has been reported that acts at a post-reverse transcription step in replication. MoMo30-plant is distinguishable from RIPs in several ways; first, its mechanism of action appears to be at the stage of attachment and fusion, as evidenced by MoMo30-plant's ability to bind to gp120 and block its interaction with cells expressing CD4 receptors. Second, from studies we did using surface plasmon resonance, MoMo30-plant appears to react specifically with glycans on the surface of gp120 (see Fig.
8). We also find that MoMo30-plant has chitinase activity (see Additional file
1: Fig. S1). Finally, its N-terminal sequence does not share homology with the N-terminal region of known RIPs from Momordica (Fig.
3B). In the current study, MoMo30-plant was isolated from water-soluble extracts made from dried leaves. In contrast, RIPs were isolated primarily from the plant's seeds. The healers partially process the dried leaves of
M. balsamina by making tea in boiling water. The exceptional heat stability of MoMo30-plant-plant may allow it to survive this process, while any other proteins in the solution would tend to denature and precipitate out of the solution. Heat stability could account for our observation that water-soluble extracts of
M. balsamina contain a single protein of 30 kDa. Therefore, we consider that MoMo30-plant protein is a previously undescribed protein present in
Momordica balsamina but distinct from RIP.
The MoMo30-plant protein has been difficult to study using typical proteomics techniques, and the genomic sequence of M. balsamina has not been reported. Attempts to do de novo sequencing of the protein were not successful. The exceptional stability of MoMo30-plant may contribute to its relative difficulty digesting with proteolytic digest by traditional enzymatic methods. Repeated attempts to analyze the protein sequence by LC/MS yielded very few or no peptides for analysis. Even intact mass measurements by MALDI/TOF were impossible as MoMo30-plant does not appear to be ionized using various matrix types. However, we obtained 15 amino acids from the N-terminus by Edman degradation (GPIVTYWGQNVXEGEL). Because of our difficulty with proteomics methods, our strategy was to isolate total cellular RNA from fresh plant tissue, perform RNAseq, and then search the transcriptome for proteins that match the N-terminal sequence.
We could germinate seeds and obtain enough growth to isolate total plant RNA. We used this to perform RNAseq (Illumina) and obtain a de novo assembly of the total transcriptome of
M. balsamina. A Diamond BLAST search for translated proteins showed significant homology to the Hevamine A-like protein from
M. charantia (see Fig.
4B). There are several possibilities why the match was not exact. The plants used for protein isolation were from Senegal and were dried. To have fresh plant material for RNA isolation, we needed to germinate seeds in our lab under artificial conditions to obtain enough plant cells for processing. Since the two sources of plant material were from different sources and conditions, this may have introduced variation from the mRNA to the original protein sequence. Another possibility is that the sequence identified by Edman determination is different than the gene identified by RNaseq. Both tissue-specific and temporal differences have been reported in the expression of defensin-like genes in Arabidopsis and Medicago [
41]. The complete genome of
Momordica balsamina has not been published to date so we can not rule out that such tissue-specific variants of the Hevamine A-like are present in
Momordica balsamina. To help ensure we had isolated the correct gene sequence, we had the gene for MoMo30-plant synthesized (GenScript) and cloned into an expression vector. In vitro
, transcription/translation by TNT wheat germ extract (Promega) was used to produce a 30 kDa protein that was reactive with an N-terminal antibody for MoMo30-plant and showed antiviral activity in a MAGI assay. Similarly, transfecting HEK293 cells with the expression plasmid produced a 30 kDa protein in the cell-free conditioned medium that was reactive with the N-terminal MoMo30-plant antibody, and the medium contained antiviral activity in a MAGI assay (see Fig.
7).
That MoMo30-plant could be a Hevamine A-like protein is consistent with its observed characteristics. Hevamine is one of several family members of plant chitinases and lysozymes produced as plant defensins against fungal infections [
32]. Plants produce intracellular and extracellular chitinases of approximately 25–35 kDa [
42]. We detected chitinase activity in MoMo30-plant (Additional file
1: Fig. S1). Hevamine is a 29 kDa protein that, according to the classification of Henrissat [
43], is a class III chitinase from family 18. Members of this family have a conserved region of amino acids from residues 120 to 130
DGX
DX
DW
EXP. The motif DDDE is highly conserved, as demonstrated by a psi BLAST search of over a thousand proteins with glycosyl transferase activity. One of the conserved aspartates (shown in bold) is replaced by arginine in Hevamine A-like proteins from both
M. balsamina and
M. charantia. This specific change was previously observed in the chitinase from
Aridopsis thaliania, which maintains chitinase activity [
44]. There are precedents for chitinases that can bind to glycoproteins. The lysine motif (LysM) is present in some chitinases from various sources [
45]. A LysM motif is also present in the carbohydrate-binding protein CyanoVirin-N [
46], which has been previously shown to have the ability to bind to gp120 on HIV and inhibit viral replication [
47].
A hypothesis for the action of carbohydrate-binding agents (CBAs) has been proposed [
23,
48]. Exposure of viruses to CBAs selects for variants expressing reduced numbers of carbohydrates on their surface (
i.e., mutants with fewer binding opportunities for MoMo30-plant-plant). Selection of viruses with reduced glycosylated gp120 proteins allows for altered immunological responses and particles with impaired infectivity, which could result in longer-term suppression. MoMo30-plant appears to act as a CBA, and exposure to the protein may induce a change in glycan patterns on the surface of virions. Changes in gp120 induced by selection in the presence of MoMo30-plant may alter the antigenicity of gp120 and allow a robust neutralizing immune response to be mounted, as proposed in the CBA theory (26). Moreover, it is generally believed that the lack of a robust immune response to HIV-1 is that gp120 is heavily glycosylated (50). Furthermore, our observation that the monosaccharide mannose can block the antiviral effect of MoMo30-plant suggests that MoMo30-plant can bind to high-mannose glycans and select for viruses with reduced glycosylation on the surface [
23].
Several other properties of MoMo30-plant make it a good candidate as a therapeutic for HIV-1. First, it has an IC
50 of 2.8 nm, similar to the IC
50 of approximately 20–40 nM for the commercially available fusion inhibitor Enfuvirtide [
35]. Secondly, it acts directly on the gp120; inhibiting attachment negates the need for penetration into the cell. In addition, once MoMo30-plant is bound to gp120, it stays bound for extended periods (Figs.
8A,
2B). The biphasic nature of the Biacore sensorgram likely reflects our observation that MoMo30-plant tends to form multimers (Additional file
1: Fig. S2). The initial off rate could reflect the disruption of multimers.
In contrast, the longer off rate is the dissociation of MoMo30-plant in complex with gp120. This prolonged “off” rate could mean less protein needs to be delivered to the blood over time to reach and maintain therapeutic levels. In addition, since the virus is prevented from entering the cell, it minimizes the possibility of integration into the host genome, avoiding potential latent infection.
It has been suggested that fewer glycosylated amino acids on g120 might allow for an altered immune response, enhancing the opportunities for broad-based neutralizing antibodies to be produced. If so, our model for MoMo30-plant action predicts that individuals treated with plants containing MoMo30-plant should have high levels of broadly neutralizing antibodies. Such studies are underway and should provide a more complete understanding of the mechanism underlying this traditional African therapeutic approach.
Methods
The preparation of water extracts from plants has been previously described [
33]. One hundred grams of dried and milled leaves from
Momordica balsamina were extracted in 1L of distilled water overnight at 4 °C. The liquid extract was separated from the solid material through centrifugation at 4000×
g for 30 min. at 4 °C. The resulting supernatant was filtered through Whatman filter paper (Cat# 3030) to remove particulates. The extract was then filter-sterilized by passing it through a 0.45-micron filter (Celltreat Cat # 229703) and was kept frozen at − 80 °C prior to lyophilization overnight.
Isolation of MoMo30-plant
The lyophilized powder was dissolved in nuclease-free water (Invitrogen Cat# AM9938) to create a 15 mg/mL solution. The solution was passed through a 30kD molecular weight cutoff filter by centrifugation at 4000×g for 10 min (Amicon ultra-15 cat# UFC903024) to remove low molecular weight contaminants. Once a retentate of 1 to 1.5 mL was obtained, the solution was passed through a 0.22-micron syringe filter (Celltreate Cat# 229747) and stored at 4 °C before use (− 20 °C for long term storage). The retentate contained one protein MoMo30-plant-plant that was > 95% pure as determined by SDS-PAGE and stained with Coomassie Brilliant Blue R-250 (Bio-Rad Cat# 161-0400).
Multinuclear activation of an indicator (MAGI) assay for infectivity
MAGI cell assays for infectivity were done as previously described [
49,
50]. MAGI cells (U-373-MAGI-CXCR4
CEM glioblastoma cells, AIDS reagent program cat# ARP-3596) were grown to 90% confluence. Cells were infected with HIV-1
NL4-3 (equivalent to 1 ng of p24; AIDS reagent program cat # 114). They were then fixed by the addition of 1% Formaldehyde (F-79–500 Fisher Chemicals) and 0.2% Glutaraldehyde (F-02957-1 Fisher Scientific) in PBS and stained in a solution that contained (14.25 mL PBS, 300 µL 0.2 M potassium ferrocyanide, 300 µL 0.2 M potassium ferricyanide, 15 µL 2 M MgCl
2 and 150 µL X-gal stock (40 mg/mL in DMSO). Two mL solution was added to each well and incubated at 37 °C for 50 min. Cells were washed twice with PBS and counted using light microscopy. Infected cells were identified as those exhibiting the development of a blue color.
Primary HIV-1 stocks
All HIV-1 strains (clade A to D) were obtained from the National Institutes of Health (NIH) AIDS Research and Reference Reagent Program (ARRRP) and were propagated at New Iberia Research Center by infecting human PBMCs 3 days after stimulation with 1.0 μg/mL ConA and culturing infected cells for 21 days after stimulation, replenishing media supplemented with 50 U/mL IL-2 twice a week. Viral supernatants were tested for HIV p24 antigen by ELISA kit (ABL Inc.) and supernatants with a high concentration were pooled. Virus-containing supernatants were clarified by centrifugation, sterile filtered and stored separately in 1-mL aliquots in liquid nitrogen.
Neutralization assay in TZM-bl cells
The neutralization activity of MoMo30-plant-plant against each HIV-1 strain (clade A to D) was measured using a standard protocol of luciferase-based HIV-1 neutralization assay in TZM-bl cells (Montefiori, Duke University). Briefly, 50 µL of fivefold serial diluted MoMo30-plant-plant and 50 µL of 1 ng virus were preincubated for 1 h at 37˚C in a 96-well flat-bottom plate. Next, 100 µL of TZM-bl cells (1 × 104/well) in 10% DMEM growth medium containing 15 µg/mL DEAE dextran (Sigma-Aldrich) were added to the preincubated each well, and the 96-well plates were incubated for 48 h. Assay controls included TZM-bl cells alone (cell control, no virus) and TZM-bl cells with virus only (virus control, no test reagent). At 48 h, the cells were lysed, and luciferase activity was measured using (Promega, Cat# E1501, 10 × 100 assays). on a BioTek Synergy HT multimode microplate reader The average background luminescence (RLU) from cell control wells was subtracted from the luminescence for each experimental well. The neutralization curves and 50% inhibitory concentration (IC50) were calculated and generated using GraphPad Prism (v7.01) software.
Determination of the effect of MoMo30-plant and Enfuvirtide on HIV-1NL4-3 infectivity
We performed a dose–response curve on MoMo30-plant and Enfuvirtide (Sigma SML0934). We used concentrations of MoMo30-plant from 0.314 to 78.25 nM. Stock solution of MoMo30-plant protein (782.5 nM) was diluted to final concentrations of 0.314 nM, 0.609 nM, 1.22 nM, 2.44 nM, 4.88 nM, 9.77 nM,19.55 nM, 39.06 nM,and 78.25 nM. For Enfurvirtide we diluted at stock solution of 782.5 nM to final concentrations of 2.2 nM, 4.43 nM, 8.86 nM, 17.73 nM, 35.44 nM, 70.91 nM, 142.05 nM, 272.73 nM, and 568.00 nM. To each 1 mL of diluted inhibitor we added 5 µL of HIV-1
NL43 (equivalent to 1 ng p24) and 10 µL of DEAE. The mixture was than added to 2 × 10
4 MAGI cells and then incubated for 48 h at 37 °C. and blue cells were counted. The IC
50 of MoMo30-plant was determined by curve fitting using the Hill equation and determined using the Dr. Fit program [
34].
Detection of MoMo30-plant in serum
To determine if the ingestion of plant extracts resulted in detectable levels of MoMo30-plant in the blood, two Rhesus macaques were given plant extracts using a scaled dosage to that typically given to humans. Basically, macaques were given the plant extracts with food. Two grams of plant was given twice a day for a period of six months. Blood samples were taken at 0 days up to 183 days. Plasma was tested by SDS -PAGE and Western blot.
MTT assay of MoMo30-plant
To determine if MoMo30-plant had significant cellular toxicity at therapeutic levels, we exposed HEK 293 cells to concentrations of MoMo30-plant from 1 to 1000 nM and performed a mitochondrial toxicity test (MTT; Sigma Cat# CGD-1) according to the manufacturer's recommendations. Percent viability was determined by comparison to an untreated control.
Stability studies on MoMo30-plant and its complex to gp120
To determine the heat stability of MoMo30-plant, we first subjected stock solutions of MoMo30-plant (4 ng/mL and 40 ng/mL) to temperatures from 25 to 120 °C for 30 min. After heating, the solution was mixed with 1 ng of HIV-1NL4-3 and was added to a MAGI cell assay, and blue cells were counted. In a separate study to determine the stability of complexes formed between virus and MoMo30-plant, we mixed virus equivalent to 1 ng p24 of HIV-1NL4-3 with a stock solution of MoMo30-plant (400 ng/mL) and kept the sample at 4 °C, we removed aliquots at time intervals of 5 min to 3 days, and centrifuged at 125,000 g through 20% sucrose cushion to remove free MoMo30-plant and tested its effect on infectivity by MAGI cell assay.
N-terminal sequencing of MoMo30-plant
Edman degradation was performed on the plant-derived MoMo30-plant protein in two separate labs (Biosynthesis, Lewisville, TX, and Creative Proteomics, New York, NY). The analysis was performed on an ABI Procise 494HT (Thermo Fisher). The procedure determines the N-terminal amino acid sequence of proteins and peptides by the Edman degradation chemistry.
RNAseq to determine the MoMo30-plant gene sequence
To help determine the gene sequence of MoMo30-plant we used RNAseq (Azenta Total RNA (~ 4 µg) purified from M. balsamina cells by the Trizol method was used for RNAseq on the Illumnina platform and the de novo.de novo transcriptome was assembled using Trinity software. The mRNA corresponding to the MoMo30-plant protein was determined by searching for the N-terminal sequence as determined by Edman degradation. Once a candidate DNA sequence had been determined we had the gene synthesized (Genscript) and cloned into the pGen-lenti vector which contains both a T7 promoter and a CMV promoter for expression in mammalian cells.
Software for gene assembly and translation
BLAST searches were done at the national center for biotechnology information (NCBI) website. Comparisons of homology to various proteins and DNA sequences were made in SnapGene 6.0.2. Prediction of secondary structure was made at the Phyre2 structure prediction web portal [
51].
Coupled transcription/translation of MoMo30-plant gene
The cloned version of the synthesized gene was expressed in the wheat germ coupled transcription/translation system (TNT T7 coupled Wheat Germ Extract System Promega Cat# L4140) according to the manufacturer's recommendations with the following modification. Instead of adding [35S] methionine, we added 1 mM of unlabeled methionine to the mixture and detected protein using Western blot. Ten µL of the product was resolved on a 4–20% SDS-PAGE gel, and a Western blot was done using an antibody to the N-terminal peptide of MoMo30-plant. Ten µL of product was also tested in a MAGI cell infectivity assay to determine any antiviral effect. Since the pGen-lenti vector also contained a CMV promoter, we transfected HEK293 cells with the plasmid using Lipofectamine 3000 (ThermoFisher Scientific Cat # L3000008) following manufacturer's protocol and grew cells for 48 h. Conditioned medium was harvested, and cells were collected and lysed by using (Pierce RIPA buffer Cat# 89901). Ten µL of cell pellet and 20 µL of conditioned medium were resolved on a 4–20% SDS-PAGE gel and blotted with N-terminal antibody to MoMo30-plant. 100 µL of conditioned medium was tested by MAGI cell assay for antiviral effects.
Fluorescent gp120 binding assay
Recombinant HIV-1 IIIB gp120 conjugated to FITC (ImmunoDX cat# 1001-F) was added to a suspension of 1 × 106/mL Jurkat cells (AIDS reagent program ARP-177 E6-1 clone) allowed to interact for 2 h at 37 °C. Free gp120 was removed by centrifugation, and the cells were resuspended in PBS. A portion was stained with Hoechst stain, and the cells were viewed under a fluorescent microscope using a neutral density filter, a blue filter, and a FITC filter.
Surface plasmon resonance (Biacore)
Surface plasmon resonance was done at the Biacore Molecular Interaction Shared Resource at Georgetown University. A Biacore T200 was used with a CM5 chip at 25 °C. Purified gp120-IIIB (Immuno Dx, 1 mg/mL) in 1 mM sodium acetate buffer at pH 5.5 was used as a ligand to immobilize onto FC2, FC3, and FC4 to the levels of 8850 RU, 283RU, and 2980RU, respectively. Standard amine coupling chemistry was used. HBS-P (10 mM Hepes, pH 7.4, 150 mM NaCl, 0.05% v/v surfactant P20) was used as the immobilization running buffer. Overnight kinetics were performed for MoMo30-plant binding to the ligand. Injected compound concentrations were 1–100 nM. Three 15 s pulses of 1:250 H3PO4 (v/v, ddH2O: H3PO4) were injected to regenerate the chip surface. All analyses were done in triplicate. The sensorgrams were obtained from overnight kinetics using 1:1 model fitting. In some experiments, gp120 was pre-treated with PNGase F (removes N-glycans) for 30 min at 50 °C before linkage to the chip surface. A control reaction was done with buffer alone. Three independent assays were done, each in triplicate.
MoMo30-plant inhibition in the presence of mannose
To determine the effect of the monosaccharide mannose on the activity of MoMo30-plant, we did infectivity assays with 2 nM of MoMo30-plant in the presence of mannose in concentrations from 0.002 to 2 nM final concentration. and determined relative inhibition using a MAGI cell assay as described. In brief, 2 nM MoMo30-plant and different concentrations of D-Mannose (Sigma cat# M6020) were mixed. After that added, virus equivalent to 1 ng of p24 of HIV-1NL4-3 was incubated at room temperature for five minutes, added to MAGI cells, and determined relative inhibition was using a MAGI cell assay as described.
Immunoblot
A rabbit antibody was produced (Genscript) from a 15-amino acid peptide with the N-terminal sequence of MoMo30-plant (GPIVTYGQNVNGELC). A separate rabbit antibody was made using a portion of the predicted sequence of the MoMo30-plant gene (LGGRSTSLRPGDC). The antibodies (at a dilution of 1:2000 for N-terminal ab; 1:3000 for the predicted sequence ab) were used to perform an immunoblot on purified protein resolved on a 4–20% SDS PAGE gel. The gels were transferred to 0.2 µm Nitrocellulose membrane using Bio-Rad Trans-Blot Turbo for 20 min and blocked with 0.5% skim milk made in Tris buffer saline with 0.1% tween 20 (TBST) for 1 h. The membrane was then incubated with primary antibody 1:2000 or 1:3000 in TBST overnight at 4 °C. The membranes were subjected to ten-minute washes with TBST and a wash with distilled water between each wash. Afterward, a secondary antibody (GE Healthcare goat anti-rabbit Cat# NA934V) was added at a dilution of 1:25,000 containing precision protein StrepTectin-HRP ( Bio-Rad Cat# 1610380) 1:10,000 and allowed to incubate for 1 h at room temperature. After this, the membrane was washed three times as previously, and chemiluminescent substrate (SuperSignal West Femto Thermo Scientific Cat# 34096) was added and incubated for 5 min. The blot was visualized by a chemiluminescent imager (ThermoFisher iBright 1500).
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.