Journal of Colloid and Interface Science 554 (2019) 296–304

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science

journal homepage: www.elsevier .com/locate / jc is

Regular Article

Towards a molecular understanding of the water purification propertiesof Moringa seed proteins

https://doi.org/10.1016/j.jcis.2019.06.0710021-9797/� 2019 The Authors. Published by Elsevier Inc.This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

⇑ Corresponding author.

M. Moulin a,b, E. Mossou a,b, L. Signor c, S. Kieffer-Jaquinod d, H.M. Kwaambwa e, F. Nermark f, P. Gutfreund a,E.P. Mitchell b,g, M. Haertlein a, V.T. Forsyth a,b, A.R. Rennie h,⇑a Institut Laue-Langevin, 71 Avenue des Martyrs, 38042 Grenoble Cedex 9, Franceb Faculty of Natural Sciences, Keele University, Staffordshire ST5 5BG, UKcUniv. Grenoble Alpes, CEA, CNRS, IBS, F-38000 Grenoble, FrancedUniv. Grenoble Alpes, CEA, Inserm, BGE U1038, F-38000 Grenoble, Francee Faculty of Health and Applied Sciences, Namibia University of Science and Technology, Private Bag 13388, 13 Jackson Kaujeua Street, Windhoek, NamibiafDepartment of Chemistry, University of Botswana, Private Bag UB00704, Gaborone, Botswanag European Synchrotron Radiation Facility, 71 Avenue des Martyrs, 38043 Grenoble Cedex 9, FrancehCentre for Neutron Scattering, Uppsala University, Box 516, 75120 Uppsala, Sweden

g r a p h i c a l a b s t r a c t

a r t i c l e i n f o

Article history:Received 23 April 2019Revised 19 June 2019Accepted 20 June 2019Available online 21 June 2019

Keywords:Moringa oleifera seedsMass spectrometryNeutron reflectometryX-ray diffractionWater treatment

a b s t r a c t

Seed extracts from Moringa oleifera are of wide interest for use in water purification where they can playan important role in flocculation; they also have potential as anti-microbial agents. Previous work hasfocused on the crude protein extract. Here we describe the detailed biophysical characterization of indi-vidual proteins from these seeds. The results provide new insights relating to the active compoundsinvolved. One fraction, designated Mo-CBP3, has been characterized at a molecular level using a rangeof biochemical and biophysical techniques including liquid chromatography, X-ray diffraction, massspectrometry, and neutron reflection. The interfacial behavior is of particular interest in consideringwater purification applications and interactions with both charged (e.g. silica) and uncharged (alumina)surfaces were studied. The reflection studies show that, in marked contrast to the crude extract, only asingle layer of the purified Mo-CBP3 binds to a silica interface and that there is no binding to an aluminainterface. These observations are consistent with the crystallographic structure of Mo-CBP3-4, which isone of the main isoforms of the Mo-CBP3 fraction. The results are put in context of previous studies of

M. Moulin et al. / Journal of Colloid and Interface Science 554 (2019) 296–304 297

the properties of the crude extract. This work shows possible routes to development of separation pro-cesses that would be based on the specific properties of individual proteins.� 2019 The Authors. Published by Elsevier Inc. This is an open access article under theCCBY-NC-ND license

(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Seeds from Moringa trees have long attracted attention as theycan contribute to rural economies in a number of ways [1–3] as asource of oil and other products. A traditional application of theseeds has been to flocculate impurities in water using crushedseeds and this has attracted scientific studies for about 40 years.Aggregation of particulate impurities is the usual first essentialstep in water purification as this allows gravitational separationand filtering that removes a large fraction of the impurities andreduces the turbidity. The efficacy in terms of reducing turbidityand bacteria in treated water has been established, and the lackof toxic residues together with the ease of supply of the materialmakes development of this process attractive to provide cleanwater to small communities, particularly in developing countries.There are 14 known species ofMoringa [2]. Most work has involvedMoringa oleifera that is native to the Indian sub-continent andwidespread in southern Africa, South America and a number ofAsian countries. However, there are some reports of the use ofMor-inga stenopetala and Moringa peregrina that are found more com-monly in other regions.

It is now clear that protein from the seeds is the active flocculat-ing agent and a number of studies have focused on establishing anunderstanding at amicroscopic level as to how thisworks. The usualmeans to achieve flocculation of colloids in water purification iseither by screening and neutralization of charge using polyvalentsalts, or by the addition of polymers that act as ‘bridging’ flocculantsand bind to multiple particles. However, Moringa proteins are notthought to act in either these ways or as a depletion flocculatingagent. The seed proteins are small, cationic and are difficult to dena-ture in solution. They show a marked tendency to self-associate inaqueous solution [4] unless the concentration is very low. Themodelthat emerges for flocculation is one that involves adsorption of pro-tein to a wide range of different particles and this will then favortheir association [5]. Essentially, the strong attractive interactionsbetween the molecules are transferred to the particles that are cov-eredwith protein. This allows effective coagulation and heterocoag-ulation of a broad range of particulate impurities.

Traditional applications of Moringa seeds would simply employcrushed seeds, and in some cases the seed oil will be separated forother uses. Investigations of the crude extract from seeds contain-ing a mixture of proteins have been reported in many studies asthis represents the material that would be exploited most fre-quently. Development of the process, optimization of the amountof material required, and studies as to how the seed protein couldbe used in practical conditions have been reported [6–13]. Thebinding of crude seed protein extracts to interfaces of silica [14]and alumina [15] from aqueous solutions has been investigatedby neutron reflection and the results give a picture of a capabilityto adsorb to a broad range of different materials as multi-layers ofprotein molecules.

The details of the extraction of protein from the seeds have beenidentified as important. The most effective flocculating agent wasobtained using a 1 M saline solution, although the specific typeof cation appears to be less significant than its concentration[16,17]. The presence of a mixture of proteins with a compositionthat might depend in part on naturally variable growth conditionsand extraction methods can explain partially the various reports ofdifferent molecular masses of the Moringa seed proteins.

It has been reported previously that there are two major proteincomponents isolated from crude extract of M. oleifera: MO2.1 thesequence of which was determined by Gassenschmidt et al. [18],and Mo-CBP3-1, which is the only Moringa seed protein with acrystal structure available [19]. Mo-CBP3-1 has been shown tobelong to the 2S albumin family of proteins [20], which is charac-terized by a compact fold due to the presence of a small chain and alarge chain linked together by four disulfide bonds. This arrange-ment is key to the marked stability of the protein and its heatand proteolysis resistance. Several studies have demonstrated thethermostability of both proteins which remain active for coagula-tion [21] and/or antifungal activities [22] after heat treatment. Thisstructural stability has been studied by circular dichroism, show-ing that there is no effect of heat treatment or pH variation onthe secondary structure [23].

Beside their use as flocculation agents, Moringa seed extractsshow antimicrobial activity [24]. This activity has been ascribedto Moringa synthesized derivatives of benzyl isothiocyanates,known antibacterial compounds, as well as to Moringa seedderived peptides [25]. Moringa seed powder can remove over 90%of the bacterial load of raw water samples. Water-soluble peptidesreleased from the crushed seed kernels belong to a group of catio-nic peptides often displaying antimicrobial activity through itsinteraction with negatively charged surfaces and its amphiphilicstructure allowing their incorporation into cellular membranes.The antibacterial activity of such a Moringa seed peptide (MO2.1also called Flo) against human pathogens including Pseudomonasaeruginosa and Streptococcus pyogenes has been studied in detail[25]. The authors suggest distinct molecular mechanisms for sedi-mentation and bactericidal activity involving membrane destabi-lization by a hydrophobic loop. Moringa seed peptides might bepromising alternatives to chemical water disinfection, chemicalfood preservatives or antibacterial treatments using antibioticsespecially when the pathogen has become antibiotic-resistant.The behavior of the proteins as an antibacterial agent has also beeninvestigated by cryo-EM on E. coli cells, showing a fusion of innerand outer membranes [26]. One strategy to exploit antibacterialactivity ofM. oleifera cationic peptide (MOCP) involves their immo-bilization on sand [27].

Comparative studies of a range of plant materials, and in partic-ular other seed proteins, have indicated that M. oleifera is the mosteffective natural flocculating agent for water purification [28–30].An understanding as to why particular seeds and particular pro-teins are more effective than others as a flocculating agent, or asan antimicrobial, will be important for the future development ofapplications from other plant varieties or from trees grown underdifferent conditions. The interaction of M. oleifera proteins withamphiphiles [4,31,32] and competitive adsorption to interfaceswith displacement from selected interfaces by rinsing with a catio-nic surfactant [15] offers the prospect of wider applications formineral separations. It has also been shown that the crude extractcan be used to bind particles to substrates [34]. That study alsoreported quartz-crystal microbalance data for adsorption that con-firmed that the bound layer of protein crude extract contained asignificant fraction of water.

The present study uses a combination of biophysical techniquesand activity assays to relate the properties and behavior of highlypositively charged proteins extracted and purified from M. oleiferato those of the crude seed extract.

298 M. Moulin et al. / Journal of Colloid and Interface Science 554 (2019) 296–304

2. Materials and methods

Unless specified otherwise, all chemicals used in this work werepurchased from Sigma-Aldrich and were of analytical grade.

2.1. Protein extraction from seeds

The initial extraction of Moringa protein involves use of petro-leum to remove oil, extraction of the protein with water, precipita-tion of protein with ammonium sulfate, filtration, dissolution ofthe precipitate in water, dialysis to remove excess of ammoniumsulfate, adsorption on a carboxymethyl cellulose column and elu-tion with 1 M NaCl, dialysis and finally freeze-drying. This proce-dure has been described by Kwaambwa and Maikokera [33].

Protein purification by ion exchange and size exclusion chro-matography: 20 mg of freeze-dried crude seed extract was resus-pended in 2 mL of carboxymethyl cellulose (CM) Sepharosebuffer (10 mM ammonium acetate, pH 6.7), and loaded on 5 mLof CM handmade Sepharose column. A gradient from 0 to 60% ofCM Sepharose buffer with 1 M NaCl for 8 column volumes (CV)was applied. The elution fractions were analyzed on a 4–20%Tris-Tricine gradient gel (Bio-Rad) and fractions corresponding tothe main peak were pooled together.

The pooled fractions from CM Sepharose were concentratedusing Amicon ultra centrifugal units (Millipore) (MWCO 3 K) to afinal volume of 1 mL and then loaded on a Superdex 75 HR 10/30 gel filtration column (GE Healthcare) which was equilibratedwith distilled water running at 0.5 mL min�1.

2.2. Mass spectrometry (MS) analysis

MALDI-TOF mass spectrometry measurements were carried atthe Institut de Biologie Structurale in Grenoble. Nano liquid chro-matography mass spectrometry (nano LC-ESI) and tandemMS datawere collected at the CEA Grenoble. Complete descriptions of theinstrumental settings that were used and the data acquisitionparameters are reported in the Supporting Information.

2.3. Determination of flocculation and coagulation activities

Simple tests of flocculation were made by adding purified Mor-inga protein or crude extract to dispersions of sulfonated polystyr-ene latex and observing cluster formation with an optical invertedmicroscope (Leica DMI1). The latex, PS3, was the same as that usedin previous studies of flocculation using ultra small-angle neutronscatteringmade byHellsing et al. [5]. The latex has been extensivelycharacterized using a variety of techniques [35] and has a zetapotential of about �35 mV. The final concentration of Moringa pro-teins was 2 mg mL�1 and the concentration of particles was 4% byweight.

To test the coagulation activity on E. coli and Nannochloropsisgaditana cells stationary phase cultures were supplemented with0.4 mg mL�1 of Moringa proteins and the coagulation observedunder a microscope.

2.4. Crystallization and X-ray diffraction

The screening conditions for crystallization of Mo-CBP3-4 pro-tein were optimized using the HTX platform of the Partnershipfor Structural Biology, Grenoble France. 6 � 96 conditions weretested at 20 �C, varying buffer, salt and protein concentration[36]. The resulting conditions for crystallization were establishedto be 0.1 M citric acid; 2.4 M sodium formate pH5.0.

For data collection at 100 K, individual crystals were brieflysoaked in 2.75 M sodium formate, 0.1 M citric acid, pH 5.0 with

10% glycerol prior to cryo-cooling. X-ray diffraction data were col-lected at ID29 [37] at the European Synchrotron Radiation Facility(ESRF), Grenoble, to a resolution of 1.68 Å for the high-resolutiondataset and to 2.06 Å for the sulfur single-wavelength anomalousdiffraction (SAD) dataset. Table S1 (Supporting Information) sum-marizes the data collection statistics. A helical data collection strat-egy was used to minimize the effects of radiation damage.

The data were indexed using space group I4122 witha = b = 108.07 Å and c = 43.62 Å, consistent with the previouslypublished crystal structure of Ullah et al. [19].

2.5. Neutron reflection

The interaction of the protein with mineral surfaces was inves-tigated by measuring neutron reflectivity with the D17 reflectome-ter [38] at the Institut Laue-Langevin in Grenoble. The time-of-flight mode was used allowing reflectivity data to be measuredat two incident beam angles, h, of 0.7 and 3.2 degrees with wave-lengths between 2.5 and 25 Å. These parameters gave a usefulrange of momentum transfer, perpendicular to the interface, Q (=(4p/k) sin h) of 0.006 to a maximum of 0.25 Å�1, which dependson the level of the signal over the background. Typical measure-ment times for each reflectivity curve were about 1 h. In order tomake measurements with a small amount of sample at two differ-ent interfaces, a special sample cell was used that had a reservoirfor solution formed by a PTFE gasket between one crystal of siliconand another of sapphire [39]. The sample holder could be rotatedby 180 degrees and translated so that either the silicon oxide sur-face on the silicon crystal or the alumina surface that were bothvertical and in contact with the liquid sample could be used forreflection measurements. Data were reduced to reflectivity curvesby normalizing to measurements of the incident spectra that weretransmitted through the relevant crystal using the cosmos software[40,41] available at the instrument. The program also subtracts thebackground that is observed on the instrument detector aroundthe specularly reflected signal.

The aqueous solvent could be exchanged automatically using aKnauer HPLC pump connected to the filling port. In order to mini-mize the volumes of protein solution, the sample solutions wereinjected manually using a syringe and a three-way valve at the bot-tom of the sample cell. Initial measurements were made using pureD2O and pure H2O to characterize the oxide layer on the silicon crys-tal and any gel layer or roughness at the alumina surface. Whenchanging contrast, at least 20 mL of solvent was flushed throughthe sample holder to an exit tube at the top of the cell. It was suffi-cient to inject 4 mL of purified protein solution at various concentra-tions to fill the sample cell or to increase the concentration.

3. Results

3.1. Protein purification by ion exchange and size exclusionchromatography

Two types of liquid chromatography were used to fractionatethe proteins present in the crude extract, as shown in Fig. 1. Thefirst step, using ion exchange chromatography results in the iden-tification of three main peaks. The peak of proteins, eluting at thehighest salt concentration (peak C) carrying the highest positivecharges, was then further purified using size exclusion chromatog-raphy (peak D). This material was used for all subsequent charac-terization apart from the mass spectrometry analysis.

These results show an unexpected change in the apparent sizedistribution of the protein components when run through the50 mMNaCl equilibrated column. This difference in the two elutionprofiles (i.e. peaks D and E) could be attributed to salt-dependentinteractions betweenMoringa proteins and the dextran polysaccha-

Table 1Identification of the main protein species observed in different samples from Moringa oleifera by mass spectrometry analysis.

Sample (1) Experimental mass (Da) (2) Theoretical mass (Da) (3) Sequence assignment (4) Modifications Proteinisoform

Without reduction (5)

Fraction 42 11959.02 11956.78 S + L Gln? Pyr(N-ter of S and L chains)

Mo-CBP3-3

Fraction 46 11786.88 11785.63 S + L Gln? Pyr(N-ter of S and L chains)

Mo-CBP3-4

Crude extract 11785.9011955.96

11785.6311956.78

S + L Gln? Pyr(N-ter of S and L chains)

Mo-CBP3-3Mo-CBP3-4

Crystals 11824 11785.63 S + L Gln? Pyr(N-ter of S and L chains) 2 oxidation sites on Lchain

Mo-CBP3-4

After reduction (5)

Fraction 42 4083.047881.99

4082.577882.27

small chain (S)

large chain (L)

Gln? Pyr(N-ter of S and L chains)

Mo-CBP3-3

Fraction 46 3777.908016.00

3777.288016.41

small chain (S)large chain (L)

Gln? Pyr(N-ter of S and L chains)

Mo-CBP3-4

Crude extract Nano LC-MS/MS analysis (6) Mo-CBP3-2Mo-CBP3-3Mo-CBP3-4

Crystals 37768045

3777.288016.41

small chain (S)large chain (L)

Gln? Pyr(N-ter of S and L chains)2 oxidation sites on L chain

Mo-CBP3-4

(1) Fractions 42 and 46 and crude extract were analyzed by nanoLC ESI MS and MS/MS; crystals were analyzed by MALDI TOF MS.(2) Mass values: the mass of the highest specie of isotopic pattern for nanoLC ESI MS spectra and average mass for MALDI TOF MS spectra are reported respectively.(3) Theoretical mass values calculated considering the formation of four disulfide bridges between the eight available cysteine residues present on the small (S) and large (L)chains.(4) Mo-CBP3-3 isoform: Small chain (S): Pyr-QQGQQQQQCRQQFLTHQRLRACQRFIRRQTQGGG. Large chain (L): Pyr-QARRPAIQRCCQQLRNIQPRCRCPSLRQAVQ-LAHQQQGQVGPQQVRQMYRLASNIPAICNLRPMSCPFG. Mo-CBP3-4 isoform: Small chain (S): Pyr-QQQRCRHQFQTQQRLRACQRVIRRWSQGGGP. Large chain (L):Pyr-QARRPPTLQRCCRQLRNVSPFCRCPSLRQAVQSAQQQQGQVGPQQVGHMYRVASRIPAICNLQPMRCPFR.(5) Samples were measured without reduction and after reduction with TCEP.(6) Nano LC-ESI-MS/MS analysis were carried out on the crude extract sample after running it on a 1D-SDS gel; twelve bands were cut and submitted to in-gel reduction andtrypsin digestion before mass spectrometry.

Fig. 1. (a) Ion exchange chromatography results that identify three main peaks from the crude extract (labeled A, B, C). Peak C eluted at a concentration of 0.6 M NaCl. (b) Sizeexclusion chromatography carried out on material from peak C using a water-equilibrated S75 column. (c) Further fractionation obtained using the same column equilibratedwith 50 mM NaCl. In all plots, the red trace shows the measured conductivity profile and the blue trace the absorbance at 280 nm.

M. Moulin et al. / Journal of Colloid and Interface Science 554 (2019) 296–304 299

Fig. 2. Microscopic images showing flocculation by the purified protein (peak D in Fig. 1) for the three different materials: (a) the algae Nannochloropsis gaditana, (b) thebacterium Escherichia coli (BL21), (c) latex particles.

300 M. Moulin et al. / Journal of Colloid and Interface Science 554 (2019) 296–304

ride component of the size exclusion column. Fractions 42 and 46were further analyzed by mass spectrometry as described below.

3.2. Mass spectrometry (MS)

Different samples obtained from M. oleifera were analyzed bymass spectrometry, using both MALDI TOF MS and nano LC-ESI MSand MS/MS (See Fig. S2 in Supporting Information), to characterizeand identify the main proteins present in the different extracts.The samples were measured before and after reduction with TCEP(tris(2-carboxyethyl)phosphine). A summary of the mass spectro-metric data obtained is given in Table 1. Analyzed fractions areshown in Fig. 1(c). Mass spectrometry analysis of the different frac-tions reveals a complexmixture ofN- andC-terminal processed spe-cies of variousM. oleifera seed protein isoforms (Mo-CBP3). TandemMSmeasurements on the reduced chains (S: small chain and L: largechain) allowed the identification of the abundant isoforms in eachfraction (see Table 1). In gel tryptic digestion of the crude extractresulted in the identification of three forms of Mo-CBP3 with theconcentration decreasing in the following order: Mo-CBP3-4 >Mo-CBP3-3>> Mo-CBP3-2. A pyroglutamate residue was identified attheN-terminal position of each individual chain for allMo-CBP3 iso-forms. Mass spectrometry analysis demonstrates the separation ofdifferent species after the size exclusion purification, with a pre-dominance of Mo-CBP3-3 component present in fraction 42 and ofMo-CBP3-4 component present in fraction 46.

3.3. Flocculation

Flocculation activity of the purified fraction D was visualized bystandard light microscopy. Fig. 2A shows that clear flocculationeffects (large visible aggregates) are introduced by the additionof the purified seed proteins for the algae N. gaditana; the sameeffect is noted in 2B for E. coli bacteria. Fig. 2C shows that, as withthe crude extract, the purified protein causes flocculation of anio-nic polystyrene latex particles. Simple inspection with opticalmicroscopy allows large aggregates to be identified. The particlesare the same as those described in previous scattering studies [5]where compact flocs with a high fractal dimension were observed.These results demonstrate the flocculating activity in relation to awide variety of materials.

3.4. Crystallographic structure of Mo-CBP3-4

In addition to the previous biochemical and biophysical charac-terization described above, crystallization trials were carried out inan attempt to get further structural insight into the water purifica-tion mechanism of theMo-CBP3 proteins. Fraction D (Fig. 1(b)) wassuccessfully crystallized, and analysis of the crystals by MALDI TOFmass spectrometry demonstrated that the material corresponds toMo-CBP3-4 (see Supporting Information, Fig. S1).

The crystal structure determined here is essentially the same asthat described by Ullah et al. [19] where the sample was mistak-enly referred to as Mo-CPB3-1 instead of Mo-CPB3-4. Both

Fig. 3. Primary sequence and the crystal structure of Mo-CPB3-4. (a) Shows theprimary sequence of Mo-CPB3-4 with the differences with Mo-CPB3-1 highlightedin red, the threonine on position 9 becomes a serine and the leucine in position 58becomes a proline, in Mo-CPB3-1. (b) and (c) Showing the helix and surfacerepresentation with polar versus non-polar areas of the protein. Red represents themost hydrophobic and white the most hydrophilic regions according to theEisenberg hydrophobicity scale [42].

M. Moulin et al. / Journal of Colloid and Interface Science 554 (2019) 296–304 301

sequences differ in two amino acid residues (highlighted in red inFig. 3a), according to Freire and co-workers [20] who have charac-terized the four isoforms of Mo-CBP3 (Mo-CPB3-1, Mo-CPB3-2,Mo-CPB3-3, Mo-CPB3-4).

The structural refinement and data statistics for the X-raydiffraction analysis are summarized in Table S1 (PDB depositioncode: 6S3F). Structure solution was determined by sulfur SAD.The values of Rmerge (0.035) completeness (99.6%), Rfree and Rwork

(0.23/0.18) are good indicators of the high quality of the data. Mostof the backbone of the protein was readily visible from the exper-imentally determined electron density map allowing unambiguousidentification of 90 residues without resorting to sequence infor-mation. The molecular structure consists of 5 a-helices formingtwo chains, each having an intra-chain disulfide bond. In addition,there are two inter-chain disulfide bonds. The structure is compactand highly stable as with other 2S albumin proteins. Fig. 3b showsthe crystal structure, highlighting the hydrophobic residues (inred). While the core of the protein appears to be mostly hydropho-bic, the solvent-exposed region, which contains a large number ofarginine residues (10 out of 90 residues) is highly hydrophilic, asseen in Fig. 4. The surface charge distribution of the protein

Fig. 4. (a) Surface charge distribution of Mo-CBP3-4 where red, white and blue represenisocontours going from �5kBT/e to +5kBT/e, where kB is the Boltzmann constant, T is tePyMOL [43] and APBS software [44] were used.

(Fig. 3c) confirms its high net positive charge, as well as the exis-tence of both cationic and anionic residues. This is consistent withthe high calculated isoelectric point of the protein (11.8). The crudeextract was also observed to have a high isoelectric point, above10, as noted by Kwaambwa and Rennie [4].

3.5. Surface behavior

3.5.1. Adsorption to silicaMeasurements of specular reflection of neutrons allow compo-

sition profiles at interfaces to be determined. Previous experiments[14] have identified that the crude extract binds to silica fromaqueous solution to form a dense layer with a thickness corre-sponding to multiple protein molecules near the surface, and a dif-fuse layer with decreasing density that extends towards the bulksolution. The results of the neutron reflection experiments showinteresting behavior that distinguishes the purified fraction fromthat found previously. The data for the purified fraction are shownin Fig. 5, and clearly demonstrate adsorption to the silica layer onthe silicon substrate for a solution having a concentration as low as0.025 mg mL�1. These results show a marked difference from thoseobserved for the clean interface, with a strong reduction in thereflectivity for the solution with protein in the Q range between0.05 and 0.15 �1.

In contrast to previous results on crude extract, there was nochange in the reflectivity as the concentration was raised. Thisindicated that there was no increase in the adsorbed amount athigher concentrations. This is illustrated in Fig. 6, which shows thatthere is no change in the reflectivity as the concentration isincreased to 0.5 mg mL�1. In comparison, the crude extract showedan increasing adsorption until a plateau amount was reached at aconcentration of 0.5 mg mL�1.

The clear change in reflectivity on adsorption of protein that isapparent in the data shown in Figs. 5 and 6 can be interpretedquantitatively using optical matrix calculations of the reflectivityas implemented in the cprof program [45]. The plot of RQ4 vs. Qallows some features in the data to be identified given that theeffect of the large contrast difference between silicon and D2O isdiminished. The minimum in the reflectivity at aboutQ = 0.12 �1 indicates a well-defined layer. Rinsing the surfacewith D2O after adsorption did not change the reflectivity (see Sup-porting Information, Fig. S3), showing that the adsorbed protein isnot displaced with pure water. This observation also allowed mea-surements to be made with a second solvent contrast that can pro-vide more details about the interfacial layer. The fit made to thecombined data for the bound layer after rinsing first with D2O

t negative, neutral and positive charge respectively with the electrostatic potentialmperature and e is the electronic charge. (b) Highlights the arginine residues. Both

Fig. 5. Neutron reflectivity data measured for the clean silicon/silica/D2O interface( ) and the surface in contact with 0.03 mg mL�1 solution of protein in D2O ( ).

Fig. 6. Reflectivity shown as RQ4 against Q for different concentrations of protein( 0.0025 mg mL�1, D 0.05 mg mL�1, 0.1 mg mL�1, h 0.5 mg mL�1) at the oxidelayer on a silicon substrate. For comparison, the data for the clean silicon/silicasubstrate are also shown ( ).

302 M. Moulin et al. / Journal of Colloid and Interface Science 554 (2019) 296–304

and then H2O is shown in Fig. 7. The fitted curves (solid lines) cor-respond to a layer of 15.3 ± 1 Å that consists of 53 ± 3% protein and47 ± 3% water. Allowance was made for exchange of protons in theprotein as well as the different contrast of water in the fit to thedata. The data are adequately modelled with a uniform proteinlayer having just 2 to 3 Å roughness at each of the interfaces.

The surface coverage of protein can be calculated from the fitparameters as the product of the volume fraction and the thicknessof the layer. It corresponds to 1.3 ± 0.2 mg m�2. This is hydratedand corresponds to a single molecular layer of protein and con-trasts with the multilayer adsorption that was found for the crudeextract.

Fig. 7. Reflection data for the adsorbed protein layer on the silica surface afterrinsing measured in D2O ( ) and H2O ( ) with the curves for the model describedin the text (1.3 mg m�2, 47% water, thickness 15 Å).

3.5.2. Adsorption to aluminaThe design of the sample holder [39] allowed measurements to

be made with the same protein solution at the alumina/solutioninterface by simply rotating the cell so as to reflect from a secondsolid/liquid interface. The data shown in Fig. S4 (Supporting infor-mation) indicate that there is no significant change in the reflectiv-ity, and this demonstrates that adsorption did not occur at thissurface. This contrasts with the results found previously for thecrude extract of Moringa seed protein [15] that had shown adsorp-tion to alumina, albeit with a lower plateau coverage thanobserved at the silica interface.

4. Discussion

These results, which exploit measurements made using a widevariety of techniques, demonstrate clear differences in the behav-ior of the purified fraction of the Moringa seed protein from thatof the crude extract. The observed flocculation activity for latexparticles is in agreement with previously published work [5].Sanchez-Martin et al. [46] have shown that the highest flocculationactivity for kaolinite corresponds to the most charged fractions ofthe extract. These observations imply that the valuable range ofinteractions that causes the Moringa seed proteins to flocculate awide variety of materials may depend on a mixture of proteinsbeing present. The surface of the latex particles has a small overallnegative charge that arises from dissociation of sulfate groups: thisexplains why the particles are stabilized in aqueous dispersion andhow flocculation occurs following combination with the positivelycharged protein. Similar interactions may also play a role in theflocculation/coagulation properties of the purified fraction for bac-teria and algae, as seen in Fig. 2.

The neutron reflection results, which show a clear interaction ofthe protein with the silica interface, and an essentially immeasur-able interactionwith the alumina interface,maybe of general signif-icance for an understanding of binding mechanisms relevant toapplications involving selective separation of different particles inaqueousdispersions. These observations canbeunderstood in termsof the positively charged nature of the purified fraction and are con-sistent with the crystal structure of Mo-CBP3-4, the major compo-nent of the analyzed fraction. It is interesting to consider thecrystal structure alongside the composition and arrangement ofthe bound protein at solid interfaces. The hydration of the crystalstructure is about 54% – very similar to the 47% identified in the pro-tein layer at the silica interface.Wenote that the interfacial behaviorof the purified fraction occurs as a protein monolayer, whereas thatfor the crude extract appears to incorporate additional diffuse layersbeyond the immediate surface, and this is apparently a consequenceof the presence of a mixture of different molecules [14,15]. Themolecular arrangement of the protein in the crystal, at neutral pH,suggests a fairly uniform distribution of positively charged groups;given the stability of the protein and its resistance to denaturationand degradation, the distribution of polar groups is likely to be sim-ilar in free solution. This may explain why the purified fraction doesnot associate as multilayers at surfaces, in contrast to the observedbehavior of the crude extract. It may also explain the lack of adsorp-tion on the uncharged alumina surface at neutral pH, in contrast tothe clear binding to negatively charged silica.

Apart from providing understanding of the behavior a singleprotein, Mo-CBP3-4, at a molecular level, the present work sug-gests that the combination of different materials in the seed extractmay be important in the overall use as a flocculating agent. Theprevious suggestions of a single sequence of peptides that give riseto flocculation [25,26,47] may be only part of an overall picture ofinteraction of multiple components together or of different pep-tides causing association of different impurities. Although thisobservation could at first be seen as a complexity in exploitation

M. Moulin et al. / Journal of Colloid and Interface Science 554 (2019) 296–304 303

of Moringa seeds, it presents the possibility that different compo-nents of seed material could be used for selective flocculationand consequent separation of specific components. This couldadd further value to the use of seed proteins in the treatment ofboth drinking water and waste-water.

5. Conclusions

The key findings of this work provide a step in understanding themolecular basis of thewater purification applications of theMoringaseedprotein system. The results demonstrate clearly the importanceof characterizing individual components of the seed proteins andtheir behavior. Whereas previous studies have focused on the char-acterization of less well-defined seed extracts, which may dependon species, harvesting conditions, and extraction methods, thisstudy has investigated the Mo-CBP3 protein fraction, which consti-tutes the largest single protein fraction from the seed extract. It con-tains predominantly 2 isoforms (Mo-CBP3-3 and Mo-CBP3-4). Incharacterizing this specific fraction, it is important to note that M.oleifera seeds contain several other constituents in addition to theproteins described here and that the physical and biochemical prop-erties of the extract such as flocculation, surface interaction, antimi-crobial activity as a whole may therefore depend on multiplebiomolecule interactions. The purified fraction adsorbs on silica asa single layer rather than as multilayers observed for the crude-extract [14]. This suggests that the presence ofmultiple componentsmay be important to causemultilayer absorption. Compared to pre-vious studies, the newfindings have yielded substantially improvedinsights beyond those concerning the crude protein extract wherematerial was found to adsorb to both silica and alumina [14,15].For example, the explicit identification of molecular characteristicsthat can control adsorption to different interfaces is important forpossible future exploitation of these materials in separation pro-cesses. Deeper understanding of which seed components bind tomaterials will also facilitate development of reusable filters, coatedwith Moringa proteins, which have been proposed for facile watertreatment [48–51].

The composition of the bound layer close to the surfaceobserved in this and in previous studies with both neutron reflec-tometry and quartz crystal microbalance measurements [34] thatconsists of almost 50% water corresponds approximately to thehydration within the crystal structure identified in the presentwork. This suggests that the maximum achievable protein densityat a surface may already be achieved with present protocols. Infuture studies, it will be interesting to explore whether differentvarieties of Moringa or different growth conditions for M. oleiferagive rise to variations in the molecular composition in the seeds.Previous work has shown differences in the flocculation of particlesin the presence of extracts from M. stenopetala and M. oleifera [5].Further work may also benefit from molecular dynamics (MD) cal-culations that exploit the crystal structure information in identify-ing interfacial interactions at a molecular level. Recent MD studiesof bovine serum albumin on silica surfaces have been described byKubiak-Ossowska et al. [52].

Additionally, recombinant expression of the different compo-nents may prove very useful for further detailed characterization;for example, the use of neutron scattering in conjunction within vivo deuteration approaches [53,54] may provide valuable infor-mation on interactions within the crude extract as well as withother relevant systems such as phospholipid bilayers.

Acknowledgements

We acknowledge the platforms of the Grenoble Instruct-ERICCenter (ISBG : UMS 3518 CNRS-CEA-UGA-EMBL) with support from

FRISBI (ANR-10-INBS-05-02) and GRAL (ANR-10-LABX-49-01)within the Grenoble Partnership for Structural Biology (PSB). Thisstudy was also supported by the ‘Investissement d’Avenir Infras-tructures Nationales en Biologie et Santé’ programme (ProFI pro-ject). We are also grateful for the allocation of beam time at the ILLfor neutron measurements (DIR-141, data available atdoi:10.5291/ILL-DATA.DIR-141) andat the ESRF formacromolecularcrystallography. HMK, FN and ARR thank the Swedish ResearchCouncil for support under Research Links project 348-2011-7241.V.T.F. acknowledges support from the EPSRC under grant numbersGR/R99393/01 and EP/C015452/1 which funded the creation of theDeuteration Laboratory (D-Lab) in the Life Sciences group of the ILL.

Appendix A. Supplementary material

Supplementary data to this article can be found online athttps://doi.org/10.1016/j.jcis.2019.06.071.

References

[1] J.F. Morton, The Horseradish Tree, Moringa-Pterygosperma (Moringaceae) – aBoon to Arid Lands, Econ. Bot. 45 (3) (1991) 318–333.

[2] J. Shindano, C. Kasase, Moringa (Moringa oleifera): a source of food andnutrition, medicine and industrial products, ACS Symp. Ser. 1021 (2009) 421–467.

[3] F. Anwar, S. Latif, M. Ashraf, A.H. Gilani, Moringa oleifera: a food plant withmultiple medicinal uses, Phytother. Res. 21 (2007) 17–25.

[4] H.M. Kwaambwa, A.R. Rennie, Interactions of surfactants with a watertreatment protein from Moringa oleifera seeds in solution studied by zeta-potential and light scattering measurements, Biopolymers 97 (4) (2012) 209–218.

[5] M.S. Hellsing, H.M. Kwaambwa, F.M. Nermark, B.B.M. Nkoane, A.J. Jackson, M.J.Wasbrough, A.R. Rennie, Structure of flocs of latex particles formed by additionof protein from Moringa seeds, Colloids Surf., A 460 (2014) 460–467.

[6] G.L. McConnachie, G.K. Folkard, M.A. Mtawali, J.P. Sutherland, Field trials ofappropriate hydraulic flocculation processes, Water Res. 33 (6) (1999) 1425–1434.

[7] J. Beltrán-Heredia, J. Sánchez-Martín, Improvement of water treatment pilotplant with Moringa oleifera extract as flocculant agent, Environ. Technol. 30(6) (2009) 525–534.

[8] M. Pritchard, T. Craven, T. Mkandawire, A.S. Edmondson, J.G. O’Neill, Acomparison between Moringa oleifera and chemical coagulants in thepurification of drinking water – an alternative sustainable solution fordeveloping countries, Phys. Chem. Earth. 35 (13–14) (2010) 798–805.

[9] M. Franco, G.K.e. Silva, J.E.S. Paterniani, Water treatment by multistagefiltration system with natural coagulant from Moringa oleifera seeds,Engenharia Agrícola 32 (5) (2012) 989–997.

[10] S. Bhatia, Z. Othman, A.L. Ahmad, Pretreatment of Palm Oil Mill Effluent(POME) using Moringa Oleifera seeds as natural coagulant, J. Hazard. Mater.145 (1–2) (2007) 120–126.

[11] R.K. Prasad, Color removal from distillery spent wash through coagulationusing Moringa oleifera seeds: use of optimum response surface methodology,J. Hazard. Mater. 165 (1–3) (2009) 804–811.

[12] A.F.S. Santos, A.C.C. Argolo, P.M.G. Paiva, L.C.B.B. Coelho, Antioxidant activity ofMoringa oleifera tissue extracts, Phytotherapy Research: PTR 26 (9) (2012)1366–1370.

[13] A.T.A. Baptista, P.F. Coldebella, P.H.F. Cardines, R.G. Gomes, M.F. Vieira, R.Bergamasco, A.M.S. Vieira, Coagulation-flocculation process with ultrafilteredsaline extract of moringa oleifera for the treatment of surface water, Chem.Eng. J. 276 (2015) 166–173.

[14] H.M. Kwaambwa, M. Hellsing, A.R. Rennie, Adsorption of a water treatmentprotein from moringa oleifera seeds to a silicon oxide surface studied byneutron reflection, Langmuir 26 (6) (2010) 3902–3910.

[15] H.M. Kwaambwa, M.S. Hellsing, A.R. Rennie, R. Barker, Interaction of moringaoleifera seed protein with a mineral surface and the influence of surfactants, J.Colloid Interface Sci. 448 (2015) 339–346.

[16] G.S. Madrona, R. Bergamasco, V.J. Seolin, M.R. Fagundes Klen, The potential ofdifferent saline solution on the extraction of the Moringa oleifera seed’s activecomponent for water treatment, Int. J. Chem. Reactor Eng. 9 (1) (2011) A31.

[17] G.S. Madrona, G.B. Serpelloni, A.M. Salcedo Vieira, L. Nishi, K.C. Cardoso, R.Bergamasco, Study of the effect of Saline solution on the extraction of theMoringa oleifera seed’s active component for water treatment, Water Air SoilPollut. 211 (1–4) (2010) 409–415.

[18] U. Gassenschmidt, K.D. Jany, T. Bernhard, H. Niebergall, Isolation andcharacterization of a flocculating protein from Moringa oleifera Lam, BBA –Gen. Subjects 1243 (3) (1995) 477–481.

[19] A. Ullah, R.B. Mariutti, R. Masood, I.P. Caruso, G.H. Gravatim Costa, C. Millenade Freita, R.K. Arni, Crystal structure of mature 2S albumin from Moringaoleifera seeds, Biochem. Biophys. Res. Commun. 468 (1–2) (2015) 365–371.

304 M. Moulin et al. / Journal of Colloid and Interface Science 554 (2019) 296–304

[20] J.E.C. Freire, I.M. Vasconcelos, F.B.M.B. Moreno, A.B. Batista, M.D.P. Lobo, M.L.Pereira, T.B. Grangeiro, Mo-CBP3, an antifungal chitin-binding protein fromMoringa oleifera seeds, is a member of the 2S albumin family, PLoS ONE 10 (3)(2015) 1–24, https://doi.org/10.1371/journal.pone.0119871.

[21] K.A. Ghebremichael, K.R. Gunaratna, H. Henriksson, H. Brumer, G. Dalhammar,A simple purification and activity assay of the coagulant protein from Moringaoleifera seed, Water Res. 39 (11) (2005) 2338–2344.

[22] J.M. Gifoni, J.T.A. Oliveira, H.P.H.D. Oliveira, A.B. Batista, M.L. Pereira, A.S.Gomes, I.M. Vasconcelos, A novel chitin-binding protein from Moringa oleiferaseed with potential for plant disease control, Biopolymers 98 (4) (2012) 406–415.

[23] A.B. Batista, J.T.A. Oliveira, J.M. Gifoni, M.L. Pereira, M.G.G. Almeida, V.M.Gomes, I.M. Vasconcelos, New insights into the structure and mode of action ofMo-CBP3, an antifungal chitin-binding protein of Moringa oleifera seeds, PLoSONE 9 (10) (2014) 1–9.

[24] M. Madsen, J. Schlundt, E.F.E. Omer, Effect of water coagulation by seeds ofMoringa oleifera on bacterial concentrations, J. Ethnopharmacol. 22 (3) (1988)329.

[25] M. Suarez, M. Haenni, S. Canarelli, F. Fisch, P. Chodanowski, C. Servis, N.Mermod, Structure-function characterization and optimization of a plant-derived antibacterial peptide, Antimicrob. Agents Chemother. 49 (9) (2005)3847–3857.

[26] K. Shebek, A.B. Schantz, I. Sines, K. Lauser, S. Velegol, M. Kumar, Theflocculating cationic polypetide from moringa oleifera seeds damagesbacterial cell membranes by causing membrane fusion, Langmuir 31 (15)(2015) 4496–4502.

[27] H.A. Jerri, K.J. Adolfsen, L.R. McCullough, D. Velegol, S.B. Velegol, Antimicrobialsand via adsorption of cationic Moringa oleifera protein, Langmuir 28 (4)(2012) 2262–2268.

[28] I. Bodlund, Coagulant Protein from plant materials: Potential Water TreatmentAgent Licentiate Thesis, KTH, Sweden, 2013.

[29] I. Bodlund, A.R. Pavankumar, R. Chelliah, S. Kasi, K. Sankaran, G.K. Rajarao,Coagulant proteins identified in Mustard: a potential water treatment agent,Int. J. Environ. Sci. Technol. 11 (4) (2014) 873–880.

[30] S. Conaghan, M.Sc. Thesis, Development and analysis of environmentallyneutral, biodegradable, novel flocculants for drinking water treatment, DublinCity University, 2013.

[31] R. Maikokera, H.M. Kwaambwa, Interfacial properties and fluorescence of acoagulating protein extracted from Moringa oleifera seeds and its interactionwith sodium dodecyl sulphate, Colloids Surf., B 55 (2) (2007) 173–178.

[32] H.M. Kwaambwa, F.M. Nermark, Interactions in Aqueous Solution of aZwitterionic Surfactant with a Water Treatment Protein from Moringaoleifera Seeds Studied by Surface Tension and Ultrasonic VelocityMeasurements, Green Sustain. Chem. 3 (4) (2013) 135–140.

[33] S. Nouhi, M. Pascual, M.S. Hellsing, H.M. Kwaambwa, M.W.A. Skoda, F. Höök, A.R. Rennie, Sticking particles to solid surfaces using Moringa oleifera proteins asa glue, Colloids Surf., B 168 (2018) 68–75.

[34] H.M. Kwaambwa, R. Maikokera, Infrared and circular dichroism spectroscopiccharacterisation of secondary structure components of a water treatmentcoagulant protein extracted from Moringa oleifera seeds, Colloids Surf., B 64(1) (2008) 118–125.

[35] A.R. Rennie, M.S. Hellsing, K. Wood, E.P. Gilbert, L. Porcar, R. Schweins, M.Malfois, Learning about SANS instruments and data reduction from roundrobin measurements on samples of polystyrene latex, J. Appl. Crystallogr. 46(5) (2013) 1289–1297.

[36] N. Dimasi, D. Flot, F. Dupeux, J.A. Márquez, Expression, crystallization and X-ray data collection from microcrystals of the extracellular domain of thehuman inhibitory receptor expressed on myeloid cells IREM-1, ActaCrystallogr., Sect. F: Struct. Biol. Cryst. Commun. 63 (3) (2007) 204–208.

[37] D. De Sanctis, A. Beteva, H. Caserotto, F. Dobias, J. Gabadinho, T. Giraud, C.Mueller-Dieckmann, ID29: A high-intensity highly automated ESRF beamlinefor macromolecular crystallography experiments exploiting anomalousscattering, J. Synchrotron Radiat. 19 (3) (2012) 455–461.

[38] R. Cubitt, G. Fragneto, D17: the new reflectometer at the ILL, Appl. Phys. A 74(Suppl) (2002) S329–S331.

[39] A.R. Rennie, M.S. Hellsing, E. Lindholm, A. Olsson, Note: Sample cells toinvestigate solid/liquid interfaces with neutrons, Rev. Sci. Instrum. 86 (1)(2015).

[40] COSMOS, 2017. <https://www.ill.eu/instruments-support/instruments-groups/instruments/d17/more/documentation/d17-lamp-book/cosmos/>.

[41] P. Gutfreund, T. Saerbeck, M.A. Gonzalez, E. Pellegrini, M. Laver, C. Dewhurst, R.Cubitt, Towards generalized data reduction on a chopper-based time-of-flightneutron reflectometer, J. Appl. Crystallogr. 51 (3) (2018).

[42] D. Eisenberg, R.M. Weiss, T.C. Terwilliger, The hydrophobic moment detectsperiodicity in protein hydrophobicity, Proc. Natl. Acad. Sci. 81 (1) (1984) 140–144.

[43] The PyMOL Molecular Graphics System, Version 1.8 Schrödinger, LLC[44] N.A. Baker, D. Sept, S. Joseph, M.J. Holst, J.A. McCammon, Electrostatics of

nanosystems: application to microtubules and the ribosome, Proc. Natl. Acad.Sci. 98 (18) (2001) 10037–10041.

[45] Rennie A.R, (2015) http://www.reflectometry.net/fitprogs/cprof.htm[46] J. Sánchez-Martín, J. Beltrán-Heredia, J.A. Peres, Improvement of the

flocculation process in water treatment by using Moringa oleifera seedsextract, Braz. J. Chem. Eng. 29 (3) (2012) 495–501.

[47] M. Suarez, J.M. Entenza, C. Doerries, E. Meyer, L. Bourquin, J. Sutherland, N.Mermod, Expression of a plant-derived peptide harboring water-cleaning andantimicrobial activities, Biotechnol. Bioeng. 81 (1) (2003) 13–20.

[48] H.A. Jerri, K.J. Adolfsen, L.R. McCullough, D. Velegol, S.B. Velegol, Antimicrobialsand via adsorption of cationic Moringa oleifera Protein, Langmuir 28 (2012)2262–2268.

[49] F.E. Williams, A.K. Lee, S. Orandi, S.K. Sims, D.M. Lewis, Moringa oleiferafunctionalised sand – reuse with non-ionic surfactant dodecyl glucoside, J.Water Health 15 (2017) 863–872.

[50] C. Mndelemani, Y. Song, F. Wei, Y. Chen, Chemically enhanced primarytreatment-trickling filter with Moringa oleifera seeds for improved energyrecovery from wastewater treatment plants in Malawi. 2018, Int. J. Environ.Protect. Policy 6 (2018) 26–31.

[51] B. Xiong, B. Piechowicz, Z. Wang, R. Marinaro, E. Clement, T. Carlin, A. Uliana,M. Kumar, S.B. Velegol, Moringa oleifera f-sand filters for sustainable waterpurification, Environ. Sci. Technol. Lett. 5 (2018) (2018) 38–42.

[52] K. Kubiak-Ossowska, K. Tokarczyk, B. Jachimska, P.A. Mulheran, Bovine serumalbumin adsorption at a silica surface explored by simulation and experiment,J. Phys. Chem. B 121 (16) (2017) 3975–3986.

[53] M. Haertlein, M. Moulin, J.M. Devos, V. Laux, O. Dunne, V.T. Forsyth, Chapterfive – biomolecular deuteration for neutron structural biology and dynamics,Methods Enzymol. 566 (2016) 113–157.

[54] O. Dunne, M. Weidenhaupt, P. Callow, A. Martel, M. Moulin, S.J. Perkins, V.T.Forsyth, Matchout deuterium labelling of proteins for small-angle neutronscattering studies using prokaryotic and eukaryotic expression systems andhigh cell-density cultures, Eur. Biophys. J.: EBJ 46 (5) (2017) 425–432.