Moringa oleifera functionalised sand – reuse with non-ionic surfactant dodecyl glucoside

Eleven percent of the world's population relies on untreated surface or ground water as a primary source of drinking water (WHO 2014). This water is often contaminated with pathogenic bacteria, and treatment for immediate use is commonly undertaken using tablets of sodium hypochlorite or calcium hypochlorite (Arnold & Colford 2007). This method is mainly used in regions which lack access to suitable water treatment infrastructure. The use of the tablets prevents recontamination, as the chlorine provides residual protection lasting from hours to days (Arnold & Colford 2007). High cost, an unpleasant taste and low availability of these chemicals have been identified as the main reasons which lead to their underuse or misuse (Islam et al. 2014).

The crushed seed kernels from Moringa oleifera are known to be an appropriate antibacterial agent for water treatment (Madsen et al. 1987; Jahn 1988). The seeds can be used directly, or as a crude water extract, to inhibit bacterial and viral organisms. They also cause coagulation of suspended particles in water (Ali et al. 2004; Kansal & Kumari 2014). However, the disadvantage of using the seeds in such a way is that the crushed seed particles quickly decompose and increase the amount of organic carbon in the water, promoting bacterial regrowth (Jerri et al. 2012).

The compound in the M. oleifera seed that causes the effects has been identified as a cationic polypeptide called MO_2.1 (Gassenschmidt et al. 1995; Suarez et al. 2005). This polypeptide is part of a dimer protein comprised of two polypeptide subunits linked by di-sulfide bonds (Ndabigengesere et al. 1995). The amino acid sequence of MO_2.1 has been documented (Gassenschmidt et al. 1995) and the molecular weight is 6,782 g/mol (http://www.uniprot.org/, accessed 28/09/2016).

MO_2.1 has been observed to adsorb to the surface of negatively charged solids, including silica nanoparticles, quartz sand, aluminum oxide and micro-emulsion prepared magnetic iron oxide nanoparticles (Kwaambwa et al. 2010; Jerri et al. 2012; Okoli et al. 2012; Hellsing et al. 2013). When the MO_2.1 is adsorbed to sand, this ‘functionalised’ sand (known as ƒ-sand) displays the bactericidal and coagulant properties of the non-bonded polypeptide (Jerri et al. 2012). This simple method of purifying MO_2.1 has scope to be used in water treatment, as it negates the previously described disadvantages of the crude extract as well as reducing the costs associated with isolating the polypeptide by other methods. The ƒ-sand can be used to attach and inhibit Escherichia coli in water (Jerri et al. 2012). In practice, the obstacle which currently prevents the use of ƒ-sand to treat water is that the E. coli does not desorb once bonded to the ƒ-sand, preventing its reuse (Jerri et al. 2012).

Surfactants are a class of amphiphilic compounds able to wet surfaces and solubilise fatty material (Glover et al. 1999). Anionic surfactants, such as sodium dodecyl sulfate, are reported to be bactericidal; affecting the permeability of cell membranes and causing cell lysis at high surfactant concentrations (Glover et al. 1999; Van Hamme et al. 2006). Non-ionic surfactants are also bactericidal and are known to solubilise the bacterial lipid membrane (le Maire et al. 2000). When solubilised in water, surfactants self-assemble into compounds called micelles, and the minimum surfactant concentration required for micelle formation is called the critical micelle concentration (Privé 2007).

M. oleifera seed extracts have been observed to interact with sodium dodecyl sulfate and other anionic surfactants (Kwaambwa & Maikokera 2007; Maikokera & Kwaambwa 2007; Beltran-Heredia & Sanchez-Martin 2009). The crude seed extract has been reported to remove up to 80% of sodium dodecyl sulfate via flocculation (Beltran-Heredia & Sanchez-Martin 2009). Both MO_2.1 and the dimer protein are cationic and readily adsorb to the anionic sodium dodecyl sulfate (Beltran-Heredia & Sanchez-Martin 2009; Beltran-Heredia et al. 2012). This has been reported to neutralize the charge (Maikokera & Kwaambwa 2007). The process begins below the critical micelle concentration of 8.2–8.3 × 10⁻³ M, at the critical aggregation concentration of 1 × 10⁻⁵ M, and can be used to remove sodium dodecyl sulfate from solution (Beltran-Heredia & Sanchez-Martin 2009).

Dodecyl glucoside is a non-ionic surfactant, and no studies investigating this surfactant and the M. oleifera crude or purified seed extracts were identified in the literature. Okoli et al. (2012) report that a 0.1 wt% solution of the non-ionic surfactant Tween 20 can be used to wash coagulated clay particles off M. oleifera protein functionalised magnetic nanoparticles. These particles can then be reused as coagulants with some loss of efficiency. Triton X, another non-ionic surfactant, does not interact with the M. oleifera coagulant protein (Kwaambwa & Maikokera 2007).

This study demonstrates the use of surfactants in regenerating ƒ-sand for continued E. coli removal. All concentrations of dodecyl glucoside tested against the ƒ-sand were above the critical micelle concentration of 1.9 × 10⁻⁴ M (Neugebauer 1990). This is a novel approach for the regeneration of ƒ-sand and has the potential to be applicable in water treatment for developing countries.

This investigation was conducted in two stages. The first stage was focused on testing two surfactants, dodecyl glucoside and sodium dodecyl sulfate, to examine their effect on attachment of the MO_2.1 to the sand surface. The second stage evaluated the efficiency of these surfactants at separating E. coli from the MO_2.1.

The functional sand (ƒ-sand) was prepared using M. oleifera seeds. The preparation was based on the method described by Jerri et al. (2012) with some modifications. Dry, whole seeds were purchased from AustraHort, Australia. The seeds were dehusked and the kernels pulverised to a fine powder. A 0.05 g/mL crude solution was prepared by mixing 20 g of the seed powder with 400 mL of a 0.1 M sodium chloride solution for 1 hour at 150 rpm. The solution was sieved and filtered to remove the seed particles.

100 g of commercially available, fine grained silica sand was thoroughly rinsed with deionised water, dried then autoclaved. The sand was combined with 400 mL of the filtered crude extract solution and mixed on a shaker plate at 150 rpm for 1 hour. After the ƒ-sand settled, the crude extract supernatant was removed and the ƒ-sand was rinsed with deionised water. The ƒ-sand was dried at 50 °C.

The mass of MO_2.1 per gram of ƒ-sand was determined by eluting the polypeptide on 1 g of ƒ-sand using 5 mL of a 0.6 M NaCl solution. A baseline for the absorbance at 280 nm was set against the absorbance at 300 nm. A Shimadzu 1601 UV-visible spectrophotometer capable of scanning from 190 nm to 1,000 nm was used to detect the absorbance throughout this study. The Beer-Lambert law, Abs = ɛlc, was applied using a path-length (l) of 1 cm and an extinction coefficient (ɛ) of 1,520 (Aitken & Learmonth 1996). The mass of MO_2.1 was determined in mg per gram ƒ-sand using a molecular weight of 6,781.6 g/mol, based upon the reported amino acid sequence of MO_2.1 (http://www.uniprot.org/, accessed 28/09/2016).

Two surfactants were examined in this study: anionic sodium dodecyl sulfate (sodium lauryl sulfate, CAS 151-21-3) and non-ionic dodecyl glucoside (lauryl glucoside, CAS 11061-47-9) at concentrations ranging between 0.0005 M and 0.1 M. The concentrations of dodecyl glucoside were determined by UV-spectrophotometry. The absorbance was measured at 223 nm and was initially determined by serially diluting the supplied surfactant from a 49.74% w/w solution. A commercially available dishwashing detergent (Earth choice dishwashing concentrate, Natures Organics, Australia) was purchased from a local Australian supermarket. The detergent contained 10%–30% dodecyl glucoside as coco glucoside and <10% sodium dodecyl sulfate, and was tested at 1% and 2% v/v dilutions. The amount of dodecyl glucoside in the 1% solution was equivalent to 0.01 M dodecyl glucoside solution as confirmed by UV-visible spectroscopy.

The effect of the surfactants sodium dodecyl sulfate and dodecyl glucoside on the attachment of MO_2.1 to the sand was measured by exposing 1 g of ƒ-sand to 5 mL surfactant at known concentrations. This was mixed for 1 hour and rinsed with deionised water. The samples were mixed between one and four times to determine the effect of multiple washes on the ƒ-sand. The amount of MO_2.1 was measured as described above and the experiment was repeated in triplicate. The concentrations of sodium dodecyl sulfate examined were 0.01, 0.001 and 0.0005 M, and the concentrations of dodecyl glucoside were 0.1, 0.01 and 0.001 M.

Synthetic water with a known bacterial concentration was prepared to simulate non-potable water. Escherichia coli (ATCC^® 25922™) was sourced from the School of Animal and Veterinary Sciences, The University of Adelaide and maintained on Bacteriological agar (Oxoid). Overnight bacterial suspensions were grown in nutrient broth (Oxoid). The suspension was centrifuged for 5 minutes at 3,000 g and the pellet was washed in a sterile 0.85% NaCl solution. This was centrifuged, and the pellet was re-suspended in sterile Milli-Q water diluted to an absorbance of 0.1 ± 0.005 at 600 nm. This was equivalent to a bacterial suspension containing 7 × 10⁷ CFU/mL of stationary phase E. coli as confirmed by cultivatable counts. Fresh synthetic water was prepared for each set of experiments.

Stage two evaluated the efficiency of the surfactants at separating E. coli from the MO_2.1. To do this, 50 g of dry ƒ-sand was poured into a vertical glass column of 1.5 cm diameter. The ƒ-sand was rinsed with Milli-Q water until the absorbance of the outgoing solution was less than 0.001 at 280 and 600 nm. The column was then subjected to a set of solutions in the following order: synthetic water, 2 × Milli-Q water, surfactant, 4 × Milli-Q water. Each set was repeated four times, so that the ƒ-sand had been exposed to four discrete synthetic water treatments. A schematic of this process is presented in Figure 1. New columns were prepared for sodium dodecyl sulfate concentrations of 0.001 and 0.0005 M and for dodecyl glucoside concentrations of 0.1, 0.01 and 0.001 M.

Sand was heated at 600 °C in a high temperature oven for 5 hours to remove any impurities and organic matter, including dead bacteria, from the surface. 10 g of ƒ-sand was prepared as above and exposed to synthetic water, 0.01 M dodecyl glucoside and synthetic water in the same manner as the columns. 2 g samples were collected after each step and were stained with a 1:500 dilution of propidium iodide in Tris buffer for 15 minutes, then rinsed twice with sterile Milli-Q water. Staining was performed on all samples, including controls of bare sand and ƒ-sand, to confirm that only the dead E. coli was being stained. The samples were viewed on a Nikon Ti Live Cell Microscope at 40× optical zoom. A 100 ms exposure for fluorescence was used on all samples with a 195 ms exposure for the live view.

The mass of MO_2.1 per gram of ƒ-sand was calculated using the Beer-Lambert Law. The maximum, minimum and average masses of eluted MO_2.1 in mg per g of ƒ-sand were 1.19, 0.48 and 0.79, respectively. Variation in turbidity was observed across all of the elution-type experiments. To overcome the variation between samples and batches, the absorbance at 300 nm was used as a baseline and subtracted from the absorbance of the protein peak at 280 nm.

The effect of sodium dodecyl sulfate on the attachment of MO_2.1 to the sand was immediately apparent. At 0.01 M, the highest concentration tested, there was an 80% reduction in the amount of MO_2.1 but there was no further significant loss after the first wash. At 0.001 M, there was a gradual reduction in the amount of MO_2.1, and at 0.0005 M, the amount present was relatively stable. The results of the interaction of sodium dodecyl sulfate and ƒ-sand are presented in Figure 2.

Washing the ƒ-sand with dodecyl glucoside caused a slight and immediate reduction in the amount of MO_2.1. The loss was not as significant as that observed with the sodium dodecyl sulfate, and for 0.01 and 0.001 M the mass of MO_2.1 remained above 80% of the initial value. The results are presented in Figure 3.

Based on the results from part one, the 0.01 M concentration of sodium dodecyl sulfate was excluded from further investigation. The treatment results of functional sand columns with sodium dodecyl sulfate, 0.001 and 0.0005 M, are presented in Figure 4 and demonstrated that the surfactant was not effective at separating the E. coli from the ƒ-sand. In the first treatment run, when the synthetic water was passed through the ƒ-sand column, between 60% and 70% of the E. coli was removed. Further treatment results did not show any reduction in E. coli and even increased the turbidity of the water.

The dodecyl glucoside was effective at separating the E. coli from the ƒ-sand without affecting the capacity of the ƒ-sand to remove E. coli from water. The 0.01 M dodecyl glucoside was the most consistently effective, as shown in Figure 5. To confirm the use of UV-visible spectroscopy in bacterial estimation, colony counts were performed on the ƒ-sand columns treated with 0.01 M dodecyl glucoside. For the first treatment, the percentage reduction in the ƒ-sand columns determined by colony counting was 12% greater than that determined by UV-visible spectroscopy. For the second, third and fourth treatments, the difference in the values was less than 4%. When testing by UV-visible spectroscopy, the absorbance includes both viable and inhibited cells, whereas with cultivatable counts, only the living cells are recorded, indicating that UV-visible spectroscopy underestimates the amount of bacterial reduction. In this study, the absorbance between 0.1 and 0.001 at 600 nm was comparable to the cultivatable count taken from the solutions. However, for concentrations below 1 × 10⁵ CFU/mL, colony counting only should be performed.

To confirm that the dodecyl glucoside was effective at separating the E. coli from the ƒ-sand, three further columns, A1, A2 and A3, were prepared and are presented in Figure 6. For A1, the column was washed with 0.01 M dodecyl glucoside after each E. coli suspension, and each treatment retains the efficiency of the first run. Two synthetic water treatments were undertaken in succession for A2 and three for A3 without exposure to dodecyl glucoside between synthetic water treatments. For A2, the column was washed with dodecyl glucoside after the second treatment, and the restoration of the column is apparent on the third treatment. A fourth synthetic treatment was immediately run through the column and the efficiency was reduced. For A3, the column was run three times before it was washed with the dodecyl glucoside, and by the third treatment the E. coli removal dropped to below 10%. After washing with the dodecyl glucoside, the ƒ-sand regained its initial efficacy.

Dodecyl glucoside is a surfactant that is widely used in commercial detergents and can be readily sourced in local supermarkets. The efficacy of a detergent mixture reported to contain between 10 and 30% v/v dodecyl glucoside was tested on the ƒ-sand columns. 1% and 2% solutions were prepared, which corresponded to approximately 0.25 and 0.57 M dodecyl glucoside, respectively, and were compared against the 0.01 M v/v dodecyl glucoside solution as shown in Figure 7. The 2% solution was comparable to the 0.01 M dodecyl glucoside, whereas the 1% solution exhibited a loss of efficacy for each run. This result further confirms the use of dodecyl glucoside as an effective treatment for the removal of E. coli from ƒ-sand.

The attachment of E. coli to the ƒ-sand was clearly visible when viewed on the live cell microscope, as presented in Figure 8. After washing with the 0.01 M dodecyl glucoside, the E. coli was no longer visible on the ƒ-sand. After further exposure to the bacterial solution, the ƒ-sand attached and inhibited more E. coli. For each image, the left-hand image corresponds to the superimposed fluorescence-photo image, and the right-hand image is the fluorescence image only. The untreated sand and unused ƒ-sand did not display any fluorescence.

MO_2.1 is a potent inhibitor of a range of bacterial and viral organisms (Broin et al. 2002). The key mechanism of action of MO_2.1 involves the interference and absorbance of the polypeptide through the hydrophobic regions of bacterial phospholipid membranes, reported in E. coli to cause lysis and cell death (Suarez et al. 2005; Shebek et al. 2015). When MO_2.1 is attached to ƒ-sand, the bacterial attachment prevents any further bacterial attachment, and the aim of this study was to develop a method of renewing the ƒ-sand. Kwaambwa et al. (2010) examined the attachment of the coagulant protein to the oxide layer on an III face of a cut silicon crystal, reporting that saturation occurred at 5.5 mg/m². Rinsing with water or exposing the surface to a 2 × 10⁻³ M sodium dodecyl sulfate solution did not cause the protein to separate from the silica surface. In a similar experiment, the coagulant protein was shown to adsorb onto the surface of aluminium oxide (Kwaambwa et al. 2015). However, in that study, the coagulant protein could be removed by rinsing with water after exposure to concentrations of sodium dodecyl sulfate below the critical micelle concentration (Kwaambwa et al. 2015). Additionally, washing the surface with a 9 × 10⁻³ M solution of the cationic surfactant cetyltrimethylammonium bromide caused all of the protein to be removed (Kwaambwa et al. 2015).

In our study, the ƒ-sand sample washed with a sodium dodecyl sulfate solution above the critical micelle concentration had an 80% reduction in the amount of MO_2.1 attached to the sand. The next sample was just below this and also showed a loss of MO_2.1. All of the samples were exposed to concentrations above the critical aggregation concentration. It is likely that the 0.01 M sodium dodecyl sulfate was able to overcome the electrostatic adsorption of most of the MO_2.1 polymers; however, some strongly bonded polypeptides remained. The third sample, which was below the critical micelle concentration but still above the critical aggregation concentration, did not exhibit a loss of MO_2.1. Kwaambwa et al. (2010) reported that the 2 × 10⁻³ M sodium dodecyl sulfate solution exposed to a cut silicon crystal co-adsorbed onto the silica surface with a ratio of four molecules per one M. oleifera protein molecule.

When the 0.001 and 0.0005 M sodium dodecyl sulfate were run through the ƒ-sand column, it became apparent that the surfactant did not affect the attachment of the E. coli. Here, the formation of surfactant-MO_2.1 complexes and the solubilising of organic contaminants previously attached to the sand surface contribute to the increase in turbidity of the eluting solutions. The increase in turbidity of the outlet solutions was the most obvious on the first surfactant wash, where the eluting solution was visibly turbid.

Studies examining the effect of dodecyl glucoside on MO_2.1 have not been undertaken previously; however, reports of other non-ionic surfactants were encouraging. While all of the ƒ-sand samples washed with dodecyl glucoside showed an immediate loss of MO_2.1, they remained stable around 80% for the subsequent washes. When the dodecyl glucoside solutions were run through the ƒ-sand column, the high percentage reduction of bacteria for both 0.01 and 0.001 M dodecyl glucoside was unsurprising, as the destabilising mechanism is similar to that which occurs when phospholipid membranes are solubilised by detergents (Helenius & Simons 1975; le Maire et al. 2000). Bacterial membranes are made of a complex and dynamic lipid bilayer (Seddon et al. 2004) and gram negative bacteria possess a high density of anionic lipids (Shebek et al. 2015). The use of detergents to solubilise bacterial lipid membranes has been thoroughly investigated and covered by a three-stage hypothesis addressed by le Maire et al. (2000). In stage one, the free detergent molecules distribute into the phospholipid membrane. In stage two, the phospholipid membranes co-exist at a thermodynamic equilibrium comprised of phospholipids and detergent. Eventually, the detergent–detergent interactions destabilize the membrane structure, causing it to fragment. This leads to stage three, where, upon exposure to more detergent, the phospholipids become fully solubilised into the detergent micelles. In our study it is assumed that this three-stage phenomenon is what is causing the E. coli to separate from the MO_2.1. As the E. coli membrane was already disrupted by the MO_2.1, the hydrophobic phospholipids would readily solubilise into the dodecyl glucoside micelles and be removed from the ƒ-sand column, leaving behind the MO_2.1. Since the dodecyl glucoside did not interfere with the attachment of the majority of the MO_2.1, this method allowed the reuse of the ƒ-sand.

The amount of E. coli which the ƒ-sand removed varied significantly between experimental runs. In some cases, it was 99%; however, in other runs it was as low as 57%. This was likely due to the sensitivity of the spectroscopy as a measurement method. When it was compared with colony counting, it was observed that the difference between the two methods was greater when the ƒ-sand had lower efficacy, and the spectroscopy method underestimated the amount of bacteria removed when compared with direct colony counting. It was still a useful measurement for this type of study, as it was a quick way of determining whether the system could be regenerated. For further scale-up studies, colony counting or similar methods should be employed.

Bacterial contamination of water is a significant problem and many people rely on untreated water for drinking purposes. MO_2.1, found within the seed of Moringa oleifera, can be readily adsorbed onto the surface of silica sand particles, which can then be used for the attachment and inhibition of bacteria such as E. coli in contaminated water. The attached bacteria prevent any further attachment and the ƒ-sand is no longer effective for water treatment. The main findings from this study are as follows. 0.01 M dodecyl glucoside, a non-ionic surfactant, is effective at separating the E. coli from the MO_2.1, allowing the reuse of the ƒ-sand. Sodium dodecyl sulfate, an anionic surfactant, causes the MO_2.1 to separate from the sand surface. A commercially available detergent containing 10–30% dodecyl glucoside was also effective at separating the E. coli from the sand.

The results demonstrated in this work offer a potentially sustainable, reusable process to combat bacterial contamination of water.

The authors would like to thank Dr Agatha Labrinidis at Adelaide Microscopy for her help and advice in using the Nikon Ti Live Cell Microscope and Dr Quang Doan for his timely advice regarding chromatography and lipid membrane–detergent interactions. They would also like to acknowledge the generous donation of dodecyl glucoside from the Australian suppliers FPI Oceania and Ingredients plus.

This research was funded through the School of Chemical Engineering, The University of Adelaide.