Analysis of organic compounds in the rhizosphere soil of using LC-MS

(1)

INOM

EXAMENSARBETE KEMIVETENSKAP, AVANCERAD NIVÅ, 30 HP

STOCKHOLM SVERIGE 2018,

Analysis of organic compounds in the rhizosphere soil of Cyperus

using LC-MS rotundus

Development of methods for the attempted

identification of possible oviposition attractants of gravid Anopheles gambiae mosquitoes in

western Kenya

DANIEL RUOTSALAINEN

KTH

(2)

Abstract

The present study aimed to develop an extraction, sample preparation and analysis method for the rhizosphere soil of the grass Cyperus rotundus in an attempt to identify possible oviposition attractants of the Anopheles gambiae mosquito. Reversed Phase High Performance Liquid Chromatography (RP-HPLC) coupled online with Electrospray Ionization Mass spectrometry (ESI-MS) was used for the separation and detection of the extracted samples. Multiple extraction batches were done, altering parameters and techniques along the way. Ultrasound-Assisted Extraction (UAE) of the soil with a 50:50 (v/v) mixture of methanol and water was determined to be the most effective of the attempted methods. Purifying soil extracts with Solid Phase Extraction (SPE) showed to have an impact on the signal intensities in the chromatogram as well as reducing the intensity of system peaks eluting at the dead-time. This indicates that more polar compounds were removed during the SPE purification. A comparison was done between an extraction of soil soaked in water for five days and soil extracted without prior wetting. A difference could be seen and was shown by comparing chromatograms and mass spectra from the two samples. Tandem MS experiments were done for multiple precursor ions in order to identify the compounds by comparison in databases Human Metabolome Database (HMDB), Metlin and Massbank, but no matches in the databases were found.

The tandem MS results were also used for comparison of consecutive chromatographic peaks with similar MS¹-spectra. An identification method should be developed before the method presented in this work is validated and optimized further.

(3)

Sammanfattning

Den presenterade studien hade som syfte att utveckla en extraktions-, provpreparerings- och analysmetod för jorden, i vilken gräset Cyperus rotundus växer, i ett försök att identifiera möjliga äggläggningsattraherande ämnen för myggan Anopheles gambiae. Högupplösande vätskekromatografi med hydrofob stationär fas (RP-HPLC, en.) med Masspektrometri med Elektrosprejjonisering (ESI-MS, en.) användes för separation och detektion av de extraherade proverna. Ett flertal extraktionssatser gjordes, där parametrar och tekniker ändrades under utvecklingen av arbetet. Ultraljudsassisterad extraktion (UAE) av jord i en 50:50 (v/v) blandning av metanol och vatten visades vara mest effektiv av de undersökta metoderna. Upprening av jordextraktioner med fastfasextraktion (SPE, en.) visades ha en positiv inverkan på signalintensiteten i kromatogrammen såväl som att reducera intensiteten av de systemtoppar som eluerar vid dödtiden. Detta indikerar att de poläraste ämnena har renats bort. En jämförelse gjordes mellan en extraktion av jord som legat i blöt i fem dagar och jord som extraherats utan att blötläggas. En skillnad kunde ses och visades genom att jämföra kromatogram och masspektra från de två proverna. Tandem-MS-experiment gjordes för ett flertal prekursorjoner med avsikt att identifiera ämnena genom jämförelse med databaserna Human Metabolome Database (HMDB), Metlin och Massbank, men inga överrensstämmande jämförelser kunde göras. Tandem-MS-resultaten användes även för att jämföra efter varandra följande kromatografiska toppar med liknande MS¹-spektrum. En identifieringsmetod bör utvecklas innan metoden som presenterats i detta arbete valideras och vidareutvecklas.

(4)

The worldwide efforts to eradicate malaria in the years 1958-1963 showed significant improvements in health gains, and parts of Africa and India were practically free of the disease. Due to changes in interest among concerned parties, the World Health Organization (WHO) formally abandoned the eradication of malaria as a goal in 1969. [1] Today, malaria is still an issue in sub-Saharan Africa, although a large decrease in infections has been reported over the past 15 years [2]. Malaria is caused by the parasite Plasmodium falciparum and is carried over between humans by certain species of mosquitoes when they blood feed. Some of the most prevalent malaria vectors are parts of the Anopheles gambiae species complex. [3] No real vaccine for malaria exists but certain strategies to control the infection from spreading have been proven effective. Pesticides and insect nets prevent mosquitoes to blood feed indoors but does not prevent a human from being bitten while outdoors. While this does hinder some gravid females from blood feeding, thus also hindering them from laying eggs, strategies which allows controlling of where eggs are laid are of importance.

The so called “attract and kill” strategy works by attracting gravid females to a controlled oviposition site to either exterminate the females or remove the eggs that are laid. Studies on oviposition cues are therefore of great interest. In addition, the present work aims to contribute to the global malaria research, with the goal of reducing the risk of infection to those vulnerable.

In 2013, a group of researchers found that water vapor acts strongly as a pre-oviposition attractant for An. gambiae sensu stricto (s.s.). This was done by comparing unfed female mosquitos and gravid female mosquitoes in a series of tests where the preference of moving towards high or low humidity was recorded. [4] In a study where oviposition behavior of certain mosquitoes was studied by infusing water with soil from a natural oviposition site in western Kenya, it was found that cedrol, a sesquiterpene alcohol, attracts gravid An. gambiae sensu lato (s.l.) mosquitoes [5]. Cedrol has also been found to be produced by fungi cultivated from rhizomes of the grass Cyperus rotundus gathered from a natural oviposition site in western Kenya. Headspace collection of volatiles above the fungi, identified as Fusarium fujikuroi and Fusarium falciforme, was analyzed by Gas Chromatography Mass Spectrometry (GC-MS) and a small amount of cedrol was detected. [6]

In the previously mentioned studies, where identification was needed, GC-MS was used due to suitability for analysis of volatile compounds. Scent related cues are without doubt the most important external stimuli for oviposition behavior although visual and environmental cues do play a role, but the distances at which cues act vary vastly [7]. The utilization of Liquid Chromatography (LC) would cover shorter olfactory cue distances of less volatile compounds as well as still being viable with Mass Spectrometry (MS) for identification. Soil is complex in its composition and contains many organic compounds in the vicinity to roots and plants. The presence of bacteria and fungi in the soil produces metabolites, which are related to odor-mediated behavior of mosquitoes, and differs when environmental factors, such as temperature, humidity and moisture content changes. [7, 8] Hence, the aim of the present work was to develop a liquid chromatographic and mass spectrometric method for the separation and identification of organic compounds in rhizosphere soil of Cyperus rotundus from a natural oviposition site of the An. Gambiae mosquito, including sample preparation.

(6)

1.2 Soil extraction

For the analysis of organics in soil, an effective and robust extraction method is required due to the often complex sample matrix. Several generally accepted methods do exist but vary in simplicity and instrumental requirements. In a study of different extraction procedures of water-extractable organics in soils, a high extraction yield was shown for a method where soil was placed in an heated and pressurized extraction cell with water [9]. For analysis of certain compounds, an extraction procedure can be very specific. An example of this is the quantification of the carcinogenic compound ptaquiloside in soil, done by Jensen et al., where ammonium acetate was used to derivatize the analyte during the extraction process although the extraction itself was done with methanol. In the study, isolated ptaquiloside was used to spike soil to a known concentration prior to analysis for proof of concept. [10] The same compound was analyzed by Zaccone, C., et al. in 2014, but was instead extracted straight from the bracken fern and the soil surrounding it. For the soil extraction, ultrasound-assisted extraction (UAE) was used with water as the extracting solvent. [11]

No studies on metabolites in soil have been found, but utilizing sonication in extraction has been widely used in the analysis of for example plants and food products. Hence, this technique was chosen for the present work.

UAE has several features that make it a valuable technique for extracting organics from soil, plant parts, food products and many other sources. A review article from 2016 describes the concept in detail with applications for extraction of food and natural products. Possible extraction mechanisms are described and some examples follow. The first is referred to as fragmentation, where raw material is fragmented due to inter-particle collisions and shockwaves from cavitation bubble collapses. Cavitation bubble implosions could also cause an erosion effect on the surface of for example leaves. A phenomenon referred to as the sonocapillary effect can occur, which increases the velocity and depth of liquid penetration into pores of certain materials. Lastly, a combination of effects is discussed and it is claimed to be the most probable scenario. The choice of solvent should primarily be based on the solubility of analytes in the sample matrix. For the cavitation phenomenon as an extraction mechanism, low vapor pressure and low viscosity is preferred although temperature will affect these parameters as well. [12] However, the choice of solvent should also be based on its environmental impact and toxicity, to be sustainable.

1.3 High performance liquid chromatography

High performance liquid chromatography (HPLC) is an analytical separation technique where the components of a sample are separated from each other in an order based on the degree to which they interact with the stationary phase of the separation column. The interaction with the stationary phase is also governed by the choice of mobile phase, i.e. the liquid used to transport the sample through the HPLC system. Many different HPLC techniques are available and the choice will depend on what kind of analytes are being separated, although reversed phase (RP)-HPLC is the most commonly used technique for the separation of small organic molecules. The system consists of a high pressure pump that first runs the mobile phase through a gradient valve that controls the composition of the mobile phase. A valve switch connected to the injector and a sample loop follows.

After the sample is injected with a syringe into the injector, turning the valve switch allows the

(7)

and lastly goes to a waste container. A schematic representation of the HPLC system is presented in Figure 1. [13]

Figure 1. Schematic representation of an HPLC system.

1.3.1 Reversed phase HPLC

In RP-HPLC, the column consists of a non-polar organic stationary phase attached to the surface of a support material of silica. Examples of stationary phases in RP-HPLC are octadecyl, octyl and phenyl.

The stationary phase examples can be seen in Figure 2. The polarities of analytes, stationary phase and mobile phase determine the extent of the separation. The mobile phase consists of a non-polar or weakly polar solvent and a polar solvent. The polarity, or more specifically the elution strength, of the mobile phase can thus be chosen. Analytes with high affinity for the column stationary phase will be retained in the column, allowing for more polar analytes to elute faster. The mobile phase elution strength will determine how fast certain analytes are eluted. A less polar mobile phase mainly elutes non-polar analytes while a more polar mobile phase will elute both polar and non-polar analytes.

[13]

Figure 2. Examples of reversed phase liquid chromatography stationary phases: Octadecyl, octyl and phenyl. Increasing retention of non-polar compounds to the left, increasing polarity to the right.

1.3.2 Detection

Common HPLC detectors are ultraviolet light (UV) absorbance and mass spectrometry (MS). UV detection can be direct and indirect. Direct UV detection relies on the analyte to be UV active, while indirect UV detection instead relies on a UV absorbing component of the solvent that elutes with the solvent. Detection is achieved as the analyte passes through the flow cell and either absorbs UV light in direct detection or lowers the absorbance of the solvent in indirect detection. The collected data is time resolved and yields a chromatogram. Mass spectrometry is a method of analysis in its own but is often used as a detector for HPLC. In MS, the mass-to-charge ratio (m/z) of a given ion is detected.

Depending on the MS-technique, the ions are either molecular ions or fragments of molecules, and sometimes a combination of both. It is the amount of ions that hits the detector of the MS instrument that is being measured when the mass spectrometer produces a chromatogram, called the total ion current (TIC). [13]

(8)

1.4 Mass spectrometry

A mass spectrometer consists of an ion source, a mass analyzer and an ion detector. First, analytes are ionized by the ion source. Analytes can be solid, liquid or gaseous and this will determine what instrument is suitable for the analysis. Electron ionization (EI), chemical ionization (CI), matrix-assisted laser desorption/ionization (MALDI) and electrospray ionization (ESI) are examples of sources. In the ion source, analytes are ionized from their original neutral state. The ions can be either positive or negative depending on which mode the MS is run in. MALDI and ESI are soft ionization techniques where molecular ions are formed, whereas EI is a hard ionization technique that transfers more energy to the analytes, causing them to fragment into multiple different ions.

The ion source is followed by the mass analyzer, which sorts the ions according to m/z. This is generally done by transport through a magnetic or electric field. Quadrupole mass filters, ion traps and time-of-flight (TOF) analyzers are examples of mass analyzers. A mass spectrometer also requires vacuum to work. It operates under vacuum to prevent gas molecules to collide or react with analytes in the mass analyzer and the ion detector. Ionization can take place at both vacuum and atmospheric pressure, depending on the ion source. Lastly, the ions are detected when ions pass by or hit a surface which records the induced charge or produced current. The recorded signals produce a mass spectrum. [14]

Tandem mass spectrometry is often utilized when identification is desired and the primary ion source causes soft ionization. Soft ionization techniques can cause some fragmentation of the analytes but seldom enough for identification. In tandem MS, a precursor ion from the primary ion source is selected, and other ions are filtered out by the mass analyzer. The precursor ion is then fragmented further. The fragmentation causes more ions to be formed, which in turn can be sorted by a second mass analyzer and detected. However, some mass spectrometers do not require multiple mass analyzers to do tandem MS, e.g. in-source fragmentation, when the ionization is powerful enough to leave ions with internal energy sufficient for fragmentation after being sorted by the mass analyzer.

The fragmentation pattern can then be used to deduce structural information of the chosen precursor ion. [14]

1.4.1 Electrospray ionization triple quadrupole mass spectrometry (ESI-TQMS) In ESI, the liquid containing the analyte is introduced in the ion source through a capillary fitted with a needle at its end. The liquid exits the needle and it is put under high voltage as it is sprayed as a mist into the desolvation chamber. Having added for example acetic or formic acid to the solution prior to spraying aids ionization and increases the conductivity of the liquid. As the solvent evaporates and the droplet becomes smaller, increasing electrostatic forces within the droplet causes the droplet to explode, forming more droplets. This process is repeated until the ions are free from solvent and are in the vapor phase. Ions are introduced to the mass analyzer, in this study a triple quadrupole, where they are sorted by m/z. Only the first quadrupole is active in the MS1 mode. After the ions are separated they go straight to the detector. In tandem MS mode, an ion is chosen by its m/z and the remaining ions are, after ionization, filtered out by the first quadrupole.

The second quadrupole is a collision cell, where the chosen analyte collides with neutral molecules, often an inert gas like nitrogen or argon. Some of the kinetic energy of the collision is converted to

(9)

Figure 3. Schematic representation of the triple quadrupole mass analyzer.

2. Experimental 2.1 Material

Rhizosphere soil of the grass Cyperus rotondus, collected from Mbita, Kenya was used for all soil extractions. HPLC-grade methanol was acquired from Merck (Stockholm, Sweden) and HPLC-grade acetonitrile, formic acid and Acrodisc CR 25 mm syringe filters with 0.2 µm PTFE membranes were acquired from Sigma-Aldrich (Stockholm, Sweden). 25 mm syringe filters with 0.2 µm polypropylene membranes were acquired from VWR (Stockholm, Sweden). Bond elut C18 solid-phase extraction (SPE) cartridges (3 ml/200 mg) from Agilent were acquired from VWR (Stockholm, Sweden).

A 150 x 4.6 mm EKA C18 HPLC column with 5 µm particle size used for the separation. A LiChroCART C18 pre-column, acquired from Merck (Stockholm, Sweden), was used. All water used in this work was purified with a Millipore Sinergy 185 (Belford, MA, USA) water purification system operating at a resistivity of 18.2 MΩ and 25°C.

2.2 Instrumentation

A Dionex Ultimate 3000 (Sunnyvale, CA, USA) HPLC system with a pump unit and variable wavelength detector was used for the separation of samples. The HPLC system was coupled online with a Waters micromass Quattro micro ESI-MS (Milford, MA, USA) with a triple quadrupole mass analyzer.

In-house produced nitrogen gas was used as desolvation gas. Argon 5.0 LAB LINE (99.999 % (v/v) purity) was acquired from Strandmöllen (Ljungby, Sweden) to use as collision gas in tandem MS mode. A 35 kHz, 80/160 W, Bandelin sonorex RK100H (Berlin, Germany) ultrasound bath was used for sample preparation and mobile phase degassing. A Biofuge pico centrifuge by Kendro laboratory products (Osterode, Germany) and an Eppendorf Concentrator 5301 (Hamburg, Germany) were used for sample preparation.

2.3 Preparation protocols 2.3.1 Sample preparation

The organic content of the soil was extracted with UAE. 11.1 g of soil was weighed and was parted by hand into smaller pieces. Soil that was extracted without prior wetting was added to a 50 ml test tube with 15 ml of water and 15 ml of methanol. Soil that was wetted prior to extraction was placed in a 50 ml test tube with 15 ml of water and was allowed to sit for five days. After five days, 15 ml of methanol was added. The container was shaken mechanically for 1 minute. The test tube was placed in a sonication bath and sonicated for 30 minutes without external heating but was allowed to heat

(10)

up as the sonication took place. The slurry was then moved to 1.5 ml centrifuge vials and centrifuged at 10000 rpm for 15 minutes. The supernatants were filtered with 0.2 µm syringe filters and stored refrigerated in a 25 ml test tube in the.

Filtered extracts were purified with C18 SPE cartridges. The SPE purification was done with an SPE manifold put under rough vacuum. The cartridge was preconditioned with 5 ml of methanol and 5 ml of 0.1 % (v/v) formic acid in water before 4 ml of filtered extract was loaded. The retained analytes were washed with 2 ml of 0.1 % (v/v) formic acid in water, and eluted with 1.5 ml of acetonitrile. The samples were evaporated until dryness in a vacuum centrifuge operated in high vapor pressure mode and reconstituted in 10 % (v/v) of its original volume with methanol. The samples were sonicated for 10 minutes to ensure proper desolvation.

The above described sample preparation method will later be referred to as Batch 6. Remaining extraction and sample preparation protocols are showed in Appendix A: Sample preparation.

Samples are named in the following way:

𝑆𝑎𝑚𝑝𝑙𝑒 𝑛𝑎𝑚𝑒: 𝐵(𝑥)𝑆(𝑦)(𝑧)

where (x) = batch number, (y) = number of days soaked in water prior to extraction and (z) = extraction solvent (M for methanol, W for water and MW for a 50:50 (v/v) water/methanol mixture), i.e. B5S0W is the sample from Batch 5 extracted with water without prior wetting of the soil.

2.3.2 Mobile phase preparation

Two bottles of mobile phase was connected to the HPLC. Mobile phase A consisted of water with 0.1 % (v/v) formic acid. Mobile phase B consisted of 98 % (v/v) acetonitrile and 2 % (v/v) water with a total of 0.1 % (v/v) formic acid. Formic acid was added and the bottled was gently shaken by hand.

Both bottles where then degassed using sonication for 10 minutes. Other mobile phases used in the present work are described in 3.1 Method development.

(11)

2.4 Separation and detection methods

2.4.1 High performance liquid chromatography method

The HPLC was operated with the software Chromeleon 7. Prior to the first run of the day, the system was purged 3 minutes each with mobile phases A and B. Upon changing mobile phase bottles, the system was purged 2x3 minutes with each mobile phase. Final soil extracts were separated using a gradient elution method programmed as follows: hold at 37 % (v/v) mobile phase B for 2 minutes, linear increase of mobile phase B to 100 % (v/v) over 16 minutes, hold at 100 % (v/v) mobile phase B for 10 minutes and hold at 37 % (v/v) mobile phase B for 5 minutes. The flow rate was kept at 0.7 ml/min. Remaining gradients used in the method development are presented in 3.1 Method development. Due to instrumental limitations, the detection could be done either by UV spectrophotometry or mass spectrometry, and not both in series. Acquisition of UV- chromatograms was done at 254 nm. The hyphenation of the HPLC and the MS was done by connecting the HPLC column output to the ESI ion source input.

2.4.2 Mass spectrometry method

The ESI-MS was operated with the software MassLynx V4.1 and was run at either full scan mode or daughter scan mode. Full scan mode was used for acquisition of chromatograms. Analytes were ionized in positive mode. The used parameters are shown in Table 1.

Table 1. Acquisition parameters used for ESI-MS.

Capillary voltage: 3.2 kV Cone voltage: 30 V

Extractor voltage: 3V RF lens voltage: 0.2 V

Ion source temperature: 100 °C Desolvation temperature: 350 °C

Desolvation gas flow: 800 dm³/hr Cone gas flow: 50 dm³/hr

Scan time: 0.5 s Inter-scan delay: 0.1 s

Daughter scan mode was used for acquisition of tandem mass spectra. The software was programmed to fragment a set precursor ion over a certain period of time. The time period at which a certain ion eluted was deduced from a full scan acquisition of the same sample. For daughter scan, the same operating parameters were used as for full scan. Precursor ions were fragmented with a collision energy of 30 eV.

(12)

3. Results

3.1 Method development 3.1.1 Separation

Initial trials of extraction and sample preparation methods were performed in several variations to determine the direction, in which to go upon further improving the methods. Using methanol and water as extraction solvents, extraction by ultrasound was compared with extracting by soaking the soil in the extraction solvent for 20 hours. The amount of soil in a certain volume of extraction solvent was chosen arbitrarily but was roughly the same in all samples prepared at this stage. After the extraction, the liquid was separated from the soil through centrifugation and filtration with 0.2 µm syringe filters, followed by vacuum centrifugation of the samples until the volume was reduced by slightly more than half for samples extracted with water and to a fifth for samples extracted in methanol. Separation was attempted with isocratic separation methods using water (mobile phase A) and methanol (mobile phase B) as mobile phases with UV detection at 254 nm.

Tested mobile phase ratios were 100, 90, 80 and 70 % (v/v) mobile phase B at a flowrate of 1 ml/min.

Most of the sample analytes co-eluted early in the separation and the highest signal intensities were found for the UAE sample with methanol as the extraction solvent. Chromatograms of analytes eluted with 90 % (v/v) and 80 % (v/v) mobile phase B of the first sample extracted in methanol by sonication are shown in Figure 4.

Figure 4. Chromatograms of soil sample B1S5M, extracted with methanol by sonication and concentrated. Isocratic elution with a) 90:10 (v/v) methanol/water and b) 80:20 (v/v) methanol/water. Detection by UV at 254 nm.

(13)

Isocratic elution was abandoned because separation could not be achieved and a gradient elution method was tested instead. Detection was also changed to ESI-MS. Due to that, the mobile phases were changed to A: water with 5 mM ammonium acetate and B: 90:10 (v/v) acetonitrile (99.8 % (v/v) purity)/water with 5 mM ammonium acetate. The development was approached by trial and error and the first attempted gradient was Gradient 1 in Figure 5 below. The flow rate of the mobile phase was lowered to 0.7 ml/min to not overflow the electrospray ion source of the mass spectrometer.

Different gradients that have been used can be seen in Figure 5. MS parameters were set by direct infusion of the sample, mixed 1:10 with mobile phase B. The sample was run and the chromatograms with UV- and MS-detection are shown in Figure 6.

Figure 5. Used elution gradients with reference to % (v/v) mobile phase B and time in minutes.

Figure 6.Chromatogram of soil sample B1S5M, extracted with methanol by sonication and concentrated.

Acetonitrile/water ratio in mobile phase B was 90:10 (v/v) and the mobile phase additive was 5 mM ammonium acetate.

Eluted with Gradient 1, see Figure 5. a) Detection by UV at 254 nm, b) Detection by ESI-MS with parameters described in section 2.4.2.

(14)

A few attempts followed, altering hold times and ramp speed and Gradient 2 was found to be the most suitable. The chromatogram from this run with MS detection is shown in Figure 7. This gradient was then used as a basis for evaluation of the progress in development of a method for sample preparation.

Figure 7. Chromatogram of soil sample B1S5M, extracted with methanol by sonication and concentrated.

Acetonitrile/water ratio in mobile phase B was 90:10 (v/v) and the mobile phase additive was 5 mM ammonium acetate.

Eluted with Gradient 2, see Figure 5. Detection by ESI-MS with parameters described in section 2.4.2.

3.1.2 Sample preparation

Six batches of samples were prepared throughout the present work, changing parameters and techniques along the way. Sample preparation protocols can be found in Appendix A: Sample preparation. In the appendix, the sample preparation methods are referred to as Batch 1-6.

Separation of samples from Batch 1 is shown in 3.1.1 Separation.

Batch 2 sample separation was investigated using Gradient 2, modified with a slightly faster ramp, and the mobile phases A: water with 0.1 % (v/v) formic acid and B: 90:10 (v/v) acetonitrile (95 % (v/v) purity)/water with 0.1 % (v/v) formic acid. The chromatogram did not show any interesting peaks and the 95 % (v/v) purity of acetonitrile was assumed to be the problem. This was the only attempt to optimize the method with less pure acetonitrile. A chromatogram from the separation attempt can be seen in Figure 8. The mobile phase was changed back to 99.8 % (v/v) purity acetonitrile with 5 mM ammonium acetate in both mobile phase reservoirs and separation was attempted again but did not show any noticeable improvements. Instead the sample was further concentrated five times and the precipitates that formed were removed by centrifuging the sample vial and separating the supernatant from the solids. The chromatogram of the further concentrated sample of Batch 2, eluted with Gradient 2, can be seen in Figure 9.

(15)

Figure 8. Chromatogram of soil sample B2S5M, extracted with methanol 23 hours of stirring and once concentrated.

Acetonitrile/water ratio in mobile phase B was 90:10 (v/v), acetonitrile purity was 95 % (v/v) and the mobile phase additive was 0.1 % (v/v) formic acid. Eluted with Gradient 2, see Figure 5, but with a slightly faster ramp. Detection by ESI-MS with parameters described in section 2.4.2

Figure 9. Chromatogram of soil sample B2S5M, extracted with methanol 23 hours of stirring and twice concentrated.

Acetonitrile/water ratio in mobile phase B was 90:10 (v/v) and the mobile phase additive was 5 mM ammonium acetate. Eluted with Gradient 2, see Figure 5. Detection by ESI-MS with parameters described in section 2.4.2.

The concentrated sample of Batch 2 was purified with SPE according to the protocol described in Appendix A. The separation of this sample was performed with Gradient 2 and the mobile phase containing ammonium acetate. The chromatogram is shown in Figure 10.

Figure 10. Chromatogram of soil sample B2S5M, extracted with methanol 23 hours of stirring, twice concentrated and purified by SPE. Acetonitrile/water ratio in mobile phase B was 90:10 (v/v) and the mobile phase additive was 5 mM ammonium acetate. Eluted with Gradient 2, see Figure 5. Detection by ESI-MS with parameters described in section 2.4.2.

(16)

Batch 3 was extracted using only water and purified by SPE. The results were not satisfactory due to that very few signals and of low intensity were detected and further improvement was not pursued.

Chromatogram from separation of the Batch 3 sample can be seen in Figure 11.

Figure 11. Chromatogram of soil sample B3S0W, extracted with water by sonication and purified by SPE.

Acetonitrile/water ratio in mobile phase B was 90:10 (v/v) and the mobile phase additive was 0.1 % (v/v) formic acid.

Batch 4 was prepared with the purpose of evaluating different SPE elution solvents. Methanol, acetonitrile and hexane were considered but when the separations had been compared, hexane was excluded from future experiments due to fewer and lower intensity signals being present. The difference between methanol and acetonitrile was slight, but acetonitrile showed most promise. A comparison between the two is shown in Figure 12.

Figure 12. Chromatogram of soil sample B4S0MW, extracted with a 50:50 (v/v) mixture of water/methanol by sonication,

(17)

In Batch 5, a final attempt of extracting with water was done as well as extracting with a 50:50 (v/v) water/methanol mixture. Extraction was performed after soaking the soil in water for five days as well as without soaking. This yielded four samples; non-soaked soil extracted with water, non-soaked with a water/methanol mixture, soil soaked for five days extracted with water and a water/methanol mixture. Samples were also volumetrically concentrated by a factor of ten after SPE purification. For this batch, several steps of the sample preparation process were analyzed by HPLC-MS; between filtering and SPE cleanup, the sample solvent that was not retained by the SPE cartridge, the SPE purified sample and the concentrated SPE purified sample. Samples extracted with water once again showed few peaks with low signal intensities and a chromatogram from one of the concentrated, SPE phase, water extracted samples is shown in Figure 13. The chromatograms of three out of the four preparation steps of the methanol and water extracted sample that was soaked for five days are shown in Figure 14. The filtered extract is not presented as it does not contain any peaks.

Figure 13. Chromatogram of soil sample B5S5W, extracted with water, purified with SPE and concentrated.

(18)

Figure 14. Chromatogram of soil sample B5S5MW, a) by SPE, non-retained analytes, b) purified with SPE and c) purified with SPE and concentrated. The sample was extracted with a 50:50 (v/v) mixture of water/methanol by sonication.

The gradient was then modified by changing the hold time at 100 % (v/v) B from 5 minutes to 10 minutes to observe if any analytes were retained in the column with the present method. An additional signal was seen at 26 minutes and the gradient method was adapted to this new longer hold time. The mobile phase was later modified by changing the ratio of acetonitrile and water in mobile phase B from 90:10 (v/v) to 95:5 (v/v) to increase the elution strength, in hopes of reducing the hold time at 100 % (v/v) mobile phase B. The gradient was changed so that the proportion of acetonitrile in the mixed mobile phase would increase with roughly the same pace as before, throughout the ramp of mobile phase B. The new gradient started with a hold at 38 % (v/v) mobile phase B for 2 minutes, linear increase of mobile phase B to 100 % (v/v) over 16 minutes, hold at 100 % (v/v) mobile phase B for 10 minutes and hold at 38 % (v/v) mobile phase B for 5 minutes.

(19)

3.2 Chromatographic separation

Here, the final steps of chromatographic method development are presented together with final chromatographic separation of samples from Batch 6. In 3.2, all samples are referring to SPE purified and tenfold volumetrically concentrated samples of Batch 6. These samples were run with a slightly changed mobile phase and gradient. The ratio of acetonitrile in mobile phase B was then increased to 98 % (v/v) and the gradient was adapted thereafter. The gradient, Gradient 3, started with a hold at 37 % (v/v) mobile phase B for 2 minutes, linear increase of mobile phase B to 100 % (v/v) over 16 minutes, hold at 100 % (v/v) mobile phase B for 10 minutes and hold at 37 % (v/v) mobile phase B for 5 minutes, i.e. the slope of the ramp was not altered. As the samples were first analyzed it was realized that the full 20 µl of the injection loop had not been injected throughout the present work up to this point. A few trials were done to check how big the difference was. The injection valve was until this point turned from the load position to the inject position and back very fast when injecting.

A 20 µl injection loop takes about 1.7 seconds to empty with a flow rate of 0.7 ml/min.

Chromatograms of the non-soaked sample with short and 2 second injection times are shown in Figure 15.

Figure 15. Chromatogram of soil sample B6S0MW, extracted with a 50:50 (v/v) mixture of water/methanol by sonication, purified with SPE and concentrated with a) a short injection and b) a 2 second injection. Acetonitrile/water ratio in mobile phase B was 98:2 (v/v) and the mobile phase additive was 0.1 % (v/v) formic acid. Eluted with Gradient 3, see Figure 5. Detection by ESI-MS with parameters described in section 2.4.2.

Higher signal-to-noise ratios than for previous samples were desired and background subtractions were deemed necessary. A background chromatogram was produced by injecting methanol. The background was subtracted from a chromatogram through MassLynx and a new chromatogram was produced. An example of a background chromatogram can be seen in Figure 16. The difference between before and after subtraction can be seen by comparing Figure 15 b) with Figure 17.

(20)

Figure 16: Chromatogram of a methanol background injection. Acetonitrile/water ratio in mobile phase B was 98:2 (v/v) and the mobile phase additive was 0.1 % (v/v) formic acid. Eluted with Gradient 3, see Figure 5. Detection by ESI-MS with parameters described in section 2.4.2.

3.2.1 Sample comparison

In the following section, the two samples B6S0MW and B6S5MW, the one extracted without prior wetting and the one soaked in water for five days are compared. Figure 17 shows a chromatogram of the separation of sample B6S0MW. Figure 18 shows a chromatogram of the separation of sample B6S0MW. The numbering has its reference in Figure 17. Signals numbered #a are present in Figure 18 but not in Figure 17. Signals present in Figure 17, but not in Figure 18 are left out from Figure 18.

Peak 1, 2 and 32 in both chromatograms are system peaks and are present even if no sample is injected.

Figure 17. Chromatogram of soil sample B6S0MW, extracted with a 50:50 (v/v) mixture of water/methanol by sonication, purified with SPE and concentrated. Acetonitrile/water ratio in mobile phase B was 98:2 (v/v) and the mobile phase additive was 0.1 % (v/v) formic acid. Eluted with Gradient 3, see Figure 5. Detection by ESI-MS with parameters described in section 2.4.2. Peaks are numbered in order of retention time.

(21)

Figure 18. Chromatogram of soil sample B6S5MW, extracted with a 50:50 (v/v) mixture of water/methanol by sonication, purified with SPE and concentrated. Acetonitrile/water ratio in mobile phase B was 98:2 (v/v) and the mobile phase additive was 0.1 % (v/v) formic acid. Eluted with Gradient 3, see Figure 5. Detection by ESI-MS with parameters described in section 2.4.2. Peaks are numbered in order of retention time and in relation to Figure 17. Peaks labeled #a are not present in Figure 17.

Peaks present in both chromatograms are, by their mass spectra, confirmed to peak-by-peak be the same compound, except the following:

• Peak 7 partially shows the same m/z signals for both samples.

• Peak 24 has a low intensity in sample B6S0MW and the mass spectrum shows no strong m/z signals.

• Peak 26 has a low intensity in sample B6S5MW and the mass spectrum shows no strong m/z signals.

• Peaks between 20 and 22 minutes + peak 31 show no strong m/z signals.

The mass spectrometric comparisons presented in the bullet list above are shown in Appendix C: MS comparison of Batch 6 separations.

Aside from confirming the similarities between the two samples, differences in peaks that appear in one but not the other can be seen. Peak 5 for B6S0MW and peaks 4a and 5 for B6S5MW are shown in Figure 19. Similarities between the mass spectrum of peak 5 in both samples can be seen, as for example the signals at 508, 485, 463 and 377 m/z. If peak 5 in B6S0MW is instead compared with peak 4a, new signals can be seen at 305, 391, 413 and 436 m/z for peak 4a, suggesting the presence of another compound eluting at a time close to that of peak 5.

(22)

Figure 19. MS¹-spectrum of a) peak 5 in Figure 17 of soil sample B6S0MW, b) peak 5 in Figure 18 of soil sample B6S5MW and c) peak 4a in Figure 18 of soil sample B6S5MW, extracted with a 50:50 (v/v) mixture of water/methanol by sonication, purified with SPE and concentrated. Acetonitrile/water ratio in mobile phase B was 98:2 (v/v) and the mobile phase additive was 0.1 % (v/v) formic acid. Eluted with Gradient 3, see Figure 5. Detection by ESI-MS with parameters described in section 2.4.2.

The repeatability of the separation method has been evaluated through runs of samples B6S0MW and B6S5MW multiple times over two days. The average retention time and relative standard deviation has been calculated on a peak-basis for the two samples. Relative intra-day standard deviations have been calculated for B6S0MW. Selected peaks, their retention times, average retention times, relative inter-day standard deviations and relative intra-day standard deviations for B6S0MW are shown in Table 2. Selected peaks, their retention times, average retention times and relative inter-day standard deviations for B6S5MW are shown in Table 3. All peaks, their retention times, average retention times, standard deviations and relative standard deviations subject to the repeatability evaluation are shown in Appendix B: Repeatability of separation method.

(23)

Table 2: Retention times, average retention times and relative inter-day standard deviations as well as relative intra-day standard deviations for day 1 and day 2 of selected signals from six separations of sample B6S0MW run over two separate days. Acetonitrile/water ratio in mobile phase B was 98:2 (v/v) and the mobile phase additive was 0.1 % (v/v) formic acid. Eluted with Gradient 3, see Figure 5. Detection by ESI-MS with parameters described in section 2.4.2.

Peak no.

Retention time, day 1 [min]

Retention times, day 2 [min]

Avg.

ret.

time [min]

%RSD inter- day

%RSD intra- day 1

%RSD intra- day 2

3 6.357 6.408 6.408 6.367 6.367 6.398 6.384 0.329 0.376 0.229

10 12.412 12.453 12.443 12.362 12.362 12.382 12.402 0.294 0.140 0.076 14 14.810 14.840 14.851 14.760 14.790 14.800 14.809 0.206 0.117 0.115 23 17.863 17.944 17.944 17.833 17.863 17.873 17.887 0.237 0.213 0.095 25 19.455 19.545 19.566 19.445 19.485 19.465 19.494 0.235 0.247 0.084

Table 3: Retention times, average retention times and relative inter-day standard deviations of selected signals from five separations of sample B6S5MW run over two separate days. Acetonitrile/water ratio in mobile phase B was 98:2 (v/v) and the mobile phase additive was 0.1 % (v/v) formic acid. Eluted with Gradient 3, see Figure 5. Detection by ESI-MS with parameters described in section 2.4.2.

Peak no.

Avg.

ret.

time [min]

%RSD inter- day

3 6.347 6.367 6.388 6.398 6.398 6.380 0.311 10 12.352 12.372 12.362 12.392 12.392 12.374 0.129 14 14.830 14.790 14.850 14.830 14.820 14.824 0.132 23 17.873 17.833 17.843 17.863 17.843 17.851 0.082 25 19.515 19.425 19.435 19.435 19.455 19.453 0.167

Data from analysis done during day 2 is incomplete due to instabilities in the baseline for all injections, including background injections, past 20 minutes. An example of this baseline instability is shown in Figure 28. Due to this, repeatability evaluations where done considering only retention times of signals up to 20 minutes. Excluding system peaks 1, 2 and 32 and peaks past 20 minutes, inter-day standard deviations throughout the chromatogram were found to vary between 0.066-0.967 % for B6S0MW and 0.082-0.802 % for B6S5MW.

(24)

Figure 20. Chromatogram of soil sample B6S0MW, extracted with a 50:50 (v/v) mixture of water/methanol by sonication, purified with SPE and concentrated. Unstable baseline past 20 minutes of separation. Acetonitrile/water ratio in mobile phase B was 98:2 (v/v) and the mobile phase additive was 0.1 % (v/v) formic acid. Eluted with Gradient 3, see Figure 5.

Detection by ESI-MS with parameters described in section 2.4.2. Peaks numbered by retention time and in relation to Figure 17.

(25)

3.3 Tandem mass spectrometry

For the tandem MS daughter scans that have been run on samples of Batch 6, MS/MS database searches have been made in the Human Metabolome database (HMDB) [15], Massbank [16] and Metlin [17] without success. The MS/MS-results were also used to show the differences between chromatographic peaks with similar MS¹-spectra. The comparisons were based on daughter scans of sample B6S5MW.

Mass spectra of peaks 17-19 are shown in Figure 21. The mass spectra are similar with signals at 279, 296, 318, 341 and 359 m/z but differ in the less abundant peaks at 141, 219 and 233 m/z in Figure 21 a), b) and c) respectively.

Figure 21. MS¹-spectrum of a) peak 17, eluting at 15.85 min, b) peak 18, eluting at 16.04 min and c) peak 19, eluting at 16.56 min from the chromatogram in Figure 18, of soil sample B6S5MW, extracted with a 50:50 (v/v) mixture of water/methanol by sonication, purified with SPE and concentrated. Acetonitrile/water ratio in mobile phase B was 98:2 (v/v) and the mobile phase additive was 0.1 % (v/v) formic acid. Eluted with Gradient 3, see Figure 5. Detection by ESI-MS with parameters described in section 2.4.2.

(26)

The MS/MS spectra of 296 m/z as precursor ion for peaks 17, 18 and 19 are shown in Figure 22. The general trend in presented mass spectra is similar, but intensities of groups of signals and within groups of signals differ. An example of differences is the one between signals at 55 and 57 m/z in respective mass spectra. In b) and c), 55 m/z is most abundant but the intensity of 57 m/z is 20 and 70 % of the 55 m/z signal respectively. In Figure 22 a), 57 m/z is ten times as abundant as 55 m/z.

Figure 22. MS²-spectrum of precursor ion 296 m/z of a) peak 17, eluting at 15.85 min, b) peak 18, eluting at 16.04 min and c) peak 19, eluting at 16.56 min from mass spectra shown in Figure 21, of soil sample B6S5MW, extracted with a 50:50 (v/v) mixture of water/methanol by sonication, purified with SPE and concentrated. Acetonitrile/water ratio in mobile phase B was 98:2 (v/v) and the mobile phase additive was 0.1 % (v/v) formic acid. Eluted with Gradient 3, see Figure 5. Collision energy: 30 eV, collision gas: argon. Detection by ESI-MS with parameters described in section 2.4.2.

(27)

4. Discussion

4.1 Method development

As both the sample preparation and the separation methods were developed throughout the presented work there was no natural way of confirming their validity individually. As the sample preparation progressed, this could only be confirmed once a separation method was found that effectively separated the components of the sample. The separation method could only be confirmed once an extraction method that gave a pure enough sample with high enough concentrations had been developed. Luckily, the first extraction batch was deemed good enough. A lot of work was then done to find an adequate separation method. Isocratic elution was abandoned early and the arbitrarily chosen gradient method Gradient 1, Figure 5, worked well. Samples of Batch 1 were extracted with methanol and water respectively to investigate if water was suitable, as it would be an environmentally and economically friendly alternative to an organic solvent like methanol. Methanol however is widely used for soil extraction in the literature, and was chosen of that reason. All samples of Batch 1 except the one extracted by sonication in methanol were obsolete in comparison, but it could not yet be concluded that water was not viable as an extraction solvent. This is because the concentrations were not the same of the UAE samples with water and methanol as the extraction solvents. The mass spectrometer was unavailable for the initial experiments in this study and detection was done with UV. Due to a certain learning curve with the mass spectrometric software, detection parameters were quite sparsely optimized once the MS became available. Proper optimization could have altered the progression of method development vastly, but cannot be said for sure. Detection parameters were still considered good enough as several clear signals were detected and an early comparison with UV at 254 nm showed similar chromatograms.

Extraction of Batch 2 was ineffective but the utilization of SPE purification was successful. The purpose of Batch 2 therefore indirectly became to show the effect of SPE purification on samples extracted from a soil matrix. It was thereafter used on all extraction batches, to in theory reduce the amount of ions and remove the most polar components of the sample. The SPE purification both increased the concentration of the samples and isolated the analytes from the soil matrix. Samples of Batch 3 showed unsatisfactory results due to few signals and low intensities, and it was originally thought that choosing water as the extracting solvent was the reason. But this conclusion could not be drawn as no comparison to another solvent was done in the same extraction batch. However, a comparison like this was done in Batch 5, and all parameters except the extracting solvent were kept constant. The comparison is presented in Figure 13 and Figure 14 c). Here, water is shown to be the inferior extracting solvent, when compared to extracting with a 50:50 (v/v) mixture of water and methanol. Herrera-Varela et al. [18] found that Anopheles gambiae s.l. was more likely to lay their eggs in water from Lake Victoria infused with soil for six days compared with the same water infused with soil for two days. This suggests that microbial or fungal activity gives rise to a more attractive oviposition site, as an autoclaved six day infusion did not give rise to the same effect. Hence, comparing extractions of soil prior to and after five days of soaking in water was done to mimic the seen effect. The reason for using a water/methanol mixture when extracting was because the water in which the soil had been soaked was assumed to contain compounds of interest and was therefore not discarded. Instead it was included in the mixture when extracting with methanol.

(28)

In Batch 4 further improvement of the SPE purification method was sought. The difference between eluting the contents of the SPE cartridge with methanol or with acetonitrile was minor and the choice would not be considered to have had a large impact on the end results of the present work.

Because hexane was included as a potential eluent candidate and it could not be injected in the HPLC at present conditions due to its immiscibility with water, all elutions were evaporated to dryness and reconstituted in methanol. This in turn gave better control over parameters that were meant to be fixed for this experiment; the volumetric concentration factor of the respective samples throughout the process, as well as the sample solvent when injecting. Conclusions about possible loss of sample during SPE purification and drying cannot be drawn without the use of for example an internal standard that can be quantified before and after sample workup, which was not done in the presented work. The evaporation to dryness and reconstitution of sample was however used in Batch 5 and Batch 6 as means to control the volumetric concentrating factor. What could have been done to evaluate possible sample loss when drying the sample, could be to compare a sample that was evaporated to dryness and reconstituted with a sample that was evaporated to a given volume, where the volumetric concentration factor would have been controlled by weighing the sample vials to ensure the same amount of solvent.

4.2 Chromatographic separation

Changing the percentage of acetonitrile in mobile phase B from 90 to 95 % (v/v) and from 95 to 98 % (v/v) in order to increase the elution strength, with the purpose of reducing the required hold time at 100 % (v/v) B due to late eluting peaks, did help with eluting the last analytes. However, the hold time was kept at 10 minutes, to ensure no carry-over to the next injection occurred. Ultimately, the separation gradient still has room for improvement, with final optimization of starting conditions, ramp speed and hold times, to minimize time of analysis and get baseline separation throughout the chromatogram.

The full 20 µl of the sample loop was not injected throughout most of the method development. This was discovered due to a mistake, where the injection valve was kept in the inject position for a slightly longer time than for other injections. It could be said that the method development until this point had accidentally been optimized to be more sensitive to sample amount than it had to be.

Once a full injection loop was injected the intensities of all signals increased heavily. Peak 31 in Figure 23 could probably be of half its height, and still contain enough information for both identification and quantification. With this said, decreasing the sample loop size to 5 or 10 µl would be a good option for further method optimization, with the current sample concentration. The amount of sample would decrease, but the injection band would be narrower. In theory, the effect of longitudinal diffusion would not be as wide as compared with a wider injection band. The narrower injection band thus promotes more effective separation.

Background subtraction was done with an injection of methanol, primarily to remove the ions formed by the mobile phase, from the chromatogram. This was done successfully to enhance the chromatogram resolution and increase relative peak intensity.

(29)

4.2.1 Sample comparison

In the comparison between samples B6S0MW and B6S5MW, both a 2 second injection and a background subtraction were done to enhance the chromatograms. By visual comparison of Figure 17 and Figure 18, it is clear that they are different but most signals elute at similar retention times in both chromatograms. By this, it can be said that most sample components are probably not affected by soaking the soil in water for five days prior to extraction and analysis. However, some peaks, like peak 4a in Figure 18, are not present in the soil that was not soaked prior to extraction and analysis.

This suggests that new compounds appear throughout the five days of soaking in water. It is therefore assumed that bacterial or fungal activity causes this but cannot be concluded by the experiments conducted in the present work. To confirm this, an experiment could be designed to compare unaltered soil with autoclaved soil soaked in water for five days and see if the differences between chromatograms remain. However, it could also be that soaking the soil for five days, allows water to penetrate pores and crevices in the soil more effectively. This would increase the extraction efficiency and cause more peaks to be detected. Solvents other than methanol and water have not been used for extraction in the present work and could be worth investigating further.

As the evaluation of the repeatability of the separation method was done, separations conducted during day 2, showed instabilities in the baseline after 20 minutes, as shown in Figure 20. It was problematic, but does not hinder one from evaluating the retention times of peaks before the instability. Inter-day comparisons showed relative standard deviations between 0.066-0.967 % for both samples. Intra-day evaluations were only done for B6S0MW because the three injections were done each day, where two and three injections were done during the two days for the 5-day sample.

Differences in inter- and intra-day deviations vary between peaks, but for example for peak 14 in Table 2 it can be seen that the relative standard deviation decreases from 0.206 % to 0.117 % and 0.115 % for both days when doing the evaluation within a day instead of between two days.

4.3 Tandem mass spectrometry

The tandem mass spectrometric analyses that were conducted were done with the goal of producing mass spectra of as many precursor ions as possible in order to have more options in the database searches. But because the database searches did not give any hits, the spectra were instead used to principally show how the tandem MS spectra can be used to compare signals with similar MS¹-spectra. As it was not the main purpose of the MS/MS analyses from the start, the comparisons could not be done throughout the separation of the samples. However, it does show how the information from a tandem MS analysis can be used for comparison, without knowing the identity of the target analytes. The peaks of the comparison in Figure 21 and Figure 22, were chosen because they elute right after each other and have similar MS¹-spectra. The similarity in the MS¹-spectra could either be because the compounds are structurally very similar or that there is co-elution of something of high abundance throughout the peaks. As identification was not possible, this conclusion cannot be drawn, but comparing the MS²-spectra suggests that they are in fact different.

This is suggested by the difference in the spectra, both at low m/z-values where alkyl chains are typically seen, but also in the slightly higher range of m/z-values where for example 149 m/z is present in Figure 22 b) and c) and 183 m/z is present in Figure 22 a) and c).

Optimization of the operating parameters for tandem MS were done very sparsely. Two collision energies were tested, but as the experiments were done on different precursor ions, with hopes of identification of compounds in the soil samples, the spectra of the two energies are not comparable.

(30)

5. Conclusions

From the present work it can be concluded that the optimized extraction method works well and that SPE purification has an impact on the quality of the data that is produced from the analysis. It has also been shown that a 50:50 (v/v) mixture of water and methanol works better than just water for the ultrasound-assisted extraction of soil samples in this application. The present liquid chromatographic separation method works well but shows some co-elution of certain peaks. The separation method can be improved to further reduce run time and reduce co-elution. Optimizing sample concentration and injection volume should also be considered. The current tandem MS method is, by the looks of mass spectra, well-functioning but it is hard to tell without identification of the compounds present in the extracted soil sample. The identification method should be a first priority in pursuit of further development of the present method. Identification should be in place before the method is validated from start to finish.

(31)

References

1. Attaran, A., Where did it all go wrong? Nature, 2004. 430(7002): p. 932.

2. Bhatt, S., D.J. Weiss, E. Cameron, D. Bisanzio, B. Mappin, U. Dalrymple, K.E. Battle, C.L.

Moyes, A. Henry, P.A. Eckhoff, et al., The effect of malaria control on Plasmodium falciparum in Africa between 2000 and 2015. Nature, 2015. 526(7572): p. 207.

3. WHO: World Malaria Report 2013. World Health Organization, Geneva [cited 2018 02 22];

Available from:

http://www.who.int/malaria/publications/world_malaria_report_2013/report/en/.

4. Okal, M.N., B. Francis, M. Herrera-Varela, U. Fillinger and S.W. Lindsay, Water vapour is a pre-oviposition attractant for the malaria vector Anopheles gambiae sensu stricto. Malaria Journal, 2013. 12(1): p. 365-372.

5. Lindh, J.M., M.N. Okal, M. Herrera-Varela, A.K. Borg-Karlson, B. Torto, S.W. Lindsay and U.

Fillinger, Discovery of an oviposition attractant for gravid malaria vectors of the Anopheles gambiae species complex. Malaria Journal, 2015. 14(1): p. 119-130.

6. Eneh, L.K., H. Saijo, A.K. Borg-Karlson, J.M. Lindh and G.K. Rajarao, Cedrol, a malaria mosquito oviposition attractant is produced by fungi isolated from rhizomes of the grass Cyperus rotundus. Malaria Journal, 2016. 15(1): p. 478-481.

7. Takken, W. and B.G.J. Knols, Odor-mediated behavior of Afrotropical malaria mosquitoes, in Annual Review of Entomology. 1999. p. 131-157.

8. Effmert, U., J. Kalderás, R. Warnke and B. Piechulla, Volatile Mediated Interactions Between Bacteria and Fungi in the Soil. Journal of Chemical Ecology, 2012. 38(6): p. 665-703.

9. Guigue, J., O. Mathieu, J. Lévêque, S. Mounier, R. Laffont, P.A. Maron, N. Navarro, C.

Chateau, P. Amiotte-Suchet and Y. Lucas, A comparison of extraction procedures for water- extractable organic matter in soils. European Journal of Soil Science, 2014. 65(4): p. 520-530.

10. Jensen, P.H., O.S. Jacobsen, H.C.B. Hansen and R.K. Juhler, Quantification of ptaquiloside and pterosin B in soil and groundwater using liquid chromatography-tandem mass spectrometry (LC-MS/MS). Journal of Agricultural and Food Chemistry, 2008. 56(21): p. 9848-9854.

11. Zaccone, C., I. Cavoski, R. Costi, G. Sarais, P. Caboni, A. Traversa and T.M. Miano, Ptaquiloside in Pteridium aquilinum subsp. aquilinum and corresponding soils from the South of Italy:

Influence of physical and chemical features of soils on its occurrence. Science of the Total Environment, 2014. 496: p. 365-372.

12. Chemat, F., N. Rombaut, A.G. Sicaire, A. Meullemiestre, A.S. Fabiano-Tixier and M. Abert- Vian, Ultrasound assisted extraction of food and natural products. Mechanisms, techniques, combinations, protocols and applications. A review. Ultrasonics Sonochemistry, 2017. 34: p.

540-560.

13. Harris, D.C., Quantitative chemical analysis. 8. ed. 2010, New York: Freeman.

14. Hoffmann, E.d., Mass spectrometry: principles and applications. 3. ed, ed. V. Stroobant. 2007, Chichester: Wiley.

15. Human Metabolome database. [cited 2018 06 26]; Available from: http://www.hmdb.ca/.

16. MassBank MS database. [cited 2018 06 26]; Available from:

https://massbank.eu/MassBank/.

17. Metlin MS database. [cited 2018 06 26]; Available from: https://metlin.scripps.edu/.

18. Herrera-Varela, M., J. Lindh, S.W. Lindsay and U. Fillinger, Habitat discrimination by gravid Anopheles gambiae sensu lato - A push-pull system. Malaria Journal, 2014. 13(1): p. 133-147.