Extração de metabolitos

Bioresource Technology 124 (2012) 311–320

Contents lists available at SciVerse ScienceDirect

Bioresource Technology

journal homepage: www.elsevier .com/locate /bior tech

Selection and optimisation of a method for efficient metabolites extractionfrom microalgae

Benoît Serive a, Raymond Kaas a,⇑, Jean-Baptiste Bérard a, Virginie Pasquet b, Laurent Picot c,Jean-Paul Cadoret a

a IFREMER, Laboratoire de Physiologie et Biotechnologie des Algues, 44311 Nantes, Franceb Mer, Molécules, Santé, Institut Universitaire Mer et Littoral, FR 3473 CNRS, LUNAM Université, Université du Maine, EA 2160, IUT de Laval, 53020 Laval, Cedex 9, Francec University of La Rochelle, UMR CNRS 7266 LIENSs, F-17042 La Rochelle, France

h i g h l i g h t s

" Nine disruption techniques weretested on two microalgae models.

" Image analysis was used to evaluatethe efficiency of disruptiontechniques.

" The best grinding method was themixer mill with polypropylengrinding jars.

" The disruption method wasoptimised in the objective of highthroughput screening.

" Pigments were good candidates tofollow extraction of fragilemetabolites.

0960-8524/$ - see front matter � 2012 Elsevier Ltd. Ahttp://dx.doi.org/10.1016/j.biortech.2012.07.105

⇑ Corresponding author. Tel.: +33 2 40 37 41 09; faE-mail address: [email protected] (R. Kaas

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 10 April 2012Received in revised form 23 July 2012Accepted 28 July 2012Available online 14 August 2012

Keywords:MetabolitesPigmentMicroalgaeDisruptionExtraction

a b s t r a c t

Over the last decade, the use of microalgae for biofuel production and carbon dioxide sequestration hasbecome a challenge worldwide. Processing costs are still too high for these methods to be profitablethough, leading to a need to find high value by-products to optimise the added value of this biomass.For high-throughput screening of such metabolites, it is essential to reach the inner content of the cell. Thispaper presents research and development of a technique enabling a high extraction yield of any metabo-lite, taking into account the difficulty of extracting bound and or inaccessible molecules with a wide vari-ety of polarities. To this end, several disruption techniques were tested at laboratory scale on twobiological models: Porphyridium purpureum and Phaeodactylum tricornutum. A mixer mill gave the bestresults, offering access to a broad diversity of metabolites from microalgae for high-throughput screening.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Microalgae are nowadays considered to be the best source forbiofuel production due to their ability to produce large amountsof triglycerides or to be converted into biogas. These photosyn-thetic micro-organisms are capable of converting carbon dioxide

ll rights reserved.

x: +33 2 40 37 40 73.).

into lipids representing up to the half of their dry weight (Chisti,2007). While primary metabolites are the result of the unity oflife on earth, secondary metabolites are the expression of its biodi-versity (Kornprobst, 2005). Many of them have a high added value(Harun et al., 2010), such as isoprenoids, alkaloids, toxins, polysac-charides, polyunsaturated fatty acids, oxylipins, enzymes, phyco-biliproteins and non-ribosomal peptides, which find applicationsin health, pharmacology, nutrition and biotechnology. Amongall new marine molecules identified, the proportion produced by

312 B. Serive et al. / Bioresource Technology 124 (2012) 311–320

microalgae is growing significantly, particularly in the case ofsecondary metabolites (Pietra, 1997; Sasso et al., 2011). One ofthe major bottlenecks in obtaining molecules from microalgae isthe difficulty of extracting deeply inaccessible metabolites, whichcan compromise results of high-throughput screening analysis.

Development of extraction techniques for microalgae has be-come a field of growing interest for the scientific community.Because of the great economic stakes for biofuel production frommicroalgae, many studies have aimed to find efficient and eco-nomic ways to recover lipids (Lee et al., 2010; Balasubramanianet al., 2011; Halim et al., 2011). The techniques developed wereonly intended for heat-stable molecules, however, and are rarelysuitable for high-throughput screening of sensitive molecules. Forthe latter, a few studies have been conducted to extract naturalbioactive products from micro-organisms since the beginning ofthe 1980s. This research is progressing and many extraction tech-niques have now been developed (Mendes et al., 2003; Barbarinoand Lourenço, 2005; Wiyarno et al., 2010; Palavra et al., 2011).Pigment studies, in particular, have been steadily increasing sincethe early 1970s, both in the field of oceanography (Jeffrey et al.,1997; Cartaxana and Brotas, 2003; Hagerthey et al., 2006; Szymc-zak- _Zyła et al., 2011) and for industrial applications (Hejazi andWijffels, 2003; Machmudah et al., 2006). These techniques are of-ten suitable for molecule purification from large quantities ofbiomass (Hosikian et al., 2010), but not for high-throughputscreening purposes. Furthermore, they can also alter the mostlabile molecules (Latasa et al., 2001).

The aim of the present paper is to identify a generic techniquethat will make it possible to extract the largest possible numberof molecules from the largest number of species, taking into ac-count microalgal biodiversity. This biodiversity includes thethick-walled green or red algae, silicified diatoms, cyanobacteriawith their multilayered walls, red algae with wall-bound exopoly-saccharides and armoured dinoflagellates, which need to be brokenbefore extraction is possible (Porra, 1991). An other important needis to enhance the solubilisation of molecules of a wide polarityrange. At the laboratory scale, it is tempting to use strong solventsto extract targeted molecules. However, acetone, chloroform,dimethyl acetamide, dimethyl formamide, dimethyl sulfoxide andmethanol are considered unsuitable at the industrial scale due tosafety considerations (low lethal dose, carcinogenic, harmful, irri-tant or toxic characteristics) (Jeffrey et al., 1997). The present studyadopted an eco-friendly approach instead, by developing a mild sol-vent method that would limit health risks and thus facilitateup-scaling. The finalised extraction method needs to be quick, easyto use and require no heavy equipment. Temperature is another sig-nificant factor that can improve analyte solubilisation in solvents(Pasquet et al., 2011) as it can, for example, fluidise membrane lipidrafts. Temperature has a double effect, however: it can act posi-tively for heat-resistant molecules but can degrade heat-labile mol-ecules such as chlorophylls. Some metabolites, however, can bestrongly bound to proteins, polysaccharides, lipids, cell membranesor organelle membranes (Kornprobst, 2005) and can only be ob-tained by high fractionation.

To bring compounds of interest and solvents into contact, it isnecessary to disrupt cells into the smallest fragments possible with-out damaging fragile molecules. The aim of this work was to com-pare various disruption techniques and identify the best of theseaccording to efficiency criteria based on disruption yield and preser-vation of labile molecule quality. Nine different methods werecompared: soaking, cryogrinding, potter grinding, dispersing,bead-beater grinding, planetary micro mill grinding, sonicationand mixer mill grinding (with stainless steel jars or polypropylenetubes). The best of these methods was then optimised through thedesign of experiment (DOE) approach. The final goal was to providean organic soluble extract that could be screened using acute

biochemical or biological biotests to seek molecules of high addedvalue in high-throughput screening strategies. Pigments were cho-sen as test molecules for their wide range of polarities, fragilityand diversity of localisation within microalgal cells.

2. Methods

2.1. Biological models

Algae were grown under continuous light (120 lmol m�2 s�1)provided by fluorescent lamps (Philips TLD 58W 865). They weremaintained at 19.5 �C in batch culture in 10-L flasks and bubbledwith 0.22-lm filtered air containing 3% CO2 (v/v). Strains of twomicroalga species were selected for the assays in order to assesstheir differing mechanical and chemical properties: Porphyridiumpurpureum CCAP 1380.3 and Phaeodactylum tricornutum CCAP1.8.6. P. purpureum was chosen because of its strong cell wall andits ability to produce exopolysaccharides, which can cause problemsduring the extraction phase. The choice of P. tricornutum was mainlymade because this species has a slight silica cell wall that showsresistance to extraction without organic solvents. Furthermore,although this species has been particularly studied for its genome,transcriptome and proteome (Maheswari et al., 2010), obtaininggood extraction yields of its desoxyribonucleic acids, ribonucleicacids or proteins without degradation remains a challenge.

Algal monocultures were harvested by centrifugation (20 min,4500g, 4 �C). The pellets of microalgae were then extracted withanhydrous ethanol, whatever the protocols required by the differentdisruption techniques.

2.2. Disruption techniques

The conditions and screened factors were not identical for allcell disruption techniques tested because each one had its ownconstraints and efficiency advantages. In addition, the physicalconstraints of some techniques required a relaxation time to avoidexcessive heating of the sample.

2.2.1. SoakingMicroalgae pellets (50 � 106 cells) were soaked in ethanol at

ambient temperature for 24 h, as proposed in many oceanographicstudies (e.g., Wright et al., 1997). Assays were conducted in 1.5-mLconical tubes with absolute ethanol (500 lL).

2.2.2. CryogrindingMicroalgae samples (50 � 106 cells) were filtered through a

47 mm GF/C filter (Millipore) and rinsed with 3 mL deionisedwater. Grinding was performed in a ceramic mortar where the fil-ter was placed with about 200 mL liquid nitrogen and 20 mL deion-ised water. The filter was ground with a ceramic pestle until theliquid nitrogen had completely evaporated and the ice had fullymelted.

2.2.3. Potter homogeniserWheaton USA (10- and 5-mL) potter homogenisers were used,

with controlled rotation by a AEG Best 12� drill (GmbH Winnend-er). Samples (300 � 106 cells) were pelleted by centrifugation(20 min, 4500g, 4 �C) to remove culture medium. Ethanol (3 mL)was then added and the DOE completed according to the followinglimits: working time: 1 and 5 min; potter homogeniser volume: 5and 10 mL. No relaxation time was needed.

2.2.4. Bead-beaterThe bead-beater 1G918 (Biospec Products Inc) is made up of a

30 mL polycarbonate chamber, a rotor assembly, a cooling jacket

B. Serive et al. / Bioresource Technology 124 (2012) 311–320 313

and a Teflon impeller. Microalgal pellets (300 � 106 cells) wereresuspended in 2.7 mL ethanol before being poured into the beadchamber. Twelve millilitres of ethanol were then added, followedby glass beads, to reach a volume of 15 mL. The ice chamber wasfilled with water-ice before running the grinding cycles, and thetemperature of the samples measured after each run. A prelimin-ary test was conducted to choose the best size of glass beads(2 or 4 mm diameter). Two factors were then studied: the millingtime and the number of cycles. The lower and upper limits were1 and 4 min, respectively, for the milling time and 2 and 4 forthe number of cycles. The relaxation time between cycles was1 min. At the end of the process, the mixture of beads and dis-rupted cells was filtered through a stainless steel sieve (250 lm).Beads were rinsed with about 45 mL deionised water and thecollected volume was adjusted to 60 mL.

2.2.5. HomogeniserTests were performed with a Polytron PT3100 equipped with a

homogeniser generator PT-DA 3012/2TM (Kinematica AG – Switzer-land). Speed was fixed at 18,000 rpm. For the factorial design, worktime limits were 10 and 180 s and cycle number limits were 2 and 4.The interval between each cycle was set at 30 s (relaxation time).While working, the samples were cooled in an ice bath. Each sampleconsisted of 3 � 108 cells in ethanol (3 mL).

2.2.6. Planetary micro millA Planetary Micro Mill Pulverisette 5 (Fritsch) was used. The

factors considered were the nature of the material and the beaddiameter. Bowl volume was 300 mL. Beads were made of zirco-nium oxide or glass. The latter had a diameter of 2 or 3 mm and200 g were used for each trial. Those made of zirconium oxidehad a diameter of 20 mm and fifteen beads were used. Two cyclesof 4 min at 400 rpm were applied with 1 min separation (relaxa-tion time). Each sample consisted of microalgae pellets represent-ing 3 � 109 cells resuspended in ethanol (30 mL).

2.2.7. SonicationA Vibra Cell 75115 (Bioblock Scientific) ultrasonic processor

with a multichannel probe (S&M 630-0559) was tested. The lowerDOE limit of the milling time was 10 min and the upper limit was20 min. The numbers of cycles tested were 2 and 4, with intervalsof 30 s (relaxation time). Power was fixed at 30% of total power(500 W), as advised by the supplier of the probe, and the apparatuswas driven in pulse mode (10 s ON, 5 s OFF). Tubes were cooled byusing an ice bath during the experiment. Each sample contained300 � 106 cells in ethanol (3 mL).

2.2.8. Mixer mill – stainless steel grinding jarsThe first step was to test an MM-400 mixer mill (Retsch) with

stainless steel grinding jars (10 mL capacity). Microalgae samples(300 � 106 cells) were added to the grinding jar with ethanol(5 mL) and 2 stainless steel balls (9 mm diameter). Each jar hadbeen previously cooled in liquid nitrogen for 5 min. The processwas run for 2 min per cycle to avoid sample overheating, and thevibration frequency was set at 30 Hz. Four cycles were carriedout, with the sample cooled in liquid nitrogen for 3 min betweencycles.

2.2.9. Mixer mill – polypropylene grinding tubesThe same mixer mill was tested equipped with small polypro-

pylene grinding tubes grouped in 2 PTFE racks. Microalgae(50 � 106 cells) were aliquoted into 1.5 mL graduated conical tubes(StarLab ref: E1415-2230) with certified standard caps with their0-ring (StarLab ref: E1480-0100). Each PTFE rack could hold 10tubes. Finally, ethanol (500 lL) was added to the grinding tube.

To protect weak metabolites from breakdown, samples were firststored for 1 h at �80 �C.

Before working on the response surface design, preliminarytests were run to improve the grinding capacity of this technique:a comparison was made of different kinds of grinding beads (glassbeads of 500 lm; zirconium beads of 1.2 and 3 mm), and optimalcell concentrations in the grinding tube were determined (100,50 and 25 � 106 cells).

In order to prevent EPS affecting the grinding of P. purpureumcultures, different ethanol concentrations (ethanol/distilled water,100:0, 50:50, 0:100 v/v) and salt water (35 psu) washing cycles (upto 8 times 700 lL) were tested. At the end, the samples were vor-texed, centrifuged (2000g, 5 min, 4 �C) and the supernatantfinally removed.

For the optimisation procedure, the tested factors were thegrinding time and the amount of beads added to the tubes. Re-sponse surface factor levels for the grinding time were 0.65, 5,15.5, 26 and 30 min and the weights of the 500-lm glass beadswere 0.076, 0.2, 0.5, 0.8 and 0.92 g. The vibration frequency appliedto samples was always 30 Hz.

2.3. Response variables linked to disruption yield measurement

During all experiments, sample temperature was measuredafter each run.

2.3.1. Cell countingTo estimate the effectiveness of each technique of fragmenta-

tion, the percentage of cells that had disappeared after grindingor milling was used as a response variable. Counting was performedusing a Malassez cell and an image analysis system. Pictures weretaken with a Leitz Diaplan microscope equipped with a 3CCD colourcamera (Donpisha) at 100� magnification and processed usingSAMBA software. In order to obtain a coefficient of variation below5% it was necessary to take 10 pictures for P. purpureum and 20 pic-tures for P. tricornutum. A yield calculation was performed by com-paring the results of counts between trials and controls. Cellcounting was used during the first phase of experimentation, whenselecting of the most appropriate method for cell destruction.

2.3.2. Particle size distributionFragment size assessments were carried out by image analysis

during the optimisation phase of the grinding procedure. The samesystem as described above was used and pictures were taken intriplicate for each sample. Pictures were taken with Ips4 softwareversion 4.21 and processed using Matlab 7.0. The area of each par-ticle was determined in pixels and a frequency table built with thefollowing size groups: For P. tricornutum, one group of small parti-cles with an area less or equal to 10 pixels, a second medium-sizedgroup concerning particles with areas between 10 and 49 pixels,accounting for the size of whole cells, and a third group of particlesrepresenting large aggregates with areas over 100 pixels. ForP. purpureum the medium-sized group, corresponding to whole cellsize, was set between 25 and 49 pixels.

2.4. Extraction yield measurement

The pigment analysis was done by hyphenated High Pressure Li-quid Chromatography with UV DAD detector. An Agilent Technolo-gies series 1200 HPLC system equipped with an auto-injector, acooled autosampler compartment, a thermostated column com-partment, a quaternary pump with on-line vacuum degasser anda photo-diode array detector was used for this study, with a reversephase column Eclipse XDB-C8 150 � 4.6 mm, 3.5 lm. Elution meth-od was as described by Van Heukelem and Thomas (2001). Forchromatographic conditions, the end of the gradient was extended

Fig. 1. Standard pareto charts for fine particle yield from (A) Phaeodactylumtricornutum and (B) Porphyridium purpureum. In the pareto charts A and B representthe linear effect of the factors. AA and BB represent the quadratic effect of factors Aand B, respectively. AB represents the interaction effect between the two factors.


with mobile phase A (5%) and mobile phase B (95%) up to 40 min in-stead of the 37 min given in the original method. HPLC grade meth-anol was purchased from Carlo Erba and HPLC grade water fromGrosseron. Tetrabutyl ammonium acetate was purchased fromSigma. The samples were filtered through glass fibre filters of0.20 lm porosity and 4 mm diameter (Whatman) using a syringe.The main detection wavelength was 436 nm and derivatives weredetermined by signal comparison between 436 and 405 nm. Abso-lute quantification was possible for zeaxanthin, chlorophyll a, phe-ophytin a, beta-carotene, fucoxanthin and diadinoxanthin frompigment standards purchased from DHI Water & Environment(Denmark). The other major pigments were analysed by relativearea quantification. Mixer mill samples were analysed at two spe-cific levels of the response surface design. The first correspondedto levels of factors where the percentage of fine fraction was lowestand indicative of a relatively coarse grinding. The second samplewas taken for factor levels corresponding to the optimum calcu-lated by the desirability index (maximum of tiny fragments, mini-mum of whole cells and minimum of aggregates, see below),characterising the best grinding performance for each microalga.HPLC analyses for P. tricornutum were achieved at 5 and 28 min,and those for P. purpureum at 5 min and at 30 min with 0.9 g of500 lm glass beads.

The comparison of these analyses can provide a great deal ofinformation on the availability of certain pigments and on the pos-sible degradation of fragile pigments when the optimal conditionsfor degradation are applied. It can also be used to determinewhether the number of molecules detected increases. For somemethods, a quantitative change would signal that some pigmentsare more strongly linked to cellular structures and released bymore efficient grinding.

2.5. Statistical analysis

Calculations were made using Statgraphics Centurion XV.I.Screening methods based on factorial designs were used for allexperiments. Boundaries of the experimental domain were workedout following advice of the suppliers and previous assays (data notshown). Main effects and interactions between factors were repre-sented by standardised Pareto Charts and data analysed by a multi-way ANOVA. For the final optimisation of the process, a res- ponsesurface method was used, based on a full central composite facto-rial design with star points. For two independent variables, thisneeded at least 4 square points, 4 star points and 3 centre points(Fig. 1).

A quadratic model was required to describe the processbehaviour:y ¼ b0þ b1�x1þ b2�x2þ b12�x1�x2þ b11�X12 þ b22�X22

where x1, . . .,Xk are the main effects for the factors, X1 �X2, ...,Xk � 1 � Xk represent the interactions and X12, ...,Xk2 the qua-dratic component.

Contour plots provide a visual representation of the model andhelp to pinpoint the optimal levels that should be used for eachfactor to maximise or minimise the observed response.

Finally, when two or more responses were followed, a desirabilityindex could be built, offering the possibility to maximise, minimiseor fix the different responses. For pigment extraction yield compar-isons statistical significance was determined by Student’s T test.

3. Results and discussion

3.1. Selection of the most effective cell disruption technique

During this first phase, the percentage reduction in whole cellswas determined and the temperature was monitored.

3.1.1. SoakingSoaking led to low disruption yields (5% for P. tricornutum), par-

ticularly for P. purpureum (0%), protected by a strong cell wall andpolysaccharides. Soaking, nevertheless, is the most common meth-od in oceanography studies. This extraction technique is valid fornaked flagellates, such as some Prasinophyceae, but only has amoderate breadth of applications, and the accuracy of some studiesis questionable. Real concentrations of difficultly accessible pig-ments could, therefore, be vastly underestimated.

3.1.2. CryogrindingCryogrinding results were also poor, as cells generally remained

whole (1% of the disruption yield) and reproducibility was low.This technique is not suitable for the needs of high-throughputscreening.

Both of these first techniques were therefore rejected.

3.1.3. Potter homogeniserThe potter homogeniser gave a 10–42% disruption yield for P.

purpureum and one from 49% to 79% for P. tricornutum. Grindingtime also had a significant effect on P. purpureum. A slight rise intemperature was observed in relation to grinding time. This tech-nique rejected due to the poor yield results for P. purpureum.

3.1.4. Bead beaterBead beater. Following the pre-test, the 4 mm beads were aban-

doned in favour of those of 2 mm, because they led to a lesser risein temperature of the product and introduced fewer foreign parti-cles in the samples. The grinding yields were between 21% and 85%for P. purpureum (2 � 1 min and 4 � 4 min grinding time, respec-tively) and between 86 and 94% for P. tricornutum (2 � 4 min and


4 � 4 min grinding time, respectively). Grinding time had a stronginfluence on P. purpureum, but a weaker one on P. tricornutum. It isimportant to note that grinding time had an important effect onthe final temperature of the sample at the end of the test. Temper-atures between 31 and 65 �C were noted for P. purpureum and be-tween 32 and 54 �C for P. tricornutum. In other technologies, suchas microwave-assisted or soxhlet extractions, where relativelyhigh temperatures are reached (Wiyarno et al., 2010; Pasquetet al., 2011), only heat resistant molecules are preserved. Fragilemetabolites might be damaged by oxidation, hydrolysis and isom-erisation. Due to the excessive temperature rise of the samples, thismethod was also abandoned.

3.1.5. HomogeniserThe homogeniser produced grinding yields between 0 and 56%

for P. purpureum and between 35% and 65% for P. tricornutum. Tem-perature elevation was moderate (12–36 �C according to time).However, the unit lost its effectiveness rapidly due to the wear ofthe cutting tool, which led the technique to be abandoned.

3.1.6. Planetary millThe planetary mill produced yields close to 100% for both algae

tested, whatever the tested conditions, but the formation ofnumerous clumps of debris altered the initial effectiveness. Thesample temperature varied between 36 and 46 �C. Because of thehigh temperatures recorded and the presence of numerous celldebris clusters, this technique was also rejected.

3.1.7. SonicationSonication proved to be almost ineffective on P. purpureum,

whatever the conditions tested. However, this technique gave gooddisruption yields ranging from 85% to 99% on P. tricornutum. Overall,the temperature of the sample was maintained because of the ice

Table 1Evaluation of disruption techniques.

Method Effectiveness(disruption yield)a

Reproducibilityb Operatingcostc

Soaking PP� PT� � +Cryogrinding PP� PT� � �

Potter homogeniser PP� PP� � +Bead Beater PP+ PT+ � +Homogeniser PP+ PT� � +

Planetary micro mill PP+ PT+ + +Sonication PP� PT+ + +Mixer mill – stainless

steel jarsPP� PT+ + �

Mixer mill – plasticgrinding tube

PP+ PT+ + +

PP: Porphyridium purpureum.PT: Phaeodactylum tricornutum.+: Positive effects.�: Negative effects.n.r.: Nothing to report.A summary of the preliminary results obtained in the first disruption evaluation (countinconclusive. Otherwise, the grinding techniques could not be considered for the rest of t

a The result is model-dependant, so each biological model is marked with a ‘+’ if theb If the technique allowed a reproducibility of results, it was marked with a ‘+’ and ifc A technique marked ‘+’ indicates that the cost of the process per each sample is acce

that supplies represent a significant cost if the technique is used regularly. In these experid A ‘+’ indicates that the sample remained at an acceptable temperature (<35 �C). A ‘�

have adverse effects on fragile metabolites or cause hot spots in the sample, e.g., with se Apart from the apparatus itself, the criterion ‘‘easy to use’’ means it is not necessar

marked ‘�’ are difficult to apply in the context of high-throughput screening, which reqf Techniques marked with a ‘�’ cannot be used in their present form to extract meta

upscaling is conceivable in techniques marked ‘+’.g This criterion means the technique allows a process with a large proportion of water

samples but require sample concentration (i.e., centrifugation) before disruption proces

bath, although, according to (Suslick, 1994), a hot spot at the tip ofthe probe can reach a temperature estimated to be about 5000 �C,due to rapid adiabatic compression of gases and vapours within bub-bles or cavities. Moreover, in sonochemistry, free radical reactionsleading to metabolite oxidation can occur if a small amount of wateris present in the treated sample (Petrier et al., 1992). This techniquewas rejected as it only proved efficient with one type of algae.

3.1.8. Mixer mill – stainless steel jarsThe milling yield was 96% for P. tricornutum and 89% for P. pur-

pureum, but numerous cell clumps were formed during the processwith the latter species. Because of the polysaccharides, the stain-less steel balls became blocked during some assays. This techniquewas efficient but not appropriate for all species and therefore notsuitable for high-throughput screening procedure.

3.1.9. Mixer mill – polypropylene grinding tubesTests were performed with glass beads (500 lm, 0.9 g) and pro-

duced milling yields around 76% for P. purpureum after 4 mingrinding and close to 99% after 32 min. P. tricornutum was groundwith an efficiency of 97% within 5 min and 100% after 32 min. Thetemperature reached at the end of the experiments never exceeded31 �C.

Results of all the methods are presented in Table 1 with theobserved advantages or disadvantages of the different methods.

Among all the disruption techniques tested, the mixer mill hadthe greatest number of positive points (Table 1). This techniquewas efficient and reproducible, whatever the biological modeltested, and samples were not overheated during the process. Inaddition, the technology/protocol is easy to use and has a low oper-ating cost. One negative point is the need to have a pellet of cells,which means it is not possible to process diluted streams without atechnique to concentrate the sample. Despite this constraint, the

Temperaturestabilityd

Easy tousee

Scalablef Dilutedstreamsg

Other considerations

+ + + � Long processing+ � + � Liquid nitrogen

handling require� + + � n.r� � � + � Particles release� + + + Blade blunted by

diatoms� � + � Particles release� � + � + ROS release, hot spots� + + � n.r.

+ + � � n.r.

g method). Percentages of ground cells are only given in the paper if the result washe study and are marked with a ‘�’ in the summary table.technique produced a sufficient yield and with a ‘�’ if the result was poor.not, it was marked with a ‘�’.ptable in a context of high-throughput screening. A technique marked ‘�’ indicatesments, it was the cost of liquid nitrogen that penalised two of the tested techniques.’ indicates that the technique caused a rise significant rise temperature that wouldonication.y to set up a great deal of equipment and use much labour. Disruption techniquesuires rapidity and ease of implementation.bolites of interest identified by screening based on the same disruption technique;

in the sample. Thus, techniques marked with a ‘�’ cannot efficiently disrupt dilutedsing.


mixer mill method is a good compromise and was therefore chosenfor optimisation in the present study.

3.2. Optimisation of a protocol using a mixer mill and polypropylenegrinding tubes

In order to optimise the grinding process for an accurate high-throughput screening of biological activities, preliminary testswere performed before the optimisation procedure.

3.2.1. Preliminary tests3.2.1.1. Bead material determination. Zirconium oxide beads (diam-eters 1.2 and 3 mm) generated large debris clusters, indicating thatthe cells had not been ground efficiently. Glass beads (500 lm)produced far smaller amounts of large clusters. As zirconium oxidebeads are also relatively expensive, the choice of glass beads wasobvious.

3.2.1.2. Cell concentration determination. The trials clearly showedthat low cell concentrations increased the grinding yield signifi-cantly and had a better reproducibility. From there on, the objec-tive was to define the best cell concentration compatible withthe threshold of subsequent biotests and HPLC analysis. In thiscontext, the extract concentration was a compromise. For med-ium-sized microalgae cells such as P. tricornutum (about 4–5 lmwidth � 16–25 lm length) and P. purpureum (4–8 lm diameter),it was determined that 50 � 106 cells per tube were necessary.

3.2.1.3. EPS elimination. Complementary experiments showed thatwashing steps before grinding are essential when working withspecies like P. purpureum, as cells tend to form clusters that limitthe efficiency of the beads. This phenomenon may be related tothe EPS secreted by the microalgae into the medium, which pro-mote cell aggregations. Among EPS, polysaccharides are well repre-sented and some are known to be strongly bound to the cell wall(Heaney-Kieras et al., 1977; Percival and Foyle, 1979). Up to 8 washcycles were tested, showing that 2 washes are sufficient to removemost of the free EPS in the medium and significantly increase thegrinding yields. Additional cycles did not lead to significant gainsfor most of the microalgae. Yet for cultures with high EPS concen-tration, it would be judicious to add 2 more washing cycles. ButEPS bound to cell walls cannot be removed simply by washing.

A

B

Par

ticle

fine

ness

(%

)

Par

ticle

fine

ness

(%

)

Fig. 2. Estimated response surface for fine particle yield from (A)

Thus, ethanol was used in this technique to solubilise metabolites,although this might also act on polysaccharides and increase theirprecipitation. The resulting polysaccharide fibres might then ab-sorb the impact of the grinding beads and reduce their efficiency.The next experiment aimed to assess the effect of ethanol concen-trations on grinding performance.

3.2.1.4. Grinding medium determination. With a 100% seawatermedium, the grinding yield was 67%. A 100% ethanol mediumreached 89% efficiency, but a 50:50 seawater/ethanol mixture ledto an average grinding yield of only 30%, with a high coefficient ofvariation (61%). Results obtained with 100% ethanol medium and100% seawater medium were significantly different from the mixedmedium (50:50) and showed the necessity to remove as muchwater as possible before grinding. For good grinding efficiency,the sample should contain only a minimum of water correspondingto the pore water associated with the cell steric hindrance. Theexclusive use of absolute ethanol also has the advantage of limitingthe activity of chlorophyllases, which are particularly commonamong the diatoms (Barrett and Jeffrey, 1971). Conversely, forapplications such as high-throughput screening, a high rate ofethanol is less appropriate if bioassays on cell lines are planned.

3.2.2. Response surface design for P. tricornutum grinding optimisationTaking into account the size of the particles makes it possible to

define the quality of the grinding more precisely by looking at theevolution of small-sized fragments (<10 pixels).

There was a significantly positive effect of bead weight on dis-ruption yield (Figs. 1A and 2A). The best performance was 50%,meaning that 50% of particles were smaller than 10 pixels. Thequadratic effect of time on the disruption yield was not significantat the 0.05 significance level. No significant interactions betweenweight and time were observed.

The adjusted R-squared statistic indicates that the model forparticle fineness as fitted explains 70.6% of the variability of the dis-ruption yield and the lack-of-fit is not significant (0.9744), meaningall influential factors were taken into account with this model.

The equation of the fitted model (Fig. 2A) was:Disruption yield = �12.9292 + 2.68125 � Time + 65.2163 �

Weight� 0.00671933 � Time2� 0.403627 � Time �Weight� 13.2827�Weight2

Analysis of Variance for Particle fineness

R-squared = 77.6 percent

Analysis of Variance for Particle fineness

SourceSum of Squares Df

Mean Square F-Ratio P-Value

A:Time 0.0208338 1 0.020833 0.01 0.9380 B:Weight 39.8776 1 39.8776 13.12 0.0223 AA 1.53546 1 1.53546 0.51 0.5164 AB 4.25424 1 4.25424 1.40 0.3022 BB 27.6853 1 27.6853 9.11 0.0392 blocks 0.0462373 1 0.046237 0.02 0.9078 Lack-of-fit 86.8134 11 7.89213 2.60 0.1850 Pure error 12.1542 4 3.03856 Total (corr.) 170.956 21

R-squared = 42.1 percent

SourceSum of Squares Df

Mean Square F-Ratio P-Value

A:Time 277.242 1 277.242 1.61 0.2739 B:Weight 3004.45 1 3004.45 17.40 0.0140 AA 619.813 1 619.813 3.59 0.1311 AB 12.9323 1 12.9323 0.07 0.7979 BB 16.1403 1 16.1403 0.09 0.7751 blocks 27.7569 1 27.7569 0.16 0.7090 Lack-of-fit 448.01 11 40.7281 0.24 0.9744 Pure error 690.768 4 172.692 Total (corr.) 5092.93 21

Phaeodactylum tricornutum and (B) Porphyridium purpureum.

Wei

ght (

g)

0 5 10 15 20 25 30 350.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1W

eigh

t (g)

Time (min)0 5 10 15 20 25 30 35

0.2

0.3

0.4

0.5

0.6

0.7

0.8

A

B

Fig. 3. Overlay of the three response plots used for the desirability index, enablingmulti-criteria optimisation for (A) Phaeodactylum tricornutum and (B) Porphyridiumpurpureum. ( ) Fragments; ( ) Whole cells; (—) Aggregates.


Calculation of the optimum value gave an optimal disruptionyield of 56% of fragment areas less than or equal to 10 pixels witha grinding time of 17 min and 0.92 g of 500 lm beads.

This first model does not take into account the percentage offragments corresponding to the size of whole cells and the largeaggregates. These two response variables were added in order tominimise their value.

Finally, a desirability index was built to try to maximise theresponse for small particles, and minimise the responses for largecell aggregates and the medium-sized group. The desirability mod-el, shown with an overlay plot in Fig. 3A, indicates an optimum at28 min grinding time with 0.92 g beads (a cross on the plot). Theoptimal predictions for the 3 size groups were 0% for the largeaggregates (100–999 pixels), 0% for the range of whole cells (10–49 pixels) and 49% for the small fragments (1–9 pixels). The opti-mal weight was the same as in the first model and the optimal timea little longer (+11 min). The optimal point was confirmed after-wards by a complementary experiment run with the calculatedoptimal values. Results were 1% for the large aggregates, 26% forwhole cells and 70% for the small fragments. Observations madeby light microscopy showed that the group designed to includewhole cells in fact contained clumps of small fragments. Thisshows the limits of a fully automated image analysis system; visualchecking of the results is always necessary. Results thereforebecame much closer to the predicted values and were even betterfor the small fragment group.

3.2.3. Response surface design for P. purpureum grinding optimisationConcerning the small size group, there was a significant qua-

dratic effect for the weight factor with a small amount of beadsshown to be the most efficient, whatever the grinding time (Fig. 1B).

The R-squared statistic indicates that the fitted model explainedonly 42% of the variability of the disruption yield and that the lack-of-fit is not significant (0.1850). The unexplained part of variabilityindicates that there is probably at least one influential factor thatwas not accounted for in this study. The cause probably lies in

the presence of residual polysaccharides that are still bound tothe cell wall and act as shock absorbers.

The equation of the fitted model (Fig. 2B) is:Disruption yield = 83.1167 + 0.222865 � Time + 15.7223 �Weight

� 0.00334446 � Time2� 0.2315 � Time �Weight� 17.3964 �Weight2

The optimal calculated value was 88% of fragments less then 10pixels with a 23-minute grinding time and 0.30 g beads.

The desirability index was built, in the same way as for the otherspecies by adding two response variables for medium-sized frag-ments (25–49 pixel) and large aggregates (100–999 pixels). Thismodel (Fig. 3B) predicts that, with a grinding time of 30 min and0.24 g of beads (a cross on the plot), it should be possible to obtain3% large aggregates (100–999 pixels), 3% whole cells (25–49 pixels)and 88% small fragments (1–pixels). In the validation experiment,the general trend was close to these predictions and results were2% large aggregates, 6% whole cells and 69% small fragments. Thecomplexity of this biological model means that these results shouldbe interpreted with caution. The coupling of the models respectedthe overall trends the differences between expected data and ob-served data are acceptable. The models for P. purpureum might beimproved, by incorporating, for example, a new independent vari-able related to the amount of EPS present, and/or by adding the con-centration of EPS as a dependent variable.

3.3. Extraction yield comparisons

The study of pigments allows one to define the relationship-between the fineness of fragments and access to metabolites. In-deed, authors took the case of pigments as a representativeexample of metabolite diversity in terms of localisation, polarityand links with other molecules and structures. Pigments can beeasily obtained if they are free in the cytoplasm, but extraction ismuch more difficult for metabolites that are strongly bound toorganelles or photosystems. The HPLC analysis was therefore agood tool to use in compliment to the fragment size class analysis.This part of the study was built from the conclusion of the optimi-sation procedure.

3.3.1. Chemodiversity extraction from P. purpureumPigment concentrations were compared between an extraction

made after 5 min of grinding and the one made after 30 min. Themost interesting results were obtained with P. purpureum, wherea long grinding time allowed the recovery of many pigments ofinterest. The zeaxanthin concentration was 47% higher after30 min grinding compared with 5 min, indicating that this pigmentis located relatively deep within the cell organelles. For the cantha-xanthin-like molecule, the difference was 58%. Chlorophyll a like-1,chlorophyll a like-2, chlorophyll a and derivative product 2 fromchlorophyll a, showed improvements of 39%, 34%, 29% and 33%,respectively. It can be assumed that these molecules were moreeasily accessible than zeaxanthin and the canthaxanthin-like mol-ecule. Derivative products 1 and 3 from chlorophyll a showed gainsof 113% and 94%, respectively, suggesting that these moleculeswere less accessible and consequently more difficult to solubilise.One molecule in particular showed a specific phenomenon: b,b-carotene showed a negative relationships between extraction yieldand grinding time. A loss of 89% was thus observed between 5 and30 min grinding.

One hypothesis for this unexpected result could be a tribolelec-tricity effect. Triboelectricity refers to the electrostatic phenome-non by which, when two materials of different nature are incontact, some electrons of the contact surface of one of the twomaterials are transferred to the other, and the transfer remainseven after separation. The triboelectric effect can be increased byproviding mechanical energy to the materials, e.g., by rubbingthem against each other. In this study, b,b-carotene is the only


linear pigment, and is highly conjugated and non-functionalized. Itneeds to be considered whether such a molecule could becomesustainably charged under the influence of mechanical energy.

The vibrations of the mixer mill system (30 Hz) would be capa-ble of creating static electricity, which is able to influence themovement of pi electrons. This imbalance would then resonateelectrostatically along the b,b-carotene chain. It can be hypothe-sised that a polarisation of the chain could occur, creating positiveand negative ends. A plastic grinding tube, glass beads and a bio-logical sample high in polysaccharides in the presence of ethanol(0.88 E0 polarity) provide a complex environment in terms of elec-trostatic charge. In such an environment, it is likely that b,b-caro-tene could potentially form micelles and thus adsorb onto theglass beads and/or the EPS and/or the plastic tube. When recover-ing the filtrate these would remain in the tube and thus be ex-cluded from the subsequent analysis. If they were to adhere tocellular debris, they would be caught on the filter during the finalfiltration and be also excluded from the HPLC pigment analysis. Itwill therefore be necessary to pay attention to this phenomenonwhen using this grinding technique to seek molecules with the fea-tures of b,b-carotene.

3.3.2. Chemodiversity extraction from P. tricornutumFor P. tricornutum, four major pigments were measured at 5 and

28 min grinding time (fucoxanthin, b,b-carotene, chlorophyll a likeand diadinoxanthin like). Quantitative differences between theworst and the optimal conditions were analysed by Student’s t-test(5% risk coefficient). For the four pigments, the H0 hypothesis couldnot be rejected, indicating that there was no significant differencebetween the two grinding conditions. From these results it can behypothesised that these pigments are rather easily accessible tosolvent in the cell and/or are not strongly linked to sub-cellularstructures. These results are consistent with those of Kim et al.(2012), showing that fucoxanthine is easily extractible.

In summary, although a grinding time of 5 min is sufficient forsome species to extract metabolites such as pigments, a longertime is needed for other more complex ones. In the context ofthe proposed high-throughput screening, it would be necessaryto apply the maximum milling time for all species to be as discrim-inating as possible.

3.4. Proposed protocol for high-throughput screening studies

The following protocol (Fig. 4), summarises the informationgained from this study. Among the techniques used here to evalu-ate disruption yield, the mixer mill showed excellent results. Dur-ing the second part of the study, the work on parameteroptimisation showed that these differ according to the microalgaespecies. In the context of high-throughput screening studies (espe-cially if these are automated), it is preferable to use one singlemethod to compare all samples. The following protocol is basedon the more difficult of the two biological models tested (P.purpureum).

First, it is necessary to evaluate the amount of cells essential fora good detection of targeted metabolites (in this case50 � 106 cells). The sample must be placed in a grinding tube(graduated conical tubes 1.5 mL – Starlab). When samples are richin polysaccharides, washing is necessary in order to improve grind-ing yield and reproducibility. To remove polysaccharides, samplesare first centrifuged for 5 min at 2000g or up to 6500g if EPS con-tent is very high, and the supernatant discarded. According to themicroalgae species, 500 lL of 0.2-lm filtered seawater for marinespecies or distilled water for freshwater species are then added tothe tube, which is vortexed and the supernatant removed. A secondaddition of 500 lL is made followed by a centrifugation at 2000gfor 5 min and the supernatant is again discarded. After the de-

crease in the overall volume of the cell mass due to the disappear-ance of much of the exopolysaccharides, 700 lL distilled water (orseawater) are finally added to the sample. It is then vortexed,gently centrifuged (2000g, 5 min) to avoid cell breakage and thesupernatant discarded. Analyses of this supernatant revealed nopresence of organo-soluble pigments (data not shown). This lastset of steps (addition of 700 lL of water, centrifugation, superna-tant elimination) must be repeated twice. If the sample is high inEPS, the last procedure must be repeated once more and the lastcentrifugation should be 4500g for 5 min before removing thesupernatant. Glass beads (500 lm, 0.24 g) and absolute ethanol(500 lL) are finally added to the tube. If the targeted metabolitesare very sensitive, the tube can be stored for one hour at �80 �C be-fore beginning the grinding process. As the melting point of etha-nol is �114 �C, the sample will not freeze before being processed.Finally, samples are ground for 30 min and can be easily recoveredusing a self-built device constructed at the laboratory: a standardcap used for grinding tubes was pierced in the middle and a pieceof phytoplankton net (e.g., 200 lm) fixed onto the cap to keep thebeads in the tube. To retrieve the sample, the centrifuge tube istapped to get the retentate out and the tube tip is cut with awire-cutter. The filtrate is recovered in a 2 mL test tube. The reten-tate is washed 3 times with ethanol (500 lL). The cell fragmentsstill present in the filtrate can be removed by a final centrifugation(4500g, 5 min, 4 �C). A last filtration through a 0.2 lm PTFE filterfor organo-soluble samples can be followed either by analysis oractivity bioassays. A summary of the protocol is given in Fig. 4and was validated by complementary assays.

Optimised parameters for P. purpureum (30 min grinding with0.24 g beads), were applied to P. tricornutum. Results were 71%for the small fragments, 29% for the medium-sized group and 0%for large aggregates. Observation of the medium-sized group, con-firmed the presence of particle clumps and no whole cells wereobserved. These results are acceptable when compared with theoptimal parameters of the P. tricornutum model.

The choice of grinding parameters must be set by consensusand this compromise inevitably reduces the grinding efficiencyobtained but advantages are offered by the universality ofapplications.

Care should be taken for the influence of triboelectric effects ona few specific molecules. Fortunately, few species contain highamounts of polysaccharides.

3.4.1. Perspectives in the field of efficient metabolite extraction frommicroalgae

Among all disruption techniques presently used to study micro-algae at the laboratory scale, the mixer mill protocol proposed inthis study can provide a solution for effective metabolite extractionfrom microalgae. Apart from ethanol, other solvents like water orbuffers can be used to extract water-soluble chemicals (e.g., DNA,RNA, phycobilins, peptides or cyclic polyethers). At present, therecovery of RNA or DNA from microalgae is often based on sonica-tion techniques to improve the extraction yield. However, thisleads to poor RNA quality because thermal and cavitation effectsof sonication break hydrogen bonds and cause rupture of DNAstrands (Elsner and Lindblad, 1989). The optimised milling tech-nique seems appropriate in this case as it causes less damage tomolecules than sonication. In oceanography, this method wouldbe useful for the accurate study of phytoplankton populations be-cause it avoids the bias that could be caused by microalgae speciesthat are difficult to process. In oceanography, this method wouldallow a more precise study of phytoplankton populations by meansof pigments, as it reduces the bias related to microalgae unsuitedto pigment analyses. Phytoplanktonic population studies are oftenperformed using the CHEMTAX program, which is based on pig-ment ratios. If part of the population escapes full grinding, some

Fig. 4. Proposed protocol to grind diverse microalgae, including EPS-rich microalgae. 1Up to 6500 g for highly EPS-rich samples. 2Step 3 repeated 4 times for highly EPS-richsamples. 3Up to 4500 g for highly EPS-rich samples.


pigments will not be detected in HPLC analysis and this can lead toerroneous conclusions.

The optimised milling technique is consistent with the work ofMolina Grima et al. (2003), supporting the idea that mechanicalgrinding is often the most effective technique to recover themetabolites of microalgae.

This technique, because it uses ethanol, is green, clean and scal-able for the solvent. Previous work showed that ethanol is a goodsolvent to extract fatty acids and volatile compounds such asfucosterol, neophytadiene and hexadecanoic acid (Plaza et al.,2010). According to Lan et al. (2011), ethanol is more efficient forextracting chlorophyll than many other organic solvents, and alsohas greater stability. Previous experiments confirm this point andshow that extraction with 100% ethanol limits the degradation ofporphyrin pigments by inhibiting chlorophyllases (Serive, unpub-

lished data). Ethanol allows extraction of a wide range of moleculeswith different polarities (e.g., pigments) and avoids the esterifica-tion of porphyrin or the loss of its bound metals, which can occurwith solvents such as methanol or acetone.

Moreover, the mixer mill technique is rigorous and allows therecovery of most pigments from microalgae samples. It fulfils mostof the criteria listed by Wright et al. (1997) for a promising extrac-tion procedure: extractability (by favouring the connection be-tween pigment and solvent), fidelity (no modification of analytes,as this method avoids the activation of chlorophyllase enzymes,de-metallation, allomerisation or epimerization of chlorophylls,and degradation or isomerisation of carotenoids), compatibilitywith HPLC analysis (because of the exclusive use of ethanol), sim-plicity of implementation and safety (as there is no need for strongsolvents or the use of liquid nitrogen).


4. Conclusions

Effective cell disruption techniques need to be rapid, generatelittle heat, use small volumes of sample and favour the solubilisa-tion of molecules of interest, even the least accessible. The bestcompromise applicable to all cases is to grind 50 million cells for30 min with 0.24 g glass beads (500 lm) using a mixer mill with1.5 mL plastic tubes. This protocol can be used with many microal-gae and leaves even the most fragile molecules intact. In a contextof high-throughput screening, this technique seems to be a goodtool to obtain valuable metabolites from microalgae for pharma-ceutical applications.

Acknowledgements

This research was supported by a doctoral fellowship grantfrom Région Pays de la Loire, European grants from the ‘‘FondsEuropéen de Développement Régional FEDER’’ and the ‘‘Contratsde Projet Etat-Region CPER’’ (Project ‘‘Extraction of anticancer pig-ments from marine microalgae’’ XPIG). We also thank the Frenchcancer league (Ligue Nationale contre le Cancer) for financial sup-port. Many thanks to Dr Claire Boisset (IFREMER Brest) for con-ducting the part of mixer mill experiment made with thestainless steel jars. The authors also thank Prof. Gilles Barnathanand Prof. Jean-Michel Kornprobst from the Faculty of Pharmacy(University of Nantes) for their opinion about the tribology phe-nomenon in our experiments.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.biortech.2012.07.105.

References

Balasubramanian, S., Allen, J.D., Kanitkar, A., Boldor, D., 2011. Oil Extraction fromScenedesmus obliquus using a continuous microwave system – Design,optimization and quality characterization. Bioresour. Technol., 102.

Barbarino, E., Lourenço, S.O., 2005. An evaluation of methods for extraction andquantification of protein from marine macro- and microalgae. J. Appl. Phycol.17, 447–460.

Barrett, J., Jeffrey, S.W., 1971. A note on the occurrence of chlorophyllase in marinealgae. J. Exp. Mar. Biol. Ecol. 7, 255–262.

Cartaxana, P., Brotas, V., 2003. Effects of extraction on HPLC quantification of majorpigments from benthic microalgae. Arch. Hydrobiol. 157, 339–349.

Chisti, Y., 2007. Biodiesel from microalgae. Biotechnol. Adv. 25, 294–306.Elsner, H.I., Lindblad, E.B., 1989. Ultrasonic degradation of DNA. DNA 8, 697–701.Hagerthey, S.E., Louda, J.W., Mongkronsri, P., 2006. Evaluation of pigment extraction

methods and a recommended protocol for periphyton chlorophyll adetermination and chemotaxonomic assessment. J. Phycol. 42, 1125–1136.

Halim, R., Gladman, B., Danquah, M.K., Webley, P.A., 2011. Oil extraction frommicroalgae for biodiesel production. Bioresour. Technol. 102, 178–185.

Harun, R., Singh, M., Forde, G.M., Danquah, M.K., 2010. Bioprocess engineering ofmicroalgae to produce a variety of consumer products. Renew. Sustain. EnergyRev. 14, 1037–1047.

Heaney-Kieras, J., Rodén, L., Chapman, D.J., 1977. The covalent linkage of protein tocarbohydrate in the extracellular protein-polysaccharide from the red algaPorphyridium cruentum. Biochem J. 165, 1–9.

Hejazi, M.A., Wijffels, R.H., 2003. Effect of light intensity on b-carotene productionand extraction by Dunaliella salina in two-phase bioreactors. Biomol. Eng. 20,171–175.

Van Heukelem, L., Thomas, C.S., 2001. Computer-assisted high-performance liquidchromatography method development with applications to the isolation andanalysis of phytoplankton pigments. J. Chromatogr. A 910, 31–49.

Hosikian, A., Lim, S., Halim, R., Danquah, M.K., 2010. Chlorophyll extraction frommicroalgae: a review on the process engineering aspects. Int. J. Chem. Eng. 2010,1–11.

Jeffrey, S.W., Mantoura, R.F.C., Wright, S.W., 1997. Phytoplankton pigments inoceanography: guidelines to modern methods. Monographs on OceanographicMethodology, UNESCO, Paris.

Kim, S.M., Jung, Y.-J., Kwon, O.-N., Cha, K.H., Um, B.-H., Chung, D., Pan, C.-H., 2012. Apotential commercial source of fucoxanthin extracted from the microalgaPhaeodactylum tricornutum. Appl. Biochem. Biotechnol..

Kornprobst, J.-M., 2005. Substances Naturelles d’origine Marine. Editions Tec & Doc.ed, Lavoisier, Paris.

Lan, S., Wu, L., Zhang, D., Hu, C., Liu, Y., 2011. Ethanol outperforms multiple solventsin the extraction of chlorophyll-a from biological soil crusts. Soil Biol. Biochem.43, 857–861.

Latasa, M., Lenning, K., Garrido, J.L., Scharek, R., Estrada, M., Rodríguez, F., Zapata, M.,2001. Losses of chlorophylls and carotenoids in aqueous acetone and methanolextracts prepared for RPHPLC analysis of pigments. Chromatographia 53, 385–391.

Lee, J.-Y., Yoo, C., Jun, S.-Y., Ahn, C.-Y., Oh, H.-M., 2010. Comparison of severalmethods for effective lipid extraction from microalgae. Bioresour. Technol. 101,S75–S77.

Machmudah, S., Shotipruk, A., Goto, M., Sasaki, M., Hirose, T., 2006. Extraction ofastaxanthin from Haematococcus pluvialis using supercritical CO2 and ethanol asentrainer. Ind. Eng. Chem. Res. 45, 3652–3657.

Maheswari, U., Jabbari, K., Petit, J.L., Porcel, B.M., Allen, A.E., Cadoret, J.P., DeMartino, A., Heijde, M., Kaas, R., La Roche, J., Lopez, P.J., Martin-Jézéquel, V.,Meichenin, A., Mock, T., Parker, M.S., Vardi, A., Armbrust, E.V., Weissenbach, J.,Katinka, M., Bowler, C., 2010. Digital expression profiling of novel diatomtranscripts provides insight into their biological functions. Genome Biol. 11,R85.

Mendes, R.L., Nobre, B.P., Cardoso, M.T., Pereira, A.P., Palavra, A.F., 2003.Supercritical carbon dioxide extraction of compounds with pharmaceuticalimportance from microalgae. Inorg. Chim. Acta 356, 328–334.

Molina Grima, E., Belarbi, E.-H., Acién Fernández, F.G., Robles Medina, A., Chisti, Y.,2003. Recovery of microalgal biomass and metabolites: process options andeconomics. Biotechnol. Adv. 20, 491–515.

Palavra, A.M.F., Coelho, J.P., Barroso, J.G., Rauter, A.P., Fareleira, J.M.N.A., Mainar, A.,Urieta, J.S., Nobre, B.P., Gouveia, L., Mendes, R.L., Cabral, J.M.S., Novais, J.M.,2011. Supercritical carbon dioxide extraction of bioactive compounds frommicroalgae and volatile oils from aromatic plants. J. Supercrit. Fluids 60, 21–27.

Pasquet, V., Chérouvrier, J.-R., Farhat, F., Thiéry, V., Piot, J.-M., Bérard, J.-B., Kaas, R.,Serive, B., Patrice, T., Cadoret, J.-P., Picot, L., 2011. Study on the microalgalpigments extraction process: performance of microwave assisted extraction.Process Biochem. 46, 59–67.

Percival, E., Foyle, R.A.J., 1979. The extracellular polysaccharides of Porphyridiumcruentum and Porphyridium aerugineum. Carbohydr. Res. 72, 165–176.

Petrier, C., Jeunet, A., Luche, J.L., Reverdy, G., 1992. Unexpected frequency effects onthe rate of oxidative processes induced by ultrasound. J. Am. Chem. Soc. 114,3148–3150.

Pietra, F., 1997. Secondary metabolites from marine microorganisms: bacteria,protozoa, algae and fungi. Achievements and prospects. Nat. Prod. Rep. 14, 453–464.

Plaza, M., Santoyo, S., Jaime, L., García-Blairsy Reina, G., Herrero, M., Señoráns, F.J.,Ibáñez, E., 2010. Screening for bioactive compounds from algae. J. Pharm.Biomed. Anal. 51, 450–455.

Porra, R.J., 1991. Recent advances and re-assessments in chlorophyll extraction andassay procedures for terrestrial, aquatic, and marine organisms, includingrecalcitrant algae, in: chlorophylls. Hugo Scheer, Boca Raton, pp. 31–57.

Sasso, S., Pohnert, G., Lohr, M., Mittag, M., Hertweck, C., 2011. Microalgae in thepostgenomic era: a blooming reservoir for new natural products. FEMSMicrobiol. Rev., 1–25.

Suslick, K.S., 1994. The Year Book of Science and Future. Encyclopedia Britannica,Chicago.

Szymczak- _Zyła, M., Kowalewska, G., Louda, J.W., 2011. Chlorophyll-a andderivatives in recent sediments as indicators of productivity and depositionalconditions. Mar. Chem. 125, 39–48.

Wiyarno, B., Yunus, R.M., Mel, M., 2010. Oil algae extraction from Nannochloropsissp.: a study of soxhlet extraction and ultrasonic assisted extraction. Presentedat the International Conference on Process Engineering and Advanced Materials2010 (ICPEAM 2010), Kuala Lumpur.

Wright, S.W., Jeffrey, S.W., Mantoura, R.F.C., 1997. Evaluation of methods andsolvents for pigment extraction. In: Phytoplankton Pigments in Oceanography:Guidelines to Modern Methods. Monographs on Oceanographic Methodology,UNESCO, pp. 261–282.

Extração de metabolitos

Documents

Transcript of Extração de metabolitos