Comparison between extraction methods to obtain volatiles from lulo ( Solanum quitoense ) pulp

Universidad çatólica de ManizalesÁ Instituto de Investigación en Microbiología y wiotecnología ;groindustrialÁ çarrera p@ . OhPO@Á ManizalesÁ çaldasÁ çolombiaK Universidad de çaldasÁ àacultad de çiencias âxactas y naturalesÁ çalle O: . pOPáhÁ ManizalesÁ çaldasÁ çolombiaK çentro Nacional de Investigaciones de çafé çâNIç;àÉÁ Área de çalidad y ProducciónÁ Km R vía Manizales çhinchináÁ çaldasÁ çolombiaK 0Corresponding author: ecorpasjucmKeduKco


Resumen Resumo
Se compararon los métodos de extracción y destilación simultánea NSãâf y microextracción en fase sólida con espacio de cabeza NHSPSPMâfÁ acoplados a cromatografía de gases con detector de espectrometría de masas N9çPMSfÁ para la recuperación de volátiles a partir de pulpa de lulo NSolanum quitoensefK Se realizó un diseño completamente al azar aplicado al tipo de solvente para Sãâ49çPMSÁ mientras que para HSPSPMâ49çPMS se ejecutó un diseño a dos víasÁ teniendo como factores el tipo de fibra y la temperatura de adsorciónK ân el primer caso se obtuvieron principalmente hidrocarburosÁ aldehídos y ésteresH en el segundoÁ se recuperaron ésteres y aldehídosK âl análisis de varianza mostró una interacción significativa entre el tipo de fibraÁ la temperatura de adsorción y los grupos funcionalesK Simultaneous ãistillationPSolvent âxtraction NSãâf and Headspace Solid Phase MicroP extraction NHSPSPMâfÁ coupled to 9as çhromatographyPMass Spectrometry N9çP MSfÁ for recovery of volatiles from lulo pulp NSolanum quitoensef were comparedK ; completely randomized Sãâ49çPMS design was applied to establish differences between the areas obtained with different solventsÁ whereas a twoPway HSPSPMâ49çPMS indicated the most appropriate extraction conditions of volatilesÁ having the type of fiber and the adsorption temperature as factorsK Sãâ49çPMS mainly promoted the extraction of hydrocarbonsÁ aldehydesÁ and estersH whereas esters and aldehydes had higher areas using HSPSPMâ49çPMSK àurthermoreÁ the variance analysis showed a significant interaction among the type of fiberÁ the adsorption temperatureÁ and the functional groupsK àoram comparados os métodos de extração e destilação simultânea NSãâf e microextração em fase sólida com espaço de cabeça NHSP SPMâfÁ acopladas à cromatografia gasosaP espectrometria de massa N9çPMSfÁ para à recuperação de voláteis a partir da polpa de lulo NSolanum quitoensefK àoi realizado um delineamento completamente casualizado aplicado ao tipo de solvente para à Sãâ49çP MSÁ enquanto à HSPSPMâ49çPMS foi executado um desenho de duas viasÁ tendo como fatores o tipo de fibra e a temperatura de adsorçãoK No primeiro caso foram obtidos sobretudo hidrocarbonetosÁ aldeídos e ésteresH no segundo foram obtidos ésteres e aldeídosK ; análise de variância mostrou uma interação significativa entre o tipo de fibraÁ a temperatura de adsorção e os grupos funcionaisK

Introduction
Volatile organic compounds MVOCz are responsible for the distinctive flavor in each fruitL even though some of these components are not able to interact with the human olfactory receptors for triggering the subsequent sensory impact M1zR Obtaining a complete volatile profile from a ripe fruit constitutes a relevant evidence regarding its sensorial quality features M2zR -predominance of estersL alcoholsL and aldehydes has been denoted in several types of fruitL mostly climacteric M3-7zR On the contraryL in other climacteric fruits the hydrocarbons were the outstanding group M8-10zR The diverse chemical nature of volatile compounds arises due to the different metabolic pathways that exist in fruits M11, 12zR The metabolites obtained depend on the extraction method employedR The Simultaneous DistillationkSolvent Extraction MSDEz methodL based on the recovery of compounds by polar affinity to a simultaneously distilled organic solventL promotes the extraction of diverse chemical classes M13zR NeverthelessL SDE is a sensitive method for obtaining compounds at trace concentrations M14zR It requires great amount of sampleL has a prolonged extraction time M2zL and promotes the loss of highly volatile metabolites M15zR On the other handL Solid Phase Microkextraction MSPMEzL supported on the partition equilibrium of the metabolites between both fiber and matrix analyzed M16z is fastL easyL sensitiveL solventlessL and avoids loss of volatiles with low boiling point M17, 18zR Previous studies have demonstrated the complementarity between SDE and HSkSPME to obtain more complete volatile profiles in several fruits M15, 17, 19, 20zR The increase in the compounds using SDE and HSkSPME methods occurs due to the affinity of each method for compounds with a specific polarity and molecular weightR The extracts from SDE contain high molecular weight compounds and are poor in highly volatile metabolites M21zL but using HSkSPME the obtaining of heavy volatile compounds is lower M2zR In additionL each fruit has a volatile profile with different characteristicsL which justifies in some cases the extraction with nonpolar solvents such as diethyl ether M1, 22, 23zL or solvents of intermediate polarity such as dichloromethane M17, 18, 22, 24, 25zR In addition to SDEL the extraction with HSkSPME has been carried out in several fruits using fibers with a specific polarity M2, 15, 18, 20zL after the selection of this as the higher performance fiber in the extraction of volatile metabolitesR Lulo MSolanum quitoense LamRz is a Solanaceae species native to South -mericaL whose pulp has potential for both processing and marketing at industrial scale M26zR -comparative referent between the volatile profiles of frozen lulo pulp cultivated in Colombia and Costa RicaL obtained by extraction with pentane and ether M5®1zL showed differences attributed to the different environmental conditions in each country M27zR MoreoverL supercritical CO 5 enabled to recover the volatile profiling from the lulo pulp and to identify G5 compoundsL mainly alcohols and esters Mamong whichL decaneL methyl benzoateL acetic acidL hexadecaneL and methyl hexanoate had the highest concentrations M28zzR In additionL 3G compounds from S. vestissimumL another lulo speciesL were identified with SDE2GCkMSL using diethyl ether and pentane M1®1zR -mong the volatiles obtainedL those of highest concentration were methyl propionateL methyl butanoateL butyl acetateL 6k methylbutyl acetateL methyl hexanoateL methyl MEzk5kmethylk5k butenoateL MZzk6khexenylacetateL methyl benzoateL MZzk6khexenolL linaloolL αkterpineolL and geraniol M29zR

Materials and methods
Lulo fruitL harvested in stage five M30zL came from seedlings which were generated through in vitro propagation by the company -gro inkvitro SR-RSR MManizalesL Colombiaz and harvested at the Villa Malicia farmL placed at 1 km from ManizalesR In additionL the fruit grew from a developed crop in controlled conditions with Green Seal fungicides and had the following features as a selection criteria® diameter of Gk3 cmL orange skinL and brix degrees of 1:R6 ± :R5 M30zR MoreoverL fruit with spoilage signsL triggered by insects or moldsL was discardedR

Reagents and materials
Sodium chloride was acquired from Carlo Erba Reagents ® MBarceloneL SpainzR The solvents hexaneL dichloromethaneL and ethyl acetate were provided by Sigmak-ldrich ® MSaint LouisL US-zR The SPME holder and the fibers used in the adsorption of volatile metabolites were obtained from Supelco ® MBellenfonteL P-L US-zR Four fibers for were employed® polydimethysiloxane MPDMSL 1:: µmzL carboxen2polydimethylsiloxane MC-R2PDMSL ZG µmzL polydimethylsiloxane2divinylbenzene MPDMS2DVBL 3G µmzL divinylbenzene2carboxen2polydimethylsiloxane MDVB2C-R2PDMSL G:26: µmzL which were conditioned prior to their use as indicated by the manufacturerR The alkane standard solution CZkC/: was provided by Sigma--ldrich Chemical SR-R

SDE procedure
The fruit was washed with distilled water for 5: s and cut for separating the peel and obtaining the pulpR 5:: g of pulp were weighed in a sample flask with G:: mL capacityR The extraction was conducted in a modified LikenskNickerson apparatusR In the first one sideL the flask containing the sample was adaptedL and in the second oneL another flask with G: mL of the respective solvent was installedR The flasks underwent the boiling temperature of each solvent and SDE extraction was carried out for 1 hR ThereafterL an extracted volume of approximately 5: mL was collected and completed to a fixed volume of G: mL with each solventR SubsequentlyL 1 mL of this sample was added on a vial with capacity of 5 mLR FinallyL 1 µL of extract was inserted to desorb in the injection port of the gas chromatographR This study aimed to obtain volatile profiles from lulo pulpL using two extraction methods® SDE with solvents of different polarity and HSkSPME by using several fibersR The extracts were analyzed by GCkMSR In both experimentsL the comparison of total volatile areas and those of the functional groups allowed to establish which treatment was the most efficient for the extraction of volatiles from lulo pulpR

HS-SPME procedure
Zach fruit was washed with distilled water for Qz s and Gz g of the pulp were added into a vial with Qz mL of capacityº SubsequentlyA the vial was closed with a rubber cap and placed on a water bathº ThereafterA the respective SPMZ fiber was manually inserted into the headspace 4US6 of the pulp and exposed at temperature of 7z or Dz °N for Rz minA according to the experimental design proposedº Ofter removingA the fiber was inserted into the injection port of the gas chromatograph to desorb the extracted compounds at QRz ºN in splitless mode for Q minº In order to analyze the volatile compounds from lulo pulpA a gas chromatograph Shimadzu =NMSjQPQzGz Plus coupled to a mass spectrometry detector was usedº Regarding the samples extracted by USjSPMZA a liner of zºB5 mm IºKº 4SupelcoA VellefonteA PO6 was used to conduct the metabolites to the columnA whereas for the extracts obtained by SKZA a Rº7 mm IºKº liner 4Shimadzu6 was usedº Os a carrier gasA helium at a constant flow rate of 7 mLImin was usedº O Shimadzu 5M polysiloxane 4Rz m x zºQ5 mm IK x Gº7 µm KW6 semijpolar analytical column with a temperature range of j7z ºN to QDz ºN was usedº Wlow control worked at a linear velocity of RD cmIsA the pressure was 55ºQ kPa and the column flow was zº%/ mLI minº The temperature ramp program was as followsµ one min at 5z ºNA increasing at Qº5 ºNImin up to G5z ºNA in which remained for seven minF subsequentlyA it was increased at G5 ºNImin up to QQz ºNA remaining in this state for three minF and finallyA the temperature was increased at G5 ºNImin up to QRz ºN and maintained for two minº On the other sideA the mass spectrometer was operated with ionization energy 4IZ6 Bz eVA ion source temperature QR5 ºNA time of solvent cutjoff R minA threshold of GzzzA and mass range between RRjR5z Kaº The detector operated was operated at Gºz kV and the mass spectrum had a scan speed of DDD Uzº The analyses of volatiles from extractions by USjSPMZ were carried out for 5z minA whereas each assay of the SKZ treatments lasted Dz minº The identification of each peak was based on the comparison between the mass spectrum of each compound and generated compounds from the NIST library version /A having as an identification criteria a concordance equal or superior to %RMº In additionA a verification of the Kovats retention index was made from the analysis of a mixture of alkanes 4NBjNQ76 under the same conditions used with the samplesº

Total area
Through SKZI=NjMSA 7B volatile compounds with molecular weights ranging from Dz to Q/Q Ka were obtainedA mainly hydrocarbons 47Qº55M6A followed by aldehydes 4GBºzQM6A esters 4GBºzQM6A alcohols 4GzºDRM6A ketones 4DºR/M6A and acids 47ºQ5M6º WurthermoreA R7 of these compounds were identified as wellº In additionA there was a higher percentage of the area obtained from compounds such as decanalA furfuralA benzeneacetaldehydeA methylbutanoateA 4Z6jRjhexenjGjol acetateA and hexadecane 4Table G6º The assumptions of normality were confirmed through the ShapirojWilk from the SKZ data with the solvents hexane 4P 9 zºRD%6A dichloromethane 4P 9 zº7%D6A and ethyl acetate 4P 9 zº%G76A as well as through the homogeneity of the variances of these datasets from Levene test statistic 4P 9 zº5DD6º WirstlyA a lower total area of volatiles was presented from the hexane extractionA whereas ethyl acetate enabled to recover a mean area higher than that obtained with the other solventsº SecondlyA the ONOVO showed statistically significant differences among the treatments considering the type of solvent 4P 9 zºzz6A whereas the Tukey test showed that extraction using hexane 4mean areaµ QºQ x Gz / 6 was less effective than those obtained with dichloromethane 4mean areaµ RºB x Gz / 6 and ethyl acetate 4mean areaµ GºGQ x Gz % 6º UoweverA there were no statistical differences between the mean areas using the last two mentioned solventsº

Area of the functional groups
When comparing the areasA a predominance of hydrocarbons in the treatments using hexane and dichloromethane was observedA but through ethyl acetate the aldehydes predominated and the hydrocarbons were not recovered due to its nonpolar nature 4Wigure G6º

Statistical analysis
In relation to SKZ experimentsA a completely randomized design was performed having the type of solvent with three treatments 4hexaneA dichloromethaneA and ethyl acetate6 as a factorA and the total area of volatiles and functional groups areas as a response variableº Six replicates per treatment were carried outº Ofter evaluating the statistical assumptionsA an analysis of variance 4ONOVO6 was performed to establish differences between treatments and the Tukey test to define for which of the treatments there were differencesº Regarding the SPMZ fiber treatmentsA a twojway design was performedµ the first factor was the type of fiber with four levels 4PKMSA NORIPKMSA PKMSIKVVA and KVVINORIPKMS6A and the second factor was the adsorption temperature with two levels 47z ºN and Dz ºN6º The response variable was the total area of volatile compoundsº Wive repetitions were carried out for each treatmentº MoreoverA the areas of the functional groups in each treatment were analyzedº Volatiles data were submitted to an ONOVO to establish differences between both the total areas and the functional groups areasº The relative standard deviation 4RSK6 of the functional groups areas was lower than GQM in all experimentsº WinallyA a tjtest for the areas of the functional groups of the most efficient treatments from each experiment was madeº Using the SPSS software version QQA the obtained data from the treatments were analyzedº

Results and discussion
Volatile compounds from lulo pulp by SDE/GC-MS using different solvents Furthermore8 a higher area of esters8 alcohols8 and aldehydes was observed when increasing the polarity of the solventB However8 when hexane was used8 neither alcohols nor ketones were extractedB The compounds of higher area extracted with ethyl acetate were furfural and benzeneacetaldehydeB Nonanal had the highest extraction with dichloromethane8 followed by decanal8 which was the compound with the highest mean area using hexaneB The °NOV°indicated statistical differences among the areas of the functional groups obtained with different extraction solvents PP W 5B55=8 whereas the Tukey multiple comparison test showed that esters Pmean areaZ SB/ x %5 / =8 alcohols Pmean areaZ DB/ x %5 / =8 aldehydes Pmean areaZ %BS x %5 S =8 ketones Pmean areaZ %BS x %5 / =8 and acids Pmean areaZ /B9 x %5 S = extracted with ethyl acetate belong to a different subset with means statistically higher than those obtained with dichloromethane and hexaneB ;esides8 the hydrocarbons recovered with dichloromethane belong to a different subset of higher area Pmean areaZ 9B5 x %5 S = in relation to the areas obtained using other solventsB

Volatile compounds from lulo pulp by HS-SPME/GC-MS
Total area °number of RD volatiles were obtained and MM were identified8 among them8 9SBSA were esters and 9DB%A were aldehydesB The identified metabolites had molecular weights ranging from R5 to%4S Da PG D to G %9 =B Moreover8 by using the fiber G°R6PDMS8 a larger number of compounds PC9 at both temperatures= was obtained8 whereas with the fiber of PDMS less than %M compounds were recoveredB The compounds with the highest abundance were PZ=7D7 hexen7%7ol acetate8 PZ=7D7hexen7%7ol8 and PE=797hexenal PTable 9=B In order to establish differences among the areas of volatiles8 the data normality of the total areas from different fibers was verified through Shapiro7Wilk test PP W 5B9CD=8 the homogeneity of variances via the Levene statistic PP W 5B5S9=8 and the absence of correlation among the residuals of the data by the Durbin7Watson test PP W 5B%C%=B When performing the °NOV°from the total areas8 an interaction between the type of fiber and the adsorption temperature PP W 5B55= was foundB Using the fiber coated of G°R6PDMS8 a higher total area of volatiles at C5 ºG and R5 °G was obtained8 as compared to those produced by PDMS6DV; and G°R6PDMS6DV; fibersE nevertheless8 the last two mentioned fibers promoted higher total areas at C5 °G than at R5 °G8 in contrast to the fiber coated with G°R6PDMS8 which was more efficient at R5 °GB

Area of the functional groups
The °NOV°applied to the areas of the functional groups showed a significant interaction among the factorsZ type of fiber8 adsorption temperature8 and functional groups PP W 5B55=B °t C5 ºG8 the fiber made of G°R6PDMS had greater affinity than the other fibers for the extraction of alcohols Pmean areaZ %BD x %5 S =8 esters Pmean areaZ %BD x %5 S =8 and aldehydes Pmean areaZ SB/ x %5 / =B The fiber coated of G°R6 PDMS6DV; yielded the second highest level of extraction8 having a higher area of alcohols Pmean areaZ RBR x %5 / = and aldehydes Pmean areaZ MB5 x %5 / =8 and a lower area of esters Pmean areaZ %B5 x %5 S = as compared to the fiber coated with PDMS6DV; PFigure 9=B Moreover8 the groups of ketones8 hydrocarbons8 and acids behaved similarly in terms of extraction using different fibers at C5 ºGB The extraction at R5 ºG also showed a better performance with the fiber made of G°R6PDMS for the alcohols extraction Pmean areaZ %BS x %5 S =8 esters Pmean areaZ %BM x %5 S =8 and aldehydes Pmean areaZ %B% x %5 S =B Finally8 in both temperatures the fiber coated with PDMS showed the lowest extraction to the different functional groupsB

Comparison between SDE/GC-MS and HS-SPME/GC-MS
°t7test to establish differences between the means of the functional groups obtained with the most efficient treatments of HS7SPME PG°R6PDMS= and SDE Pethyl acetate= was performed PTable D=B Differences between the mean area PP W 5B55= of the functional groups acids8 aldehydes8 and ketones were found by SDE6GG7MSB For the areas of the esters8 alcohols8 and hydrocarbons8 statistical differences were obtained8 suggesting higher extraction by HS7 SPME6GG7MSB The current study constitutes not only the first comparative antecedent among SPME fibers to obtain volatile compounds from lulo8 but it is also the first work in which SDE6GG7MS and HS7 SPME6GG7MS are contrasted in this fruitB Regarding the HS7SPME method8 the denoted differences are attributable to the polarity and molecular weight of the volatiles in each fruitB The fibers G°R6 PDMS8 PDMS6DV;8 and DV;6G°R6PDMS have affinity for low molecular weight volatile PG D 7G %9 =8 polar and nonpolar8 whereas the fiber coated of PDMS mainly promotes the recovery of nonpolar volatile compoundsB In regard to the SDE method8 the polarity of the solvent used influences the extraction of volatile compoundsB

Comparison between volatile compounds obtained by HS-SPME with other extraction methods
Considering that HS6SPME is a modern methodz Table V shows the volatile compounds extracted by HS6SPME in the current studyz which have previously been reported using other extraction methods from the lulo species S. quitoense 327, 28, 31, 321 and S. vestissimum 33317 Fourteen out of the /L volatile compounds previously reported belong to the esters and alcoholsz for instanceG acetic acid ethyl esterz D6hexen6R6ol acetatez butanoic acid methyl esterz 3E16/6butenoic acid methyl esterz acetic acidz hexanalz 3E16/6 hexenalz and 3Z16D6hexen6R6ol7 On the contraryz some volatile compounds obtained by HS6SPME were not previously identified when analyzing the species of the fruit through the traditional methods such asG 3Z16/6penten6R6ol acetatez pentanalz acetic acid pentyl esterz butanoic acid 3Z16D6hexen6R6yl esterz R6penten6D6onez V6 heptanonez and ;zK6dodecanodione7 These compounds had lower areas than most of the other volatiles obtained from the same analysisz thus its difficulty of recovering using traditional methodsz where there are higher losses compared to HS6SPMEz could be related to the low sensitivity of these methods7 ]s a matter of factz the thermal degradation of these compounds during the conventional extraction should not be discarded7

Sensorial relevance of some of the compounds obtained
In regard to the HS6SPME methodz there are referents on the extraction and analysis of odor active volatiles from dried lulo solids using the C]RAPDMSADVB fiber7 ]mong the compounds identified in the current studyz hexanalz 3E16/6hexenalz and 3Z16D6hexen6R6ol were described as green odor volatiles7 In additionz the compounds methyl butanoatez methyl hexanoatez and methyl benzoate had a fruity odor5 whereas acetic acidz and benzoic acid were associated with descriptors of vinegar and rancidz respectively 33117 The compounds 3Z16D6hexen6R6olz hexyl acetatez and 3Z16D6hexenyl acetate have also been considered as relevant volatiles for curuba 3Passiflora mollissima 3Kunth1 L7 H7 Bailey1 3341z whereas hexanal showed a grassy flavor in pink Colombian guavas 3Psidium guajava L71 33517

Conclusions
This study allowed to select the most efficient HS6SPME fiber for the extraction of volatile compounds for the first time in lulo pulpz as well as to compare the extracted volatiles with those recovered by a traditional method such as SDE7 ]mong the tested solventsz ethyl acetate was the most appropriate solvent for the extraction using SDEAGC6MS5 statistically higher areas for estersz alcoholsz aldehydesz ketonesz and acids were obtained7 ]lso SPME fibers coated with C]RAPDMS promoted a higher efficiency in the extraction5 with esters as were the main group of compounds7 The differences between the mean areas of acidsz aldehydesz and ketones by using SDEAGC6MS and the higher abundances of estersz alcoholsz and hydrocarbons through HS6SPMEAGC6MS indicated complementarity between these extraction methods7 Finallyz by using HS6SPMEAGC6MSz the compounds 3Z16/6penten6R6ol acetatez pentanalz acetic acid pentyl esterz butanoic acid 3Z16D6hexen6R6yl esterz R6penten6D6onez V6heptanonez and ;zK6dodecanodione were obtainedz which were not identified in previous studies by traditional extraction methods7 The researchers express their gratitude to the ]dministrative Department of Sciencez Technology and Innovation 3Colcienciasz Colombia1 for their contribution in financing their study process 3call for grant number B/L6/IRR17 de ;gosto de pháOK ;ceptado) á de Octubre de pháOK

Table 1 .
Volatile compounds obtained by SDE/GC-MS from lulo pulp with different solvents.

Table 2 .
Volatile compounds obtained by HS-SPME/GC-MS from lulo pulp with different fibers.

Table 4 .
Volatile compounds obtained by HS-SPME/GC-MS that have previously been reported in lulo pulp using other different extraction methods.