Determination of the extraction kinetics for the quantification of polyhydroxyalkanoate monomers in mixed microbial systems

Page 1

Dp

AMAa

b

G

a

ARRAA

KPPPPE

1

im[iepenrmc

pmi

1h

Process Biochemistry 48 (2013) 1626–1634

Contents lists available at ScienceDirect

Process Biochemistry

jo u r n al homep age: www.elsev ier .com/ locate /procbio

etermination of the extraction kinetics for the quantification ofolyhydroxyalkanoate monomers in mixed microbial systems

na B. Lanhama, Ana Rita Ricardoa, Maria G.E. Albuquerquea, Filipa Pardelhaa,ónica Carvalheiraa, Marta Comab, Joana Fradinhoa, Gilda Carvalhoa,

drian Oehmena,∗, Maria A.M. Reisa

REQUIMTE/CQFB, Departamento de Química, Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa, Caparica, PortugalLaboratory of Chemical and Environmental Engineering (LEQUIA-UdG), Institute of the Environment, University of Girona, Campus Montilivi s/n, E-17071irona, Spain

r t i c l e i n f o

rticle history:eceived 16 May 2013eceived in revised form 17 July 2013ccepted 26 July 2013vailable online 2 August 2013

a b s t r a c t

For the first time, a systematic approach was conducted to determine the key factors influencing thekinetics of hydroxyalkanote (HA) extraction in biological systems. Six mixed microbial systems wherepolyhydroxyalkanoate (PHA) is produced were evaluated. Experiments were carried out for full-scaleand lab-scale activated sludge systems using different configurations (containing floccular or granularsludge), as well as specific PHA accumulating cultures that contain high or low intracellular PHA fractions.

eywords:olyhydroxybutyrate (PHB)olyhydroxyvalerate (PHV)olyhydroxy-2-methylbutyrate (PH2MB)olyhydroxy-2-methylvalerate (PH2MV)xtraction and recovery

The overall reaction was limited by the kinetics of the PHA hydrolysis in floccular cultures, whereas ingranular cultures, it was limited by the cell lysis step. The monomeric composition of the polymer alsohad an impact on the HA extraction rate: higher acid concentration and a longer digestion time should beemployed when cells accumulate monomers with more substituents, such as hydroxy-2-methylbutyrate(H2MB) and hydroxy-2-methylvalerate (H2MV). This study optimised the method for HA extraction,which impacts the assessment of the quantity and quality of PHA biopolymers.

. Introduction

Polyhydroxyalkanoate (PHA) are polymers composed by a fam-ly of polyester monomers which include over 110 different

olecules with varying backbone length from 4 to 16 carbons1]. PHA synthesis from an external carbon source has beendentified in a great number of different bacteria (e.g. Ralstoniautropha, Azotobacter vinelandii, Candidatus Accumulibacter phos-hatis), which utilise this polymer as a carbon, energy and reducingquivalents storage product when facing limiting conditions (e.g.utrients or oxygen) [2,3]. PHA is accumulated inside the bacte-ial cells as discrete granules that are often composed of differentonomers, forming a co-polymer, e.g., poly-(3-hydroxybutyrate-

o-3-hydroxyvalerate) (co-PHB-PHV).PHA is a biodegradable and renewable alternative to the fossil-

roduced polyesters that currently hold the majority of the plastics

arket. Using pure or mixed cultures fed on waste material or

ndustrial by-products, several authors were able to show the

∗ Corresponding author. Tel.: +351 212948571; fax: +351 212948550.E-mail address: [email protected] (A. Oehmen).

359-5113/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.procbio.2013.07.023

© 2013 Elsevier Ltd. All rights reserved.

feasibility of producing a new generation of more sustainable andbiocompatible plastics [4–6].

Additionally, PHA also plays a significant role as a storage poly-mer in environmentally engineered processes [3], where variousorganisms rely on PHA in order to survive in these systems. There-fore, PHA is included in biological wastewater treatment models[7] and is often a key compound when addressing the metabolismof bacteria present in such systems [8,9].

While PHA is a highly important biopolymer to both the bio-plastics and wastewater treatment industries, very little focus hasbeen paid to the extraction step necessary to recover hydroxyalka-noate (HA), particularly from mixed microbial cultures [4]. This isof particular importance for the quantification of the individualmonomers, which have a high impact on the mechanical propertiesof the polymer, and furthermore impact the determination of thekinetic and stoichiometric parameters associated with mathemat-ical models. The method that is most widely used for HA extractioninvolves an acidic methanolysis reaction, which was initially devel-oped by Braunegg et al. [10] for a PHB producing pure culture, andlater adapted for HV and medium-chain length (MCL) monomers

[11–13]. In mixed microbial cultures, microorganisms are oftennot in suspension but form complex and heterogeneous structures,such as flocs or granules, aggregated due to extracellular polymericsubstances. The biomass properties are likely to affect HA recovery,

Page 2

iochem

wwtb

stwpm2((wiics

2

2

sca

Maf[owoH

baPtaf

eutpgami

lawmImH

m(p(

A.B. Lanham et al. / Process B

here the effect of e.g. higher cell packing in the case of granulesas also found to impact glycogen extraction [14]. Nevertheless,

he factors affecting HA extraction kinetics have never been studiedefore.

This work used a comparative approach in different microbialystems to systematise the factors impacting HA extraction. Forhe first time, a systematic approach using design of experimentsas conducted for six mixed microbial cultures from differentrocesses, with different microbial characteristics producing theost common HA monomers found in these systems (HB, hydroxy-

-methylbutyrate (H2MB), HV and hydroxy-2-methylvalerateH2MV)). These cultures included PHA accumulating organismshigher PHA content), activated sludge (lower PHA content) asell as floccular and granular systems. The determination of the

mpact of these factors on different microbial systems accumulat-ng various co-polymers is a valuable tool to guide researchers inharacterising the quantity and quality of the PHA in their specificystem.

. Materials and methods

.1. Microbial cultures tested

Six different mixed microbial cultures were selected for thistudy, which displayed different PHA accumulation capacities andompositions. These cultures result from ongoing work at theuthors’ laboratories.

The first and the second microbial cultures tested (MC1 andC2) were from the same PHA producing culture, maintained in

800 mL sequencing batch reactor (SBR), run in aerobic dynamiceeding conditions and fed with fermented sugar cane molasses15]. MC1 was obtained through a PHA accumulation batch test, inrder to achieve a higher content of internal PHA (30 wt%). MC2as withdrawn from the selection SBR and contained only 1 wt%

f PHA. MC1 and MC2 contained a co-polymer composed of mainlyB (72–78 wt%) and HV (22–26 wt%).

The third microbial culture (MC3), consisted of a mixture ofoth bacteria and algae, obtained from a 4 L lab-scale SBR, oper-ted under anaerobic and photosynthetic conditions [16]. ThisHA accumulating culture was fed with a synthetic wastewa-er, containing acetate as the sole carbon source. MC3 containedpproximately 5 wt% PHA, mainly composed of HB (97 wt%) and araction of HV (3 wt%).

The fourth microbial culture (MC4) was collected from a 2 Lnhanced biological phosphorus removal (EBPR) SBR, operatednder anaerobic–aerobic conditions, fed with synthetic wastewa-er containing 75% acetate and 25% propionate [17]. Its microbialopulation was aggregated into a mixture of flocs and small-sizedranules and consisted of a high enrichment in polyphosphateccumulating organisms. MC4 contained approximately 6 wt% PHA,ainly composed of HB (57 wt%) and HV (32 wt%) but also contain-

ng 3 wt% H2MB and 8 wt% H2MV.The fifth microbial culture (MC5) was obtained from a 30 L

ab-scale SBR performing biological nutrient removal (BNR), oper-ting under alternating anaerobic–anoxic–aerobic conditions, fedith domestic wastewater (Quart WWTP, Girona, Spain) [18]. Theicrobial population was aggregated in both flocs and granules.

t contained a low amount of PHA (less than 1 wt%) composed ofainly HB (66 wt%) and HV (29 wt%), but also of small fractions of2MB (3 wt%) and H2MV (2 wt%).

The sixth microbial culture was sludge from a wastewater treat-

ent plant (WWTP), sampled either from Aalborg West WWTP

Aalborg, Denmark), operated under a BiodeniphoTM system cou-led with side-stream hydrolysis (MC6a), or from Beirolas WWTPLisbon, Portugal) and operated under sequential anaerobic, anoxic

istry 48 (2013) 1626–1634 1627

and aerobic conditions (MC6b) [19]. Both of these cultures were fedwith mainly domestic wastewater, and contained approximately2 wt% PHA composed mainly of HB (80 wt%) and HV (20 wt%). MC6balso contained a small fraction of less than 3 wt% of H2MB andH2MV.

2.2. HA extraction and analysis method

The PHA contained in each biomass underwent a digestion pro-cedure, where the HA were extracted, prior to analysis by gaschromatography. The biomass digestion involves (1) cell lysis, (2)hydrolysis of PHA into methyl-ester monomers (HA), and (3) amethanolysis reaction.

All microbial samples were centrifuged at 10,000 × g, the super-natant was discarded and the pellet was freeze-dried in a VarianDS102 freeze-drier over night. A precise amount of lyophilisedbiomass (error of 0.01 mg) was weighed into a Pyrex® tube, where1 mL of acidic methanol (3–20% sulphuric acid v/v) and 1 mL ofchloroform (Fluka) were added. The chloroform solution contained1 mg/mL heptadecane (Fluka) as internal standard. The tubes weresealed with an air tight Teflon®-lined screw cap and incubated at100 ◦C in a dry-heat thermo-block for the necessary time (1–20 h).The tubes were then cooled on ice for 30 min. Water (0.5 mL) wasadded to aid the two phase separation and the phases were mixedusing a vortex for 1 min. The lower phase, containing the chloro-form, was extracted into a GC vial and dried using molecular sieves(4 A, Prolabo) to remove traces of water.

2 �L of sample were injected in a Varian CP-3800 gas chromato-graph (Varian, CA, USA) equipped with a FID detector and a ZB-WAXplus column (60 m, 0.53 mm internal diameter, 1 �m film thickness,Phenomenex, USA) coupled with a guard-column (0.32 mm inter-nal diameter). Helium was used as a carrier gas, at constant pressure(14.5 psi). The temperature of injection was 280 ◦C, the tempera-ture of the detector was 230 ◦C and the temperature ramp startedat 40 ◦C, then increased at a rate of 20 ◦C/min until 100 ◦C, furtherincreased at a rate of 3 ◦C/min until 175 ◦C, and finally increasedagain at 20 ◦C/min until 220 ◦C, to ensure a cleaning step of thecolumn after each injection.

A co-polymer of PHB-PHV (88:12 wt, Aldrich) was used as astandard for HB, HV and H2MB, while 2-hydroxy-caproic acid(Aldrich) was used as a standard for H2MV. Standards wereprocessed in the same way as the samples, after being dissolvedinto a chloroform solution.

Results were presented as the response factor for each com-pound in terms of area of the selected peak (A), divided by the areaof the internal standard peak (Ais) – A/Ais. This procedure correctedfor error in the volume measurement in the reaction process as wellas errors in the volume of injection in the gas chromatograph. Theresults were then normalised with the amount of biomass weighed.

To find the concentration of polymer, the area of each peak,divided by the area of the internal standard, was calibrated using a6 point calibration curve.

2.3. Factors impacting extraction, design of experiments andresponse surface modelling

Initially, a series of tests were performed to assess the individ-ual effect of key factors affecting HA extraction kinetics, and also todefine their appropriate range for design of experiments (DOE) test-ing, which was employed to assess the interaction between each

factor, as described through response surface modelling (RSM). Tothe best of our knowledge, this is the first time that the DOE andRSM tools have been employed to describe the factors affecting HAextraction.

Page 3

1628 A.B. Lanham et al. / Process Biochem

Table 1Description of the conditions of the experiments conducted for each microbial cul-ture; each experiment was performed in duplicate.

Exp. no. Acid concentration (%) Digestionduration (h)

Biomass concentration(mg/mL) CHCl3

1 3 3 22 11.5 11.5 23 3 3 84 3 20 25 20 3 26 11.5 11.5 87 11.5 3 58 20 3 89 3 20 8

10 20 20 211 3 11.5 512 20 20 813 20 11.5 514 11.5 20 5

2

tdptuctlfcTa(fwpoittti

2

toStcfiswosuU

acr

shown in Fig. 2. However, when calculating the ratio of the HB

15 11.5 11.5 5

.3.1. Factors impacting extractionThe 3 main impacting factors chosen for study were the concen-

ration of sulphuric acid in the methanol solution, the length of theigestion time and the concentration of biomass in the chloroformhase used in the analysis. According to the literature [10,13,20,21],he sulphuric acid concentrations and digestion times most oftensed ranged from 3% to 20% acid, which were the two extremeshosen for testing. A range of digestion times (1–20 h) were choseno evaluate the extraction kinetics for initial tests at these two acidevels. Since culture MC4 contained the highest H2MB and H2MVractions, it was chosen for the evaluation of the effect of acid con-entration on the extraction kinetics of the different HA fractions.he impact of granulation on the extraction kinetics was also evalu-ted, with a comparison performed between two floccular culturesa lab-scale (MC2) and a full-scale (MC6a) culture), and two samplesrom a culture containing granules and flocs (MC4 1 and MC4 2),here the second sample was homogenised prior to the extractionrocedure in order to disrupt the granular biomass. Finally, a rangef biomass concentrations was also tested (1–50 mg/mL CHCl3),n order to define an appropriate range for the DOE study, sincehe effect of this parameter had not previously been evaluated. Inhis case, a comparison was performed between the floccular cul-ures (MC1, MC2, MC3 and MC6a), in order to assess the resultsndependently from granulation effects.

.3.2. Design of experiments (DOE)This study used the DOE tool to optimise the methanolysis reac-

ion of different HA monomers for 6 different microbial cultures inrder to assess the conditions where full digestion was achieved.ix DOE were conducted, one for each of the microbial culturesested. Each DOE consisted of 15 experiments performed in dupli-ate, whose conditions were determined by varying the 3 inputactors (acid, digestion time and biomass quantity) to assess theirmpact on 2–4 output factors depending on the polymer compo-ition in each culture (HB, H2MB, HV, H2MV). The output factorsere introduced into the DOE as the area produced by each peak

btained in the chromatogram, divided by the area of the internaltandard – A/Ais and normalised by the weight of the biomass pelletsed. The DOE was carried out in Matlab 2006b (Mathworks Inc.,SA). A full list of the 15 experiments is shown in Table 1.

For each microbial culture tested, the combination of factorsnd the number of experiments were determined using a central

omposite face-centred design with 3 levels, 1 central point and 1eplicate for each experiment [22].

istry 48 (2013) 1626–1634

2.3.3. Model buildingUsing the experimental results obtained through the DOE, the

effect of each variable on the area of HB, HV, H2MB and H2MV wasdescribed by determining the model with best fit. A response sur-face modelling approach (RSM) was used to fit a quadratic equation(q) to the experimental values of HA (Eq. (1)), considering not onlylinear and quadratic coefficients (a) of the variables (x) but alsoall possible interactions between them [23], with i, j and k vary-ing between 1 and 3 according to the three variables considered.Coefficients were determined by multiple linear regression (MLR).

q(x1, x2, x3) = a0 +3∑

k=1

akxk +3∑

i,j=1

aijxixj + a123x1x2x3 (1)

The regstats function in Matlab 2006b (Mathworks Inc., USA)was used to calculate the model’s coefficients. The validation of themodels was based on three statistical tests: the coefficient of deter-mination (R2) and two other tests based on the analysis of variance(ANOVA): the regression’s goodness of fit and the lack of fit (LOF).The values of all the variables and outputs were normalised usingthe standard score method, i.e., the overall mean was subtractedfrom each value and then the result was divided by the overallstandard deviation, so that the weight of each variable could becompared. The average of the replicate error obtained for all sam-ples was used to calculate the standard error of the method. A moredetailed description of the statistical methods used can be found in[14].

3. Results

3.1. The effect of monomeric composition and sludge structure onthe HA extraction rate

The impact of the acid concentration on extraction kinetics wasassessed by quantifying the different HA monomers at increas-ing digestion times. Two parallel hydrolysis curves were obtainedusing acidic methanol with 3 and 20% sulphuric acid (Fig. 1) andapproximately 3 mg/mL of biomass. The GC peak areas were nor-malised with the internal standard and the biomass weighed foreach sample.

The extraction rates increased from 2 to 4 fold when using20% acidic methanol instead of 3%. For example, the HB extractionrate with 3% (43.2 A/(AIS.gTS.s)) doubled to 90 A/(AIS.gTS.s) at 20%acid and the rate of H2MV increased 4-fold when using 20% (14.4A/(AIS.gTS.s)) instead of 3% (3.6 A/(AIS.gTS.s)).

Also, for the same acid concentration, the extraction rate var-ied depending on the type of monomer. Hence, at 3% acid, HB andHV had the lowest digestion times, of approximately 3–4 h, whileH2MB and H2MV required a much longer digestion time of approx-imately 8–15 h.

For HB analysis, the peak areas obtained with 20% acid wereapproximately 20% lower than the ones obtained with 3% acid. Adecrease in HB peak areas, when increasing the acid concentration,was also observed for culture MC2, only containing HB and HV, in asimilar set of experiments (results not shown). This effect was notobserved for any of the other monomers, suggesting it was specificor more enhanced for HB in comparison with other monomers.

When hydrolysing the co-polymer standards with acid concen-trations of 3 and 20%, HB, but not HV, displayed the same effectsobserved in the sludge samples: a decrease in peak areas and anincrease in the extraction rate for higher acid concentrations, as

area in the sample vs. the standard (Asample/Astandard) throughoutthe different digestion times, the difference between the aver-age ratios obtained at 3 and 20% acid were not statistically

Page 4

A.B. Lanham et al. / Process Biochemistry 48 (2013) 1626–1634 1629

F ng 3%w ). Erro

s0

s3p

d(dftP2o

ufi

Fct

ig. 1. Recovery of each HA monomer throughout the digestion of MC4 samples usiith the internal standard and the biomass weighed for each sample (approx. 3 mg

ignificant (0.113 ± 0.008 and 0.102 ± 0.007 for MC2 and.25 ± 0.04 and 0.24 ± 0.03 for MC4 at 3 and 20% acid respectively).

Concerning the extraction rates, the difference observed for HBtandards hydrolysed using 3 or 20% acid was only approximately0%, which was lower than what was observed for the sludge sam-les.

The profiles of HB, HV, H2MB and H2MV extraction rates wereetermined using 3% acid concentration in a PHA producing cultureMC2) and full-scale activated sludge (MC6a), as well as in stan-ards. These results were compared with the ones already obtainedor MC4 (Fig. 3a and b). However, while MC2 and MC6a only con-ained a co-polymer of PHB:PHV, MC4 contained a tetra-polymer ofHB:PHV:PH2MB:PH2MV and therefore results are also shown at0% acid concentration in comparison with the methanolysis ratef the standard for H2MV, caproic acid (Fig. 3c).

The monomer concentration increased during the digestionntil reaching a maximum value. The profile of the HA recoveredraction for each microbial culture and standards tend to agree,ndicating a common optimal digestion time, except in the case of

ig. 2. Recovery of HB monomers in a PHB–PHV copolymer standard using 3% (cir-les) and 20% (squares) acidic methanol. Error bars represent the standard error ofhe method.

(circles) and 20% (squares) acidic methanol. Peak areas presented were normalisedr bars represent the standard error of the method.

caproic acid, which hydrolyses much faster at 20% acid concentra-tion than H2MB or H2MV.

In order to further clarify the mechanism of the reaction, thekinetic constant, k, and the order of the reaction, n, were deter-mined using the iterative Euler method programmed into Excel(Office 2007, Microsoft) and using the solver function. This methoddetermined that the reactions were of first order (n = 1). The kineticconstants for each culture, indicated in Table 2, decreased fromHB to H2MV, i.e., kHB > kHV > kH2MB > kH2MV. This was verifiedfor hydrolyses done in both acidic conditions, 3 and 20%. Further-more, the effect of acid on the kinetics of the extraction, alreadyobserved in the previous section, was confirmed since k3%

HA < k20%HA

for all monomers in culture MC4.The kinetic constant for hydrolysis of PHB and PHV standards

was similar to the kinetic constant of the polymers accumulated bycultures MC2 and MC6a, but higher than the one of culture MC4(Table 2). The lower kinetic constant for polymers from cultureMC4, composed of granules and flocs, was confirmed in anotherexperiment undertaken with a different MC4 sample (MC4 2). Infact, the MC4 2 sample was homogenised by mechanical shear todisrupt the granules, and the kinetic constant was even lower.

Using the kinetic constants, it was possible to predict the opti-mum time needed, at these conditions, to achieve at least 90% HAextraction. The extraction of HB and HV in standards and culturesMC2 and MC6a should be completed in between 2–3 h, when usinga 3% acidic methanol solution. For MC4 a higher digestion time,between 3–6 h, was needed. This digestion time could be decreasedto approximately 2 h for both MC4 samples if the acid concentra-tion was increased to 20%. Results for H2MB and H2MV were onlyavailable for culture MC4, which determined that at 20% acid con-centration a digestion time between 3 and 6 h could be employedfor enhanced recovery.

3.2. The effect of biomass concentration on HA extraction

The effect of the biomass concentration on the digestion wasassessed using the optimum conditions for HB extraction in

Page 5

1630 A.B. Lanham et al. / Process Biochemistry 48 (2013) 1626–1634

Table 2Kinetic constants (k) and optimum digestion times for HA monomers in different microbial cultures and for standards using 3% acid and 20% acid methanolysis.

3% acid

HB HV H2MB H2MV

k R2 t (h) k R2 t (h) k R2 t (h) k R2 t (h)

Standard 0.88 0.89 2.6 0.71 0.88 3.2 – – – – – –MC2 0.88 0.98 2.6 0.77 0.95 3.0 – – – – – –MC4 1 0.64 0.93 3.6 0.41 0.95 5.7 0.22 0.81 10.3 0.15 0.83 15.7MC6a 0.92 0.98 2.5 0.77 0.94 3.0 – – – – – –

20% acid

HB HV H2MB H2MV

k R2 t (h) k R2 t (h) k R2 t (h) k R2 t (h)

MC4 1a 2.0 0.95 1.1 1.5 0.96 1.6 0.95 0.95 2.4 0.52 0.99 4.42.3

cdiaRcu

fm(tatts

3

mipMo

fdtT

Njtottempsv

ecte

MC4 2a 1.2 0.92 1.9 1 0.98

a MC4 1 and MC4 2 refer to different samples of microbial culture MC4.

ultures MC2 and MC6a (3 h digestion and 3% acidic methanol). Fourifferent floccular microbial cultures were tested: PHA accumulat-

ng organisms with higher and lower PHA content (MC1 and MC2), floccular lab-scale culture (MC3), and WWTP sludge (MC6a).esults, shown in Fig. 4, reveal a similar trend for all the microbialultures, where a 95% recovery of HB was obtained for analysessing between approximately 3–8 mg biomass/mL.

For biomass concentrations equal or lower than 2 mg/mL, theraction of HB recovered was only in the range of 60–80% of the

aximum value. This was verified both for higher (MC1) and lowerMC2 and MC3) PHB contents. For biomass concentrations higherhan 8 mg/mL, the fraction of HA recovered slowly decreased untilpproximately 80% at 50 mg/mL. The same effect was observed onhe quantification of HV in cultures MC1, MC2 and MC6a, suggestinghat this effect was independent of the HA monomer (results nothown).

.3. Design of experiments (DOE)

In addition to the experimental results carried out for someicrobial cultures, an experimental design strategy was conducted

n six different microbial cultures, some of them already testedreviously but using different sample batches: MC1–MC5 andC6b. This allowed for a comprehensive confirmation of the trends

bserved previously.A quadratic model was built for each microbial culture and

or each individual monomer (HB, HV, H2MB and H2MV) whichescribed well the results as indicated by the R2 higher than 0.8,he p-value lower than 0.05 and the LOF value higher than 0.05 (cf.able A1 of the Supplementary material).

Two DOE conditions were not fulfilled: condition No. 10 ando. 12 (cf. Table 1), due to extraction difficulties for samples sub-

ected to high digestion periods and high acid concentration. Underhese conditions, a greater instability between the two phases wasbserved, which reduced significantly the feasibility of the extrac-ion step and the concentration of monomeric ester quantified. Forhis reason, although the models satisfied the statistical criteriastablished, they lacked sensitivity to accurately predict the opti-um digestion time for each microbial culture. The accuracy of the

rediction for the optimum digestion time decreased even furtherince a wide time interval was chosen from 3 to 20 h, in order toerify the optimum conditions for all HA monomers.

The model’s coefficients indicated the effect and the weight of

ach parameter on the output result, in this case, the HA monomeroncentration. An average of the model coefficients for all 6 cul-ures is shown in Fig. 5 for each monomer, with the correspondingrror bars indicating their significance. Coefficients were only

0.58 0.93 4.0 0.41 0.97 5.6

considered significant if the error bar was lower than the coeffi-cient’s value. The coefficients were normalised between −1 and1 and therefore their values may be directly compared to eachother. Positive coefficients imply a positive effect and negativecoefficients, a negative contribution to the HA monomer concen-tration.

The most obvious aspects can be derived from the linearcoefficients associated to acid concentration, digestion time andbiomass concentration, as separate parameters. Acid has a strongnegative effect for HB whereas it brings a moderate positive effectto H2MB and H2MV. The digestion time brings a positive effect toall HA monomers, but with higher values for H2MB and H2MV, asexpected. The biomass coefficient was also positive and significantfor all monomers except for H2MV, meaning that higher biomassconcentrations, in the tested range of 2–8 mg/mL, improved therecovery of HA monomers.

The quadratic coefficients dictate the form of the quadraticcurve obtained from the model. When considering normalisedcoefficients between −1 and 1, and considering a positive linearcoefficient, a positive quadratic coefficient implies a curve with apositive growth, whereas a negative quadratic coefficient impliesa curve with a negative growth. When the linear and the quadraticcoefficients have the same sign, they reinforce either a positive ora negative effect. When they have opposite signs, the effect of thelinear coefficient is cancelled and inversed for a certain value of theparameter equal to −ai/2aii, where ai and aii are the values of thelinear and the quadratic coefficients for each parameter (cf. Eq. (1)).

The quadratic coefficient for acid was not significant formost monomers, except in the case of H2MB, where it wasslightly negative. Since the linear coefficient for acid was mod-erately positive, the result was that the positive effect ofacid was progressively diminished for higher concentrations ofacid.

The quadratic coefficient for time was moderately negative forall monomers except in the case of H2MV. For HB and HV, sincethe linear coefficient was positive and in the same range as thequadratic one, this means that the overall positive effect of timewas very slight and that a change in trend can be anticipated forlong digestion periods, as expected. For H2MB, the linear coeffi-cient was approximately 3 times higher than the quadratic one,therefore stressing the positive effect of this parameter and indi-cating that a turning point was not observed in the time intervalchosen. The fact that H2MV does not have a significant negative

quadratic coefficient indicated that the positive linear effect wasnot reduced for the time interval chosen.

The quadratic coefficient for biomass concentration was mod-erately negative for all monomers and in the same range as the

Page 6

A.B. Lanham et al. / Process Biochemistry 48 (2013) 1626–1634 1631

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 2 4 6 8 10 12 14 16 18 20

Reco

vere

d HA

frac

�on

HV_std MC2 MC4 MC6a

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 2 4 6 8 10 12 14 16 18 20

Reco

vere

d HA

frac

�on

HB_std MC4 MC2 MC6a

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 2 4 6 8 10 12 14 16 18 20

Reco

vere

d HA

frac

�on

H2MB H2MV Caproic acid

Digestion time (h)

a

b

c

Fig. 3. Extraction kinetic profile for HB, HV, H2MB and H2MV monomers; (a) HBextraction profile with 3% acidic methanol; (b) HV extraction profile with 3% acidicmethanol; (c) H2MB and H2MV extraction profile in MC4 with 20% acidic methanol;sgo

lsw

ooacCaam

Hai

0

0.2

0.4

0.6

0.8

1

1.2

5550454035302520151050Biomass concentra�on (mg/mL)

Reco

vere

d HB

frac

�on

MC1 MC2 MC3 MC6a

Fig. 4. Effect of the biomass concentration on HB extraction for 4 different cultures(3%, 3 h): MC1 (diamonds); MC2 (squares); MC3 (circles); MC6a (triangles). Error

sample used being not representative of the entire sample.

Fig. 5. Averaged model coefficients for all the microbial cultures and for each

tandards (std), squares; for (a) and (b) MC2, circles, MC4, diamonds, MC6a, trian-les; for (c): H2MB, circles; H2MV, squares. Error bars represent the standard errorf the method.

inear coefficient. Hence, the biomass concentration had an overalllight positive effect, in the interval of 2–8 mg/mL, and this effectas diminished for higher biomass values.

The interaction coefficients, implicating 2 different variables atnce, describe the synergistic effect that each variable has on thether. A positive interaction coefficient will further enhance anlready positive variable. On the other hand, a negative interactionoefficient will moderate and eventually reduce the positive trend.onsidering the acid × time coefficient, it was slightly negative forll monomers. Therefore, considering that the digestion time hasn overall positive effect for all monomers, this tendency will beoderated with the increase of the acid concentration.

The other significant interaction coefficients only implicate

2MB and apply to the time × biomass coefficient and to thecid × biomass coefficient, which are both moderately positiventeractions. Considering the overall positive effect of time and

bars represent the standard error of the method.

acid on the hydrolysis of H2MB, this positive effect was evenmore noticeable for higher biomass concentrations. This maybe explained by the fact that the low concentrations of H2MBobserved (3 wt%, in cultures MC4, MC5 and MC6) will be more eas-ily extracted not only at conditions using higher acid and time butalso at conditions where the biomass content was higher in orderto yield a detectable GC peak. However, the same effect was notnoticeable for H2MV.

During the design of experiments procedure, a total of approx-imately 160 samples were processed. All of these samples wereperformed in duplicate by a total of 7 different operators overa 2-week period. The average of the replicate error obtained forall samples was 5%. When comparing the replicate error for eachbiomass, there were no significant differences i.e., all average errorsfor each culture were the same within the standard deviation ofthe biomass from the other cultures (results not shown). Whencomparing the error from different operators, also no significantdifferences were noted. However, when comparing the conditionsof the samples, the ones where the biomass amount was 2 mg/mLhad a higher error (7 ± 2%) than samples with 5 or 8 mg/mL (4 ± 2%),likely due to a higher weighing error, or due to the small portion of

HA monomer. Error bars represent the significance of each coefficient therefore,coefficients with an error bar higher than the value itself were not considered sig-nificant to the model. HB, HV, H2MB and H2MV model coefficients are presented inincreasingly darker shades of blue. (For interpretation of the references to colour inthis figure legend, the reader is referred to the web version of this article.)

Page 7

1 iochemistry 48 (2013) 1626–1634

4

4q

hfc

atrmtg

owaHasfcn

iewps

4

feMspst

ctscfgsastiioFtMoadw[

0.0

0.2

0.4

0.6

0.8

1.0

2520151050Time (h)

HB

rec

over

ed fr

actio

n

experimental results (this study)Granular MC4 and MC5 (this study)biomass conc = 2mg/mL (this study)Oehmen et al. (2005)Braunegg et al. (1978)Jan et al. (1995)

Fig. 6. Extraction kinetic profile for HB using all the results collected during thisstudy (empty circles crosses and plus signs), complemented with results from 8,pure culture (squares); 9, pure culture (diamonds) and 17, EBPR floccular culture(circles); during a hydrolysis with 3% acid concentration. Experimental results from

632 A.B. Lanham et al. / Process B

. Discussion

.1. The overall effect of each parameter on the extraction anduantification of PHA monomers

Acid concentration, time of digestion and biomass concentrationad an effect on the HA extraction rate and recovery. The results

rom the hydrolysis profiles, performed on only some cultures, wereonsistent with the results from the DOE.

The biomass concentration limited the recovered fraction of HBnd HV in values lower than 3 mg/mL, which was supported byhe DOE results since they indicated that in the 2–8 mg/mL testedange, biomass concentration had a positive effect, although only aoderate effect at higher values. For biomass concentrations higher

han 10 mg/mL, the recovered HA fraction started to decrease, sug-esting that there was incomplete digestion.

Although an increase in acid concentration had a negative effectn HB recovery, it increased the extraction rates of all monomers,hich could be particularly useful in the determination of H2MB

nd H2MV, since they possess considerably slower kinetics thanB and HV. The DOE results showed a positive impact of time onll monomers, however, the combined effect of time and acid waslightly negative, suggesting that finding the optimum conditionsor each monomer implies finding a compromise between the acidoncentration and the digestion time: the more acid, the less timeeeded or the reverse.

Apart from the acid concentration and the digestion timenteraction, the different parameters, in general, do not have a syn-rgistic effect, i.e., the interaction coefficients were not significantithin a 95% confidence interval. This indicated that, if needed, eacharameter can be further optimised independently for a particularystem.

.2. Different cultures, different extraction methods?

In the previous experimental sections, slower kinetics wereound for MC4, in comparison with the kinetics of standards andven of other microbial cultures such as MC2 and MC6a. CultureC4 has shown a dynamic behaviour in terms of the aggregation

tate of the biomass, alternating between floccular and granularhases. For instance, the sample used in the DOE (MC4 1) corre-ponded to a semi-granulation state, whereas MC4 2 correspondedo a complete granulated state.

While floccular cultures MC2 and MC6a showed a similar kineticonstant as the standards, suggesting that the limiting reaction washe hydrolysis of the polymer, the granular culture MC4 displayedlower kinetics, indicating that the limiting mechanism could beell disaggregation and lysis, likely indicating internal mass trans-er limitations of acid within the granules. Additionally, the moreranulated the sludge, the slower the kinetics, as shown by thelower kinetics of MC4 2 compared with MC4 1. This effect waslso noticeable in the DOE results for cultures MC4 and MC5 (alsoemi-granular), which revealed the highest value for the digestionime coefficients in the models of each monomer (results shownn Supplementary material in Table A2). The effect of slower kinet-cs could be compensated by either increasing the digestion timer by increasing the acid concentration. In fact, results shown inigs. 1 and 2 indicated that, when increasing the acid concentra-ion from 3 to 20%, the extraction rate increased more in culture

C4 than in the standards, confirming that the acid impacts notnly the PHA hydrolysis kinetic constant, as shown in Table 2, but

lso the cell lysis/disaggregation mechanism. A difference in con-itions for granular and floccular biomasses for glycogen recoveryas also found when hydrolysing these two types of aggregates

22].

granular cultures (MC4 and MC5) (crosses) and using low biomass concentrations(approx. 2 mg/mL) (plus signs) were singled out with different markers. Error barsrepresent the standard error of the method.

4.3. The combined extraction of different monomers

Results from the kinetic profiles indicated that the longer andthe more substituents contained in the chain, the slower the poly-mer digestion (Table 2). Also, the DOE experiments revealed thatacid concentration and digestion time have a greater positive effecton H2MB and H2MV than on HB and HV, which suggested thatthese monomers require longer digestion at higher acid concen-trations in order to achieve a complete polymer hydrolysis. Theacidic methanolysis reaction implies that the acid attacks the mosthydrophilic part of the poly-ester molecule, which lies in the car-boxylic group – COOH. With the increase in size and number ofthe substituents, e.g. from a methyl group in PHB (carbon 3), toan ethyl group in PHV (carbon 3) and to a methyl and an ethylgroup in PH2MV (carbons 2 and 3, respectively), the hydrophobic-ity around the main carbon chain increases, which leads to a highersteric hindrance to the acid attack. The steric hindrance was evenmore accentuated in the case of PH2MB and PH2MV because ofthe added substituents in carbon 2, which are closer to the car-boxyl group. This effect on HA recovery had not been previouslyidentified.

In terms of PHB, the negative impact of an increased acid concen-tration, observed in the hydrolysis profiles and in the DOE results,has been often discussed in the literature [10,13,24]. The acid effecton PHB has been suggested to derive from its degradation into cro-tonic acid or other degradation products [10,24,25]. However, bothJan et al. [24] and Huijberts et al. [25] have shown that no secondaryhydrolysis products were formed in the reaction and Jan et al. [24]suggested that this decrease in HB recovery was due to a shift inthe partition coefficient of HB between the chloroform and themethanol phase during the extraction procedure, most likely dueto pH. Therefore, the recovery of this monomer should preferablyuse lower acid concentrations. However, this effect was equivalentin samples and in standards, as shown by the same values of theAsample/Astandard ratios at 3 and 20% acid concentrations. Hence, itcan be corrected at higher acid concentrations, as long as the stan-dards for the calibration curves have been subjected to the sameconditions as the samples.

The time needed for achieving maximal HB recovery was inagreement with conditions described in literature. For HB and HV,most authors used a common procedure [11,13,16,24] of 2–4 h and3% acid concentration. According to Table 2, these conditions would

also be sufficient for the recovery of HB and HV in all standardsand floccular cultures. Granular cultures might need a longer diges-tion period, up to 6 h. Most of the results in the literature availablefor HB analysis using 3% acid concentration are displayed in Fig. 6,

Page 8

A.B. Lanham et al. / Process Biochem

Ff

tDtuc2

4

aedt

ig. 7. Ideal HA extraction conditions (acid concentration and digestion time) as aunction of HA fraction for (a) floccular biomass and (b) granular biomass.

ogether with all the results obtained in this study, including theOE results. It is interesting to note that the lower recovered frac-

ions (<0.8), obtained in this study, for digestion times of 3 h andsing an acid concentration of 3%, were all either from granularultures, from samples with low biomass concentrations (approx.

mg/mL) or from both, as pointed out in Fig. 6.

.4. Choosing an optimised HA extraction method

An optimised method should use the least resources (lowest

cid concentrations) and be the least time consuming (short-st digestion times), while achieving the highest recovery. Wheneveloping a HA extraction method, the type of biomass and theype of monomers analysed should be considered: the higher the

istry 48 (2013) 1626–1634 1633

complexity of the monomers and the number and size of granules,the longer the digestion time and the higher the acid concentration(see Fig. 7). The reaction kinetics, combined with a statistical con-firmation of the trends observed in 6 different microbial cultures,pointed out the importance of choosing the appropriate conditionsfor HA quantification in different systems. Floccular cultures con-taining PHB and PHV, which constitute the majority of the casesin mixed microbial cultures, can be analysed using a 3-h diges-tion period with 3% acid concentration. If the culture also containsPH2MB and PH2MV, the acid concentration should be increased to20% and the samples should be digested at least 4 h. When dealingwith granular biomass, the best option could be either to still use3% acid and longer digestion periods, e.g., 6 h, or to increase theacid concentration. Using an acid concentration of 20% and in thesituation of higher chain monomers such as H2MB and H2MV, thedigestion time should be increased to 4–6 h. The major mechanismsuncovered in this work related to the impact of the type of biomassand the HA composition, should also be applicable for other acidicalcoholysis methods (e.g. [26]), however further work is needed toconfirm this hypothesis.

5. Conclusions

This study systematises, for the first time, the factors deter-mining the efficiency of HA extraction using acidic methanolysis.Floccular or granular cultures should be treated differently, sincethe HA recovery is limited by the hydrolysis kinetics in the formerand the cell lysis step in the latter. Hydrolysis kinetics decreasewith an increase in the size and number of substituent groups ofthe co-polymer, due to steric hindrance mechanisms, an effect thatcan be counterbalanced with an increase in the acid concentrationand digestion time. Considering the high diversity of PHA producingsystems and the diversity in their monomers, special care should betaken to tailor the best extraction method to the system under anal-ysis, since this procedure affects the quantification and compositionof PHA. This, in turn, affects the determination of kinetic and stoi-chiometric parameters and the correlation of polymer compositionwith physical/chemical properties.

Acknowledgments

The authors wish to thankfully acknowledge funding fromthe Fundac ão para a Ciência e Tecnologia through the PhD grantsSFRH/BD/25275/2005, SFRH/BD29477/2006, SFRH/BD38763/2007,SFRH/BD42085/2007, SFRH/BD65113/2009, the Post-Docgrants SFRH/BPD/88382/2012, SFRH/BPD/70185/2010 andthe research projects PTDC/EBB-EBI/103147/2008 and PEst-C/EQB/LA0006/2011. Marta Coma also wishes to thank thefinancial support of the Spanish Government through the researchproject MICINN-PT2009-0047 and the Luso-Espanhola integratedaction E-94/10.

Appendix A. Supplementary data

Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.procbio.2013.07.023.

References

[1] Rijk T, Meer P, Eggink G. Methods for analysis of poly (3-hydroxyalkanoate)composition. In: Doi Y, Steinbüchel A, editors. Biopolymers online. Wiley Online

Library; 2005. p. 1–12.

[2] Lee SY. Bacterial polyhydroxyalkanoates. Biotechnology and Bioengineering1996;49:1–14.

[3] van Loosdrecht MCM, Pot MA, Heijnen JJ. Importance of bacterial storage poly-mers in bioprocesses. Water Science and Technology 1997;35:41–7.

Page 9

1 iochem

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

634 A.B. Lanham et al. / Process B

[4] Reis MAM, Albuquerque MGE, Villano M, Majone M. Mixed culture processesfor polyhydroxyalkanoate production from agro-industrial surplus/wastes asfeedstocks. In: Fava F, Agathos S, Young M, editors. Comprehensive Biotechnol-ogy, vol. 6. Elsevier; 2011. p. 669–83.

[5] Jiang Y, Marang L, Tamis J, van Loosdrecht MCM, Dijkman H, KleerebezemR. Waste to resource: converting paper mill wastewater to bioplastic. WaterResearch 2012;46:5517–30.

[6] Braunegg G, Lefebvre G, Genser KF. Polyhydroxyalkanoates, biopolyestersfrom renewable resources: physiological and engineering aspects. Journal ofBiotechnology 1998;65:127–61.

[7] Henze M, Gujer W, Mino T, van Loosdrecht MCM. Activated Sludge ModelsASM1, ASM2, ASM2d and ASM3. London, UK: IWA Publishing; 2000.

[8] Mino T, Van Loosdrecht MCM, Heijnen JJ. Microbiology and biochemistryof the enhanced biological phosphate removal process. Water Research1998;32:3193–207.

[9] Oehmen A, Lemos PC, Carvalho G, Yuan ZG, Keller J, Blackall LL, Reis MAM.Advances in enhanced biological phosphorus removal: from micro to macroscale. Water Research 2007;41:2271–300.

10] Braunegg G, Sonnleitner B, Lafferty RM. Rapid gas-chromatographic method fordetermination of poly-beta-hydroxybutyric acid in microbial biomass. Euro-pean Journal of Applied Microbiology and Biotechnology 1978;6:29–37.

11] Comeau Y, Hall KJ, Oldham WK. Determination of poly-beta-hydroxybutyrateand poly-beta-hydroxyvalerate in activated-sludge by gas-liquid-chromatography. Applied and Environment Microbiology 1988;54:2325–7.

12] Gross RA, Demello C, Lenz RW, Brandl H, Fuller RC. Biosynthesis and character-ization of poly(beta-hydroxyalkanoates) produced by Pseudomonas oleovorans.Macromolecules 1989;22:1106–15.

13] Oehmen A, Keller-Lehmann B, Zeng RJ, Yuan ZG, Keller J. Optimisa-tion of poly-�-hydroxyalkanoate analysis using gas chromatography forenhanced biological phosphorus removal systems. Journal of ChromatographyA 2005;1070:131–6.

14] Lanham AB, Ricardo AR, Coma M, Fradinho J, Carvalheira M, Oehmen A, Car-valho G, Reis MAM. Optimisation of glycogen quantification in mixed microbialcultures. Bioresource Technology 2012;118:518–25.

15] Albuquerque MGE, Carvalho G, Kragelund C, Silva AF, Barreto-CrespoMT, Reis MAM, Nielsen PH. Link between microbial composition and

[

istry 48 (2013) 1626–1634

carbon substrate-uptake preferences in a PHA-storing community. ISME Jour-nal 2013;7:1–12.

16] Fradinho JC, Domingos JMB, Carvalho G, Oehmen A, Reis MAM. Polyhydrox-yalkanoates production by a mixed photosynthetic consortium of bacteria andalgae. Bioresource Technology 2013;132:146–53.

17] Carvalheira M, Oehmen A, Carvalho G, Reis MAM. The effect of dissolved oxy-gen concentration on the competition between polyphosphate accumulatingorganisms and glycogen accumulating organisms. In: Nutrient Removal andRecovery 2013, 28–31July. WEF/IWA; 2013.

18] Coma M. Biological Nutrient Removal in SBR Technology: from Floccular toGranular Sludge. Girona, Spain: Universitat de Girona; 2011. PhD thesis.

19] Lanham AB. Full-Scale Biological Phosphorus Removal: Quantification ofStorage Polymers, Microbial Performance and Metabolic Modelling. Lisboa,Portugal: Universidade Nova de Lisboa; 2012. PhD thesis.

20] Ng K-S, Wong Y-M, Tsuge T, Sudesh K. Biosynthesis and characterization ofpoly(3-hydroxybutyrate-co-3-hydroxyvalerate) and poly(3-hydroxybutyrate-co-3-hydroxyhexanoate) copolymers using jatropha oil as the main carbonsource. Process Biochemistry 2011;46:1572–8.

21] Serafim LS, Lemos PC, Oliveira R, Reis MAM. Optimization of polyhydroxybu-tyrate production by mixed cultures submitted to aerobic dynamic feedingconditions. Biotechnology and Bioengineering 2004;87:145–60.

22] Eriksson L, Johansson E, Kettaneh-Wold N, Wikström C, Wold S. Design ofExperiments: Principles and Applications. 3rd ed. Sweden: Umetrics AB; 2008.p. 425.

23] Montgomery D. Design and Analysis of Experiments. 5th ed. New York: Wiley;2000. p. 680.

24] Jan S, Roblot C, Goethals G, Courtois J, Courtois B, Saucedo JEN, SeguinJP, Barbotin JN. Study of parameters affecting poly(3-hydroxybutyrate)quantification by gas-chromatography. Analytical Biochemistry 1995;225:258–63.

25] Huijberts GNM, Vanderwal H, Wilkinson C, Eggink G. Gas-chromatographic

analysis of poly(3-hydroxyalkanoates) in bacteria. Biotechnology Techniques1994;8:187–92.

26] Werker A, Lind P, Bengtsson S, Nordstrom F. Chlorinated-solvent-free gas chro-matographic analysis of biomass containing polyhydroxyalkanoates. WaterResearch 2008;42:2517–26.