Characterization of enzymatically synthesized structured lipids containing eicosapentaenoic, docosahexaenoic, and caprylic acids

lable at ScienceDirect

Polymer 55 (2014) 6271e6277

Contents lists avai

Polymer

journal homepage: www.elsevier .com/locate/polymer

Characterization of enzymatically synthesized amylopectin analogsvia asymmetrical flow field flow fractionation

Jelena Ciric a, b, Agn�es Rolland-Sabat�e c, Sophie Guilois c, Katja Loos a, b, *

a Department of Polymer Chemistry & Zernike Institute for Advanced Materials, University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlandsb Dutch Polymer Institute (DPI), P.O. Box 902, 5600 AX Eindhoven, The Netherlandsc UR1268 Biopolym�eres Interactions Assemblages, INRA, F-44300 Nantes, France

a r t i c l e i n f o

Article history:Received 7 August 2014Received in revised form27 September 2014Accepted 29 September 2014Available online 5 October 2014

Keywords:Asymmetrical flow field flow fractionationAmylopectinEnzymatic polymerization

* Corresponding author. University of Groningen,ningen, The Netherlands. Tel.: þ31 50 3636867.

E-mail addresses: [email protected] (J. Ciric)(A. Rolland-Sabat�e), [email protected] ([email protected] (K. Loos).

http://dx.doi.org/10.1016/j.polymer.2014.09.0590032-3861/© 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

Asymmetrical flow field flow fractionation (AF4), when coupled with multi-angle laser light scattering(MALLS), is a very powerful technique for determination of the macromolecular structure of high molarmass (branched) polysaccharides. AF4 is a size fractionation technique just as size exclusion chroma-tography (SEC), nevertheless can overcome some crucial problems found in SEC analysis especially instarch like structures. This paper describes a detailed investigation of the macromolecular structure oftwo groups of well-defined synthetic amylopectin analogs e synthesized via an in vitro enzyme-catalyzed reaction using the enzymes phosphorylase b from rabbit muscle and Deinococcus geo-thermalis glycogen branching enzyme (Dg GBE). Size, molar mass distributions and structural data werestudied by AF4 coupled with online quasi-elastic light scattering (QELS) and multi-angle light scattering(MALLS).

© 2014 Elsevier Ltd. All rights reserved.

1. Introduction

The most popular and the most developed method for deter-mination of the size distributions of starch and starch-like poly-mers is size-exclusion chromatography (SEC) [1]. However, for goodseparation of branched and large polysaccharides in SEC the usedcolumns and their limitations can be an obstacle. Currently usedcolumns have low exclusion limits for extremely large poly-saccharides, and lead to interactions of polysaccharides with thecolumnmaterial - especially inwater; shear scission, and inevitableband broadening [1,2]. A technique that can overcome theseproblems is asymmetrical flow field flow fractionation (AF4) thathas already been used for the analysis of natural and syntheticbranched polysaccharides [3e5]. The separation process is based ondifferences in the diffusion coefficients of the analyzed components(separation depends on the size and shape of particles and mole-cules) [7e9]. AF4 selectivity is based on the diffusion coefficient ofthe sample; hence the hydrodynamic radius (Rh) can be determinedvia the diffusion coefficient from the StokeseEinstein equation.

Nijenborgh 4, 9747 AG Gro-

, [email protected]. Guilois), [email protected],

Rh≡kBT

6phDT(1)

where DT represents the translational diffusion coefficient, kBBoltzman's constant, T the temperature, and h the viscosity of thesolvent.

The retention ratio in the normal mode (small particles areeluted first) with constant crossflow according to the AF4 theory is[10]:

R ¼ t0triz

6DiV0

Fcw2 (2)

where t0 represents the void time (the time needed for the carriersolvent to pass from inlet to outlet in the channel), tri elution time,Di diffusion coefficient of the ist exponent slice, Fc is the crossflowrate, w the channel thickness, and V0 the void volume (the geo-metric volume of the channel).

For samples with high size dispersities, like the enzymaticallysynthesized amylopectin analogs studied here, a good fractionationof the whole sample can be achieved using a crossflow gradientinstead of a constant crossflow. For an exponential regression of acrossflow tri can be expressed as [4]:

J. Ciric et al. / Polymer 55 (2014) 6271e62776272

8>>< 266 t0w2Fcstart

6DV0� tstart

3779>>=

tri ¼ tstart þ ðtend � tstartÞ>>:1� exp4�tend � tstart 5>>; (3)

where tstart is the time corresponding to the start and tend the timecorresponding to the end of the gradient.

Size distributions can be obtained by determining the rela-tionship between DT and tri using established equations formethods with constant crossflow [11], and their modifications forgradient crossflow [12,13]. If possible, Rh can be determined onlineby means of quasi-elastic light scattering (QELS), providing exper-imental size distributions.

In our previous paper we have solved parts of the molecularmechanismof the enzymatic polymerization of amylopectin analogs,see Fig. 1. These analogs were synthesized in vitro using the enzymesphosphorylase b and Deinococcus geothermalis glycogen branchingenzyme (Dg GBE). The synthesized analogs had tunable degree ofbranching, obtained by regulation of the reaction time, and tunabledegree of polymerizationobtainedby regulationof the ratio betweenthe monomer glucose-1-phosphate (G1P) and the primer malto-heptaose (G7) (i.e. regulation of monomer concentration) [6]. Wehave seen that with an increase of the reaction time (average degreeof branching detected by 1H NMR), an unexpected constant increaseof themolecularmass is detected by SEC after all availablemonomeris consumed in the reaction. We explained this anomaly by the hy-pothesis that somechains servemostlyasbranchdonors and someasacceptors during the one-pot enzymatic synthesis. Furthermore, wehave noticed that with an increase of the monomer concentration(degree of polymerization, followedby released inorganic phosphatefrom the used monomer G1P), an unexpected constant decrease ofthe molecular mass occurs. This atypical behavior was explained byhindered diffusion which changes the preferences of the enzymesand the fact that some of the cleaved short oligosaccharides serve asprimers instead of becoming branches [6].

For this reason, in order to verify the conclusions previouslymade based on the analysis, we compare and verify them by

Fig. 1. Proposed mechanism of the in vitro enzymatic synthesis of branched polysaccharidestime was varied. b) represents the second group of samples e concentration was varied.

analyzing the same enzymatically synthesized amylopectin analogswith AF4.

The system used for the separation and characterization ofamylopectin analogs was an AF4-MALLS-QELS setup. We success-fully determined the macromolecular characteristics and thestructural information of the branched polysaccharides andconcluded to have highly branched glycogen-like particles. Hereweshow that AF4 can superbly be used for the fractionation of highlybranched polysaccharides with various sizes and degrees ofbranching. Moreover, we confirm all the previously made as-sumptions based on SEC analysis concerning the enzymatic syn-thesis used in this work. By comparing, fulfilling, correcting andcombining the results from two powerful techniques such as SECand AF4withmulti detection, wewere able to show how importantthe establishment of improved characterization protocols forbranched polysaccharides is.

2. Experimental

2.1. Materials and methods

All chemicals used for the synthesis and AF4 (glucose-1-phosphate (G-1-P), tris(hydroxymethyl)aminomethane (Tris),dithiothreitol (DTT), adenosine monophosphate (AMP), phosphor-ylase b, sulfuric acid, NaN3, Orcinol) were purchased from Sigma-eAldrich and used without further purification. Glycogenbranching enzyme from D. geothermalis (Dg GBE) was kindly pro-vided by R.J. Leemhuis and L. Dijkhuizen, whereas maltoheptaose(G-7) was synthesized as explained elsewhere [14]. The water usedfor analysis and sample preparationwas produced by a RiOsTM andSynergy purification system (Millipore, Bedford, MA, USA).

2.2. Synthesis of well-defined branched polysaccharides [6]

G-1-P was dissolved in Tris buffer (100 mM, pH 6.7, 0.02% NaN3)containing G-7 (0.7 mM) as a primer, DTT (1.3 mM) as a reducingagent, and AMP (3.5mM) as a phosphorylase b activator, and the pH

using phosphorylase b and Dg GBE. a) represents the first group of samples e reaction

J. Ciric et al. / Polymer 55 (2014) 6271e6277 6273

was adjusted to 7. The polymerization was catalyzed by addition ofrabbit-muscle phosphorylase b (1.5 U/mL) and the branching wasinitiated by Dg GBE (250 U mL�1) at 37 �C. The reaction time was 1,3, 6, 9, 24, and 72 h, respectively, to obtain the average degree ofbranching 1�3, 3�5, 5�7, 7�9, 9�11, and 11�13%. The concentra-tion of G-1 P was 35e420mM to obtain the number average degreeof polymerization (DP) 30�350. Termination was done by a 5 minheat-treatment. The samples were dialyzed to remove the excess ofG-1-P, AMP, and DTT. Afterward, the samples were freeze-dried.Estimation of the number average DP was performed by UV-spectrometry while the average degree of branching was deter-mined by 1H NMR as explained in the literature [14].

2.3. AF4 analysis

The AF4 equipment, including the asymmetrical channel,Control-Box V3, Flow box P2.1, and the valve box, was obtainedfrom Consenxus (Ober-Hilbersheim, Germany). The channel ge-ometry was trapezoidal with a tip-to-tip length of 286 mm andbreadths at the inlet and outlet of 21.2 and 4.7 mm, respectively.

A 350 mm polyester spacer and a pure cellulose membrane witha cutoff point of 10,000 Da from Celgard LLB (Charlotte, NC) wereused. During all AF4 experiments, the sample was introduced intothe channel using a 100 mL loop injector (Valco Instruments Co.,Inc., Houston, TX). A Dawn® Heleos MALLS system fitted with a K5flow cell and a GaAs laser (l ¼ 658 nm) from Wyatt TechnologyCorporation (Santa Barbara, CA, USA) and an RID-10A refractometerfrom Shimadzu (Kyoto, Japan) were used as detectors. Online QELSmeasurements were performed at 142.5� for a time interval of 7 susing a WyattQELS® system (Wyatt Technology Corporation). Priorto use, the carrier (Millipore water containing 0.2% NaN3) wascarefully degassed and filtered through Durapore GV (0.22 mm)membranes (Millipore). The carrier was eluted initially at a flowrate of 1 mL min�1 for channel flow in (Fin). The crossflow (Fc) wasthen set at 1 mL min�1, and the channel flow rate (Fout) was set at0.2 mL min�1 for the sample introduction and relaxation/focusingperiod. The sample was injected at 0.1 mLmin�1 for 600 s. After theinjection pump was stopped, the sample was allowed to relax andfocus for 60 s.

For elution, Fout was set at 1 mL min�1, and Fc was reduced from1.3 to 0.1 mL min�1 for 600 s exponentially, after which it wasmaintained at 0.1 mL min�1 for 900 s. In order to elute the wholesample in case some aggregates existed, the flow was reduced to0 mL min�1 and maintained there for 300 s.

Prior to analysis, the samples were solubilized for 3 h at 50 �Cand overnight at room temperature in a water bath. All sampleswere filtered through 0.45 mm filters after dissolving. Sample re-coveries were calculated from the ratio of the mass eluted from thechannel (integration of the differential refractometric index (DRI)signal) and the injected mass. The injected masses were deter-mined using the sulfuric acid-orcinol colorimetric method (seebelow).

2.4. Data processing

The number and weight averagemolar masses (Mn ¼ Pci=

P ciMi

and Mw ¼ PciMi=

Pci, respectively), the dispersity Mw=Mn, and

R2gz ¼

PciMiR2gi=

PciMi were established using ASTRA software

from WTC (version astra 6.1.1 for Windows). The quantities ci, Mi,Rgi were obtained after processing the light scattering (LS) and DRIprofiles. The refractive index increment dn/dc used for all the cal-culations was 0.146 mL g�1. The normalization of photodiodes wasaccomplishedwith a lowmolarmass pullulan standard (P20). Berryextrapolation of the light-scattering equation (Equation (4)) for thescattered light to an angle of zero was used to obtain Mi and Rgi.

ffiffiffiffiffiffiffiffiffiffiffiffiffiffi�KcRq

�i

s¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1Mi

�1þ 16p2n2

3l2R2gi sin

2�q=2

��s(4)

where K is the optical constant, Rq is the excess Rayleigh ratio of thesolute, l is the wavelength of the incident laser beam, and q is theangle of observation. The Berry extrapolation method was usedrather than the classic Zimm plot since it allows a more accurateextrapolation for a very large polymer size [15]. Only the sevenlower angles (from 29� to 90�) were used for extrapolation. Thehydrodynamic radius of the slice i Rhi was calculated using DT fromonline QELS measurements with the StokeseEinstein equation (seeEquation (1)) as previously defined [16,19].

2.5. Sulfuric acid-orcinol colorimetric method

For the preparation of the sulfuric-orcinol reagent, firstly1400 mL of concentrated sulfuric acid (98%, d ¼ 1.83) was dilutedwith 650 mL of Millipore water in an ice bath. 2 g of orcinol dis-solved in 50 mL of water was added to the diluted sulfuric acid. Thesulfuric-orcinol reagent must be kept in the dark, and not usedlonger than two weeks.

For every sample the glucose content was determined beforeand after filtration prior to AF4 analysis, in order to determinewhether the polysaccharides were properly dissolved. Twentysamples per one calibration should be the maximum number ofsamples analyzed. Degradation of polysaccharides to glucose andspectroscopic colorimetric measurements was done on a contin-uous flow system from Bran et Luebbe (Plaisir, France), using thesulfuric-orcinol reagent [17]. Samples are injected into this systemvia an auto-sampler and a pumpwas used to carry all the fluids andto make intentional air bubbles every second in the analytical line,the samples were heated up to 98 �C andmixedwith the reagent. Atthe end of the line a colorimeter measures the absorbance at420 nm. The glucose concentration was determined using a cali-bration in the range of 0 and 100 mg mL�1 established with thesolutions of glucose standards with different concentrations usingsulfuric-orcinol colorimetric method.

3. Results and discussion

All analyzed samples were divided into the same two groups ofbranched polysaccharides as previously reported and analyzedwith SEC [6]. The first group consisted of polymers synthesized atdifferent reaction times in order to achieve different degrees ofbranching. From the SEC analysis a constant increase of the molarmass is expected with increase of reaction time. Whereas the sec-ond group consisted of polymers synthesized with different ratiosbetween the monomer and the primer (different monomer con-centrations) in order to influence molar mass. The constantdecrease of the molar mass is expected according to SEC.

3.1. Different reaction time e different degree of branching

The enzymatically synthesized amylopectin analogs that hadthe same starting ratio of monomer (G-1-P) and primer (G-7), butdifferent reaction times (different degree of branching) during thesynthesis were tested using AF4 with DRI, MALLS and QELS. Theaverage degree of branching increased with increase of reactiontime as previously shown by 1H NMR [18]. The characteristics of thetested samples are shown in Table 1, containing the average degreeof branching (DB) [14], and the weight-average and numberaverage molar masses ðMw; MnÞ, determined by AF4/DRI/MALLScoupling, for each sample; the hydrodynamic coefficient nh that was

Table 1Properties of the enzymatically synthesized branched polysaccharides withdifferent reaction times (different degree of branching).

S (Time) h G�1�PG�7 DBa, % Mn , g mol�1

AF4 waterMw , g mol�1

AF4 watervh

b

1.4 (1) 300 3 1.36 � 105 5.04 � 105 ND2.4 (3) 300 7 2.08 � 105 1.03 � 106 0.273.4 (6) 300 7 2.57 � 105 2.20 � 106 0.334.4 (9) 300 8 3.91 � 105 3.35 � 106 0.305.4 (24) 300 10 5.66 � 105 4.56 � 106 0.326.4 (72) 300 12 7.18 � 105 6.33 � 106 0.32

a Determined via 1H NMR spectroscopy.b Experimental slope of the logelog plot of Rh versus molar mass.

J. Ciric et al. / Polymer 55 (2014) 6271e62776274

determined from Rh, calculated using online QELS measurements,as explained further on in this paper.

AF4 analyses of all analyzed polysaccharides from this groupshowed elution recoveries higher than 97%, which indicates thatthe fractionation response was quantitative for all samples.

The Mn obtained with AF4 was in the same range or slightlyhigher than the Mn, measured with SEC in different solvents andmonotonically increased with the increase of the reaction time andthe average degree of branching. These results confirm theassumption that not all chains uniformly serve as branching donorsand acceptors; some linear amylose chains that are created by theaction of phosphorylase plainly serve as branch donors and thereforeget completely consumed in the end. This results in fewerchains thanthe used primer sequences (G-7) at the end of the reaction.

The plots of the LS signal versus the elution volume (Ve) for allsamples from this group (Fig. 2 (a)) showed a shift of the peaks tohigher Ve with increase of the reaction time, meaning that thelonger the reaction time the bigger the molecules are. The samplesynthesized for the shortest time, showed an additional LS signal, atlong elution time, that is most probably an artifact (no concentra-tion for this size) and could correspond to associates in very lowconcentrations.

Size (Rh) distributions measured for each sample, using onlineQELS, can be seen in Fig. 2 (b). The Rh was fitted with polynomial fit(fit based on all measured samples) for the improvement of thevalues for samples measured with lower quality. With increase ofthe reaction time the Rh values shifted to higher values, indicatingan increase in size. The distributions were bimodal for all samples.These results fully agree with the different size distributionsobserved via SEC measurements. The first population existed in thelow Rh area up to 10 nm, whereas the second population existed inthe high Rh area with Rh up to 40 nm. Fig. 2 (b) clearly shows that

Fig. 2. (a) Elugrams, (b) size, and molar mass distributions of the enzymatically synthesized b(Sample codes from Table 1).

the amount of low Rh molecules decreases with increase of thereaction time, whereas the amount of molecules with high Rh in-creases. In both areas, the most dominant Rh shifted towards highervalues with increase of reaction time.

Structural information can be obtained from the exponents nh orng using the power-law equations:Rh ¼ KhM

nhw and Rg ¼ KgM

ngw ,

respectively, in which Kh and Kg represents the corresponding co-efficients. The exponent n should remain the same in the context ofthese laws, for linear polymers or self-similar objects [19]. Its valuedepends upon temperature, polymeresolvent interactions andpolymer shape, and: nh¼ ng¼ 0.33, for a sphere, nh¼ ng¼ 0.5e0.6 fora linear randomcoil [20], and nh¼ ng¼1 for a rod. The correspondingcoefficients are not the same and they are dependent on the chosensolvent and the detailed monomer structure [21].

For the majority of the polysaccharides, for which the Rh valueswere measurable, structural information was determined from theexponent nh (slope of the logelog plot of the hydrodynamic radiusversus molar mass) using the power-law equation: Rhi ¼ KhM

nhi ,

and are listed in Table 1. Fig. 3 (a) compiles the experimental datafor Rh of all samples from this group, and it is clear that a quality fitcan be achieved using the power law equation for Rh > 10 nm. Theparameter nh varied between 0.27 and 0.33, which is characteristicfor compact spheres [22]. This was expected, since these hyper(-branched) amylopectin analogs are very similar to glycogens [19].

Unfortunately, due to many different molecular sizes in onemeasured polysaccharide, and a number of the lowmolecular sizes(Rg < 10 nm) of measured polysaccharides (in particular for samplesafter short reaction times), determination of Rg was not possibleusing MALLS for the whole distribution therefore ng (Rgi ¼ KgM

ngi

slope of the logelog plot of the radius of gyration versus molarmass) values were not good enough due to an inadequate fit.Sample 6.4 was an exception: two slopes along the logelog plot ofthe Rg versusmolarmasswere determined for Rg higher than 10 nm(Rh > 20 nm, molar mass >>106 g mol�1), corresponding to ng of0.68 for molar masses <1.88 � 107 g mol�1 and 0.43 for molarmasses >1.88 � 107 g mol�1. Such behavior had already been re-ported for rabbit liver glycogen by Rolland-Sabat�e et al [4,19]. Thedifferences between Rg and Rh, and in particular the Rg/Rh ratio,provides information on the solution conformation of the macro-molecule studied [19,23]. By plotting the Rg and Rh, versus molarmass on the same graph for sample 6.4 (results not shown) one canobserve that for the high molar mass fraction, the Rg’s are slightlysmaller than the Rh, giving a Rg/Rh ratio of about 0.8 which corre-sponds theoretically to the homogeneous hard sphere structure(0.778) [23] in linewith previous data on glycogens [4,19]; whereas,for the low molar mass fraction, the Rg are very small compared to

ranched polysaccharides with different reaction times (different degrees of branching).

Fig. 3. (a) Rh (from online QELS measurements) versus molar mass (b) the apparent molecular density (dHappLS) versus molar mass.

Table 2Properties of the enzymatically synthesized branched polysaccharides withdifferent monomer concentrations (different degree of polymerization).

S G�1�PG�7 DBa, % Mn , g mol�1

AF4 waterMw , g mol�1

AF4 watervh

6.1 50 12 5.52 � 106 4.62 � 107 0.256.2 100 12 4.34 � 106 3.64 � 107 0.256.3 200 12 2.84 � 106 2.18 � 107 0.286.4 300 12 7.18 � 105 6.33 � 106 0.326.5 400 11 4.22 � 105 6.02 � 106 0.326.6 500 10 1.82 � 105 1.31 � 106 0.336.7 600 10 2.06 � 105 2.48 � 106 0.32

a Determined via 1H NMR spectroscopy.

J. Ciric et al. / Polymer 55 (2014) 6271e6277 6275

the Rh and tend to increase with molar mass. This leads to Rg/Rhratios smaller than 0.6 which could correspond to microgel struc-tures [23]. Samples 4.4 and 5.4 tend to have the same behavior.

The apparent molecular density distribution dHappLSi¼ Mwi=

4p3 R

3hi

calculated with the fitted Rh can be seen in Fig. 3 (b). For everysample, for the whole Rh and molar mass range the density wasvery high, confirming the highly dense branched structure of thepolysaccharides. One can also see a slight increase in the densitywith reaction time, hence an increase in the branching. Sample 1.4should have the lowest density of all samples since it was synthe-sized for only one hour. The observed higher than expected densityis due to the fitting, since the experimental data of this sample wasnot very good due to small size of the molecules. In the low Rh andmolar mass area (molar mass <106 g mol�1) the density increasewith the molar mass for the majority of the samples. However, theobserved increase of dHappLSi

with the molar mass was not expectedfor a single spheroidal molecule with an homogeneous structurealong the molar mass distribution [19]. This unexpected behavior isnot linked to aggregation, as the molecules obtained are very sol-uble; and it is then surely the result of an important increase ofbranching with size. In the high Rh area the density becomes thesame after a specific reaction time, which means that even thoughthe average degree of branching increases the molecules stop to befurther branched in this area after a certain time.

3.2. Different monomer concentration e different degree ofpolymerization

The amylopectin analogs that had the same reaction time (72 h)but a different starting G� 1� P=G� 7 ratio (different degree ofpolymerization) during the synthesis, were analyzed in the sameway as outlined above. From SEC analysis a bimodal distribution ofhighly branched molecules decreasing in the average molar massand size with increase of the monomer concentration was ex-pected. The properties measured via AF4 are shown in Table 2. BothMn and Mw decreased constantly with increase of the monomerconcentration. The only exception is the sample in which thehighest monomer concentration was used (sample 6.7;G� 1� P=G� 7 ratio ¼ 600) � the molar mass is slightly higherthan for the G� 1� P=G� 7 ratio ¼ 500. The decrease in the molarmass confirmed that due to increase of the concentration andtherefore hindered diffusion some cleaved oligosaccharides did notreach the final destination (other polysaccharide chains) andbecame branches, but served as new primers for new linear chains.The slight increase in the molar mass of the sample in which thehighest monomer concentration was used could be an artifact,however it is more likely an indication for the change of theenzymatic polymerization mechanism as already observed by SEC.

This means that no more oligosaccharides are cleaved, and that theexisting branched polysaccharides are simply serving as primersand are being enlarged by G-1-P with phosphorylase as catalyst(see Fig. 1).

Polysaccharides from this group with a ratio of monomer toprimer higher than 100, showed elution recoveries higher than95%, which indicates that the fractionation response wasquantitative.

The plots of LS signal versus Ve for all samples from this group(Fig. 4 (a)) showed a shift of the peak to lower Ve with increase ofmonomer concentration, meaning that the higher the usedmonomer concentration the lower the molecular size is.

Size distributions made for each sample, using the onlinemeasured Rh (QELS), fitted with a polynomial fit (fit based on allmeasured samples) can be seen in Fig. 4 (b). The distributions werebimodal for all samples. With increase of the monomer concen-tration Rh values were shifted to lower values in the high Rh area,whereas in the low Rh area the Rh values stayed the same. Theseresults fully agree with the different size distributions observedwith SEC measurements. The first population existed in the low Rharea up to 10 nm, whereas the second population existed in thehigh Rh area with Rh up to 80 nm. Fig. 4 (b) clearly showed that theamount of the polysaccharides in the low Rh area increased withincrease of the monomer concentration, whereas the amount ofpolysaccharides in a high Rh area increased till the critical monomerconcentration (G� 1� P=G� 7 ratio ¼ 500) after which theirconcentration decreased.

For all polysaccharides, structural information was determinedfrom the exponent nh as explained above, and can be seen in Table 2.Fig. 5 (a) shows all experimental data for Rh of all the samples fromthis group, and it is clear that a quality fit can be achieved. Theparameter nh varied between 0.25 and 0.33, which is characteristicfor compact spheres. For Rg and ng values the same problems asdiscussed above occurred for the majority of samples.

Fig. 4. (a) Elugrams, (b) size, and molar mass distributions of the enzymatically synthesized branched polysaccharides with different monomer concentration (different degree ofpolymerization). (Sample codes from Table 2).

J. Ciric et al. / Polymer 55 (2014) 6271e62776276

Nevertheless it was possible to determine ng values for Rg higherthan 10 nm (Rh > 20 nm, molar mass >>106 g mol�1) in the “largersamples” synthetized using G� 1� P=G� 7 ratio < 300. For molarmasses <1.88� 107 gmol�1, ng of 0.68 and 0.55were obtained for 6.4and 6.5, respectively and for molar masses >1.88� 107 g mol�1, ng of0.42, 0.41, 0.46 and 0.43 were obtained for 6.1, 6.2, 6.3 and 6.4,respectively. These values are close together indicating similar globalstructures and they are in agreement with those previously obtainedfor natural glycogens [4,19], in line with a spherical structure in so-lution. Moreover, the plot of the Rg/Rh ratio versus molar mass in thehighRh area (resultsnot shown) showRg/Rhvalues smaller than0.6, asfor6.4 in thefirstgroup, formolarmass<1.88�107gmol�1 andabout0.7e0.8 for G� 1� P=G� 7 ratio < 500 in the high molar mass area(molar mass >1.88 � 107 g mol�1) indicating microgel like structure,on one hand, and spherical structure, on the other hand [23].

The apparent molecular density (distribution calculated withfitted Rh Fig. 5 (b)), was very high for all the samples throughout thewhole distribution, as expected. Moreover, an unexpected increaseof dHappLSi

with the molar mass was also observed for this group ofsamples, which is to be linked to an important increase ofbranching with size.

For the samples with a lower ratio (G� 1� P=G� 7 ratio < 400),while the branching degree remain stable (Table 2), an unexpectedincrease of the density with increase of the monomer concentra-tion in the low Rh andmolarmass area (molarmass <106 gmol�1) isobserved.

For the samples with a lower ratio (G� 1� P=G� 7 ratio < 300)we see an unexpected increase of the density with increase of themonomer concentration, in the low Rh and molar mass area.Whereas a slight decrease in the density for the samemolar mass ofthe samples synthesized with G� 1� P=G� 7 ratio ¼ 300 andhigher is visible, which agrees well with the decrease in the degree

Fig. 5. (a) The Rh from online QELS measurements versus molar mass (b) the apparen

of branching of themolecules (Table 2). The result for the branchingbehavior in the low molar mass (molar mass <106 g mol�1), sinceSEC analysis showed constant decrease in molar mass for the samesize for every sample, indicating constant decrease in branching.The difference between SEC results and AF4 results in this area forthe lower ratio could be due to the use of universal calibration forSEC analysis and the use of online QELS for AF4 analysis. Addi-tionally, the concentration of these molecules is very low; hencethe correct detection can always be questioned. In the high Rh areawe can clearly see the confirmation of the SEC results [6]. Thedecrease of the apparent molecular density with increasing G� 1�P=G� 7 ratio is clear, and after a critical monomer concentration anincrease is visible, which indicates the change of the mechanism ofthe branching enzyme, as previously assumed. In addition, thevariations in density are in line with Rg/Rh values calculated in thehigh molar mass area (molar mass <1.88 � 107 g mol�1) whichslightly increase with the G� 1� P=G� 7 ratio, in line with adecrease of branching in the molecule, and then decrease for 6.7 inthe high molar mass area (molar mass >1.88 � 107 g mol�1).

In our previous work (combination of SEC and MALDI analyses)we proposed that a critical number of molecules existed(G� 1� P=G� 7 ratio > 400), after which the process of diffusionbecomes hindered and the branching pattern changes. Unfortu-nately, solely fromAF4 analysis of thewhole polysaccharide sampleit is not possible to confirm the difference in the branching patternfor samples with a critical monomer concentration. Fraction anal-ysis, or partial degradation of the polysaccharides and further an-alyses could support the speculations about the different branchingpattern. Nevertheless, the critical point after which a change startsto occur is obvious, when the branching density starts to increase.

Recently, very similar branched polysaccharides were synthe-sized with amylosucrase from Neisseria polysaccharea and the

t molecular density (dHappLS) versus molar mass. (Sample codes from Table 2).

J. Ciric et al. / Polymer 55 (2014) 6271e6277 6277

branching enzyme (BE) from Rhodothermus obamensis using su-crose as a unique substrate [24], and the effect of the monomerconcentration was analyzed with the same AF4 system undersimilar conditions [5]. It is interesting that the increase of themonomer concentration has a completely opposite effect on thesize and structure of the synthesized polysaccharides. In this case,as expected, with increase of the monomer concentration, themolar mass also increased. Whereas, in the low Rh area, the amountof fractions decreased, and in the high Rh area the amount offractions increased with the monomer concentration. nh valueswere slightly higher but still in the range for a compact sphere.Additionally for some samples, authors were not able to determineRg, like in the case of the polysaccharides studied in our work.However, for the majority of the samples they were able to deter-mine Rg, hence ng which was the same as nh and Rg/Rh ratio whichwas close to the theoretical values of a dense spherical structure.The differences between the two polysaccharide samples areprobably due to different enzymes, their mechanisms and sub-strates used in the polymerizations. Moreover the size distributionswere completely different, and the bimodal distribution was morepronounced in our research. Problems with the determination of Rgin our case are due to highly compact structures leading to verysmall Rg then not measurable bymeans of MALLS, as well as varietyof branched structures inside one polysaccharide.

4. Conclusions

AF4 with multi detection has been confirmed to be a verypowerful technique for the analysis of the macromolecular struc-ture of branched polysaccharides [4,5,19]. Not only did we suc-cessfully determine the size and molar mass distributions ofenzymatically synthesized amylopectin analogs, but we alsoconfirmed their highly branched structure. The assumptions con-cerning the molecular mechanism of the enzymatic polymerization(Fig. 1) e used in our previous SEC work [6] proved to be correct.

With increase of the reaction time/degree of branching, a con-stant increase of the molar masses confirmed that some chainsserve mostly as branch donors and some as acceptors during theone-pot enzymatic synthesis. Additionally, we clearly showed thatin the high Rh area after 6 h amylopectin analogs stop beingbranched further on, even though the average degree of branchingincreases with time.

With increase of the monomer concentration/degree of poly-merization, we observed a clear decrease of the molar masseswhich can be explained by the fact that some of the short oligo-saccharides cleaved by Dg GBE serve as primers instead ofbecoming branches due to hindered diffusion.

Using AF4-MALLS-QELS we clearly determined the macromo-lecular characteristics of the branched polysaccharides. Concerningstructural information based on hydrodynamic radius, weconcluded to have highly branched glycogen-like particles in allcases. Unfortunately, due to the low molecular sizes and difficultdetermination of the radius of gyration, structural informationbased on this radius could not be taken into account for the wholedistributions.

In this paper we showed that AF4 could superbly be used for thefractionation of highly branched polysaccharides with various sizes

and degrees of branching. Moreover, we confirmed all the as-sumptions based on SEC analysis concerning the enzymatic syn-thesis used in this work.

When SEC and AF4 are compared for the aqueous system, due tothe absence of columns (no stationary phase) and no interaction ofpolysaccharides with the column material, preferably AF4 shouldbe used, especially for bigger molecules. Unsurprisingly, if afford-able and accessible, the combination of these two techniques is thebest one for the characterization of branched polysaccharides.

Acknowledgments

This research was supported by the Dutch Polymer Institute(DPI) under Project no. 727 a. Financial support of this research byThe Netherlands Organisation for Scientific Research (NWO) via aVIDI innovational research grant (700.59.424) is greatly acknowl-edged. The authors would like to thank Prof. Dr. L. Dijkhuizen andDr. R.J. Leemhuis from the Groningen Biomolecular Sciences andBiotechnology Institute for kindly providing Dg GBE. We are alsovery grateful to INRA for providing the possibility to perform theresearch in their facilities with their materials and equipment.

References

[1] Cave RA, Seabrook SA, Gidley MJ, Gilbert RG. Biomacromolecules 2009;10:2245e53.

[2] Vilaplana F, Gilbert RG. J Sep Sci 2010;33:3537e54.[3] Rolland-Sabat�e A, Guilois S, Jaillais B, Colonna P. Anal Bioanal Chem 2011;399:

1493e505.[4] Rolland-Sabat�e A, Colonna P, Mendez-Montealvo MG, Planchot V. Bio-

macromolecules 2007;8:2520e32.[5] Rolland-Sabat�e A, Guilois S, Grimaud F, Lancelon-Pin C, Roussel X, Laguerre S,

et al. Anal Bioanal Chem 2014;406:1607e12618.[6] Ciric J, Oostland J, de Vries JW, Woortman AJ, Loos K. Anal Chem 2012;84:

10463e70.[7] Schimpf M, Caldwell K, Giddings JC, editors. Field-flow fractionation hand-

book. new York, USA: Wiley interscience; 2000.[8] Wahlund KG. J Chromatogr A 2013;1287:97e112.[9] Ratanathanawongs Williams SK, Lee D. J Sep Sci 2006;29:1720e32.

[10] Wittgren B, Wahlund K-G. J Chromatogr A 1997;760:205e18.[11] Wittgren B, Wahlund K-G, D�erand H, Wessl�en B. Macromolecules 1996;29:

268e76.[12] Håkansson A, Magnusson E, Bergenståhl B, Nilsson L. J Chromatogr A

2012;1253:120e6.[13] Magnusson E, Håkansson A, Janiak J, Bergenståhl B, Nilsson L. J Chromatogr A

2012;1253:127e33.[14] van der Vlist J, Reixach MP, Maarel Mvd, Dijkhuizen L, Schouten AJ, Loos K.

Macromol Rapid Commun 2008;29:1293e7.[15] Aberle T, Burchard W, Vorwerg W, Radosta S. Starch e St€arke 1994;46:

329e35.[16] Wittgren B, Wahlund K-G. J Chromatogr A 1997;791:135e49.[17] Planchot V, Colonna P, Saulnier L. Dosage des glucides et des amyloses. Paris,

France: Lavoisier; 1997. p. 341e98.[18] Ciric J, Loos K. Carbohydr Polym 2013;93:31e7.[19] Rolland-Sabat�e A, Mendez-Montealvo MG, Colonna P, Planchot V. Bio-

macromolecules 2008;9:1719e30.[20] Flory PJ. Principles of polymer chemistry. Cornell University Press; 1953.[21] De Gennes PG. Scaling concepts in polymer physics. Cornell university press;

1979.[22] Burchard W. Static and dynamic light scattering from branched polymers and

biopolymers. In: Light scattering from polymers. Berlin Heidelberg: Springer;1983. p. 1e124.

[23] Burchard W. Advances in polymer science, vol. 143. Berlin,: Springer-Verlag;1999. p. 113e94.

[24] Roussel X, Lancelon-Pin C, Viksø-Nielsen A, Rolland-Sabat�e A, Grimaud F,Potocki-V�eron�ese G, et al. Biochimica Biophysica Acta (BBA) e General Subj2013;1830:2167e77.