Effect of Polysaccharide Conformation on Ultrafiltration Separation Performance

The manifold array of saccharide linkages leads to a great variety of polysaccharide architectures, comprising three conformations in aqueous solution: compact sphere, random coil, and rigid rod.

Available online 17 February 2021

( http://creativecommons.org/licenses/by-nc-nd/4.0/ ).

Effect of Polysaccharide Conformation on Ultrafiltration

Separation Performance

Severin Edera, Patrick Zueblina, Michael Dienerb, Mohammad Peydayeshb, Samy Boulosa,

Raffaele Mezzengab, Laura Nystr¨oma,*

aETH Zurich, Department of Health Science and Technology, Institute of Food, Nutrition and Health, Laboratory of Food Biochemistry, Schmelzbergstrasse 9, 8092

Zurich, Switzerland

bETH Zurich, Department of Health Science and Technology, Institute of Food, Nutrition and Health, Laboratory of Food and Soft Materials, Schmelzbergstrasse 9, 8092

Zurich, Switzerland

A R T I C L E I N F O

Keywords:

Ultrafiltration

Polysaccharide conformation

Polysaccharide separation

Molecular weight cut-off deviation

Glucose-based polysaccharide

A B S T R A C T The manifold array of saccharide linkages leads to a great variety of polysaccharide architectures, comprising three conformations in aqueous solution: compact sphere, random coil, and rigid rod This conformational variation limits the suitability of the commonly applied molecular weight cut-off (MWCO) as selection criteria for

polysaccharide ultrafiltration membranes, as it is based on globular marker proteins with narrow M w and hy-drodynamic volume relation Here we show the effect of conformation on ultrafiltration performance using randomly coiled pullulan and rigid rod-like scleroglucan as model polysaccharides for membrane rejection and molecular weight distribution Ultrafiltration with a 10 kDa polyethersulfone membrane yielded significant different recoveries for pullulan and scleroglucan showing 1% and 71%, respectively We found deviations

greater than 77-fold between nominal MWCO and apparent M w of pullulan and scleroglucan, while recovering

over 90% polysaccharide with unchanged M w We anticipate our work as starting point towards an optimized membrane selection for polysaccharide applications

1 Introduction

The global production of polysaccharides in nature considerably

exceeds the production volume of any other polymer Polysaccharides

constitute the central carbon source for living organisms and provide a

basis for all life on our planet (Navard & Navard, 2012) In recent

de-cades, polysaccharides aroused great interest in research and across

various industries owing to their unique biological and physiological

properties, such as biocompatibility and –degradability paired with

atoxic characteristics (Muzzarelli, 2012) In particular, polysaccharide

purity becomes a crucial product criterion in applications involving

humans, such as biomedicine or food technology (Pinelo, Jonsson, &

Meyer, 2009) Furthermore, as the bioactive potential of

polysaccharides is distinctly related to their chemical structure and molecular weight, selective purification processes become indispensable (Wang et al., 2017)

Common purification techniques such as chromatography, evapo-ration and ion exchange require resource-intensive opeevapo-ration and maintenance, involve substantial investment costs, and lack scalability

In comparison to these conventional separation and purification pro-cesses, membrane filtration offers several merits, including high effi-ciency, simple modification of operating variables and low energy requirements (Cano & Palet, 2007; Chen et al., 2020) In addition, membrane separation is especially suited for heat-sensitive bio-molecules as it is operated at room temperature (RT) and without phase transfer (Sun, Qi, Xu, Juan, & Zhe, 2011)

Abbreviations: α, Mark-Houwink parameter; AFM, atomic force microscopy; AN-scleroglucan, alkaline-treated and neutralized scleroglucan; ANS-scleroglucan,

alkaline-treated, neutralized and sonicated scleroglucan; Đ, dispersity index; DLS, dynamic light scattering; dn/dc, refractive index increment; HPAEC,

high-per-formance anion-exchange chromatography; HPSEC, high-perhigh-per-formance size exclusion chromatography; HY, hydrosart; IEP, isoelectric point; LALS, low-angle light scattering; Δ%M w , percentage difference in M w between the retentate and feed solution; M n , number average molecular weight; M w, weight average molecular weight; MWCO, molecular weight cut-off; PAD, pulsed amperometric detection; PES, polyethersulfone; RALS, right-angle light scattering; RI, refractive index; RT, room temperature; SSE, sum of squared error; ζ, zeta potential; [η], intrinsic viscosity

* Corresponding author

E-mail address: laura.nystroem@hest.ethz.ch (L Nystr¨om)

Contents lists available at ScienceDirect Carbohydrate Polymers

https://doi.org/10.1016/j.carbpol.2021.117830

Received 9 November 2020; Received in revised form 12 February 2021; Accepted 13 February 2021

Over the past years, the advances in ultrafiltration technology have

led to the development of refined membranes, enabling selective

sepa-ration of saccharides with molecular weights as low as 3 kDa (Pinelo

et al., 2009; Sun et al., 2011) The rejection properties of ultrafiltration

membranes are reflected in the molecular weight cutoff (MWCO) Its

arbitrary definition comprises the lowest molecular weight at which

90% of the solute is retained by the membrane (Koros, Ma, & Shimidzu,

glob-ular marker proteins for calibration, although an industry-wide standard

is still lacking (Scott, 1995) Since globular proteins fold to sphere-like

structures in solution, they present a narrow relation of molecular

weight to hydrodynamic volume, influencing the separation factor of

membrane filtration that is governed decisively by hydrodynamic

vol-ume (Pinelo et al., 2009) Unlike the linear sequences found in peptide

bonds, polysaccharides form diverse primary structures with a variety of

condensation linkages (Liu, Brameld, Brant, & Goddard, 2002) This

extra dimension of geometry conjunct with variable saccharide units

leads to a remarkable world of polymeric architecture (Atkins, 1985) It

is generally recognized that the conformation of polysaccharides in

so-lution comprises three distinct patterns with increasing rigidity:

compact sphere, random coil, and rigid rod (Harding, Abdelhameed, &

Morris, 2011) The respective glycosidic linkage geometry of a

poly-saccharide mainly defines its conformation in aqueous solution, which

in turn determines the hydrodynamic volume (M Q Guo, Hu, Wang, &

Ai, 2017) The polysaccharides pullulan and scleroglucan can be

considered as extreme representatives of their respective

conforma-tional cluster owing to their diverse glycosidic linkage patterns

Pul-lulan, a linear water-soluble (1→4;1→6)-α-D-glucan produced by the

polymorphic fungus Aureobasidium pullulans, behaves as random coil in

aqueous solution (Nishinari et al., 1991) The rotational freedom

pro-vided by the α-(1→6)-linkages enables flexible folding along the

poly-mer chain (Gidley & Nishinari, 2009) Scleroglucan, a water-soluble

(1→3;1→6)-β-D-glucan produced by fungi of the genus Sclerotium

(Coviello et al., 2005), exhibits the (1→6)-linkage only in the sidechain

of D-glucopyranosyl residues attached to the (1→3)linked backbone

(Castillo, Valdez, & Farina, 2015) The alignment of scleroglucan strands

in aqueous solution results in a stiff rigid rod-like conformation (

Slet-moen & Stokke, 2008) Ultimately, distinct hydrodynamic volume and

spatial orientation of polysaccharides with comparable molecular

weight may restrict the applicability of MWCO as suitable selection

guide for polysaccharide ultrafiltration membranes Numerous studies

reported discrepancies between the apparent and the nominal MWCO

provided by the manufacturer (Kim et al., 1994; Platt, Mauramo,

Butylina, & Nystrom, 2002) Platt et al (2002) found apparent MWCOs

lower than the nominal when filtering polyethylene glycol solutions and

excluded fouling and concentration polarization as underlying cause

Sun et al (2011) observed a considerable loss of a polysaccharide

mixture subjected to different ultrafiltration membranes that should

have retained the investigated fraction according to the manufacturer’s

MWCO However, governing factors for the discrepancies in membrane

separation obtained were not suggested

So far, the conformation of polysaccharides in ultrafiltration

appli-cations was assessed mostly in terms of the resulting hydrodynamic

volume for membrane and MWCO selection The ultimate effect of

polysaccharide conformation on membrane transport during

ultrafil-tration attracted little attention until now Mathematical simulations

focusing on the adaptation of steric pore models to include capsular-

shaped molecules, or the assessment of the probability of elongated

shapes entering a membrane pore compared to spherical particles

showed the necessity to consider also conformation for a better

under-standing of ultrafiltration separation (Montesdeoca, Bakker, Boom,

Janssen, & Van der Padt, 2019; Vinther, Pinelo, Brons, Jonsson, &

Meyer, 2012) However, the systematic assessment of the effect of

polysaccharide conformation on the resulting ultrafiltration

perfor-mance remains yet unaddressed

The present work seeks to describe the effect of polysaccharide

conformation on the separation performance of ultrafiltration mem-branes with an empirical approach For this purpose, pullulan and scleroglucan as model polysaccharides were subjected to crossflow ul-trafiltration with two membrane materials, namely Hydrosart (HY), a low-binding regenerated cellulose material, and polyethersulfone (PES), exhibiting various MWCO (2, 3, 5, and 10 kDa) Particular attention was paid to the exclusion of any potential effect on membrane separation beside the polysaccharide conformation The requirements for the polysaccharide model solution to guarantee an unambiguous assign-ment of ultrafiltration variation owing to the effect of conformation encompassed: (i) identical monomeric units; (ii) uniform distinct conformation in aqueous solution; (iii) solubility and conformation stability in aqueous solution; (iv) comparable weight average molecular

weight (M w ) with ΔM w ≈ 100 kDa Moreover, state-of-the-art size- exclusion chromatography coupled to light scattering and viscometer detectors (HPSEC-triple detection), high-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD), and high-resolution atomic force microscopy imaging (AFM) assured comprehensive evaluation of the ultrafiltration retentate in terms of polysaccharide yield, conformation, and molecular weight

2 Materials & methods

2.1 Chemicals

Pullulan powder was purchased from Carbosynth (Berkshire, United Kingdom) Scleroglucan powder was obtained from Elicityl (Crolles, France) D-Glucose anhydrous (≥ 99.5%), sodium azide (NaN3; >99%),

sodium hydroxide (NaOH, ≥ 98%), sodium hypochlorite solution (NaClO), sodium nitrate (NaNO3; ≥ 99.5%), D-sorbitol (99%) and

tri-fluoroacetic acid (TFA, >99.9%) were purchased from Sigma-Aldrich (St Louis, United States) Hydrochloric acid (HCl, >37%) was

ob-tained from VWR International (Radnor, United States) All solutions were prepared with purified water using a Millipore MilliQ-system (Billerica, United States)

2.2 Preparation of polysaccharide standard solutions

Pullulan was dissolved at RT under stirring for 1 h Scleroglucan was dissolved at 80 ◦C under stirring for 24 h We selected the mildest possible conditions facilitating complete dissolution of both poly-saccharides Pullulan and scleroglucan exhibit different flexibilities in their structure that affect the strength of intermolecular interactions and thus require adapted dissolution procedures Pullulan and scleroglucan solutions were prepared at 0.1% (w/v) for polysaccharide character-ization and at 0.025% (w/v) for ultrafiltration feed solutions Poly-saccharide solutions were prepared taking into consideration the purity assessment of the crude polysaccharide powder (w/w) (see Section 2.5)

To unify the dispersity and to reduce the molecular weight to a com-parable level with pullulan, scleroglucan feed solution was preliminary treated with 0.2 M NaOH at RT for 10 min and subsequently neutralized with HCl, followed by centrifugation at 9000 rpm for 15 min, resulting

in alkaline-treated and neutralized scleroglucan (AN-scleroglucan) The solution was then subjected to ultrasonic treatment over a total duration

of 180 min with a probe sonicator (UP200H, Hielscher, Germany), operated at 100% pulsation with 80% amplitude, resulting in alkaline- treated, neutralized and sonicated scleroglucan (ANS-scleroglucan) During the sonication procedure, the solution was cooled in an ice bath and kept under stirring to prevent heating Prepared pullulan feed so-lution was used without further treatment All polysaccharide soso-lutions were filtered through a 0.45 μm Nylon filter prior to analysis or ultrafiltration

2.3 Crossflow ultrafiltration set-up and procedure

Crossflow ultrafiltration was conducted with a Vivaflow 200

cross-flow device equipped with Hydrosart (HY) or polyethersulfone (PES)

crossflow membrane cassettes (Sartorius AG, G¨ottingen, Germany) The

nominal molecular weight cut-offs (MWCO) provided by the

manufac-turer were 2, 5 and 10 kDa for HY and 3 and 10 kDa for PES membranes

Further characteristics of the membrane can be found in the

supple-mentary information (Table S1, Fig S1) Ultrafiltration was performed

in constant volume diafiltration operation mode at a constant pressure

of 2.5 bar set with a Masterflex L/S peristaltic pump (Cole-Parmer

GmbH, Wertheim, Germany) (Fig 1) The resulting circulation flowrates

were between 20.9–24.5 L/h (Fig S2) In each trial, 250 mL of

poly-saccharide feed solution (0.025%, w/v) were subjected to diafiltration

for 1 h at RT MilliQ water, connected to the feed tank, was used as

exchange solution in order to maintain a constant volume of 250 mL

The number of diavolumes exchanged during the crossflow diafiltration

were recorded for each membrane (Table S2) Crossflow ultrafiltrations

were conducted in triplicates for each membrane material with distinct

MWCO After each ultrafiltration run, the ultrafiltration device was

washed with the corresponding washing solution to avoid carry-over

Washing solutions were 0.5 M NaOH and 0.5 mM NaOCl in 0.5 M

NaOH for HY and PES membranes, respectively The feed solutions and

retentates were analyzed in terms of molecular weight distribution and

conformational parameters using size-exclusion chromatography

coupled to light scattering and viscometer detectors (HPSEC-triple

detection) (see Section 2.4) The resulting yield of the respective

poly-saccharide in the retentate was obtained as mass-ratio according to the

following formula:

m feed

where m retentate and m feed are the masses of available polysaccharide

in the retentate and feed solution, respectively, determined by high-

performance anion-exchange chromatography with pulsed

ampero-metric detection (HPAEC-PAD) after TFA hydrolysis (see Section 2.5)

(Cheryan, 1998) The standard deviation of Eq (1) was calculated,

assuming independent variables, according to (Taylor, 1982):

σYield= Yield

̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅

((

σm retentate

m retentate

)2 +

(

σm feed

m feed

)2)

√

√

2.4 Molecular weight determination and conformation analysis

The weight average molecular weight (M w ), dispersity Đ (M w /M n),

intrinsic viscosity [η], and the Mark-Houwink plot of the polysaccharide feed and retentate solutions were determined using high-performance size exclusion chromatography (HPSEC) equipped with triple detec-tion (OMNISEC, Malvern Panalytical Ltd., Malvern, United Kingdom) according to the procedure described by Demuth, Betschart, and Nystr¨om (2020) In short, the HPSEC-triple detection system consisted of

a OMNISEC resolve unit (OMNISEC, Malvern Panalytical Ltd, Malvern, United Kingdom) coupled to the multi-detector module OMNISEC reveal (OMNISEC, Malvern Panalytical Ltd, Malvern, United Kingdom) encompassing a refractive index (RI), right-angle light scattering (RALS)

at 90◦, low-angle light scattering (LALS) at 7◦, and a viscometer detec-tor Two A6000M columns with an exclusion limit of 20 000 000 Da connected in series (Malvern Panalytical Ltd., Malvern, United Kingdom) were maintained at 30 ◦C Polysaccharide solutions were filtered through a 0.45 μm Nylon syringe filter prior to analysis Sample injections of 100 μL were eluted with 0.1 M aq NaNO3 containing 0.02% (w/v) NaN3 at a flow rate of 0.7 mL/min The system was calibrated with

a one-point calibration using a Malvern PolyCAL™ polyethylene glycol (PEO24 K) standard and verified with a dextran (DEX-T70 K) standard Data analysis was performed using the OMNISEC 10.30 software (Mal-vern Panalytical Ltd., Mal(Mal-vern, United Kingdom) using a refractive index increment value (dn/dc) of 0.145 mL/g for both polysaccharide stan-dards The Mark-Houwink equation was used to investigate the conformation:

where M is the molecular weight at a given point within the

mo-lecular weight distribution; [η ] is intrinsic viscosity; K is a constant, and

α is a scalar related to the conformation in solution The value of α

re-sults from the slope of the Mark-Houwink plot The data for M and [ η] were extracted from HPSEC measurements In general, the value of α is below 0.5 for spherical-like (theoretically 0 for a fully collapsed coil in a poor solvent, as predicted by the Einstein equation), between 0.5–0.8 for random coil, and larger than 0.8 for rigid rod conformation (Q Guo

et al., 2013; He, Zhang, Wang, Qu, & Sun, 2017) Mark-Houwink plots with two fractions were evaluated for the value of α using a MATLAB script computing the optimal breakpoint on a given data set for two

linear fits by minimizing the overall sum of squared errors (SSE) The SSE was calculated using:

N i=1

(

2.5 Determination of purity and yield of polysaccharides

Purity of polysaccharides and membrane rejection in terms of poly-saccharide yield was analyzed by quantifying monopoly-saccharides using high-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD) after TFA hydrolysis based on the method described by Boual, Abdellah, Aminata, Michaud, and Hadj (2012) In brief, 1 mL aqueous polysaccharide solution was incubated with 1.5 mL 3.3 M TFA at 100 ◦C for 4 h After complete evaporation under N2 gas stream at RT, the dried hydrolysate was dissolved in 10 mL water Hydrolysate solutions were filtered through a hydrophilic 0.45

μm PTFE syringe filter prior to analysis For the polysaccharide analysis,

a Dionex ICS-5000+ System (Thermo Scientific, Sunnyvale, United States) equipped with a Dionex CarboPac PA1 (4 × 250 mm) column and

a CarboPac PA1 (4 × 50 mm) guard column operating at 25 ◦C was used Injection volume of the samples was 10 μL and eluted using a combi-nation of the two mobile phases: (A) 200 mM NaOH and (B) purified water at a flow rate of 1 mL/min The applied gradient program was adapted from the method described by Rohrer, Cooper, and Townsend (1993) with slight modifications The resulting gradient program was: 0–20 min, isocratic 8% A and 92% B; 20–30 min, isocratic 100% A; and 30–39 min, isocratic 8% A and 92% B Eluents were kept under helium atmosphere Quantification of samples was performed with the internal

Fig 1 Scheme of the cross-flow diafiltration set-up Reprinted with permission

from Sartorius© (Directions for Use Vivaflow 50 | 50R | 200, 2016)

calibration method using the Chromeleon Chromatography Data System

(CDS) Version 7 (Thermo Scientific, Sunnyvale, United States)

D-Glucose at seven concentration levels between 1.25–30 mg/L was used

as external standard and D-Sorbitol as internal standard D-Glucose was

the only identified monomeric sugar for both polysaccharides in

HPEAC-PAD analysis The purity of the polysaccharides on a w/w basis

was determined as follows:

m polysaccharide

where m polysaccharide and m hydrolysate represent the masses of crude

poly-saccharide powder and polypoly-saccharide available after TFA hydrolysis

determined by HPAEC-PAD, respectively

2.6 Atomic force microscopy

Imaging of the polysaccharides was conducted by high-resolution

atomic force microscopy (AFM) For the sample preparation, 20 μL of

1 μg/mL filtered scleroglucan solution were deposited on freshly cleaved

mica, left to adsorb for 30 s and subsequently gently dried with

pres-surized air AFM height images were then obtained using a Nanoscope

VIII Multimode Scanning Force Microscope (Bruker AXS, Karlsruhe,

Germany) equipped with commercial silicon nitride cantilevers in

tap-ping mode at ambient conditions Images are presented after a 3rd order

flattening and without any further processing

2.7 Dynamic light scattering

Correlation function and zeta potential (ζ) were measured by dy-namic light scattering (DLS) using a Zetasizer Nano (Malvern Panalytical Ltd., Malvern, United Kingdom) The experiments were performed at 25

mea-surement The results were processed using the Zetasizer software

2.8 Statistical analysis

All experiments were performed at least in triplicates and the data were expressed as mean values ± standard deviation One-way analysis

of variance (ANOVA) with Tukey’s post-hoc test was performed to compare mean group values An alpha value of 0.05 was considered significant We analyzed the data using Origin, Version 2018 (OriginLab Corporation, Northampton, United States)

3 Results & discussion

3.1 Characterization of polysaccharides

Pullulan and scleroglucan standards required comprehensive

Fig 2 (A) DLS correlation functions showing the solubility of native pullulan and scleroglucan in aqueous solutions and respective purities determined with HPSEC-

RI and HPAEC-PAD (pullulan, n = 15; scleroglucan, n = 17) (B) HPSEC-RI signal overlay for scleroglucan after dissolution at varying temperature and duration The dotted arrows indicate increased signal intensity upon prolonged dissolution time (C) Representative HPSEC-LALS chromatograms for native pullulan and

scle-roglucan solutions with M w and Đ indications of the fractions observed (pullulan and scleroglucan, n = 6) (D) Representative Mark-Houwink plot for pullulan and

scleroglucan solutions The α values for the scleroglucan fractions were calculated using a MATLAB script computing the optimal breakpoint on a given data set for

two linear fits by minimizing overall SSE Simplistic illustrations of the respective conformation given by the Mark-Houwink α value and dotted lines are included for visualization purposes

characterization to ensure the molecular comparability of both

poly-saccharides and to constrain ultrafiltration separation variations

exclu-sively to the respective conformation in solution The characterization

focused on the solubility, purity, molecular weight, and the

conforma-tion in aqueous soluconforma-tion of both polysaccharides under investigaconforma-tion

(Fig 2) A fast decay in the correlation functions obtained by DLS

illustrated the complete solubility in aqueous solution for pullulan and

scleroglucan (Fig 2A) The slower decay observed for the scleroglucan

correlation curve demonstrated the larger hydrodynamic radius of

scleroglucan compared to pullulan The consistency of the

poly-saccharide purity in solution measured by HPSEC-RI along with the

polysaccharide powder purities obtained by HPAEC-PAD after TFA

hy-drolysis corroborated the complete solubility of pullulan and

scle-roglucan standards The pullulan purity of 73 ± 2% observed with

HPSEC-RI matched the purity of 72 ± 4% determined by HPAEC-PAD

after dissolving pullulan for 1 h at RT under constant stirring

(Fig 2A) This observation is in accordance with previous work on the

high solubility and stability of pullulan in aqueous solution (Adolphi &

Kulicke, 1997) Scleroglucan dissolution trials at various combinations

of temperature and incubation duration provided the optimal

dissolu-tion procedure and ensured the complete solubility as monitored by

HPSEC-RI (Fig 2B) Incubation for 24 h at 80 ◦C resulted in a purity in

solution of 32 ± 1% with HPSEC-RI, which agreed with the purity

determined of 36 ± 4% by HPAEC-PAD (Fig 2A) Hence, a dissolution

procedure of 1 h at RT for pullulan and 24 h at 80 ◦C for scleroglucan

were adopted (see Section 2.2)

HPSEC-triple detection revealed a uniform pullulan population with

a M w of 270 ± 7 kDa and moderate dispersity (Đ) of 1.52 ± 0.02,

whereas scleroglucan exhibited distinct high-M w and low-M w fractions,

with 3730 ± 60 kDa and 1510 ± 50 kDa, respectively (Fig 2C) Both

scleroglucan fractions showed uniform Đ of 1.017 ± 0.002 and 1.092 ±

0.024, respectively The Mark-Houwink plot derived from HPSEC-triple

detection analysis provides a valuable measure for polysaccharide

conformational elucidation Pullulan exhibited a random coil

confor-mation across the total polysaccharide population indicated by α =0.68

(Fig 2D) The two fractions present in scleroglucan showed two

distinctly different conformations in solution The Mark-Houwink plot

indicated a spherical conformation in the high-M w fraction and a rigid

rod-like conformation in the low-M w fraction, reflected by α =0.03 and

α =2.5, respectively Literature suggests that the low-M w fraction might

be composed of several scleroglucan strands coordinated to rigid rod-

like entities in solution (Sletmoen & Stokke, 2008; Zhang, Zhang, &

Xu, 2004) The low α value of the high-M w scleroglucan fraction

indi-cated the presence of aggregates (Q Guo et al., 2013) Yanaki and

Norisuye (1983) confirmed the presence of two fractions of scleroglucan

in aqueous solution Furthermore, their study proposed that the high-M w

fraction consists of two or more linear rigid rod entities, in line with our

observation of high-M w scleroglucan aggregates with uniform Đ

Ultra-filtration separation evaluation based on conformation requires the breakdown of aggregates and the presence of the total scleroglucan population in a rigid rod-like conformation beside an adjustment of the molecular weight

3.2 Treatment of scleroglucan solution 3.2.1 Aggregate breakdown with alkaline treatment and subsequent neutralization

Alkaline treatment with subsequent neutralization of native scle-roglucan solution (AN-sclescle-roglucan) was evaluated for its suitability to

break down high-M w aggregates and to induce a rigid rod-like confor-mation across the total scleroglucan population The successive break-down of aggregates upon increasing NaOH concentration up to 0.2 M prior to neutralization resulted in a distinct shift of the molecular weight

distribution of the total scleroglucan population towards lower M w

(Fig 3A) Treatment with 0.2 M NaOH followed by neutralization suc-cessfully induced the transition to a rigid rod-like conformation over the entire scleroglucan population Previous work showed that scleroglucan strands in rigid rod-like entities undergo a conformational transition from rigid rod-like structures to random coil at 0.1– 0.2 M NaOH induced by electrostatic repulsion owing to high ionic strength ( Slet-moen & Stokke, 2008; Zhang et al., 2004) Furthermore, the introduced charges destabilize hydrogen bonds and lead to the breakdown of ag-gregates The transition to a rigid rod-like conformation observed is consistent with the described ability of alkaline-treated and denaturated random coil scleroglucan strands to spontaneously renaturate and form rigid rod-like structures after subsequent neutralization (Sletmoen & Stokke, 2008; Zhang et al., 2004) The overlay of the Mark-Houwink plots of the alkaline treatment at various NaOH concentration after neutralization revealed increasing slopes and hence increasing α values with higher NaOH concentration (Fig.3 B) Consequently, more scle-roglucan strands were separated and structures with higher rigidity renaturated after neutralization as the NaOH concentration in the treatment increased The treatment with 0.2 M NaOH showed an increased uniformity in the resulting molecular weight distribution of

scleroglucan with high rigidity, displayed by a Đ value of 1.31 ± 0.02,

and a Mark-Houwink α of 2.03 ± 0.04, compared to treatments with 0.01 M and 0.1 M NaOH (Fig 3A, B) At concentrations equal to or lower than 0.1 M NaOH, scleroglucan exhibited distinct fractions composed of

Fig 3 (A) HPSEC-RI monitoring for molecular weight distribution and Đ alteration (n = 6) of scleroglucan after alkaline treatment for 10 min at RT with varying

NaOH concentration followed by neutralization and (B) corresponding Mark-Houwink plot illustrating conformational transitions Colored areas depict corre-sponding molecular weight fractions in panel (A) and (B) for enhanced visualization guiding

aggregates and rigid rod-like entities Interestingly, aggregate formation

first increased during treatment with low NaOH concentration, as

observed after alkaline treatment with 0.01 M NaOH and subsequent

neutralization in comparison to the untreated control (Fig.3 A) This

observation was consistent with the expected equilibrium of co-existing

aggregates and linear rigid rod-like structures in solution below 0.1–0.2

M NaOH (Sletmoen & Stokke, 2008) Ding, Jiang, Zhang, and Wu

(1998)) observed a similar phenomenon for pachyman, a (1,

3)-β-D-glucan, in aqueous NaOH solution and concluded that large

ag-gregates are formed

Alkaline treatment with 0.2 M NaOH for 10 min and subsequent

neutralization facilitated the breakdown of the majority of aggregates

and the conformational transition to rigid rod-like structures across the

entire scleroglucan population The optimal treatment conditions

enabled the preparation of AN-scleroglucan solutions with uniform

molecular weight distribution and conformation

3.2.2 Molecular weight adjustment by sonication

The assessment of the effect of polysaccharide conformation on

ul-trafiltration performance requires comparable molecular weights with

distinct conformations in solution For this purpose, the adjustment of

molecular weight by sonication and its effect on the Đ and conformation

were investigated Sonication of AN-scleroglucan (see Section 3.2.1) for

180 min resulted in a significant decrease of M w from 1860 ± 130 kDa to

387 ± 14 kDa and yielded alkaline-treated, neutralized and sonicated

scleroglucan (ANS-scleroglucan) (Fig 4A) The gradual shift of the

respective molecular weight distribution to higher retention volume

(Fig 4B), which is inversely proportional to the molecular weight,

illustrated the successful M w reduction with increasing sonication time

(Fig 4A) The uniform Đ of AN-scleroglucan solution remained

un-changed, without any significant variation, in the ANS-scleroglucan

solution, as shown by the Đ values of 1.31 ± 0.02 and 1.36 ± 0.07,

respectively (Fig 4A) Simultaneously, the value of Mark-Houwink α

decreased significantly from 2.03 ± 0.04 to 1.1 ± 0.1 during the

soni-cation process However, α values above 0.8 are ascribed to a rigid rod-

like conformation, indicating that the conformation of scleroglucan was

maintained (Q Guo et al., 2013; He et al., 2017)

The cleavage of glycosidic linkages owing to shear forces in the fluid

caused by imploding cavitation bubbles presumably governs the

sonication-driven polysaccharide degradation (Cizova, Bystricky, &

Bystricky, 2015) Consequently, aggregates formed by hydrogen bonds

are more susceptible to sonication degradation and readily disentangled

Hence, the breakdown of remaining aggregates accounted for the

considerable decrease of M w within the first 5 min of sonication

(Fig 4A) The shoulder of the elution profile around retention volume = 12.25 mL at t = 0 vanishes after 5 min sonication, thus illustrating the complete breakdown of aggregates (Fig 4B) Furthermore, the unaltered

Đ observed for ANS-scleroglucan indicates the absence of any structural selectivity of the applied treatment The change of α values might originate from an altered higher-order structure within the rigid rod-like entities owing to the polysaccharide degradation (Sletmoen & Stokke,

2008), yet with no observed effect on the resulting conformation in solution (Fig 4A) The simultaneous decrease in α values and M w

observed matches the controversially discussed presence of rigid

rod-like structure of scleroglucan below a critical M w (Li, Xu, & Zhang,

2010; Wang et al., 2017) Denaturation of scleroglucan polymer chain ends or incomplete strand breakage might hinder the sterical alignment and contribute to the abated rigidity through sonication treatment The findings on sonication-induced scleroglucan degradation ob-tained pushes further studies on the biological activities of native (1,3)-

β-D-glucans, as controlled sonication is a promising approach to reduce their viscosity (Sletmoen & Stokke, 2008) and was already successfully applied for cellulose (Arcari et al., 2020) Overall, preparation of ANS-scleroglucan enabled a controlled and reproducible molecular

weight adjustment with unaltered Đ and remaining rigid rod-like

conformation of scleroglucan in solution over the total treatment period investigated Furthermore, ANS-scleroglucan showed structural stability over the entire ultrafiltration and analysis period, facilitating a meaningful ultrafiltration evaluation and subsequent chromatographic investigation (Fig S3) For more information on the structural stability

of ANS-scleroglucan solution, see supplementary information

3.3 Visualization of scleroglucan treatment by AFM

AFM imaging was used to visualize the morphology and changes thereof during the preliminary preparation of the ANS-scleroglucan feed solution Scleroglucan in the native state covered the complete surface revealing a mesh of branched, overlapping and intertwined poly-saccharides chains with varying heights (Fig 5A), resembling observa-tions made in previous studies (McIntire & Brant, 1997; Vuppu, Garcia,

& Vernia, 1997) After the alkaline treatment and neutralization, the majority of aggregates were separated with individual aggregates still being observable, supporting the observation from HPSEC-RI of a decreasing effect on the molecular weight (Fig 5B) As fuzzy ends and branching points were observable, the interaction of multiple poly-saccharide chains was confirmed, as already reported in other linear polysaccharides such as the carrageenans and gellan gum (Diener et al.,

2019, 2020) The effect of sonication was visually confirmed as the

differences (one-way ANOVA + Tukey’s post hoc test, p < 0.05, n = 3) (B) HPSEC-RI overlay of sonicated AN-scleroglucan illustrating gradual reduction of M w with prolonged sonication time

remaining aggregated structures, and thus the detected molecular

weight, were further disintegrated (Fig 5C) The sonication treatment of

the AN-scleroglucan solution resulted in the liberation of rigid rod-like,

linear polysaccharide chains, confirming the expected conformation of

scleroglucan Similarly to the ANS-scleroglucan solution, mainly rigid

rod-like, linear polysaccharide chains were observed in the retentate

solution after ultrafiltration (Fig 5D) The rigidity of the polysaccharide

chains may explain the indifference of the ANS-scleroglucan solution

before and after application of ultrafiltration and points again at the

importance of the conformation of the polysaccharide conformation for

the assessment of a membrane The apparent alignment of the polymers

and aggregates is presumably caused by the drying step in the sample

preparation and emphasizes the rigidity of the scleroglucan polymers

(Stokke & Brant, 1990) Interestingly, a small number of the single

polysaccharides were ring-like shaped, also observed in ι-carrageenan

owing to their chiral secondary structure (Fig 5C, D) (Schefer, Usov, &

Mezzenga, 2015) In our observations, AFM imaging provided a simple

characterization pathway to explore conformational changes and verify

the effect of the pretreatments

3.4 Evaluation of pullulan and ANS-scleroglucan ultrafiltration

Ultrafiltration investigations of polysaccharides with identical

monomeric units, comparable molecular weight, and distinct

confor-mation in aqueous solution reveal insight into the effect of conforconfor-mation

on the membrane filtration process Pullulan and ANS-scleroglucan

so-lutions were subjected to ultrafiltration using Hydrosart (HY) and

pol-yethersulfone (PES) membranes with distinct molecular weight cut-offs

of 2, 5 and 10 kDa, and 3, and 10 kDa, respectively Ultrafiltration

separation performance was evaluated considering percentage

differ-ence in M w of retentate and feed solution (Δ%Mw), revealing the impact

on the molecular weight distribution, and the corresponding recovery

yield of pullulan and ANS-scleroglucan achieved with the membranes

studied The comparison of pullulan and ANS-scleroglucan for each

membrane and the separate consideration of pullulan and ANS-

scleroglucan ultrafiltration performance across all membranes studied

provides comprehensive inferences on the separation processes

observed

3.4.1 Characterization of pullulan and ANS-scleroglucan feed solutions

The scleroglucan solution pretreatment described (see Section 3.2) facilitated the elimination of potential influencing factors on the ultra-filtration separation beside the polysaccharide conformation in solution and ensured the comparability with pullulan The feed solutions of

pullulan and ANS-scleroglucan showed a M w of 271 ± 9 and 383 ± 22 kDa, respectively, fulfilling supposed molecular weight comparability

with ΔM w ≈100 kDa (Fig 6A, B) Moreover, the α values of 0.68 ± 0.05 and 1.08 ± 0.04 depict the random coil and rigid rod-like conformation

of pullulan and ANS-scleroglucan in solution, respectively The Đ of 1.59

±0.08 for pullulan and 1.34 ± 0.05 for ANS-scleroglucan reflect the

narrow to moderate Đ of the desired conformation considering the

respective polysaccharide Additionally, the intrinsic viscosity [η] of pullulan, 0.84 ± 0.27 dL/g, and ANS-scleroglucan, 1.13 ± 0.13 dL/g, corroborate the elucidated conformation (Fig 6A, B) The [η] might be considered as “inverse density“, with higher values indicating a more extended and less dense polymer in solution Hence, the low [η] value of pullulan illustrates the compact random coil conformation in compari-son to scleroglucan The high [η] of ANS-scleroglucan substantiates the extended polymer arrangement corresponding to the linear conforma-tion The ζ-potential measurement of the feed solutions reflected the neutral character of pullulan and ANS-scleroglucan and excluded any charge-induced differences among them (Fig 6A, B) Furthermore, density and viscosity values of both polysaccharide solutions were comparable to water (Table S3, Fig S4) Therefore, influences on the diafiltration process due to physical properties of the polysaccharide solutions could be excluded

3.4.2 Separation efficiency of pullulan and ANS-scleroglucan by ultrafiltration

The comparison of pullulan and ANS-scleroglucan for each mem-brane provided insight into the separation efficiency for both poly-saccharides Pullulan and ANS-scleroglucan recoveries after ultrafiltration mostly revealed no statistically significant differences, irrespective of the membrane used or MWCO selected (Fig 7), apart from the 10 kDa PES membrane The filtration process with this particular membrane resulted in a substantial difference between ANS- scleroglucan and pullulan, with a yield of 71% and a marginal recovery yield of 1%, respectively The rejection coefficients and permeability

Fig 5 Representative AFM height images of aqueous scleroglucan deposited on mica: (A) In its native state; (B) after alkaline treatment and neutralization (AN-

scleroglucan); (C) after subsequent sonication (ANS-scleroglucan); (D) retentate after ultrafiltration Colored squares highlight the location of the enlarged AFM images displayed right below Height applies to all images

values for pullulan and ANS-scleroglucan considering each membrane

were in line with the yields presented (Table S2, Fig 7) This striking

difference in remaining yield demonstrates the fundamental effect of

polysaccharide conformation on ultrafiltration separation Moreover,

the significantly higher values for Δ%Mw of ANS-scleroglucan after

ul-trafiltration with 5 kDa HY, 10 kDa HY and 3 kDa PES indicate the

conformational effect on membrane separation Interestingly, Δ%Mw of

pullulan and ANS-scleroglucan did not differ significantly for filtrations

with 2 kDa HY and 10 kDa PES membranes (Fig 7) It appears that the

10 kDa PES membrane offers potential merits to selectively separate

pullulan and scleroglucan and provides great potential for other

poly-saccharide applications with similar conformational differences

The variations observed can be ascribed to the distinct conformations

in solution The higher chain flexibility of pullulan permits a more

compact spatial alignment in solution Considering the applied

trans-membrane pressure during ultrafiltration, the hydrodynamic volume of

flexible polymers can be additionally decreased by shear-induced

deformation at the membrane interface (Fried, 1997) Ultimately,

these circumstances enhance the transport across the membrane and

reduce the pullulan yield On the contrary, ANS-scleroglucan possesses a

linear rigid rod-like conformation, which entails an increased

hydrodynamic volume relative to the molecular weight in case of a spatial consideration along the polysaccharide chain The results ob-tained are consistent with the statistical model proposed by Vinther

et al (2012), claiming that linear shapes have a lower probability of entering a membrane pore compared to spherical shapes Our observa-tions affirm the importance to consider conformation in ultrafiltration separation, since the physical separation directly relies on the hydro-dynamic volume under ultrafiltration conditions of the particles to be retained

3.4.3 Membrane performance for pullulan and ANS-scleroglucan yields

The evaluation of pullulan and ANS-scleroglucan ultrafiltration across all membranes studied permits comprehensive inferences on membrane performance for the considered polysaccharide The com-parison within a respective polysaccharide for all membranes studied showed no significant effect of membrane selection on pullulan yield, except for the aforementioned 10 kDa PES membrane (Fig 8) ANS- scleroglucan exhibited a trend of decreasing yield with increasing MWCO, although solely the difference between 3 kDa PES, showing 94% yield, and 10 kDa PES, showing 71% yield, was statistically significant Moreover, the yield of 98% achieved with 2 kDa HY was significantly

Fig 6 Characteristics of prepared (A) pullulan and (B) ANS-scleroglucan feed solutions used to investigate the conformational effect of polysaccharides on

ul-trafiltration separation (pullulan, n = 10; scleroglucan, n = 15) Simplistic illustrations of the respective conformation given by the Mark-Houwink α value are included for visualization purposes Mark-Houwink α values were derived from areas indicated by the dotted lines, corresponding to the major weight fraction of the

respective polysaccharide, and excluding software extrapolation at the border areas within the molecular weight distribution M w , Đ, and [ η] values were derived from areas within the molecular weight distribution indicated by the brackets

the pullulan and ANS-scleroglucan retentate and feed solutions (Δ%M w) and corresponding recovery yields of the respective Hydrosart (HY) and polyethersulfone (PES) membranes observed The statistical evaluation allows for comparison between pullulan and ANS- scleroglucan for each membrane studied Different letters denote significant differences

of recovery yields (oneway ANOVA + Tukey’s

post hoc test, p < 0.05, n = 3) Asterisks

indi-cate significant differences of Δ%M w (oneway

ANOVA + Tukey’s post hoc test, *p < 0.05, n =

3)

higher than the ANS-scleroglucan yield of 71% obtained with 10 kDa

PES Furthermore, the extrapolation of the yield remaining after

assuming the highest ND observed (ND = 20 for pullulan diafiltration

with 10 kDa PES, see Table S2), corroborated the trends of

poly-saccharide yields observed (Table S2, Fig 8) Pullulan ultrafiltration

revealed significant differences in Δ%Mw between distinct MWCO within

a given membrane material, except for 5 kDa HY (Fig 8) However, 2

opposite effect was observed for PES membranes Moreover, Δ%Mw of 10

kDa PES membrane exhibited a significant difference to all other

pul-lulan ultrafiltrations ANS-scleroglucan showed significant differences

in Δ%Mw between HY and PES membranes after ultrafiltration, but no

differences within the same membrane material were observed

Overall, the recovery yields revealed remarkable deviations between

the nominal MWCO and the actual M w of the respective feed solution

The MWCO represents the lowest molecular weight of a considered

molecule that is 90% rejected by the membrane (Koros et al., 1996) A

yield of 90% for pullulan and ANS-scleroglucan with PES membranes

were achieved with a 3 kDa MWCO only, which implies a 90–fold and

128–fold deviation between nominal MWCO and actual M w,

respec-tively Generally, selection of a membrane with a MWCO 3 to 6 times

smaller than the molecular weight of the molecule to be retained is

recommended in order to assure complete retention (Schwartz, 2003)

Since the M w of the prepared pullulan and ANS-scleroglucan feed

solu-tions were 27–fold and 38–fold greater than the highest MWCO chosen

(Figs 7 and 8), respectively, a recovery yield of 90% was expected for all

membranes studied These observations demonstrate that MWCOs based

on globular proteins are not applicable to polysaccharides Pullulan and

ANS-scleroglucan ultrafiltration with HY membranes achieved over

90% or insignificant lower yields irrespective of the MWCO selected

observed for pullulan contradicts the principle of MWCO rating,

indi-cating an effect of membrane material on separation performance

Higher MWCO are expected to result in the rejection of larger molecules

with a concomitant increase in M w of the retentate Pullulan

ultrafil-tration with 10 kDa PES demonstrated clearly the rejection of larger

molecules with higher MWCO corresponding to the expected effect and

to the conformational effect discussed above However, the Δ%Mw

re-sults obtained for ANS-scleroglucan, showing that differences occurred

solely among membrane materials, corroborate the considered effect of

membrane material on the molecular weight distribution

scleroglucan rejection

Globular proteins usually exhibit a narrow relation of molecular

weight and hydrodynamic volume In this case, ultrafiltration perfor-mance evaluation based on resulting yields is an adequate measure to reflect the membranes suitability for a considered application However, polysaccharide often present broader molecular weight distribution and thus specific molecular weight fractions might get lost despite a satis-factory yield Ideally, polysaccharide filtration achieves the highest yield whilst maintaining an unchanged molecular weight distribution of the desired polysaccharide The combined assessment of Δ%Mw and yield elucidates the effect of membrane selection on the molecular weight distribution and separation efficiency of the considered polysaccharide

Pullulan ultrafiltration with yields of at least 90% without adverse alteration of the molecular weight distribution was achieved with 2 kDa

HY or 3 kDa PES (Fig 8) This observation corresponds to a 136-fold and

90-fold deviation between nominal MWCO and pullulan M w, respec-tively ANS-scleroglucan ultrafiltration with yields of at least 90% and smallest alteration of molecular weight distribution was observed for 2 and 5 kDa HY membranes (Fig 8), corresponding to a 192–fold and

respectively The significant difference in ANS-scleroglucan yield after ultrafiltration with 3 kDa and 10 kDa PES was not accompanied with any significant change in Δ%Mw (Fig 8) Since ANS-scleroglucan possesses a linear rigid rod-like structure, the ability to pass the membrane may also depend on the spatial orientation Under ultrafiltration conditions with elevated transmembrane pressure, linear structures might be forced through the membrane pore irrespective of their molecular weight,

whereas spherical structures with high M w are retained Such a phe-nomenon would result in an unchanged Δ%Mw of the rigid rod-like polymer with simultaneously decreasing yield upon an increasing MWCO, as observed Interestingly, pullulan ultrafiltration with HY membranes displayed no significant differences in terms of yield, but the significant reduction in Δ%Mw with increasing MWCO indicated the loss

of higher-M w pullulan fractions (Fig 8) Furthermore, the reverse pattern of significant differences between pullulan Δ%Mw and yield with

10 kDa HY and PES reinforce the effect of membrane material on M w

alteration and recovery yield Moreover, the ultrafiltration of ANS- scleroglucan with HY membranes, where neither significant difference

in Δ%Mw nor in the yield were observed, substantiates a greater effect of membrane material rather than the nominal MWCO on resulting retentate properties In addition, the distinct observations within Pul-lulan and ANS-scleroglucan filtration emphasize the fundamental effect

of polysaccharide conformation on the resulting separation process The smallest MWCOs of HY and PES membranes provided the desired

membranes performance regarding yield and M w for pullulan In case of ANS-scleroglucan, ultrafiltration with HY membranes yielded the

the pullulan and ANS-scleroglucan retentate and feed solutions (Δ%M w) and corresponding recovery yields of the respective Hydrosart (HY) and polyethersulfone (PES) membranes observed The statistical evaluation allows for comparison within a respective polysaccharide across the membranes studied Different letters denote significant differences of recovery yields

(oneway ANOVA + Tukey’s post hoc test, p <

0.05, n = 3) Asterisks indicate significant dif-ferences of Δ%M w (oneway ANOVA + Tukey’s

post hoc test, *p < 0.05, n = 3)

highest recovery yields while maintaining the closest weight

distribu-tion to the initial feed soludistribu-tion observed, irrespective of the selected

MWCO

The impact of membrane material on ultrafiltration performance

within a considered polysaccharide is of particular interest since both

membrane materials have been extensively used in filtration of

poly-saccharides (Kothari et al., 2014; Susanto, Arafat, Janssen, & Ulbricht,

2008) Saha, Balakrishnan, and Ulbricht (2007) found that

cellulose-based membranes are more prone to fouling than PES

mem-branes when filtering a high molecular weight fraction of 130 kDa

containing arabinogalactan For this study, the monitoring of the

permeate flow obviously indicated that fouling didn`t occur during the

time of the diafiltrations, irrespective the membrane utilized (Fig S5)

Many factors possibly contribute to fouling, such as the concentration of

solutes or the interaction of solutes and membrane e.g electrostatic

in-teractions, hydrophobic inin-teractions, and hydrogen bonding Given the

diluted concentration of the polysaccharide feed solutions (0.025%

(w/v)) utilized in the diafiltration, any adverse fouling effect due to

solute concentration can be neglected In particular, charges on the

membrane and solute surface are an important factor to consider in

ultrafiltration since electrostatic interactions can influence the

separa-tion process (Hu et al., 2018) Electrostatic attraction owing to

oppo-sitely charged surfaces of membrane and solute might induce fouling,

whereas same charges suppress fouling by repulsive effects (Breite,

Went, Thomas, Prager, & Schulze, 2016) The isoelectric point (IEP) of

the PES membranes in this study is reported to be around 5.5 (Salgin,

Salgin, & Soyer, 2013) Membranes based on regenerated cellulose, such

as the HY membranes used, have IEP`s between 3–5 (Pontie, Chasseray,

Lemordant, & Laine, 1997; Pontie, Durand-Bourlier, Lemordant, &

Laine, 1998) Since pullulan and ANS-scleroglucan feed solutions were

neutral, both membrane materials are operated above their IEP, hence

exhibiting slightly negative surfaces charges during ultrafiltration

Furthermore, it could be assumed that the ionic strength of μ =0.2 in the

ANS-scleroglucan solution resulting from NaCl after alkaline treatment

and neutralization had no effect on the IEP (Salgin et al., 2013)

Consequently, potential constraints due to adverse charge-charge

in-teractions at the membrane surface, such as adsorptive effects, could be

neglected owing to the ζ-potential measurements and the membrane

IEPs reported Hence, fouling cannot explain the differences observed

between e.g 10 kDa HY and 10 kDa PES in the ultrafiltration of pullulan

However, further investigations are needed to ascertain the underlying

mechanistic cause for the observed difference between membrane

ma-terials for a given polysaccharide

The assembled data suggest that polysaccharide conformation

sub-stantially affects ultrafiltration performance when separating distinct

polysaccharide geometries Considering a respective polysaccharide

conformation, the membrane material seems to influence largely the

rejection behavior However, conformation might play a decisive role as

the separation variations in terms of membrane material differ for both

glucose-based polysaccharides Considering polysaccharide

purifica-tions, we recommend choosing the smallest MWCO applicable for a

desired application In case of pullulan, HY and PES membranes proved

to be suitable selections, whereas HY showed superior performance in

terms of yields without M w alteration for ANS-scleroglucan Based on

our results, ultrafiltration with 10 kDa PES membrane might be an asset

for the selective separation of pullulan and ANS-scleroglucan in future

studies

4 Conclusions

Polysaccharide conformation in aqueous solution showed a

remarkable effect on ultrafiltration retentates in terms of recovery yield

and molecular weight distribution Overall, the impact of

poly-saccharide conformation and membrane material, considering a distinct

conformation, were more decisive on the membrane separation than the

manufacturer’s declared MWCO Consequently, large deviations

between apparent and nominal MWCO were observed for certain membranes The conformation as crucial factor was evidenced by a higher molecular weight and yield in the retentate of rigid rod-like ANS- scleroglucan compared to randomly coiled pullulan Furthermore, the effect of spatial orientation of linear molecules on the transport across the membrane was illustrated with ANS-scleroglucan While the mo-lecular weight remained unchanged after ultrafiltration, the yield significantly decreased, indicating membrane transport irrespective of molecular weight for linear polysaccharides Eventually, Hydrosart membranes may be recommended for purification purposes of glucose- based polysaccharides with comparable conformation and molecular weight as in this study, to ensure high polysaccharide yield and smallest possible effects on the molecular weight distribution Moreover, the smallest MWCO feasible for the considered application should be cho-sen We anticipate that polyethersulfone membranes with elevated MWCO will facilitate the selective separation of pullulan and ANS- scleroglucan and offer great potential for polysaccharides with similar structural feature and conformation This work provides the empirical framework for the development of an improved membrane selection for polysaccharide filtration, paving the way to revised membrane guide-lines in general and high-performance separation of polysaccharides in particular

CRediT authorship contribution statement Severin Eder: Conceptualization, Methodology, Formal analysis,

Investigation, Writing - original draft, Writing - review & editing,

Visualization, Supervision Patrick Zueblin: Methodology, Formal analysis, Investigation, Writing - original draft Michael Diener:

Conceptualization, Formal analysis, Investigation, Writing - original

draft, Visualization Mohammad Peydayesh: Conceptualization, Writing - review & editing Samy Boulos: Conceptualization, Method-ology, Writing - review & editing Raffaele Mezzenga: Resources, Writing - review & editing Laura Nystr¨om: Resources, Writing - review

& editing, Supervision, Project administration, Funding acquisition

Declaration of Competing Interest

The authors reported no declarations of interest

Acknowledgment

The authors gratefully thank Dr Pascal Bertsch from the Laboratory

of Food Process Engineering, ETH Zürich for his assistance in the ultrasonication setup and sharing his expertise The authors acknowl-edge Dr Jo¨el Zink from the Laboratory of Food Process Engineering, ETH Zürich for supporting the viscosity and density measurements and his help This work was supported by European Research Council ERC, under the European Union’s Horizon 2020 research and innovation programme (Grant agreement No 679037), and ETH Zurich

Appendix A Supplementary data

Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.carbpol.2021.117830

References

Adolphi, U., & Kulicke, W M (1997) Coil dimensions and conformation of macromolecules in aqueous media from flow field-flow fractionation/multi-angle

laser light scattering illustrated by studies on pullulan Polymer, 38(7), 1513–1519

Arcari, M., Axelrod, R., Adamcik, J., Handschin, S., S´anchez-Ferrer, A., Mezzenga, R., & Nystr¨om, G (2020) Structure–property relationships of cellulose nanofibril hydro-

and aerogels and their building blocks Nanoscale, 12(21), 11638–11646

Atkins, E (1985) Conformations in polysaccharides and complex carbohydrates Journal

of Biosciences, 8(1–2), 375–387

Boual, Z., Abdellah, K., Aminata, K., Michaud, P., & Hadj, M (2012) Partial characterization and hydrolysis procedure of water soluble polysaccharides