new approaches to aqueous polymer systems: theory, thermodynamics and applications to biomolecular...

8/13/2019 New approaches to aqueous polymer systems: Theory, thermodynamics and applications to biomolecular separations

1/8

Pure & Appl. Chem. Vol. 67 No. 6 pp. 955-962 1995.Printed in Great Britain.Q 1995 IUPAC

New approaches to aqueous polymer systems:

Theory thermodynamics and applications tobiomolecular separations

agbY, Guan, aT.H . Lilley, aM.N. Garcia-Lisbona and bT .E .Treffry

aBiothermodynamics Laboratory, Chemistry Department, The University, Shefield,S3 7HF, .K.bDepartment of Molecular Biology and Biotechnology, The University, Sheffield,10 2UH, U.K.

Abstract A summary is presented of the main qualitative features of aqueous two-phase systems. Some aspects of the application of such systems to biomolecularseparation are discussed. A brief description of a recent theoretical approach toaqueous two-phase systems based on statistical geometrical concepts is consideredand some links are made to measurable thermodynamic properties.

Some polymer-polymer pairs or polymer-inorganic salt pairs, when in a good solvent, can form twoimmiscible homogeneous liquid If the solvent is water, we call the systems formed aqueous two-phase systems. Fig.1 illustrates these two types of aqueous two-phase systems with typical compositions.

Examples of other pairs of polymers which form aqueous two-phase systems include: Dextran+ Ficoll,polyvinylalcohol or hydroxypropylstarch,and polyethylene glycol + Ficoll,polyvinylalcohol.' Other examples ofpolymer + salt aqueous two-phasesystems are (see also ref. 2 and 3):methoxypolyethylene glycol+ potassiumphosphate, polyvinylpyrrolidone +

potassium phosphate, polyethyleneglycol+ sodium sulphate, polyethyleneglycol + sodium carbonate andpolyethylene glycol+ sodium chloride.

Special types of aqueous two-phasesystems are those containing polyoxylenedetergents.4 Th e formation of thesesystems arises from the inversetemperatureholubility behaviour of non-ionic detergents carrying polyoxyethylene

PEG60002 Dexsoo

9 3 H 2 0

PEG60007 Dex500

9 0 % H z O

19 PEG60006 phosphate

75 H

1 PEG600014 phosphate8 5 H z 0

Figure 1 An illustration of the typical phase compositions ofPEG 6000+Dex500 and PEG 6000+phosphate aqueous two-phase systems. The concentrations shown are based o n weight.

groups as the hydrophilic moiety. On raising the temperature above room temperature phase separationoccurs. In the tw o resulting phases, oneis a detergent rich coacervated phase and the other is the waterrich depleted phase. Terstappenet al. have systematically and quantitatively studied protein partitioning in

Usually polyethylene glycol (PEG) is a linear polymer and has the structural unit CH,C&O) -. Methoxypolyethyleneglycol has one end of the PEG molecule methoxylised. Dextran is composed predominantly by poly(a-1,6 -glucose), Ficoll is anonionic synthetic polymer of sucrose, polyvinyl alcohol has the structural unit CH,CHOH) -.

955


2/8

956 Y .GUAN e t a / .

such detergent-based aqueous two-phase systems, employing a series of similar polyoxyethylene detergentsand proteins of varying hydrophobicity.

Some oth er similar aqueou s two-p hase systems6 composed o f aliphatic alcohols, salts and w atere.g.,ethanol-phosphate-water, ethanol-sulphate-water, ethanol-citrate-water,1-butanol-phosphate-water, 1-propanol-phosphate-water, 2-propanol-phosphate-water,2-methyl-2-propanol-phosphate-water ave also

been reported.These aqueous two-phase systems can provide benign and non-disruptive environments for the separationand purification of biopolymers.They are potentially attractivesince, as has been shown in someexamples, they can give a meansof separation which is easy tomanipulate, reliable in scale-upand is simple and effective forprocess operation.2 A procedurefor the purification of p-interferon from tissue culturesofhuman fibroblasts7f8 has beensuccessfblly used in thepharmaceutical industry for someyears and this work has beenrecently extended to arecombinant human E. coliinterferon.9 We have recentlydeveloped a novel bioseparationprocesslOJ1 in which the roleofaqueous two-phase extraction isessential and which is shown inFig. 2.

E. coliFermentation

Supertiatan

SolidPEG' @

Fieelelltratiltration Thaw

Settling

cPhosnhate and

Solid WasteKetentate Permeate

PEG' forJ. RecyclingFreezePEG'=PEG + PEG-N (CHS)~

Drier

PenicilliiiAcylase

J.

Figure 2 The extraction process for penicillin acylase recovery from

E. coli.lo

The understanding of the phase diagram of a particular aqueous two-phase system is necessary for thescientific application of this system to biological, biochemical, biotechnological or even metallurgicalapplications. The phase diagrams shown in Fig.3 represent two types of commonly encountered aqueoustwo-phase systems.

FROM PHASE DIAGRAM TO SOLUTE PARTITIONING

As indicated earlier, one of the important features of aqueous two-phase systems is that in the twoequilibrated phases water is the dominant component. If one adds biomolecules, especiallybiomacromolecules or cell organelles, into an aqueous two-phase system, uneven partitioning of themolecules between the two phases will usually result and this is illustrated in Fig.4 Conventionally, thisphenomenon is described by the partition coefficient of the added species and this is given as :

K p = C; C:

where C and C, are the concentrations of the added solute in the top and bottom phases, respectively.The reason for the uneven distribution can be attributed to several factors such as the size of the solute,electrostatic interactions, hydrophobic (hydrophilic) interactions, biospecific affinity interactions andconformational effects in the different chemical and physical environments of the two phases. Generally,

this partition coefficient is relatedto the facto rs above by*

1995 IUPAC Pure and Appl ied Chemist ry 67 955-962


3/8

New approaches to aqueous polymer systems

In K p = In KpO +In K,, +In KMObIn KbloSpIn Ksi, +In KCod . .2)

where K, is the intrinsic partition coefficient and el, hfob,biosp, size, conf represent the contributions to the partitioncoefficient from electrostatic, hydrophobic, biospecific, size andconformational effects.

The partition coefficients of biomacromolecules show a widevariation and it is this variation which is the foundation forusing tw o phase systems for the separation of biologically activemolecules or cell organelles.

If small amounts of proteins or enzymes are added to anaqueous two-phase system, usually the partition coefficient isindependent on the tota l concentration of this component. Theyield, Y,, of the target product, if it is enriched in the top phase,is given by

where C, and C, are the enzyme activities or proteinconcentrations in the to p and in the bottom phases respectively,R is the volume ratio of the top and bottom phases.

In practical situations, the systems are complex and biomass isusually present in aqueous two-phase systems. We observedfor many cases that cell mass (cell debris) is collected at theinterface between the phases. The nature of this aggregation issimilar to precipitation rather than the formation of anotherliquid phase. Therefore, aqueous two-phase systemscharacteristically both separate the target protein fromimpurities through equilibrium partitioning and coagulate celland cell debris at the interface of the two liquid phases.

f 6

O

Dertran D48 (w/w)

957

o 10 2 ophosphate Yo w/w)

Figure 3 Phase diagrams of aqueoustwo-phase systems: (A) PEG 6000-Dextran D48 at 20C (adapted from ref.2); (B) PEG 4000 + tripotassiumphosphate in the absence and in thepresence of E. coli cell mass: ...without cell mass, A ... with cell mass(adapted from ref.10).- -

Interestingly, there are two extreme cases:i) the topphase is so small that the interface appears to be thetop phase, (ii) the bottom phase isso small that theinterface appears to be the bottom phase

To illustrate these effects we consider some

experiments on PEG+ phosphate aqueous two-phasesystems containingE. coli cell mass. Tw o broad typesof observation are seen: firstly, the top phase looksclear but the bottom phase is turbid (referred to as I),and secondly, both top and bottom phases looktransparent but cell mass is coagulated at the interface(referred to a s 11). T able1 shows the influence of PEGconcentration on the transition between these twophenomenon. High concen trations of PE G andlor highof cell mass loadings always tend to give the lattersituation (case 11) Although we have categorised the

observations a s being distinct and different, the main difference between the above tw o phenomenon is thatin the former the bot tom ph ase volume isso small that it becomes negligible.

Figure 4 Distribution of a homogeneous proteinspecies in an aqueous two-pha se system

1995 IUPAC Pure and Applied Chemistry 67 955-962


4/8

958 Y . GUAN et al.

For systems where cell mass is present, the loss of some of enzyme activity by enclosure in the cellorganelles is unavoidable. Th e enzyme yield in, for example, the top phasec j eqn. 2) changes to:

YT= = 100 (4)where M T is the total amount o f enzyme added to the entire partitioning system.

Table 1 The influence of PEG concentration and cell mass load on P EG+ potassium phosphateaqueous two-phase systemst

Cell mass YO)

10 5 20 25EG 4000 ( w/w)

10 I I I I I12 I I I I I1

14 I I I I1 I116 I I I1 I I1

22 I I1 I1 I1 I1

or cell

tPotassium phosphate concentration was keptat 20 (w/w) nd pH at 7.1. See text for the meaningof categoriesI and I1

The industrial applications o f these systems need a fundamental understanding of the factors involved in theformation of two-phase systems and of the features which determine the partitioning behaviour ofbiomolecules. It has been indicated (eqn.2) that the partition of solutes, or indeed cells, between the twoaqueous phases is affected by many factors such as hydrophobicity and charge, and to date, optimisation inseparation processes has largely been obtained by empirical manipulation of such properties. This is

generally unsatisfactory and inefficient and the object of our work has been to study, from a theoreticalviewpoint, the behaviour of aqueous two-phase systems. The ultimate objective is to obtain methods for therational design of commercial purification and separation processes which use such systems.

SOM E NEW APPROACHES IN DEALING WITH AQUEOUS TWO-PHASE SYSTE MS

If one is to have a predictive method which allows the rational design of commercial purification andseparation processes there are several problems which must be addressed and these include the following.(i) The establishment of the molecular criteria necessary to describe the coexistence curves (binodals) of thetwo phases in the absence of biomaterials. (ii) The establishment of the tie-line relationships in the systems.(iii) The prediction of how biomolecules partition when present at concentrations such that no significantperturbation of the phase behaviour occurs. (iv) The prediction of the behaviour of systems when the targetspecies and/or impurities are present at high concentrations. Superimposed upon these are problems onissues associated with polymer modification, scale-up and multi-stage partitioning. Some of the outstandingaspects are as follows:

Coexistence curvesThere have been several attempts to use the Flory-Huggins theoryof polymer solutions to describe bothcoexistence curves and tie-line relationships. Such approaches are doomed at the outset, at least if one isaiming for any real molecular understanding of the phenomena involved, since some of the fundamentalassumptions made in the formulation by Flory and Huggins'* are transparently not applicable to systemscontaining water. Consequently, we have developed a new approach, which we have termed the Binodal

Model , to polymer-containing systems which is based on the concepts of statistical geometry. Its wideadaptability has also been illu ~t rat ed .1 ~n many cases, the Binodal Model14 takes the form

1995 IUPAC Pure and Appl ied Chemistry 67 955-962


5/8

New approaches to aqueous polymer systems 959

where w , and w2 are the w eight concentrations of two phase-forming species along the binodal respectively,

(V)210 is the effective excluded volume (EEV) of species2 in the species 1 aque ous solution, p is the

solution density, N , is Avogadros number, for polymer molecules M ) I and M ) , are mean molecularweights for species 1 and 2, and usually the root-mean-square average molecular weights for polydispersecomponents are taken.

This novel theory is an adva nce towards the solution of phase separation problems and allow s the predictionof the coexistence behaviour of two-phase systems using only one parameter, the EEV, and the physicalmeaning of this parameter has been demonstrated.A historical outline of efforts in describing binodals hasbeen given elsewhere.13

It is appropriate to mention here that this theory also has potential applications to many other problems inpolymer and colloid science and we are currently extending this treatment t o the important and outstanding

problems of the partitioning of soluble particulates and the precipitation o f macromolecules by polymers orsalts. In practice, w e are aw are that a correct understanding of phase separation and partition in partlyaqueous systems is importantto those who wish to avoid phase separation. This situation often occurs infood technology, surfactant and oil recovery industries. The theory is therefore expected to find moreapplications to many other problems in polymer and colloid science.

Tie-line relationshipsThe Binodal Model, although able to predict coexistence curves, cannot be applied to tie-line relationships,i.e. to linking the com positions of the phases in equilibrium. Most of the approaches which have been usedto obtain relationships betwe en phase compositions have invoked, in one form o r another, virial expansionsin solute concentrations. Although such expansions have been frequently used, we have recently shown15that there are some fimdamental problems in their use to describe aqueous two-phase systems. Briefly, the

situation is that there is a conflict between the virial expansion formulation and molecular events in that, atthe solute concentrations necessary to give two phases, third and higher order virial coefficients are requiredto represent system properties. How ever, when these coefficients are given their proper physical meaning,there is always a hndamental inconsistency. A means of circumventing this difficulty by, e.g., theintroduction of w hat w e have termed effective second virial coefficients has been suggested.15

Protein partitioning at low concentrationsStrictly speaking, the protein partitioning coefficient in eqn. l a will be a constant only at very low proteinconcentration in the system,i.e.

Kp = lim C,/C, = constantCp+O

In recent years this has been on e of the most popular subjects in this area and some progress has been madeusing both nonelectrostatic and electrostatic approaches.16J7J8 This workis summarised in a recent reviewarticle.19

The combination of these two types of molecular interactions were firstly realised by Albertssonz and hegave the following expression

In Kp = In K i + FZpAy/RT 6 )where Kp and K; are the protein partition coefficients at a given pH and at the protein isoelectric point,respectively, Ay is the electrostatic potential difference between the tw o phases andZ is the charge on theprotein.

1995 IUPAC Pure and Applied Chem istry 67 955-962


6/8

96 Y. GUAN e t a l .

Equation 6 shows that protein partitioning can be represented as two factors, one from electrostaticinteractions and the othe r from short range interactions. In other words, it is possible to deal with shortrange (van der Waals) and long range (electrostatic) molecular interactions separately. W e will also developa protein partitioning model which is consistent with the concepts used in developing our Binodal Model.20

Protein partitioning at high solute and/or high cell mass concentrations

If the experimental conditions are such that the phase system is perturbed significantly by the addedmaterials, the problems will be more complicated. There are three possible scenarios:

(i) the soluble target protein could be present at very high concentrations.(ii) the target protein could be present at low concentrations but there is also present a high concentration of

(iii) a large amount o f insoluble cell and/or cell debris could be present.

In the first and second situations, rather than having a system with two phase-forming components, ineffect, such systems now consist of at least three phase-forming components.As a consequence, thebinodal will shift from its original position, but the tie-line slope will be little changed. In the third situation,most of the cell mass will be precipitated or will coagulated at the interface of the two phases and thebinodal will be less influenced, but the total phase composition will alter due to the removal of some bulkwater by the cell mass. The comm on feature in all three cases will be the alteration of the initial tie-linelength. There is little coherent information available on any of the above aspects, but such information isnecessary to complement the theoretical studies. Comprehensive studies of all three areas need to beundertaken.

Affinity P artitioningSome usefil results on affinity partitioning were obtained from studies of metal affinity extraction in bothpolymer+ polymer21 and polymer+ salt aqueous two-phase systems.22

Theories of affinity partitioning stemmed from the original work of Flanagan and B a r o n d e ~ . ~ ~or proteinswith n ndependent binding sites and very high ligand concentration, they derived

soluble impurities.

where K and K,, are the protein partition coefficient in the presence and absence of polymer ligand, KL isthe partition coefficient of the polymer-bound ligand, KaTand KaBare the binding constants for th e protein-ligand complex in the to p and b ottom phases respectively.

Later Brookset obtained the m ore general expression where the limitation to the ligand concentrationis removed and this is:

Equation 8 was also obtained by Cordeset a ZS nterestingly, as was observed by Flanagan and Barondes23

a linear relationship exists between ln(K,/K ) and L The following expression

derived by Johansson but show n by Cordeset a1.25 fits the experimental results very well. We also found, inour study of the recognition o f affinity interactions, that this relationship holds.

Equation 8 has been made mo re generalz6 and applied to the situation where on e has a protein withm

binding sites for metal ion association or hydrogen ion association, the association constants being Kb and

1995 IUPAC Pure and Applied Chemistry 67 955-962


7/8

New approaches to aqueous polymer systems 961

KH ,b espectively, andn sites with corresponding constants Ka and K H ,a . The following expression wasderived:

where the subscripts T andB refer to top and bottom phases respectively.

This model takes into accoun t inhibition of ligand binding by hydrogen ion and the possibility that m ore thanone type of ligand site con tributes t o the affinity partitioning.

When one comes to consider the incorporation of chemical finctionality on phase-forming polymersi. .affinity partitioning), eqns.7 10 show that the influence of affinity interactions can be superimposed on thepartition coefficient in a simple way. This amalgamation could be very useful in practical situations.

Thermodynamic measurementsWe are cu rrently investigating thermodynamically based measurements of the pertinent molecularinteractions to provide independently measurable model parameters. Since the EEV is related to

thermodynamic variables by:4

where is the difference between the Helmholtz free energy necessary to create a cavity with volume

and that o f the mean value

The determination of EEV's and their variation with composition, temperature and with other measurablethermodynamic variables is an important part of our studies. We are determining the solubility of a range ofproteins under a varietyof conditions (pH and solution concentrations) in both binary (water one polymer

or salt) and ternary (water+ two polymers and/or salt) in one phase regions This will give some measuresof the interplay between th e tendencies proteins have to self-interact and their propensities t o be solvated bythe solvent comp onents In essence, the problems we are addressing experimentally and how these arerelated to ourpredictive models areillustrated in Fig. naqueous two-phasesystem for proteinseparation in abiotechnologicalprocess contains at

least four components,but molecular

interactions between Figure 5 A schematic experimental strategy for the prediction of proteincomponents can be extraction in aqueous two-phase systems, showing that pertinentobtained by studying experimental information for aqueous two-ph ase systems is obtained fromsome three-componentand two-componentsystems. Links couldbe investigated (once a reasonable body of information is obtained) between the volumetric properties of thesolvent systems investigated and the Effective Excluded Volume Some link must exist since the EEV is a

function of the spacing among the molecular centres of the neighbouring species in the system, but wedonot know what this is at the moment However, it would be very usef il in practiceto establish the link

polymerl + polymer2 + water + protein (two phases)

information on polymer1 + water I Ipolymer2 + water I protein + water

simpler systems containing only someof the components

1995 IUPAC Pure andA ppl ied Chemistry 67 955-962


8/8

962 Y . GUAN e t a / .

between the EEV and solute volumetric properties since the latter are easily measured and the formerdetermines the coexistence curve in two-phase systems.

We are also studying the volum etric propertiesof mixed solutions containing proteins and exploiting linksbetween changes in vo lumetric properties and preferential solvation.

Acknowledgement

We thank the financial support from Science and Engineering Research Council under grantNo.GWGl9824.

1.2.3 .

4.5.6.7.8.9.

10.11.12.13.

14.15.16.17.18.19.20.21.22.23.24.

25.26.

REFERENCES

M.W . Beijerinck,Kolloid. Z. 7, 12-16 (1910).P.-A, Albertsson,Partition of Cell Particles andMacromolecules, 3rd ed., Wiley, New York, 1986.S.M. Snyder, K.D . Cole and D.C . Szlag,J . Chem. Eng. Data 37, 268-274 (1992).

C. Bordier, J . Biol. Chem.256, 1604-1607 (1981).G.C. T erstappen, R .A . Ramelmeier and M.-R. Kula,J Biotechnol. 28, 263-275 (1993).A. G reve and M .-R. Kula,Fluidphase Equilibria 62, 53-63 (1991).U. Menge, M . Morr, M .-R . Kula and K. Anastassiadis, German patent DE2943016 (1981).M. M orr and M.- R. Kula, German patent DE2935134Al (1981).Y. Guan, T .H . Lilley, T.E . Treffry, C.-L. Zhou and P .B. Wilkinson, submitted toEnzyme Microb.Technol.Y. Guan, T .E. T reffiy and T.H . Lilley,Biosepuration 4, 89-99 (1994 ).Y. Guan, X .-Y . Wu, T .E. Treffiy and T .H. Lilley,Biotechnol. Bioeng. 40, 517-524 (1992).P.J. Flory, Principles of Polymer Chemistry,Cornell University Press, Ithaca, New York(1953).Y. Guan, T.E . Treffiy and T .H . Lilley,J. Chromatoq. A 668, 3 1-45 (1994).

Y. G uan, T .H . Lilley and T.E. Treffry,Macromolecules25,3971-3979 (1993).Y. Guan, T .H . Lilley and T.E. Treffiy,J. Chem. Soc ., Furuday Trans. 89,428 3-4298 (1993).C.A. H aynes, F.J . Benitez, H.W . Blanch and J.M . Prausnitz,AIChEJ. 39, 1539-1557 (1993).P. Neogi, J . Colloidinterfcrce Sci. 159, 261-274, (1993).C.L . DeLigny and W .J. Gelsema,Sep. Sci. Technol. 17, 375-380 (1982).H . Walter, G . Johansson and D .E . BrooksAnal. Biochem. 197, 1-18 (1991).Y. Guan , T .H . Lilley and T .E. Treffry, results to be published.M E . Van D am, G .E . Wuenschell and F.H. Arnold,Biotechnol. Appl. B iochem.11,492 -502 (1989)S.D . Plunkett and F.H . Arnold,Biotechnol. Techniq. 4,45 -48 (1990).S.D . Flanagan and S.H . Barondes,J Biol. Chem.,250, 1484-1489 (1975).D.E . Brooks, K.A . Sharp andD. Fisher, inPartitioning in Aqueous Two-Phase Systems: Theory,Methods, Uses and Applications to Biotechnology,eds. H. Walter, D.E Brooks and D . Fisher,Academic, London, 1985; ch. 2.A. Cordes, J Flossdorfand M.-R. Kula,Biotechnol. Bioeng. 30, 5 14-520 (1985).S.-S. Suh and F. -H . Arnold,Biotechnol. Bioeng. 35, 682-690 (1990).

1995 IUPAC Pure and App lied Chemistry 67 955-962

new approaches to aqueous polymer systems: theory, thermodynamics and applications to biomolecular...

Documents