Prediction of retention time of cutinases tagged with hydrophobic peptides in hydrophobic interaction chromatography.

Abstract

Hydrophobic interaction chromatography (HIC) is an important technique for protein purification, which exploits the separation of proteins based on hydrophobic interactions between the stationary phase ligands and hydrophobic regions on the protein surface. One way of enhancing the purification efficiency by HIC is the addition of short sequences of peptide tags to the target protein by genetic engineering, which could reduce the need for extra and expensive chromatographic steps. In the present work, a methodology for predicting retention times of cutinases tagged with hydrophobic peptides in HIC is presented. Cutinase from Fusarium solani pisi fused to tryptophan-proline (WP) tags, namely (WP)2 and (WP)4, and produced in Saccharomyces cerevisiae strains, were used as model proteins. From the simulations, the methodology based on tagged hydrophobic definition proposed by Simeonidis et al. (Phitagged), associated to a quadratic model for predicting dimensionless retention times, showed small differences (RMSE<0.022) between observed and estimated retention times. The difference between observed and calculated retention times being lower than 2.0% (RMSE<0.022) for the two tagged cutinases at three different stationary phases, except for the case of cut_(wp)2 in octyl sepharose-2 M ammonium sulphate. Therefore, we consider that the proposed strategy, based on tagged surface hydrophobicity, allows prediction of acceptable retention times of cutinases tagged with hydrophobic peptides in HIC.

Full text

Page 1

Short communication
Prediction of retention time of cutinases tagged with hydrophobic
peptides in hydrophobic interaction chromatography
M.E. Lienqueoa,∗, O. Salazara, K. Henriqueza, C.R.C. Caladob,c,
L.P. Fonsecab, J.M.S. Cabralb
aCentre for Biochemical Engineering and Biotechnology, University of Chile, Santiago, Chile
bCentro de Engenharia Biol´ ogica e Qu´ ımica, Instituto Superior T´ ecnico, 1049-001 Lisboa, Portugal
cFaculdade de Engenharia da Universidade Cat´ olica Portuguesa, Rio de Mouro, Portugal
Abstract
Hydrophobic interaction chromatography (HIC) is an important technique for protein purification, which exploits the separation of proteins
based on hydrophobic interactions between the stationary phase ligands and hydrophobic regions on the protein surface. One way of enhancing the
purification efficiency by HIC is the addition of short sequences of peptide tags to the target protein by genetic engineering, which could reduce
the need for extra and expensive chromatographic steps. In the present work, a methodology for predicting retention times of cutinases tagged
with hydrophobic peptides in HIC is presented. Cutinase from Fusarium solani pisi fused to tryptophan–proline (WP) tags, namely (WP)2and
(WP)4, and produced in Saccharomyces cerevisiae strains, were used as model proteins. From the simulations, the methodology based on tagged
hydrophobicdefinitionproposedbySimeonidisetal.(Φtagged),associatedtoaquadraticmodelforpredictingdimensionlessretentiontimes,showed
smalldifferences(RMSE<0.022)betweenobservedandestimatedretentiontimes.Thedifferencebetweenobservedandcalculatedretentiontimes
being lower than 2.0% (RMSE<0.022) for the two tagged cutinases at three different stationary phases, except for the case of cut (wp)2in octyl
sepharose–2M ammonium sulphate. Therefore, we consider that the proposed strategy, based on tagged surface hydrophobicity, allows prediction
of acceptable retention times of cutinases tagged with hydrophobic peptides in HIC.
Keywords: Hydrophobic interaction chromatography; Protein hydrophobicity; Retention time prediction; Hydrophobic peptide tags
1. Introduction
Hydrophobic interaction chromatography (HIC) has become
widely used as a bioseparation technique for the laboratory and
industrial-scale purification of biomolecules [1,2]. HIC exploits
the separation of proteins based on hydrophobic interactions
between the stationary phase ligands and hydrophobic regions
on the protein surface.
Sincehydrophobicinteractionscouldbeveryselective,small
differences in surface hydrophobicities between proteins can be
used as an efficient means to perform protein purification eas-
ily [3]. The hydrophobicity of a protein can be modified with
∗Correspondingauthorat:DepartmentofChemicalandBiotechnologyEngi-
neering, University of Chile, Beauchef 861, Santiago, Chile.
Tel.: +56 2 9784709; fax: +56 2 6991084.
E-mail address: mlienque@ing.uchile.cl (M.E. Lienqueo).
genetic engineering, such as site-directed mutagenesis or fusion
of hydrophobic peptide tags. Examples of short hydrophobic
tags that presented a strong effect on the relative hydrophobic-
ity of the tagged protein are (WP)2, (WP)4, T3, (TP)3, T3P2,
T4, (TP)4, T6, T6P2, T8 [4]. Of these tags, the most com-
monlyusedaretryptophane-containingtags(e.g.(WP)2,(WP)4)
[5–9]. The advantage of fusion of a tag over site-direct muta-
genesis is that the structure/function changes are minimised in
relation to the original structure/function of the native protein.
Furthermore, if necessary, the fused tag could be enzymati-
cally removed after purification. An important advantage of
hydrophobic polypeptide tags over traditional affinity tags is
the possibility of exploring simple and much less expensive
bioseparation materials.
A mathematical model that predicts the chromatographic
behaviour in HIC of the tagged-protein could be very useful.
Bydefiningtherelevantparametersthatinfluencethechromato-
graphic behaviour of the tagged-protein in relation to the native

Page 2

M.E. Lienqueo et al. / J. Chromatogr. A
one, the model can constitute a tool to design the optimal tag
for the protein purification process. In this way, it is possible to
design tags focused on improving the protein purification pro-
cess, while the possible negative effect of a random design on
thefunctionortheproductionofthetargetproteinisminimised.
The mechanism of protein binding to and elution from HIC
adsorbents has been studied with the goal to increase the recov-
ery and resolution [10–13]. Several authors have developed
models to predict the chromatographic behaviour of proteins in
HIC based on the surface hydrophobicity of proteins [14–20].
There are many ways to estimate the surface hydrophobicity of
tagged proteins [14]. Bergreen et al. [21] proposed that each
amino acid in the protein has a relative contribution to surface
hydrophobicity Φsurface, and the amino acids in the tag presents
a full exposed surface. The average surface hydrophobicity of
proteins is calculated by Eq. (1).
Φsurface=
20
?
i=1
?Saai
Sp
× φaai
?
(1)
wherei(i=1,...,20)indicatesthe20differentaminoacids,φaai
thevalueofthehydrophobicityassignedtoaminoacid“i”using
the Miyazawa-Jernigan scale [22], saaithe total exposed area of
the amino acid residue “i” in the tagged protein and spis the
total surface of the tagged protein. These values were calculated
using the Graphical Representation and Analysis of Structural
Properties (Grasp) program [23].
The second method to estimate the surface hydrophobic-
ity, designated tagged surface hydrophobicity (Φtagged), was
proposed by Simeonidis et al. [24]. It calculates the surface
hydrophobicity of the tagged protein as an average surface
hydrophobicity of the original protein (without the tag) plus
the hydrophobicity of the peptide tag. A fully exposed sur-
face of the amino acids in the tag is assumed [25]. Then, the
tagged surface hydrophobicity of proteins is calculated by Eq.
(2).
Φtagged=
20
?
i=1
?saai
?
sp
× φaai
?
+
20
?
k=1
stag aak× nk
sp+?(stag aak× nk)× φaak
?
(2)
where nkis the number of amino acids “k” in the tag and stag aak
isthefullyexposedsurfaceofaminoacid“k”inthetag;[24,25].
In cases where this is not applicable, selecting to place the pep-
tide tag on the other terminus (the N-terminus instead of the
C-terminus of the protein product or vice versa) can solve this
problem [24].
On the other hand, the methodology for predicting dimen-
sionless retention time (DRT) of single proteins [26] and
mixtures of proteins [27] in HIC, includes three steps: (i),
obtainingthe3DstructureoftheoriginalproteinsusingthePro-
tein Data Bank File (PDB); (ii), determine the average surface
hydrophobicity of the protein, Φsurface, considering that each
amino acid has a relative contribution to surface properties, i.e.,
usingEq.(1);(iii),finally,aquadraticmodelisusedtopredictthe
dimensionless retention time (DRT) of proteins using the pro-
teins average surface hydrophobicity, Φsurface. The model can
be written as follows:
DRT = AΦ2
where DRT is defined as:
surface+ BΦsurface+ C
(3)
DRT =tR− t0
tf− t0
(4)
where tR is the time corresponding to the peak of the chro-
matogram, t0 the time corresponding to the start of the salt
gradient, and tfis the time corresponding to the end of the salt
gradient. The values of A, B and C, for several stationary phases
are summarized in Table 1.
Inpreviouspublications[26,27]thismethodologywastested
andvalidatedwithanindividual,standard,andrecombinantmix-
ture of proteins with well known three-dimensional structure.
In this work, we extend the methodology for predicting dimen-
sionlessproteinretentiontimesinHICoftaggedproteins.Using
this definition, we analyse which method of hydrophobic calcu-
lation describes more adequately the tagged-protein behaviour
in hydrophobic interaction chromatography.
2. Experimental
2.1. Microorganism
The wild type (wt) cutinase, cutinase-(WP)2(fusion peptide
composedoftwotryptophanresiduesinterspersedwithtwopro-
line residues) and cutinase-(WP)4(fusion peptide composed of
fourtryptophanresiduesinterspersedwithfourprolineresidues)
producingSaccharomycescerevisiaeMM01strains(Mata,leu2-
3, ura3, gal1: URA3, MAL-8, MAL3, SUC3), containing the
expression vectors pUR7320, pUR807, and pUR806, respec-
tively, were constructed and provided by Unilever Research
Laboratory, Vlaardingen, The Netherlands within the Euro-
pean Union project: Integrated bioprocess design for large scale
production and isolation of recombinant proteins [28] (BIO4-
CT96-0435).
Table 1
Constantsandcorrelationcoefficientsforpredictingthedimensionlessretention
time in HIC (adaptation of Lienqueo et al. [27]a)
Operating conditions
ABCr2
Butyl sepharose–2M
ammonium sulphate
Octyl sepharose–2M
ammonium sulphate
Phenyl sepharose–2M
ammonium sulphate
−3.647.33
−1.170.97
−11.8111.59
−1.720.92
−26.9519.40
−2.500.97
In this adaptation DRT is equal to 1 for an extremely hydrophobic protein, in
this case the Neisserial surface protein A (NspA), which showed bigger sur-
face hydrophobicity than protein ankyrin, protein previously used as the most
hydrophobic protein.
aThe quadratic model is:
DRT = Aφ2+ Bφ + C.
(3)

Page 3

M.E. Lienqueo et al. / J. Chromatogr. A
Table 2
Comparison between predicted and observed protein retention times in HIC at different operating conditions
ProteinDRTobserveda
Prediction based on average
surface hydrophobicity, Φsurface
Φsurfaceb
Prediction based on tagged
surface hydrophobicity, Φtagged
Φtaggede
DRTcalculatedc
Deviationd(%)DRTcalculatedf
Differenced(%)
Butyl sepharose (4FF) 2M ammonium sulphate
Cut (wp)2
Cut (wp)4
0.88
0.98
0.317
0.335
0.79
0.88
10.36
10.39
0.338
0.379
0.89
1.00
2.04
1.75
Average deviation
RMSEg
10.37
0.135
1.46
0.022
Octyl sepharose (4FF) 2M ammonium sulphate
Cut (wp)2
Cut (wp)4
0.97
1.00
0.317
0.335
0.77
0.84
20.6
15.9
0.338
0.379
0.85
0.89
12.20
1.90
Average deviation
RMSEg
18.2
0.256
7.10
0.163
Phenyl sepharose (6FF, high sub) 2M ammonium sulphate
Cut (wp)2
1.00
Cut (wp)4
1.00
0.317
0.335
0.94
0.98
6.0
2.0
0.338
0.379
0.98
0.99
2.00
1.00
Average deviation
RMSEg
4.0
0.063
1.50
0.022
aObserved dimensionless retention times.
bAverage surface hydrophobicity calculated using Eq. (1).
cEstimated dimensionless retention times calculated using average surface hydrophobicity, Eq. (3) and constants of Table 1.
dDifference =|DRTobserved−DRTcalculated|
eTagged hydrophobicity calculated using Eq. (2).
fEstimated dimensionless retention times calculated by using tagged hydrophobicity, Eq. (3) and constants of Table 1.
gRMSE: root mean squared error.
|DRTobserved|
× 100.
2.2. Cultivation and cell harvesting
Cultivation,cellharvestingandactivityassaywascarriedout
as described in Calado et al. [29].
2.3. Hydrophobic interaction chromatography
HICwasperformedinaPharmaciaFastProteinLiquidChro-
matography(FPLC)system.HR5columns(50mm×5mmi.d.)
were filled with approximately 1ml sorbent (butyl sepharose
4FF, octyl sepharose 4FF and phenyl sepharose 6FF by
Amersham-Pharmacia). The experiments were performed at
room temperature, using a flow rate equal to 0.75ml/min. The
absorbanceoftheeffluentwasmonitoredwithanUVdetectorat
280nm.Five-hunderdmicroliterfractionswerecollectedandthe
enzyme activity and protein concentration was measured over
the entire chromatogram. Retention time of cutinases (RT) was
recordedanddimensionlessretentiontime(DRT)wascalculated
using Eq. (4).
Elution was obtained by a decreasing concentration gradient
of analytical-reagent grade ammonium sulphate for 10ml (i.e.
10columnvolumes).Theinitialeluentwas20mMBis–Tris,pH
7.0, plus 2M ammonium sulphate (solvent A). The final eluent
was20mMBis–Tris,pH7.0(solventB).Thegradientsteepness
used was 7.5% B/min.
2.4. Determination of the hydrophobicity of proteins
Determination of the hydrophobicity of proteins was carried
out as described in Lienqueo et al. [27].
3. Results and discussion
Theproposedmethodologywasvalidatedusingtwodifferent
mutantsofextracellularcutinases:cutinase-(WP)2andcutinase-
(WP)4inthreedifferentresins:butylsepharose–2Mammonium
sulphate, octyl sepharose–2M ammonium sulphate, and phenyl
sepharose–2M ammonium sulphate. Table 2 shows observed
dimensionless retention times; predicted dimensionless reten-
tion times calculated using the average surface hydrophobicity
and the tagged surface hydrophobicity models, and the varia-
tion between observed and calculated dimensionless retention
times for two tagged cutinases at three different stationary
phases.
It was observed that both hydrophobicity definitions result in
DRT estimations inside the 95% confidence intervals (data not
shown) in all the sorbents under study. However, the difference
from the estimated DRT using the tagged hydrophobicity des-
ignation in relation to the experimental DRT were under 2.0%
(RMSE<0.022) in all stationary phases under study, except for
the case of cut (wp)2in octyl sepharose–2M ammonium sul-
phate, where the deviation was 12.2% (RMSE<0.163). On the
other hand, when using the average surface hydrophobicity def-
inition, the difference between the experimental and calculated
DRT was between 2.0% and 20.6%; where the biggest diver-
gence was for the case of cut (wp)2 in octyl sepharose–2M
ammonium sulphate.
Since the tagged hydrophobicity definition considers pri-
marily the influence of the tag (the second term of Eq. (2)),
and based on the results obtained from the comparison of both
hydrophobicity definitions in different resins, we suggest that

Page 4

M.E. Lienqueo et al. / J. Chromatogr. A
theinteractionbetweentagged-proteinandthechromatographic
ligand occurs in or near the tagged zone.
4. Conclusions
Themethodologyforpredictingdimensionlessretentiontime
in hydrophobic interaction chromatography, using hydrophobic
descriptions(Φ)anda“quadraticmodel”wasappliedtotagged-
cutinase. In general small differences (RMSE<0.022) between
observed and estimated retention times was obtained when
applyingboththeaveragesurfacehydrophobicityandthetagged
surface hydrophobicity definitions. However, the methodology
based on tagged hydrophobic definition (Φtagged) proposed by
Simeonidisetal.[28],whichassumesthataminoacidsinthetag
haveafullyexposedsurface,provedtobemoreadequate,sinceit
presentedalowerdivergencebetweenpredictedandexperimen-
tal retention times (RMSE<0.022) for the two tagged cutinases
evaluated in three different resins. Therefore, we consider that
the proposed strategy, based on tagged surface hydrophobic-
ity, allows prediction of acceptable retention times of cutinases
tagged with hydrophobic peptides in HIC.
Acknowledgments
The authors would like to acknowledge the financial support
of Fondecyt 1030668, the Programa de Cooperaci´ on Cient´ ıfica
Internacional GRICES/CONICYT 2002-6-152 and Proyecto
Enlace ENL06/14 of the Departamento de Investigaci´ on,
Universidad de Chile. Cec´ ılia Calado acknowledges a fellow-
ship (SFRH/BPD/11626/2002) from program POCTI-Formar e
qualificar-Medida 1.1, Minist´ erio da Ciˆ encia e Tecnologia of
Portugal. The authors also wish to thank Drs. M. Egmond, A.
Fellinger, and M. Mannesse of Unilever Research Laboratory
for providing the yeast strains.
References
[1] J.A. Queiroz, C.T. Tomaz, J.M. Cabral, J. Biotechnol. 87 (2001) 143.
[2] M.M. Diogo, J.A. Queiroz, G.A. Monteiro, S.A. Martins, G.N. Ferreira,
D.M. Prazeres, Biotechnol. Bioeng. 68 (2000) 576.
[3] S. Fexby, A. Nilsson, G. Hambraeus, F. Tjerneld, L. Bulow, Biotechnol.
Prog. 20 (2004) 793.
[4] S. Fexby, L. Bulow, Trends Biotechnol. 22 (2004) 511.
[5] K. Berggren, A. Veide, P.A. Nygren, F. Tjerneld, Biotechnol. Bioeng. 62
(1999) 135.
[6] N. Bandmann, E. Collet, J. Leijen, M. Uhlen, A. Veide, P.A. Nygren, J.
Biotechnol. 79 (2000) 161.
[7] A. Rodenbrock, K. Selber, M.R. Egmond, M.R. Kula, Bioseparation 9
(2000) 269.
[8] S. Fexby, L. Bulow, Protein Expr. Purif. 25 (2002) 263.
[9] A. Collen, M. Ward, F. Tjerneld, H. Stalbrand, J. Biotechnol. 87 (2001)
179.
[10] J. Chen, Y. Sun, J. Chromatogr. A 992 (2003) 29.
[11] H. Jennissen, Meth. Mol. Biol. 305 (2005) 81.
[12] C. Machold, K. Deinhofer, R. Hahn, A. Jungbauer, J. Chromatogr. A 972
(2002) 3.
[13] R.Hahn,K.Deinhofer,C.Machold,A.Jungbauer,J.Chromatogr.B:Anal.
Technol. Biomed. Life Sci. 790 (2003) 99.
[14] M. Lienqueo, A. Mahn, Chem. Eng. Technol. 28 (2005) 1326.
[15] M. Lienqueo, A. Mahn, G. Navarro, J. Salgado, T. Perez-Acle, I. Rapaport,
J. Asenjo, J. Mol. Recogn. 19 (2006) 260.
[16] H.P. Jennissen, A. Demiroglou, J. Chromatogr. A 1109 (2006)
197.
[17] A. Mahn, G. Zapata-Torres, J.A. Asenjo, J. Chromatogr. A 1066 (2005)
81.
[18] A. Ladiwala, F. Xia, Q. Luo, C.M. Breneman, S.M. Cramer, Biotechnol.
Bioeng. 93 (2006) 836.
[19] F. Xia, D. Nagrath, S.M. Cramer, J. Chromatogr. A 1079 (2005)
229.
[20] A. Mahn, M.E. Lienqueo, J.A. Asenjo, J. Chromatogr. B (2006).
[21] K. Berggren, A. Wolf, J.A. Asenjo, B.A. Andrews, F. Tjerneld, Biochim.
Biophys. Acta 1596 (2002) 253.
[22] S. Miyazawa, R. Jernigan, Macromolecules 18 (1985) 534.
[23] A. Nicholls, K.A. Sharp, B. Honig, Proteins 11 (1991) 281.
[24] E. Simeonidis, J.M. Pinto, M.E. Lienqueo, S. Tsoka, L.G. Papageorgiou,
Biotechnol. Progr. 21 (2005) 875.
[25] C. Chothia, J. Mol. Biol. 105 (1976) 1.
[26] M.E. Lienqueo, A. Mahn, J.A. Asenjo, J. Chromatogr. A 978 (2002)
71.
[27] M.E. Lienqueo, A. Mahn, L. Vasquez, J.A. Asenjo, J. Chromatogr. A 1009
(2003) 189.
[28] C.R. Calado, M. Mannesse, M. Egmond, J.M. Cabral, L.P. Fonseca,
Biotechnol. Bioeng. 78 (2002) 692.
[29] C.R.C. Calado, M.A. Taipa, J.M.S. Cabral, L.P. Fonseca, Enzyme Microb.
Technol. 31 (2002) 161.