(Carvalho, 2014) Phenylboronate chromatography selectively separates glycoproteins

1250 © 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Biotechnol. J. 2014, 9, 1250–1258 DOI 10.1002/biot.201400170
www.biotechnology-journal.com
Biotechnology
Journal
1 Introduction
Many studies on boronic acids (BA) have focused on their
ability to interact specifically with cis-diol-containing bio-
molecules through the formation of a reversible pair of
covalent ester bonds [1–3]. This rather unique feature has
been widely explored during the last 30 years in the devel-
opment of carbohydrate sensors [4, 5], extraction of cis-
diol-containing molecules [6], labeling of proteins, and cell
surfaces [7, 8] and in the affinity chromatographic sepa-
ration of biomolecules such as glycoproteins, glycopep-
tides, nucleosides, and nucleic acids [9–14].
BA ligands can be found in two different stable con-
formations depending on the pH of the medium. Typical-
ly at neutral or acidic environments, i.e. below the BA’s pI,
the BA adopts a trigonal conformation while at alkaline
environments the ligand becomes hydroxylated and the
resulting boronate anion adopts a tetrahedral conforma-
tion (Supporting information, Fig. S1) [3]. This hydroxyla-
tion is possible since trigonal BAs exhibit a Lewis acid
behavior due to the deficient valence of the boron atom,
which has a free p orbital that enables it to coordinate
with basic molecules [15], such as hydroxyls, resulting in
a highly stable tetrahedral conformation, which is nega-
Biotech Method
Phenylboronate chromatography selectively separates
glycoproteins through the manipulation of electrostatic,
charge transfer, and cis-diol interactions
Rimenys J. Carvalho1,2, James Woo2, M. Raquel Aires-Barros1, Steven M. Cramer2 and Ana M. Azevedo1
1 Department of Bioengineering, IBB – Institute for Biotechnology and Bioengineering, Centre for Biological and Chemical
Engineering, Instituto Superior Técnico, Universidade de Lisboa, Lisbon, Portugal
2 Department of Chemical and Biological Engineering, Center for Biotechnology and Interdisciplinary Studies,
Rensselaer Polytechnic Institute, Troy, NY, USA
Phenylboronate chromatography (PBC) has been applied for several years, however details regard-
ing the mechanisms of interactions between the ligand and biomolecules are still scarce. The goal
of this work is to investigate the various chemical interactions between proteins and their ligands,
using a protein library containing both glycosylated and nonglycosylated proteins. Differences in
the adsorption of these proteins over a pH range from 4 to 9 were related to two main properties:
charge and presence of glycans. Acidic or neutral proteins were strongly adsorbed below pH 8
although the uncharged trigonal form of phenylboronate (PB) is less susceptible to forming elec-
trostatic and cis-diol interactions with proteins. The glycosylated proteins were only adsorbed
above pH 8 when the electrostatic repulsion between the boronate anion and the protein surface
was mitigated (at 200 mM NaCl). All basic proteins were highly adsorbed above pH 8 with PB also
acting as a cation-exchanger with binding occurring through electrostatic interactions. Batch
adsorption performed at acidic conditions in the presence of Lewis base showed that charge-trans-
fer interactions are critical for protein retention. This study demonstrates the multimodal interac-
tion of PBC, which can be a selective tool for separation of different classes of proteins.
Keywords: cis-Diol groups · Glycoproteins · Lewis bases · Multimodal chromatography · Phenylboronic acid chromatography
Correspondence: Dr. Ana M. Azevedo, Department of Bioengineering,
IBB – Institute for Biotechnology and Bioengineering, Centre for Biological
and Chemical Engineering, Instituto Superior Técnico, Universidade de
Lisboa, Av. Rovisco Pais, 1049-001 Lisbon, Portugal
E-mail: a.azevedo@tecnico.ulisboa.pt
Abbreviations: BA, boronic acid; CPG, control pore glass; PB, phenyl-
boronate; PBC, phenylboronate chromatography; pI, isoelectric point;
RNase A, ribonuclease A; RNase B, ribonuclease B
Biotechnology
Journal
Received 28 MAR 2014
Revised 22 JUN 2014
Accepted 07 AUG 2014
Accepted
article online 11 AUG 2014
Supporting information
available online

© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1251
www.biotecvisions.comwww.biotechnology-journal.com
Biotechnology
Journal
Biotechnol. J. 2014, 9, 1250–1258
tively charged at high pH. This simple change in BA con-
formation increases its affinity for cis-diol compounds,
resulting in slower dissociation from tetrahedral complex-
es than from trigonal complexes [2, 16]. As a result, the
tetrahedral conformation is considered the most reactive
species, although some controversy exists in the litera-
ture [17].
Phenylboronate chromatography (PBC) has been
broadly applied for glycoprotein separation and identifi-
cation since it can specifically retain these proteins via
the glycans present in their structures, especially at pH
values higher than the pKa (8.8) of the phenylboronate (PB)
ligand. PBC has been used in a clinical assay to measure
levels of glycated hemoglobin species in blood [18, 19];
other potential applications include the determination of
glycan patterns [20, 21] and the separation of complex
biological mixtures containing a glycosylated product
[9, 22, 23]. This type of chromatography has important
advantages such as low cost, high chemical stability, and
the specificity to replace established processes such as
Protein A chromatography for monoclonal antibodies
purification [9, 23] and lectin affinity chromatography for
isolation of glycoproteins for glycoproteomics [22]. As pro-
tein-based ligands, Protein A and lectins are more sus-
ceptible to chemical and thermal degradation and are
most costly to produce than PB; however, they are also
very effective due to highly specific interactions with the
target compound [24, 25].
In addition to the affinity interaction between PB and
cis-diol molecules, which is unique to PBC, the Lewis acid
character of the ligand also allows it to interact with hard
Lewis bases whenever the pH is below the ligand pKa [26].
This secondary interaction could be exploited to increase
the selectivity of cis-diol interactions or as an orthogonal
driving force for protein separations. Lewis acid–base
interactions have already been explored in chromatogra-
phy using zirconia as stationary phase. This support is a
Lewis acid that can complex with strong Lewis bases
buffers and form a negatively charged surface [27]. In
PBC, Lewis base groups are present in proteins, in the
form of negatively charged carboxylates (aspartate and
glutamate) or unprotonated amino groups (asparagine
and glutamine); both can easily interact with boron at low
pH [28]. Hydrophobic interactions via the phenyl moiety
have been shown to increase the retention of hydropho-
bic solutes [29], though these interactions were shown to
be minimal at low ionic strength. Lastly, when the ligand
is in the negatively charged tetrahedral conformation, the
hydroxyboronate anion can act as a weak cation-
exchanger [29, 30].
Ren et al. [29] used small, organic molecules to demon-
strate these various interactions (electrostatic, Lewis
acid–base, hydrophobic, and hydrogen bonding) that
occur alongside the affinity cis-diol interaction and
showed that these non-specific interactions can be mini-
mized by altering the solution conditions (i.e. increasing
ionic strength, adding organic solvents, or hydrogen
bonding co-solutes). However, with larger biomolecules
(e.g. proteins), the frequency of these non-specific inter-
actions will increase and some of the mitigation strate-
gies (organic modifiers or urea) are incompatible with pro-
teins as they will induce denaturation. Interestingly,
Ghose et al. [31] have shown that even in specific affinity
interactions such as the one between Protein A and anti-
bodies, these various non-specific interactions are also
present and are responsible for the different elution pH of
the different antibody classes.
Using a library of proteins containing glycosylated and
nonglycosylated species with isoelectric points (pI) rang-
ing from 2.5 to 9.5, the adsorption behavior was measured
over a pH range of 4–9. From these studies, we charac-
terized dominant interactions between PB and different
classes of proteins (acidic, neutral, basic, glycosylated,
and nonglycosylated) and determined which solution
conditions (pH, ionic strength, and modifiers) can mini-
mize non-specific interactions with the ligand.
2 Materials and methods
2.1 Materials
The protein library was composed by the following pro-
teins obtained from Sigma–Aldrich® (St. Louis, MO,
USA): amyloglucosidase from Aspergillus niger, carbonic
anhydrase from bovine erythrocytes, cellulase from Tri-
choderma reesei, conalbumin from chicken egg white,
pepsin from porcine stomach, ribonuclease A (RNase A)
from bovine pancreas, and ribonuclease B (RNase B) from
bovine pancreas.
2-(Cyclohexylamino)ethanesulfonic acid (CHES),
sodium phosphate mono and dibasic anhydrous, sodium
citrate anhydrous, sodium acetate anhydrous, sodium flu-
oride, tris(hydroxymethyl)aminomethane (Tris), sodium
chloride, and glycine were also purchased from Sigma–
Aldrich. The chromatographic media used in this study
were controlled porous glass (CPG®) and aminophenyl-
boronic acid functionalized CPG® (ProSep®-PB, EMD Mil-
lipore, UK) and Agarose P6XL (ProMetic Biosciences, UK).
All reagents used had a purity ≥98%, were pro analysis or
of HPLC grade. All the water used in the experiments was
deionized water.
Äkta Explorer and Purifier systems (GE Healthcare,
Uppsala, Sweden) were used to perform all column exper-
iments, both systems were operated online by the soft-
ware Unicorn 5.11. Tricorn™ empty glass columns (i.d.
5 mm; max. bed height 109 mm) were used for packing
ProSep®-PB and Agarose P6XL to a column volume of
2 mL (CV). Batch experiments were performed with Mul-
tiScreen™-HV 0.45 μm Durapore 96-well membrane
plates (EMD Millipore, USA).
Biotechnology
Journal

www.biotechnology-journal.com www.biotecvisions.com
Biotechnology
Journal
Biotechnol. J. 2014, 9, 1250–1258
2.2 Adsorption experiments
2.2.1 Column experiments
Proteins were loaded in the column in amounts between
135 and 216 μg; lower loadings were performed in order to
avoid saturation of the column as well as aggregation. The
pH range used in these studies was from 4 to 9 and the
adsorption buffers used were (i) 50 mM sodium acetate
from pH 4 to 6, (ii) 10 mM sodium phosphate from pH 7
and 8, and (iii) 20 mM CHES for pH 9, with the addition of
either 0 or 200 mM NaCl. Elution of adsorbed proteins was
triggered by changing the eluent to 300 mM Tris–HCl, pH
8.5 under a 10 CV gradient elution based on previous work
[9]. According to these authors, the addition of Tris was
found to be effective in eluting all protein from the col-
umn. After elution, the column was cleaned with 30 mM
acetic acid, pH 3 in order to verify if total recovery was
obtained. The total peak area of each protein was ana-
lyzed and compared to confirm total recovery of each
studied protein. Columns of 2 mL of either CPG®,
ProSep®-PB and P6XL were employed in all column exper-
iments.
The percentage of adsorption of each protein was
determined based on the area of the Tris–HCl elution peak
divided by the total peak area in the chromatogram. Ran-
domly selected conditions were duplicated and showed a
standard deviation of 3% or less.
2.2.2 Batch experiments
In order to investigate the dominant interactions at the
acidic/neutral pH range, batch experiments were per-
formed in which the adsorption buffer was supplemented
with different Lewis bases buffers, with the intention to
disrupt charge transfer interactions and thus identify the
dominant interactions for protein adsorption to PB.
Pepsin, amyloglucosidase, carbonic anhydrase, conalbu-
min, and cellulase were used in these studies. The Lewis
bases selected were citrate and fluoride, a hard and medi-
um strength base, respectively. Their effect was studied
at concentrations ranging from 10 to 200 mM either pre-
pared in 20 mM acetate pH 4 or 20 mM phosphate pH 6.
The adsorptions were performed in 96-well micro-
plates with membranes, 200 μL of protein solution was
added in 20 μL of ProSep®-PB and agitated for 5 h using
shaker-incubator Stat Fax™ 2200 (Bio-Rad, USA) at max-
imum velocity and at room temperature. The supernatant
was then vacuum aspirated into UV 96-well microplate
and the values of samples at UV 280 nm was obtained
with SPECTRAmax Plus 384 microplate reader (Molecular
Devices, USA). The percentage of adsorption was calcu-
lated based on Eq. (1):
(1)

P
C C
C
100i f
i
where P is the percentage of protein adsorption, Ci is the
initial concentration of protein (mg/mL), and Cf is the con-
centration of supernatant obtained after adsorption
(mg/mL). The experiments were performed in triplicate.
Experiments with RNase A were also performed using
same equipment and materials using 10–200 mM sodium
citrate as binding condition at pH 4 with and without
300 mM NaCl. The binding condition 50 mM sodium
acetate pH 4 was used as a negative control and 20 mM
CHES pH 9 as positive control. This study was also per-
formed in triplicate and the binding capacity (q) was cal-
culated as Eq. (2):
(2)
where Ci is the initial concentration of protein (mg/mL),
Cf is the concentration of protein in the supernatant
obtained after adsorption (mg/mL), Vb is the volume of
beads (mL), and Vs is the volume of sample.
2.3 Protein surface properties analysis
The 3D structure files for each protein were obtained from
Protein Data Bank (PDB) of Research Collaboratory Struc-
tural Bioinformatics (RCSB). All proteins were pretreated
to remove solvents and glycans from the structure file and
later protonated at the desired pH using the Protonate 3D
tool and Amber99 force-field found in the Molecular Oper-
ating Environment (MOE) 2010.10 software (Chemical
Computing Group, Montreal, Canada). After creating a
structure file for each pH in the study, the protein surface
properties (electrostatic potential [EP] and charged amino
acids maps) were visualized using PyMol 1.5 software
(Schrodinger, Inc., NY, USA).
3 Results and discussion
3.1 Influence of charge and glycosylation
on protein adsorption in phenylboronate
chromatography
Seven proteins were selected as standards for the protein
library among several proteins. According to their pI, pro-
teins were categorized as acidic, neutral, or basic and
each group contained both a glycosylated and nonglyco-
sylated protein (Table 1).
These proteins were selected in order to understand
the general behavior of proteins on PBC in the pH range
from 4 to 9. As described in Section 2, two chromato-
graphic resins with the same aminophenylboronic acid
functionalization were used in this study; ProSep®-PB
with a CPG matrix and Agarose P6XL with an agarose
matrix. Adsorption of nonglycosylated proteins on bare
CPG®, was also studied as a control.


q
C C
V
Vi f
b
s
Biotechnology
Journal

Biotechnology
Journal
Biotechnol. J. 2014, 9, 1250–1258
3.1.1 Adsorption profile of basic proteins
The adsorption behavior of the basic proteins RNase A
(nonglycosylated) and RNase B (glycosylated) were exam-
ined as shown in Fig. 1A and 1B. As can be observed in
the figure, the adsorption of these proteins increased with
higher pH, independent of the presence of glycans. Curi-
ously, the onset of adsorption occurred at a lower pH value
for the ProSep®-PB matrix (pH 5–6) than for the Agarose
P6XL matrix. These results suggest the existence of non-
specific interactions with the base matrix. As mentioned
in the introduction, the PB ligand can participate in sev-
eral non-specific interactions with target proteins. With-
out these interactions, the nonglycosylated protein
(RNase A) should not interact with the PB resin and the
Biotechnology
Journal
Table 1. Isoelectric point (pI), Protein Data Bank Identity (PBD ID),
and classification within the protein library used in this work
Protein pI PDB ID Classification
Pepsin 2.6 3PEP Acidic, NGlya)
Amyloglucosidase 3.6 3GLY Acidic, Glyb)
Cellulase 5.5 3QR3 Neutral, Gly
Conalbumin 6.1 1AIV Neutral, NGly
Carbonic anhydrase 6.6 1V9E Neutral, NGly
RNase A 9.6 1RBX Basic, NGly
RNase B 9.7 1RBB Basic, Gly
a) NGly = nonglycosylated.
b) Gly = glycosylated.
Figure 1. Percentage of adsorption of basic proteins (A) nonglycosylated RNase A and (B) glycosylated RNase B, acidic proteins (C) nonglycosylated
pepsin and (D) glycosylated amyloglycosidase, and neutral proteins (E) nonglycosylated carbonic anhydrase and (F) glycosylated cellulase obtained in an
Äkta Explorer system using different packed-bed columns: ProSep®-PB (straight lines, black circles ●), Prosep®-PB with 200 mM NaCl in adsorption buffer
(straight lines, gray circles ●), Agarose P6XL (dash lines, black squares Ⅲ), and bare CPG® (full area, white circles ▫). The runs were performed in dupli-
cates and all results provide deviation of 3% or less.

Biotechnology
Journal
Biotechnol. J. 2014, 9, 1250–1258
glycosylated RNase B should bind solely through the
esterification of the cis-diol groups present in its carbo-
hydrate chain. The esterification reaction equilibrium is
considerably higher at alkaline pH (pH >8) explaining the
onset of adsorption observed in Fig. 1A for the Agarose
P6XL. The earlier onset observed for the ProSep®-PB
media was probably due to electrostatic interactions with
surface silanol groups on the silica support, which have
pKa values ranging from 4.9 to 8.5, depending on their
microenvironment [32]. In addition, it is known that BA is
in equilibrium with the hydroxyboronate anion and that
the anion is also favored at alkaline pH values. Thus, PB
was also able to act as a weak cation exchange ligand at
high pH, explained why the positively charged, yet ung-
lycosylated RNase A is retained on the column at high pH.
In fact, the onset of RNase A adsorption was consistent
with the pKas of the charged moieties on the resin, namely
the matrix for ProSep®-PB and the ligand on the Agarose
P6XL resin. Further, the adsorption of the nonglycosylated
RNase A to bare CPG showed that the protein can adsorb
to charged silanol groups on the CPG matrix at low ionic
strength, suggesting that this mode of interaction is
extending the range of adsorption on the ProSep®-PB
resin. To confirm these hypotheses, adsorption experi-
ments were performed in the presence of 200 mM NaCl in
order to mitigate the proposed electrostatic interactions.
Low concentrations of NaCl (>50 mM) have been shown
to mitigate electrostatic interactions with PB [29], yet the
affinity cis-diol interactions have been shown to be NaCl
independent [9, 20]. This strategy has been applied pre-
viously to prevent non-specific interactions in the purifi-
cation of recombinant antibodies [20]. As observed in
Fig. 1A, only 10% of the nonglycosylated RNase A was
observed to bind to the resin, while the adsorption of
RNase B was minimally affected by the presence of
NaCl.
3.1.2 Adsorption profile of acidic proteins
For acidic proteins, adsorption decreased with an
increase in the pH value (Fig. 1C and 1D). Furthermore,
adsorption of the nonglycosylated pepsin (Fig. 1C) per-
sisted over a relatively wider pH range (above 80%
between pH 4 and 7), whereas adsorption of the glycosy-
lated amyloglucosidase (Fig. 1D) began to decrease
around pH 6. The onset of decreased adsorption on
Agarose P6XL was earlier than on ProSep®-PB, analogous
to the effect seen with basic proteins. Because adsorption
occurred predominantly outside the pH range of cis-diol
affinity interactions, protein adsorption must have
occurred via non-specific interactions.
Adsorption of nonglycosylated pepsin to bare CPG
was also measured but no interaction was observed. This
was expected since negatively charged acidic proteins
would be repelled from the negatively charged silanols
(pKa 4.9–8.5) on the CPG matrix. Electrostatic repulsion
can also explain the low adsorption of the glycosylated
amyloglucosidase at high pH; repulsion from both the
negatively charged silanols and the hydroxyboronate
anion would prevent the protein from making the surface
contact required to form covalent cis-diol interactions.
As shown previously with the basic proteins, 200 mM
NaCl was added to mitigate electrostatic interactions;
this resulted in the enhanced adsorption of both proteins
at alkaline pH values. The behavior of amyloglucosidase
was particularly interesting because the enhanced
adsorption at higher pH and ionic strength indicated that
electrostatic repulsion between the protein and resin
was impeding the esterification of the glycan cis-diol
groups.
In the absence of electrostatic interactions, pepsin
was found to adsorb strongly for pH <8. This was attrib-
uted to the formation of charge transfer interactions
between Lewis acid and base groups; these interactions
were more resistant than electrostatics to an increase in
ionic strength [28]. At low pH, the BA ligand is a weak
Lewis acid [15] with an incomplete set of valence elec-
trons; this enables BA to interact with Lewis bases by
donating a pair of electrons to form a covalent bond.
Because of the covalent nature of this interaction, the
two interacting groups must be closer than the typical
interaction distance for electrostatic forces; this means
that electrostatic repulsion would need to be mitigated
for charge transfer interactions to occur. Proteins con-
tain several functional groups that are Lewis bases,
including carboxylate groups (e.g. aspartate and gluta-
mate) and uncharged primary amines (e.g. N-terminus
and lysine). At pH >8, the ligand converted to the
hydroxylated tetrahedral form; this reduced its affinity
for Lewis bases as a hydroxide (OH–) ion has interacted
with the boron atom. Pepsin has many surface-exposed
carboxylate groups, which supports the hypothesis that
charge transfer interactions play a significant role in the
adsorption of this protein onto PB for pH <8. By increas-
ing the ionic strength, electrostatic repulsion was miti-
gated, which allowed these close-range interactions to
form.
3.1.3 Adsorption profile of neutral proteins
As shown in Fig. 1E and 1F, neutral proteins (carbonic
anhydrase and cellulase) were adsorbed over most of the
pH range (pH <9). Adsorption of the nonglycosylated car-
bonic anhydrase to ProSep®-PB was very strong (above
90%) between pH 5 and 8, even in the presence of 200 mM
NaCl. Protein adsorption to bare CPG at lower pH (below
the silanol pKas of 4.9 and 8.5) suggested that binding at
low pH is not governed by electrostatic interactions, yet
contributes significantly to protein retention. For the gly-
cosylated cellulase, high adsorption to ProSep®-PB was
persistent up to pH 8 and extended to pH 9 in the pres-
ence of 200 mM NaCl; this mitigated the electrostatic
repulsion between the protein (pI 5.5) and boronate anion,
which allowed esterification to occur.
Biotechnology
Journal

Biotechnology
Journal
Biotechnol. J. 2014, 9, 1250–1258
3.1.4 Comparison of CPG versus agarose media:
Silanol groups enhance binding of proteins
This library study showed that proteins adsorbed over a
wider range of pH values on ProSep®-PB than on the
Agarose P6XL matrix (Fig. 1). In addition to the contribu-
tions of electrostatic interactions referred to previously,
lowered steric hindrance and contribution of electrostat-
ic interactions from negatively charged silanol groups
could have extended the pH adsorption range on the CPG
matrix. This matrix has a larger and more controlled pore
size (1000 Å) than P6XL and the larger porosity of CPG
matrices could have increased the number of available
ligands accessible to larger solutes (e.g. proteins) as com-
pared to agarose. Bovine serum albumin (BSA; 66 kDa)
has a distribution coefficient value (Kav) of 0.55 (55% of
BSA in solution would penetrate into the pores of the
P6XL column), which demonstrated that the proteins
used in this study may not fully penetrate into the pore
space.
3.2 Batch adsorption studies: Exploiting interactions
at lower pH reveal the importance of charge
transfer interactions
The initial column adsorption studies showed electrostat-
ic and charge-transfer interactions could be as important
as cis-diol interactions in the adsorption of proteins in
PBC. While electrostatic interactions were mitigated at
high ionic strength (200 mM NaCl), charge-transfer inter-
actions resulted in the retention of nonglycosylated pro-
teins at low pH and were not mitigated with the addition
of salt. Lewis base buffers (citrate and fluoride) were used
to selectively mitigate charge-transfer interactions in
order to evaluate their contribution to protein–ligand
interactions with the acidic and neutral proteins in this
library (Table 1).
As a Lewis base and also an α-hydroxyacid, citrate is
able to form a stable cyclic complex through a charge-
transfer interaction followed by an esterification reaction
(Supporting information, Fig. S2A) [26]. Fluoride is classi-
fied as a “hard” Lewis base compound and has also been
reported to strongly interact with BAs [29, 33]. Even in
acidic conditions, fluoride readily forms a complex with
PB and will shift equilibrium toward the anionic tetrahe-
dral boronate (PB(OH)2F–) form (Supporting information,
Fig. S2B). This was observed in a previous experiment
with catechol; the small, uncharged molecule was unre-
tained on PB under acidic conditions but was able to form
a cis-diol complex in the presence of fluoride [29]. At high-
er concentrations, fluoride can also displace the remain-
ing hydroxyl groups on the boron atom to generate high-
ly fluorinated species (PB(OH)F2
– and PBF3
–) [33].
In these batch adsorption experiments (Fig. 2), citrate
was observed to reduce the adsorption of acidic proteins,
Biotechnology
Journal
Figure 2. Effect of citrate (A and B) and fluoride (C and D) at pH 4 and 6, respectively, on the percentage of adsorption of amyloglucosidase (acidic and
glycosylated) (Ⅲ); pepsin (acidic and nonglycosylated) (Ⅲ); cellulase (neutral and glycosylated) (Ⅲ); carbonic anhydrase (neutral and nonglycosylated) (Ⅲ);
and conalbumin (neutral and nonglycosylated) (Ⅲ), all obtained in batch adsorption experiments. The experiments were performed in the presence of dif-
ferent concentration of citrate and fluoride: 10, 50, 100, and 200 mM. The positive controls were performed with 10 mM acetate at pH 4 and 6. *Result
from carbonic anhydrase in 200 mM citrate pH 4 was inconclusive due to precipitation of this protein under this condition. The error bars represent the
deviation of a triplicate experiment.

Biotechnology
Journal
Biotechnol. J. 2014, 9, 1250–1258
retaining 40% or less of either pepsin or amyloglucosi-
dase, while the effect on cellulase (less acidic) was less
pronounced. Increased adsorption of the neutral carbonic
anhydrase and conalbumin proteins was observed at pH
4 with the addition of citrate; a moderate decrease in
adsorption was only observed at pH 6 with 200 mM cit-
rate.
Fluoride also reduced the adsorption of acidic proteins
and promoted the retention of neutral species. However,
this effect was weaker than that observed with citrate;
concentrations above 50 mM were required at pH 4 in
order to reduce adsorption and was less effective at pH 6
at all concentrations tested. In both cases, protein adsorp-
tion appeared to be correlated with its pI; as shown in
Fig. 3, negatively charged species (pepsin) were repulsed
in the presence of modifiers whereas neutral species that
were retained (carbonic anhydrase) displayed patches of
positive charge.
Both modifiers introduced negative charge to the
resin surface by forming Lewis acid/base complexes with
the PB ligand. Citrate has three carboxylic acid groups
with pKa values of 3.17, 4.58, and 5.94 [26]. Therefore, the
charge state of the PB–citrate complex on the resin sur-
face varied from −1 at pH 4 to −3 at pH 6. Fluoride induced
a charge state of −1 at both pH conditions. This modifier-
induced surface charge was able to explain the observed
electrostatic character of protein adsorption in the pres-
ence of fluoride and citrate. Although citrate could theo-
retically disrupt cis-diol interactions with the competing
esterification reaction, fluoride should have promoted cis-
diol interactions under these conditions. The lack of selec-
tivity for glycosylated species suggests that cis-diol inter-
actions were inhibited by electrostatic repulsion; this was
observed previously at high pH in Sections 3.1.2 and 3.1.3.
In order to confirm this surface-modification hypothe-
sis, the adsorption of RNase A was measured in a range
of citrate concentrations at pH 4. As seen in Section 3.1.1,
Biotechnology
Journal
the introduction of negative charge improved the adsorp-
tion of this protein from pH 6 to 9, yet only minimal
adsorption was observed after the addition of 200 mM
NaCl. At pH 4, around 10% of the protein is adsorbed, but
this increased to 90% adsorption in the presence of
10–200 mM citrate (Fig. 4), which confirmed the surface
modification hypothesis. However, even 300 mM NaCl
was not enough to mitigate the electrostatic interactions
with the surface-adsorbed citrate, despite the fact that
200 mM NaCl was sufficient to mitigate electrostatic
interactions at high pH. This could be attributed to the
fact that multiple charges were introduced with the cit-
rate-modification of the surface, which increased the
strength of attraction and therefore required a higher ion-
ic strength to screen the extra charge groups.
Possible hydrophobic effects were also investigated
analyzing the retention/adsorption of the different pro-
Figure 3. Surface properties of pepsin (A) and carbonic anhydrase (B) represented by charged amino acids (aa) and electrostatic potential (EP) maps at
pH 4. Both front (1) and back (2) surfaces are represented by a 180° rotation.
Figure 4. Binding capacity (q) of RNase A, at equilibrium, on Prosep®-PB
matrix under different concentrations of citrate at pH 4 in the presence
of 0 mM (Ⅲ) and 300 mM (Ⅲ) NaCl. Two control experiments were per-
formed: 50 mM acetate pH 4 was used as negative control and 20 mM
CHES pH 9 as positive control. The error bars represent the deviation
of a triplicate experiment.

Biotechnology
Journal
Biotechnol. J. 2014, 9, 1250–1258
teins from the library on a pre-packed phenyl-sepharose
column using the same binding conditions as the ones
used for PBC. No significant retention was obtained for
any protein, suggesting that hydrophobic interactions are
not responsible for increasing the retention on PBC. Fur-
thermore, the binding conditions studied in this work are
not the most favorable for inducing hydrophobicity since
a low conductivity was used. Several studies in the liter-
ature use the same strategy of low conductivity in order
to avoid this secondary interaction [9, 20, 30].
3.3 Role of electrostatic, charge transfer,
and cis-diol interactions in phenylboronate
chromatography
This work has shown that the adsorption of proteins to PB
is influenced by electrostatic and charge transfer interac-
tions that can strengthen or compete with cis-diol inter-
actions to alter the selectivity of a PBC separation. These
forces were shown to be modulated by the pH, modifiers,
and ionic strength of the surrounding medium; this can
be used to enhance the achievable resolution of this mul-
timodal ligand.
The presence of charged silanol groups on the CPG
base matrix of the ProSep®-PB resin was found to intro-
duce electrostatic interactions that changed the selectiv-
ity of the PB surface as compared to a charge-neutral
agarose base matrix. As already observed by Zhang et al.
[20], 200 mM NaCl was sufficient to eliminate electrostat-
ic interactions (attractive and repulsive) between pro-
teins and negative charges on the PB ligand and matrix.
This reduced the electrostatic repulsion of negatively
charged glycosylated proteins and allowed them to
approach the surface and form cis-diol affinity interac-
tions.
After eliminating electrostatic interactions, stronger
charge transfer interactions were also observed between
proteins and the PB ligand. These interactions were found
to form independent of ionic strength and appeared to be
responsible for the retention of several nonglycosylated
proteins. This reduced the selectivity of the material for
cis-diol groups and suggested a different modality for the
fractionation of protein mixtures on this material. The
introduction of citrate or fluoride (strong Lewis bases)
reduced the adsorption of selected proteins at low pH
(pH 4 and 6), competing for charge transfer interactions
with the PB ligand. In addition to this competition effect,
these Lewis bases introduced negative charge to the lig-
and as they adsorbed to the ligand. This re-introduces
electrostatic forces that can retain basic proteins (RNase
A) and reject acidic species (pepsin).
4 Concluding remarks
This work has provided further insight into the multiple
modes of interaction present when proteins adsorb to PB
resins and identified possibilities for altering protein
selectivity from the selectivity generated from cis-diol
interactions. Particularly, it has suggested methods
where glycoproteins captured via cis-diol esterification
could be further fractionated during elution based on elec-
trostatic (via an ionic strength gradient) or charge trans-
fer interactions (via pH or Lewis base gradients). Charge
transfer interactions can also be used to separate biomol-
ecules on PB at lower pH, potentially using Lewis base
buffer species to reversibly modify the ligand surface
properties. By rationally exploiting the full range of inter-
actions present on this ligand, novel and highly selective
windows of separation could be achieved for the fraction-
ation of complex biological mixtures and enhance the role
of PB materials in the industrial-scale manufacturing of
biopharmaceuticals, although some further studies on
capacity and ligand density optimization are needed.
R. J. Carvalho would like to acknowledge Rensselaer Poly-
technic Institute for kindly receiving him during 11
months of collaboration, especially the research group of
Professor S. M. Cramer. R. J. Carvalho also acknowledges
Siddharth Parimal and Karim A. Nakamura for helping
with experimental work and the Fundação para Ciência e
Tecnologia (FCT) and MIT-Portugal programs for Ph.D.
financial support with fellowship BD/33725/2009. The
authors would also like to acknowledge FCT for the finan-
cial support through the contract PTDC/EQU-EPR/117427/
2010.
The authors declare no financial or commercial conflict of
interest.
5 References
[1] Minaga, T., Mclick, J., Pattabiraman, N., Kun, E., Steric inhibition of
phenylboronate complex-formation of 2′-(5′-phosphoribosyl)-5′-amp.
J. Biol. Chem. 1982, 257, 1942–1945.
[2] Springsteen, G., Wang, B. H., A., detailed examination of boronic
acid-diol complexation. Tetrahedron 2002, 58, 5291–5300.
[3] Yan, J., Springsteen, G., Deeter, S., Wang, B. H., The relationship
among pK(a), pH, and binding constants in the interactions between
boronic acids and diols – it is not as simple as it appears. Tetrahe-
dron 2004, 60, 11205–11209.
[4] Kuralay, F., Sattayasamitsathit, S., Gao, W., Uygun, A. et al., Self-pro-
pelled carbohydrate-sensitive microtransporters with built-in boron-
ic acid recognition for isolating sugars and cells. J. Am. Chem. Soc.
2012, 134, 15217–15220.
[5] Guo, Z. Q., Shin, I., Yoon, J., Recognition and sensing of various
species using boronic acid derivatives. Chem. Commun. 2012, 48,
5956–5967.
Biotechnology
Journal

Biotechnology
Journal
Biotechnol. J. 2014, 9, 1250–1258
[6] Matsumoto, M., Kado, A., Shiraki, T., Kondo, K., Yoshizuka, K., Reac-
tive extraction of diols with phenyl boronic acid and trioctylmethy-
lammonium chloride as coextractants and quantitative structure–
property relationship of their extraction behaviors. J. Chem. Technol.
Biotechnol. 2009, 84, 1712–1716.
[7] Saito, S., Massie, T. L., Maeda, T., Nakazumi, H., Colyer, C. L., On-
column labeling of gram-positive bacteria with a boronic acid func-
tionalized squarylium cyanine dye for analysis by polymer-enhanced
capillary transient isotachophoresis. Anal. Chem. 2012, 84, 2452–
2458.
[8] Kielmas, M., Kijewska, M., Stefanowicz, P., Szewczuk, Z., Testing
isotopic labeling with C-13(6) glucose as a method of advanced gly-
cation sites identification. Anal. Biochem. 2012, 431, 57–65.
[9] Azevedo, A. M., Gomes, A. G., Borlido, L., Santos, I. F. et al., Capture
of human monoclonal antibodies from a clarified cell culture super-
natant by phenyl boronate chromatography. J. Mol. Recognit. 2010,
23, 569–576.
[10] Brena, B. M., Batistaviera, F., Ryden, L., Porath, J., Selective adsorp-
tion of immunoglobulins and glucosylated proteins on phenyl-
boronate agarose. J. Chromatogr. 1992, 604, 109–115.
[11] Srivastava, A., Shakya, A. K., Kumar, A., Boronate affinity chro-
matography of cells and biomacromolecules using cryogel matrices.
Enzyme Microb. Technol. 2012, 51, 373–381.
[12] Gomes, A. G., Azevedo, A. M., Aires-Barros, M. R., Prazeres, D. M.,
Clearance of host cell impurities from plasmid-containing lysates by
boronate adsorption. J. Chromatogr. A 2010, 1217, 2262–2266.
[13] Tuytten, R., Lemiere, F., Esmans, E. L., Herrebout, W. A. et al., Role
of nitrogen Lewis basicity in boronate affinity chromatography of
nucleosides. Anal. Chem. 2007, 79, 6662–6669.
[14] Li, Q., Lu, C., Li, H., Liu, Y. et al., Preparation of organic-silica hybrid
boronate affinity monolithic column for the specific capture and sep-
aration of cis-diol containing compounds. J. Chromatogr. A 2012,
1256, 114–120.
[15] Hall, D. G., Structure, properties, and preparation of boronic acid
derivatives. Overview of their reactions and applications. in: D.G.
Hall (Ed.), Boronic Acids: Preparation and Applications in Organic
Synthesis and Medicine, Wiley-VCH Verlag, Weinheim 2005,
pp. 1–99.
[16] Bromba, C., Carrie, P., Chui, J. K. W., Fyles, T. M., Phenyl boronic acid
complexes of diols and hydroxyacids. Supramol. Chem. 2009, 21,
81–88.
[17] Siegel, D., Applications of reversible covalent chemistry in analyti-
cal sample preparation. Analyst 2012, 137, 5457–5482.
[18] Wood, J. R., Kaminski, B. M., Kollman, C., Beck, R. W. et al., Accura-
cy and precision of the Axis-Shield Afinion hemoglobin A1c meas-
urement device. J. Diabetes Sci. Technol. 2012, 6, 380–386.
[19] Garlick, R. L., Mazer, J. S., Higgins, P. J., Bunn, H. F., Characteriza-
tion of glycosylated hemoglobins. Relevance to monitoring of dia-
betic control and analysis of other proteins. J. Clin. Invest. 1983, 71,
1062–1072.
[20] Zhang, B., Mathewson, S., Chen, H., Two-dimensional liquid chro-
matographic methods to examine phenylboronate interactions with
recombinant antibodies. J. Chromatogr. A 2009, 1216, 5676–5686.
[21] Li, Y., Jeppsson, J. O., Jornten-Karlsson, M., Linne Larsson, E. et al.,
Application of shielding boronate affinity chromatography in the
study of the glycation pattern of haemoglobin. J. Chromatogr. B
2002, 776, 149–160.
[22] Lu, Y., Bie, Z. J., Liu, Y. C., Liu, Z., Fine-tuning the specificity of
boronate affinity monoliths toward glycoproteins through pH
manipulation. Analyst 2013, 138, 290–298.
[23] Borlido, L., Azevedo, A. M., Roque, A. C., Aires-Barros, M. R., Poten-
tial of boronic acid functionalized magnetic particles in the adsorp-
tion of human antibodies under mammalian cell culture conditions.
J. Chromatogr. A 2011, 1218, 7821–7827.
[24] Swinnen, K., Krul, A., Van Goidsenhoven, I., Van Tichelt, N. et al.,
Performance comparison of protein A affinity resins for the purifica-
tion of monoclonal antibodies. J. Chromatogr. B 2007, 848, 97–107.
[25] Drake, R. R., Schwegler, E. E., Malik, G., Diaz, J. et al., Lectin capture
strategies combined with mass spectrometry for the discovery of
serum glycoprotein biomarkers. Mol. Cell. Proteomics 2006, 5, 1957–
1967.
[26] Bosch, L. I., Fyles, T. M., James, T. D., Binary and ternary phenyl-
boronic acid complexes with saccharides and Lewis bases. Tetra-
hedron 2004, 60, 11175–11190.
[27] Hu, Y., Yang, X. Q., Carr, P. W., Mixed-mode reversed-phase and ion-
exchange separations of cationic analytes on polybutadiene-coated
zirconia. J. Chromatogr. A 2002, 968, 17–29.
[28] Smith, M. B., March, J., March’s Advanced Organic Chemistry –
Reactions, Mechanisms, and Structure, 6th Edn., John Wiley &
Sons, 2007.
[29] Ren, L., Liu, Z., Dong, M., Ye, M., Zou, H., Synthesis and character-
ization of a new boronate affinity monolithic capillary for specific
capture of cis-diol-containing compounds. J. Chromatogr. A 2009,
1216, 4768–4774.
[30] Liu, X. C., Boronic acids as ligands for affinity chromatography. Se
Pu = Chin. J. Chromatogr./Zhongguo hua xue hui 2006, 24, 73–80.
[31] Ghose, S., Allen, M., Hubbard, B., Brooks, C., Cramer, S. M., Anti-
body variable region interactions with Protein A: Implications for the
development of generic purification processes. Biotechnol. Bioeng.
2005, 92, 665–673.
[32] Ong, S. W., Zhao, X. L., Eisenthal, K. B., Polarization of water-mole-
cules at a charged interface – 2nd harmonic studies of the silica
water interface. Chem. Phys. Lett. 1992, 191, 327–335.
[33] Janczyk, M., Adamczyk-Wozniak, A., Sporzynski, A., Wroblewski,
W., Organoboron compounds as Lewis acid receptors of fluoride ions
in polymeric membranes. Anal. Chim. Acta 2012, 733, 71–77.
Biotechnology
Journal

© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.biotechnology-journal.com
Editorial: “Biotech Methods” − bringing methods to the forefront
of biotechnology
Judy Peng
http://dx.doi.org/10.1002/biot.201400492
Commentary
Biosilica-loaded poly(ε-caprolactone) nanofibers:
A step closer to bioprinted materials with tunable properties
Georg M. Guebitz and Enrique Herrero Acero
Biotech Method
Rapid one-step purification of single-cells encapsulated in alginate
microcapsules from oil to aqueous phase using a hydrophobic filter
paper: Implications for single-cell experiments
Do-Hyun Lee, Miran Jang and Je-Kyun Park
Biotech Method
Size-selective DNA separation: Recovery spectra help determine
the sodium chloride (NaCl) and polyethylene glycol (PEG)
concentrations required
Zhangyong He, Hong Xu, Min Xiong and Hongchen Gu
Biotech Method
Phenylboronate chromatography selectively separates
glycoproteins through the manipulation of electrostatic,
charge transfer, and cis-diol interactions
Rimenys J. Carvalho, James Woo, M. Raquel Aires-Barros,
Steven M. Cramer and Ana M. Azevedo
Biotech Method
Fluorescent dye ProteoStat to detect and discriminate intracellular
amyloid-like aggregates in Escherichia coli
Susanna Navarro and Salvador Ventura
Biotech Method
Phosphorylation of silk fibroins improves the cytocompatibility
of silk fibroin derived materials: A platform for the production
of tuneable materials
Vadim Volkov, Andreia Vasconcelos, Marisa P. Sárria,
Andreia C. Gomes, Artur Cavaco-Paulo
Research Article
A functional high-content miRNA screen identifies miR-30 family
to boost recombinant protein production in CHO cells
Simon Fischer ,Theresa Buck, Andreas Wagner,
Carolin Ehrhart, Julia Giancaterino, Samuel Mang,
Matthias Schad, Sven Mathias, Armaz Aschrafi,
René Handrick and Kerstin Otte
Research Articles
Feeding strategies enhance high cell density cultivation
and protein expression in milliliter scale bioreactors
Georg Faust, Nils H. Janzen, Christoph Bendig, Lin Römer,
Klaus Kaufmann and Dirk Weuster-Botz
Research Article
Bioactive nanoparticles stimulate bone tissue formation in
bioprinted three-dimensional scaffold and human mesenchymal
stem cells
Guifang Gao, Arndt F. Schilling,Tomo Yonezawa, Jiang Wang,
Guohao Dai and Xiaofeng Cui
Research Article
Biosilica-loaded poly(e-caprolactone) nanofibers mats provide
a morphogenetically active surface scaffold for the growth and
mineralization of the osteoclast-related SaOS-2 cells
Werner E.G. Müller, Emad Tolba, Heinz C. Schröder,
Bärbel Diehl-Seifert,Thorben Link and Xiaohong Wang
Research Article
High-level conversion of L-lysine into 5-aminovalerate
that can be used for nylon 6,5 synthesis
Si Jae Park,Young Hoon Oh, Won Noh, Hye Young Kim,
Jae Ho Shin, Eun Gyo Lee, Seungwoon Lee,Yokimiko David,
Mary Grace Baylon, Bong Keun Song, Jonggeon Jegal,
Sang Yup Lee and Seung Hwan Lee
Research Article
Synergistic effect of Aspergillus niger and Trichoderma reesei
enzyme sets on the saccharification of wheat straw and sugarcane
bagasse
Joost van den Brink, Gabriela Piccolo Maitan-Alfenas,
Gen Zou, Chengshu Wang, Zhihua Zhou,Valéria Monteze
Guimarães and Ronald P. de Vries
Research Article
Solid support membrane-aerated catalytic biofilm reactor
for the continuous synthesis of (S)-styrene oxide at gram scale
Babu Halan,Thomas Letzel, Andreas Schmid and Katja
Buehler
Research Article
Metabolic modeling of spatial heterogeneity of biofilms in microbial
fuel cells reveals substrate limitations in electrical current
generation
Nadeera Jayasinghe, Ashley Franks, Kelly P. Nevin
and Radhakrishnan Mahadevan
Biotechnology Journal – list of articles published in the October 2014 issue.
This regular issue of BTJ includes the new section “Biotech Methods”, and features articles on nanofibers and biofilms.
The cover illustrates the technique of electrospinning, which is applied for the production of artificial filaments that can be
organized to mats, which has many applications such as wound coverning or for stabilization of bioprinting/3D tissue units.
It is described that silica [bio-silica] converts those mats to morphogenetically active materials. Image by Werner Mülller.
See the article by Müller et al.
Systems & Synthetic Biology ·
Nanobiotech · Medicine
ISSN 1860-6768 · BJIOAM 9 (10) 1225–1362 (2014) · Vol. 9 · October 2014
10/2014
Biotech Methods
Nanofibers
Biofilms
www.biotechnology-journal.com

(Carvalho, 2014) Phenylboronate chromatography selectively separates glycoproteins

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Similar to (Carvalho, 2014) Phenylboronate chromatography selectively separates glycoproteins

Similar to (Carvalho, 2014) Phenylboronate chromatography selectively separates glycoproteins (20)

(Carvalho, 2014) Phenylboronate chromatography selectively separates glycoproteins