Synthesis and characterization of fluorescent displacers for online monitoring of displacement chromatography

Article abstract of DOI:10.1021/ja806279x

One of the major impediments to the implementation of displacement chromatography for the purification of biomolecules is the need to collect fractions from the column effluent for time-consuming offline analysis. The ability to employ direct online monit

Full text of DOI:10.1021/ja806279x

Published on Web 11/17/2008
Synthesis and Characterization of Fluorescent Displacers for
Online Monitoring of Displacement Chromatography
Christopher J. Morrison,† Sun Kyu Park,‡ Chester Simocko,‡ Scott A. McCallum,§
Steven M. Cramer,*,† and J. A. Moore‡
Departments of Chemical and Biological Engineering and of Biology, Center for Biotechnology
and Interdisciplinary Studies, and Department of Chemistry and Chemical Biology,
Rensselaer Polytechnic Institute, 110 Eighth Street, Troy, New York 12180
Received August 8, 2008; E-mail: crames@rpi.edu
Abstract: One of the major impediments to the implementation of displacement chromatography for the
purification of biomolecules is the need to collect fractions from the column effluent for time-consuming
offline analysis. The ability to employ direct online monitoring of displacement chromatography would have
significant implications for applications ranging from analytical to preparative bioseparations. To this end,
a set of novel fluorescent displacers were rationally designed using known chemically selective displacers
as a template. Fluorescent cores were functionalized with different charge moieties, creating a homologous
library of displacers. These compounds were then tested on two protein pairs, R-chymotrypsinogen
A/ribonuclease A and cytochrome c/lysozyme, using batch and column displacement experiments. Of the
synthesized displacers, two were found to be highly selective while one was determined to be a high-
affinity displacer. Column displacements using one of the selective displacers yielded complete separation
of both protein pairs while facilitating direct online detection using UV and fluorescence detection. Saturation
transfer difference NMR was also carried out to investigate the binding of the fluorescent displacers to
proteins. The results indicated a selective binding between the selective displacers and R-chymotrypsinogen
A, while no binding was observed for ribonuclease A, confirming that protein-displacer binding is responsible
for the selectivity in these systems. This work demonstrates the utility of fluorescent displacers to enable
online monitoring of displacer breakthroughs while also acting as efficient displacers for protein purification.
technique for the purification of biomolecules.8,9,17 While
previous work has employed high molecular weight displacers
Introduction
Displacement chromatography is a powerful separation
technique that enables the simultaneous concentration and
purification of biomolecules from complex mixtures in a single
step.1 Displacement chromatography has been successfully
employed for the purification of proteins using hydroxyapatite,2,3
hydrophobic interaction,4 and ion exchange chromatographic
systems.5-16 In particular, ion exchange displacement chroma-
tography has attracted significant attention as a powerful
for protein purification in ion exchange systems,2,3,10,18-21 our
laboratory has been actively involved in the development of
low molecular mass displacers.8,13-17,22
To add an orthogonal degree of separation to displacement
chromatography, we have recently developed a modified
(11) Jayaraman, G.; Li, Y. F.; Moore, J. A.; Cramer, S. M. J. Chromatogr.,
A 1995, 702 (1-2), 143–155.
(12) Barnthouse, K. A.; Trompeter, W.; Jones, R.; Inampudi, P.; Rupp,
R.; Cramer, S. M. J. Biotechnol. 1998, 66 (2-3), 125–136.
(13) Shukla, A. A.; Deshmukh, R. R.; Moore, J. A.; Cramer, S. M.
Biotechnol. Prog. 2000, 16 (6), 1064–1070.
† Department of Chemical and Biological Engineering, Center for
Biotechnology and Interdisciplinary Studies.
‡ Department of Chemistry and Chemical Biology.
§ Department of Biology, Center for Biotechnology and Interdisciplinary
Studies.
(14) Tugcu, N.; Deshmukh, R. R.; Sanghvi, Y. S.; Moore, J. A.; Cramer,
S. M. J. Chromatogr., A 2001, 923 (1-2), 65–73.
(15) Rege, K.; Ladiwala, A.; Cramer, S. M. Anal. Chem. 2005, 77, 6818–
27.
(16) Rege, K.; Hu, S.; Moore, J. A.; Dordick, J. S.; Cramer, S. M. J. Am.
Chem. Soc. 2004, 126 (39), 12306–12315.
(1) Frenz, J.; Horvath, C. High-Perform. Liq. Chromatogr. 1988, 5, 211–
314.
(2) Vogt, S.; Freitag, R. J. Chromatogr., A 1997, 760 (1), 125–137.
(3) Freitag, R.; Breier, J. J. Chromatogr., A 1995, 691 (1), 101–112.
(4) Shukla, A. A.; Sunasara, K. M.; Rupp, R. G.; Cramer, S. M. Biotechnol.
Bioeng. 2000, 68 (6), 672–680.
(17) Tugcu, N.; Deshmukh, R. R.; Sanghvi, Y. S.; Cramer, S. M. React.
Funct. Polym. 2003, 54 (1-3), 37–47.
(5) Barnthouse, K. A.; Trompeter, W.; Jones, R.; Inampudi, P.; Rupp,
R.; Cramer, S. M. J. Biotechnol. 1998, 66 (2-3), 125–136.
(6) Gerstner, J. A.; Morris, J.; Hunt, T.; Hamilton, R.; Afeyan, N. B.
J. Chromatogr., A 1995, 695 (2), 195–204.
(18) Antia, F. D.; Fellegvari, I.; Horvath, C. Ind. Eng. Chem. Res. 1995,
34 (8), 2796–804.
(19) Frenz, J.; Horvath, C. High-Perform. Liq. Chromatogr. 1998, 5, 211–
314.
(7) Shukla, A. A.; Hopfer, R. L.; Chakravarti, D. N.; Bortell, E.; Cramer,
S. M. Biotechnol. Prog. 1998, 14 (1), 92–101.
(20) Vogt, S.; Freitag, R. Biotechnol. Prog. 1998, 14 (5), 742–748.
(21) Schmidt, B.; Wandrey, C.; Freitag, R. J. Chromatogr., A 2002, 944
(1-2), 149–159.
(8) Kundu, A.; Cramer, S. M. Anal. Biochem. 1997, 248 (1), 111–116.
(9) Jayaraman, G.; Gadam, S. D.; Cramer, S. M. J. Chromatogr., A 1993,
630 (1-2), 53–68.
(22) Gallant, S. R.; Cramer, S. M. J. Chromatogr., A 1997, 771 (1-2),
9–22.
(10) Peterson, E. A.; Torres, A. R. Anal. Biochem. 1983, 130 (1), 271–
282.
(23) Tugcu, N.; Ladiwala, A.; Breneman, C. M.; Cramer, S. M. Anal. Chem.
2003, 75, 5806–16.
9
10.1021/ja806279x CCC: $40.75
2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 17029–17037 17029

A R T I C L E S
Morrison et al.
technique, namely, chemically selective displacement.24,25 In
this technique, certain biomolecules are selectively “retained”
behind the displacer front. In a recent study, saturation transfer
difference (STD) NMR26 was employed to investigate binding
events between chemically selective displacers and proteins.27
This technique takes advantage of intraprotein spin diffusion,
intermolecular dipolar coupling mechanisms, and ligand ex-
change between the bound and free states to characterize binding
interactions. In that study it was directly verified that chemically
selective displacement occurs by a selective binding between
the displacer and protein. From this mechanism selective
displacers are able to be designed rationally on the basis of the
combination of a protein binding affinity and a resin binding
affinity.
A robotic high-throughput screening has also been carried
out to identify and characterize a large set of cationic chemically
selective displacers.28 From that study several compounds were
identified as being both selective and exclusive in separating
multiple protein mixtures. Trends in chemical moieties of
selective displacers from this study indicated that selective
displacers contained distinct hydrophobic regions, indicating that
protein-displacer binding via hydrophobic interactions may be
playing a role in these systems.
One of the major impediments to the implementation of
displacement chromatography is the need to collect fractions
from the column effluent for time-consuming offline analysis.
The ability to employ direct online monitoring of displacement
chromatography would have significant implications for ap-
plications ranging from analytical to preparative protein separa-
tions. In this work we use the known mechanism and structures
of chemically selective displacers to design different fluorescent,
chemically selective displacers to enable direct, online monitor-
ing of the displacement separation. Chosen fluorescent cores
are functionalized into displacers using different, known resin-
binding functional groups. DC-50 and selectivity pathway
plots28,29 from batch experiments and column displacements are
used to determine the separation performance of these new
fluorescent displacers. Finally, STD-NMR is employed to study
the binding between the new fluorescent displacers and proteins.
and tris(2-aminoethyl)amine were purchased from Fluka (Seelze,
Germany). NMR sample tubes, 5 mm thin wall, 600 MHz, were
purchased from Wilmad Laboratory Glass (Buena, NJ). Deuterium
oxide (99.96% purity) was purchased from Cambridge Isotope
Laboratories, Inc. (Andover, MA). All other unlisted chemicals were
obtained from Sigma-Aldrich. All reagents and solvents were used
as received without further purification.
Equipment. A Buchi Rotavapor (Flawil, Switzerland) and a
Napco vacuum oven (Waltham, MA) were used for chemical
synthesis. NMR spectra for the determination of product purity and
validation were obtained using a Varian Unity 500 NMR spec-
trometer (Palo Alto, CA). MS spectra were obtained using an
Agilent 1100 series LC/MSD-SL ion trap system (Santa Clara, CA).
Instrument control, data acquisition, and data processing were
performed on these machines using ChemStation 10.01 and Ion
Trap 5.2 software from Agilent. High-throughput batch screening
experiments were carried out on a Biomek 2000 robotic liquid
handling system from Beckman Coulter (Fullerton, CA). Super-
natant solution was collected from the membrane plates using a
vacuum manifold donated by Millipore. Supernatant solution
analysis and column separations were carried out on an HPLC
system consisting of a 600E HPLC pump and controller, a 712
WISP autoinjector, and a 484 absorbance detector from Waters
(Milford, MA). The HPLC system utilized Millennium v.2.15.01
software for data acquisition, also from Waters. Fractions from the
column runs were collected using an Advantec SF-2120 autosampler
(Tokyo, Japan). Freeze-drying of NMR samples was done on a
Labconco freeze-dry system from Labconco (Kansas City, MO).
STD-NMR spectra were obtained using a Bruker 600 MHz
spectrometer equipped with a cryogenically cooled 5 mm triple-
resonance probe head with z-axis gradients (Billerica, MA). NMR
data acquisition and analysis was carried out using the Topsin 2.1
software package, also from Bruker.
1
Procedures. 1. Product Characterization. H and 13C NMR
spectral data were collected using the Varian Unity 500 NMR
spectrometer. The sample solutions were prepared in CDCl3 and
D2O and were referenced in parts per million relative to 7.27 ppm
and 4.8 ppm each for 1H NMR. Electrospray ionization mass
spectrometry (ESI-MS) spectra were obtained for the sample
solutions using the Agilent 1100 series LC/MSD-SL ion trap system.
Samples were introduced into the ion source using a syringe pump
at a flow rate of around 7 µL/min or by using an autosampler with
a solvent flow rate of 200 µL/min. Data were collected in the
positive ion mode. Data processing was then performed using the
ChemStation 10.01 and IonTrap 5.2 software.
Experimental Section
2. High-Throughput Batch Screen of Fluorescent Displacers.
A robotic HTS protocol that had been previously developed was
used to determine the performance of the fluorescent displacers.28
The bulk stationary phase (SP Sepharose HP) was first washed twice
with deionized water and then twice with the appropriate buffer
and was allowed to equilibrate for 3 h between each wash. Buffers
for the batch displacement runs were as follows: RNaseA and
R-ChyA, 50 mM sodium phosphate at pH 6; Cytc and Lys, 50 mM
sodium acetate at pH 5.
After gravity settling of the stationary phase, the supernatant
liquid was removed and 1.3 mL of the remaining stationary phase
slurry was equilibrated with 15.6 mL of protein mixture at a total
protein concentration of 5 mg/mL (2.5 mg/mL RNaseA and R-ChyA
each or Cytc and Lys each). The protein solution was equilibrated
with the resin for 10 h, during which the stationary phase was
allowed to settle under gravity. Upon settling, the supernatant liquid
was removed, and the protein content in the supernatant solution
was determined by reversed-phase liquid chromatography (RPLC)
analysis (see below). The mass of the protein adsorbed on the
stationary phase was then calculated by mass balance.
Materials. SP Sepharose HP bulk resin and Pharmacia empty
glass columns (4.6 mm × 100 mm) were purchased from GE
Healthcare (Uppsala, Sweden). Ninety-six-well Multiscreen-HV
Durapore membrane bottomed plates were purchased from Millipore
(Bedford, MA). A Jupiter 5 µm C4 300A column (4.6 mm × 50
mm) was purchased from Phenomenex (Torrance, CA). Ribonu-
clease A (RNaseA), R-chymotrypsinogen A (R-ChyA), horse
cytochrome c (Cytc), lysozyme (Lys), sodium phosphate (monoba-
sic and dibasic), acetonitrile (ACN), trifluoroacetic acid (TFA),
acetic acid (glacial), sodium acetate, sodium chloride, hydrochloric
acid, sodium hydroxide, trizma acid, trizma base, diethylenetri-
amine, ethylenediamine, and 2-amino-5-(diethylamino)pentane were
purchased from Sigma-Aldrich (St. Louis, MO). Dansyl chloride
(24) Tugcu, N.; Cramer, S. M. J. Chromatogr., A 2005, 1063 (1-2), 15–
23.
(25) Mayer, M.; Meyer, B. J. Am. Chem. Soc. 2001, 123, 6108–6117.
(26) Morrison, C. J.; Godawat, R.; McCallum, S. A.; Garde, S.; Cramer,
S. M. Biotechnol. Bioeng., accepted for publication.
(27) Morrison, C. J.; Cramer, S. M. Biotechnol. Prog., accepted for
publication.
The protein-saturated resin was than resuspended as a 50:50
(v/v) slurry using some of the previously removed supernatant from
the protein loading step (to prevent any desorption), and the resin
slurry was loaded into a reservoir on the robotic liquid handler
(28) Rege, K.; Tugcu, N.; Cramer, S. M. Sep. Sci. Technol. 2003, 38 (7),
1499–1517.
(29) Rule G. S.; Hitchens, T. K. Fundamentals of Protein NMR Spectros-
copy; Springer: Dordrecht, The Netherlands, 2006.
9
17030 J. AM. CHEM. SOC. VOL. 130, NO. 50, 2008

Fluorescent Displacer Synthesis and Characterization
A R T I C L E S
(Biomek). In addition, 10 mM solutions of each displacer (in the
appropriate buffer) were loaded into separate reservoirs, and the
buffer was loaded into its own reservoir. A premade protocol was
then employed to run the batch screen. The Biomek dispensed 20
µL of the resin slurry into each well of a 96-well membrane plate;
the resin slurry reservoir was mixed before each aspiration. Once
the liquid from the slurry was placed in each well, a vacuum
manifold was used to filter the slurries through the membrane plate,
leaving 10 µL of settled resin in each well. The Biomek performed
serial dilutions of each displacer and then dispensed 120 µL of 10,
9, 8, 6, 5, 4, 2, and 1 mM displacer solutions into each individual
well. Each displacer concentration was done in triplicate. After all
solutions were loaded into the wells, the suspensions were mixed
by repeated pipetting and allowed to equilibrate for 3 h and then
mixed again and allowed to equilibrate for an additional 2 h. After
the final equilibration the supernatant solutions from the wells were
collected for analysis using the vacuum manifold.
3. Analysis of the Batch Supernatant Solutions. The batch
supernatant solutions were analyzed using RPLC. RPLC was carried
out using a C4 column (50 × 4.6 mm) with an A buffer of deionized
water with 0.1% TFA and a B buffer of 90% ACN, 10% deionized
water, and 0.1% TFA (all by volume). The column was first pre-
equilibrated with 20% B. A linear gradient was then carried out
from 20% to 100% B in 10 column volumes followed by 100% B
for 2.5 column volumes. The flow rate was 1 mL/min, the column
effluent was monitored at 280 nm, and 40 µL portions of the
supernatant solutions were injected. This protocol resulted in the
flow through of the displacers and the elution of the proteins during
the linear gradient.
4. Fluorescent Column Displacements. For the displacement
experiments a SP Sepharose HP column (100 mm × 4.6 mm i.d.)
was initially equilibrated with the appropriate carrier buffer (50
mM sodium phosphate, pH 6 for R-ChyA/RNaseA and 50 mM
sodium acetate, pH 5 for Cytc/Lys). The column was then perfused
sequentially with feed, displacer, and regeneration solutions. The
experimental conditions, such as feed load, flow rate, and displacer
concentration can be found in the figure captions. Regeneration
was carried out using a linear gradient from the carrier buffer to
the regeneration buffer (50 mM Tris, pH 10 with 2 M sodium
chloride). (Note: the efficacy of this regeneration protocol was found
to be effective at removing all of the bound displacers.) Fractions
of the column effluent (200 µL) were collected during the dis-
placement experiment for subsequent analysis. The column effluent
was monitored by both UV (280 nm) and fluorescence (excitation
383 nm, emission 553 nm).
5. Analysis of the Column Effluent Fractions. After the
fractions were collected, they were analyzed using RPLC. RPLC
was carried out using the same column and buffers as described
for the analysis of the batch experiments. The column was first
pre-equilibrated with 100% A. A linear gradient was then carried
out from 0 to 100% B in 20 column volumes followed by 100% B
for 2.5 column volumes. The flow rate was 1 mL/min, the column
effluent was monitored by both UV (280 nm) and fluorescence
(excitation 383 nm, emission 553 nm), and 40 µL portions of the
fractions were injected. This protocol resulted in the elution of both
the displacer and proteins during the linear gradient. The protein
amounts were determined by the UV readings, while the displacer
amounts were determined by the fluorescence readings.
Samples were analyzed using the STD method discussed in the
Introduction. This protocol is described in detail elsewhere26 and
will be briefly summarized here. In this study the methyl protons
within the protein were directly irradiated using a train of highly
selective RF pulses, such that ligand (displacer) resonances were
not directly perturbed. The extended saturation period provided by
the RF pulse train permitted spin diffusion to spread the saturation
of methyl proton magnetization to all protons throughout the entire
protein. However, ligand protons near the protein surface (<5 Å)
also experienced the saturation effect due to intermolecular dipolar
coupling. Transfer of magnetization from the protein to the ligand
resulted in an increase in signal intensity for ligand protons at the
protein interaction surface relative to those distant. A control
spectrum was used where the RF irradiation was applied to a
spectral region devoid of ligand and protein resonances. The final
STD spectrum was obtained after subtraction of the control from
the experimental spectra. Ligand exchange between the bound and
free states during this saturation period is also a key element of
this technique and permitted measurement of the STD perturbations
in the free ligand. Ligand concentrations were present in excess
relative to the protein to ensure that saturated ligand remained
in the unbound state for detection. A filter was also applied to
remove the protein signal, thus leaving only a spectrum of ligand
resonances that underwent an STD effect.
Mixtures of 50:1 (10 mM:0.2 mM) of displacer and protein,
respectively, were analyzed utilizing the experimental parameters
stated below. Additional control STD spectra were acquired on
individual displacer and protein samples of the same concentration
analyzed in the mixtures. A saturation period of 2.4 s was applied
as a train of 8 ms Gaussian pulses applied at 40 ms intervals with
a 50 Hz B1 field. The experimental and control spectra were
acquired in an interleaved fashion with the saturating frequency
centered at -0.25 and -10 ppm, respectively. Spectra were acquired
using a 2 s acquisition time with a total of 1024 scans. An 80 ms
T1rho filter was used to remove residual signal from the protein.
Resonance assignments for displacers CBZEHTA, DEDA, and
1
1
DTAEA were obtained by analyzing 2D H-13C HSQC, H-1H
TOCSY, COSY, and NOESY spectra.30
Displacer Synthesis. 1. Synthesis of Dansyl-Based Cationic
Displacers (Scheme 1). 1.1. DEDA ·HCl (DEDA). To excess
ethylenediamine (12 mL) was added dropwise dansyl chloride (1.50
g, 5.56 mmol) in methylene chloride at 0 °C under nitrogen over
30 min. The reaction mixture was stirred overnight with the reaction
temperature steadily increasing to room temperature. A saturated
aqueous solution of sodium carbonate was added to the reaction
mixture with stirring. The organic layer was extracted with
methylene chloride and then washed three times with water. The
combined organic layer was then dried with Na2CO3. Solvent was
removed using a Rotavapor. The resulting viscous product was
further purified by recrystallization from a mixture of methylene
chloride, toluene, and n-hexane in a ratio of 1:6:0.5, respectively.
The resulting solid was obtained by filtration and dried in a vacuum
oven overnight, yielding a yellow solid. The product (1.00 g, 3.40
mmol) was dissolved in methylene chloride and acidified with HCl
gas for 12 h. The formed precipitate was filtered, dissolved in a
small amount of water, and acidified further by addition of 2 mL
of HCl (aqueous, 37%). The solution was then filtered and dried
in a vacuum oven. After the resulting solid was washed with
acetone, it was again dried in a vacuum oven to give a yellow solid
final product, DEDA ·HCl (0.90 g, 72% yield): 1H NMR (500 MHz,
D2O) δ 8.63 (d, 1H), 8.42 (d, 1H), 8.30 (d, 1H), 8.50 (d, 1H), 7.85
(m, 2H), 3.45 (s, 6H), 3.10 (m, 4H); 13C NMR (500 MHz, D2O) δ
139.24, 134.53, 130.81, 128.76, 128.51, 127.08, 126.35, 126.22,
126.05, 119.69, 46.97, 40.05, 39.46; ESI-MS (m/z) for C14H19N3O2S
[MH]+ calcd 294.12, found 294.0.
6. STD-NMR Sample Preparation and Experimental Pro-
tocol. Protein and displacer stock solutions for NMR analysis were
first suspended in 50 mM sodium phosphate buffer at pH 6. The
solutions were pH adjusted to ensure a pH of 6.0 and then
lyophilized. The resulting pellets were suspended in deuterium oxide
and lyophilized twice for two cycles before a final suspension was
made in deuterium oxide. Actual NMR samples were prepared by
diluting the stock solutions to their final analysis concentrations
(10 mM for displacers and 0.2 mM for proteins) in D2O-based
buffer and were then placed in the NMR sample tubes.
(30) ChemAxon. Marvin Calculator and Plugin, 2008. http://www.
chemaxon.com/marvin/sketch/index.html.
9
J. AM. CHEM. SOC. VOL. 130, NO. 50, 2008 17031

A R T I C L E S
Morrison et al.
Scheme 1
Scheme 2
°C. After the solution was cooled to room temperature, an aqueous
sodium carbonate solution (20%) was added with stirring. The
solution was then extracted with methylene chloride, washed five
times with water, dried over sodium sulfate, and filtered. The solvent
was removed using a Rotavapor. The resulting solid was then dried
in a vacuum oven overnight, yielding a yellow solid. The sticky
product was further purified by precipitation from diethyl ether and
carbon tetrachloride. The resulting product was dried in a vacuum
overnight at 40 °C. The obtained product (1.00 g, 3.95 mmol) was
acidified in methanol (30 mL) with HCl gas for 12 h in an
ice-water bath. The precipitate was filtered and washed with
chloroform, methanol, and acetone. The obtained solid was dried
in a vacuum oven overnight to give CBZEHTA ·HCl: yield 0.86 g
1.2. DETA ·HCl (DETA). The reaction procedure was identical
to the synthetic procedure of DEDA ·HCl except for using dieth-
ylenetriamine (15 mL) instead of ethylenediamine: yield 0.91 g
1
(68%); H NMR (500 MHz, D2O) δ 8.79 (d, 1H), 8.53 (d, 1H),
8.43 (d, 1H), 8.11 (d, 1H), 7.95 (m, 2H), 3.53 (m, 10H), 3.49-3.28
(m, 4H); 13C NMR (500 MHz, D2O) δ 139.17, 134.37, 130.86,
128.78, 128.55, 127.06, 126.44, 126.22, 126.05, 119.73, 47.85,
46.94, 44.47, 39.05, 35.57; ESI-MS (m/z) for C16H24N4O2S [MH]+
calcd 337.17, found 337.10.
1
(58%); H NMR (500 MHz, D2O) δ 8.16 (d, 2H), 7.52 (m, 4H),
7.31 (t, 2H), 4.63 (t, 2H), 3.63 (t, 2H), 3.41-3.34 (m, 8H); 13C
NMR (500 MHz, D2O) δ 139.85, 126.54, 122.80, 120.71, 120.00,
108.84, 46.76, 44.57, 43.71, 43.46, 39.08, 35.38; ESI-MS (m/z)
for C18H24N4 [MH]+ calcd 297.20, found 297.20.
1.3. DTAEA·HCl (DTAEA). The reaction procedure was id-
entical to the synthetic procedure of DEDA ·HCl except for using
tris(2-aminoethyl)amine (15 mL) instead of ethylenediamine: yield
1
Results and Discussion
1.00 g (72%); H NMR (500 MHz, D2O) δ 8.75 (d, 1H), 8.45 (d,
1H), 8.39 (d, 1H), 8.16 (m, 1H), 7.90 (m, 2H), 3.53 (s, 6H),
3.33-3.20 (m, 8H), 3.19-3.10 (m, 4H); 13C NMR (500 MHz, D2O)
δ 138.57, 134.19, 131.00, 128.60, 128.45, 127.16, 126.48, 126.21,
125.78, 119.83, 52.88, 50.28, 47.01, 38.59, 35.09; ESI-MS (m/z)
for C18H29N5O2S [MH]+ calcd 380.20, found 380.20.
As stated in the Introduction, there is a significant need for
the development of displacers which would permit direct online
monitoring of displacement chromatographic processes. Further,
the ability to design chemically selective displacers that enabled
online monitoring would be particularly advantageous. Previous
work in our laboratory has identified a double aromatic ring
head and a positively charged amine tail as important structural
motifs for chemically selective displacers on a strong cation
exchange resin.28 Accordingly, the fluorescent displacers in the
current work were designed to incorporate these motifs. Car-
bazole and dansyl chloride were chosen as building blocks
because of their fluorescent nature, exposed double aromatic
ring structure, low cost, and ease of chemical modification.
These building blocks were then functionalized with amine tails
of various lengths and charges (which would interact with the
resin) as described in the Experimental Section, creating a
homologous library of fluorescent displacers.
2. Synthesis of a Carbazole-Based Cationic Displacer (Scheme
2). 2.1. Tosylated Carbazole. A solution of 9H-carbazole-9-ethanol
(5.00 g, 23.70 mmol) in pyridine (40 mL) was added dropwise to
a solution of p-toluenesulfonyl chloride (6.70 g, 35.50 mmol) also
in pyridine (40 mL) at room temperature under nitrogen. The
reaction mixture was stirred overnight. The suspension was then
poured into ice-water with stirring, resulting in a sticky product.
The resulting product was then washed five times with water and
dissolved in acetone. Water was then slowly added dropwise to
the solution, yielding a white precipitate. After the solid was filtered
and dried in a vacuum oven, it was washed with diethyl ether and
redried in a vacuum oven over phosphorus pentoxide overnight,
resulting in a white solid final product: yield 5.70 g (66%); 1H NMR
(500 MHz, D2O) δ 8.01 (d, 2H), 7.41 (t, 2H), 7.25-7.20 (m, 6H),
6.84 (d, 2H), 4.52 (t, 2H), 4.41 (t, 2H), 2.25 (s, 3H).
2.2. CBZEtEDA·HCl (CBZEHTA). The tosylated carbazole
(1.00 g, 2.74 mmol) and diethylenetriamine (15 mL) were placed
in a round-bottom flask under nitrogen. The reaction mixture was
stirred for 6 h at room temperature and then heated for 6 h at 60
The four fluorescent displacers synthesized as described in
Schemes 1 and 2 are presented in Figure 1. The individual amine
pKa values were determined for these displacers using the
9
17032 J. AM. CHEM. SOC. VOL. 130, NO. 50, 2008

Fluorescent Displacer Synthesis and Characterization
A R T I C L E S
Figure 1. Amine pKa values for the fluorescent displacers: (A) CBZEHTA,
(B) DEDA, (C) DETA, (D) DTAEA.
Table 1. Net Molecular Charge of the Fluorescent Displacers at
the Experimental pH Values
displacer
pH 6.0
pH 5.0
CBZEHTA
DEDA
DETA
+2.25
+1
+1.75
+2
+2.75
+1.25
+2.2
DTAEA
+2.75
ChemAxon Marvin Calculator31 and are indicated in the figure.
These values were then used to determine the overall molecular
charge of the displacers at the experimental pH values and are
given in Table 1. As seen in the table, the amount of charge
varied between the different displacers, with DEDA having the
least amount of charge and CBZEHTA possessing the highest
amount of charge. Because the aromatic portion of these
molecules stays constant (the dansyl chloride-derived molecules)
while the amount of charge varies, any differences in displace-
ment behavior between the displacers can most likely be
attributed to the difference in charge and charge distribution.
While CBZEHTA had its charge concentrated on the amine tail,
the three dansyl chloride-derived displacers had charge on both
the amine tail and the other side of the molecule at pH 5.
Batch Experiments with Fluorescent Displacers. To investi-
gate the performance of these compounds, a robotic high-
throughput batch screen previously developed in our laboratory
was employed. In this screen the resin was loaded with equal
amounts of each protein from a model protein pair to determine
the displacer’s efficacy in a multicomponent mixture. (Note:
the proteins in each protein pair exhibited very similar retention
times in linear gradient chromatography under the same pH and
column conditions.28) These batch experiments were carried out
at several displacer concentrations, and the results were used
to generate DC-50 plots and selectivity pathway plots28,29 for
the displacer compounds (Figures 2 and 3). DC-50 plots present
data for the percentage of protein displaced in batch experiments
as a function of the displacer concentration. In selectivity
Figure 2. DC-50 plots of the fluorescent displacers for the protein pairs
R-ChyA/RNaseA and Cytc/Lys: (A) CBZEHTA, (B) DEDA, (C) DETA,
(D) DTAEA. The given error bars represent (1 standard deviation of the
triplicate values.
pathway plots, the percentage of one protein displaced from a
pair is plotted as a function of the percentage of the other
displaced protein in the pair. The trajectory is then constructed
at increasing displacer concentrations and different regions of
the plot are defined.
(31) Liu, J.; Hilton, Z. A.; Cramer, S. M. Anal. Chem. 2008, 80 (9), 3357–
3364.
9
J. AM. CHEM. SOC. VOL. 130, NO. 50, 2008 17033

A R T I C L E S
Morrison et al.
Figure 4. Selective displacement of the Cytc/Lys protein pair using 10
mM DTAEA: column, 100 mm × 4.6 mm i.d., HP Seph SP resin; carrier,
50 mM sodium acetate buffer, pH 5.0; protein loading, 10 mg each of Cytc
and Lys; flow rate, 0.2 mL/min; online UV, 280 nm; online fluorescence,
excitation 383 nm, emission 553 nm.
amount of the protein RNaseA or Cytc was displaced in the
batch experiments. These results indicate that DETA acts as an
effective selective displacer, which is indicative of an appropriate
balance of affinities for the resin and the selected protein.
Because the only difference between DETA and DEDA
(Scheme 1) is the amount of positive charge on the amine tails,
it is clear that the latter did not possess sufficient charge to give
a selective separation.
The results for DTAEA, which has slightly more charge than
DETA (Table 1), are given in Figure 2D. As seen in the figure,
DTAEA also acted as an efficient selective displacer for both
protein pairs. While the performances of DETA and DTAEA
were similar, DTAEA was slightly more efficient at displacing
the proteins, in general.
Figure 3. Selectivity pathway plots of the fluorescent displacers for the
two protein pairs: (A) R-ChyA/RNaseA, (B) Cytc/Lys.
As can be seen in Figure 2A the carbazole-derived fluorescent
displacer, CBZEHTA, did not give a selective separation
between the proteins of each pair. In fact, at higher concentra-
tions this displacer behaves as a high-affinity displacer rather
than a selective one. This result was unexpected because the
design of CBZEHTA fits the standard design of a chemically
selective displacer for these protein pairs, namely, a double
aromatic ring head with a positively charged amine tail.
However, in our previous work,28 it was established that the
design of a chemically selective displacer is a delicate balance
between the displacer’s affinity for the resin and its affinity for
binding to the “selected protein”, in this case R-ChyA or Lys.
It is believed that the nonselective result shown with CBZEHTA
was caused by too much charge being present on the amine
tail, a total of 2.25 and 2.75 positive charges at the experimental
conditions of pH 5 and 6, respectively. In addition, as indicated
above, all of this charge was located on the amine tail. This
higher charge density makes the CBZEHTA more efficient at
displacing proteins from the resin, thus not enabling efficient
selective displacement to take place. This hypothesis is sup-
ported by Figure 3B, where CBZEHTA acts as a selective
displacer for a specific range of concentrations. On the other
hand, at higher displacer concentrations, this molecule acts as
a fluorescent high-affinity displacer.
It is instructive to evaluate the selectivity pathway plots for
the homologous series of dansyl chloride displacers, Figure 3.
As seen in the figure, DEDA was unable to give a selective
separation at the concentrations evaluated, most likely due to
insufficient charge. On the other hand, DETA, with an increased
amount of charge, gave not only a selective separation but an
exclusive separation27 (where there is a greater than 50%
separation in the pair and one protein remains strongly bound
onto the resin with less than 10% displaced) at certain
concentrations as well. While DTAEA also produced exclusive
separations, the concentration range of exclusivity was more
limited than for DETA. From these data it is clear that the dansyl
chloride-derived displacers are indeed selective. In fact, DETA
and DTAEA have actually exhibited some of the best batch
experiment separations of any chemically selective displacer
tested in our laboratory to date.27,28,31
Column Displacements using Fluorescent Displacers. One of
the major challenges of displacement chromatography is the
determination of the displacer breakthrough during column
experiments. Typically, the breakthrough is determined by
offline analysis of fractions from the column effluent, which is
cumbersome and time-consuming. Accordingly, experiments
were conducted to determine the performance of the fluorescent
displacers in a column setting. In particular, it was desired to
gauge the performance of the fluorescent displacers as a direct,
online detection method in column displacement chromatogra-
phy. The resulting selective column displacements of the two
protein pairs Cytc/Lys and R-ChyA/RNaseA are shown in
Figures 4 and 5, respectively. Online detection methods (UV
and fluorescence) are shown as thin dotted lines, while
concentrations determined by fraction collection are shown as
thick dark lines.
The plots shown in Figure 2B-D indicate the performance
of the homologous series of dansyl chloride displacers. In Figure
2B, almost no R-ChyA or Lys is displaced while only a small
amount of RNaseA or Cytc is displaced. These results indicate
that while DEDA might possess some selectivity for these
protein pairs, the displacer has insufficient charge (+1 and
+1.25 at pH 5 and 6, respectively) to displace proteins efficiently
from the resin, making it an ineffective displacer.
The results with DETA are given in Figure 2C. As seen in
the figure, while minimal R-ChyA or Lys was displaced, a large
As can be seen in Figure 4, a complete separation between
Cytc and Lys was accomplished using 10 mM DTAEA as a
9
17034 J. AM. CHEM. SOC. VOL. 130, NO. 50, 2008

Fluorescent Displacer Synthesis and Characterization
A R T I C L E S
DEDA, and DTAEA are shown in Figure 6. Proton peak
assignments for the displacers are shown in the figure as well.
To simplify the discussion of the results in Figure 6, the
spectra for each displacer will be evaluated in series. The proton
peak assignments for the CBZEHTA displacer (A), shown on
the left side of Figure 6, were made as follows: 6-1 on the
amine tail and 10-7 on the aromatic rings (note that the peak
for proton 6 is hidden in the water peak at 4.7 ppm). As can be
seen from the STD spectra given in (B) and (C), a binding signal
was clearly observed for the R-ChyA/CBZEHTA mixture while
no signal was observed for the RNaseA mixture. Furthermore,
a lack of signal from the single-component proteins (E, F) and
single-displacer spectra (D) verified that any signal seen in the
mixture experiments was caused by a binding event. Upon closer
analysis of the R-ChyA/CBZEHTA mixture signal, it can be
seen that the binding is centered on the aromatic portion of the
displacer, with only a small amount of signal detected for
protons within the amine tail, suggesting a hydrophobic binding
event. This selective binding to R-ChyA for CBZEHTA was
unexpected because the displacer gave a nonselective result from
the analysis of the DC-50 plots. However, as stated previously,
this nonselective result was most likely due to the differences
in charge and charge distribution as compared to those of the
other dansyl chloride-based displacers (Table 1). This makes
the CBZEHTA more efficient at displacing proteins from the
resin, thus not enabling efficient selective displacement to take
place. If the amount of charge on CBZEHTA were reduced, a
more selective result would most likely occur, as indicated by
the selective binding event shown from the STD-NMR results.
Figure 5. Selective displacement of the RNaseA/R-ChyA protein pair using
10 mM DTAEA: column, 100 mm × 4.6 mm i.d., HP Seph SP resin; carrier,
50 mM sodium phosphate buffer, pH 6.0; protein loading, 10 mg each of
RNaseA and R-ChyA; flow rate, 0.2 mL/min; online UV, 280 nm; online
fluorescence, excitation 383 nm, emission 553 nm.
displacer. Cytochrome c was completely displaced ahead of the
displacement zone, as confirmed by mass balance, while no
detectable amounts of Lys were seen in the fractions from the
column effluent. This result demonstrates a complete separation
of two components that directly overlapped under linear gradient
ion exchange conditions. Importantly, when only the online
detection methods are employed, significant information is
provided regarding this two-protein displacement separation. The
UV trace distinctly shows the protein zone ahead of the dis-
placement front. while the fluorescence trace shows only the
displacer breakthrough. Also, it can be seen that the online
detection results directly reflect the results determined by
fraction collection and offline analysis.
The selective column displacement of RNaseA/R-ChyA seen
in Figure 5 indicates a result similar to that shown in Figure 4.
Ribonuclease A was completely displaced ahead of the dis-
placement zone, as confirmed by mass balance, while no dete-
ctable amounts of R-ChyA were seen in fractions from the
column effluent. This result also demonstrates a complete sepa-
ration of these two components. It is important to note these
two proteins were inseparable under linear gradient ion exchange
conditions carried out at the same pH. The small overlap
between the displaced protein zones and the DTAEA zone in
Figures 4 and 5 is comparable to that obtained previously with
displacement chromatography on the SP Sepharose HP resin
used. It should also be noted that no fluorescent signal quenching
of the displacer by the proteins was seen in these column
experiments. The results presented in Figures 4 and 5 provide
proof of concept for the utility of online detection of separate
protein and displacer zones in a column setting using fluorescent
selective displacers.
The NMR results for the displacer DEDA are shown in the
center of Figure 6. The proton peak assignments for this
displacer (A) were made as follows: 1 and 2 on the amine tail,
3 on the aromatic ring ortho to the sulfone group with 4 meta
and 5 para on the same ring, 6 on the dimethylamine off the
other aromatic ring, and 7 on the aromatic ring ortho to the
dimethylamine group with 8 meta and 9 para on the same ring.
As can be seen from the STD spectra given in (B) and (C), a
binding signal was clearly observed for the R-ChyA/DEDA
mixture while the RNaseA mixture produced a relatively weak
signal. The protein control experiments (E, F) showed no signal;
however, a small amount of signal was seen in the STD
spectrum of this displacer alone (D). This small residual STD
signal is attributed to the direct saturation of the displacer,
resulting in a slight amount of STD signal in the mixtures (B,
C) as well as the control (D). If one compares the spectra in
(B)-(D), it is clear that the signal obtained for the RNaseA
mixture can be attributed primarily to the direct saturation of
the displacer. In contrast, the signal obtained for the R-ChyA
mixture is significantly higher than that of the displacer control,
confirming displacer-protein binding. Upon closer analysis of
the R-ChyA/DEDA mixture signal, it can be seen that the
binding is centered on the aromatic portion of the displacer and
the methyl groups of the dimethylamine group, while only a
weak binding signal was seen for the amine tail. This would
again suggest that the binding event is occurring primarily due
to hydrophobic interactions. These NMR results indicate that
DEDA binds selectively to R-ChyA while not binding to
RNaseA, thus selectively displacing RNaseA while retaining
R-ChyA on the resin surface, as indicated by the DC-50 plots.
However, as shown in Figures 2 and 3, even though this
displacer has some selectivity, it does not possess sufficient
charge (Table 1) to efficiently displace any proteins from the
resin, thus being categorized as an ineffective displacer.
STD-NMR Experiments on Fluorescent Displacers. Previous
work done in our laboratory has shown that selective binding
between a protein and displacer is the mechanism by which
chemically selective displacement occurs.27,28,31 STD-NMR
experiments were employed to directly determine whether the
fluorescent displacers bound selectively to proteins. Because the
STD experiment can also give insight into where the binding
occurs on the displacers at atomic resolution, proton peak
assignments were also determined for the fluorescent displacers.
These assignments were obtained using 2D NMR experiments
(not shown) described in the Experimental Section using
standard methodologies. Mixtures of a single protein and a single
displacer in a 1:50 ratio were analyzed with the STD protocol.
Solutions of the same concentration of protein and displacer
were also analyzed as controls to verify the method and results.
The resultant spectra from these experiments for CBZEHTA,
9
J. AM. CHEM. SOC. VOL. 130, NO. 50, 2008 17035

A R T I C L E S
Morrison et al.
Figure 6. STD-NMR results for the protein/displacer experiments with proton peak assignments shown: (A) 1D proton spectra for the displacer (blue), (B)
STD for the R-ChyA/displacer mixture (red), (C) STD for the RNaseA/displacer mixture (green), (D) STD control for the displacer (violet), (E) STD control
for R-ChyA (yellow), (F) STD control for RNaseA (orange).
The NMR results for the displacer DTAEA are shown on
the right side of Figure 6. The proton peak assignments for this
displacer (A) were made as follows: 1 and 2 on the forked
extension of the amine tail, 3 and 4 on the amine tail, 5 on the
aromatic ring ortho to the sulfone group with 6 meta and 7 para
on the same ring, 8 on the dimethylamine off the other aromatic
ring, and 9 on the aromatic ring ortho to the dimethylamine
group with 10 meta and 11 para on the same ring. As can be
seen from the STD spectra given in (B) and (C), a binding signal
was clearly observed for the R-ChyA/DTAEA mixture while
the RNaseA mixture again produced a relatively weak signal.
The protein control experiments (E, F) showed no signal;
however, a small amount of signal was seen in the STD
spectrum of this displacer alone (D). As with the DEDA results,
if one compares the spectra in (B)-(D), it is clear that the signal
obtained for the RNaseA mixture can be attributed primarily to
the direct saturation of the displacer. In contrast, the signal
obtained for the R-ChyA mixture is significantly higher than
that of the displacer control, confirming displacer-protein
binding. Upon closer analysis of the R-ChyA/DTAEA mixture
signal, it can be seen that the binding is centered on the aromatic
portion of the displacer and the methyl groups of the dimethy-
lamine group, while only a weak binding signal was seen for
the forked amine tail. This would also suggest that the binding
event is occurring primarily due to hydrophobic interactions.
These NMR results indicate that DTAEA binds selectively to
R-ChyA while not binding to RNaseA, thus selectively displac-
ing RNaseA while retaining R-ChyA on the resin surface, as
indicated by the DC-50 plots.
The results from the STD-NMR confirm that all three of these
displacers bind selectively to one of the proteins in each protein
pair. However, as stated above, for the displacer to act as an
effective selective displacer, it must both bind to one of the
proteins and possess sufficient charge to displace the other
protein from the resin surface. DEDA has too little charge and
thus cannot effectively displace proteins. On the other hand,
CBZEHTA has too much charge and displaces both proteins,
even the one to which it binds. Finally, DTAEA has the
appropriate balance between protein binding and resin affinity
and can act as an efficient selective displacer for this specific
class of separation problems, namely, the separation of proteins
with similar ion exchange affinity but differing surface hydro-
phobicities.
Conclusion
This work has addressed one of the major obstacles to the
widespread implementation of displacement chromatography,
namely, the ability to carry out real time detection of the
separation process. To address this challenge, fluorescent
displacers were designed using known chemically selective
displacers as templates. Fluorescent cores were chosen on the
basis of their exposed aromatic ring structure, low cost, and
9
17036 J. AM. CHEM. SOC. VOL. 130, NO. 50, 2008

Fluorescent Displacer Synthesis and Characterization
A R T I C L E S
ease of chemical modification. The chosen cores were then
functionalized with different charged moieties, creating a homo-
logous library. These compounds were then tested on two
protein pairs, R-ChyA/RNaseA and Cytc/Lys, using batch and
column displacement experiments. The batch results indicated
that the fluorescent displacers acted as high-affinity, selective,
or ineffective displacers depending on their charge. While the
more charged fluorescent displacer CBZEHTA was generally
found to be a nonselective, high-affinity displacer, two of the
dansyl chloride-derived displacers, DETA and DTAEA, were
highly selective and exclusive at multiple concentrations.
Column displacements using 10 mM DTAEA yielded complete
separation of both protein pairs while facilitating online detection
using UV and fluorescence detection. STD-NMR was then
carried out to investigate the binding of the fluorescent displacers
to proteins. The NMR results indicated a selective binding of
CBZEHTA, DEDA, and DTAEA to R-ChyA, while no binding
was observed for RNaseA. These results indicate that if a
molecule is to act as an effective selective displacer, it must
bind to one of the proteins, while also possessing sufficient
charge to displace the other protein from the resin surface.
This work demonstrates the ability to synthesize and employ
fluorescent chemically selective displacers for protein purifica-
tion. These displacers could be custom designed for different
classes of protein separations using a similar methodology
presented in this paper. The ability to directly monitor displace-
ment separations online represents a significant advance in the
field of displacement chromatography and promises to have a
major impact on the implementation of this technology for
applications ranging from the large-scale production of biop-
harmaceuticals to complex bench-scale bioseparations.
Acknowledgment. This work was partially supported by NIH
Grant 5R01 GM047372 and NSF Grant CBET-0730830.
JA806279X
9
J. AM. CHEM. SOC. VOL. 130, NO. 50, 2008 17037