Peptide microarrays for the discovery of bioactive surfaces that guide cellular processes: A single step azide-alkyne click chemistry approach

Article abstract of DOI:10.1039/c4tb00375f

Cell behavior in vivo is guided by a complex microenvironment containing many different molecules including extracellular matrix (ECM) proteins, growth factors, and proteoglycans. Controlling the interaction between these various components at the cell-material interface will be invaluable in developing new materials for biomedical devices and tissue engineering applications. We report a single step approach to forming mixed peptide conjugated self-assembled monolayers on gold using copper-catalyzed azide-alkyne cycloaddition chemistry to study the combinatorial effects of different peptide ligands on cellular processes. We synthesized ECM adhesion peptides (YIGSR, GRGDS), a bone morphogenetic protein 7 (BMP-7) derived peptide (KPSSAPTQLN), and a heparin binding peptide (KRSR), and arrayed them, alone and in combination, onto gold coated coverslips. SAMs were characterized by X-ray photoelectron spectroscopy (XPS) and matrix-assisted laser desorption/ionization (MALDI) mass spectrometry, and arrayed peptide combinations were seen to differentially bind to adipose derived stem cells (ADSCs) and mouse embryonic fibroblasts (MEFs). We further investigated the osteogenesis of ADSCs on SAMs containing combinations of adhesion peptide and BMP-7 peptide in both standard culture and osteogenic differentiation media. We demonstrate enhanced expression of osteogenic markers Runx2 and osteopontin when ADSCs are adherent to BMP-7 derived peptide alone or in combination with ECM adhesion peptides. The platform presented here enables immobilization of multiple peptides in a single step using a commercially available microarray spotter which will prove useful in fabricating biomolecule interfaces for cell biology studies and biochemical assays. This journal is the Partner Organisations 2014.

Full text of DOI:10.1039/c4tb00375f

Journal of
Materials Chemistry B
View Article Online
View Journal | View Issue
PAPER
Peptide microarrays for the discovery of bioactive
surfaces that guide cellular processes: a single step
azide–alkyne “click” chemistry approach†
Cite this: J. Mater. Chem. B, 2014, 2,
4280
*
Douglas Zhang and Kristopher A. Kilian
Cell behavior in vivo is guided by a complex microenvironment containing many different molecules
including extracellular matrix (ECM) proteins, growth factors, and proteoglycans. Controlling the
interaction between these various components at the cell–material interface will be invaluable in
developing new materials for biomedical devices and tissue engineering applications. We report a single
step approach to forming mixed peptide conjugated self-assembled monolayers on gold using copper-
catalyzed azide–alkyne cycloaddition chemistry to study the combinatorial effects of different peptide
ligands on cellular processes. We synthesized ECM adhesion peptides (YIGSR, GRGDS),
a bone
morphogenetic protein 7 (BMP-7) derived peptide (KPSSAPTQLN), and a heparin binding peptide (KRSR),
and arrayed them, alone and in combination, onto gold coated coverslips. SAMs were characterized by
X-ray photoelectron spectroscopy (XPS) and matrix-assisted laser desorption/ionization (MALDI) mass
spectrometry, and arrayed peptide combinations were seen to differentially bind to adipose derived stem
cells (ADSCs) and mouse embryonic fibroblasts (MEFs). We further investigated the osteogenesis of
ADSCs on SAMs containing combinations of adhesion peptide and BMP-7 peptide in both standard
culture and osteogenic differentiation media. We demonstrate enhanced expression of osteogenic
markers Runx2 and osteopontin when ADSCs are adherent to BMP-7 derived peptide alone or in
combination with ECM adhesion peptides. The platform presented here enables immobilization of
multiple peptides in a single step using a commercially available microarray spotter which will prove
useful in fabricating biomolecule interfaces for cell biology studies and biochemical assays.
Received 7th March 2014
Accepted 19th May 2014
DOI: 10.1039/c4tb00375f
www.rsc.org/MaterialsB
hold much promise, precise control and understanding of the
complex signaling between the cell and the extracellular matrix
Introduction
(ECM) will be important for clinical translation. When bioma-
terials are exposed to biological environments in vivo or growth
media in vitro, ECM matrix proteins can nonspecically adsorb
to the surface, allowing cells to interact with the biomaterial
surface through the adsorbed ECM proteins.9 ECM proteins
such as bronectin, vitronectin, and laminin are oen used for
surface modications, and combinations of these and other
proteins are an attractive strategy for recapitulating the complex
composition of the in vivo microenvironment. For instance,
Langer and colleagues used combinations of leukemia-inhibi-
tory factor, bronectin, laminin, and broblast growth factor 4
substrates to investigate murine embryonic stem cell self-
renewal and differentiation.10 Bhatia and colleagues spotted 32
combinations of ve ECM molecules (collagen I, collagen III,
collagen IV, laminin, and bronectin) onto acrylamide hydro-
gels and assessed the differentiation of mature, primary rat
hepatocytes, and mouse ES cells.11 Both studies showed that
ECM molecules can have synergistic as well as antagonistic
effects on self-renewal and differentiation, demonstrating the
importance of cross-talk between integrin, growth factor, and
matrix molecules.
Autologous stem cells – multipotent cells derived from a
patient's own body – offer broad therapeutic potential due to
their immunocompatibility, ability to differentiate into multiple
cell types, and lack of ethical issues related to their derivatiza-
tion. Bone marrow derived stem cells (BMSCs), rst observed by
Friedenstein and co-workers,1 are a well characterized cell type
known to differentiate into multiple cell types important to
tissue engineering, including cartilage, bone, tendon, and
muscle.2–4 Recently, these mesenchymal stem cells have been
isolated in nearly all tissues and organs in the body.5 Multi-
potent stem cells within adipose tissue, adipose-derived stem
cells (ADSCs) have recently gained popularity as adipose tissue
is widely available and easily obtainable.6–8 While these cells
Department of Materials Science and Engineering, University of Illinois,
Urbana-Champaign, Illinois 61801, USA. E-mail: kakilian@illinois.edu
† Electronic supplementary information (ESI) available: Additional information
about HPLC purication; AFM measurements of monolayers; 10 day image
demonstrating long term culture capabilities of spotted areas; example of how
surface coverage is quantied; and ellipsometry measurements is provided. See
DOI: 10.1039/c4tb00375f
4280 | J. Mater. Chem. B, 2014, 2, 4280–4288
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Journal of Materials Chemistry B
The discovery that the arginine-glycine-aspartic acid (RGD) biofunctionalization strategy can be applied to any peptide or
amino acid sequence on the binding domain of bronectin small molecule containing a pendant alkyne group. We further
could form an essential recognition site for cells12 spurred investigate the differentiation of ADSCs on a combination of
research into discovering other minimum binding sequences peptides and show that surfaces presenting the BMP-7 derived
that had similar selectivity for cell integrins. Although less peptide alone or in combination with other adhesion peptides
effective than whole proteins alone, peptides offer the advan- can enhance osteogenic markers aer 3 weeks in standard
tage of being short, easily synthesized, and integrin-specic.13 culture media.
Peptides are cost-efficient, easy to synthesize and purify, and
able to be conjugated in a spatially dened manner.14 In addi-
Experimental section
Materials
tion, very low concentrations of peptides are needed for
maximal spreading (ꢀ1 fmol cmꢁ2).15 Materials to model ECM
must incorporate several design elements. They should be well
Cover glass substrates were purchased from Fisher Scientic.
All chemicals, unless noted otherwise, were purchased from
Sigma Aldrich, of analytical grade and used as received. 11-(2-{2-
[2-(2-Azido-ethoxy)-ethoxy]-ethoxy}-ethoxy)-undecane-1-thiol
dened, allow for ECM protein or peptide immobilization, and
present an inert background to nonspecic adsorption. One of
the best and most studied systems are self-assembled mono-
layers (SAMs) of alkanethiolates on gold.16 SAMs form well-
(referred to herein as HS-C11-EG4-N3) was purchased from
ordered close-packed structures in the presence of gold, and
Prochimia (Sopot, Poland, TH 008-m11.n4-0.2). Triethylene
ECM protein mimics can be introduced by a simple mixture of
glycol mono-11-mercaptoundecyl ether (referred to herein as
terminally substituted alkanethiols.14,17 Hudalla and Murphy for
HS-C11-EG3) was purchased from Sigma Aldrich (673110).
Tissue culture plastic ware was purchased from USA Scientic.
instance, used a mixture of carboxyl and azide terminated
alkanethiolates to immobilize adhesion peptide RGDSP as well
Peptide synthesis reagents and amino acids were purchased
as TYRSRKY in order to explore the synergistic effects of these
from Anaspec. Cell culture media and reagents were purchased
from Gibco. Sulfo-NHS-LC-biotin (Pierce) and streptavidin-Cy3
(Life Technologies) were a kind gi from Dr Paul R. Selvin in the
Department of Physics at the University of Illinois at Urbana-
peptides on cell adhesion.18
Ligand presentation clearly plays a role in stem cell fate
decision and function, and testing single candidate peptides
and proteins is instructive, but in order to discover novel
Champaign. Human adipose-derived stem cells (ADSCs) (PT-
ligands, high-throughput approaches to model ECM are
5006) and differentiation media were purchased from Lonza.
Mouse embryonic broblasts (MEFs) (C57BL/6 4M untreated)
were purchased from globalstem. ADSCs were isolated from
lipoaspirates collected during surgical liposuction procedures.
required.19 Early work by Langer and colleagues exploited the
use of robotic uid handling to create arrays of polyacrylate
monomers to study the effect of polymer–cell interactions.10
Lutolf and co-workers used a DNA spotter to create cell niche
ADSCs were tested for purity by Lonza, and were positive for
microarray spots with modular stiffness (1–50 kPa) per well,
along with various combinations of proteins. They found that
certain combinations of ECM adhesion protein, growth factor,
and stiffness could inuence proliferation, quiescence, and
death of neural stem cells.20 Kiessling and colleagues applied
CD13, CD29, CD44, CD73, CD90, CD105, CD166 by ow
cytometry, and had the ability to differentiate into osteogenic,
chondrogenic, adipogenic lineages (http://www.lonza.com). The
use of human ADSCs in this work was reviewed and approved by
the University of Illinois at Urbana-Champaign Biological Safety
Institutional Review Board.
SAMs on gold into an array type format investigating the effects
of various peptide ligands on stem cell culture.21 Using phage
display based on the “biopanning and rapid analysis of selective
interactive ligands” (BRASIL) technique,22 they were able to
Peptide synthesis
identify six novel peptides that possessed embryonal carcinoma Peptides were synthesized manually by standard Fmoc solid-
cell binding capabilities.23 Recently, Yousaf and colleagues phase methodology. N-terminal uorenylmethyloxycarbonyl
microarrayed a variety of small molecules and peptides and (Fmoc) protected rink amide resin was deprotected with 20%
investigated ligand density effects on the differentiation of piperidine in N0,N0-dimethylformamide (DMF) for 15 min. The
MSCs.24
solvent was ltered with a vacuum manifold, and resin was
In this paper we present a microarray strategy using self- washed 4 times with DMF. A solution containing 3 equivalents of
assembled monolayers of alkanethiolates on gold to form well- the amino acid, benzotriazol-1-yl-oxytripyrrolidinophosphonium
dened regions of various combinations of peptides in a single hexauorophosphate (PyBOP), and N-methylmorpholine in DMF
step. The peptides are conjugated to the SAM via copper
I
was added to each tube for a 1 h incubation at room tempera-
catalyzed azide–alkyne “click” cycloaddition (CuAAC). We ture. The solutions were then ltered, washed 4 times with
investigated a cell adhesion peptide (YIGSR), a bone morpho- DMF, and the next Fmoc was deprotected. Aer all amino acid
genetic protein (BMP) growth factor derived peptide residues were coupled, the peptides were capped with a prop-
(KPSSAPTQLN), and a heparin binding peptide (KRSR). By argyl-PEG-NHS ester (Quanta Biodesign, 10511) in DMF over-
varying the ratio of peptides before spotting, we can investigate night. The resin was then washed 4 times with DMF and 4 times
combinatorial effects of these peptides on the adhesion char- with ethanol before cleaving with a cocktail containing 95%
acteristics on two different cell types, mouse embryonic bro- triuoroacetic acid (TFA), 2.5% H2O, and 2.5% triisopropylsi-
blasts (MEFs) and adipose derived stem cells (ADSCs). This lane (TIS) and incubating at room temperature for 3 h. The
This journal is © The Royal Society of Chemistry 2014
J. Mater. Chem. B, 2014, 2, 4280–4288 | 4281

View Article Online
Journal of Materials Chemistry B
Paper
resulting solution was evaporated by owing air over the tube monolayers by incubation of click solution and peptide ligand
for 30 minutes. The resultant precipitate was re-dissolved in 1 (1 : 1, 10 mL) at room temperature for 12 h.
mL TFA, and re-precipitated in 9 mL ice-cold diethyl ether. Aer
3 additional dissolve-precipitate steps, the residues were dis-
solved in water and lyophilized overnight. Bioactive peptides
Tyr-Ile-Gly-Ser-Arg (YIGSR), Lys-Pro-Ser-Ser-Ala-Pro-Thr-Gln-
Leu-Asn (KPSSAPTQLN, herein referred to as KPSS), Lys-Arg-Ser-
Arg (KRSR), Gly-Arg-Gly-Asp-Ser (GRGDS) and scrambled
control peptide Gly-Arg-Asp-Gly-Ser (GRDGS) were synthesized.
Final products were analyzed with low resolution electrospray
ionization (ESI) (Waters Quattro II) and semipreparative
reversed-phase high-performance liquid chromatography (RP-
HPLC) (Perkin-Elmer Flexar). All peptides used were puried to
>90% purity as assessed by HPLC.
Cell culture
Human ADSCs were cultured in Dulbecco's Modied Eagle's
Medium (DMEM) low glucose (1 g mLꢁ1) media supplemented
with 10% fetal bovine serum (Invitrogen), 1% penicillin/strep-
tomycin (p/s), media changed every 3–4 days and passaged at
ꢀ80% conuency using 0.25% trypsin : EDTA (Gibco). MEFs
were cultured in Dulbecco's Modied Eagle's Medium (DMEM)
high glucose (4.5 g mLꢁ1) media supplemented with 10% fetal
bovine serum (Invitrogen), 1% penicillin/streptomycin (p/s) and
passaged at ꢀ80% conuency using 0.05% trypsin : EDTA
(Gibco). Passage 4–6 ADSCs and passage 6 MEFs were seeded
onto microarrayed and non-microarrayed surfaces at a cell
density ꢀ15 000 cells per cm2 for ADSCs and ꢀ30 000 cells per
Surface preparation
2
ꢃ
cm for MEFs. Aer 1 h incubation at 37 C and 5% CO2, non-
adherent cells were aspirated, surfaces were washed, and fresh
media added.
Glass coverslips were cleaned by sonicating 20 min in ethanol
and dried under air. 5 nm of Ti followed by 20 nm of Au were
then deposited on the surface by electron beam deposition.
Gold surfaces were stored in a dessicator for up to two weeks
before use. Prior to microarray spotting, gold surfaces were cut Cell differentiation
to t into 6-well plates, cleaned by briey sonicating for 1 min in
For long-term differentiation studies, human ADSCs were
glacial acetic acid and 1 min in ethanol, and mounted onto 75 ꢂ
25 mm microscope slides by applying a thin layer of ethanol to
the interface.
seeded on non-microarrayed SAM surfaces (ꢀ10 000 cells per
cm2) and cultured in basal media (low glucose DMEM supple-
mented with 1% p/s and 10% FBS) or osteogenic media (low
glucose DMEM containing FBS, gentamicin/amphotericin,
L-glutamine, dexamethasone, ascorbate and b-glycer-
ophosphate). Media was changed every 3–4 days for 21 days.
Microarray printing
Stock solutions of peptide ligand (1 mM in H2O), Tris[(1-benzyl-
1H-1,2,3-triazol-4-yl)methyl]amine (TBTA) (10 mM in DMSO–t-
butyl alcohol (3 : 1)), and azide-terminated alkanethiolate
solution (HS-C11-EG4-N3/HS-C11-EG3, 15% azide mole fraction
Cell attachment quantication
Quantication of ADSC and MEF spot coverage in brighteld
images was performed using ImageJ (NIH). The “Image Edge”
plugin developed by Thomas Boudier (available at http://
www.snv.jussieu.fr/ꢀwboudier/sos.html) was applied to
images of array spots and then a threshold was applied to select
areas marked by cells. Circular regions of interest (ROIs) were
sized to the same dimensions of an array spot and were used to
dene areas for measurement (Fig. S3†). A minimum of 3 sub-
arrays, each containing at least 4 spots per peptide combina-
tion, were measured.
ꢃ
in ethanol) were prepared and stored at ꢁ20 C. Copper solu-
tion (10 mM CuBr, 10 mM sodium ascorbate in DMSO) was
prepared fresh prior to click modication. Click solution was
prepared by combining stock TBTA solution with fresh CuBr
solution (2 : 1 v/v). Reaction vials containing peptide ligand
(5 mL), click solution (5 mL), and azide-terminated alka-
nethiolate solution (10 mL) were prepared and incubated at
37 ꢃC for 1 h. These solutions were transferred to a 384-well
plate and printed in subarray format on the gold-coated
surfaces using a Gene Machines OGR-03 OmniGrid Micro-
arrayer. The substrate is removed from the microscope slide
and thoroughly rinsed with DI H2O followed by ethanol. Aer 4
rinse steps, printed substrates were immersed in a 0.1% ethyl-
enediaminetetraacetic acid (EDTA) solution for 20 minutes,
followed by another 4 rinse steps. Substrates were then
immersed in a HS-C11-EG3 solution to render the non-printed
regions inert to nonspecic adsorption.
Immunohistochemistry and uorescence assays
Cells were xed with 4% paraformaldehyde for 20 min, per-
meabilized with 0.1% Triton X-100 in PBS for 30 min, and
blocked with 1% BSA in PBS for 15 min. Primary antibody
labeling was performed in 1% BSA in PBS solution overnight at
4 ꢃC with rabbit anti-osteopontin (Abcam, 1 : 500 dilution) and
mouse anti-Runx2 (Abcam, 1 : 500 dilution). Secondary anti-
body labeling was performed similarly in 5% goat serum con-
taining 1% BSA in PBS solution with tetramethylrhodamine-
Long-term SAMs
For assessing differentiation of ADSCs over 21 days, we cultured conjugated anti-rabbit IgG antibody and Alexa Fluor 647-
cells on selected peptide combinations. Au surfaces were conjugated anti-mouse IgG antibody (1 : 500 dilution) along
immersed in 15% HS-C11-EG4-N3/HS-C11-EG3 overnight to with Alexa Fluor 488-phalloidin (1 : 250 dilution) and DAPI
ꢃ
form monolayers. Surfaces are rinsed with ethanol and cut to t (1 : 5000 dilution) for 20 min at 37 C. Peptide self-assembled
into 24-well culture plates. Peptides are conjugated to the monolayers with terminal lysine residues were exposed to a
4282 | J. Mater. Chem. B, 2014, 2, 4280–4288
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Journal of Materials Chemistry B
1 mM solution of sulfo-NHS-LC-biotin in 1 ꢂ PBS for 30 background subtraction using a Shirley routine and a subse-
minutes followed by rinsing. The functionalized surfaces were quent tting of Gaussian–Lorentzian line-curves using a Mar-
incubated with Cy3-Streptavidin for 30 minutes followed by quardt–Levenberg optimization algorithm.
copious rinsing in 1 ꢂ PBS prior to microscopy.
Ellipsometry
Microscopy
Ellipsometric measurements were made with a J. A. Woollam
VASE spectroscopic ellipsometer with wavelength of 500–
800 nm and incident angles of 65, 70, and 75ꢃ. Thicknesses were
computed with a planar two-layer (substrate, monolayer) Cau-
chy model with refractive index set at 1.40 for the monolayer.25
Brighteld images were taken with a MOTIC AE31 inverted
microscope with a Moticam 3 Digital Color Camera. Immuno-
uorescent imaging was performed on an IN Cell Analyzer 2000
(GE). A minimum of 16 elds of view were taken for each sample
condition. Immunouorescent images were analyzed using
ImageJ. Regions of interest for Runx2 and osteopontin were
selected by outlining nuclei in the DAPI channel. Mean inte-
grated density of every cell localized to the region of interest was
averaged for each replicate. To normalize uorescence intensity
for cells of differing sizes, we chose to only report the average
uorescent intensity localized at the nucleus. At least three
independent experiments each with triplicate samples were
performed to verify results. One-way ANOVA was performed
using OriginPro soware.
Atomic force microscopy (AFM)
The measurements were carried out with an Asylum Research
MFP-3D AFM in non-contact (tapping) mode. AFM scans were
performed in air. A rotated, monolithic silicon tip (Budget-
Sensors) was used.
Results and discussion
Development of peptide derivatized self-assembled
monolayers on gold surfaces
Infrared spectra (IR)
Peptides were synthesized manually using standard Fmoc solid-
IR spectra was collected on a Perkin Elmer 100 serial FTIR phase methodology and puried via HPLC (Fig. S1†). We chose
spectrophotometer calibrated with polystyrene lm.
to cap our peptides with an alkyne-containing capping reagent
to yield a terminal alkyne that is amenable to bioconjugation to
an azide terminated surface. This strategy allows us to investi-
gate combinations of peptides by simply mixing stock peptide
solutions in the desired ratios.26
Mass spectrometry
Mass analysis was performed using a Bruker UltraeXtreme
time-of-ight (TOF) mass spectrometer (Bruker) in positive
reector mode. 1 mL of each spotting solution was manually
pipetted onto a small Au substrate and allowed to sit 30 min.
SAM surfaces were rinsed and attached to a custom MALDI
plate using a small spot of high vacuum grease (Dow Corning).
1 mL of a-cyano-4-hydroxycinnamic acid (HCCA, Bruker) matrix
dissolved in mixture of 1 : 2 acetonitrile (ACN) : 0.1% tri-
uoroacetic acid (TFA) in ultra pure water was spotted to the
surface of each SAM. Mass spectra were obtained at 100 mm
resolution with each spot summed from 200 laser shots. Mass
calibration was done with peptide calibration standard II
(Bruker). Spectra was analyzed with exAnalysis soware
(Bruker) and replotted with Origin (OriginLab).
Within each well of a 384-well plate, we mixed together a
stock solution of HS-C11-EG4-N3, HS-C11-EG3, a peptide or
combination of peptides capped with an alkyne reagent, and a
buffer containing Cu(I), TBTA, and sodium ascorbate to cata-
lyze the cycloaddition reaction. The plate was incubated for
one hour at 37 ꢃC before printing. Using an OmniGrid
Microarrayer, the microarray tips act as features, depositing
small nanoliter scale volumes onto the surface of the gold.
Alkanethiols in the spotted solutions adsorb immediately to
the gold to form self-assembled monolayers within the printed
circular regions (Fig. 1). The substrate was subsequently
washed and immersed in a background solution of HS-C11-
EG3 to render the non-spotted regions inert to protein
adsorption.
X-ray photoelectron spectroscopy (XPS)
FTIR was used to monitor the completion of the peptide
XPS data were acquired using a Kratos Axis ULTRA Imaging X- conjugation to the HS-C11-EG4-N3 in solution (Fig. 2) prior to
ray Photoelectron Spectrometer (Kratos Analytical Ltd.) with a spotting. Aer 1 h at 37 ꢃC we observed that the azide peak at l
monochromatic Al Ka source (1486.6 eV). Spectra was obtained 2110 cmꢁ1 completely disappears, demonstrating fast conju-
in spectroscopy mode with a base pressure below 1 ꢂ 10ꢁ8 Torr. gation of the alkyne-terminated peptide with the azide-termi-
The take-off angle was set to 90ꢃ, and an analyzer slot aperture nated alkanethiolate. Aer spotting the mixture on the gold
of 0.3 mm ꢂ 0.7 mm. Survey spectra were collected from 0 to substrate, XPS was used to conrm the presence of a mono-
1100 eV with a 1.0 eV step size, 100 ms dwell time and an layer as well as conjugation of peptides. The underivatized
analyzer pass energy of 160 eV. High-resolution spectra were monolayer surface containing 15% HS-C11-EG4-N3 to HS-C11-
obtained for photoelectrons emitted from C 1s, N 1s, and Au 4f EG3 shows low atomic % of nitrogen species (Fig. 3a and
with a 0.1 eV step size, a dwell time of 100 ms, and analyzer pass Table 1) as expected for the diluted azide monomer. The C 1s
energy of 40 eV. Electron binding energies were calibrated with narrow scan of this surface reveals the presence of C–C to C–O
respect to the Au 4f peak at 84.0 eV. All XPS analysis was per- peaks in percentages of 65.2% to 34.8% respectively (Fig. 3b),
formed with CasaXPS soware. Analysis of spectra involved in close agreement with an expected percentages of 65%
This journal is © The Royal Society of Chemistry 2014
J. Mater. Chem. B, 2014, 2, 4280–4288 | 4283

View Article Online
Journal of Materials Chemistry B
Paper
Fig. 1 Scheme for spotting of microarray surfaces. A 384-well plate containing mixed ratios of ECM, proteoglycan, and morphogen peptides (i)
defines the spot features on gold substrates (ii). After spotting, the surfaces are washed and backfilled with a non-fouling background layer (iii).
Fig. 2 Conjugation of alkyne-terminated peptide KPSSAPTQLN to
azide-terminated SAM assessed in solution via FTIR. A control solution
of HS-C11-EG4-N3 dissolved in H2O : ethanol : DMSO at concentra-
tion 40 mM (solid black line) shows clear azide peak at 2110 cmꢁ1
.
Fig. 3 XPS data for azide-terminating surfaces and peptide-conju-
gated surfaces: N 1s narrow scan of surfaces before (a) and after (c)
peptide conjugation; C 1s narrow scan of surfaces before (b) and after
(d) peptide conjugation.
Control peptide in H2O at concentration 40 mM (solid red line) and
conjugate after 1 hour (solid blue line) demonstrate disappearance of
azide groups (n ¼ 2110 cmꢁ1). Reactions were performed under the
same microarray printing experimental conditions.
Table 1 XPS quantification of peak areas for surfaces before and after
conjugation
to 35%. Aer cycloaddition reaction, surfaces included
noticeable increases in the N 1s narrow scan spectra (Fig. 3c),
as well as the appearance of a C]O peak in the C 1s spectra
(Fig. 3d).
Surface
Surface
Binding energy (eV)
concentration (%)
Azide SAM
Peptide SAM
C 1s
C 1s
C 1s
C 1s
C 1s
N 1s
285.3
286.9
284.8
286.3
288.4
400.0
65.2 (C–C)
34.8 (C–O)
63.0 (C–C)
30.5 (C–O/C–N)
6.5 (C]O)
Mass spectrometry of self-assembled monolayers on gold, a
technique termed self-assembled monolayer desorption ioni-
zation (SAMDI) by Mrksich and co-workers,27 is an effective
method for characterizing the products from interfacial reac-
tions on SAMs.28 To demonstrate that spotting the reaction
solution directly onto a gold slide will enable robust monolayer
formation, we simulated printing conditions by transferring
1 mL of spotting solution onto the surface of a cleaned gold two conjugated peptide-alkanethiolate species, or with the HS-
surface. Aer 30 minutes, this surface was cleaned and moun- C11-EG3 diluent. For instance, spotted KPSSAPTQLN shows a
ted on a custom plate for matrix assisted laser desorption single peptide-alkanethiolate conjugate at m/z 1562.2 as well as
ionization (MALDI) mass spectrometry. MALDI analysis reveals the disulde at 3114.2, and another disulde corresponding to
peaks corresponding to the conjugated peptide-alkanethiolate the addition of the EG3-terminated background alkanethiolate
species, as well as several disulde peaks formed either between at 1893.6 (Fig. 4a).
4284 | J. Mater. Chem. B, 2014, 2, 4280–4288
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Journal of Materials Chemistry B
substrate, in good agreement with our ellipsometry
measurements (Fig. S2†).
To assess whether the peptides are homogenously distrib-
uted when SAMs are formed, we spotted arrays of a peptide
containing a lysine residue in proximity to the distal end
(FHRRIKA). We biotinylated the free amine of the lysine with a
sulfo-NHS-biotin reagent followed by incubation in the pres-
ence of a Cy3-conjugated streptavidin. A control array of the
peptide was incubated with the Cy3-streptavidin without prior
biotinylation (Fig. 4b). We see uniform uorescence across our
SAM spots demonstrating that the peptide is evenly distributed.
However, we cannot rule out the possibility that the peptide-
conjugated alkanethiols may phase separate into domains, as
has been reported for two component SAMs.30,31 No uores-
cence is seen on peptide spots without prior biotinylation,
highlighting the ability of the SAMs at resisting non-specic
protein adsorption.32
Previous work by Chelmowski et al. demonstrated that click
chemistry is well suited for the production of peptide-based
SAMs due to its high selectivity and tolerance for all functional
side groups in peptides.32 Furthermore, they showed that per-
forming the click reaction in solution is more efficient than
conjugation to a pre-formed azide or alkyne-terminated SAM.33
Our strategy provides great exibility over the peptide or peptide
combination presented at each spot without the need to
generate entire peptide-conjugated alkanethiol molecules. By
forming peptide-conjugates in solution, we avoid complications
involving steric effects of bulky molecules assembling on the
surface. In this way, any small molecule or ligand containing an
alkyne group can be tethered to the microarray surface.
Fig. 4 Characteristic SAMDI spectra for peptide KPSSAPTQLN. We
generally observed up to three peaks for each peptide conjugated SAM
(a). Species (i) corresponds to the conjugated peptide species, while
species (ii) and (iii) represent the disulfides formed with either the HS-
C11-EG3 background, or two peptide conjugated alkanethiolates
respectively. Spotted regions of a peptide (FHRRIKA) containing a
lysine adjacent to the distal end of the SAM (b) is biotinylated with a
biotin-NHS reagent. After incubation in the presence of streptavidin-
Cy3 (SA) for 30 min at room temperature, we see uniform fluorescence
across the peptide spots. Non-biotinylated spotted regions incubated
with SA show no fluorescence.
Cell adhesion and spreading on peptide microarrays
To optimize our spot size and reproducibility, we experimented
with a number of buffer solutions and incubation conditions
for the generation of well-dened arrays. We initially found that
the mixed buffer containing H2O, ethanol, and DMSO evapo-
rated rapidly within 1 minute aer spotting, giving poor spot
delity aer cell seeding. We judged spot delity by how well
the cells adhered and remain localized in circular spots of
peptides known to mediate adhesion (YIGSR). To improve spot
delity, we investigated various buffer additives, incubation and
To further verify that we are forming SAMs using the cleaning methods. We experimented with a number of additives
microarraying strategy, we performed ellipsometry and and found that a small amount of glycerol (3–10%) greatly
atomic force microscopy on microarray spots. The ellipso- improved spot delity. Substrate and incubation time also had
metric thickness of a SAM containing just HS-C11-EG3 was large effects on the microarray spots. Long incubation times
measured to be approximately 1.22 nm. This thickness post-spotting generally lead to non-specic adhesion and spots
increased to 1.49 nm for the monolayer surface containing with low denition, possibly due to spreading of the alkanethiol
15% HS-C11-EG4-N3 to HS-C11-EG3, and 1.66 nm when a spots. We observed optimal results with low concentrations of
peptide (KPSSAPTQLN) was conjugated (Table S1†). The glycerol (3–10%) with spotted surfaces washed immediately
thickness was calculated assuming a parallel, homogeneous, (Fig. 5). Our trials with various spotting techniques indicate the
two-layer (monolayer, substrate) model with an assumed importance of substrate cleanliness in the formation of well-
refractive index of 1.4 for the organic layer.25 Our observed dened spot sizes. Minimal exposure to atmospheric conditions
thicknesses are comparable to those of previously reported as well as minimal disruption to the deposited gold improved
SAMs of similar structure.25,29 AFM measurements at the spot size and homogeneity. However, long post-spotting incu-
edge of one of the SAM spots show a height difference of bation times even in dessicator and low vacuum conditions
approximately 1.5 nm between the formed SAM and the gold appeared to favor breakdown of well-dened microarray spots.
This journal is © The Royal Society of Chemistry 2014
J. Mater. Chem. B, 2014, 2, 4280–4288 | 4285

View Article Online
Journal of Materials Chemistry B
Paper
the spot occupied by a given cell type (Fig. S4†). For ADSCs we
see similar levels of surface coverage on YIGSR and KRSR
surfaces, as well as on all combinations (Fig. 5a and c). Inter-
estingly, KPSS surfaces, which have been shown to enhance
osteoblast spreading,36 had low levels of adhesion of ADSCs,
while KRSR, also shown to promote osteoblast attachment,37
had high levels of ADSC coverage. Even though both peptides
may promote adhesion of osteoblasts, the difference in
spreading of ADSCs may suggest that they bind to osteoblasts
via different mechanisms. For MEFs we observed relatively high
coverage on KPSS peptides, and very low adherence to KRSR
peptide (Fig. 5b). The combination of YIGSR and KPSS dis-
played the highest surface coverage, while a combination of
KPSS and KRSR displayed signicantly lower coverage (ꢀ50% of
coverage on KPSS). The combination of all three peptides
had coverage similar to just KPSS. We reason that these differ-
ences in surface coverage are due to approximately one-half and
one-third of the surface occupied by the KRSR peptide respec-
tively, which does not support broblast adhesion.37 This is in
contrast to ADSCs, in which KRSR peptide and combinations
containing this peptide show high surface coverage. Also unlike
ADSCs, MEFs seem to much more readily adhere to surfaces
presenting the BMP7 peptide. Importantly, we spotted a
scrambled peptide GRDGS that showed no cell adhesion for
either cell types (Fig. 5), indicating that adhesion characteristics
are based on the dened peptide ligand rather than non-specic
protein adsorption. These results demonstrate that this micro-
array platform can be a useful tool to screen the adhesion
properties of various cell types to different peptide ligands.
The contrasting spreading properties of ADSCs and MEFs on
KPSS and KRSR peptides suggest that these peptides bind to
different cellular receptors that are differentially regulated
between these cell types.
Fig. 5 Brightfield images and quantification of % surface coverage of
ADSC (a) and MEF (b) on various peptide spots. Scale bar ¼ 200 mm.
*p-value < 0.05, **p-value < 0.01, ***p-value < 0.001 by one-way
ANOVA.
ADSC differentiation on combinatorial peptide surfaces
This may be related to non-specic spreading of alkanethiols to BMP-7 has been successfully tested in animal models and
the intervening background regions.
clinical trials for bone fusion and regeneration.38,39 We
The synthesis of novel biomimetic materials requires an hypothesized that a surface presenting a BMP-7 derived peptide
understanding of the complex components of the extracellular may preferentially promote multipotent ADSCs to differentiate
matrix (ECM) and their interactions with one another. In a into bone. Since we determined based on our microarray assay
recent review, Hudalla and Murphy highlight strategies wherein that KPSS only weakly supports ADSC adhesion, we chose to
growth factors and glycosaminoglycans can be incorporated examine cell adhesion peptides GRGDS12 and YIGSR, and their
into biomaterials to regulate stem cell behavior.34 We used our combination with BMP-7 peptide KPSS. To assess differences in
peptide microarrays to screen cell adhesion characteristics on differentiation for ADSCs cultured on these various peptide
combinations of an adhesion peptide YIGSR,35 a growth factor combinations for extended times, we cultured cells on ꢀ1 cm2
peptide derived from bone morphogenetic protein 7 (BMP-7),36 peptide conjugated surfaces for 21 days in basal and osteogenic
and a heparan sulfate binding peptide (KRSR) known to media. Cells were subsequently xed and stained for runt-
promote attachment of osteoblasts.37 We spotted each of these related transcription factor 2 (Runx2) (Fig. 6a), a master tran-
peptides individually as well as in combination and examined scription factor associated with osteogenic differentiation, and
the adhesion characteristics of human adipose derived stem osteopontin (OPN) (Fig. 6b), a later stage osteogenic marker. We
cells (ADSCs) as well as mouse embryonic broblasts (MEFs) observed low levels of adhesion on KPSS peptide surfaces, in
aer 3 days in culture. Cells displayed differential adhesion agreement with our microarray screen. However in basal media
characteristics depending on the peptide ligands they were conditions, adherent cells on these surfaces had signicantly
seeded on and remained conned in spots for over 10 days in higher intensity of Runx2 (Fig. 6c) and OPN (Fig. 6d) compared
culture (Fig. S3†). We quantied “% surface coverage” within to the adhesion peptides GRGDS and YIGSR. Combinations of
each spot, where surface coverage is dened by the fraction of GRGDS with KPSS and YIGSR with KPSS also had signicantly
4286 | J. Mater. Chem. B, 2014, 2, 4280–4288
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Journal of Materials Chemistry B
to that of Zouani et al., who graed RGD peptides on poly(-
ethylene terephthalate) with biomimetic peptides derived from
the knuckle epitope of BMP-2,7, and 9 and saw upregulation of
Runx2 and production of extracellular matrix.45 He and
colleagues reported similar results using a BMP-2 derived peptide
in conjunction with RGD on a hydrogel substrate.46 Biomaterials
and drug delivery systems incorporating BMPs appear to be a
promising way to improve bone formation while reducing costs
of large-scale production of recombinant BMPs.47,48 Limitations
to these materials involve determining optimal density of
peptides, as well as preventing non-specic adsorption of
proteins.47 Our self-assembled monolayer based microarray
platform allows us to address both these issues. This approach
allows us to investigate the specic effects of ligand combina-
tions, while giving us exibility to easily tune ligand density ratios
and compositions. For instance, Kilian and Mrksich investigated
the effect of ligand density on mesenchymal stromal cells. They
reported that high affinity ligand promoted osteogenic fate
decisions, while low affinity ligands could promote either muscle
or neural differentiation based on ligand density.49 Rezania and
Healy also examined the importance of ligand density by
demonstrating that adhesion and differentiation on mixed
peptide surfaces are dependent on cell type as well as the ratio of
peptide ligands. They investigated an RGD peptide with a
heparin-binding peptide FHRRIKA in 25 : 75, 50 : 50, and 75 : 25
ratios and found a 50 : 50 or 75 : 25 ratio of RGD to FHRRIKA
promoted greater spreading and mineralization for RCO cells.50
The importance of peptide ligand density when presented in
combination has been further illuminated by Murphy and
colleagues who demonstrated that growth factor binding
peptides can be agonists or antagonists depending on the density
of presentation.51 With the platform presented here, researchers
Fig. 6 Immunofluorescence images of ADSCs cultured 3 weeks on
peptide combinations. Green ¼ actin, blue ¼ nuclei, red ¼ Runx2 (a) or
OPN (b). Scale bar ¼ 200 mm. In basal media, quantification of Runx2
mean intensity (c) and OPN mean intensity (d) between the peptide
surfaces. (*) denotes significant difference between groups (p < 0.05).
In osteogenic media, the differences in Runx2 (e) and OPN (f)
expression between the surfaces were not significant.
higher expression of Runx2 and OPN than the adhesion peptide may tune ligand affinity and density, as well as investigating
alone. In osteogenic media, we observed no signicant differ- effects of combinations of ligands in different ratios.
ences in either Runx2 (Fig. 6e) or OPN (Fig. 6f) on these various
peptide surfaces. Classically, osteogenic differentiation in vitro
is performed using supplemented media containing a milieu of
Conclusions
pro-osteogenic factors, especially dexamethasone, beta-glycer- Our results demonstrate a simple procedure to form peptide
ophosphate, ascorbic acid, and other growth factors.40 We microarrays in a single-step using the robust azide–alkyne
hypothesize that the effects of these factors in the osteogenic cycloaddition reaction. Cells seeded onto the spotted regions
media condition abrogated any differences that the peptide remain localized for over a week, allowing for the use of this
surfaces may have had on cultured cells.
method for long-term culture experiments. We demonstrated
BMP-7 has been shown to upregulate Runx2 and promote an the utility of this platform for cell biology investigations by
osteogenic phenotype in precursor cells.41,42 A study by Knip- exploring a combination of an adhesion peptide and BMP-7
penberg et al. using BMP-2 and BMP-7 and adipose derived derived peptide combination and found that the presence of the
stem cells found BMP-7 actually downregulated osteogenesis BMP-7 peptide promoted the expression of osteogenic markers
and stimulated a chondrogenic phenotype.43 However, the role Runx2 and OPN. We anticipate this technique will be useful in
of BMP-7 in promoting both chondrogenesis and osteogenesis the future for screening many different growth factor and
is documented and is dependent upon cell, media, and culture adhesion peptides, at variable densities and in combination, for
conditions.44 Using biomimetic peptides rather than entire a variety of cell types.
bone morphogenetic proteins may provide a more controlled
condition to investigate ligand effects on cell fate since proteins
may have multiple competing interaction sites as well as tertiary
Acknowledgements
structures. We observed elevated levels of Runx2 and OPN This work was supported by the American Cancer Society, Illi-
expression on BMP-7 derived KPSS peptide alone and in nois Division Grant #281225. We gratefully acknowledge
combination with bronectin and laminin derived adhesion Michael Sun for assistance in generating SAM surfaces, and Dr
peptides GRGDS and YIGSR (Fig. 6c and d). This result is similar Jianjun Cheng for assistance with FTIR. XPS, Ellipsometry, and
This journal is © The Royal Society of Chemistry 2014
J. Mater. Chem. B, 2014, 2, 4280–4288 | 4287

View Article Online
Journal of Materials Chemistry B
Paper
AFM was carried out in the Frederick Seitz Materials Research 24 W. Luo, E. W. L. Chan and M. N. Yousaf, J. Am. Chem. Soc.,
Laboratory Central Facilities, University of Illinois. We thank Dr 2010, 132, 2614–2621.
Rick Haasch in the center for Microanalysis of Materials at the 25 C. Pale-Grosdemange, E. S. Simon, K. L. Prime and
Frederick Seitz Materials Research lab at UIUC for assistance G. M. Whitesides, J. Am. Chem. Soc., 1991, 113, 12–20.
with XPS measurements, Amr Abdeen for assistance with AFM 26 S. Ciampi, T. Bocking, K. A. Kilian, M. James, J. B. Harper
measurements, and Neil Kruger for assistance with ellipsometry and J. J. Gooding, Langmuir, 2007, 23, 9320–9329.
¨
measurements. The Bruker UltraeXtreme MALDI TOFTOF 27 M. Mrksich, ACS Nano, 2008, 2, 7–18.
mass spectrometer was purchased in part with a grant from the 28 J. Su and M. Mrksich, Langmuir, 2003, 19, 4867–4870.
National Center for Research Resources, National Institutes of 29 B. Zhu, T. Eurell, R. Gunawan and D. Leckband, J. Biomed.
Health (S10 RR027109 A).
Mater. Res., 2001, 56, 406–416.
30 S. J. Stranick, A. N. Parikh, Y.-T. Tao, D. L. Allara and
P. S. Weiss, J. Phys. Chem., 1994, 98, 7636–7646.
31 R. K. Smith, P. A. Lewis and P. S. Weiss, Prog. Surf. Sci., 2004,
75, 1–68.
Notes and references
1 A. J. Friedenstein, R. K. Chailakhlan, K. S. Lalykina and
R. K. Chailakhjan, Cell Tissue Kinet., 1970, 3, 393–403.
2 M. F. Pittenger, A. M. Mackay, S. C. Beck, R. K. Jaiswal,
R. Douglas, J. D. Mosca, M. A. Moorman, D. W. Simonetti,
S. Craig and D. R. Marshak, Science, 1999, 284, 143–147.
3 F. P. Barry and J. M. Murphy, Int. J. Biochem. Cell Biol., 2004,
36, 568–584.
¨
32 R. Chelmowski, S. D. Koster, A. Kerstan, A. Prekelt,
C. Grunwald, T. Winkler, N. Metzler-Nolte, A. Terfort and
¨
C. Woll, J. Am. Chem. Soc., 2008, 130, 14952–14953.
¨
¨
33 R. Chelmowski, D. Kafer, S. D. Koster, T. Klasen, T. Winkler,
¨
A. Terfort, N. Metzler-Nolte and C. Woll, Langmuir, 2009, 25,
11480–11485.
4 A. I. Caplan, J. Pathol., 2009, 217, 318–324.
5 L. da Silva Meirelles, P. C. Chagastelles and N. B. Nardi, J.
Cell Sci., 2006, 119, 2204–2213.
34 G. A. Hudalla and W. L. Murphy, Adv. Funct. Mater., 2011, 21,
1754–1768.
35 J. Graf, R. C. Ogle, F. A. Robey, M. Sasaki, G. R. Martin,
Y. Yamada and H. K. Kleinman, Biochemistry, 1987, 26,
6896–6900.
¨
¨
6 A. Schaffler and C. Buchler, Stem Cells, 2007, 25, 818–827.
7 H. Mizuno, M. Tobita and A. C. Uysal, Stem Cells, 2012, 30,
804–810.
36 Y. Chen and T. J. Webster, J. Biomed. Mater. Res., Part A, 2009,
91, 296–304.
8 P. A. Zuk, M. Zhu, H. Mizuno, J. Huang, J. W. Futrell,
A. J. Katz, P. Benhaim, H. P. Lorenz and M. H. Hedrick, 37 K. C. Dee, T. T. Andersen and R. Bizios, J. Biomed. Mater. Res.,
Tissue Eng., 2001, 7, 211–228.
1998, 40, 371–377.
9 H. Shin, S. Jo and A. G. Mikos, Biomaterials, 2003, 24, 4353– 38 A. P. White, A. R. Vaccaro, J. A. Hall, P. G. Whang, B. C. Friel
4364. and M. D. McKee, Int. Orthop., 2007, 31, 735–741.
10 D. G. Anderson, S. Levenberg and R. Langer, Nat. Biotechnol., 39 P. Simic and S. Vukicevic, EMBO Rep., 2007, 8, 327–331.
2004, 22, 863–866.
11 C. J. Flaim, S. Chien and S. N. Bhatia, Nat. Methods, 2005, 2,
40 C. Vater, P. Kasten and M. Stiehler, Acta Biomater., 2011, 7,
463–477.
119–125.
41 P. Ducy, R. Zhang, V. Geoffroy, A. L. Ridall and G. Karsenty,
Cell, 1997, 89, 747–754.
12 M. D. Pierschbacher and E. Ruoslahti, Nature, 1984, 309, 30–
33.
42 K. Gu, L. Zhang, T. Jin and R. B. Rutherford, Cells Tissues
Organs, 2004, 176, 28–40.
13 E. Ruoslahti, Annu. Rev. Cell Dev. Biol., 1996, 12, 697–715.
14 G. A. Hudalla and W. L. Murphy, Langmuir, 2009, 25, 5737– 43 M. Knippenberg, M. N. Helder, B. Zandieh Doulabi,
5746.
P. I. J. M. Wuisman and J. Klein-Nulend, Biochem. Biophys.
Res. Commun., 2006, 342, 902–908.
44 B. Shen, A. Wei, S. Whittaker, L. A. Williams, H. Tao,
D. D. F. Ma and A. D. Diwan, J. Cell. Biochem., 2010, 109,
406–416.
15 S. P. Massia and J. A. Hubbell, J. Cell Biol., 1991, 114, 1089–
1100.
16 J. C. Love, L. A. Estroff, J. K. Kriebel, R. G. Nuzzo and
G. M. Whitesides, Chem. Rev., 2005, 105, 1103–1169.
17 B. T. Houseman and M. Mrksich, Biomaterials, 2001, 22, 45 O. F. Zouani, C. Chollet, B. Guillotin and M.-C. Durrieu,
943–955. Biomaterials, 2010, 31, 8245–8253.
18 G. A. Hudalla and W. L. Murphy, Langmuir, 2010, 26, 6449–6456. 46 X. He, J. Ma and E. Jabbari, Langmuir, 2008, 24, 12508–12516.
19 M. P. Lutolf, P. M. Gilbert and H. M. Blau, Nature, 2009, 462, 47 H. Senta, H. Park, E. Bergeron, O. Drevelle, D. Fong,
433–441.
E. Leblanc, F. Cabana, S. Roux, G. Grenier and
N. Faucheux, Cytokine Growth Factor Rev., 2009, 20, 213–222.
48 H. Senta, E. Bergeron, O. Drevelle, H. Park and N. Faucheux,
Can. J. Chem. Eng., 2011, 89, 227–239.
49 K. A. Kilian and M. Mrksich, Angew. Chem., Int. Ed., 2012, 51,
4891–4895.
20 S. Gobaa, S. Hoehnel, M. Roccio, A. Negro, S. Kobel and
M. P. Lutolf, Nat. Methods, 2011, 8, 949–955.
21 B. P. Orner, R. Derda, R. L. Lewis, J. A. Thomson and
L. L. Kiessling, J. Am. Chem. Soc., 2004, 126, 10808–10809.
´
22 R. J. Giordano, M. Cardo-Vila, J. Lahdenranta, R. Pasqualini
and W. Arap, Nat. Med., 2001, 7, 1249–1253.
50 A. Rezania and K. E. Healy, Biotechnol. Prog., 1999, 15, 19–32.
23 R. Derda, S. Musah, B. P. Orner, J. R. Klim, L. Li and 51 J. T. Koepsel, E. H. Nguyen and W. L. Murphy, Integr. Biol.,
L. L. Kiessling, J. Am. Chem. Soc., 2010, 132, 1289–1295.
2012, 4, 914–924.
4288 | J. Mater. Chem. B, 2014, 2, 4280–4288
This journal is © The Royal Society of Chemistry 2014