Acta Biomaterialia 86 (2019) 158–170

Contents lists available at ScienceDirect

Acta Biomaterialia
journal homepage: www.elsevier.com/locate/actabiomat

Full length article

Cellular response to collagen-elastin composite materials

Daniel V. Bax a,b,1,⇑, Helen E. Smalley a,1, Richard W. Farndale b, Serena M. Best a,
Ruth E. Cameron a
a
Department of Materials Science and Metallurgy, University of Cambridge, 27 Charles Babbage Road, Cambridge CB3 0FS, United Kingdom
b
Department of Biochemistry, University of Cambridge, Downing Site, Cambridge CB2 1QW, United Kingdom

a r t i c l e i n f o a b s t r a c t

Article history: Collagen is used extensively in tissue engineering due to its biocompatibility, near-universal tissue dis-
Received 13 August 2018 tribution, low cost and purity. However, native tissues are composites that include diverse extracellular
Received in revised form 3 December 2018 matrix components, which influence strongly their mechanical and biological properties. Here, we pro-
Accepted 21 December 2018
vide important new findings on the differential regulation, by collagen and elastin, of the bio-response
Available online 23 December 2018
to the composite material. Soluble and insoluble elastin had differing effects on the stiffness and failure
strength of the composite films. We established that Rugli cells bind elastin via EDTA-sensitive receptors,
Keywords:
whilst HT1080 cells do not. These cells allowed us to probe the contribution of collagen alone (HT1080)
Collagen
Elastin
and collagen plus elastin (Rugli) to the cellular response. In the presence of elastin, Rugli cell attachment,
Cellular response spreading and proliferation increased, presumably through elastin-binding receptors. By comparison, the
attachment and spreading of HT1080 cells was modified by elastin inclusion, but without affecting their
proliferation, indicating indirect modulation by elastin of the response of cells to collagen. These new
insights highlight that access to elastin dominates the cellular response when elastin-binding receptors
are present. In the absence of these receptors, modification of the collagen component and/or physical
properties dictate the cellular response. Therefore, we can attribute the contribution of each constituent
on the ultimate bioactivity of heterogeneous collagen-composite materials, permitting informed, system-
atic biomaterials design.

Statement of Significance

In recent years there has been a desire to replicate the complex extracellular matrix composition of tis-
sues more closely, necessitating the need for composite protein-based materials. In this case both the
physical and biochemical properties are altered with the addition of each component, with potential con-
sequences on the cell. To date, the different contributions of each component have not been deconvolved,
and instead the cell response to the scaffold as a whole has been observed. Instead, here, we have used
specific cell lines, that are sensitive to specific components of an elastin-collagen composite, to resolve
the bio-activity of each protein. This has shown that elastin-induced alteration of the collagen component
can modulate early stage cell behaviour. By comparison the elastin component directly alters the cell
response over the short and long term, but only where appropriate receptors are present on the cell.
Due to the widespread use of collagen and elastin, we feel that this data permits, for the first time, the
ability to systematically design collagen-composite materials to promote desired cell behaviour with
associated advantages for biomaterials fabrication.
Ó 2019 Acta Materialia Inc. Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).

1. Introduction

Purified extracellular matrix (ECM) components are often uti-

lised for scaffold fabrication in tissue engineering applications as
⇑ Corresponding author at: Department of Materials Science and Metallurgy, they provide both physical and biochemical cues to guide tissue
University of Cambridge, 27 Charles Babbage Road, Cambridge CB3 0FS, United development [1]. As such they can recreate the cellular environ-
Kingdom. ment of native tissue. The ECM of tissue is comprised of a complex
E-mail address: dvb24@cam.ac.uk (D.V. Bax). organisation of macromolecules such as collagen (CN), laminin,
1
These authors contributed equally.

https://doi.org/10.1016/j.actbio.2018.12.033
1742-7061/Ó 2019 Acta Materialia Inc. Published by Elsevier Ltd.
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
 D.V. Bax et al. / Acta Biomaterialia 86 (2019) 158–170 159

elastin, and glycosaminoglycans (GAGs). Fibrillar CN type I is the tin addition. This discrepancy is presumably due to the different
most abundant of these ECM proteins comprising
30% of the total assay parameters utilised for each study [33]. For example short
protein mass in humans [2], where it provides strength, stiffness term studies on fibroblasts and myoblasts showed little effect of
and cell ligating motifs. Additionally isolated CN possesses appro- IE addition to CN [7,18] whilst others have observed decreased cell
priate biocompatibility, biodegradability, purity and cost which proliferation [34,35]. Similarly vascular smooth muscle cell prolif-
makes it an ideal candidate as a scaffold precursor material [1]. eration has been inhibited [31] or promoted [36] with IE addition,
Chemical or physical crosslinkers are frequently used to although the overall consensus is that addition of elastin reduces
enhance the physical integrity of CN-based biomaterials. Of these, SMC proliferation [37]. Likewise, SE has been shown to increase
the most commonly used employs 1-ethyl-3-(3-dimethylaminopro fibroblast proliferation [38] whilst others observe selective stimu-
pyl)-carbodiimide hydrochloride (EDC) in the presence of N- lation of keratinocyte but not fibroblast proliferation [39]. We
hydroxysuccinimide (NHS), as any cytotoxic reagents and products hypothesise that this wide range of responses is due to cell-type
are simply removed by washing [3]. This allows discrete control specific ligation with the CN and elastin components of these com-
over the mechanical properties and degradation kinetics of the posites. In particular, we note that the relative contribution of the
CN-based material [3–5]. Evaluation of thermal shrinkage and pri- CN and elastin components to the cellular response, and the time-
mary amine content has led to the commonly used ratio of 5 EDC:2 scale over which these effects occur, is unclear. Therefore, the aim
NHS:1 COOH group on CN in a 75% ethanol solution for 2–4 h as an of this study was to understand the cell-biological response to the
optimal condition. We have chosen to use this ratio here as it yields CN and elastin components in IE- and SE-CN composite materials.
a maximal crosslinking density [4,6–8]. EDC cross-links proteins To undertake this analysis we utilised two different model cell
through carboxylic acid groups on aspartic and glutamic acid resi- lines, rat glioma cells (Ruglis) and human fibrosarcoma cells
dues and neighbouring primary amine groups, however this can be (HT1080s), which exhibited differential binding properties to elas-
detrimental to cell interaction [9]. In particular native-like cell tin, allowing us to delineate the cellular behaviour in the presence
adhesion through cell-surface heterodimeric, transmembrane inte- and absence of elastin ligation by the cell. This has enabled us to
grins is lost, which is replaced with non-native interactions that do probe the specific bio-activity of CN-elastin composites over bio-
not support cell proliferation [9]. Of the 24 different integrin pairs logically important time frames, illustrating the importance of
[10], CN associates with integrins a1b1, a2b1, a10b1, and a11b1, via the elastin-engagement in this response.
an inserted A domain (I domain) within the a subunit of the inte-
grin [11,12]. These I domains bind to the consensus Gxx’GEx” (sin- 2. Materials and methods
gle amino acid code) sequences in CN [13–16]. The critical
dependence on the carboxylic group contained on the glutamic 2.1. Materials
acid (E) residue [17] of this sequence gives the intriguing hypoth-
esis that chemical crosslinking of CN competes with integrin adhe- Unless stated otherwise all reagents were purchased from
sion for the same chemical groups [9,18], necessitating Sigma Aldrich UK. 4% v/v acetic acid was purchased from Alfa
refunctionalisation of CN films after cross-linking [19]. Aesar. Silicone moulds for producing films were purchased from
Despite the recent interest in CN-based materials, it is ambigu- Lakeland Ltd. (Windermere, UK).
ous whether materials composed of CN-only adequately recapitu-
late the complex mechanical and biochemical properties of native
tissue. Therefore, inclusion of additional ECM components, such as 2.2. Amino acid analysis
elastin, offers the opportunity to tailor the bio-response. Elastin
represents an attractive additive as it is provides elastic recoil to Approximately 1 mg samples were subjected to amino acid
a range of tissues [20] and interacts with cells via a number of cell analysis (Department of Biochemistry, University of Cambridge,
surface receptors. These include the elastin-binding protein (EBP) UK). The percentage content of each amino acid was determined
[21,22], cell surface heparan and chondroitin sulphate-containing by correcting against a norleucine standard and then dividing the
glycosaminoglycans (GAGs) [23] and cell surface integrins aVb3 measured mmole of each amino acid/mg of sample by the total
and aVb5 [24–26]. As such, elastin has been examined for biomate- mmole of all measured amino acids. Cystine and tryptophan were
rials fabrication. To date a variety of elastin precursors have been not measured, however they are absent from CN and represent
employed, including native elastin, insoluble elastin (IE), soluble trace amino acids in elastin. The predicted amino acid content
elastin (SE), tropoelastin and synthetic elastin-like polypeptides was derived from the primary amino acid sequence of bovine elas-
(ELPs) [27]. When fabricated solely from purified elastin, for exam- tin (UniProtKB-P04985), bovine CN alpha I (I) (UniProtKB-P02453)
ple through electrospinning [28], or by decellularising natural tis- and bovine CN alpha 2 (I) (UniProtKB-P02465).
sues [29], elastin-based materials interact with a range of cell
types, exhibit non-thrombogenic potential and are remodelled 2.3. Slurry preparation
in vivo [30]. As for materials fabricated solely from CN, these
elastin-only materials do not fully replicate the complex mechan- 1 wt% suspensions of bovine insoluble Achilles tendon CN,
ics and biochemical attributes of tissues. Therefore, elastin has insoluble bovine neck ligament elastin (IE) and soluble bovine neck
been incorporated into composites containing other ECM proteins, ligament elastin (E6527 [40] – SE) were swollen in 0.05 M acetic
in particular CN I, offering huge potential as they possess complex acid overnight at 4 °C. These were homogenised in a beaker on
mechanical properties that are similar to native tissue. For example ice using a VWR VDI25 homogeniser for 5 min at 6500 rpm fol-
inclusion of elastin into CN-based materials reduces the specific lowed by 25 min at 13500 rpm for CN, for 5 min at 6500 rpm fol-
tensile and compressive moduli [31]. To further recreate the fibril- lowed by 10 min at 13500 rpm for IE, and for 5 min at 6500 rpm
lar tissue architecture, elastin and CN have been simultaneously followed by 5 min at 13500 rpm for SE. Mixed compositions were
electrospun to produce co-deposited fibrous materials [32], how- obtained by manually mixing appropriate volumes of the 1 wt%
ever, to reduce complexity, here we have explored amorphous stock suspensions. All of the protein suspensions had a final con-
materials of insoluble CN and elastin. centration of 1 wt%. Suspensions that did not contain IE were
Although the physical properties of CN-elastin composite mate- degassed by centrifuging for 5 min at 3250 g (2500 rpm) in a
rials are well established [31], there is conflicting biochemical data Hermle Z300 centrifuge. Suspensions containing IE were degassed
showing both increased and decreased cell proliferation upon elas- for 2 min at 14 mTorr in a VirTis advantage freeze-dryer to prevent
160 D.V. Bax et al. / Acta Biomaterialia 86 (2019) 158–170

sedimentation of the IE component. We define SE alone as a solu- applied force in N, A is the sample cross-sectional area in m2 and e
tion and SE in combination with insoluble CN as a suspension. is the strain of the material. The failure stress and strain were cal-
culated from the highest force that was exerted on the sample
2.4. Film preparation before it failed and its equivalent strain.

F fail
Films for SEM and mechanical testing were cast in 40 mm diam- rfail ¼ ð1Þ
eter circular silicone moulds. 1 mL of suspension was added to A
each mould, and any bubbles were removed. Films for cell adhe-
sion and proliferation analysis were prepared by casting 400 lL
eðin%Þ
eðdimensionlessÞ ¼ ð2Þ
or 200 lL of slurry into wells of 24- and 48-well plates respec- 100
tively. Films for cell imaging were cast by applying 50 lL protein
The stiffness was obtained by plotting a line of best fit onto the
slurry onto the centre of 13 mm diameter glass coverslips, placed
high strain linear portion of the force-strain curve. The stiffness, E
in a 24 well plate. All samples were dried in a fume hood for 2 days.
in Pa, was derived from the gradient using Eq. (3), where m is the
Films were crosslinked using a solution containing a molar ratio of
gradient in N%-1 of the resultant force vs strain graph.
5 EDC:2 NHS in 70% ethanol with a final EDC concentration of
12 mM. 800 lL, 400 lL or 5 mL of crosslinking solution was added
100m
to each sample in 24-well plates, 48-well plates or 40 mm silicone E¼ ð3Þ
moulds respectively. The samples were covered and incubated at A
room temperature on a rocker at 50 rpm for 2 h. Once cross-
To account for noise, average values from 5 force values
linking was completed the samples were washed for 5  5 min in
were calculated, and each force value was set to an initial value
H2O whilst rocking at 50 rpm. They were then dried in a fume hood
of zero.
for 48 h.

2.5. Enzyme-linked immunosorbent analysis (ELISA) 2.8. Cell culture

Films were washed with 3  1 mL aliquots of phosphate buf- HT1080 cells, derived from a human fibrosarcoma, were
fered saline (PBS) then blocked with 2% (w/v) bovine serum albu- obtained from the European Collection of Animal Cell Cultures,
min (BSA) in wash buffer (0.1%(v/v) Tween-20, 1 mg/mL BSA in Porton Down, UK. Rugli cells, derived from a rat glioma, were from
PBS) for 1 h at room temperature. The samples were washed with Dr. J. Gavrilovic, University of East Anglia, UK. All cell lines were
3  1 mL wash buffer and then incubated in 0.75 mL of 1:2000 cultured on tissue culture plastic flasks maintained in a humidified
diluted mouse anti-elastin antibody (clone BA-4) in wash buffer incubator with 5% CO2 at 37 °C in Dulbecco’s modified Eagle’s med-
for 1 h at room temperature. The antibody was aspirated, and the ium (DMEM) containing 10% (v/v) fetal bovine serum and 1% (v/v)
samples were washed in 3  1 mL wash buffer before incubation streptomycin/penicillin (complete media). Once 70–80% confluent
in 0.75 mL of 1:10000 diluted goat anti-mouse IgG-HRP conjugated the cells were passaged into new flasks at 1/10th of the original cell
secondary antibody (DAKO) in wash buffer for 45 min at room density. Cells were prepared for cell adhesion, spreading or prolif-
temperature. The secondary antibody was removed, and the sam- eration analysis by detaching from the cell culture flasks with
ples washed in 4  1 mL PBS for 20 min each wash. The samples 0.05% (w/v) trypsin/0.02% (w/v) EDTA and re-suspending in an
were transferred to a new 24-well plate and 0.4 mL TMB solution appropriate volume of DMEM or complete media.
(Thermo) added for 20 min at room temperature. 100 lL aliquots
of solution were transferred to a clean 96-well plate and the absor- 2.9. Cell adhesion analysis
bance was read at 652 nm (A652) on a SPECTROstar Nano plate
reader (BMG labtech). Values represent means of quadruplicate Where stated, wells of tissue culture plates were coated with SE
measurements ± standard deviation. diluted from a 1 wt% stock solution to a final concentration of
between 1 and 20 lg/mL or with 5 lg/mL of soluble CN I (First Link
2.6. Scanning electron microscopy (SEM) (UK) Ltd.). The solutions were removed, the wells were washed 3x
with PBS then non-specific adsorption to the film or well was
Samples were attached to metal SEM stubs using double-sided blocked with 600 lL or 150 lL of 2% w/v BSA in PBS for 24- or
conductive carbon tape. They were coated with gold for 2.5 min at 96-well plates respectively. After 60 min at room temperature
20 mA using an Emitech K550 sputter coater. SEM images were the wells were washed 3x with PBS then 400 lL or 100 lL of cells
obtained using a Camscan MX 2600 FEGSEM using an accelerating were added at a density of 1x106 cells/mL in DMEM to 24- or 96-
voltage of 10 kV and magnifications of 500. well plates respectively. The plates were incubated at 37 °C/5%
CO2 for 1 h and non-adherent cells were removed with 3 X PBS
2.7. Tensile testing washes. 300 lL or 75 lL of lysis buffer (81 mM TriSodium Citrate,
31 mM Citric Acid, 0.1% v/v Triton X-100, 1.85 mg/mL p-
Films were cut into 5 mm wide strips using a scalpel. The thick- nitrophenyl phosphate (PNP) substrate, pH 5.4) was added to 24-
ness of each sample was measured using a digital micrometer. or 96-well plates respectively and incubated for 16 h at 4 °C. Either
Before extension, samples were soaked with deionised water until 200 lL or 50 lL of 2 M sodium hydroxide solution was then added
they became fully hydrated. Tensile testing was performed using a to 24- or 96-wells respectively. 100 lL of solution was transferred
Hounsfield uniaxial tester with a 5 N load cell. Each composition to a clean 96-well plate and the absorbance read at 405 nm (A405)
was tested five times. A gauge length of 15 mm and an extension on a SPECTROstar Nano plate reader (BMG labtech).
rate of 6 mm min1 were used. For inhibition assays, wells were SE or soluble CN I coated then
Strain values were measured using a laser on the testing rig. BSA blocked as for cell adhesion analysis. 12.5 mL of 40 mM EDTA
Force vs strain data were plotted to derive the stiffness at high (final concentration 5 mM), 80 mM a-lactose (final concentration
strain, failure stress and failure strain of each material (Supple- 10 mM), 80 mM b-lactose (final concentration 10 mM), and/or
mentary Fig. 1). The failure stress and failure strain were calculated 80 mg/mL heparan sulphate (final concentration 10 mg/mL) in
using Eqs. (1) and (2), where r is the applied stress in Pa, F is the DMEM were added to each well. The volume in each well was
 D.V. Bax et al. / Acta Biomaterialia 86 (2019) 158–170 161

made up to 50 mL then 50 mL of Rugli cells was added at a density of slides using Vectashield mounting media and visualised on a Zeiss
2x106 cell/mL (final density 1x106 cell/mL) for 1 h at 37 °C/5% CO2. Observer Z1 fluorescent microscope.
Loosely adherent cells were removed, and bound cells were
detected using a PNP substrate as for cell adhesion analysis. 2.13. Statistical analysis
Values are means of quadruplicate measurements ± standard
deviation. Unless otherwise stated all error bars indicate standard devia-
tions from the mean. Statistical significance was determined with
2.10. Cell spreading analysis a student t-test with unequal variance where N/S indicates >0.05,
* indicates p  0.05, ** indicates p  0.01, *** indicates p  0.001
Films and soluble CN coated wells were prepared in a 24-well and **** indicates p  0.0001.
plate then BSA blocked as for adhesion analysis. 400 lL of cells at
a concentration of 3  105 cells/mL in DMEM were added to each 3. Results
well and incubated at 37 °C/5% CO2 for 210 min for Rugli cells
and 180 min for HT1080 cells. These time points were chosen as 3.1. Film characterisation
they represent the duration required to observe spreading onto
the positive control, soluble CN I coated, tissue culture wells. Composites were generated by blending appropriate volumes of
100 lL of 25% w/v glutaraldehyde stock solution was added to each 1 %w/v CN or elastin to produce films with increasing, 25%, 50% or
well to achieve a final concentration of 5% w/v then incubated at 75%, content of soluble (SE) or insoluble (IE) elastin. ELISA analysis
room temperature for 20 min. The fixed cells were washed with using an elastin-specific antibody confirmed the presence of elas-
3  1 mL PBS then permeabilised with 600 lL of 0.5 %w/v Triton tin in these resultant materials (Fig. 1). This showed that the
X-100 in PBS for 5 min at room temperature. The samples were amount of elastin incorporated into the composite was propor-
washed 3x with PBS then 600 lL of 0.01% rhodamine phalloidin tional to the amount of elastin in the slurry. As these films were
(Molecular Probes, made up to manufacturer’s instructions) in crosslinked, this indicates that the elastin component was retained
PBS was added for 30 min at room temperature in the dark. After during EDC/NHS crosslinking in 75% ethanol. Although the absor-
washing 3x with PBS the cell nuclei were stained with 600 lL of bance measurements were above background levels for all the
3.5 lM DAPI in H2O for 2 min then washed 3x with H2O. The sam- composite materials, detection was approximately 2.5-fold higher
ples were mounted onto glass slides using Vectashield antifade for 75% SE compared to 75% IE containing composites. Therefore,
mounting medium, sealed with nail varnish, and visualised on a although the same mass of elastin was added, the degree of elastin
Zeiss Observer Z1 fluorescent microscope using a 20x magnifica- detection varied. This is presumably due to the fibrillar morphol-
tion objective lens. The cell area was obtained by thresholding ogy of IE which, effectively prevents the antibody from accessing
the rhodamine-phalloidin stained images and measuring the cell- the centre of the elastic fibre. This theory was tested by anti-
derived fluorescent area in ImageJ. The cell number for each image elastin fluorescent microscopy of the IE composite (Fig. 2). Autoflu-
was derived from the DAPI stained images by using the nucleus orescence, due to the presence of a crosslinking tricarboxylic
counter plug-in feature of ImageJ. To calculate the average cell area
the total fluorescent area for each image was divided by corre-
sponding cell number. Values represent means of 5 measure-
ments ± standard deviation.

2.11. Cell growth

Samples in 48-well plates were sterilised using ultraviolet (UV)

light for 20 min then blocked using 300 lL of filter-sterilised 2 %w/
v BSA for 1 h at room temperature. 300 lL of cells were added at a
density of 3.5x104 cells/mL in complete media to each well. Plates
were incubated at 37 °C/5% CO2 for 1, 2 or 4 days, at which point
75 lL of 25% w/v glutaraldehyde stock solution was added to each
well to achieve a final concentration of 5% w/v for 20 mins at room
temperature. The wells were washed 3x with PBS and then perme-
abilised for 5 min with 300 lL of 0.5 %w/v Triton X-100 in PBS. The
wells were washed three times with PBS and 300 lL of 3.5 lM
DAPI in H2O was added for 2 min in the dark. Following 3x H2O
washes the cell nuclei were visualised using a Zeiss Observer Z1
fluorescent microscope and a 10x objective lens. The number of
nuclei per field of view, of a known area, were counted manually
and expressed as a cell count/mm2. Values represent means of qua-
druplicate measurements ± standard deviation.

2.12. Elastin fluorescence microscopy

Films were BSA blocked as for adhesion analysis then incubated

in 1:1000 mouse anti-elastin antibody (clone BA-4, Abcam, UK) in
PBS for 1 h. The primary antibody was aspirated, washed 3xPBS Fig. 1. Enzyme linked immunosorbent analysis of the elastin content of IE- (light
grey bars) and SE- (dark grey bars) containing CN films as detected by the BA-4 anti-
then incubated in 1:500 AlexaFluor 594 conjugated donkey-anti- elastin antibody (y-axis). The composite composition in relative % composition is
mouse secondary antibody (Jackson Immuno Research) for 1 h. detailed on the x-axis. The negative control indicates a no-film reference sample
The samples were washed 3x with PBS then mounted onto glass (unfilled bar). Error bars indicate S.D. of quadruplicate measurements.
162 D.V. Bax et al. / Acta Biomaterialia 86 (2019) 158–170

Fig. 2. Phase contrast microscopy (A) and fluorescent microscopy (B–D) of 75%IE-25%CN composites. Autofluorescence of the IE fibres [41] is shown in (B) and green in the
merged image (D). BA-4 anti-elastin antibody detection is shown in (C) and red in the merged image (D). Arrows highlight peripheral antibody detection on IE fibres in (D).
The scale bar indicates 200 mm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

amino acid with pyridinium ring [41] was used to visualise the capacity. To directly test the elastin-binding profile of these cell
entire elastin inclusion. This was clearly visible throughout the lines, adhesion to SE coated onto tissue culture plastic, was exam-
entirety of the large IE inclusions (green in Fig. 2). Conversely, ined (Fig. 5). These data show a dramatic difference in the adhesive
detection by the BA-4 anti-elastin antibody was restricted to the response between these lines. Whilst SE can support Rugli cell
periphery of these inclusions (red in Fig. 2). This suggests that only adhesion at a coating concentration of 2 lg/ml, little HT1080 cell
a sub-proportion of the elastin in the IE containing scaffolds was adhesion was observed with SE coating concentrations of up to
accessible by the detection antibody which could explain the lower 20 lg/ml. Inhibition analysis using well-known elastin-receptor
IE to SE detection by ELISA for same nominal mass of elastin. SEM blocking molecules (EDTA, b-lactose and heparan sulphate for inte-
was used to image the protein films to determine how the grin, elastin binding protein and GAG-mediated adhesion respec-
microstructure was affected by the addition of SE or IE (Fig. 3). This tively) showed that Rugli cell adhesion was blocked in the
shows that films composed exclusively of CN did not contain large presence of EDTA (Fig. 5C). By contrast a-lactose alone (a low-
fibres but instead implies a roughened morphology. Addition of IE activity control for b-lactose) resulted in a slight decrease in elastin
resulted in the presence of large 5–6 lm diameter inclusions, indi- engagement, however when combined with b-lactose or heparan
cated by arrows in Fig. 3. These inclusions were not evenly dis- sulphate it did not alter the cellular response. When all four inhi-
tributed with some agglomeration apparent. By comparison, the bitors were added in combination, the degree of cell adhesion
addition of SE to films resulted in a smooth and relatively feature- was similar to EDTA inhibition alone, indicating an integrin-
less film. mediated cell binding mechanism in Rugli cells. Similarly, Rugli
Tensile testing of these films showed that all the material com- cell adhesion to soluble CN was inhibited by EDTA, but not lactose
positions possessed a J-shaped force-strain curve (example shown or heparan sulphate, highlighting an integrin-mediated response to
in Supplementary Fig. 1) which is frequently observed in a wide CN (Fig. 5D). Therefore Rugli but not HT1080 cells can bind to elas-
range of soft tissues. These showed a general trend that addition tin via integrin receptors, making comparison of these model cell
of IE decreased the material stiffness (Fig. 4A), however due to lines a convenient method to delineate the contribution of CN-
spread of values for the 100%CN samples this was not statistically only (HT1080) and CN-elastin (Rugli) in our elastin-CN composites.
significant. Conversely, SE increased the stiffness of the composites Both HT1080 and Rugli cells could adhere to films fabricated
in a concentration dependent manner with a 3.3-fold increase in solely from insoluble CN (Fig. 6) giving absorbance values of
the modulus at 50% SE content (p = 0.0008 against 100%CN). Values 0.84 ± 0.12 and 0.65 ± 0.10 respectively. It should be noted that
are not shown for 75% SE inclusion as these films were not suffi- the entire well surface was covered with the film material, ensur-
ciently robust for analysis. The failure stress was unaffected by ing that cell adhesion is solely through the film material. These
the addition of IE and increased by the addition of SE (Fig. 4B). absorbance values were higher than BSA coated control wells,
By contrast the failure strain increased with IE addition, particu- showing values of between 0.11 ± 0.02 and 0.2 ± 0.1 on a low-
larly at the highest 75% elastin content (p = 0.0002 against 100% adhesion surface. Adhesion to CN films was lower than the soluble
CN), with a failure strain of
21% (Fig. 4C). Addition of SE did not CN (Abs.405 1.61 ± 0.17 and 2.42 ± 0.12) and tissue culture plastic
influence the failure strain. It should be noted that all of these controls (Abs.405 1.5 ± 0.04 and 2.27 ± 0.11) for HT1080 and Rugli
materials were cross-linked with EDC/NHS as this is necessary cells respectively. The addition of up to 50% SE markedly increased
for stability in cell culture conditions. As cross-linking is known Rugli cell adhesion by over 2-fold. This showed significance
to change the mechanical properties of CN-based materials the (p = 0.007) over the CN-only control. Consistent with the lack of
same crosslinking parameters were utilised across all of the SE adhesion, inclusion of SE did not significantly alter the degree
samples. of HT1080 cell adhesion. Although data are shown for 75% SE inclu-
sion, these films were not fully resistant to the washing regime
3.2. Cell adhesion employed to remove non-adherent cells, and so only fragments
remained attached to the wells. As film fragmentation was a dom-
HT1080 and Rugli cells were chosen to analyse the cell binding inant effect, masking the cell-binding properties, the values from
properties of the elastin-CN composite materials as they contain these samples are shown with a dashed line and are excluded from
well defined CN-binding integrins, utilising a2b1 and a1b1 respec- analysis here. IE inclusion increased the degree of Rugli cell adhe-
tively [16,42,43], but with potentially differing elastin binding sion by
2-fold at 75% IE content. Again, this showed significance
 D.V. Bax et al. / Acta Biomaterialia 86 (2019) 158–170 163

lular engagement to promote cell spreading. Supplementary

Fig. 2A and B show representative fluorescent microscopy images
of rhodamine phalloidin stained actin (red) and DAPI stained nuclei
(blue) for HT1080 and Rugli cells respectively. For IE containing
films the IE fibres are non-specifically stained with DAPI, however
the nuclei are still apparent due to their relative intensity and cir-
cularity. Consistent with our previous reports showing that EDC/
NHS crosslinked CN does not support cell spreading [9], HT1080
cells and Rugli cells both possessed a rounded morphology on
CN-only films. Addition of IE did not affect Rugli cell spreading,
regardless of the concentration added. Conversely SE addition
resulted in the formation of thin cellular projections that were
not observed on the CN-only films. This appeared independent of
the SE concentration with a similar morphology on films composed
of all SE densities. Similarly, HT1080 cell spreading was observed
on films containing SE, which, qualitatively, appeared to be depen-
dent upon the amount of elastin present. The addition of IE did not
lead to HT1080 spreading with all cells being round at all IE con-
centrations. Both HT1080 and Rugli cells possessed a spread mor-
phology on the positive soluble CN I control and a rounded
morphology on the negative BSA control.
Quantitative analysis of the cell area on each material (Fig. 7) is
consistent with the qualitative observations. This shows that the
cell area is similar on both the CN-only films and BSA controls
for both HT1080 (205 ± 37 and 172 ± 26 mm2) and Rugli cells
(186 ± 11 and 179 ± 14 mm2), confirming a lack of spreading on
EDC/NHS crosslinked CN. Addition of SE dose dependently
increased the HT1080 cell area to 362 ± 52 mm2 at 75% SE content.
This was significantly different to the CN-only films (p = 0.00086 at
75% SE). By comparison HT1080 cells possessed a cell area of
between 157 ± 18 and 198 ± 15 mm2 on all IE containing compos-
ites. The Rugli cell area was sensitive to SE, although this was sim-
ilar for all SE densities with a cell area of between 206 ± 9 and
329 ± 57 mm2. This showed significance against the CN-only films
(p = 0.0053 at 75% SE). The Rugli cell area was insensitive to IE
addition with cell areas of between 166 ± 7 and 195 ± 11 mm2 for
all compositions. Both cells lines spread onto soluble CN I coated
glass with cell areas of 627 ± 70 and 512 ± 51 mm2 for HT1080
and Rugli cells respectively.

3.4. Cell proliferation

We have previously shown that a loss of divalent cation-

dependent cell adhesion to EDC/NHS crosslinked insoluble CN I
inhibits cell proliferation [9]. As SE and IE incorporation alters
the cell adhesive properties of CN-based materials, we have mea-
sured cell proliferation on SE and IE composites (Fig. 8). Both cell
lines could proliferate on a positive soluble CN I coated control,
with the number of cells/mm2 of material increasing
6- and

18-fold over 4 days in culture for HT1080 and Rugli cells respec-
tively. This confirms the proliferative capacity of these cells.
Consistent with our previous observations, HT1080 cells did not
proliferate on EDC/NHS crosslinked CN-only films with
40–50%
fewer cells at day 4 compared to day 1. Inclusion of IE or SE had
Fig. 3. Scanning Electron Micrographs of CN only (A), CN-IE (B) and CN-SE no significant effect on HT1080 cell proliferation over the no-
composites (C). The composition is shown as a relative % of each component.
elastin controls. For example, there were
50 and 40% fewer cells
Arrows highlight IE inclusions. Scale bar indicates 50 mm.
after 4 days in culture on 75% IE and 75% SE containing samples
respectively. Rugli cells showed low levels of cell proliferation on
(p = 0.00034) over the CN-only material. Interestingly IE inclusion CN-only samples with an approximately 4-fold increase in the cell
increased HT1080 cell adhesion by
1.6-fold, presumably due to density between day 1 and day 4 in culture. This was significantly
changes in the mechanics, roughness or surface area. below the
18-fold increase in cell number observed on the sol-
uble CN I control. Therefore, although Rugli cells showed prolifer-
3.3. Cell spreading ative capacity on EDC/NHS crosslinked CN, this was
4–5 times
lower than for non-crosslinked CN. Interestingly, inclusion of IE
A short-term cell morphological analysis was conducted to and SE could dose-dependently increase the proliferative response
determine if the CN-elastin composites could elicit appropriate cel- of Rugli cells. This showed significance over the CN-only control
164 D.V. Bax et al. / Acta Biomaterialia 86 (2019) 158–170

Fig. 4. Stiffness (A), failure stress (B) and failure strain (C) of CN-IE (i) or CN-SE composites (ii). The relative % of CN and elastin is shown on the x-axis. No data are shown for
25%CN-75%SE films as these adhered to the mould and could not be removed for analysis. Error bars indicate S.D. (n = 5). N/S, * and *** indicate >0.05, <0.05, <0.001
significance from the 100%CN/0%elastin values respectively.

with p = 0.0032 and 0.00017 for 75% SE and 75% IE containing films ties of CN-based materials can be tailored by incorporating elastin.
respectively. Rugli proliferation was noticeably higher on the SE Furthermore, by comparing 2 complementary cell lines, one of
compared to IE composites. This was evident at all densities of which utilises elastin-binding receptors, and the other which does
elastin inclusion. For example, inclusion of 25%, 50% or 75% IE not, we determined the influence of elastin- and CN-receptor
resulted in a
4-, 7- or 12-fold increase in the cell number between engagement on the cell response over short-term (adhesion,
day 1 and day 4. By comparison inclusion of 25%, 50% or 75% SE spreading) and long-term (proliferation) culture.
resulted in a
6-, 16- and 16-fold increase in the Rugli cell number J-shaped force-strain plots were observed for all films, in which
over the same duration. Indeed with 75% SE incorporation the the gradient of the force-strain profile increases with strain. The
degree of cell proliferation was approaching that observed on the initial extension phase of this profile is presumably dominated
soluble CN I positive control. Therefore, elastin incorporation can by CN fibril realignment towards the tensile axis, after which fur-
selectively increase the proliferation of cell types that possess the ther extension is via the stiffer CN fibres. These profiles yielded
appropriate elastin-binding receptors. stiffness values that are comparable to the literature values of 2–
31 MPa, depending on the composition [18,36]. In our hands, inclu-
4. Discussion sion of up to 75% IE decreased the stiffness and failure stress of
films but increased the failure strain. This is consistent, to a reason-
Matching the mechanical and biochemical properties of soft tis- able approximation, with disruption of the CN network by the large
sue implants to the surrounding tissue is functionally important and discrete IE inclusions [18]. By comparison SE had the opposite
[44,45]. As major components of numerous soft tissues, IE and effect to IE inclusion. SE was spread across the entire CN film, with-
SE, in combination with CN, was investigated. Analysis of these out large inclusions, and so it is possible that the SE component
composites showed that the mechanical and cell-ligating proper- resisted the applied force in parallel with the CN. Our data contra-
 D.V. Bax et al. / Acta Biomaterialia 86 (2019) 158–170 165

Fig. 5. HT1080 (A) or Rugli (B) cell adhesion to SE coated onto tissue culture plastic surfaces. The coating concentration of SE is shown on the x-axis and cell derived
absorbance on the y-axis. Bovine serum albumin (BSA) was used to block non-specific adhesion to the tissue culture plastic substrate. The degree of non-specific adhesion to
BSA is shown with a dashed line. Inhibition of Rugli cell adhesion to SE (C) or soluble CN (D) using EDTA (inhibits integrins), b-lactose (b-lac; inhibits elastin binding protein),
a-lactose (a-lac; non-functional control for b-lactose) and heparan sulphate (H/S; inhibits GAG binding). Inhibitor inclusion is detailed under the x-axis. Error bars indicate S.
D. of quadruplicate measurements. N/S, *, **, *** and **** indicate >0.05, <0.05, <0.01, <0.001 and <0.0001 significance from the 0 lg/ml SE values (A,B) or no-inhibitor controls
(C,D) respectively.

dicts others showing that SE decreased the stiffness of CN-based that there is a complex interplay between the distribution and
films, however, the differing fabrication processes [36] and elastin density these components, with different intrinsic mechanical
concentration [18] prevent direct extrapolation. No significant properties. Others have used electrospinning to further refine the
influence on the yield strain was observed with up to 50% IE or physical properties of elastin-CN composites, however the archi-
SE content, agreeing with CN-based films manufactured using elec- tectural properties of these fibres heavily influence cell behaviour,
trochemical alignment [36]. Interestingly we observed an increase and so amorphous blended films were chosen here to reduce the
in the yield strain with 75% IE inclusion, presumably as the loading number of interconnected parameters, allowing direct comparison
was resisted by the IE component. Alternatively, it is possible that to the cell response.
the physical properties are due to the lower CN content upon Currently the cell-binding properties of CN-elastin composites
increasing addition of IE or SE to the total 1% w/v slurry protein are relatively poorly understood. Although elastin addition influ-
content. This fits well with the lack of IE or SE influence on the fail- ences smooth muscle cell behaviour on CN-based materials
ure strain as the alignment and tensile stretch of each CN fibre does [31,35], the contribution from the elastin and CN component is
not change with lower CN density. However this cannot describe unknown. The major aim of this study was to explore this complex
the differences between SE and IE composites. Therefore, it is likely fundamental interplay by assigning the cellular response to the CN
166 D.V. Bax et al. / Acta Biomaterialia 86 (2019) 158–170

Fig. 7. Cell area quantification for HT1080 (A) or Rugli (B) cells on films with
increasing SE (blue bars) or IE (red bars) content. The relative % CN and elastin
content is shown beneath each pair of bars. Non-specific adhesion was blocked with
bovine serum albumin (BSA). The cell area on a negative control (BSA), a positive
control (5 lg/ml soluble CN I coated glass) and CN-only films are shown with grey
bars. Error bars indicate S.D. (n = 5). N/S, **, *** and **** indicate >0.05, <0.01, <0.001
and <0.0001 significance when compared to the 100%CN-0% elastin values
respectively. (For interpretation of the references to colour in this figure legend,
the reader is referred to the web version of this article.)

capacity of each on SE coated tissue culture plastic. SE solution

Fig. 6. HT1080 (A) or Rugli (B) cell adhesion to CN films containing an increasing
content of SE (blue, cross) or IE (red, diamond). The relative % of CN or elastin is was chosen over IE suspension to enable careful control over the
shown on the x-axis and cell derived absorbance is on the y-axis. Adhesion to 25% coating concentration. This analysis clearly showed that Rugli,
CN–75% SE films is shown with a dashed cross and line as these composites but not HT1080 cells, could adhere to SE in an EDTA-sensitive man-
fragmented during the washing regime used to remove loosely bound cells. Non-
ner, suggesting integrin-dependence of Rugli adhesion. We should
specific adhesion was blocked with bovine serum albumin (BSA) for all composites.
Cell binding to 5 lg/ml soluble CN I coated tissue culture plastic (square), tissue
point out that enzymatic-based cell detection, and not cell staining,
culture plastic (circle) and the negative control BSA-blocked tissue culture plastic was used here to avoid high non-specific, CN-film-associated,
(triangle) are shown in grey. Error bars indicate S.D. of quadruplicate measure- background values. Although we cannot entirely preclude material
ments. N/S, ** and *** indicate >0.05, <0.01 and <0.001 significance from the 100% induced effects on the cell-derived enzymatic activity, cell number
CN-0% elastin values respectively. (For interpretation of the references to colour in
linear regressions and microscopic observations were used to con-
this figure legend, the reader is referred to the web version of this article.)
trol for this possibility. As such enzymatic detection of these cell
lines was used to examine a series of IE and SE-CN composites.
and elastin elements. Our approach was to compare two different Consistent with EDC-induced modulation of cell adhesion,
model cell lines, one of which utilises elastin binding receptors and spreading and proliferation on CN-based materials, both Rugli
the other which is insensitive to elastin. For this reason HT1080 and HT1080 cell adhesion to CN-only films was lower than a sol-
and Rugli cells were chosen as they have well described CN- uble CN I control. This has been attributed to EDC-dependent
binding receptors (a2b1 and a1b1 respectively [13]) and the influ- blocking of native-like integrin binding to critical carboxylic acid
ence that EDC crosslinking exerts on their response to CN-only side chains in GxOGER motifs, and induction of non-native cell
films is known [9]. We therefore examined the elastin-binding binding mechanisms that do not support cell proliferation [9].
 D.V. Bax et al. / Acta Biomaterialia 86 (2019) 158–170 167

Fig. 8. Cell count over 4 days in culture for HT1080 (A,B) and Rugli (C,D) cells on SE (B,D) or IE (A,C) containing composites. The relative % CN and elastin content is identical
for all graphs and the legend is shown at the bottom of the figure. A 5 lg/ml soluble CN I coated tissue culture plastic positive control is shown (black, cross). Error bars
indicate S.D. of quadruplicate measurements. N/S, **, *** and **** indicate >0.05, <0.01, <0.001 and <0.0001 significance when compared to the 100%CN-0% elastin values at
Day 4.

Therefore, it is highly likely that adhesion on the CN-only films was monomer, tropoelastin, may be lost in mature elastic fibres [46]. As
non-native, which correlates well to the subsequent lack of cell such it is possible that the C-terminal RKRK motif integrin binding
spreading and proliferation. site may be absent in our elastin-CN composites. In addition,
Inclusion of elastin into CN films affects numerous different although amino acid analysis indicates that CN and elastin are
cell-responsive facets simultaneously, including the availability the major constituent of our protein preparations (Supplementary
of cell-binding motifs, roughness and stiffness. Comparison of Rugli Fig. 3), we cannot completely discount the possibility that each
against HT1080 cells allowed us to deconvolute the contribution of could contain traces of other bioactive molecules. For example,
elastin-derived cell-adhesive motifs. The Rugli response was dra- microfibril-associated glycoprotein can contaminate elastin prepa-
matically altered by the addition of elastin to CN-based materials rations. Notwithstanding these caveats, our data clearly show that
with increased attachment, spreading and particularly pronounced elastin addition offers huge potential to integrate cell-biological
proliferation. The integrin-dependent SE-binding mechanism of activity in to EDC/NHS crosslinked CN-based materials.
Rugli cells indicate that elastin-derived integrin-binding motifs The Rugli cell response was modulated by both IE and SE, how-
within elastin promote this potent biological effect. These could ever by mass, SE promoted increased cell adhesion, spreading and
include the integrin-binding motifs RKRK and AAAAAAAAAAKAA- proliferation than IE. Interestingly this trend is similar to the
KYGAAAGL [26] that ligate with integrins aVb3 and aVb5 [24,25]. degree of BA-4 detection by ELISA. One possible explanation is
However there are reports that the C-terminal region of the elastin the innate, densely packed, fibrous architecture of IE, where the
168 D.V. Bax et al. / Acta Biomaterialia 86 (2019) 158–170

elastin molecules within the fibre core are not solvent accessible. response is through direct cell-anchorage to elastin, it could be
This was clearly evident from anti-elastin fluorescent microscopy via SE-modulation of the CN component and/or via the physical
of IE samples. As a large proportion of the elastin mass in the IE properties of the material. For example, it is possible that elastin
films is not available for cell engagement, comparing the cellular alters EDC-dependent inhibition of CN-cell interactions [9] due to
response against elastin mass potentially under-represents the altered stoichiometry of the crosslinking reaction by the amine-
bio-activity of the IE component. Instead Rugli proliferation and rich regions on elastin. Alternatively, the influence of SE on the
attachment correlate linearly with SE and IE detection (R2 between mechanical properties and roughness could alter the cellular
0.992 and 0.882) by ELISA (Fig. 9). When compared in this manner, response. As the mechanical properties and chemistry of the CN
the cellular response to the IE and SE addition are very similar. This films change in tandem with SE addition these two explanations
is particularly important as the other characteristics of IE and SE cannot be separated here. It should also be noted that the relevance
containing films, such as mechanical properties and roughness of bulk mechanical properties to cell behaviour is often not a sim-
profiles, do not correlate with the cellular response. Instead, once ple correlation as the small-scale stiffness sensed by the cells is not
the relative solvent exposure is accounted for, the Rugli response necessarily the same as the bulk stiffness measured by stress-
is predominated by the elastin inclusions. strain techniques. It is also possible that SE addition alters the
We have shown that HT1080 cells do not adhere to elastin. CN spacing (i.e. by separating CN fibres) which influences the cell
Despite this HT1080 cell adhesion and spreading were elevated response. One surprising finding was that elastin inclusion induced
with the inclusion of SE but not IE. Whilst it is unlikely that this HT1080 cell spreading. Although we have shown that HT1080 cells

Fig. 9. Direct comparison of the degree of SE (C,D) and IE (A,B) detection by ELISA analysis (x-axis - taken from Fig. 1) and Rugli cell count at Day 4 (A,C – y-axis - taken from
Fig. 8) or adhesion (B,D – y-axis - taken from Fig. 6). Associated background values on the bovine serum albumin negative controls were deducted for all values. Linear
regressions fit with R2 values of 0.992 (A), 0.955 (B), 0.882 (C), 0.9386 (D).
 D.V. Bax et al. / Acta Biomaterialia 86 (2019) 158–170 169

do not adhere to elastin, it is possible that elastin is engaging with [3] L.H. Olde Damink, P.J. Dijkstra, M.J. van Luyn, P.B. van Wachem, P.
Nieuwenhuis, J. Feijen, Cross-linking of dermal sheep collagen using a water-
signalling molecules on these cells. One such example is the elastin
soluble carbodiimide, Biomaterials 17 (1996) 765–773.
binding protein which would not possess sufficient affinity to sup- [4] L. Buttafoco, P. Engbers-Buijtenhuijs, A.A. Poot, P.J. Dijkstra, W.F. Daamen, T.H.
port cell adhesion but has been shown to activate signalling cas- van Kuppevelt, I. Vermes, J. Feijen, First steps towards tissue engineering of
cades [47,48]. small-diameter blood vessels: preparation of flat scaffolds of collagen and
elastin by means of freeze drying, J. Biomed. Mater. Res. B Appl. Biomater. 77
SE or IE inclusion dramatically increased Rugli but not HT1080 (2006) 357–368.
cell proliferation. This differential response between the cell lines [5] I. Rault, V. Frei, D. Herbage, N. Abdul-Malak, A. Huc, Evaluation of different
suggests that the cell proliferative response is not solely due to chemical methods for cross-linking collagen gel, films and sponges, J. Mater.
Sci. - Mater. Med. 7 (1996) 215–221.
the CN component or physical attributes of the films as these [6] C.N. Grover, J.H. Gwynne, N. Pugh, S. Hamaia, R.W. Farndale, S.M. Best, R.E.
would influence both cell lines similarly. Instead it is likely Cameron, Crosslinking and composition influence the surface properties,
elastin-binding integrin-associated signalling cascades promote mechanical stiffness and cell reactivity of collagen-based films, Acta
Biomater. 8 (2012) 3080–3090.
cell proliferation [49]. [7] C.N. Grover, R.E. Cameron, S.M. Best, Investigating the morphological,
In summary, comparison of two independent cell lines, each mechanical and degradation properties of scaffolds comprising collagen,
with a differing affinity for elastin have shown that addition of gelatin and elastin for use in soft tissue engineering, J. Mech. Behav. Biomed.
Mater. 10 (2012) 62–74.
elastin-specific cell-binding motifs, or modulation of the CN com- [8] N. Davidenko, T. Gibb, C. Schuster, S.M. Best, J.J. Campbell, C.J. Watson, R.E.
ponent in an elastin-CN composite both contribute to the ultimate Cameron, Biomimetic collagen scaffolds with anisotropic pore architecture,
cellular response. Moreover, for cells that possess elastin-binding Acta Biomater. 8 (2012) 667–676.
[9] D.V. Bax, N. Davidenko, D. Gullberg, S.W. Hamaia, R.W. Farndale, S.M. Best, R.E.
integrins, the elastin component can dominate over the EDC cross-
Cameron, Fundamental insight into the effect of carbodiimide crosslinking on
linked CN component leading to a cell-inductive long-term cellular recognition of collagen-based scaffolds, Acta Biomater. 49 (2017) 218–
response. 234.
[10] D. Sheppard, The role of integrins in pulmonary fibrosis, Eur. Respir. Rev. 17
(2008) 157–162.
[11] J.D. Humphries, A. Byron, M.J. Humphries, Integrin ligands at a glance, J. Cell
5. Conclusion Sci. 119 (2006) 3901–3903.
[12] M. Barczyk, S. Carracedo, D. Gullberg, Integrins, Cell Tissue Res. 339 (2010)
SE and IE were successfully incorporated into CN-based materi- 269–280.
[13] S. Hamaia, R.W. Farndale, Integrin recognition motifs in the human collagens,
als. Both SE and IE significantly altered the mechanical properties
Adv. Exp. Med. Biol. 819 (2014) 127–142.
of CN-based films where SE increased but IE decreased the stiffness [14] R.W. Farndale, T. Lisman, D. Bihan, S. Hamaia, C.S. Smerling, N. Pugh, A.
and failure stress. Two cell lines were identified which possessed Konitsiotis, B. Leitinger, P.G. de Groot, G.E. Jarvis, N. Raynal, Cell–collagen
interactions: the use of peptide Toolkits to investigate collagen–receptor
differing elastin-binding properties. These allowed delineation of
interactions, Biochem. Soc. Trans. 36 (2008) 241–250.
the cellular response to the CN and elastin components of the com- [15] P.R.-M. Siljander, S. Hamaia, A.R. Peachey, D.A. Slatter, P.A. Smethurst, W.H.
posites. The solvent accessibility of the elastin element of the films Ouwehand, C.G. Knight, R.W. Farndale, Integrin activation state determines
heavily influenced the behaviour of cells which possessed elastin- selectivity for novel recognition sites in fibrillar collagens, J. Biol. Chem. 279
(2004) 47763–47772.
binding integrins. By comparison, if cells were devoid of elastin- [16] C.G. Knight, L.F. Morton, A.R. Peachey, D.S. Tuckwell, R.W. Farndale, M.J.
binding integrins then it was the indirect influence of elastin, over Barnes, The collagen-binding A-domains of integrins alpha(1)beta(1) and
the physical properties and CN component, that dictated the cellu- alpha(2)beta(1) recognize the same specific amino acid sequence, GFOGER, in
native (triple-helical) collagens, J. Biol. Chem. 275 (2000) 35–40.
lar response. Therefore, we have established that the cellular [17] J. Emsley, C.G. Knight, R.W. Farndale, M.J. Barnes, R.C. Liddington,
response to CN-elastin composites is highly dependent upon the Structural basis of collagen recognition by integrin a2b1, Cell 101 (2000)
cell surface receptor expression profile. CN and elastin are the most 47–56.
[18] C.N. Grover, R.W. Farndale, S.M. Best, R.E. Cameron, The interplay between
abundant protein components of tissues and are ubiquitously physical and chemical properties of protein films affects their bioactivity, J.
employed for tissue engineering. Consequently, these data offer Biomed. Mater. Res. Part A 100 (2012) 2401–2411.
the opportunity to control the bioactivity that each of the compo- [19] J.-D. Malcor, D. Bax, S.W. Hamaia, N. Davidenko, S.M. Best, R.E. Cameron, R.W.
Farndale, D. Bihan, The synthesis and coupling of photoreactive collagen-based
nents bestows upon these complex heterogeneous materials.
peptides to restore integrin reactivity to an inert substrate, chemically-
crosslinked collagen, Biomaterials 85 (2016) 65–77.
[20] J. Uitto, Biochemistry of the elastic fibers in normal connective tissues and its
Acknowledgements alterations in diseases, J. Invest. Dermatol. 72 (1979) 1–10.
[21] R.P. Mecham, A. Hinek, R. Entwistle, D.S. Wrenn, G.L. Griffin, R.M. Senior,
This work was supported by the ERC Advanced Grant 320598 Elastin binds to a multifunctional 67-kilodalton peripheral membrane protein,
Biochemistry 28 (1989) 3716–3722.
3D-E, EPSRC fellowship EP/N019938/1, and British Heart Founda- [22] R.M. Senior, G.L. Griffin, R.P. Mecham, D.S. Wrenn, K.U. Prasad, D.W. Urry, Val-
tion Special Project SP/15/7/31561 D. V. Bax was funded by the Gly-Val-Ala-Pro-Gly, a repeating peptide in elastin, is chemotactic for
Peoples Programme of the EU 7th Framework Programme (RAE fibroblasts and monocytes, J. Cell Biol. 99 (1984) 870–874.
[23] T.J. Broekelmann, B.A. Kozel, H. Ishibashi, C.C. Werneck, F.W. Keeley, L. Zhang,
no: PIIF-GA-2013-624904). R.P. Mecham, Tropoelastin interacts with cell-surface glycosaminoglycans via
its COOH-terminal domain, J. Biol. Chem. 280 (2005) 40939–40947.
[24] D.V. Bax, U.R. Rodgers, M.M.M. Bilek, A.S. Weiss, Cell adhesion to tropoelastin
Appendix A. Supplementary data is mediated via the C-terminal GRKRK Motif and integrin aVb3, J. Biol. Chem.
284 (2009) 28616–28623.
[25] P. Lee, D.V. Bax, M.M.M. Bilek, A.S. Weiss, A novel cell adhesion region in
Supplementary data to this article can be found online at tropoelastin mediates attachment to integrin aVb5, J. Biol. Chem. 289 (2014)
https://doi.org/10.1016/j.actbio.2018.12.033. The underlying data 1467–1477.
for this article may be found at https://doi.org/10.17863/CAM. [26] P. Lee, G.C. Yeo, A.S. Weiss, A cell adhesive peptide from tropoelastin promotes
sequential cell attachment and spreading via distinct receptors, FEBS J. 284
35187 (2017) 2216–2230.
[27] J.F. Almine, D.V. Bax, S.M. Mithieux, L. Nivison-Smith, J. Rnjak, A. Waterhouse,
S.G. Wise, A.S. Weiss, Elastin-based materials, Chem. Soc. Rev. 39 (2010) 3371–
References 3379.
[28] J. Rnjak, Z. Li, P.K.M. Maitz, S.G. Wise, A.S. Weiss, Primary human dermal
[1] C.H. Lee, A. Singla, Y. Lee, Biomedical applications of collagen, Int. J. Pharm. 221 fibroblast interactions with open weave three-dimensional scaffolds prepared
(2001) 1–22. from synthetic human elastin, Biomaterials 30 (2009) 6469–6477.
[2] L.C. Abraham, E. Zuena, B. Perez-Ramirez, D.L. Kaplan, Guide to collagen [29] Q. Lu, K. Ganesan, D.T. Simionescu, N.R. Vyavahare, Novel porous aortic elastin
characterization for biomaterial studies, J. Biomed. Mater. Res. B Appl. and collagen scaffolds for tissue engineering, Biomaterials 25 (2004) 5227–
Biomater. 87 (2008) 264–285. 5237.
170 D.V. Bax et al. / Acta Biomaterialia 86 (2019) 158–170

[30] D.T. Simionescu, Q. Lu, Y. Song, J.S. Lee, T.N. Rosenbalm, C. Kelley, N.R. [40] S.M. Partridge, H.F. Davis, G.S. Adair, The chemistry of connective tissues. 2.
Vyavahare, Biocompatibility and remodeling potential of pure arterial elastin Soluble proteins derived from partial hydrolysis of elastin, Biochem. J. 61
and collagen scaffolds, Biomaterials 27 (2006) 702–713. (1955) 11–21.
[31] A.J. Ryan, F.J. O’Brien, Insoluble elastin reduces collagen scaffold stiffness, [41] Z. Deyl, K. Macek, M. Adam, O. Vancíková, Studies on the chemical nature of
improves viscoelastic properties, and induces a contractile phenotype in elastin fluorescence, Biochim. Biophys. Acta, Mol. Cell. Biol. Lipids 625 (1980)
smooth muscle cells, Biomaterials 73 (2015) 296–307. 248–254.
[32] J. Rnjak-Kovacina, S.G. Wise, Z. Li, P.K.M. Maitz, C.J. Young, Y. Wang, A.S. Weiss, [42] N. Raynal, S.W. Hamaia, P.R.-M. Siljander, B. Maddox, A.R. Peachey, R.
Tailoring the porosity and pore size of electrospun synthetic human elastin Fernandez, L.J. Foley, D.A. Slatter, G.E. Jarvis, R.W. Farndale, Use of synthetic
scaffolds for dermal tissue engineering, Biomaterials 32 (2011) 6729–6736. peptides to locate novel integrin alpha2beta1-binding motifs in human
[33] C.B. Saitow, S.G. Wise, A.S. Weiss, J.J. Castellot, D.L. Kaplan, Elastin biology and collagen III, J. Biol. Chem. 281 (2006) 3821–3831.
tissue engineering with adult cells, Biomol. Concepts. 4 (2013) 173–185. [43] S. Perret, J.A. Eble, P.R.-M. Siljander, C. Merle, R.W. Farndale, M. Theisen, F.
[34] W.F. Daamen, S.T.M. Nillesen, T. Hafmans, J.H. Veerkamp, M.J.A. van Luyn, T.H. Ruggiero, Prolyl hydroxylation of collagen type I is required for efficient
van Kuppevelt, Tissue response of defined collagen-elastin scaffolds in young binding to integrin alpha 1 beta 1 and platelet glycoprotein VI but not to alpha
and adult rats with special attention to calcification, Biomaterials 26 (2005) 2 beta 1, J. Biol. Chem. 278 (2003) 29873–29879.
81–92. [44] J.D. Berglund, R.M. Nerem, A. Sambanis, Viscoelastic testing methodologies for
[35] W.F. Daamen, H.T.B. van Moerkerk, T. Hafmans, L. Buttafoco, A.A. Poot, J.H. tissue engineered blood vessels, J. Biomech. Eng. 127 (2005) 1176–1184.
Veerkamp, T.H. van Kuppevelt, Preparation and evaluation of molecularly- [45] T. Sugiura, R. Agarwal, S. Tara, T. Yi, Y.-U. Lee, C.K. Breuer, A.S. Weiss, T.
defined collagen-elastin-glycosaminoglycan scaffolds for tissue engineering, Shinoka, Tropoelastin inhibits intimal hyperplasia of mouse bioresorbable
Biomaterials 24 (2003) 4001–4009. arterial vascular grafts, Acta Biomater. 52 (2017) 74–80.
[36] T.-U. Nguyen, C.A. Bashur, V. Kishore, Impact of elastin incorporation into [46] T. Broekelmann, C. Ciliberto, A. Shifren, R. Mecham, Modification and
electrochemically aligned collagen fibers on mechanical properties and functional inactivation of the tropoelastin carboxy-terminal domain in cross-
smooth muscle cell phenotype, Biomed. Mater. 11 (2016) 025008. linked elastin, Matrix Biol. 27 (2008) 631–639.
[37] A. Waterhouse, S.G. Wise, M.K.C. Ng, A.S. Weiss, Elastin as a nonthrombogenic [47] S. Mochizuki, B. Brassart, A. Hinek, Signaling pathways transduced through the
biomaterial, Tissue Eng. Part B. Rev. 17 (2011) 93–99. elastin receptor facilitate proliferation of arterial smooth muscle cells, J. Biol.
[38] J. Skopinska-Wisniewska, A. Sionkowska, A. Kaminska, A. Kaznica, R. Chem. 277 (2002) 44854–44863.
Jachimiak, T. Drewa, Surface characterization of collagen/elastin based [48] M. Donet, S. Brassart-Pasco, S. Salesse, F.-X. Maquart, B. Brassart, Elastin
biomaterials for tissue regeneration, Appl. Surf. Sci. 255 (2009) 8286–8292. peptides regulate HT-1080 fibrosarcoma cell migration and invasion through
[39] G. Lammers, G.S. Tjabringa, J. Schalkwijk, W.F. Daamen, T.H. van Kuppevelt, A an Hsp90-dependent mechanism, Br. J. Cancer 111 (2014) 139–148.
molecularly defined array based on native fibrillar collagen for the assessment [49] P. Moreno-Layseca, C.H. Streuli, Signalling pathways linking integrins with cell
of skin tissue engineering biomaterials, Biomaterials 30 (2009) 6213–6220. cycle progression, Matrix Biol. 34 (2014) 144–153.