(337f) Decoupling Cellular Response to Topography and Stiffness in Three Dimensions

Introduction: Biophysical
aspects of in vivo tissue microenvironments include microscale mechanical
properties, architecture or topography of the extracellular matrix (ECM), and
the repertoire of ECM ligands present, all of which provide cues to drive
cellular response. Cell-ECM interactions are important regulators of both
normal tissue homeostasis and malignancy. Thus, understanding both
extracellular cues and the cellular responses they elicit is fundamental to
developing therapeutic strategies. Recapitulating the diversity of tissue
architectures present in vivo in a controlled, well-defined, and
well-characterized manner in three-dimensional tissue mimetics is challenging. We
recently developed a method whereby functionalized paramagnetic nanoparticles
are magnetically aligned in 3D hydrogels to create fibrils that span microns in
length and 10s of nms in width (1). Here, we used this approach to study the
response of healthy (human foreskin fibroblasts [HFF]) and malignant (U87
glioblastoma) human cells to topography and ECM ligand presentation in
topographically controlled environments.

Materials and Methods: Human ECM
proteins (fibronectin, tenascin C, or laminin) or bovine serum albumin were
labeled with a fluorescent marker and conjugated to 300 nm-diameter carboxylated
superparamagnetic particles. Cells were mixed with Matrigel and the conjugated nanoparticles
and plated in coverslip-bottom chamber slides. The slide was then either placed
on a magnet to generate aligned fibrils or kept far from the magnet to maintain
a random dispersion of particles. Finally, the slide was kept at 37°C to gel and set
the nanoparticle topography
(Fig. 1A). Composite gels were mechanically characterized using both
parallel plate small angle oscillatory shear (SAOS) bulk rheology and a custom
optical trap-based active microrheology setup. To examine cell response to aligned
topography, cells were fixed and stained 24 h after seeding and imaged via
confocal microscopy, and protrusion length and direction were measured. In
aligned matrices, cells were treated with inhibitors of contractility and
integrin binding to mechanistically understand cellular response to
topographical cues. To isolate the effects of stiffness on cell morphology,
cells were also plated on silicone substrates of defined stiffness (0.5, 2, and
64 kPa), fixed, and imaged.

Results and Discussion: Magnetic
field-driven assembly was used to generate fibrils decorated with human ECM
proteins in three-dimensional environments. Fibroblasts embedded in aligned
matrices were spindle shaped and could form long, actin-rich protrusions (Fig.
1B). In contrast, cells embedded in matrices containing unaligned
nanoparticles (and thus, the same absolute amount of human ECM proteins but
lacking topographical cues) remained largely spherical (Fig. 1C). The
length of protrusions of cells in aligned matrices 24 h after seeding
significantly increased compared to cells in unaligned matrices, where
protrusion lengths were similar to those seen in Matrigel alone (Fig. 1D).
The local angle between fibers and cell protrusions in aligned matrices was
preferentially either 0°
or 90° (Fig.
1E),
indicating protrusions sent out parallel or perpendicular to the fibers (see
insets in Fig. 1B). This preference was not observed in unaligned
matrices, where protrusions were randomly distributed around the cell body (Fig.
1F). These trends held across all of the ECM protein coatings tested,
including BSA, suggesting that the presence of a topographical cue was more
important in the observed cell extensions than specific integrin-ECM
interactions. Similar results were observed for U87 glioblastoma cells. Inhibition
of integrin-mediated binding using a function-blocking antibody directed
against integrin β1 abrogated cell protrusion generation along or
perpendicular to the fibrils.

Mechanical
characterization of the composite hydrogels revealed that while the bulk
mechanics of hydrogels containing either aligned fibers or randomly distributed
colloidal particles were similar, aligned hydrogels exhibited spatial
heterogeneities in microscale mechanical properties near aligned fibers. Bulk
rheological measurements demonstrated that the complex moduli (G* = G’ + iG”,
where G’ = elastic component and G” = viscous component) of aligned vs
unaligned (random) matrices were comparable independently of the ECM coating
used (Fig. 1G).  These hydrogels were mostly
elastic, with the shear elastic moduli ranging from 10-30 Pa with very little
viscous component over the range of ~0.1-100Hz (Fig. 1G). However,
at length scales relevant to cellular protrusions, active microrheology
measurements showed a gradient in the complex modulus as a function of distance
from the assembled fiber (Fig. 1H,I). Regions within 1 μm of the
fibers, which were comprised of rigid paramagnetic particles, were stiffer than the
unaligned gels. At further distances of 2–4 µm from the nearest fiber, local
stiffness decreased (Fig. 1I). Complex modulus values at distances of
2–4 µm from the nearest fiber were slightly less those than in unaligned gels (Fig.
1I). Having determined that there were local heterogeneities in microscale
mechanics, we next sought to determine how cells would respond to substrates of
differing mechanical properties. Over a range of stiffnesses greater than the
magnitude of the gradients in the vicinity of the aligned fibrils, cell aspect
ratios were unchanged. Therefore, we concluded that topographical cues were
dominant in causing the aligned cell morphologies observed in patterned 3D
environments.

(A) Schematic of 3D matrix patterning
process. Human cells and superparamagnetic nanoparticles were suspended in
Matrigel, plated on a glass slide containing a base layer of Matrigel, and
either aligned in a magnetic field (aligned gels) or left unaligned and
dispersed throughout the matrix (unaligned gels). Gels were then formed by
heating at 37°C to set the matrix topography. Cells were fixed for analysis 24
h after seeding. Representative images of human foreskin fibroblasts (HFF)
embedded in (B) aligned or (C) unaligned matrices containing nanoparticles
conjugated to fibronectin. Insets show detailed morphology of cells. Scales are
indicated. (D) Protrusion length (mean ± standard deviation) of HFF cells as a
function of matrix alignment status and nanoparticle ECM protein conjugation.
Average protrusion length in Matrigel lacking nanoparticles is also shown.
****, p<0.0001 by Sidak’s multiple comparisons test following two-way ANOVA.
Relative frequency distribution of the angle between (E) HFF protrusions and
nearest neighbor fibers or (F) HFF cells in unaligned matrices and the vertical
for matrices containing nanoparticles conjugated to human fibronectin, tenascin
C, laminin, or BSA. Protrusion direction distribution in Matrigel lacking
nanoparticles is also shown. (G) Bulk rheology measurements of elastic (G’) and
viscous (G’’) components of complex moduli of gels made with nanoparticles
coated in human fibronectin (red), tenascin C (green), or BSA (black), either
unaligned [U] or aligned [A] within a Matrigel matrix and averaged across
frequency. Measurements were carried out in duplicate. (H) Schematic and
summary of microrheology experiments. In aligned gels, local stiffness
increased closer to the fibers. In unaligned gels, stiffness was the same
throughout the gel. (I) Trend is evident by plotting the complex modulus (mean
± standard deviation), normalized to the complex modulus in unaligned gels, as
a function of distance from the nearest fiber. The percentage was averaged
across all measured frequencies for a given fiber distance to obtain the mean
and standard deviation.

Conclusions: Here, we use a
bottom-up approach to build fibrillar architecture into 3D amorphous hydrogels
using magnetic-field driven assembly of paramagnetic colloidal particles
functionalized with human ECM proteins and examined the response of normal and
malignant human cells topographical and chemical cues. Micromechanical and
topographical characterization of the resulting composite hydrogels enabled
characterization of the cell-scale biophysical properties of the system.
Topographical cues proved more important than chemical (ECM protein) or
stiffness differences in driving alignment to topographical features.

References:
(1)
Kim, J., Staunton, J.R., and Tanner, K. Independent control of topography for
3D patterning of the ECM microenvironment. Adv Mater 28: 132-137 (2016).