Ester Comellas
my research
Publications
Projects
A computational tool to elucidate the mechanobiological regulation of limb development
My second postdoc brought me to Prof. Sandra Shefelbine's group at Northeastern University in Boston. Prof. Shefelbine is an expert in multiscale bone biomechanics and mechanoadaptation of bone and joint, and her group combines experimental and computational research. I had the opportunity to join her ongoing project in collaboration with the developmental biologist Prof. James Monaghan, who is well known for his cutting-edge research on axolotl regeneration. Just after starting, I won a Marie Curie Global Fellowship, with host supervisor Prof. Jose Muñoz at the Universitat Politècnica de Catalunya in Barcelona. Prof. Muñoz is a recognized researcher in computational biomechanics and has vast experience in developing numerical tools to explore the biomechanical forces that drive morphogenesis.
About the project
Understanding the roles of motion and mechanotransduction in joint formation holds promise for the study and treatment ofjoint deformities in humans. Joint development has been widely studied in axolotls (Ambystoma mexicanum), as these animals regrow whole limbs throughout their life. Axolotl limbs are morphologically similar to human limbs and utilize the same biological rubrics as ontogenic growth. To draw from the therapeutic potential of these similarities, we propose to build a multi-scale multi-physics computational model for the prediction of vertebrate limb development. Our model will be based on in vivo data obtained using novel imaging techniques via NSF-funded experiments on axolotl limb growth, and will be utilised to determine the physical mechanisms of normal and pathological joint morphogenesis. To this end, in AIM 1 we will build a finite element model of growth at the tissue level to study how specific changes in limb motion regulate joint morphology. Next, in AIM 2 we will build a model of growth at the molecular level to determine how biochemical and biomechanical signalling pathways interact during normal and pathological joint development. Finally, in AIM 3 we will integrate both experimental and computational data from the different length scales into a single multi-scale mechanobiochemical model of vertebrate limb growth. A computational model that links the biomechanics and biochemistry of normal and pathological limb development at the subcellular, cellular and tissue scales is a powerful predictive tool. We envisage this tool will be utilised to optimise treatment therapies for joint deformities and better inform the preventive screening of congenital defects in humans.
Key findings
Movement-induced forces are critical to proper joint formation, but it is unclear how biophysical stimuli drive the shaping of the joint. We examined the role of mechanical stimuli in joint formation in axolotl salamanders during limb regeneration. Transient receptor potential vanilloid 4 (TRPV4) channel desensitization during the joint-shaping phase of forelimb regeneration resulted in reduced cell proliferation and an altered elbow joint shape. To link TRPV4 desensitization to impaired mechanosensitivity in chondrocytes, we developed a poroelastic model of joint morphogenesis. Computational results indicated fluid pressure is a reasonable predictor of local tissue growth and may influence local joint shape.
Salamander limbs regrow after amputation by dedifferentiation of cells near the injury, generating a specialized tissue called a blastema. The regeneration blastema has increases in DNA, RNA, and protein synthesis, but how these processes are regulated is a fundamental question in regeneration biology. Imaging these processes in entire tissues is challenging. However, a convergence of technologies now make it possible to label newly synthesized biomolecules in animals using click-it based chemical fluorescent probes and visualize them via light sheet fluorescence microscopy. We set out to utilize these new technologies to develop a robust protocol for labeling synthesis of DNA, RNA, protein, and glycoproteins in a large whole mount tissue.
Modelling the porous and viscous responses of brain tissue behaviour
I did my first postdoc abroad in Prof. Paul Steinmann's group at the Institute of Applied Mechanics of Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU). Prof. Steinmann is an recognized expert in computational mechanics. During the two years I spent in Erlangen (Germany), I discovered and "converted" to deal.II thanks to my colleague Jean-Paul Pelteret, and learnt almost everything I know about brain mechanics from my colleague Silvia Budday.
About the project
Computational modelling can provide important insights into the underlying mechanisms of both healthy and pathological processes in the brain that go far beyond the possibilities of traditional methods. Adequate characterization of human brain tissue rheology is crucial to obtaining meaningful computational predictions within a clinical context. Experimental evidence shows that the essential features of brain tissue are nonlinearity, preconditioning, hysteresis, and tension-compression asymmetry. Past work by Budday and Steinmann's group had characterized brain tissue using a finite viscoelastic material model. Their results support the hypothesis that brain tissue behaviour is characterized by at least two different time scales. Herein, the longer time scale seems to be associated with the poroelastic interaction of fluid flowing within the solid network of cells and extracellular matrix in the tissue. The shorter one seems to be attributed to the viscoelastic nature of the solid skeleton itself, primarily due to fluid flow inside the cells. The monophasic viscoelastic model can only implicitly capture the porous effects of the longer time scale. To address this shortcoming, we have developed a biphasic nonlinear poro-viscoelastic model to further explore the role of viscous and porous effects in human brain tissue.
Key findings
The mechanical response of brain tissue is ultra-soft and heterogeneous, which has led to apparently contradicting experimental results in the literature. In this study, we demonstrate that a poro-viscoelastic model can explain why indentation experiments suggest that white matter tissue in the human brain is stiffer than gray matter tissue, while large-strain compression experiments show the opposite trend. This study highlights the potential of nonlinear continuum mechanics modeling and finite element simulations to standardize and help interpret experimental observations in the future through the use of material models that capture the behavior of brain tissue across different time scales and loading conditions.
Here, we present a novel material model that combines finite viscoelasticity with a nonlinear biphasic poroelastic formulation, developed within the context of the Theory of Porous Media. Embedded in a finite element framework, our model is capable of predicting the brain tissue response under multiple loading conditions. We show that our model can capture both experimentally observed fluid flow and conditioning aspects of brain tissue behavior in addition to its well-established nonlinear and compression–tension asymmetric characteristics. Our results support the notion that porous and viscous effects are highly interrelated and that additional experimental data are required to reliably identify the model parameters.
Numerical modelling of growth and remodelling in soft biological tissues
The aim of my doctoral studies was to develop a general constitutive formulation to reproduce the behaviour of soft biological tissues through finite element modelling, using the in-house Fortran code PLCd. I spent five years in the School of Civil Engineering at the Universitat Politècnica de Catalunya in Barcelona, working under the supervision of Prof. Sergio Oller. Prof. Oller is an expert in computational mechanics and composite material modelling. My co-supervisor, Prof. Facundo Bellomo, based at the National University of Salta (Argentina), is experienced in the constitutive modelling of biological tissue.
During my doctoral studies, I participated in activities at CIMNE and TA'd in Continuum Mechanics under the supervision of Prof. Xavier Oliver. I also had the opportunity to do a 2-month research stay with Prof. Salvador Botello's group at CIMAT in Guanajuato (Mexico) and a 6-month stay with Prof. Christian Gasser's group at KTH in Stockholm (Sweden).
About the project
Living biological tissues are complex structures that have the capacity of evolving in response to external loads and environmental stimuli. The adequate modelling of soft biological tissue behaviour is a key issue in successfully reproducing biomechanical problems through computational analysis. This study presents a general constitutive formulation capable of representing the behaviour of these tissues through finite element simulation. It is based on phenomenological models that, used in combination with the generalized mixing theory, can numerically reproduce a wide range of material behaviours. First, the passive behaviour of tissues is characterized by means of hyperelastic and finite-strain damage models. A generalized damage model is proposed, providing a flexible and versatile formulation that can reproduce a wide range of tissue behaviour. It can be particularized to any hyperelastic model and requires identifying only two material parameters. Then, the use of these constitutive models with generalized mixing theory in a finite strain framework is described and tools to account for the anisotropic behaviour of tissues are put forth. The active behaviour of tissues is characterized through constitutive models capable of reproducing the growth and remodelling phenomena. These are built on the hyperelastic and damage formulations described above and, thus, represent the active extension of the passive tissue behaviour. A growth model considering biological availability is used and extended to include directional growth. In ad-dition, a novel constitutive model for homeostatic-driven turnover remodelling is presented and discussed. This model captures the stiffness recovery that occurs in healing tissues, understood as a recovery or reversal of damage in the tissue, which is driven by both mechanical and biochemical stimuli. Finally, the issue of correctly identifying the material parameters for computational modelling is addressed. An inverse method using optimization techniques is developed to facilitate the identification of these parameters.