ABSTRACT
We demonstrate that the sessile tunicate Botryllus schlosseri is remarkably resilient to applied loads by attaching the animals to an extensile substrate subjected to quasistatic equiradial loads. Animals can withstand radial extension of the substrate to strain values as high as 20% before they spontaneously detach. In the small to moderate strain regime, we found no relationship between the dynamic size of the external vascular bed and the magnitude of applied stretch, despite known force sensitivities of the vascular tissue at the cellular level. We attribute this resilience to the presence and mechanical properties of the tunic, the cellulose-enriched gel-like substance that encases the animal bodies and surrounding vasculature.
INTRODUCTION
Botryllus schlosseri is a tunicate – a colonial marine chordate and close ancestor of vertebrates (Delsuc et al., 2006; Kocot et al., 2018) – consisting of genetically identical individuals (zooids) interconnected by a large colonial vascular system. Each colony forms a flat sheet encased within a soft, transparent, cellulose-based tunic that serves as a barrier layer that assists in wound healing and provides protection against predation and microbial attack (Di Bella et al., 2015; Hirose, 2009). This unusual body structure has enabled unprecedented investigations of the dynamic and mechanical properties of the vasculature within the context of a living organism (Gasparini et al., 2014; Madhu et al., 2020; Rodriguez et al., 2017, 2021). We previously demonstrated the ability to stimulate global retraction of the B. schlosseri extracorporeal vasculature upon loss of extracellular matrix (ECM) contact, either by disruption of collagen crosslinking by inhibition of lysyl oxidase (LOX) or by disruption of the focal adhesion kinase (FAK)-dependent cellular force sensing machinery (Madhu et al., 2020; Rodriguez et al., 2017). This indicates the importance of internal mechanical signaling within the tissue.
Yet, B. schlosseri and related organisms thrive in the shallow intertidal zone and are subjected to frequent waves exerting significant compressive and shear hydrodynamic stresses with fluid velocities that can exceed 10 m s−1 and accelerations on the order of 100 m s−2 (Denny and Shibata, 1989; Gaylord, 1999; Koehl et al., 2013). Such forces necessarily couple with extensional stress, stretching the adherent colonies and the soft structures within. Sub-lethal impact forces owing to wave action and predators are known to negatively influence other sessile marine organisms (Burnett and Sarà, 2019), and the influence of turbulent ocean flows has been studied in the context of reproduction and settlement (Abelson and Denny, 1997; Denny and Shibata, 1989; Mead and Denny, 1995). However, the effects of external forces on the B. schlosseri vasculature have not been explored.
Here, we used custom-built mechanical testing tools with in situ imaging to determine the interplay between loading, mechanics and vascular response of B. schlosseri organisms attached to an extensible substrate subjected to mechanical stretch. We investigated whether the application of external forces could induce changes to the mechanosensitive vasculature (Huang et al., 2010). In natural settings, wave action is complex, often leading to shear, compression and impact forces. Considering the B. schlosseri anatomy and assuming that the animal volume is roughly constant on the time scale of wave action, we expected that such forces would deform the soft planar animal body, causing both normal and tangential boundary stresses on the adherent surface to develop. Experimentally, we employed time-lapsed imaging of living organisms before and after the application of radially symmetric extension at the adherent boundary, and assessed the resultant changes to the vascular bed. This loading captures important aspects of the natural mechanical environment, while providing key advantages: access to high-resolution imaging and tracking throughout mechanical testing and roughly uniform strain throughout the animal, simplifying the analysis and interpretation of results. We found the animals to be extremely robust when the extensible substrate to which they are attached was stretched to up to 20% strain. We attribute this response to the colony morphology and mechanical properties of the tunic itself, which we determined using a custom-built microindenter (Urueña et al., 2018). Our results provide insight into the natural protective strategies developed by soft-bodied sessile marine organisms and offer motivation for future development of biomimetic materials.
MATERIALS AND METHODS
Design of the mechanical stretcher
A radial stretching device was designed and constructed to apply equiradial mechanical strain to extensible silicone membranes to which the organisms attach, while enabling unobstructed imaging during application of stretch (Dan et al., 2016; Huang et al., 2010). To achieve this, the silicone membrane was mounted onto a circular substrate holder (inner diameter 70 mm) and was held in place by a custom O-ring (Fig. S1A). This assembly was attached to a custom 3D-printed support capable of vertical translation by use of a linear carriage (McMaster-Carr, 60585K91) and single C-Beam linear actuator (Openbuilds, 995–250 mm) (Fig. S1B). The linear motion was driven by the rotation of the lead screw, which provides an 8 mm translation per revolution. As the substrate holder was lowered, the membrane was stretched over a separate indenter ring with the outer diameter 65 mm, leading to a uniform radial deformation of the silicone membrane. The substrate holder, O-ring and indenter ring were designed using Computer-Aided Design (CAD) software (Creo Parametric) and generated by a polyjet 3D printer (Objet 30, Stratasys) using Rigur (RGD 450, Stratasys), an advanced simulated polypropylene, as the print material owing to high resistance to saltwater corrosion. The entire system was mounted on a rigid T-slotted frame (McMaster, 47065T101) using standard brackets (McMaster, 47065T833) and fasteners (McMaster, 47065T142) and fixed to an aluminum breadboard which served as a base (Thorlabs, MB1218A). In all studies, an elastic, transparent and biocompatible polydimethylsiloxane (PDMS) membrane (Q7-4840, 0.005 inch thickness, Specialty Manufacturing, Inc.) was used as the substrate upon which the animals attached (Palchesko et al., 2012). Per the manufacturer (Specialty Manufacturing Inc., Saginaw, MI, USA), the substrate consists of a sheet formed from a SILASTIC BioMedical Grade Liquid Silicone Rubber, with a 5-min press cure and no post cure processing (elastic modulus of 2.6 MPa; tensile strength 9.4 MPa; elongation to failure 540%).
Animal preparation
Botryllus schlosseri (Pallas 1766) colonies were retrieved from the harbor in Santa Barbara, CA, USA, and allowed to spawn on 7.6×5 cm glass slides (Fig. S1D). Progeny were subsequently grown in a mariculture system according to established protocols (Boyd et al., 1986), with circulating 0.5-mm-filtered seawater at 18–21°C and daily feedings with live algae for a period of 1–2 months, until they reached 5–8 zooids in size, as described previously (Rodriguez et al., 2017). Prior to experiments, the animals, which are typically 5–10 mm in diameter, were transferred from the glass slide to the silicone membrane. A clean razor blade was used to gently detach the colonies from the glass slide without damaging the vasculature. The detached animal was placed at the center of a dry PDMS membrane that was secured to the substrate holder. A brush was used to place each animal at a desired location and to remove any feces or contaminants that could interfere with adherence. After positioning, the animal was left in a dry condition for no more than 10 min to facilitate attachment. After this initial dry incubation, the animal, which had now attached to the substrate-holder-mounted membrane, was quickly placed on a 3D-printed support stand and returned to the mariculture tank. A hose supplying circulating seawater was positioned to supply water to the animal (Fig. S1G). The system was left undisturbed for approximately 1 week to allow adhesion maturation.
Adult B. schlosseri initiate a 7-day asexual reproductive process (called blastogenesis) that results in the regeneration of a new individual with an identical body plan. These asexually derived bodies are called blastozooids (herein shortened to zooids) and also reproduce asexually every 7 days, resulting in a continuously expanding colony of zooids. The asexual cycle includes a 24 h period, called takeover (stage D) (Lauzon et al., 2002), during which zooids migrate within the colony, and there is a natural remodeling of the extracorporeal vasculature. In order to avoid this period, which could potentially confound our measurements, all colonies in these studies were tested at an equivalent stage in the blastogenic cycle (stage C1), which occurs 72 h prior to the takeover event.
Strain calculation and stretching tests
The mechanical strain was applied based on this calculation, by manually rotating the knob of the linear actuator, which translates 8 mm per revolution. In all cases, this motion was completed in less than 10 s. The applied strain was then held at a fixed value, and the vascular response was observed for ∼16 h, the time scale of action determined via chemical perturbation of the vessels in prior work (Rodriguez et al., 2017). The animal was entirely submerged in filtered seawater during testing.
Dynamic imaging was performed before, during and after stretching using a stereomicroscope (Stemi 2000-C, Zeiss) with a 10 MP USB 2.0 Color CMOS C-Mount (MU1000, AmScope, 3584×2748) camera. An LED-based Microscope Illuminator Light (LED-11C AmScope) with two gooseneck lamps provided the adjustable positioning and brightness required to produce a high contrast image. We found empirically that it was possible to produce a bright and nearly uniform light field by positioning the lamps at opposed diagonal directions (top right and bottom left). A portion of the indenter ring side wall was removed to enable unimpeded light throughput. Images in tagged image file format (TIFF) were collected over a typical imaging period 6–8 h at a collection rate of one image every 5 min (i.e. 1/300 frames s−1).
Characterization of membrane stretching
We experimentally validated the full-field displacement of the substrate under applied strain using DIC analysis. In this approach, a random grayscale intensity distribution (i.e. the speckle pattern) was applied to the substrate using black spray paint (Krylon). The silicone membrane was then mounted and stretched, while images of the specimen surface were recorded, with typically 8–10 images collected from a single trial (Canon EOS Rebel SL2 camera: 100 mm, f/2.8 Macro USM fixed lens, 1× magnification, 30 frames s−1). Between each image pair, a 2% target strain was applied, so the total targeted strain of the membrane was 20–25% over the course of the measurement (Fig. S2E,F). In practice, we found that the measured strain was somewhat smaller than the targeted strain (typically <20% difference in value at the largest strain), which we attribute to frame compliance. Here, targeted strains are reported. After image acquisition, we used a standard execution of the open source DIC code on the MATLAB File Exchange (Digital Image Correlation and Tracking: www.mathworks.com/matlabcentral/fileexchange/12413-digital-imagecorrelation-and-tracking) to obtain the displacement field by tracking the motion of the speckles (Fig. S2).
Experimentally, we confirmed the equiradial strain profiles of the deformed substrate up to targeted strain values of ∼20% (see Fig. S2 for details). Above ∼20% targeted strains, we observed skewing of the strain field, which we attributed to the fact that the substrate holder was driven by the linear actuator from only one side (Fig. S1B); this regime was avoided for animal studies. It should be noted that the DIC test was performed without animals. Given that the silicone membrane is stiffer than the animal by several orders of magnitude, we did not anticipate that the adherence of the animal to the membrane would affect the applied strain.
Finally, to confirm that mechanical stretching of the membrane was transmitted to the organism, two images acquired before and after the application of stretch were overlaid utilizing image registration and compared (Fig. 1B). Details of the method are described in the Supplementary Materials and Methods. Although the deformation field is not perfectly equiradial, the composite image clearly shows the deformation of the tunic and animal (yellow arrows). These small deviations in stretch uniformity could arise from imperfections in image registration, translocation of the animal from the membrane center during the time for adhesion maturation, or non-uniform adhesion or deformation of the animal.
Analysis of images of stretched animals
Images were collected as described and analyzed offline. Each image was read as an RGB image (Fig. 1C) in MATLAB and converted into a binary image through thresholding, where a value of 1 represented cellular regions of the vasculature and a value of 0 represented regions outside of the vascular tissue that were excluded from analysis. A series of image processing steps were used to further refine the identification of vascular tissue. Unwanted signals owing to feces or contaminants were inevitable and needed to be removed for accurate analysis. As a first step, one representative image (typically the first or last image in the movie) was selected and used to create a binary mask that was defined by simply drawing a polygon that separated regions of the animal that would be analyzed in subsequent steps from noisy or otherwise undesirable regions of the image. The subtraction of this mask was applied to all images to eliminate the noisy and undesirable regions while maintaining the region corresponding to the animal. A secondary clean-up step was then applied using the built-in function ‘bwareafilt’, which identified and removed small regions of noise based on the area of detected objects within the representative image. Next, the function ‘bwconvhull’ was applied to generate a polygon that detected the edges of the vessel bed and connected the vertices, and the area was calculated (Fig. 1D). Once the analysis of the single representative image was completed, the same procedure was repeatedly executed for all images in the movie. This enabled quantitative analysis of the time-lapsed series of images to identify the force-induced responses (e.g. regression or expansion) of animals subjected to various states of stretch and enabled reproducible tracking without reliance on manual feature identification, which tends to lead to systematic variation among different users. In addition to the area of the polygon, the boundaries were tracked as time series contour plots to observe qualitative responses, such as animal movements or localized vascular changes (Movie 1). This post-processing was applied to all experimental images.
Mechanical testing and analysis
To assess the mechanical resilience of B. schlosseri tunics, we determined the compressive elastic modulus, E, which is a measure of a material's resistance to compressive deformations. We employed microindentation using a custom-built microindenter, as previously described (Urueña et al., 2018). Briefly, a polished steel spherical probe (radius of curvature, R=1.6 mm) was mounted to a titanium double-leaf cantilever assembly with a stiffness of Kn=190 µN µm−1 in the normal direction. A capacitance sensor (Lion Precision, sensitivity: 5 µm V−1, range: 20 V) was used to convert cantilever displacements into normal forces. Microindentation measurements were performed on three different B. schlosseri animals. Indentations were conducted in a submerged environment at a constant indentation velocity, v=10 µm s−1, and only the tunic portions of the animals were probed (Fig. 3; Fig. S3). In each case, the animals were indented multiple times in the same position with varying maximum applied normal forces (Fmax=550–1570 µN), leading to 16 total microindentations. The largest applied force used in the contact mechanics analysis was limited to 500 µN to ensure that the mechanical response was measured in the small to moderate strain regime.
Hertzian contact mechanics theory was used to estimate the reduced compressive elastic modulus, E*, of the tunic assuming that the elastic modulus of the steel was significantly higher than that of the tunic. In this limit, E*=E/(1–ν2), where E is the compressive elastic modulus and ν is the Poisson ratio of the tunic. In practice, identifying the first point of contact between the probe and a soft biological sample is challenging and can greatly affect the estimated value of the sample's elastic modulus, E, and subsequently E* (Dimitriadis et al., 2002). To eliminate the need to determine the initial contact point and the associated uncertainty this introduces, we exploited the generalized relationship between indentation depth, u, and force, F, which can be written as F=KuP, where K=4/3E*R1/2 and P=3/2 for Hertzian contact (Garcia and Angelini, 2019; Garcia et al., 2017).
RESULTS AND DISCUSSION
Response of the animal to applied stretch
To determine the response of animals to applied stretch at their adherent boundary, 3–4 individuals were subjected to controlled extension of the membrane leading to targeted strains of 7%, 10%, 15% and 20%. Stresses applied at this substrate–tunic interface at the lower, adherent boundary of the animal are expected to propagate through the tunic to the vascular tissue (Fig. 1B). Although the molecular mechanisms of physical coupling are unknown, we observed that motile cells in the tunic often extend long actin-rich processes, reminiscent of filopodia, toward the external surface of the blood vessel (Rodriguez et al., 2017; Madhu et al., 2020), suggesting one mode of possible interaction. Additionally, when strains of >20% were applied, B. schlosseri lost adhesion to the membrane and partial or complete detachment occurred, indicating that mechanical coupling is compromised by such large extension. Thus, we limited our measurements and analysis to applied strains ranging from 0% to 20%, a regime where animal health is reasonably maintained and the stretching system operates well, as designed.
The time-dependent values of αlocal were calculated for every segment for every time series. Then, maximum and minimum values, denoting local elongation (positive) or retraction (negative), respectively, were obtained for all experiments. Maxima and minima were averaged by the sample sizes for each strain condition and compared (Fig. 2C). We found no obvious quantitative trend in local extrema as a function of applied strain. Additionally, animals showed no reproducible qualitative changes in response to mechanical perturbation.
Mechanical analysis of tunic
To investigate the physical origins of the observed mechanical resilience, we next measured the mechanical properties of the tunic, which encases the living vasculature, using a custom-built microindenter device using a steel spherical probe (radius of curvature, R=1.6 mm). A representative indentation curve is shown in Fig. 3. We observed high sample-to-sample variability and report the average stiffness and dissipated energy, Ud, using 16 microindentations across three independent animals. We found the average reduced compressive elastic modulus was E*=5.0±1.2 kPa (see Materials and Methods for details). For biomaterials, the Poisson ratio ν typically ranges between 0.3 and 0.5 (Zhang et al., 2013). Assuming ν=0.4, the average compressive elastic modulus is E=4.2±1.0 (n=16 tests from three tunics). For each individual tunic, E=3.8±0.6 kPa (n=8), E=4.5±0.3 kPa (n=3) and E=4.7±1.5 (n=5). These values are similar to those of other soft tissues and biomimetic hydrogels (Luo et al., 2022; Tejo-Otero et al., 2022; Urueña et al., 2018). The estimated contact diameter at the maximum normal force of 500 µN is approximately 1 mm, which corresponds to an approximate contact pressure of 1 kPa. Because the contact pressures do not exceed the elastic modulus of the tunic even at the highest applied loads (Fmax=1570 µN), we assume the test is non-destructive (Hart et al., 2019; Schulze et al., 2017).
We next considered the dissipated energy, Ud, as a function of the applied normal load and indentation depth (Cappella et al., 2005; Mu et al., 2022). The area between the loading and unloading curve above Fn=0 indicates the dissipated (hysteresis) energy (Fig. 3D). Quantitatively, Ud increased from 2.9×10−8 to 16.5×10−8 J as the maximum indentation depth increased during indentation (Fig. 3E). Interestingly, we found that under repeated loading, the mechanical response was identical from cycle to cycle, despite the presence of significant energy dissipation (Fig. 3F). There was no evidence of stress-softening (i.e. the Mullins effect), where subsequent loading cycles follow the unloading curve of the previous cycle (Shaw and MacKnight, 2018). This indicates that the source of energy dissipation within the tunic cannot be attributed to irreversible damage, i.e. rupture of covalent crosslinks (Fung et al., 2019; Mu et al., 2022). This could be attributed to viscoelasticity (Brown et al., 1998). However, we found that the tunics recovered quickly between indentations (<5 min) and demonstrated repeatable energy dissipation across cycles. Moreover, there was no evidence of unrecovered strain between cycles, which would indicate a viscous response. Hence, we attribute the energy dissipation to the disruption of dynamic bonds within the tunic. Such bonds can rupture upon loading, allowing the tunic network to rearrange and dissipate stress quickly, but quickly heal after unloading to form mechanically equivalent networks that demonstrate similar elastic responses upon reloading. Importantly, these network dynamics are operating on the time scales of minutes, similar to the time scales of wave actions. The presence of dissipation without loss of elasticity was observed in crosslinked networks of rigid protein filaments when the number of ruptured bonds remains small compared with the total number of crosslinkers (Yang et al., 2013; Vaca et al., 2015). The stiff cellulose structures of the tunic may provide similar responses.
We demonstrated that the vasculature of B. schlosseri is remarkably resilient to the application of quasistatic equiradial stretch at the attached basal surface at strain values as high as 20%, approaching its failure strain. On the one hand, this mechanical resilience is unexpected, as the B. schlosseri vasculature is known to be sensitive to the mechanical cell–ECM interface (Rodriguez et al., 2017). On the other hand, in their natural environment, B. schlosseri are repeatedly subjected to chaotic waves that can exert considerable hydrodynamic stresses that would necessarily couple to compressive and extensile deformations, so it is not surprising that some protective strategy would have evolved to provide organismal resilience. It has been argued that the relatively weak adhesion of soft-bodied compound ascidians provides advantages: preventing injury upon force-induced detachment (Edlund and Koehl, 1998) and as a means of asexual reproduction and dispersal (Bullard et al., 2007).
Our results indicate that below the detachment stress, the gel-like tunic may also provide mechanical protection by dissipating energy through the rupture of dynamic bonds that reform quickly after the stress is removed. We postulate that the cellulose fibers forming the tunic are sufficiently stiff and have sufficient crosslinks remaining that upon unloading, a mechanically equivalent network state is achieved, with little to no unrecovered strain observed between cycles. Our results suggest that the tunic functions to dissipate mechanical energy and suppress force propagation to the vasculature, which provides a protective benefit without compromising the previously established force-sensing capabilities of the vessel cells. Although not explicitly analyzed here, the tunic may provide similar benefit to the animal bodies themselves. The cellulose-enriched tunic thus offers inspiration for the development of soft yet resilient bio-inspired materials, such as biomimetic adhesives and soft tissue replacements (Almeida et al., 2020; Oh et al., 2015; Zhan et al., 2017; Capadona et al., 2008; Shanmuganathan et al., 2010).
Acknowledgements
Authors acknowledge use of the Mechanical Engineering Test Laboratory for tensile tests on silicone membranes and DIC set up, and the Microfluidics Laboratory within the California NanoSystems Institute, supported by the University of California, Santa Barbara and the University of California, Office of the President for 3D printing.
Footnotes
Author contributions
Conceptualization: Y.K., D.R., A.A.P., A.W.D.T., M.T.V.; Methodology: Y.K., S.S., D.R., A.L.C., A.A.P., M.T.V.; Software: Y.K.; Formal analysis: Y.K., A.L.C., A.A.P., M.T.V.; Investigation: Y.K., S.S., D.R., A.L.C.; Resources: A.A.P., A.W.D.T., M.T.V.; Data curation: Y.K.; Writing - original draft: Y.K., A.L.C., M.T.V.; Writing - review & editing: S.S., D.R., A.A.P., A.W.D.T.; Visualization: Y.K., A.L.C., M.T.V.; Supervision: D.R., A.A.P., A.W.D.T., M.T.V.; Project administration: M.T.V.; Funding acquisition: A.A.P., A.W.D.T., M.T.V.
Funding
This work was partially supported by the G. Harold and Leila Y. Mathers Foundation Grant SB170066 (to M.T.V. and A.W.D.T.), the MRSEC Program of the National Science Foundation Award No. DMR-2308708, NSF Graduate Research Fellowship Program Award No. 1650114 (to A.L.C.), NSF CAREER award CMMI-CAREER-2048043 (to A.A.P.), and National Institutes of Health MIRA Grant (R35 GM139649) (to A.W.D.T.). Deposited in PMC for release after 12 months.
Data availability
All relevant data can be found within the article and its supplementary information.
References
Competing interests
The authors declare no competing or financial interests.