Here, we present an assembloid model system to mimic tendon cellular crosstalk between the load-bearing tendon core tissue and an extrinsic compartment containing cell populations activated by disease and injury. As an important use case, we demonstrate how the system can be deployed to probe disease-relevant activation of extrinsic endothelial cells.
Tendons enable locomotion by transferring muscle forces to bones. They rely on a tough tendon core comprising collagen fibers and stromal cell populations. This load-bearing core is encompassed, nourished, and repaired by a synovial-like tissue layer comprising the extrinsic tendon compartment. Despite this sophisticated design, tendon injuries are common, and clinical treatment still relies on physiotherapy and surgery. The limitations of available experimental model systems have slowed the development of novel disease-modifying treatments and relapse-preventing clinical regimes.
In vivo human studies are limited to comparing healthy tendons to end-stage diseased or ruptured tissues sampled during repair surgery and do not allow the longitudinal study of the underlying tendon disease. In vivo animal models also present important limits regarding opaque physiological complexity, the ethical burden on the animals, and large economic costs associated with their use. Further, in vivo animal models are poorly suited to systematic probing of drugs and multicellular, multi-tissue interaction pathways. Simpler in vitro model systems have also fallen short. One major reason is a failure to adequately replicate the three-dimensional mechanical loading necessary to meaningfully study tendon cells and their function.
The new 3D model system presented here alleviates some of these issues by exploiting murine tail tendon core explants. Importantly, these explants are easily accessible in large numbers from a single mouse, retain 3D in situ loading patterns at the cellular level, and feature an in vivo-like extracellular matrix. In this protocol, step-by-step instructions are given on how to augment tendon core explants with collagen hydrogels laden with muscle-derived endothelial cells, tendon-derived fibroblasts, and bone marrow-derived macrophages to substitute disease- and injury-activated cell populations within the extrinsic tendon compartment. It is demonstrated how the resulting tendon assembloids can be challenged mechanically or through defined microenvironmental stimuli to investigate emerging multicellular crosstalk during disease and injury.
In their function of transferring muscle forces to the bones to enable movement, tendons face some of the most extreme mechanical stresses occurring in the human body1,2,3. Due to aging societies, increasing obesity prevalence, and the growing popularity of mechanically demanding sports activities, the prevalence of tendon diseases and injuries is projected to climb in developed countries4,5,6. The development of novel evidence-based and disease-modifying treatment regimens to combat this increase has been hindered by limitations of currently available model systems1,7,8.
Ideally, disease and injury repair models would allow studying how the target organ processes a defined set of input parameters (imitating disease triggers, Table 1) into measurable output parameters (representing disease hallmarks, Table 2) while controlling for confounding factors. Studies using such model systems would then be able to identify the (patho-) physiological processes underlying disease and injury repair and gain knowledge that could be exploited to prevent or reduce disease and injury hallmarks in the clinics. Applying this principle to tendons, a useful model system should recapitulate central parts of the in vivo tendon response to disease and injury, which encompass the following hallmarks: microdamage, inflammation, neovascularization, hypercellularity, accelerated matrix turnover, and decompartmentalization9,10,11,12,13,14,15. Using these hallmarks as a base, the following requirements for a successful tendon disease and injury repair model system can be inferred.
Mechanical overloading is hypothesized to be a central factor in tendon injury and disease pathogenesis and is thus a commonly used experimental approach to create microdamage16. Controllable mechanical loadability is, therefore, a prime prerequisite for tendon disease and injury repair models. Ideally, the model system enables three main modes: single stretch-to-damage loading, fatigue loading, and unloading8,17,18. Upon mechanical deformation, tissue-resident cells experience a complex combination of tensional forces, shear forces (due to the sliding of collagen fibers surrounding the cells), and compression forces occurring during unloading or near the enthesis19,20. Model systems should recreate these complex loading patterns as closely as possible.
An alternative way to introduce matrix microdamage is to leverage biochemical stressors that mimic systemic predispositions for tendon disease and injury, such as (pro-)inflammatory cytokines, oxidative stress, or high glucose concentrations21,22,23. Consequently, a controllable niche microenvironment is advantageous for a tendon disease and injury repair model system.
A common prerequisite for model systems to be able to recapitulate inflammation, neovascularization, and hypercellularity is the selective presence of cell populations that drive these processes24. For inflammatory processes, these populations include neutrophils, T-cells, and macrophages, while endothelial cells and pericytes would be needed to study neovascularization25,26,27,28,29. Tendon fibroblasts are not only vital for tendon repair but, as proliferative and migrating cells, also partially responsible for the local hypercellularity observed in tendon disease30,31,32,33,34,35,36.
Besides changes in resident cell populations, the tendon matrix composition is altered in tendon disease and injury as well7,37,38,39,40. To present the right disease-relevant microenvironmental cues, model systems should be able to integrate an extracellular matrix composition matched to the targeted disease or injury stage, for example, by enabling relevant proportional combinations of collagen-1, collagen-3, and cellular fibronectin41.
The compartmentalization of healthy tendons into the tendon core and the extrinsic compartments (i.e., endotenon, epitenon, and paratenon) is central to their function and often disturbed in diseased or injured tendons1,42,43,44,45,46,47. Incorporating 3D tendon compartmentalization into tendon model systems is therefore not only required to simulate the processes underlying de- and re-compartmentalization more closely but also helps to establish the correct spatiotemporal gradients of cytokines and nutrients48,49.
Finally, modularity is another central asset of model systems, allowing researchers to combine the correct relative contribution of and interplay between the previously described stressors during the investigated processes8,17.
Besides selecting the optimal input modalities, an important step is being able to measure, observe, and track changes in the resulting output. The mechanical properties of the model system (i.e., toe region length, linear elastic modulus, maximum tensile strain, maximum tensile stress, fatigue strength, and stress relaxation) are central here, as they characterize the tendon's main function50,51,52. To link these functional changes to tissue-level changes, it is important to enable methods detecting (collagen) structural matrix damage and tracking proliferation and recruitment of disease- and repair-relevant cell populations30,53,54,55,56,57,58,59,60.
To study emerging cell-cell and cell-matrix crosstalk, one should be able to isolate or mark proteins in adequate amounts for quantification (i.e., ELISA, proteomics, immunohistochemistry, flow cytometry)14,21,61,62. Population- or at least compartment-specific gene expression analysis should be possible as well (i.e., fluorescence-activated cell sorting [FACS], single-cell/bulk RNA- sequencing, and real-time quantitative polymerase chain reaction (RT-qPCR))21,24,27,63. The model system should allow measuring as many of the aforementioned output parameters on the same specimen and on multiple specimens in a manner fast enough to unlock high-throughput studies.
Among model systems currently available to study human tendon disease and injury repair, the human body itself is, of course, the most representative one. It is also the least compatible with experimental intervention. While patients with acute tendon injuries are abundantly available for clinical studies, patients with early tendinopathy (the most common tendon disease) are largely symptom-free and often go clinically undetected until more severe changes manifest14,64,65. This makes it hard to pinpoint the critical moment when tendon homeostasis derails and the mechanisms behind this derailment16,66,67,68,69. In addition, extracting biopsies from healthy tendons is ethically challenging, as it may result in persisting damage. Hamstring tendon remnants from anterior cruciate ligament reconstruction surgery are often used as healthy controls but arguably differ in function, mechanical properties, cell populations, and matrix composition compared to the rotator cuff, Achilles, and patellar tendons commonly affected by tendon disease and injury70,71,72,73.
In vivo animal models are more accessible and tractable, but their usage imposes a significant ethical burden on the animals and economic cost on the researchers. In addition, most of the popular model animals either do not develop tendinopathic lesions spontaneously (i.e., rats, mice, rabbits) or lack the primers and genetically modified strains necessary to track the multicellular communication pathways involved in it (i.e., horses, rabbits).
Simple 2D in vitro model systems are on the other side of the complexity/tractability spectrum and better allow controlled, time-efficient study of specific intercellular communication pathways in response to a more controllable set of triggers8,74. However, these simplified systems commonly fail to recapitulate the multi-dimensional mechanical loading (i.e., tension, compression, and shear) that is central to tendon functionality. In addition, the (too) high stiffnesses of tissue culture plastic tend to override any matrix cues provided by coatings intended to mimic the disease state of interest75,76.
To overcome this drawback, increasingly sophisticated tissue-engineered 3D model systems have been developed to provide a loadable matrix whose composition can at least be partially matched to the desired disease state77,78,79. Still, these systems not only struggle to accurately replicate the complex in vivo extracellular matrix compositions and cellular loading patterns but generally lack long-term loadability and the compartmental interfaces required to study the cross-compartmental communication pathways that coordinate tendon disease and injury repair48,49,80.
Ex vivo tendon explant model systems have the distinct advantage of a built-in in vivo-like matrix composition that comprises pericellular niches, cross-compartmental barriers, as well as spatiotemporal cytokine/nutrient gradients and recapitulates complex loading patterns when stretched8. As a result of size-dependent nutrient diffusion limits, explants from larger animal models (i.e., horses) are difficult to keep alive for the long-term study of tendon disease and injury repair81,82,83. Meanwhile, smaller explants from murine species (i.e., Achilles tendon, patellar tendon) are challenging to reproducibly clamp and mechanically load. Their size also constrains the amount of material that can be gathered for cell-, protein-, and gene-level readouts without pooling samples and decreasing throughput. In this sense, murine tail tendon fascicles offer the potential to unlock high-throughput study of tendon disease and injury repair as they are readily available in large quantities from a single mouse, preserve the complex in vivo pericellular matrix composition, and recapitulate cellular loading patterns. During the extraction process, however, they lose most of their extrinsic compartment and the therein contained vascular, immune, and fibroblast populations that are now considered to drive tendon disease and repair8,18.
To bridge this gap, a model system combining the advantages of murine tail tendon-derived core explants with the advantages of 3D hydrogel-based model systems has been developed. This model system consists of a cell-laden (collagen-1) hydrogel cast around tail tendon explants84,85. In this paper, the necessary manufacturing steps are provided in detail alongside useful readouts that can be obtained by co-culturing core explants (intrinsic compartment) within an endothelial cell-laden type-1 collagen hydrogel (extrinsic compartment).
All methods described here were approved by the responsible authorities (Canton Zurich license numbers ZH104-18 and ZH058-21). An overview is presented in Figure 1.
1. Isolation of tendon assembloid components from 12-15 week-old mice (i.e., B6/J-Rj)
2. Collagen isolation from Wistar or Sprague-Dawley rats
3. Production of the culture system components
4. Clamping the core explants
5. Collagen hydrogel preparation and casting
6. Available readout methods
Component isolation (Figure 1 and Figure 2)
Before utilizing the core explants and cell populations in assembloid co-culture, these components are to be checked under the microscope (Figure 1). Core explants should have a uniform diameter (100-200 µm) and no visible kinks or wrinkles. Endothelial cells should present an elongated shape in contact with other cells, which they do not when seeded at a too-low density because of a low initial yield from the isolation. In this case, the endothelial cells assume a more roundish shape with cytoskeletal extensions and proliferate markedly slower. Split them 1:5 after 7-10 days. Tendon fibroblasts isolated from the Achilles tendons assume a more roundish morphology compared to their human counterparts within 1-2 passages (10-14 days each) when they were split 1:6. Macrophages are much smaller than fibroblasts or endothelial cells and do not proliferate after the isolation. Depending on the batch, their shape can vary from pyramidal to round.
The phenotypes of the cellular components were verified with flow cytometry. A conjugated CD31 antibody was used as a marker for endothelial cells (Figure 2A). Setting the fluorescence threshold based on an unstained control sample (grey), 90.1% of passage 1 (P1) and 48.7% of passage 2 (P2) endothelial cells were identified as CD31-positive. A genetically modified mouse line co-expressing the tendon fibroblast marker Scleraxis alongside a green fluorescent protein (ScxGFP) and a conjugated CD146 antibody was used to characterize the tendon fibroblasts (Figure 2B)35,60. After one passage (P1), 37.3% of the fibroblasts were ScxGFP+CD146–, 0.2% were ScxGFP+CD146+, 4.3% were ScxGFP–CD146+, and 58% were ScxGFP–CD146–. After two passages (P2), the percentage of ScxGFP+CD146– cells decreased to 27.6%, the percentage of ScxGFP+CD146+ cells increased to 6.9%, the percentage of ScxGFP–CD146+ cells increased to 10.6%, and the percentage of ScxGFP–CD146– cells decreased to 54.9%. To identify and characterize the macrophages, a F4/80 antibody was used in combination with a CD86 and a CD206 antibody (Figure 2C). After isolation and culture, 96.4% of the bone marrow-derived cells were F4/80-positive. Among these F4/80-positive cells, 8.6% were CD206+CD86–, 23.6% were CD206+CD86+, 28.3% were CD206–CD86+, and 39.4% were CD206–CD86–. Collagen crosslinking speed may vary from batch to batch and is to be tested before starting experiments.
Assembloid appearance (Figure 3)
In lesion-like culture conditions (36 °C, 20% O2), the core explant remained mechanically stretchable, did not change in appearance, and continued to be visually distinguishable and physically separatable from the surrounding hydrogel over at least 21 days (Figure 3A,B). The surrounding hydrogel was compacted over time, with the compaction speed depending on the cell population seeded into it. Achilles tendon-derived fibroblasts contracted their surrounding hydrogel the fastest and did so radially when in a hydrogel cast around a core explant and in all directions when not (Figure 3B,C). Initially, cell-free hydrogels placed around a core explant compacted, as well. This contraction was likely caused by migrating cells from the core explant, indicating a dynamic cross-compartmental interface. As cell-free hydrogels without an embedded core explant did not compact detectably, the contribution of water loss-induced shrinkage appears to be negligible (Figure 3B and Supplementary File 6).
A lack of hydrogel compaction can, therefore, be used to detect mistakes in the assembloid assembly (i.e., low cell concentrations) and should be checked before continuing with more expensive readout methods. While establishing this method, common mistakes reducing the cell concentration included dying cells in the extrinsic hydrogel because they were left for too long in the relatively harsh crosslinking solution (high pH, low temperature) and drying core explants because the time between medium aspiration and hydrogel injection was too long, or because the core explant was clamped too high to be embedded in the collagen.
Confocal fluorescence microscopy: Viability and morphology analysis (Figure 3)
Once removed from the clamps with scissors (Figure 3B), assembloids can be fixed, stained, and imaged with a confocal microscope as a whole without sectioning. Here, core // endothelial cell, core // macrophage, and core // fibroblast assembloids were stained with DAPI (NucBlue) and Ethidium Homodimer (EthD-1) to analyze the viability and DAPI and F-actin to analyze morphology and cell spreading in the 3D collagen hydrogel (Figure 3D). The viability of core // endothelial cell assembloids (Figure 3E) was quantified and found to be generally lower after assembloid assembly than previously reported for core // macrophage and core // fibroblast assembloids84. However, the viability remained stable during assembloid culture until at least day 7.
Mechanically induced microdamage and measurement of mechanical properties (Figure 4)
The screws and pins attached to the clamp holders allow the fixation of clamped assembloids to uniaxial stretching devices. The custom-made stretching device used here is equipped with a 10 N load cell and has been described in previous publications (Figure 4A)22. All samples were pre-conditioned with five stretch cycles to 1% strain prior to the measurements.
Recording the full stress-strain curve of core explants or assembloids (Figure 4B) would allow quantification of the linear elastic modulus (α), the maximum stress (β), and the maximum strain (у). However, it also irreversibly damages the core explant or the assembloid, which makes it impossible to assess the longitudinal development of the maximum stress (β) and the maximum strain (у) for the same samples (Figure 4B). Here, the linear elastic modulus was used as a measure for the sample's ability to withstand forces, as this measurement requires stretching the sample to only 2% strain, which has been shown previously to not cause permanent reductions in the linear elastic modulus18. In particular, core // endothelial cell assembloids were exposed to the clamping procedure to 2% strain (approximately the end of the linear elastic region) or 6% strain (approximately the maximum strain). The resulting microdamage was assessed by measuring the linear elastic modulus before and after the procedure (Figure 4C).
In line with previously conducted experiments exploiting mono-cultured core explants, core // endothelial cell assembloids retained their linear elastic modulus for at least 14 days when cultured in quasi-homeostatic niche conditions (29 °C, 3% O2) and exposed to strains no higher than 2%18,21. Regarding mechanical baseline stimulation, the static stretch applied through the clamps seemed to sufficiently mimic native strain levels experienced by tendon core units in vivo to prevent catabolic processes generally associated with matrix unloading87. Indeed, the progressive and statistically significant decline of the linear elastic modulus observed in core // endothelial cell assembloids exposed to 6% strain could be attributed to the matrix unloading stemming from mechanically-induced matrix microdamage.
When performing these experiments, it is important to prevent the drying of the assembloid. Here, they were encased in autoclaved and wetted paper, but other methods could also be viable depending on their compatibility with the stretching device used. As the friction between the metal clamps and the core explant is limited, add small pieces of paper between the metal and the core explant during clamping to prevent slippage and closely monitor the stretching process to detect and exclude slipped core explants and assembloids.
Compartment-specific transcriptome and assembloid-specific secretome analysis (Figure 5 and Figure 6)
In the first set of core mono-culture experiments presented here, the stability of core gene expression after explant isolation was assessed to decouple isolation from experimental effects (Figure 5A). Although higher replicate numbers are necessary for precise conclusions, the expression of Vegfa and Mmps increased strongly in freshly isolated core explants within hours after the explant isolation when cultured in lesion-like niche conditions (37 °C, 20% O2).
Neovascularization is a central hallmark of tendon disease and repair that could, in part, be driven by endothelial cells activated by pro-angiogenic factors (i.e., vascular endothelial growth factor, Vegfa) secreted by the tendon core under hypoxia88. Examining the first step of this potential crosstalk (Figure 5B), the expression of both Vegfa and the hypoxia marker carbonic anhydrase 9 (Ca9) was found to be increased statistically significant in explants mono-cultured under low oxygen tension (3% O2) in contrast to those mono-cultured under high oxygen tension (20% O2). Meanwhile, the lower oxygen tension did not seem to cause changes in the expression of tendon fibroblast markers such as Scleraxis (Scx) and collagen-1 (Col1a1). Together, these results identify core-resident cells as plausible contributors to pro-angiogenic signaling in a hypoxic niche.
Next, the activation of endothelial cells by pro-angiogenic core signaling was assessed in core // endothelial cell assembloid co-culture under high (20% O2) and low (3% O2) oxygen tension. Fortunately, the modular composition of assembloids allows compartment-specific transcriptome analysis after culture by physically separating the core explant from the extrinsic collagen hydrogel (Figure 6A). In the core explant (Figure 6D), Vegfa expression was again confirmed to increase under low oxygen tension, although the effect on other hypoxic markers such as Fgf2 was less clear and requires higher replicate numbers for precise conclusions. In addition, the expression of pro-inflammatory markers such as Tnf-α and markers for extracellular matrix degradation such as Mmp3 were decreased in the core under low oxygen tension. In the extrinsic hydrogel initially seeded with endothelial cells (Figure 6E), the presence of an alive core explant (aC) decreased Vegfa expression under low oxygen tension, but not under high oxygen tension. In addition, the presence of a devitalized (dC) core explant under low oxygen tension did not decrease Vegfa expression either. Under low oxygen tension, Tnf-α expression in the extrinsic hydrogel was comparable around aC/dC but increased under high oxygen tension around alive core explants. Fgf2 expression was decreased in all conditions compared to the extrinsic endothelial cell-laden hydrogel cultured around a devitalized core explant under high oxygen tension but most under low oxygen tension. Mmp3 expression was highest around alive core explants under high oxygen tension and lowest around devitalized core explants under low oxygen tension. Overall, the co-cultured endothelial cells seem responsive to both the active core explant, which is capable of initiating crosstalk and variations in oxygen levels. A more comprehensive transcriptome analysis would facilitate the elucidation of their respective contributions.
The modularity of the assembloid system allows the integration of genetically modified cells containing fluorescent reporter genes. Here, endothelial cells isolated from Pdgfb-iCreER mG mice89 were seeded into the hydrogel compartment. These cells co-express the endothelial cell marker platelet-derived growth factor subunit b (Pdgfb) alongside the enhanced green fluorescent protein (EGFP), which makes Pdgfb-expressing endothelial cells appear green under the microscopy (Figure 6C). Using this method, the presence of Pdgfb-expressing endothelial cells was confirmed to be maintained over 7 days in culture (37 °C) and appeared to be independent of oxygen tension (20% O2 compared to 3% O2).
To analyze the secretome of assembloids, the culture medium used respectively for core // cell-free and core // fibroblast, core // macrophage, or core // endothelial cell co-culture was replaced with its serum-free counterpart three days before aspirating and freezing the supernatant now enriched with the secretome (Figure 6A). This enrichment time was sufficient to detect cytokines such as vascular endothelial growth factor (VEGF) with an MSD assay, as shown here for core explants and core // fibroblast assembloids cultured in lesion-like niche conditions (Figure 6B).
Important considerations when analyzing the secretomes and transcriptomes of core explants and assembloids concern the usage of proper controls. Freshly isolated core explants have limited value, as especially their expression of Vegfa and Mmps increases strongly within hours after the isolation (Figure 5A). Time-matched explants surrounded by an initially cell-free hydrogel are more suitable as controls for the core compartment gene expression. For the extrinsic hydrogel, cell-laden hydrogels cultured without a core explant are inferior controls compared to cell-laden hydrogels cultured around devitalized core explants (Supplementary File 7), mainly because they compact into roundish shapes instead of elongated hydrogels which greatly changes cell morphology (Figure 3A).
Figure 1: Assembloid component isolation and assembly to model in vivo crosstalk. Tendon core explants were extracted from mouse tails, cut, and clamped. Mouse leg muscles (i.e., quadriceps femoris (QF), gastrocnemius (G), and tibialis anterior (TA)) were digested to isolate endothelial cells that were then cultured on tissue culture plastic. The Achilles tendons (AT) were digested as well to isolate tendon fibroblasts, which were then cultured on tissue culture plastic. The bone marrow from the tibia and the femur was flushed out of the bones. Then, the isolated monocytes were cultured on tissue culture plastic and differentiated into naïve macrophages. The light microscopy images (10x) depict the appearance of core explants, endothelial cells, tendon fibroblasts, and macrophages immediately before their integration into assembloids. During the assembly, the cells cultured on plastic were put in suspension and then seeded into a collagen-1 solution (1.6 mg/mL). Then, the cell-hydrogel mixture was cast around the clamped core explant and polymerized for 50 min at 37 °C before adding culture medium. Culture conditions were controlled via the clamps (mechanical tension) and the incubator settings (oxygen concentration, temperature). Please click here to view a larger version of this figure.
Figure 2: Characterization of cellular assembloid components. (A) Representative flow cytometric analysis of muscle-derived endothelial cells after one passage (P1, top row) and two passages (P2, bottom row). The counts for unstained (grey) and CD31-stained (green) cells were normalized to modal. The percentages are given for the CD31-stained group. (B) Representative flow cytometric analysis of Achilles tendon-derived fibroblasts after one passage (P1, top row) and two passages (P2, bottom row). The axes report fluorescence intensities of unstained cells (grey) and cells both expressing ScxGFP and stained with CD146 antibodies (rainbow colors). (C) Representative flow cytometric analysis of bone marrow-derived macrophages after culture. In the top row, the counts for unstained (grey) and F4/80-stained (green) cells were normalized to modal. The percentages are given for the F4/80-stained group. The graph in the bottom row reports fluorescence intensities of unstained cells (grey) and the F4/80+ subset of cells stained with CD206 antibodies and CD86 antibodies (rainbow colors). Please click here to view a larger version of this figure.
Figure 3: Assembloid imaging and appearance. (A) Representative photographs taken at day 0 (d0) and day 21 (d21) of culture (37 °C, 20% O2) show a multi-dimensional contraction of a hydrogel containing extrinsic fibroblasts without an embedded core explant and strong radial compaction of a hydrogel containing extrinsic fibroblasts around a core explant. (B) Representative photographs taken at day 21 (d21) of culture (37 °C, 20% O2) show differences in compaction speed between cell-free hydrogels, cell-free hydrogels cast around a core explant, and tendon fibroblast-laden hydrogels cast around a core explant. (C) The representative light microscopy images (10x) taken at day 0 (d0) and day 21 (d21) of culture (37 °C, 20% O2) indicate longitudinal changes in the presence of cell populations and the compaction speed of the collagen hydrogel (HG) around the core explant (E) in core // cell-free and core // fibroblast assembloid co-culture. The schematic representation depicts the differences in hydrogel compaction between core // cell-free assembloid and core // fibroblast assembloid co-culture. (D) Representative confocal microscopy images taken at day 7 (d7) of core // endothelial cell, core // macrophage, and core // fibroblast assembloid co-culture (37 °C, 20 % O2). Images in the left row depict assembloids with cell nuclei stained in blue (DAPI) and dead cells stained in pink (Ethidium homodimer-1). The other two rows depict assembloids with cell nuclei stained in blue (DAPI) and actin filaments in green (F-actin). (E) Boxplots depicting the quantified viability of core // endothelial cell assembloids at day 1 (d1) and day 7 (d7) of co-culture. N = 5. The upper and lower hinges correspond to the first and third quartiles (25th and 75th percentiles) and the middle one to the median. Whiskers extend from the upper/lower hinge to the largest/smallest value no further than 1.5 times the interquartile range. P-values: n.s.p > 0.05. Please click here to view a larger version of this figure.
Figure 4: Mechanical stimulation of assembloids and measurement of assembloid mechanical properties. (A) Graphical depiction of the custom-made stretching device comprising the clamp holder platforms, a force sensor, and a stepper motor. The photographic image shows an assembloid mounted to the stretching device with clamps. The lid of a 15 mL plastic tube (Ø: 17 mm) used for scale. (B) Graph depicting representative stress/strain curves for core explants (light blue) and assembloids (light red). The linear elastic modulus (α), maximum stress (β), and maximum strain (у) can be extracted from the data to mechanically characterize the core explant or assembloid. (C) Graph showing the linear elastic modulus (Emod) of core // endothelial cell assembloids co-cultured (29 °C, 3% O2) over a 14-day time course after being clamped (solid line), clamped and stretched to 2% L0 strain (dotted line), or clamped and stretched to 6% L0 strain (dashed line) at the start of the experiment. N = 5. The data points were normalized to the initial modulus linear elastic modulus before the stretching and are all displayed as mean (±sem). P-values: *p < 0.05, **p < 0.01. Please click here to view a larger version of this figure.
Figure 5: Changes in the core transcriptome after isolation and culture under different niche conditions. (A) Scatterplot depicting the fold changes in Vegfa, Mmp13, and Mmp3 gene expression in mono-cultured (37 °C, 20% O2) murine core explants 2 h, 4 h, 6 h, and 8 h after isolating them from the tail. The fold changes at the respective time points were normalized to the gene expression 2 hours after the isolation. N = 2. (B) Boxplots depicting the fold changes in Ca9, Vegfa, Scx, and Col1a1 gene expression in core explants mono-cultured under low oxygen tension (3% O2) normalized and compared to those mono-cultured under high oxygen tension (20% O2). N = 5-6. The upper and lower hinges of the boxplots correspond to the first and third quartiles (25th and 75th percentiles) and the middle one to the median. Whiskers extend from the upper/lower hinge to the largest/smallest value no further than 1.5 times the interquartile range. Datapoints used for normalization are depicted as black dots and individual datapoints as red dots. P-values: **p < 0.01, ***p < 0.001. Please click here to view a larger version of this figure.
Figure 6: Assembloid-specific secretome and compartment-specific transcriptome analysis. (A) Representative photograph showing the assembloid at day 7 (d7), when secretome and transcriptome samples were taken, and depiction of the underlying workflow. (B) VEGF concentration (pg/mL) in the supernatant of core // cell-free and core // fibroblast assembloids after 7 days of co-culture (37 °C, 20% O2) depicted as boxplots. N = 6. (C) Representative confocal microscopy images of core // endothelial cell assembloids after 7 days of co-culture (37 °C) under high oxygen tension (20% O2) and low oxygen tension (3% O2). Cell nuclei are stained in blue (DAPI), and the embedded endothelial cells co-express enhanced green fluorescent protein (EGFP) alongside the endothelial cell marker platelet-derived growth factor subunit b (Pdgfb). The dotted line indicates the compartmental interface between the core explant (E) and the endothelial cell-laden hydrogel (HG). (D) Scatterplot depicting the fold changes in Vegfa, Tnf-α, Fgf2, and Mmp3 gene expression in the core compartment from core // endothelial cell assembloids co-cultured under low oxygen tension (3% O2) normalized and compared to those cultured under high oxygen tension (20% O2). N = 2. (E) Scatterplot depicting the fold changes in Vegfa, Tnf-α, Fgf2, and Mmp3 gene expression in the extrinsic compartment of core // endothelial cell assembloids with an alive core (aC) or a devitalized core (dC) co-cultured under high oxygen tension (20% O2) and low oxygen tension (3% O2). The fold changes in the respective conditions were normalized to the extrinsic compartment of a core // endothelial cell assembloid with a devitalized core (dC) co-cultured under high oxygen tension (20% O2). N = 3-4. In B, the upper and lower hinges of the boxplots correspond to the first and third quartiles (25th and 75th percentiles) and the middle one to the median. Whiskers extend from the upper / lower hinge to the largest / smallest value no further than 1.5 times the interquartile range. Outliers are depicted as black dots. P-values: *p < 0.05. In D and E, the datapoints used for normalization are depicted as black dots, and individual datapoints are depicted as red dots. Please click here to view a larger version of this figure.
Table 1: Input requirements for tendon disease and injury model systems. A list of primary tendon disease triggers and secondary drivers matched to a selection of input parameters whose tractability is central for modeling tendon disease and injury. Please click here to download this Table.
Table 2: Output requirements for tendon disease and injury model systems. A selection of tendon disease hallmarks matched to a selection of output parameters whose quantifiability is central for the interpretation of tendon disease and injury model behavior. Please click here to download this Table.
Supplementary File 1: .stl file for the clamp holders, the mounting station, and the chamber molds. Please click here to download this File.
Supplementary File 2: Plan of right clamp holder. Please click here to download this File.
Supplementary File 3: Plan of left clamp holder. Please click here to download this File.
Supplementary File 4: Plan of the mounting platform Please click here to download this File.
Supplementary File 5: Plan of metal clamps. Please click here to download this File.
Supplementary File 6: Image showing cell-free hydrogel shrinkage. Please click here to download this File.
Supplementary File 7: Image showing a devitalized core explant. Please click here to download this File.
Overall, the assembloid model system presented here has several critical steps to highlight. First, the model system is only as good as the quality of its components. It is vital to check the core explant and the to-be-seeded cell populations under the microscope prior to initiating the assembly process. It is similarly important to verify the phenotype of the isolated cell populations at least once with flow cytometry. Especially when a new batch of collagen-1 is used for the first time, it is advantageous to check the crosslinking speed in a trial run before embedding cells into it. The assembloid assembly requires a lot of manual handling, which increases the risk of infections. To minimize the risk of infections, work in a sterile biosafety hood with laminar air flow, exchange gloves often, and decontaminate the gloves as well as the working space with 80% ethanol. For similar reasons, do not use the 3D-printed clamp holders more than once. Before the embedding process itself, it is important to keep all the hydrogel components (crosslinking solution, collagen-1 solution) on ice to prevent premature crosslinking. Consequently, one must work quickly once the cells are added to the crosslinking solution to limit cell death due to the high pH and low temperature of the crosslinking solution. To prevent drying-related cell death in the core explant, aspirate the medium covering the clamped core explants immediately before mixing the crosslinking solution with the collagen-1 solution. To guarantee the central placement of the core explant within the hydrogel, it is ideal to cast the hydrogel around a clamped core explant that is slightly tensioned. To do so, use the dowel pin and the M3 x 16 mm bolt screw to fixate the clamp holders to a (3D-printed) plate set with holes at the appropriate lengths. After the 50 min polymerization time, the embedded core explant can be detensioned again depending on the desired culture conditions. The amount of tension the assembloid experiences during culture has a profound impact on experimental outcomes and is to be kept uniform across samples and conditions21.
Nevertheless, the large impact of mechanical (un-)loading on experimental outcomes is a main advantage of the assembloid model over most tissue-engineered alternatives, especially since the maintained matrix composition of the core explant should also recreate the complex in vivo loading patterns on the cellular level90. While in practice, only the measurement of the linear elastic modulus, the maximum tensile strain, and the maximum tensile stress of assembloids have been demonstrated so far, protocols for fatigue strength and stress relaxation measurements have been described for tendon core explants elsewhere and should be applicable to the assembloids91,92. In addition to the in vivo-like loading patterns, the assembloid's multi-level modularity is likely its biggest advantage. Thanks to the individual culture chambers, a controllable set of niche conditions can be set for each sample separately (i.e., temperature, oxygen tension, glucose concentration, supplementation, stimulators, inhibitors, and static stretch with a plate). Next, matrix stiffness and matrix composition of the extrinsic compartment are customizable through the hydrogel composition and would, for example, allow for studying the impact of an increasingly diseased tissue microenvironment by incorporating more collagen-3 and cellular fibronectin93,94,95. The assessed cell populations in the extrinsic compartment are easily adaptable by selecting which cells to seed but can also be modified in the tendon core explant by leveraging established genetically modified cell lines and mouse lines (i.e., ScxLin cell depletion)96. The differing matrix and cell composition of the two compartments further provides a unique compartmentalized 3D structure that is another central tendon hallmark1,30,46.
When using this system, it is important to consider the consequences of the system's modularity for the granularity of the outcome parameters. While cell proliferation and recruitment can be assessed for each compartment separately, the mechanical properties, secretome components, and degradation products are currently only measurable for the complete assembloid. Regarding throughput, one properly trained person can prepare up to 50 assembloids in a regular workday, with the main bottleneck being the clamping procedure. While some of the readout methods are mutually exclusive, it is possible to assess mechanical properties and secretome components repetitively on the same sample as well as either cell population composition (flow cytometry), cell transcriptome (RT-qPCR, RNA-sequencing), or matrix and cell distribution (immunocytochemistry/fluorescence microscopy) at endpoints. In previous publications, these methods have been deployed to extensively characterize intercellular, cross-compartmental interactions in core // fibroblast and core // macrophage assembloids exposed to a lesion-like niche84,85. In this work, the capability of the assembloid model system to probe the cross-compartmental interplay between the core and extrinsic endothelial cells under different microenvironmental stimuli has been explored.
The modularity of the model system allows for future refinement of the method, which is necessary to overcome the following limitations of the current design iteration. The flow cytometric analysis presented in this work and single-cell RNA-sequencing data published recently revealed that the tendon core-resident tenocytes and Achilles tendon-derived populations are more heterogeneous than previously assumed24,34,59,84,97. In addition, the migratory behavior of initially core- or hydrogel-resident cell populations blur assembloid compartmentalization during culture. Both factors together make it challenging to attribute transcriptomic differences to specific cell types and to separate proliferation- from migration-based processes. This limitation could be overcome by refining the input population with fluorescence-activated cell sorting (FACS) based on the cellular composition of healthy or diseased tendons characterized in recent in vivo studies, improving the readout by implementing single-cell RNA-sequencing, and integrating proliferation markers such as an EdU (5-ethynyl-2'-deoxyuridine) staining during microscopy.
The assembloids presented here also share a weakness with most of the currently available in vitro systems that simulate diseased organs disconnected from the rest of the body98,99. However, the culture chamber-based platform used here positions the model system well for integration into a multi-organ platform where assembloids mimicking different organs are connected and interorgan interactions can be studied.
At its core, the model system is based on positional rodent tendons, which results in its own unique set of drawbacks. First, the translatability of results is hampered by wild-type mice not developing or suffering from tendon diseases8,100,101. Integrating tissues and cells from humans or newly developed mouse strains that exhibit aspects of tendon disease could alleviate this issue102. The switch towards a human-based assembloid is particularly interesting, as it would enable studies with patient-derived tissues from differently diseased tendons (i.e., tendinitis, tendinosis, or peritendinitis) and even treatment-resistant donors that could unlock more personalized treatment programs. Second, murine tail tendon explants do not handle overload-induced microdamage particularly well, which limits the applicability of the model system for the study of acute tendon damage.
For all these reasons, explant // hydrogel assembloids are in a prime position to study tendon core biology, matrix structure-function interactions, and cross-compartmental interactions between specific cell populations in response to niche-induced microdamage. Insights gathered from these rather high-throughput studies could give direction to in vivo research and treatment development.
The authors have nothing to disclose.
This work was funded by the ETH Grant 1-005733
0.4 mm x 25 mm injection needle (G27) | Sterican | 9186174 | |
3D printing filament: Clear polylactic acid prusament | Prusa | NA | |
4% formaldehyde | Roti-Histofix | P087.4 | |
Accutase cell detachment solution | Sigma-Aldrich | A6964-100ML | |
Amphotericin | VWR | L0009-100 | |
Attachable digital C-mount camera: Moticam 2 | Motic | NA | |
Bolt screw M3 x 16 mm, stainless steel | RS PRO | 1871235 | |
Bolt screw M3 x 6 mm, stainless steel | RS PRO | 1871207 | |
CaCl2 | Sigma-Aldrich | C5670 | |
CD146 antibody: PE anti-mouse | BioLegend | 134703 | |
CD206 antibody: Alexa Fluor 488 anti-mouse | BioLegend | 141709 | |
CD31 antibody: Alexa Fluor 488 anti-mouse | BioLegend | 102413 | |
CD86 antibody: PE anti-mouse | BioLegend | 105007 | |
Collagenase I | Thermo Fisher Scientific | 17100017 | |
Collagenase IV | Gibco | 17104-019 | |
Dialyzed Fetal Bovine Serum (FBS) | Sigma-Aldrich | F0392-100ML | |
Dimethyl sulfoxide (DMSO) | Sigma-Aldrich | 7000183 | |
Dispase II | Sigma-Aldrich | D4693-1G | |
DMEM/F12 | Sigma | 7002211 | |
Dowel Pin, 3 mm x 16 mm, stainless steel | Accu | HDP-3-16-A1 | |
Dragon Skin 10 Slow/1 silicone | KauPO | 09301-004-000001 | |
Endopan 3 Kit | Pan-Biotech | P04-0010K | |
Endothelial cell growth supplement | Lonza | CC-3162 | |
Eppendorf safe-lock plastic tubes (1.5 mL) | Eppendorf | 30121023 | |
Ethidium homodimer, EthD-1, 2 mM stock in DMSO | Sigma-Aldrich | 46043-1MG-F | |
F4/80 antibody: Apc/fire 750 anti-mouse | BioLegend | 123151 | |
Falcon plastic tube (15 mL) | Corning | 352096 | |
Falcon plastic tube (50 mL) | Corning | 352070 | |
Flow cytometer: LSR II Fortessa | BD Bioscience | 23-11617-02 | |
Gelatin | Invitrogen | D12054 | |
Hellmanex III alkaline cleaning concentrate | Sigma | Z805939-1EA | |
Heparin | Sigma-Aldrich | H3149-10KU | |
Hydroxyproline assay | Sigma-Aldrich | MAK008 | |
Image analysis software: Motic Images Plus 3.0 ML | Motic | NA | |
L-Ascorbic Acid Phosphate Magnesium Salt n-Hydrate | Wako Chemicals | 013-19641 | |
LSE Low Speed Orbital Shaker | Corning | 6780-FP | |
MEM non-essential amino acids | Sigma | 7002231 | |
Mouse macrophage-stimulating factor (m-CSF) | PeproTech | 315-02-50ug | |
MSD assay | Mesoscale Discovery | various | |
NucBlue | Thermo Fisher Scientific | R37605 | |
Nylon mesh strainer cap, 100 µm | Corning | 734-2761 | |
Original Prusa i3 MK3S 3D printer | Prusa | i3 MK3S | |
Penicillin-Streptomycin | Sigma-Aldrich | P4333 | |
Phosphate-buffered saline (PBS), ph 7.4, sterile, 10 L | Gibco | 10010001 | |
Puromycin | Gibco | A1113803 | |
RBC lysis buffer | VWR | 786-650 | |
recombinant m-CSF | PeproTech | 315-02 | |
RNA extraction kit: Rneasy plus Micro | Qiagen | 74034 | |
Slicing software: PrusaSlicer | Prusa | NA | Version 2.6.0 or higher |
Sterile Cell Strainer 100 µm | Fisherbrand | 22363549 | |
Surgical scalpel blade No. 21 | Swann-Morton | 307 | |
Trizol reagent | Thermo Fisher Scientific | 15596018 | |
Trypsin-EDTA (0.5 %) | Gibco | 15400054 |