An overview of bio-actuation in collagen hydrogels: a mechanobiological phenomenon

Graphical abstract

Due to their congruity with the native extracellular matrix and their ability to assist in soft tissue repair, hydrogels have been touted as a matrix mimicking biomaterial. Hydrogels are one of the prevalent scaffolds used for 3D cell culture. They can exhibit actuation in response to various stimuli like a magnetic field, electric field, mechanical force, temperature, or pH. In 3D cell culture, the traction exerted by cells on hydrogel can induce non-periodic mechanobiological movements (shrinking or folding) called ‘bio-actuation’. Interestingly, this hydrogel ‘tropism’ phenomenon in 3D cell cultures can be exploited to devise hydrogel-cell-based actuators for tissue engineering. This review briefs about the discrepancies in 2D vs. 3D cell culturing on hydrogels and discusses on different types of cell migration occurring inside the hydrogel matrix. It substantiates the role of mechanical stimuli (such as stiffness) exhibited by the collagen-based hydrogel used for 3D cell culture and its influence in governing the lineage commitment of stem cells. Lastly, the review also audits the cytoskeleton proteins present in cells responsible for influencing the actuation of collagen hydrogel and also elaborates on the cellular signaling pathways responsible for actuation of collagen hydrogels.

Introduction

An actuator is a machine component responsible for moving or ‘actuating’ a mechanism within the system (e.g. a solenoid). An actuator usually has restricted kinetics when activated by external stimuli such as physical, electrical, hydraulic, and air. On the other hand, a bio-actuator is a component of a biological system that can induce movement within that system, powered by a biological stimulus such as an insect, muscle, and cells (Albert Basson 2012). Hydrogel-based actuators are categorized under soft actuators, which use the property of swelling or shrinking of the hydrogel to achieve the actuation (Velders et al. 2017). The naturally occurring phenomenon of the shrinking or self-folding of hydrogels was observed in the 1980s (Allen and Schor 1983). Due to their similarity with the native extracellular matrix (ECM), hydrogels have been suggested as possible matrices mimicking biomaterials with the potential ability to assist in soft tissue repair (Mantha et al. 2019). Hydrogels are crosslinked polymers possessing a substantial amount of ‘bound’ liquid in 3 dimensions (Kopecek 2007). The aptness of hydrogels to associate with aqueous solvents arises from hydrogen bonding, as shown in Fig. 1 (Bahram et al. 2016). Hydrogels possess the ability to absorb a solute without disintegrating in it (Singh et al. 2016). Many natural and synthetic polymer-based hydrogels have been identified as potential biomaterials for tissue reparation (Bahram et al. 2016). Hydrogels can be categorized according to the different classification systems depicted in Fig. 2.

Fig. 1.

Uptake of water in a hydrogel. The functional group in the polymer backbone chain undergoes hydrogen bonding with water molecules and, as a result, swells up

Fig. 2.

Classification of hydrogels based on material, polymer composition, and gelation stimuli

Collagen is the primary structural protein present in vertebrates (Silver and Jaffe 2009). It is a naturally occurring protein-based polymer touted for its excellent biocompatibility, beneficial interaction with cells, and its superior biodegradation properties. Collagen is an essential protein that exhibits a triple helical structure as the functional unit (Shoulders and Raines 2009). Collagen is found in the skin and connective tissue and has a vital role in maintaining the structural integrity of ECM architecture (Shoulders and Raines 2009). The outstanding ability of collagen gel to mimic and restore the structural integrity of ECM is the reason why collagen-based scaffolds are receiving much attention in the tissue regeneration field (Mantha et al. 2019). During the wound-healing process, collagen plays a vital role in the induction of clotting, cell proliferation, migration, and final scar formation (Lodish et al. 2000). There are more than twenty-eight types of collagen (Lodish et al. 2000; Shoulders and Raines 2009), of which collagen Type 1 is the most widely used collagen in tissue engineering (Wu and Crane 2018). Gelation of collagen hydrogels can be induced by different stimuli such as temperature (Yunoki et al. 2013), pH (de Moraes and Cunha 2013), and ionic strength (Achilli and Mantovani 2010). Numerous reports have demonstrated that collagen hydrogels slowly shrink under in vitro conditions (Liu et al. 1998; Velegol and Lanni 2001; Galois et al. 2006; El-Fiqi et al. 2013). Cells cultured on collagen type 1 hydrogel have exhibited contraction, shrinkage, and/or actuation of the hydrogel due to traction exerted by cells onto the surface of the hydrogel (Bacakova et al. 2019). Additionally, the intrinsic mechanical properties of hydrogel such as compressive strain energy, tensile modulus, and compressive modulus can get altered by cells cultured on the hydrogels and vice a versa (Velegol and Lanni 2001). On culturing bovine chondrocytes on collagen type 1 gels, 50–60% of contraction was observed after 12 days of incubation. The expression of collagen type 2 plays an integral role in cartilage formation and with regards to this point the expression of type 2 collagen was higher in cells cultured on type 1 collagen hydrogel than on plastics. This finding implies that chondrocyte phenotype preferred hydrogel environment over plastic (Galois et al. 2006).

The present review aims to provide an overview of 3D cell culture in/on collagen hydrogels, describe the different types of migration that cells undergo in a hydrogel, and review the influence of stiffness as a biophysical factor in determining the lineage commitment of stem cells. This article also focuses on actuation caused by cells on hydrogels in terms of contraction/shrinkage/folding of hydrogels under cell traction forces. It audits the cytoskeleton proteins present in cells responsible for influencing the actuation of collagen hydrogel and also elaborates on the cellular signaling pathways responsible for actuation of collagen hydrogels. Lastly, the paper highlights actuation expressed by collagen hydrogels and collagen blended with other polymer hydrogels.

Discrepancies in 2D vs. 3D cell culturing on hydrogel

Although 2D cell culture was long regarded as the best in vitro culturing technique, 3D cell culture methods have emerged as the next-generation models for regenerative medicine (Bacakova et al. 2019). The ability of 3D culture environments to mimic the endogenous niche of native tissues by providing essential spatial microenvironments makes them superior to 2D culture techniques (Tibbitt and Anseth 2009). In terms of feasibility, 2D cell culture studies are easier to carry out because cells adhere faster and proliferate better than 3D. However, 3D cell culture provides better cell-to-cell interaction by allowing enhanced cell signaling in all three dimensions and not just at the ventral surface as experienced in 2D culture (Caliari and Burdick 2016a). Cells cultured in 2D tend to appear more relaxed and stretched than their natural state (Bacakova et al. 2019). This behavior is due to the availability of space in the latitudinal plane of the culture plate—a situation not present in in vivo conditions. In addition, 3D cell culture better mimics the natural ECM by promoting the formation of cell growth niches, thereby providing results closer to the in vivo investigation (Kapałczyńska et al. 2018). It has been observed that 3D cell culture influences cell fate, morphology, adhesion, migration, and also their behavior with neighboring cells (Baker and Chen n.d.). In contrast, 2D cultures do not suitably mimic the in vivo milieu, giving rise to discrepancies in results compared with 3D (Jensen and Teng 2020). Indeed, the growing interest in 3D cell culturing for stem cells is due to the enhanced stem cell differentiation into cardiomyocytes (Fleischer et al. 2019), chondrocytes, osteocytes (Yan et al. 2019), and neurons as observed by researchers (Tibbitt and Anseth 2009; Farrukh et al. 2017).

Prior research has shown a drastic difference in the results of 2D and 3D culture models. For example, relatively lower metabolic activities of the AKT (also known as Protein kinase B - PKB), mammalian target of rapamycin (mTOR), and S6K (also known as RPS6KB1) signaling pathways were observed for colon cancer cell lines when cultured in 3D than 2D (Riedl et al. 2017). Cultured murine cardiac fibroblasts cells grown in a 3D collagen gel were quiescent, resembling those found in normal myocardium. In contrast, the same cells grown in 2D culture in polystyrene dishes closely resemble myofibroblasts found in healing infarcts (Darby et al. 2014). Such phenotypical changes in 2D culture occur due to myofibroblast transdifferentiation-based generation of stress fibers from the upregulation of α-Smooth Muscle Actin (α-SMA); whereas, cells cultured in 3D acquire dendritic and circular morphology with extremely low levels of α-SMA and therefore no concomitant production of stress fibers (Darby et al. 2014). In addition, 3D cultured fibroblasts generated low levels of collagen proteins but high levels of MMPs (Shinde et al. 2017). Cardiomyocyte differentiation of human-induced pluripotent stem cells (hIPCs) in a 3D culture has several advantages over those grown in 2D. These include a long-term cultivating platform, an accelerated maturation, and a phenotype shift from atrial to ventricular cardiomyocytes (Fleischer et al. 2019). However, some 2D cell culture studies have shown better outcomes than 3D gels, such as those involving valvular interstitial cells (VICs) used for reparation of a heart valve. Indeed, culturing VICs in a 2D polystyrene plate led to a seven-fold increase in myofibroblast activation vs. that seen for VICs cultured in a 3D environment (Mabry et al. 2016).

In an effort to close the gap between in vivo models and 2D cell culture, 3D tumor models provide the best of both worlds. 3D culture provides microenvironments with spatial architecture similar to the in vivo model at comparatively lower cost, enhanced ease of handling, and simple optical cell tracking, all without the need for animal sacrifice. Furthermore, it is noted that the real-life clinical efficacy of anti-cancer drugs is often lower than that suggested from tests carried out in 2D culture models and more in line with studies made on 3D culture (Riedl et al. 2017; Lv et al. 2017; Melissaridou et al. 2019). In a recent report, the therapeutic effect of Doxorubicin and Resveratrol (DOX & RES) for the treatment of pancreatic cancer using 2D and 3D models were conducted, in which the cells grown in the 3D model showed greater resistance to DOX & RES due to upregulation of the anti-apoptotic gene Bcl2 (Barros et al. 2018). Similar results were obtained with different anti-cancerous drugs such as 5-Fluorouracil (5-FU), AXP-107-11, DOX, Gemcitabine, and H107 in the pancreatic cancer cell lines (MIAPaCa-2 and PANC-1) (Longati et al. 2013; Wen et al. 2013; Yeon et al. 2013) which suggest that 3D models are the more realistic for drug testing. Studies involving the in vitro culturing of breast cancer cell lines on collagen gel indicated that these cells carried out regular cellular activities such as forming tubular structures like that observed in in vivo (Yang et al. 1979). It is of interest to note that the presence of collagen ligands (α1β1, α2β1, α10β1, and α11β1) on collagen hydrogel activates β1-integrin in breast cancer cell lines via protein kinase pathway; a mechanism completely absent in 2D culturing (Wang et al. 1998).

Cell proliferation rates in 2D vs. 3D cultures have shown contradictory results depending on the cell line. For instance, rectal cancer cell lines CACO-2, the endometrial cancer cell lines EN-1078D and KLE, colorectal cancer cell lines DLD-1, prostate cancer cell line DU145(Adcock et al. 2015), and the glioblastoma cell line, U87 (Lv et al. 2016) all showed increased proliferation rates in 2D compared to that of 3D. However, other cell lines such as breast cancer cell line JIMT1, oral cancer cell line CAL27 (Adcock et al. 2015), and human embryonic kidney cells 293FT (Ouyang et al. 2015) all grew faster in 3D versus 2D (Edmondson et al. 2014).

Migration mechanism in hydrogel

Significant differences in spatial concentrations and deposition of various types of protein within the 3D hydrogel matrix govern the cell migration, and this phenomenon is termed gel polarization. Polarization can be influenced by an external stimulus such as pH, chemical, optical, or mechanical-based perturbations (Vu et al. 2015). Cellular protrusions such as lamellipodia and lobopodia located at the cell’s leading edge assist in cell motility for normal cells (Vu et al. 2015), while amoeboid cellular migration is mostly observed in cancer cell lines when cultured in 3D. Lobopodia is a blunt cylindrical protrusion that responds to polarised signaling, while lamellipodia are thin sheet/fan shape protrusions, which are not responsive to polarized signaling (Petrie and Yamada 2012). In 2D culturing, cells exhibit multiple lamellipodia and thereby undergo lamellipodia-based migration. In 3D culture, a cell respectively shows lamellipodia and lobopodia, depending on the intrinsic and extrinsic properties of the cell and matrix—causing the cell to adapt to either a lobopodia-based or lamellipodia-based migration. The ingrained route depends on RhoA-ROCK-myosin II signaling, while the extrinsic pathway is regulated by matrix stiffness and elastic behavior of the 3D hydrogel (Petrie et al. 2012). Since lobopodia-based migration is dependent on RhoA signaling, RhoA active cells grown in a 3D linearly elastic matrix undergoes lobopodian migration. However, if either of the criteria is not fulfilled, i.e., the 3D matrix is not linearly elastic, or the RhoA activity is reduced, the cells undergo lamellipodia-based migration (Petrie et al. 2012).

Extrinsic properties governing cell migration have been studied by culturing human foreskin fibroblasts (HFFs) on collagen gel with varying stiffness (Vu et al. 2015). On a stiff gel (0.6 to 6.4 kPa), the HFFs formed lobopodia protrusions for migrations, while on a soft gel (∼0.015 kPa), several lamellipodia were formed on the surface of cells to promote lamellipodia-based migration (Vu et al. 2015). In another study, human vocal fold fibroblasts (hVFFs) cultured on hyaluronic acid (HA)–gelatin (Ge) microgels demonstrated the coexistence of both lobopodia-based and lamellipodia-based mechanisms. The stress and strain behavior of the microgels exhibited a linear elastic modulus up to 0.75 strain and was then non-linear, implying that lobopodia-based migration occurred within the linear range and lamellipodia migration in the non-linear elastic region (Heris et al. 2016). In addition to the migration mechanism, a relationship between the pore size and migration mechanism has also been highlighted by Heris et al. (Heris et al. 2016). Greater pore size promoted lobopodia-based migration, while a lower porosity promoted the lamellipodia-based mechanism (Heris et al. 2016). Pelham et al., in their study on cell locomotion on collagen-coated polyacrylamide hydrogel, concluded that cells grown on softer hydrogels tend to follow lamellipodia-based migration (Pelham and Wang 1997). The different migration mechanisms are shown in Fig. 3. Furthermore, the different conformation of collagen hydrogel (monomeric and fibrillar) has also been shown influenced the lamellipodial and filopodial projections (Pelham and Wang 1997). Chinese hamster ovary (CHO) cells, Saos-2, and human fibroblasts grown on monomeric collagen promoted the formation of wide lamellipodia while fibrillar collagen promoted long thin projections similar to filopodia (Mercier et al. 1996). In addition, upon blocking β1-integrin in CHO and Saos-2 cells, filopodia formation was completely impeded (Mercier et al. 1996; Jokinen et al. 2004).

Fig. 3.

Different types of cell migration mechanisms experience by cells in 3D

Cancer cells such as fibrosarcoma, osteosarcoma, and breast cancer cell lines have demonstrated migratory behavior when cultured in the 3D hydrogel. Cancer cells have been postulated to migrate in an amoeboid pattern (Krakhmal et al. 2015). Glioblastoma cells on treatment with Rac1 inhibitor stimulated mesenchymal amoeboid transition in hydrogel but not in 2D culture due to lack of 3D architecture to promote mesenchymal amoeboid migration (Huang et al. 2018). Upon further blocking of RhoA (responsible for influencing cell migration), the cross-talk between Rac1 and RhoA signals was affected, resulting in inhibition of mesenchymal amoeboid transition. In addition, the migration speed of cells cultured in 2D with Rac1 inhibitor migrated the slowest, while cells in 3D (otherwise receiving the same treatment) migrated the fastest in an amoeboid pattern (Huang et al. 2018).

Cells undergo phenotypical changes when transferred from 2D to 3D platforms for culturing (Kapałczyńska et al. 2018). These changes can be regarded as a response to the structural/fibril arrangement and mechanical stiffness of the hydrogel. The influence of a particular environmental niche on the cell phenotype is reflected by the expression of the focal adhesion genes (Vinculin, Talin, Filamin ), cytoskeleton components (Actin, Myosin, tropomyosin), and extracellular matrix (ECM) proteins (collagen, metal matrix protein, fibrillin, heparanase, matrix metalloproteases) (Pineda et al. 2013). Valve interstitial cells (VICs), for example, grown in 2D, exhibited a much higher fold change in these genes than VICs grown in 3D (Mabry et al. 2016). In another study by Pineda et al., embryonic stem cells showed a higher degree of cytoskeletal elements and extracellular matrix proteins when grown in 2D than when cultured in 3D (Pineda et al. 2013).

Modulus of hydrogel controls morphology and the lineage commitment of human mesenchymal stem cells (hMSCs)

Cell morphology is dependent on the matrix stiffness of the hydrogel (Engler et al. 2006; Caliari and Burdick 2016b; Tong et al. 2016). Engler et al. were the first to show that the cells sense the microenvironment and that stiffness determines the lineage and morphology (Engler et al. 2006). Gels with a compressive modulus similar to the modulus of native niche present in the body tend to direct cells to their respective native lineage. For example, matrigel coated onto a polyacrylamide gel of low modulus (0.05–0.3 kPa) tends to mimic a softer environmental niche such as that found in the brain (Georges et al. 2006) and causes the development of more filopodia protrusion resembling nerve cells than on the stiffer matrix (0.3–5.5kPa) (Flanagan et al. 2002). Engler et al. further demonstrated that collagen-coated polyacrylamide (PA) gel of stiffer modulus (8–17kPa) could influence stem cells to express myocadiac markers (Engler et al. 2004). Stem cells cultured on PA-gel coated with fibronectin having a stiffness of 62–68 KPa induced osteogenic differentiation of stem cells by integrin α5/β1 signaling pathway through a presumed mechanotransduction effect (Sun et al. 2018). In the absence of osteogenic media, stem cells cultured on collagen-I coated PA-gels of different stiffness influenced the stem cell fate. In hydrogels possessing a stiffness of 25kPa, a significant increase of 4-fold in RunX2 expression was observed compared to soft hydrogel (2kPa) (Sun et al. 2018). An increase in osteocalcin expression was also observed; however, the data was not statistically significant (Sun et al. 2018). On the other hand, when induced to undergo osteogenesis, a significant increase (16 fold) in RunX2 expression was observed (Sun et al. 2018). Chaudhuri et al. discovered a relationship between stress relaxation of the hydrogel and governance of stem cell fate (Chaudhuri et al. 2016). The force that a cell exerts when cultured on top of gel results in a reactional force of comparable magnitude. This force is translated within the gel leading to a remodeling of the matrix, which causes the stress to diminish in magnitude over time—a process termed stress relaxation (Chaudhuri et al. 2016). An alginate gel with low stiffness (9KPa) and higher stress relaxation time facilitated a high degree of adipogenesis while a high degree of adipogenesis , a stiffer gel (17KPa) with lesser stress relaxation time facilitated enhanced osteogenesis. These observations indicated that adipogenesis of hMSC is proportional to stress relaxation time while osteogenesis is inversely proportioned to stress relaxation time (Chaudhuri et al. 2016). Farrukh et al. demonstrated that progenitor cells from different sources require dissimilar physical milieus to differentiate into the neuronal lineage (Farrukh et al. 2017). Their study highlighted that a biofunctionalized hydrogel made up of polylysine or laminin, respectively possessing a stiffness of 2kPa and 20kPa, differentiates embryonic and adult neural progenitors into neurons respectfully (Farrukh et al. 2017).

Lee et al. made an interesting discovery concerning how dynamic changes in the stiffness of a hydrogel can affect a stem cell's propensity to differentiate (Lee et al. 2015). MSCs cultured on PA-gel having ~0.5KPa and then transferred to gel with minimum 30KPa stiffness for ten days; reflected an increase in osteogenic marker (RunX2 and osteopontin). On the other hand, the transfer of cells from a high to low stiffness gel increased neurogenic markers (β3tubulin and MAP2). An exception of the study was the expression of RunX2, which increased as shown in Fig. 4 and was observed to be independent of the gel's stiffness even when the cells were transferred from high stiffness to low stiffness and vice-versa (Lee et al. 2015).

Fig. 4.

Expression of a) osteogenic (runx2 and osteopontin) and b) neurogenic (β-tubulin and MAP2) markers before and after switching the substrate (0.5 and 40 kPa) reproduced from reference (Lee et al. 2015). Reproduced with permission under license number 4990111306962 from Springer Nature

Some reports suggest that YAP (Yes-associated protein) localization plays a crucial role in

mechanotransduction, and it is affected by the microenvironment. It controls the spreading of cells and changes in their conformation caused by traction forces exerted by cells along with differentiation of progenitor cells (Dupont et al. 2011). Experimental results have revealed that in 3D collagen gels, YAP localization in the cytoplasmic region is correlated with well-spread fibroblasts, while on a 2D substrate, YAP localization in the cell nucleus was correlated with well-spread cells. MSCs seeded on collagen hydrogel with varied stiffness were seen to encourage cell spreading with an increase in stiffness. Collagen hydrogels with 0.3kPa stiffness showed a failure to spread across the gel, while hydrogel with 19kPa stiffness enabled cells to spread in a lamellae-like fashion (Baker et al. 2015).

Highlighting the role of cytoskeleton proteins

TGF-β Actuation in collagen hydrogel

Each cell line has a slightly varying signaling mechanism than the other, which is controlled by a set of cytoskeleton proteins which in turn regulates hydrogel contraction (Liu et al. 1998; Albert Basson 2012). A simple demonstration of this point can be shown by experiments blocking the fibronectin protein in fibroblasts cultured on collagen hydrogel. This situation restricts hydrogel contraction for fibroblasts, but not in the case of epithelial cells (Liu et al. 1998). When bovine retinal pigment epithelial (RPE) cells were cultured on collagen gel, it was stipulated that protein kinase C (PKC) is responsible for the contraction of gels. This study showed that the PKC pathway could mediate gel contraction induced by cytokine Transforming Growth Factor-β (TGF-β) (Sakamoto et al. 1994). Raymond et al. provided proof that RPE-mediated collagen gel contraction is stimulated by TGF-β (Raymond and Thompson 1990). Different cell lines such as rat mesangial cells (Kagami et al. 1999) and trabecular meshwork cells (Nakamura et al. 2002) have exhibited TGF-β stimulated contraction collagen gel. TGF-β has been shown to influence collagen-binding integrin α1β1 in regulating the migration of cells and cell-induced contraction of gel (Nakamura et al. 2002). An upregulation in protein and mRNA expression of α1β1 under the influence of TGF-β has also been noted. In MG-63 cell lines, the induction of TGF-β upregulated α2β1 expression (Riikonen et al. 1995).

In addition to that described above, an upregulation in α5β1 and ECM proteins such as fibronectin, collagen1 were shown to be regulated by TGF-β (Kagami et al. 1996). The signaling mechanism of TGF-β for activation of ECM genes is thought to occur through both Smad-dependent and Smad-independent pathways (Van Caam et al. 2020). Lai et al. discovered that the upregulation of β5 integrins upon treatment with TGF- β can be assisted through the sp1 and Smad signaling pathways (Lai et al. 2000). In contrast, the synthesis of ECM proteins like fibronectin is likely activated via the Smad-independent pathway (Hocevar et al. 1999.). Studies have also shown that any modulation inactivation of TGF-β -Smad and Rho/ROCK signaling results in inhibition of collagen production and reduced contraction of myofibroblasts (Jiang et al. 2015).

Smad3 has been shown to mediate the TGF-β1-induced collagen gel contraction in human lung fibroblasts (Kobayashi et al. 2006). A relationship between Smad3, TGF-β, and α-SMA was established by Kobayashi et al. suggesting that for stimulation of α-SMA through TGF-β, Smad3 is required (Kobayashi et al. 2006). Moreover, the ability of TGF-β1 to initiate fibroblast-mediated collagen gel actuation is dependent on the Smad3 signaling pathway. The blocking of the Smad3 pathway completely restricts the increase in the α-SMA expression, which is in turn co-related to a cell's ability to induce contractibility (Kobayashi et al. 2006).

Role of collagen-binding integrin receptors in collagen matrix contraction

Integrins are integral extracellular proteins present on the cell membrane, which act as the main matrix receptors responsible for binding and responding to ECM proteins such as collagen, fibronectins, laminins (Schultz and Wysocki 2009). Integrins act as the first point of interface between the cytoplasm and the ECM and act to both signal and mediate cytoskeleton arrangement for adhesion and interaction with the surrounding environmental niche (Schultz and Wysocki 2009). Briefly, every integrin fragment comprises two glycoprotein units (one α and one β subunit), which vary in their ligand-binding specificity between different cell types. Integrins can be classified into subsets based on their detection capabilities, as shown in Fig. 5 (Alberts et al. 2002; Margadant et al. 2011). There are four collagen-binding integrins present in mammalian cells having the same β1 submit with unique α subunits, mainly α1, α2, α10, and α11, making up the four heterodimers (Margadant et al. 2011).

Fig. 5.

Classification of integrins subsets based on their detection capabilities. Inspired from reference (Margadant et al. 2011) under license number 4990120673192 from Elsevier

The function of different collagen-binding integrins is tabulated in Table 1. Of the different alpha subgroups of integrin-β1 which bind to collagen, α1 and α2 have been studied the most extensively (Lowell and Mayadas 2011). The role of integrin α1β1 and integrin α2 β1 on collagen reorganization and cell spreading has revealed contrasting results. On studying the interaction of collagen fibrils with integrin through immunoelectron microscopy, it was observed that in comparison to monomeric collagen, integrin α1β1 could mediate cell spreading better fibrillar type 1 collagen (Lowell and Mayadas 2011). In contrast, integrin α2β1 encouraged the formation of long cellular projections (Lowell and Mayadas 2011). These results suggest that integrin α2β1 plays a cellular receptor role while integrin α1β1 plays a crucial role in binding with type 1 collagen monomers (Jokinen et al. 2004). In vitro investigation in transfected CHO cell line were able to express either α2β1or α1β1, reiterated that integrin α1β1 did not play any crucial role in cell spreading. In contrast, integrin α2β1-mediated gel contraction was 1.6 times more than CHOα1β1cells (Jokinen et al. 2004).

Table 1.

Function of different collagen-binding integrins

S.No	Integrin/receptor	Description	Ligand	Effect on contraction of collagen gel upon blocking or expressing
1.	α1β1/ Collagen	Present abundantly in smooth muscle cells vascular, visceral, pericytes and endothelial cells and connective tissue cells (fibroblasts, chondrocytes, mesenchymal stem cells, and white blood cells)mesangial cells (Kaneko et al.)	Col I, III, IV, IX, XIII, XVI and non-fibrillar Col I-IV and monomer Col I	Blocking β1 almost completely stops gel contraction,; α1 blocking strongly reduces contraction but not as much as β1blocking (Carver et al. 1995; Kagami et al. 1999, 2000)
2.	α2β1/ Collagen	Major collagen receptor in epithelial cells and platelets, Keratinocytes, interstitial collagen-I-rich matrices (such as fibroblasts, T-cells, myeloid cells and megakaryocytes )	Col I, III, IV, V and XI, XVI and XXIII proteoglycans, (lumican and decorin) Endorepellin (HSPG2)	Blocking β1 almost completely stops gel contraction ; α2 blocking reduces contraction rate slightly but not as much as α 1 and β1 blocking (Schiro et al. 1991; Jokinen et al. 2004)
3.	α10β1/ Collagen	Mainly found in cartilage (chondrocytes) & junctional fibroblasts, chondrogenic mesenchymal stem cells	ColIX,XXII, IV/VI Cannot recognisenon fibrillar collagen	Dysfunction of growth plate chondrocytes (Bengtsson et al. 2005)
4.	α11β1/ Collagen	Present subsets of fibroblasts and to mesenchymal stem cells (undergoing osteogenesis). mesodermally derived cells. 518 human (foreskin fibroblasts, human satellite cells XXVI a lesser degree keratocytes and HT1080 fibrosarcoma cells (Tiger et al. 2001)	Col IX,I IV	C2C12 lacks α chain. On transfection with α11β1, C2C12+ α11β1 cells mediated collagen gel contraction. (Tiger et al. 2001)

*Col collagen

The propensity of collagen gel to provide an appropriate environmental niche in 3D cell culture experiments allows the cells to interact as a collective unit. This theory was justified by an experiment conducted by Ishida et al. (Ishida et al. 2014). They cultured Madin-Darby canine kidney (MDCK) cells to form a sheet followed by deposition of collagen gel on top. On interaction with the hydrogel, the cells adhered and exerted traction forces resulting in the folding of the hydrogel (Ishida et al. 2014). This study highlighted the role of cytoskeleton protein integrin-β1in the actuation of collagen hydrogel. Inhibition of integrin-β1 using inhibitor delayed the gel folding. This result strengthens the claim that cytoskeleton protein integrin-β1 plays a crucial role in the actuation of hydrogels (Ishida et al. 2014).

In another study, human bronchial epithelial cells (HBECs) were shown to cause a contraction in collagen gels (Liu et al. 1998). Upon variation of the concentration of collagen, the percentage of shrinkage varied inversely. For instance, 91% shrinkage with 0.5 mg/ml collagen, while 43% with 1.5 mg/ml collagen. It is interesting to learn that, for epithelial cell lines when cultured on collagen type 1 hydrogel, the blocking of β-1 integrin resulted in a decrease in the extent of contraction by more than 50 %, while the α integrin subfamily (α2, α3, α-vβ5) did not inhibit the gel contraction. However, the combination of alpha subfamily integrins with β-1 was able to inhibit the contraction of gel completely (Liu et al. 1998).

Role of α-SMA in fibroblast, hsp90 in contraction ability of different cell lines cultured on collagen gel

In a study carried out by Li et al. using hepatic stellate cells (HSCs) grown in 3D collagen gels, metformin (a drug used for the treatment of diabetes) attenuated gel contraction, whereas platelet-derived growth factor (PDGF) counteracted the drug’s effect (Fig. 6) (Li et al. 2018). Increasing the concentration of metformin resulted in a decreased expression of vascular endothelial growth factor (VEGF), thereby implying a role of VEGF in cell-induced collagen gel contraction (Li et al. 2018). In the same study, α-SMA, a hallmark marker of mature myofibroblasts and HSC activation, was inhibited inversely proportional to increasing dosages of metformin (Li et al. 2018). In another study, Zhang et al. highlighted the role of VEGF and TGF in the contraction of collagen gels upon seeding the gel with retinal epithelial cells (Zhang et al. 2012). They concluded that upon treatment with VEGF and TGF, the contraction rate was approximately twice and four times, respectively, more significant than the control group. However, administration of TGF formed more permanent contraction and had a significant increase in protein expression of α-SMA than VEGF (Zhang et al. 2012). Concordant with Li et al.’s work, a significant change in protein expression of α-SMA was observed upon administration of VEGF, but an overexpression was observed on treatment with TGF (Zhang et al. 2012).

Fig. 6.

Metformin restricted the gel contraction (e), whereas PDGF enhanced the gel contraction (b). upon addition of both metformin and PDGF, the gel contraction property was hampered, as can be seen in images c, d, and e compared to control gel (a). Adapted from reference (Li et al. 2018) under license CC BY-NC 4.0

Similarly, investigating murine cardiac fibroblast cells cultured on collagen gels with serum/fibroblast growth factor (FGF)/TGF resulted in inconsistent expression of α-SMA (Shinde et al. 2017). On administration of serum/FGF, the contraction of collagen gel was chaperoned with decreased expression of α-SMA and matrix metalloproteinase than the control group, leading the authors to conclude that contraction ability of fibroblasts are not associated with increased expression of α-SMA; implying that contraction of the gel is proportional to the downregulation of α-SMA expression (Shinde et al. 2017). However, after knocking down α-SMA, the contraction ability of the collagen gel decreased in serum condition, contradicting the above-reported conclusion (Shinde et al. 2017). Consistent with other reports, fibroblasts on induction with TGF resulted in collagen gel contraction assisted with upregulation of α-SMA.

A comparative study of subcutaneous fibroblasts (SCFs) and lung fibroblasts (LFs) to understand the role of α-SMA in fibroblast contractility when cultured on collagen gels was carried out by Hinz et al. (Hinz et al. 2001). Their findings suggested that upon addition of TGF-β1, SCFs (showing low expression levels of α-SMA in the quiescent state) increased α-SMA expression while LFs (showing high expression levels of α-SMA in the quiescent state) attenuated expression of α-SMA. Transfection of α-SMA cDNA in 3T3 fibroblasts led to a significantly higher collagen gel contraction, implying increased expression of α-SMA is ample enough to enhance fibroblast contractility (Hinz et al. 2001).

Interestingly, an increase in the gel contractility properties of dermal fibroblast cells was observed after exposing the cells to an electric field of 50 and 200mV (Rouabhia et al. 2013). A significant increase was observed between cells exposed to the electric field and those not exposed (control), as shown in Fig. 7. At the same time, an increase in the secretion of FGF-1 and FGF-2 was also observed relative to the control cell, implying that upregulation of growth factors leads to increased contractility (Rouabhia et al. 2013). Subsequently, an increase in the expression of α-SMA, proportional to exposure time and electric field intensity, was detected, thereby validating the role of α-SMA in regulating the contraction of gel cultured with fibroblasts (Rouabhia et al. 2013).

Fig. 7.

An increase in the contractility of dermal fibroblast, undergone electric stimulation was observed to be proportional to exposure time and electric field intensity. Adapted from reference (Rouabhia et al. 2013) under Creative Commons Attribution License from Plus One

Fibronectin, an ECM protein that plays an essential role in wound healing (and copiously expressed in fibroblasts), has been shown to influence the expression of α-SMA (Beyeler et al. 2019). In a fibronectin null (FN^{−/− )} cell line, cultured over fibrin hydrogel, the expression of α-SMA was minimal, whereas in positive fibronectin cells (FN^f/f ), expression of α-SMA was detected as shown in Fig. 8 (Beyeler et al. 2019). It is interesting to note that on culturing FNf/f and FN−/− fibroblast cells over collagen gel, FN−/− fibroblast failed to spread and adhere to the gel. In contrast, FN^f/f Fibroblasts spread widely onto the collagen hydrogel surface. Upon exogenous addition of fibronectin, the number of FN^−/− fibroblast adhered on collagen gel surface increased (Beyeler et al. 2019). These experiments emphasize the role of fibronectin in gel contraction, an aspect that is often overlooked.

Fig. 8.

Immunofluorescence staining of α-SMA actin. FN^f/f and FN^−/− fibroblasts were grown on fibrin hydrogel. Immunofluorescence staining of cells cultured on fibrin gel was carried out 24 h after gel detachment from the tissue culture plate. Several FN^f/f fibroblasts were positive for α-SMA, indicating the existence of myofibroblasts, which were not observed in FN^−/− fibroblast cultures. Adapted from reference (Beyeler et al. 2019) under license CC BY

The interaction of Interleukin (IL-1β), a pro-inflammatory cytokine and bacalin, a drug derived from a Chinese herbal medicine shown to have anti-fibrotic and anti-inflammatory properties, was investigated on nasal fibroblast cultured on collagen gel (Shin et al. 2016). IL-1β is known to stimulate extracellular matrix production, upregulating α-SMA levels (Shin et al. 2016) However, the addition of bacalin, showed downregulation of α-SMA expression followed by attenuation of collagen contraction via MAPK and Akt/ NF-κB pathways (Shin et al. 2016).

Heat shock protein 90 (hsp90) is a chaperone protein responsible for essential cell signaling and maintenance of the cell's proteostasis response (Hoter et al. 2018). Downregulation of hsp90 is targeted in cancer cell lines as an anti-cancer treatment (Darimont 1999). The inhibition of hsp90 in cancer cell lines has been shown to decrease cell motility, as observed by Taiyab et al. (Taiyab and Rao 2011). Additionally, the inhibition of hsp90 has also been shown to reorganize the F-actin component of the cytoskeleton and induce loss of integrin-linked kinase(ILK) from focal adhesions to form elongated cells (Taiyab and Rao 2011). Hsp90 inhibitors activate a non-hsp90 receptor protein paxillin, which in turn increased fibrillar adhesion, thereby restricting cell movements (Radovanac et al. 2013). Thus, hsp90 inhibitors such as 17AAG, 17DMAG, and radicicol have been shown to restrict the contraction (Taiyab and Rao 2011; Henke et al. 2016). Radicicol and 17-DMAG in cancer-associated fibroblasts act through the Rho/ROCK pathway to restrict the contraction (Henke et al. 2016). Studies have revealed that blocking the ROCK pathway by ROCK inhibitors downregulates the α-SMA expression and subsequently inhibits collagen gel contraction (Ibrahim et al. 2019). Along with hsp90, (ILK) has also been shown to govern cell migration and ECM modeling. Upon inhibiting hsp90, the stability of ILK is also impaired. Since gel contraction is orchestrated by cell traction forces, in ILK -/- cells, contraction of the hydrogel in the ILK-/- group did not take place (Radovanac et al. 2013).

Influence of Wnt/β-catenin pathway in actuation of gels

To illuminate the role of the Wnt/β-catenin pathway in the actuation of the hydrogel, researchers inhibited the Wnt/β-catenin signaling using an inhibitor IGC-001 (Kim et al. 2017). IGC-001 is a small molecule formerly used as a colorectal cancer drug (Eguchi et al. 2005). It is reported that IGC-001 binds to the CREB-binding protein and blocks the interaction between CREB- binding protein and β-catenin, thereby inhibiting Wnt/β-catenin signaling (Emami et al. 2004). The treatment of human fibroblasts with ICG-001 resulted in a significant decrease in the production of type-I collagens in a dose and time-dependent manner (Kim et al. 2017). The inhibition of contraction of gels seeded with fibroblast upon treatment with IGC-001 was also significant, as shown in Fig. 9. This confirms that the Wnt/β-catenin pathway also plays a crucial role in collagen actuation (Kim et al. 2017).

Fig. 9.

The Effects of ICG-001 (inhibitor of Wnt/b-catenin pathway) human fibroblast when cultured on collagen hydrogel. The Non-treated cells are labeled as N, cells treated with DMSO, and labeled D. The expression of type I collagen was determined by western blotting (a), where ICG-001 noticeably inhibited collagen synthesis. Collagen gel contraction(b) was significantly inhibited in the ICG-001-treated group compared to control groups (non-treated and DMSO-treated groups). Adapted from reference (Kim et al. 2017) with permission under license number 4990120116426 from Elsevier

The role of Wnt/β-catenin signaling in the contraction of gel by dermal fibroblast was also demonstrated by Mullin et al. with the help of a novel non-coding RNA Wincr1 (Mullin et al. 2017). It was reported that Wincr1 shows gene-regulatory activity and plays a critical functional role in dermal fibroblast behavior, especially in cell migration and collagen contraction (Mullin et al. 2017). To elucidate the function of Wincr1, both overexpressed and knocked-down cell lines were used (Mullin et al. 2017). For overexpressed cell lines, lentivirus was used as a construct (thus named LV-Wincr1). After culturing cells on collagen for 48hr, the gels were detached manually from the sides of the well. After detaching within 14 h of incubation, a significant contraction produced by LV-Wincr1 cells was observed. Hence, the role of Wincr1 in the actuation of collagen gel was confirmed, as shown in Fig. 10 (Mullin et al. 2017).

Fig. 10.

The role of non-coding RNA Wincr1 in collagen gel contraction demonstrates LV-Wincr1 fibroblasts have a positive and correlating role in collagen contraction. Adapted from reference (Mullin et al. 2017) under Creative Commons Attribution License (CC BY)

Contraction behavior in fibrin-collagen gel

Fibrin is a natural polymer that can be made from purified plasma proteins thrombin and fibrinogen (indeed, the name ‘fibrin’ itself is an amalgamation of the name of these two proteins). Fibrin plays an important role in hemostasis cascades and serves as a natural matrix to specific cells during wound healing (Li et al. 2015). Fibrin sealant has been used extensively in surgeries as a hemostatic adhesive for cardiovascular (Rousou 2013), thoracic (Jessen and Sharma 1985), bone, cartilage (de Barros et al. 2016), and skin (Schuh et al. 2014) and for promoting nerve repair (Janmey et al. 2009; Spotnitz 2014). In addition to sealants, coagulation of fibrin at a site of injury helps to provide a biodegradable platform for angiogenesis (Hadjipanayi et al. 2015) The biocompatibility of fibrin gel, controllable degradation rate, non-linear elasticity, and adhesive property, make fibrin gel an attractive biomaterial for tissue engineering applications. Fibrin gels can also be used as conduits for cell proliferation, differentiation, and tissue regeneration (Li et al. 2015).

As highlighted by Nakamura et al., the morphology of cells is governed by the blend of the hydrogel they are grown over. For example, human trabecular meshwork (HTM) cells grown on a blend of fibronectin and collagen hydrogels showed different behavior on fibronectin collagen hydrogel and pure collagen hydrogel (Nakamura et al. 2003). The addition of fibronectin facilitated the spreading of the HTM cells. Consequently, this resulted in the formation of actin stress fibers and an increase in the expression of integrin α5 (Nakamura et al. 2003). The 3D cell culturing of fibroblast on gels is somewhat controversial. Some studies have shown that fibroblasts can exhibit spread morphology, well-defined actin stress fibers in collagen gels having 104 Pa and 391 Pa modulus (Ali et al. 2014). In contrast, other studies report that fibroblasts cultured on collagen gel have more spindle-shaped phenotype with fewer lateral protrusions and reduced actin stress fiber relative to those grown on 2D matrices (Hakkinen et al. 2011).

Through its action as a protease, thrombin converts fibrinogen into fibrin. Its concentration plays a critical role in determining the mechanical properties of the fibrin hydrogel that is formed. In a study led by Shaneen et al., rat aortic smooth muscle cells (rSMCs) were seeded on fibrin hydrogel prepared at varied thrombin concentrations ranging from 1.0, 0.1, 0.01, and 0.001 units of thrombin/mg of fibrinogen. After 7 days of incubation with rSMCs, it was observed that the fibrin gel underwent more contraction when prepared at lower thrombin concentration (Rowe et al. 2007). In addition, decreases in thrombin concentration led to an increase in the diameter of the fibrin gel fibers (as observed by scanning electron microscopy), thereby confirming that thrombin concentration influences the mechanical properties of the fibrin hydrogels (Rowe et al. 2007). An interpenetration polymer network (IPN) hydrogel (constituted by a blend of fibrin, hyaluronic acid, and tyramine) was used to investigate the effect of fibrin gel contraction on the cell traction forces exerted by fibroblasts (Lee and Kurisawa 2013). Upon enzymatic degradation (using plasmin), the INP hydrogels inhibited gel contraction due to the interaction of Hyaluronic acid and the Tyramine network (Lee and Kurisawa 2013).

A novel technique for creating a hollow tube by careful direction of the traction force exerted by cells was discovered by Huang et al. (Shabnam Virji et al. 2004). The method involved pinning the gel at the two opposite ends of the culture dish and seeding the myoblast cells directly onto the gel resulting in the contraction of the gel leading to the formation of a conduit (Shabnam Virji et al. 2004). The formation of these ‘myotubes’ after 10 days of incubation was regarded as a result of cell-mediated contraction forces. In addition to the folding of gel, a twitch force of up to 329 ± 26.3μN and a tetanic force of up to 805.8 ±55 μN was generated by these muscles (Huang et al. 2005).

For neuronal repair, hydrogel conduits embedded with cells of interest represent a promising alternative to autologous nerve grafts (ANG). The main drawback of the ANG' gold standard' treatment is both the limited availability and the loss of healthy donor nerves (Kornfeld et al. 2018). Schuh et al. created a fibrin-PLGA based electrospun sheet loaded with fibrin gel encapsulating rat adipose-derived stem cells (rASCs) (Schuh et al. 2014). This study aimed to differentiate rASCs into Schwann cell-like cells (SCLs) on the fibril gel to allow proliferation and transplantation of SCLs through a conduit to treat peripheral nerve defects. On culturing rASCs in a fibrin gel, the self-folding mechanism of the gel was observed after seven days of incubation, as shown in Fig. 11. The ability of the gel to actuate into a tubular structure (supported by pinning the gel) is related to the biomechanical forces exerted by the rASCs and SCLs cultured within the matrix. However, the study lacked an in-depth analysis of the protein responsible for the folding mechanism (Schuh et al. 2014).

Fig. 11.

Fibrin gel scaffold construct with SCLs seeded on the electrospun scaffold on days 0 ( A ), 7 ( B ), and 14 ( C ) after cell seeding. D Scheme of the fibrin gel scaffold construct. Adapted from reference (Schuh et al. 2014) with permission under license number 4990111025542 from Karger Publishers

Although 3D cell culturing on fibrin blended with collagen, hyaluronic acid, and polyglycolic acid has been shown to submit to cell-mediated contraction; a mixture of fibrin, collagen, and glycosaminoglycan can resist contraction produced by cultured cells (Brougham et al. 2015). In comparison to fibrin gel, collagen-fibrin-glycosaminoglycan (CFG) gel respectively showed a 6- and 30-fold increase in the compressive modulus and tensile modulus. Upon seeding of human vascular smooth muscle cells (hSMCs), the CFG gel did not undergo contraction after 7 days, whereas the fibrin gel underwent a 90% shrinkage in comparison to its original diameter (Brougham et al. 2015). In a comparative study done between pure collagen gel, pure fibrin gel, and collagen-fibrin blend of 1:1 ratio; both pure gels underwent a higher degree of contraction than the collagen-fibrin blend upon the seeding of rat aortic smooth muscle cells (Cummings et al. 2004). Additionally, the collagen-fibrin blend presented itself with moderate values of linear modulus, while pure collagen and fibrin gels were at the higher and lower end of the series, respectfully. The linear modulus of pure collagen gel at 2 and 4 mg/ml protein content was 191 and 242 kPa, respectively, while pure fibrin exhibited the least linear modulus values of 28kPa and 19 kPa at 2 and 4 mg/ml protein content, respectively (Cummings et al. 2004). At the same time, collagen-fibrin blend demonstrated moderate moduli of 153 and 116 kPa at 2 and 4 mg/ml protein content, respectively. Contrarily, the highest ultimate tensile strength of 49.7 kPa, was achieved by equal collagen–fibrin blend at 2 mg/ml, which was significantly greater than the UTS of pure collagen (36.1 kPa) and pure fibrin (15.6 kPa) at 2mg/ml. A new experimental technique to help understand the mechanobiological forces exerted by cells causing actuation of fibrin hydrogel was demonstrated by De Jesus et al. (De Jesús and Sander 2014). They placed fibroblast explants on fibrin gel in a microscope-mounted bioreactor. The fiber realignment and contraction process was captured using differential interference contrast (DIC) microscopy. This technique allowed them to capture the alignment of the fibers within the gel. The alignment of fibers was a result of traction forces exerted by fibroblast cells. Although the technique allowed measurement of strain within the gel, a wide range of properties such as boundary conditions, in situ gelation, and migration of cells, the geometry of gel, modulus of gel and stress distribution are yet to be investigated (De Jesús and Sander 2014). Beyeler et al., used fibronectin null cell lines on a fibrin hydrogel to assess the gel contraction. Despite lacking the fibronectin gene, essential in contraction of wounds, fibronectin null cell lines were able to adopt fibrin hydrogel niche to execute contraction. Formation of fibronectin fiber assembly was confirmed by immunofluorecence (Beyeler et al. 2019).

Contraction behavior in hyaluronic acid

Hyaluronic acid (HA) or hyaluronan is a non-sulfated glycosaminoglycan comprised of disaccharide units of glucuronic acid and N-acetylglucosamine forming a linear polysaccharide. It is found abundantly in the skin, followed by synovial fluid, vitreous of the eye, and bodily fluids (Papakonstantinou et al. 2012; Liu et al. 2017). Using hyaluronic acid for treating burn wounds, combined with other biopolymers, has been shown to reduce scarring and wound contraction and enhance the mechanical properties of biomaterials. Various HA derivatives such as thiol functionalized hyaluronic acid (HA-DTPH), polyethylene glycol diacrylate (PEGDA)-crosslinked HA-DTPH (HA-DTPH-PEGD) have shown limited contraction. However, conversely, studies have also shown that HA-collagen hydrogels contract in vitro, with the degree of contraction directly proportionate to the HA concentration (Mehra et al. 2006).

Future applications

One possible approach to the building cell-hydrogel actuators could follow the tissue origami route (Kim et al. 2015). Indeed, using myosin-II-mediated contraction in mesenchymal stem cells, Hughes et al. developed a cell-driven tissue folding technique (Hughes et al. 2018). For such an approach, it is possible that both collagen fibers and gel could be used as a motif for folding in the 'tissue origami' philosophy. The principal objective is to take a collagen gel slab and to fold it along a predefined trajectory using mesenchymal condensation—a critical stage in intramembranous ossification leading to chondrogenesis (chondrogenic differentiation of stem cells during embryonic development) (de Barros et al. 2016). The basic idea is to use the condensates to direct the tissue folding. Cells with the potential to exert high traction forces are paired with low traction forces so as to be controlled by the former. Mouse embryonic stem cells and primary human mammary fibroblasts have a higher rate of contraction than epithelial and endothelial cell lines, allowing the former to guide the folding trajectories without any hindrance by the latter cell lines (Hughes et al. 2018).

Concluding remarks

This study has provided an elementary framework for conceptualizing new ways to actuate hydrogels using cells capable of degrading collagen hydrogels. Such understanding is important in many bioengineering fields, including controlled bioactive molecule delivery, cell encapsulation for controlled three-dimensional culture, and tissue engineering.

Declarations

Conflict of interest

The authors declare no competing interests.