The synthesis of the cRGD-DWIVA {[(cyclic(Arg-Gly-Asp-D-Phe-Glu)-Ahx-Ahx) (Ac-Asp-Trp-Ile-Val-Ala-Ahx-Ahx)]-Lys-βAla-Cys-NH ₂ } and cRGD-cDWIVA {[(cyclic(Arg-Gly-Asp-D-Phe-Glu)-Ahx-Ahx) (cyclic(Asp-Trp-Ile-Val-Ala-Glu)-Ahx-Ahx)]-Lys-βAla-Cys-NH ₂ } biomimetic peptides was performed by solid-phase peptide synthesis (SPPS). Fmoc-Rink Amide MBHA resin (164 mg, 0.4 mmol/g for the cRGD-DWIVA, and 150 mg, 0.04 mmol/g for the cRGD/cDWIVA) was used as a solid support. After placing the resin in a propylene syringe, the Fmoc group was removed with piperidine (20% piperidine in DMF, v/v) (1 × 1 min, 1 × 5 min and 1 × 10 min), followed by the addition of Fmoc-Cys (Trt)-OH (0.5 eq.), using OxymaPure (0.5 eq.) and DIC (0.5 eq.) for 90 min in DMF. The excess of reactive positions of the resin were capped with 31 µL of Ac₂O and 57 µL DIEA in DMF for 30 min. Subsequently, the building block (Fmoc-Ahx-Ahx-Lys (Alloc)-βAla) was incorporated stepwise using standard Fmoc/tBu chemistry (5 eq. of each Fmoc-protected amino acid) and OxymaPure/DIC as coupling reagents (5 eq. each).

In the case of the cRGD-DWIVA peptide, the partially protected cyclic RGD peptide {cyclic [R(Pbf)GD(OtBu)fE], 2.5 eq.—its synthesis and characterization were already published in (Oliver-Cervelló et al., 2022)} was incorporated using PyBOP (4 eq.), HOAt (4 eq.) and DIEA (8 eq.) at pH = 8 for 1 h in DMF. To ensure a quantitative yield, this reaction was followed by a resin washing and a re-coupling of the cyclic peptide [PyBOP (2 eq.), HOAt (2 eq.) and DIEA (4 eq.)]. The Alloc group of the Lys was then removed using catalytic amounts of palladium and two units of Fmoc-Ahx-OH were sequentially coupled to build the second peptidic branch. Finally, the DWIVA sequence was elongated using standard Fmoc/tBu chemistry and the N-terminus acetylated by treatment with Ac₂O/DIEA/DMF (1:2:7, v/v/v) (1 × 5, 2 × 10 min). Cleavage and side-chain deprotection of the peptide were done with TFA/TIS/H₂O (95:2.5:2.5, v/v/v) for 90 min. The obtained crude was dissolved in H₂O/ACN (1:1, v/v) and lyophilized to yield 96.2 mg of crude peptide. The peptide was purified by semipreparative HPLC.

For the cRGD-cDWIVA peptide, the coupling of cyclic RGD, Alloc removal and the insertion of Fmoc-Ahx-OH residues was performed as described for cRGD-DWIVA above. Next, 3 eq. of the partially protected cyclic [D(OtBu)W(Boc)IVAE] [details published elsewhere (Oliver-Cervelló et al., 2022)] were coupled to the peptidyl-resin using PyBOP (4 eq.), HOAt (4 eq.) and DIEA (8 eq) in DMF for 90 min. A recoupling of the cyclic peptide using the same conditions was performed to ensure the reaction completion. Finally, the cleavage of the peptide was carried out as previously described yielding 76.6 mg of the crude peptide, which was purified by semipreparative HPLC.

PEG hydrogels were fabricated by Michael-type addition reaction between maleimide-functionalized 4-arm PEG (PEG-4Mal) and dithiol cross-linkers, following a modified protocol described elsewhere (Trujillo et al., 2020). In detail, the required amount of PEG-4Mal (20 kDa, Biochempeg, United States) was weighted and dissolved in phosphate buffered saline (PBS) to have a final hydrogel concentration of 5% (w/v). Then, the corresponding amount of either cRGD-DWIVA or cRGD-cDWIVA biomimetic peptides was added to the PEG-4Mal solution to have a final concentration of 1 mM of peptide in the hydrogel. The functionalized PEG-4Mal solution was quickly mixed and incubated during 15 min at room temperature (RT) to ensure the peptide-maleimide reaction. Simultaneously, cross-linking solutions of PEG-diSH (3.4 kDa, CreativePEGWorks, United States) or mixtures of 50:50 molar ratio of PEG-diSH and a protease-degradable peptide (VPM peptide, GCRDVPMSMRGGDRCG, purity 95.5%, 1,696.96 g/mol, GenScript, United States) were prepared. Afterwards, 50 μL hydrogels (or 400 μL hydrogels for rheological measurements) were produced by mixing PEG-4 Mal or functionalized PEG-4Mal with the cross-linking solutions to have a 1:1 molar ratio between thiols and maleimide groups. After adding the cross-linkers, hydrogels were allowed to gel during 30 min at RT. For biological assays, cells were always mixed with the PEG-4Mal-peptide before adding the cross-linker at a density of 30,000 cells/hydrogel (600,000 cells/mL). Biomimetic hydrogels, i.e., hydrogels functionalized with either the cRGD-DWIVA or the cRGD-cDWIVA, were always cross-linked at 50:50 molar ratio PEG-diSH:VPM. As negative controls, non-functionalized but protease-degradable PEG hydrogels at 50:50 molar ratio PEG-diSH:VPM (PEG-50 condition), and PEG-only hydrogels (without VPM nor peptides, PEG-0 condition) were also designed. Of note, all hydrogel conditions were designed to present the same degree of cross-linking. Table 1 summarizes all the hydrogel conditions used in the present study.

All data presented in this work are given as mean values ± standard deviation. SPSS Statistics 24.0 software (IBM, United States) was used for statistical analysis. When normal distribution was satisfied, one-way ANOVA test with a post hoc pairwise comparison using Tukey’s (for homogeneous variances) or Tamhane test (for non-homogeneous variances) was performed. Otherwise, the non-parametric Kruskal–Wallis test was used. p values were considered significant if p < 0.05. For physicochemical characterization three (n = 3) samples per condition were used, while for biological characterization, each condition was replicated in triplets in each experiment (n = 3) and, for quantification, five pictures per sample were used to calculate cell area and viability.

Protease-degradable PEG hydrogels were designed with the main objective of recreating the 3D microenvironment of bone ECM. To this end, the hydrogels incorporated i) cell instructive peptides combining the integrin binding peptide RGD and the BMP-2-derived peptide DWIVA, previously shown by us to promote synergistic integrin-growth factor signaling (Oliver-Cervelló et al., 2021); and ii) the MMP-degradable sequence VPM (Turk et al., 2001). Thus, in order to produce the hydrogels, PEG-4Mal was first modified with the thiolated multifunctional biomimetic peptides (either with cRGD-DWIVA or cRGD-cDWIVA, Figure 1). Of note the use of the individual peptides (either RGD or DWIVA) or their combination as a mixture, without controlling their geometrical disposition, failed to support synergistic signaling (Oliver-Cervelló et al., 2021; Oliver-Cervelló et al., 2022), and thus these peptides were not included in the present study. The rationale for using cyclic peptides relies on the fact that conformational restriction is known to enhance the peptide’s receptor affinity and provide higher biological potential. Indeed, cyclization of RGD has been demonstrated to enhance the selectivity of this peptide towards integrins involved in cell adhesion and osteodifferentiation, such as αvβ3 (Mas-Moruno et al., 2016). On the other hand, cyclization of the BMP-2-derived DWIVA was shown to retain or even enhance the potential of the linear counterpart, although the exact effect of the conformation in its binding to BMP receptors has not been elucidated (Oliver-Cervelló et al., 2022). The peptide-functionalized PEG-4Mal was then cross-linked with dithiolated cross-linkers (PEG-diSH and/or VPM peptide—Figure 1A) according to a recently published protocol (Trujillo et al., 2020), to allow hydrogel formation and endow the system with protease-degradable properties. It should be mentioned that biodegradable hydrogels were always fabricated at a 50:50 molar ratio between the PEG-diSH and VPM cross-linkers. Non-degradable PEG hydrogels (PEG-0) were also used as negative (non-degradable) controls with the same degree of cross-linking as the functionalized hydrogels. Moreover, non-functionalized but biodegradable PEG hydrogels, i.e., with the VPM sequence (PEG-50), were also included as controls. It is important to remark that all hydrogels were always formed using the same concentration of cross-linkers, ensuring both the same degree of cross-linking for all hydrogel conditions and a stoichiometric balance between the free thiols present in the cross-linkers and the remaining free maleimide groups (unreacted) on the PEG-functionalized molecules. PEG hydrogels were formed by Michael-type addition, which is considered a very efficient reaction (Darling et al., 2016; Martínez-Jothar et al., 2018; Ravasco et al., 2019). In this regard, the thiol-maleimide conjugation is a click reaction, being fast, straight forward and easily controlled by pH modification (Darling et al., 2016; Martínez-Jothar et al., 2018; Ravasco et al., 2019). Interestingly, such Michael-type addition has been previously used in PEG-based hydrogels without having any detrimental effect in the bioactivity, compatibility and cell viability, mainly, because the reaction takes place under physiological conditions, without generating any by-products and with no requirements of adding initiating chemicals (which may be toxic) to control the reaction (Nair et al., 2014; Kim et al., 2016; Jansen et al., 2018). The high efficiency of the Michael reaction allows the control of the hydrogel stiffness, resulting in broader stiffness ranges compared to other cross-linking reactions (Phelps et al., 2012). It should also be mentioned that the thiol-maleimide reaction is not only used to cross-link the hydrogels, but also to functionalize them (as in the present study). Thus, cysteine-containing molecules, i.e., proteins, peptides or GFs, have commonly been used to modify the bulk PEG structure, providing hydrogels with high bioactivity (Lutolf and Hubbell, 2003; Cambria et al., 2015). In this regard, functionalized PEG hydrogels have been widely employed in biomedical applications, including drug delivery, regenerative strategies or surface modifications (Peyton et al., 2006; Zhu, 2010; Li et al., 2018; Jansen et al., 2022). For instance, VEGF-loaded-PEG hydrogels were engineered as a release platform to promote pancreatic islet vascularization (Phelps et al., 2015), while PEG-based systems modified with chitosan, enhanced proliferation and differentiation of neurospheres-like progenitors cultured in the self-healing hydrogels (Tseng et al., 2015). Furthermore, PEG is characterized by its high biocompatibility, easily tunable mechanical properties, resistance to protein adsorption and non-immunogenic reactions, which makes it an ideal candidate for tissue engineering (Alcantar et al., 2000; Zhu and Marchant, 2011; Chang et al., 2019; Mandal et al., 2020).

Physicochemical properties of the engineered hydrogels were first evaluated by SEM (Figure 2A). It is worth noting that the porosity values obtained from SEM images do not probably correspond to the actual porosity of the samples in the hydrated state due to the freeze-drying of hydrogels. However, SEM images allowed us to compare the internal structure and pore sizes in the different hydrogel conditions. The four hydrogels showed similar structural morphology with interconnectivity between different pores, indicating that neither the functionalization of the PEG with the biomimetic peptides nor the addition of the VPM during the cross-linking process had any significant effect in terms of structural morphology. Furthermore, the hydrogels were highly porous, with values about 98% of bulk porosity (Figure 2A). These results ensure that the possible different cellular behavior observed in the functionalized hydrogels can be attributed to the presence of the biomimetic peptides and not to their structure or porosity.

Rheological measurements were next performed to assess the viscoelastic properties of the hydrogels (Figures 2B, C). Such viscoelasticity is characteristic of this kind of materials, due to the intrinsic properties of the polymer and the great amount of water entrapped in its network. Having systems with viscoelastic properties may be a great advantage, as it has been demonstrated that many soft tissues and ECMs in the human body present this type of mechanical behavior (Chaudhuri et al., 2016; 2020; Chaudhuri, 2017). Thus, strain sweep measurements in the LVR were performed, showing that bulk non-degradable PEG hydrogels without functionalization exhibited a G’ of about 200 Pa (Figure 2B), which corresponds to a stiffness of about 600 Pa. Similar values were obtained when adding the biodegradable cross-linker and/or the biomimetic peptides in the hydrogels, indicating that none of both components had any significant effect in the G’ values. This value of stiffness may seem rather low for designing hydrogels for bone tissue engineering. Indeed, hydrogel stiffnesses for bone regeneration have been traditionally set in the range of >20 kPa, with values much more similar to the ones presented by osteoid tissues (Sen et al., 2009; Huebsch et al., 2015). Although such values may have a positive effect in the mechanotransduction phenomenon, at such range of stiffness, the ability of the cells to spread, grow and initiate osteodifferentiation through biochemical cues may be hindered. (Bao et al., 2017; Major et al., 2019). Furthermore, such difference in stiffness between our hydrogels (G’≈ 200 Pa) and the traditional ones for bone regeneration (G’ > 20 kPa) is probably translated in differences in degradability and structure of the systems, both having an important effect in cell traction forces developed during hydrogel degradation and the remodeling process of the cells. These cellular tractions have been demonstrated to be crucial in MSCs osteogenic differentiation, being mediated by the cell degradation process of the matrices and, more important, being independent of matrix mechanics (Khetan et al., 2013; Clark et al., 2020). Of note, finding the best approach depending on the application is paramount to ensure the successful performance of the hydrogel. Another interesting feature derived from the rheological measurements, is the similar damping factor presented by the different hydrogels (Figure 2C), indicating that the ratio between G” and G’ for all the conditions was very similar, without observing statistically significant differences. Thus, the viscoelastic properties of the four conditions were comparable, obtaining G” values of about 5 Pa for the four conditions.

The swelling ratio of the hydrogels was also measured (Figure 3A). This parameter is important to shed light on the hydrogel network structure as well as on how the distance between cross-linking affects the structure. In our system, the four hydrogels presented a similar swelling behavior, reaching values of 600%–650% at 24 h, with no statistically significant differences between conditions. In general, longer cross-linkers (PEG-diSH) will restrain less the swelling capacity of the hydrogels in comparison to shorter chains (VPM). Hence, degradable hydrogels (PEG-50, cRGD-DWIVA and cRGD-cDWIVA) would be expected to swell less than the PEG-0; however, this was not the case. On one hand, this could be explained by the fact that the differences in molecular weight between cross-linkers are not large enough to be reflected on the swelling of the hydrogels. In addition, the VPM peptide is slightly positive charged, which may enhance the affinity for water molecules in comparison to the PEG-diSH. Nonetheless, the swelling ratios obtained in this study are in the same order of magnitude than the ones found in the literature working with PEG-4Mal networks with similar concentrations as ours (Clark et al., 2020; Wei et al., 2020).

Moreover, mesh size of the hydrogels was also calculated either by applying the rubber elasticity theory combined with the G’ values from rheology or by the Flory-Rehner equations and the M ¯ c values (either theoretical or experimental—obtained from swelling assays). The results are summarized in Table 4. Mesh sizes calculated by the rubber elasticity theory were of ≈27–29 nm for all the conditions, with no differences between them. Interestingly, from the Flory-Rehner equation, slightly bigger mesh sizes for each condition were obtained using the theoretical M ¯ c in comparison to the experimental one. Nonetheless, the mesh size obtained by the three different approaches were very similar, indicating a good correlation between the assays. Of note, the mesh size of the hydrogels is in accordance with other works (Clark et al., 2020; Dobre et al., 2021). Such mesh sizes would allow the diffusion of nutrients and oxygen through the hydrogels and thus, support cell growth.

In addition, degradability of the hydrogels was studied by incubating them in a collagenase type I solution (Figure 3B). The degradation assay was initiated when hydrogels reached swelling equilibrium (after overnight incubation), considering the mass of the swollen hydrogels as the initial mass for the experiment. As expected, non-degradable hydrogels (PEG-0) were stable during 6 days of incubation. However, when introducing the VPM cross-linker, which is a protease-degradable sequence, hydrogels (PEG-50, cRGD-DWIVA and cRGD-cDWIVA) began to degrade, reaching a plateau at 72 h. After 2 h, the degradation rate was already noticeable, and PEG-50 hydrogels exhibited about 8% of mass loss, whereas cRGD-DWIVA and cRGD-cDWIVA hydrogels presented a higher, 15% of degradation. Of note, the differences observed in the degradation rate between PEG-50 and the biomimetic hydrogels (cRGD-DWIVA and cRGD-cDWIVA) were more evident with increasing incubation times, and, for instance, at 8 h, PEG-50 had a ≈25% of mass loss in contrast to the ≈55% of the biomimetic hydrogels, being twice faster their degradation behavior. Indeed, after 6 days, PEG-50 exhibited about 50% of mass loss, while the biomimetic hydrogels showed about 80% degradation. More insights would be required to better understand such differences in degradability; nonetheless, similar degradation rates have been observed in PEG hydrogels modified with biological cues and cross-linked with the VPM sequence (Trujillo et al., 2020).

Together, the data obtained from the physicochemical characterization of the hydrogels demonstrated the capacity to modify the intrinsic properties of the systems to control their stiffness, swelling and degradation rates. Of note, the similar properties, i.e., swelling, porosity and stiffness, shown in the different hydrogel conditions ensure that the possible changes observed in the biological characterization can be attributed to the presence of the protease-degradable sequences as well as the biomimetic peptides, and not to the physicochemical properties of the hydrogels.

After the physicochemical characterization, hydrogels laden with human MSCs were produced and their biocompatibility was assessed by cell viability studies (Figure 4). The hydrogels were prepared similarly to the ones for physicochemical assays, but in this case, cells were mixed directly in the PEG-4Mal solution (already functionalized with the biomimetic peptides) previous to cross-linking (see section 2.2 for details). Notably, the encapsulation of the cells in the hydrogels did not have any effect in the cross-linking process. Then, live/dead staining (live cells in green and dead cells in red) at 1, 3, 7, and 14 days was performed (Figure 4A). At each time point, most of the cells were alive, indicating the great biocompatibility of PEG. The highest number of dead cells was observed at day 1, which was associated to the fabrication process of the hydrogels, in which cells were subjected to high stress. On one hand, gelation of the hydrogels was allowed during 30 min at RT, and additionally, cells were thoroughly mixed with the PEG-4Mal solutions to achieve homogeneous distribution of the cells in the loaded hydrogels, both processes decreasing cell viability. For quantification purposes, cell viability was calculated as the ratio between live cells and the total number of cells within the hydrogel (Figure 4B). As expected, the lowest cell viability was observed at day 1, with values of about 80% for all the conditions. Nevertheless, cell viability improved over time, reaching values higher than 95% after 14 days in culture, with no significant differences observed between conditions. Furthermore, in PEG-0 hydrogels (not degradable nor biomimetic) cells presented a roundish-like morphology, characteristic of 3D microenvironments where physical constraints due to the polymeric dense matrix do not allow cell spreading (Major et al., 2019). However, from day 3 onwards, biomimetic hydrogels, i.e., cRGD-DWIVA and cRGD-cDWIVA, exhibited cell spreading, especially at 7 and 14 days (Figure 4C). Also, PEG-50 at day 7 started to present some spread cells, although to a lower extent in comparison to the biomimetic peptides, which confirms the importance of installing both biodegradability and bioactivity in the hydrogels to promote cell spreading.

To further study cell morphology, actin staining was performed at day 7 and 14 (Figure 5A). Cytoskeletal actin labeling confirmed the capacity of the cRGD-DWIVA and cRGD-cDWIVA biomimetic peptides to promote cell spreading due to the degradation of the hydrogel as well as the stimulation of the cells, showing well-defined actin fibers. On the contrary, cells embedded in PEG-0 were totally round. Although at 7 days PEG-50 hydrogels presented some spread cells, degradability alone was not enough to promote cell spreading and it was only when combined with the biomimetic peptides that cells were able to effectively spread. Indeed, quantification of cell morphology at day 7 (Figure 5B) showed that both biomimetic peptides promoted the highest cell spreading, reaching cell area values of about two times bigger than the PEG-50 control. Notably, no significant differences were observed between PEG-0 and PEG-50, verifying that degradability is not enough to support cell spreading if bioactive cues are not present in the systems. Moreover, cells were classified as round, spread and super spread depending on their area at day 7, considering a round cell every cell with smaller area than the biggest one in the PEG-0 condition (Figure 5C). Such classification showed that introducing degradability alone increased up to 33% the number of spread cells. Interestingly, when functionalizing the hydrogels with the cRGD-DWIVA all the cells were classified as spread, while the modification with the cRGD-cDWIVA resulted in 40% of spread cells and 60% of super spread cells, indicating a positive effect in cell area by the cyclization of DWIVA.

Interestingly, higher magnification of the embedded cells in the biomimetic conditions showed the characteristic morphology of osteoblasts, which is an initial indicator of the successful differentiation of the human MSCs towards the osteogenic lineage. Such cell morphologies have been also observed in PEG hydrogels loaded with MSCs with comparable stiffness values to our systems and presenting osteogenic differentiation cues (Shekaran et al., 2014; Clark et al., 2020; Nasello et al., 2020). Of note, only in the biomimetic hydrogels, MSCs were able to form “cell branches,” a characteristic phenomenon that has been also observed in other soft hydrogels for bone tissue engineering (Campos et al., 2016). We hypothesize that such “cell branches” may be related with the mesh size of the hydrogels, which was of ≈24–30 nm. This mesh size is not big enough to initially allow cell migration around the hydrogel. However, cell protrusions may be generated through the mesh due to the degradation process of the hydrogels (in which mesh size will progressively become bigger) and the presence of the biomimetic peptides.

The osteogenic lineage commitment of human MSCs was first evaluated by studying osteospecific gene expression at 7 and 14 days by means of RT-qPCR. In detail, Runx2, COL1A1, ALP, Osterix and OCN genes were measured (Figure 6A). At day 7, an overexpression of COL1A1, ALP, Osterix and OCN genes was observed for the cRGD-DWIVA condition in comparison to the negative PEG-0 control. Such differences were also statistically significant in comparison to the biodegradable hydrogels without peptides (PEG-50), except for the Osterix gene, which was also overexpressed in PEG-50, reaching similar values to the cRGD-DWIVA. Nonetheless, the overexpression of Osterix in PEG-50 hydrogels was not surprising, as some studies have related the overexpression of such gene with the activation of some MMPs, such as MMP13, MMP9 or MMP2, the last two highly involved in matrix remodeling (Dai et al., 2015; Liu et al., 2020; Nishimura et al., 2012). On the other hand, only ALP and OCN genes were overexpressed for cRGD-cDWIVA, indicating that cyclization of the DWIVA motif may have a detrimental effect (at least at 7 days) on the osteogenic capacity of the biomimetic peptide. The highest osteogenic expression at day 7 was observed for the ALP gene, for both cRGD-DWIVA (with almost a 7-fold increase compared to PEG-0) and cRGD-cDWIVA (with a 5-fold change). Remarkably, Runx2 was not expressed in any of the conditions at 7 days, although at day 14, it was observed an 8-fold increase for the cRGD-DWIVA and 2-fold for cRGD-cDWIVA, both showing a significant increase in comparison to PEG-0 and PEG-50. Noteworthy, at day 14, cRGD-DWIVA promoted a high expression of all the characteristic osteogenic genes, reaching values of 50-fold change for COL1A1. These results demonstrate the osteogenic capacity of the cRGD-DWIVA biomimetic peptide. However, the cRGD-cDWIVA was only able to express Runx2 at day 14, confirming the previous observation at day 7, in which the cDWIVA lost its bioactivity in terms of osteodifferentiation. We previously reported that in 2D (Oliver-Cervelló et al., 2022), the cRGD-cDWIVA peptide was able to maintain the osteogenic differentiation capacity in comparison to cRGD-DWIVA and, in some cases, even improve it. Such discrepancies reflect how critical is the translation of in vitro testing from 2D to 3D environments, in which cell behavior may substantially vary (Duval et al., 2017; Mirbagheri et al., 2019; Jensen and Teng, 2020).

Moreover, the overexpression of ALP and Osterix genes is paramount in our system, as they are involved in the Smad-independent pathway (Wei et al., 2015; Sano et al., 2017) [pathway triggered by our biomimetic peptides (Oliver-Cervelló et al., 2021)]. This is in agreement with the high expression of ALP at day 7 and Osterix at day 14 for the cRGD-DWIVA. Interestingly, PEG-50 hydrogels (incorporating the biodegradable cross-linker but not the biomimetic peptides), were also able to promote ALP expression, especially at day 7, which could be explained by the fact that such gene has been demonstrated to be involved in the formation of the organic phase of bone ECM. In this regard, COL1A1 and OCN genes are also influenced by bone ECM formation, which contributes to the high expression of both genes for the cRGD-DWIVA condition at day 14 (Frank et al., 2002; Granéli et al., 2014; Viti et al., 2016).

As biodegradable hydrogels containing the VPM sequence were engineered in this study, MMPs expression was also assessed. The full degradable sequence incorporated in the hydrogels was the GCRDVPMS↓MRGGDRCG peptide, the ↓ indicating the cleavable site for MMPs. Among the MMP family, MMP2 and MMP9 have been reported for being highly sensitive to the VPM cleavable site (Jha et al., 2016; Chen Weikai et al., 2021). Thus, MMP2 gene expression was also analyzed by RT-qPCR. In this regard, no expression of MMP2 was observed after 7 days in culture, which could be explained by the fact that cells did not have enough time to degrade the hydrogel network and produce new ECM. Indeed, at day 7, hydrogels still presented a good consistency. On the contrary, at day 14, PEG-50 hydrogels showed a 3-fold expression in comparison to PEG-0 (which did not include the degradable sequence), suggesting the activation of MMP2 and ultimately leading to VPM cleavage. More relevantly, the incorporation of both biomimetic peptides, either cRGD-DWIVA or cRGD-cDWIVA, significantly increased the MMP2 gene expression, highlighting the importance of adding biologically active cues, i.e., cRGD and DWIVA/cDWIVA, to stimulate cell behavior. These results are in accordance with the cell spreading data (Figure 5), in which it was shown that the degradation of the hydrogel (and consequently decreasing the physical restriction for the cells) was not enough to promote full cell spreading and that the presence of the biomimetic peptides was paramount to orchestrate cell behavior. However, the 3-fold expression of MMP2 observed in the PEG-50 at 14 days (as well as the 5-fold change expression of COL1A1) could be an indication of nascent protein secretion by the cells due to the presence of the VPM sequence, as both genes are involved in the remodeling of ECM and the production of proteins, which could contribute (together with the degradation—and thus the softening of the hydrogel) to the spreading of some cells in that condition (Figure 5). Of note, MMPs are transcriptional targets of Runx2 (Wessely et al., 2019). Hence, MMP2 expression may be dependent on Runx2 expression, as observed at day 14. Moreover, at that time point, cells have been able to degrade to a higher extent the hydrogels and thus, produce more ECM, contributing to the higher expression of MMP2, especially in the biomimetic conditions.

Gene expression results were summarized in a heatmap (Figure 6B), clearly showing the activation of a higher number of genes for the cRGD-DWIVA and cRGD-cDWIVA conditions, especially for the former at day 14.

To further verify the osteogenic differentiation as well as to correlate the change of the cell genotype with a modification of its phenotype, ALP activity was measured at 14 days (Figure 6C). The biomimetic peptides significantly enhanced ALP activity with regards to PEG-0. However, PEG-50 also expressed ALP activity, although to a lower extent compared to the biomimetic conditions. cRGD-DWIVA and cRGD-cDWIVA exhibited the highest ALP activity values, both presenting statistically significant differences in comparison to PEG-50. These ALP activity results are in accordance with the ones from RT-qPCR.

The results presented here differ from the work of Madl et al., where alginate hydrogels functionalized with an RGD and DWIVA mixture were not able to promote ALP activity (Madl et al., 2014). However, in their work, the spatial disposition between both motifs was not controlled, which seemed to be a key factor to promote synergistic signaling between both peptides, as we recently demonstrated (Oliver-Cervelló et al., 2021). Moreover, the biomaterial employed to produce the hydrogels was also different, which may influence cell behavior as well.

Very recent studies have demonstrated the importance of installing biodegradability in hydrogels for bone tissue engineering, being such parameter even more critical than the material stiffness to regulate osteogenesis (Sun et al., 2022). In this regard, Peng et al. showed that fast degradation of soft PEG hydrogels was essential to stimulate MSCs and to further trigger their osteogenic commitment (Peng et al., 2018). Similarly, Lutolf et al. demonstrated that bone remodeling in a critical defect in rat cranium was totally dependent on the proteolytic sensitivity of the gels (Lutolf et al., 2003). Moreover, the degradation of the hydrogels facilitated cell spreading and cellular traction, which contributed to the human MSCs osteogenic differentiation (Khetan et al., 2013). Such findings have not only been observed in bone regeneration but also in neurogenesis, where neural progenitor cell stemness was strongly related to the hydrogel biodegradability rather than the matrix stiffness, revealing that degradability may enhance cell-mediated matrix remodeling and thus, influence cell mechanoresponse (Madl et al., 2017). All these examples support our findings, in which, with low stiffness hydrogels, osteogenic differentiation is possible due to both the presence of the biomimetic peptides (that act as biochemical cues for the cells) and degradability (that allows cell spreading and induces cellular traction). Indeed, other hydrogels with soft properties have also shown the capacity to differentiate cells towards the osteogenic lineage (Mullen et al., 2013; Jha et al., 2014; Clark et al., 2020; Wei et al., 2020; Sun et al., 2022).

On the whole, our PCR data and the values of ALP activity clearly indicate the osteogenic commitment of MSCs on the functionalized hydrogels; nonetheless, taking into account that MSCs showed a notable elongated morphology and numerous cell-cell contacts, we wanted to exclude any possible (partial) differentiation towards the myogenic lineage. Thus, characteristic myogenic markers, namely, MHC staining and MyOD and Desmin gene expression, were also studied (Figure 7) (Xu et al., 2015; Yi et al., 2017). Human AoSMCs cultured on glass were used as positive control. MHC staining of the cells demonstrated that only the AoSMCs cultured on glass exhibited myosin, while the biomimetic hydrogels functionalized with either the cRGD-DWIVA or the cRGD-cDWIVA did not present any. The expression of the characteristic myogenic genes MyOD and Desmin was also investigated. Interestingly, none of the biomimetic hydrogels expressed these myogenic genes. Similar results were obtained by Wei et al., in which soft PEG hydrogels containing osteogenic cues, i.e., BMP-2, did not express myogenic markers (Wei et al., 2020).

All in all, these biological results demonstrate the capacity of the protease-degradable and biomimetic PEG hydrogels to promote cell spreading and human MSCs osteogenic differentiation.