Fibrillogenesis in collagen hydrogels accelerated by carboxylated microbeads

Collagen type I is a material widely used for 3D cell culture and tissue engineering. Different architectures, such as gels, sponges, membranes, and nanofibers, can be fabricated with it. In collagen hydrogels, the formation of fibrils and fibers depends on various parameters, such as the source of collagen, pH, temperature, concentration, age, etc. In this work, we study the fibrillogenesis process in collagen type I hydrogels with different types of microbeads embedded, using optical techniques such as turbidity assay and confocal reflectance microscopy. We observe that microbeads embedded in the collagen matrix hydrogels modify the fibrillogenesis. Our results show that carboxylated fluorescent microbeads accelerate 3.6 times the gelation, while silica microbeads slow down the formation of collagen fibrils by a factor of 1.9, both compared to pure collagen hydrogels. Our observations suggest that carboxylate microbeads act as nucleation sites and the early collagen fibrils bind to the microbeads.


Introduction
Cell culture is a powerful tool for studying and understanding biological systems.Two-dimensional cell culture has been the principal technique used for this goal; however, these models only represent a cellular monolayer cultured on glass or plastic substrates.Moreover, in tissues, cells do not reside in 2D monolayers; they live within a complex 3D matrix; thus, three-dimensional cell culture represents a more realistic model for mimicking the microenvironment of cells in living organisms [1].In 3D culture, cells can grow on different kinds of scaffolds and matrices made of natural extracellular matrix (ECM) biomaterials, synthetic ECM, ceramics, bioactive glass, metals, etc [2][3][4].Synthetic polymers have become popular candidates for these applications due to their controllable properties [5].In particular, hydrogels have been used extensively as synthetic ECMs to reveal the impact of specific physical and biochemical signals on cell behaviors [6].However, these materials cannot mimic fundamental cellular processes such as adhesion, proliferation, differentiation, or migration.Natural biomaterials, such as collagen, represent a better option for 3D cell culture.
Collagen type I is the most abundant protein in mammalian ECM.Due to its natural abundance, properties, and cellular interactions, fibrillar collagen matrices have been used as scaffolds in tissue engineering, three-dimensional environments for biophysical studies, and to develop biomedical devices [2,4,7,8].Both in vivo and in vitro, collagen type I can form viscoelastic gels with varied network structures and mechanical properties [9].Biomimetic fibrillar hydrogels form from collagen molecules via different pathways, and the mechanism of gel formation depends on the types of molecules [10].Fibrils can form a hydrogel as individual entities owing to association and/or entanglement, and the fibrils can form a bundle from fibers, which subsequently branch and/or entangle to form a hydrogel [10].Collagen molecules form fibrils with a length of 300 nm and a diameter of 1.5 nm [2], and the fibrils are assembled into fibers of tens of micrometers in length and 10-200 nm in diameter [10].
Fibrillogenesis, the formation of collagen fibrils, can be monitored by measuring turbidity, which is, to a close approximation, proportional to the amount of fibrillar material formed [8,9,11,12].In this technique, UV or visible light is passed through the hydrogel, and the light attenuation is measured over time, providing information such as gelation time.Turbidity increases are a direct result of lateral assemblage of linear aggregates and the formation of wide fibrils [13].However, with the turbidity assay, it is not possible to visualize the fibrils.For this goal, optical microscopy is used.In confocal fluorescence microscopy (CFM), staining the fibrils with fluorescent labels is required.Alternatively, collagen fibrils can be visualized using confocal reflectance microscopy (CRM).With this technique, it is possible to observe the formation process of collagen fibrils in real time and study their microstructure [8,9,11,14].
Additionally, collagen hydrogels with embedded microbeads have been used as models to study local remodeling of the ECM in 3D culture [15][16][17].In these studies, fluorescent or nonfluorescent microbeads are used as tracers to visualize the movement of the matrix around a cell using the particle tracking technique [15,16,[18][19][20].In this work, we studied the fibrillogenesis process in collagen type I hydrogels with different embedded microbeads using turbidity assay and CRM.We observed that microbeads embedded in the collagen hydrogels modify the gelation time and gelation rate, and our observations suggest that the collagen fibrils formed are binding to carboxylated microbeads, acting as nucleation sites.
Pure collagen hydrogels were prepared using nine parts of collagen and one part of 10X PBS, adjusting the pH of the mixture to 7.4 with NaOH 0.1 N [21].The collagen mixture and all reagents were kept at 4 • C to prevent gelation.To prepare collagen hydrogels with microbeads, the same methodology was followed, except that the beads were mixed with 10X PBS at high density (∼1.455 × 10 6 ), see figure 1.Finally, all samples were incubated at 37 • C to form the collagen hydrogels by the fibrillogenesis process.

Turbidity assay
Gelation kinetics for all the collagen hydrogels were quantified using a turbidity assay.For this purpose, 100 µl of the collagen samples were transferred to a 96-well plate.Absorbance was measured using a BioTek Epoch 2 microplate spectrophotometer (Agilent) and the data were recorded at 405 nm and 37 • C (n = 6 for each sample) at a sampling interval of 30 s for 180 min.Gelation time (t gel ), lag time (t lag ) and gelation rate (dAbs/dt) of all collagen hydrogel samples were calculated from the typical turbidity data.

Microscopy
Collagen hydrogel microstructure was visualized using the CRM technique, while fluorescent beads were recorded with CFM.For this study, an inverted confocal laser scanning microscope Zeiss Axio Observer, equipped with an LSM 880 confocal module (Carl Zeiss), was used.The microscope was equipped with an incubator (130-800 005, PECON).To generate CRM images, the samples were illuminated with a laser at 488 nm and the reflected light was collected with a photomultiplier tube [22].On the other hand, for CFM images, the sample was excited with a laser at 543 nm and emission light was collected using the settings for the spectra of Rhodamine B. All images were recorded using a 63X/1.4NA oil objective.The image size was 512 × 512 pixels and a 12 bit depth.
For fibrillogenesis experiments, a glass-bottom dish was placed in an incubator at 37 • C with a humidified environment for 30 min.Afterward, 25 µl of samples were transferred to a glass-bottom dish (see figure 1).Immediately, CRM and CFM images were recorded in the nearby area at the center of the sample every 30 s for 120 min.Images were taken 15 µm above the coverslip.Experiments were performed in triplicate.
Moreover, Z-stacks were acquired after fibrillogenesis experiments in different areas.CRM and CFM images were acquired in gelled samples.Image slices were collected for each sample with a step size of 0.34 µm and the thickness of the Z-stacks was approximately 20 µm.To obtain confocal images (CFM and CRM), the pinhole was set to one Airy unit for all images.For microstructure studies, 3D reconstructions were generated using 2D slides using ZEN Blue (Zeiss) and ImageJ software programs.

Image analysis
Raw images were processed and analyzed to estimate the pore diameter.All images were processed in the same way to define the parameters to measure.Firstly, in an image, the blue channel was separated, and brightness and contrast were adjusted.The minimum and maximum displayed values were averaged for the three samples for each collagen hydrogel type studied.The threshold image was adjusted using these values for each image.Then, four regions of interest (ROI) were chosen inside the image, avoiding microbeads.The area of each ROI was 128 × 128 pixels.Thus, we analyzed 4 ROI for 3 samples, i.e. 12 images for each collagen hydrogel type.
To determine the pore diameter of collagen networks, we developed a routine in Python based on the erosion method described by Franke et al [23].First, ROIs are binarized using a threshold value; pixels with intensity above the threshold are given a value of 0 (black), corresponding to a fibril pixel, while the pixels with intensity below the threshold receive a value of 1 (white), corresponding to a pore pixel.Next, a non-linear filter is applied to reduce the noise; this takes the mean values of neighboring pixels in a square of 3 × 3 pixels.Subsequently, the erosion algorithm fills up all pores with circles.In the image, a white pixel (pore pixel) is found, and a circle is drawn using Bresenham's algorithm; if the total circular area contains white pixels, then a gray circle is drawn.If the algorithm finds a white zone surrounding a gray circle, then all zones are converted to gray zones.Bresenham's algorithm starts with a radius equal to one pixel and increases the radius until the radius of the circle is greater than the pore.Finally, the algorithm evaluates the percentage of the pore surface (white pixels) that is filled with circles.Following Franke et al [23], the pore diameter in an image is set as the circle radius that fills 50% of the pore surface.Figure S1(a) in supplementary material illustrates this process.
Furthermore, the diameter of the fibrils was determined by coding a Python routine that is based on an autocorrelation analysis method described by Franke et al [23].This involved the computation of the 2D autocorrelation function of the intensities of pixels in the ROIs.Slices of the 2D autocorrelation function along X and Y were then fitted each with the sum of two Gaussian functions.One Gaussian fits the large-scale structure and the other the short-scale structure of the autocorrelation function.The width of the short-scale Gaussian fit is used as an estimate of the fibril diameter.Figure S1(b) in supplementary material illustrates this process.

Characterization of microbeads
For the microbeads employed in this study, carboxylate fluorescent polystyrene microbeads for 4.5 µm and silica microbeads of 2.5 µm, an aliquot of each type was diluted in PBS 10X, dried at room temperature and combined with analytical grade KBr.The resulting mixture was then pressed into tablet form for Fourier transform infrared spectroscopy analysis.Microbeads morphology and particle size were assessed using scanning electron microscopy (SEM, JEOL 7610F).Microbeads were dried onto Si substrates and coated with gold for further examination.

Turbidity assay
The kinetic gelation process was measured by assessing the turbidity of the hydrogel samples as a result of the optical density changes that occur during collagen gelation.In this work, a UV-Vis spectrophotometer was used in absorbance mode, at 405 nm, to study the kinetic gelation process.Data showed that the turbidity increases and reaches a plateau when the gelation process of hydrogel scaffolds was complete, see figure 2. The experiments were repeated six times.For each absorbance profile, a sigmoidal Boltzmann function was fitted [12].For each hydrogel type, the average of the six Boltzmann functions fitted is represented in figure 2. The shaded area of each curve represents the standard deviation of the measured samples.
In turbidity curves, three zones can be distinguished: a zone 1 in which no turbidity change is observed (lag phase), a zone 2 where the turbidity increases very rapidly (growth phase) and a zone 3, a plateau zone where the turbidity reaches its maximum value (plateau phase) [24,25].Figure 2 shows the turbidity curves obtained for the hydrogel samples; they all show a sigmoidal typical turbidity behavior with the three zones mentioned above.Nevertheless, the hydrogels show different behaviors depending on the microbeads mixed with the collagen hydrogel.Considering pure collagen (BCH, green curve) as the control sample, hydrogels with carboxylate microbeads (BCH/CFMB4.5, blue curve) have a faster gelation process, while the hydrogels with silica microbeads (BCH/MB2.5,gray curve) show a slower gelation process.To verify our hypothesis that carboxylated microbeads act as nucleation sites, we prepared two hydrogels mixed with fluorescent carboxylated beads and silica beads.On the one hand, we combined different particle sizes: 4.5 µm carboxylated fluorescent microbeads and 2.5 µm silica microbeads (BCH/CFMB4.5/MB2.5).On the other hand, we used beads with similar size: 2 µm carboxylated fluorescent microbeads and 2.5 µm silica microbeads (BCH/CFMB2/MB2.5).The turbidity curves obtained for these hydrogels are shown in figure 2. Both hydrogels exhibited a similar behavior; BCH/CFMB4.5/MB2.5 gelled in approximately 51 min, while BCH/CFMB2/MB2.5 gelled in 48 min, respectively.It is interesting to note that these hydrogels had a gelation time, lag time and gelation rate values that fall between those of BCH and BCH/CFMB4.5.This suggests that carboxylated beads act as nucleating sites, accelerating fibrillogenesis, while silica beads tend to delay that same process.These observations can support the results of other studies that used collagen matrix [7,26,27].
From the turbidity curves (Boltzmann fits), we calculated the gelation time (t gel ), defined as twice the time at which half the total change in absorbance was reached (t 1/2 ), the lag time (t lag ), defined as the intersection of a tangent to the growth portion of the curve with the time-axis, and the gelation rate dAbs/dt, defined as the slope of the absorbance profile at t 1/2 [24,25].Table 1 shows the values found for these parameters and their calculated standard deviation; the values are presented from lowest to highest gelation time.

Optical microscopy
To visualize the fibrillogenesis process and the structure of the hydrogels, we use CRM.CRM images reveal fibril formation over time; in our case, we follow this process for 120 min.Figure 3 shows examples of such observations, performed on  representative samples containing fluorescent or nonfluorescent particles.In the early stages (4 min), no fibrils are visualized in the images; only fluorescent microbeads are distinguished.After 30 min, fibrils in BCH/CFMB4.5 samples were visible, while in BCH and BCH/MB2.5 samples no structures were observed.At 60 min, in BCH/CFMB4.5 samples, a defined network was observed, but in the other samples, no fibrils were distinguished.After 90 min, fibrils appeared in BCH and BCH/MB2.5 samples, and at 120 min, a dense network was visualized in all samples, see movie S1 in supplementary information.A more detailed analysis, measuring frame by frame in movie S1, shows that in BCH hydrogels, fibrils started to be seen from 55 min, while in BCH/CFMB4.5 hydrogels, the time was 10 min, and 60 min for BCH/MB2.5 hydrogels.
After the gelling period, collagen hydrogels appear whitish on a macroscopic scale, confirming that the fibrillogenesis process is complete.At this point, CRM images reveal a well-defined homogeneous network for all collagen hydrogels, as shown in the column corresponding to 120 min in figure 3.For the sample with fluorescent beads (BCH/CFMB4.5), it is easy to distinguish the particles embedded in the collagen network.On the other hand, for the hydrogel with silica microbeads, particles are not fluorescent, so the microbeads are localized using transmission images.CFM images show particles in the observation plane.For clarity, only the merge of the CRM image and the CFM image is presented in figure 3, and the particle positions are indicated by red arrows.For the BCH/CFMB4.5images, in the column corresponding to 4 min, fluorescent particles appear yellow due to Rhodamine emission, but in the following columns the particles are white because the reflectance from the collagen fibrils increases in the blue channel.
Additionally, 3D reconstructions of collagen hydrogels were obtained through z-stacks, with 60 planes and a vertical step of 344 nm.In all samples, a dense collagen network was observed (see figure S2).It is interesting to note that around the carboxylated fluorescent microbead a concentration of collagen fibrils is observed, as seen in plane XY in figures 4(d) and (e).This point will be examined in the Discussion Section.

Collagen structure analysis
The structure of the collagen hydrogels was studied from CRM images.We chose ROIs in the samples avoiding microbeads.In total, 12 ROIs were analyzed for each collagen hydrogel type.To determine the pore diameter, the erode method was used as explained in section 2.5. Figure 5 shows the mean pore diameters.Error bars represent the standard deviation over the 12 ROIs.The data shows that the pore diameter remains fairly constant from case to case.This is consistent with the pore size being dependent on the collagen concentration [28] or pH [12,29,30].This suggests that the particles embedded into the network do not alter the pore size; they only accelerate or slow down the gelation rate.The mean pore sizes are given in table S1 in the supplementary information.
A similar behavior is noticed concerning the fibril diameter, which is represented as the yellow line in figure 5.This parameter does not show a significant variation for the different conditions.

Discussion
Multiple studies have used fluorescent and nonfluorescent particles embedded in collagen hydrogels to study local remodeling of the ECM or alignment of collagen fibers using particle tracking technique [15-19, 31, 32].However, these reports do not study fibrillogenesis.In this work, we studied the effect of embedded microbeads on the fibrillogenesis kinetics of bovine collagen type I hydrogels.For this reason, we used two different types of microbeads: fluorescent carboxylated polystyrene microbeads with a diameter of 4.5 µm (CFMB4.5)and silica microbeads with a diameter of 2.5 µm (MB2.5).
Typically, when a pure collagen is neutralized, adjusting the pH to 7.4, to form a hydrogel, molecules spontaneously self-assemble into fibrils in a stepwise process.Initially, collagen monomers aggregate into small and short fibrils (nano/micro-fibrils).At this point, small structures do not scatter or reflect light and the turbidity (or absorbance) is negligible, also, structures are not distinguishable in CRM images.During the lag phase, nucleating structures are formed and linear growth occurs to form fibrils. Structures larger provoke light scattering and turbidity increases, and small fibrils are recognized in CRM images.During the growth phase, fibrils become longer and wider due to lateral assembly, larger clusters are formed, and fibrils associate into a network.At this stage, turbidity increases rapidly, and more fibrils are visualized in CRM images.Finally, collagen monomers are depleted and the fibrils formation and growth end, and collagen forms a network of interconnected fibrils upon gelation; the number of fibrils is constant and turbidity plateaus, and a homogeneous and dense network is revealed by CRM [8,9,11,[33][34][35][36].
The rate of assembly of fibrils was monitored by measuring turbidity, which is, to a close approximation, proportional to the amount of fibrillar material formed.The turbidity assay showed that the gelation time (t gel ) was modified in the presence of microbeads, see table 1.For pure collagen (BCH), t gel was 86 min, while for collagen with carboxylated fluorescent microbeads (BCH/CFMB4.5) it was faster, at around 24 min.In contrast, for collagen with silica microbeads (BCH/MB2.5),t gel was slower than pure collagen, taking approximately 162 min for a complete gelation.BCH/CFMB4.5 had a gelation time 3.58 times faster than BCH and 6.75 times faster than BCH/MB2.5.This behavior was also noticed in the lag time and gelation rate.BCH/CFMB4.5 had the shortest lag time, followed by BCH, and finally for BCH/MB2.5.During the growth phase, the rate of turbidity changes (dAbs/dt) in the BCH/CFMB4.5condition is 2.26 and 17 times larger than in the BCH and BCH/MB2.5 conditions, respectively.In CRM, we observed a similar behavior: first, fibrils were observed in BCH/CFMB4.5,then in BCH, and finally in BCH/MB2.5 (see figure 3).
The formation of fibrils was monitored using CRM.Images from CRM showed that fibrils were self-assembling and growing.Initially, fibrils were not distinguishable due to their small size, as seen in insets 4 min of the figure 3.In the early stages, small fibrils appeared, followed by rapid growth in length resulting in longer and thicker fibrils (inset 30 min for BCH/CFMB4.5 and insets 90 min for BCH and BCH/MB2.5 in figure 3).Finally, a dense network of fibrils was formed, as seen in inset 120 min of the figure 3.This behavior was observed in all samples, but at different points in time for each collagen hydrogel type, which is consistent with the turbidity measurements.See figure 3 and movie S1 in the supporting material.
Our observations made in turbidity and timelapse CRM are consistent with the nucleation and growth model for pure collagen gelation, described above.However, the difference in gelation time, lag time and gelation rate between the pure collagen hydrogel and collagen hydrogels with microbeads is not explain with this model.For this reason, we studied the vicinity around the microbeads using CRM and z-stacks.Figure 6 shows representative regions of interest for each collagen hydrogel studied.Movie S2, in the supporting material, shows a sequence of zstacks for the different microscopies used.It is interesting to note that, in the CRM image a bright zone around CFMB4.5 was observed, and in the transmission image, filaments can also be seen around the carboxylated microbead.The merged image shows a good match on the filaments.On the other hand, for MB2.5, CRM revealed a less bright zone and fewer filaments around the silica microbead.In contrast, for pure collagen, collagen fibril clusters were not observed.This can also be observed in figure 4(e).
We think that the bright zone corresponds to the collagen fibrils around the microbeads and provide additional information to the diffraction patterns of beads [16].This suggest that a greater number of fibrils were observed around the carboxylated fluorescent beads than the silica beads, see movie S2.Thus, we hypothesize that the collagen fibrils bind to the carboxyl group on microbeads, and they could act as nucleation sites, accelerating the fibrillogenesis process.Newman et al studied the viscosity and elastic modulus of collagen matrices in the presence and absence of polystyrene latex beads (6 µm); they observed an acceleration of fibrillogenesis in the presence of a very dilute suspension of particles after the lag phase was completed.The authors hypothesized that the beads act as nucleation centers for collagen assembly due to the recruitment of collagen to the bead surface [26].Moreover, Tan and coworkers studied collagen-nanotube composite constructs with carboxylated nanotubes and their results suggested that these composites improved nanotube integration during collagen fibrillogenesis due to carboxylic group on carbon nanotube providing grafting sites for collagen molecules to interact with, thus serving as a sites for nucleation [27].
In the case of collagen hydrogel with the 2.5 silica microbeads, we observe a reduced number of fibrils around the silica beads, see movie S2.We think this is related to the fact that the gelation rate and gelation time are the smallest values compared with the other samples.Eglin and coworkers studied the effect of silica nanoparticles (12 nm) on the collagen fibrillogenesis process, and they indicated that silica nanoparticles inhibit fibrillogenesis; at high concentrations, silica nanoparticles aggregate along the collagen molecules, hindering their association and decreasing the amount of free collagen until the self-assembly process is inhibited [7].
A representative comparison between microbeads of similar diameter in the same hydrogel (BCH/CFMB2/MB2.5) is shown in figure 7. The figure presents the sum of 19 slices (6.5 µm thick) of a region around a 2 µm carboxylated bead (left) and 2.5 µm silica bead (right).The z projection shows a greater number of fibrils around the CFMB2 than in MB2.5, in accordance with our previous observations; see movie S3 in the supplementary information.The nucleation of collagen fibrils appears to be independent of the microbead size, however a more detailed studied is necessary.
Furthermore, we determined the average pore size of the collagen network for the different samples.Figure 5 reveals that, despite the variation in the embedded microbead sizes and composition, but under the same collagen concentration, no significant differences in pore size are observed between samples.These values align with the pore size of collagenous tissues in vivo, which can vary from 1 to 20 µm [25].The measured pore size values are comparable to pepsin-digested collagens, known to have pore diameters from 3 to 5 µm [37].This process often used to denature highly cross-linked collagens derived from bovine or porcine skin.Our analysis indicates that particles embedded in type I collagen hydrogels do not alter pore size, it only modulates the kinetics of hydrogel formation.
Similar results were obtained for the fibril diameter, with a maximum value of 319.5 nm for BCH/MB2.5 and a minimum value of 253.6 nm for BCH.A significant difference was not observed, as the difference between the maximum and minimum values is only 65.9 nm.This is smaller than the lateral resolution of the CRM (212.62 nm), suggesting that the embedded particles do not significantly alter the fibril diameter.

Conclusions
Collagen type I hydrogels represent a powerful option for studying cells in a 3D biomimetic environment, and their kinetics of polymerization depend on various parameters such as collagen type [33], pH [12,29,30], concentration [11,28,30], and temperature [9,30].We have demonstrated that the addition of carboxylated and silica microbeads can also modify the kinetics of polymerization of the collagen hydrogels.Using a turbidity assay, we observed that the gelation time, gelation rate, and lag time increased in the presence of carboxylated polystyrene microbeads, while these parameters decreased for silica microbeads.When we combined the two types of microbeads, we observed that the fibrillogenesis was slowed down compared with only carboxylated beads, indicating that the collagen fibrils interacted with both particles.Additionally, we monitored the fibril formation over time using CRM as a complementary technique.We observed how the fibrils grew until they formed a dense network, and the microbeads were embedded into the hydrogel.Moreover, we distinguished an accumulation of fibrils around the carboxylated microbeads, independently of their diameter, suggesting that the collagen fibrils bound to the carboxyl groups in the particles and grew linearly.Thus, carboxylated beads acted as nucleation sites, promoting the fibrillogenesis, while silica beads inhibited this process.Furthermore, we observed that microbeads did not alter the collagen network; the pore size was similar in all samples.
One important aspect that should be addressed in the near future is the effect of microbeads on the mechanical properties of hydrogels, especially when they are used as ECM models.For example, any eventual modification of the gel stiffness will need to be taken into account if tracking the forces of cells is to be estimated from the displacement of the beads.

Figure 1 .
Figure 1.Preparation of collagen type I hydrogels with microbeads.Gelation process was analyzed using a turbidity assay, and fibrillogenesis was visualized using confocal reflectance microscopy.

Figure 4 .
Figure 4. 3D views of regions of interest in a collagen hydrogel with carboxylated fluorescent microbeads (BCH/CFMB4.5).Reflected light is seen in blue while fluorescence appears in yellow-white.(a) Central plane of a gelled hydrogel volume (225 µm × 225 µm × 20 µm).(b) 3D reconstruction of the orange ROI; a fluorescent microbead is embedded in the collagen network.Orthogonal views for two ROIs, a zone without microbeads ((c), green ROI) and a zone with a microbead ((d), light blue ROI).(e) Two microbeads (CFMB4.5)at different heights; left picture displays a transmission image, while the right picture shows a merge of fluorescent and reflectance images.Scale bars: 50 µm (yellow) and 5 µm (red).
with three ROIs indicated by orange, green and light blue squares.Images were obtained from overlapping CRM images and CFM images.A 3D reconstruction of the orange ROI is presented in figure 4(b); in this volume a fluorescent microbead within the collagen fibrils network are visualized.Additionally, two orthogonal views into the volume are shown (square green and light blue).A zoom of carboxylated fluorescent microbeads is observed in figures 4(b)and (e), while a zoom of a zone without microbeads is presented in figure4(c).Beads appear to be elongated particles due to the axial resolution of the microscope.

Figure 5 .
Figure 5. Size of collagen fibrils and network of the hydrogels.The mean pore diameter in the collagen networks is similar for all hydrogels.Error bars represent the standard deviation.N = 12 images.

Figure 7 .
Figure 7. Sum of 19 slices of a region around a 2 µm carboxylated bead (left) and a 2.5 µm silica bead (right).Yellow circles indicate the position of the microbeads.More fibrils around the carboxylated microbead are observed.The volume of each is 25 µm × 25 µm × 6.5 µm.Scale bar: 5 µm.

Table 1 .
Gelation and morphology parameters obtained for collagen hydrogels with microbeads embedded.