2.1. Parameters of Microspheres
Ca
2+-crosslinked ALG MPs are often used for delivering protein factors since they are able to retain their bioactivity well [
23,
24,
26]. In addition, the release pattern of the loaded factor can be regulated by majorly changing the applied amount of Ca
2+ ions [
23,
26]. In this study, MPs with potency in injectable applications are needed for administering the release of PDGF-BB and BMP-2 in a sequential fashion in which PDGF-BB was released earlier than BMP-2. Hence, ALG MPs were employed as a carrier for delivering PDGF-BB. Blank ALG MPs were first produced to optimize their size and crosslinking in order to save the costly PDGF-BB. An emulsification method was used to prepare blank ALG MPs by controlling two major parameters: the concentration of ALG solutions and the amount of used CaCl
2. The optimal compositional proportion and processing conditions for these MPs are provided in the experimental section.
Figure 1A shows a representative SEM image for the blank ALG MPs. The image exhibits that these MPs had good sphericity and smooth surfaces with varied sizes ranging from several microns to around 20 microns.
By grafting hydrophobic polylactide (PLA) oligomers onto the CH backbone, the resulting chitosan-polylactide (CH-PLA) copolymers could be soluble in water-based solvents while having some merits stemmed from CH and PLA [
30,
31]. To this end, a series of CH-PLAs with varied PLA weight percentages was synthesized by selectively grafting PLA to the C-6 sites of CH units and leaving the amino groups of the CH backbone free for subsequent crosslinking. CH-PLA with a PLA content of around 31 wt% was selected for the preparation of CH-PLA MPs in view of the solubility of CH-PLA in 1.0% acetic acid aqueous solutions. Blank CH-PLA MPs were also prepared with an emulsification method by using sodium tripolyphosphate (TPP) as a crosslinker considering the good in vivo safety of TPP [
32].
Figure 1B presents a SEM image for the blank CH-PLA MPs. These MPs are seen to be spherical and have their size less than 20 μm. Several sets of blank ALG MPs and blank CH-PLA MPs were measured for their mean size and zeta (ζ) potential, and the obtained data are listed in
Table 1. It can be observed from
Table 1 that in contrast to BM-I MPs, BM-II MPs had a mean size of around 8 μm, which is significantly less than that of BM-I. Additionally, BM-II MPs exhibited a positively charged surface feature, which is indicated by their high positive ζ potential. As mentioned earlier, the factor-encapsulated MPs need to be embedded into the CH/GP gel for constructing composite gels. A concern was thus found to arise from the positively charged surface property of BM-II MPs because our tentative experiments revealed that the BM-II-embedded composite gels had a pronouncedly elevated gelling temperature and considerably prolonged gelling time, making these gels inapt for the potential use in the clinic.
ALG MPs had a negative ζ potential, and accordingly, it would be feasible to alter the surface charging nature of CH-PLA MPs by coating the CH-PLA MPs with an ALG layer in the light of anionic features of ALG. In order to endow the blank ALG-coated CH-PLA MPs with a mean size similar to that of the ALG MPs, an attempt was made to prepare smaller blank CH-PLA MPs so that they could become similar to ALG MPs in mean size after being coated. Data in
Table 1 indicate that the blank ALG-coated CH-PLA MPs indeed had a mean size very similar to that of ALG MPs without significant difference (
p > 0.05), and meanwhile, their ζ potential also became similar to that of ALG MPs (
p > 0.05).
Figure 1C shows a SEM image for BM-III MPs. These MPs bear similarity to BM-I MPs in size and morphology, and clearly, the inserted fluorescent image further demonstrates the presence of a coating layer on the surface of these MPs.
Table 1 shows that BM-I MPs had a large swelling index (SI) whereas BM-II MPs and BM-III MPs had much smaller SI. As described in the experimental section, BM-II MPs were prepared using TPP as an ionic crosslinker, and the CH component in the BM-II MPs has hydrophilic and swelling nature. As a result, BM-II MPs could be swollen to some extent because the TPP-connected network in the MPs could have become loose when the MPs were exposed to an aqueous environment. On the other hand, the CH-PLA used for preparing BM-II MPs contains around 31 wt% of PLA. Hence, a large number of PLA side chains in CH-PLA molecules would entangle CH-PLA chains together to prevent the MPs from swelling because PLA is highly hydrophobic and non-swelling in water [
33]. Therefore, the relatively small SI for BM-II MPs can be mainly ascribed to the effect of the PLA component in these MPs. When compared with BM-II MPs, BM-III MPs have an ALG coating layer, but this layer could exert a very limited impact on SI of BM-III MPs because the coating layer is very thin (see
Figure 1C) and has been hardened by Ca
2+ ions. Accordingly, BM-III MPs have their SI similar to that for BM-II MPs without significant difference (
p > 0.05). In the case of BM-I MPs, although they were crosslinked by Ca
2+ ions, the employed ALG component is water-soluble, and their ALG chain network built by ionic linkages would be loose in a wet sate; these two factors would jointly lead to a large SI for BM-I MPs.
On the basis of investigations for blank MPs, PDGF-BB and BMP-2 were separately loaded into ALG MPs and ALG-coated CH-PLA MPs, using the same methods and formulations respectively applied to the preparation of BM-I and BM-III MPs, in order to achieve the desired MPs with a mean size and electrical surface property similar to their corresponding blank MP counterparts. Some results for these factor-loaded MPs are presented in
Figure 2 and
Table 2. The two SEM images signify that both kinds of factor-loaded MPs had their size and morphology similar to their blank MP counterparts.
Figure 2B reveals that the coating layer of these MPs is thin and looks similar to that for BM-III MPs, their blank counterpart (see
Figure 1C). Data in
Table 2 indicate that the addition of growth factor into these MPs seems to not exert significant impacts (
p > 0.05) on their mean size, ζ potential, and SI when compared to their corresponding blank MP counterparts. These results are reasonable, since the same preparation methods and the same formulations that were applied to their respective blank MP counterparts were employed, and in addition, the mass of the encapsulated PDGF-BB or BMP-2 in these MPs is very low in comparison to the mass of the MP matrices themselves.
It was found that the encapsulation efficiency (EE) for factor-loaded MPs was significantly affected by the fed amount of factor. The EE of factor-loaded MPs was therefore optimized by altering the fed amount of factor while keeping the composition of MPs and the preparation conditions constant to achieve possibly higher EE. The obtained EE for two types of MPs is shown in
Table 2. It can be seen that LM-1 MPs had an EE of around 59%, which is much lower than that for LM-2 MPs. The difference in EE between LM-1 and LM-2 MPs is understandable if more details for these MPs are revealed. It is known that ALG is a water-soluble polysaccharide, and hence, a certain amount of PDGF-BB in the superficial layer of LM-1 MPs would be easily washed away during the preparation of LM-1 MPs even though these MPs were crosslinked by Ca
2+ ions. Unlike LM-1 MPs, LM-2 MPs are composed of a core formed by BMP-2-encapsulated CH-PLA MPs and a shell formed by an ALG layer. Inside their core CH-PLA MPs, PLA branched chains in the CH-PLA component are hydrophobic and could be shaped as rigid hooks for griping BMP-2 molecules and CH-PLA chains together to form entanglements during the preparation of CH-PLA MPs. These entanglements would certainly prevent the loss of BMP-2 and result in a high EE for the CH-PLA MPs. The ALG coating layer for the BMP-2-encapsulated CH-PLA MPs was solidified by Ca
2+ ions, and this layer would further act as a barrier to prevent BMP-2 molecules from being washed away during the preparation of LM-2 MPs. Consequently, LM-2 MPs have a high EE for loading BMP-2.
2.2. Gelation Properties of Hydrogels without Factor Load
In view of the similarity in sizes and ζ potential between the factor-loaded MPs and their blank MP counterparts, a series of MP-embedded composite gels was first prepared by embedding BM-I and BM-III MPs into the CH/GP gel in order to save the costly factors, and the resulting gels were used for evaluation of composite gels. CH/GP gel is known to have thermo-sensitive features near physiological temperature and pH [
25], and the amount of MPs inside the composite gels was thus regulated within a proper range to retain the thermo-sensitive properties of the composite gels.
Figure 3 shows variations of modulus versus temperature for several gels as well as the typical sol-gel transitions for two kinds of gels. T
i for GL-0 gel was around 36 °C, whereas T
i for GL-1, GL-2, and GL-3 gels was around 35, 35.2, and 34.3 °C, respectively, suggesting that the gelling temperature of composite gels does not significantly change when compared to that for CH/GP gel.
Figure 3F shows that the gelling time of GL-3 gels became significantly shorter than that of GL-0. Several sets of gels were measured to determine their pH, T
i, and gelling time, and the obtained results are summarized in
Table 3. Data listed in
Table 3 indicate that the MP-embedded gels had a pH and T
i that is very similar to that for CH/GP gel without significant differences (
p > 0.05). These results demonstrate that MP-embedded composite solutions are capable of forming into gels near physiological temperature and pH.
Table 3 shows that GL-1, GL-2, and GL-3 had significantly shorter gelling times when compared with GL-0; the difference in gelling time could be attributed to the concentration effect of the employed composite solutions. It can be seen that the total amount of MPs embedded into GL-1, GL-2, and GL-3 gels was the same, and it is comparable to the amount of CH, meaning that the composite solutions used for preparing GL-1, GL-2, and GL-3 gels had the same concentration but were significantly thicker than that used for the preparation of GL-0 gel. Taking account of the commonness of gel matrix for all these gels, it can be inferred that the shortened gelling time for GL-1, GL-2, and GL-3 gels should be correlated to the increased concentration of the corresponding composite solutions when compared to GL-0 gel.
In principle, an injectable gel needs a fast sol-gel transition with durative stability because otherwise it could transude from the applied site and randomly migrate from one site to another, leading to unwanted effects [
22]. On the other hand, it usually takes time to prepare the injectable mixture if the gel is used to load cells or drugs, which requires the resulting mixture to maintain good fluidity prior to gelling for easy injection [
34]. In the present instance, the MP-embedded gels should be applicable for injection applications because their gelling time is only about 2 min shorter than that of the CH/GP gel, and the latter has been largely investigated for a variety of injection applications [
22,
25,
29].
2.3. In Vitro Release
Based on the results illustrated in
Table 3, MP-embedded gels loaded with varied amounts of PDGF-BB and BMP-2 were prepared using the protocol similar to that applied to the preparation of GL-3 gel. Two kinds of MPs respectively loaded with PDGF-BB and BMP-2 were embedded into the CH/GP gel in two different ways, as formulated in
Table 4 and
Table 5. Release patterns for three kinds of gels that were loaded with different amounts of growth factors but the same amount of MPs are shown in
Figure 4. The curves for PDGF-BB show that more than 10% of PDGF-BB was released in the first day, and at the end of two weeks, around 55% of PDGF-BB was released from these gels. In contrast to the patterns for PDGF-BB, BMP-2 was released from the gels in a considerably delayed way. Less than 10% of BMP-2 was released during the first week and the cumulative release amount of BMP-2 reached about 50% at the end of five weeks. These results reveal that such produced composite gels meet an important design requirement, namely the temporal presentation of the released PDGF-BB and BMP-2.
In principle, the release of drugs or bioactive reagents from polymeric matrices usually is correlated to swelling, diffusion, swelling followed by diffusion, and erosion [
35]. In the present situation, the factor release was mediated by both MPs and the gel matrix. As shown in
Table 4, these gels had the same gel matrix, and thus, the difference in the release behavior of PDGF-BB and BMP-2 should be critically ascribed to the effect of their corresponding MPs. PDGF-BB loaded ALG MPs had high SI, as shown in the example (see LM-1 MPs) in
Table 2, due to the presence of loose ionic linkages among ALG chains when these MPs were exposed to the aqueous media. The Ca
2+ ion crosslinked chain network inside the MPs thus has a limited ability to resist the loaded PDGF-BB molecules from being rapidly defused into the gel matrix, which would lead to high initial burst and subsequent fast release of PDGF-BB. In contrast to PDGF-BB, BMP-2 inside the core-shell MPs faces a rather different environment. As described in the experimental section, BMP-2 was first encapsulated into the CH-PLA MPs, and the resulting MPs were then coated by ALG. The CH-PLA chains in the core CH-PLA MPs have highly hydrophobic and degradation-tolerant PLA branches, and hence, they would grip BMP-2 molecules to form certain entanglements, preventing BMP-2 from getting out of the MPs. Moreover, the BMP-2 molecules that are detached from the core CH-PLA MPs have to be transported cross the ALG coating layer to reach the gel matrix. Accordingly, the double resistance stemmed from hydrophobic PLA branches inside the CH-PLA MPs and the barrier layer on the surface of the core-shell MPs would greatly hinder the diffusion of BMP-2 into the gel matrix, resulting in remarkably delayed release of BMP-2 from the composite gels.
The gels illustrated in
Table 4 were produced by embedding two types of MPs at a fixed weight ratio while being endued with various loads of PDGF-BB and BMP-2 in attempt to regulate the release rate of the factors, since an increase in the factor load may facilitate the factor to diffuse out of the MPs and possibly result in faster factor release from these gels. However, it can be observed from
Figure 4 that the initial factor load in these gels seemed to not have substantial impacts on their release profiles. A possible reason could be attributed to the factor-loaded MPs themselves. Data in
Table 2 show that LM-1 and LM-2 MPs have capacities in carrying a high amount of PDGF-BB or BMP-2, respectively. The presently employed ALG MPs and core-shell MPs for the composite gels were loaded with much lower amounts of PDGF-BB and BMP-2 when compared to LM-1 and LM-2 MPs (see
Table 2 and
Table 4), respectively, since such loads for PDGF-BB and BMP-2 are enough for their applications due to the greatly reduced initial release for PDGF-BB and greatly delayed release for BMP-2 [
19,
20,
21,
36,
37,
38,
39]. As a result, the varied loads of PDGF-BB and BMP-2 in the presently employed ALG MPs and core-shell MPs may not be high enough to significantly increase the concentration gradient of PDGF-BB or BMP-2 in their corresponding MPs, leading to insignificant changes in release rates of their matched composite gels.
Taking into account the lack of effective regulation related to above produced gels, an attempt was further made to formulate another set of composite gels in which the amount of the embedded MPs and the factor load were both varied in order to effectively modulate the release rate of the loaded factors, and the relevant results are represented in
Figure 5 and
Table 5.
The curves in
Figure 5 indicate that these composite gels showed capacities in administering the release of both PDGF-BB and BMP-2 in a temporally separated manner while effectively regulating the release rates of the factors. In the present situations, the factor load for the two types of MPs was maintained as a constant, but the embedded amount of MPs in the gels was changed. When a higher amount of MPs was dispersed in the gel matrix, the contact area between the MPs and the gel matrix will increase, which would promote more loaded PDGF-BB and BMP-2 to diffuse into the gel matrix and result in their faster release from the gels when compared with those gels containing a lower amount of MPs. It is worth pointing out that the release patterns shown in
Figure 5 rely on the initial factor load in MPs. The initial factor load for the two types of MPs was devised as around 85 (ng/mg) for PDGF-BB and 152 (ng/mg) for BMP-2 (see
Table 5) in order to make comparisons between the gel set illustrated in
Table 4 and the gel set presented in
Table 5. In fact, by changing the initial factor load within an appropriate range for two kinds of MPs or increasing the total amount of the embedded MPs up to an upper limit of about 2.0% (
w/
v), the release rate of PDGF-BB and BMP-2 can be further regulated in a wider range and in different ways. Results in
Figure 5 and
Table 5 suggest that these composite gels have potential for certain bone repair and regeneration where the sequential administration of PDGF-BB and BMP-2 is required while the applied dosages of the factors need to be well controlled. Based on the details described in the experimental section for the preparation of different types of MPs and the resulting composite gels, it can be inferred that the presently developed gel systems can actually function as a platform for sequentially delivering different signal molecules or hydrophilic drugs provided that the adopted molecules or drugs could be efficiently loaded into the employed MPs.
2.4. Bioactivity Assessment of Released Factors
One of important biological functions of PDGF-BB is its chemotactic activity [
40,
41]. Hence, the released PDGF-BB was compared with free PDGF-BB to see their ability for inducing the migration of Balb/c 3T3 cells using Boyden-type chambers [
42]. Since the release medium contained both PDGF-BB and BMP-2, the positive control employed for these measurements was prepared by mixing both free PDGF-BB and free BMP-2, and the applied amount of free PDGF-BB and free BMP-2 was precisely matched with the amount of equivalent PDGF-BB and BMP-2 in the release medium. By doing so, the effect of the released BMP-2 on the migration of Balb/c 3T3 cells can be deduced. The measured numbers of migrated cells are graphed in
Figure 6. The results show that the chemotactic response of Balb/c 3T3 cells to the released PDGF-BB was dependent on the applied equivalent PDGF-BB doses, and PDGF-BB-induced cell migration was detected even though the dose of the applied PDGF-BB was as low as 0.5 ng/mL. The number of migrated cells increased as the dose of applied PDGF-BB rose; and there were no significant differences (
p > 0.05) in the number of the migrated cells between the released PDGF-BB group and the free PDGF-BB group. The results shown in
Figure 6 signify that the bioactivity of the loaded PDGF-BB in all these composite gels is well preserved.
BMP-2 is known to be capable of reprograming some types of myoblasts through the transdifferentiation pathway [
43,
44]. In this study, a biological assay based on the BMP-2-induced ALP synthesis in C2C12 cells was used for detecting the activity of the released BMP-2, since BMP-2 has the ability to reprogram C2C12 cells to an osteogenic lineage [
37,
43,
44,
45]. Similar to the case of PDGF-BB activity testing, the positive control used here was also prepared by mixing both free BMP-2 and free PDGF-BB while controlling the applied amount of free PDGF-BB and free BMP-2 as the same as that of the equivalent PDGF-BB and BMP-2 in the release medium. The obtained data are depicted in
Figure 7. The bar-graphs show a dose-dependency trend for the BMP-2-induced ALP activity in C2C12 cells, and the released BMP-2 from all composite gels had almost equal efficiency in inducing ALP activity when compared with the free BMP-2 without significant difference (
p > 0.05); the activity of the loaded BMP-2 is effectively retained.
In the present study, CaCl2 and TPP, two types of ionic crosslinkers, were used to prepare the PDGF-BB-loaded ALG MPs and the BMP-2-loaded core-shell MPs, respectively. The preparation of these MPs was conducted using an emulsion technique under mild processing conditions, and hence, the activity of the loaded PDGF-BB or BMP-2 in MPs can be effectively preserved. On the other hand, the gel matrix contains highly biocompatible CH and GP components, and its preparation does not involved adverse factors such as covalent crosslinkers, harsh organic solvents, and unsuitable processing temperature and pH,; thus, the CH/GP matrix would not impair the bioactivity of PDGF-BB or BMP-2 molecules that diffuse from the MPs and pass through the gel matrix to reach the release medium. Based on the above-presented results, it can be concluded that the presently developed composite gels are reliable for sustaining the release of PDGF-BB or BMP-2 in a sequential manner with tunable release rates while effectively preserving the bioactivity of the loaded PDGF-BB and BMP-2.