*3.1. Analysis of ALG–POL*

ALG–POL copolymer was synthesized by grafting MAPT onto the ALG backbone through EDC/NHS chemistry. Figure 1 presents a schematic illustration to show the synthesis route for MAPT and ALG–POL copolymer. A representative NMR spectrum for MAPT is provided in Figure S1 in which the typical peaks respectively assigning to methyl and methylene confirm the successful synthesis of MAPT. The preliminary experimental results indicated that the composition of ALG–POL exerted significant e ffects on the thermoresponsivity, degradation tolerance, and strength of resulting ALG–POL/SF gels. Accordingly, the proportion of ALG and POL components in ALG–POL copolymer was optimized via orthogonal design method. By changing the ratio of MATP to ALG in a range between 20 and 35 at a step size of 5 during the synthesis of ALG–POL copolymers, one of ALG–POL copolymers was thus screened out by setting the MATP/ALG ratio at 30. The POL content in such optimized ALG–POL copolymer was thus measured to be around 66 wt% via elemental analysis. Figure S2 presents FTIR spectra for POL, ALG and ALG–POL. The spectrum of POL is characterized by three typical bands at 2891 (C–H stretch aliphatic), 1345 (in-plane O-H bend) and 1112 cm<sup>−</sup><sup>1</sup> (C–O stretching). The ALG spectrum shows specific absorbance bands of its COOH stretching at 1610 cm<sup>−</sup><sup>1</sup> and C–O–C stretching at 1305 cm<sup>−</sup><sup>1</sup> [33], respectively. In the spectrum for ALG–POL, the carbonyl absorption band for carboxylate sodium salt originally showing in the ALG spectrum disappeared while a new characteristic amide I band appeared at around 1637 cm<sup>−</sup>1, suggesting that amide bonds have formed between ALG and POL [9,33]. FTIR spectra demonstrate that the ALG–POL copolymer has been successfully synthesized.

**Figure 1.** Schematic illustration for synthesis of alginate-poloxamer copolymer.

## *3.2. Parameters for Mesoporous BG Nanoparticles*

Two kinds of mesoporous BG NPs were produced under slightly di fferent processing conditions in order to attain certain BG NPs having proper pore-sizes and pore volumes for gaining high LE. Panels A and B in Figure 2 display two typical TEM micrographs for the prepared BG NPs, and these

spherical BG NPs were seen to be porous with their size of about 200 nm. A mass of BG NPs were measured for their average particle size and ζ-potential, and relevant data are listed in Table 1. There were significant di fferences (*p* < 0.05) in the average size and ζ-potential of these BG NPs, signifying that the processing conditions significantly modulated their structure and property even though they were formulated with the same chemical composition [28,29]. Figure 2C displays the recorded N2 adsorption–desorption isotherms for MBG-1 and MBG-2 NPs. Two isotherms exhibited their respective hysteresis loops of the desorption branch, indicative of the existence of large pores inside BG NPs [28,29]. In comparison to MBG-1 NPs, the hysteresis loop for MBG-2 NPs was shown to be steeper and its inception turning point was closer to the high pressure end of the N2 isotherm, suggesting that MBG-2 NPs have more pores with larger size than MBG-1 NPs. Besides these di fferences, MBG-2 NPs also quite di ffered from MBG-1 NPs in pore volume and pore size distribution (Figure 2D). Two sets of BG NPs were measured for their major parameters and the obtained data are summarized in Table 1. It can be seen that the presently developed BG NPs had a high level of pore volume, large average pore size and negative ζ-potential, and thus, they could act as a practical vehicle for the IGF-1 delivery since IGF-1 is somewhat positively charged (isoelectric point, 8.6) at physiological pH and has its molecular weight of about 7.6 kD [34,35].

**Figure 2.** Images of MBG-1 ( **A**) and MBG-2 (**B**) nanoparticles (NPs); N2 adsorption isotherms ( **C**) and pore size distribution ( **D**) of NPs.

## *3.3. IGF-1 Release from BG Nanoparticles*

Blank MBG-1 and MBG-2 NPs were loaded with IGF-1 under the condition of varying IGF-1 feed amounts, and two sets of IGF-1-loaded BG NPs were thus produced (Table 2). BS*i* (*i* = 1, 2 and 3) sample set was prepared by loading IGF-1 into MBG-1 NPs having smaller average pore size than that for MBG-2 NPs (see Table 1) whereas BL*j* (*j* = 1, 2 and 3) sample set was prepared by loading IGF-1 into MBG-2 NPs. Data in Table 2 reveal that these IGF-1-loaded NPs had similar average particle size (*p* > 0.05) but significantly higher ζ-potential (*p* < 0.001) as compared to their respective blank counterparts (see Table 1). The nearly unchanged average size for IGF-1-loaded NPs shown in Table 2 can be attributed to the very small IGF-1 mass when compared to NPs themselves, whereas the significantly increasing ζ-potential should be ascribed to the slightly positively charged nature of IGF-1 [34]. Table 2 indicates that the IGF-1-loaded NPs in BL*j* (*j* = 1, 2 and 3) set had significantly higher (*p* < 0.05) LE as compared to the counterpart in BS*i* (*i* = 1, 2 and 3) set. These differences are rational because the blank MBG-2 NPs used for preparing BL*j* (*j* = 1, 2 and 3) set have notably higher pore volume and larger pore size when compared to blank MBG-1 NPs used in BS*i* (*i* = 1, 2 and 3) set (see Table 1). Table 2 also shows that the IGF-1 feed amount exerted certain effects on LE, and this kind of effect would become insignificant once the IGF-1 feed amount reached 100 ng/mL or higher. It is worth mentioning that IGF-1 feed amounts were designated as such in order to test the LE for NPs. Actually, the IGF-1 load in these NPs can be effectively regulated by changing the feed amount of IGF-1.

The IGF-1 release patterns from IGF-1-loaded BG NPs are shown in Figure 3A,B. BS1, BS2, and BS3 NPs released around 50% of their initial IGF-1 load on the first day, and after 4-day release, the cumulative IGF-1 release amounted to about 70%. The IGF-1 load in these NPs did not impose any significant impacts on their release profiles. With respect to the cases associated with BL1, BL2, and BL3 NPs, their release profiles looked quite similar to that assigned for BS1, BS2, and BS3 NPs, respectively, with somewhat faster release rates (Figure 3B). Figure 3 verifies that these mesoporous BG NPs themselves are not able to effectively control the release kinetics of IGF-1 on account of their initial burst release features. LE is a key issue that is correlated to the rational use of IGF-1 because of high cost of IGF-1. In consideration of the similar release profiles illustrated in Figure 3 for both MBG-1 and MBG-2 NPs but significantly higher LE detected from MBG-2 NPs (see Table 2), MBG-2 NPs were thus chosen for the follow-up gel preparation.

**Figure 3.** IGF-1 release profiles from 5 mg of BS1, BS2, and BS3 NPs ((**A**), initial IGF-1 load in BS1, BS2 and BS3 was 1.93, 4.61, and 7.27 ng/mg) and BL1, BL2, and BL3 NPs ((**B**), initial IGF-1 load in BL1, BL2, and BL3 was 2.29, 5.72, and 9.21 ng/mg) in 500 μL PBS (see Table 2 for their parameters).

## *3.4. Rheological Properties of BG*/*ALG–POL*/*SF Gels without IGF-1 Load*

ALG–POL is a thermoresponsive copolymer and the thermal transition of ALG–POL solutions had strong concentration dependence. A previous study reported that ALG–POL was gelable when its solution concentration reached about 15 wt% or higher [9]. In the present study, the optimally synthesized ALG–POL copolymer was found to have clear sol-gel transition during a rational gelation period when its solution concentration reached 12 wt% or higher. Several optical images are presented in Figure S3 for showing changes of ALG–POL fluids after incubation. It can be noticed that the ALG–POL solution with its concentration of 9 wt% was remained as a fluid even though it was incubated at 37 ◦C for 60 min, and on the other hand, a 12 wt% ALG–POL solution was able to

°

turn into gel via incubation at the same temperature during 14 min. Results in Figure S3 demonstrate that ALG–POL alone could be thermally gelable at 37 ◦C when its solution concentration is higher than a certain threshold.

In view of independent gelable mechanisms respectively belonging to ALG–POL and SF components, dual network gels with mechanically strong nature could be constructed by using ALG–POL and SF together. To achieve a ALG–POL/SF gel with required properties, a series of ALG–POL/SF composite solutions having their weight proportions of 4/8, 5/7, 6/6, and 7/5 was formulated for the preparation of blank ALG–POL/SF gels, and the optimal gel was sought out as 5 wt% for ALG–POL and 7 wt% for SF. Based on such designed composition for the ALG–POL/SF gel, blank MBG-2 NPs were embedded into the gel to fabricate three kinds of BG/ALG–POL/SF gels without IGF-1 load and the resulting gels were utilized to evaluate the rheological properties in order to save IGF-1. Major parameters for these blank BG/ALG–POL/SF gels are given in Table 3.

As seen from Table 3, G-A, G-B, and G-C gels had the same matrix and the difference in their composition was the percentage of blank MBG-2 NPs. Panels A, B, and C in Figure 4 elucidate the representative temperature sweep curves of G and G" for different gels, and these gels were seen to respond to the thermal stimulus at different inception temperatures (Ti). G-A gel had its Ti at around 36 ◦C, whereas G-B and G-C gels showed their Ti near 35 ◦C, connoting that the introduction of BG NPs into the ALG–POL/SF gel has a very limited effect on their gelation temperature. It can be observed from Table 3 that the pH value and Ti of these gels were quite close to the physiological pH and temperature, and meanwhile, their gelation time was seen to be rational [5,6], suggesting their applicability under physiological conditions. Figure 4D presents the shear dependence of viscosity for different gels. Their viscosity was shown to be lower than 70 pa.s at 23 ◦C, and progressively decreased with rising shear rate, indicating their shear-thinning features. Given that the gel injection is usually performed at room temperature, curves graphed in Figure 4D validate that the presently formulated BG/ALG–POL/SF gels have well-defined injectability.

**Figure 4.** Temperature sweep curves (**A**–**C**) and shear viscosity ((**D**), 23 ◦C) for BG/ALG–POL/SF gels.

In principle, magnitude of G and G" of hydrogels in the linear viscoelastic region (LVR) of their frequency sweep spectra together with G/G" ratio can be used to assess the gel strength [36]. In general, a strong hydrogel is characterized by high G, and meanwhile, its G should be 1–2 orders of magnitude greater than its G" [36]. Figure 5A,B show that at a fixed frequency in their respective LVR, for example, 1.0 HZ, three kinds of gels had their G of around 5 kPa or higher and their G" greater than 300 Pa. The incorporation of BG NPs into the ALG–POL/SF gel seemed not to exert marked effects on G and G" of the resulting gels. To make quantitative comparisons, G and G" at 1.0 Hz for these gels were measured, and obtained average values are graphed in Figure 5C. The bar-graphs explicate that these gels had their G higher 5.5 kPa with the G/G" ratio greater than 15, verifying their mechanically strong features.

**Figure 5.** Frequency dependent functions (**A**,**B**) of modulus and average values (**C**) of modulus at 1.0 Hz and 37 ◦C for BG/ALG–POL/SF gels (\*\*, *p* < 0.001; N.S., no significance).

## *3.5. Morphological Analysis of Dry Gels*

The presently developed BG/ALG–POL/SF gels need to be porous because they are intended for use in bone repair where they will function as injectable materials for housing cells. G-A and G-C gels in Table 3 were selected and their lyophilized samples were examined to see their internal structures. A few SEM images for the dry gels are represented in Figure 6. These images show that dry gels were highly porous and their pore size changed from several tens of microns to more than two hundred microns with pore-interconnected characteristics. With respect to G-C gel, the incorporation of BG NPs did not significantly affect its pore structure when compared with G-A gel. The image with a larger magnification (Figure 6D) displays that the wall of pores in the gel was stuck or attached with many size-varied granules and these granules should be assigned to BG NPs or their aggregates. The average pore size for these dry gels is shown in Figure 6E. The dry gels had large average pore size without significant difference, which is advantageous for bone repair where large pore size and high porosity in the requisite gels are concurrently required.

SI of dry gels has been used as an approximate estimation for their porosity since the channels shaped inside the gels can regulate their swelling and deswelling behavior via water convection [37]. In general, dry gels with open-cell pores and high porosity have high SI and short swelling equilibrium time due to fast water convection [38]. The bar-graph in Figure 6F illustrates that these dry gels had their SI higher than 5, and meanwhile, the composition of the gels seemed not to exert significant impacts on their SI. In addition, it was found that these dry gels reached their respective swelling equilibrium in PBS less than 30 min. The similar SI together with their rapidly swollen features connotes that these gels have similar porosity.

**Figure 6.** SEM micrographs ((**A**,**B**), G-A gel; (**C**,**D**), G-C gel), average pore-size (**E**) and swelling index (**F**) of BG/ALG–POL/SF dry gels (see Table 3 for parameters of gels; N.S., no significance).

## *3.6. IGF-1 Release of Gels*

The gels were loaded with varied amounts IGF-1 in a designated way as illustrated in Table 4 and they were detected to access their capacity for administration of IGF-1 release. Release profiles for these gels are presented in Figure 7. Curves in Figure 7A exhibit that two kinds of gels directly loading with IGF-1 had fast IGF-1 release in the first few days at varied rates somewhat depending on their initial IGF-1 load, and their IGF-1 release became visibly slower after one-week release with similar release rate in the light of approximately constant distance between the two curves. In marked contrast to this observation, two gels embedded with IGF-1-loaded BG NPs behaved in quite different ways (Figure 7B). IGF-1 load of around 7% or less was released from the gels in the first day, and afterwards, the release patterns followed approximately linear behavior for a few weeks at various release rates. In comparison to the patterns shown in Figure 7A, the significantly reduced initial IGF-1 release and the subsequent release slowdown in Figure 7B can be attributed to the joint contribution of the gel matrix and BG NPs. As denoted in Table 4, GEL-3 and GEL-4 gels were prepared by embedding IGF-1-loaded BG NPs into ALG–POL/SF. In comparison to GEL-1 and GEL-2 gels, IGF-1 molecules in GEL-3 and GEL-4 gels will encounter increasing resistance derived from both BG NPs and gel matrix, which will certainly result in their release slowdown. When the release patterns in Figure 7B are

compared each other, it can be observed that IGF-1 content in the gels remarkably affected the release rate. A possible reason could be ascribed to that the higher IGF-1 loading in a gel would form a larger IGF-1 concentration gradient inside the gel, which would force IGF-1 molecules to transport through the gel matrix faster and to reach the media earlier, leading to higher cumulative IGF-1 amount.

**Figure 7.** IGF-1 release patterns for gels directly loaded with IGF-1 (**A**) and for IGF-loaded BG/ALG–POL/SF gels (**B**).

## *3.7. Bioactivity Assessment of IGF-1*

Bioactivity preservation of released IGF-1 is an important issue because it is closely correlated to the biological effects of IGF-1. In this study, MC3T3-E1 cells were used for assessment of IGF-1 bioactivity because IGF-1 is capable of promoting the proliferation of osteoid cells in dose-dependent manners [30–32]. To evaluate these gels on the same baseline, GEL-2 and GEL-4 gels were selected taking account of their similar and higher initial IGF-1 load. MC3T3-E1 cells were cultured with equivalent amount of released IGF-1 or free IGF-1 for varied durations up to 72 h and relevant results are elucidated in Figure 8. At a low level of IGF-1 (5 ng/mL), OD values matching with IGF-1-applied cell groups were remarkably higher than that of control group but no significant differences were detected among cell groups that were exposed to free IGF-1 or released IGF-1 as sampling time advanced (Figure 8A). By increasing the applied IGF-1 amount by 10 times (Figure 8B), the variation trend of OD values and their differences looked similar to that detected at the IGF-1 dosage of 5 ng/mL. In addition, by comparing each OD value in Figure 8B with the corresponding one in Figure 8A, IGF-1-dose dependent characteristics can be detected when the culture time reached 72h. These results support that the released IGF-1 was able to promote the proliferation of MC3T3-E1 cells in the way of dose-regulation, confirming that bioactivity of released IGF-1 can be well preserved.

**Figure 8.** Response of MC3T3-E1 cells to 5 ng/mL (**A**) or 50 ng/mL (**B**) of free or released IGF-1 during various culture periods (\*, *p* < 0.05; \*\*, *p* < 0.001; N.S., no significance).

Besides these tests, MC3T3-E1 cells were also cultured on the surface of GEL-2 and GEL-4 gels for three days to further evaluate their growth, and the measured OD values are depicted in Figure 9. In these cases, the cumulated IGF-1 amount in the culture media would be dynamically altered because GEL-2 and GEL-4 gels would incessantly release IGF-1 at different rates despite their similar initial IGF-1 load. As shown in Figure 7, the cumulative amount of IGF-1 released from GEL-2 gel on the first, second and third days reached around 27, 39, and 50%, respectively; and the corresponding cumulative IGF-1 release from GEL-4 gel was about 7, 10, and 13%. Considering the patterns shown in Figure 7, it can be envisioned that in the current situation, the amount of available IGF-1 in GEL-2 group was significantly higher than that in GEL-4 group in the first three days. It can be seen that OD values measured from two gel groups were markedly higher than that detected from control group during the three sampling days, demonstrating that the released IGF-1 is bioactive and able to promote the growth of MC3T3-E1 cells (Figure 9). When GEL-2 and GEL-4 groups were compared each other, it shows that there was no significant difference in their OD value on the first day, but on the second and third days, OD value measured from GEL-2 group was notably larger than that detected from GEL-4 group. The possible reasons for these observations could be attributed to that (1) cells need a certain period of time to attach to the gels and to undergo recovery growth with low responsiveness to the released IGF-1 on the first day, resulting in insignificant difference in their OD value; and (2) after fully attaching and returning to their normal growth, cells seeded on GEL-2 gel would grow faster than those on GEL-4 gel because GEL-2 gel can release a notable higher IGF-1 amount than GEL-4 gel. These results further confirm that presently devised gels have ability to promote the proliferation of MC3T3-E1 cells.

**Figure 9.** Optical density (OD) values of MC3T3-E1 cells cultured on the surface of gels during various periods (\*, *p* < 0.05; \*\*, *p* < 0.001; N.S., no significance).
