Introducing antibacterial drugs or nanomaterials to bone scaffolds is a promising strategy to deal with implant-associated infections [1,2]. Currently, commonly used methods to prepare antibacterial polymer scaffolds are generally through directly blending or coating antibacterial materials [3,4]. Ong et al. [5] prepared curcumin and gentamicin-encapsulated poly(lactic-co-glycolic acid) (PLGA) microsphere scaffolds by oil/water emulsion and supercritical foaming, whose results indicated that approximately half of the drugs was released from the blending scaffolds in the first day. Visscher et al. [6] drop-coated cefazolin solution on poly (l-lactide) (PLLA) scaffolds and dip-coated them with gelatin methacrylate solution, whose results indicated that most of the drugs was released from the coated scaffolds in the first three days. Resulting from the direct erosion of the aqueous solution and the direct diffusion of the antibacterial drug, it is still challenging for polymer scaffolds to achieve a more effective drug sustained-release.

Utilizing inorganic nanomaterials to load drugs is attracting much attention in drug delivery [7]. Zheng et al. [8] loaded amoxicillin (AMX) on the surface of nano-hydroxyapatite (nHA) by physical adsorption and then incorporated the drug-loaded nHA into electrospun PLGA nanofibers, whose results indicated that the release of AMX from the PLGA/nHA-AMX nanofibers was slower than the control groups. Scaffaro et al. [9] loaded carvacrol on graphene via π-π stacking interaction and then introduced the drug-loaded graphene into PLLA films and electrospun nanofibers, whose results indicated that graphene slowed down the release of the drug from the films and nanofibers. Although the adsorption of drugs on the surface of inorganic carriers could sustain the drug release to some extent, it was still unable to avoid the direct erosion of aqueous solution and the direct diffusion of the drug. We are thinking if the drugs could be loaded inside the inorganic carrier, it may protect the drug from the erosion of the aqueous solution and slow the release of the drug.

Having large pore volume and specific surface area, porous inorganic nanomaterials such as halloysite nanotubes (HNTs) [10], zeolite [11], mesoporous silica [12] and mesoporous glass [13] are receiving increasing attention in many fields such as water treatment and drug delivery. Among them, HNTs have many advantages including naturally occurring nanotubular structure, good adsorption property, and superior biocompatibility [14,15]. Generally, it has an inner diameter of about 15-20 nm, an outer diameter of about 50-70 nm, and a length of about 100-1500 nm [16]. More importantly, it consists of a roll of 15-20 alumosilicate sheets. Taking these characteristics into consideration, it is possible to utilize HNTs as a carrier to load antibacterial substances into the lumen for a sustained release. In this condition, the tube walls of HNTs may act as a 'shield' which not only retard the erosion of the external aqueous solution on the internal drug but also postpone the internal drug to diffuse outward, thereby effectively slowing the release of the drug.

There have been some studies on using HNTs as a carrier for the sustained-release of some antibiotics such as cationic chlorhexidine, anionic povidone iodine, amoxicillin, tetracycline, gentamicin, etc [17,18]. Compared with these antibiotics, inorganic antibacterial nanomaterials such as Ag and ZnO have much higher stability with much less drug resistance problems, which are receiving increasing attention. Nevertheless, there were few studies on loading inorganic antibacterial nanomaterials into HNTs, especially for tissue engineering applications. Abdullayev et al. [19] synthesized nano Ag inside HNTs and then added the Ag loaded HNTs into a blue oil-based polymer paint for antibacterial function without the color change of the paint resulting from the oxidation of Ag under light exposure due to a shielding effect of HNTs on Ag. Besides, HNTs have the potential to improve several properties of polymers, including mechanical properties, biocompatibility, wettability, thermal stability, etc [20].

In this study, HNTs were used as a carrier to load nano Ag and then introduced to PLLA scaffolds for the sustained-release of Ag⁺. Nano Ag was loaded into the lumen of HNTs through vacuum negative-pressure suction & injection and thermal decomposition of silver acetate, and then PLLA-HNTs@Ag nanocomposite scaffolds were prepared by additive manufacturing. The Ag⁺ release behavior of the nanocomposite scaffolds was studied, and a sustained-release mechanism was proposed. Moreover, the antibacterial properties and cytocompatibility of the nanocomposite scaffolds were evaluated. Besides, the degradation properties, biomineralization ability, and mechanical properties were investigated.

HNTs (purity ≥ 98.0%), silver acetate (CH₃COOAg) (purity ≥ 99.0%) and PLLA (inherent viscosity ~ 2.5-3.0 dl) were purchased from Yuanxin Nano Technology Co., Ltd. (Guangzhou, China), Sinopharm Chemical Reagent Co., Ltd. (Beijing, China) and Shenzhen Polymtek Biomaterial Co., Ltd. (Shenzhen, China), respectively.

Nano Ag was loaded into the lumens of HNTs via vacuum negative-pressure suction & injection and subsequent thermal decomposition of CH₃COOAg, as schematically illustrated in Fig. 1. In a typical run, 5 g CH₃COOAg was added to a beaker containing 500 mL deionized water under magnetic stirring for 30 min. Then, 1 g HNTs were added to the CH₃COOAg solution under magnetically stirring and ultrasonically dispersed for 30 min. Next, the beaker containing the HNTs/CH₃COOAg suspension was placed in a vacuum tank and vacuumized with a vacuum pump (vacuum degree of about -1 Bar) for 30 min. During vacuumizing, large amounts of small air bubbles accompanied by sizzling sounds can be observed, indicating that the air in the lumens of HNTs was sucked out. Afterwards, the vacuum was broken to allow the CH₃COOAg solution to be injected into the lumens of HNTs by atmospheric pressure with magnetically stirring and ultrasonically dispersing for 30 min. The process of vacuum negative-pressure suction & injection was repeated three times to allow the loading of CH₃COOAg solution as much as possible. After the loading process, CH₃COOAg loaded HNTs (HNTs@CH₃COOAg) was separated by centrifuging at 10,000 rpm for 15 min, followed by washing with deionized water. In order to avoid as far as possible the synthesis of Ag on the outside surface of halloysite nanotubes, we repeated the centrifugal washing process [21] three times to remove excessive silver acetate absorbed on the outside surface of HNTs to avoid the synthesis of Ag on the outside surface as far as possible. Finally, the precipitate of HNTs@CH₃COOAg was dried at 50 °C for 48 h, followed by thermally decomposing at 380 °C for 1 h to transform CH₃COOAg to metallic Ag to obtain Ag loaded HNTs (HNTs@Ag) according to the chemical Eq. (1) [22]:

A solution mixing method was applied to prepare PLLA-HNTs@Ag nanocomposite powders: (a) a given mass of PLLA and HNTs@Ag powders at a specific mass ratio were magnetically stirred and ultrasonically dispersed in anhydrous ethanol for 30 min, forming PLLA and HNTs@Ag suspensions, respectively; (b) the HNTs@Ag suspension was added into the PLLA suspension, forming a PLLA-HNTs@Ag mixing suspension with magnetically stirring and ultrasonically dispersing for 30 min, respectively; (c) the mixing suspension was filtered, and the PLLA-HNTs@Ag precipitates were collected; (d) the precipitates were dried at 60 °C for 48 h, followed by ball-milling to obtain PLLA-HNTs@Ag fine powders for the preparation of scaffolds. A series of PLLA-HNTs@Ag powders containing 2, 4, 6, and 8% HNTs@Ag were prepared with labeling as PLLA-2%HNTs@Ag, PLLA-4%HNTs@Ag, PLLA-6%HNTs@Ag, and PLLA-8%HNTs@Ag, respectively. PLLA powder was also prepared as blank control according to the above method except for without adding HNTs@Ag.

A self-built selective laser sintering (SLS) system was used to prepare 3D porous scaffolds [23,24]. The scaffolds were built in a layer-by-layer method as an additive manufacturing technology, including (a) spreading a certain amount of powders from the reservoir piston to the building piston with a roller, forming a powder layer; (b) selectively scanning the powder layer in accordance with the section profiles of the model with a laser, where the irradiated powders were sintered and fused together, while the unirradiated powders kept loose; (c) ascending the reservoir piston while descending the building piston to a layer height, respectively; (d) subsequent powder layers were spread, sintered and fused on the previous built layers successively; (e) repeating the steps of (a-d) until all the layers of the section profiles were finished building; (f) removing the loose powders in the pores the scaffold with compressed air to obtain the final porous scaffold. The primary processing parameters of SLS were set as scanning speed 200 mm/s, laser power 2 W, layer thickness 0.1 mm, and scanning space 0.1 mm.

The phase composition of HNTs@Ag powders and PLLA-HNTs@Ag scaffolds was detected by an X-ray diffractometer (XRD) under a 2θ angle range of 5°-70° and a scan speed of 4°/min. The functional group composition of the HNTs@Ag powders and PLLA-HNTs@Ag scaffolds (ground to powders and tableted with KBr) was analyzed by Fourier transform infrared (FTIR) spectroscopy under 1200 to 400 cm^-1.

The particle morphology of PLLA-HNTs@Ag powders and the distribution morphology of HNTs@Ag in the PLLA matrix were observed by Phenom ProX scanning electron microscope (SEM) equipped with INCA energy disperse spectroscopy (EDS) after drying and spurting with gold. The morphology of HNTs@Ag was observed by a Tecnai G² F20 transmission electron microscopy (TEM). The specimens were ultrasonically dispersed in anhydrous ethanol and dropped on carbon meshes before TEM.

The Ag⁺ release behaviors of PLLA-HNTs@Ag scaffolds were studied by soaking in deionized water for different intervals for 28 days. At predetermined time intervals (1, 2, 2, 2, 7, 7, and 7 days), the scaffold specimens were removed to another fresh deionized water. Meanwhile, the aqueous solution after soaking for different time intervals was collected. The non-cumulative release concentration of Ag⁺ of different time intervals was measured with an inductively coupled plasma optical emission spectrometer (ICP-OES). The cumulative release concentration of Ag⁺ of 1, 3, 5, 7, 14, 21 and 28 days were obtained by accumulating the non-cumulative release concentration; for example, the cumulative release concentration at 7 days was the sum of the non-cumulative release concentration at the time intervals 1, 2, 2 and 2 days. Meanwhile, the average release concentration per day during different time intervals was calculated. The soaking tests were performed in triplicate.

The antibacterial properties of PLLA-HNTs@Ag scaffolds were studied with disk diffusion and anti-adhesion tests with selecting E. coli as a model bacterium (a common bacterium in orthopedic infection. The bacteria were cultured in the Lysogeny broth (LB) culture medium and diluted to approximately 1 × 10⁷ CFU/mL with McFarland standard and gradient dilution before use. The disk diffusion test was used to assess the long-lasting antibacterial properties of the scaffolds. The diluted bacterial suspension was mixed with the culture medium at v/v 1:9, resulting in a 1 × 10⁶ CFU/mL mixing suspension of bacteria/culture medium [25]. The mixing suspensions were evenly spread over a 90 mm plate, followed by equidistantly placing the disk scaffolds onto the plate and culturing in an incubator at 37 °C. After the predetermined time (3, 7, and 14 days), the culture plate was photographed, and the inhibition zones were measured. For the anti-adhesion test, the scaffolds were seeded with 1 × 10⁶ CFU/mL bacterial suspension and incubated at 37 °C. After 1 day of culture, the bacteria-scaffold constructs were collected and gently rinsed with phosphate buffer saline (PBS), followed by fixing with 2.5% glutaraldehyde and washing using PBS. Next, the specimens were dehydrated using a series of gradient ethanol and dried at 37 °C for 24 h. Finally, the adhesion morphology was observed by SEM after drying and sputtering with gold.

The cytocompatibility of PLLA-HNTs@Ag scaffolds was assessed with fluorescence staining, CCK-8 assay, and alkaline phosphatase (ALP) staining [57] using MG63 cells to evaluate the cell responses. The scaffolds were sterilized by immersing in anhydrous alcohol for 30 min and then putting in an autoclave at 121 °C for 10 min. The cells were cultured in low glucose Dulbecco’s Modified Eagle Medium (DMEM) supplied with 10% FBS and 1% antibiotic solution in an incubator at 37 °C in a humidified atmosphere of 5% CO₂. The cell suspensions were dropped onto the scaffolds in culture plate containing culture medium and cultured for predetermined time. For fluorescence staining, the cells were stained with a PBS solution containing 4 mM calcein-AM for 30 min. For CCK-8 assay, 500 μL culture medium containing 50 μL CCK-8 solution was added to each sample after predetermined culture time. After incubation at 37 °C for 3 h, the optical density (OD) was measured at 450 nm using a microplate reader. For ALP staining, the cells were incubated with Nitro-Blue-Tetrazolium (NBT)/5-bromo-4-chloro-3-indolyl-phosphate (BCIP) using a P0321 ALP color development kit according to the manufacturer’s protocol. The cells were photographed with a light microscope.

The degradable properties of PLLA/HNTs@Ag scaffolds were evaluated by PBS soaking (pH = 7.4) for 7, 14, 21 and 28 days. The scaffold specimens were dried to constant weight before the test, and the initial weight was measured. Then, they were soaked in PBS at 37 °C. After tests, the specimens and PBS were collected. The pH of PBS was measured. The specimens were dried to constant weight and weighted; the mass loss was obtained according to Eq. (2) [26]:

where M₀ is the initial weight of the scaffolds while M_t is the residual weight after t days of soaking. The tests were performed in triplicate.

The biomineralization ability of the PLLA/HNTs@Ag scaffolds was evaluated by simulated body fluid (SBF) soaking test for 28 days with observing if bone-like apatite deposited. The scaffold specimens were soaked in SBF at 37 °C with renewing SBF every other day. After the test, the specimens were collected, dried, and spurted with gold for SEM-EDS characterization.

The compressive properties of the scaffolds were measured by a universal tester at a loading rate of 1 mm/min at room temperature. The compressive strength and modulus were obtained from the stress‒strain curves. Besides, the hardness of the specimens was determined with a Vickers microhardness tester at 2.94 N and 15 s. The mechanical tests were carried out in triplicate.

The quantitative data were expressed as the mean ± standard deviation. Levene’s test was used to examine variance equality. Unpaired two-tailed Student’s t-test was performed to determine the statistical significance, where *p < 0.05 was considered to be significantly different.

XRD was performed to detect whether Ag was successfully synthesized (Fig. 2(a)). Compared with HNTs, HNTs@Ag showed three new diffraction peaks at 2θ = 38.3°, 44.4° and 64.6°, which were assigned to the (111), (200) and (220) crystalline planes of cubic Ag (JCPDS No. 04-0783) [27,28], respectively, clearly verifying the successful synthesis of Ag. FTIR was performed to detect the purity of the synthesized Ag (Fig. 2(b)). Compared with HNTs, there were no additional absorption peaks in HNTs@Ag such as CH₃COOAg or the Ag₂O, indicating a high purity of the synthesized Ag. Scanning transmission electron microscopy (STEM) and TEM were performed to verify that Ag was synthesized into the lumens of HNTs (Fig. 2(c-j)). It can be seen from the STEM image (Fig. 2(g)) that some bright dots appeared in the lumens of HNTs distributing along the axial direction. The TEM image (Fig. 2(h)) showed a similar result to that of STEM, where there existed some dark dots in the lumens of HNTs along the axial direction. High-resolution transmission electron microscope (HRTEM) was performed to identify the crystalline structure of these newly formed nanoparticles in the lumens (Fig. 2(i)). It is visible that there were some lattice fringes, indicating a monocrystalline structure. The periodicity of the lattice is approximately 0.236 nm, which was consistent with (111) d-spacing of Ag [29,30], demonstrating that the dark dots were just Ag. Selected area electron diffraction (SEAD) was performed to further identify the crystalline structure of Ag (Fig. 2(j)). Four diffraction rings appeared in the pattern, corresponding to (111), (200) and (220) reflections of cubic Ag, respectively, which was consistent with the result of XRD. Combining the results of XRD, FTIR, and TEM, it was demonstrated that nano Ag with a high purity was successfully synthesized into the lumens of HNTs. The content of Ag in HNTs@Ag was calculated by a TGA method [31]. According to the TGA result (Fig. S2 in Supplementary Data), the content of Ag in HNT@Ag was deduced to be approximately 7.5 wt%.

A solution mixing method was applied to mix PLLA and HNTs@Ag powders (Fig. 3(a)). PLLA powders showed an irregular shape with particle size about 20-100 μm, while HNTs@Ag powders showed a rod-like appearance with a diameter and length of about 50 nm and 1 μm, respectively. As for the PLLA-HNTs@Ag powders, it can be seen that many bright dots of HNTs@Ag uniformly distributed on the PLLA particles, indicating a well-mixed condition. A schematic diagram illustrating the preparation of the PLLA-HNTs@Ag scaffolds by SLS and a representative cylindrical scaffold are shown in Fig. 3(b). The scaffold showed a periodic porous structure with a pore size of about 600 μm. The porous structure was suitable for supporting the transport of oxygen, nutrients, and wastes, and the ingrowth of bone cells and tissues [32]. XRD was carried out to analyze the phase composition of PLLA-HNTs@Ag scaffolds, as shown in Fig. 3(c). PLLA showed a broad diffraction peak at 16.6° [33,34], corresponding to the (110) or (200) crystalline planes. Compared with PLLA scaffolds, PLLA-HNTs@Ag scaffolds showed additional diffraction peaks at 2θ = 38.3°, 44.4°, and 64.6°, which were all assigned to the reflections of Ag. It seems that the diffraction peaks of HNTs in the PLLA-HNTs@Ag scaffolds disappeared, which may be ascribed to a fact that the diffraction peaks of metallic Ag were so strong that its covered those of HNTs. FTIR was performed to demonstrate the presence of HNTs in the scaffolds (Fig. 3(d)). Compared with PLLA scaffolds, PLLA-HNTs@Ag scaffolds showed additional absorption peaks at about 467 and 534 cm^-1, which were corresponding Al-O-Si and Si-O-Si deformations [35], respectively, clearly demonstrating the presence of HNTs in the scaffolds. Combining these results, the presence of HNTs@Ag in the scaffolds was demonstrated.

The release behaviors of Ag⁺ of PLLA-HNTs@Ag scaffolds were studied by soaking test and regularly detecting the release concentration of Ag⁺ with ICP. As clearly seen from Fig. 4(a), the cumulative release concentration of Ag⁺ increased with soaking time. More importantly, the release of Ag⁺ continued to the end of the tests, indicating a sustained-release behavior over 28 days. Besides, it is obvious that the release concentration of Ag⁺ also increased with increasing HNTs@Ag content, suggesting that the release amount of Ag⁺ can be adjusted by simply changing the content of HNTs@Ag. The average release concentration of Ag⁺ during different time intervals decreased with prolonging soaking time (Fig. 4(b)), which was ascribed to the decrease in the total content of Ag during the test.

The release kinetics analysis was carried out by fitting the cumulative curves of Ag⁺ release from the scaffolds into Ritger-Peppas equation, which is a commonly used release kinetic model for biodegradable drug delivery systems [36,37]. The Ritger-Peppas equation is given as Q_t/Q_∞=ktⁿ, where t is the release time (h), Q_t is the fraction of drug released at time t, k is the release rate constant, and Q_t/Q_∞ is the fraction of drug release at time t. The release kinetic parameters of Ag⁺ from PLLA/HNTs@Ag scaffolds by fitting with the Ritger-Peppas model using Origin 8.0 software are given in Table 1. According to the results, all of the correlation coefficients R² were close to 1, indicating a good fitting effect. In the Ritger-Peppas model, the drug release mechanisms are characterized using the release exponent (n value), where n ≤ 0.43 was considered to be Fickian release, while 0.43 ≤ n ≤ 0.85 was considered to be non-Fickian or anomalous release [37]. As the n values for the scaffolds were lower than 0.43, the Ag⁺ release from the scaffolds was considered to be controlled by Fickian diffusion.

A possible mechanism of sustained-release of Ag⁺ from the scaffolds was proposed and schematically depicted in Fig. 4(c). On the one hand, the tube walls of HNTs acted as 'shields', hindered the diffusion of the aqueous solution into the lumens and its erosion on Ag. On the other hand, the 'shields' inhibited the release of Ag⁺ from the lumens to the matrix. That is to say, the aqueous solution and Ag⁺ cannot diffuse through the tube walls but can only possible through the ends of the tubes. As a result, the generation and release of Ag⁺ was restrained with undergoing a five-stage process, including: (1) diffusion of H₂O & O₂ from the aqueous solution to the matrix; (2) diffusion of H₂O & O₂ from the matrix to the HNTs via the tube ends and reaction with Ag to generate Ag⁺; (3) diffusion of Ag⁺ from the ends of HNTs to the matrix; (4) diffusion of Ag⁺ across the scaffold matrix; and (5) diffusion of Ag⁺ from the scaffold matrix to the aqueous solution.

The antibacterial properties of PLLA-HNTs@Ag scaffolds were studied by disk diffusion and anti-adhesion tests (Fig. 5). It can be seen that inhibition zones formed around all the PLLA/HNTs@Ag scaffolds while not for PLLA scaffolds (Fig. 5(a)). Meanwhile, after 14 days of incubation, a large bacterial colony agglomerate formed around the PLLA scaffolds while not for the other scaffolds. Besides, the sizes of the inhibition zones increased with increasing HNTs@Ag content at the same culture time (Fig. 5(d)). More importantly, the inhibition zones remained surrounding the PLLA-HNTs@Ag scaffolds during the whole test of 14 days (Fig. 5(b, c)). The results indicated that the scaffolds had a long-lasting antibacterial property, which was owing to the sustained-release of Ag⁺ from the scaffolds. The adhesion morphology of E. coil after culture with the scaffolds for 1 day is shown in Fig. 5(e-i). There were large amounts of E. coil adhering on the PLLA scaffolds with normal rod-like morphology (Fig. 5(e)). In contrast, it was found that there were almost no E. coil adhering on PLLA-HNTs@Ag scaffolds (Fig. 5(f-i)), indicating that the scaffolds inhibited the adhesion of E. coil owing to their high antibacterial property. The generation and release of Ag⁺ was the key antibacterial mechanism of nano Ag. It was reported that Ag⁺ could react with the thiol groups in the respiratory and transport proteins in the cytomembrane of bacteria, blocking cellular respiration and electron transfer and disrupting membrane potential and permeability [[38], [39], [40]]. Moreover, Ag⁺ ions can enter into the bacterial cells due to the change of the cytomembrane permeability or through some ion channels [38,41]. After the entrance, they react with the nucleobases in DNA and RNA, resulting in their condensation and thereby the loss of replication and transcription ability [38,42]. Hence, the adhesion and proliferation of the bacteria were inhibited.

The viability, proliferation, and osteogenic differentiation of MG63 cells were tested using fluorescence staining, CCK-8 assay, and ALP staining to evaluate the cytocompatibility of the scaffolds (Fig. 6). From fluorescence staining, the cell number increased on all the scaffolds with prolonging the culture time from 1 to 3 days (Fig. 6(a-j)), indicating that the cell proliferation was supported. The results of the CCK-8 assay further quantitatively demonstrates the cell proliferation on the scaffolds, as shown in Fig. 6(k). The optical density is proportional to the cell number. Obviously, the optical density increased for all the scaffolds with culture time increasing from 1 to 5 days, suggesting that all the scaffolds supported the cell proliferation. Moreover, it seems that the cell proliferation level on the PLLA/HNTs@Ag scaffolds is higher than that on the PLLA scaffolds. From the results of ALP staining (Fig. 6(l-o)), the cells cultured with PLLA-HNTs@Ag scaffolds were more positively stained than those cultured with PLLA scaffolds, suggesting that HNTs@Ag has positive effects on the osteogenic differentiation. The material composition and surface structure of scaffolds play significant roles in stimulating cell responses [43,44]. A study indicated Si was in support of osteoblast-like cells’ proliferation and can positively stimulate MG63 cells’ biological responses through promoting bone-specific proteins’ production [45]. Another study indicated that Si is involved in regulating the proliferation and osteogenic differentiation of BMSCs with the involvement of some signaling pathways [46]. Therefore, it may be inferred that the Si component in HNTs positively stimulated the cell responses of MG63 cells.

The degradation properties of the PLLA-HNTs@Ag scaffolds were evaluated by soaking test in PBS with measuring the mass loss and pH after different days (Fig. 7(a, b)). The mass loss increased with prolonging the soaking time, indicating a good degradability (Fig. 7(a)). More importantly, the PLLA-HNTs@Ag scaffolds showed higher mass loss than PLLA scaffolds, indicating that HNTs@Ag increased the degradability of the scaffolds. After soaking for 28 days, the mass losses were 2.6, 3.3, 3.9, 4.6, and 5.0% for the scaffolds with 0, 2, 4, 6, and 8% HNTs@Ag, respectively. The pH decreased with prolonging soaking time (Fig. 7(b)), due to the production of acid products from PLLA degradation during PBS soaking. Moreover, the decrease in the pH of PBS for PLLA-HNTs@Ag scaffolds was larger than that for PLLA scaffolds, indicating a faster degradation, being consistent with the results of mass loss. The biomineralization ability of the PLLA-HNTs@Ag scaffolds was evaluated by SBF soaking and characterize the surface morphology by SEM-EDS after 28 days (Fig. 7(c-h)). It can be seen that flower-like clusters deposited on all of the scaffolds with HNTs@Ag, while the PLLA scaffolds remained a smooth surface. According to the EDS spectrum, the clusters consisted mainly of O, P, and Ca elements, which indicated that they were bone-like apatite. Additional signals of the elements C, Ag, Al, Si were assigned to the PLLA-HNTs@Ag scaffolds below the clusters due to the penetration of X-ray photons (PLLA consisted of C and O elements, and HNTs-Ag consisted of O, Al, Si, Ag). It was noted that the amounts of the bone-like apatite increased with increasing the content of HNTs@Ag, indicating an increased biomineralization. The nucleation and growth of apatite on biomaterials were associated with the surface characteristics such as surface composition and surface charge. As a ceramic nanoparticle, HNTs exposed and distributed the surface of the nanocomposite scaffolds can act as nucleation sites for apatite nucleation and growth. Considering that the surface of HNTs was negatively charged, whose zeta potential was about -50 mV [47], they may electrostatically attract positively charged Ca²⁺ in SBF. Meanwhile, Ca²⁺ can attract negatively charged HPO₄^- in SBF. When Ca²⁺ and HPO₄^- achieved a local supersaturation, they began to form apatite nuclei, which then spontaneously grew by continuing to attract Ca²⁺ and HPO₄^- in SBF [[48], [49], [50]]. As a result, the clusters of bone-like apatite formed and covered on the scaffolds.

The distribution of HNTs@Ag in the PLLA matrix was observed by SEM (Fig. 8(a-f)). Compared with PLLA scaffolds (Fig. 8(a)), there were some bright dots in the matrix of PLLA-HNTs@Ag scaffolds (Fig. 8(b-e)), which were just the incorporated HNTs@Ag according to the EDS results where the elements of Ag, Si, Al, and O were all detected (Fig. 8(f)). With the content increasing from 2% to 6% (Fig. 8(b-d)), more and more HNTs@Ag particles were observed, which showed a relatively uniform distribution in the matrix. However, when the content continued to increase up to 8%, some obvious HNTs@Ag aggregates appeared in the matrix (Fig. 8(e)). The mechanical properties of PLLA-HNTs@Ag scaffolds vs. HNTs@Ag content were plotted in Fig. 8(g-h). With the content from 0-6%, the compressive properties (compressive strength and modulus) increased. Nevertheless, they started to decrease when the content continued to increase up to 8%. The changing trend of the hardness as a function of the content of HNTs@Ag was similar to that of the compressive properties. The optimal compressive strength, compressive modulus and hardness were 33.42 MPa, 2.98 GPa, and 85.88 MPa, respectively, which were increased by 78.0, 67.4 and 52.1% compared with those of PLLA scaffolds, respectively. It was well known that the content and distribution state of nanoparticles in polymer composites has significant effects on the final mechanical properties [51,52]. At a uniform distribution, nanoparticles, acting as rigid filler, can effectively absorb and transfer stresses in the matrix under external force loading [53,54]. Therefore, the composites can be effectively reinforced, and the reinforcing effects increased with increasing the nanoparticle content. Nevertheless, once the content of nanoparticles was excessive, they tended to form severe aggregates, resulting in stress concentration, which instead decreased the reinforcing effects with increasing the content [55,56]. In this work, a relatively uniform distribution of HNTs@Ag in the PLLA matrix can be obtained when the content was between 0-6%, thereby keeping increase of the mechanical properties (compressive properties and hardness) with increasing the content. Nevertheless, when the content of HNT@Ag increased up to 8%, obvious aggregates formed, which instead weakened the reinforcing effects.

In this study, HNTs were used as a carrier to load nano Ag into the lumen through vacuum negative-pressure suction & injection and thermal decomposition of silver acetate. Then, HNTs@Ag was introduced to PLLA scaffolds for the sustained-release of Ag⁺. The results indicated that the PLLA-HNTs@Ag scaffolds showed a sustained-release of Ag⁺ over 28 days without obvious initial burst release, which was owing to the shielding effects of the tube walls of HNTs on the erosion of the external aqueous solution and the out-diffusion of the internal Ag⁺. Meanwhile, the PLLA-HNTs@Ag scaffolds showed a desired long-lasting antibacterial property. Besides, the scaffolds supported cellular proliferation and osteogenic differentiation. In addition, the degradation properties, biomineralization ability, and mechanical properties of the scaffolds were increased.

The authors declare no competing interest.

This work was financially supported by the National Natural Science Foundation of China (Nos. 51935014, 51905553, 81871494, 81871498, 51705540); the Hunan Provincial Natural Science Foundation of China (Nos. 2019JJ50774, 2018JJ3671 and 2019JJ50588); the JiangXi Provincial Natural Science Foundation of China (No. 20192ACB20005); the Guangdong Province Higher Vocational Colleges & Schools Pearl River Scholar Funded Scheme (2018); the Open Sharing Fund for the Large-scale Instruments and Equipments of Central South University; the Project of Hunan Provincial Science and Technology Plan (No. 2017RS3008); the Hunan Provincial Innovation Foundation For Postgraduate (No. CX2018B093) and the Fundamental Research Funds for the Central Universities of Central South University (Nos. 2018zzts022 and 2019zzts725).

Supplementary material related to this article can be found, in the online version, at doihttps://doi.org/10.1016/j.jmst.2019.11.019.