Fe₃O₄ magnetic nanoparticles were prepared by a coprecipitation method (Xu et al., 2004; Xu et al., 2007). The obtained nanoparticles were collected through a magnetic field and washed repeatedly with deionized water. The Fe₃O₄ magnetic nanoparticles were observed on a transmission electron microscope (T20, FEI, United States).

Fe₃O₄@Agar microspheres were produced by an inverse suspension method. Typically, 0.24 g Fe₃O₄ (wet weight), 0.25 g agarose powder, and 5.76 g deionized water were added in a 25-ml flask. The mixture was sonicated (JP-010T, SKYMEN, China) for 5 min and heated to 95°C, then 0.5 ml of 2 M NaOH was added. The resultant suspension was mechanically stirred at 250 rpm for 20 min to dissolve the agarose. The emulsifier span-80 of 0.5 g was added in 25 ml n-octane. The n-octane solution was mechanically stirred at 1,000 rpm and meanwhile heated to 65°C, and then the Fe₃O₄ and the agarose mixed aqueous suspension was transferred into the n-octane solution. After stirring at 1,000 rpm and 65°C for 0.5 h, the suspension was cooled to 55°C, and 0.5 g of the cross-linking agent epichlorohydrin was added. The cross-linking reaction proceeded at a stir speed of 300 rpm for 1 h. After the reaction, the suspension was stationary at room temperature. The cross-linked magnetic agarose microspheres in the lower layer were collected and washed with 20 v% ethanol–water solution several times, filtered through a sieving sieve, and washed with deionized water successively to obtain purified Fe₃O₄@Agar microspheres with diameters of 80–250 µm. The morphology and size of the purified and undried microspheres were observed on a microscope (BX51, Olympus, Japan).

The purified and undried Fe₃O₄@Agar microspheres were added in deionized water with the concentration equivalent to 50 mg/ml dried microspheres, and then hyaluronic acid powder was added with the concentration of 15 mg/ml. The mixture was mechanically stirred overnight to obtain a uniform Fe₃O₄@Agar/HA magnetic hydrogel. The magnetic hydrogel was sterilized by autoclaving and dispensed into a 1-ml or 2-ml syringe before use.

The Fe₃O₄@Agar/HA samples with the final Fe concentrations of 0, 0.0008, 0.0016, 0.0032, 0.0064, 0.0128, 0.0256, and 0.0512 mM were, respectively, prepared by adding 0, 0.0075, 0.015, 0.03, 0.06, 0.12, 0.24, and 0.48 g of the Fe₃O₄ nanoparticles in the Fe₃O₄@Agar microspheres, while the agarose and hyaluronic acid concentrations were unchanged. The transverse relaxation time (T₂)-weighted magnetic resonance signal intensities of the Fe₃O₄@Agar/HA samples were acquired on an MRI instrument (BioSpec 70/20 USR, Bruker, Germany). The parameters were repeating time 3,000 ms, echo time 105.4 ms, and field of view 4.00 cm. The transverse relaxation rate of Fe₃O₄@Agar/HA was obtained by a linear fitting of 1/T₂ versus Fe concentration.

Cytotoxicity was evaluated using the MTT method (AQSIQ and SAC, 2017a). The cell line was mouse fibroblast L929 from the Cell Bank of Chinese Academy of Science (Shanghai, China). The Fe₃O₄@Agar/HA extract solution was prepared by immersing 0.2 g Fe₃O₄@Agar/HA in a 1 ml MEM medium containing 10% fetal bovine serum (Fisher Scientific International Inc., Logan, UT, United States) at 37°C and 5% CO₂ for 24 h, and the solution was immediately used after filtration. The positive control was a polyurethane film containing 0.1% zinc diethyl dithiocarbamate, and the negative control was a high-density polyethylene film (AQSIQ and SAC, 2017a). Their extract solutions were prepared as the solution of Fe₃O₄@Agar/HA. The cells were incubated with one of the extract solutions for 24 h to evaluate the cytotoxicity. Six parallel experiments were performed for each condition.

Salmonella typhimurium histidine-deficient strains TA98, TA100, TA102, TA1535, and TA1537 were obtained from MOLTOX^® Molecular Toxicology, Inc. (Boone, NC, United States). The extract solution was prepared by immersing 0.2 g Fe₃O₄@Agar/HA in 1 ml saline and incubating the solution at 50°C for 72 h with shaking. The extract solution was used without filtration within 6 h. The metabolic activation, rat liver homogenate (S9), was from CHI Scientific (Boston, MA, United States). The genotoxicity test was performed using the bacterial reverse mutation test (Ames test) by a plate incorporation method (Hamada et al., 1994; Zeiger, 2019). Three parallel experiments were performed for each condition.

The animal experiments of this study were reviewed and approved by the Ethics Committee of Children’s Hospital, Fudan University [ethical approval number: (2018) 119]. All the animals were housed in full compliance with the Chinese National Standard GB14925-2010 (AQSIQ and SAC, 2010) and acclimated to the environment for more than 5 days before the experiment. The animals had free access to standard diet and water.

Three healthy New Zealand young adult male rabbits from the Songlian Laboratory Animal Farm (Shanghai, China) were used for the single exposure test. The fur of the both sides of the spine (approximately 10 cm × 15 cm) was clipped at 24 h before the test. For each rabbit, the two right sites were covered with 0.5 ml Fe₃O₄@Agar/HA, and the two left sides were blank control. Each of the application sites was covered with 2.5 cm × 2.5 cm sterile gauze and wrapped with a bandage for 4 h. After that the bulking agent was removed by washing with saline, and the sites were carefully wiped dry. The appearance of each application site was recorded at 1, 24, 48, and 72 h after the removal of Fe₃O₄@Agar/HA. The primary irritation index of Fe₃O₄@Agar/HA was calculated as described in GB/T 16886.10-2017/ISO 10993-10:2010 (AQSIQ and SAC, 2017b).

Healthy Harley guinea pigs in their early adulthood (260–430 g) were obtained from the Laboratory Animal Center of Southern Medical University (Guangzhou, Guangdong, China). The Fe₃O₄@Agar/HA extract solution was prepared as described in the in vitro genotoxicity section, and used without filtration within 24 h. Freund’s complete adjuvant was from STC (Dongguan) Company Limited (Dongguan, Guangdong, China). Five samples were used in the guinea pig maximization test (AQSIQ and SAC, 2017b). Sample A was the Freund’s complete emulsion prepared by mixing the adjuvant with saline at a 1:1 volume ratio. Sample B was the extract solution without dilution. Sample C was the mixed solution of sample A and sample B with a 1:1 volume ratio. Sample D was saline. Sample E was the mixed solution of sample A and saline with a 1:1 volume ratio. Ten guinea pigs were treated with the samples A, B, and C (two sites for each sample and six sites for each guinea pig), and five guinea pigs were treated with samples A, D, and E as the control. The test procedure including an intradermal induction phase, a topical induction phase, and a challenge phase was the same as described in GB/T 16886.10-2017/ISO 10993-10:2010 (AQSIQ and SAC, 2017b). At 24 and 48 h after the removal of the dressings, the erythema and edema of the challenge skin sites were observed and graded according to the Magnusson and Kligman grading (AQSIQ and SAC, 2017b).

Healthy young adult KM mice (nulliparous female) were purchased from the Laboratory Animal Center of Southern Medical University. The mice were divided into two groups with five in each. The Fe₃O₄@Agar/HA extract solution was prepared as described in the in vitro genotoxicity section, and used without filtration within 24 h. In the extract solution group, the extract solution was injected via the caudal vein at a single dosage of 50 ml/kg. In the control group, the same volume saline was injected. Immediately after the injection and at 4, 24, 48, and 72 h postinjection, death or systemic toxicity was observed (AQSIQ and SAC, 2011b). The mice were weighed before as well as 1, 2, and 3 days after the injection. After that, the mice were sacrificed, the gross pathological changes were observed, and the organs such as the heart, liver, spleen, lung, and brain were taken out for pathological analysis by paraffin section and microscopic examination methods.

Healthy New Zealand male rabbits (8–10 weeks, 2.2–2.6 kg) were acclimated to the environment for more than 1 week. The rabbit was anesthetized with an intramuscular injection of 25 mg/kg ketamine (Jiangsu Hengrui Medicine Co., Ltd., Lianyungang, Jiangsu, China), and then the roof of the left intravesical ureter was incised to create VUR (Baek et al., 2010; Mangera and Edhem, 2012). Four weeks after the surgery, voiding cystourethrogram (VCUG) was acquired on an X-Ray System (AXIOM Iconos R200, SIEMENS, Germany) to confirm the reflux (Sencan et al., 2008).

A total of 14 VUR rabbits (Grade II-III) were randomly divided into three groups to receive the following operations: 1) six rabbits in the bulking agent injection group were injected with Fe₃O₄@Agar/HA; 2) four rabbits in the saline control group were injected with saline; and 3) four rabbits in the sham-operation control group were anesthetized, and their bladders were opened but no injections were performed. After being anesthetized with 25 mg/kg ketamine by intramuscular injection, the bladder of the VUR rabbit was opened through the original surgical incision and the trigone was exposed. Then, 0.2–0.3 ml of the bulking agent or saline was immediately injected through a 0.45-mm ID needle and 1 ml syringe into the submucosal plane beneath the left ureteral orifice at the 6 o’clock position (O’donnell and Puri, 1984). A dosage of 50 mg/kg/day ceftriaxone (Shanghai Roche Pharmaceuticals Ltd., Shanghai, China) was given intramuscularly at 30 min before the operation as well as on the first and second days after the operation as a prophylactic antibiotic therapy.

After 4 weeks of the operation, the following investigations were performed in succession for the rabbits: 1) VCUG examination to evaluate the therapeutic effect; 2) T₂-weighted MRI (Siemens Prisma 3.0T, Germany) to observe the injected bulking agent; the rabbits being anaesthetized by intraperitoneal injection with 10% chloral hydrate at a dosage of 3.5 ml/kg and examined at the conditions of field strength 3.00 T, slice thickness 2 mm, repetition time 6,000 ms, and echo time 108.0 ms; 3) pathological analysis of the paraffin sections of the organs to evaluate the biocompatibility and migration of the bulking agent; three samples being randomly acquired from each organ of the liver, spleen, gallbladder, pancreas, heart, brain, bladder, and bilateral kidneys, ureteropelvic junctions, ureterovesical junctions, and middle ureters.

Table 1 shows the cell viabilities after 24 h incubation in the media containing different volume fractions of the Fe₃O₄@Agar/HA extract solution. The cell viability decreased with the increase of the volume fraction of the extract solution. According to GB/T 16886.5-2017/ISO 10993-5:2009 (AQSIQ and SAC, 2017a), the material has a cytotoxic potential when the cell viability is lower than 70% of the blank. In our study, the cell viability was 80% in 100% extract solution, no cell lysis was observed, and the cells were in a good growth state, indicating that Fe₃O₄@Agar/HA has good cytocompatibility.

The Ames test was performed to detect the point mutations (Hamada et al., 1994; Zeiger, 2019) caused by the Fe₃O₄@Agar/HA extract solution. For TA98, TA100, and TA102 strains, the data in Table 2 show that the revertant colony number of the extract solution group was less than two-fold of the number of the negative control group. For TA1535 and TA1537 strains, the revertant colony number of the extract solution group was less than three-fold of the number of the negative control group. These results mean that the Fe₃O₄@Agar/HA extract solution did not cause mutations in the stains tested (Hamada et al., 1994), indicating that Fe₃O₄@Agar/HA has no genotoxicity potential.

Three young adult rabbits were used for a single exposure test to assess the dermal irritation potential (AQSIQ and SAC, 2017b). Table 3 shows the erythema grade plus edema grade of each application site after the removal of Fe₃O₄@Agar/HA. Only one rabbit had transient minimal erythema, and no other adverse changes were observed. The primary irritation index of the Fe₃O₄@Agar/HA group was 0.17, calculated by dividing the sum of all the scores by 18 (three rabbits, two test sites, and three time points observed at 24, 48, and 72 h). According to GB/T 16886.10-2017/ISO 10993-10:2010 (AQSIQ and SAC, 2017b), less than 0.4 of the primary irritation index indicates that the dermal irritation potential of Fe₃O₄@Agar/HA is negligible.

Young adult albino guinea pigs were used to assess the skin sensitization potential of the Fe₃O₄@Agar/HA extract solution through the guinea pig maximization test (AQSIQ and SAC, 2017b). The clinical manifestations of the guinea pigs treated with the test samples and the control samples were all normal. At 24 and 48 h after the removal of the dressings, no visible changes were observed at the challenge skin sites, suggesting that Fe₃O₄@Agar/HA has no skin sensitization potential.

Young adult KM mice were injected with the Fe₃O₄@Agar/HA extract solution at a single dosage of 50 ml/kg via the caudal vein to assess the acute systemic toxicity. During the observation period, no mouse died, and all the mice were clinically normal. The body weight changes of the extract solution group were similar to the changes of the saline group, and the loss of the body weight was less than 5% of the corresponding weight before the injection (Table 4). After 72 h of the observations, the mice were euthanasia and dissected. No significant systemic toxicity and gross pathology changes were observed (Supplementary Figure S2). Except that some paraffin sections of the organs presented slight lymphocytic infiltration (≤20%) for both test and control groups, no significant histopathology changes were observed, as shown in Table 5. These results indicate that Fe₃O₄@Agar/HA has no significant acute systemic toxicity.