Materials and animals Gelatin, methacrylic anhydride (MA), and 2-hydroxy-2-methylpropylenone (HMPP) were obtained from Shanghai Aladin Co., Ltd., China. Bletilla polysaccharide (BSP) was purchased from Bai chuan Bio-technology Co., Ltd, China. Carbomer 940 was acquired from Beijing Solaibao Technology Co., Ltd, China. rRtEGF was bought from Shanghai PrimeGene Bio-tech Co., Ltd, China. Yunnan Baiyao (BY) was purchased from Yunnan Baiyao Group. Gelatin methacrylate (GelMA) was self-synthesized in our laboratory. Deionized water (DIW, 18.25 MΩ·cm− 1, Millipore) was used through all experiments. Male Sprague-Dawley rats weighted 125–150 g were provided by Vital River Laboratory Animal Technology Co., Ltd. Any animal experimental protocols were approved by the Animal Investigation Ethics Committee of Wenzhou Institute of University of Chinese Academy of Sciences (No. WIUCAS22092203).

Synthesis of GelMA GelMA was synthesized in our laboratory as we previously published [32]. Typically, the gelatin (10%, w/v) and Na2CO3 (5%, w/v) were dissolved in deionized water (DIW) and stirred for 2 h. 5 mL of methacrylic anhydride was dropwise added into 200 mL of gelatin solution within 30 min and stirred for another 2 h. During the reaction process, the pH value of the mixed solution was adjusted by NaOH solution (1 M) for maintaining 8–9. The resultant solution was dialyzed against DIW before lyophilization to obtain GelMA. The degree of substitution was examined from 1 H NMR spectra (QUANTUM-I-400 MHz, Q.One Instruments, China).

Fabrication of (BY + EGF) @MNs GelMA (20%, g/mL), EGF (0.002%, g/mL) and HMPP (1%, g/mL) were dissolved in the DIW to obtain the EGF-GelMA pre-gels, which 100 µL EGF-GelMA pre-gels were used to fill the MN tip cavity in the polydimethylsiloxane (PDMS) mold using a vacuum pump. Excess solution was removed (≥ 50 µL) and then cured under UV light (365 nm, 10 W, 10 s). BSP (10% w/v), Carbomer 940 (0.6%, g/mL) and BY (0.8%, g/mL) were dissolved in the DIW to obtain the BY-BSP pre-gels, which about 100 µL BY-BSP pre-gels were used to fill MN base cavity in the PDMS mold and subsequently frozen at –80 ℃. The (BY + EGF)@MNs were finally obtained after lyophilization.

Characterization The bright-field and fluorescence micrographs of (BY + EGF)@MNs were observed using a stereomicroscope (Olympus BX51, Tokyo, Japan). The microscopic structures of (BY + EGF)@MNs were recorded by a scanning electron microscope (SU8010, Hitachi, Japan). The compression and tensile tests of (BY + EGF)@MNs were performed on a universal mechanical testing machine (5944, Instron, USA).

Mechanical characterizations The puncture strength of MN tips was measured by a compression test. The MN patches with different GelMA concentrations were laid on a platform, where the MN tips pointed to the descending pressure sensor. The sensor gradually moved to the tips at 2 mm/min, and the force began to be recorded as 0.01 N when touching the MN tips. The force-displacement curves were recorded, and the force at 500 nm for different MN tips were calculated. In addition, the (EGF + BY)@MNs were inserted into the rat skin and liver tissues, which were stained with H&E to evaluate the puncture intensity.

Tissue adhesion tests The adhesion strength between the skin tissues and MN bases was verified by a tensile test. The pig skin tissues were firstly fixed to the horizontal stage. MN bases with gradient Carbomer contents were contacted with the skin tissues under a pre-pressure of 5 N for 1 min. The upside pressure sensor attached to the MN bases was stretched upwards at a speed of 20 mm/min. The stretching process was terminated manually when the MN bases were separated from the skin tissues. The maximum tensile force could be obtained from the force-displacement curves as an index to evaluate the adhesion ability of the MN bases.

BSP dissolution and BY release The release kinetics of BSP and BY were assessed using a UV absorption spectroscope (obtained by Agilen CARY5000, USA). BSP and BY had the highest absorbance at 280 nm and 260 nm. The BSP patches with different Carbomer concentrations were placed in 5 mL of PBS (PH 5.6). Subsequently, 0.5 mL of PBS leaching solution was taken out every 1 min, and then an equal amount of identical PBS was added immediately. By measuring the absorbance at 280 nm, the BSP dissolution rate could be calculated based on the standard curve of BSP solution. Similarly, the release kinetics curves of BY were obtained by treating BY-containing BSP film (BY@FN) in PBS solution under different pH conditions (i.e., pH = 5.6, 6.5, and 7.4) with the same way.

GelMA Degradation and drug release The degradation rate of GelMA MN tips was evaluated by recording their weight loss. We prepared cube-shaped GelMA hydrogel samples (size: 16 mm× 16 mm× 1 mm) to avoid interference of BSP-based MN bases. All samples were immersed in 1 ml of PBS with collagenase II (2.6 U/mL), kept shaking at 37 °C. The PBS solution was replaced at regular intervals after centrifugation, and then 1 mL of fresh PBS was supplemented. The MNs were dried completely at 60 °C and their dry weights were recorded at designed time pints. The degradation curve was plotted based on the weight loss. For drug release test, BSA was used as a drug model and be incorporated into the same cube-shaped GelMA hydrogel to avoid interference of BSP-based MN bases. With the same manner as above, the absorbance of BSA released in the leaching solution were detected using UV spectroscopy at each time point. The BSA release profiles were obtained under different pH conditions.

RBC and platelet adhesion assays Gauze (clinical cotton gauze), BSP@FN (BSP flat patches, fabricated by 100 µL pre-gels with 10% BSP and 0.6% carbomer (w/v) after freeze-drying), BY@FN (BY-loading BSP flat patches, fabricated by 100 µL BY-BSP pre-gels (g/mL) containing 0.8% BY, 10% BSP and 0.6% carbomer), and (EGF + BY)@MNs (MN patches MN patches, fabricated by 100 µL BY-BSP pre-gels (g/mL) containing 0.8% BY, 10% BSP and 0.6% carbomer and 50 µL EGF-GelMA pre-gels (g/mL) containing 0.002% EGF and 20% GelMA ) were used as experimental groups. The adhesion experiments were carried out as published [33]. RBC suspension and platelet-rich plasma (PRP) were separated from the citrated whole blood (CWB) by centrifugation (1500 rpm, 10 min).

For RBC adhesion assay, the RBCs suspensions (100 µL) were paved on the MN patch surface. After 1 min, they were washed to remove non-adherent RBCs by using PBS solution, and then soaked in DIW (4 mL) to release hemoglobin. After 2 h, the supernatant (100 µL) was pipetted, and the OD value was measured at 540 nm (OD experimental). The OD value of DIW (4 mL) with RBCs suspension (100 µL) and PBS (100 µL) was measured as positive value (OD positive) and negative value (OD negative), respectively. The rate of adhered RBCs was calculated by the following Eq. (1):

$$_}}}(\% ) = \frac_}}}-_}}}}}}_}}}-_}}}}}} \times 100\%$$

(1)

For platelet adhesion assay,100 µL of PRP was paved onto the sample surfaces for 1 min. Non-adherent platelets were removed with PBS, and then the samples were soaked in 2 mL of Triton X-100 solution to free the lactate dehydrogenase (LDH). The LDH contents in the supernatants were determined using a LDH kit (Solarbio, China). The OD450 nm value of the Triton X-100 solution containing 100 µL of PBS and PRP was used as negative value (OD negative) and positive value (OD positive), respectively. The ratio of adhered platelets was calculated by the following Eq. (2):

$$_}}}(\% ) = \frac_}}}-_}}}}}}_}}}-_}}}}}} \times 100\%$$

(2)

In addition, the adhered RBCs and platelets were observed by SEM. The SEM samples were prepared by being incubated with 100 µL RBC or PRP for 1 min, washed with PBS, fixed in 3% glutaraldehyde for 12 h, dehydrated in gradient ethanol solutions, dried with a supercritical dryer, and then sprayed with gold by a high-vacuum ion sputtering machine (Leica, EM ACE600).

Coagulation assessments The blood clotting test of (EGF + BY)@MNs was performed according to previous studies [34]. Normal (without treatment), BY (BY powers, 1 mg), BY@FN (BY-loading BSP flat patches, fabricated by 100 µL of BY-BSP pre-gels containing 0.8% BY, 10% BSP and 0.6% carbomer), and (EGF + BY)@MNs (MN patches with a BY-loading BSP base and EGF-loading GelMA tips, fabricated by 100 µL of BY-BSP pre-gels and 50 µL of EGF-GelMA pre-gels (g/mL) containing 0.002% EGF and 20% GelMA) were ground and mixed with the 1 mL of serum from CWB. The clotting times of different groups were recorded at 37 ℃ by the PT/APTT assay reagent kits (Beijing ZONCI Technology Development Co., Ltd, China).

Platelet activation assessments P-selectin (CD 62p) immunofluorescence staining was performed to evaluate the activation effect of (EGF + BY)@MNs. The samples preparation in different groups were as same as RBC and platelet adhesion assays, which were fabricated with co-incubated with PRP for 10 min, horizontally placed on glass slides, and then fixed in glutaraldehyde for 12 h. The staining was performed sequentially by CD 62p monoclonal antibody, and then examined using a fluorescence microscope.

Hemocompatibility assessments Hemocompatibility of (EGF + BY)@MNs was assessed by observation and quantification of hemoglobin released from the RBCs incubated with different samples. RBC suspensions were acquired from CWB by centrifugation. 100 µL of RBCs suspension was mixed with 900 µL of leaching solution for different samples, which was centrifuged after incubation for 1 h. This hemolysis process was photographed using a digital camera. The OD values in different experimental groups were recorded as OD experimental at 550 nm. 100 µL RBCs suspension was dropped into DIW and used as positive control (OD positive), while PBS was used as negative control (OD negative). The hemolysis rate was obtained by following Eq. (3):

$$_}}}(\% ) = \frac_}}}-_}}}}}}_}}}-_}}}}}} \times 100\%$$

(3)

Cytocompatibility assessments 0 and 1 mg/ml of leaching medium of (BY + EGF)@MNs were used as control and experimental group, respectively. NH3T3 cells were added in a 96-well plate (800/well) and cultivated by different solutions. On days 1, 2, and 3, the OD value was detected at 450 nm after incubating with CCK-8 solution (10% v/v, Beyotime Biotechnology) for 4 h. The NH3T3 cells were stained with a live/dead staining kit for morphology observation.

Hemostasis and liver wound healing in vivo In all animal experiments, SD rats were divided into 7 groups: Ctrl (without treatment), BY (treated with BY powders, 1 mg), BY@FN (treated with BY-loading BSP flat patches, fabricated by 100 µL of BY-BSP pre-gels containing 0.8% BY, 10% BSP and 0.6% carbomer), EGF (treated with 10 µL of 0.01% g/mL EGF solution), EGF@FN (treated with EGF-loading GelMA flat patches, fabricated by 50 µL of EGF-GelMA pre-gels containing 0.002% EGF and 20% GelMA), (EGF + BY)@FN (treated with EGF-loading GelMA and BY-loading BSP flat patches, fabricated by 50 µL of EGF-GelMA pre-gels and 100 µL of BY-BSP pre-gels), (EGF + BY)@MNs (treated with MN patches with EGF-loading GelMA tips and a BY-loading BSP base, fabricated by 50 µL of EGF-GelMA pre-gels and 100 µL of BY-BSP pre-gels). The drug or patch in each group were administered only once on day 0. Additionally, before traumatic procedures, all animals were anesthetized by injecting atropine sulfate (0.01 ml/100 g, 0.4 mg/mL, Shanghai full woo Biotechnology Co., Ltd., China) into muscular and Zoletil®50 (0.15 ml/100 g, 50 mg/mL, Virbac Co., Ltd., France.) into abdominal cavity.

The in vivo hemostatic and liver regenerative capacities of the (EGF + BY)@MNs were evaluated using a liver puncture hemorrhage model in SD rats. The rat abdomen was opened after anesthetization. The livers were laid onto a pre-weighed sterile filter paper. Next, a “cross” wound with a length of 5 mm and a depth about 3 mm was made for hepatic hemorrhage. The liver wounds were covered with different samples, and the hemostasis process was recorded. The hemostasis time was measured by a timer. The blood loss was determined by weighting the filter papers. The live tissues in different groups were harvested after 28 days. H&E staining (Solarbio, Beijing), PAS staining (Beyotime, Shanghai) and IL-6 immunohistochemical staining (Servicebio, Wuhan) were conducted for histological analysis.

In vivo tissue repair evaluation A full-thickness cutaneous wound mode was established in rats to demonstrate the tissue repair effect of (EGF + BY)@MNs. The cutaneous tissues on the back of anesthetized rats were excised to form a circle with a radius of 5 mm and treated with different ways in various groups. The wounds were divided into 7 groups: Ctrl (without treatment), BY (treated with BY powders, 1 mg), BY@FN (treated with BY-loading BSP flat patches, fabricated by 100 µL of BY-BSP pre-gels (g/mL) containing 0.8% BY, 10% BSP and 0.6% carbomer), EGF (treated with 10 µL of 0.01% EGF solution), EGF@FN (treated with EGF-loading GelMA flat patches, fabricated by 50 µL of EGF-GelMA pre-gels containing 0.002% EGF and 20% GelMA), (EGF + BY)@FN (treated with EGF-loading GelMA and BY-loading BSP flat patches, fabricated by 50 µL of EGF-GelMA pre-gels and 100 µL of BY-BSP pre-gels), (EGF + BY)@MNs ( treated with MN patches with EGF-loading GelMA tips and a BY-loading BSP base, fabricated by 50 µL of EGF-GelMA pre-gels and 100 µL of BY-BSP pre-gels). Subsequently, all wounds were covered with medical breathable tapes to avoid unwanted scratches. For tracking the wound healing process, the wounds were photographed on days 0, 2, 6, and 10. The wound dressings were not renewed at the designed intervals, which could be beneficial for monitoring the in vivo degradation of (EGF + BY)@MNs. Regenerated skin tissues were collected from the sacrificed rats on day 10. Microanatomy analysis was conducted by typical H&E/Masson staining, and immunofluorescence staining of CD31, VEGF, α-SMA and VIM.

Statistical analysis All quantitative analysis were conducted as means with standard deviations (n ≥ 3). Differences of two groups and multiple groups were analyzed by unpaired Student’s t-tests and one-way analysis-of-variance (ANOVA), respectively. Significant differences were considered when *p < 0.05, **p < 0.01, and ***p < 0.001, and ‘ns’ indicated no significant difference.