All bacterial strains used in this study are listed in Supplementary Table 1. The multidrug-resistant A. baumannii strain MDR-AB2, isolated from the sputum samples of a patient with pneumonia at PLA Hospital 307 was supplied by the Beijing Institute of Microbiology and Epidemiology (Peng et al., 2014). The antibiotic susceptibility profile of this strain was determined, which confirmed its resistance to most of the commonly used antibiotics (Peng et al., 2014). All the bacterial strains were grown in Nutrient Broth (NB) medium at 37°C. Colistin was bought from J&K Scientific (Beijing, China) (Cat No. 437689, >19,000 IU/mg).

The plasmid was constructed using standard cloning methods. Genes encoded Dpo71 (Protein id: YP_009592222.1) was synthesized by BGI (Shenzhen, China) and cloned into the pET28a plasmid using BamHI and XhoI site and the protein sequence was listed in supporting information (Supplementary Table 2).

The constructed plasmid was transformed into Escherichia coli BL21 (DE3) cells and colonies were grown overnight at 37°C in LB media supplemented with 50 μg/ml kanamycin. The start culture was grown overnight, and then was used to inoculate LB media supplemented with antibiotics at 1:100 ratio. The cell culture was grown at 37°C to reach OD₆₀₀ ∼0.6 before 0.25 mM IPTG was added to induce protein expression. After grown at 16°C overnight, cells were harvested for protein purification. The enzyme was purified by nickel affinity chromatography using HisTrap™ HP column (GE Healthcare, Chicago, United States). Briefly, harvested cells were re-suspended in lysis buffer containing 10 mM imidazole, 50 mM phosphate/300 mM sodium chloride (pH 8.0). The cell suspension was lysed by sonication and centrifuged. The supernatant was collected, filtered, and loaded into the column. The bound protein was eluted by imidazole gradient from 10 to 500 mM. Pure protein fractions eluted with imidazole gradient were collected and exchanged with PBS (pH 7.4). After purification, all proteins were flash frozen under liquid nitrogen and stored at −80°C. Protein concentration was determined using NanoDrop (Thermo Fisher Scientific, Massachusetts, United States).

Purified depolymerase Dpo71 as well as two gel filtration markers with molecular weight of 150 and 200 kDa (MWGF200, Sigma-Aldrich, Missouri, United States) were used for analysis, performed on an AKTA FPLC instrument installed with a Superose 6 Increase prepacked column (GE Healthcare, Illinois, United States). Proteins were filtered with 0.2 μm membrane and injected into the FPLC system equilibrated with PBS.

Far-UV CD spectroscopy is commonly used to analyze the secondary structures of proteins with a Jasco J810 CD spectrometer (Oliveira et al., 2020). The spectrum measurement was performed with the Dpo71 of 0.30 mg/ml in PBS buffer (pH 7.4) using a wavelength range from 190 to 260 nm. Thermal denaturation with 1°C/min increments was also employed to measure the secondary structure unfolding at 215 nm, from 20 to 90°C. The melting curves were fitted into a Boltzmann sigmoidal function.

The depolymerase activity of Dpo71 was qualitatively assayed by a modified single-spot assay. In brief, 100 μl of MDR-AB2 overnight bacterial culture was added to 5 ml of molten soft nutrient agar (0.7%) and incubated at 37°C for 3 h to form a bacterial lawn in plates. The purified enzyme was serially diluted, then 5 μl of each dilution (from 0.001 to 10 μg) was dropped onto the MDR-AB2 bacterial lawn for incubation at 37°C overnight. The plates were monitored for the formation of semi-clear spots as a confirmation of the depolymerase activity.

The bacterial polysaccharide extracts (containing both CPS and LPS) were purified, via a modified hot water-phenol method as described previously (Liu et al., 2020). Briefly, A. baumannii were cultured overnight in LB with 0.25% glucose. A volume of 1 ml culture was centrifuged (10,000 rpm, 5 min) and resuspended in 200 μl of double distilled water (ddH₂O). An equal volume of water-saturated phenol (pH 6.6; Thermo Fisher Scientific) was added to the bacterial suspension. The mixture was vortexed and incubated at 65°C for 20 min, centrifuged at 10,000 rpm for 10 min. Then the supernatant was extracted with chloroform to remove bacterial debris. The obtained bacterial CPS/LPS were lyophilized and stored at –20°C before use.

The enzymatic activity of Dpo71 degrading the bacterial polysaccharides was determined as described in Liu et al. (2020) with minor modifications. The extracted CPS/LPS powder of A. baumannii was resuspended in ddH2O (1 mg/ml) and mixed with Dpo71 (30 μg/ml) or deactivated Dpo71 (by heating at 90°C for 15 min) to a final reaction volume of 200 μl. The extracted CPS/LPS or enzyme alone served as the controls. After 2 h incubation at 37°C, cetylpyridinium chloride (CPC, Sigma-Aldrich, Missouri, United States) was added to the mixture at the final concentration of 5 mg/ml, which was further incubated at room temperature for 5 min. Absorbance was measured at 600 nm using a microplate reader (Multiskan Sky, Thermo Fisher Scientific, Massachusetts, United States). The experiment was performed in triplicate and repeated at least in two independent experiments.

The extracted CPS/LPS powder was dissolved in 100 mM citric acid-Na₂HPO₄ buffer (pH 3.0–8.0) or 100 mM Glycine-NaOH buffer (pH 9.0–10.0) to a final concentration of 1 mg/ml. The CPS/LPS solutions of A. baumannii were mixed with Dpo71 (30 μg/ml) to a final reaction volume of 200 μl, respectively. After 2 h incubation at 37°C, the turbidity of residual CPS/LPS in various pH buffers was determined as described above. The effect of pH on the enzymatic activity was determined by this method. The storage stability of Dpo71 at 4°C was determined by measuring the CPS/LPS degradation activity after 1, 3, and 6 months storage. All assays were performed in triplicate and repeated at least in two independent experiments.

The serum killing assay was performed as previously described (Oliveira et al., 2020). Log phase bacteria were prepared by inoculating overnight culture at 1:100 ratio in NB medium and shaking 180 rpm for 3–4 h at 37°C. Then the cells of each culture (AB#1, AB#2, AB#3, and AB#4) were harvested via centrifugation, washed, and resuspended in PBS, then adjusted to OD₆₀₀ = 0.6. Human serum (Sigma-Aldrich, Missouri, United States) was mixed with (i) only bacteria or (ii) bacteria and Dpo71 mixture. The Dpo71 was fixed at final concentration of 10 μg/ml and the volume ratio of human serum was set at 50%. Experiments with heat-inactivated serum (at 56°C for 30 min) were served as controls. The mixtures were then incubated at 37°C for 4 h for viable bacterial counting. Time-killing assays were also performed for the two sensitive strains (AB#1, AB#2) with the volume ratio of human serum varied from 1 to 50% and at 100 μg/ml Dpo71. Samples were withdrawn at 1, 3, 5, and 24 h for bacterial counting. All assays were performed in triplicate and repeated at least in two independent experiments.

Scanning Electron Microscopy (SEM) was conducted as described previously (Oliveira et al., 2019a). Log phase A. baumannii bacteria were washed twice with PBS and resuspended in PBS buffer at OD₆₀₀ = 0.6. Then approximately 10⁸ cells were incubated at 37°C with PBS or Dpo71 for 3 h. Cells were then fixed with 2.5% (v/v) glutaraldehyde at 4°C overnight. Thereafter, the fixed cells were washed twice with PBS and dehydrated with a graded ethanol series (15, 30, 50, 70, 85, and 100% for twice). The bacterial suspensions were then spotted on glass and dried with vacuum. Finally, the samples were coated with gold and observed with Quanta 400F SEM (FEI, Oregon, United States).

For the investigation of outer membrane permeability, 1-N-phenylnaphthylamine (NPN) uptake assay was performed (Helander and Mattila-Sandholm, 2000; Chen et al., 2021). A. baumannii cells were grown to mid-log phase (OD₆₀₀ = 0.6–0.8), centrifuged, and resuspended in PBS. Then PBS (control), Dpo71 (10 μg/ml), colistin (1 μg/ml) or their combination (Dpo71 + colistin) were incubated with 10⁷ CFU/ml cells in the presence of 10 μM NPN for 5 min. The fluorescence intensities were recorded using a microplate reader (CLARIOstar, BMG Labtech, Ortenberg, Germany) with 350 ± 7.5 nm for excitation and 420 ± 10 nm for emission. All experiments were performed in three biological replicates and repeated at least in two independent times.

The assay described by Campos et al. (2004) was done with minor modifications. Briefly, cells grown as described above were resuspended in PBS buffer about 10¹⁰ CFU/ml. Then, the bacteria were first treated with Dpo71 or PBS at 37°C for 2 h. Then 50 μg/ml colistin was added, after 5 min of incubation, the cells with the bound colistin were sedimented (12,000 g, 10 min), the supernatant was centrifuged two more times under the same conditions. Finally, the unbound colistin was quantified with HPLC. The integrated areas were recorded and the concentrations of the unbound colistin were calculated from the standard curve of colistin (16–500 μg/ml). All experiments were performed in three biological replicates and repeated at least in two independent times.

A. baumannii were grown in NB medium overnight at 37°C with continuous shaking 180 rpm. The overnight bacterial culture was diluted with fresh NB medium to a final density of 10⁶ CFU/ml. Then the diluted bacteria cultures were treated with PBS (control), Dpo71 (1, 10, or 40 μg/ml), colistin (1 μg/ml) or their combination (Dpo71 + colistin) with a final volume of 150 μl/well at 37°C for 24 h with gentle shaking (100 rpm). At the end of the incubation time, all media were removed and the wells were stained with 200 μl 0.1% (w/v) crystal violet for 1 h. After staining, the crystal violet solution was removed and the wells were washed with 200 μl PBS for three times. Then, 200 μl of 70% ethanol was added to dissolve the crystal violet and 100 μl solution was transferred to a new plate for quantification of the residual biofilm biomass using a microplate reader (CLARIOstar, BMG Labtech, Ortenberg, Germany) at 570 nm. All experiments were performed in three biological replicates and repeated at least in two independent times.

A. baumannii were grown in NB medium overnight at 37°C with continuous shaking 100 rpm. The overnight bacterial culture was diluted with fresh NB medium to a final density of OD₆₀₀ = 0.2. To initiate the biofilm growth, diluted culture was aliquoted into a 96-well plate at 100 μl/well (Costar, Corning Incorporated, New York, United States) and incubated at 37°C for 24 h at 100 rpm. Biofilm was washed twice with PBS and treated with PBS (control), Dpo71 (10 μg/ml), colistin (4 μg/ml) or their combination (Dpo71 + colistin) with a final volume of 150 μl/well at 37°C for 24 h with gentle shaking (100 rpm). At the end of the incubation time, the residual biomass was quantified using crystal violet as described in the inhibition assay above. The antibiofilm activity was also evaluated by the counting the viable bacteria in the biofilm (Wilson et al., 2017). Briefly, the biofilm was grown and treated as above, then the wells were washed three times with PBS. Then the biofilm-containing wells were mixed fully with a pipetting device, making the biofilm cells become planktonic cells. Each sample was serially diluted and plated for bacterial counting. All experiments were performed in triplicate and repeated at least in two independent times.

Confocal dish (NEST, Wuxi NEST Biotechnology, Jiangsu, China) was used instead of the 96-well plate for the antibiofilm study described above. Before microimaging, the biofilms were stained by LIVE/DEAD™ BacLight™ Bacterial Viability Kit (Thermo Fisher Scientific, Massachusetts, United States) for 60 min following the manufacturer’s instructions. Then the biofilms were washed three times with PBS before the confocal laser scanning microscopy (Nikon, C2+ confocal, Tokyo, Japan) study. The excitation maximum and emission maximum of SYTO 9 is at 483 and 503 nm, respectively. The excitation maximum and emission maximum of propidium iodide is at 493 and 636 nm, respectively.

The effect of Dpo71 on the hemolysis of human red blood cells was performed using previously described methods with minor modifications (Liu et al., 2019a). The human blood sample from a healthy donor was centrifuged at 1,000 rpm for 10 min to remove the serum. The red blood cell pellets were washed with PBS (pH = 7.4) at least three times and then diluted to a concentration of 5% (volume ratio) with PBS. The Dpo71 (10, 100, and 500 μg/ml, final concentration) was added to the red blood cells and incubated at 37°C for 1 h, followed by centrifugation at 1,000 rpm for 10 min. Then 100 μl supernatant was transferred to a 96-well microplate and topped up with another 100 μl of PBS to get a final volume of 200 μl. The erythrocytes in PBS and 0.1% Triton X-100 were served as negative and positive controls, respectively. The hemoglobin in supernatant was determined by measuring absorbance at 540 nm using a microplate reader (Multiskan Sky, Thermo Fisher Scientific, Massachusetts, United States). All experiments were performed in three biological replicates and repeated at least in two independent times.

BEAS-2B (Human Normal Lung Epithelial Cells) cells were cultured in DMEM (GIBCO, New York, United States) containing 10% FBS (GIBCO, New York, United States) under standard conditions in a humidified incubator with 5% CO₂ at 37°C. The cytotoxic effect of the Dpo71 on BEAS-2B cells was measured by Cell Counting Kit-8. The BEAS-2B cells were seeded at density of 10⁴ cells/well in a 96-well plate containing 200 μl of culture medium and incubated at 37°C for 24 h. Next, the cells were incubated with Dpo71 for 12 h followed by incubating with 10 μl of WST-8 solution (Beyotime, Shanghai, China) for another 2 h at 37°C. Absorbance was measured at a wavelength of 450 nm using a microplate reader (Multiskan Sky, Thermo Fisher Scientific, Massachusetts, United States). The PBS group was served as a negative control. All experiments were performed in three biological replicates and repeated at least in two independent times.

The G. mellonella model was conducted following the procedures described by Peleg et al. (2009) with some minor modifications, and referring in other G. mellonella studies (Yang et al., 2015; Blasco et al., 2020). The G. mellonella larvae were acquired from WAGA company in Hong Kong and the injection was performed with a 10 μl SGE syringe (Sigma-Aldrich, Missouri, United States). To infect the G. mellonella, the larvae were first injected with 10⁶ CFU A. baumannii (MDR-AB2 strain) into the last left proleg. Then the PBS buffer (control group) or 5 μg Dpo71; 1 μg colistin; or 1 μg colistin + 5 μg Dpo71 (treatment group), were injected into the last right proleg within 30 min. The G. mellonella were then incubated at 37°C and observed at 24 h intervals over 4 days. The G. mellonella which did not respond to physical stimuli were considered dead. Each group included nine G. mellonella with individual experiments repeated two times (n = 18).

All experimental data are presented as means ± standard deviation (SD), and significance was determined using independent Student’s t tests and the one-way analysis of variance (ANOVA), assuming equal variance at a significance level of 0.05. Comparison of the survival rates of G. mellonella between groups was determined by Kaplan–Meier survival analysis with a log-rank test. All statistical analysis was performed using GraphPad Prism software.

The IME-AB2 phage exhibits plaque surrounded by a halo-zone, suggesting the presence of a depolymerase protein. Bioinformatic analysis reveals that gp71 is a tail fiber protein (Peng et al., 2014) with 43% sequence similarity as the depolymerase encoded by another Acinetobacter phage vB_AbaP_AS12 (Protein Data Bank number 6EU4) (Figure 1A). Expression of the ORF71 sequence in E. coli. yields a protein with more than 95% purity and a molecular mass of about 80 kDa (Figure 1B), matching the calculated value of 80.2 kDa. Size exclusion chromatography shows that the purified Dpo71 elutes as a single peak at a molecular weight larger than 200 kDa (Figure 1C). This corresponds to a trimer, consistent with the expected oligomeric form of phage tail fiber protein which is believed to endure extreme conditions for phage infection and survival (Oliveira et al., 2019a,2020). Circular dichroism (CD) reveals that the Dpo71 protein adopts a well-folded conformation rich in β-sheet structures with a negative dichroic minimum at 215-nm and a positive maximum around 195-nm characteristic peaks (Figure 1D). The melting curves following the CD signal at 215 nm show a melting temperature (Tm) of 58.5°C (Figure 1E). Spot tests were performed next to confirm the ability of Dpo71 in degrading bacterial capsules of the host bacteria MDR-AB2 of the parent phage. Semi-clear spot formation was observed on the bacterial lawn with the spot sizes increasing with the dose of depolymerase from 0.001 to 10 μg range (Figure 1F). The bacterial CPS/LPS degradation activity of Dpo71 was also confirmed by the reduced CPS/LPS concentration and Alcian blue staining (Supplementary Figure 1) using protocols documented in previous studies (Liu et al., 2020).

To evaluate the therapeutic applicability of Dpo71, the enzymatic activity of Dpo71 at various pH values was evaluated by monitoring the turbidity of the residual polysaccharide extracts (CPS/LPS). Figure 2A shows that the enzyme remained active in the range of pH 4–8, covering most of the physiological conditions. Then we measured the toxicity of Dpo71 to mammalian cells, human red blood cells and lung bronchial epithelial cell line (BEAS-2B cells). No hemolytic activity was detected even at a high dose of 500 μg/ml (Figure 2B). Figure 2C also shows Dpo71 has no cytotoxicity against BEAS-2B cells. The result suggests that Dpo71 may be a safe treatment, likely for systemic or pulmonary infections. As stability upon storage is critical for the development of commercially viable protein therapeutics, the storage stability of Dpo71 at 4°C has been evaluated. Results show that Dpo71 is stable for at least 6 months without noticeable activity loss (Figure 2D).

As depolymerase can disintegrate bacterial capsules and thus sensitize bacteria to be killed by host immune system (Yoshida et al., 2000; Spinosa et al., 2007; Oliveira et al., 2020), we first measured whether serum could kill the Dpo71-treated bacterial cells. Two A. baumannii strains (AB#1 and AB#2), belonging to ST208 sequence type (Liu et al., 2019a), sensitive to the parent IME-AB2 phage and two insensitive strains (AB#3 and AB#4) were chosen for the serum killing assays. The four tested strains were resistant to serum killing and continue to grow in human serum without the presence of depolymerase. Bacteria treated with 10 μg/ml Dpo71 and inactivated serum also showed no antibacterial effect (Figure 3A). On the contrary, a remarkable bacterial reduction (8 log) was noted for the two sensitive strains (AB#1 and AB#2), but not for the insensitive ones (AB#3 and AB#4) when the bacteria were treated with active human serum (with a volume ratio of 50%) and Dpo71 (Figure 3A). Furthermore, the time-killing assay on the two sensitive strains was performed with a serum ratio of 1–50%. Complete bacterial eradication was observed after a 5-h treatment at 50% human serum for both AB#1 and AB#2 (Figure 3B and Supplementary Figure 2). It is noteworthy that a 5% serum was sufficient to achieve around 4-log bacterial reduction after 5 h and with minor regrowth after 24 h. This efficacy is significantly higher as compared with previous reports, in which at least 25% human serum is needed to achieve the same killing efficiency (Majkowska-Skrobek et al., 2018; Liu et al., 2019a,2020; Mi et al., 2019; Oliveira et al., 2020).

We next examined the antibiotic adjuvant effect of Dpo71 against MDR-AB2. Colistin was chosen because it was the only tested antibiotic that the MDR-AB2 strain was susceptible to with a minimum inhibition concentration (MIC) of 2 μg/ml (Peng et al., 2014; Supplementary Table 3). The colistin concentration was set at 1 μg/ml, half of the MIC value. Figure 3C shows that 1 μg/ml colistin alone or colistin with 5% serum had no antibacterial effect against the inoculation of 10⁸ CFU/ml. Dpo71 combined with 5% serum could achieve around 4-log bacterial reduction and the antibacterial effect was further enhanced to nearly complete eradiation (residual viable bacteria reduced from 4.4 ± 0.2 log to 0.7 ± 1.1 log) when Dpo71 was used in combination with 1 μg/ml colistin in the presence of 5% serum (Figure 3C). This boosting effect was consistent with a modified checkerboard assay examining the synergy between Dpo71 and colistin that the MIC of colistin dropped from 2 to 0.5 μg/ml with the addition of 5% serum (Data not shown). Notably, these results indicated that Dpo71 could act as an adjuvant to boost the antibacterial activity of colistin in low serum condition.

Although the combined administration of depolymerase and antibiotics was superior to the individual treatments, the underlying mechanisms responsible for the adjuvant effect of Dpo71 have not been fully investigated (Bansal et al., 2014, 2015; Olsen et al., 2018; Wu et al., 2019; Latka and Drulis-Kawa, 2020). We first confirmed that Dpo71 could strip the bacterial capsule using scanning electron microscopy (SEM). Figure 4A shows clearly the morphological changes of the bacterial surface after Dpo71 treatment. The bacterial surface of untreated bacteria possessed a complex polysaccharide capsule with pilus-shaped protrusions, while the surface of the Dpo71-treated group showed the absence of these pilus-shaped protrusions, suggesting the loss of the capsule. Then we evaluated the impact of the capsule on the outer membrane (OM) destabilization capacity of colistin using the 1-N-phenylnaphthylamine (NPN) uptake assay. The fluorescence intensity of NPN for the Dpo71-alone group was comparable with the control group and that of the Dpo71-colistin treated bacteria (decapsulated) was significantly higher than the colistin-alone groups (Figure 4B). These results indicate that Dpo71 depolymerase only acted on bacterial capsules and the removal of the capsule could significantly enhance the OM destabilization capability of colistin. The binding efficiency of colistin with native and decapsulated bacteria was estimated. The Dpo71 treated bacteria had a lower level of free colistin in the supernatant compared with the untreated group, indicating that more colistin binds onto the capsule-stripped bacteria (Figure 4C). These results all suggested that depolymerase can function as an antibiotic adjuvant by disintegrating the bacteria capsule to promote interaction between antibiotics and the bacteria, and hence their entry to the bacterial host.

We next measured the inhibition effect of Dpo71 on biofilm formation (Wilson et al., 2017). Figure 5A shows Dpo71 inhibits biofilm formation in a dose-dependent manner. At 1 μg/ml Dpo71, the residual biomass was 80% as compared with the PBS-treated control. The residual biomass was further reduced to 60.0 ± 7.2% at 10 μg/ml Dpo71 and 58.2 ± 7.0% at 40 μg/ml. Therefore, 10 μg/ml Dpo71 was chosen for the evaluation of adjuvant effects with 1 μg/ml colistin (1/2 MIC) in inhibiting biofilm formation. Colistin alone brought down the biomass to 43.5 ± 4.9%, and further reduced it to 28.9 ± 3.1% when used in combination with Dpo71 (Figure 5B). The biofilm was visualized by the LIVE/DEAD staining, in which live cells were stained with green fluorescence and dead cells with damaged membrane were stained red (Figure 5C). Consistent with the results from the CV staining assay, confocal imaging showed Dpo71 alone and colistin alone could prevent the biofilm formation to a certain extent compared with the PBS-treated sample, but the combination of Dpo71 and colistin yielded the most effective biofilm inhibition. Our results confirmed the capability of both Dpo71-alone and co-treatment with colistin in preventing A. baumannii biofilm formation.

We next assessed whether Dpo71 or its combination with colistin can remove pre-formed biofilm (Wilson et al., 2017). According to the CV staining assay, the biomass could be disrupted by a single treatment of Dpo71 (10 μg/ml) or colistin (4 μg/ml, 2× MIC) to around 60% residual biomass of the PBS control (Figure 6A). The combined treatment could further reduce the residual biomass to 41.5 ± 6.6%. We reached the same conclusion when we compared the number of viable bacterial cells in the dispersed biofilms (Figure 6B). In the absence of Dpo71, colistin was inefficient in killing bacteria embedded in the biofilm (<0.5 log reduction), whereas with the help of Dpo71, more than 90% of the bacterial cells in biofilm were killed (from 7.3 ± 0.1 to 6.2 ± 0.2 in log scale). In the LIVE/DEAD viability assay, the pre-formed biofilm network was dismantled by Dpo71 (Figure 6C). These results show that Dpo71 can efficiently disrupt per-formed biofilm. For biofilms treated with the combination of Dpo71 and colistin, both 2D and 3D confocal images showed only weak fluorescence signals, confirming that the biofilm was effectively removed (Figure 6C). These data indicate that Dpo71 could boost the antibiofilm activity of colistin.

We next evaluated the in vivo efficacy of Dpo71 and colistin in combating bacterial infections in a Galleria mellonella infection model (Figure 7A). In the control group, approximately 70% of the G. mellonella died within 18 h and the death rate increased to 90% at 48 h (Figure 7B). The survival rate of the colistin-treated group increased to 50% after 24 h post-infection, and around 30% endured to the end of the monitoring period (72 h post-infection). Although depolymerase itself is not bactericidal, the Dpo71-alone treatment was found to be effective in rescuing the infected worms with 40% of the G. mellonella surviving for 72 h. The combination treatment increased the survival rate of the infected worms to 80% till the end of the monitoring period, significantly higher than the monotherapy groups (^**p < 0.01, log-rank test).