The strains, plasmids and oligonucleotide primers used in this study are shown in Supplementary Table S1 . All strains except those containing the pCaskp plasmid were grown in liquid (shaking at 220 rpm) or solid LB medium at 37°C in incubator or shaker (Zhichu).

The DNA sequences for glgB and glgX from K. pneumoniae MGH78578 (GCF_000016305.1) were retrieved from the GeneBank database respectively. Utilizing the genomic DNA of K. pneumoniae MGH78578 as a template, the glgB and glgX genes were amplified via PCR using the PrimerStar DNA Polymerase (Takara). Subsequently, the amplified glgB and glgX sequences were cloned into the pET-28a vector through the In-Fusion Snap Assembly Master Mix (Takara).

E.coli BL21 (DE3) cells (Transgen Biotech) containing pET28a-glgB or pET28a-glgX vector were inoculated into LB medium with kanamycin and incubated at 37°C until OD600 was 0.8, and then incubated at 16°C for 20 h with isopropyl β-D-1-thiogalactoside of 0.5 mM. Bacterial cells were resuspended with binding buffer (50 mM Tris, 300 mM NaCl, pH 7.5) and lysed by FastPrep-24™ 5G homogenizer (MP Biomedicals). Recombinant proteins were enriched by Ni-NTA resins and eluted with elution buffer (50 mM Tris, 500 mM Imidazole, 300 mM NaCl, pH 7.5). Fractions were concentrated by an Amicon Ultra-15 centrifugal filter (Millipore) with a molecular weight cutoff of 10 kDa.

The assessment of enzyme activity was modified based on previous study (Schumacher et al., 2022). Briefly, the enzyme activity is assessed by measuring the increase in reducing sugars in the reaction solution. After purification, the enzyme is mixed with phosphate buffer and a final concentration of 0.05% glycogen solution, and incubated at 37°C for one hour. Following incubation, an equal volume of bicinchoninic acid (BCA) reagent is added to the reaction mixture and vortexed for 60 seconds on a vortex mixer (IKA), followed by incubation at 80°C for 30 minutes. After the reaction solution has cooled to room temperature, 125 µL of the solution is transferred to a 96-well plate for the measurement of absorbance at 562 nm. D-glucose is used to create the standard curve, and the absorbance values at 562 nm are converted to reducing ends [µM glucose equivalents]. All measurements are performed in triplicate.

To determine the optimal temperature, the enzyme is placed in phosphate buffer at pH 7.5, and its activity is measured at different temperatures in increments of 5°C. For the determination of the optimal pH, the phosphate buffer is replaced with sodium dihydrogen phosphate (100 mM, pH 6.0-7.5) and tris(hydroxymethyl)aminomethane (50 mM, pH 8.5-9.0) buffer solutions, respectively, and the enzyme activity is measured at the optimal temperature. Regarding temperature stability, the enzyme is incubated at the optimal temperature and pH for durations ranging from 4 to 28 hours, after which the activity is assessed.

A dual-plasmid CRISPR/Cas9 system was employed to delete the glgB and glgX genes in K. pneumoniae MGH78578 (Wang et al., 2018). The kanamycin resistance gene in the pSGKP plasmid was replaced with a hygromycin B resistance gene to circumvent the kanamycin resistance of K. pneumoniae MGH78578. The pCasKP plasmid was first electroporated into K. pneumoniae MGH78578 and the correct transformants were selected on LB plates containing apramycin. Subsequently, the pSGKP plasmid carrying spacer sequences along with ssDNA or dsDNA homology arms was co-transformed into K. pneumoniae MGH78578 containing the pCasKP plasmid, which had been induced with L-arabinose. The transformants were then screened on plates containing both apramycin and hygromycin B to further validate the deletion of the glgB or glgX genes. Strains with successful deletion of glgB or glgX were cultured overnight at 37°C on plates containing 5% sucrose to cure the plasmids. Strains that lost the pSGKP plasmid alone were cultured on plates with 5% sucrose at 30°C for subsequent gene deletion, and the strain with the simultaneous deletion of glgB and glgX was named ΔglgBX.

To determine the growth rate of the strains, they were cultured overnight in LB medium and then inoculated at a 1% dilution into M63+ medium. The Microscreen HT growth curve analyzer (Jieling Instruments) was used for the measurement of growth curves. The culture medium was added to the 48-well deep-well plate in 1 mL per well and placed in the instrument for incubation at 37°C and 800 rpm. The growth was measured every 20 minutes for a continuous period of 24 hours. All reactions were performed in triplicate.

The M9 minimal medium was employed for the cultivation of various K. pneumoniae. After the bacterial cells were harvested by centrifugation, the glycogen/protein ratio was determined using the method described in previously study (Li et al., 2023). Bacteria cells were lysed by FastPrep-24™ 5G homogenizer (MP Biomedicals). After centrifugation, the supernatant was reacted with 60% KOH at 95°C for 2 hours. After treatment, the supernatant was divided into two equal parts, one of which was used to determine protein concentration using the BCA method, while the other was used for the detection of glycogen content. For the determination of glycogen content, pre-cooled ethanol was added to the samples, and the mixture was thoroughly vortexed to facilitate glycogen precipitation. The mixture was then placed at -20°C for 20 hours. The precipitates containing glycogen granules were collected by centrifugation at 4°C and then treated with ethyl alcohol to wash away impurities, with the washing procedure repeated three times. Thereafter, the supernatant was discarded, and the glycogen granules were placed at 60°C to dry any remaining moisture. All glycogen granules were then resuspended in a sodium acetate solution (50 mM, pH 4.8, containing 0.3 mg/mL amyloglucosidase) and incubated overnight at 37°C to completely convert the glycogen granules into glucose. The glucose content in the reaction system was measured using a glucose detection kit (Solarbio) to estimate the glycogen content and calculate the glycogen/protein ratio.

The measurement of biofilm biomass was performed using a modified method described in previous study (Haney et al., 2021). K. pneumoniae MGH78578 and three deletion mutants were incubated overnight and then diluted to 1% in M63+ medium. The diluted bacterial solution was incubated with 96-well plate at 37°C for 24h. After incubation, the 96-well plate was washed twice with deionized water to remove floating bacteria and medium. The biofilm of each well was stained with 200 μL of 0.1% (w/v) crystal violet for 15 min, and washed twice with deionized water. It was dissolved with 150 μL of 33% (v/v) glacial acetic acid for 15 min, and the absorbance at 600 nm was measured using BioTek 800 TS microplate detector (Agilent).

The bacterial cells were stained with SYTO-9 prior to observation with the confocal laser scanning microscope (CLSM), as described in previous studies with some modifications (Qiao et al., 2022). K. pneumoniae MGH78578 and deletion mutant strains were incubated overnight and then diluted to 1% in M63+ medium. The diluted bacterial solution was placed in Lab-Tek™ Chamber Slide (Thermo Scientific) at a volume of 400 µL per well and incubated statically at 37°C for 12 hours. After incubation, the medium was removed, and ddH₂O was used to wash away the planktonic bacteria, with the washing procedure being repeated once. SYTO-9 dye was then added to wells and incubated in the dark for 20 minutes. After discarding the dye, the wells were washed to remove any residual dye. A Zeiss LSM 710 Confocal Laser Scanning Microscope (CLSM) was utilized to observe the stained biofilms. Subsequent to acquisition, the images were processed with ZEN 2009 software and subjected to analysis with COMSTAT 2.1 software for the quantification of biofilm biomass.

The virulence of K. pneumoniae is assessed using Galleria mellonella larvae model (Schaefer et al., 2024). G. mellonella larvae were randomly divided into four groups, with ten in each group. Three groups were injected with 5 µL of wild type, ΔglgB and ΔglgX mutants, and the control group was injected with 5 µL of PBS solution. Larvae were incubated in an incubator at 37°C and 60% humidity. The survival of the larvae was observed from day 0 to day 4, with larvae that were unable to right themselves being considered deceased.

Statistical analyses were conducted utilizing GraphPad Prism software, wherein a one-way ANOVA was employed to assess the significance of differences. Each experimental was conducted with a minimum of three replicates to ensure reliability of the results.

GlgX and GlgB both belong to the GH13 family, which contains an α-amylase domain, and they can catalyze the degradation of glycosidic bonds. The GlgX and GlgB proteins were expressed in E. coli and purified with Ni-NTA resin to obtain soluble proteins ( Figure 1A ). The degradation activities of GlgX and GlgB against glycogen, pullulan, amylose, and amylopectin were measured using the BCA method. We observed that GlgB catalyzed the hydrolysis of all substrates to a nearly equivalent extent ( Figure 1B ). In contrast, GlgX exhibits a pronounced substrate specificity, with the highest degradation activity towards glycogen and pullulan, while its activity towards amylose and amylopectin is significantly reduced ( Figure 1C ). Previous studies have indicated that Bis-(3′-5′)-cyclic dimeric GMP (c-di-GMP) enhances the hydrolytic activity of GlgX from E. coli towards glycogen (Schumacher et al., 2022). However, the addition of c-di-GMP at a final concentration of 50 μM did not increase the enzyme activity ( Figure 1D ). Subsequently, the enzymatic properties were characterized. Notably, both enzymes showed activity in a broad optimal temperature range, with peak activity observed between 35-40°C ( Figure 2A ). The optimal pH for GlgB was found to be 7.5, whereas for GlgX it was 7.0 ( Figure 2B ). When pH exceeding 7.5, the activity of both enzymes was markedly reduced ( Figure 2B ). To evaluate the stability of the enzymes, they were incubated at 37°C for 28 hours. A significant decrease in activity was noted after 4 hours of incubation, followed by a gradual decline in activity thereafter ( Figure 2C ). Two proteins have been shown to possess glycogenolytic activity in vitro, prompting us to further examine their activity in vivo.

GlgB and GlgX are annotated as glycogen branching enzyme and debranching enzyme, respectively, in the UNIPROT database. It has been evidenced that these enzymes play a crucial role in bacteria, such as bacterial growth and the synthesis of glycogen (Wang et al., 2020). To investigate the role of protein in K.pneumoniae, we deleted the active center of GlgX and GlgB, resulting in the ΔglgX, ΔglgB, and ΔglgBX mutants ( Supplementary Figure S1 ). Analysis of growth curves revealed that the ΔglgX mutant exhibited an accelerated growth rate, while the ΔglgB mutants and the ΔglgBX mutant displayed a slower growth pattern compared to the wild-type strain ( Figure 3A ). As glycogen serves as a crucial energy reserve for bacterial growth, the growth rate of a strain may be influenced by its glycogen content. Therefore, we quantified glycogen levels at different times throughout the bacterial growth cycle. The glycogen content in both the wild-type and ΔglgX mutants reached a peak at 8 hours post-inoculation and subsequently decreased, with the ΔglgX mutant maintaining a slightly higher glycogen level than the wild type ( Figure 3B ). The glycogen content in the ΔglgB and ΔglgBX mutants was comparable and remained relatively stable during the growth phase, consistently lower than that of the wild type ( Figure 3B ). The correlation between glycogen content and growth rate among the strains suggests that the amount of glycogen present impacts the growth rate of the bacterial strains. Additionally, since the glgB gene is located upstream of the glgX gene, which leads to the phenotypic similarity between ΔglgB and ΔglgBX mutants, future research will be directed exclusively towards investigating the ΔglgB and ΔglgX mutant.

Previous study has indicated that bacterial glycogen can influence biofilm formulation (Wang et al., 2020). To quantify the impact of glycogen on biofilm formation, we utilized crystal violet staining to assess biofilm biomass. The data revealed a reduction in biofilm biomass following the deletion of the glgB gene, while the deletion of the glgX gene did not alter biofilm biomass ( Figure 4A ). To further investigate the effects of glycogen metabolism on biofilm formation, we employed CLSM and utilized SYTO9 stain the bacterial cells, respectively. In the corresponding images, both ΔglgX and ΔglgB mutants showed a decrease in bacterial biomass compared to the wild type, with the ΔglgB mutant showing a greater decrease in biomass than the ΔglgX mutant ( Figure 4B ). Subsequently, the COMSTAT software was employed to quantify the bacterial biomass within the images and to assess the significance of the biomass reduction. We found that the biomass of the ΔglgB mutant was significantly lower than that of the wild type, whereas the decrease in biomass of the ΔglgX mutant was not significant, indicating that the deletion of glgB more profoundly affects the biofilm formation capacity of K. pneumoniae ( Figure 4C ). This indicates that the deletion of glgB reduces the biofilm-forming ability of K. pneumoniae.

The deletion of glgB and glgX genes in bacteria resulted in alterations to the growth rate and biofilm biomass compared to the wild type, leading us to hypothesize that these changes might also affect the virulence of the bacteria. Given that the G. mellonella larval model has been widely used in numerous studies to assess bacterial virulence and the efficacy of antimicrobial agents, we adopted this model to assess the virulence profiles of distinct K. pneumoniae isolates (Jin et al., 2024; Firacative et al., 2020). Following the inoculation of larvae with 1×10⁶ CFU of the respective K. pneumoniae strains, a marked difference in survival rates was observed. Larvae injected with the ΔglgB mutant strain exhibited a consistent survival rate of 90% from the initial to the fourth day post-inoculation ( Figure 5 ). In contrast, the survival rate of larvae exposed to the wild-type strain progressively declined from 80% to 60% over the same period, signifying a reduced virulence in the ΔglgB mutant strain compared to the wild-type ( Figure 5 ). Notably, the survival rate of larvae inoculated with the ΔglgX mutant strain decreased from 60% to 50% between the initial day and the fourth day following inoculation, implying a virulence slightly higher than the wild-type strain ( Figure 5 ). Considering that the biofilm formation capacity of the ΔglgB strain is significantly reduced compared to the wild-type strain, this suggests that the absence of glycogen in bacteria leads to a decrease in key factors such as biofilm formation, which in turn diminishes the virulence of the organism.