Cronobacter sakazakii (ATCC 29544) used in this study was purchased from Beijing BeNa Culture Collection (Beijing, China). Strains were maintained at −80°C and the working culture was obtained following the instructions of the supplier. The frozen stock culture was transferred into a 5-mL sterile tryptic soy broth (TSB; Land Bridge Technology, Beijing, China), and the overnight culture was inoculated to a fresh TSB medium at a ratio of 1% (v/v) and incubated at 37°C with agitation, till cell concentration reached 10⁸~10⁹ CFU/mL. The growth curve was depicted by measuring OD_600nm values at different growth timepoints. E. coli β2155 was derived from the strain repository of this lab and grown at 37°C under aeration for cloning of recombinant vectors.

Hypocrellin B (HB) purchased from Yua-nye Biotech (Shanghai, China) was first dissolved in dimethyl sulphoxide (DMSO) at a concentration of 1 mM and stored at 4°C in the dark before the photosensitizer was used for further experiments. The light source used here was LED with a wavelength of 460 nm and a power adjusted to 50.8 mW/cm². The LED system (CREE; Durham, NC, United States) was set up in a closed chamber to avoid the interference of ambient visible light. The optical output power was calibrated by a laser power meter (VLP-2000, Beijing, China). The 460 nm LED was attached to an aluminum heat sink to dissipate the heat generated.

Antimicrobial photodynamic inactivation treatment was performed as previously described (Demidova and Hamblin, 2005). The bacteria at a mid-log phase were centrifuged to remove the supernatant and incubated in PBS with varying concentrations of photosensitizers (0, 10, 15, 20, 25, and 30 μM for HB) in a 37°C incubator under dark environment for 30 min, reaching an optical density of 0.8. The bacterial preparation (1 mL) was added to a 24-well plate and irradiated with 460 nm-LED light at room temperature. The light intensity ranged from 0 to 60 J/cm² at a power density of 50.8 mW/cm². All aPDI experiments were compared with three control groups: blank control, light only or photosensitizer only. All samples were serially diluted with sterile PBS (10⁻¹~10⁻⁶). To determine the appropriate dilution factor, 1 μL of aliquot with various dilution was subjected to TSB solid media (supplemented with 1.3% agar powder) for dot plating. Aliquots of 50 μL were plated in TSB media for enumeration of viable cells. The plates were incubated in a 37°C incubator for 24 h and the emerging colonies were counted. The survival fraction of C. sakazakii was then calculated relative to the control group.

Propidium iodide (PI) and LIVE/DEAD^® BacLight^™ Bacterial Viability Kit (Molecular Probes, Invitrogen, Carlsbad, United States) were used here to detect bacterial membrane integrity and cell viability, respectively. Double-staining performs well in deciphering live/dead ratio, while PI alone could better detect membrane penetrance without interference of other dyes. Following aPDI exposure, the bacteria were centrifuged at 5,000 g for 3 min and washed twice with sterile PBS. The harvested cells were then incubated with PI (3 μL per mL) in the dark at 37°C for 15~20 min. After washing twice with PBS to remove excess dye, the suspension was then transferred to a glass slide for fluorescence microscopy examination (ECLIPSE Ti2-U, Nikon, Tokyo, Japan) using a 590 nm bandpass filter. The fluorescence intensity of each image was quantified by Image J software (NIH, Bethesda, MD, United States). Double staining results are expressed by the ratio of live (green) and dead bacteria (red; Zou et al., 2020).

The bacterial cells were collected by a centrifugation at 8,000 g for 3 min after aPDI treatment. The supernatant preparations were filtered with a 0.22 μm microporous membrane. The concentration of nucleic acids was then determined using a microplate spectrophotometer (Infinite 200 Professional, Tecan, Mannedorf, Switzerland) by OD_260nm values, according to the manufacturer’s instructions.

SEM was performed as described previously (Jan et al., 2019). In brief, C. sakazakii cells were collected, added to the fixative buffer for 2 h at room temperature, and then fixed at 4°C overnight. The sample was then sequentially washed 3 times and eluted with 50, 70, 80, and 90% ethanol, respectively. After another round of centrifugation, ethanol was added to the pellet and incubated for 30 min. Upon ethanol dehydration, the samples were then freeze-dried, coated with gold and observed using the scanning electron microscope (X650, Hitachi, Tokyo, Japan).

After aPDI treatment, 3 μL DCFH-DA (2′, 7′-dichlorofluorescein diacetate) fluorescent dye (Sigma-Aldrich, Shanghai, China) was added to C. sakazakii suspension and incubated at 37°C for 30 min. The excess dye was then washed off and the cells were transferred to a glass slide. Fluorescence signals were observed with fluorescence microscope (Nikon, Tokyo, Japan), and intracellular ROS levels were analyzed using Image J software.

The aPDI-treated (30 μM HB under irradiation of various light doses) C. sakazakii was centrifuged at 3,000 g for 7 min to remove the supernatant and resuspended in an equal volume of sterile PBS. The enzymatic activity of CAT (catalase) was determined with a CAT detection kit (A007-1-1, Jiancheng, Nanjing, China) according to the manufacturer’s instructions. Absorbance at 405 nm was then measured using a microplate spectrometer (infinite 200 pro, Tecan, Mannedorf, Switzerland).

Biofilm quantification was performed by crystal violet (CV) assay. The aPDI-treated C. sakazakii was diluted in TSB and 200 μL of aliquot was transferred to 96-well plates (Corning Life Sciences, Acton, MA) for biofilm formation. The incubation was performed in a 37°C incubator for various length of time (12 h, 24 h, 36 h, and 48 h), to indicate different stages of biofilm dynamics. After incubation, the culture was washed twice and fixed at 60°C for 1 h. The biofilms were then stained with 200 μL of 0.1% crystal violet (CV) for 20 min in the dark, washing 3 times with PBS to remove excess dye. The stained biofilms were then imaged using fluorescence microscope (ECLIPSE Ti2, Nikon, Tokyo, Japan). The OD value of each well was determined at a wavelength of 590 nm using microplate spectrophotometer (Infinite 200 PRO, Tecan, Mannedorf, Switzerland), following biofilm dissolution in 33% acetic acid.

Total RNAs were extracted from aPDI-treated C. sakazakii using the RNeasy Protect Bacteria Mini Kit (Qiagen, Shanghai, China). Subsequently, the reverse transcription reaction was performed using First-Stand cDNA Synthesis SuperMix (Transgen Biotech, Beijing, China) according to the manufacturer’s instructions, to generate cDNA and stored at-20°C until further experiments. qPCR was performed based on cDNA templates using LightCycle96 Q-PCR (Roche, Basel, Switzerland). The relative transcription levels of the objective genes were detected by real-time quantitative reverse transcription PCR, with primers listed in Table 1. Relative expression of individual genes was quantified using 16S rRNA as an internal control. The PCR process consisted of pre-denaturation and 40 cycles of amplification (denaturation at 95°C for 30 s and annealing at 55°C for 30 s).

Suicide plasmid was constructed by ligating upstream and downstream fragments of CpxA into pCVD442, with the aid of Gibson cloning. The homologous arms were amplified by PCR from C. sakazakii ATCC 29544 genomic DNA using primer pairs CpxA5F/R and CpxA3F/R, respectively. The recombinant plasmids were then introduced by electroporation into E. coli strain β2155. After ligation, the resultant plasmid was transformed into the recipient C. sakazakii strain by conjugation. The deletion mutants were selected from sequencing the trans formant colonies, and the isolated genomic DNA was subjected to PCR verification by the primer pair CpxAVF/R (Table 1).

The PIF trial was referenced to a previous study (Zheng et al., 2021), with some modifications: the C. sakazakii culture at a mid-log phase was centrifuged at 8,000 g for 3 min at 4°C, washed twice, mixed with powdered infant formula (PIF, Good Start Soy, Nestle, Beijing, China) with a ratio of 1:20 (w/w). The used PIF contains 12.5% protein, 8.1% polyunsaturated fatty acid and 55.6% carbohydrate (w/w), along with other ingredients. The mixture was incubated without a lid in a sterile incubator at 37°C for 2 day to air-dry before being transferred to a 24-well plate for aPDI treatment. The plate counting was then carried out following aPDI exposure to check the validity of PIF decontamination: after samples were withdrawn and reconstituted in PBS, appropriate dilutions were plated on TSB medium, and the bacterial colonies were then counted after the growth of 24 h at 37°C.

In aPDI-treated PIF samples, pH values were determined to assess food quality by a pH meter (Leitz, Shanghai, China), which was pre-calibrated with standard solution before use.

Cronobacter sakazakii can survive under a range of environmental conditions, represented by desiccation and high temperature, which contributes to its prevalence in PIF. Hence, to investigate effect of aPDI on the bacterial resistance to desiccation, the aPDI-treated PIF was transferred to 24-well plates and stored without a lid in a 37°C sterile incubator pre-treated with air-drying for 2 weeks (Amalaradjou and Venkitanarayanan, 2011). The samples were collected at 0, 1, 2, 4, 7, and 14 day during storage, reconstituted in PBS, and plated in TSB medium for counting. The experimental results are expressed as log₁₀ (N/N0). N is the microbial population after drying (CFU/g), while N0 is the initial cell number before drying (CFU/g).

The ability of C. sakazakii in PIF to survive heat stress was investigated at 45°C, 50°C, and 55°C. The 1.5 g formula was reconstituted in 10 mL sterile distilled water, and 1 mL aliquot was pipetted into 1.5 mL Eppendorf tube, which was subsequently incubated in a temperature-controlled water bath. At various time intervals (0, 20, 40, 60, and 90 min), PIF preparations were collected and subjected for plate counting of viable cells. The data was obtained using the same formula as desiccation trials.

Values are expressed as mean ± SEM (standard error of mean) from at least three independent biological replicates. Statistical analysis was performed using SPSS software (version 19.0, IBM, NY, United States). Independent-sample t-test was used to indicate two group comparisons, and one-way analysis of variance (ANOVA) with post hoc test was adopted to perform multiple comparisons. p < 0.05 indicates a significant difference among groups.

The inhibitory effect of aPDI on the survival rate of C. sakazakii was first evaluated. The aPDI was carried out based on the energy conversion activity of HB, a natural pigment with perylenequinone structure (Figure 1A). The growth curve of C. sakazakii was determined (Supplementary Figure S1) and cells in the mid-log phase were harvested for aPDI test. The doses of photosensitizer were set as 10, 15, 20, 25, and 30 μM, respectively and light intensities of 460-nm LED were adjusted to 20, 40, or 60 J/cm². Under these conditions, the survival rate of C. sakazakii was sharply decreased, reaching a maximum 3.6 log when 30 μM HB was irradiated by 60 J/cm² of blue light (Figures 1B–D). Besides, it’s obvious that the aPDI efficacy was highly dependent on the HB dosage and light intensity used. The representative dot-plating was shown in Figures 1C,D. The bacterial injury was not induced by DMSO, as the solvent alone did not display antimicrobial effect (Supplementary Figure S2). Besides, no dark toxicity of HB was detected in the studied settings (Figures 1B–D).

We next used LIVE/DEAD double staining to consolidate the efficacy of aPDI. 30 μM HB was selected for the subsequent experiments due to the prominent photokilling effect. As evidenced by Figure 1E, the ratio of live/dead C. sakazakii cells was diminished at an irradiation of 20 J/cm², and a rigorous condition (60 J/cm²) augmented the inhibitory impact. To summarize, the HB-based aPDI could inactivate C. sakazakii in a significant way.

In order to explore mechanisms underlying the aPDI-induced injury of C. sakazakii, PI staining was performed to examine the bacterial membrane integrity. According to the results, the membrane integrity was severely damaged by aPDI (Figures 2A,B). This phenomenon was further consolidated by the enhanced DNA leakage (Figure 2C), which is an iconic molecular event of membrane damage.

Membrane damage is closely associated with surface morphological changes (Hu et al., 2018). According to SEM graphs, after aPDI treatment, the anatomical features of C. sakazakii was prone to a series of changes, manifested by distorted, wrinkled and shrunken surface morphology (Figure 2D). Of note, bacterial cells tended to auto-aggregate as light intensity increased, a process involving the attachment of the released substances. Taken together, aPDI has detrimental effect on the membrane integrity of C. sakazakii.

Antimicrobial photodynamic inactivation normally acts by generating a unique combination of ROS molecules. To test if this is the causative agent of C. sakazakii inactivation, we quantified ROS in vivo via DCFH-DA staining. As shown in Figures 3A,B, intracellular ROS was significantly enhanced by aPDI treatment, with the alteration of 2.5 folds under the irradiation of 60 J/cm². This result might implicate ROS generation in the resulting bacterial death. Interestingly, the imposed oxidative stress augmented the enzymatic activity of bacterial catalase (Figure 3C). This change might be attributed to the intrinsic antioxidative response of C. sakazakii, which, however, did not prevent the light-induced accumulation of ROS. ROS was thought to attack proteins, lipids and nucleic acids inside the microbial cells (Hu et al., 2018). Thus, aPDI promotes the generation of ROS, which might play part in the impairment of C. sakazakii cells.

Cronobacter sakazakii can form biofilms by adhering to the surface of materials, which allows it to persist in tough environments (Xu et al., 2021). So we next examined the effect of aPDI treatment on the assembled biofilms. The fluorescence microscopy showed that the photo-activated HB significantly disrupted the biofilm structure at various stages, which were observed at 12 h, 24 h, 36 h, and 48 h, respectively, (Figures 4A,B). It was seen from the graphs that the biofilm tended to be dispersed and disintegrated by the continued light irradiation, suggesting that aPDI was able to attack both planktonic and biofilm cells of C. sakazakii.

CpxRA signaling plays important roles in controlling sensing and adhesion of E. coli (Otto and Silhavy, 2002), thus this TCS system has relevance to biofilm formation. To further inspect the molecular mechanisms underlying biofilm disruption of C. sakazakii, the expression levels of CpxA, CpxR, and CpxP were measured in response to aPDI. According to the results, this treatment reduced the mRNA transcripts of CpxA while upregulating CpxR and CpxP (Figure 4C), which may indicate that the CpxRA TCS was deregulated by the imposed oxidative stress. By profiling the CpxRA-targeting genes, some flagella-related genes were found increased, in parallel with an outer membrane porin, OmpC (Figure 4C), suggesting that CpxRA pathway was aberrantly stimulated by aPDI. In summary, aPDI has deleterious impact on the biofilm of C. sakazakii.

In order to investigate if CpxRA is implicated in the aPDI-mediated removal of C. sakazakii, the mutant strain, ΔCpxA, was established by the homologous recombination (Figure 5A), and its growth curve was shown in Supplementary Figure S1: no evident changes were observed compared to wildtype strain. The mutant strain was susceptible to aPDI treatment, as demonstrated by the addition of 30 μM HB excited by 20 J/cm² of illumination (Figures 5B,C). Of note, compared to wildtype strain (0.9 log), the viability of the mutant was curbed to a larger extent (2.0 log), suggesting that CpxA abrogation diminished the antioxidative capacity of C. sakazakii under the studied context. This finding was then consolidated by the LIVE/DEAD double staining (Figures 5D,E).

Considering membrane integrity, PI staining indicated an aggravated permeability due to the deletion of CpxA (Figures 6A,B). In line with the hampered cell envelope morphology (Figure 6C), it might deduce that CpxA plays pivotal roles in modulating the aPDI potency against C. sakazakii. Moreover, as unveiled by DCFH-DA staining, an increasing amount of ROS (3 folds) was produced as irradiation prolonged (Supplementary Figures S3A,B).

By determining the expression profiles, the mRNA level of CpxR was upregulated in ΔCpxA, indicating the negative mode of interaction between the molecular partners (Figure 7A). In addition, OmpC expression was also promoted upon CpxA deletion, and this trend was reinforced by the addition of light-activated HB (Figure 7B). This data might implicate a CpxRA-OmpC pathway in the studied sterilization course. In the event of biofilm formation (Figures 7C–E), the ΔCpxA biofilm displayed a more dispersed status, a proof that CpxA deletion rendered C. sakazakii prone to the oxidative stress brought by aPDI in either planktonic or biofilm forms. These results demonstrate that CpxRA modulates the effect of aPDI on C. sakazakii.

Cronobacter sakazakii is a well-defined foodborne pathogen contaminating PIF (Zhang et al., 2020). To this end, a laboratory PIF model was constructed to investigate the validity of aPDI in reducing the risk of C. sakazakii infection. PIF was first inoculated with C. sakazakii at a mid-log stage, mixed with hypocrellin B and then subjected to 460-nm LED illumination. After this transient manipulation, the contaminated PIF was placed at room temperature for a variable length of time, followed by bacterial enumeration. The results showed that aPDI treatment decreased the viable number of C. sakazakii in the infant formula with a margin of 85% (Figure 8A), in the studied contamination model. Of note, no pH values of PIF were altered upon treatment (Figure 8B). Next we examined the PIF preservation under various storage conditions. Seen from Figures 8C–F, desiccation or heated stress reduced the contamination of C. sakazakii on a case-to-case basis, whereas in all cases their survival rate further declined with an early exposure of aPDI. This data might suggest that aPDI has prophylactic effect on PIF persistence. In summary, aPDI can decontaminate PIF from C. sakazakii while maintaining the normal food property.