A Method to Directly Identify Cronobacter sakazakii in Liquid Medium by MALDI-TOF MS

Matrix-assisted laser desorption ionization time-of-flight mass spectrometry has been widely used as an emerging technology for the rapid identification of microorganisms. Cronobacter sakazakii (C. sakazakii) is a food-borne pathogen of particular importance to the powdered infant formula (PIF) processing environment due to its high lethality in infants. However, the traditional solid spotting detection method of pretreating samples for MALDI-TOF MS leads only to qualitative detection of C. sakazakii. We developed a new, low-cost, robust liquid spotting pretreatment method and used a response surface methodology to optimize its parameters. The applicability, accuracy, and quantitative potential were measured for different types of samples. The optimal parameters of this method were as follows: a volume of 70% formic acid of 25 μL, treatment with ultrasound at 350 W for 3 min, and a volume of acetonitrile added of 75 μL. These conditions led to the highest identification score for C. sakazakii (1926.42 ± 48.497). This method was found to detect bacteria accurately and reproducibly. When 70 strains of C. sakazakii isolates were analyzed with this method, the identification accuracy was 100%. The detection limit of C. sakazakii in environmental and PIF samples was 4.1 × 101 cfu/mL and 2.72 × 103 cfu/mL, respectively.


Introduction
Food-borne pathogens are of intense interest, as they are often strongly pathogenic and can persist in a variety of environments [1][2][3]. Among these pathogens, C. sakazakii is of particular importance to the infant formula industry due to its high lethality in infants and because infant formula manufacturing plants are susceptible to contamination [4,5]. Accordingly, industries dealing with the production and processing of powdered infant formula (PIF) tend to be highly regulated throughout the world.
Effective control of C. sakazakii can be established by ensuring rapid throughput of susceptible materials. In addition, continuous monitoring of the processing environment can provide regulators with an understanding of the level of risk present in the final product. However, incidents of C. sakazakii contamination of PIF continue to emerge globally without interruption [6][7][8][9]. There is an urgent need to develop a rapid and reliable identification method to monitor C. sakazakii in PIF processing environments and in final products.
Traditional pathogen detection methods, such as biochemical [10], colorimetric [11], and enzyme-linked immunosorbent assays [12], are limited due to their cumbersome operations and requirements for highly trained technicians [13]. Other problems, including challenges with distinguishing similar species and genera and the identification of microorganisms that are difficult to cultivate, further make these methods less than optimal [14,15]. used to evaluate the accuracy of direct identification from liquid culture media. All strains were cultured in LB liquid media overnight at 37 • C with shaking at 150 rpm. Cells in these cultures were isolated by centrifugation at 12,000× g for 2 min at 4 • C, and the pellets were resuspended in sterilized PBS (0.1 M, pH 7.4). The number of bacterial cells was determined by the TSB agar plating method.

Direct Identification of C. sakazakii from Liquid Culture Media Pretreatment Method Process by MALDI-TOF MS
Pretreatment of samples was performed by transferring 500 µL of a suspension containing bacteria at a density of 10 8 cfu/mL into a 1.5 mL sterile centrifuge tube, which was centrifuged at 12,000× g for 2 min. The supernatant was removed, and the pellet was washed twice with 500 µL of sterile PBS. FA was added, and the mixture was subjected to ultrasonic disruption. ACN was then added, and the mixture was mixed by vortexing for 2 min and then centrifuged at 12,000× g for 2 min. A sample (1.5 µL) of the supernatant was applied to the target plate and allowed to incubate for 10 min to air dry. Next, an aliquot (1.5 µL) of a MALDI matrix solution containing CHCA was added, and the mixture was air dried prior to mass spectrometric analysis. The overall process is shown schematically in Figure 1.
(including C. universalis, C. turicensis, C. muytjensii, C. dublinensis, and C. malonaticus) were used to compare the performance of analytical methods involving solid spotting and liquid spotting. The 66 C. sakazakii strains used in this study were all isolated from various environments within PIF production facilities, specifically fluidized beds, fixed beds, and U-shaped valves, as well as spray-dried powder samples. These PIF-related cultures were used to evaluate the accuracy of direct identification from liquid culture media. All strains were cultured in LB liquid media overnight at 37 °C with shaking at 150 rpm. Cells in these cultures were isolated by centrifugation at 12,000× g for 2 min at 4 °C, and the pellets were resuspended in sterilized PBS (0.1 M, pH 7.4). The number of bacterial cells was determined by the TSB agar plating method.

Direct Identification of C. sakazakii from Liquid Culture Media Pretreatment Method Process by MALDI-TOF MS
Pretreatment of samples was performed by transferring 500 µL of a suspension containing bacteria at a density of 10 8 cfu/mL into a 1.5 mL sterile centrifuge tube, which was centrifuged at 12,000× g for 2 min. The supernatant was removed, and the pellet was washed twice with 500 µL of sterile PBS. FA was added, and the mixture was subjected to ultrasonic disruption. ACN was then added, and the mixture was mixed by vortexing for 2 min and then centrifuged at 12,000× g for 2 min. A sample (1.5 µL) of the supernatant was applied to the target plate and allowed to incubate for 10 min to air dry. Next, an aliquot (1.5 µL) of a MALDI matrix solution containing CHCA was added, and the mixture was air dried prior to mass spectrometric analysis. The overall process is shown schematically in Figure 1.

Single-Factor Test
The parameters that were analyzed as single factors were the volume of 70% FA added (10,20,30,40, and 50 µL), the ultrasound power (300, 350, 400, 450, and 500 W), the time of ultrasound exposure (1, 2, 3, 4, and 5 min), and the volume of ACN added (20,40,60,80, and 100 µL). The MALDI-TOF MS spectral data acquisition was performed under the following conditions: laser energy, 5 µJ; laser frequency, 5000 Hz; detector voltage, −0.56 kV; and focus mass, 10,000 u. The spectral data were imported into the QuanID DB database for comparison, and the identification of microorganisms was accomplished by analyzing the mass-to-charge ratios of characteristic peaks. The scores that represented the accuracy of the identification were used to determine the optimal parameters for each of the single factors. Each experiment was repeated three times.

Response Surface Methodology
According to the optimal parameters determined in the single factor experiments and based on the Box-Behnken experimental design principle, a response surface analysis method with four factors and three levels was designed (Table 1). Using the MALDI-TOF MS microbial identification score as the evaluation index, the parameters used for sample pretreatment of liquid cultures prior to MALDI-TOF MS were tested systematically. Each experiment was repeated three times. Values generated in two conditions were compared using one-way ANOVA with SPSS 17.0, and the least significant difference test was used to compare multiple results.

Evaluation of the Accuracy of Identification of the MALDI-TOF MS Liquid Spotting Pretreatment Method
The effects of different types of liquid growth media (LB, NB, and TSB) on the identification method were analyzed using identical culture conditions. In this series of experiments, 66 strains of C. sakazakii that were isolated from different sources and four reference strains of C. sakazakii were subjected to the identification method utilizing the optimized liquid spotting pretreatment method. The QuanID DB microbial database was used as a reference for this evaluation.

Comparison of Solid and Liquid Spotting Methods
Strains of Cronobacter spp. were streaked onto solid agar and incubated for 12 h. The solid spotting method was performed by using a 1 µL inoculation loop to isolate a single colony from this plate. This colony was transferred to a target plate, and 2 µL of 70% FA was added dropwise. After the spot was air-dried, MALDI matrix solution (1.5 µL) was added dropwise, and mass spectrometric identification was performed after this sample was dry. The liquid spotting method was performed using each optimized parameter as described above. The two methods were compared using cluster analyses of the spectral data as performed using Origin 2018 software (version number: OriginPro 2018C).

Direct Detection and Analysis of the Linear Relationship between the Cell Number of C. sakazakii in Environmental Samples
The detection of C. sakazakii simulated in environmental samples is performed by adding C. sakazakii cells to sterile PBS by gradient dilution as follows: Cells obtained during the collection of environmental pathogens or the reference strain C. sakazakii ATCC 29544 were subjected to serial 10-fold dilutions in sterile PBS to obtain 1 mL samples with cell numbers ranging from 10 8 to 10 0 cfu/mL. Identification was carried out by MALDI-TOF MS using the optimized liquid pretreatment method, and the detection limit was determined. Under the culture conditions of 37 • C and 150 rpm, using TSB, LB, and NB liquid medium to optimize the pre-enrichment medium, medium addition amount (400, 600, 800, and 1000 µL), and pre-enrichment time (0, 2, 4, and 6 h), and finally determine the optimal conditions for C. sakazakii ATCC 29544 in environmental collection samples. According to the linear regression equation with the number of C. sakazakii cells as the abscissa and the peak intensity in the obtained identification spectrum as the ordinate, each experiment was repeated three times. The relative standard deviation (RSD) was calculated to evaluate the repeatability and stability of the method.

Direct Detection and Analysis of the Linear Relationship between the Cell Number of C. sakazakii in PIF Samples
The feasibility of simulating the MALDI-TOF MS pretreatment method for the number of C. sakazakii cells in collected PIF samples is as follows: Samples of PIF (20 mg/mL) verified to be free of C. sakazakii contamination were spiked with a target strain of bacteria. The samples were serially diluted 10-fold in sterile PBS to generate samples containing various cell densities, from 10 8 to 10 0 cfu/mL. The cells were collected by centrifugation for 2 min at 12,000× g, and the pellets were added to 1 mL samples of commercial PIF. After a pre-enrichment incubation for 0, 2, 4, or 6 h at 37 • C and shaking at 150 rpm, these samples were mixed by vortexing. Portions (500 µL) of the resulting samples were centrifuged at 5000× g for 5 min, and the supernatants were removed. The pellets were washed only several times with sterile PBS to completely eliminate the lipid layer. The other components in the PIF sample are not further isolated and purified, minimizing the loss of C. sakazakii cell numbers. MALDI-TOF MS identification was then performed using the optimized liquid pretreatment method. The peak intensities in the resulting spectra were plotted as a function of the number of C. sakazakii cells, and the linear regression equation defining the relationship was obtained. Each experiment was repeated three times. The RSD was calculated to evaluate the repeatability and stability of the method.

Effects of Single Factors on the Identification Score of C. sakazakii from Liquid Culture Media by MALDI-TOF MS
The effects of several pretreatment factors were tested individually. Specifically, these factors were the amount of 70% FA added, the power of the ultrasound used to disrupt the cells, the time of ultrasonic treatment, and the amount of ACN added. When the MALDI-TOF MS technique was used to process the cultures, the accuracy of the identification was assigned a score. The relationships between this score and each factor are shown in Figure 2.
As shown in Figure 2A, the MALDI-TOF MS identification score of C. sakazakii in liquid culture was optimal when the volume of 70% FA added was 20 µL. When the amount of 70% FA exceeded 20 µL, the identification score was significantly lower than the score obtained following the addition of 20 µL of 70% FA (p < 0.05). This negative impact of excess FA addition may be due to the degradation of ribosomal proteins released by C. sakazakii upon lysis of the cell membrane, resulting in differences in the experimental and reference spectra. In support of this mechanism, higher volumes of 70% FA led to fewer ribosomal protein matches, resulting in a lower identification score. Therefore, 20 µL was selected as the optimal amount of 70% FA to be added.
The relationship of identification score with ultrasonic power showed an upward trend between 300 and 350 W, which may be due to increased cavitation and enhanced cellular lysis. An increased permeability of C. sakazakii cells would be expected to increase the release of ribosomal proteins. When ultrasound was applied at 350 W, the outer membrane of the cell was likely completely broken, resulting in complete dissolution of the ribosomal proteins and maximization of the MALDI-TOF MS identification score. Higher ultrasonic power may have changed or destroyed ribosomal proteins, affecting the positions of their correlated peaks in the spectrum and thereby reducing the identification score [29]. Therefore, 350 W was selected as the optimal pretreatment ultrasonic power. As shown in Figure 2A, the MALDI-TOF MS identification score of C. sakazakii in liquid culture was optimal when the volume of 70% FA added was 20 µL. When the amount of 70% FA exceeded 20 µL, the identification score was significantly lower than the score obtained following the addition of 20 µL of 70% FA (p < 0.05). This negative impact of excess FA addition may be due to the degradation of ribosomal proteins released by C. sakazakii upon lysis of the cell membrane, resulting in differences in the experimental and reference spectra. In support of this mechanism, higher volumes of 70% FA led to fewer ribosomal protein matches, resulting in a lower identification score. Therefore, 20 µL was selected as the optimal amount of 70% FA to be added.
The relationship of identification score with ultrasonic power showed an upward trend between 300 and 350 W, which may be due to increased cavitation and enhanced cellular lysis. An increased permeability of C. sakazakii cells would be expected to increase the release of ribosomal proteins. When ultrasound was applied at 350 W, the outer membrane of the cell was likely completely broken, resulting in complete dissolution of the ribosomal proteins and maximization of the MALDI-TOF MS identification score. Higher ultrasonic power may have changed or destroyed ribosomal proteins, affecting the positions of their correlated peaks in the spectrum and thereby reducing the identification score [29]. Therefore, 350 W was selected as the optimal pretreatment ultrasonic power.
The effect of the time of ultrasound treatment on the MALDI-TOF MS identification of C. sakazakii in liquid culture was also investigated ( Figure 2C). The identification scores first increased until the treatment time reached 2 min, and then the score decreased with longer processing times until it ultimately reached a stable value. While longer ultrasound treatments would be expected to result in a more efficient release of ribosomal proteins, the thermal effect caused by continuous sonication may lead to the degradation of certain The effect of the time of ultrasound treatment on the MALDI-TOF MS identification of C. sakazakii in liquid culture was also investigated ( Figure 2C). The identification scores first increased until the treatment time reached 2 min, and then the score decreased with longer processing times until it ultimately reached a stable value. While longer ultrasound treatments would be expected to result in a more efficient release of ribosomal proteins, the thermal effect caused by continuous sonication may lead to the degradation of certain largemolecular-weight proteins, thereby affecting the accuracy of identification [30]. Therefore, an ultrasonic treatment time of 2 min was determined to be optimal.
As shown in Figure 2D, the influence of the volume of ACN addition on the identification score showed an upward trend in the range of 20 to 60 µL. At 60 µL ACN, the identification score reached a plateau that continued over a range of 60 to 100 µL ACN. These results suggest that 60 µL of ACN leads to complete disruption of the membrane and release of the ribosomal proteins of C. sakazakii, as adding additional ACN did not significantly improve the identification score. In consideration of reagent cost minimization, the optimal volume of ACN was 60 µL.
The results of the test of significance for the regression equation coefficients are shown in Table 3. The order of influence of each factor on the identification score of C. sakazakii was: volume of ACN added (D) > ultrasonic time (C) > volume of 70% FA added (A) > ultrasonic power (B). Among these factors, the effect of A 2 was extremely significant (p < 0.01), and the effects of C 2 and D 2 were significant (p < 0.05).

Analysis of Response Surface and Two-Dimensional Contour Plots
Two-dimensional contour plots were developed as graphical representations of the regression model ( Figure 3). The shapes of such plots can be used as an indication of the significance of the interactions between two tested variables, where circular contour plots indicate that the interactions are not significant and elliptical or saddle-shaped contour plots suggest that the interactions are significant.   Figure 3a,b compare the effects of the volume of 70% FA added and the volume of ACN added on the identification scores. In these plots, when the amount of FA was constant, the MALDI-TOF MS identification score was affected by the volume of ACN added; here, the identification score increased until it reached a maximum and then decreased. The contour arrangement of the relationship between these two factors was relatively loose, and the elliptic curvature of the plot was small, indicating that the interaction between the two factors was not significant.   the identification score increased until it reached a maximum and then decreased. The contour arrangement of the relationship between these two factors was relatively loose, and the elliptic curvature of the plot was small, indicating that the interaction between the two factors was not significant.
The effects of the volume of ACN added and ultrasonic power on the identification score were also tested (Figure 3c,d). Here, the MALDI-TOF MS identification score increased with the addition of ACN when the ultrasonic power was kept constant. The contour lines of the ACN addition axis are more dense than those on the ultrasonic power axis, indicating that ACN addition had a greater impact on the MALDI-TOF MS identification score and that the impact of ultrasonic power on the identification score was small [31]. When tested as single factors, the influences of ultrasonic time and volume of ACN added on the identification score were found to be the same, as both increased up to a maximum point and then decreased (Figure 3e,f).
As is evident from Figure 3g,h, the contour line of the axis relating to the volume of 70% FA added is significantly denser than the contour line of the ultrasonic power axis, indicating that the effect of FA addition on the identification score is greater than that of the ultrasonic power. The effect of ultrasonic treatment time on the MALDI-TOF MS identification score tends to increase first and then decrease, and vice versa (Figure 3i,j). These data also demonstrate that the interaction between these two factors was not significant (p > 0.05). Similarly, when the ultrasonic time was held constant and the ultrasonic power was increased, the slope of the graph of the relationship to the identification score was negative (Figure 3k,l), indicating that the score increased. This lower identification score may be due to disruption of ribosomal proteins due to the thermal effect of excessive sonication time or power [30].

Optimization of Pretreatment Parameters and Validation of the Optimized Conditions
When we analyzed our model with Design-Expert 8.0.6.1 software, the optimal pretreatment conditions for the liquid spotting MALDI-TOF MS method were found to be: 25.62 µL of 70% FA added, ultrasonic power of 359.15 W, time of ultrasonic of 3 min, and 75 µL of ACN added. These conditions led to an identification score of 1926.42 when applied to a liquid culture of C. sakazakii. To test the reliability of the results from the response surface method, the optimal pretreatment conditions were applied to identification tests using other strains grown in liquid culture. In consideration of the practical limitations of the procedure, the pretreatment parameters were adjusted to 25 µL of 70% FA added, 350 W of ultrasonic power, 3 min of ultrasonic treatment time, and 75 µL of ACN added.
Three parallel tests were carried out under the adjusted conditions. The average strain identification score was determined to be 1972.00 ± 23.356 after the liquid sample pretreatment method. This value was consistent with the theoretical value of 1926.42 ± 48.497, for a relative deviation of 2.3%. In other words, our results demonstrated that the response surface method optimization model was well fitted with the experimental observations, indicating that the model is reliable.

Evaluation of the Accuracy of Identification Using the Liquid Spotting Pretreatment Method
The growth and metabolism of microorganisms can be affected by the composition of the growth medium [32]. Therefore, we endeavored to investigate the influence of different medium components on the identification score of the MALDI-TOF MS microbial liquid spotting method. Here, we cultivated C. sakazakii ATCC 29544 to a density of 10 8 cfu/mL in different media mixtures (LB, NB, and TSB). We then used the optimized processing method for MALDI-TOF MS-based identification.
The MALDI-TOF MS identification scores for this comparison are shown in Table 4. There was no significant difference in the identification scores of cultures grown in different media (p > 0.05). This result indicates that the composition of the medium used to culture target microorganisms is not an important factor in the identification process.  (Table 5). Therefore, we conclude that the MALDI-TOF MS liquid pretreatment method that was developed and optimized can be applied to the identification of C. sakazakii strains from various sources.

16S rDNA Identification Result
CE30 Plant product

Comparison of Solid Spotting/Liquid Spotting Methods
The reproducibility of the liquid spotting pretreatment method was evaluated with a clustering analysis. Clustering is the process of categorizing data into different classes or clusters, permitting a comparison of the similarity of individual data points and thus how well a method of microbial identification is able to distinguish similar subjects [33].
This strategy was used to investigate the reproducibility of the liquid spotting pretreatment method and the ability of the method to differentiate different organisms. Figure 4 shows the results of cluster analyses that compared the MALDI-TOF MS spectral data generated from Cronobacter spp. at different species levels with data generated from different strains at the same species level. As shown in Figure 4A, after pretreatment using the liquid spotting technique, the different species levels that characterize Cronobacter spp. can be well distinguished, and the results have good reproducibility. Compared with the cluster analysis data generated from liquid spotting ( Figure 4A), lower reproducibility was observed in a cluster analysis of MALDI-TOF MS data generated using the traditional solid spotting method ( Figure  4B). In addition, it was impossible to distinguish different levels and different strains of microorganisms using the solid spotting method. This result is somewhat different from the results obtained by Wang et al. [34]. They treated single colonies of Cronobacter cultured on TSA medium using the ethanol/formic acid method. Although different levels of Cronobacter can be distinguished, this is based on microbial colonies. The data were not uniform, and the thickness of the target plate site was found to be variable, suggesting that it is strongly dependent on the spotting method. These inconsistent results may be due in part to the fact that some microbial colonies formed are small and therefore difficult to spot and detect in the operation of the identification method.

Identification of the Analysis Relationship between the Cell Number of C. sakazakii and Peak Intensities Using Environmental and PIF Samples
The MALDI-TOF MS liquid pretreatment and sample loading method was used to detect C. sakazakii contamination in samples from the production environment and in samples of PIF spiked with the microorganism. Using pre-treatment methods without pre- As shown in Figure 4A, after pretreatment using the liquid spotting technique, the different species levels that characterize Cronobacter spp. can be well distinguished, and the results have good reproducibility. Compared with the cluster analysis data generated from liquid spotting ( Figure 4A), lower reproducibility was observed in a cluster analysis of MALDI-TOF MS data generated using the traditional solid spotting method ( Figure 4B). In addition, it was impossible to distinguish different levels and different strains of microorganisms using the solid spotting method. This result is somewhat different from the results obtained by Wang et al. [34]. They treated single colonies of Cronobacter cultured on TSA medium using the ethanol/formic acid method. Although different levels of Cronobacter can be distinguished, this is based on microbial colonies. The data were not uniform, and the thickness of the target plate site was found to be variable, suggesting that it is strongly dependent on the spotting method. These inconsistent results may be due in part to the fact that some microbial colonies formed are small and therefore difficult to spot and detect in the operation of the identification method.

Identification of the Analysis Relationship between the Cell Number of C. sakazakii and Peak Intensities Using Environmental and PIF Samples
The MALDI-TOF MS liquid pretreatment and sample loading method was used to detect C. sakazakii contamination in samples from the production environment and in samples of PIF spiked with the microorganism. Using pre-treatment methods without pre-enrichment, the detection limit was found to be 1.18 × 10 6 cfu/mL. We then determined whether a pre-enrichment method would improve detection. After pre-enrichment culturing of C. sakazakii in different liquid media, MALDI-TOF MS detection was carried out by the liquid spotting pretreatment method. Among the types of media tested, we focused on the use of LB liquid medium (600 µL) and performed the detection procedure after pre-enrichment for 6 h; in this case, the detection limit reached 4.1 × 10 1 cfu/mL. When TSB and NB liquid media were used for pre-enrichment, the detection limit reached 1.42 × 10 2 cfu/mL after 6 h of pre-enrichment. By analyzing the MALDI-TOF MS spectra generated from samples with different initial numbers of C. sakazakii cells (10 1 -10 8 cfu/mL) after pre-enrichment ( Figure 5A), it was found that at 9476 m/z, there was a linear relationship between the number of bacterial cells; the relationship between relative intensity (y) and the number of bacterial cells (x) was found to be y = 0.0005 x − 0.00009, and the correlation coefficient was 0.9852.  [37]. All three methods require pre-enrichment medium (BPW or mLST) to encourage PIF samples. This is more expensive than the pre-enrichment medium (LB) used in this study. However, the novel MALDI-TOF MS method developed in this study only takes 8 h for the detection of C. sakazakii in PIF (6 h for pre-enrichment + 2 h for pretreatment and detection). The spectrogram acquisition time for each point was about 0.4 min. The time to process the spectra and analyze the whole sample plate (96 spots) was about 10 min. The detection limit was good.
In addition, the newly described operation is simple and more environmentally friendly than the traditional method. This is because there will be less contamination of C. sakazakii that may occur due to the use of inoculation rings to pick single colonies.

Conclusions
In this study, we developed and optimized a liquid spotting pretreatment method for the MALDI-TOF MS-based detection and identification of Cronobacter spp. The optimal parameter conditions were the addition of 25 µL of 70% FA, the application of ultrasound with a power of 350 W and a time of 3 min, and the addition of 75 µL of ACN. Under these conditions, the identification score of C. sakazakii reached 1926.42 ± 48.497. This method was associated with consistently strong identification accuracy, and it is applicable to the identification of multiple strains of C. sakazakii. Compared with the traditional MALDI-TOF MS method, the detection time of this method can be shortened to 8 h. Meanwhile, the method was evaluated at different frequencies and with different samples, and the overall results showed good stability. Thus, we conclude that this new method is superior to the traditional solid spotting method.
This study also reports the first direct identification and detection of C. sakazakii in environmental samples and PIF samples by MALDI-TOF MS. Overall, this method is a useful alternative to traditional MALDI-TOF MS-based pathogen detection and identification methods. Future studies will adapt this method to the detection of mixed samples and samples containing multiple strains of microorganisms.
Supplementary Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/foods12101981/s1, Figure S1: MALDI-TOF MS At the same time, the relative standard deviation values corresponding to the number of different microbial cells were calculated (Table S1). The relative standard deviation values corresponding to the number of different microbial cells were less than 3.29%, which was within the credible range. The results show that the method has good repeatability.
According to a visual inspection of the MALDI-TOF MS profiles ( Figure 5A), LB liquid medium seems to have had a more beneficial pre-enrichment effect on the detection of C. sakazakii in the simulated environment, and the obtained MALDI-TOF MS spectrum shows a linear relationship between the peak intensities and the number of bacterial cells; this relationship, then, may be useful for practical applications. In previous studies, it has been demonstrated that the different ribosomal protein peaks in MALDI-TOF MS spectra can be used to discriminate pathogenic bacterial strains at the same species level with different molecular typing [35]. The present study, however, provided for the first time an analysis of the linear relationship between the bacterial cell number of C. sakazakii and the MALDI-TOF MS peak intensity. This linear relationship suggests the feasibility of the use of MALDI-TOF MS for quantitative as well as qualitative detection of C. sakazakii, which is consistent with the quantitative results obtained by Hsieh et al. regarding the quantitative identification of several other classes of pathogenic bacteria [36]. Specifically, the method described here can be applied to the quantitative detection of C. sakazakii in environmental samples from PIF production and processing factories. Figure S1 shows that there are some differences in MALDI-TOF MS spectra between the contaminated PIF samples and the uncontaminated PIF samples after pretreatment. The number of peaks in uncontaminated PIF samples was less than that in contaminated PIF samples, and the spectrum of uncontaminated PIF samples could not be identified, but the spectrum of contaminated PIF samples could be identified as C. sakazakii.
The results showed that the accuracy of MALDI-TOF MS identification was still guaranteed by washing only PIF samples contaminated with C. sakazakii with PBS and removing the upper layer of fat. Furthermore, other components in PIF do not have a large impact on subsequent MALDI-TOF MS detection. This demonstrated the feasibility of this method for the detection of C. sakazakii in PIF samples by MALDI-TOF MS.
The pre-enrichment method was further used to detect C. sakazakii in actual PIF samples using MALDI-TOF MS, as shown in Figure 5B. After 4 h of pre-enrichment treatment, the detection limit reached 1.9 × 10 4 cfu/mL; after 6 h of pre-enrichment treatment, the detection limit was found to reach 2.72 × 10 3 cfu/mL. At the same time, after analyzing the data of different numbers of bacterial cells under 6 h pre-enrichment, it was found that there was a linear relationship between the number of bacterial cells over the range of 10 3 to 10 8 cfu/mL and the intensity of the characteristic MS peak at 9476 m/z. Here, the relationship was described by the equation y = 0.0009x + 0.0016, and the correlation coefficient was 0.9693.
To further investigate the stability of the method, we calculated the relative standard deviation of the results obtained after multiple tests, and the results are shown in Table S2. When the method was used to detect three different artificially contaminated PIF samples on the market (the detection frequency was five days), the RSD values were all less than 1.78%, indicating that the method had good stability.
Javůrková et al. compared the timeliness of three methods of Cronobacter detection (ISO), immunochromatographic test (ICT), and traditional MALDI-TOF MS methods: (1) The traditional ISO method takes 140 h for the detection of Cronobacter in PIF samples; (2) ICT methods take 24 h; and (3) the traditional MALDI-TOF MS method takes 46 h [37]. All three methods require pre-enrichment medium (BPW or mLST) to encourage PIF samples. This is more expensive than the pre-enrichment medium (LB) used in this study. However, the novel MALDI-TOF MS method developed in this study only takes 8 h for the detection of C. sakazakii in PIF (6 h for pre-enrichment + 2 h for pretreatment and detection). The spectrogram acquisition time for each point was about 0.4 min. The time to process the spectra and analyze the whole sample plate (96 spots) was about 10 min. The detection limit was good.
In addition, the newly described operation is simple and more environmentally friendly than the traditional method. This is because there will be less contamination of C. sakazakii that may occur due to the use of inoculation rings to pick single colonies.

Conclusions
In this study, we developed and optimized a liquid spotting pretreatment method for the MALDI-TOF MS-based detection and identification of Cronobacter spp. The optimal parameter conditions were the addition of 25 µL of 70% FA, the application of ultrasound with a power of 350 W and a time of 3 min, and the addition of 75 µL of ACN. Under these conditions, the identification score of C. sakazakii reached 1926.42 ± 48.497. This method was associated with consistently strong identification accuracy, and it is applicable to the identification of multiple strains of C. sakazakii. Compared with the traditional MALDI-TOF MS method, the detection time of this method can be shortened to 8 h. Meanwhile, the method was evaluated at different frequencies and with different samples, and the overall results showed good stability. Thus, we conclude that this new method is superior to the traditional solid spotting method.
This study also reports the first direct identification and detection of C. sakazakii in environmental samples and PIF samples by MALDI-TOF MS. Overall, this method is a useful alternative to traditional MALDI-TOF MS-based pathogen detection and identification methods. Future studies will adapt this method to the detection of mixed samples and samples containing multiple strains of microorganisms.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/foods12101981/s1, Figure S1: MALDI-TOF MS spectra of uncontaminated and contaminated PIF samples after pretreatment; Table S1: Relative standard deviations after pretreatment of different artificially contaminated environment sample; Table S2: Relative standard deviations after pretreatment of different artificially contaminated PIF samples in the market.