The WT C. sakazakii G4023, a food-isolated strain, has been proved to belong to sequence type 4 (ST4), and the pathogenicity has also been confirmed in our previous work (Wang M. et al., 2019). In this study, we compared the whole-genome sequence (WGS) of C. sakazakii G4023 with that of ATCC 29544. ATCC 29544 was clinically isolated and identified as a type strain of C. sakazakii (Farmer et al., 1980), and detailed information of both strains are summarized in Supplementary Table 1. The WGS comparisons showed there were 3,648 homologous genes, which account for 90.9 and 83.3% of total genes in G4023 and ATCC 29544, respectively (Figure 1A). Further analyses suggested the specific genes in G4023 were found mainly localized in genomic islands (Figure 1B), which were normally horizontally acquired to provided accessory genes for niche adaptation and pathogenicity (Schmidt and Hensel, 2004). In addition, whole-genome average nucleotide identity (ANI), average amino acid identity (AAI), and tetranucleotide usage deviation (TUD) between C. sakazakii G4023 and ATCC 29544 also demonstrated they belong to the same bacterial species (97.86%, 98.77%, and 0.999, respectively).

In addition to the genomic comparisons, biochemical identifications of C. sakazakii G4023 were also performed, with ATCC 29544 strain as a control. Both the typical features of C. sakazakii, green/blue colonies on selective Druggan Forsythe Iversen (DFI) agar and yellow pigment production on Luria–Bertani (LB) agar (Cawthorn et al., 2008), were observed in G4023 and ATCC 29544 bacteria; EHEC O157: H7 EDL933 and Salmonella Typhimurium ATCC 14028 bacteria presented white colonies on both DFI agar and LB agar (Figure 1C and Table 1). Meanwhile, a commercial kit for C. sakazakii identification was applied to G4023 bacteria, with ATCC 29544 as a control. Both of the strains reacted exactly like the typical C. sakazakii according to instruction (Table 1 and Supplementary Figure 1), which further proved that G4023 belongs to C. sakazakii.

Growth curves of C. sakazakii G4023 and ATCC 29544 strains cultured in LB medium exhibited no significant differences [approximately 2 × 10⁵ colony forming units (CFU) were initially inoculated; Figure 2A]. To measure the desiccation tolerance of C. sakazakii G4023 strain, we desiccated the bacteria for up to 50 days and set C. sakazakii ATCC 29544 as a control. We dispensed the bacteria into individual wells of 12-well microtiter plates separately for desiccation treatment. Every 5 days, the viable count (CFU/ml) of each strain was documented to assess the desiccation tolerance ability. As shown in Figure 2B, viable counts of both G4023 and ATCC 29544 were sharply decreased during the first 5-day interval and tended to be stable after then. These results indicated that G4023 and ATCC 29544 strains might have adapted to desiccation during the first 5 days, and G4023 has better resistance to desiccation than ATCC 29544. Due to the strong desiccation tolerance and pathogenicity of G4023 (Wang M. et al., 2019), it could be a potential threat to infants. Therefore, we performed transcriptomic and proteomic analyses of desiccated C. sakazakii G4023 to provide insights into the desiccation adaption mechanisms.

As suggested above, C. sakazakii G4023 bacteria have adapted to desiccation during the first 5 days, and similar desiccation treatment for 4 days has been reported earlier; so samples used for RNA-seq and quantitative proteomic analyses were desiccated for 4 days as previously described (Shaker et al., 2008; Chaibenjawong and Foster, 2011). Bacteria cultured in LB medium were collected as the control group. High-throughput Illumina RNA-seq system (Novaseq 6000) was used to systematically catalogue the transcripts. After filtering, a total of 18,914,098 and 20,738,460 clean reads were obtained for the desiccated and control sets of C. sakazakii G4023, respectively. Approximately 98.11% of the total reads for desiccated C. sakazakii G4023 and 97.64% of those for control set were uniquely mapped to the respective reference genome. We observed 1,573 genes to be differentially expressed (≥2-fold change), out of which 609 were upregulated and 964 were downregulated (Figure 3A). Detailed differential expression profiles are shown in Supplementary Table 2.

The RNA-seq results were further confirmed by quantitative real-time PCR (qRT-PCR) analysis. The housekeeping gene glnS was used as the internal control gene as its expression did not change significantly under desiccation (Supplementary Figure 2). Ten genes were selected randomly to detect the fold-changes under desiccation. Although the fold-change values were not exactly alike, the regulation trends agreed with the RNA-seq results (Figure 3B), indicating the validity of the RNA-seq data.

The differentially expressed genes (DEGs) in the transcriptome were assigned annotations to three primary gene ontology (GO) domains: biological process (BP), cellular component (CC), and molecular function (MF). A total of 1,140 DEGs were successfully annotated against the GO database, of which 77 GO terms were significantly enriched (36 terms assigned to BP, 19 to CC, and 22 to MF, listed in Supplementary Table 2), and the most enriched 30 GO terms are shown in Figure 4. Genes involved with gene expression, macromolecule biosynthetic process, intracellular components, and cofactor binding were observed the most significantly differentially expressed in desiccated C. sakazakii G4023.

The DEGs were also subjected to Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis. Only three pathways, namely, citrate cycle, oxidative phosphorylation, and carbon metabolism, were significantly enriched out of the 84 enriched ones (Supplementary Table 2), which suggested that the basal metabolisms were the most challenged in desiccated C. sakazakii G4023.

Label-free quantitative proteomic analyses were simultaneously performed to investigate the regulations that happened at the protein level of desiccated C. sakazakii G4023, with LB-cultured bacteria set as the control group as well. Proteomics analysis showed that a total of 209,357 spectra were matched to 18,763 peptides, and 2,225 proteins were identified, out of which 1,754 were quantifiable. After data filtering (≥1.5-fold change), 360 proteins were observed upregulated and 322 were downregulated (Figure 5A). All the differentially expressed proteins (DEPs) are listed in Supplementary Table 3.

In order to obtain a better understanding of DEPs in desiccated C. sakazakii G4023, we performed the GO annotation and functional classification, subcellular location classification, and functional classification of clusters of orthologous groups of proteins (COG). Five hundred forty-seven identified DEPs (80.20% of the total) were annotated to GO terms (details shown in Supplementary Figure 3), and 30 hierarchically structured GO classes were significantly enriched (Figure 5B and Supplementary Table 3). According to the GO enrichment results, proteins related to proteolysis regulation, pyrimidine biosynthesis, propionate metabolic process, and acid-ammonia (or amide) ligase activity achieved the greatest fold enrichment. Subcellular location analysis of the 682 DEPs revealed that most of them (63%) are located in the cytoplasm, whereas outer membranal DEPs and extracellular DEPs were only 3.37 and 2.49%, respectively (Supplementary Figure 3). COG analyses showed that except for the functionally unknown proteins, DEPs were mostly categorized into translation, ribosomal structure and biogenesis (70/74 proteins were upregulated), cell wall/membrane/envelope biogenesis (50/73 proteins were downregulated), and amino acid transport and metabolism (49/71 proteins were upregulated) (Figure 5C and Supplementary Table 4). These indicated that a large abundance of proteins was needed by C. sakazakii G4023 to cope with desiccation, which was also revealed by Hu et al. (2017), and transport and metabolism of amino acids might contribute to the G4023 bacterial survival under desiccation stress.

Enrichment of KEGG pathways and protein domains of DEPs were performed as well. Three hundred sixty-three DEPs were allocated to the reference pathways in KEGG, and seven pathways were significantly enriched (p ≤ 0.05; Figure 5D and Supplementary Table 3). Three of the seven (42.86%) were related to biosynthesis and metabolism of amino acids (arginine, alanine, aspartate, glutamate, taurine, and hypotaurine). Correspondingly, amino acid biosynthesis and metabolism-related protein domains were also significantly enriched (Figure 5E and Supplementary Table 3).

Based on the transcriptome and proteome of desiccated C. sakazakii G4023, we performed an integrated analysis of these two omics to promote a better understanding of bacterial responses to desiccation. Overall, 331 genes were found to be regulated at both transcriptional level (≥2.0-fold) and protein levels (≥1.5-fold) under desiccation stress (Figure 6A). COG analysis of these 331 genes suggested that 325 genes were classified into 18 COG categories (Figure 6B, left). Besides those whose functions are unknown, genes were mostly involved in translation (44 genes), amino acid metabolism and transport (39 genes), energy production and conversion (30 genes), cell wall/membrane/envelope biogenesis (27 genes), and carbohydrate metabolism and transport (27 genes) categories, which were highly consistent with COG categorizing of DEPs. Further analysis suggested that 191 out of the 331 genes were regulated in matching tendencies, which were upregulated or downregulated at both transcriptional and protein levels (87 up/upregulated and 104 down/downregulated; Supplementary Table 5), and the fold changes are shown in the heatmap (Supplementary Figure 4). The COG classifications showed the up/upregulated group was dominantly sub-categorized into the translation category (Figure 6B, middle), and the down/downregulated group was mostly sub-categorized into the cell wall/membrane/envelope biogenesis.

In desiccated C. sakazakii G4023, protein levels of CheA, CheW, CheY, CheB, and CheR were upregulated 1. 78-, 1. 25-, 2. 00-, 3. 22-, and 1.23-fold, respectively, while the transcriptional expressions of the encoding genes (cheA, cheW, cheB, and cheR) were decreased (2. 22-, 3. 37-, 3. 54-, and 4.25-fold, respectively), and expression of gene cheY was not detected in transcriptome (Table 2). CheA is a histidine kinase and plays a central role in signal transduction of bacterial chemotaxis. The coupling protein CheW, cognate response regulator CheY, methyltransferase CheR, and methylesterase CheB are also involved (Kehry et al., 1984; Stephens et al., 2006; Muok et al., 2020).

To further investigate what role che system plays in desiccated C. sakazakii G4023, we constructed cheA, cheW, cheY, cheB, and cheR mutant strains of G4023, and their growth rates exhibited no significant differences in vitro (Figure 7A). Desiccation tolerance tests were then performed. Four-day desiccation results suggested that the G4023ΔcheR strain exhibited stronger desiccation tolerance than C. sakazakii G4023, while G4023ΔcheA and G4023ΔcheW showed significantly decreased tolerance (Figure 7B). The 50-day desiccation tests showed that G4023ΔcheA, G4023ΔcheW, and G4023ΔcheB exhibited significantly decreased resistance to desiccation compared with that of C. sakazakii G4023, especially G4023ΔcheA (Figure 7C). These results together with the upregulations in CheAWYBR protein levels suggested that the chemotaxis signaling cascade, especially the core protein CheA, might be important in the desiccation survival of C. sakazakii G4023.

It has been reported that flagellum synthesis can be regulated by a chemotaxis signal transduction pathway to optimize swim cell–swarm cell differentiation in response to changing environmental conditions in Rhodospirillum centenum (Berleman and Bauer, 2005). Since the chemotaxis system relevant proteins were upregulated in desiccated C. sakazakii G4023, we wondered whether the flagellar synthesis was affected as well. Flagellar biosynthesis genes are organized in a three-tiered cascade allowing for regulated expression (Romilly et al., 2020). The products of class I operon, FlhD and FlhC, are regulators positively controlling the expression of class II operons. However, the transcriptional expressions of genes flhD and flhC showed no significant changes (0.89- and 1.01-fold), and regulations at protein levels were not determined as only FlhD was detectable in the desiccated sample, which could result from the fact that the protein amount is not enough to reach the detection threshold of label-free quantitative proteome analysis.

Although expression regulations of class I operon were not suggested in the omics analyses, FliA and FlgM encoded by class II genes were found upregulated 1.86- and 2.59-fold, respectively at protein level; transcriptional expressions of fliA and flgM genes were downregulated 4.43- and 1.12-fold, respectively. FliA is a sigma factor σ²⁸, whereas FlgM is an anti-sigma factor which binds to FliA to prevent its association with RNA polymerase core enzyme (Hughes and Mathee, 1998). The FliA–FlgM regulatory system is involved not only in the expression of class III but also in that of class II operons, and the relative concentration of FliA and FlgM may play an important role in the production of flagella (Kutsukake and Iino, 1994). In addition to these two regulators, class II genes also encode basal body and hook components (Kapatral et al., 1996). Unlike the upregulations of FliA and FlgM, the hook-associated proteins FlgK, FlgL, FliC, and FliD were all found downregulated over 2-fold at both transcriptional and protein levels (Table 2). The downregulation of basal body and hook components of flagella suggested that flagellum demand in C. sakazakii G4023 was decreased in desiccation, and energies could be saved for better survival.

Class III genes regulated by FliA–FlgM regulatory system encode flagellar motor switch proteins and flagellin, such as FliG, FliM, and FliN, which were found increased 1. 44-, 2. 03-, and 2.86-fold, respectively, at protein level (Table 2). It has been reported that FliG, FliM, and FliN interact with one another to form a switch complex, which in turn is thought to interact with chemotaxis protein CheY, to induce clockwise rotation of the flagellar motors and tumbling of the cell (Toker and Macnab, 1997). From these results, we guessed that when C. sakazakii G4023 bacteria encounter the desiccation, flagellum numbers needed by each cell is decreased, but the flagellar motors and tumbling ability of bacteria are needed to be increased to favor survival in desiccation stress.

LPS, which is composed of lipid A, a core oligosaccharide, and O-antigen repeats (Liu et al., 2019), anchors in the outer membrane of gram-negative bacteria and serves as an effective barrier between the bacterium and its environment (Sun et al., 2011). It has been reported that cleavage of LPS offered C. sakazakii bacteria higher resistance to the hostile environment as it caused outer membrane defects and increased biofilm formation (Wang L. et al., 2012).

In this study, we focused on the O-antigen, which is generally not necessary for the survival of the microorganism (Gronow and Brade, 2001), to investigate its role in adaptation to desiccation stress of C. sakazakii G4023. The O-antigen biosynthesis genes in the genome of C. sakazakii are normally located between galF and gnd (Sun et al., 2011), and the transcriptional expressions of these genes were found downregulated in desiccated C. sakazakii G4023 (Table 2). However, most of the proteins encoded by these genes were not able to be quantified at the protein level; this may be due to their low abundance as they were mostly located on cell membranes (Okuda et al., 2016). We tried to mutate the O-antigen biosynthesis essential gene, wzy or wzx (Wang et al., 2009), in C. sakazakii G4023 but failed; we then turned to two other important genes for O-antigen biosynthesis, wzzB and fepE, which are responsible for the biosynthesis of long-chain and very-long-chain O-antigens, respectively (Murray et al., 2003). O-antigen chain length has been reported to affect survival and infection procedures in several bacteria (Murray et al., 2003; Kintz and Goldberg, 2011; da Silva et al., 2018). The expression of wzzB in this study was found decreased 1.70- and 1.72-fold at transcriptional and protein levels, respectively; while fepE was found downregulated 2.67-fold at transcriptional level but upregulated 1.74-fold at protein level.

To further investigate whether FepE and WzzB played roles in desiccation tolerance of C. sakazakii G4023, we constructed the fepE and wzzB mutant strains (G4023ΔfepE and G4023ΔwzzB). Both mutant strains showed no growth defects in vitro (Figure 8A), and the desiccation tolerance tests were performed. Results of a 4-day term of desiccation showed that both G4023ΔfepE and G4023ΔwzzB exhibited better survival than G4023 (Figure 8B), whereas the survival advantages faded with desiccation time, especially G4023ΔfepE. By the 50th day of desiccation, bacterial recovery counts of G4023ΔfepE were significantly less than those of G4023 (Figure 8C). These results indicated that shorter O-antigen chain length biosynthesis aids in the survival of C. sakazakii G4023 in the short-term rather than the long-term desiccation.

Moreover, crystal violet staining assays were also performed to quantify the biofilm formed by G4023, G4023ΔfepE, and G4023ΔwzzB. There was no significant difference in biofilm formation between G4023 and the mutant strains when they were cultured in LB. However, after desiccation treatment for 4 days, all of them exhibited significant increases in biofilm formation. Notably, biofilm formed by desiccated G4023ΔfepE and G4023ΔwzzB was significantly more than that in desiccated G4023 (Figure 8D). These results indicated that C. sakazakii G4023 increased its biofilm formation to overcome desiccation, and shortening of O-antigen chain length could further facilitate the biofilm formation yet only happened when bacteria were suffering desiccation.

COG and KEGG enrichment analyses of DEPs, as well as COG classifications of genes regulated at both transcriptional and protein levels, suggested that proteins related to transport and metabolism of amino acids were significantly regulated in desiccated C. sakazakii G4023. Detailed analyses revealed that proteins involved in arginine biosynthesis, such as ArgABCDEFGH, CarAB, and the regulator ArgR (Caldara et al., 2008), were upregulated; furthermore, ArtI and ArtJ involved in arginine ABC transporter (Bai et al., 2021) were also upregulated 1.7- and 2.22-fold at protein level, respectively (Table 2). These results indicated that biosynthesis and transport of arginine may contribute to the survival of C. sakazakii G4023 in desiccation, but whether and how the arginine functions required further investigations.

During the analyses, we found the transcriptional expressions of arginine biosynthesis genes (argABCDEFGH) were downregulated (Table 2). As it has been reported that productions of all enzymes involved in arginine biosynthesis are subjected to repression by arginine, mediated by the repressor ArgR (Vyas and Maas, 1963), these downregulations could have resulted from the upregulated protein level of ArgR.

Iron/sulfur cluster is involved in all kingdoms of life to be catalysts or redox sensors in diverse BPs and is formed by Fe/S biogenesis systems NIF, ISC, and SUF (Yang et al., 2006; Roche et al., 2013), of which ISC and SUF systems allow the maturation of all Fe/S proteins in cells, while NIF is dedicated to the maturation of nitrogenase (Ayala-Castro et al., 2008). Genes encoding ISC and SUF are present in the genome of C. sakazakii G4024, while those for NIF are absent. The operon iscSUA-hscBA, encoding the ISC system (Roche et al., 2013), were observed upregulated 9. 2-, 6. 09-, 6. 40-, 7. 27-, and 5.21-fold, respectively, at the transcriptional level in desiccated C. sakazakii G4024. Meanwhile, protein amounts of IscS, IscU, and HscA were also increased 2. 31-, 2. 06-, and 1.96-fold, respectively; whereas IscA and HscB were not detectable in proteome (Table 2). Additionally, genes encoding the two sub-complexes, SufBCD and SufSE, of SUF system were found transcriptionally regulated (sufABCDS were increased 3. 59-, 4. 43-, 2. 71-, 3. 72-, and 2.83-fold, respectively, while sufE was decreased 1.22-fold). Besides, the protein levels of SufABCDSE were 0. 50-, 0. 76-, 0. 53-, 0. 82-, 0. 68-, and 1.18-fold, respectively, in desiccated C. sakazakii G4023 compared with that in LB-cultured (Table 2). Considering the ISC systems were upregulated at both transcriptional and protein levels and the SUF system was increased at the transcriptional level but decreased at the protein level, we speculated that when C. sakazakii G4023 suffered desiccation, the bacteria would prefer the ISC system instead of the SUF to synthesize F/S cluster to cope with the stress. However, these surmises need to be confirmed with further experimental investigations.

All the strains, plasmids, and primers used in this study are listed in Supplementary Table 6. The C. sakazakii G4023 strain is a food-isolated strain kindly offered by Beijing Entry--Exit Inspection and Quarantine Bureau and belongs to ST4. The ST type of G4023 was defined by the Cronobacter spp. multilocus sequence typing (MLST) Database.¹ Genomic comparisons between G4023 and ATCC 29544 (ST8) were performed to identify G4023 as a C. sakazakii strain. The whole genome comparison was performed by local BLASTp (v2.3.0)² with an E-value cutoff of <1e−10, identity >80%, and coverage >80%, and genomic islands were identified by software IslandViewer 4. Then the results were visualized by software BLAST Ring Image Generator (BRIG). The ANI and TUD were calculated via online server JspeciesWS³ (Richter et al., 2016); The AAI was calculated by software toolkit CompareM.⁴ Overnight LB-cultured bacteria of C. sakazakii G4023, C. sakazakii ATCC 29544, EHEC O157: H7 EDL933, and Salmonella Typhimurium ATCC 14028 were steak cultured on the selective DFI agar and the LB agar; the plates were then statically cultured in 37°C overnight to observe the phenotypes of colonies. Meanwhile, biochemical identifications of G4023 and ATCC 29544 were performed with commercial biochemical identification kit [Biochemical identification strip of E. sakazakii (GB), Hopebio, Qingdao, China], and the positive or negative responses were judged by the color changes of mediums according to the manufacturer’s instructions.

Bacteria in this study were grown at 37°C in LB medium under standard bacterial growth conditions unless otherwise specified. For growth curves, three replicates per strain were grown in 200 μl LB medium in a 96-well plate and inoculated aerobically in the ratio of 1:1,000 from overnight cultures at 37°C. The initial inoculum was approximately 2 × 10⁵ CFU and monitored by measuring the OD₆₀₀ value using a Multiskan GO 1510 plate reader (Thermo Fisher Scientific, Vantaa, FIN) every 10 min.

Mutant strains were generated using the λ Red recombination system described previously (Datta et al., 2006) with the plasmid pSIM17 generated by Chan et al. (2007). The plasmid pSIM17 is a temperature-sensitive, low-copy plasmid and contains blasticidin resistance. In brief, we first electroporated pSIM17 into C. sakazakii G4023 bacteria and selected on blasticidin-containing (50 μg/ml) LB agars at 30°C. Then the kanamycin cassette used to replace the target genes of C. sakazakii G4023 was amplified from pKD4 with primers listed in Supplementary Table 6. Each primer contains a 5′ end homologous to the flanking regions of the target gene and a 3′ end that primes the kanamycin cassette for PCR amplification. The amplified PCR products were purified with SanPrep Column DNA Gel Extraction Kit (Sangon Biotech, China) and electroporated into pSIM17-containing C. sakazakii G4023 strain, which had been induced for red expression by growth at 42°C for 15 min. After electroporation, 1 ml LB was added, and the bacteria were incubated at 37°C with shaking for 3–4 h. Then the bacteria were cultured on LB agar containing 50 μg/ml of kanamycin to select drug resistant recombinants at 37°C. Several colonies on each plate were picked on the following day and amplified with mutant identification primers, after which the amplified products were sequenced to verify the replacement of target genes with kanamycin cassette (Njoroge et al., 2012). The verified mutant colonies were re-cultured in LB medium supplemented with kanamycin (50 μg/ml) at 37°C and preserved in glycerin at −80°C until use.

Strains used for desiccation tolerance test were grown overnight and transferred to fresh LB medium at a ratio of 1:100 for 4 h until mid-exponential phase, and 50-μl aliquots were dispensed into individual wells of a 12-well microtiter plate, and three replicates were set for each strain. Thereafter, the bacteria were desiccated as previously described (Shaker et al., 2008; Hu et al., 2017). In brief, the plates were incubated statically in a 37°C incubator without a lid for 4 h for drying. Dehydrated silica gel was placed in the incubator. After drying, the plate was covered with a lid and kept at 30°C statically for 4 days. After desiccation, the cells were rehydrated with 1 ml sterile phosphate buffered saline (PBS, pH 7.4), and 100-μl aliquots were re-cultured in LB agar plates at 37°C for 24 h to count the cell population (CFU/ml).

RNA samples from desiccated C. sakazakii G4023 were prepared using TRIzol^® LS Reagent (Invitrogen, Carlsbad, CA, United States) and treated with RNase-Free DNase I (Fermentas, Burlington, Canada) to eliminate genomic DNA contamination. C. sakazakii G4023 cultured in LB medium was set as control. Purified RNA samples were stored at -80°C till further use. The library preparation and RNA-seq were carried out commercially (Novogene, Tianjin, China). In brief, a total of 3 μg RNA per sample was used as input material for the RNA sample preparations. Sequencing libraries were generated using NEBNext^® Ultra™ Directional RNA Library Prep Kit for Illumina^® (NEB, United States) following manufacturer’s recommendations. In order to select cDNA fragments of preferentially 150–200 bp in length, the library fragments were purified with AMPure XP system (Beckman Coulter, Beverly, United States). Then 3 μl USER Enzyme (NEB, United States) was used with size-selected, adaptor-ligated cDNA at 37°C for 15 min followed by 5 min at 95°C before PCR. Then PCR was performed with Phusion High-Fidelity DNA polymerase, Universal PCR primers and Index (X) Primer. At last, products were purified (AMPure XP system), and library quality was assessed on the Agilent Bioanalyzer 2100 system. The clustering of the index-coded samples was performed on a cBot Cluster Generation System using TruSeq PE Cluster Kit v3-cBot-HS (Illumia) according to the manufacturer’s instructions. After cluster generation, the library preparations were sequenced on the NovaSeq 6000 (Illumina), and paired-end reads were generated.

For proteomic analysis, desiccated C. sakazakii G4023 cells were harvested and washed with ice-cold PBS five times. Cells were stored in liquid nitrogen overnight and then stored at −80°C until required. Samples were sonicated three times on ice using a high-intensity ultrasonic processor (Scientz) in lysis buffer (8 M urea, 1% Protease Inhibitor Cocktail). The remaining debris was removed by centrifugation at 12,000 g at 4°C for 10 min. Finally, the supernatant was collected, and the protein concentration was determined with BCA Protein Assay Kit (abcam, Cambridge, England) according to the manufacturer’s instructions. For digestion, the protein solution was treated as previously described (Ondrej et al., 2020). Accordingly, the protein solution was reduced with 5 mM dithiothreitol for 30 min at 56°C and alkylated with 11 mM iodoacetamide for 15 min at room temperature in darkness. The protein sample was then diluted by adding 100 mM triethylammonium bicarbonate (TEAB) to urea concentration less than 2 M. Finally, trypsin was added at 1:50 trypsin-to-protein mass ratio for the first digestion overnight and 1:100 trypsin-to-protein mass ratio for a second 4-h digestion. C. sakazakii G4023 cultured in LB medium was set as control. Three biological replicates were prepared. The tryptic peptides were then submitted to conduct the label-free quantitative mass spectrometry analyses commercially (PTM BIO, Hangzhou, China); high-performance liquid chromatography (HPLC) fractionation and liquid chromatographic-tandem mass spectrometry (LC-MS/MS) analyses were performed as previously described (Ondrej et al., 2020). Briefly, the tryptic peptides were fractionated into fractions by high pH reverse-phase HPLC using Thermo Betasil C18 column [5 μm particles, 10 mm inner diameter (ID), 250 mm length]. Peptides were first separated with a gradient of 8–32% acetonitrile (pH 9.0) over 60 min into 60 fractions. Then, the peptides were combined into six fractions and dried by vacuum centrifuging. The tryptic peptides were then dissolved in 0.1% formic acid (solvent A), directly loaded onto a home-made reversed-phase analytical column (15-cm length, 75-μm ID). The gradient comprised an increase from 6 to 23% solvent B (0.1% formic acid in 98% acetonitrile) over 26 min, 23–35% in 8 min, climbing to 80% in 3 min, and then holding at 80% for the last 3 min, all at a constant flow rate of 400 nl/min on an EASY-nLC 1000 ultra-performance liquid chromatography (UPLC) system. The peptides were subjected to nanospray ionization (NSI) source followed by MS/MS in Q ExactiveTM Plus (Thermo Fisher Scientific) coupled online to the UPLC. The electrospray voltage applied was 2.0 kV. The m/z scan range was 350–1,800 for full scan, and intact peptides were detected in the Orbitrap at a resolution of 70,000. Peptides were then selected for MS/MS using normalized collision energies (NCE) setting of 28, and the fragments were detected in the Orbitrap at a resolution of 17,500. A data-dependent procedure alternated between one MS scan and 20 MS/MS scans with 15.0 s dynamic exclusion. Automatic gain control (AGC) was set at 5E4. Fixed first mass was set at 100 m/z. Finally, the resulting MS/MS spectrums were searched against UniProt database Cronobacter_sakazakii_strain_ATCC_BAA-894_290339_PR_20191018.

RNA-seq data were validated using qRT-PCR. RNA purified from the desiccated and control C. sakazakii G4023 cells was transcribed to cDNA using the PrimeScript 1st Strand cDNA Synthesis Kit (Takara, Japan). qRT-PCR analysis was performed in an Applied Biosystems ABI 7300 Real-Time PCR System. Each qRT-PCR reaction was carried out in a total volume of 20 μl in a 96-well optical reaction plate (Applied Biosystems, Foster, CA, United States) containing 10 μl Power SYBR Green PCR Master Mix (Applied Biosystems, Foster, CA, United States), 1 μl cDNA, and two gene-specific primers with a final concentration of 0.3 mM each. The housekeeping gene glnS was used as a reference control for normalization as its transcriptional expression levels did not change dramatically when C. sakazakii G4023 bacteria suffered desiccation (Supplementary Figure 2). Relative difference in gene expression was calculated as the fold change using the 2^–ΔΔCt method. At least three biological replicates were performed for each qRT-PCR analysis.

Crystal violet staining assay was performed to assess the formation of biofilm as described previously (Fernández-Gómez et al., 2020). In brief, 100-μl overnight strains were cultured in 96-well polystyrene microtiter plates and subjected to desiccation. On the third day of desiccation, the 96-well plates were taken out, and overnight-cultured strains were diluted 1:100 in fresh LB medium in the remaining blank wells. The plate was then statically cultured at 30°C for 24 h. The desiccated and LB cultured bacteria were stained with 0.5% crystal violet for 5 min after washing with double-distilled water. Next, 200 μl of 95% ethanol was added after removing the unbound crystal violet to dissolve the biofilm-bound dye, and the ability of biofilm formation was determined by measuring the absorbance at 590 nm with an enzyme-linked immunosorbent assay plate reader. Three wells of each strain were studied in parallel, and each experiment was independently performed thrice.

Desiccation tolerance test data are presented as the mean ± standard deviation (SD) of at least three trials, and the differences between two groups were evaluated in the independent samples by two-tailed Student’s t-test. A p-value < 0.05 was considered statistically significant. The asterisks indicate significant differences (*p < 0.05; ^**p < 0.01; ^***p < 0.001).