Curing of the cryptic plasmids by using the incompatible nature of the plasmid

To investigate the impact of the curing of cryptic plasmids on the EcN strain, we explored the differences at the transcriptional expression level. To obtain an EcN strain with the cryptic plasmids curing, we constructed two recombinant suicide plasmids which contained the same origin of replication (ori) with the cryptic plamids. These plasmids were sequentially transformed into EcN to cure the cryptic plasmids based on plasmid incompatibility, and then the sucrose was added to trigger their self-destruction (Fig. 1a). The recombinant suicide plasmids contains the sacB gene encoding levansucrase. When the EcN strain harboring the recombinant suicide plasmids is cultivated on plates containing 10% sucrose, levan can be produced by the hydrolysis of sucrose through levansucrase, and accumulates in the periplasmic space of EcN, hindering bacterial growth, thus EcNc strains that do not harbor any plasmids could be screened (Ying et al. 2016). Subsequently, the curing of cryptic plasmids was confirmed using PCR with specific primers designed based on previous studies (Blum-Oehler et al. 2003). The results showed bands of 429 and 361 bp in size for EcN, while absent in EcNc, indicating the successful curing of pMUT1 and pMUT2 in EcNc strain (Fig. 1b).

Fig. 1

Cryptic plasmid curing. a The schematic diagram of the cryptic plasmid curing process. b The curing of the cryptic plasmids was identified by PCR. PCR was performed using two pairs of primers (muta5/6-F, muta5/6-R, muta7/8-F, muta7/8-R) that were listed in Additional file 2: Table S1

Identification of EcNc strain

At present, the identification of the EcN strain is commonly performed using three pairs of specific primers that target the combination of two cryptic plasmids (Blum-Oehler et al. 2003). However, these three primer sets fail to identify EcNc strains. Additionally, the identification based on 16S rRNA can only provide basic identification at the species level. The EcNc strain is a non-antibiotic resistant strain of E. coli. Therefore, there is a need for a more convenient, rapid, and accurate identification method to identify EcNc.

Three pairs of primers were designed based on the EcN genome for the identification of EcN and EcNc strains through multiplex PCR. Initially, 115 unique genes were identified in EcN, which not existed in the BL21 and DH5α genomes. Subsequently, two genes, ipuB and tcpC, were screened from these 115 genes. The product of ipuB is tyrosine-type DNA invertase, while the product of tcpC is NAD (+) hydrolase. These two genes were selected because the number of E. coli strains harboring both them is the smallest (only 56, Fig. 2a, Additional file 2: Table S3). Consequently, primers were designed for the genes, ipuB and tcpC, and the length of PCR products was 293 bp and 464 bp, respectively. Additionally, a pair of primers were designed for the housekeeping gene gapA in E. coli, producing a PCR product of 629 bp. The results of PCR revealed that the bands were appeared at 293 bp, 464 bp, and 629 bp for both EcN and EcNc, whereas only the 629 bp band of the housekeeping gene was exhibited for BL21 and DH5α (Fig. 2b). This suggested that the designed primers can be used for the specific identification of EcN and EcNc strains. To further differentiate between EcN and EcNc, one pair of primers targeting a cryptic plasmid was selected (muta5/6-F and mata5/6-R). The PCR amplification exhibited a 361 bp band of the cryptic plasmid for EcN, and no band exhibited for EcNc strain (Fig. 2c), showing the ability to distinguish between EcN and EcNc.

Fig. 2

Identification of primer design for EcN genome. a Venn diagram of blast results for ipuB and tcpC. The blast results for ipuB revealed similarities in only 184 strains, while the blast results for tcpC showed similarities in only 160 strains (as of September 9, 2023). Of these, 56 strains were found to overlap between the 184 strains with ipuB similarity and the 160 strains with tcpC similarity. b Primer verification. PCR amplification was performed on the BL21, DH5α, EcN, and EcNc strains using three pairs of designed primers. c Identification of EcN and EcNc strains. Incorporating a pair of primers (muta5/6-F and mata5/6-R) designed for identifying cryptic plasmids into the three primer pairs, PCR amplification was performed to further distinguish between EcN and EcNc

Transcriptome analysis and qPCR verification of EcN and EcNc

Comparison of the transcriptome data between EcN and EcNc revealed some impact of cryptic plasmids curing on the genome transcription, affecting a small subset of genes—10 in total, with 6 downregulated and 4 upregulated (Fig. 3a). Gene Ontology (GO) enrichment analysis of these differentially expressed genes showed a predominant enrichment in the molecular function, mainly in amino acid metabolism, such as carboxy-lyase activity, carbon–carbon lyase activity, tartronate-semialdehyde synthase activity, lysine:cadaverine antiporter activity, and lysine decarboxylase activity (Fig. 3b).

Fig. 3

Transcriptome data analysis and qPCR verification of EcN and EcNc. a Differential expressed gene volcano map on the genome. b GO enrichment analysis of differential expressed genes on the genome. BP biological process, CC cellular component, MF molecular function. c Statistical visualization of fluC reads. The reads were normalized as Fragments Per Kilobase per Million (FPKM). d Four genes were verified by qPCR. Gcl, fadA, fluC, and yfjQ encoded glyoxylate carboligase, 3-ketoacyl-CoA thiolase, self recognizing antigen 43 (Ag43) autotransporter, and DUF932 domain-containing protein, respectively

The gene (flu/agn43) encoding the Ag43/Cah family adhesin exhibited the most significant differential expression (Fig. 3c). Ag43 is widely distributed in E. coli strains and typically encoded by multiple gene copies within a single strain (Roche et al. 2001). The EcN genome harbors three copies of the Ag43 gene (named fluA, fluB, fluC), expressing three Ag43 variants (named Ag43a, Ag43b, Ag43c), with the lengths of 1040 aa, 1039 aa, and 948 aa, respectively. Recent research has classified Ag43 variants into six distinct categories according to phylogenetic analyses, denoted as C1, C2, C3, C4, C5, and C6 (Ageorges et al. 2023). In EcN, Ag43a and Ag43b fall into the C3 category, while Ag43c belongs to the C6 category. Notably, after the cryptic plasmids curing, the expression level of the C6 category Ag43c significantly increased at the transcriptional level, while there was no apparent difference in the expression of the other two C3 category Ag43 genes, Ag43a, Ag43b.

Subsequently, we further validated the transcriptome data using qPCR, selecting four genes from the differentially expressed gene. Among these, two were significantly upregulated (fluC and yfjQ), and two were significantly downregulated (fadA and gcl). The gapA gene was used as an internal reference. The qPCR results exhibited a consistent trend with the transcriptome results (Fig. 3d), providing further confirmation that the cryptic plasmids curing led to a significant increase in the transcriptional expression of Ag43c.

Massive increase in protein level of Ag43c after cryptic plasmids curing

The structure of Ag43c was predicted using Robetta (https://robetta.bakerlab.org/). Ag43c, like other variants of Ag43, possesses a modular structure (Fig. 4a), including: the N-terminal signal sequence (SP), the L-type passenger domain (α domain), the autochaperone domain (AC), the C-terminal translocator (β domain) (Ageorges et al. 2023; Benz and Schmidt 2011; Koh et al. 2022). Upon crossing the inner membrane, the SP sequence is cleaved (Ageorges et al. 2019; Koh et al. 2022). After traversing the outer membrane, the region between the AC and β domain is cleaved by an unknown protease (Ageorges et al. 2023). The Ag43α domain attaches to the cell surface through non-covalent interactions and can be easily released through simple heat treatment (Henderson et al. 2004; Klemm et al. 2003; Owen et al. 1996).

Fig. 4

The differential level of Ag43 protein and mass spectrometry analysis. a Ag43c structure diagram. Ag43 features a modular structure, including: (1) the N-terminal signal sequence (SP), encompassing a highly conserved extension region (ESPR); (2) the passenger (α domain) responsible for protein function, which can be divided into three subdomains: SL (stem of the L shape), EJ (elbow joining), and BL (bottom of the L shape); (3) the autochaperone domain (AC); (4) the C-terminal β domain. The Ag43c structure was predicted by Robetta (https://robetta.bakerlab.org/). b The analysis of whole bacteria or heat-extracted proteins of EcN and EcNc by SDS-PAGE. c–e MS/MS spectra of peptides in Ag43c

The cultures of EcN and EcNc were incubated at 60 °C for 30 min, followed by centrifugation. The supernatant was collected as the heat-extracted proteins and was detected by SDS-PAGE. Interestingly, by comparing the heat-extracted proteins from the EcN and EcNc strains, a distinct band in the range of 50 kDa to 70 kDa is observed, consistent with the anticipated range of 54.4 kDa to 60.1 kDa of Ag43c (Fig. 4b). Subsequent mass spectrometry analysis for this distinct band showed the presence of three proteins, Ag43a, Ag43b, and Ag43c (Fig. 4c–e, Additional file 1: Fig. S1). However, the relative abundance of Ag43c was 462.9 times that of Ag43a and 2740.7 times that of Ag43b. Furthermore, the relative abundance of Ag43c was at least 9.5 times higher than that of other proteins (Additional file 2: Table S4). Thus, we concluded that the predominant protein within this distinct band is Ag43c, indicating that the elevated content of Ag43c is responsible for the observed differential band. Additionally, two recombinant cryptic plasmids expressing SOD gene were constructed, and the EcN mutant strain containing only the recombinant cryptic plasmids exhibited no differential band caused by Ag43c, similar to EcN strain, further confirming that the distinct band is a consequence of the cryptic plasmid curing (Additional file 1: Fig. S2). Based on the above results, we suggested that the expression of the Ag43c protein in EcNc is significantly higher compared to EcN, and this difference is attributed to the cryptic plasmids curing.

Ag43 does not mediate cell self-aggregation

The above results have demonstrated a significant increase in Ag43 expression at the protein level following the cryptic plasmids curing. Studies have indicated that Ag43 can induce autoaggregation and sedimentation, thereby influencing bacterial colonization and infection (Henderson et al. 1997). Therefore, we conducted an autoaggregation assay to compare the self-aggregation abilities of EcN and EcNc. The results revealed that there was no significant change in the autoaggregation ability between EcNc and EcN (Fig. 5a), and no sedimentation was observed in these two strains (Fig. 5b).

Fig. 5

Aggregation test of strains. a Aggregation effect of EcN and EcNc strains (n = 3). b Self-sedimentation visualization of EcN and EcNc strains. The effects observed after a 2-h standing at room temperature for two bacterial strains with similar initial OD600 values