Luria-Bertani (LB), Roche, Kligler Iron Agar (KIA), MacConkey Agar Media (MAC), Motility Indol Urea Media Agar (MIU), and Salmonella and Shigella (SS) Agar media were prepared in our lab. The Gram stain and acid-fast staining kit were purchased from Sigma (USA). Taq and PFU DNA polymerase, as well as all restriction endonucleases, were purchased from NEB (USA). Primers were synthesized by Tsingke Biotechnology Co. (China) and the information were listed in Table S1 . E. coli DH5α was purchased from Takara (Japan). S. flexneri 2a, E. coli (ATCC 25922), Staphylococcus saprophyticus (ATCC 15305), S. aureus (ATCC 8095), Mycobacterium smegmatis (ATCC 14468), and M. tuberculosis H37Rv (ATCC 27294) were obtained from the experimental teaching center of Chongqing Medical University. The diagnostic sera kit for Shigella was purchased from Ningbo Tianrun Bio-pharmaceutical Co., LTD (China). The pOcus2 plasmid and simulated plastic specimens were kept in our lab and described previously (Jia et al., 2022; Zhang et al., 2013). The key enzymes of tryptophan metabolism pathway and the genes encoding flagella of E. coli were analyzed on Kyoto Encyclopedia of Genes and Genomes (https://www.kegg.jp/).

The generation of tnaA and FliC gene deficient E. coli DH5α△TnaA△FliC were described in supplementary data . On this base, a plasmid pOcus2-Sf2aO bearing the O-antigen determinant of S. flexneri 2a was constructed and transformed into the E. coli DH5α△TnaA△FliC to generate E. coli DH5α△TnaA△FliC2a which consistently express the S. flexneri 2a O antigen. The expression of S. flexneri 2a O antigen was confirmed using a serological agglutination test following the protocol of the kit and the plasmid pOcus2-Sf2aO were constructed in the following steps.

First, the backbone of the pOcus2 plasmid, including the replication origin sites and the ampicillin-resistant gene, was amplified using the primers T7-ter-F(ln) and pOcus2-R(ln). Then, a 2443bp fragment containing the bgt and grtII genes, and a 3984bp fragment containing the rfbB-C gene were amplified separately from the genome of S. flexneri 2a using the primer pairs bgt-F(ln)/grtII-R(ln) and rfbB-F/rfbC-R, respectively.Thereafter, all three fragments were ligated together using the Seamless Cloning Kit (Biorun, China) and transformed into normal E. coli DH5α competent cells to generate the intermediate plasmid pOcus-bgtII-rfbC. Finally, a 6233bp fragment was amplified from the genome of S. flexneri 2a using the primers frbC-F and prMD4-2A and inserted into the XhoI site of the intermediate plasmid pOcus-bgtII-rfbC using the Seamless Cloning Kit to generate the S. flexneri 2a O antigen expression plasmid, pOcus2-Sf2aO.

Female guinea pigs, three weeks old, were purchased from the Laboratory Animal Center at Chongqing Medical University and housed in cages that comply with animal welfare standards. After anesthesia, both eyes of each guinea pig (n = 4 per group) were inoculated with 5x10⁶/20 μL bacteria in PBS. The severity of eye inflammation was monitored for three days. All animal experiments were approved by the Ethics Committee of Chongqing Medical University, department Laboratory Animal Management and Use Committee of Chongqing Medical University (IACUC-CQMU) and were conducted in accordance with the regulations for experimental animal management. The acid tolerance test was performed by inoculating the bacteria in acidified LB media. The bacteria were cultured in LB media until the logarithmic growth phase (OD = 0.6), after which 1 μL of the culture was added to 100 μL of fresh acidified LB media (pH = 4.0) and incubated at 37°C for 0, 10, 20, and 30 hours. Samples were taken at different time points and spread on LB plates. After 18 hours, the number of colonies was counted, and the PFU of the bacteria was calculated. Each group was repeated three times.

A total of 10 teachers, 5 medical laboratory staff from the hospital, and students from three experimental classes were invited to identify E. coli DH5α△FliC△tnaA2a and S. flexneri 2a bacteria using standard identification methods, including biochemical reaction experiments and serological tests. Specifically, the bacteria were first streaked on MAC and SS plates and suspected pathogenic bacterial colonies were then picked for preliminary identification with Gram staining and puncture inoculation in KIA and MIU. Finally, the suspected bacteria were identified based on biochemical phenomena and serological results, and the results were recorded.To test the effectiveness of using E. coli DH5α△FliC△tnaA2a as a substitute for real S. flexneri 2a in experimental teaching, students with odd student numbers were required to practice with real S. flexneri 2a, while the other half practiced with E. coli DH5α△FliC△tnaA2a. Students who could not come to school due to the COVID-19 epidemic were required to practice with a virtual experiment system online. At the end of the semester, all students in the three groups were invited to take a mock examination on the identification of Enterobacteriaceae bacteria. The examination consisted of items such as zoning streak plating, selection of the correct colony, and preliminary identification with Gram staining, culture results in KIA and MIU, and results and discussion, with a maximum of 20 points available for each item. Finally, the scores of the three groups were tallied and analyzed.

In our investigation, we surveyed students from the medical laboratory technique and clinical medicine, as well as their teachers. The survey asked students to provide feedback on their experiences in the microbiology laboratory, including their feelings about laboratory classes, safety concerns, and attitudes toward substitute materials used in class. Surveys were conducted using either the school’s Superstar system or paper questionnaires. This study was approved by the Ethics Committee of Chongqing Medical University.

According to our study proposal, the survey of students and teachers was conducted by the researchers who recruited volunteers during the break of the laboratory classes and obtained their informed verbal consent. No incentives were provided to participate and the data obtained from the simulation test does not involve the privacy of participants, nor does it affect the official exam results. The Ethics Committee of Chongqing Medical University reviewed and approved our proposal and provided us with a supporting document (Reference Number: 2023027).

Data from the surveys and the survival rate of the bacteria in acid were analyzed using Graphpad Prism 8.0. The show as average with standard error of mean. Statistical analysis was performed using Student’s t-test and p values less than 0.05 were considered significantly different and indicated by asterisks in the figures.

The attitudes of teachers and students towards microbiology laboratory classes were evaluated through questionnaires, including biosafety and the most interesting parts of the class.

Regarding biosafety during class, the students were asked, ‘Do you worry about microbial infections during the experiment?’ The students were told to rate their level of worry from 1 to 10; 1 indicating the least worry. The results show that 24.2, 21.5, 10.1, 12.7, 17.7, and 6.3% of the students chose 1, 2, 3, 4, 5, and 6 points, respectively, while the remaining, 13.8% of the students, chose 7–10 points. Similar to the student’s choice, 30, 30, 10, and 20% of teachers chose one, two, three, and four points, respectively, while 10% chose more than five points, indicating that most students and teachers were moderately worried ( Figure 1A ). When asked about the bacteria which is most likely to cause infection accidents, 26.1% of the students surveyed responded Staphylococcus aureus, followed by Shigella and Salmonella, both with 16.1%. The teachers’ responses were similar to those of the students; 40, 30, 20, and 10% of the teachers chose S. aureus, Salmonella, Shigella, and others, respectively ( Figure 1B ). This showed that both teachers and students had a good understanding of bacterial pathogenicity, and the concerns about biosafety.

To evaluate potential biosafety issues arising during medical microbiology experiments conducted by students, classroom observations and questionnaires on potential biosafety issues were collected. It was observed that student behavior during the class was not appropriate; approximately 1/5^th of the classes had incidents requiring help from the teacher, such as a fire caused by an alcohol lamp, and injuries, which may have led to biological contamination. Statistical analysis revealed that majority of the students encountered situations that may have led to infection during the semester-long course. Many students experienced glove tearing (27.4%), hand injuries caused by broken glass (6.2%), direct contact with bacterial cultures (10.6%), being splashed with bacteria-containing materials (34.5%), broken tubes (1.8%), and other hazards (19.5%) ( Figure 1C ).

In addition, students’ awareness about personal responsibility was evaluated by investigating whether they washed their hands after class. Although most students washed their hands after every class (74.6%), some students washed their hands after classes (24.1%) or never (1.3%), indicating that not every student conducted appropriate biosafety procedures ( Figure 1D ).

Owing to the COVID-19 pandemic, some laboratory courses were completed online. Therefore, students’ and teachers’ attitudes towards online virtual experiments was investigated. Most students showed very little interest in purely virtual experiments, instead preferring in-person classes. Approximately 91.1% and 90% of the students and teachers, respectively, opposed purely online virtual experiments.

Finally, a survey to identify the course content that most interested students was conducted. Results showed that Enterobacteriaceae were of the greatest interest, scoring higher than Pyrococcus and non-fermentative bacteria, mainly because the Enterobacteriaceae test procedure is more complex and interesting. It was also the main focus of the course. In comparison, students were least interested in viruses since there were no in-person virology-focused experiments. There was only one virology-based experiment which was completed by viewing videos and pictures taken by the teachers ( Figure 1E ).

In conclusion, most teachers and students prefer in-person medical microbiology laboratory classes, although most of them have encountered potential biosafety problems. Therefore, the use of safe biological materials is necessary. Given the students’ keen interest and, this study focused to simulate Shigella by modifying E. coli DH5α strain.

E. coli DH5α strain was adapted to mimic key features of S. flexneri 2a by deleting the tnaA and FliC genes and introducing an O-antigen expressing plasmid, resulting in E. coli DH5α△tnaA△FliC2a. Additional details are provided in Supplementary Data 1 .

As expected, E. coli DH5α△tnaA△FliC2a had characteristics similar to S. flexneri 2a, as shown by Gram staining and biochemical experiments. On MAC and SS agar plates, E. coli DH5α derived strains and S. flexneri 2a were colorless and translucent, while E. coli produced pink colonies ( Figure 2A ). This demonstrated the loss of lactose decomposition ability in E. coli DH5α△tnaA△FliC2a and other DH5α-derived strains, which are representative of intestinal pathogenic bacteria, such as Shigella and Salmonella.

Under a light microscope, E. coli, E. coli DH5α, S. flexneri 2a, and E. coli DH5α△tnaA△FliC2a showed identical Gram-negative bacilli scattered in the array. E. coli DH5α-derived cells were slightly larger than Shigella cells, but shorter than E. coli; however, this size difference was not enough to distinguish them ( Figure 2B ).

The characteristics observed in KIA and MIU media were key diagnostic features used to differentiate Shigella from other Enterobacteriaceae. Generally, the characteristics of E. coli on KIA are AA+-, while those of Shigella are KA-/+-. In this part, E. coli, E. coli DH5α, E. coli DH5α△tnaA, E. coli DH5α△tnaA△FliC2a, and Shigella in KIA media were tested and compared. The results were similar to those expected, except that compared to Shigella, the E. coli DH5α-derived strains produced lesser gas, but significantly less as compared with E. coli. Generally, only some Shigella strains produce gas in KIA. These DH5α-derived strains are not perfect but they closely simulate the characteristics of Shigella in KIA ( Figure 2C ). Their characteristics in the MIU media were as anticipated; E. coli and E. coli DH5α in MIU media showed ++-, those of E. coli DH5α△tnaA were +–, and of E. coli DH5α△tnaA△FliC2a and Shigella were —. This illustrates that E. coli DH5α△tnaA△FliC2a can effectively replace S. flexneri 2a in MIU experiments ( Figure 2D ).

Further, the serological characteristics of E. coli, E. coli DH5α, S. flexneri 2a, E. coli DH5α△tnaA△FliC, and E. coli DH5α△tnaA△FliC2a were tested with the Shigella serological test kit. Only S. flexneri 2a and E. coli DH5α△tnaA△FliC2a agglutinated with Shigella multivalent serum and S. flexneri 2a serum ( Figure 2E ). Additionally, E. coli DH5α△tnaA△FliC2a samples were sent to five hospital laboratories for blind testing in order to further confirm its characteristics, where each hospital identified it as S. flexneri.

These results indicate that E. coli DH5α△tnaA△FliC2a can effectively simulate S. flexneri 2a in terms of its key biological characteristics.

Although E. coli DH5α△tnaA△FliC2a effectively simulated S. flexneri 2a, it was necessary to confirm its safety. Given Shigella’s intolerance to acid and pathogenicity to guinea pigs, acid resistance and guinea pig eye toxicity testing were carried out to determine the resistance and toxicity of E. coli DH5α△tnaA△FliC2a, respectively.

E. coli, E. coli DH5α, E. coli DH5α△FliC△tnaA, E. coli DH5α△FliC△tnaA2a, and S. flexneri 2a were placed in acidic LB media (pH 4.0) to monitor their survival rate for 30 h; this showed that bacterial colonies formed, decreased with increased incubation time. E. coli had the highest survival rate after 30 h (23.7%), followed by E. coli DH5α (0.42%), E. coli DH5α△FliC△tnaA (0.82%), and E. coli DH5α△FliC△tnaA2a (0.25%). Almost all S. flexneri 2a were killed after 30 h ( Figure 3A ). This indicated that DH5α-derived strains had intermediate resistance to acid, that is between S. flexneri 2a and E. coli. Since E. coli is considered safe when used in student experiments, E. coli DH5α and its derived strains should be safer.

To further confirm the safety of E. coli DH5α△FliC△tnaA2a, guinea pig eye toxicity tests were carried out. Each strain was inoculated into the eyes of guinea pigs. After 72 h, all four guinea pigs inoculated with S. flexneri 2a had inflammatory symptoms, such as redness, blurred eyes, and massive secretions in both eyes, while guinea pigs inoculated with the other strains showed no inflammatory symptoms ( Figure 3B ). This shows that E. coli DH5α△tnaA△FliC2a, similar to E. coli and E. coli DH5, had lesser bio-hazard than S. flexneri 2a.

Since E. coli DH5α△tnaA△FliC2a had greater biosafety and analogous biochemical and biological characteristics as S. flexneri 2a, its effectiveness in teaching in comparison to S. flexneri 2a and plastic specimens was evaluated. Experimental reports submitted by students in three independent groups showed that most students accurately identified S. flexneri 2a (86.84%), E. coli DH5α△tnaA△FliC2a (86.99%), and plastic specimens (81.33%) as Shigella sp., and only a few incorrectly identified the specimens as E. coli (4.66%, 6.44%, and 6.54%, respectively) or other Enterobacteriaceae (85.0%, 65.70%, and 12.15%, respectively) ( Figure 4A ).

Further, the attitudes of teachers and students towards E. coli DH5α△tnaA△FliC2a and plastic specimens was investigated. According to the survey of students’ attitudes towards simulation materials, 17.72% of the students believed that plastic specimens should be used instead of real bacteria, 17.72% had no opinion, while 64.66% opposed the use of plastic specimens. The major objections were that the plastic specimens looked too fake and that it was impossible to carry out certain operations, such as streaking, inoculation and bacterial culture. In contrast, 74.68% of the students agreed to substitution of Shigella with E. coli DH5α△tnaA△FliC2a, while 18.19% had no opinion, and only 6.33% opposed the substitution ( Figure 4B ). Subsequently, a survey on teachers’ attitudes was conducted, which showed that 10% of the teachers supported plastic specimen use, 20% had no opinion while 70% opposed the idea. For E. coli DH5α△tnaA△FliC2a, half of the teachers supported its use, 30% had no opinion, and 20% opposed its use ( Figure 4C ). Teachers had reasons similar to students, including that plastic specimens can only be observed, but some who supported the use of plastic specimens also cited safety considerations.

Students’ scores also reflected the objective effect of E. coli DH5α△tnaA△FliC2a in laboratory teaching. Students were divided into three groups to practice Enterobacteriaceae identification. Groups A and B practiced with E. coli DH5α△tnaA△FliC2a and S. flexneri 2a, respectively, while group C was composed of students who could not return to school because of the COVID-19 pandemic and therefore practiced through online virtual experiments. A simulation test was conducted and the scores were analyzed. The test involved various operations, including Gram staining, single colony isolation (at least five single colonies), selecting suspicious colonies for Gram staining and culture, inoculation, cultivation, interpreting KIA and MIU media results, and explaining the experimental results to teachers, each item was assigned a score out of 20. As expected, the average scores of groups A and B for each criterion were similar but significantly higher than that of group C.

Students in groups A and B scored an average of 19.22 and 19.32 points, respectively, in zoning streak plating, while the students in group C only scored 16.9 points. For selecting suspicious colonies and subsequent preliminary identification after Gram staining, groups A and B scored 18.87 and 18.68 points on average, respectively, while group C scored 17.10 points. In terms of interpretation and results, the scores were relatively close between the three groups; the average scores of group C were 18.20 and 18.30, lower than those of groups A (19.25 and 19.52 points) and B (19.63 and 19.53 points). The average scores of groups A, B, and C were the closest for interpretation and discussion of the results at 19.83, 19.68, and 19.20 points, respectively ( Figures 4D , S2 ). In terms of overall grades, students in groups A and B achieved average scores of 96.78% and 96.84%, respectively, while group C obtained an average score of 89.70% ( Figure S2 ).