In the string test, a sterile loop was applied to a single colony on an agar plate, which had been cultured at varying temperatures (28°C, 37°C, 40°C, 42°C) for a duration of 24 h. The loop was gently lifted, and if there was a 5 mm extension without breaking, the string test was considered positive, which characterizes hypermucoviscosity (Fang et al., 2004). Furthermore, the hypermucoviscosity was confirmed with the sedimentation assay as described previously (Mike et al., 2021). In short, 50 µL of a standardized bacterial suspension (0.5 McFarland standard turbidity in 0.9% (w/v) aqueous NaCl solution) was transferred into 5 mL LB broth and was incubated at the desired temperature (RT, 28°C, 37°C, 40°C, 42°C) for 24 h. Afterwards, 1.5 mL of the 24 h culture was transferred into a 2 mL reaction tube and centrifuged for 1000 x g for 5 minutes, 200 µL from the supernatant as well as from the 24 h culture were pipetted as triplicates into a 96 well plate and the OD₆₀₀ was measured with a plate reader (CLARIOstarplus, BMG Labtech, Ortenberg, Germany) (Eger et al., 2021).

The ability to generate exopolysaccharides was assessed using BHI agar plates supplemented with 5% sucrose (w/v) and 0.08% congo-red (w/v), following the methodology outlined in a prior study (Freeman et al., 1989). A single colony was picked up from an LB agar plate, streaked onto the stained BHI agar plates, and subsequently incubated overnight for 24 h at (RT, 28°C, 37°C, 38°C, 39°C, 40°C, 42°C). If the single colonies color black, they would be classified as positive for exopolysaccharides, while single colonies which are white to yellow would be considered negative for exopolysaccharide production.

The temperature influence on attachment factors was investigated with the crystal violet (CV) assay, as described previously (Schaufler et al., 2016; Eger et al., 2022) In short 50 µL of an overnight culture was transferred to 5 mL LB. After a visible turbidity samples were adjusted to an OD₆₀₀ of 0.01 and 200 µL was transferred as into a 96 well plate. Afterwards, they were incubated at 28°C, 37°C, 40°C and 42°C for 24 h. Following the incubation, the OD₆₀₀ was measured to determine the overall growth. Then, the supernatant of each well was discarded, the plate was washed three times with deionized water and dried for 10 min. Subsequently, fixed with 250 µL methanol for 15 min and then air dried for 15 min. Now the remaining cells were stained with 250 µL 1% crystal violet solution (w/v) for 30 min. This was followed by three washing steps with deionized water and drying for 10 min. The stained bacteria were dissolved with 300 µL of an ethanol acetone mixture with a ratio of 80 to 20 (v/v) and shaken for 30 min at 220 rpm at room temperature. Then the OD₅₇₀ was measured. The specific biofilm formation (SBF) (Niu and Gilbert, 2004) was then calculated. The negative control was subtracted from the OD₅₇₀ of the sample and then divided by the OD₆₀₀ of the sample.

To test the influence of different temperatures on the pathogenic potential the Galleria mellonella in vivo infection model was used, as previously described (Insua et al., 2013) with slight adjustments of the experiment setup in respect to this temperature dependent study. Briefly, 2 mL of overnight culture of PBIO1953 was harvested and pelleted at 16000 x g for 5 min at RT. The pellet was once washed with PBS and diluted to an OD₆₀₀ of 1.0. The 2 × 10⁹ CFU/mL were further adjusted to 2 × 10⁶ CFU/mL. The larvae (from Peter Zoopalast, Kiel, Germany) were randomly divided into 6 different groups of 10 individuals. 10 µL of the bacterial suspension was injected into the right proleg of the larvae. For the control groups 10 µL of PBS solution was used as injection, to check if the infection of Galleria mellonella larvae was affected by the traumata or the altered incubation temperature. The sample and control groups were incubated in a 90 mm glass petri dish at 28°C, 37°C and 40°C each. The vital state of the larvae was controlled every 24 h. The larvae were considered dead, if they showed strong pigmentation accompanied by not responding to physical stimuli. This assay was performed three independent times the results were pooled for each condition. a Kaplan-Meier plot was generated where the survival was plotted to visualize the mortality rates of the Galleria mellonella at different temperatures.

To determine the plasmid copy numbers of the three largest plasmids (plasmid 1 bla _NDM-1 and rmpA, plasmid 2 bla _CTX-M-15, plasmid 3) at different temperatures (RT, 28°C, 37°C, 40°C, 42°C) qPCR was used. Therefore, sets of different Primers seen in Supplementary Table 1 were designed to amplify a specific region on the gene constructs of interest. The primers were designed according to the manufacture’s instruction for Luna qPCR Kit (New England Biolabs GmbH, Ipswich, MA, US) and ordered from Eurofins (Eurofins Genomics Europe Shared Services GmbH, Ebersberg, Germany).

The calibration for the plasmids (1–3) and the chromosome was performed with the Q5 High-Fidelity DNA Polymerase (New England Biolabs GmbH, Ipswich, MA, US) according to the manufacture’s guideline, and isolated with NucleoSpin Gel and PCR Clean−up (MACHEREY-NAGEL GmbH & Co. KG, Düren, Germany) and quantified with the Qubit 4 fluorometer using the dsDNA HS Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA). The qPCR was performed with the Luna qPCR protocol (New England Biolabs GmbH, Ipswich, MA, US) in regard to the qPCR cycler CFX Opus 96 Real-Time PCR System (Bio-Rad, Hercules, CA, US). Each sample was set up as seen in the Supplementary Table 2 and each sample was measured in triplicates following the protocol shown in Supplementary Table 3 .

The measured concentration of the samples was related to the overall length of the gene structure and the plasmid copy number was calculated as

PCN: Plasmid copy number per genome equivalent P: Overall amount of the plasmid (in ng) C: Overall amount of the chromosome (in ng) Lc: Overall length of the chromosome (in base pairs) Lp: Overall length of the plasmid (in base pairs).

A single colony was inoculated into 30 mL of LB in an Erlenmeyer flask and cells were incubated at 28°C, 37°C, 40°C and 42°C for 24 hours. Subsequently, the bacterial suspension was adjusted to OD₆₀₀ = 1, and 1 mL of the suspension was diluted with 9 mL of 0.9% NaCl. The resulting suspension was then filtered through a hydrophilic polycarbonate filter (0.2 μm pore size, Merck Millipore). A segment of the filter was fixed with 1% glutaraldehyde and 4% paraformaldehyde in washing buffer (10 mM cacodylate buffer, 1 mM CaCl₂, 0.075% ruthenium red; pH 7) and then the samples were stored at 4°C until further processing (not more than 16 hours).

Samples were washed with washing buffer three times for 10 min each time, treated with 2% tannic acid in washing buffer for 1 h at RT, and then washed again with washing buffer three times for 15 minutes each time. Afterwards, the samples were dehydrated in a graded series of aqueous ethanol solutions (10, 30, 50, 70, 90, 100%) on ice for 15 min each step. Before the final change to 100% ethanol, samples were allowed to reach RT and then critical point-dried with liquid CO₂. Finally, the samples were mounted on aluminum stubs, sputtered with gold/palladium, and examined with a field emission scanning electron microscope Supra 40VP (Carl Zeiss Microscopy Deutschland GmbH, Oberkochen, Germany) using the Everhart-Thornley SE detector and the in-lens detector in a 70:30 ratio at an acceleration voltage of 5 kV. All micrographs were edited by using Adobe Photoshop CS6.

The DNA isolation for the qPCR was performed by either 1 mL of 24 h cultures in LB at 28°C, 37°C, 40°C and 42°C or directly from the BHI agar plates supplemented with congo−red and sucrose incubated at the 4 different temperatures, where 4 to 6 single colonies were scraped off and resuspended via vortexing in 500 µL of PBS in a 1.5 mL test tube. The DNA isolation was then performed with the MasterPure DNA Purification Kit for Blood according to the manufacturer’s specifications (Lucigen, Middleton, WI, USA). The quantification of the DNA was ensured by using the Qubit 4 fluorometer using the dsDNA HS Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA).

For transcriptomic analysis, the RNA was isolated from 50 mL liquid 24 h LB medium cultures (n=3) at 28°C, 37°C, 40°C and 42°C. 800 µL of the cultures were harvested in a 1.5 mL reaction tube and instantly chilled with liquid nitrogen for 5 s, to inhibit changes in the transcriptome. After the centrifugation (16,000 × g for 3 min at 2°C), the following RNA isolation was performed immediately with Monarch™ Total RNA Miniprep Kit (New England Biolabs GmbH, Ipswich, MA, US) according to the manufacture’s instruction. The quality of the isolated RNA was tested using the Bioanalyzer with the Total RNA Nano Chip (Agilent, Santa Clara, CA, US). The samples were shipped on dry ice to Novogene (NOVOGENE COMPANY LTD., Cambridge, UK) for RNA sequencing (Illumina NovaSeq 6000; paired-end 150 bp).

The closed genome sequence of PBIO1953 was annotated with Bakta v. 1.7.0 and bakta database v. 5. In order to process the raw reads and to further quality trim the data, Trim Galore v. 0.6.8 (https://github.com/FelixKrueger/TrimGalore) was used. The trimmed reads were then mapped with Bowtie 2 v. 2.5.1 (mode: –very-sensitive-local), where the assembly of PBIO1953 was used as a reference. The gene counts were calculated using featureCounts v.2.0.1/stranded-mode) based on PBIO1953s´annotation. In the following step, the count table was imported into R v. 4.3.1 (https://www.R-project.org/), and differentially expressed genes were identified with DESeq2 v. 1.40.0 in default mode, with one exception; genes with rowSums of <10 in the count table were excluded before the analysis. Within our analyses, we used an absolute log2 fold change 1.5 thresholds combined with an adjusted e value lower than 0.05 to determine differences within gene expression, between the different incubation conditions. We excluded one replicate which was incubated at 37°C due to a shift on principal component 1 (PC1) and PC2, which can be seen in Supplementary Figure S2 where principal-component analysis is plotted. Finally, all genes were annotated using the online tool eggnog-mapper v. 2.1.9 to extract COG (cluster of orthologous groups) categories.

Statistical analyses were conducted using GraphPad Prism v. 9.5.1 for Windows, developed by GraphPad Software (San Diego, CA US). To assess qPCR results, the Bio-Rad CFX Maestro 2.3 v. 5.3.0 22.1030 (Bio-Rad, Hercules, CA, US) was utilized. All phenotypic experiments consisted of three or more independent biological replicates. Unless otherwise specified, data were presented as the mean and standard deviation. Statistical significance was determined using p values below 0.05 to indicate significant differences among the results.

Ethical approval for this study was obtained from the ethics committee of the University of Greifswald, Germany (Approval number: BB 133/20), ensuring compliance with established ethical standards. Informed consent from patients was waived as samples were collected under a hospital surveillance framework for routine sampling. The study strictly adheres to the principles outlined in the Helsinki Declaration. No identifiable patient data, such as names and ages, was included in the study, and no tissue samples were utilized during isolation procedures.

To explore the underlying mechanisms, PCN-variations were measured using qPCR. Previously, we have shown that the convergent PBIO1953 strain harbors five different plasmids, one of which is a hybrid plasmid (plasmid 1) encoding both AMR and hypervirulence genes (Heiden et al., 2020). The three largest PBIO1953 plasmids were included in the subsequent analysis: plasmid 1 (360,596 bp), positive for the metallo-β-lactamase gene bla _NDM-1 and the regulator of the mucoid phenotype rmpA, plasmid 2 (130,131 bp) encoding the extended-spectrum β-lactamase (ESBL) gene bla _CTX-M-15, and plasmid 3 (72,679 bp) without any resistance genes ( Figure 2 ).

As the hypermucoviscosity switch was observed above 37°C, we normalized all PCN to 37°C. Interestingly, the PCN differed not only based on the different temperatures but also regarding the respective plasmid and experimental set-ups ( Figure 2 ). For the hybrid plasmid 1, the PCN increased with higher temperatures, with the highest value obtained at 42°C. The PCN of plasmid 2 did not show a clear temperature decency, but a PCN reduction was apparent at both 28°C and 42°C in the liquid culture set-up. Plasmid 3 showed a PCN decreasing with higher temperatures.

For transcriptomics, we first focused on genes displaying differential gene expression (DGE) at 28°C, 40°C and 42°C (e-value < 0.05, |log2fold| change > 1.5) in comparison to 37°C. DGE appeared mostly between 28°C and 37°C ( Supplementary Figures S4 , S5 ), while, at 37°C compared to 40°C and 42°C, we only noticed few differentially expressed genes. Our initial hypothesis that rmpA would be differentially expressed was rejected and, genes related to the capsular gene cluster were unaffected.

When comparing 28°C-associated gene expression to 37°C (and 40°C and 42°C), we observed various, significant changes. Notably, the chromosomally located ibpAB gene and two small heat shock proteins encoded on plasmid 2, exhibited differential expression at increased temperatures. These genes are known for mitigating growth defects under thermal stress (Kitagawa et al., 2000). Furthermore, we observed the upregulation of several stress-related genes at 37°C and higher, including the Fur repressor, which possesses protective properties against reactive oxygen species (Achenbach and Yang, 1997) and sodA, a gene involved in radical scavenging. Of particular interest was the upregulation of acrAB, known for its role in antimicrobial resistance and virulence (Padilla et al., 2010). Remarkably, the expression of mrkABCD, which encodes the type 3 fimbriae system, increased at 37°C and above. Previous research has suggested that environmental factors (Johnson et al., 2011), like temperature, can impact attachment factors, making this observation particularly intriguing. Conversely, we noted a downregulation of fimHGFDCIA, responsible for the expression of fimbriae type 1 (Struve et al., 2008), and the downregulation of the rfb gene, which encodes O-antigens (Kelly and Whitfield, 1996). Additionally, ompA, a known virulence factor and key element in immune system recognition (Llobet et al., 2009) was downregulated at increased temperatures. On plasmid 1, the partitioning system parAB responsible for segregation and plasmid stability (Bignell and Thomas, 2001) demonstrated increased expression at higher temperatures. Notably, genes associated with transposon activation and IS sequences were also upregulated on plasmid 1. The gene iutA, encoding aerobactin and contributing to bacterial virulence, was upregulated at 37°C (and 40°C and 42°C). Moreover, on plasmid 2, the metallo-resistance gene arsR was upregulated (Xu et al., 1996). Finally, plasmids 4 and 5 showed increased expression of genes involved in conjugative transfer, including traA (Yang et al., 2007) and mobC (Zhang and Meyer, 1997). When comparing the transcriptomes of 40°C to those at 37°C, DGE was observed in six genes. Notably, two downregulated genes, metR (associated with the methionine pathway) and argC (linked to the arginine pathway), were particularly interesting as they are important for general growth. Only two genes were upregulated, with betB standing out as this gene is involved in the biosynthesis of osmoprotective choline-glycine betadine (Landfald and Str0m, 1986). Similarly, the comparison of 42°C to 37°C revealed six differentially expressed genes. BetB demonstrated concordance with the upregulated genes at 40°C. An intriguing finding was the upregulation of pecM, a gene previously implicated as a potential deactivator of the fim operon and a member of the permease of the drug/metabolite transporter (DMT) superfamily.