Wild-type (WT) K. pneumoniae (ATCC BAA-1705), as well as its two subpopulations M1, M2 isolated from bacterial culture in simulated microgravity environment, were used in this study. WT strains were aliquoted, one as a control strain was cultured continuously for 2 weeks under the standard culture condition of normal gravity and the other was cultured in a HARV bioreactor filled with LB medium at 37°C with a rotational of 25 r.p.m, and passaged every 24 h (Wang et al., 2017). After 2 weeks of continuous incubation, the bacterial cultures were collected and transferred to LB agar plate and incubated at 37°C for 24 h under standard culture conditions. From the SMG group, M1 and M2 subpopulations were isolated according to the different morphology of the colonies. However, the colony morphology of the control group did not change compared with 2 weeks ago. Furthermore, the experiments also found that the colony morphology and bacterial cell arrangement of M1 and M2 subpopulations remain unchanged at least within five passages under the standard culture condition in vitro. So the control strain, M1 and M2 subpopulation strains were separately inoculated into LB medium and kept at 37°C with 200 r.p.m for 24 h under standard culture conditions. Then bacteria cultured were collected and flash-frozen in liquid nitrogen at –80°C for subsequent experiments. In our experiments, cryopreserved strains were activated under standard culture conditions and collected the bacteria cultured to the early logarithmic phase to prepare samples.

The protocol for the Hi-C assay was performed as previously described (Yin et al., 2020). Bacteria cultured to the early logarithmic phase were collected by centrifugation and washed twice with cold PBS. A final concentration of 3% fresh formaldehyde was used to crosslink the bacteria (1 × 10⁹ cells) at room temperature (RT) for 10 min. The formaldehyde was quenched by incubation with a final concentration of 0.375 M glycine at 4°C for 20 min and the sample was stored at –80°C or used directly for subsequent experiments. The cross-linked bacterial samples were resuspended in Tris-EDTA (TE), followed by addition of lysis buffer (50 U/μL). Lysis was performed at 37°C in an incubator for 10 min and followed by addition of SDS to a final concentration of 0.5% to accelerate lysis and the samples were left for a further 10 min at RT. 10% Triton X-100 was used to terminate the lysis reaction and the samples were incubated at RT for 10 min. After centrifugation, the supernatant was discarded and the DNA that collected in the pellet was digested by adding the digestion system (10 × NEB buffer 2.1 and 100 U of Sau3AI) and placing in an incubator at 37°C. The end-complementation system (10 × 2.0 NEB buffer, Klenow, 10 mM dATP/dTTP/dGTP, 5 mM biotin-16-AA-2′dCTP) was added and the reaction was left at RT for 4 h. Blunt-end ligation was performed using T4 DNA ligase with 10.0 mL final volume (Thermo Fisher Scientific, United States) at 16°C overnight. After ligation, RNaseA (Thermo Fisher Scientific, United States) was added to remove residual RNA and 200 mg/mL proteinase K (Thermo Fisher Scientific, United States) was added before incubation at 65°C for 5 h. DNA was purified and recovered using AMpure beads (Beckman, United States) according to the manufacturer’s protocol. The samples were sonicated for 60 s using Covaris M220 and the DNA was sheared to ∼400 bp length. Then, AMpure beads (Beckman) with an equal volume was used to purify and recover the DNA according to the manufacturer’s protocol. The target biotinylated fragments were captured using Dynabeads™ MyOne™ Streptavidin C1 (Thermo Fisher Scientific, United States). Hi-C libraries were constructed with 200 ng DNA as input amounts using the NEBNext^® Ultra™ II DNA Library Prep Kit for Illumina (NEB, United States) according to the manufacturer’s protocol. Fragment lengths from 400 to 600 bp were analyzed on an Illumina HiSeq X Ten platform (San Diego, CA, United States) with paired -end sequencing and two replicates per set of samples, for a total of 6 libraries.

We got about 20 million reads for each library. For the downstream data from high-throughput sequencing, we first filtered the raw data to remove unqualified reads, the data quality control were checked using MultiQC with default parameters (Ewels et al., 2016), and used bowtie2 (Langmead and Salzberg, 2012) to map the clean data to the reference genome (GCF_000240185.1) with default parameters. The resulting files were subsequently processed using HiCExplorer (Wolff et al., 2020) to generate the DNA-DNA interaction matrix files and the correlation between each sample and each replicate was analyzed. Then the two biological replicates were merged for subsequent analysis. The original Hi-C matrices were further normalized and corrected for data bias (including uneven distribution of restriction fragment lengths, ligation efficiency, and nucleotide composition) using the HiCExplorer tools. 2 kb resolution was used to identify local features of chromatin and for subsequent identification of relatively active and inactive regions of chromosomes (A/B compartments) (Takemata et al., 2019; Misteli, 2020), as well as CID analysis. Calling chromosome A/B compartments using cooltools with default parameters (Venev et al., 2021) and a script of cworld, matrix2insulation.pl (with parameters –is 100000 –ids 52000 –nt 0.2) was used to analyze CID boundaries (Lajoie and Venev, 2016).

Samples were collected by centrifugation and discarded the supernatant. The cells (approximately 5 × 10⁶) were treated with TRIzol reagent (Sigma-Aldrich, United States) and pipetted up and down until the liquid was mixed and incubated for 5 min at RT. Added 0.2 mL of chloroform per 1 mL of TRIzol Reagent used for lysis, then thoroughly mixed by shaking and incubated for 10 min at RT. Centrifuged the sample for 15 min at 12,000 × g at 4°C. Transferred the aqueous phase containing the RNA to a new tube. Added 0.5 mL of isopropanol to the aqueous phase and incubated for 20 min at RT. Centrifuged for 20 min at 12,000 × g at 4°C and discarded the supernatant. Washed RNA with 75% ethanol and discarded the supernatant by centrifugation, then air dry the RNA pellet for 5–10 min. Resuspended the pellet in 20 μL of RNase-free water by pipetting up and down. NanoPhotometer spectrophotometer^® (Bio-Rad, Hercules, CA, United States) was used to check the purity of RNA. Then RNA concentration was measured using Qubit^® RNA Assay Kit in Qubit 3.0 Fluorometer (Life Technologies, CA, United States). RNA integrity was evaluated using the RNA Nano 6000 Assay Kit of the Agilent Bioanalyzer 2100 system (Agilent Technologies, CA, United States). Next, cDNA library was constructed using NEBNext^® Ultra™ RNA Library Prep Kit for Illumina (NEB, United States) with 1 μg RNA as input amounts. With two replicates per sample, a total of 6 libraries were constructed. Libraries were deeply sequenced using an Illumina Hiseq X Ten platform and generated 150 bp paired-end reads.

Mapping values per kilobase of transcript per million fragments were calculated to determine the transcript expression levels. DESeq2 (Love et al., 2014) was used to analyze gene expression and the absolute value of the log2FoldChange ≥ 1 criterion was used to identify differentially expressed genes in the WT, M1, and M2 strains. GO and KEGG functional annotation analyses were performed using DAVID (Huang et al., 2009a,b) and KOBAS-I (Bu et al., 2021).

Cells (about 1 × 10⁹ cells) were Collected and used the QIAamp DNA Mini Kit (Qiagen, Germany) to extracted DNA according to the manufacturer’s instructions. Then DNA concentration was measured using Qubit^® DNA Assay Kit in Qubit 3.0 Fluorometer (Life Technologies, CA, United States) and DNA quality was assessed by 1% agarose gel electrophoresis. Methylation libraries were constructed by using the NEBNext^® Enzymatic Methyl-seq Kit (NEB) with 200 ng DNA as input amounts. And referred to the manufacturer’s protocol, we prepped libraries with fragments size of ∼400 bp for each sample (no biological replicates, a total of 3 libraries). High-throughput sequencing was performed using the Illumina sequencing platform, and the sequencing read length was 150 bp paired-end reads.

We got 3.3G of raw data in total. And bismark (Krueger and Andrews, 2011) was used to identify and extract methylation sites. The R package DSS (Wu et al., 2015) was used to identify differentially methylated regions of different components, using a threshold of P-value < 0.005.

Hi-C experiments were performed to explore differences in chromosomal conformations between the control strain and subpopulations M1 and M2 obtained by continuous incubation in SMG. Chromosome interaction heat maps were drawn at 2 kb resolution to identify different chromosome interactions at the genome-wide level. Compared to WT, the chromosome interaction patterns of M1 and M2 were altered (Figure 1A). Under SMG, short-range interactions decreased while long-range interactions increased, indicating that the chromosome folding and compact of M1 and M2 had become looser. To confirm this phenomenon, we analyzed the relationship between chromatin interaction intensity and one-dimensional sequence distance. Compared to WT, M1, and M2 had significantly fewer proximity interactions and more distant interactions, while the negative correlation between the strength of chromosome interactions and distance was diminished (Figure 1B). Referring to a previous study (Lioy et al., 2018), we plotted a scalogram to illustrate the cumulative amount of contact between each bin and its flanking region, reflecting the relative tightness of the contact distribution. Taken together, all the results showed that chromosomes in M1 and M2 were looser compared to the WT (Figure 1C). In addition, 3D simulation of chromosomes showed the same results (Supplementary Figure 1).

Next, we considered whether there was a corresponding increase in the open chromosome regions and changes in chromosome interactions when chromosome compact structures became looser. Chromosomal domains generally tend to be divided into distinct compartments, labeled “A” and “B.” The more open regions where gene expression is active are defined as A compartments, and the opposite expression-inactive regions are defined as B compartments (Takemata et al., 2019; Misteli, 2020). Therefore, we analyzed the A/B compartments along the chromosome of K. pneumoniae. The results showed that compartment A increased and compartment B decreased in M1 and M2, indicating that the active region of the chromosome increased after SMG culture (Figure 2A). In addition, a clear shift from B compartment to A compartment occurred in the replication terminus (ter) region found around position 2.4–3.2 M of the genome. This was consistent with our observation (Figure 1B) that the compacted region in the WT disappeared in M1 and M2. Next, we used the index of insulation score to delineate the CIDs and identified 21, 13, and 10 potential CIDs in WT, M2, and M1, respectively (Table 1). The CIDs were relatively reduced after SMG, and several originally smaller CIDs observed in the WT merged to form a larger CID in M1 and M2 (Figure 2B). These structural changes are closely related to gene expression, which will be analyzed in later sections.

To investigate the effect of SMG on K. pneumoniae gene expression, we performed transcriptome analysis. Using a twofold threshold, M1 and M2 had 215 and 304 DEGs, respectively, compared to the WT (Supplementary Tables 1, 2). Although parts of the genes had not been annotated by GO or KEGG databases, for M1 we still found GO and KEGG annotations of 23 upregulated genes related to the integral component of membrane and further upregulation of the genes involved in ribosome, protein transporter activity, transferase activity, cell adhesion, and pilus (Figure 3A). Downregulated genes in the M1 strain were those involved in transporter activity and cell adhesion (Supplementary Table 1). In M2, 57 upregulated genes were related to the integral components of the membrane and other upregulated genes were mainly involved in transport, cell adhesion, and pilus (Figure 3B). Downregulated genes in M2 were also related to integral components of the membrane (Supplementary Table 2).

In addition, cellulose synthesis-related genes (KPHS_03110, KPHS_03880, KPHS_03890, KPHS_43460, and KPHS_43590), as well as biofilm formation-related glycosyltransferase genes (KPHS_03620, KPHS_28150, and KPHS_30480) were upregulated in M2. These findings are consistent with previous studies showing that M2 has a high capacity for biofilm formation, with the biofilm being mainly composed of cellulose (Wang et al., 2017). The expression of five genes (KPHS_28260, KPHS_28270, KPHS_30010, KPHS_30020, and KPHS_30030) associated with quorum sensing, one of the important regulatory mechanisms of bacterial biofilm formation (Solano et al., 2014), was also upregulated in M2 (Table 2).

The spatial organization of the genome is essential for modulating gene regulation (Kong and Zhang, 2019). In mammalian T cells, there is a non-random overlap between regions of chromosome structural changes and regions of DEGs in response to gravity changes (Vahlensieck et al., 2021). In addition, our previous study found that under SMG, most DEGs of Vibrio natriegens were located in regions with changeable chromosome structures (Yin et al., 2020).

To further investigate the relationship between DEGs and chromosome structure in K. pneumoniae under SMG, we analyzed the distribution of DEGs on chromosomes using a 40 kb window and a 2 kb sliding step. DEGs were significantly enriched near some changeable CID boundaries, with more than twice the enrichment fold than their distribution at the genome-wide level (Figures 4A,B and Supplementary Figure 2). These variable CID boundaries were categorized into two main types: the first was CID boundaries present in WT but disappearing in M1 or M2 (CID boundaries ID-1,2,7,10,11,12,13,18,19,20 of WT), and the second was relatively conserved boundaries in all three strains (CID boundaries ID-5,21 of WT). DEGs were enriched in these regions and were mainly upregulated, and some of them were involved in physiological activities related to membrane components, pili, and phagosomes, while others were mostly putative genes of unknown function.

Several studies have shown that DNA methylation plays an important role in regulating gene expression during space flight (Singh et al., 2010; Kamal et al., 2018; Lei et al., 2020). We analyzed the methylation levels of WT, M1, and M2, and counted their differentially methylated regions (DMRs). Overall, the M1 and M2 genomes were in a hypomethylated state (Supplementary Figure 3 and Supplementary Tables 3, 4). However, we did not observe a significant association between these regions and the DEGs. In addition, although methylation levels are reduced near some regions with boundary changes, whether they are related to domain formation remains to be explored.