L. rhamnosus ATCC 53103 was from the American Type Culture Collection (Manassas, VA, United States) and cultured in MRS medium at 37°C. It was revitalized in MRS medium at 37°C three times before use. Cultures were stored at –80°C in MRS medium containing 10% glycerol. The growth of L. rhamnosus ATCC 53103 was monitored by an automatic growth curve analyzer (Bioscreen Cpro; OY Growth Curves, Finland) at 600 nm (OD₆₀₀). For osmotic stress, the cells (OD₆₀₀ ∼1.8) were collected by centrifugation (6,000 × g, 10 min, 4°C) and then resuspended in MRS medium containing 0.2, 0.4, 0.6, and 0.8 M sodium lactate. The effect of osmotic stress on cell growth was further monitored by an automatic growth curve analyzer (Bioscreen Cpro; OY Growth Curves, Finland). The relevant results were shown in Supplementary Figure 1. Because a higher sodium lactate concentration (0.8 M) in the MRS inhibited the growth of L. rhamnosus ATCC 53103. When the sodium lactate concentration was 0.6 M, the growth rate of L. rhamnosus ATCC 53103 decreased significantly, while the viable counts reached 8.66 log₁₀CFU/mL after 24 h culture. Therefore, the final concentration of sodium lactate in this study was 0.6 M.

The cultures of control group and osmotic stress group were harvested by centrifugation (6,000 × g 15 min, 4°C) after 3.5 and 6.5 h (OD₆₀₀ ∼1.0), respectively. Total RNAs were extracted using a TRIzol-based method (Life Technologies, CA, United States). rRNAs were removed using the Ribo-Zero Magnetic Gold Kit (Epicenter Biotechnologies, Madison, WI, United States). RNA quality was checked using the Agilent 2200 TapeStation system (Agilent Technologies, Inc., Santa Clara, CA, United States). The library was constructed using a TruSeq RNA Sample Prep Kit v2 (Illumina, San Diego, CA, United States), and then sequenced by using the Illumina HiSeqi^TM 2500 platform with pair-end 150 base reads.

For Ribo-Seq, chloramphenicol (200 μM) was added into the cultures when OD₆₀₀ reached 1.0, and then cells were harvested by centrifugation (6,000 × g, 10 min, 4°C) after shaking for 2 min. Bacterial sludge was washed using resuspension buffer (20 mL) composed of NH₄Cl (100 mM), MgCl₂ (10 mM), chloramphenicol (1 mM), and Tris-HCl (pH 8.0, 20 mM). The cells were collected by centrifugation (4,000 × g, 5 min, 4°C), and then were mixed immediately with cell lysis buffer. The resuspended extracts were transferred from lysis buffer to new microtubes, pipetted several times then incubated on ice for 10 min. The cells were triturated ten times through a 26-G needle. The lysate was collected by centrifugation (20,000 × g, 10 min, 4°C. To prepare RPFs, RNase I (7.5 μL) and DNase I (5 μL) were added to lysate (300 μL) to incubate for 45 min with gentle mixing. Nuclease digestion was stopped by adding RNase inhibitor (10 μL). RPFs were isolated using the RNA Clean and Concentrator-25 Kit (R1017, Zymo Research, Orange County, CA, United States). The Ribo-seq libraries were constructed using NEBNext^® Multiple Small RNA Library Prep Set from Illumina^® (catalog no. E7300S, E7300L). 140–160 bp PCR products were enriched to generate cDNA libraries and then sequenced using Illumina HiSeq^TM 2500.

The gene expression level was normalized by the fragments per kilobase of transcript per million (FPKM) of mapped reads to control the influence of gene lengths and amount of sequencing data in calculation of gene expression. The edgeR package¹ was used to identify DEGs across sample groups. Gene ontology (GO) annotation and Kyoto Encyclopedia for Genes and Genomes (KEGG) pathway of DEGs were analyzed. According to expressions levels in translation and transcription, DEGs were classified into five different groups: unchanged (DEGs were not regulated at both two levels), homodirection (DEGs were regulated at the two levels with consistent trends), opposite (DEGs were regulated at the two levels with opposite trends), translation (DEGs were only regulated at the translation level) and transcription (DEGs were only regulated at the transcription level).

The TEs of all DEGs were identified, calculated, and compared using RiboDiff (Zhong et al., 2017). According to DEGs expression with the changes of TE at the transcription level, DEGs were classified into five different groups as described above: Unchanged; Homodirection; Opposite; TE (DEGs only had change at the TE level); and Transcription.

Raw data were filtered following the previous study (Langmead and Salzberg, 2012). Trimmed reads were mapped to L. rhamnosus reference transcriptome² allowing no mismatches using Bowtie2 (version 2.2.8). Retained reads were aligned to the reference genome using Bowtie2. Genes and gene expression were identified and calculated using RSEM. Genes with|log₂(fold change)| > 1 and a false discovery rate (FDR) < 0.05 were considered as significant DEGs. Both Ribo-seq and RNA-seq data types had two biological duplications.

To systematically investigate the effect of osmotic stress on transcription and translation regulation, RNA-seq and ribosome profiling were analyzed in the same two sets of parallel populations of L. rhamnosus ATCC 53103 cells with two different concentrations of sodium lactate (control vs. 0.6 M). Each experiment was divided into control group (CG) and osmotic stress group (OS), respectively (Supplementary Figure 2). The regulation of transcription and translation under osmotic stress were examined by deep sequencing of cellular total mRNAs and RPFs, respectively. A large number of reads were produced by ribosome footprints and RNA-seq transcripts. The average read was around 30 bp.

The total reads, ranging from 11,356,406 to 81,788,555 per DNA library, were produced by deep sequencing (Supplementary Table 1). After filtering, around 30 and 20 million reads were generated from RPF of CG and OS samples, respectively. Around 26 and 24 million mRNA-seq reads were generated from CG and OS samples, respectively. These reads were mapped to L. rhamnosus reference transcriptome data. The mapping efficiency of RPF samples was ∼20%. There are high correlations between the two biological duplication (R²>0.85) for both Ribo-seq and RNA-seq data. A total of 5578 and 5616 mapped genes were obtained from RPF samples of CG and OS, respectively. A total of 5,318 and 5,370 mapped genes were obtained from CG and OS mRNA samples, respectively.

The length of RPFs in OS and CG samples was around 30 nt (Figures 1A,B). The triplet periodicity between OS and CG samples were compared by scanning from the start codon to stop codon, and a strong three-nucleotide periodicity in OS and CG samples was observed (Figures 1C,D).

To explore how L. rhamnosus ATCC 53103 cells deal with osmotic stress, both translation and transcription relative variations between CG and OS were examined. The results showed that osmotic stress changed gene expression at the translation and transcription levels (Figure 2). A total of 289 and 614 DEGs were upregulated while 156 and 336 DEGs were downregulated at the translation and transcription levels, respectively (Figures 2A–C). About a quarter (11.5 and 12.1%) of the regulated DEGs were shared between the two levels (Figures 2D,E). At the translation and transcription levels, the expression of genes from samples of OS was moderately correlated (R² = 0.7757; Supplementary Figure 3). GO annotation and KEGG pathway analysis also showed that DEGs largely overlapped in GO and many KEGG pathways which also illustrated a big overlap at these two levels including ABC transporters, biosynthesis of antibiotics, secondary metabolites, and metabolic pathways etc. (Figures 3A,B). It is worthwhile to note that large numbers of DEGs showed different expression trends and many DEGs of enriched pathways showed a different response at the two levels. These results demonstrated that the regulation of DEGs at the translation level played an important role under osmotic stress in LGG.

To obtain the changes of gene expression trends at the translation and transcription levels simultaneously, DEGs were divided into different groups based on the fold change of gene expression (|log₂(fold change)| > 1) between CG and OS (Figure 4 and Supplementary Table 2).

A total of 93 DEGs were upregulated at the two levels and 53 DEGs were downregulated at the two levels (yellow dots). These DEGs were mainly involved in the fatty acid biosynthesis and metabolism, ribosome assembly, and purine metabolism pathways. The DEGs purD, purH, purN, purM, purQ, purC, and purK involved in purine metabolism were downregulated from 99.37- to 249.25-fold and from 12.25- to 205.53-fold at the translation level and transcription level, respectively. The DEGs accA, accD, accC1, accB, pksA, and fabH involved in fatty acid biosynthesis were downregulated from 4.56- to 14.47-fold and from 3.52- to 7.35-fold at the translation level and transcription level, respectively.

A total of 24 DEGs were downregulated at the transcription level and upregulated at the translation level, while 28 DEGs were upregulated at the transcription level and downregulated at the translation level (purple dots). These DEGs were mainly involved in biosynthesis of phenylalanine, tyrosine, and tryptophan, protein export, phosphotransferase system (PTS), ABC transporters, galactose metabolism, and pyrimidine metabolism pathways. The DEGs fruA and lacF involved in the PTS pathway were upregulated and downregulated from 2.09- to 2.57-fold and from 7.63- to 8.43-fold at the translation level and the transcription level, respectively.

A total of 752 DEGs were regulated at the transcription level only (red dots). These DEGs were mainly involved in biosynthesis and metabolism of amino acid, the two-component system, ABC transporters and pyruvate metabolism pathways. The DEGs tauA, tauB, tcyJ, and macB2 involved in the ABC transporters were upregulated from 4.00- to 13.64-fold, respectively. The DEGs cydA, ciaR, dltC, citF, and citC involved in the two-component system were downregulated from 2.26- to 8.07-fold, respectively.

A total of 247 DEGs were regulated at the translation level only (blue dots). The DEGs levE, fruK, rhaD, sorA, and rhaB involved in fructose and mannose metabolism were upregulated. The DEGs licB, bglP, sorA, manX, and mtlF involved in the PTS pathway also were upregulated. The DEGs rplL, rpmD, rplF, rplN, and rplP involved in ribosome assembly were upregulated about 2–3-fold. The DEGs potA and potB involved in the ABC transporters pathway were upregulated by 10.42- and 7.69-fold, respectively. Meanwhile, the DEGs cysE, cysK, hisG, hisZ, nodI, oppA, and macB involved in biosynthesis of amino acids and antibiotics, and ABC transporters pathway also were downregulated.

TEs of 363 and 386 DEGs were upregulated and downregulated, respectively, which suggested that TE played an important role in response to osmotic stress in LGG (Figure 5A and Supplementary Table 3). GO annotation showed that these DEGs were mainly enriched in cell part, membrane part, binding, catalytic activity, metabolic process, biological regulation, localization, cellular, and single-organism process (Figure 5B). KEGG pathway analysis indicated that these DEGs were mainly involved in biosynthesis of secondary metabolites, biosynthesis of antibiotics, amino acids metabolism, pyruvate metabolism, carbon metabolism, purine metabolism, glycolysis pathway, ribosome assembly, the two-component system, ABC transporters, PTS pathways (Supplementary Table 4). The DEGs rpsD, rpsT, rpmG, rpsQ, rpsC, rpsS, rpsJ, rpsZ, purD, purH, purN, purK, desR, yvfT, bceB, ywqE, iphP, htrA, opuCD, opuCC, opuCB, and opuCA were involved in ribosome assembly, purine metabolism, the two-component system, and ABC transporters pathways, and their TEs were downregulated. Other DEGs, such as secG, dppC, comA, yidC, manZ, and lacF were involved in quorum sensing and PTS pathways and their TEs were upregulated.

The calculation and analysis showed that the Pearson correlation coefficient between mRNA abundance and TE was –0.5845 (Supplementary Figure 4). This showed that transcription abundance affected TE. Hence, the distribution of DEGs was classified into different groups based on the regulation of TE and gene expression at the transcription level (Figure 5C). A total of 434 DEGs were regulated only at the transcription level while their TEs were not regulated, such as rpsN, rplJ, accA, fabZ, pksA, tauA, tcyJ, nodI, purF, and trpA. They were mainly involved in the assembly of ribosomes, fatty acid biosynthesis and metabolism, PTS, ABC transporters, and metabolism of starch, sucrose, purine, fructose, mannose, amino sugar, and nucleotide sugar pathways (Supplementary Table 5). TEs of 233 DEGs were regulated while their expressions were not regulated at the transcription level, such as cysE, cysK, sdaAB, ilvE, tpiA, glcK, kdgK, rhaD, and ssdA. They were mainly involved in the metabolism of cysteine and methionine, glycolysis, and microbial metabolism in diverse environment pathways (Supplementary Table 6). Only 11 DEGs were regulated at the transcription level, while their TEs were also regulated, and their regulations had similar trends. These DEGs included fhs, purH, purN, accD, accB, purD, purK, and potC. They were mainly involved in the metabolism of purine, fatty acid, propanoate, carbon, and pyruvate pathways (Supplementary Table 7). TEs of 286 DEGs were downregulated while their expressions were upregulated at the transcription level. These DEGs were mainly involved in pathways of biosynthesis and metabolism of amino acids, antibiotic biosynthesis, the citrate cycle, glycolysis, glutathione metabolism, carbon metabolism, amino sugar metabolism, nucleotide sugar metabolism, pyrimidine metabolism, and ribosome assembly. Such as the DEGs dapB and dapH were involved in lysine biosynthesis via the succinyl-DAP and acetyl-DAP pathways, respectively. SpeF was involved in arginine and proline metabolism. The DEGs rpmG, rpsC, rpsD, and rpsT were involved in ribosome functional group assembly. The mapped reads of these DEGs are shown in Figures 6A–D. TEs of 219 DEGs were upregulated and their expressions were downregulated at the transcription level. These DEGs mainly involved in pathways of protein export, ABC transporters, and two-component system (Supplementary Table 8). The DEGs included lspA, yheI, ecfT, and dltC, and the mapped reads are shown in Figures 6E–H.

Codon resolution is one of the advantages of ribosome profiling. To evaluate the translation events, the average occupancy of ribosomes was calculated around the start and stop codons. The footprints increased significantly under osmotic stress in L. rhamnosus ATCC 53103, which mapped to the initiation region around the 5′-end approximately at the –15 position (Figure 7A). In addition, ribosome occupancy from the nucleotide –100 to the end codon was also higher in OS than that in CG (Figure 7B). A few typical DEGs had higher ribosome occupancy under osmotic stress, such as pdhB, pdhD, and metF (Figures 7C–E). TEs of these DEGs were downregulated under osmotic stress.