Cell samples of L. paraplantarum RX-8 were obtained in co-culture and mono-culture at 24 h by centrifugation (10,000 rpm, 10 min, 4°C). After removing the supernatant, cells were immediately cleaned by phosphate buffered saline (PBS) two times, frozen in liquid nitrogen for 15 min, and then stored at −80°C.

Total RNA extractions were performed using Magen HiPure Universal RNA Kit (Magen, China). RNA concentration and purity were measured using Qubit 3.0 (Thermo Fisher Scientific, MA, USA) and Nanodrop One (Thermo Fisher Scientific, MA, USA), and integrity was confirmed using the Agilent 4200 system (Agilent Technologies, Waldbron, Germany). The RNA samples with an RNA integrity number (RIN) >6.5, and 280/260 ratios >1.5 were further used for RNA-sequencing purposes. Libraries were generated from three replicates using NEB Next^®Ultra™ Directional RNA Library Prep Kit for Illumina^® (New England Biolabs, MA, USA). From these libraries, the PE150 reads were produced with the Illumina Novaseq6000 platform by Guangdong Magigene Biotechnology Co., Ltd. (Guangzhou, China). All raw RNA-Seq data were submitted to NCBI under BioProject PRJNA901346.

The raw reads were trimmed using Trimmomatic (version 0.36) to obtain the qualified reads (Bolger et al., 2014). Then, the rRNA sequences were removed in the qualified reads mapping with NCBI Rfam datasets using Bowtie2 (version 2.33) (Langdon, 2015). Initially, we mapped the mRNA reads of these samples to the reference genome L. plantarum WCFS1 downloaded from NCBI, which was the strain that was most closely related to L. paraplantarum RX-8, but the acquired alignment rate was mostly low at around 50% and limited to transcriptomic analysis. Therefore, to obtain a higher mapping rate, we assembled the draft genome of L. paraplantarum RX-8 using Unicycler (version 0.4.7) to assemble the combined Illumina NovaSeq PE150 data and Nanopore PromethION data (Wick et al., 2017). The new reference genome was uploaded to the NCBI shared database for research retrieval in NCBI GenBank CP111117-CP111119. Eventually, the transcriptomic data were mapped to the new reference genome using Hisat2 (version 2.1.0) (Kim et al., 2019), and the mapping rate was improved to above 90%. Read counts per genome assembly annotated transcript were calculated using HTseq (version 0.9.1) (Anders et al., 2015).

The differentially expressed genes (DEGs) between groups (RY and R, three biological replicates) were identified from all genes according to the expression levels of the transcripts, based on the statistics of the fragments per kilobase of each reading, per million mapped reads (FPKM). The DEGs were detected using edgeR (version 3.16.5) with default parameters (Robinson et al., 2010). We then used Benjamini and Hochberg's correction approach to adjust the p-values to control the false discovery rate (FDR). The genes were regarded as DEGs if they satisfied the threshold with FDR ≤ 0.05 and |Log₂FC|≥1. These genes were used for further enrichment analysis.

The enrichment analyses of DEGs with Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases were implemented using clusterProfiler (version 3.4.4), in which gene length bias was corrected. The GO terms and KEGG pathways with FDR ≤ 0.05 were regarded as candidates for enrichment functional annotation via the DEGs.

Cell samples of L. paraplantarum RX-8 in co-culture and mono-culture were collected at 24 h according to Section 2.5.1. The protein was extracted by using a lysis buffer (8 M urea, 50 mM Tris8.0, 1% NP40, 1% sodium deoxycholate, 5 mM dithiothreitol (DTT), 2 mM EDTA, 30 mM nicotinamide, and 3 μm trichostatin A), and, after sonication on ice, the total protein concentration of the supernatant, which was obtained by centrifugation (20,000 rpm, 10 min, 4°C), was determined by using a BCA Protein Assay kit. The protein sample was reduced by DTT (5 mM, 45 min, 30°C), later alkylated with 30 mM iodoacetamide (30 mM, 1 h, RT) in darkness, and then precipitated with ice-cold acetone. After being washed thrice with acetone, the precipitate was suspended in 0.1 M triethylammonium bicarbonate (TEAB) and digested with trypsin (1/25 protein mass, Promega) for 12 h at 37°C. Finally, the reaction was ended with 1% trifluoroacetic acid (TFA), and the resulting peptide was desalted with Strata X C18 SPE column (Phenomenex, Torrance, CA, USA) and vacuum-dried in Scanvac maxi-beta (Labogene, Alleroed, Denmark).

After reduction and alkylation, six digested peptide samples (RY and R, three biological replicates, 100 μg/sample) were transferred to the TMT Reagent vial (Thermo Fisher Scientific, 90066) and incubated for 2 h at room temperature. Then, 8 μl of 5% hydroxylamine was added to the mixture and incubated for 15 min to quench the reaction. Finally, the labeled samples were fractionated into 15 fractions by XBridge Shield C18 RP column (Waters, Milford, MA, USA) in an LC20AD HPLC system (Shimadzu, Kyoto, Japan) for proteomic analysis.

The sample was loaded to the column equilibrated with buffers A (0.1% formic acid in water) and B (0.1% formic acid in acetonitrile). The Acclaim PepMap 100 C18 trap column (75 μm × 2 cm, Dionex) was equilibrated with liquid A by Ultimate 3000 nanoUPLC (Dionex), and the sample was eluted onto an Acclaim PepMap RSLC C18 analytical column (75 μm × 25 cm, Dionex) at a flow rate of 300 nl/min. The liquid-phase gradient was given as follows: 0–6 min, where the linear gradient of buffer B was from 2 to 10%; 7–51 min, where the linear gradient of buffer B was from 10 to 20%; 51–53 min, where the linear gradient of buffer B was from 20 to 80%; 53–57 min, where the linear gradient of buffer B solution was held at 80%.

Mass analysis was performed by nano-spray ionization-mass spectrometry (NSI-MS) and Q-Exactive HF mass spectrometry (Thermo Scientific). The intact peptide was detected by the Orbitrap, the scanning range was set to 250–1,500 m/z, the automatic gain control (AGC) target was 3E6, the resolution was 70,000, the max injection time (IT) was 250 ms, and the dynamic exclusion time was 15 s. The peptide was selected and fragmented for MS/MS using 28% NCE; ion fragments were detected in the Orbitrap, the resolution was 17,500, AGC was 1E5 or 5E4, and the maximum IT was 100 or 200 ms. LC-MS was performed by Micrometer Biotech (Hangzhou, China).

To improve the accuracy of protein identification, RNA-Seq data as mentioned earlier were used to construct specific reference protein databases using DIAMOND (version 2.0.15) (Buchfink et al., 2015). The MS files were searched against the sample-specific L. paraplantarum RX-8 database using MaxQuant (version 1.5.2.8) (Tyanova et al., 2016). The mass error of precursor and fragment ions was set as 10 ppm and 0.02 Da, respectively. Trypsin was selected for enzyme specificity and two missed cleavages were allowed. Carbamidomethylation on Cys, TMT 6-plex tag of Lys, and peptide N-terminal was specified as a fixed modification. The variable modifications were oxidation on Met and TMT 6-plex tag on Tyr. FDR was estimated by reverse-decoy strategy, percolator algorithm, and peptide–spectrum match (PSM). P < 0.05 and e < 0.05 were used to calculate the results. For quantitation, a protein must have at least two unique peptides with the above identity. The protein ratio type was median, and the normalization method was median. All the raw files were uploaded to PRIDE with the accession number PXD038315.

The proteins identified as differentially expressed proteins (DEPs) should satisfy the threshold of p ≤ 0.05 and |Fold Change|≥1.3. The enrichment analysis of DEPs used the clusterProfiler (version 3.4.4) for GO function annotation and KEGG pathways (Yu et al., 2012). The GO terms and KEGG pathways were regarded as significant protein enrichment annotations using the DEPs.

Compared with L. paraplantarum RX-8 in mono-culture, L. paraplantarum RX-8 in co-culture presented a much stronger carbohydrate utilization ability, which contributes to its great adaptability in competition with W. anomalus Y-5. In the process of carbohydrate transport and metabolism, the following metabolic pathways are involved in L. paraplantarum RX-8, with or without co-culture: PTS, fructose and mannose metabolism, galactose metabolism, glycolysis, TCA cycle, and amino sugar and nucleotide sugar metabolism.

The phosphotransferase system is responsible for the absorption of sugars, which depends on ATP and the 50 DEGs and 13 DEPs enriched in the PTS. In the PTS, the genes and proteins of mannose transport (manXaYZ) and galactitol transport (gatABC) are upregulated, while genes and proteins of fructose transport (FruA/B) are downregulated. In Lactobacillus pentosus, the mannose PTS EIIB participates in the inhibition of the fructose PTS (Chaillou et al., 2001). The mannose PTS in L. paraplantarum RX-8 may have a similar effect on inhibiting fructose transport (FruA/B). Aside from mannose, other carbohydrates, such as glucose, cellobiose, and maltose, could be transported by the mannose PTS. The sensing of exogenous AI-2 in Streptococcus pneumoniae is dependent on FruA; then, AI-2 promotes the transition of the pneumococcus from colonization to invasion by facilitating the utilization of galactose (Trappetti et al., 2017). In Escherichia coli, the internalization of AI-2 also depends on the PTS (Ha et al., 2018). AI-2 plays an important role in the inducing effect under a bacteria–bacteria co-culture (Chanos and Mygind, 2016; Liu et al., 2022), while AI-2's role in inducing this effect under fungi–bacteria co-culture needs further investigation. The proteins celA, celB, and celC, comprising the cellobiose transport system permease, take charge of the cellobiose utilization in L. paraplantarum RX-8, and are significantly upregulated, by 2.639-, 1.375-, and 2.71-fold, compared to that in mono-culture. However, the gene celA showed the opposite expression trend and the gene celC did not change. This indicates that post-transcriptional regulation may occur in L. paraplantarum RX-8. The gene crr encodes the sugar PTS EIIA component; when it is non-phosphorylated, Crr can inhibit the uptake of certain sugars, such as maltose, melibiose, lactose, and glycerol. However, when Crr is phosphorylated, it may activate adenylate cyclase (Saier and Roseman, 1976). The downregulation of the gene crr suggests that sugar uptake restrictions are lifted. Moreover, the mannose PTS EIIC acts as a target protein to confer class II bacteriocin sensitivity in L. monocytogenes, Enterococcus faecalis, and Latilactobacillus sakei (Kjos et al., 2009; Jeckelmann and Erni, 2020). The class IIa bacteriocin, immunity protein, and receptor proteins (mannose PTS EIIC and EIID) build up a tight complex to prevent pore formation and block cell death (Diep et al., 2007). However, the relationship between the mannose PTS and the immunity of class IIb bacteriocin produced by L. plantarum has not been reported to date (Kjos et al., 2010). Apart from its dual role in carbohydrate transport, the PTS is involved in the coupling of the sensory and regulatory mechanisms (Lengeler and Jahreis, 2009). The PTS transporter subunit IIB may be related to the effect of β-glucooligosaccharides on enhancing nisin production in L. lactis I2 (Lee et al., 2020). According to the results of transcriptomic and proteomic analyses, the PTS seems to be involved in the regulation of the response of L. paraplantarum to W. anomalus Y-5, leading to an increase in bacteriocin production.

Under co-culture conditions, the genes and proteins responsible for the utilization of cellobiose, lactose, galactose, fructose, sorbitol, galactitol, and ribose were also upregulated in L. paraplantarum RX-8. These sugars undergo glycolysis to pyruvate and ATP. Glucose-6-phosphate isomerase (GPI) was upregulated by 2.63-fold, which catalyzed the mutual conversion of glucose-6-phosphate and fructose-6-phosphate and helped sugars to enter the glycolysis pathway. The upregulated phosphoglycerate kinase (PGK), which directly generates ATP by phosphorylating the ADP, suggested that L. paraplantarum RX-8 produced more energy under co-culture. The pyruvate produced by glycolysis is a major metabolic intersection connecting the utilization and biosynthetic pathways of carbohydrates or amino acids; it linked the L-cysteine and carbon metabolism (Figure 5). After pyruvate dehydrogenase complex catalysis, pyruvate is oxidatively decarboxylated with the loss of one carbon and converted into acetyl-CoA, before entering the TCA cycle (Tian et al., 2022). The upregulation of DEPs in the TCA cycle enhances glucose oxidation and energy production. However, the increase in TCA decreased the transformation from phosphoenolpyruvate to aspartate, thereby weakening the serine synthesis. This leads to the downregulation of aspartate, serine, and L-cysteine metabolism.

Amino acids could serve as the materials in plantaricin synthesis. Aminoacyl-tRNA was responsible for the transportation of amino acids to ribosomes for incorporation into polypeptide chains of plantaricin. Under co-culture, the aspartate–ammonia ligase (asnA) and asparagine–tRNA ligase (asnS) increased by 2.02- and 2.35-fold, respectively. The aspartate–ammonia ligase catalyzed the production of L-asparagine; then, it was converted by asparaginyl–tRNA synthetase into L-asparaginyl–tRNA, which played a vital role in the translation of RNA into protein (Yeom et al., 2020). In addition, the enzymes that catalyzed the conversion of L-aspartate into fumarate were downregulated. The L-aspartate concentration in Lactobacillus casei increased when it faced acid resistance, and the transcriptional levels of argG and argH in L. casei increased with the decrease in pH (Wu et al., 2013). In our study, the pH of L. paraplantarum RX-8 in mono-culture was lower than that in co-culture, and a higher expression of argG and argH was detected in mono-culture.

At the same time, glutamate decarboxylase (gad), which was responsible for the mutual conversion of L-glutamate and γ-aminobutyrate (GABA), increased by 2.02-fold. The gad process converted L-glutamate into GABA, accompanied by the consumption of H⁺. This may explain why the pH of the co-culture was higher than that of the mono-culture of L. paraplantarum RX-8. Considering the upregulation of glutamate decarboxylase (gdhA), it is possible that L. paraplantarum RX-8 under co-culture produces more ammonia or proton consumption via glutamate decarboxylase (Teixeira et al., 2014). Moreover, the repression of GlnR could increase the expression of glutamine- and glutamate-metabolism-related genes (glnP, glnQ, amtB, glnK, gltB, and gad), promoting cell growth, acid resistance, and nisin yield (Miao et al., 2018). However, direct interactions between the anti41, an sRNA with a strong inhibitory effect on glnR, and the nisin gene cluster were not found (Miao et al., 2018). In addition, purine metabolism was weakened when the gene purF was downregulated, while pyrimidine metabolism was enhanced when the gene CAD was upregulated (Figure 5). The expression of carbamoyl–phosphate synthase (CAD) converted L-glutamine into carbamoyl phosphate, which is an intermediate in the biosynthesis of nucleotides through the orotate pathway (Namrak et al., 2022).

Meanwhile, the related arginine biosynthesis (argACDJGH) genes were downregulated, especially the genes controlling the transportation from glutamate to arginine (Figure 5). It was indicated that L. paraplantarum RX-8, under co-culture, led to the accumulation of glutamate by reducing arginine biosynthesis. The gene metC, which converted L-cysteine into L-methionine and was a precursor of S-ribosomal homocysteine (SRH), was downregulated. The gene luxS, a key enzyme, converted SRH into 4,5-dihydroxy-2,3-pentanedione (DPD). Then, DPD underwent spontaneous rearrangements to form AI-2. The downregulation of luxS influences the biosynthesis of AI-2, and a similar tendency was observed in co-culture AI-2 activity (Supplementary Figure 2). It was reported that the decrease in the L-methionine and L-cysteine cycling pathways was caused by the knockout of the luxS gene (Qian et al., 2022b). Similar results of changes in L-methionine- and L-cysteine-related genes were found in co-culture.

Some genes related to the quorum-sensing system and ABC-transport system were upregulated in L. paraplantarum RX-8 under co-culture. OppA is an oligopeptide-binding protein that is responsible for peptides' capture (Kieliszek et al., 2021). Therefore, the strong upregulation of oppA suggested an increase in the transportation of oligopeptides. OppA could increase Ox-bile resistance in L. salivarius Ren and salt resistance in Lactobacillus paracasei ATCC 334 (Ma et al., 2018). It is possible that OppA helps L. paraplantarum RX-8 to survive in the presence of yeast. When PlnA or Lactobacillus sanfranciscensis DPPMA174 is co-cultured with L. plantarum DC400, OppA is upregulated compared with L. plantarum DC400 mono-culture (Calasso et al., 2013). Although oppA did not show a difference in the plantaricin Q7 biosynthesis of L. plantarum Q7, oppD-, and oppF-encoding intracellular ATPases were upregulated when plantaricin Q7 production increased and downregulated when plantaricin Q7 was hydrolyzed to support the nutrients needed to promote the growth of L. plantarum Q7 (Bu et al., 2021a). Moreover, the Opp system in E. faecalis and Streptococcus mutans has specific pheromone binding activity (Leonard et al., 1996; Li and Tian, 2012). It can be hypothesized that L. plantarum DC400 transports PlnA via the Opp system (Calasso et al., 2013). The upregulation of phosphate-inducible transport system (pstSCAB) suggested that the phosphate uptake was increased to meet the increase in the phosphorylated nucleotides and proteins, which control the biosynthesis of many secondary metabolites. In addition, fructose, galactose, or mannose in media could promote the expression of the pstSCAB operon and significantly increase the formation of PstS (Martín and Liras, 2021). In this result, the mannose transport system and galactose metabolism were also upregulated.

The production of plantaricin was regulated by the quorum-sensing system (encoded by plnABCDEF gene clusters) (Zhao et al., 2021; Liu et al., 2022). Although plnABCDEF gene clusters synthesizing bacteriocin did not show significant differences (|Log₂FC|≥1, p < 0.05) in transcriptomic analysis, the plnABCDEF gene clusters were found to show 2.51-, 1.61-, 1.45-, 1.27-, 1.03-, and 1.41-fold changes at 24 h in transcriptomic analysis, and the results of RT-qPCR indicated that W. anomalus Y-5 stimulated plantaricin production via plnABCDEF. In the QS system, the gene agrA, an RR, was upregulated under co-culture. The Agr-mediated QS system (agrBDCA) might participate in commensal host–microbe interactions, especially agrA-encoded RR, which performs important functions in cell adhesion and biofilm formation in L. plantarum WCFS1 (Sturme et al., 2005). In addition, the expression of agrA was related to a global cellular response, such as the upregulated pyrimidine biosynthesis genes (pyrB, pyrD, pyrF, and pyrR) in L. plantarum WCFS1 (Sturme et al., 2005). Interestingly, similar results were also found for L. paraplantarum RX-8 in co-culture, suggesting that agrA controls pyrimidine biosynthesis under the interactions of L. paraplantarum RX-8 and W. anomalus Y-5. agrBDCA may influence plantaricin Q7 biosynthesis; specifically, agrC and agrA were upregulated in the synthesis of plantaricin Q7 (Bu et al., 2021a). The gene agrB of L. paraplantarum RX-8 was upregulated by 1.5-fold under co-culture. Protein AgrB also increased when L. plantarum DC400 was co-cultured with L. plantarum DPPMA20 or L. sanfranciscensis DPPMA174 (Calasso et al., 2013). This indicates that the agrBCDA quorum-sensing system might enhance plantaricin production when L. paraplantarum RX-8 is induced by W. anomalus Y-5.

To repair DNA damage, there are different classes of DNA repair systems. The nucleotide excision repair (uvrABCD) system recognizes the damaged strand, removes it, and then fills the gap (Goosen and Moolenaar, 2001). The gene uvrABCD was downregulated, which indicates that there is less DNA synthesis or less DNA damage when L. paraplantarum RX-8 is under co-culture. It is important for a tightly regulated RNA metabolism to adapt to environmental change and utilize nutrients. RNA helicases could locally unwind double-stranded RNA, or they can clamp protein complexes to a substrate RNA in an ATP-dependent manner. Gene cshA-encoded RNA helicase was downregulated in L. paraplantarum RX-8 under co-culture. It was reported that the cshA mutant of Staphylococcus aureus SA564 led to increased stability in agr mRNA, and it was hypothesized that the agrBDCA mRNAs were unable to correctly degrade in the absence of cshA (Oun et al., 2013). This would account for the lower cshA and higher agrA expression in L. paraplantarum RX-8 under co-culture. GroEL could partially unfold the misfolded RNA or proteins in an ATP-dependent manner. DnaK is a major chaperone, and the binding and release of protein clients by DnaK are regulated by ATP's hydrolytic activity. The downregulation of groEL and dnaK may indicate that there is less RNA degradation and a more stable RNA in co-culture.