The L. casei Zhang descendent strain (L. casei Zhang-A-600) was more resistant to amoxicillin resistance with a minimum inhibitory concentration (MIC) of 8 μg/mL (vs. 2 μg/mL for the parental strain; Wang et al., 2017). The resistant strain was isolated from long-term propagation of the parental strain in lactic acid bacteria (LAB) susceptibility test medium broth (LSM), consisting of 90% Iso-sensitest medium (IST; OXOID, CM0473) and 10% MRS (Klare et al., 2005), supplemented with 0.5 μg/mL amoxicillin (Wang et al., 2017). At this antibiotics concentration, the bacterial growth was suppressed by 50%. The growth of bacteria (optical density, pH, and viable counts) was monitored every 2 h (from 0 to 30 h of cultivation). All experiments were performed in triplicate.

To minimize the interfering effect between the antimicrobial compound and the growth medium components (Klare et al., 2005), the minimal growth medium, LSM, was chosen for the current study. For proteomic analysis, the parental and descendent L. casei Zhang cells were collected after 24 h of cultivation in amoxicillin-containing LSM (0.5 μg/mL of amoxicillin). In each case, 4 biological replicates of samples were prepared. The culture conditions used here were the same as our previous study which aimed to characterize the adaptation of the resistant strain at the genomic level (Wang et al., 2017). Briefly, the bacterial cells were pelleted by centrifugation and washed 4 times with phosphate buffered saline (PBS). One milliliter of lysis buffer (7 M urea, 4% SDS, and 1x protease inhibitor cocktail) was added to each sample. The mixtures were then sonicated on ice and spun at 13,000 rpm for 10 min at 4°C. The supernatant of each sample was separately collected.

The protein concentration of the supernatants was estimated by the bicinchoninic acid protein assay. One hundred microgram of protein of each sample was adjusted to a final volume of 100 μL with 100 mM triethylammonium bicarbonate (TEAB), followed by adding 5 μL of 200 mM DTT and incubating at 55°C for 1 h. Afterwards, 5 μL of iodoacetamide (375 mM) was added to each sample, followed by 30 min incubation in dark at room temperature. Then the protein was precipitated with ice-cold acetone and redissolved in TEAB (20 μL). Proteins were digested with sequence-grade modified trypsin (Promega, Madison, WI) and labeled using the iTRAQ reagents kit. The labeled samples were combined, desalted (Sep-Pak C18 SPE column, Waters, Milford, MA), and vacuum dried.

Phase separation was performed as described by Gilar with some modifications (Gilar et al., 2005). Briefly, the peptide mixture was redissolved in buffer A (buffer A: 10 mM ammonium formate in water, pH 10.0, adjusted with ammonium hydroxide). The dissolved peptide mixtures were then fractionated by high pH separation using the Aquity UPLC system (Waters Corporation, Milford, MA) connected to a reverse phase column (BEH C18 column, 2.1 × 150 mm, 1.7 μm, 300 Å, Waters Corporation, Milford, MA). A linear gradient, starting from 0% B to 45% B in 35 min (B: 10 mM ammonium formate in 90% ACN, pH 10.0, adjusted with ammonium hydroxide), was used in the high pH separation. The column flow rate and temperature were maintained at 250 μL/min and 45°C, respectively. Sixteen fractions were separately collected and dried in a vacuum concentrator.

The fractions were redissolved in a solvent composed of solvents C and D (C: 0.1% formic acid in water; D: 0.1% formic acid in ACN), separated by nano LC and analyzed by on-line electrospray tandem mass spectrometry. The experiments were performed on a Nano Aquity UPLC system (Waters Corporation, Milford, MA) connected to a quadrupole-Orbitrap mass spectrometer (Q-Exactive) (Thermo Fisher Scientific, Bremen, Germany) with an online nano-electrospray ion source. Eight microliters of each peptide sample were loaded onto the trap column (Thermo Scientific Acclaim PepMap C18, 100 μm × 2 cm), with a flow of 10 μl/min for 3 min, to be separated on a 75 μm × 25 cm Acclaim PepMap C18 analytical column. A linear gradient, from 5% D to 30% D in 95 min, was used. The column was re-equilibrated at initial conditions for 15 min. The column flow rate and temperature were maintained at 300 μL/min and 45°C, respectively. An electrospray voltage of 2.0 kV was applied to the mass spectrometer inlet.

The Q-Exactive mass spectrometer was operated in the data-dependent mode, switching automatically between MS and MS/MS acquisition. Survey full-scan MS spectra (m/z 350–1,600) of mass resolution of 70K were acquired, followed by 15 sequential higher-energy collisional dissociation (HCD) MS/MS scans of 17.5K resolution. In all cases, one 30-s dynamic exclusion micro-scan was recorded. The MS/MS fixed first mass was set to 100.

Tandem mass spectra were extracted by ProteoWizard (version 3.0.5126; Thermo Fisher Scientific) using the Proteome Discoverer software (version 1.4.0.288; Thermo Fisher Scientific). All MS/MS samples were analyzed using Mascot (version 2.3; Matrix Science, London, UK), which was used to search the NCBI database (Taxonomy: Lactobacillus casei Zhang, 2804 entries; trypsin digestion; 0.050 Da fragment ion mass tolerance and 10.0 PPM parent ion tolerance). Moreover, in the Mascot search, cysteine carbamidomethylation and iTRAQ 8plex of lysine and the n-terminus were opted as fixed modifications, while methionine oxidation and iTRAQ 8plex of tyrosine were specified as variable modifications.

Statistical analyses were performed following the recommendations of Predrag Radivojac and Olga Vitek (Radivojac and Vitek, 2012). The percolator algorithm of <1% was used to control the false discovery rate. Only unique peptides were quantified. Experimental biases were corrected by normalization with the protein median. The minimum number of observed proteins was 1000. Statistical analysis was performed under the R environment (Student's t-tests, p < 0.05 was considered statistically significant). A cut-off level of 2.0-fold change was applied to select differentially expressed proteins; and only those showing a consistent expression change in all 4 biological replicates were considered as differentially expressed proteins. They were functionally assigned by the clusters of orthologous genes (COGs) and the Kyoto Encyclopedia of Genes and Genomes databases (Tatusov et al., 1997).

Two genes, LCAZH_0490 and LCAZH_0521, were selected as target candidates to be genetically disrupted. They putatively encoded an OmpR family DNA-binding response regulator and a surface-associated protein, respectively. The plasmids and primers used for constructing the gene disruption mutants are listed in Table 1.

The gene disruption mutants were constructed using a cre-lox-based system (Lambert et al., 2007). To disrupt the LCAZH_0490 gene, the upstream and downstream flanking regions of the LCAZH_0490 gene were amplified by PCR by two primer pairs (0490upF and 0490upR; 0490downF and 0490downR) using the genomic DNA of L. casei Zhang-A-600 as template. These fragments were then cloned between the Xho I or Pme I and Ecol53 KI or Bgl II restriction sites of the suicide vector pNZ5319 to create the recombinant mutagenesis vector, pNZ5319-0490 Up-Down, which was introduced into L. casei Zhang-A-600 by electroporation. Colonies harboring the anticipated inserts in the desired orientation were identified by the combined use of the primers 85, 87, and an insert-specific primer. Chloramphenicol-resistant transformants were selected and replica plated to check for erythromycin-sensitive clones. Candidate double-crossover mutant clones were first analyzed by PCR (with the primer pairs, catR and catF, EryintF, and EryintR), followed by verifying the correct integration of the 0490::lox66-P32-cat-lox71 cassette into the genome using the primer combination of 0490upF or catR and catF or 0490downR. In order to excise the P32-cat selectable marker cassette, the cre expression plasmid, pMSPcre, was transformed into the 0490::lox66-P32-cat-lox71 gene replacement mutant. Erythromycin-resistant and chloramphenicol-sensitive colonies were selected by another round of replica plating. The cre-mediated recombination and correct excision of the P32-cat cassette were confirmed by PCR using primers spanning the recombination locus (0490upF and 0490downR). The pMSPcre vector was cured from L. casei Zhang-A-600 Δ0490 colonies by growth without erythromycin selection pressure. This plasmid was constructed from pMSP3535 (provided by Professor Jian Kong, Shandong University, unpublished). Similar procedures were used to inactivate the LCAZH_0521 gene in L. casei Zhang-A-600.

The amoxicillin-resistant phenotype of the gene disrupted mutants were evaluated by determining their MICs of amoxicillin (Guo et al., 2017). Moreover, the OD value at the time of observing the MIC by witness was determined to quantify the growth performance of the wild-type and mutant strains. The Student's t-test at a confidence level of 0.05 was used to evaluate any difference in the growth performance between different strains.

The gene disruption plasmids were designed and constructed based on the genome sequence of L. casei Zhang (GenBank database accession number: CP001084.2). The sequences of gene disruption regions of the mutant strains, Zhang-A-600-delta-0490 and Zhang-A-600-delta-0521, have been deposited to the GenBank database under the accession numbers MG021089 and MG021090.

The growth characteristics (viable counts, pH, and OD values) of the parental L. casei Zhang strain and the amoxicillin-resistant Zhang-A-600 descendent were monitored. The growth performance of L. casei Zhang-A-600 was better than that of the parental strain when amoxicillin was present in the culture medium (Figures 1A–C). Both the maximum viable count and cell density of L. casei Zhang-A-600 were significantly higher (P < 0.05) than that of the parental strain, which were 6.3 × 10⁸ CFU/mL and 1.36, respectively. The L. casei Zhang-A-600 culture entered stationary phase at 12 h (vs. 10 h for the parental strain; Figure 1A). Moreover, the pH of the L. casei Zhang-A-600 culture dropped much faster than that of the parental strain (Figure 1B).

When L. casei Zhang-A-600 was grown in the presence of amoxicillin, the expression of 38 proteins significantly increased (>2.0-folds, P < 0.05) compared with the parental cells (Table 2). Most of these proteins could be assigned to specific COG functional categories.

Six of the highly expressed proteins (15.8%) were involved in amino acid transport and metabolism (E), including a transporter component for oligopeptides (LCAZH_0339), an amino acid ABC transporter substrate-binding protein (LCAZH_0419), as well as a set of proteins responsible for amino acid transport (LCAZH_1957–LCAZH_1960). Five other highly expressed proteins (12.8%) were associated with carbohydrate transport and metabolism (G), namely the PTS system cellobiose-specific transporter subunit IIC (LCAZH_0295), the PTS system fructose-specific transporter subunit IIABC (LCAZH_0393), a trehalose-6-phosphate hydrolase (LCAZH_2653), and a transaldolase (LCAZH_2725).

Some of the highly expressed proteins fell into the COG classes T and M, which were connected with cellular stress response. These included the OmpR family DNA-binding response regulator (LCAZH_0490), a signal transduction histidine kinase (LCAZH_0491), a conjugated bile salt hydrolase-like amidase (LCAZH_0447), a nucleoside-diphosphate-sugar epimerase (LCAZH_0562), and a cyclopropane fatty acid synthase-like methyltransferase (LCAZH_2067). The functions of several other differentially expressed proteins were unknown.

In contrast to the spectrum of highly expressed proteins, the majority of the lowly expressed proteins (35.3%) belonged to the COG class G (Table 3). Among them, 4 putative proteins, namely the PTS system cellobiose-specific transporter subunits IIA and IIB (LCAZH_2637, LCAZH_2638), a triosephosphate isomerase (LCAZH_2697), and a fructose/tagatose bisphosphate aldolase (LCAZH_2698), were encoded by genes located within an operon-like structure in the L. casei Zhang genome. Two other lowly expressed proteins were assigned to the COG class V, i.e., the antimicrobial peptide ABC transporter permease and ATPase (LCAZH_1927, LCAZH_1928).

The target knockout genes, LCAZH_0490 and LCAZH_0521, were selected because of their predicted molecular functions in stress response. No significant difference (P > 0.05) was observed in the MICs of amoxicillin between the gene disruption mutants and the wild-type, L. casei Zhang-A-600. However, the mutants, particularly Δ0490 that lacked the response regulator, grew slower than L. casei Zhang-A-600 in the presence of amoxicillin. As shown in Figure 2, the OD of the L. casei Zhang-A-600 culture was 1.31-fold higher (P < 0.05) than that of the mutant Δ0490 cultivated in LSM with 4 μg/mL amoxicillin.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.