Using the published protocol (Lai et al., 2016), A. baumannii SK17-S and SK17-R were grown to the mid-exponential phase (OD_600nm = 0.8) before extraction of proteins. Extracted proteins were then subjected to in-solution tryptic digestion to yield dried peptide fragments as described before (Lai et al., 2016). Peptides were re-dissolved in NETN buffer (100 mM NaCl, 50 mM Tris-HCl, 1 mM EDTA, 0.5% w/v NP-40, pH 8.0). In parallel, 100 μl of the anti-acetyllysine antibody (Okanishi et al., 2013) at the concentration of 1 mg/ml were conjugated to 100 μg of protein G beads. The conjugated protein G beads were added into the peptide solution and incubated overnight at 4°C with gentle shaking. The beads were washed once with 1 ml ETN buffer (100 mM NaCl, 50 mM Tris-HCl, 1 mM EDTA, pH 8.0) and twice with 1 ml NETN buffer. The enriched peptides were eluted with 0.1% trifluoroacetic acid solution (400 μl × 3), and the eluent were passed through C18 Sep-Pak column before vacuum drying. Finally, ZipTip containing C18 reversed-phase media were used to desalt and concentrate peptides, which were then subjected to nano-scale liquid chromatography with tandem MS (nanoLC-MS/MS) analysis as described before (Lai et al., 2016).

All MS and MS/MS raw data were processed with Proteome Discoverer version 1.4 (Thermo Scientific), and the peptides were identified from the MS/MS data searched using the MASCOT search engine version 2.4.1 (Matrix Science) in combination with Sequest HT to search against the A. baumannii SK17 database (Lai et al., 2016). For each experiment, two raw files of MS/MS spectra generated from analyses of each protein fraction were merged into a single contiguous input file to facilitate bioinformatics analysis. The combined data files were searched by using the protein databases with the following parameters: a precursor mass within ±5 ppm of the theoretical mass, a fragment ion tolerance of <0.6 Da, and allowed a maximum of two missed cleavages sites for trypsin. Carbamidomethylation on cysteine was specified as a fixed modification, whereas oxidation on methionine and acetylation on lysine were set as variable modifications. The maximum false discovery rate (FDR) threshold of peptides and proteins was set to 0.05, for those FDR <0.01 were indicated with additional marks. Charge states of +2, +3, and +4 were considered for parent ions. Only rank-one peptides were counted for filtering the output data set. Peptides containing acetyllysine residue(s) with MASCOT score >20.0 were manually verified as described previously (Chen et al., 2005). Bioinformatics analysis of the identified peptides was carried out as described before (Lai et al., 2016).

The evolutionary history was inferred by using the Maximum Likelihood method based on the James-Taylor-Thornton matrix-based model. The percentage of trees in which the associated taxa clustered together is shown next to the branches. Initial tree(s) for the heuristic search were obtained automatically by applying Neighbor-Join (NJ) and BioNJ algorithms to a matrix of pairwise distances estimated using a James-Taylor-Thornton model, and then selecting the topology with superior log likelihood value. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. The analysis involved 33 amino acid sequences. All positions containing gaps and missing data were eliminated. There was a total of 90 positions in the final dataset. Evolutionary analysis was conducted in MEGA7 (Kumar et al., 2016).

Amino acid sequence of A. baumannii HU (Swiss-Prot/TrEMBL accession number B0VNW6) was used to build all homology models by SWISS-MODEL (Biasini et al., 2014; Bienert et al., 2017) or I-TASSER (Yang et al., 2015). For SWISS-MODEL, pdb 1P51 (Anabaena HU), and 4P3V (E. coli Huβ) were used as the templates to generate homology models with and without DNA bound to the protein, respectively. For I-TASSER, default setting was applied, and the generated model has C-score = 0.74, estimated TM-score = 0.81 ± 0.09, and estimated RMSD = 2.4 ± 1.8 Å. The models were visualized with PyMOL (Schrödinger).

Chromosomal DNA from both A. baumannii SK17-S and SK17-R were isolated by phenol extraction (Chan and Goodwin, 1995). The hupB gene was PCR amplified from the chromosome DNA. Sanger sequencing confirmed that the hupB gene in both SK17-S and SK17-R strains has identical sequence. The hupB gene was then cloned into pET-21a(+) vector (Novagen) between the NdeI and XhoI sites to provide a plasmid that expresses wild-type HU with a C-terminal His-tag. The plasmid, pET-21a hupB, was then PCR amplified using the two primers: TTCAGCAATTGCATCGATTAATTCTGATTTATTCATATG and TAATCGATGCAATTGCTGAAtagGGGGGAGTATCTAAGACTGATGCAG. These primers replace the original lysine codon (AAA) with an amber stop codon (TAG) for site-specific incorporation of acetyllysine. The PCR product (5,657 bp) was transformed into chemically competent E. coli cells to afford plasmid expressing HU(K13ac).

To construct a plasmid that expresses wild-type HU without a His-tag, a 342 bp PCR amplimer generated using primers TGAGCGATTACGACATCCCCACTACTGAGAATCTTTATTTTCAGGGCGCCATGAATAAATCAGAATTAATCGATGC and GCTTTGTTAGCAGCCGGATCTCAAGCAACTGAATCTTTAAGACC from pET-21a hupB was joined by Gibson assembly to a 5,397 bp PCR amplimer generated using primers AGTGGGGATGTCGTAATCGCTCATGGGGTGATGGTGATGGTGATGTTTCATATGTATATCTCCTTCTTAAAGTTAAAC and GCTTGAGATCCGGCTGCTAACAAAGCCCGAAAGGAAG from pET-21a hupB. This construct contains a N-terminal His-tag, followed by a PMSDYDIPTT linker and a TEV protease cleavage sequence (ENLYFQG).

To construct a plasmid that expresses acetyllysyl-tRNA synthetase/tRNA_CUA pair, the plasmids pCDF PylST (Neumann et al., 2010) and pBK AcKRS (Neumann et al., 2009), gifts from Jason Chin, were digested with BamHI and Eco147I, and the digested vector (4,630 bp) from pCDF PylST and the digested insert (1,382 bp) from pBK AcKRS were purified by gel extraction, followed by ligation to afford the plasmid pCDF AcKST.

To generate the wild-type protein with or without His-tag, E. coli BL21(DE3) cells containing the appropriate plasmid were grown at 37°C in terrific broth (TB) medium containing ampicillin (100 μg/ml). Fresh TB medium containing ampicillin (100 μg/ml) was inoculated with 2% v/v overnight culture and was incubated at 37°C. At OD₆₀₀ 0.6, isopropyl β-D-thiogalactopyranoside (IPTG) was added into the culture to a final concentration of 0.5 mM to induce gene expression at 20°C for 18 h.

To express the acetylated protein, E. coli BL21(DE3) cells containing pET21a hupB(K13TAG) and pCDF AcKST were grown at 37°C in TB medium containing ampicillin (50 μg/ml) and spectinomycin (25 μg/ml). Fresh TB medium containing ampicillin (50 μg/ml) and spectinomycin (25 μg/ml) was inoculated with 2% v/v overnight culture and was incubated at 37°C. At OD₆₀₀ 0.5, the culture was supplemented with nicotinamide (final concentration = 20 mM) and N_ε-acetyl-L-lysine (final concentration = 5 mM). Gene expression was induced 30 min later by the addition of IPTG and incubation at 20°C for 18 h.

Cells were harvested by centrifugation at 6,000 × g at 4°C for 30 min, re-suspended in lysis buffer (500 mM NaCl, 25 mM Tris-HCl, 20 mM imidazole, pH 7.5) containing lysozyme (1 mg/ml) and phenylmethanesulfonyl fluoride (100 μM), and lysed by sonication. Cell debris was removed by centrifugation at 39,000 × g for 30 min. The clear supernatant containing the hexahistidine-tagged proteins was purified using a Ni-NTA column, washed with 50 mM immidazole and eluted in 250 mM imidazole. Fractions containing the desired protein were identified by SDS-PAGE analysis and pooled for further treatment.

For wild-type and K13ac HU proteins containing a C-terminus His-tag, the pooled fractions were subjected to cation-exchange chromatography on a NGC Chromatography System (Bio-Rad). Resource S column (GE Healthcare, #17-1180-01) was equilibrated with potassium phosphate buffer (10 mM, pH 7.0, 5% v/v gycerol). Sample was loaded on the column and eluted using the same buffer with an increasing amount of NaCl (0–500 mM).

To generate wild-type HU without the His-tag, the protein purified by Ni-NTA column was exchanged into Tris buffer (25 mM, pH 8.0) to a concentration of 1–2 mg/ml. The protein was incubated with recombinant His-tagged TEV protease (10% w/w) in the presence of 5 mM DTT for 18 h at 20°C. The sample was loaded onto a Ni-NTA column, and the flow through was collected.

After purification, proteins were concentrated by Amicon stirred cell using 3 kDa cutoff membrane (Millipore, #PLBC04310). Protein concentration was determined by BCA assay. Protein yields were 15–20 mg/l for wild-type HU with a C-terminal His-tag, 1–1.5 mg/l for K13ac HU with a C-terminal His-tag, and 12–15 mg/l for wild-type HU without His-tag. For storage, 20% v/v glycerol was added into the buffer, and proteins were kept at −80°C.

Size-exclusion chromatography was performed on a NGC Chromatography System (Bio-Rad). Superdex 75 10/300 GL (24 ml, GE Healthcare) was equilibrated with PBS. Sample was then loaded on the column, followed by elution with PBS at a flow rate of 0.5 ml/min. Fractions were collected and analyzed by SDS-PAGE to confirm the peak corresponding to HU protein.

CD experiments were performed on an Applied PhotoPhysics Chirascan spectrometer using 11 μM protein in deoxygenated potassium phosphate buffer (10 mM, pH 7.0). Spectra were measured between 200 and 400 nm in 1 mm quartz cuvettes under N₂ with a 50 nm/min scan speed, 0.5 nm data pitch, 1 nm bandwidth, and 0.5 s response time. To measure the thermal melting point, CD spectra were collected every 2°C as temperature increased from 5 to 85°C. The rate of temperature increase was 0.5°C/min with 300 s equilibration time at each temperature.

DNA (250 ng) was incubated with varied concentrations of protein at 37°C for 1 h in binding buffer in a 10 μl reaction volume (Ghosh et al., 2016). Following incubation, the samples were mixed with 5x GelPilot DNA Loading Dye (Qiagen) and loaded on 1% agarose gel containing SYBR Safe (Thermo Fisher) in 1x TAE buffer (40 mM Tris-acetate and 1 mM EDTA pH 8.0). The gel was electrophoresed at 130 V at 20°C for 45 min, visualized and analyzed using a ChemiDoc XRS+ System (Bio-Rad).

Plasmids pBK AcKRS (3,219 bp) and pCX eGFP (5,514 bp) were chosen due to their small size, and they only have identical sequence around the origin of replication (725 bp). Restriction enzyme BamHI (for pBK AcKRS) or BcuI (for pCX eGFP) was used to generate linear DNA. The plasmids were gifts from Jason Chin (Neumann et al., 2009) and Anthony Perry (Perry et al., 1999), respectively.

All the biolayer interferometry experiments were performed on an Octet RED96 System (Pall ForteBio). Samples in buffer were dispensed into polypropylene 96-well black flat-bottom plates (#655209, Greiner Bio-One) at a total volume of 200 μl per well, and all measurements were performed at 25°C with agitation at 1,000 rpm. Streptavidin-coated biosensor tips (#18-5019, Pall ForteBio) were pre-wetted with assay buffer (30 mM Tris, pH 7.5, 100 mM NaCl, 0.02% v/v Tween20, 0.5 mg/ml BSA). Biosensors were then loaded with biotin-labeled dsDNA (5′-biotin-ATTAATATTTTCTTAAACTAATTTAAAAT) by incubation at the concentration of 0.1 μM for 300 s, and excess DNA were removed by incubation with assay buffer for 300 s. To measure association and dissociation kinetics, biosensors were transferred into fresh assay buffer for 300 s to collect a baseline read, followed by incubation with the protein at a specific concentration for 300 s to record protein association and incubation with fresh assay buffer for 300 s to record protein dissociation. Measurements were performed from low to high protein concentration. Tips were regenerated between each measurement by repeatedly incubation in regeneration buffer (10 mM glycine, pH 2.0, 20 seconds) and fresh assay buffer (20 s) for three times.

Data were analyzed using ForteBio Data Analysis Software version 8.1. The binding profile of each sample was recorded as an “nm shift” (the wavelength or spectral shift in nanometers), which represented the difference between the start and end of the kinetic cycle. All sensorgrams were referenced for buffer effects, and the Y axes were aligned to the association step with an inter-step correction of aligning to the dissociation step. Kinetic responses were fitted to a 1:1 Langmuir binding model to obtain values for association (k_on), dissociation (k_off) rate constants, and the equilibrium dissociation constant (K_D). All fitted curves have R² > 0.90.

Sample solutions contained linear pCX eGFP DNA (by treatment with BcuI) at 10 ng/μl and/or the designated HU protein (3 μM) in buffer (10 mM HEPES, pH 7.5, 4 mM MgCl₂). This protein: DNA ratio equals to the presence of 1 HU homodimer for every 10 bp DNA, if all proteins bind. The solution was either incubated at 37°C for 1 h or without incubation. Ten microliters of sample solution was applied to freshly cleaved mica by spin coating (15 s ramp 0–2,000 rpm; 30 s at 2,000 rpm), followed by washing with water (2 ml, full wetting with intervals at 3,000 rpm) and drying (4 min at 3,000 rpm).

Atomic force microscopy measurements were carried out in air at 293 K, using a Nanoscope V instrument (Veeco), type Multimode 8. Probes PPP-NCHR POINTPROBE-PLUS® Silicon-SPM-Sensor were used (resonance frequency = 330 kHz; force constant = 42 N/m, length = 125 μm, NanosensorsTM) operating in tapping mode at a frequency around 321.5 kHz to image the surfaces through ScanAsyst®-air mode. The data were processed using the WS × M software (Horcas et al., 2007).

Tier 3 measurements, as defined by the National Cancer Institute's Clinical Proteomic Tumor Analysis Consortium (CPTAC) Program, were performed here. Acetylated peptides from whole cell lysates of A. baumannii SK17-S and SK17-R were fractionated and enriched using immunoaffinity enrichment strategies (Okanishi et al., 2013). Each strain was analyzed by high-accuracy nanoflow LC-MS/MS technology. Yields for the recombinant protein expression were based on three independent experiments. The same pools of proteins were subjected to biochemical characterizations. For analytical size-exclusion chromatography, identical samples were run twice on the instrument. For electrophoretic mobility shift assay, a representative gel from duplicates was shown for each condition. For biolayer interferometry analysis, three independent measurements were carried out for each protein. Mean and standard deviation calculated from the three experiments are presented.

Two clinical isolates of A. baumannii, SK17-S and SK17-R (Chen et al., 2010), were used in this study. Strain SK17-S isolated first from a male hospitalized patient is susceptible to antibiotic imipenem, whereas strain SK17-R isolated later from the same patient is resistant to this antibiotic. To investigate the acetylomes of these two strains, cells were harvested at mid-exponential phase, and a polyclonal anti-acetyllysine antibody (Okanishi et al., 2013) was employed to enrich acetylated peptides from the whole-cell protein tryptic digests before analysis by high-resolution mass spectrometry (MS).

In total, we identified 145 unique lysine acetylation sites in 132 unique acetylated peptides from 125 proteins that have diverse functions and localize in different part of the cells (Tables S1–S4). Among these, there are 103 sites on 88 proteins in SK17-S and 67 sites on 60 proteins in SK17-R, including 25 sites on 22 proteins in both strains (Figures 1A,B). It is noteworthy that only four identical acetylation sites were previously reported in A. baumannii ATCC19606 harvested at stationary phase (Kentache et al., 2016), including DNA-binding protein HU (Table S1). Furthermore, most of the identified proteins locate in cytoplasm (Figure 1C). From the functional perspective, 77 of the 125 identified proteins could be functionally annotated to metabolism of nucleotide, amino acid, protein, carbohydrate, or energy (Figure 1D). This finding is consistent with the previous studies of bacterial acetylome (Ouidir et al., 2016), suggesting a close link between protein acetylation and metabolic processes.

Lysine acetylation on DNA-binding protein HU is of particular interest because of its potential to change the biophysical properties and consequently the function of this protein which is known to regulate cell growth and pathogenicity (Lee et al., 2013; Okanishi et al., 2013; Zhang et al., 2013; Castano-Cerezo et al., 2014; Pan et al., 2014; Kosono et al., 2015; Ouidir et al., 2015; Xie et al., 2015; Ghosh et al., 2016; Kentache et al., 2016). MS/MS spectrum of peptide fragment SELIDAIAEKacGGVSK (corresponding to residues 4–18 of HU) clearly showed acetylation at Lys13 (Figure 2A). This acetylated peptide was found in both A. baumannii SK17-S and SK17-R, indicating that Lys13 acetylation is a common phenomenon in this organism. Molecular phylogenetic analysis showed that A. baumannii HU protein is closely related to homologs in other bacteria (Figure S1), and sequence alignment revealed that Lys13 is conserved in many HU homologs (Figure 2B). In fact, Lys13 acetylation (K13ac) was found in A. baumannii ATCC19606 (Kentache et al., 2016), E. coli K12 (Zhang et al., 2013) and BW25113 (Castano-Cerezo et al., 2014), M. tuberculosis H37Ra (Xie et al., 2015; Ghosh et al., 2016), and Vibrio parahaemolyticus (Pan et al., 2014; Figure 2B and Figure S2). The presence of K13ac in both SK17-S and SK17-R as well as other pathogenic bacteria indicates this site-specific modification may play critical role in bacterial growth or infection. In addition, bacterial HU has analogous function to eukaryotic histones on DNA compaction (Grove, 2011), so HU acetylation may affect DNA binding and serve as epigenetic modulator.

Since no crystal structure of A. baumannii HU is available, SWISS-MODEL (Biasini et al., 2014; Bienert et al., 2017) and I-TASSER (Yang et al., 2015) were used to generate homology models. SWISS-MODEL was used to build the homodimeric structure of the 90-amino-acid A. baumannii HU with or without DNA (Figure 2C). The DNA-bound structure is based on the 94-amino-acid Anabaena HU that shares 56% sequence identity and 72% similarity, whereas the DNA-free structure is based on the 90-amino-acid E. coli HUβ that shares 70% sequence identity and 83% similarity. On the other hand, I-TASSER generated a monomeric structure that aligned well to the subunit of DNA-bound structure produced by SWISS-MODEL (Figure S3). In fact, the only discrepancy among the three models is between the amino acid residue 55 and 73, the region involving in DNA binding. The homology model indicated that Lys13 does not directly interact with the DNA. Such a phenomenon is commonly observed in eukaryotic histone proteins (Shahbazian and Grunstein, 2007), where most histone acetylation sites locate in the N-terminal tails that do not directly interact with DNA and are accessible in nucleosome complexes. Acetylation at those sites can regulate nucleosome assembly, DNA accessibility by other proteins and gene transcription in eukaryotes (Shahbazian and Grunstein, 2007). Therefore, we decided to investigate if Lys13 acetylation changes the properties and/or function of A. baumannii HU.

To pinpoint the effect of Lys13 acetylation, the technique of genetic code expansion was used to produce homogenously acetylated protein, HU(K13ac) that has acetyllysine at the amino acid residue 13. Site-specific incorporation of acetyllysine was achieved through an orthogonal acetyllysyl-tRNA synthetase/tRNA_CUA pair (Neumann et al., 2009). A His tag was introduced at the C-terminus to facilitate purification (Figure S4). HU and HU(K13ac) were overexpressed in E. coli BL21(DE3), purified by Ni-NTA column and cation-exchange chromatography. Yield of 15–20 and 1–1.5 mg/l was obtained for the wild-type and K13ac HU, respectively.

SDS-PAGE analysis of the purified proteins showed bands at the desired size (Figure 3A), and MS spectra confirmed the incorporation of acetyllysine (Figure 3B and Figure S4). Although there has been no prior experimental study of A. baumannii HU, bacterial HU proteins characterized to date normally exist in a dimeric form (Grove, 2011). However, it is also known that two point mutations of E. coli Huα can transform the dimeric protein into an octamer (Kar et al., 2006). Thus, we applied analytical size-exclusion chromatography to determine the oligomeric states of A. baumannii HU. Using Superdex 75 10/300 GL column, both wild-type and K13ac HU migrate between the 44-kDa ovalbumin and 17-kDa myoglobin (Figure 3C) marker with the estimated molecular weight of about 25 kDa (Figure S5), indicating that A. baumannii HU exists as a dimer and Lys13 acetylation does not change the oligomeric state. Lastly, the effect of Lys13 acetylation on protein secondary structure and thermal stability was investigated by circular dichroism (CD). At 20°C, there is a small change of the CD spectra around 209 nm (Figure 3D) likely due to acetylation at Lys13 that localizes on an α-helix (Figure 2C). Indeed, the overall CD spectra indicate that both proteins exist as mainly α-helix, which is in accordance with the homology models (Figure 2C). On the other hand, by measuring CD spectra at different temperature, we were able to calculate the melting temperature and found that Lys13 acetylation lowers the melting temperature by 4°C (Figure 3E). It is likely that either acetylation renders the protein intrinsically less stable or destabilizes the dimeric complex of HU, consequently decrease the protein thermal stability.

HU homologs are known to undergo sequence-independent DNA binding and regulation (Grove, 2011). Thus, electrophoretic mobility shift assay was performed to investigate if Lys13 acetylation affects DNA binding. A linear double-stranded DNA (dsDNA) from BamHI-digested plasmid, pBK AcKRS (Neumann et al., 2009), was employed as the binding partner. Mixture containing different amount of recombinant proteins was subjected to agarose gel electrophoresis. The binding of protein to DNA leads to a complex with larger molecular weight than that of DNA itself, hence a slower shift in mobility. Decrease of DNA mobility was observed in samples with higher amount of protein (Figure 4A). However, both wild-type and K13ac proteins showed similar degree of change in DNA mobility. In addition, this phenomenon is independent of DNA sequence or shape and composition of binding buffer (Figure S6). Notably, whereas high salt level (>300 mM NaCl) abolishes binding of E. coli HU to DNA (Xiao et al., 2010), the binding of A. baumannii HU to DNA remained up to 800 mM NaCl (Figure S6F).

To rule out the possibility that the observed DNA binding is due to the presence of C-terminal His-tag (Ceci et al., 2005), wild-type HU protein without a His-tag was also prepared. This protein was expressed with a TEV cleavage sequence between the HU sequence and N-terminal His-tag. SDS-PAGE and MS were used to monitor and confirm the TEV cleavage (Figure S4). This protein showed comparable DNA-binding ability to that of the C-terminally His-tagged wild-type HU (Figure 4A); accordingly, we concluded that binding of DNA by wt and K13ac HU is unlikely due to the His-tag.

To quantify the protein-DNA interaction, we used biolayer interferometry to study the binding kinetics (i.e. association and dissociation rate constants, k_on and k_off) and the dissociation equilibrium constant K_D. In the sensorgrams (Figures 4B–D), we noticed profound difference in the signal intensity between wild-type and K13ac proteins. Nevertheless, the K_D values calculated from experimental triplicate showed negligible difference among the modified and unmodified proteins (Table S5). However, both k_on and k_off values of HU(K13ac) are close to an order of magnitude smaller than those of the wild-type. Such difference in k_on likely explains the slow signal increase for the HU(K13ac) sample in comparison to the wild-type protein in the sensorgrams (Figures 4C,D). Thus, it is likely that Lys13 acetylation of free HU slows the step of DNA complexation, whereas acetylation of HU-DNA complex decreases the dissociation of DNA from the complex.

To examine the effect of acetylation on DNA architectural organization, atomic force microscopy was applied to visualize DNA-protein complexes at the stoichiometry of 10 bp/HU dimer. At this ratio, if all proteins bind to DNA, there will be one HU homodimer for every 10 bp DNA. This is the ratio we observed most profound interaction between DNA and HU proteins on electrophoretic mobility shift assay (Figure 4A and Figure S6). Samples in buffer solution (Ghosh et al., 2016) with or without pre-incubation at 37°C were deposited on freshly cleaved mica by spin coating, followed by imaging with the tapping mode (Figures S7–S15). Without pre-incubation, DNA seems to form bundles (Figure S8), and there is no significant morphology difference in samples containing HU proteins (Figures S9–S11). On the other hand, pre-incubating the sample solution at 37°C before depositing on mica resulted in well-separated DNA strands (Figure S12), and DNA strands appear significantly higher and wider in samples containing either wild-type or K13ac HU (Figures S13–S15), indicating binding of HU proteins to DNA strands. Nevertheless, DNA does not form compact aggregates with HU proteins, suggesting limited compaction ability of HU proteins under the experimental conditions.

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. The reviewer XDL and handling Editor declared their shared affiliation.