Advantages of label free method in comparison with 2DE proteomic analysis of Butyrivibrio fibrisolvens 3071 grown on different carbon sources

Abstract The aim of this study was the comparison of label-free method with 2DE to other analytical method of bacterium Butyrivibrio fibrisolvens extracellular protein samples. Label-free quantification (LF) method performed in this work was compared with previously obtained results of proteomic analysis using two-dimensional electrophoresis (2DE) and nano-liquid chromatography coupled with mass spectroscopy (nLC/MS). B. fibrisolvens as an important plant fibre degrader was cultivated on four different carbon sources (including xylan, xylose, glucose, and a mixture of xylan with glucose). The impact of growth substrate on protein profile was assessed by six pair-wise comparisons evaluating the significantly differently abundant extracellular proteins. Gel-free and gel-based methods resulted in substantially dissimilar results. The LC-MS/MS approach detected substrate-dependent differences in transport and binding membrane proteins (TBP) and nucleotidase, while the 2DE approach detected substrate effect on proteins included in protein synthesis and butyrate synthesis. On the other hand, both methods observed differentially regulated proteins involved in the glycolytic pathway, however, the only shared enzyme (protein detected both by 2DE and LC-MS/MS approach) was fructose-bisphosphate aldolase. The LC-MS/MS approach detected a high abundance of separated peptides. However, it cannot be easily considered as supreme to 2DE analysis. Both methods differ in sample preparation, their advantages and limitations. The study findings indicate these methods can complement each other and together they elucidate better the metabolic functions of B. fibrisolvens. HIGHLIGHTS Comparison of LF and 2DE analysis showed limitations connected with a narrow optimised pH interval and number of affected proteins in 2DE in contrary to LF results. Butyrivibrio fibrisolvens culture showed the same proteomic changes in the presence of different saccharides as the whole rumen fluid in vivo. Selected proteins could be used for the monitoring of polysaccharides fermentation in the rumen and be used for optimisation hemicellulose degradation and overall feed utilisation.


Introduction
Ruminants are directly dependent on the activity of microbiota in their forestomachs. There is a wide range of rumen microbes, including bacteriophages, archaea, protozoa, anaerobic fungi, and most crucial bacteria. They play a key role in plant biomass degradation. Due to the rumen conversion of plant carbohydrates they can convert them into products with nutritional value and providing the host animal with energy (Owens and Basalan 2016). The understanding of the role and functioning of this complex microbial meta-metabolome action is essential for the animal's health, feed efficiency, and modulation of the microbiome to control methane emissions. In nature, hemicelluloses are the second most important plant polymers and they represent an important energy source for protein synthesis in the rumen (Hungate 1966). Over the last ten years, 'omics' methods have been widely used to characterise the mechanisms of plant cell wall degradation by ruminants (Kamke et al. 2016;Artegoitia et al. 2017;Li and Guan 2017;Denman et al. 2018;Guo et al. 2019;Li et al. 2019;Xue et al. 2020;). The majority of 'omics' technologies are based on nucleic acids analysis, while proteomic approaches to study the rumen environment has been applied only several gel-based studies (Deusch et al. 2017;Snelling and Wallace 2017;Hart et al. 2018). Recent meta-proteomic study (Honan and Greenwood 2020) have enlarged notably the knowledge and understanding of the active microbial metabolic pathways in the rumen and protein-mediated pathway dynamics. The meta-proteomic analysis of the whole rumen fluid is however complicated by the complexity of the sample. Development of optimised protein extraction protocols is a real challenge, if an enhanced amount of prokaryotic should be retrieved instead of plant-and bovine-derived cells only (Deusch and Seifert 2015). Proteomic studies of axenic cultures of rumen bacteria could elucidate in more detail the role of particular species in rumen metabolism and should be performed to complement, support, and verify meta-proteomic data. The greatest attention has been devoted to rumen bacterial strains with superior cellulolytic activity, such as Ruminococcus flavefaciens (Rincon et al. 2007;Vodovnik et al. 2013), Ruminococcus albus (Rakotoarivonina et al. 2009), Fibrobacter succinogenes (Burnet et al. 2015;Raut et al. 2015Raut et al. , 2019 and hemicellulolytic strains of Butyrivibrio species (Devillard et al. 2006;Kelly et al. 2010;Bond et al. 2012;Dunne et al. 2012Dunne et al. , 2015Sechovcov a et al. 2019). These bacteria were selectively specialised for the rumen niche (Mart ınez-Alvaro et al. 2020). Butyrivibria are quantitatively and functionally important members of the anaerobic gut microbiome of herbivores all around the world, as they belong to the seven most abundant rumen bacterial groups and are considered as part of a core heritable bacterial microbiome (Henderson et al. 2015;Wallace et al. 2019). Butyrivibria are supposed to be the dominant hemicellulolytic bacteria in the rumen producing a large and diverse spectrum of hydrolases degrading xylan (endo-1,4-xylanase, exo-1,4-oligoxylanase), glucuronoxylan (a-glucuronidase, endo-1,4-xylanase), arabinoxylan (xylosidase/arabinofuranosidase), glucomannan, and xyloglucan (Emerson and Weimer 2017;Moraïs and Mizrahi 2019;Palevich et al. 2019).Feed hemicelluloses are a part of the insoluble plant fibre. Microbial degrading enzymes, due to the compact plant cell-wall structure, approach them from soluble fraction (Kruger and den Haan 2022). Most species are able to utilise storage polysaccharides such as starch, inulin, and glycogen as carbon sources (Willems and Collins 2009) while cellulolytic activity is not a common feature (Palevich et al. 2019). Butyrivibrio importance is also due to their ability to butyrate production which is an energy source for the gut epithelial cells (Donohoe et al. 2011) and the host (Liu et al. 2018).
Up to date, proteomic studies of Butyrivibria have been based on two-dimensional gel electrophoresis (2DE) followed by mass spectrometry (MS), which is a well-established technique widely used to analyse protein mixtures of a broad range of biological samples. Over the last three decades, this method represented an essential tool for proteomic research in general (Domon and Aebersold 2006;Issaq and Veenstra 2008;Jorrin-Novo et al. 2019). Limitations of this approach are inability to resolve certain classes of proteins, reproducibility, and quantitation have been however addressed (Carbonara et al. 2021). Therefore, quantitative proteomic approach nLC-MS proteomics techniques have emerged as an alternative (Patel et al. 2009;Zhu et al. 2010;Neilson et al. 2011;Abdallah et al. 2012;Sch€ akermann et al. 2017;Toymentseva et al. 2020;Andersen et al. 2021). Thus, scientists are gradually switching to label-free (LF) shotgun proteomics technique as fast, rigorous, reliable and versatile tool. But this tool is often not able to quantify the changes in proteoform abundance, whereas 2DE can quantify changes in proteoform abundance. This is of extreme importance when trying to determine changes in phenotype (Carbonara et al. 2021).
In this study, we have applied label-free proteomic analysis using LC-MS/MS to elucidate the impact of four different carbon sources on the extracellular protein profile of B. fibrisolvens. The present work aimed mainly to compare this new approach with data retrieved previously by (Sechovcov a et al. 2019) using protein fractionation through two-dimensional gel electrophoresis (2DE). The analyses of the significantly differently abundant proteins of B. fibrisolvens cultured on simple as well as complex polysaccharides through two diverse strategies indicate that gel-based and gelfree methods can bring substantially dissimilar results, the differences were caused by the pH range of focussing stripes. The value range of pI IPG strips for IEF was selected according to optimisation method for Butyrivibrio fibrisolvens species.

Material and methods
This study was focussed on comparison of already published 2DE analysis and new results obtained with label-free LC-MS/MS method. Therefore, culture conditions and experiment arrangement are shortened (Sechovcov a et al. 2019).

Extracellular proteins preparation
The cell culture supernatants were obtained by centrifugation of 200 mL of respective cultivation media at 10,000 g for 20 minutes at 4 C. Supernatants were concentrated in a stirred ultrafiltration cell (Millipore) with Millipore PES membrane with a 10 kDa cut off at 4 C. These 6-fold concentrated extracellular enzyme extracts were immediately frozen and stored at À24 C until use. Further protein processing differed according to the type of analysis as described below.

Two-dimensional electrophoresis (2-DE)
Sample preparation, 2 DE-electrophoresis procedure, and statistical evaluation were performed according to (Sechovcov a et al. 2019). Briefly, proteins were precipitated for 1 h using 10% trichloroacetic acid (w/v) and pelleted by centrifugation (4 C, 7500 Â g for 15 min). Pellets were washed and incubated with acetonitrile (4 C for 1 h) and centrifuged (4 C, 7500 Â g for 15 min). After repeated washing pellets were dried (1 h at room temperature) and resuspended in lysis buffer (7 M urea, 4 M thiourea, 4% (w/v) CHAPS, 0.6% (w/v) Biolyt and 1% (w/v) DTT) to final protein concentration of 1 mg/mL. One hundred and forty lL of the sample was applied to a 7-cm IPG strip with a linear gradient of pH 4-7 (ReadyStrip TM IPG Strips, Bio-Rad). Isoelectric focussing (Protean IEF cell system, Bio-Rad), followed by second-dimension SDS-PAGE (Mini-Protean Tetra cell system, Bio-Rad), and staining with Bio-Safe Coomassie R-250 Staining Solution (Bio-Rad) was performed as described previously (Sechovcov a et al. 2019). Two gels were prepared for each growth condition (four different carbon sources) and the comparison was based on pair-wise matching (eight gels in total). Stained gels were scanned by GS-800 TM Calibrated Imaging Densitometer (Bio-Rad) and protein spots intensities were calculated with PDQuest TM software (version 8.0.1. Bio-Rad). Two-tailed Student t-test (p < 0.05) was used to assess statistical significance of protein abundance changes. Significantly different spots (p < 0.05) were excised from the gels, processed according to (Shevchenko et al. 2006) and subjected to digestion in a solution containing NH 4 HCO 3 (50 mmol/L) and trypsin (0.02 mg/mL) at 37 C for 16 h. Stage Tips (Empore TM Octadecyl C18, 47 mm Extraction Discs, Empore TM Supelco, Solid Phase Extraction Discs) were used for peptide purification according to (Rappsilber et al. 2007). Extracted solutions were lyophilised and dissolved in 20 lL of 2% formic acid (v/v) for consequent nLC MS/MS analysis.

Sample preparation for label-free LC-MS/MS method (LC-MS/MS)
The concentration of proteins in condensed extracellular protein mixture was determined by the method of Bradford (Sechovcov a et al. 2019) using bovine serum albumin as the standard. The protein concentration was determined using a Nanodrop ND-1000 spectrometer (ThermoFisher, Wilmington, DE). Samples were diluted to a concentration of 1 mg/mL. One mL of samples was lyophilised (Freeze drying, Nawah Scientific). Trypsin (0.2 mg/mL) in 0.5 M NH 4 HCO 3 was added to lyophilised sample in ratio 100:1 (trypsin/sample). Protein cleavage was performed at 37 C for 16 h with gentle shaking using Mixing Block MB-102 (Bioer, China). After cleavage, the samples were lyophilised. The lyophilised samples were diluted in 40 lL of 2% formic acid and sonicated for 10 minutes. Samples were subjected to Stage Tips analysis (Empore TM Octadecyl C18, 47 mm Extraction Discs, Empore TM Supelco, Solid Phase Extraction Discs) according to (Rappsilber et al. 2007).

Mass-spectrometry (MS)
Protein digests, resulting from both 2DE and LF sample preparation approach were analysed using a nanoliquid chromatography using a Proxeon Easy-nLC (Proxeon, Odense, Denmark) device coupled to a MaXis quadrupole time-of-flight (Q-TOF) mass spectrometer (Bruker Daltonics, Bremen, Germany). NS-AC-11-C18 Biosphere column with an NS-MP-10 Biosphere C18 precolumn (NanoSeparations, Netherlands) was used for separation of peptides under the conditions described by (O st' adal et al. 2015). The collected data were analysed with the software packages ProteinScape 3.0 (Bruker Daltonics, Bremen, Germany). Proteins were identified by correlating tandem mass spectra to the extracted database for Butyrivibrio fibrisolvens from the NCBI database using the MASCOT search engine v. 2.3.0 (http://www.matrixscience.com) as described previously (Sechovcov a et al. 2019). Only hits (Mascot score ! 80 for proteins; ! 20 for peptides, http://www.matrixscience.com) were accepted.

Label-free quantification
Label-free quantification was based on determination of the relative amount of proteins in two extracellular matrix (without stable isotopes). Our experiment was set up on the comparison of MS signal intensities of individual particular peptides at various sets of samples. A combination of nano-high-pressure liquid chromatography (HPLC) and a maXis (Q-TOF) ultrahighresolution mass spectrometer (Bruker Daltonics, Bremen, Germany) with a nanospray was used to analyse proteins cleaved by trypsin. The nLC-MS/MS was controlled with HyStar 3.2 software (Bruker Daltonics) and otofControl 3.4 (Bruker Daltonics). Data were acquired and processed using ProteinScape 3.0.0.446 and DataAnalysis 4.2 software (Bruker Daltonics). The prepared peptide mixture (20 lL) was injected per column (NS-AC À11-C18 Biosphere C18 column, particle size 5 lm, pore size 12 nm, length 20 mm, inner projection 75 lm, and with NS-MP À10 Biosphere C18 column, particle size 5 lm, pore size 12 nm, length 20 mm, inner projection 100 lm, NanoSeparation, The Netherlands).
Peptide separation was performed in a linear gradient between mobile phase A (water) and B (acetonitrile), with 0.1% (v/v) formic acid added to both phases. A constant flow rate of 0.25 mL/min was set and a constant temperature of 20 C was maintained. The separation was started with a 5% mobile phase B content, followed by a change in the elution gradient to 30% for 70 min. Then, the mobile phase ratio was changed to 50% for 10 minutes, followed by 100% mobile phase for 10 minutes. Finally, the column was eluted with 100% mobile phase B for 30 minutes. Equilibration before the next sample was performed by washing the column with 5% mobile phase B for 5 minutes. MS chromatogram was measured for each sample, and the MS/MS auto was turned off for this measurement for quantitative purposes (LF quantification). The collected data were processed using the Profile Analysis software. The mass spectrometer settings were the same as for the MS and MS/MS analyses (for LF quantification evaluation). Experiments were performed by scanning from 50 to 2200 m/z. The 'monocharged ion' C24H19F36N3O6P3 (m/z 1221.9906) was used as a reference ion (internal mass lock). For the MS/MS analyses, an automatic MS/MS with active exclusion (after 1 spectrum and release after 0.3 min) was set up. Nano-electrospray ionisation (nano-ESI) in positive mode was used for ionisation. The nano-ESI voltage was set at þ4.5 kV and the scan time at 1.3 Hz. The operating conditions for ionisation were: Drying gas N2, temperature 160 C and nebulisation pressure 0.4 bar (Kulhav a et al. 2020). The measurement of each MS chromatogram was used with the switched off auto MS/MS mode. The peptide composition values were obtained by analysis of collected data using Profile Analysis software v. 2.1 (Bruker Daltonics GmbH) and ProteinScape software (Bruker). Each peptide must be found in all of samples in each group (comparison of each substrate), it was fulfilled criterion for statistical processing. Individual values of intensities for the given peptides (obtained from MS data) in the compared samples were processed statistically using Student's T-test (p ˂ 0.05). The sample compilation (MS/MS data) was created for each substrate comparison and only significantly differently expressed peptides were taken into consideration (Kulhav a et al. 2020).
Database searches were performed as described with the taxonomy restricted to Butyrivibrio fibrisolvens from the NCBI database (downloaded on 26th February 2018; 33,312 proteins), using the MASCOT search engine v. 2.3.0 (http://www.matrixscience.com) to remove protein identification redundancy.

Results
The present study aimed to compare and evaluate two common proteomics methods, namely label-free LC-MS/MS and two-dimensional electrophoresis (2DE) methods, to identify the significant differences in protein profiles of B. fibrisolvens 3071 cultivated on four different carbon sources. Substrates were compared to each other resulting in the following six comparisons: glucose vs. xylan þ glucose (I vs. II), glucose vs. xylan (I vs. III), glucose vs. xylose (I vs. IV), xylan þ glucose vs. xylan (II vs. III), xylan þ glucose vs. xylose (II vs. IV), xylan vs. xylose (III vs. IV). Using LF method, 18 valid proteins were found in substrate comparison I vs. II (glucose vs. xylan) and at least one significantly altered peptide were found in 7 proteins. The 16 valid proteins were found in substrate comparison I vs. III (glucose vs. xylan þ glucose) and at least one significantly altered peptide were found in 3 proteins. The 14 valid proteins were found in substrate comparison I vs. IV (glucose vs. xylose) and at least one significantly altered peptide were found in 5 proteins. The 20 valid proteins were found in substrate comparison II vs. III (xylan vs. xylan þ glucose) and at least one significantly altered peptide were found in 3 proteins. The 22 valid proteins were found in substrate comparison II vs. IV (xylan vs. xylose) and at least one significantly altered peptide were found in 2 proteins. The 20 valid proteins were found in substrate comparison III vs. IV (xylan vs. glucose þ xylan) and no protein was found to have at least one significantly altered peptide. The significant proteins are shown in Table 1. Numbers of valid peptides, significant changes in comparison, proteins assignment, and mascot scores of LF analysis are shown in Supplementary Table S1.
Thus, 20 proteins in LF analysis fulfilled the MASCOT Score Criteria and were found to be differentially abundant depending on the used carbon source. This outcome was compared with the conclusions of the 2-DE approach (Sechovcov a et al. 2019) and the comparison is summarised in Table 2. Surprisingly, the two proteomic methods applied on the same set of samples resulted in different outcomes, and no match among proteins in the same substrate comparison was found. Fructose-bisphosphate aldolase, class II, was the only enzyme detected by both methods as significantly increased, but never in the same substrate comparison ( Table 2).
The LC-MS/MS analysis identified significant changes in proteins of three functional groups. Nine peptides belonged to the cluster of transport and binding membrane proteins (TBP) or their precursors (green colour in Table 1). Eight of them belonged to an extracellular solute-binding protein, and three unspecified peptides similar to ChvE 1 protein. Six proteins were involved in glycolysis (blue colour in Table  1) and 3 proteins were fragments of a nucleotidase (red colour in Table 1). TBP were affected in The following substrates were compared: (I) glucose/xylan, (II) glucose/xylan þ glucose, (III) glucose/xylose, (IV) xylan/xylan þ glucose, (V) xylan/xylose, (VI) xylan þ glucose/xylose. R -Range, the difference intensities (peptides fulfilling the criteria of the T-test (p < 0.05)) between maximum and minimum values. The specific peptides that fulfilled these conditions are listed in the Supplement S1. RI -Average value of the first comparative group. RII -Average value of the second comparative group. " -Indicates the increase by the appropriate substrate.
comparisons of all substrates. Nucleotidase was either increased when B. fibrisolvens Bf 3071 was cultivated either with a more complex substrate or not affected. Enzymes of glycolytic group (GP) were effected in xylan medium, where phosphoenolpyruvate carboxykinase was increased, while fructose-bisphosphate aldolase was decreased. The LC-MS/MS analysis did not detect any proteins included in protein synthesis and butyrate synthesis, which is contrary to the results of the 2DE proteomic approach of (Sechovcov a et al. 2019). On the other hand, both methods identified differentially regulated protein involved in the glycolytic pathway, however the only shared enzyme (protein detected both by 2DE and LF approach) was fructosebisphosphate aldolase (Figure 1). The proteins detected by 2DE are highlighted by blue colour, green colour was used for mark proteins obtained by LF and red colour indicated enzyme confirmed in both methods.

Discussion
In this work, same protocols were used for preparation of extracellular proteins of B. fibrisolvens grown on different substrates and two distinct protocols were used to process identical proteome samples. This bacterium is known for its important hemicellulolytic activities (Sewell et al. 1988;Lin and Thomson 1991;Dalrymple et al. 1999). Therefore we focussed our interest on the effect of different substrates, including xylan and xylose, on the expression of extracellular proteins (Kruger and den Haan 2022). The gel-based proteomics approach using a simple way of protein pre-fractionation through 2DE was compared LF liquid chromatography and high-resolution mass spectrometry.
Protein separation by electrophoretic methods (2DE) and subsequent identification by mass spectrometry (MS) or tandem mass spectrometry (MS/MS) represents a proteomic method, which has many advantages but also several drawbacks such as proteoform level quantification (L opez 2007;Ning et al. 2016;Schaffer et al. 2019;Carbonara et al. 2021). The biggest advantage lie in the possibility to separate proteoforms based on different pI and molecular weight. This step enables the separation of proteins with posttranslational modifications, e.g. glycosylation, phosphorylation, which play a fundamental role in protein activity (Ujcikova et al. 2016). The other advantage of Arabinogalactan oligomer / maltooligosaccharide transport system substrate-binding protein pI 3,5 Phosphoenolpyruvate carboxykinase (ATP) pI 4,9 5 0 -nucleotidase pI 4,5 Putative multiple sugar transport system substrate-binding protein pI 3,9 Fructose-bisphosphate aldolase, class II pI 5,2 Listeria/Bacteroides repeat-containing protein, partial pI 4,1 Formate C-acetyltransferase pI 5,4 I vs. III Glucose-6-phosphate isomerase pI 5,1 Putative multiple sugar transport system substrate-binding protein pI 3,9 N-acetylmuramoyl-L-alanine amidase pI 5,2 5 0 -nucleotidase pI 4,5 Phosphoglycerate kinase pI 4,7 Sugar ABC transporter substrate-binding protein pI 3,5 1-phosphofructokinase pI 5,2 I vs. IV Beta-hydroxybutyryl-CoA dehydrogenase pI 5,7 Putative multiple sugar transport system substrate-binding protein pI 3,9 5 0 -nucleotidase pI 4,5 Sugar ABC transporter substrate-binding protein pI 3,5 Phosphoenolpyruvate carboxykinase (ATP) pI 4,9 Fructose-bisphosphate aldolase, class II pI 5,2 II vs. III Fructose-bisphosphate aldolase, class II pI 5,2 Putative multiple sugar transport system substrate-binding protein pI 3,9 sugar ABC transporter substrate-binding protein pI 3, gel-based approach is an application of steps removing extracellular polysaccharides (EPS). Butyrivibrio strains produce viscous EPS (Ha et al. 1991), which can interact with proteins (and also DNA). TCA protein precipitation (Rajalingam et al. 2009) and rinsing with acetonitrile and acetone removes polysaccharide residues and other substances, which are mostly insoluble in water, but soluble in some organic solvents (El Seoud et al. 2013), which results in higher purity of a sample. On the other hand, spots obtained from 2D gel could contain more than one protein and thus the quantification becomes ambiguous. Further limitations of gel-methods include limited dynamic range, determination of proteins with extreme molecular weights (low or high) and pI values, problems associated with hydrophobic proteins, quantitative reproducibility, and limitations on the ability to study certain classes of proteins (Bunai and Yamane 2005;L opez 2007;Wu et al. 2007;Issaq and Veenstra 2008;Mosley et al. 2011;Marcus and Rabilloud 2020).
The important outcome of this study is the dissimilarity of results achieved by these two different proteomic approaches. In our opinion, this disparity can be attributed primarily to the different sample preparation, mainly to the precipitation of proteins with TCA, washing with acetonitrile, application of lysis solution, application of IPG strip with a linear gradient of pH 4-7 and isoelectric focussing during the sample preparation for 2DE analysis, which steps, however, have not been used in the sample preparation for LC-MS/ MS quantification. The pI values of proteins obtained by the 2DE method ranged from 4.8 to 6.0, which was the result of method optimisation (Sechovcov a et al. 2019). On the other hand, the pI of proteins identified by the LC-MS/MS method ranged from 3.5 to 5.2. The differences of proteome coverage between the gelbased and gel-free approaches are contrary to the results of (Patel et al. 2009) who compared several proteomic approaches on bacterium Methyllocella and found satisfactory results of their comparison. In their study was used 1 D SDS PAGE, only. Whereas in our study was used 2DE. However other studies (e.g. Li et al. 2012or Megger et al. 2014) described differences depending on the used method showing LF proteomics to be superior to others. However, the mentioned studies used very distinct type of samples and also methodological procedures were different. The evaluation of this finding is difficult due to the limited number of proteomic methods comparative studies on bacteria and the non-existence of similar works on rumen bacteria.
The LC-MS/MS method identified significantly changed peptides of three membrane proteins. The first was a Putative multiple sugar-binding protein, which exhibits 45% amino acid similarity with periplasmic receptor ChvE precursor involved in bacteria chemotaxis and attachment sugar ABC transporter substrate-binding protein (Petrovicheva et al. 2017), and is a key determinant of the substrate specificity in the high affinity of ABC uptake systems in bacteria and usually recognise a broad range of sugars (Maqbool et al. 2015).
Annotation revealed the second protein detected in this study as an arabinogalactan oligomer/maltooligosaccharide transport system, however this protein was always increased in abundance in the presence of glucose, which could indicate a possible role in the glucose transport. The protein is widespread in many Butyrivibrio isolates with similarity over 95% and is similar to substrate-binding protein in Eubacterium ruminantium by 10%.
The third protein sequence, described in Table 1 as a partial Listeria/Bacteroides repeat-containing protein, was further identified as Internalin-J with 83% amino acid (AA) similarity to B. fibrisolvens Inl-J precursor. These proteins are members of the protein family harbouring leucine-rich repeats known as internalins (Maia et al. 2019). Internalin was identified as an invasion protein of intracellular pathogen Listeria monocytogenes (Braun et al. 1997) and its action was described by Bleym€ uller et al. (2016). Originally, internalin sequences were mentioned to be present in other pathogenic bacteria (Bierne et al. 2007), but have been identified also in non-pathogenic bacteria, such as Enterococcus faecalis (Brinster et al. 2007), or Lactobacillus plantarum (Siezen et al. 2006). Our work brings the first indication, that this type of protein can be present in B. fibrisolvens.
All 3 peptides assigned to fragments of the same nucleosidase with only 13% AA similarity to Trifunctional nucleotide phosphodiesterase protein YfkN, therefore have to be considered as unspecified enzyme. YfkN phosphodiesterase exhibits 2 0 ,3 0 cyclic nucleotide phosphodiesterase, 2 0 (or 3 0 ) nucleotidase and 5 0 nucleotidase activities. The multifunctional activity of YfkN was studied mainly in Bacillus subtilis suggesting that this enzyme plays a role in the cellular reprocessing of nucleotide phosphates present in the culture medium. This hypothesis was proposed based on the increase in YfkN production when the bacteria were grown in phosphate restricted medium (Chambert et al. 2003). Study of Nguyen et al. (2011) identified YfkN as a surface protein, which can be anchored on the cell wall by the B. subtilis sortases (prokaryotic enzymes that modify surface proteins by recognising and cleaving a carboxyl-terminal sorting signal). In pathogenic bacteria, the surface proteins promote interaction between the pathogen and animal tissues, but can also provide ingenious strategies for bacterial escape from the host's immune response (Mazmanian et al. 2001).
Furthermore, the LC-MS/MS analysis identified 4 significantly differently expressed proteins involved in the glycolytic pathway, which is in agreement with the results of the 2DE analysis (Sechovcov a et al. 2019). However, each method detected different types of enzymes (see Table 2, Figure 1), and only fructosebisphosphate aldolase (class II) was found by both proteomic approaches (Figure 1). This enzyme takes a central position in glycolysis and gluconeogenesis pathways and has been found in high transcript numbers in the transcriptomic study of Wirth et al. (2018) and detected in the whole rumen fluid proteomic studies (Snelling and Wallace 2017;Hart et al. 2018;Honan and Greenwood 2020). This indicates the core functional position of this enzyme in the active metabolic system of the rumen. Wirth and colleagues (Wirth et al. 2018) also described pyruvate phosphate dikinase (PPdK) and phosphoenolpyruvate carboxykinase ATP (PEPCK). PPdk and PEPCK are the two most frequently identified sequences in the metatranscriptomic profile of cattle rumen fluid. Both PPdK and PEPC enzymes are a part of phosphoenolpyruvate (PEP)-pyruvate-oxaloacetate node (POP-node), which is a meeting point of glycolysis and metabolite synthesis (Koendjbiharie et al. 2021). PPdK is implied to be gluconeogenic and PEPCK catalyses reversible decarboxylation and phosphorylation (using CO 2 instead of HCO 3 -) between phosphoenolpyruvate and oxaloacetate. Both are highly significant enzymes in the rumen and were confirmed by LC-MS/MS analysis of B. fibrisolvens performed in this study as proteins increased in abundance. However, they were not detected by 2DE approach in the previous study of Sechovcov a et al. (2019). On the other hand, gel-based proteomic work of Sechovcov a et al. (2019) detected upregulated glyceraldehyde phosphate dehydrogenase, what was not observed with LC-MS/MS approach, but described by (Wirth et al. 2018) as an important part of core transcripts. In general, proteomic approaches involve four steps (Deracinois et al. 2013). For this study, the first step was 'Bacterial strain culture condition and extracellular proteins isolation'. This step was the same for both methods. The second step, and a crucial one, was 'Sample preparation'. The results dissimilarity was caused by sample preparation procedure. The 'in-gel' approach, used in the 2DE method, depended on optimised range of pI and condition of electrophoresis; the 'off-gel' approach, used in LF quantification, was based on chromatography. The third step was 'Separation'. And the last was 'Protein identification'. The last two steps did not play such a key role in obtaining results in this case. Although the 'in-gel' method used a protein-based approach and the 'offgel' method used a peptide-based approach. Regardless of the proteomic approach used, the identification of proteins or peptides was always done via MS. Basically, these are two methods with a different principle for sample analysis. It was these different principles that led to the different results in the observed bacterium. From these findings can be derived that neither method (2DE vs LC-MS/MS) is better nor worse. On the contrary, both methods complement each other and create a more complete picture of the important metabolic role of B. fibrisolvens in the polysaccharide fermentation process.

Conclusions
Comparison of two proteomic methods of Butyrivibrio fibrisolvens revealed advantages of LF method which provided more proteoforms than 2DE gel-based method. The results showed that hemicellulose presence significantly increased enzymes of phosphoenolpyruvate (PEP)-pyruvate-oxaloacetate node connected with gluconeogenesis and phosphoenolpyruvate metabolism. There were significantly phosphoenolpyruvate carboxykinase (4.1.1.49), pyruvate phosphate dikinase (2.7.9.1) and formate C-acetyltransferase (2.3.1.54). The changes could be connected with location of xylan utilising enzymes with phosphoenolpyruvate:sugar phosphotransferase system similar to Butyrivibrio proteoclasticus fructans breakdown . This is in agreement with a wide upregulation of several membrane sugar transport systems.
Proteomics, an important 'omics' tool complementary to genomics and transcriptomics, brings new information and detailed insight into the understanding of the mechanisms of degradation of plant cell wall polysaccharides by rumen bacteria. The application of conventional gel-based proteomics method has rapidly increased in recent years, however, the gel-free approach in this field is still in its infancy. Our study dealing with the influence of carbon source on the differentially expressed protein profile of B. fibrisolvens compares gel-based and gel-free methods and highlights the differences in obtained results. Each technique had dissimilar coverage of proteins indicating the importance of sample preparation and method optimisation with respect to pI values. Each approach has its advantages and drawbacks, and it seems that the results of both methods can complement each other.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Availability of data and materials
We are ready to share our data.