Proteomic Analysis of Bifidobacteriumlongum subsp.
`infantis Reveals the Metabolic Insight on Consumption
`of Prebiotics and Host Glycans
`
`Jae-Han Kim1, Hyun Joo An2, Daniel Garrido3,4, J. Bruce German4,5, Carlito B. Lebrilla4,6,
`David A. Mills3,4,5*
`
`1 Department of Food Nutrition, Chungnam National University, Yuseong-gu, Daejeon, Korea, 2 Graduate School of Analytical Science and Technology, Chungnam
`National University, Yuseong-gu, Daejeon, Korea, 3 Department of Viticulture and Enology, University of California Davis, Davis, California, United States of America,
`4 Foods for Health Institute, University of California Davis, Davis, California, United States of America, 5 Department of Food Science and Technology, University of
`California Davis, Davis, California, United States of America, 6 Department of Chemistry, University of California Davis, Davis, California, United States of America
`
`Abstract
`
`Bifidobacterium longum subsp. infantis is a common member of the intestinal microbiota in breast-fed infants and capable
`of metabolizing human milk oligosaccharides (HMO). To investigate the bacterial response to different prebiotics, we
`analyzed both cell wall associated and whole cell proteins in B. infantis. Proteins were identified by LC-MS/MS followed by
`comparative proteomics to deduce the protein localization within the cell. Enzymes involved in the metabolism of lactose,
`glucose, galactooligosaccharides, fructooligosaccharides and HMO were constitutively expressed exhibiting less than two-
`fold change regardless of the sugar used. In contrast, enzymes in N-Acetylglucosamine and sucrose catabolism were
`induced by HMO and fructans, respectively. Galactose-metabolizing enzymes phosphoglucomutase, UDP-glucose 4-
`epimerase and UTP glucose-1-P uridylytransferase were expressed constitutively, while galactokinase and galactose-1-
`phosphate uridylyltransferase, increased their expression three fold when HMO and lactose were used as substrates for cell
`growth. Cell wall-associated proteomics also revealed ATP-dependent sugar transport systems associated with consumption
`of different prebiotics. In addition, the expression of 16 glycosyl hydrolases revealed the complete metabolic route for each
`substrate. Mucin, which possesses O-glycans that are structurally similar to HMO did not induced the expression of transport
`proteins, hydrolysis or sugar metabolic pathway indicating B. infantis do not utilize these glycoconjugates.
`
`Citation: Kim J-H, An HJ, Garrido D, German JB, Lebrilla CB, et al. (2013) Proteomic Analysis of Bifidobacterium longum subsp. infantis Reveals the Metabolic
`Insight on Consumption of Prebiotics and Host Glycans. PLoS ONE 8(2): e57535. doi:10.1371/journal.pone.0057535
`
`Editor: Valerie de Cre´ cy-Lagard, University of Florida, United States of America
`
`Received August 21, 2012; Accepted January 25, 2013; Published February 26, 2013
`Copyright: ß 2013 Kim et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted
`use, distribution, and reproduction in any medium, provided the original author and source are credited.
`
`Funding: This work was supported in part by the University of California Discovery Grant Program, the California Dairy Research Foundation, and National
`Institute of Health Awards R01HD061923 (C.B.L.) and R01AT007079 (D.A.M.). The research was also supported by the Converging Research Center Program
`through the Ministry of Education, Science and Technology (2011K000985 for J.H.K.). The funders had no role in study design, data collection and analysis,
`decision to publish, or preparation of the manuscript.
`
`Competing Interests: The authors have declared that no competing interests exist.
`
`* E-mail: damills@ucdavis.edu
`
`Introduction
`
`infantis)
`infantis (B.
`Bifidobacterium longum subsp.
`is a common
`member of the gastrointestinal tract (GIT) of breast-fed infants [1].
`The establishment of a bifidobacterial-dominant microbiota has
`received considerable attention regarding the development of the
`infant [2]. One potential means by which specific bifidobacteria
`succeed in colonizing the infant GIT is
`through prebiotic
`enrichment by human milk oligosaccharides (HMO). Human
`milk contains a significant amount of oligosaccharides (,15 g l21),
`in contrast to bovine or formula milk [3,4]. HMO is a term that
`collectively refers to approximately 200 different types of glycans
`with diverse structures with a length smaller than 20 units of
`carbohydrates [5,6]. These HMO are composed of hexoses (Hex)
`and N-Acetylhexosamines (HexNAc) connected through b1-3 or
`b1-4-glycosidic linkage with additional decoration of fucose (Fuc)
`and N-Acetylneuraminic acid (NeuAc). Since the human host does
`not have the enzymatic capacity to degrade these polymers, HMO
`are not considered to be metabolized by infant, and have been
`shown to arrive to the lower GIT [7]. HMO are considered part of
`
`the innate immune system being used as decoy binding sites for
`intestinal pathogens in the developing newborn [8]. Another role
`is the enrichment of bifidobacteria in the GIT [9]. While B. infantis
`and other bifidobacteria can actively utilize HMO as growth
`substrate, other intestinal bacteria such as streptococci, entero-
`cocci, E. coli and Clostridium sp. cannot use HMO as a carbon
`source, emphasizing the bifidogenic potential of HMO [10].
`Genome sequencing showed that B.
`infantis has a 40 kb gene
`cluster, termed HMO cluster I, that encodes several enzymes and
`transport systems required for HMO catabolism including a-
`fucosidase, a-sialidase, b-hexosaminidase, b-galactosidase and the
`ABC transport systems with six family 1, extracellular solute
`binding proteins (SBPs) predicted to bind oligosaccharides [11].
`These clusters of HMO-linked genes suggested a broader evolu-
`tionary partnership among B. infantis, the developing infant and
`human milk [12].
`While current commercial prebiotics used in infant formulas,
`such as fructooligosaccharides (FOS) and galactooligosaccharides
`[13], have been shown to be bifidogenic,
`these commercial
`prebiotics do not replicate the many different roles that HMO
`
`PLOS ONE | www.plosone.org
`
`1
`
`February 2013 | Volume 8 |
`
`Issue 2 | e57535
`
`Genome & Co. v. Univ. of Chicago, PGR2019-00002
`UNIV. CHICAGO EX. 2001 - 1/13
`
`

`

`carry out in the developing infant. Since B. infantis displays several
`mechanistic preferences for HMO consumption, this subspecies
`represents a useful reference with which to assess the similarity of
`commercial prebiotics to actual HMO. This in turn will aid in the
`design of more effective prebiotics for use in infant formulas as well
`as other conditions where very selective symbiotic applications
`may be useful.
`Global expression analysis of whole cell proteome by LC-MS/
`MS can provide direct information relative to biological systems
`such as metabolic pathways and regulatory networks. Despite the
`advantages of whole cell proteomics,
`it is still challenging to
`reliably identify cell wall associated (CWA) proteins due to the fact
`that these proteins are often sequestered into the insoluble material
`upon cell lysis. The lack of CWA proteins in proteomic analyses
`results in significant loss of biological information since membrane
`proteins or cell wall-anchored proteins also play important roles in
`transport, exopolysaccharide production, hydrolysis of macromo-
`lecules, and sensing of extracellular signals. Several proteomic
`methods have been developed to specifically identify the CWA
`proteins, including a 2-D gel based proteomic analysis followed by
`fractionation [14,15,16], or labeling the cell surface proteins via
`biotinylation followed by affinity capture [17,18]. An alternative
`approach is the tryptic digestion of exposed epitopes on cell surface
`proteins from intact cells in isotonic solution followed by the MS
`analysis, which often called ‘‘shaving and shedding’’ [19,20].
`In this study, we have investigated the impact of various
`prebiotics on the proteomic expression profiles of B. infantis ATCC
`15697. The microorganism was cultured with seven different
`sugars and prebiotics including HMO and mucin, and more than
`500 proteins were quantitatively identified enabling us to explore
`the entire activity cell
`in a given substrate. Importantly, we
`developed a proteomic analysis method that allowed reliable
`determination of cell wall associated proteins. Through this
`method,
`the expression of cell
`surface/membrane associated
`proteins can be determined, maximizing the total protein
`identification per cell. This has allowed a more comprehensive
`comparison of the B. infantis physiology on these prebiotics.
`
`Materials and Methods
`
`Cell Culture
`Bifidobacterium longum subsp. infantis ATCC 15697 was obtained
`from the American Type Culture Collection (Manassas, VA).
`Cultures were routinely maintained in de Mann-Rogosa-Sharp
`(MRS) medium with no carbon source, and supplemented with
`0.05% w/v L-cysteine (Sigma-Aldrich, St. Louis, MO), and 2%
`w/v of either lactose (Sigma Aldrich, MO), purified HMO [21],
`mucin from porcine stomach type III (Sigma Aldrich, MO), FOS
`(raftilose Synergy 1, Orafti, Malvern, PA), inulin (raftiline HP,
`Orafti, Malvern, PA) or GOS (Purimune, GTC Nutrition, Golden,
`CO). These experiments were performed in duplicates. Cells were
`anaerobically grown in a vinyl chamber
`(Coy Laboratory
`Products, Grass Lake, MI) at 37uC for 24 h, in an atmosphere
`consisting of 5% carbon dioxide, 5% hydrogen, and 90% nitrogen.
`Optical density was assayed using a PowerWave microplate
`spectrophotometer (BioTek Instruments, Inc., Winoosky, VT).
`Due to the significant high turbidity of medium, cell growth on
`mucin was evaluated separately. Mucin was autoclaved for 10
`minutes and added on the media (50 ml) at a final concentration of
`20 g l21. Cell growth was monitored by measuring optical density
`of cell at 600 nm in a Shimadzu UV1601 spectrophotometer
`(Shimadzu Scientific Instruments, Columbia MD). Cell cultures
`supplemented with 2% (w/v) of glucose and 2% (w/v) of HMO
`were obtained in similar conditions as control experiments.
`
`Whole Cell Proteomics of B. infants
`
`Proteomic Sample Preparation
`B. infantis cells were taken at the exponential phase of growth
`and normalized at an OD of 1.0 by dilution or concentration.
`After centrifugation to remove the spent media, 15 mL of cells
`were washed three times with phosphate buffer saline (PBS)
`followed by three more washes with urea buffer (8 M Urea/0.1 M
`Tris, pH 8.5). The cell pellet was resuspended in 600 mL of urea
`buffer, then mechanically disrupted by silica beads in a bead-
`beater (FastPrep; QBiogene, Carlsbad, CA, USA) for eight cycles
`of 30 s pulses and 30 s on ice. Upon centrifugation, the soluble
`and the insoluble fractions were stored separately at 280uC until
`further analysis.
`For proteome analysis of the soluble fraction, a sample volume
`containing 200 mg of protein was transferred to a new micro-
`centrifuge tube and precipitated by the addition of ethanol (75%
`(v/v) final) at 220uC. After centrifugation, the protein pellet was
`resuspended in 100 mL of 0.1 M Tris/1 M urea buffer (pH 8.5).
`Proteins were then digested using 5 mg of mass spectrometry grade
`trypsin (Promega, Madison, WI, USA) overnight at 37uC. In order
`to obtain the proteome of insoluble fraction, the insoluble cell
`debris pellet was resuspended in 1 mL of PBS and washed three
`times with urea buffer (8 M Urea/0.1 M Tris). The resulting pellet
`was resuspended again in 100 mL of 0.1 M Tris/1M urea buffer
`(pH 8.5) and digested with trypsin overnight. These two fractions,
`soluble and CWA, were analyzed independently. The tryptic
`peptides from two fractions were cleaned independently using
`a Macro Trap with Peptide concentration and Desalting cartridge
`(Michrom, Aurburn, CA, USA) according to the manufacturer’s
`instructions. The resultant peptides were eluted in 98% acetoni-
`trile in water and then dried prior to mass spectrometry analysis.
`
`Mass Spectrometry Analysis
`Peptides were reconstituted in water at concentrations corre-
`sponding to between 40 and 200 ng of the original protein per
`2 mL injection. Nano LC/MS and nano LC/MS/MS analyses
`were performed on an Agilent HPLC-Chip Quadrupole Time-of-
`Flight
`(Q-TOF) MS system equipped with a microwell plate
`autosampler (maintained at 6uC), capillary sample loading pump,
`nano pump, HPLC Chip/MS interface, and Agilent 6520 Q-TOF
`MS detector. The chip used consisted of a 960.075-mm i.d.
`enrichment column and a 15060.075-mm i.d. analytical column.
`For sample loading, the capillary pump delivered 0.1% formic
`acid in 3.0% acetonitrile/water (v/v) isocratically at 4.0 mL/min.
`Following sample injection, a nano pump gradient was delivered
`at 0.4 mL/min using (A) 0.1% formic acid in 3.0% acetonitrile/
`water (v/v) and (B) 0.1% formic acid in 90.0% acetonitrile/water
`(v/v). Samples were eluted with 0% B, 0.00–2.50 min; 0 to 16% B,
`2.50–10.00 min; 16 to 44% B, 10.00–30.00 min; 44 to 100% B,
`30.00–35.00 min; and 100% B, 35.00–45.00 min. This was
`followed by equilibration at 0% B, 45.01–65.00 min. The drying
`gas temperature was set at 325uC with a flow rate of 5.0 L/min
`(2.5 L of filtered nitrogen gas and 2.5 L of filtered dry compressed
`air).
`Single-stage MS spectra were acquired in the positive ionization
`mode over a mass range of m/z 400–3,000 with an acquisition
`time of 1 s per spectrum. Mass correction was enabled using
`reference masses of m/z 322.048, 622.029, 922.010, 1,221.991,
`and 1,521.971 (ESI-TOF Calibrant Mix G1969-85000, Agilent
`Technologies, Santa Clara, CA, USA).
`
`Protein Identification
`All MS/MS samples were analyzed by X! Tandem (GPM-XE
`manager, ver. 2.2.1). X! Tandem was set up to search against the
`proteome of B. infantis ATCC 15697 and common contaminant
`
`PLOS ONE | www.plosone.org
`
`2
`
`February 2013 | Volume 8 |
`
`Issue 2 | e57535
`
`Genome & Co. v. Univ. of Chicago, PGR2019-00002
`UNIV. CHICAGO EX. 2001 - 2/13
`
`

`

`proteins and searched with a fragment ion mass tolerance of
`100 ppm. Oxidation of methionine was specified as a variable
`modification in X! Tandem. The probability (log(e)) cutoff for
`peptide assignment was set at 22. For the identification, two
`conditions were required: (a) more than two unique peptides per
`protein (Puniq$2) and (b) the probability cutoff for protein less than
`26 (log(e)#26). Additional algorithms for protein identification
`validated the protein identification results by X! Tandem. All
`proteins identified by the X! Tandem were also assigned by the
`peptide prophet [22] and protein prophet algorithm [23] with the
`probability greater than 95% and 99% respectively. Since the
`experiments were performed on biological duplicates an additional
`rule was applied for the decision of protein expression. If a protein
`was identified in one of the duplicates with high probability
`(Puniq$2 and log(e)#26) then it was considered expressed. If
`a protein had a log(e) value between 26 and24 and a Puniq$2 but
`was found in both duplicates, it was also considered as expressed.
`
`Protein Quantification
`Relative amounts of protein in a sample were calculated by the
`spectral counting method [24] using the proteome on lactose
`grown cells as control. Normalized spectral abundance factor
`(NSAF), which represented the relative abundance of protein in
`a sample, was estimated from the total number of spectra used for
`the identification of protein (SpC) and the number of aminoacids of
`the protein (L). For accurate estimations, the relative amount of
`protein was calculated only if the protein NSAF value was higher
`than 1.0 or SpC of both proteins were $6.
`
`Determination of Protein Localization
`introduced that
`For
`this purpose, a scoring system was
`employed the NSAF values of each protein in the soluble and
`insoluble fraction (Table S1). If the protein was observed only at
`the insoluble fraction, or the NSAF value of that protein in the
`insoluble fraction was more than 3-fold higher than the NSAF
`value of the same protein in the soluble fraction, the protein had
`score +1. If the NSAF ratio was between 2- and 3-fold, a protein
`was given a score of +0.5. When the NSAFs values of the same
`protein in the soluble and insoluble fractions were within 2-fold
`difference, the protein was given a score of zero. If the NSAF of
`a protein at the soluble fraction was larger than that of at insoluble
`fraction, the scores were attributed using the same criteria above,
`however the values were negative. Once scores were determined,
`values from the experiments with six different carbon sources were
`averaged. If a protein had an average score between +0.5 to +1.0 it
`was classified as a CWA protein. If the average score of a protein
`was between 20.5 to 21.0, it was considered as cytosolic (CYT)
`protein (Table S2).
`
`Results
`
`Cell Growth and Overall Protein Expression Profiles
`B. infantis was able to utilize various mono and oligosaccharides
`as growth substrates. The specific cell growth rates on GOS, FOS,
`inulin, and HMO were in a range of 0.14,0.19 (hr21) which are
`similar to that of glucose and lactose (Figure S1A). Mucin is a large
`extracellular protein that is heavily glycosylated. The O-glycan
`attached on the serine residue of mucin protein is composed of N-
`Acetylgalactosamine, N-Acetylglucosamine, fucose, galactose and
`sialic acid and has similar structures to HMO. It has been reported
`that several intestinal bacteria are able to interact with the mucin
`glycans [25,26]. Using mucin as the sole carbon source (Figure
`S1B), an increase in OD600 was appreciated for B. infantis albeit
`with the high initial cell density values.
`
`Whole Cell Proteomics of B. infants
`
`identified from the soluble and
`The number of proteins
`insoluble fractions is summarized in Table S3. Between 350 and
`400 proteins were found in any individual sample grown on each
`carbon source, and 100,170 proteins were commonly observed in
`both the soluble and insoluble fractions in each sample. A sum
`total of 540 proteins were identified during growth on the seven
`different carbon sources among 2416 ORFs encoded by B. infantis.
`Using the normalized spectrum abundance factor (NSAF) for
`each protein (representing the normalized protein quantity in
`a whole proteome of a single sample), quantitative expression
`profiles of whole proteomes were compared between cells grown
`on different carbon sources. While growth on most substrates
`exhibited similar proteome expression patterns, growth on
`galactooligosaccharides resulted in a separate cluster as witnessed
`by hierarchical cluster analysis (Figure S2). In addition,
`the
`correlation coefficients (Pearson’s product) between the GOS and
`the other proteomes were low (0.6,0.8),
`indicating that
`the
`protein expression observed during growth on GOS is significantly
`different from growth on other tested substrates (Table S4).
`
`Validation of Cell Wall Associated (CWA) Proteomics
`Most proteins observed in current shotgun whole cell proteo-
`mics experiments are cytosolic, while cell wall associated (CWA)
`proteins (both cell wall attached and membrane proteins) remain
`in the insoluble fraction with cell debris. Even if released from the
`soluble fraction, they are often missed due to the low ionization
`efficiency during MS analysis. Another constraint affecting their
`detection is their low (and/or variable) abundance by comparison
`to more readily obtained cytosolic proteins. To better understand
`the expression of CWA proteins, the insoluble fraction obtained
`after cell disruption was analyzed separately followed by the
`extensive washing with a strong denaturing agent (8 M urea/0.1M
`Tris). After removal of cytosolic proteins by washing, peptides
`from the CWA proteins were released by tryptic digestion. Protein
`localization was estimated by the quantitative ratio of
`their
`amounts in the soluble fraction and insoluble fraction.
`To validate this approach, the expression of representative
`CWA proteins such as extracellular solute binding proteins (SBPs)
`from B.
`infantis ATCC 15697 was examined [27]. These are
`common lipoproteins often associated with the transport of various
`ligands. Overall, 71 incidences of SBP expression were observed
`from the proteome of B.
`infantis grown on the seven different
`carbon sources. Among these, 23 cases of SBP expression were
`only found in the insoluble fraction (Figure 1, black bar).
`Moreover, 93% of the observed SBPs exhibited in average 2.9-
`fold higher NSAF value in the insoluble fraction by comparison to
`the soluble fraction (Figure 1). In addition to the SBPs, expression
`of other known cell wall proteins was more clearly, or solely,
`witnessed in the insoluble fraction. As shown in Table S5,
`Blon_1259, a V5/Tpx-1 family protein was not identified on the
`soluble fraction during growth on any carbon source, but it was
`one of the most abundant proteins observed in insoluble fraction.
`Also a cell surface protein involved in capsular exopolysaccharide
`synthesis
`(Blon_2082) and the NlpA lipoprotein (Blon_1721)
`exhibited 2 and 14 fold higher NSAF values in the insoluble
`fractions than that observed in the soluble fractions respectively.
`Therefore, by comparing the NSAF values of the soluble and
`insoluble protein fractions this approach provided insight into the
`cellular localization of each protein identified (Table 1 and 2). As
`shown on Table 1, 196 proteins were classified as CWA proteins
`and 93 were cytosolic (CYT) protein. However, the remaining 281
`proteins could not be conclusively classified in either cellular
`location (UI: unidentified). To further evaluate our method, the
`number of a theoretical CWA protein was predicted bioinforma-
`
`PLOS ONE | www.plosone.org
`
`3
`
`February 2013 | Volume 8 |
`
`Issue 2 | e57535
`
`Genome & Co. v. Univ. of Chicago, PGR2019-00002
`UNIV. CHICAGO EX. 2001 - 3/13
`
`

`

`Whole Cell Proteomics of B. infants
`
`Figure 1. Distribution of the NSAF ratio of each SBP between soluble and insoluble fraction (grey bars; yNSAFinsol/NSAFsol). White
`bars (soluble: left corner) and black bars (insoluble: right corner) indicate the number of SBP observed solely in soluble and insoluble fraction,
`respectively.
`doi:10.1371/journal.pone.0057535.g001
`
`tically [28]. Determination of transmembrane domains and signal
`peptides by TMHMM 2.0 [29] and the SignalP 2.0 HMM [30]
`indicated that 127 proteins are putative cell surface associated
`proteins. Among this pool of proteins, 108 were validated as cell
`wall associated by this proteomic approach. Only two of the cell
`surface associated proteins
`identified bioinformatically were
`categorized as cytosolic. In addition, 17 out of 127 theoretical
`surface proteins were found in the UI group. Thus while 281
`proteins were classified into this group by the above scoring
`criteria, the latter results on putative cell surface proteins suggests
`that the UI group is more likely to contain cytosolic proteins.
`
`Carbohydrate Metabolism
`Proteins involved in carbohydrate metabolism in B. infantis are
`summarized in Figure 2. Enzymes in the bifid shunt and glycolytic
`pathway, which are common to the utilization of various carbon
`sources, exhibited constitutive expression on the substrates used in
`this study (Figure S3). When compared to lactose-grown cells, the
`amount of each protein in these pathways varied less than two-fold
`among the different substrates. Interestingly, the genome of B.
`infantis ATCC 15697 indicates
`that
`several enzymes have
`
`isozymes. For example it contains seven genes encoding putative
`phosphoglycerate mutases, but only the expression of a single gene
`was appreciated, Blon_2152 (Figure S3).
`B.
`infantis metabolizes galactose through the Leloir pathway.
`Proteins of this pathway were observed in the proteomes of cells
`grown on all seven carbon sources, albeit their amounts varied. As
`shown in Figure 2A, galactokinase (galK: Blon_2062), Gal-1-P
`uridylyltransferase (galT: Blon_2063) and UDP-glc epimerase
`(galE: Blon_0538) were highly expressed on lactose, HMO and
`FOS. However, the alternative galE (Blon_2171) and phosphoglu-
`comutase (pgm: Blon_2184) were constitutively expressed across all
`substrates.
`N-Acetylglucosamine (GlcNAc) is one of the major constituents
`of HMO. To metabolize this monosaccharide, B. infantis possibly
`converts GlcNAc to fructose-6-P (Fru-6-P) by deacetylation
`followed by deamidation. GalNAc is another hexosamine found
`in O-linked glycans, and it could be first isomerized to GlcNAc by
`(nanE:
`an N-Acetylglucosamine
`6-phosphate
`2-epimerase
`Blon_0645). The acetyl and amine groups of GlcNAc-6-P could
`then be removed by GlcNAc-6-P deacetylase (nagA: Blon_0882)
`and glucosamine-6-P isomerase (nagB: Blon_0881), respectively.
`
`Table 1. Distribution of B. infantis proteins in cell wall or cytosolic fractions.
`
`Location
`
`Cell wall association (CWA)
`
`Cytosol (CYT)
`
`Unidentified (UI)
`
`Total
`
`Proteins identifieda
`
`Theoretical cell wall/membrane associated proteinsb
`
`196
`
`93
`
`281
`
`540
`
`108
`
`2
`
`17
`
`127
`
`aNumber of the protein for which the location was determined by the score system (See Materials and Methods).
`bTheoretical cell wall/membrane associated proteins were determined by the Signal prediction by SignalP 2.0 HMM and the transmembrane helices prediction by
`TMHMM2.0. Bioinformatic information including pfam classification was obtained from the Integrated Microbial Genomes database.
`doi:10.1371/journal.pone.0057535.t001
`
`PLOS ONE | www.plosone.org
`
`4
`
`February 2013 | Volume 8 |
`
`Issue 2 | e57535
`
`Genome & Co. v. Univ. of Chicago, PGR2019-00002
`UNIV. CHICAGO EX. 2001 - 4/13
`
`

`

`Whole Cell Proteomics of B. infants
`
`Figure 2. Central metabolic pathways reconstructed from proteomic datasets. Sugar substrates are presented inside the red boxes.
`Pathways regulated or induced by the presence of a specific sugar are indicated by colored background and their expressions on different sugars are
`presented next to the pathway as bar graphs: (a) Leloir pathway, (b) HexNAc catabolic pathway, (c) pyruvate fermentation pathway, (d) FOS/Inulin
`glycosyl hydrolase. The relative amounts of enzymes involved in the bifid shunt are presented in Figure S3. Numbers in the parenthesis next to the
`enzyme are the locus tag of each protein expressed.
`doi:10.1371/journal.pone.0057535.g002
`
`These three enzymes are expressed significantly only when B.
`infantis was cultivated on HMO, suggesting the specific induction
`of catabolic pathway for HexNAc (Figure 2B).
`While the proteins
`in the main glycolytic pathway were
`constitutively expressed, enzymes involved in pyruvate fermenta-
`tion changed their expression level depending on the carbon
`source. As shown in Figure 2C, pyruvate is reduced to lactate by
`lactate dehydrogenase (ldh: Blon_0840) with the regeneration of
`+
`NAD
`. Alternatively, pyruvate is converted to formate and acetate
`by the combination of pyruvate-formate lyase (pfl: Blon_1715)
`cleaving pyruvate to formate and acetyl CoA, and acetate kinase
`(ack: Blon_1731) converting acetyl CoA to acetate with the
`production of ATP. Expression of acetate kinase and formate C-
`acetyltransferse was consistent across all growth substrates.
`However, lactate dehydrogenase exhibited a different expression
`pattern depending on the carbon sources used. During growth on
`lactose or FOS, the amounts of LDH were 2 to 3.5-fold smaller
`than those when B.
`infantis grew on glucose, GOS, inulin and
`HMO. Interestingly, LDH was expressed 5.5-fold more on mucin
`compared to lactose. Alternative route enzymes that can produce
`acetyl CoA from pyruvate,
`i.e. pyruvate dehydrogenase or
`pyruvate oxidase, were not expressed in any of the proteomes
`(data not shown).
`
`Inulin and FOS are fructose polymers with different degrees of
`polymerization (DP). When FOS and inulin were used as carbon
`sources, a specific glycosyl hydrolase (GH) was induced (Fig-
`ure 2D). Blon_0128 is a family 13 GH annotated as a sucrose
`phosphorylase. These enzymes cleave sucrose to glucose and
`fructose-6-phosphate. The NSAF of Blon_0128 in the soluble
`fraction is higher than insoluble fraction suggesting a cytosolic
`localization. Two family 32 GH proteins
`(Blon_2056 and
`Blon_0787) were also expressed exclusively on FOS and/or
`inulin. Known activities of the family 32 GH includes inulinase
`(EC:3.2.1.7),
`levanase
`(EC:3.2.1.65),
`and
`exo-inulinase
`(EC:3.2.1.80). While Blon_0128 and Blon_2056 were expressed
`on both inulin and FOS,
`the expression of Blon_0787 was
`observed only in the proteome of FOS grown cells. The
`localization of Blon_0787 was determined in the UI group but is
`most likely a cytosolic protein.
`
`Solute Binding Proteins (SBP)
`Extracellular SBPs are typically linked to specific ABC transport
`systems consisting also in two inner membrane transporter
`components and an ATPase. The B. infantis genome contains 39
`SBPs of
`three different classes: Family 1, 3, and 5 [31]. In
`particular, Family 1 extracellular SBPs are thought
`to be
`
`PLOS ONE | www.plosone.org
`
`5
`
`February 2013 | Volume 8 |
`
`Issue 2 | e57535
`
`Genome & Co. v. Univ. of Chicago, PGR2019-00002
`UNIV. CHICAGO EX. 2001 - 5/13
`
`

`

`Whole Cell Proteomics of B. infants
`
`Figure 3. The expression of Family 1 extracellular SBPs on different carbon sources. The quantitative expression of each SBP is described
`by NSAF value. Black and gray bars indicate the NSAF values obtained in insoluble and soluble fraction, respectively. Box indicates the SBPs in the
`HMO cluster.
`doi:10.1371/journal.pone.0057535.g003
`
`responsible to the translocation of oligosaccharides [31]. As shown
`in Figure 3, different SBPs were expressed during growth of B.
`infantis on different carbohydrates. As indicated above, all Family 1
`SBPs were categorized as CWA.
`
`infantis was grown on GOS, the amounts of SBP
`When B.
`Blon_2414 were notably high, and FOS and inulin induced the
`expression of Blon_2061. In the latter case, low amounts of other
`SBPs were also observed, however, the expression of Blon_2061
`was 10,20 fold higher than the others. When HMO was added
`
`PLOS ONE | www.plosone.org
`
`6
`
`February 2013 | Volume 8 |
`
`Issue 2 | e57535
`
`Genome & Co. v. Univ. of Chicago, PGR2019-00002
`UNIV. CHICAGO EX. 2001 - 6/13
`
`

`

`on the media, SBPs Blon_2344 and Blon_2347 were mainly
`expressed. Significant expression of SBPs was not observed when
`B. infantis grew on glucose and lactose. Interestingly, mucin, whose
`glycan structure is similar to HMO, did not induce the expression
`of any of Family 1 SBPs.
`B. infantis has seven Family 3 and nine Family 5 SBPs, with
`broad affinity for di- and oligopeptides. Family 3 SBPs were
`actively expressed throughout cell growth (Figure S4). Blon_0747
`and Blon_0760 were commonly expressed regardless of carbon
`sources. Blon_2021 was induced by glucose, lactose, GOS, FOS
`and mucin but not HMO or inulin. Expression of Blon_0710 and
`Blon_2022 was observed when B. infantis was cultivated on glucose
`and lactose. Among Family 5 SBPs, Blon_0922 was highly
`expressed on all carbon sources examined. The amount of
`Blon_0922 expression was consistent and very high: it was one
`of the most highly expressed SBPs on all conditions (with the
`exception of Blon_2061 during growth on inulin). NSAF of
`Blon_0834 was high but specific in GOS proteome. Blon_0053
`and Blon_2419 appeared to be expressed constitutively across all
`substrates but at a low level.
`
`HMO Cluster I Protein Expression
`B. infantis contains a 43 kb cluster (HMO cluster I) that encodes
`genes for the catabolism of HMO ([11]; Figure 4A). During
`growth on HMO, expression of all different glycosyl hydrolases
`required for the degradation of HMO was observed. A b-
`galactosidase (Blon_2334) was expressed on all seven proteomes
`but the expression during growth on GOS was 2,3 times higher
`than that of other carbon sources (Figure 4B). The expression of
`the a-1/3,4 fucosidase Blon_2336 [32] was observed on lactose,
`FOS, inulin, and HMO indicating a more constitutive expression
`the expression of an a-sialidase
`(Figure 4C).
`Interestingly,
`Blon_2348
`[33] was
`exclusive
`to
`the HMO proteome
`(Figure 4D). The expression of a b-hexosaminidase (Blon_2355)
`was observed only during growth on HMO and inulin (Figure 4E).
`Using the scoring criteria described above, the cellular location of
`b-galactosidase and b-hexosaminidase were inconclusive however
`the NSAF of these proteins in the insoluble fraction was more than
`2 times higher than those at soluble fraction suggesting possible
`membrane association. As described in Figure 4A, putative HMO-
`related transporter SBPs Blon_2344 and Blon_2347 were adjacent
`to genes for two inner membrane transporter components, while
`SBPs Blon_2350, Blon_2351, Blon_2352, and Blon_2354 were
`positioned elsewhere in the HMO Cluster 1 without adjacent
`permease components. Protein from Blon_2344 was identified
`during growth on most of carbon sources but its amount was low
`(Figure 4F). During growth on HMO, however, expr