Studies of mucus in mouse stomach, small intestine, and colon. II. Gastrointestinal mucus proteome reveals Muc2 and Muc5ac accompanied by a set of core proteins

Abstract

The mucus that protects the surface of the gastrointestinal tract is rich in specialized O-glycoproteins called mucins, but little is known about other mucus proteins or their variability along the gastrointestinal tract. To ensure that only mucus was analyzed, we combined collection from explant tissues mounted in perfusion chambers, liquid sample preparation, single-shot mass spectrometry, and specific bioinformatics tools, to characterize the proteome of the murine mucus from stomach to distal colon. With our approach, we identified ∼1,300 proteins in the mucus. We found no differences in the protein composition or abundance between sexes, but there were clear differences in mucus along the tract. Noticeably, mucus from duodenum showed similarities to the stomach, probably reflecting the normal distal transport. Qualitatively, there were, however, fewer differences than might had been anticipated, suggesting a relatively stable core proteome (∼80% of the total proteins identified). Quantitatively, we found significant differences (∼40% of the proteins) that could reflect mucus specialization throughout the gastrointestinal tract. Hierarchical clustering pinpointed a number of such proteins that correlated with Muc2 (e.g., Clca1, Zg16, Klk1). This study provides a deeper knowledge of the gastrointestinal mucus proteome that will be important in further understanding this poorly studied mucosal protection system.

the gastrointestinal tract (GIT) is a highly specialized system that allows animals to catabolize the food ingested to obtain the essential molecules required for life and disposes of pernicious or waste components. Besides ingested material and its own digestive secretions, the GIT bears an enormous number of bacteria, estimated between 10³ and 10¹² per gram of luminal content depending on the segment, increasing from duodenum to colon (22). To cope with this potentially harmful load, as well as with the mechanical and chemical stress caused during the digestive function, the GIT is protected by mucus that covers the epithelial surface from stomach to colon. The mucus and its properties along the tract are described in an accompanying article (10).

The GIT mucus has recently come more into focus after finding that it is structurally and functionally divided into two layers in colon (16, 18): an inner layer in proximity to the mucosal surface, free from bacteria and physically resistant to mechanical disturbances; and an outer mucus layer, looser in structure, colonized by bacteria and most probably other types of commensals, which provides an environment for them to interact with the host without necessarily eliciting an immune response (19, 22). Both inner and outer layers share both Muc2 as major mucin and the associated proteins (19). However, the remaining components of the mucus are just now beginning to be discovered and understood.

Mucin proteins are large polymers that are highly glycosylated, synthesized, and secreted together with other proteins in specialized cells called goblet cells (32). These cells together with the remaining types of intestinal cells contribute to the composition of the mucus. We previously surveyed the colonic mucus protein composition, but only in the distal colon (19). To date, no study had directly addressed the composition of the gastrointestinal mucus along the length of the system. Proteomic technologies have been widely applied to characterize whole cells in culture, specific organelles, etc. In mouse intestine, a label-free proteomic study was recently applied to study the development of primary cells taken from jejunum (13). Given the recent improvements in proteomic technologies, in this work we combined filter-aided sample preparation (FASP) (38) and single-shot mass spectrometry in an Orbitrap system with detailed statistical analyses, to describe the composition of the mucus along the whole murine GIT, from stomach to colon.

MATERIALS AND METHODS

Animals and mucus collection.

All animal procedures were approved by the local Laboratory Animal Ethics Committee, Gothenburg, Sweden. Tissues from stomach to colon obtained from conventional C57BL/6 mice (Taconic, in-house bred) were prepared as described (10). Luminal mucus, present on the explants without stimulation, was collected from the stomach, duodenum, jejunum, ileum, proximal colon, and distal colon of six mice (3 males and 3 females) (10, 12). Tissues were mounted in a horizontal perfusion chamber with a circular opening of 4.9 mm², and the loose mucus (outer layer in stomach and colon) was collected apically.

FASP.

The collected mucus was processed according to the FASP method (38) modified so that 6 M guanidinium hydrochloride (GuHCl) was used instead of urea. Samples were thus solubilized, reduced by adding 100 mM DTT, and alkylated on the filter with 0.05 M iodoacetamide. Ten labeled peptides (HeavyPeptide AQUA Standards, Thermo Scientific) were synthesized for six of the most abundant proteins we found in preliminary MS analyses of the mucus (Muc2, Fcgbp, Clca1, Agr2, Zg16, EnoA; Table 1). A standard mixture of these peptides was titrated and added to each sample before digestion with 10 ng/μl of porcine trypsin (Promega) in 50 mM ammonium bicarbonate. Peptides were then eluted following the FASP protocol and cleaned up in StageTips (30).

Table 1.

Heavy peptides used as a standard mixture for the MS analyses

Peptide No.	Sequence	Protein
P1	SGDFELIK*	Muc2
P2	HETQEVQIK*	Muc2
P3	LPASLSEGR*	Fcgbp
P4	ISVINGGSK*	Fcgbp
P5	ASNYIIR*	Clca1
P6	SEISNIAR*	Clca1
P7	LPQTLSR*	Agr2
P8	IVFVDPSLTVR*	Agr2
P9	ASGTSFNAVPLHPNTVLR*	Zg16
P10	GVSQAVEHINK*	EnoA

Mass spectrometry.

The protein composition of the mucus samples was studied by nanoRPLC-ESI-MS/MS as described before (3, 32). In short, 2 μl of tryptic digest were injected with an HTC-PAL autosampler (CTC Analytics) connected to an HPLC system (Agilent 1100, Agilent). Peptides were loaded into a precolumn (4 cm length × 100 μm inner diameter) for 6 min and then eluted over an analytic column (20 cm length × 50 μm inner diameter) for 110 min by a gradient between 0.2% formic acid and 100% acetonitrile at a split flow rate of ∼100 nl/min. Both columns were packed with 3 μm ReproSil-Pur C18-AQ resin (Dr. Maisch). Data were acquired in a hybrid LTQ-Orbitrap XL instrument (Thermo Scientific) in dependent mode, measuring full MS in the Orbitrap and selecting the eight most abundant multiply charged ions for collision-induced dissociation (CID, 30% normalized collision energy) and acquisition in the LTQ. Full MS scans were performed in the m/z 350–2,000 range, with internal calibration by lock mass (m/z 371.1012, siloxane), and resolution of 60,000. MS/MS scans were set at a target value of 100,000, with isolation width of 3 amu. After fragmentation, ions were excluded for 45 s. The peptides were analyzed in nonconsecutive duplicates, making a total of 72 nanoRPLC-ESI-MS/MS experiments. The overall reproducibility of MS/MS identified was estimated as high by Pearson correlation (r = 0.7877), and removal of as little as the three “worst” runs increased the correlation to r = 0.9021. Both of these coefficients reflect the accuracy of the quantification and are in line with recently reported measurements of reproducibility (34).

Protein identification and statistical evaluation.

The raw files obtained after MS were analyzed in the MaxQuant 1.2.2.5 environment (8). Data were searched with the Andromeda search engine integrated in MaxQuant (9) against an in-house database containing all the mucin sequences available (http://www.medkem.gu.se/mucinbiology/databases/), the UniProt-SwissProt mouse database (version 1203, reviewed sequences), and the standard MaxQuant contaminant database. Oxidation of methionines and acetylation of the protein NH₂-terminus were set as variable modifications, and carbamidomethylation of cysteines as fixed; enzyme cleavage rules were defined for trypsin/P, with a maximum of two missed cleavages. Tolerances were limited as maximum five modifications per peptide, 20 ppm error for the first search and 6 ppm for the main search. Isoleucine and leucine were considered indistinguishable. The false discovery rate was calculated from searches against a reversed database, and set to 0.01 for proteins, peptides, and modified sites. The identification rate was improved by matching between runs through remapped retention time (window of 2 min). The mass spectrometry data have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository (37) with the dataset identifier PXD000271.

The resulting data were loaded as protein groups into Perseus 1.3.0.4 (J. Cox, available at http://maxquant.org/index.htm). Protein quantities were calculated as intensity values [from the extracted ion current (XIC)] and normalized in parts per million to the total XIC for the standard labeled peptides in each run. Reverse hits (n = 14, 1%) and proteins with no unique peptides assigned (n = 2, 0.15%) were removed from the analysis. From the remaining proteins (98.85%), the ones identified only by a modified site and the contaminants (e.g., the porcine trypsin used in digestion) were also discarded. For the analyses described below, the samples were categorized per replicate, sex, animal (mouse), organ, and segment.

The protein distribution in the different samples was examined by plotting frequency histograms, where data were log2 transformed, replicates were averaged (except for reproducibility tests), and groups (“organ,” “sex,” etc.) were represented by their median. Different sample groups were directly compared by multiscatter plots, calculating the corresponding Pearson correlations (P threshold value of 0.01) with the SPSS Statistics 17.0 software. To examine the level of abundance of certain proteins in relation to others, data were normalized by ranking, averaged by segment, and plotted into histograms in logarithmic (ln) form. Presence and absence comparisons were done by numerical Venn diagrams, obtained from the log2-transformed data, averaging the samples by the median of each segment. Unique and common proteins were graphically represented through Venn diagrams created with the software Venny (Oliveros, 2007, http://bioinfogp.cnb.csic.es/tools/venny/index.html). Differences in abundance values were tested for significance by analysis of variance (ANOVA) in case of multiple samples, and t-test in case of two-sample comparisons. For both ANOVA and t-test, permutation-based false discovery rate was used for truncation, with a threshold value of 0.05. The t-tests were two-sided, and 250 randomizations (not preserving the grouping) were used for truncation. For principal component analysis (PCA), missing values were replaced by values imputed following the normal distribution, with the standard parameters set in Perseus (width 0.3 and down shift 1.8). Hierarchical clustering was done by Euclidean distance with average linkage after Z-scoring data.

Production of anti-ZG16 antibody.

Total RNA was extracted from the colon adenocarcinoma cell line LS-174T with RNeasy Mini Kit (Qiagen). The open reading frame of ZG16 was reversed transcribed and amplified with SuperScript One-Step RT-PCR with Platinum Taq (Invitrogen). The obtained cDNA was amplified and inserted in the pS-IgG vector (2). CHO-K1 cells were cultured in Iscove's Modified Dulbecco's Medium (Lonza), containing 10% (vol/vol) fetal bovine serum and supplemented with sodium pyruvate (110 mg/l), l-arginine (116 mg/l), l-glutamine (290 mg/l), l-asparagine, and folic acid (10 mg/l). CHO-K1 cells were seeded in a 12-well plate at 90% confluence and transfected with the expression plasmid pS-ZG16-IgG by use of Lipofectamine 2000 (Invitrogen). To purify the recombinant ZG16, spent media was collected, centrifuged at 4,000 g for 5 min, and passed through a 0.22-μm filter to remove cellular debris. Filtered medium was then diluted 1:1 with protein G binding buffer (20 mM sodium phosphate buffer, pH 7.0) and applied to a HiTrap Protein G column (GE Healthcare). The bound protein was eluted with 0.1 M glycine-HCl pH 2.7, and pH in the eluted fractions was adjusted to 7.0 with 1 M Tris·HCl pH 9.0. Fractions containing ZG16-IgG were pooled and concentrated by use of Vivaspin 6, MWCO 10,000 (Sartorius) spin columns. Next, the IgG-tag was cleaved off from ZG16 for 18 h at 37°C with 50 U of enterokinase (EKmax, Invitrogen) per milligram of recombinant protein. The buffer was exchanged to 50 mM HEPES pH 8.0 with a PD-10 column (GE Healthcare), and impurities were separated from ZG16 by ion-exchange chromatography on an ÄKTA Purifier system (GE Healthcare). ZG16 was loaded onto a MONO S 5/5 column equilibrated in 50 mM HEPES pH 8.0 and was eluted (∼400 mM NaCl) with a linear gradient of 0–1.0 M NaCl in the same buffer. The purified protein was used to immunize rabbits (Agrisera, Umeå, Sweden). The resulting antiserum allowed the detection of the human ZG16 and cross-reacted with the murine Zg16 (the human and mouse sequences are 89.8% similar, being 83.2% of the sequence identical) and therefore was titrated for both types of antigens. Staining of a Zg16 knockout mouse strain showed no specific signal.

Immunostaining.

Methanol-based Carnoy-fixed tissues (17) were stained with the following antibodies: 1/500 anti-Muc2C3 antiserum (19), 1/1,500 anti-Clca1 (Abcam), 1/600 anti-ZG16 antiserum, 1/2,000 secondary antibodies Alexa Fluor 488 (green) or 555 (red) anti-rabbit IgG (H+L) (Invitrogen). DNA was stained with 1/20,000 DAPI (Sigma). The sections were imaged with a fluorescence Eclipse E-1000 microscope (Nikon).

RESULTS AND DISCUSSION

Quality of the protein identifications.

The physical properties of the GIT mucus vary along the GIT (4, 10, 36). To find out whether the protein composition was also different, we profiled the protein content of the mucus in six different segments of the GIT, following the pipeline described in Fig. 1A. To make sure that only the mucus was analyzed, the tissues were mounted in an open horizontal Ussing-type chamber and allowed to secrete. The mucus was then carefully collected without disturbing the epithelial cells. The mucus amount sampled varied from location to location depending on the thickness and density of the mucus layer in each segment but in all cases was from a tissue area of 4.9 mm². Mucus was collected at six locations from six conventional C57BL/6 WT mice (3 females and 3 males). In the large intestine (colon), this mucus that could be easily aspirated corresponded to the outer mucus layer, also called the “loose” layer. The inner layer (previously called “firm” layer) was not sampled, to obtain only mucus equivalent in properties to the mucus aspirated from the small intestinal locations. Cecum was not analyzed because it was not possible to separate the luminal content from the mucus in a reproducible way. After collection, the 36 mucus samples were randomized, dissolved, digested into peptides, and analyzed by nanoRPLC-ESI-MS/MS. In total, we identified 6,934 unique peptides, assigned to 1,339 proteins. Reverse entries, contaminants, and proteins identified only by modified sites were filtered out. Eventually 1,276 mouse proteins with at least one unique peptide identified (∼95.3%) were compared through the data set. Searches against bacterial databases gave numerous hits but, because of, on the one hand, the high homology of some proteins, and on the other hand the lack of sequences for many commensal bacteria, these were not analyzed further.

Fig. 1.

The gastrointestinal mucus proteome varies along the gastrointestinal tract (GIT). Mucus peptides obtained by tryptic digestion were analyzed by an Orbitrap mass spectrometer. A: schematic description of the preparation scheme used for mucus proteome analysis. B: scatter plots depicting correlations in observed protein relative abundances between sexes, organs, and segments. The Pearson correlation coefficient is shown for each plot. C: principal component analysis could classify the mucus samples on the basis of the total protein fingerprint. In particular, component 2 could separate samples from stomach (negative values) from large intestine (proximal and distal colon; positive values). Samples are represented as stomach (■), duodenum (+), jejunum (□), ileum (△), proximal colon (•), distal colon (▲).

Characterization of GIT locations by their mucus protein profile.

Multivariate techniques were first used to analyze the variability between groups by calculating the correlation of the relative abundance of all the proteins identified. Comparing replicate samples showed a high degree of correlation (r = 0.914, P < 0.001), in agreement with the strong correlation for the number of MS/MS spectra identified. Comparing across males and females, and among different GIT regions, strong and significant correlations were observed (Fig. 1B), with the highest correlation between all regions in males and females (r = 0.841, P < 0.0001) and between stomach and small intestine (r = 0.641, P < 0.0001) and the lowest correlation between stomach and large intestine (r = 0.402, P < 0.0001). The mucus protein patterns were then analyzed by PCA, revealing that the stomach and large intestine (y-axis) were well separated and that proximal and distal colon tended to be grouped together (Fig. 1C). No grouping according to MS replicate, sex, or animal was observed. Samples from the small intestinal segments (duodenum, jejunum, and ileum) were, however, intermixed with the other locations, which agrees with this organ showing a mucus transitioning between the phenotype observed in the stomach and the one in the large intestine.

The specific proteins composing the mucus were first assessed by ranking them on the basis of their relative abundance in each sample. In all the samples, actin (Actg) appeared consistently in the top 10 most abundant proteins of the mucus (Supplemental Table S1). Other proteins, like keratin-8 (Krt8), were highly abundant in most of the samples, decreasing in specific samples, especially in proximal colon. Albumin (Alb) was abundant in all the samples (top 25 hits), especially in ileum and the colonic segments, probably owing to contamination during dissection. Noticeably, the proteins Clca1, Fcgbp, and Zg16, which we found associated with Muc2 in a previous study of the distal colon mucus (19), were very abundant (top 10) in this segment. Furthermore, these proteins were now found in the mucus in all the locations analyzed, and correlated with Muc2 (see cluster analysis below), thus showing that the major proteins found in our previous study were also present in the mucus at other intestinal locations.

Ranking proteins on the basis of their abundance has a drawback, since proteins absent in one sample will still be ranked with abundance = 0 in the lowest bin of the distribution. Therefore, to find out which proteins were completely absent from any of the groups studied, we built Venn diagrams (Fig. 2). Figure 2A represents the number of proteins present in stomach, small intestine, and large intestine. Figure 2B compares proteins in duodenum, jejunum, and ileum and Fig. 2C proximal and distal colon. Table 2 and Supplemental Table S2 show the results of numerical Venn diagrams, where all the proteins identified in the six segments have been summarized on the basis of their presence in specific segments. Table 2 displays the total number of proteins found in each segment, as well as the number of proteins only identified in that location (“exclusive”); 752 proteins out of 1,276 (79%) were present in all the samples.

Fig. 2.

Venn diagrams showing the number of shared and exclusive proteins found in the mucus of the 3 organs from the GIT (A), the small intestine (B), and the large intestine (C). The total number of proteins in these Venn diagrams is 1,471, 4 proteins less than mentioned in the text, owing to the necessary statistical filtration in the process of generation of the diagrams. The proteins removed were Niban, Hook2, Nrn1l, and Drg2.

Table 2.

Number of proteins identified in total for each of the locations of the gastrointestinal tract, and number of proteins found exclusively in each of these locations

Location	N Total Proteins	N Exclusive Proteins
Stomach	1161	8
Duodenum	1069	2
Jejunum	1169	3
Ileum	1176	4
Proximal colon	1180	4
Distal colon	890	1
All	752

To corroborate the reproducibility of our approach, we applied a t-test to the 1,276 proteins identified, finding no proteins with significant differences between the replicates (Supplemental Table S3, column: t-test R01-R02; replicate no. 1: R01, replicate no. 2: R02). More importantly, we investigated whether there was any difference in the protein expression due to the sex of the animals, finding none (column: t-test M-F; M, male; F, female). Differences among the animals were studied by ANOVA for multiple samples, interestingly finding no protein with a significantly different level among the six mice studied (column: ANOVA Mouse). Although there was an individual variability in mucus thickness or penetrability (10), there is no such a difference in the presence or absence of specific proteins. When samples were grouped by organ (stomach—small intestine—large intestine; column: ANOVA Organ), we found 369 proteins with significantly different abundance in at least one of the locations. Taking into consideration the segments (organs subdivided into stomach—duodenum—jejunum—ileum—proximal colon—distal colon; column: ANOVA Segment), the number of varying proteins was 389, of which 99 were new regarding the organ comparison. Thus combined, these analyses showed a total of 488 proteins that differed in amounts, 99 among segments, 119 among organs, and 270 differential regardless of the classification of the samples. These 488 proteins represent ∼38% of the proteins identified in the mucus and probably reflect functional differences in the separate parts of the GIT.

A t-test was also employed to study significant differences between specific pairs of samples. Comparisons between organs reported 51, 307, and 120 proteins with significantly different levels between stomach and small intestine, stomach and large intestine, and small and large intestine, respectively. When looking for differences between segments, jejunum and ileum showed no differences, whereas duodenum and jejunum had only one significant protein difference (septin-9; Supplemental Table S3, columns: t-test Jej-Ile and t-test Duo-Jej, respectively). Comparing mucus from ileum and proximal colon showed just 12 significantly different proteins (column: t-test Ile-PC). On the other hand and as expected, mucus from distant segments as the stomach and the proximal or distal colon showed the largest differences with 328 and 321 proteins, respectively (columns: t-test Sto-PC and t-test Sto-DC). Interestingly, this number quickly decreases once we move into the small intestine, and thus the duodenum and distal colon just showed differences in the protein abundance of 151 proteins (column: t-test Duo-DC).

As the Muc2 mucin is the scaffold of the intestinal mucus, it is interesting to find proteins that follow the same trends as Muc2, as these could take part in the mucus formation. With this in mind, we applied hierarchical clustering to our total dataset. Figure 3 highlights the node that clustered around Muc2. The levels of some proteins were specifically studied along the GIT and are shown in Fig. 4 as boxplots. These figures depict both the average levels and the individual variability. Except Muc5ac, which was plotted to show the distribution of a mucin other than Muc2, all the proteins shown in Fig. 4 follow the same trends in their relative levels, and that is why they were all grouped together in the hierarchical clustering.

Fig. 3.

The proteins observed by proteomic analysis of the GIT mucus were hierarchically clustered. A detail of this clustering around the node for Muc2 is displayed, with the proteins closely related to Muc2 highlighted in blue.

Fig. 4.

Boxplots of the relative levels of mucins (logarithm of the normalized intensity) along the GIT mucins and some of the proteins observed to cluster around Muc2 (Fig. 3). Circles represent outliers, i.e., values between 1.5 and 3 times the interquartile range (IQR; difference between the ends of the box). Stars represent extreme values, defined as those larger than 3 times the IQR. Sto, stomach; Duo, duodenum; Jej, jejunum; Ile, ileum; PC, proximal colon; DC, distal colon; SI, small intestine; LI, large intestine.

Some of the proteins found in the proteomic lists are typical abundant intracellular proteins as for example histones. This is not surprising since the epithelial cells are continuously renewed and expelled out into the lumen and mucus. Proteins belonging to the apical cytoskeleton and vesicle trafficking (actin, cytokeratin, 14–3-3 proteins, etc.) were also identified, suggesting that the enterocyte microvillus generates ample amounts of vesicles which are then found in the mucus, as was recently suggested (23). Several of the proteins found in the mucus were shown to be enriched in these vesicles. These type of proteins are also involved in mucus secretion of the goblet cells (27). It was recently proposed that the goblet cell can open its inner cytoplasm to the intestinal content upon secretion, something that should also contribute significantly to the presence of cytoplasmic proteins in the mucus (24). Several of these cytoskeletal proteins, which tend to be thought of as simply contaminants, are known to be regulated by bacteria (15), being significantly different in the intestine of germ-free and conventional mice (7).

Mucins in the GIT mucus.

Muc5ac was one of the most abundant proteins in the stomach mucus (Supplemental Table S1), but it appeared also in the small intestine and proximal colon of most of the animals tested (Fig. 4B and D), although not in the distal colon. This was reflected in the number and type of peptides found. Muc5ac mRNA expression and immunostaining is only found in the surface epithelium of the stomach, and the presence of the Muc5ac protein all the way down to the proximal colon thus reflects the distal propulsion in the GIT (26, 31). It also shows that Muc5ac, like Muc2, is relatively resistant to endogenous digestive enzymes, and it is not until Muc5ac reaches the commensal bacteria in colon that it is degraded. This starts with the removal of one monosaccharide at a time until the mucin protein core is exposed and degraded, probably explaining that Muc5ac is still found in proximal colon.

Muc2 was one of the most abundant proteins in our mucus samples except in the stomach, where its role is taken up by Muc5ac (Supplemental Table S1). Its abundance was high in the small intestine (especially in the duodenum and the ileum), but much higher and less variable in the large intestine (Fig. 4, A and C), where the mucus layer is thicker (10), a pattern similar for males and females (Fig. 4E). It seems then that the amount of Muc2 actually increases from beginning to end of the GIT, as the total number of Muc2 peptides increased steadily from stomach to distal colon (24, 58, 101, 181, 363, 413 in order of the segment). To support the proteomics data, we stained tissue sections from all the segments of the GIT studied for Muc2 (Fig. 5). Muc2 (in green) could be observed both inside the goblet cells and in the mucus, with intensities that reflect the relative amounts suggested from Fig. 4.

Fig. 5.

Localization of Muc2 and proteins that clustered together with Muc2 in the hierarchical clustering (Fig. 3). The Muc2 mucin (green; left), Clca1 (red; middle), and Zg16 (red, right) were specifically detected in tissues from the 6 locations from where mucus was sampled. Prox., proximal; Dist., distal. Bar: 100 μm.

We found no peptides from the mucins Muc5b, which is normally high in saliva and lungs, or Muc6, indicating that they are not major components of the mucus. Muc6 is the major mucin in the stomach glands (11), but this and a previous study did not identify Muc6 in the surface mucus of mouse stomach (29). This may suggest that Muc6 does not mix with the Muc5ac mucus layer as has been suggested previously (14).

Muc13, a small transmembrane mucin (33), was also found in the mucus with a lower level in the stomach than in the rest of the GIT (Supplemental Table S3). Comparison of segments by pairs revealed that stomach, duodenum, ileum, and proximal colon all showed a significant difference compared with the distal colon (Fig. 4F). This difference is probably due to the higher variability of Muc13 in the proximal segments and could depend on different levels of vesicle shedding since Muc13 is enriched in the shed vesicles (23). Searching specifically for mucins, we identified one peptide from another mucin, Muc3 (17), in four of six samples of distal colon. This is a transmembrane mucin found in enterocytes, and a highly glycosylated protein, which can thus theoretically generate only a few peptides that could be identified by MS. Peptides from this protein have been found before in the mucus of distal colon (20). We did not detect additional transmembrane mucins, as for example Muc1 and Muc4, probably reflecting that these are membrane bound and minor components in the luminal mucus (25). However, it should also be pointed out that the extracellular parts of these and other transmembrane mucins contain few peptides that are nonglycosylated and thus possible to detect by current approaches.

The colonic mucus is built around the Muc2 mucin, whereas the stomach mucus is built around the Muc5ac mucin. A model for how the Muc2 mucin is packed in the goblet cells and how the released mucin is forming large netlike rings that can stack on top of each other was recently introduced (1). Although we do not yet know the polymeric organization of the Muc5ac mucin, it is interesting to note that the large NH₂- and COOH-terminal parts of the Muc2 and Muc5ac mucins are the most homologous among the gel-forming mucins. The Muc5ac mucin can thus be suggested to have a similar organization to that of the Muc2 in the large intestine, something that is supported by the observed two-layered mucus system in both stomach and colon (10).

Other mucus associated proteins in the GIT.

A number of other proteins identified are probably important components of the GIT mucus. Among these are the Clca1 protein (Fig. 4G) that was previously believed to be a calcium channel, but more recent information suggests other yet-unraveled functions (5, 28). Immunostaining for Clca1 (Fig. 5) revealed that it appeared specifically located to the upper third of the intestinal crypts. Fcgbp (Fig. 4H) had previously been identified as an abundant mucus protein that at least in the distal colon was covalently attached to the Muc2 mucin (20). This protein contains 13 von Willebrand D domains, of which 11 can be cleaved by an autocatalytic mechanism and by this generate a reactive group that can anchor and cross-link other proteins. Kallikrein 1, Klk-1 (Fig. 4I), is a serine proteinase found at the beginning of the intrinsic coagulation pathway. This protein was highlighted in the cluster analysis, being for a still-unknown reason abundant in the intestinal mucus. Zg16 was first identified in the pancreatic zymogen granulae (6) and later suggested to bind negatively charged glycans (21, 35). Immunostaining of Zg16 using a novel antiserum showed it located to both the goblet cells and the mucus (Fig. 5, red). Zg16 was only faintly stained in the stomach, in accordance to the levels quantified by proteomics (Fig. 4J).

Conclusion.

The composition of the mucus along the GIT was studied by a proteomic approach that revealed a comprehensive list of proteins. It can be concluded that the GIT mucus contains a relatively conserved mucus protein core along the alimentary tract. This is surprising taking into account the different mucus properties observed in the accompanying article (10), suggesting that mucus differences are due to processing of especially the gel-forming mucins. A number of proteins were found to be mucus associated, but the functions of most of these are not understood yet. The information presented here will be the basis for further insights on the molecular perturbation found in animal models of GIT diseases.

GRANTS

This work was supported by the Swedish Research Council (grants 7461, 21027), Swedish Cancer Foundation, Knut and Alice Wallenberg Foundation, IngaBritt and Arne Lundberg Foundation, Sahlgren's University Hospital (LUA-ALF), Wilhelm and Martina Lundgren Foundation, Torsten and Ragnar Söderberg Foundation, Adlerbert Research Foundation, the Sahlgrenska Academy, the National Institute of Allergy and Infectious Diseases (U01AI095473; the content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH), and the Swedish Foundation for Strategic Research—Mucus, Bacteria, Colitis Center (MBC, Innate Immunity Program).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

A.M.R.-P., J.H.B., A.E., J.K.G., A.S., M.E.J., and G.C.H. conception and design of research; A.M.R.-P., J.H.B., A.E., J.K.G., A.S., and M.E.J. performed experiments; A.M.R.-P., J.H.B., A.E., J.K.G., A.S., M.E.J., and G.C.H. analyzed data; A.M.R.-P., J.H.B., and G.C.H. interpreted results of experiments; A.M.R.-P. and J.H.B. prepared figures; A.M.R.-P., J.H.B., and G.C.H. drafted manuscript; A.M.R.-P. and G.C.H. edited and revised manuscript; A.M.R.-P., J.H.B., A.E., J.K.G., A.S., M.E.J., and G.C.H. approved final version of manuscript.