Recruitment of patients and sample collection

Patients ≥18 years who were referred to Sahlgrenska University Hospital (Gothenburg, Sweden) for colonoscopy, were eligible for inclusion and subject to the provision of written informed consent. Patients with macroscopic/microscopic evidence of ileocolonic pathology other than Inflammatory bowel disease were excluded. Eight biopsies were obtained from the terminal ileum of each patient. The study protocol was approved by the regional ethics committee (Ethical permit #2020–03196) and complied with the Declaration of Helsinki.

Phylogenetic data

Phylogenetic trees and molecular time estimates were extracted from TimeTree [6, 13].

Sequence alignments

Local sequence similarity search and identity measurements of MUC genes were performed using NCBI BLAST [14]. Primer specificity was analyzed using primer BLAST [15]. Multiple sequence alignment of MUC gene and protein homologs was conducted using CLUSTALW [16]. Perl scripts were used for all data extraction (see supplementary methods in S1 File). Promotor regions -1 kb from the transcription start site of MUC3A and MUC3B in Cercopithecoid and Hominoid superfamilies were aligned using Multiple Alignment using Fast Fourier Transform (MAFFT) high-speed multiple sequence alignment tool [17].

Generation of dot plots for pairwise sequence alignment and sequence logo representations

Dot plots representing pairwise sequence alignments were generated using Genome Pair Rapid Dotter (GEPARD) version 1.40 [18]. Sequence logos of perfect tandem repeats were generated using WebLogo3 [19].

Mapping of DNase-seq and ChIP-seq data to the human genome

DNase hypersensitive sequences upstream of MUC3A in GRCh38.p13 and Chromatin immunoprecipitation (ChIP) sequencing of the human small intestine, colon, and stomach samples are summarized in S1 Table [20]. Graphical representation of epigenetic signatures was prepared by aggregating multiple segment-sorted tracks using the Matplot function in Washington University Epigenome Browser v53.5.0 [21].

Single-cell expression of transcription factors

Expression profiles for transcription factors were extracted from the following data sets available at Single Cell Portal (Broad institute): single-cell transcriptome analysis of human small intestine (GSE148829) [22], human colon (GSE178341) [23], and mouse small intestine (GSE92332) [24].

Mapping of RNA-sequencing data to T2T-CHM13 human genome assembly

The T2T-CHM13 human genome assembly was downloaded from NCBI BioProject PRJNA559484. Fastq-dump was used to obtain RNA-sequencing reads. Burrows-Wheeler Aligner (BWA) software package [25] was used to align RNA-sequencing reads to exonic sequences of genes belonging to the MUC3 cluster. Perl scripts were used to perform quality control and measure read number (see supplementary methods in S1 File). The following publicly available data sets were used to determine MUC gene expression in human tissues: single-cell transcriptome analysis of human ileum, colon, rectum (GSE125970) [26], human liver (GSE124395) [27], and human kidney (GSE131685) [27]. Gene expression of individual MUC genes in the MUC3 cluster was calculated as transcripts per million (TPM) as previously described [28].

RNA extraction from human ileum, cDNA synthesis, and RT-qPCR

RNA from human ileal biopsies was extracted using RNeasy Mini Kit (Qiagen). 500 ng of RNA was reverse transcribed to cDNA with TaqMan Reverse Transcription kit (#N8080234, Applied Biosystems), using 2.5 μM random primers and the cycling parameters 25.0°C for 10 min, 37.0°C for 30 min, and 95.0°C for 5 min. 750 ng of cDNA was used for downstream reverse transcription quantitative PCR (RT-qPCR) with 0.3 μM MUC3A-specific primers (forward 5’-TGGGGGTCAGTGGGATGGCCTCAAA-3’; reverse 5’-CACGTGGGACCGCTCGTCTCC) or MUC3B-specific primers (forward 5’-CGGGGGCCAGTGGGATGGCCTCAAG-3’; reverse 5’-CACGCGGGACCGCTCGTCTCT-3’) using SsoFast EvaGreen Supermix (#1725200, Bio-Rad) on a CFX96 Real-Time PCR Detection System (Bio-Rad) with the cycling parameters 95.0°C for 3 min, 39 cycles of 95.0°C for 10 s, 63.5°C for 10 s, 72.0°C for 20 s. Melting curve analysis was performed at 95.0°C for 10 s, and 65.0°C to 95.0°C at an increment of 5°C for 5 s.

Restriction site analysis and agarose gel electrophoresis

5 μL of the RT-qPCR reaction was digested with 1 μL FastDigest PstI restriction enzyme (#FD0614, ThermoFisher Scientific) for 1 h at 37°C. Full-length amplicons and digestion products were separated on 1.5% agarose gel with ethidium bromide.

Statistics

Statistical analysis and graphical illustrations were performed using GraphPad PRISM 8.3.1 (GraphPad Software). Statistical tests were applied using two-way ANOVA and corrected for multiple comparisons using Tukey´s test. Data are presented as mean ± standard deviation (SD). For all statistical analyses: * p<0.05, ns = Not significant.

The evolution of a MUC3 cluster in Cercopithecoids and Hominoids

The human chromosome locus 7q22 contains three MUC genes MUC3, MUC12, and MUC17, arranged in a MUC3 cluster flanked by ACHE at its 5’ end, and TRIM56 and SERPINE1 at its 3’ end (Fig 1A). Using ACHE, TRIM56, and SERPINE1 as genomic markers, we identified the MUC3 cluster in species belonging to the Catarrhini parvorder, namely Cercopithecoid (Old World monkeys) and Hominoid superfamilies, the latter including the genera Pongo (orangutang), Gorilla, Pan (chimpanzee and bonobo) and Homo (Fig 1B). In Cercopithecoids we identified a MUC3 cluster with a length of 153 kilobase pairs (kbp) in Macaca mulatta (rhesus), while the corresponding gene cluster in the Papio Anubis (baboon) consisted of two mapped sequences with a total length of 138 kbp (Fig 1B). In Hominoids, MUC3 cluster length ranged from 106 kbp in the Nomascus leucogenys (gibbon) to 203 kbp in Pongo abelii (orangutang). In the Homininae subfamily, we observed striking differences between MUC3 cluster length in Pan troglodytes (chimpanzee) and its two closest relatives; the gene cluster in H. sapiens GRCh38.p13 assembly was 73 kbp shorter and in G. gorilla (gorilla) 66 kbp shorter than in the chimpanzee (Fig 1B). Thus, we hypothesized that the MUC3 cluster within the human GRCh38.p13 assembly contains significant sequence gaps that may obscure unknown MUC genes.

To test our hypothesis, we used a set of defined criteria when exploring available primate genome assemblies for unidentified MUC3 cluster genes. We scanned the clusters for 1) start codons, 2) long mucin-type PTS-encoding exons, 3) SEA domains conserved in membrane mucins and, 4) unique intronic and exonic sequences that separate individual MUC genes. Our analysis revealed that all Cercopithecoids carried a MUC3 cluster consisting of MUC3, MUC12, and MUC17 genes (Fig 1B). Strikingly, the primate MUC3 gene existed as two distinct MUC3A and MUC3B genes, although only partial sequences of the MUC3B gene were identified in P. anubis. The Hominoid superfamily, except for H. sapiens and G. gorilla, carried a MUC3A gene and full or partial sequences of MUC3B. Thus, we identified a MUC3B gene exclusively in species belonging to the Catarrhini parvorder, which diverged from Platyrrhini (New World monkeys) around 43 million years ago (Mya) (Fig 1C). However, because of inadequate sequence coverage of the MUC3 cluster in Platyrrhini, and Scandentia (treeshrew) and Dermoptera (colugos) orders that constitute the closest relatives of primates, we were not able to determine when MUC3B first emerged during vertebrate mammalian evolution.

The human 7q22 locus contains a MUC3B gene

Since humans and chimpanzees share 98.8% of their genomic DNA and the chimpanzee genome carries a MUC3B gene in the MUC3 cluster, we hypothesized that the absence of a MUC3B gene in humans is a result of sequence gaps in the GRCh38.p13 assembly. In the quest for a human MUC3B gene, we explored PacBio Single Molecule Real-Time (SMRT) reads from a human HX1 [29] and identified 3 individual reads that covered the 3’ end region of MUC3A (encoding the C-terminal region of MUC3A protein, designated MUC3A C-term), an intergenic region, and the 5’ end region of a putative MUC3B gene (designated MUC3B N-term) (Fig 2A). Strikingly, the length of the intergenic region was on average 10,939 bp, which corresponded to the length of the MUC3A-MUC3B intergenic region in Catarrhines (average of 11,810 bp). Moreover, we identified 5 SMRT reads covering MUC3B C-term, an intergenic region, and MUC12 N-term (Fig 2A). The average length of the MUC3B-MUC12 intergenic region was 2469 bp and conserved in Catarrhines (average of 2491 bp). This initial exploration provided evidence for the existence of a distinct human MUC3B gene. However, because MUC3A and MUC3B share high sequence identity (87% and 94% for N-term and C-term across catarrhines) and the error rate of the SMRT reads was 70–85%, the HX1 assembly could not with high confidence distinguish between the two MUC3 genes. Moreover, the reads failed to capture the length and sequence of a predicted single PTS-encoding exon in MUC3B.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Evidence of a putative MUC3B gene in recent human genome assemblies.

(A) Exploration of PacBio sequencing of HX1 genome identified SMRT reads covering the intergenic region between MUC3A and putative MUC3B, an incomplete PTS sequence, and intergenic sequences between putative MUC3B and MUC12. (B) The T2T-CHM13 assembly contains a 60 kb gap between MUC3A and MUC12. (C) Sequence alignments of SEA, transmembrane (TM), and cytoplasmic tails (CT) of MUC3A and putative MUC3B show high sequence identity, nucleotide mismatches, and a conserved PDZ binding motif.

https://doi.org/10.1371/journal.pone.0275671.g002

The current GRCh38.p13 draft covers lightly packed euchromatic regions corresponding to 92% of the human genome, while more complex regions including long tandem repeats in MUC genes are underrepresented. A recently published CHM13 T2T v1.1 assembly, based on long-read genome sequencing of homozygous complete hydatidiform mole (CHM) cells followed by gapless telomere-to-telomere assembly, adds approximately 200 Mbp to the GRCh38.p13 assembly [12]. Importantly, the T2T-CHM13 assembly filled a 60 kbp gap between MUC3A and MUC12 at locus 7q22 (Fig 2B). Within this gap, we identified a 39,267 bp long PTS-encoding exon flanked upstream by a 2,187 bp long sequence with 87% identity to MUC3A N-term. Downstream of the PTS-encoding exon, we identified a 6,303 bp long sequence that was 92% identical to MUC3A C-term and contained a SEA domain, a transmembrane domain, and a cytoplasmic tail with a conserved PDZ motif [30] (Fig 2C). Thus, our findings suggest that the T2T-CHM13 assembly contains a putative MUC3B gene at locus 7q22 with a high sequence identity with MUC3A. Although previous studies have proposed the existence of a human MUC3B gene [31, 32], most recently in the African pan-genome [33] that contains a 22,827 bp contig aligning with MUC3B exon 2 (S1A Fig in S1 File), the complete length, sequence and exon-intron architecture of MUC3B, including its repetitive PTS-coding exon 2, remain unresolved.

Distinct human MUC3A and MUC3B genes share high sequence homology

To better characterize the putative human MUC3B gene, we compared the exon-intron architecture of MUC3B to MUC3A. MUC3A has been reported to contain 11 exons, including a PTS-encoding exon with a length of at least 6 kbp [34]. Our analysis showed that MUC3A and MUC3B both have 12 exons, revealing a previously overlooked exon 4 (S1B, S1C Fig in S1 File). Exons of the two MUC3 genes have nearly identical nucleotide lengths, except the single PTS-encoding exon 2 which measures 15,873 bp (5,291 amino acids) in MUC3A and 39,267 bp (13,089 amino acids) in MUC3B (Fig 3A). Nucleotide sequence identity between MUC3A and MUC3B was on average 93% for exons, and 92% for introns. The superfamilies of Hominoids (apes and humans) and Cercopithecoids (Old World monkeys) diverged around 29 Mya [6]. Sequence alignments between N- and C-termini of MUC genes in the MUC3 cluster showed a high degree of conservation between H. sapiens and members of the Cercopithecoid and Hominoid branches. Human MUC3A N-term was 99% identical to chimpanzee MUC3A N-term and 90–91% identical to MUC3A N-term in Cercopithecoid members rhesus and baboon. MUC3A C-term showed a slightly higher degree of divergence compared to MUC3A N-term (Fig 3B and S2 Table). The same trend was observed for MUC3B, in which the MUC3B C-term was less conserved than MUC3B N-term. Tandem repeat regions are prone to duplications and deletions caused by recombination [35]. Accordingly, we observed higher evolutionary sequence divergence in the PTS-encoding exon 2 of MUC genes in the MUC3 cluster (Fig 3B and S2 Table). Moreover, pairwise alignment of Catarrhini MUC3 cluster genes revealed a general trend toward the expansion of tandem repeats during primate evolution (S1D, S1E Fig in S1 File). Specifically, within exon 2 of human MUC3A and MUC3B, we identified imperfect repeats with 87% amino acid sequence identity between MUC3A and MUC3B. In addition, MUC3B contained an additional 1368 amino acids of imperfect repeats (Fig 3C). MUC3A and MUC3B also harbored 166 and 549 perfect tandem repeats, respectively, consisting of a 17 amino acids long consensus sequence (ITTTETTSHSTPSFTSS) (Fig 3D). We conclude that the genetic structure of MUC3B is highly similar to MUC3A and that the two MUC3 genes are likely paralogous genes characterized by variable number of tandem repeats.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Comparison of genetic and structural features of MUC3A and MUC3B genes.

(A) Exon structure and length of exon 2 of MUC3A and MUC3B. (B) The evolutionary rate of N-terminal-, PTS- and C-terminal-encoding exons in MUC3A, MUC3B, MUC12, and MUC17 measured as gene content conservation (%) versus evolutionary distance (Mya). (C) Dot plot of pairwise sequence alignment of MUC3A and MUC3B identified imperfect (blue) and perfect (red) tandem repeat sequences in exon 2. (D) Sequence logo representation of 17 amino acids long consensus sequence in 166 and 549 perfect tandem repeats (TRs) in exon 2 of MUC3A and MUC3B, respectively.

https://doi.org/10.1371/journal.pone.0275671.g003

MUC3A and MUC3B are regulated by conserved regulatory elements

Sequences upstream of transcription start sites (TSS) contain regulatory elements that dictate gene expression. Promotor activity within positions -1 –-242 upstream of MUC3A TSS has been reported previously [32], yet evolutionary conservation of the promotor regions and transcription factors that potentially regulate MUC3 genes are largely unknown. Sequence analysis of presumed regulatory sequences -1 kbp upstream of human MUC3A TSS identified a candidate cis-Regulatory Element (cCRE) at position -1 –-403 bp (Fig 4A), which shared 83% identity with the corresponding region in MUC3B. Published DNase I hypersensitive site sequencing (DNase-seq) data sets from the human small intestine and colon revealed high signals within the cCRE (Fig 4A). Moreover, we identified high signals for active chromatin markers H3K9ac and H3K4me3 within the MUC3A cCRE in the human small intestine and colon, while active chromatin signals in the stomach were either low or not detected. By predicting transcription factor binding sites (TFBSs) using the JASPAR CORE vertebrate collection [36, 37], we identified putative TFBSs in cCRE of MUC3A (S2 Fig in S1 File). Seven of these transcription factors (ELF3, HNF4A, HNF4G, KLF4, PPARA, STAT3, and XBP1) were enriched in transporting IECs in human and mouse intestines (Fig 4B and S3 Fig in S1 File). Alignment of putative promoter regions upstream of MUC3A and MUC3B genes in Hominoid and Cercopithecoid superfamilies identified conserved TFBSs for HNF4A, HNF4G, and STAT3, strongly suggesting that the two MUC3 genes share an evolutionarily conserved regulatory expression program in the small intestine and colon (Fig 4C, S4 Fig in S1 File and S3 Table).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Conserved regulatory elements upstream of MUC3A and MUC3B genes.

(A) Epigenetic analysis of the human small intestine and colon reveals a DNase I-sensitive cCRE and specific histone modifications surrounding the MUC3A transcription start site. (B) Single-cell analysis of human and mouse intestines shows gene expression of transcription factors in transporting IECs, with conserved binding sites upstream of MUC3A and MUC3B. (C) Binding sites for transcription factors STAT3 and HNF4A/G are completely conserved upstream of MUC3A and MUC3B in Cercopithecoid and Hominoid superfamilies.

https://doi.org/10.1371/journal.pone.0275671.g004

Expression of human MUC3A and MUC3B genes in the human intestine

To determine whether MUC3B is transcribed into messenger RNA, we mapped published RNA-sequencing data sets from the human intestine [26], liver [27], and kidneys [38] to the T2T-CHM13 assembly. A considerable number of sequenced reads from MUC3A and MUC3B transcripts were detected in the human ileum, colon, and rectum (Fig 5A), while the liver and kidneys were devoid of transcripts from the MUC3 cluster genes (S4 Table). Our findings are supported by the human cell atlas [39], which shows that MUC3A is mainly expressed in epithelial cells of the small intestine and colon (1446 of 2316 MUC3A⁺ cells) (S5 Fig in S1 File). Our data contradict a previous observation of MUC3A transcripts in the heart, liver, prostate, and thymus, and MUC3B transcripts in the small intestine and colon [40]. Notably, primer BLAST analysis showed that the MUC3B probe used in the study was specific for exon 3 of MUC17, whereas the MUC3A probe was specific for MUC3A exon 2 (S4 Table). Since these exons encode the repetitive mucin-type PTS-domain, we cannot exclude unspecific detection of mucin transcripts in other tissues.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Expression of MUC3B gene in the human intestine.

(A) Unique reads for MUC3A and MUC3B in RNA-sequencing data from human ileum, colon, and rectum mapped to T2T-CHM13 human genome assembly. (B) Gene expression of MUC3 cluster genes in human ileum, colon, and rectum. 2 samples per tissue segment. * p<0.05 as determined by two-way ANOVA, corrected for multiple comparisons using Tukey´s test. Data are presented as mean ± standard deviation (SD). (C) Specific primers amplify a 646 bp cDNA spanning exons 3–8 in MUC3A and MUC3B transcripts from the ileum of five individuals. MUC3A cDNA carries a PstI restriction site in exon 6 that distinguishes MUC3A from MUC3B transcripts. Agarose gel electrophoresis of PstI restriction digests of amplified cDNA from MUC3A and MUC3B transcripts results in 380 bp and 266 bp fragments from MUC3A cDNA. (D) Quantification of bands from agarose gel in C. n = 5 individuals. Data are presented as mean ± standard deviation (SD).

https://doi.org/10.1371/journal.pone.0275671.g005

Because exons encoding the N-terminal, PTS, and C-terminal regions of MUC3A and MUC3B share 87%, 83%, and 92% identity and PTS-encoding exons are highly repetitive, it is challenging to detect unique reads that accurately distinguish between the two MUC3 genes. Therefore, we turned our attention to reads that map to exons 3–12, where we identified an average of 3.5±1.6 unique reads per kilobase transcript (RPK) of MUC3A and 13.0±4.4 unique RPK of MUC3B (Fig 5A and S4 Table). We next used unique and shared reads in the C-terminal region to calculate normalized gene expression of MUC3A, MUC3B, MUC12, and MUC17 in the human intestine. In the ileum, MUC17 showed significantly higher expression than MUC3A, MUC3B, and MUC12, while MUC12 showed a trend towards higher expression in the rectum compared to the ileum. We detected comparable numbers of MUC3A and MUC3B transcripts in all three intestinal segments (Fig 5B).

Next, we applied targeted reverse transcriptase quantitative polymerase chain reaction (RT-qPCR) to validate the presence of unique MUC3A and MUC3B transcripts in ileum collected from five human patients (S5 Table). For this purpose, we designed gene-specific primer pairs that target exons 3 and 8 with a 9.5–12.0% mismatch between MUC3A and MUC3B. The resulting 646 bp cDNA amplicons from each gene transcript were further distinguishable by a unique PstI restriction site in the MUC3A cDNA amplicon (Fig 5C). RT-qPCR from all five patients resulted in the expected 646 bp cDNA amplicon and subsequent PstI-digestion produced 380 bp and 266 bp restriction fragments (Fig 5C). Notably, we observed significant differences in intensities of PstI-sensitive and PstI-resistant fragments produced by the two gene-specific primer pairs. 75% of amplicons generated by MUC3A-specific primers were PstI-sensitive and therefore originated from MUC3A transcripts (Fig 5D). Similarly, 76% of amplicons generated by MUC3B-specific primers were PstI-resistant MUC3B transcripts. Thus, despite significant sequence similarity between the two MUC3 genes, we successfully identified and distinguished between MUC3A and MUC3B transcripts in the human intestine.

Completion of a gapless human MUC3 cluster at locus 7q22

Finally, based on the T2T-CHM13 assembly, we revised the gapless length and sequence of all four membrane mucins genes in the MUC3 cluster at locus 7q22. In this new assembly, we identified a longer PTS-encoding exon 2 of 15,870 bp in MUC3A compared to 8805 bp in GRCh38.p13. The PTS-encoding exon of human MUC12 was 32,428 bp long compared to 14,935 bp in GRCh38.p13 (S6 Table). The complete gapless sequences of MUC3A, MUC3B, MUC12, and MUC17 genes at the 7q22 locus are publicly available via Mucin Biology Groups’ Mucin database (http://www.medkem.gu.se/mucinbiology/databases/index.html).