INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: airway epithelium; asthma; digital cloning; gene expression; mouse model; mucin gene

Mucus, which covers the airway lumenal surface, is essential in protecting the epithelial surface from environmental insults. Overproduction of mucus, which occurs frequently in chronic airway diseases, causes airway obstruction and contributes to the morbidity and mortality of patients with asthma, bronchitis, and cystic fibrosis (1, 2). The major macromolecular components of mucus are large and heavily glycosylated mucin proteins that are encoded by various MUC genes.

Twelve human MUC genes had been cloned and named as MUC1, MUC2, MUC3, MUC4, MUC5AC, MUC5B, MUC6, MUC7, MUC8, MUC9, MUC11, and MUC12 (3, 4, and review in Rose and Gendler [5]). Among these human MUC genes, MUC2, MUC5AC, and MUC5B share extensive similarities in their DNA sequences and they have been thought to evolve from a common ancestor gene (6). All three genes are located in the ~ 400-kb region of chromosome 11p15 in the order of tel-5'-MUC2-MUC5AC-MUC5B-3'-cen (7). We have isolated a genomic clone, Cos-1, that contains the region between the 3' end of MUC5AC and the large central exon of MUC5B. From the restriction enzyme map and nucleotide sequence, we found that MUC5AC and MUC5B were close to each other, with only ~ 20 kb in between (Y. Chen, Y. H. Zhao, Y. P. Di, and R. Wu, unpublished).

MUC5AC and MUC5B gene products have been demonstrated to be the major components in human airway mucous secretions (8). In normal adult human tissues, MUC5AC message is expressed in the goblet cells of airway surface epithelium, whereas MUC5B is mainly expressed in the mucous cells of the submucosal gland regions (12). However, we have demonstrated the presence of MUC5B message in surface goblet cells of airway tissues with chronic airway diseases (Y. Chen, Y. H. Zhao, Y. P. Di, and R. Wu, unpublished). The nature of this transdifferentiation is unclear. Other reports have shown that MUC5B gene product was one of the major components in mucus obtained from subjects with asthma (11) and cystic fibrosis (13). These results implicate the involvement of MUC5B gene expression in the pathogenesis of airway diseases.

In contrast to the human gene, there is no information regarding the mouse Muc5b gene. In the mouse, only six Muc genes have been identified. Five of them were named after their human homologs as Muc1 (14), Muc2 (GenBank; AF016695), Muc3 (15), Muc4 (GenBank; AF218265), and Muc5ac (16); mouse Muc10 has no human homolog (17). In addition, the histology of mouse airway tissue is different from that of human airway tissue. In the mouse, surface goblet cells are absent from the trachea and relatively sparse in the proximal bronchi of normal mouse airway, and mucus-producing glands in the mouse are condsiderably fewer than in humans and restricted to the first few cartilaginous rings of the trachea.

Because the mouse has become an essential animal model for various human airway diseases, it becomes necessary to elucidate the gene structure of Muc5b and its expression. In the present study, we used the BLAST search and DNA-sequencing approach to identify a mouse Muc5b cDNA clone, 3pmmuc5b-1. Probes generated from this clone hybridized to normal mouse tracheal submucosal cells but not the surface epithelium. However, strong hybridization signals were obtained on the surface epithelium of allergen-induced asthmatic mouse airways.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

BLAST Searching of Mouse EST and Genomic Databases

The cDNA sequence from the 3' end of the human MUC5B gene (GenBank, Y09788) was extracted and assembled according to its exon table. This DNA sequence information was used as a probe to search the mouse expressed sequence tag (EST) database (http://www.ncbi.nlm.nih.gov/dbEST/) for a homologous clone, using the BLAST 2.1 program (http://www.ncbi.nlm.nih.gov/blast/). This approach led to the identification of an EST clone, AA038672, as a putative mouse Muc5b cDNA clone. This EST clone was later named 3pmmuc5b-1 after complete DNA sequencing.

For the genomic database search, the full sequence of the 3pmmuc5b-1 clone was used as a probe to search the HTGS (High- Throughput Genomic Sequence) database (http://www.ncbi.nlm.nih.gov/HTGS/) BLAST with the 2.1 program.

Construction of Phylogenic Tree

The phylogenic tree was constructed by using the following sequences containing the cystine-knot domain: human MUC2 (GenBank, NP_002448), mouse Muc2 (GenBank, AA615833), rat Muc2 (GenBank, P98089), human MUC5AC (GenBank, P98088), mouse Muc5ac (GenBank, CAA09365), rat Muc5ac (GenBank, AAC53312), porcine gastric mucin (pgm) (GenBank, AAD19833), and human MUC5B (GenBank, AAC51343). The multisequence alignment was performed by the CLUSTAL W program (18).

Comparison of Human MUC5B and Putative Mouse Muc5b Genomic Sequences

The human MUC5B cDNA full-length sequence was assembled from the GenBank database entries AF107890, AJ004862, Y09788, and Z72496 to generate a 17.5-kb cDNA sequence. The Dotplot program in the Lasergene package (DNAStar, Madison, WI) was used to compare full-length human MUC5B with a 30.1-kb genomic sequence upstream of 3pmmuc5b-1 obtained from AC020817.

DNA Sequencing

The EST clone was sequenced according to the manufacturer protocol by an ABI Prism model 377 automated DNA sequencer (Applied Biosystems, Foster City, CA).

Ovalbumin-induced Mouse Asthma Model

Female BALB/c mice (8 wk old) were purchased from Charles River Laboratories (Wilmington, MA) and housed in an environmentally controlled, pathogen-free animal facility. The ovalbumin sensitization and challenge protocol was based on the protocol described by Temelkovski and coworkers (19).

RNA Isolation and Northern Blot Hybridization

Total RNA was isolated from 16 different mouse tissues by a single-step phenol-chloroform extraction method (20). For Northern blot hybridization, equal amounts of total RNA (20 µg/lane) were subjected to electrophoresis on a 1.2% agarose gel in the presence of 2.2 mM formaldehyde and transblotted onto Nytran membranes. The RNA was cross-linked to membrane by a UV Stratalinker 2400 (Stratagene, La Jolla, CA). Membrane prehybridization and hybridization with ³²P-labeled probes were carried out as described previously (21). Two single-stranded antisense oligonucleotide probes, corresponding to the 3' and 5' ends of putative mouse Muc5b message (GCA GGA ACC CTC GCA GAA GGT GAT GTT GAC CTC TGT CTC ACA GCC CTT and AAT GGA AGT CAC CCC ACG TGC TGC ACA CTC TCC CAT TGT G, respectively) were synthesized. The 3' end oligonucleotide corresponds to the sequence of 3pmmuc5b-1 (from nucleotide 45 to 92), whereas the 5' end is deduced from the mouse genomic clone AC020817 (from nucleotide 69319 to 69358), which contains the putative mouse Muc5b gene. Sense oligomers were also synthesized and used as controls for the hybridization. These oligonucleotide probes were end labeled with [-³²P]ATP by T4 polynucleotide kinase (New England BioLabs, Beverly, MA). The relative abundance of Muc5b message in Northern blot was normalized with the 18S rRNA band, as described previously (21).

In Situ Hybridization

Both sense and antisense oligonucleotides (100 pmol each), corresponding to the 3' end of the 3pmmuc5b-1 clone (from nucleotide 45 to 92), were end labeled with the digoxigenin (Dig) oligonucleotide tailing kit (Roche Molecular Biochemicals, Indianapolis, IN), according to the manufacturer protocol. In situ hybridization was carried out as described previously (22). Briefly, slides were digested with proteinase K (10 µg/ml) in 50 mM Tris-HCl (pH 8.0), 50 mM EDTA for 15 min at 37° C, rinsed twice in 0.2× saline-sodium citrate (SSC), and then postfixed in 4% paraformaldehyde-phosphate-buffered saline (PBS) for 20 min. Slides were treated twice for 5 min each with 0.1 M triethanolamine (TEA), pH 8.0, and then blocked by 0.25% acetic anhydride in 0.1 M TEA. The sections were then dehydrated through an ethanol gradient series. For each section, 0.5 pmol of digoxigenin-labeled oligonucleotide probe in 50 µl of hybridization buffer was applied. The hybridization buffer contains 2× SSC, 1× Denhardt's solution, 10% dextran sulfate, 50 mM phosphate buffer (pH 7.0), 50 mM dithiothreitol (DTT), yeast tRNA (250 µg/ml), poly(A) (100 µg/ml), and salmon sperm DNA (500 µg/ml). The section was hybridized at 45° C overnight in a humidified chamber. After hybridization, the section was washed twice for 15 min each at 37° C with 2× SSC, twice for 15 min each with 1× SSC, and twice for 15 min each with 0.25× SSC. After the wash, the slide was reacted with anti-Dig primary antibody conjugated with alkaline phosphatase. After several washes, the reacted probes in the slide were color developed with the digoxigenin nucleic acid detection kit obtained from the same manufacturer (Roche Molecular Biochemicals).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Identification of EST Clone Containing Part of Putative Mouse Muc5b Gene

It has been reported previously that extensive similarities exist in the nonrepetitive regions (e.g., von Willebrand factor D domain and 3'-end cystine-knot domain) between human MUC genes and their orthologous genes of other species (23, 24). Because most EST clones were generated by oligo (dT) primer-based reverse transcription, 3'-end gene sequences are most abundant in the EST database. Considering these features, we chose a 3'-end sequence of the human MUC5B gene (GenBank, Y09788) as a probe to screen the mouse EST database by BLAST. An EST clone, AA038672, showing 80% similarity with this human MUC5B gene probe, was obtained. This EST clone was generated from the cDNA library of whole mouse fetus (Figure 1). After completely sequencing the insert of this clone, we found that it was 810 bp in length plus a polyadenylation site. A hypothetical translation of the DNA sequence in this clone generated a 97-amino acid peptide containing the cystine-knot domain, which has nine conserved cysteine residues completely identical to those of human MUC5B 3' end, in regard to their positions (Figure 1). We subsequently named this EST clone 3pmmuc5b-1, standing for putative 3' end of mouse Muc5b, clone 1. 

View larger version (31K): [in this window] [in a new window]   Figure 1.   A sequence comparison between the 3' end of the human MUC5B gene and the 3pmmuc5b-1 clone insert. The 3' end of human MUC5B cDNA was assembled from the sequence database (GenBank, Y09788). The hypothetical translation of the 3pmmuc5b-1 insert demonstrated a similar cystine-knot domain feature found at the 3' end of the human MUC5B gene.

To confirm further that the newly identified EST clone belongs to the mouse homolog of human MUC5B, we constructed a phylogenic tree by using the amino acid sequences derived from similar regions (i.e., all contain cystine-knot domains) of various mucin genes of different species, including MUC5AC (human, mouse, rat, porcine), MUC2 (human, mouse, rat), MUC5B (human), and also the sequences of von Willebrand factors (human, mouse, porcine). All sequences were obtained from the GenBank database with the accession numbers as indicated in METHODS. As shown in Figure 2, the phylogenic tree reflected perfectly the previously proposed evolutionary relationships among von Willebrand factors and mucin genes (6). In this tree, all mucin genes are close to each other and distant from three von Willebrand factors. Among these mucin genes, MUC5AC and MUC5B are more similar than MUC2, which is consistent with a report from another laboratory (6). Interestingly, although MUC2, MUC5AC, and MUC5B are considered as one gene family that evolved from the common ancestor of the von Willebrand factor gene (6), they share fewer similarities than their orthologous genes in different species. For example, the cystine-knot domain of the 3' end of human MUC5AC is more similar to porcine Muc5ac than to human MUC5B. Furthermore, the newly identified mouse EST clone is much closer to human MUC5B than to other mucin genes in this phylogenic tree.

View larger version (11K): [in this window] [in a new window]   Figure 2.   A phylogenic tree analysis of the von Willebrand factor (vWF) and various mucin genes (MUC2, MUC5AC, and MUC5B from different species). h = human; m = mouse; r = rat; p = porcine; pgm = porcine gastric mucin, suggested to be the porcine homolog of human MUC5AC. All the sequences were obtained from GenBank as indicated in METHODS. Large brackets indicate orthologous gene clusters named according to their human orthologs.

Identifying Two Genomic Clones Containing Mouse Muc5ac and Muc5b

To confirm further that the 3pmmuc5b-1 clone insert is the 3' end of the mouse Muc5b gene, we took advantage of current progress in the mouse genome project by BLAST searching the mouse genome database (HTGS), using the 3pmmuc5b-1 sequence as a probe. Two related genomic clones (GenBank, AC020817 and AC020794) were found. On the basis of the alignment analysis, we found that these two clones actually overlapped with each other and both showed a perfect match (100%) with the 3pmmuc5b-1 sequence in corresponding regions (Figure 3). This analysis revealed that the 3pmmuc5b-1 sequence was composed of two exons, of which one contained the 5' part of the cystine-knot domain and the other contained the 3' part of the cystine-knot domain as well as the polyadenylation site and the 3'-end untranslated region (Figure 3). In humans, the genomic position of the MUC5AC and MUC5B regions has been suggested as an arrangement of tel-5'- MUC5AC-MUC5B-3'-cen (7), with a small (~ 20-kb) spacer sequence in between (Y. Chen, unpublished). Furthermore, this genomic region was conserved between human and mouse with regard to the gene order, based on the comparative map of human and mouse (http://www.ncbi.nlm.nih.gov/Homology/). Because AC020817 and AC020794 have insert sizes of 219,218 and 191,672 bp, respectively, and contain the 3pmmuc5b-1 sequence at one end, we reasoned that these two genomic clones should also contain the 3'-end sequence of the previously cloned mouse Muc5ac gene (GenBank, CAA09365). Sequence alignment analysis of these two genomic clones demonstrated that they did contain these sequences (Figure 3). On the basis of this analysis, the genomic sequence of mouse in this region was indeed found to be 5'-Muc5ac-3pmmuc5b-1-3' (Figure 3).

View larger version (13K): [in this window] [in a new window]   Figure 3.   The putative location of the 3pmmuc5b-1 sequence in two mouse genomic clones, AC020817 and AC020794. The two boxes represent two exons. The numbers (100824, 100929, etc.) represent nucleotides positioned at each junction.

Comparison between Human MUC5B cDNA and Mouse Genomic Sequence Upstream of 3pmmuc5b-1 by Dotplot Program

To confirm further the authenticity of the mouse Muc5b gene in these two genomic clones (AC020817 and AC020794), the nucleotide sequence upstream of the 3pmmuc5b-1 locus was deduced from these clones and used to compare the cDNA sequence in the human MUC5B gene by the Dotplot program. The comparison stringency parameter was set as 70%, which meant that only the regions sharing 70% or more similarity between the two sequences would be drawn on the Dotplot figure. As shown in Figure 4, these two nucleotide sequences of mouse and human did share an impressive similarity along with their sequences. The disruption on the axis of the mouse genomic sequence, but not on the axis of the human MUC5B cDNA sequence, was related to the mouse intron sequence. Interestingly, a region indicated by a bracket in Figure 4 showed a scattering or irregular alignment between these two sequences, and this region (10 kb) actually corresponds to the large central exon known as human MUC5B, which contains many repeats with different lengths. Because repetitive regions in various mucins are not conserved in all species, it is reasonable to expect a large gap in between human and mouse in the Dotplot graph. On the basis of this comparison, the putative 5' end of mouse Muc5b is about 25 kb downstream of the 3' end of Muc5ac, which was similar to the human sequence.

View larger version (28K): [in this window] [in a new window]   Figure 4.   A Dotplot analysis comparing sequence homology between human MUC5B and the putative mouse Muc5b genomic sequence upstream of 3pmmuc5b-1. The full-length human MUC5B cDNA sequence was assembled as indicated in METHODS. The position of the mouse genomic sequence used for comparison is indicated. The vertical axis is the human MUC5B cDNA sequence while the horizontal axis is the mouse genomic sequence. The Dotplot program was set to a stringency of 70% with a search window of 70 bp per comparison, which meant that all the small lines drawn on the graph were the regions at least 70 bp long and having 70% similarity or more between these two sequences. The large bracket indicates a region with an absence of similarity or an abnormal alignment. This region corresponds to the 10-kb size of the large central exon region of the human MUC5B gene.

Characterization of Gene Expression of Mouse Muc5b Gene

On the basis of the preceding information, an oligonucleotide corresponding to the cDNA sequence of the 5' end of mouse Muc5b was synthesized. This oligonucleotide, together with the oligonucleotide deduced from the 3pmmuc5b-1 clone, was used in Northern blot hybridization with RNA samples from 16 mouse tissues and organs. The stained gel showed an equal loading of these tissue RNA samples, as well as the integrity of these RNA preparations (Figure 5C). Hybridization of these RNAs with radioactive antisense cDNA probes revealed that only two tissues had a detectable level of putative Muc5b message (Figure 5A and 5B), the tongue and trachea. RNA from other tissues, such as liver, heart, spleen, etc., had no detectable level of message under these experimental conditions. In essence, these two probes, one generated from 3pmmuc5b-1 (3' end of Muc5b) and the other from the putative 5' end of Muc5b, deduced by sequence alignment, showed exactly the same hybridization pattern. This result further confirmed that these two cDNA sequences belonged to the same gene.

View larger version (98K): [in this window] [in a new window]   Figure 5.   Northern blot analysis of mouse Muc5b gene expression. Mouse tissues 1 to 16 are as follows: heart, esophagus, kidney, prostate, bladder, lung, trachea, spleen, parotid gland, pancreas, stomach, colon, small intestine, ovary, tongue, and liver, respectively. RNA samples were prepared as described in text and hybridized to the 32P-labeled cDNA probe derived from the insert of 3pmmuc5b-1 (A), whereas in (B), the probe was deduced from the putative 5' end of the mouse Muc5b gene as described in text. (C ) Staining of RNA gel with ethidium bromide to indicate the integrity and equal loading of these RNA samples.

To elucidate further the tissue specificity of Muc5b gene expression, in situ hybridization was carried out on some of these mouse tissue sections. Consistent with the Northern blot hybridization, those tissue sections, such as liver, spleen, etc., had no positive hybridization signal with the putative Muc5b 3'-end cDNA probe (data not shown), whereas positive signals were seen in both tongue and tracheal sections (Figure 6). As shown in Fig. 6, the hybridization signals were located exclusively in the submucosal gland region of the trachea (Figure 6, B and C) and the glandular cells of the sublingual gland region of the tongue (Figure 6E). Furthermore, the hybridization signals in the sublingual gland were relatively higher than in the submucosal gland region of the trachea. For the surface epithelium of the airway section, there was no significant hybridization signal (Figure 6B, arrow; Figure 7, C and D). Similarly, the control sense probe had no hybridization in these tissues (data not shown).

View larger version (59K): [in this window] [in a new window]   Figure 6.   In situ hybridization of mouse Muc5b expression in the glandular regions. Normal mouse tracheal (A and B ) and tongue (D and E ) sections were prepared as described in text. (C ) Tracheal section from an ovalbumin-induced asthmatic mouse, as also described in text. Tissue sections in both (A) and (D) were histochemically stained with alcian blue-PAS; (B, C, and E ) in situ hybridization results with the antisense oligonucleotide probe derived from the 3pmmuc5b-1 insert, as described in text. The arrow in (B) indicates the surface epithelium of mouse trachea, which had no hybridization signal with the probe.

View larger version (70K): [in this window] [in a new window]   Figure 7.   In situ hybridization of mouse Muc5b expression on the surface epithelia of respiratory tracts. Tracheal and lung tissue sections were prepared from sham-treated (A-D) and ovalbumin-induced (E-J ) asthmatic mice as described in text. (A, B, E, F, and G) Tissue sections histochemically stained with alcian blue-PAS; (C, D, H, I, and J ) sections hybridized in situ with the antisense oligonucleotide probe of mouse Muc5b, as described in Figure 6. (A, C, E, and H ) Tracheal sections; (F and I ) bronchial sections; (B, D, G, and J) bronchiolar sections. Because serial tissue sections are presented here, the boxes in (E), (F ), and (G) indicate the corresponding regions enlarged for showing in situ hybridization results (H, I, and J ).

Elevation of Muc5b Gene Expression in Mouse Asthma Model

We (Y. Chen, Y. H. Zhao, and R. Wu, unpublished) have observed that MUC5B message can be expressed on the airway surface epithelium in human patients with various airway diseases, in addition to the submucosal gland region. This phenomenon was often associated with the goblet hyperplasia/ metaplasia in various airway diseases. Thus, the lack of Muc5b message expression on the surface airway epithelium might be related to the lack of goblet cell differentiation in normal mouse airways. The question is, then, whether Muc5b message can be expressed on the surface epithelium when there are many goblet cells present in mouse airways. To address this question, an ovalbumin-induced mouse asthmatic model (19) was used. In sham-treated control mice, Muc5b gene expression was restricted to the submucosal gland region (Figure 6B), and no messages were detected on the surface epithelial cells of airways (Figure 7, C and D). However, the expression was dramatically elevated in the surface epithelia from trachea to bronchiole (Figure 7, H, I, and J), as well as in the submucosal gland region (Figure 6C) of asthmatic mice. This elevation of Muc5b message coincided with a large increase in the alcian blue-PAS-positive goblet cell-like population (Figure 7, E, F, and G) on the surface epithelia of asthmatic mouse airways, whereas those cells were absent from control mice (Figure 7, A and B). These results suggest that the Muc5b gene expression pattern in mouse airways is similar to that in human tissue sections derived from patients.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The mouse has been extensively used as a model in mimicking the pathogenesis of various human diseases. Concerning the mouse airway, there is a lack of information about mucin genes and their expression in the normal mouse, as well as in mice with diseases. In this report, we utilized a bioinformatic strategy to quickly clone and identify the mouse Muc5b gene, using its human homolog as a probe. This is different from the time-consuming, traditional homologous gene cloning approach, which needs degenerate-priming RT-PCR or low-stringency screening of both cDNA and genomic libraries of mouse. The bioinformatic (or in silico) approach takes advantage of genetic and sequence information available for both humans and mice from public databases, and the conserved nature of mucin genes among different species. By searching those databases with the BLAST program, it was found that the mouse EST clone AA038672 contains an insert with a DNA sequence similar to the 3' end of the human MUC5B gene. Further DNA sequencing of the entire insert in this clone confirmed the putative nature of the 3' end of the mouse Muc5b gene. We therefore named this clone 3pmmuc5b-1.

The first evidence that 3pmmuc5b-1 is the 3' end of the mouse Muc5b gene rests on the sequence similarity between this clone and the 3' end sequence of the human MUC5B gene. Essentially, 3pmmuc5b-1 contains a complete cystine-knot domain that is similar to the corresponding domain at the 3' end of human MUC5B. Further evidence comes from phylogenic analysis. Because it has been reported that there are similarities among the cystine-knot domains of various mucin genes at the 11p15 locus (e.g., MUC2, MUC5AC, and MUC5B) (6), a phylogenic tree generated by a multisequence alignment is the best tool to demonstrate the evolutionary relationship among these sequences (18). The phylogenic tree showed in Figure 2 not only demonstrated a close evolutionary relationship between 3pmmuc5b-1 and human MUC5B, but also pinpointed the fact that orthologous mucin genes in different species shared more similarities that those from the different MUC genes in the same species.

The other evidence that the insert in the 3pmmuc5b-1 clone is indeed the 3' end of the mouse Muc5b gene comes from its genomic position. In humans, there is ~ 60 kb between the 3' end of MUC5AC and the 3' end of MUC5B, and ~ 20 kb between the 3' end of MUC5AC and the 5' end of MUC5B (Y. Chen, Y. H. Zhao, Y. P. Di, and R. Wu, unpublished). With regard to gene order, the human/mouse homolog map (NCBI), from the CD151-encoding gene to H-ras (flanking the MUC5AC gene), a region of ~ 48.7 cM (1 cM = ~ 1 million bp), has been shown to be conserved between human chromosome 11 and mouse chromosome 7, without any reported inversion, translocation, or other events that can disrupt the gene order (http://www.ncbi.nlm.nih.gov/Homology/). The two independent genomic clones (GenBank, AC020817 and AC020794) identified by searching the High-Throughput Genomic Sequence database contain both the mouse Muc5ac and 3pmmuc5b-1 sequences in the order 5'-Muc5ac- 3pmmuc5b-1-3' end. This arrangement is identical to that of human MUC5AC and MUC5B genes at the 11p15 locus. The distance between the 3' end of mouse Muc5ac and the 3pmmuc5b-1 sequence is ~ 55.9 kb, which is comparable to the genomic position of the 3' end of human MUC5B relative to the 3' end of MUC5AC. Because the gene positions are maintained in two different mouse genomic clones, it is less likely that the ordering of these genes is due to a cloning artifact. We also compared the sequence upstream of 3pmmuc5b-1 with the full-length human MUC5B genomic sequence, using the Dotplot program. The Dotplot program demonstrated a high similarity (> 70%) in various regions between those two sequences, except in the intron regions and in the large central exon, which are known to be variable between species. Furthermore, the conserved region of the mouse genomic sequence upstream of the 3pmmuc5b-1 sequence stops about 25 kb downstream of mouse Muc5ac. The ~ 25-kb distance between this putative 5' end of mouse Muc5b and the 3' end of Muc5b is also comparable with the distance of ~ 20 kb in humans.

Using this information, an oligonucleotide probe with a DNA sequence corresponding to the 5' end of the putative mouse Muc5b gene was produced and used in a hybridization study. Northern blot hybridization with 16 mouse tissue RNA samples demonstrated that this oligonucleotide probe had the same hybridization result as the cDNA probe from the 3' end of the Muc5b 3pmmuc5b-1 sequence. Both probes hybridize to tracheal and tongue RNA samples, producing identical molecular weight bands. Both probes would not hybridize to the rest of the tissue RNA samples. These results further suggest that these two probes are the same gene. In situ hybridization with the antisense sequence of 3pmmuc5b-1 demonstrated the expression of this message in the submucosal gland region, but not in the surface epithelium of normal mouse airway sections. These results are also consistent with MUC5B expression in the human airway. On the basis of all these analyses, we conclude that we identified the mouse Muc5b gene and that the 3pmmuc5b-1 clone indeed belongs to the 3' end of the mouse Muc5b gene.

In humans, the MUC5B gene is expressed in many tissues (airway, gall bladder, tongue, endocervix, etc.). In some tissues, the MUC5B messages could be detected mainly in the gland (e.g., conducting airway and tongue) (12, 25); in other tissues, MUC5B gene expression could be observed both on the surface and in the glandular regions (e.g., endocervix) (25). However, in the case of mouse Muc5b expression, the expression so far has been observed only in trachea and tongue samples. This suggests that mouse Muc5b is more restricted in its tissue-specific expression than its human homolog. This finding is in accord with the study of MUC5AC/Muc5ac gene expression patterns, which also showed more restricted expression in the mouse (16).

Using this newly cloned mouse Muc5b gene as a probe, we found that the expression of this gene in normal mouse airways was mainly on the mucous cells of the submucosal gland. However, the expression was elevated both in the submucosal gland and on the surface epithelium of ovalbumin-induced mouse asthmatic airways. And this phenomenon is similar to MUC5B expression in human airway tissue sections (Y. Chen, Y. H. Zhao, Y. P. Di, and R. Wu, unpublished). This is the first demonstration of elevated Muc5b gene expression associated with the mouse airway disease model. Previously, it was shown that Muc5ac gene expression was also elevated in this allergen-induced asthmatic airway model (26). In accordance with this study, we have also observed a significant increase in the alcian blue-PAS-stained goblet cell population. Essentially, the elevated Muc5b gene expression on the surface epithelia was exclusively restricted to those cells. Because the normal mouse airway epithelium is devoid of any goblet cell type, the appearance of large numbers of goblet cells represents a metaplastic change. This "plastic" change in epithelial cells has been addressed previously (27). A transdifferentiation mechanism seems possible considering the lack of goblet cells in the airway of the normal mouse. The studies of Muc5b by us and of Muc5ac by other researchers (26) suggest that an elevated expression of both Muc5b and Muc5ac genes might be involved in this metaplastic change. Furthermore, the observation that the superficial MUC5B/Muc5b gene expression occurring with both the human disease and mouse disease models strengthens the notion that this gene might be a good marker for studying airway disease-related metaplastic change. Therefore, identifying and elucidating the mouse Muc5b gene in this communication may help to discover the mechanism underlying goblet cell metaplasia and eventually contribute to successfully treating such chronic airway diseases as asthma, chronic obstructive pulmonary disease, etc.