Mucus hypersecretion is a cardinal feature of chronic airway diseases and mucins are the major component of airway mucus [1–4, 17, 18, 23]. Despite the significant progress in understanding the transcriptional control, biosynthesis and secretion of mucins using murine models, advances in understanding the coordinated gene expression relevant to producing and secreting mucins in the human airway have been limited due to the lack of methods to isolate and culture pure populations of primary human airway mucin-producing cells. Taking advantage of the knowledge that MUC5AC is the major secreted mucin of the small airway epithelium, and our observation that there is variable small airway epithelium MUC5AC expression among healthy nonsmokers, we hypothesized that genome-wide comparison of small airway gene expression of healthy individuals with “high” MUC5AC expression to those with “low” MUC5AC expression would identify genes whose regulation paralleled that of MUC5AC.
From a list of 528 genes up-regulated in healthy high MUC5AC expressors compared to healthy low MUC5AC expressors, we identified 73 MUC5AC core genes that had literature support for a role in mucin production/secretion, from murine studies of the airways, immunohistochemical or in situ hybridization studies of human airway epithelium, studies from mucin-secreting cell lines, or studies relating to mucin production/secretion in organs other than lung. These genes included those categorized as mucus components, mucus-producing cell differentiation-related transcription factors, mucus-producing cell differentiation-related pathways or mediators, post-translational modification of mucin, vesicle transport, ER stress-associated, secretory granule-associated, mucus secretion-related regulators and mucus hypersecretory related ion channels. While some of these 73 human small airway epithelium core MUC5AC genes have been previously associated with human airway MUC5AC expression, for most, this is the first demonstration that, in the human airways, these genes are coordinately up-regulated along with MUC5AC expression. As an independent validation set, we assessed the expression of these 73 core MUC5AC genes in the small airway epithelium of healthy smokers, where there is a stress on the airway epithelium associated with up-regulation of MUC5AC expression [1–4, 17, 18, 22, 23]. The data demonstrates that, as in the “high” vs “low” healthy nonsmokers, the healthy smokers up-regulate 38% of the MUC5AC-core genes. Finally, and consistent with studies in knockout and transgenic mice, MUC5AC expression in the human small airway epithelium was linked to several mucus production/secretion-related transcription factors (SOX2, KLF4, SPDEF, XBP1 and CREB3L1), as well as the Notch, Wnt and MAPK pathways, that, in turn, control coordinated expression of many of the MUC5AC-core genes identified in the present study. These findings expand the knowledge regarding molecular events underlying human airway MUC5AC-related production/secretion and provide several potential targets for medical intervention for mucus hypersecretion.
MUC5AC expression in the human small airway epithelium
MUC5AC is one of the major secretory mucins expressed by surface airway epithelial cells, and represents a marker for the airway surface mucus-producing cells [1–4, 17, 18, 22, 23]. As demonstrated in the present study and in several other studies [25, 28, 29], there is up-regulation of MUC5AC expression in the airway epithelium of smokers compared to nonsmokers. This observation is consistent with the knowledge that, on the average, smoking is associated with increased airway mucus production [1–4, 17, 18, 22, 23]. However, novel to the present study is the observation that MUC5AC expression in the small airway epithelium, assessed at both the mRNA and protein levels, is highly variable among healthy nonsmokers, implying that there are some individuals more prone to airway mucus production than others. We capitalized on this observation to categorize healthy nonsmokers into “high” and “low” MUC5AC expressors. Then, using genome wide assessment of gene expression, we compared “high” vs “low” MUC5AC expressors to identify genes, the expression of which correlates with MUC5AC expression. Using the literature to identify genes already known to be associated with the airway epithelium (mostly from murine studies), as well as to identify genes associated with mucin expression in other organs, we were able to define “MUC5AC-core” genes i.e., the genes plausibly linked to MUC5AC gene expression in the human small airway epithelium. To test the validity of the MUC5AC-core genes, we hypothesized that many of these genes would be up-regulated further in the small airway epithelium of smokers, since smokers have, on the average, up-regulation of MUC5AC. This proved to be correct.
Mucus-producing cell differentiation-related transcription factors, pathways and mediators
Studies of MUC5AC expression in experimental non-human models have identified several transcription factors that modulate mucus-producing cell differentiation, including Stat6, Spdef, Klf4, Sox2 and Foxa2 [4, 5, 20, 31, 35]. In addition to these transcription factors, murine studies have identified several signaling pathways, including Wnt , Notch [7, 8] and MAPK  pathways, which have been implicated in differentiation of MUC5AC-producing cells.
Consistent with these observations, the MUC5AC-core genes linked to MUC5AC expression in the human small airway epithelium in the present study include the mucus-producing cell differentiation-related transcription factors, SPDEF, KLF4, SOX2 and FOXA3. Up-regulation of FOXA3 and SPDEF have been reported during mucus overproduction in human airways cells in vitro and in vivo. SPDEF can also drive differentiation of MUC5AC-producing cells in primary human airway epithelial cells in vitro. The demonstration that SOX2 and KLF4 are linked to MUC5AC in the human airway epithelium is novel to the present study. The MUC5AC-core genes also included the Wnt pathway gene LRRFIP2, Notch pathway genes HES1 and TSTA3 , and the MAPK pathways genes MAPK13, KRAS , and RPS6KA3 (RSK2)  as linked to MUC5AC expression in the human small airway epithelium. Together, this suggests that a gene network, instead of single gene or pathway, controls differentiation of MUC5AC-producting cells.
The EGF receptor family mediated signaling pathway is involved in mucus production . One of the EGF receptor family members, ERBB2, has been shown to be overexpressed in lung mucinous adenocarcinomas [33, 34]. However, no correlation between ERBB2 and MUC5AC at transcription level was found in the current study using normal airway epithelium. These results suggest that MUC5AC expression is not dependent on the expression level of EGF receptors in the normal airway epithelium. Instead, activation status of EGF receptors might be more important, while in mucinous adenocarcinomas, the constitutive activation of the ERBB2 pathway, which is caused by persistent up-regulation of ERBB2 and ERBB2 mutations, might contribute to the MUC5AC overexpression.
In addition to these transcription factors and pathways, it is known that several enzyme-related genes can mediate mucus-producing cell differentiation [38–40]. Consistent with these observations, we found that SERPINB4 (serpin peptidase inhibitor), CTSC (peptidase) and PLA2G4A (phospholipase) are up-regulated in high MUC5AC expressors. Serpinb3a (SERPINB4 homolog) deletion markedly attenuates mucus-producing cell hyperplasia in mouse airway after allergen challenge ; ozone and Staphylococcal enterotoxin B exposure is associated with CTSC-driven mucus production in the mouse ; and PLA2G4A activation is associated with mucus overproduction in murine airway . Both SERPINB4 and CTSC are up-regulated in IL13 treated human primary bronchial epithelial cells on air liquid interface culture , and PLA2G4A is more active in bronchial explants from individuals with cystic fibrosis . It is a novel observation that these genes are linked to small airway MUC5AC production/secretion in healthy humans.
MUC5AC is a polymeric mucin which is heavily glycosylated and thus the ER and Golgi apparatus play a central role in biosynthesis of MUC5AC. To form the mature MUC5AC protein, a stepwise posttranslational modification is needed, including N-glycosylation, dimer formation, O-glycosylation, fucose/sialic acid/sulfate modification and polymer formation, processes involving several specific enzymes [2, 3]. Among these, several were identified among the MUC5AC core genes.
Consistent with the findings from mouse or human in vitro studies, we found that AGR2 (protein disulfide isomerases) , GALNT4 and GALNT7 (GalNAc-T members)  and GNE (required for normal sialylation)  were up-regulated in high-MUC5AC expressors. Besides these genes, we found some novel enzyme genes that are likely involved in human MUC5AC posttranslational modifications, including PDIA5 (disulfide isomerase), GALNT5, GALNT6 and GALNT12 (GalNAc-T members), FUT3 and FUT6 (fucosyltransferase), ST6GAL1 and ST8SIA1 (sialic acid transferase). Interestingly, PDIA5 is also up-regulated in Barrett’s esophagus, a disorder in which mucus-producing cells are a basic pathologic change .
Along the secretory pathway, vesicle-mediated transportation is important to carry newly synthesized proteins from the ER to the Golgi . To maintain the function of the ER, reverse transportation is also active, which balances the loss of protein embedded in the vesicles from the ER . Consistent with this concept, we found 15 vesicle transport related genes associated with high MUC5AC expression, including MIA3, SURF4, KDELR2, KDELR3, ITSN1, ERGIC1, CKAP4, GOSR1, SYNJ2BP, MPPE1, SEC31A, ARF4, VPS13D, SEC22B, and TPD52. None have been associated with mucus production in the literature.
Mucin biosynthesis places a large metabolic burden on the cell, with potential ER and Golgi apparatus stress . In agreement with the concept that mucus deregulation might induce ER stress, EIF2AK3 and XBP1 (classical unfolded protein response genes), EDEM3 (endoplasmic reticulum-associated degradation gene), and CREB3L1 (non-canonic unfolded protein response gene) were up-regulated in high-MUC5AC expressors. EIF2AK3 can repress global protein synthesis , XBP1 can induce chaperone synthesis , and EDEM3 can promote the degradation of poor-quality proteins . Finally, the Drosophila ortholog of CREB3L1, CrebA, has been shown to be the major and direct regulator of secretory capacity in the salivary gland . Our findings raise the possibility that the human airway surface MUC5AC-producing cells likely use an ER stress-associated signal as feedback to finely control mucus production.
MUC5AC secretion is highly regulated . As shown in the “secretory granule-associated” and “mucus secretion-related regulator” categories, activation of the mechanisms underlying mucin secretion was reflected in the function of MUC5AC-core genes. For example, we found PRSS23, PLCE, PRKCD, DKGA, ITPR3, RAB3D, RAB27B, MYO5B, MYO5C, SYTL2, SYTL4, SYTL5, SCIN, CDC42EP5, ATP6V0A4, STXBP6, PAM, RIMS1, PLCO and CASK were up-regulated in high-MUC5AC expressors. PRSS23 is a serine protease, which can stimulate mucus glycoprotein release from hamster tracheal ring organ culture . PLCE is a PLC family member that might be involved in generating the intracellular second messengers, IP3 and DAG, in mucus-producing cells . ITPR3 is the receptor of IP3 . PRKCD is a protein kinase C family member. RAB3D and RAB27B belong to RAB3 and RAB27 family, respectively . ATP6V0A4 and PAM are associated with maturation of secretion granules . Both MYO5B and MYO5C are type V myosins, and SYTL2 (SLP2A) , SYTL4 and SYTL5 are granulophilins . KIF5B is associated with granule movement . SCIN (scinderin) is an enzyme that is needed for actin disruption  and gelsolin can affect actin remodeling . RIMS1 (RIM), PCLO (Piccolo), and CASK can help secretory granule tethering and docking . STXBP6 belongs to Munc18 family. Among this group of MUC5AC-core genes, SYTL4, SYTL5, RIMS1, PCLO, CASK, KIF5B, CDC42EP5, ATP6V0A4, STXBP6, PAM, DKGA, and PLCE have not been previously identified in association with mucus-producing cells.
Mucus hypersecretion-related ion channels
Several studies have suggested that ion channels are involved in mucus production, including CFTR , SLC26A4 (pendrin) , SCNN1B (ENaC subunit) , CLCA family , GABAergic system receptors  and SLC12A2 . In the current study, we found SCNN1A (ENaC alpha subunit), CLCA2 (CLCA family member), GABAP (a subunit of the GABAergic system receptor), SLC12A2 (Slc12a2 homolog) were up-regulated in high MUC5AC expressors. Among them, SLC12A2 expression is known to be restricted to human airway mucus-producing cells . The links between SCNN1A, CLCA2, GABAP and MUC5AC expression in human airway are novel. Since multiple ion channel genes are found to be associated with MUC5AC expression, these observations suggest an active water/ion exchange in the airway epithelial cell membrane during human airway mucus production.
Several limitations of this study should be pointed out. First, all MUC5AC-core genes were identified based on the literature, and thus, important genes might be ignored because we are lacking clues from current knowledge. Second, our strategy was based on the transcriptome level, so regulation mechanisms depending on protein level modulation are underestimated.