To our knowledge, this study is the first to demonstrate the presence of multiple GR isoforms in human nasal tissues with a combination of different GR antibodies. Our study found that at least five GR isoforms with MWs of approximately 91–94, 80, 67, 60 and 52 kDa exist in NPs and control mucosae in addition to the archetypal full-length 94 kDa GRα-A. According to research on multiple GR isoforms, we established that band 1 corresponded to GRα-A/B, band 2 represented GRα-C, and band 5 represented GRα-D. In addition, we also identified two previously unknown isoforms, the 67 kDa isoform and 60 kDa isoform, which were detected with both batches of GR antibodies. Importantly, we found that GR isoforms presented disease and tissue-specific expression profiles either between the CRSwNP and control groups or between the E-CRSwNP and NE-CRSwNP subgroups.

Recently, several translational isoforms of GR have been reported to exist in different species, various tissues and several cell lines [11, 13, 34, 35]. According to the previous study of Nick, when the human GRα was expressed in COS-1 cells, which lack endogenous GR, multiple proteins of 94, 91, 82–84, and 53–56 kDa were detected by using different selective anti-peptide GR antibodies [13]. In a series of Australian studies on the expression of multiple GR isoforms in different species with a total GR antibody, multiple known GR isoforms (such as GRα-A, GRα-C, GRα-D) and some unknown GR isoforms were identified in the placenta of humans, guinea pigs, mice and a sheep model of maternal allergic asthma [16, 24, 33, 34]. However, some variations in the expression patterns of GR isoforms have still been observed among species and tissues. For example, GRα-B is not expressed from the GR gene in guinea pigs because a second start codon on exon 2 of the guinea pig GR gene encodes isoleucine instead of methionine [24]. In another study on GR isoforms of human dorsolateral prefrontal cortex (DLPFC) detected with the GRα-specific antibody, five bands were identified by WB analysis and further verified in Hela cells by transfecting with GRα-A, GRα-C 1and GRα-D1variants [14]. Sinclair et al. concluded that band1 corresponded to GRα-A, the 67 kDa band represented a newly uncharacterized GR isoform, the 50 kDa band represented the truncated GRα-D1 isoform in human DLPFC tissue [14]. In our study, the previously identified isoforms GRα-A, GRα-B, GRα-C and GRα-D were observed in nasal tissues. In consistent with Sinclair’s study, previously unknown 67 kDa isoform was also identified in our study.

Previous studies on GR protein isoforms in CRS have focused mainly on total GR, GRα and/or GRβ isoforms, and the results have been controversial and inconsistent [22, 25, 26]. Pujols and coworkers revealed that GR protein expression determined by immunohistochemistry was reduced in NPs but upregulated after short-term oral GC and intranasal budesonide treatment [22]. In contrast, Watanabe and Suzaki revealed markedly increased GRα expression in inflammatory cell infiltrates in the NPs of individuals with asthma and found that the expression of GRα was significantly reduced following GC treatment [26]. In our study, GR presented different protein expression profiles either between the CRSwNP and control groups or between different endotypes of CRSwNP. We speculate that the inconsistency of the above-mentioned results could be, to some extent, that different GR protein isoforms profiles may be implicated in different responses to GC between different endotypes of CRS [3].

With regard to different translational GR isoforms, the expression levels of the GRα-A/B and GRα-C isoforms did not significantly differ between the CRSwNP and control groups or between the E-CRSwNP and NE-CRSwNP subgroups. Surprisingly, relative to the other GR isoforms, GRα-A was weakly expressed, constituting less than 3% of the total expression (the sum of the intensities of bands 1–5), although it was expected to be the predominant isoform of GR. Consistent with our findings, Saif and coworkers also reported that GRα-A, which was detectable in 57–70% of all preterm placentae, was weakly expressed, constituting less than 2% of the total expression of all isoforms [33]. This could explain in part why significantly different expression levels of GRs were not observed between diseases and responses to GC when only GRα-A was considered.

In our study, the expression of GRα-D, which was stable and abundant in each individual, was much higher in the NP group than in the control group and was much higher in the NE-CRSwNP subgroup than in the E-CRSwNP subgroup. The differences in GR isoforms among the groups may be partially attributed to possible geographic differences in different CRSwNP endotypes [1]. However, Sinclair et al. also found that patients with schizophrenia and bipolar disorder have selective increases in GRα-D isoform expression in certain brain regions [14, 36]. Isoform-specific gene regulatory profiles have been found to produce functional differences in GC-induced apoptosis in osteosarcoma cells [12]. Cells expressing GRα-C3 exhibit the highest sensitivity to glucocorticoid-induced apoptosis, while GRα-D3-expressing cells are the most resistant [12]. In one study, when the individual isoforms were expressed at similar levels in U2OS osteosarcoma or Jurkat T lymphoblastic leukemia cells, they each regulated a unique set of genes [21]. Fewer than 10% of the genes were commonly regulated by all the subtypes, indicating that the vast majority of genes were selectively regulated by different GRα isoforms [21]. These findings suggest that GRα-D, which is considered a “resistant” isoform, deserves further functional investigation in the future.

In addition, we found that the intensities of the unknown 67 kDa and 60 kDa isoforms were both drastically higher in the noneosinophilic CRSwNP subgroup than in the E-CRSwNP subgroup. Similar unknown isoforms were also found in a series of Australian studies on the expression of multiple GR isoforms among different species detected with a total GR antibody [16, 24, 33, 34]. Clifton et al. reported that translational isoforms for GRβ, GRγ, GR-A and GR-P variants might account for the presence of these unknown bands. Given that some of these unknown isoforms are also altered in relation to fetal sex and gestational age, they have some physiological activity. Furthermore, when total GR gene translation is silenced, the protein expression of these unknown isoforms is reduced along with that of the known isoforms, supporting the hypothesis that all protein bands identified by WB analysis are GR proteins. However, confirmation is required, as these unknown bands could also be degraded forms of GR [33]. Similarly, the unknown 67 kDa isoform was also found to be predominant in the DLPFC by the detection of GRα antibody [14]. Among the other splice GR isoforms, only GR-A is truncated but retains exon 9a, which contains the epitope targeted by the anti-GRα antibody in this study [14]. According to the MWs of GR proteins, the 67 kDa isoform may be a GR-A isoform, or it may be an unknown GRα translational isoform or GR-associated protein. Given its abundance in NPs, the 67 kDa GRα-related isoform could play a critical role in inflammatory signaling in NPs. However, its identity and functional properties remain to be determined [33].

Our results should be interpreted with several limitations. This study was limited by its small sample size and retrospective method. Therefore, we could not obtain separate cytosolic and nuclear protein fractions. As GC signaling is mediated mainly by GRs, a major barrier to understanding the changes in the expression of multiple GR isoforms that occur in relation to GCs is the development of glucocorticoid sensitivity or resistance in certain NP groups; thus, hormone stimulation research should be performed on NPs with different sensitivity statuses. In addition, the unknown GR isoforms and its functional properties of GR isoforms and the relationships among them remain to be further investigated.