Aberrant localization of FOXJ1 correlates with the disease severity and comorbidities in patients with nasal polyps

Study protocol approval was obtained from the institutional review boards of Guangdong General Hospital and the Second Hospital of Shandong University in China, and the National Healthcare Group Domain Specific Review Board of Singapore. All subjects provided informed consent.

Control subjects and patients with NPs were recruited from individual sites. CRSwNP diagnoses were made according to the European Position Paper on Rhinosinusitis and Nasal Polyps 2012 [3], and AR was diagnosed according to Allergic Rhinitis and its Impact on Asthma (ARIA) [12] based on the clinical symptoms combined with skin prick test (Dermatophagoides farinae, Dermatophagoides pteronyssinus, pollens of oak, ragweed, mugwort, Humulus japonicus, cat and dog dander allergens) (Allergopharma, Reinbek, Germany) or serum total/specific IgE detected by using AllergyScreen^® (Mediwiss Analytic GmbH, Moers, Germany). The grades (G) of allergen-specific IgE were defined as follows: G0 (< 0.35 IU/ml), G1 (0.35–0.70 IU/ml), G2 (0.71–3.5 IU/ml), G3 (5.6–17.5 IU/ml), G4 (17.6–50 IU/ml), G5 (51–100 IU/ml) and G6 (> 100 IU/ml). The diagnosis of asthma was based on the history inquiry and physician’s diagnosis. Co-existence with asthma was deemed a more severe form of NPs [13]. The sinus opacity was scored radiologically and the severity and the size of NPs were assessed with the Lund–Mackay (LM) score [14]. Patients with antrochoanal polyps, fungal sinusitis, or susceptible PCD (an autosomal recessive congenital disease in which the structure and function of cilia are affected) were excluded. Control subjects were recruited from patients undergoing septoplasty due to anatomic variations and those who did not have allergic or inflammatory airway diseases, and biopsies were taken from the inferior turbinated mucosa (n = 40) during septal plastic surgery. Of these 40 control subjects, 20 were recruited for immunologic study and 34 for gene expression study (including 14 subjects for both assays). NP biopsies (n = 111) were performed during functional endoscopic sinus surgery. Among them, 83 subjects were recruited for immunologic study and 77 for gene expression study (including 39 subjects for both assays). Primary single-cell suspensions (1–2 × 10⁵ cells) were dissociated from control subjects (n = 4) and NPs (n = 6) for cytospin preparations. Details are shown in Table 1.

Primary single-cell suspensions were harvested from nasal brushing (HydraFlock Flocked Swab, Puritan, Guilford, ME) or fresh nasal biopsy specimens. The specimens were initially treated with 10 mg/ml of Dispase II (Sigma-Aldrich, St. Louis, MO) at 4 °C overnight, followed by incubation with 1× trypspin/ethylenediaminetetraacetic acid at 37 °C for 15 min to acquire dissolved cells. Brushed or dissolved cells were subsequently fixed with 4% formaldehyde at room temperature for 10 min, followed by washing with 1× Dulbecco’s Phosphate Buffered Saline, and centrifugation at 1200 rpm for 5 min. Cytospin (1–2 × 10⁴ cells/slide) preparations were made at 500 rpm for 5 min with cytospin techniques (Shandon Cytospin 3 Cytocentrifuge, Thermo Fisher Scientific; Thermo Fisher Scientific, Waltham, MA).

We performed hematoxylin–eosin staining, immunohistochemistry staining and evaluation of the epithelial structure as described previously [15]. Eosinophils were enumerated with hematoxylin–eosin staining, whereas neutrophils were stained with murine anti-human neutrophil elastase monoclonal antibodies (clone NP57; Dako A/S, Glostrup, Denmark) for immunohistochemistry staining. Epithelium with more than four layers was defined as having epithelial hyperplasia. The presence of stratified squamous epithelium was defined as squamous metaplasia. We evaluated the inflammatory cell types based on the proportions of eosinophils and neutrophils at 400× magnifications (high-power field, HPF), as described previously [15]. Briefly, the proportion of each type of cells in the 3 HPFs with respect to the total number of inflammatory cells in each HPF was calculated using the following formula: The proportion of the type of cells = (n1 + n2 + n3)/(m1 + m2 + m3) × 100%, where n1, n2, and n3 denoted the numbers of the type of inflammatory cells investigated in 3 HPFs, and m1, m2, and m3 denoted the total number of inflammatory cells in the 3 identical HPFs. All samples were coded confidentially and evaluated by two independent examiners by following the same protocol. NP with the percentage of eosinophils or neutrophils exceeding 10% was categorized as being eosinophilic or neutrophilic, respectively.

Protein expression of FOXJ1 and acetylated alpha-tubulin was examined by using IF staining on paraffin sections and primary single-cell cytospin sections of nasal biopsies. The sections were incubated with primary polyclonal antibodies [rabbit anti-human FOXJ1 [HPA005714] (Sigma, Ronkonkoma, NY) and mouse anti-human acetylated alpha tubulin [clone ab24610] (Abcam, Cambridge, MA)] overnight at 4 °C, followed by incubation with Alexa Fluor 488- or Alexa Fluor 594-conjugated secondary antibodies (Life Technologies, Carlsbad, CA) at 37 °C for 1 h.

Cellular nuclei were visualized by staining with 4′ 6-diamidino-2-phenylindole (DAPI) (Life Technologies, Carlsbad, CA). For negative controls, primary antibodies were substituted with species- and subtype-matched antibodies of the same concentration. Images were acquired with fluorescence microscopy (Olympus IX51, Tokyo, Japan).

To elucidate the different localization patterns, FOXJ1 (red) and acetylated alpha tubulin (green) were evaluated according to a semi-quantitative scoring system. The nuclei were stained with DAPI (blue). By following our recently published study [11], the localization of FOXJ1 expression was defined as follows: Presence of FOXJ1 in the nuclei only (Normal; N); Presence of FOXJ1 in both the nuclei and axoneme (Intermediate; I); Presence of FOXJ1 within the axoneme only (Mislocalization; M); Absence of FOXJ1 in the nuclei and axoneme (Absence; A).

We randomly selected five areas of cilia staining from each paraffin sample in a blinded manner to elucidate the localization of FOXJ1 at 400× magnification. Based on the criterion proposed by Shoemark et al. [16] and our recently published study [11], each area was assigned a score between 0 and 2, with 0 corresponding to the field containing > 70% N-localization; 1 for > 70% I + N-localization; and 2 for ≥ 30% with M- or A-localization. Each field was scored by two reviewers blinded to the grouping. We then averaged the score of five areas for subsequent reporting. We randomly selected single ciliated cells from primary specimens for cytospin preparations, and the localization pattern of individual cell was evaluated at 1000× magnification under oil immersion lens.

The mRNA expression levels of FOXJ1 were validated with quantitative real-time polymerase chain reaction (PCR) (SYBR Green, Promega, Madison, WI, USA) in accordance with the “The Minimum Information for Publication of Quantitative Real-Time PCR Experiments (MIQE) guidelines” (Additional file 1: Table S2) [17]. The relative gene expression was calculated using the 2^−ΔΔCt method normalized against the housekeeping gene, glyceraldehyde 3-phosphate dehydrogenase (GAPDH). Selection of the appropriate reference gene depends on the individual experimental setting. Despite the use of a single reference gene, we have confirmed the suitability of GAPDH as the reference gene in our previous studies [18, 19]. Amplification of FOXJ1 and GAPDH was performed with the following primers: FOXJ1 forward (5′-GTGAAGCCTCCCTACTC-3′), FOXJ1 reverse (5′-AATTCTGCCAGGTGGG-3′) (PrimerPair ID: H_FOXJ1_1, Gene ID: 2302; Sigma-Aldrich, St. Louis, MO); GAPDH forward (5′-ACAGTTGCCATGTAGACC-3′), GAPDH reverse (5′-TTT TTGGTTGAGCACAGG-3′) (PrimerPair ID: H_GAPDH_1, Gene ID: 2597; Sigma-Aldrich, St. Louis, MO).

Statistical analyses were conducted with SPSS software (version 18.0, SPSS, Chicago, IL, USA). Mann–Whitney two-sided non-parametric test was applied to compare FOXJ1 localization score and FOXJ1 mRNA expression levels in paraffin-embedded specimens. The Fisher’s exact test was employed to assess for the differences in FOXJ1 localization patterns between control subjects and patients with NPs for cytospin specimens. P < 0.05 was deemed statistically significant for all analyses.

The clinical characteristics of control subjects and patients with NPs are summarized Table 1. For paraffin specimens, all CRSwNP patients with AR (37.3%, 31/83) were diagnosed with AllergyScreen^®, and were graded as having G1 (n = 3), G2 (n = 11), G3 (n = 4) and G4 (n = 13) AR (Additional file 1: Table S1). The hematoxylin–eosin staining showed that 90.4% (75/83) of specimens from NPs displayed varying degrees of epithelial remodeling.

We initially examined the localization patterns of FOXJ1 in nasal epithelial cilia by performing IF staining of samples from control subjects and patients with NPs. As shown in Fig. 1, four distinct localization patterns (N, I, M and A, Fig. 1a–c) were observed in isolated primary single cells. Based on our semi-quantitative FOXJ1 scoring system (Fig. 1d–g), we scored five areas and obtained the average score of each paraffin sections (Fig. 2a–e).

The median (the 1st and 3rd quartile) score was 1.00 (0.40, 1.80) for samples from NPs (n = 83), and 0.10 (0, 0.40) for samples from control subjects (n = 20) (P < 0.001, Fig. 3a). We also evaluated 10 single cilia cells from each primary samples from cytospin preparations. The N, I, M and A pattern was observed in 15.0% (9/60), 3.3% (2/60), 53.3% (32/60) and 28.3% (17/60) of ciliated cells in patients with NPs (n = 6), respectively. The corresponding percentage was 82.5% (33/40), 5.0% (2/40), 5.0% (2/40), and 7.5% (3/40) in control subjects (n = 4), respectively (P < 0.001, Table 2).

We interrogated if there were major differences between FOXJ1 localization score and the co-existence of AR or asthma in patients with NPs. Note that patients with co-existing AR and asthma have been excluded from our study. The median (1st, 3rd quartile) localization score was 0.55 (0.40, 1.15) for NPs with neither AR nor asthma (n = 34), 1.40 (0.80, 2.00) for NPs with isolated AR (n = 31), and 1.50 (0.80, 1.80) for NPs with isolated asthma (n = 18). The FOXJ1 localization score was significantly increased in both NPs with isolated AR, and NPs with isolated asthma (both P < 0.05, Fig. 3b).

Next, we examined the difference and association of the FOXJ1 localization score and the grading (G) of allergen-specific IgE in patients with NPs and AR. Compared with NPs without AR or asthma, the FOXJ1 localization score was significantly increased in patients with NPs who had more greater degree of allergy (G2-G4) (P < 0.05, Fig. 3c). However, the FOXJ1 localization score did not differ significantly between NPs and those with minor allergy (G1) (P > 0.05, Fig. 3c).

In our study, the percentage of eosinophils or neutrophils exceeding 10% was categorized as eosinophilia or neutrophilia [15, 20]. Of these, 56.6% (47/83) of NPs were eosinophilic and 47.0% (39/83) were neutrophilic, whereas 14 specimens have mixed eosinophilic and neutrophilic inflammatory phenotypes. We sought to determine if there were underlying associations between FOXJ1 localization score and the presence of eosinophilic or neutrophilic airway inflammation. The median (1st, 3rd quartile) FOXJ1 localization score was 1.00 (0.45, 1.60) for eosinophilic NPs (n = 47) and 1.00 (0.40, 2.00) for non-eosinophilic NPs (n = 36) (P > 0.05, Fig. 3d). When stratified according to the presence of neutrophilia only, we also noted no marked difference in the median (1st and 3rd quartile) FOXJ1 localization score between neutrophilic (n = 39) and non-neutrophilic NPs (n = 44) [1.00 (0.40, 1.90) vs. 1.20 (0.60, 1.80), P > 0.05, Fig. 3e).

The Lund–Mackay scores were available in 54 patients with NPs. The FOXJ1 localization score positively correlated with the Lund–Mackay CT score (r = 0.362, P = 0.007, Fig. 3f). The finding confirms that FOXJ1 localization score is useful for predicting NP severity and presence of comorbidities, but does not distinguish between the inflammatory phenotypes.

We compared FOXJ1 gene expression with quantitative real-time PCR from controls (n = 34) and patients with NPs (n = 77). We initially evaluated the FOXJ1 mRNA expression levels in patients with NPs (n = 77) compared with controls. However, FOXJ1 mRNA expression levels did not differ statistically (P = 0.634, Fig. 4a). However, the FOXJ1 mRNA expression levels were markedly higher in NPs without AR or asthma (n = 32) than in control subjects (P = 0.035, Fig. 4b). Furthermore, the FOXJ1 mRNA expression levels were significantly down-regulated in NPs with isolated AR (n = 25), and NPs with isolated asthma (n = 20) (both P < 0.05, Fig. 4c).