Between May 2019 and July 2021, this study serially included 132 active patients with AS. The inclusion criteria were as follows: (i) confirmed as AS according to Assessment of Spondylarthritis International Society (ASAS) criteria (20); (ii) age ≥18 years; (iii) Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) score ≥4 [on a 0–10 cm Visual Analog Scale (VAS)]; (iv) Ankylosing Spondylitis Disease Activity Score (ASDAS) ≥2.1; (v) were about to receive TNFi treatment or NSAID treatment; and (vi) willing to comply with the study protocol for data collection, sample collection, and follow-up. The exclusion criteria were as follows: (i) concomitant with tuberculosis or hepatitis; (ii) had severe joint deformity; (iii) had a chronic or recurrent infection; (iv) complicated with cancer or hematologic malignancy; and (v) pregnant and lactating female patient.

In addition, during the same period, 30 patients with osteoarthritis (OA) were enrolled as disease controls (DCs) in the study. All patients with OA, aged over 18 years, were diagnosed by imageological examinations. Additionally, 30 healthy subjects were also recruited in the study as healthy controls (HCs). The DCs and HCs who had an infection, tuberculosis, hepatitis, severe joint deformity, cancer, or hematologic malignancy were excluded, and the subjects who were pregnant or lactating were also ineligible.

The written informed consents were collected from all subjects. The study was permitted by Ethics Committee.

For patients with AS, clinical characteristics were obtained after admission. Peripheral blood (PB) samples were collected from all 132 patients with AS at W0 (before treatment initiation), from 127 patients with AS at W4 (4 weeks after treatment initiation), from 116 patients with AS at W8 (8 weeks after treatment initiation), and from 104 patients with AS at W12 (12 weeks after treatment initiation). Then, peripheral blood mononuclear cell (PBMC) and serum samples were isolated. For DCs and HCs, PB samples were collected after recruitment, and then, PBMC samples were separated.

The collected PBMC samples of all subjects were used to determine the expression of HDAC4 by reverse transcription-quantitative polymerase chain reaction (RT-qPCR). A total of 69 fresh PB samples of patients with AS were available to detect T-helper 1 (Th1) cell percentage and T-helper 17 (Th17) cell percentage by flow cytometry. Serum samples of all AS patients were used to assess the level of proinflammatory factors [e.g., tumor necrosis factor (TNF), interleukin (IL)-1β, and IL-6], the cytokine secreted by Th1 cells [interferon-gamma (IFN-γ)], and the cytokine secreted by Th17 cells (IL-17A) by enzyme-linked immunosorbent assay (ELISA).

A total of 54 patients with AS were treated with TNFi, including adalimumab and etanercept. The detailed regimens were as follows: adalimumab was administered subcutaneously at a dose of 40 mg every 2 weeks; etanercept was administered subcutaneously at a dose of 50 mg every week. Besides, a total of 78 patients with AS were treated with NSAID, including celecoxib, meloxicam, and imrecoxib. The detailed regimens were as follows: celecoxib was administered orally at a dose of 200 mg/day; meloxicam was administered orally at a dose of 7.5 mg/day; imrecoxib was administered orally at a dose of 100 mg two times a day.

At W4, W8, and W12, treatment response was assessed according to the ASAS40 response criteria (20). During the follow-up, a total of 28 patients dropped out early from the study. Of these, 22 cases were lost to follow-up, three cases had poor efficacy, one case had an abnormal liver function, one case had an active infection, and one case was no longer in the study. For the assessment of the ASAS40 response, the missing data were processed using the last observation carried forward (LOCF) method. In the analysis, patients who achieved ASAS40 response at W12 were categorized as response patients, and the others without the achievement of ASAS40 response at W12 were categorized as non-response patients.

Human naïve CD4⁺ T cells were isolated from the PBMC of patients with AS using a naïve CD4⁺ T cell isolation kit (Miltenyi, Germany). Cells were maintained in GT-T551 T-cell culture and expansion medium (Takara, Japan) and transfected with HDAC4 overexpression lentivirus (HDAC4 group), HDAC4 knockdown lentivirus (anti-HDAC4 group), or control lentivirus (Scramble group) (Genepharma, China) according to the manual. Un-transfected naïve CD4⁺ T cells were used as the control (control group). After 48 h incubation, cells were harvested for detecting the expression of HDAC4 by RT-qPCR and Western blot.

The polarization of Th1, T-helper 2 (Th2), and Th17 cells was performed as described previously (21, 22). Briefly, transfected naïve CD4⁺ T cells were stimulated with the following polarizing conditions for 3 days: IL-12 (10 ng/ml, Sigma, USA) and anti–IL-4 (5 μg/ml, Affinity, USA) for Th1 polarization; IL-4 (2.5 ng/ml, Sigma, USA) and anti-IFN-γ (5 μg/ml, Affinity, USA) for Th2 polarization; TGF-β (10 ng/ml, Sigma, USA), IL-6 (10 ng/ml, Sigma, USA), IL-1β (10 ng/ml, Sigma, USA), IL-23 (10 ng/ml, Sigma, USA), anti–IL-4 (5 μg/ml), and anti-IFN-γ (5 μg/ml) for Th17 polarization. After polarization, the cells and supernatants were collected for flow cytometry and ELISA, respectively.

Total RNA of PBMC samples or transfected cells was purified using TRIzol (Invitrogen, USA) and quantified using NanoDrop (Thermo, USA). The cDNA Synthesis Kit (Bio-Rad, USA) and PCR Master Mix (Beyotime, China) were used for RT-qPCR according to the manufacturer's instructions. The expression of HADC4 was normalized to GAPDH using the 2^−ΔΔCt method. The primers used were as follows: HDAC4, 5'-AGAATGGCTTTGCTGTGGTC-3' (forward) and 5'-ATCTTGCTCACGCTCAACCT-3' (reverse); GAPDH, 5'-GAGTCCACTGGCGTCTTCAC-3' (forward) and 5'-ATCTTGAGGCTGTTGTCATACTTCT-3' (reverse).

Total proteins were extracted using RIPA (Beyotime, China) and quantified with a protein quantification kit (Beyotime, China). The SDS-PAGE was performed for protein separation using 10% precast gels (Thermo, USA). The polyvinylidene difluoride membranes (Millipore, USA) were applied for protein transfer. After being blocked with 5% bovine serum albumin (BSA) buffer, the membrane was incubated with rabbit anti-HDAC4 antibody (1:2,000, Affinity, USA) or rabbit anti-GAPDH antibody (1:10,000, Affinity, USA). Bands were revealed using an ECL kit (Servicebio, China) after secondary antibody incubation (1:10,000, Affinity, USA). Quantification of HDAC4 was normalized to GAPDH by densitometry using ImageJ software (NIH, USA).

Th1 cell percentage and Th17 cell percentage among patients with AS were measured by flow cytometry using Human Th1/Th17 Phenotyping Kit (Thermo, USA), whereas in in vitro experiment, the percentages of Th1, Th2, and Th17 cells were determined using Human Th1/Th2/Th17 Phenotyping Kit (BD, USA). Briefly, PBMC samples or polarized cells were stimulated with ionomycin (5 μM, Sigma, USA), phorbol myristic acetate (PMA) (1 μM, Beyotime, China), and BD GolgiStop (1 μl/ml, BD, USA) for 6 h before collection. For PBMC samples, cells were then stained with FITC-conjugated anti-CD4, PE-conjugated anti-IFN-γ, and APC-conjugated anti-IL-17A. For polarized cells, cells were then stained with PerCP-Cy5.5-conjugated anti-CD4 and FITC-conjugated anti-IFN-γ for the assessment of Th1 cells; PerCP-Cy5.5-conjugated anti-CD4 and APC-conjugated anti-IL-4 for the assessment of Th2 cells; and PerCP-Cy5.5-conjugated anti-CD4 and APC-conjugated anti-IL-17A for the assessment of Th17 cells. After being stained for 0.5 h, cells were analyzed using a FACSCanto II (BD, USA). The data were analyzed using FlowJo7.6 software (BD, USA). Briefly, CD4⁺ cells were first gated, then IFN-γ⁺, IL-4⁺, and IL-17A⁺ cells were gated for calculating the percentages of Th1, Th2, and Th17 cells in CD4⁺ cells, respectively.

The levels of TNF, IL-1β, IL-6, IFN-γ, and IL-17A among patients with AS were determined by ELISA using a commercial Human ELISA Kit (Sangon Biotech, China). For in vitro experiment, ELISA detection was performed using Th1/Th2/Th17 cytokine Human ProcartaPlex™ Panel (Thermo, USA). The experimentations were performed according to the manuals provided by the manufacturers.

SPSS V.21.0 (IBM Corp., USA) was employed for data analysis. GraphPad Prism V.7.0 (GraphPad Software Inc., USA) was implemented for graph plotting. The difference of HDAC4 expression among the three groups was analyzed using the Kruskal-Wallis H rank sum test, followed by post-hoc comparisons using the Bonferroni test. The differentiation performance of HDAC4 expression was estimated using the receiver operating characteristic (ROC) curve. The correlation of HDAC4 expression with clinical characteristics was determined using Spearman's rank correlation test or Wilcoxon rank sum test. Change of HDAC4 expression over time was determined using the Friedman test. Comparison analysis between the two groups was executed using the chi-square test or Wilcoxon rank sum test. In the study of in vitro experiment, statistical significance among groups was evaluated using one-way ANOVA with Tukey's post-hoc test. P-value <0.05 was considered statistically significant.

The mean age of the enrolled patients with AS was 37.1 ± 9.7 years with 14 (10.6%) female patients and 118 (89.4%) male patients. Meanwhile, there were 115 (87.1%) patients with positive HLA-B27. Regarding the biochemical indexes, the median [interquartile range (IQR)] C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR) levels were 23.6 (16.4–38.7) mg/L and 28.0 (20.1–41.4) mm/h, respectively. As for disease activity scores, the median (IQR) BASDAI score was 5.3 (4.6–6.3), and the median ASDAS was 3.7 (2.9–4.2). Besides, the median (IQR) levels of TNF, IL-1β, and IL-6 were 75.5 (60.9–99.3) pg/ml, 3.1 (2.1–3.8) pg/ml, and 30.2 (24.4–39.4) pg/ml, respectively. Moreover, the median (IQR) proportion of Th1 and Th17 cells was 17.2 (14.5–22.2)% and 6.1 (4.8–7.6)%, respectively. Regarding treatment information, 120 (90.9%) patients had a history of NDAID treatment and 34 (25.8%) patients had a history of TNFi treatment; currently, 78 (59.1%) patients initiated NSAID treatment and 54 (40.9%) patients started TNFi treatment. More detailed clinical characteristics of patients with AS are shown in Table 1.

Histone deacetylase 4 was decreased in patients with AS [median (IQR) level: 0.502 (0.313–0.758)] compared with DCs [median (IQR) level: 0.796 (0.546–1.262)] (P = 0.001) and HCs [median (IQR) level: 1.010 (0.800–1.524)] (P < 0.001) (Figure 1A). Subsequent ROC curve analysis showed that HDAC4 presented good ability in identifying AS patients from HCs [area under the curve (AUC): 0.820, 95% confidence interval (CI): 0.736–0.904]; meanwhile, HDAC4 also showed acceptable value in identifying patients with AS from DCs (AUC: 0.721, 95% CI: 0.622–0.821) (Figure 1B).

Regarding the correlation between HDAC4 and proinflammatory cytokines in AS patients, it was found that HDAC4 showed a strong, negative correlation with TNF (r_s = −0.370, P < 0.001) (Figure 2A) and a weak, negative correlation with IL-1β (r_s = −0.260, P = 0.003) (Figure 2B). However, HDAC4 did not correlate with IL-6 in patients with AS (r_s = 17 polarization 0.134, P = 0.125) (Figure 2C).

As to the correlation of HDAC4 with Th1 cells, Th17 cells, and their secreted cytokines in patients with AS, HDAC4 showed no correlation with Th1 cells (r_s = −0.190, P = 0.118) (Figure 3A) and their secreted cytokine IFN-γ (r_s = −0.138, P = 0.115) (Figure 3B), whereas HDAC4 was negatively correlated with Th17 cells (r_s = 0.316, P = 0.008) (Figure 3C) and their secreted cytokine IL-17A (r_s = −0.276, P = 0.001) (Figure 3D) in patients with AS.

In terms of the association of HDAC4 with categorized characteristics of AS patients, no association was found in HDAC4 with gender, HLA-B27, and history of NSAID or TNFi (all P > 0.05) (Table 2). Concerning the correlation of HDAC4 with continuous characteristics of patients with AS, HDAC4 was negatively correlated with CRP (r_s = −0.349, P < 0.001) and ASDAS (r_s = −0.181, P = 0.038), but not with other characteristics including ESR, disease duration, BASDAI score, BASFI score, total back pain score, or patient global assessment of disease activity score (all P > 0.05) (Table 2).

Histone deacetylase 4 was different among W0, W4, W8, and W12, which presented an increasing trend from W0 to W12 in patients with (P < 0.001) (Figure 4A). Meanwhile, comparison analysis was conducted between patients with AS receiving TNFi or NSAID treatment, which presented that HDAC4 at W12 was increased in patients with AS receiving TNFi treatment compared with those receiving NSAID treatment (P = 0.049); however, HDAC4 at W0, W4, and W8 remained similar between these two types of patients (all P > 0.05) (Figure 4B).

The number of patients who achieved ASAS40 response at W0, W4, W8, and W12 were 0 (0.0%), 16 (29.6%), 27 (50.0%), and 31 (57.4%), respectively, in TNFi-treated patients and 0 (0.0%), 18 (23.1%), 24 (30.8%), and 30 (38.5%) in NSAID-treated patients. Further comparison analysis revealed that the proportion of patients who achieved ASAS40 response was higher at W8 (P = 0.026) and W12 (P = 0.032) but remained similar at W4 (P = 0.397) in patients receiving TNFi treatment compared with those receiving NSAID treatment (Figure 5A).

Subsequently, the relationship between HDAC4 and treatment response in patients with AS was analyzed. Data revealed that among all patients with AS, HDAC4 at W8 (P = 0.014) and W12 (P = 0.006) was higher in response patients compared with non-response patients (Figure 5B). By subgroup analysis, after NSAID treatment, HDAC4 at all time points remained similar between response and non-response patients (all P > 0.05) (Figure 5C); meanwhile, after TNFi treatment, HDAC4 at W12 was elevated in response patients compared with non-response patients (P = 0.016) (Figure 5D).

Th cell polarization is one of the critical regulators of AS; meanwhile, the data mentioned earlier presented that HDAC4 was correlated with Th17 cells in patients with AS. Therefore, whether HDAC4 could affect Th cell polarization in AS was further explored. Initially, naïve CD4⁺ T cells from patients with AS were isolated and transfected with HDAC4 overexpression or knockdown lentivirus. Then, RT-qPCR and Western blot analysis confirmed high transfection efficiency (Figures 6A–C). Subsequently, the transfected naïve CD4⁺ T cells were, respectively, polarized into Th1, Th2, and Th17 cells. The flow cytometry analysis revealed that HDAC4 overexpression reduced Th1 (P = 0.039) and Th17 (P = 0.008) cell percentage, but not Th2 cell percentage (P = 0.083); whereas HDAC4 knockdown only promoted Th17 cell percentage (P = 0.006) (Figures 6D,E). Meanwhile, the ELISA analysis revealed that HDAC4 overexpression decreased, while its knockdown increased IL-17A level (both P < 0.05); however, they could not affect IFN-γ or IL-4 level (all P > 0.05) (Figure 6F). These data suggested that HDAC4 negatively regulated Th17 polarization but has a weak effect on Th1 or Th2 polarization in AS.