MG-HS, a virulent strain, is reported in detail in our previous studies (30). The concentration of MG-HS in this study was 10¹⁰CCU/mL. DF-1 cells and CP-II cells were cultured in a 6-wells plate and treated MG. After 24h infection, we isolated the total RNA or protein from infected cells. In addition, White Leghorn specific-pathogen-free (SPF) chicken embryos were used for MG-HS infection experiments. The experimental method was detailed in our previous reports (31). In short, the chick embryo allantoic cavity was injected with 300 ul of MG on 9-day of incubation. Then, samples of chicken embryonic lung tissue were collected at 6, 7, 8, and 9 days post-infection (15, 16, 17 and 18 days after hatching).

The wild-type and mutant 3’UTR DNA fragments of JNK1 covering the predicted binding sites of miR-33-5p were successfully replicated. The psiCHECK™-2-JNK1-3’UTR (wild type and mutant) vector was constructed by combining the luciferase vector psi-CHECK™-2 (Promega, Madison, WI, USA) with JNK1-3’UTR (wild type and mutant). Similarly, we cloned the CDS of JNK1 or Lnc90386 into the pcDNA3.1 vector. The primers are described in Table 1 .

Dual-luciferase reporter assay is detailed in our previous reports (16). In a nutshell, when cells reached 80-90% confluence, using Lipofectamine 2000 (Invitrogen Life Technologies, USA) each co-transfected cells with wild-type or mutant reporter plasmid (200 ng) and 10 pmol of the indicated RNA oligonucleotides. Then, the luciferase activity in each group was detected by using an automatic microplate reader (Bio-Rad, Hercules, CA, USA) in accordance with the dual luciferase reporter gene detection kit instructions (Promega, Madison, WI, USA) according to the manufacturer’s protocol.

In this experiment, total sequences of DNA primers were adopted, which are presented in Table 1 . In addition, miR-33-5p mimics (denoted as miR-33-5p) are double-stranded RNAs synthesized to simulate naturally occurring mature miR-33-5p, whereas the inhibitor (denoted as miR-33-5p-Inh) are chemically modified antisense single-stranded RNAs that silence the endogenous miRNAs by sequence complementarity. A random miRNA mimics that had not been found to suppress any chicken target genes (denoted as miR-33-5p-NC), and a random miRNA inhibitor that had not been found to promote any chicken target genes (denoted as miR-33-5p-Inh-NC) were also designed and synthesized to serve as the negative controls. All RNA oligonucleotides were designed and synthesized by GenePharm (Shanghai, China). The RNA oligonucleotides sequences are shown in Table 2 .

Once 80-90% confluence was achieved, each group of cells was transfected with mimics, inhibitor, plasmids and siRNA respectively. At 48h after transfection, TRNzol (TIANGEN, Beijing, China) was used to harvest total cells.

According to the manufacturer’s instructions, total RNA was isolated from post-infected and non-infected cells via TRNzol Universal Reagent kit (TIANGEN, Beijing, China). Then, RNA was inverse transcribed to cDNA with the first strand cDNA synthesis kit (Cat No.11119-11141; Yeasen, Shanghai, China) and reverse transcription PCR (RT-PCR).

The Cell Counting Kit-8 (CCK-8, DOJINDO, Shanghai, China) was used for cell proliferation experiments. Cells were inoculated on a 96-well plate at 2x10⁴ cells per well. Each group of cells was separately transfected with different oligonucleotides or plasmids using Lipofectamine™3000 and each group had 6 biological replicates. Next, MG-HS (7µL, 10¹⁰ CCU/mL) was utilized to infect DF-1 cells for 2 hours. At 12h, 24h, and 36h post-transfection, the cell proliferation curve was measured by the CCK-8 kit according to the manufacturer’s instructions. Transfection treatments were described above. Annexin V, FITC apoptosis detection kit (DOJINDO) was used to test the cell apoptosis. Each group was repeated three times.

The grouping of transfection treatments was described above. Forty-eight hours after transfection, the supernatants were collected and the pro-inflammatory cytokines (IL-1β and TNF-α) levels were detected with enzyme-linked immunosorbent assay kits (Bio Legend, San Diego, CA) according to the manufacturer’s directions.

Forty-eight hours after transfection, DF-1 cells were lysed in RIPA buffer (Beyotime, Nantong, China) supplemented with 100 mM phenyl methane sulfonyl fluoride (PMSF) to exact total protein. Protein concentrations were measured by the Pierce BCA Protein Assay Kit (Transgen, Shanghai, China). An equal amount of protein was separated by 12% SDS-polyacrylamide gel electrophoresis (Beyotime, China) and blocked with 5% skim milk for 1h. Then, the membranes were separately probed with p-JUN (ABclonal, AP0048), p-FOS (ABclonal, AP0038), p-JNK1 (ABclonal, AP0631), Bcl-2 (ABclonal, A19693), Caspase8 (ABclonal, A0215), Caspase9 (ABclonal, A18676), Caspase3(ABclonal, A19654), GAPDH (Abmart, M20024) overnight at 4°C with a final dilution of 1:5000 (v/v). Finally, the membrane was incubated with the secondary antibody for 1h after TBST washing. The enhanced chemiluminescence (ECL) detection system (Bio-Rad) was used to detect protein expression.

Data were presented as the mean ± SD. Student’s t-test was used to determine significant differences between groups. A value of p < 0.05 was considered statistically significant and p < 0.01 considered extremely significant. (*p < 0.05, **p < 0.01).

Previous studies found that miR-33-5p was significantly up-regulated in MG infection by deep sequencing (21). The qPCR result showed that miR-33-5p was remarkably up-regulated both in DF-1 cells and CP-II cells with MG infection compared with the control group ( Figure 1A ). In addition, we observed that miR-33-5p expression was significantly higher in the lungs of the MG-infected chicken embryos than in the non-infected group on the 7th–9th days of post-infection (equivalent to the 16th–18th days of egg hatching) ( Figure 1B ).

pMGA1.2, the major adhesin protein of MG-HS, is required for MG-HS infection (12). To further evaluate the effect of miR-33-5p on the MG infection in DF-1 cells, miR-33-5p, miR-33-5p-NC, miR-33-5p-Inh or miR-33-5p-Inh-NC were transfected into DF-1 cells. Then, the DF-1 cells were treated with 200 µL of MG at 10¹⁰ CCU/mL for 24h. qPCR and Western blot were used to respectively determine the expression levels of miR-33-5p and pMGA1.2. As expected, miR-33-5p mimics had further up-regulated the expression of miR-33-5p induced by MG infection, whereas miR-33-5p inhibitor diminished its expression ( Figure 2A ). Furthermore, we found that overexpression of miR-33-5p led to a down-regulation of pMGA1.2 protein expression in DF-1 cells, while the opposite result was obtained when miR-33-5p was suppressed ( Figure 2B ).

Next, we further explored the biological functions of miR-33-5p on MG infection. qPCR and ELISA results showed that the expressions of pro-inflammatory cytokines IL-1β and TNF-α were significantly down-regulated in the miR-33-5p mimics group compared with the miR-33-5p NC group ( Figure 3A ). Conversely, when the expression of endogenous miR-33-5p was inhibited, the release of pro-inflammatory cytokines was significantly increased ( Figure 3B ). In the MG-infected group, transfection with miR-33-5p mimic significantly down-regulated the expressions of MG-induced IL-1β and TNF-α, but transfection with miR-33-5p inhibitor showed the opposite results ( Figures 3A, B ).

In addition, we examined the effect of miR-33-5p on cell proliferation by using the CCK-8 kit. Data showed that MG infection significantly inhibited DF-1 cell proliferation. Overexpression of miR-33-5p alleviated the inhibition of MG-induced cell proliferation, whereas suppression of miR-33-5p exacerbated the loss of MG-induced cell proliferation ( Figure 3C ). Furthermore, cell apoptosis assay showed that upon MG infection, overexpression of miR-33-5p resulted in a reduction in apoptosis, while inhibition of miR-33-5p led to an opposite result ( Figure 3D ). To further explore how miR-33-5p regulates MG-induced cell apoptosis, the expression of apoptosis-related genes (including pro-apoptotic and anti-apoptotic genes) were examined by qPCR and Western blot. We found that pro-apoptotic genes including FAS, Casp3, Casp8 and Casp9 were significantly increased, while the anti-apoptotic gene BCL-2 was significantly suppressed after MG infection. Overexpression of miR-33-5p inhibited the expression of MG-induced FAS, Casp3, Casp8 and Casp9, but up-regulated the BCL-2 expression ( Figure 3E ). However, in the miR-33-5p inhibitor group, the opposite trend was observed ( Figure 3F ).

It is well known that miRNAs mediate complex biological processes by partially binding to the 3’UTR of target genes (32). In order to explore the function of miR-33-5p in MG infection, potential target genes were predicted by TargetScan (http://www.targetscan.org) and miRDB (http://www.mirdb.org/miRDB/index.html). Analysis of the KEGG database identified that the target genes of miR-33-5p were mainly enriched in the MAPK pathway ( Figure 4A ). The prediction result of TargetScan showed that the target site’s sequence in the 3’UTR of JNK1 is highly conserved across species ( Figure 4B ). The minimum free energy (MFE) of the RNA duplex was -19.2 kcal/mol ( Figure 4C ) by RNAhybrid, which suggested a powerful combination between miR-33-5p and JNK1. Meanwhile, we also found that JNK1 expression was down-regulated in both MG-infected cells and chicken embryo lungs, which was exactly opposite to the expression pattern of miR-33-5p ( Figures 4D, E ). To further confirm the targeting relationship between miR-33-5p and JNK1, the dual-luciferase reporter vectors containing either the wild-type or the mutant 3’UTR of JNK1 were constructed ( Figure 4F ). Compared with the control group, miR-33-5p mimics could significantly reduce the luciferase activity of the reporter vector containing wild-type 3’UTR, but had no obvious effect on the luciferase activity of mutant 3’UTR reporter vector ( Figure 4G ). These results indicated that miR-33-5p could be complementarily bound to 3’UTR of JNK1. In addition, the results of qPCR and Western blot showed that miR-33-5p mimics significantly inhibited JNK1 expression in both MG-infected and uninfected DF-1 cells, while miR-33-5p inhibitor markedly up-regulated JNK1 expression compared with the control group ( Figure 4H ).

Next, we further studied the functional mechanism of miR-33-5p in MG infection. It has been well reported that the JNK signaling pathway is the core pathway through which JNK1 exerts a regulatory role (33, 34). The results of qPCR and Western blot showed that the mRNA and phosphorylated protein expressions of JUN and FOS in the MG infection group were significantly decreased compared with the normal group. Overexpression of miR-33-5p reduced the expression of JUN and FOS after MG infection ( Figures 5A, C ), while transfection with miR-33-5p inhibitor resulted in an opposite result ( Figures 5B, D ). These results show that miR-33-5p inhibits the JNK signaling pathway.

To further confirm that miR-33-5p mediates the JNK signaling pathway by negatively regulating JNK1, we assessed the effects of JNK1, JNK1 inhibitor (SP600125) and JNK1 over-expressing pcDNA3.1 vector. First, we confirmed that overexpression of JNK1 substantially increased mRNA and protein expression of JNK1, but SP600125 significantly inhibited JNK1 expression in DF-1 cells ( Figures 5E, F ). The results of Western blot and qPCR showed that overexpression of JNK1 up-regulated the expressions of JUN and FOS to activate the JNK pathway in both MG-infected and uninfected DF-1 cells ( Figures 5E, G ). Consistently, overexpression of miR-33-5p significantly reversed the elevated expression levels of p-JUN and p-FOS caused by overexpression of JNK1, whereas knockdown of JNK1 resulted in opposite results ( Figures 5F, H ).

To further understand the regulatory role of miR-33-5p/JNK1 on MG infection, we over-expressed or down-regulated JNK1 in MG-infected DF-1 cells through the transfection of JNK1 overexpression vector or the treatment of JNK1 inhibitor (SP600125) respectively. The results of qPCR and Western blot showed that upon MG infection, overexpression of JNK1 led to the release of pro-inflammatory factors TNF-α and IL-1β, which was opposed by the overexpression of miR-33-5p ( Figure 6A ). As expected, downregulation of JNK1 by SP600125 obviously decreased the expressions of pro-inflammatory cytokines. Interestingly, the up-regulation of TNF-α and IL-1β induced by miR-33-5p inhibitor were counteracted when co-transfected with SP600125 in MG infection groups ( Figure 6B ), indicating that miR-33-5p down-regulates the inflammation by decreasing JNK1 expression.

In addition, we examined the effect of JNK1 on cell proliferation by using the CCK-8 kit. Results showed that overexpression of JNK1 inhibited the MG-infected cell proliferation, which was partly counteracted by transfection of miR-33-5p mimics ( Figure 6C ). Then, we also examined the effect of JNK1 on cell apoptosis induced by MG infection. Flow cytometry results found that overexpression of JNK1 aggravated MG-induced apoptosis, but miR-33-5p significantly reversed this pro-apoptotic effect ( Figure 6D ). Besides, the results of Western blot and qPCR showed that JNK1 promoted the expressions of pro-apoptotic genes including FAS, Casp3, Casp8 and Casp9, but suppressed the expression of anti-apoptotic BCL-2 ( Figures 6E, F ). These results demonstrate that miR-33-5p binds to JNK1 to inhibit MG-induced cellular inflammation and apoptosis.

To further investigate the regulatory role of miR-33-5p/JNK1 on MG infection, DF-1 cells were transfected with oligonucleotides and then co-cultured with MG for 48h. As shown in Figure 6G , apart from the blank group, pMGA1.2 expression was significantly high in all infected groups. Compared with the control group, overexpression of JNK1 distinctly increased the pMGA1.2 expression, but knockdown of JNK1 had an inverse effect. Importantly, miR-33-5p mimics significantly reversed the elevated pMGA1.2 expression caused by JNK1 overexpression, while the miR-33-5p inhibitor attenuated the inhibitory effect of SP600125 on pMGA1.2 expression.

LncRNAs have been recently identified as ceRNAs that sponge miRNAs by complementary base pairing in animal diseases (35). To determine whether lncRNA is associated with miR-33-5p and regulates miR-33-5p expression, we obtained 14,674 chicken-related lncRNA sequences from LncRBase V.2 (http://dibresources.jcbose.ac.in/zhumur/lncrbase2/start2.php). Next, we analyzed and identified overlapping 1533 potential lncRNAs that were predicted to target miR-33-5p by both miRanda and TargetScan ( Figure 7A ). Then, eight lincRNAs were screened for candidate lncRNAs by a series of indicators including the length of lncRNAs, the score of miRanda and the localization of lncRNAs ( Figures 7B, C ). qPCR results showed that Lnc90386, Lnc56449, Lnc68634, Lnc86360, Lnc88478 and Lnc71825 were significantly down-regulated in MG-infected DF-1 cells ( Figure 7D ). Moreover, the results in MG-infected chicken embryo lung tissues showed that only Lnc90386 was extremely decreased in all MG-infected groups compared with the blank groups ( Figure 7E ). Therefore, Lnc90386 was selected for this study.

To investigate whether Lnc90386 could function as a ceRNA in MG infection, we first constructed a luciferase construct of wild-type Lnc90386 (Luc-Lnc90386) and a mutated form [Luc-Lnc90386(Mut)] to verify the binding relationship between miR-33-5p and Lnc90386 ( Figure 8A ). Results showed that miR-33-5p mimics significantly reduced the luciferase activity of Luc-Lnc90386, yet had no effect on mutant Luc-Lnc90386(Mut) ( Figure 8B ). These results reveal that Lnc90386 might interact with miR-33-5p by this putative binding site.

Subsequently, to further verify the relationship between Lnc90386 and miR-33-5p-JNK1, wild-type/mutant vectors overexpressing Lnc90386 [named Over-Lnc90386/Over-Lnc90386(Mut)] were successfully constructed. The Lnc90386 overexpression vector had an excellent overexpression effect ( Figure 8C ). Over-Lnc90386 or Over-Lnc90386(Mut) was co-transfected with miR-33-5p mimics and JNK1-3’UTR into DF-1 cells for 24h, respectively. The results showed that the activity of JNK1-3’UTR was significantly increased when Lnc90386 was overexpressed. However, upon mutation of the miR-33-5p and Lnc90386 binding site, the fluorescence activity of JNK1-3’UTR was significantly reduced to the point where there was no statistical difference with the miR-33-5p/JNK1-3’UTR co-transfection group ( Figure 8C ). Meanwhile, the Lnc90386 knockdown (Si-Lnc90386) experiment was also performed. Si-Lnc90386 significantly decreased the relative fluorescence activity of JNK1-3’UTR. However, Over-Lnc90386 counteracted the reduced relative fluorescence activity of JNK1-3’UTR by Si-Lnc90386 ( Figure 8D ). Thus, these results show that the presence of Lnc90386 hinders the inhibitory effect of miR-33-5p on JNK1-3’UTR.

To demonstrate that Lnc90386 can act as a ceRNA to regulate JNK1 expression by competitively binding miR-33-5p. miR-33-5p mimics, miR-33-5p inhibitor, miR-33-5p mimics + Over-Lnc90386, miR-33-5p inhibitor + Si-Lnc90386 were transfected into DF-1 cells, respectively. Subsequently, DF-1 cells were infected with 200 µL of MG at 10¹⁰ CCU/mL at 24h post-transfection. As expected, overexpression of Lnc90386 significantly repressed the expression of miR-33-5p, but this inhibitory effect could be partially counteracted by the co-transfection of miR-33-5p. In contrast, the Si-Lnc90386 group significantly increased the expression of miR-33-5p, while the Si-Lnc90386+miR-33-5p inhibitor group significantly down-regulated the miR-33-5p expression ( Figure 8E ). Meanwhile, we also tested the JNK1 expression under the same treatment by qPCR and Western blot. As shown in Figures 8F, G , overexpression of Lnc90386 significantly enhanced the JNK1 expression, while knockdown of Lnc90386 had the opposite effect. When cells were co-transfected with miR-33-5p mimics and Lnc90386, miR-33-5p mimics reduced the effect of over-Lnc90386, but knockdown of Lnc90386 had the opposite effect ( Figures 8F, G ). Interestingly, compared with all uninfected groups, miR-33-5p was significantly up-regulated while JNK1 was down-regulated in all infected groups, which indicated that down-regulated Lnc90386 could promote the miR-33-5p expression and inhibit the JNK1 expression during the MG infection. The above results indicate that Lnc90386 regulates JNK1 expression by sponging miR-33-5p.

As Lnc90386 can interact with miR-33-5p, we then tested whether Lnc90386 was capable of regulating MG infection. DF-1 cells were transfected with Over-Lnc90386, Si-Lnc90386 or co-transfected with Lnc90386 plasmid and miR-33-5p oligonucleotide, and MG was infected after 24h of the treatment. The results of qPCR and ELISA showed that overexpression of Lnc90386 resulted in up-regulation of IL-1β and TNF-α ( Figure 9A ). Interestingly, after MG infection, the up-regulation of pro-inflammatory factors induced by overexpression of Lnc90386 was partially offset by overexpression of miR-33-5p ( Figure 9A ). In contrast, knockdown of Lnc90386 down-regulated the expressions of MG-induced pro-inflammatory factors, but this trend was significantly attenuated by the miR-33-5p inhibitor ( Figure 9B ). These results indicate that Lnc90386 affects the secretion of pro-inflammatory factors by adsorbing miR-33-5p.

Subsequently, we further analyzed the effect of Lnc90386 on cell proliferation and apoptosis. As shown in Figure 9C , overexpression of Lnc90386 aggravated the loss of the MG-induced cell viability, which was attenuated by miR-33-5p mimic. Consistently, knockdown of Lnc90386 had an opposite effect ( Figure 9C ). Furthermore, enforced expression of Lnc90386 aggravated MG-induced apoptosis ( Figure 9D ), increased the expression of pro-apoptotic genes including FAS, Casp3, Casp8 and Casp9, and decreased the expression of the anti-apoptotic gene BCL-2 ( Figures 9E, F ). As expected, Lnc90386 siRNA had an exactly opposite effect ( Figures 9D–F ). It is worth mentioning that the miR-33-5p mimics partially reversed Lnc90386-induced cell apoptosis. Similarly, the miR-33-5p inhibitor attenuated the Lnc90386 siRNA-induced cell apoptosis inhibition.

More importantly, overexpressed Lnc90386 facilitated the expression of the MG adhesion protein pMGA1.2, while the opposite result was obtained when Lnc90386 was suppressed ( Figure 9G ). As expected, miR-33-5p inhibited the promotion effect of Lnc90386 on pMGA1.2.