The molecular mechanism of cell cycle arrest in the Bursa of Fabricius in chick exposed to Aflatoxin B₁

Abstract

Aflatoxin B₁ shows potent hepatotoxic, carcinogenic, genotoxic, immunotoxic potential in humans and many species of animals. The aim of this study was to clarify the underlying mechanism of G₀G₁ phase and G₂M phase arrest of cell cycle in the bursa of Fabricius in broilers exposed to dietary AFB₁. 144 one-day-old healthy Cobb broilers were randomly divided into two groups and fed on control diet and 0.6 mg·Kg⁻¹ AFB₁ diet for 3 weeks. Histological observation showed that AFB₁ induced the increase of nuclear debris and vacuoles in lymphoid follicle of BF. Results of flow cytometry studies showed that bursal cells arrested in G₂M phase at 7 days of age and blocked in G₀G₁ phase at 14 and 21 days of age following exposure to AFB₁. The qRT-PCR analysis indicated that cell cycle arrested in G₂M phase via ATM-Chk2-cdc25-cyclin B/cdc2 pathway, and blocked in G₀G₁ phase through ATM-Chk2-cdc25-cyclin D/CDK6 pathway and ATM-Chk2-p21-cyclin D/CDK6 route. In a word, our results provided new insights that AFB₁ diet induced G₂M and G₀G₁ phase blockage of BF cells in different periods, and different pathways were activated in different arrested cell cycle phase.

Introduction

Aflatoxin B₁ (AFB₁), secondary metabolites generated by the fungi Aspergillus flavus and Aspergillus parasiticus, usually can contaminate agricultural products and threaten food safety¹. AFB₁ also presents potent hepatotoxic, carcinogenic, genotoxic, immunotoxic potential^2,3 and other adverse effects in many species of animals, including rodents, fish, humans and non-human primates⁴. Immunosuppression is a major effect of AFB₁, which is characterized by injuries of mucosal immunity, cellular immunity and humoral immunity. These injuries include alteration of organ morphology and immune organ weights, reduction of T or B lymphocytes number, inhibition of lymphocyte activity^5,6, decrease of antibody production⁷, changes of T lymphocyte subsets of peripheral blood^8,9, and increased sensitivity of poultry to bacterial¹⁰, viral and protozoan diseases¹¹.

In order to clarify the mechanisms of AFB₁-induced toxicity, many researches have been focused on the mechanism of cell cycle arrest in different cells. Previous studies have shown that AFB₁ could induce G₂M phase arrest in broiler’s jejunum in vivo^12,13; S-phase accumulation in human bronchial epithelial cell in vitro¹⁴; G₀G₁ phase blockage in hepatocytes of rat in vivo¹⁵. Moreover, AFB₁ can arrest immune cells particularly lymphocytes growth at G₂M phase in vivo¹⁶. Further researches indicated that AFB₁ can affect cell cycle through different signaling pathways. For example, AFB₁-induced S-phase arrest might be mediated via inhibiting Wnt/β-catenin signaling route in HepG2 cells in vitro¹⁷, however, it may be triggered by activating ATM/ATR, Chk2, and p53 signaling pathways in human bronchial epithelial cells¹⁴. Yin et al.’s study indicated that AFB₁ induced G₂M phase arrest via ATM-Chk2-cdc25-cyclin B/cdc2 route in jejunum of broilers in vivo¹³. In rat models¹⁵, upregulation of miR-34a-5p led to cell cycle arrest at G₀G₁ phases via inhibiting cell cycle-related genes (CCND1, CCNE2 and MET) after exposed to AFB₁. Although previous studies have shown that aflatoxin-contaminated corn induced G₂M phase blockage in bursal cells¹⁸ and splenocytes⁸ of chickens, and G₀G₁ phase arrest in thymocytes¹⁸ of broilers, the molecular mechanisms and signaling pathways of cell cycle arrest of bursal cells have not been mentioned.

Bursa of Fabricius (BF), peculiar central lymphoid organ of birds, has major roles in establishment and maintenance of B cell compartment and humoral immunity¹⁹. In the present investigation, a broiler model was used to clarify the signaling pathway related molecular mechanisms involved in the cell cycle arrest of BF cells after dietary AFB₁ treatement. We analyzed the histological lesions of BF, cell cycle phase distribution of bursal cells, mRNA expression levels of regulatory molecules involved in G₀G₁, S and G₂M transitions, and the protein expression level of proliferating cell nuclear antigen (PCNA).

Results

Histopathological lesions of BF

Histopathologically, there were much more nuclear debris and vacuoles in cortical and medullary areas of bursal follicles in the AFB₁ group when compared with those of the control group at 7 and 14 days of age. At 21 days of age, the population of lymphocytes was decreased, and tissue cells in the medulla loosely arranged, but there were not so much nuclear debris as those at 7 days of age (see Fig. 1).

Figure 1.

The impact of AFB₁ exposure on BF’s histopathological (nuclear debris and vacuoles). Histological assessment of H&E-stained BF tissues of broilers at 7, 14 and 21 days of age in the control group and AFB₁ group. ^↑Marks nuclear debris in BF, *increased number of vacuoles.

Cell cycle phase distribution of bursal cells

In the AFB₁ group, the percentage of cells in G₀G₁ phase was lower than that of the control group at 7 days of age (p < 0.05), but was higher at 14 and 21 days of age (p < 0.05 or 0.01). The percentage of bursal cells in G₂M phase was increased when compared with that of the control group at 7 days of age (p < 0.05), but was significantly decreased at 14 days of age (p < 0.01). In comparison to the control group, the percentage of bursal cells in S phase were decreased at 14 and 21 days of age (p < 0.05 or 0.01), but the change was not obvious at 7 days of age (p > 0.05). There were decreased tendency on PI values between the control group and AFB₁ group at 14 and 21 days of age (p < 0.05 or 0.01), but the PI value was increased at 7 days of age (p > 0.05). These results showed that G₂M phase was arrested at 7 days of age and G₀G₁ phase was blocked at 14 and 21 days of age in chicken’s BF. Histograms of cell cycle distribution by flow cytometer were shown in Fig. 2.

Figure 2.

Effect of AFB₁ on cell cycle phase distribution of BF in chickens. Bar graph (A) shows the change of percentages of G₀G₁, S and G₂M phase distribution of BF cells. Data are presented as means ± standard deviation (n = 6). Letters *mean p < 0.05 and **mean p < 0.01 between the AFB₁ group and control group, respectively. Histograms by flow cytometry show that cell cycle arrest in G₂M phase at 7 days of age (C) and blockage in G₀G₁ phase at 14 and 21 days of age (D and E) in the AFB₁ group when compared with those of the control group (B).

Relative expressions of cell cycle-related genes

At 7 days of age, compared with the control group, the mRNA level of cdc2 was obviously decreased (p < 0.01), however, there were not significantly changes on the mRNA expressions of cyclin D₁, cyclin E₁, cyclin B₃, CDK6 and CDK2 in the AFB₁ group (p > 0.05). At 14 days of age, the mRNA contents of cyclin D₁ and cyclin E₁ were increased (p < 0.05), nevertheless, the expressions of CDK6, CDK2 and cdc2 were obviously decreased when compared with those of the control group (p < 0.01). In the AFB₁ group, the mRNA expressions of CDK6 and CDK2 were obviously decreased when compared with those of the control group at 21 days of age (p < 0.01), but there were no significant changes for cyclin D₁, cyclin E₁, cyclin B₃ and cdc2 (p > 0.05).

Compared with those of the control group, the mRNA levels of PCNA were significantly increased at 7 and 21 days of age (p < 0.01), the mRNA levels of p53 were obviously decreased at 7 and 14 days of age (p < 0.01). However, the expressions of p21 were increased at 14 and 21 days of age (p < 0.05), but there was no obvious change at 7 days of age. Furthermore, the mRNA expressions of ATM and Chk2 were higher at 7, 14 and 21 days of age than those in the control group (p < 0.05 or 0.01), and the mRNA levels of cdc25 were lower at 14 and 21 days of age (p > 0.05). The results were shown in Fig. 3 and Supplementary Fig. S1. The representatives of amplification curves of aforementioned 12 genes at 14 days of age were displayed in Supplementary Fig. S2.

Figure 3.

The relative expressions of mRNA in bursal cells from the broilers in the AFB₁ group. Bar graph (A). The mRNA expressions of cyclin D₁, CDK6, cyclin E₁, CDK2, cyclin B₃ and cdc2 in the bursal cells of the AFB₁-fed broilers are expressed as fold change relative to the control-fed broilers. Bar graph (B). The mRNA levels of ATM, Chk2, cdc25, p53, p21 and PCNA in the bursal cells of the AFB₁-fed broilers are expressed as fold change relative to the control-fed broilers. All data are expressed as means ± standard deviation. *p < 0.05, **p < 0.01 vs control, n = 6 for each group.

PCNA Expression

As shown in Fig. 4, the nuclei of PCNA-positive cells were stained with brown color by immunohistochemical method. Compared with the control group, the numbers of PCNA-positive cells were increased in the AFB₁ group. According to Fig. 4C, the IOD of PCNA-positive cells were found to be significantly increased in the AFB₁ group at 7 and 21 days of age.

Figure 4.

Protein expression of PCNA in bursal cells from the broilers in the control group and AFB₁ group. Image (A and B) are representatives of the negative control staining in the control group and AFB₁ group at 7 days of age, respectively. Line chart (C). shows the change of integrated optical density (IOD) of PCNA-positive cells by immunohistochemical method. All data are expressed as means ± standard deviation. **p < 0.01 vs control, n = 6 for each group.

Discussion

BF is primary central lymphoid organ of chicken²⁰, which is related to diversification of B cells¹⁹. Histopathologically, consistent with our earlier studies^8,18,21, obvious lymphocyte depletion, more vacuoles and debris were observed in the BF in the AFB₁ group. It has been reported that chickens with poorly developed bursal follicles are expected to be very sensitive to bacterial²² and viral²³ diseases, because lymphoid follicles of BF have a crucial roles in humoral immune reactions²⁴. So these histological lesions could finally impaire the humoral immune function in chickens after exposed to AFB₁.

Cell cycle is divided into different phases, including G₀, G₁, S, G₂ and M phases. By FCM method, different phases of cell cycle are normally determined based on DNA content²⁵. Many studies showed that AFB₁ induce cell cycle arrest at different phases depending on the cell types, such as the accumulation of G₂M phase cell in thymocytes²⁶ and broiler’s jejunal epithelia¹³, the increase of S-phase cell population in human bronchial epithelial cells¹⁴, and the arrest at G₀G₁ phases in lymphocytes¹⁶ and liver cells²⁷. Our previous study have shown that the percentage of bursal cells in the G₂M phase was increased after treated with AFB-contaminated corn¹⁸. However, the present study showed that consumption of 0.6 mg·kg⁻¹ AFB₁ diet induced the arrest at G₂M phase in bursal cells of broilers at 7 days of age and the blockage at G₀G₁ phases at 14 and 21 days of age. The results showed an interesting finding that the characteristics of cell cycle arrest could be changeable in the same kinds of cells with the increase of the AFB₁ exposure time. Similar results have also been reported in splenocytes and thymocytes^8,18. Why were the bursal cells arrested in different cell cycle phase at different period after exposed to AFB₁? The transcription and translation of different cyclin genes may be related to the mechanisms of different cell cycle phase arrest. To further investigate the possible mechanism, the expressions of some cyclin genes were detected with qRT-PCR method.

Traditionally, as we know that the ATM-Chk2-cdc25 route and ATM-p21-p53-dependent pathway are the two classical pathways in cell cycle progression. Briefly, Ataxia telangiectasia mutated (ATM) is the primary kinase activated by DNA double-strand breaks (DSBs), which mediates the downstream signal cascade leading to cell cycle slowdown, DNA repair, and chromatin remodeling²⁸. At the same time, in the process of DNA-damage, Chk2 is phosphorylated by phospho-ATM²⁹, and then the phospho-Chk2 inhibits the activity of cdc25^30,31. On the other hand, cyclin/CDK complexes are able to activate cdc25. More importantly, AFB₁ can generally be metabolized by cytochrome P450 (CYP450) enzymes to generate active AFB₁-8,9-epoxide (AFBO)³², which can react with DNA to form AFB₁-FAPY adducts after forming AFB₁-N7-Gua adducts⁴. The accumulation of AFB₁-DNA adducts often induce DNA damage¹⁴. In the present study, the increase of ATM and Chk2 mRNA expressions, and the decrease of cdc25 mRNA level were observed at 7 days of age. Notably, the cyclin B₃/cdc2 complexes were maintained at a relatively low level when compared with those of the control group. These results suggested that AFB₁ activated ATM-Chk2-cdc25 pathway to inhibit cyclin B₃/cdc2 expression in G₂M phase. Similar results from other researchers were reported. Yin. et al.¹³ and Yang et al.¹⁴ showed that AFB₁ induced accumulation of ATM and Chk2 in broiler’s jejunal epithelia and human bronchial epithelia in vivo and vitro, respectively.

p53, a tumor suppressor gene, has been shown to mediate cell cycle arrest, apoptosis and senescence in response to DNA damage^33,34. p21 gene expression is generally under the transcriptional control of p53 protein³⁵. However, the accumulated evidence indicates that p21-mediated cell cycle arrest can be either p53-dependent or p53-independent pathway^36–38. Recent studies suggested that other unconventional cell cycle pathways are also involved in cell cycle progression. Namely, The full activation of Chk2 kinase can induce Chk2-dependent accumulation of p21³⁹, and then induce cell cycle arrest at G₁ phase through activating ATM-Chk2-p21 pathway in HaCaT cells in vitro⁴⁰. In the current study, we found that the mRNA expressions of ATM, Chk2 and p21 were increased, and the expressions of p53 and cyclin D₁/CDK6 complexes were decreased in the AFB₁ group at 14 and 21 days of age. Interestingly, the expressions of cdc25 were also decreased when compared with that of the control group. These results indicated that AFB₁ induce cell cycle arrest in G₀G₁ phase via two different routes at 14 and 21 days of age, namely ATM-Chk2-cdc25 pathway and ATM-Chk2-p21-p53-independent route.

p21 is a cyclin-dependent kinase inhibitor (CKI), which possesses the highest binding affinity among all PCNA-interacting partners⁴¹. Protracted p21 expression abrogated the property of PCNA to up-regulate CDKs and suppressed the PCNA mono-ubiquitination. Our results showed that AFB₁ induced the increase of PCNA proteins and mRNA expressions. Because PCNA needs to be ubiquitinated for performing its biological function, these up-regulations of PCNA could not be certainly related to its promoted functions. On the other hand, p21 induces cell cycle arrest at G₁⁴² and G₂M⁴³ phases by inhibiting the activity of CDKs such as CDK6, CDK2 and cdc2^42,44. In this study, we found that the expressions of cyclin B₃/cdc2 and cyclin D₁/CDK6 complexs were decreased in the AFB₁ group during the experiment. Therefore, maintaining a relatively low amounts of active cyclin/CDK complexes appears to be a major factor to regulate cell cycle progression⁴⁵.

In summary, our results showed that 0.6 mg·kg⁻¹ dietary AFB₁ can induce histopathological lesions of bursa of Fabricius and cell cycle arrest in the bursal cells of broilers. Briefly, AFB₁ induced G₂M phase arrest via ATM-Chk2-cdc25-cyclin B/cdc2 pathway at 7 days of age, and G₀G₁ phase blockage through ATM-Chk2-cdc25-cyclin D/CDK6 pathway and ATM-Chk2-p21-cyclin D/CDK6 route at 14 and 21 days of age (see Fig. 5). Our results suggested that different mechanisms of G₀G₁ and G₂M cell cycle might be involved in different stages of the development of BF, and could provide reference for deeper understanding of the mechanism of AFB₁ induced immunosuppression for the same or similar studies in both human and other animals in the future.

Figure 5.

Schematic diagram of the proposed mechanisms of AFB₁ induced cell cycle arrest in BF.

Materials and Methods

Animals and diets

One hundred and forty-four one-day-old healthy Cobb broilers from Chia Tai Group (Wenjiang, Sichuan, China), weighed 40 ± 5 g, were randomly divided into two groups, namely control group (0 mg·kg⁻¹ AFB₁) and AFB₁ group (0.6 mg·kg⁻¹ AFB₁). Each group consisted of three replications with 24 birds per replication. The control group received corn-soybean basal diet, which was formulated according to the National Research Council (NRC, 1994)⁴⁶ and the Chinese Feeding Standard of Chicken (NY/T33-2004) recommendations, and the AFB₁ group was fed with AFB₁ diet. The AFB₁ diet was made according to the method described by Kaoud et al.⁴⁷. Briefly, 27 mg AFB₁ (A6636, Sigma-Aldrich, USA) farinose solid was dissolved into 30 ml methanol completely, and then the mixture was mixed into 45 kg basal diet to formulate the AFB₁ diet. The equivalent methanol was mixed into the basal diet to produce control diet. Afterward, the methanol of diets was evaporated at 98 °F (37 °C). The AFB₁ concentrations were analyzed by HPLC with fluorescence detection (Waters Model 2475), and the AFB₁ concentrations were determined as <0.01 mg·kg⁻¹ for the control diet and 0.601 mg·kg⁻¹ for the AFB₁ diet. Broilers were housed in cages with electrically heated units and provided with water as well as aforementioned diet ad libitum for 21 days. Previous studies indicated that deficient or a complete lack of a functional glutathione-S-transferase (GST) with affinity toward AFBO appears to be a major reason that poultry are extremely susceptible to AFB₁⁴⁸. Additionally, 0.6 mg·kg⁻¹ AFB₁ in diet had obvious adverse effects on the immune organs of broilers, such as spleen⁴⁹, thymus and BF⁵⁰. Based on these information, experimental model (broiler) and toxin concentrations (0.6 mg·kg⁻¹ AFB₁) were chosen.

All study procedures followed the medical ethics according to national and international guidelines and has been approved by Sichuan Agricultural University Animal Care and Use Committee (Approval No: 2012-024).

Histopathological examination

At 7, 14, and 21 days of age during the experiment, six broiler chickens were randomly selected from each group, and euthanized. The BF were immediately removed and fixed in 4% paraformaldehyde (PFA), embedded in paraffin, and sectioned at 5 μm. Some sections were prepared for immunohistochemistry, and some were stained with hematoxylin and eosin Y (H.E) for light microscopic observations. Typical histological changes were photographed with a digital camera (Nikon DS-Ril, Japan).

Flow cytometry assay

At 7, 14 and 21 days of age during the experiment, six broilers in each group were sampled to determine the cell cycle phase distribution of BF cells by Flow Cytometry, with a similar method described by Chen et al.⁵¹. Single cell suspension of BF was harvested by dissecting each sample into small pieces and filtered through 300-mesh nylon gauze. Then, the cells were washed and suspended in phosphate buffered saline (PBS, pH 7.2~7.4) at a concentration of 1 × 10⁶ cells/mL. A total of 1 mL cell suspension was transferred into 5 mL culture tube and centrifuged at 800 × g for 5 min. Afterward, the cell suspension was permeabilized with 500 µL 0.25% Tritonx-100 at 4 °C for 20 min and centrifuged at 800 × g for 5 min. The supernatant was discarded, and then 1.5 µL of staining solution (Propidium iodide, BD Pharmingen, USA, 51-66211E) were added and incubated for 30 min at 4 °C in a dark room. Finally 400 µL of PBS was added and the cell cycle phase distribution was assayed by flow cytometry (BD FACSCalibur) within 45 minutes and analyzed by Mod Fit LT software for Mac V3.0 computer program.

$$\mathrm{Proliferating}\mathrm{index}\mathrm{value}\left(\mathrm{PI}\mathrm{value}\right)=\frac{S+G_{2}M}{G_0{G}_1+S+G_{2}M}\times 100\%$$

qRT-PCR

BF from six birds in each group was removed and immediately stored in liquid nitrogen at 7, 14, and 21 days of age. Then, these BF samples were homogenized by crushing with a pestle and mortar until powdery. The powdered tissues were collected into eppendorf tubes and stored at −80 °C. As previously described¹³, total RNA was extracted from the powdery using TriPure Isolation Reagent (Cat No. 11667165001, Roche Applied Science, Germany) following the manufacturer’s instruction. Next, 1 µg of total RNA was used for reverse transcription using Transcriptor First Strand cDNA Synthesis Kit (Cat No. 04897030001, Roche Applied Science, Germany). For qRT-PCR reactions, cDNA was amplified using Bestar^® SybrGreen qPCR mastermix (Cat No. DBI-2043, DBI^® Bioscience, Germany) according to the manufacturer’s instruction for 40 cycles on a Bio-Rad C1000 Thermal Cycler (Step One Plus, Applied Biosystems, USA). Chicken β-actin was used as an reference gene⁵². Sequences for target genes (Table 1) were obtained from GenBank of NCBI and synthesized by Sangon Biotech (Shanghai, China). qRT-PCR data were analyzed by using the 2^−∆∆Ct calculation method⁵³ and hierarchical cluster of gene expression data were analyzed by using HemI 1.0 software (Heatmap Illustrator, China).

Table 1.

Primer sequence for proliferation genes.

Gene symbol	Accession number	Forward Primer (5′-3′)	Reverse Primer (5′-3′)	Product size
ATM	NM001162400.1	TTGCCACACTCTTTCCATGT	CCCACTGCATATTCCTCCAT	110bp
cdc25	NM001199572.1	AGCGAAGATGATGACGGATT	GCAGAGATGAAGAGCCAAAGA	163bp
CDK2	NM_001199857.1	TCCGTATCTTCCGCACGTTG	GCTTGTTGGGATCGTAGTGC	276bp
CDK6	NM_001007892.2	CCAGACCCGCACAACCTATT	TCTTGGCTGGATTGAACGCT	96bp
cdc2	NM205314.1	TCTGCTCTGTATTCCACTCCTG	ATTGTTGGGTGTCCCTAAAGC	144bp
Chk2	NM001080107	AGACCAAATCACTCGTGGAGAATAC	GATGCTCTAAGGCTTCCTCTATTGT	140bp
cyclin D1	NM_205381.1	GACTTTTGTGGCTCTGTGCG	CTGTTCTTGGCAGGCTCGTA	202bp
cyclin E1	NM_001031358.1	CGCCACCACAAAGCAGTAAG	TCACCGGCAGCATTTCCATA	137bp
cyclin B3	NM205239.2	ATCACCAACGCTCACAAGAAC	AGGCTCCACAGGAACATCTG	171bp
p21	AF513031.1	TCCCTGCCCTGTACTGTCTAA	GCGTGGGCTCTTCCTATACAT	123bp
p53	NM_205264.1	TGGAACCATTGCTGGAACCC	AGTTGCTGTGATCCTCAGGG	127bp
PCNA	AB053163.1	GATGTTCCTCTCGTTGTGGAG	CAGTGCAGTTAAGAGCCTTCC	104bp
β-actin	L08165	TGCTGTGTTCCCATCTATCG	TTGGTGACAATACCGTGTTCA	178bp

Immunohistochemistry (IHC)

The immunohistochemistry technique used for PCNA was performed on 5 µm thickness sections according to the report by Yu et al.⁵⁴. Briefly, the sections were deparaffinized and rehydrated. After washed three times with PBS (0.1 M, pH 7.2~7.4), the sections were treated with 3.0% hydrogen peroxide in PBS at room temperature for 10 min to quench the endogenous peroxidase activity. Following washed with PBS, the sections were exposed to normal goat sera for 30 min to block nonspecific antibody binding. In a humidified chamber, the sections were incubated the rabbit anti-PCNA polyclonal antibody (bs-0754R, Bioss, Beijing, China) for 20 h at 4 °C (working dilution: 1:100). After three successive washings in PBS, secondary antibody biotinylated goat anti-rabbit IgG and streptavidin-biotin complex (SA1020, Boster, Wuhan, China) were, in turn, applied onto sections for 1 h and 30 min at 37 °C, respectively. Slides were visualized with diaminobenzidine hydrochloride (AR1022, Boster, Wuhan, China) under the microscope and stopped by immersion in distilled water, as soon as brown staining was visible. Finally, the sections were lightly counterstained with hematoxylin and placed in absolute ethylalcohol and xylene for 3 min following coverslipped. Negative controls were performed in the same way, except that PBS was used as a substitute for the primary antibody.

The stained sections were photographed using a digital camera at ×1000 magnification. For each section, six randomly selected fields were used for analyzing integrated optical density (IOD) using Image Pro Plus 5.0 software (USA)⁵⁴.

Statistical Analysis

Statistical analyses were performed using SPSS 18.0 (SPSS Inc, Chicago, IL, USA). The experimental data were expressed as mean ± standard deviation ($\overline{X}$ ± SD) and independent sample test followed by post hoc t test was applied to determine the level of significance. Statistical significance was considered at p < 0.05 and markedly significant was considered at p < 0.01.

Author Contributions

P.H., H.L. and F.Y.W. carried out the majority of this work (experiment design, sample collection, statistical analysis and manuscript modification); Z.C.Z., X.P. and J.F. assisted in manuscript preparation and modification; H.M.C., C.X.G., H.T.S., Y.Z. and Z.L.C. directed research; P.H. wrote the manuscript and all the authors provided comments and feedback on the manuscript.

Competing Interests

The authors declare that they have no competing interests.