Endothelin increases the proliferation of rat olfactory mucosa cells - 图文

NEURAL REGENERATION RESEARCH www.nrronline.orgRESEARCH ARTICLE

Endothelin increases the proliferation of rat olfactory mucosa cells

Bertrand Bryche1, 2, Audrey Saint-Albin1, Claire Le Poupon Schlegel1, Christine Baly1, Patrice Congar1, Nicolas Meunier1, 2, *

Funding: This work was funded by the Institut National de la Recherche Agronomique (INRA).

1 Neurobiologie de l’olfaction, Institut National de Recherche Agronomique (INRA), Université Paris-Saclay, Jouy-en-Josas, France2 Université de Versailles Saint-Quentin en Yvelines, Département de Biologie, Versailles, France

Abstract

*Correspondence to:

The olfactory mucosa holds olfactory sensory neurons directly in contact with an aggressive environment. In order to maintain its integrity, it is one of the few neural zones which are continuously renewed during the whole animal life. Among several factors regulating this renewal, endothelin acts as an anti-apoptotic factor in the rat olfactory epithelium. In the present study, we explored whether endothelin could also act as a proliferative factor. Using primary culture of the olfactory mucosa, we found that an early treatment with endothelin increased its growth. Consistently, a treatment with a mixture of BQ123 and BQ788 (endothelin re-ceptor antagonists) decreased the primary culture growth without affecting the cellular death level. We then used combined approaches of calcium imaging, reverse transcriptase-quantitative polymerase chain reaction and protein level measurements to show that endothelin was locally synthetized by the primary culture until it reached confluency. Furthermore, in vivo intranasal instillation of endothelin receptor antagonists led to a decrease of olfactory mucosa cell expressing proliferating cell nuclear antigen (PCNA), a marker of prolifer-ation. Only short-term treatment reduced the PCNA level in the olfactory mucosa cells. When the treatment was prolonged, the PCNA level was not statistically affected but the expression level of endothelin was in-creased. Overall, our results show that endothelin plays a proliferative role in the olfactory mucosa and that its level is dynamically regulated. This study was approved by the Comité d’éthique en expérimentation ani-male COMETHEA (COMETHEA C2EA -45; protocol approval #12-058) on November 28, 2012. Key Words: autocrine factor; cell culture; cellular dynamics; endothelin; olfaction; olfactory basal cells; olfactory epithelium; olfactory mucosa primary cultureChinese Library Classification No. R452; R364; R741

Nicolas Meunier, PhD, nicolas.meunier@inra.fr.orcid:

0000-0002-2769-647X (Nicolas Meunier)

doi: 10.4103/1673-5374.265558 Received: February 26, 2024Accepted: June 5, 2024

Introduction

The first step of odour detection takes place in the olfactory epithelium. This sensory neuroepithelium is mainly com-posed of olfactory sensory neurons (OSNs) surrounded by sustentacular cells. It lies on a lamina propria composed of Bowman’s glands, blood vessels, fibroblasts and OSNs axons wrapped in olfactory ensheathing cells. The lamina propria and the olfactory epithelium form the olfactory mucosa (OM). OSNs possess cilia-holding olfactory receptors ca-pable of binding odorants. These cilia are bathed in mucus directly in contact with the environment. They are exposed to oxidative stress, noxious substances and pathogens that gradually damage the OSNs. While neuronal injury in the central nervous system cannot be repaired, the olfactory epithelium undergoes regular apoptosis and proliferation throughout life to maintain the sense of smell (Graziadei, 1973; Schwob, 2002; Brann and Firestein, 2014). This unique capacity has given rise to interest in this mechanism as a po-tential way to regenerate other injured neuronal tissue espe-cially as it involves the targeting of newly formed OSN axons to the central nervous system (Yao et al., 2024). The renewal of the OM relies on the proliferation of two stem cells type: globose and horizontal basal cells which are engaged dif-ferently in the process according to the severity of the OM damages (Schwob et al., 2017). Numerous growth and sur-352

vival factors are involved in this process and studies of their characterization are increasing in number. Many such factors have been described since the review by Mackay-Sim and Chuah (Mackay-Sima and Chuahb, 2000), from pituitary adenylyl cyclase-activating polypeptide (Hansel et al., 2001), to C-type natriuretic peptide, along with the brain-derived neurotrophic factor (Simpson et al., 2002), neurotrophin (Simpson et al., 2003), the leukemia inhibiting factor acting on immature OSNs (Moon et al., 2009), and more recently BDNF (Ortiz-Lopez et al., 2017).

Endothelin (ET) was first characterized as the strongest vasoconstrictive peptide (Yanagisawa et al., 1988) but has since been described as a pleiotropic peptide, acting on various cellular processes including cellular population dy-namics in different tissues (Schinelli, 2006). Three peptides of 21 amino acids named ET-1 to ET-3 (Edn1 to Edn3) have been characterized, ET-1 being the most widely distribut-ed (Khodorova et al., 2009). These peptides are secreted as pro-peptides (big-ETs), which are matured locally through peptidase endothelin-converting enzymes (mainly Ece1). They then act autocrinally and/or paracrinally on two dif-ferent G protein-coupled receptors, named ETA and ETB (Ednra and Ednrb, respectively). The peptides along with their converting enzymes and receptors are expressed in the olfactory epithelium. This expression is particularly Bryche B, Saint-Albin A, Le Poupon Schlegel C, Baly C, Congar P, Meunier N (2024) Endothelin increases the proliferation of rat olfactory

mucosa cells. Neural Regen Res 15(2):352-360. doi:10.4103/1673-5374.265558

strong in the basal cell and neuronal cell layers in 10-day-old rats, with a similar level of expression for ETBourhis et al., 2014). Endothelin acts as an anti-apoptotic A and ETB (Le factor in the OM (Laziz et al., 2011) but ET-1 is also known to act as a proliferative factor in the central nervous system (Vidovic et al., 2008). In the present study, we explored this potential role in the rat OM using primary culture and vivoin

approaches.Animals

Material and Methods

Male Wistar rat pups (from multiparous females, and housed in our local animal Rattus norvegicus) were obtained care facilities in 12-hour light, 12-hour dark cycles. Breed-ers were fed standard chow instillations on pups as described previously (Francois et al., ad libitum. We performed nasal 2013), either for one-week starting on day 3 or for 24 hours at the age of 9 days. For the one-week treatment, instillations were performed twice a day (at 9 a.m. and 6 p.m.) with 5 to 8 μL (increasing the volume by 0.5 μL every day) per nostril of either a mixture of endothelin receptor antagonists (BQand BQ123 at 10–5788for proliferating cell nuclear antigen (PCNA) and quantita- M in PBS; referred to BQs in the following; , Sigma Aldrich, Saint-Quentin Fallavier, France) n = 13/6 tive polymerase chain reaction (qPCR) experiments, respec-tively) or PBS with 10–4into account the initial dilution of BQ788 in 10% of NaOH (pH 7.6; vehicle taking –2= 15/7 for PCNA and qPCR experiments, respectively). The % NaOH; n higher concentration used in the nasal instillation compared to the treatment of the primary culture took into account the dilution in the mucus, as well as the low amount reaching the olfactory epithelium. Similar differences in treatment ef-fectiveness have been observed among others for vasopressin treatments (Ludwig et al., 2013). For the one-day treatment starting at the age of 9 days, instillations were performed at 9 a.m., 6 p.m. and again at 9 p.m. the next day (for 10?4% of NaOH/10?5n = 5/6 randomly allocated to treatments taking into account their M BQs respectively). Pups were weight (17.4 ± 0.53 g at the age of 9 days) and sex. Pups were euthanized 1h after the last instillation. For the OM primary culture, we used 2 pups for the quantification of OM prima-ry culture confluence, six pups to prepare two independent experiments for the quantification of cellular population by lactate dehydrogenase content, two pups for the ET-1 quanti-fication by enzyme immunoassay (EIA), two pups for reverse transcriptase-qPCR (RT-qPCR) and 6 pups for the calcium imaging.

with the European Communities Council Directive 2010/63/All animal experiments were conducted in accordance EU on the protection of animals used for scientific purposes, and approved by the Comité d’animale COMETHEA (COMETHEA C2EA -45; protocol éthique en expérimentation approval #12-058) on November 28, 2012. Investigators hold the Individual Authorization for Performing Experiments in Animals, including the animal’s experiments conducted in the present study (agreements #78-154/R-94ENVA-F1-12).

OM primary cultures

We used a long-term primary culture of rat OM cells as previously described (Gouadon et al., 2010). Briefly, for each experiment, two to three Wistar rats (7 to 10 day old) were euthanized by decapitation and their entire OM was dissected from the turbinates and the olfactory part of the septum. After enzymatic digestion and mechanical dissociation, the suspension was passed through a 40 μm filter to remove undigested material. After centrifugation through 10% new-born calf serum gradient (NCS; 150 × 10 minutes at room temperature), cells were suspended in g, sterile Dulbecco’s Modified Eagle Medium/ HamF12 based medium (DMEM/Ham F12; Eurobio, les Ulis, France) con-taining 10% NCS (Eurobio). They were plated at a density of approximately 120 cells/mm2assay), 12-well plates filled with a glass coverslip previously on 6-well plates (for qPCR coated with 0.01% poly L-lysine (for confluence evolution measurement and calcium imaging), or 96-well plates (for lactate dehydrogenase (LDH) measurements). Cells were grown at 37°C under 5% COto limit the influence of external growth factors in order to 2. NCS level was reduced to 2% evaluate the impact of ET-1 and endothelin receptor antag-onists on cell proliferation.Quantification of OM primary culture confluence

To measure the degree of confluence of the cellular culture, we took an average of 18 images per glass coverslip for each day of culture from day 2 to day 8. We used ImageJ (Rasband, WMD, USA, http://imagej.nih.gov/ij/, 1997-2012) to quantify .S., ImageJ, U. S. National Institutes of Health, Bethesda, the percentage of total area filled by cells (each day).

n = 3 coverslips for Quantification of cellular population by LDH contentWe assessed the growth of the OM primary culture follow-ing the different treatments by quantifying the content of LDH in the cells as described previously (Allen et al., 1994; Stander et al., 2024). The LDH concentration is proportional to the rate of NADH oxidation, and we thus measured activ-ities of both the extracellular (related to the level of cellular death) and the intracellular LDH (related to the cellular population). This was respectively measured in the extracel-lular culture medium and in the primary cultured adherent cells extemporaneously lysed with PBT (PBS + 0.5% Triton X-100). Results were normalized to the control condition (CTL) for each culture (cultures).

n = 8 from two independent primary ET-1 measurement by EIA

ET-1 concentration in the OM primary culture medium was measured using a commercial EIA kit (Cayman Chem-ical, Ann Arbor, MI, USA) according to the manufacturer instructions. Plates were read at 405 standard curve. Data are presented either as the raw values nm and compared to a of ET-1 concentrations or divided by the slope of NADH oxidation in the presence of intracellular LDH content. This last measure accounts for the production of ET-1 relative to

353

Bryche B, Saint-Albin A, Le Poupon Schlegel C, Baly C, Congar P, Meunier N (2024) Endothelin increases the proliferation of rat olfactory mucosa cells. Neural Regen Res 15(2):352-360. doi:10.4103/1673-5374.265558

the cellular population (DIV5 and DIV7 respectively from two independent primary n = 5/4/5 for day in vitro 3 (DIV3), cultures).

qPCR

Total RNA was extracted from OM primary culture cells us-ing the Trizol method and subsequently treated with DNase I. OligodT first strand cDNA were synthesized from 5 μg total RNA by the Superscript II reverse transcriptase (Invitrogen, Cergy Pontoise, France) following the manufacturer recom-mendations. 5 μL of 200-fold diluted cDNA templates were added to a 15 μL reaction mixture containing 200 nM prim-ers (sequences are shown in Green GoTaq?France). The expression levels of target genes were measured qPCR Master Mix (Promega, Charbonnieres, Additional Table 1) and SYBR using the CFX Connect qPCR platform (BioRad, Hercules, CA, USA). A dissociation curve was carried out at the end of the PCR cycle to verify efficiency of primers to produce a single and specific PCR amplification. Quantification was achieved using the ΔΔCt method, and mRNA expression was normalized to the expression level of β-actin (Francois et al., 2016). An efficiency corrective factor was applied for each primer pair.

Calcium imaging

OM primary culture responses to stimulations with ET-1 (10–7M) and adenosine triphosphate (ATP tored using microscopic epifluorescence Ca, 10–4 M) were moni-2+were loaded with Fura-2 AM, using a dedicated Xcellence RT imaging. Cells imaging station (Olympus Soft Imaging Solutions – OSIS, Münster, Germany) as previously described (Gouadon et al., 2010). Standard image pairs of Fura-2 fluorescence, excited at 340 and 380 nm, were acquired at 510 nm at a frequency of 1 Hz, with short exposure times (200 ms) to minimize bleaching, using an UPlanApo oil immersion objective with a 10× magnification. Images were background subtracted and kinetics of Fura-2 fluorescence ratios 340/380 nm of defined regions of interest (ROI) were calculated offline and normal-ized to the baseline averaged ratio. Changes in fluorescence (seconds before stimulation) and expressed as F) were calculated relative to the averaged baseline (from 30 F%?F/F ([(F–Soft Imaging Solutions OSIS, Münster, Germany).baseline)/Fbaseline] × 100) using Xcellence RT software (Olympus Immunohistochemistry

Immunohistochemistry of OM tissue sections was per-formed as described previously (Laziz et al., 2011). Briefly, the nasal septum and turbinates were removed as a block and post-fixed overnight at 4°C in 4% paraformaldehyde PBS. Blocks were cryoprotected with sucrose (30%) and cryo-sectioned sagitally (14 μm thickness). Sections were kept frozen at antigen retrieval in a citrate buffer (pH 6) at 95°C for 30 –80°C until use. PCNA staining requires an minutes. Non-specific staining was blocked by incubation with 10% non-immune goat serum diluted in PBS contain-ing 2% bovine serum albumin and 0.3% Triton X-100. The sections were then incubated overnight at 4°C with primary

354

antibodies directed against PCNA (1:200; mouse monoclo-nal PC10, GeneTex; Tebu-bio SAS, Le Perray-en-Yvelines, France). Fluorescence staining was performed with Al-exa-Fluor-488-conjugated goat secondary antibodies (1:1000; Molecular Probes; Invitrogen, Cergy Pontoise, France) for 150 minutes at room temperature. Immunohistochemistry was performed on medial transversal sections of the OM. We took 6 to 8 images located either dorso-medially at the base of the septum or on lateral turbinates similar to a re-cent study (Hasegawa-Ishii et al., 2017). Images were taken blind to the animal origin at 100× magnification using an Olympus IX71 inverted microscope equipped with an Orca ER Hamamatsu cooled CCD camera (Hamamatsu Photon-ics France, Massy, France). Images were quantified blind to the animal origin using ImageJ to threshold specific PCNA staining (Laziz et al., 2011). As the antigen retrieval treat-ment required for PCNA staining does not allow sufficient nuclear staining, we used the auto fluorescence of the olfac-tory epithelium at 555 nm excitation to measure its area. On each image, we measure the PCNA staining from an OE area averaging ~0.08 mm2quantify the percentage of PCNA staining area in the olfac-. These measurements allowed us to tory epithelium reflecting the level of proliferation.

Statistical analysis

Group data presented in the text and figures are expressed as mean ± standard error of the mean (SEM). Statistical signif-icance was assessed using non-parametric Mann-Whitney Ufollowed by Bonferroni multiple comparison tests, one-way or two-way analysis of variance (ANOVA) (GraphPad Prism 5.0, GraphPad software Inc., La Jolla, San post-hoc tests Jose, CA, USA). A probability value of significant differences.

P < 0.05 was set for ET-1 acts as a proliferative factor on OM primary culture Results

cells

In order to confirm and quantify a potential proliferative action of ET-1 on OM cells, we treated an OM primary cul-ture with ET-1 at 10–9in vitro M every 48 hours from day 0 to day 4 primary culture which is growing rapidly for one week (DIV; Figure 1A). We used a previously developed vitroin At DIV6, we measured the LDH content related to the cell before reaching confluence around DIV7 (Figure 1B). population (Allen et al., 1994; Laziz et al., 2011). We ob-served a slight but significant increase in the LDH content after ET-1 treatment (= 0.016) showing that ET-1 could modify the cellular dy-Figure 1C; Mann-Whitney U test, P namic of the mucosa, resulting in increased cell density in the culture. Proliferative factors can be provided directly by the primary culture cells (Price, 2017) and big ET-1 and its converting enzyme to ET-1 are expressed by an OM primary culture (Gouadon et al., 2010). We thus hypothesized that a substantial local production of ET-1 could contribute to the primary culture growth. In order to test this possibility, we treated the primary culture with a mixture of BQreceptor antagonist) and BQ123 (ETA 788 (ETB receptor antagonist) in

Bryche B, Saint-Albin A, Le Poupon Schlegel C, Baly C, Congar P, Meunier N (2024) Endothelin increases the proliferation of rat olfactory

mucosa cells. Neural Regen Res 15(2):352-360. doi:10.4103/1673-5374.265558

the same conditions from DIV0 to DIV4. For simplicity, this mixture known to efficiently block both ET-1 receptors will be referred to BQs in the following. Treatment with BQs at 10–9of = 0.023). This decrease in intracellular LDH could be related in vitro M significantly decreased the LDH content after 6 days development (Figure 1D; Mann-Whitney U test, P to a BQs-induced antagonism in ET-1 proliferative action. Another possibility relies on a BQs-induced cellular death in the OM primary culture cells. Such cellular death would lead to an increase in LDH present in the extracellular medi-um (Laziz et al., 2011). We thus measured the level of LDH present in the primary culture medium after BQs treatment, finding great variability in the measurement as it was at the limit of detection (to the control condition and in the same range as basal LDH Figure 1E). LDH levels were also similar activity measured in the cell-free culture medium containing 2% NCS. The absence of cellular LDH release shows that the BQs treatments did not impact the cellular death level in the primary culture. Overall, these results suggest that ET-1 can be produced locally by the OM primary culture cells and ac-tively promote cell growth.

OM primary culture produces ET-1 until reaching confluence

In order to characterize the level of local ET-1 production by the OM primary culture cells, we measured the presence of ET-1 in the medium of culture at DIV3, DIV5 and DIV7 directly by EIA (primary culture medium without cells (0.03 ± 0.1 pg/mL, i.eFigure 2A). ET-1 was not present in the 0.012 pM; i.e.pM; , 14.8 pM; n = 4), but was present at DIV3 (36.9 ± 3.4 pg/mL, ., n = 5) and DIV5 (37.1 ± 3.1 pg/mL, i.e., (0.2 ± 0.5 pg/mL; n = 4). However at DIV7, ET-1 was no longer detectable 14.9 the medium relative to the cellular density estimated from n = 5). We normalized the level of ET-1 in LDH content to express the production of ET-1 relative to the cellular density (by Bonferroni multiple comparison test showed that the Figure 2B). One-way ANOVA followed ET-1 production significantly decreased from DIV3 to DIV7 (showing that big-ET1 expression similarly decreased from P < 0.001). This was consistent with RT-qPCR analysis DIV3 to DIV7 (pression of ETFigure 2B; P = 0.003). By contrast, the ex-culture age (Figure 2CA and ET; PB receptors increased gradually with = 0.017 and 0.003, respectively).OM primary culture cells do not respond to ET-1 stimulation until its endogenous production level decreases

Primary culture of OM cells are strongly desensitized after ET-1 stimulation (Gouadon et al., 2010). If ET-1 was secret-ed in the culture medium by OM cells, then they should not be responsive to ET-1 stimulation until ET-1 level decreases. We recorded from DIV1 to DIV8 the response of the OM primary culture cells to exogenous stimulation by ET-1 (10–7M) and ATP (10–4 OM cells were already sensitive to ATP at DIV1, they were M) used as a positive control. While some not responsive to ET-1 before DIV6 and the largest differ-ences between these two agonists were observed at DIV5

(significant differences between the percentage of cells re-Figure 3A and B). Two-way ANOVA analysis revealed sponding according to the age of the culture (P F(7,43) = 45.74, and ET-1 were also significantly different (< 0.001). The percentage of cells responding between ATP < 0.001) especially at DIV5 and DIV6 (Bonferroni multiple F(7,43) = 11.46, P comparison of responding cells clearly varied according to the culture post-hoc tests, P < 0.01). While the percentage ages, no obvious effect of treatment was observed for the intensity of the calcium rise in the sensitive cells following ATP or ET-1 stimulation (um response intensity increased with culture age which is Figure 3C). Globally, the calci-consistent with the increasing degree of differentiation and maturity of the OM primary culture cells.

Treatment of the OMimpact on cellular proliferation

in vivo with ET receptor antagonists To evaluate the potential proliferative role of ET-1 we next treated 3-day-old rat pups with intranasal instilla-in vivo, tions of BQs at 10–5proliferation level by immunohistochemistry against PCNA, M during 1 week. We then evaluate the a marker of dividing cells (Ohta and Ichimura, 2000). We focused on two different zones of the olfactory epithelium based on a recent study showing that nasal instillation was differentially effective according to the OM localization (Hasegawa-Ishii et al., 2017). This epithelium can be divid-ed in three sub-areas with the lowest one containing basal cells and young neurons (Moon et al., 2009). As expected, we observed a staining mainly above the lamina propria (in the olfactory epithelium was not significantly modified Figure 4A). However, the percentage of PCNA stained area by a treatment with a saline solution or the mixture of ET receptors antagonists in both zones evaluated (Mann-Whitney Figure 4B; 2, respectively). Wondering if our treatment could impact U test; P = 0.317 and 0.903 for zones 1 and the local production of ET-1, we measured by RT-qPCR the expression of big ET-1 and ET receptors in these conditions. Big ET-1 mRNA was indeed increased after the BQs treat-ment (sion of ET receptors was not significantly altered despite a Figure 4C; Mann-Whitney U test; P = 0.004). Expres-slight tendency for a decrease in the expression of ETB in BQ-treated animals (Mann-Whitney 0.072 for ETU test; P = 0.522 and cate that the ET system in the OM adapted unexpectedly fast A and ETB, respectively). This result would indi-to the BQs treatment, so we next examined the impact of a 24-hour treatment with BQs on 9-day-old rat pups, which were euthanized at the same age as the previous group, i.e., at 10-day-old (ration, we observed that the proliferation rate was decreased Figure 5A). With this shorter treatment du-by the BQs treatment in the OM in both zones evaluated (zones 1 and 2, respectively).

Figure 5B; Mann-Whitney U test; P = 0.004 and 0.017 for Discussion

Since the first description of ET-1 as a potent vasoconstric-tive peptide (Yanagisawa et al., 1988), multiple roles have been identified for this pleiotropic peptide (Davenport et al.,

355

Bryche B, Saint-Albin A, Le Poupon Schlegel C, Baly C, Congar P, Meunier N (2024) Endothelin increases the proliferation of rat olfactory mucosa cells. Neural Regen Res 15(2):352-360. doi:10.4103/1673-5374.265558

AB

C

**

D

**

E

Figure 1 Endothelin-1 (ET-1) is involved in olfactory mucosa (OM) primary culture cell growth.

(A) Experimental design. The OM primary culture was grown for 6 days in culture me-dium supplemented on days 0, 2 and 4 with either control medium (CTL), endothelin receptor agonist (ET-1) or antagonists (BQs, mixture of BQ123 and BQ788). (B) Dot plot rep-resenting the evolution of the confluence of the primary culture in the control condition (n = 3). One-way analysis of variance followed by Bonferroni multiple comparison post-hoc tests (different letters indicate statistically significant differences). (C, D) Dot plot rep-resenting the relative cellular density of the primary culture for various treatments. It was estimated by measuring the cellular lactate dehydrogenase (LDH) content and expressed as a relative value to the control condition. (E) Dot plot representing the level of extracellular LDH reflecting cellular death level. Data are presented as the mean ± SEM (n = 8, from two independent cultures for the relative cellular density measurement and n = 4/6/6 for the relative extracellular LDH level measurement). Mann-Whitney U test (**P < 0.01).

AB

******

***

**ns

*

C

*

ns

ns

**

*

ns

Figure 2 Endothelin-1 (ET-1) is produced by olfactory mucosa primary culture cells until confluence.

(A) Phase contrast views of ol-factory mucosa primary culture at different ages. The primary culture gradually grows until reaching confluence at day in vitro 7 (DIV7). Scale bar: 50 μm. (B, C) Dot plot representing the evolu-tion of the relative ET-1 concen-tration in the OM primary culture extracellular medium and the mRNA relative level of big ET-1 or ET-1 receptors expressed by the OM primary culture cells. Data are presented as the mean ± SEM (n = 5/4/5 for DIV3, DIV5 and DIV7, respectively, for ET-1 pro-tein level assay; n = 6 for mRNA quantification). One-way analysis of variance (ns: non-significant; *P < 0.05, **P < 0.01, ***P < 0.001).

356