SKL2001

Role of the β-Catenin/REG Iα Axis in the Proliferation of Sessile Serrated Adenoma/Polyps Associated with Fusobacterium nucleatum

Abstract: Although sessile serrated adenoma/polyps (SSA/Ps) may arise through a pathway differ- ent from the traditional adenoma–carcinoma sequence, details of SSA/P tumorigenesis still remain unclear. Fusobacterium nucleatum (Fn) is frequently detected in colorectal cancer (CRC) tissues and may play a pivotal role in colorectal carcinogenesis. Here, we investigated the relationship between Fn and the β-catenin/REG Iα axis in SSA/Ps and their involvement in the proliferation of these lesions. Fn was detected in SSA/Ps by fluorescence in situ hybridization using a Fn-targeted probe, and expression of β-catenin, REG Iα and Ki67 was examined using immunohistochemistry. Sixteen of 30 SSA/P lesions (53.3%) were positive for Fn. Eighteen SSA/P lesions (60%) showed β-catenin immunoreactivity in the tumor cell nuclei. A significant majority of Fn-positive lesions showed nuclear expression of β-catenin (87.5%) and higher REG Iα scores and Ki67 labeling indices relative to Fn-negative lesions. The SSA/P lesions expressing β-catenin in nuclei had significantly higher REG Iα scores and Ki67 labeling indices than those expressing β-catenin on cytomembranes. The REG Iα score was positively correlated with the Ki67 labeling index in SSA/P lesions. The treatment with Wnt agonist SKL2001 promoted nuclear β-catenin translocation and enhanced REG Ia expression in Caco2 cells. Fn may play a role in the proliferation of SSA/P lesions through promotion of β-catenin nuclear translocation and REG Iα expression.

1.Introduction
Although it is widely accepted that colorectal cancers (CRCs) arise from adenomas [1], the pathway responsible has recently been shown to be far from simple [2]. Among col- orectal adenomas, sessile serrated adenoma/polyps (SSA/Ps) are classified as a subgroup exhibiting a specific saw-toothed colonic crypt morphology and genetic alterations such as BRAF mutation and microsatellite instability [3,4]. Furthermore, SSA/Ps lesions show high- labeling index for Ki67 expression [5], suggesting that those lesions have high ability in cell proliferation. These morphologic and genetic alterations are quite different from those in conventional adenomas, from which CRCs arise through accumulation of APC, KRAS and p53 mutations in multiple steps [1]. Thus, the molecular mechanism whereby CRCs arise and progress from adenomas is not fully clear. In addition to genetic alterations, the role of the gut microbiome in colorectal carcinogenesis has been highlighted; microbiome imbalance (dysbiosis) is closely associated with the development and progression of CRCs through promotion of chronic inflammatory conditions and production of carcinogenic metabolites [6]. Among various candidate pathogenic bacteria, Fusobacterium nucleatum (Fn) has been gathering the most attention, since numerous studies have reported that a higher abundance of Fn is associated with a more advanced stage, a higher risk of recurrence, and shorter survival in patients with CRC [7–9].

However, it is still debatable whether Fn directly plays a role in the pathogenesis of CRC patients although Fn may promote colorectal carcinogenesis by activating β-catenin signaling in vitro experiments [10,11]. The regenerating gene (REG) was first isolated from rat regenerating pancreatic islets [12], and since then, many REG-related genes have been isolated, currently con- stituting a family with multiple members (types I-IV) [13,14]. We and others have reported that REG family proteins play some roles in not only inflammatory [15–18] but also neo- plastic [19–21] diseases of the gastrointestinal (GI) tract. In fact, REG Iα protein, as well as other REG family proteins [22–24], are suggested to exert a cell proliferative and/or anti-apoptotic effect in inflamed and neoplastic lesions in the GI tract [25–27]. In relation to the up-regulation of REG Iα expression in GI neoplasia, we have previously clarified that STAT3-associated cytokines play a pivotal role in REG Iα overexpression in gastric cancer cells [28]. However, the mechanism by which REG Iα overexpression is regulated in neoplastic lesions remains unclear. On the other hand, it is well known that β-catenin signaling is one of the major pathways of carcinogenesis in colorectal tumors, including SSA/Ps [29,30]. In this regard, it is interesting to speculate that β-catenin mutation and aberrant β-catenin expression may be linked to REG Iα overexpression in hepatocellular carcinoma, colorectal tumors, or salivary tumors [31,32]. In the present study, therefore, we focused on the relationship between Fn abundance and the β-catenin/REG Iα axis and investigated its significance for the proliferative ability of SSA/P lesions.

2.Results
Macrophages play key roles in inflammation, wound healing,angiogenesis, and immune responses. Resting macrophagesregulate the maintenance of tissue integrity but, in responseto inflammatory stimuli, change their gene expression profileto produce inflammatory or anti-inflammatory cytokines,growth, and angiogenic factors. The rapid and profoundchanges in the expression of genes in macrophages are mediated at several levels, including transcriptional control, aswell as posttranscriptional regulation of translation andmRNA and protein stability.Classical (“M1”) activation of macrophages is induced byIFN-γ and/or TLR agonists such as LPS (TLR4 agonist) andis characterized by the rapid and transient induction ofinflammatory cytokines such as TNFα and IL-12 andNOS-3 (iNOS). In contrast, alternative activation pathwaysthat induce an anti-inflammatory phenotype have beendescribed. These have generally been termed “M2” macrophages [11, 39]. Activation by IL-4, for example, induces ananti-inflammatory phenotype termed M2a, characterized bylow expression of inflammatory cytokines and elevatedexpression of the anti-inflammatory cytokine IL-10, IL-1Rα,as well as markers such as CD206 (MR), CD163, MHCII,Ym1, FIZZ-1, and arginase-1 [39–41]. We have previouslydescribed an “alternatively activated” macrophage phenotypethat we have termed “M2d” [15, 16, 42]. This phenotype isinduced by TLR2, 4, 7, and 9 agonists in a MyD88-dependent manner, in synergy with agonists of adenosineA2A and A2B receptors [15, 18–20]. M2d macrophagesexpress low levels of inflammatory cytokines and high levelsof IL-10 and the angiogenic growth factor VEGF. Inductionof this phenotype involves transcriptional upregulation ofHIF1α and posttranscriptional suppression of phospholipase-Cβ2 (PLCβ2) [42, 43].To determine the mechanism of PLCβ2 suppression inresponse to M2a activation, we performed a global screeningof miRNAs expressed in response to LPS, to NECA (an ARagonist), and to LPS together with NECA (M2d activationC8 L8 C20 L20Kat6b (196 kDa)NPM (39 kDa)3′UTR of the Kat6b mRNA. The Kat6b 3′UTR was cloneddownstream of the luciferase open reading frame in thepLightswitch-3′UTR reporter plasmid (pKat6bLuc-3′UTR).

Asecond plasmid was prepared with the miR-487b binding sitedeleted (pKat6bLucΔ3′UTR). RAW264.7 macrophages weretransfected with either pKat6bLuc-3′UTR or pKat6bLucΔ3′UTR.Transfected cells were then treated with LPS for 6 hr (100 ng/ml)and analyzed for luciferase expression (n = 3). RAW264.7 cellswere also cotransfected with either pKat6bLuc-3′UTR orpKat6bLucΔ3′UTR together with either a synthetic miR-487bmimic or a nonspecific miR mimic (miR-433). The cells were thentreated with LPS (100 ng/ml for 6 hr and analyzed for luciferaseexpression (n = 3). Data represent the mean ± SE. ∗ indicatessamples with luciferase expression significantly different fromcontrol luciferase expression (p < 0 05).Mediators of Inflammation 7conditions). MiRNAs that are regulated by LPS have beenpublished in prior studies [24, 33, 36, 44–47]. In the currentstudy, miRNAs specifically modulated by LPS with NECAversus LPS alone were identified. As shown in Table 3, a subgroup of miRNAs was either up- or downregulated inresponse to LPS/NECA versus LPS alone. We confirmedthe changes in expression of miR-487b, which was found tobe mildly induced by LPS, but strongly suppressed by LPSwith NECA (Table 3 and Figure 2). Bioinformatic analysisof potential targets of miR-487b using TargetScan andhttp://www.mirdb.org identified a group of genes with putative miR-487b target sites conserved across mammalian species (Table 3). The HAT Kat6b was identified in theseanalyses as one of a small group of genes that are potentialtargets for miR-487b.There has been much interest recently in the role ofepigenetic modulators in the regulation of macrophageactivation pathways [48–53]. In particular, chromatinremodeling induced by targeted epigenetic modificationssuch as histone methylation or demethylation, as wellas acetylation or deacetylation, may lead to gene activation or repression [40, 54]. Histone deacetylases(HDACs) have been shown to play an important rolein macrophage M1 and M2 activation; however, the roleof histone acetyl transferases (HATs) in regulating macrophage activation has received little attention. HATsand HDACs are families of enzymes that modulate chromatin structure, thus affecting inflammatory gene expression[55, 56]. Mice lacking HDAC3 display a polarizationphenotype similar to IL-4 induced alternative activationand are hyperresponsive to IL-4 stimulation, suggesting thatHDAC3 is an epigenomic brake in macrophage M2a activation [53, 57–59]. By extension, this would suggest that HATsmight provide a stimulus to M2 activation, in contrast to theeffects of HDACs. However, the role of HATs in regulatingmacrophage M1/M2 polarization remains to be determined.HAT complexes of the MYST family are named after thefour founding family members, MOZ, Ybf2 (Sas3), Sas3, andTip60 [60, 61]. Other members of this family include Esa1,MOF, MORF, MSL, and HBO1 (Table 2). MYST familyHATs are typically characterized by the presence of zincfingers and chromodomains and are involved in acetylationof lysine residues on histones H2A, H3, and H4 [62–64].As the HAT Kat6b was identified in this study as a potentialtarget of miR-487b, we examined the effects of LPS on theexpression of Kat6b and also of the other members of theKat family of HATs (Kat6a, Kat5, Kat7, and Kat8). Kat6aand Kat6B form stable multisubunit complexes, MOZ andMORF, respectively [65]. The MOZ/MORF complex isresponsible for acetylation of a substantial portion of histoneH3, and possibly of other histones. The HAT activity of theMOZ/MORF complex is required for normal development,including hematopoiesis and skeletogenesis. Mutations ofKat6B have been identified in patients with Say-BarberBiesecker syndrome and with genitopatellar syndrome[66–68]. In a form of acute myeloid leukemia, there is atranslocation of the N-terminal portion of Kat6b in framewith CBP [62]. A translocation resulting in fusion to TAFIIalso leads to acute myeloid leukemia [63]. Disruption ofKat6b also leads to a Noonan syndrome-like phenotypeand hyperactivated MAPK signaling in both humans andmice [69]. Mutant mice deficient in Kat6b are reported todevelop poorly, exhibiting growth retardation, facial dysmorphism, skeletal abnormalities, and developmental brainanomalies, leading to their designation as “Querkopf”(“Strange head”) mice [69]. No studies on the inflammatoryand immunological responses of these mice have beenreported to date.Since Kat6B (MYST4, MORF) was identified as apotential target of miR-487b (Figure 1), we studied theexpression of Kat6b, as well as the other members ofthe MYST family of HATs, in the response of macrophages to LPS (M1) and LPS/NECA (M2d) activation.As shown in Figure 3, LPS induced a rapid, strong, andsustained suppression of Kat6b mRNA expression. Strongsuppression (>80%) was observed by 3 hours followingLPS treatment and sustained through at least 12 hours.After 24 hours, 50% suppression was still apparent. Incontrast, Kat6A (MYST5, MOZ) expression was stimulated by LPS and showed a reciprocal pattern of expression to that of Kat6A. The other members of the MYSTfamily (Kats 5, 7, and 8) were not affected by LPS treatment. Surprisingly, the expression patterns of Kat6A andKat6B in response to LPS/NECA were the same as thosewith LPS alone, despite the fact that miR-487b expressionis strongly suppressed by LPS. We propose in the light ofpublished literature implicating HDACs in M2 activation[53, 57–59] that the strong downregulation of the HATKat6b induced by LPS may play a reciprocal role withHDACs in regulating macrophage polarization. We arecurrently testing this hypothesis.To determine the role of miR-487b in the regulation ofKat6B suppression by LPS in macrophages, we cloned theintact Kat6B 3′UTR and a mutated 3′UTR lacking themiR-487b core binding sequence into a luciferase reporterplasmid. These plasmids were transfected into the macrophage cell line RAW264.7 either alone or together with amiR-487b mimic.

LPS suppressed luciferase expression inthe intact plasmid, and this suppression was only mildlyabrogated by loss of the miR-487b binding site; however,the synthetic miR-487b mimic markedly suppressed luciferase activity from the wild-type vector, while the nonspecific mimic had little effect (Figure 5). The suppressiveeffect of the miR-487b mimic was lost in the mutant vector. Together, these results suggest that while the miR487b site in the Kat6B 3′UTR plays a role in the LPSmediated suppression of Kat6B, other factors in additionto miR-487b must also contribute. As LPS/NECA stronglysuppresses miR-487b expression in comparison to LPS alone,the lack of effect of LPS/NECA on the expression of Kat6Balso clearly suggests that miR-487b is not the sole factorinvolved in the suppression of Kat6B by LPS. In this context,it is of interest to note that miR-487b was recently implicatedas a negative regulator of bone marrow-derived macrophageactivation by targeting IL-33 production. miR-487b suppressed IL-33 production during the differentiation of bonemarrow-derived macrophages by binding to the 3′UTR of8 Mediators of InflammationIL-13 mRNA and suppressing its translation [70]. Nevertheless, the role of miR-487b modulation of macrophage M1/M2polarization remains unclear.

3.Discussion
The mechanism of CRC development has been largely studied by focusing on genetic alterations, but recent investigators have begun to recognize the role of the gut microbiome in this respect [6], similarly to the role of Helicobacter pylori in gastric cancer develop- ment [33]. So far, comprehensive analyses of the gut microbiome have identified several candidate bacteria that may play a role in the development of CRC, and among them, Fn has received special attention [7–9]. In the present study, we were able to detect the presence of Fn in approximately half of the SSA/P lesions we examined, in agreement with a previous report [34]. One limitation in this study was that we were unable to address the mechanism whereby Fn affects the carcinogenesis of CRC. However, since SSA/P lesions are known to show high proliferative ability [5], we investigated the rela- tionship between Fn positivity and proliferative ability in SSA/P lesions and subsequently clarified that Fn-positive lesions had higher proliferative ability than Fn-negative lesions. In addition to proliferative ability, Fn infection appears to accelerate inflammation and DNA damage in colonic epithelial cells, and those accelerations may be regulated by a specific DNA glycosylase [35]. These findings suggest that not only cell proliferation, but also inflammation-associated DNA damage may be a key to understand the effect of Fn infection on the progression of malignant potential in SSA/P lesions.

Of note, β-catenin nuclear translocation is likely to occur in SSA/P lesions at the transition to dysplasia [29,36], being compatible with our obtained data in this study. This finding may suggest that nuclear β-catenin expression may be a marker of high risk of malignant progression in SSA/P lesions. Interestingly, recent studies have clarified that Fn is likely to promote CRC growth through the formation of a FadA-E-cadherin-annexin A1-β-catenin complex to activate the nuclear translocation of β-catenin [10,11]. Further- more, another study has demonstrated that Fn may promote the nuclear translocation of β-catenin via a TLR4/P-PAK1 cascade in colorectal cancers [37]. In this respect, it was noteworthy in this study that Fn-positive SSA/P lesions showed significant nuclear immunoreactivity for β-catenin. Moreover, we found that SSA/P lesions with a nuclear β-catenin immunostaining pattern had a significantly higher Ki67 labeling index than lesions with cytomembrane immunostaining, indicating that SSA/P lesions with nuclear β-catenin expression had higher proliferative ability. As we demonstrated in this study, β-catenin is normally localized on the cytomembranes of non-neoplastic epithelial cells in the colonic mucosa. However, nuclear translocation of β-catenin is often observed in various tumors, and such translocated β-catenin acts as a transcriptional factor to regulate the expression of its target genes [38,39]. Although we cannot explain exactly why SSA/P lesions with a nuclear β-catenin immunostaining pattern have higher proliferative ability, we speculate that Fn-associated β-catenin nuclear translocation may play at least some role in the proliferation of SSA/P lesions. Moreover, it has been known that APC mutations and APC-related abnormalities are less common in SSA/Ps [40,41]. Conversely, this suggests that APC-unrelated β-catenin activation, such as Fn-associated ones, may play a pivotal role in progression of SSA/P lesions.

Vigorous mucin production is one characteristic of SSA/P lesions [41,42]. Of note, REG Iα protein is overexpressed in precancerous metaplasia and adenoma that actively express various mucin phenotypes [19,26,43]. Moreover, REG Iα protein functions as a growth and/or anti-apoptotic factor in gastrointestinal tumors [26,27,43], which prompted us to investigate REG Iα expression in SSA/P lesions. In non-neoplastic crypts, REG Iα is expressed in a few endocrine cells with an ovoid or pyramidal morphology, whereas all of the present SSA/P lesions apparently overexpressed REG Iα protein. Interestingly, the REG Iα expression score was positively correlated with the Ki67 labeling index in SSA/P lesions. This suggests that REG Iα expression is associated with the proliferative ability of these lesions, which would be reasonable in view of the known cell growth effect of REG Iα protein. However, one might be concerned whether REG Iα overexpression is specific in SSA/Ps that predominantly arise in the right-side colon. In this regard, a few studies have reported that a proportion of colonic conventional adenomas also overexpress REG Iα protein [44,45]. Therefore, we may have to investigate REG Iα expression in not only SSA/Ps but also conventional adenomas while analyzing the relationship between REG Iα expression and various clinicopathological features.

Finally, we investigated the relationship between β-catenin and REG Iα protein in SSA/P lesions and clarified for the first time that Fn positivity was positively correlated with not only nuclear β-catenin expression but also the REG Iα expression score. Moreover, it was noteworthy that SSA/P lesions with nuclear β-catenin expression had a significantly higher REG Iα expression score. These finding suggest that nuclear translocation of β- catenin may be linked to enhancement of REG Iα expression in SSA/P lesions. In this context, previous in vitro studies have shown that β-catenin signaling is responsible for REG Iα expression [46], and that β-catenin mutation may be linked to aberrant REG Iα overexpression in hepatocellular carcinoma [31]. Thus, since both β-catenin and REG Iα are certainly involved in the progression of CRCs, the β-catenin/REG Iα axis may play a pivotal role in the growth of SSA/P lesions. One limitation of the present study was that signaling from Fn to the β-catenin/REG Iα axis was not fully investigated. However, we preliminarily showed that nuclear β-catenin translocation by Wnt signaling activation may be closely associated with the enhancement of REG Iα expression in colon cancer cells in vitro. Interestingly, recent studies have suggested that Wnt signaling may be a key pathway for Fn-associated cell growth in colon cancer [11,47,48]. Thus, since REG Iα is able to function as a growth factor [25,26], we would like to speculate that Fn-associated β-catenin-nuclear translocation may play a role in the growth of SSA/P lesions, at least in part, via the growth-promoting effect of REG Iα protein.
In summary, we have demonstrated that nuclear β-catenin expression and REG Iα overexpression are simultaneously evident in Fn-positive SSA/P lesions and that these are commonly correlated with the proliferative ability in such lesions. Although we were
unable to investigate the pathway involved in signaling from Fn to the β-catenin/REG Iα axis, the present findings at least suggest that Fn affects the proliferation of SSA/P lesions accompanied by promotion of β-catenin nuclear translocation and REG Iα expression.

4.Materials and Methods
Thirty patients with SSA/P who underwent endoscopic submucosal dissection at Hyogo College of Medicine Hospital between 2016 and 2019 were enrolled. The SSA/P lesions examined were collected by endoscopic submucosal dissection (n =16), endoscopic mucosal resection (n = 10) or polypectomy (n = 4). The tissue specimens obtained were fixed in 10% buffered formalin and embedded in paraffin, then cut into sections for pathological examination and immunohistochemistry. The characteristics of the patients and their lesions are listed in Table 3. The histological features of dysplastic change were assessed according to the previous descriptions [29,36]. This work was done with the approval of the Ethics Committee of Hyogo College of Medicine.Immunohistochemical staining for REG Iα, Ki67 and β-catenin was performed with an Envision Kit (Dako Agilent Technologies, Tokyo, Japan) as described previously [49], using anti-REG Iα antibody (dilution; 1:2000), anti-Ki67 antibody (Dako Agilent Tech- nologies, dilution; 1:50), and anti-β-catenin antibody (Cell Signaling Technology, Danvers, MA, USA; dilution; 1:500). In brief, the rehydrated sections were treated by microwave heating for 20 min in 1×Dako REAL Target Retrieval Solution (Dako Agilent Technologies) and then preincubated with 0.3% H2O2 in methanol for 20 min at room temperature to quench endogenous peroxidase activity. The sections were then incubated with primaryantibodies for 60 min at room temperature, washed in PBS, and incubated with horseradish peroxidase-conjugated secondary antibody for 30 min. The slides were visualized using 3,3′-diaminobenzidine tetrahydrochloride with 0.05% H2O2 for 3 min and then counter- stained with Mayer’s hematoxylin.For evaluation of immunohistochemical expression, crypts that were well oriented perpendicularly from the bottom to the surface of the colorectal epithelium were selected. The expression of REG Iα was graded as previously described [50] with minor modification. Thus, it was scored according to the percentage of positive cells in a crypt as follows: score 0, a few cells; score 1, <10%; score 2, 10–50%; score 3, >50%. The Ki67 labeling index was expressed as the percentage of positive cells in a crypt.

At least five different visual fields for each SSA/P lesion were observed, and the average REG Iα score and Ki67 labeling index were calculated. β-catenin immunoreactivity was detected in the cytomembranes and nuclei of neoplastic cells in SSA/P lesions. When an SSA/P lesion showed β-catenin immunoreactivity not only in the cytomembranes but also the nucleus, it was classified as the nuclear type, whereas if β-catenin immunoreactivity was evident in the cytomembranes but not in the nucleus, the lesion was classified as the cytomembrane type.The sequence of the Fn-targeted probe, FUS664 (Cy3 labeled), was 5′-CTTGTAGTT CCGC(C/T)TACCTC-3′-(Chromosome Science Labo Inc., Sapporo, Japan). The 16 rRNA- targeted oligonucleotide probe was obtained from probe-Base (http://www.microbial- ecology.net/probebase/) [34,51]. The tissue sections were deparaffinized and then hy- bridized with the Fn-targeted probe at 46 °C for 2 h in accordance with the manufacturer’s protocol. After hybridization, the slides were washed in buffer (40% formamide, 0.9 M NaCl, 0.01% SDS, 20 mM Tris-HCl) for 10 min, then in phosphate-buffered saline, and counterstained with DAPI (Thermo Fisher Scientific, Tokyo, Japan).SSA/P lesions were observed using a fluorescence microscope (DP72; Olympus, Tokyo, Japan; magnification ×200) and fluorescent signals were recorded throughout each lesion (at least five different visual fields). The number of signal dots for Fn was counted in each field and the average was calculated. When the average Fn signal count was greater than 3.0/field, the lesion was considered to be positive.Human intestinal epithelial cell line Caco2 was obtained from ATCC (Manassas, VA, USA) and cultured in RPMI 1640 medium (Invitrogen, Carlsbad, CA, USA) with 10% fetal bovine serum (Biowest, Nuaillé, France) in a humidified incubator at 37 ◦C with an atmosphere of 5% CO2. The cells were treated with Wnt agonist SKL2001 (40 µM, Selleck, Houston, USA) for 24 h.

Total RNA was isolated from the cells and reverse-transcribed using oligo-dT primer (Applied Biosystems, Branchburg, NJ). Thereafter, real-time RT-PCR was performed as previously reported [18]. In brief, the following sets of primers for human REG Iα and glyceraldehydes-3-phosphate dehydrogenase (GAPDH) were prepared: human REG Iα 5′- CTAGAGGCAACTGGAAAATACATGTCT-3′ (sense), 5′-GTTGGAGAGATGGTCCGGTTT-3′ (antisense), human GAPDH 5′-GAGTCAACGGATTTGGTCGT-3′ (sense), 5′-TTGATTTTGGAGGGATCTCG-3′ (antisense). The intensity of the fluorescent dye was determined, and the expression levels of REG Iα mRNA were normalized to those of GAPDH mRNA.The treated Caco2 cell were also subjected to immunostaining of β-catenin. Details are mentioned in the Figure 4 legend. Briefly, the cells were fixed with methanol, incubatedwith anti-β-catenin antibody (dilution 1:100; Cell Signaling), followed by TRITC-labeled secondary anti-rabbit antibodies (dilution 1:1000; DAKO), and counterstained using An- tifade Mountant with DAPI (Life Technologies, Carlsbad, CA, USA).All values were expressed as the mean ± SE. Significance of differences between two animal groups was analyzed by Mann–Whitney U-test. Correlations between two parameters were assessed by linear regression analysis. Differences were considered to be significant at p < 0.05.