Integrative Analysis of miRNA and mRNA Expression Profiles in Mammary Glands of Holstein Cows Artificially Infected with Staphylococcus aureus

Staphylococcus aureus- induced mastitis is one of the most intractable problems for the dairy industry, which causes loss of milk yield and early slaughter of cows worldwide. Few studies have used a comprehensive approach based on the integrative analysis of miRNA and mRNA expression profiles to explore molecular mechanism in bovine mastitis caused by S. aureus. In this study, S. aureus (A1, B1 and C1) and sterile phosphate buffered saline (PBS) (A2, B2 and C2) were introduced to different udder quarters of three individual cows, and transcriptome sequencing and microarrays were utilized to detected miRNA and gene expression in mammary glands from the challenged and control groups. A total of 77 differentially expressed microRNAs (DE miRNAs) and 1625 differentially expressed genes (DEGs) were identified. Gene Ontology (GO) annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis showed that multiple DEGs were enriched in significant terms and pathways associated with immunity and inflammation. Integrative analysis between DE miRNAs and DEGs proved that miR-664b, miR-23b-3p, miR-331-5p, miR-19b and miR-2431-3p were potential factors regulating the expression levels of CD14 Molecule (CD14), G protein subunit gamma 2 (GNG2), interleukin 17A (IL17A), collagen type IV alpha 1 chain (COL4A1), microtubule associated protein RP/EB family member 2 (MAPRE2), member of RAS oncogene family (RAP1B), LDOC1 regulator of NFKB signaling (LDOC1), low-density lipoprotein receptor (LDLR) and S100 calcium binding protein A9 (S100A9) in bovine mastitis caused by S. aureus. These findings could enhance the understanding of the underlying immune response in bovine mammary glands against S. aureus infection and provide a useful foundation for future application of the miRNA–mRNA-based genetic regulatory network in the breeding cows resistant to S. aureus.


Introduction
Bovine mastitis compromises the health and welfare of dairy cattle, as well as decreases the quality and quantity of milk production, causing huge economic losses in the global dairy industry [1]. Staphylococcus aureus is a major etiological pathogen of bovine mastitis, especially subclinical mastitis, causing a persistent and chronic infection, and antibiotic therapies are largely ineffective [2][3][4]. The infectivity and antibiotic resistance of S. aureus and other causative agents make bovine mastitis more difficult to control, which is also a risk of public health [5][6][7][8][9]. By breeding dairy cattle resistance to udder diseases, the risk of mastitis may be reduced in the dairy cow population [10]. Therefore, the identification of specific genes related to mastitis susceptibility or resistance can provide a new way to control mastitis through genetic selection [11,12].
In recent years, numerous studies have shown that bovine mammary epithelial cells (BMECs) respond to the invasion of bacteria or bacterial products by altering the expression levels of several genes involved in inflammation and immunity in vitro [13][14][15]. However, one limitation of these studies is that the conclusions drawn at cellular levels are not necessarily consistent with those of individuals [16]. Although some transcriptomewide association studies have been carried out on S. aureus-induced mastitis in vivo, these studies always analyzed the expression levels of mRNAs or microRNAs (miRNAs) separately [17][18][19][20][21]. Few studies used a comprehensive approach based on the integrative analysis of miRNA and mRNA expression profiles to improve the understanding of the underlying molecular mechanism of cow mastitis caused by S. aureus.
To investigate various interaction networks and regulatory modes of mRNAs and miRNAs, we constructed a S. aureus-type bovine mastitis model and integrated the analysis of miRNAs and mRNAs between the S. aureus-infected quarters and the control ones. These findings will provide new insights into the mechanism of S. aureus-induced cow mastitis.

The Establishment of Bovine S. aureus-Induced Mastitis Model
Indicators of the three cows were measured and recorded after bacterial infection. At 48 h post inoculation, the dairy cattle suffered from obvious pain and had a drastic reduction (25.8% reduction in average) in milk yield. In addition, the temperature of the cows raised (1.7 • C in average), and their mammary glands and lymph nodes were swollen and hard. At the same time, the alteration of the biophysical properties of milk (grey-white color) was observed. There were significant increases of somatic cell count (SCC) of the milk from inoculated quarters (A1: 1,790,000/mL; B1: 1,920,000/mL; and C1: 2,080,000/mL), while those from the controls remained below 100,000/mL.

Differential Expressed miRNA Identification
A total of 21,293,853 and 18,588,177 raw reads were generated from the control and S. aureus-inoculated groups, respectively, by miRNA sequencing (Table S1). After raw reads were disposed, there were 20,847,000 and 18,504,775 clean reads for length distribution assessment. The assessment results revealed that the 78.76% and 71.79% of clean reads were 20-24 nucleotides in length in the two groups ( Figure S1). Principal component analysis (PCA) showed the miRNAs in the challenged and control groups can be classified into different clusters, respectively, indicating sequencing data is qualified for further analysis (Figure 2A). A total of 77 DE miRNAs, including 30 up-regulated and 47 downregulated miRNAs (p ≤ 0.05 and |log2FC| ≥ 1), were identified in the S. aureus-inoculated group, compared with control group ( Figure 3A).    The up-regulated and down-regulated miRNAs are shown in red and green dots, respectively, while the miRNAs with no significant difference in the two groups are shown in black dots. (B) DEGs in bovine mammary gland between the control group and S. aureus-inoculated group. The up-regulated and down-regulated mRNAs are indicated by red and green dots, respectively, while the mRNAs with no significant difference in the two groups are indicated by black dots.

Differential Expressed mRNA Identification
The values of 2100 RIN and 28S/18S were between 7.5-8.9 and 1.3-2.1, respectively (Table S2), indicating that the RNA quality met the requirement and could be used for marker hybridization.
In this study, the CV values of all samples ranged from 3.389% to 4.821% (Table S3), indicating that the detection results of the microarray are reliable. The up-regulated and down-regulated miRNAs are shown in red and green dots, respectively, while the miRNAs with no significant difference in the two groups are shown in black dots. (B) DEGs in bovine mammary gland between the control group and S. aureus-inoculated group. The up-regulated and down-regulated mRNAs are indicated by red and green dots, respectively, while the mRNAs with no significant difference in the two groups are indicated by black dots.

Differential Expressed mRNA Identification
The values of 2100 RIN and 28S/18S were between 7.5-8.9 and 1.3-2.1, respectively (Table S2), indicating that the RNA quality met the requirement and could be used for marker hybridization.
In this study, the CV values of all samples ranged from 3.389% to 4.821% (Table S3), indicating that the detection results of the microarray are reliable.
The PCA was also performed to evaluate the sample distribution. Two separate clusters were found, representing the S. aureus inoculation and control groups, respectively ( Figure 2B). The transcriptional sequences of the same group were assembled in the same cluster, indicating that the main differences in the mRNA expression profiles occurred between different groups.
A total of 1030 up-regulated genes and 595 down-regulated genes (p ≤ 0.05 and |log 2 FC| ≥ 1) were identified in the S. aureus inoculation group versus control group ( Figure 3B).

Interaction Analysis of the miRNAs and mRNAs
Three up-regulated and ten down-regulated DE miRNAs (p ≤ 0.05 and |log 2 FC| ≥ 2) were selected for the miRNA-mRNA interactive analysis. Among all potential target genes predicted by TargetScan, 143 up-regulated and 63 down-regulated genes identified in this study were employed for the construction of miRNA-mRNA interaction networks ( Figure 4).

Interaction Analysis of the miRNAs and mRNAs
Three up-regulated and ten down-regulated DE miRNAs (p ≤ 0.05 and |log2FC| ≥ 2) were selected for the miRNA-mRNA interactive analysis. Among all potential target genes predicted by TargetScan, 143 up-regulated and 63 down-regulated genes identified in this study were employed for the construction of miRNA-mRNA interaction networks ( Figure 4).

Functional Analysis of Differentially Expressed Genes
The Gene Ontology (GO) annotation based on three categories (biological processes (BP), molecular functions (MF) and cellular component (CC)) was performed to explore biological functions of DEGs regulated by DE miRNAs, in which there were 721 up-regulated and 381 down-regulated genes. The 721 up-regulated genes were significantly enriched in 174 BP terms, 31 MF terms and 25 CC terms. Among them, 68 up-regulated genes

Functional Analysis of Differentially Expressed Genes
The Gene Ontology (GO) annotation based on three categories (biological processes (BP), molecular functions (MF) and cellular component (CC)) was performed to explore biological functions of DEGs regulated by DE miRNAs, in which there were 721 up-regulated and 381 down-regulated genes. The 721 up-regulated genes were significantly enriched in 174 BP terms, 31 MF terms and 25 CC terms. Among them, 68 up-regulated genes of 19 terms were involved in inflammation and immune response ( Table 1). The 381 downregulated genes were significantly enriched in 199 BP terms, 23 MF terms and 37 CC terms. Among them, 21 down-regulated genes of 25 terms were involved in inflammation and immune response. Only the top 10 up-regulated and down-regulated terms in each category are listed in Figure 5. Features of DEGs enriched in the top 9 significant GO terms are shown in Figure 6.   of 19 terms were involved in inflammation and immune response ( Table 1). The 381 downregulated genes were significantly enriched in 199 BP terms, 23 MF terms and 37 CC terms. Among them, 21 down-regulated genes of 25 terms were involved in inflammation and immune response. Only the top 10 up-regulated and down-regulated terms in each category are listed in Figure 5. Features of DEGs enriched in the top 9 significant GO terms are shown in Figure 6.    The 721 up-regulated genes were significantly enriched in 65 KEGG pathways, in which 22 pathways containing 119 up-regulated genes were involved in inflammation and immune response ( Table 2). The 381 down-regulated genes are significantly enriched in 26 KEGG pathways, in which 10 KEGG pathways containing 51 down-regulated genes were involved in inflammation and immune response ( Table 2). The top 30 up-regulated and down-regulated pathways are listed in Figure 7. Features of DEGs enriched in the top 9 significant KEGG terms are shown in Figure 8.

Validation of DE miRNAs and DEGs by qRT-PCR
To verify the accuracy of RNA sequencing and microarray, qRT-PCR was performed to detect the expression levels of miRNA and DEGs. The results showed that the relative expression levels of selected miRNAs and mRNAs identified by qRT-PCR were consistent with RNA sequencing and microarray results, respectively (Tables S4 and S5), indicating a high reliability of the study.

Discussion
To date, more than 150 pathogenic bacteria have been identified in dairy cows with mastitis; among them, Escherichia coli, Streptococcus spp. and S. aureus are most frequently isolated from cows with clinical or subclinical mastitis [9,32]. In this study, the S. aureustype bovine mastitis model was constructed to explore interaction patterns of mRNAs and miRNAs in the S. aureus-infected quarters and the control ones. One quarter of the mammary gland of each cow received the inoculation of S. aureus, and the remaining quarters with the inoculation of PBS served as control group. In this way, the systematic errors could be well minimized when we analyzed and compared the expression levels of mRNAs and miRNAs between inoculated and control groups [33,34]. In total, 77 DE miRNAs and 1625 DEGs were identified in the S. aureus-challenged quarters, compared with the healthy ones ( Figure 9). A previous study showed that miR-664b is a promising candidate involved in response to pathogen infection, which was down-regulated in S. aureus-infected quarters (0.450-fold change, p < 0.001) [35]. Accordingly, CD14 Molecule (CD14), a lipopolysaccharide-binding protein enriched significantly in several inflammation-related terms (cellular response to organic substance/oxygen-containing compound/biotic stimulus/biotic stimulus/molecule of bacterial origin terms), which was identified as a predicted target of miR-664b, was up-regulated in S. aureus-infected quarters (2.151-fold change, p = 0.002) (Table  S6). This result is consistent with previous studies, in which CD14 was measured as an up-regulated trend as an early innate immune response gene in bacterial infections of mammary gland [13,36,37]. This finding potentially supports that miR-664b negatively regulates its target gene, CD14, to mediate inflammation in mammary gland of dairy cattle infected by S. aureus. G protein subunit gamma 2 (GNG2), another target gene of miR-664b, was up-regulated in S. aureus-inoculated quarters (3.246-fold change, p = 0.020), which is significantly en- A previous study showed that miR-664b is a promising candidate involved in response to pathogen infection, which was down-regulated in S. aureus-infected quarters (0.450-fold change, p < 0.001) [35]. Accordingly, CD14 Molecule (CD14), a lipopolysaccharide-binding protein enriched significantly in several inflammation-related terms (cellular response to organic substance/oxygen-containing compound/biotic stimulus/biotic stimulus/molecule of bacterial origin terms), which was identified as a predicted target of miR-664b, was upregulated in S. aureus-infected quarters (2.151-fold change, p = 0.002) (Table S6). This result is consistent with previous studies, in which CD14 was measured as an up-regulated trend as an early innate immune response gene in bacterial infections of mammary gland [13,36,37]. This finding potentially supports that miR-664b negatively regulates its target gene, CD14, to mediate inflammation in mammary gland of dairy cattle infected by S. aureus. G protein subunit gamma 2 (GNG2), another target gene of miR-664b, was up-regulated in S. aureus-inoculated quarters (3.246-fold change, p = 0.020), which is significantly enriched in three significant terms (cellular response to organic substance term, cellular response to oxygen-containing compound term and cellular response to acid chemical term) and four significant pathways (PI3K-Akt signaling pathway, chemokine signaling pathway, Kaposi sarcoma-associated herpesvirus infection pathway and Ras signaling pathway) (Table S6). These terms and pathways are mainly involved in inflammation response. Previous studies mainly focused on functional analysis of GNG2 in human malignant melanoma cells [38][39][40]. However, there is no direct evidence to prove the association between the up-regulation of GNG2 and the infection of S. aureus in mammary glands. The highly expressed GNG2 may also be associated with the down-regulation of miR-23b-3p (0.223-fold change, p < 0.001), which was identified to be associated with various cancers, such as cervical cancer, renal cancer and pancreatic cancer [41][42][43][44]. Other up-regulated DEGs regulated by miR-23b-3p in the S. aureus infection group were collagen type IV alpha 1 chain (COL4A1) (2.272-fold change, p = 0.007), microtubule associated protein RP/EB family member 2 (MAPRE2) (5.500-fold change, p = 0.001) and member of RAS oncogene family (RAP1B) (2.548-fold change, p = 0.008). Although COL4A1, MAPRE2 and RAP1B are respectively enriched in various inflammation-related terms and pathways, to our knowledge, there is no evidence to prove that they have a bearing on bovine mastitis infected by S. aureus.
The down-regulation of miR-664b has a potential association with the extremely significant up-regulation of interleukin 17A (IL17A) (18.584-fold change, p < 0.001) in S. aureusinoculated quarters, which plays a crucial role in the defense of Gram-positive bacterial infection and inflammation development [45][46][47]. IL17A is significantly enriched in the terms of cellular response to organic substance, leukocyte migration and inflammatory response and the pathways of IL-17 signaling and rheumatoid arthritis, which indicated that IL17A potentially acts as a functional gene in the defense of S. aureus infection in bovine mammary glands. Generally known, the expression level of a single gene can be regulated by multiple miRNAs [48]. As shown in this study, miR-331-5p, which targets IL17A, was down-regulated in S. aureus-inoculated quarters (0.273-fold change, p < 0.001). At the same time, LDOC1 regulator of NFKB signaling (LDOC1), the target gene of miR-331-5p, was up-regulated in the infected group (2.114-fold change, p = 0.002). LDOC1 is significantly enriched in cellular response to organic substance term, cellular response to oxygen-containing compound term, cellular response to biotic stimulus term, cellular response to lipopolysaccharide term, response to lipopolysaccharide term, cellular response to molecule of bacterial origin term and response to molecule of bacterial origin term. Previous studies have suggested that LDOC1 regulated the expression of nuclear factor kappa-B (NF-κB), which plays a significant role in cellular inflammatory and immune responses [49]. Additionally, multiple studies have shown that LDOC1 can induce apoptosis [50][51][52]. Thus, it remains to be clarified the role of LDOC1 in S. aureus-induced apoptosis.
The down-regulation of miR-19b (0.397-fold change, p < 0.001) is potentially responsible for the up-regulation of LDOC1 in S. aureus-induced mastitis, which has been identified to be the candidate marker for lung cancer and diabetes [53,54]. The down-regulation of miR-19b is also observed to account for the down-regulation of low-density lipoprotein receptor (LDLR) (2.976-fold change, p = 0.024), which was significantly enriched in cellular response to organic substance term, cellular response to oxygen-containing compound term, cellular response to acid chemical term, inflammatory response term and toxoplasmosis pathway and can develop inflammatory atherosclerosis [55]. S100 calcium binding protein A9 (S100A9) is a kind of pro-inflammatory factor, and the protein from exosomes in follicular fluid causes inflammation by NF-κB pathway activation in polycystic ovary syndrome [56,57]. In this study, the up-regulated S100A9 (10.631-fold change, p = 0.006) and down-regulated predicted target miRNA-2431-3p (0.459-fold change, p = 0.005) were screened in S. aureus-inoculated quarters. S100A9 was enriched in multiple significant inflammatory and immune-related pathways, including positive regulation of hydrolase activity pathway, leukocyte migration pathway, neutrophil chemotaxis pathway and inflammatory response pathway.

Ethics Statement and Animals Selection
All experimental protocols in this study were reviewed and approved by the Institutional Animal Care and Use Committee of Yangzhou University (ZZCX2019-SYXY-056). All methods in this study were carried out in accordance with the Administration of Affairs Concerning Experimental Animals published by the Ministry of Science and Technology of China.
Three apparently half-sib, healthy and mastitis-free Holstein dairy cattle (A, B and C) were chosen from a dairy farm in Yangzhou, China. All the three cows were in the middle lactation term of first parity with a consistent history of milk somatic cell count (SCC) below 100,000/mL. In particular, the employed cows were detected to be in absence of Mycobacterium bovis, Brucella abortus, Anaplasma spp., Babesia spp., Theileria spp., bovine leukemia virus, bovine herpesvirus-1, bovine viral diarrhea virus and bovine respiratory syncytial virus with commercial or in-house molecular diagnostic kits [58][59][60][61]. Then, the experiment was performed after one week in quarantine.

Mastitis Model Construction
For challenge infection study, aliquots from frozen stock cultures (S. aureus, ATCC29213) were plated on sheep blood agar and incubated at 37 • C for 18 h under 10% CO2-enriched conditions. Bacterial suspensions for each pure culture were diluted in sterile phosphate buffered saline (PBS) (Biosharp, Hefei, China) to 1 × 10 7 Colony-Forming Units (CFU)/mL, using a spectrophotometer (Eppendorf, Germany) with a wavelength of 600 nm. For challenged group, one quarter (A1, B1 and C1) of the mammary gland of the three individuals received a dose of 5 × 10 7 CFU of S. aureus, and one of the remaining quarters (A2, B2 and C2) not administered with the S. aureus inoculation served as control group that received 5 mL of sterile PBS [20,62]. The milk yield, SCC (Shanghai DHI Test Center, Shanghai, China) and temperature of cows were recorded before and at 24 h post-inoculation.

Sample Collection and Total RNA Extraction
The mammary tissues (1-2 g per quarter) were collected by sterile surgery from two quarters per dairy cattle at 48 h post-inoculation. Samples from challenged (A1, B1 and C1) and control (A2, B2 and C2) quarters were immediately frozen in liquid nitrogen before RNA extraction or stored in 10% formalin for hematoxylin and eosin (HE) staining.
Total RNA was extracted from 250 mg mammary tissues with mirVanaTM RNA Isolation Kit (Applied Biosystems, Carlsbad, CA, USA) and purified with QIAGEN RNeasy ® Kit (QIAGEN, Dusseldorf, Germany). The RNA quality was assessed using Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, USA) and NanoDrop spectrophotometer (Thermo Fisher, USA). Total RNA samples were stored at −70 • C. A total of 10 µg per RNA sample was sent to a commercial sequencing laboratory (Oebiotech, Shanghai, China) for evaluating the expression levels of miRNA with HiSeq 2000 System (single-end) (Illumina, San Diego, CA, USA) and mRNA with microarray (G2519F-023647, Agilent Technologies, Santa Clara, CA, USA).

Pathological Tests
After 48 h of soaking, the samples were rinsed with water for 12 h and subjected to gradient alcohol dehydration, wax impregnation and embedding. Hematoxylin-eosin (HE) staining was performed for 15 min after dewaxing and adequate washing. The pathological changes were visualized with a microscope (M152, Mshot, Guangzhou, China) at different magnifications.
The miRNA counts were normalized as transcript per million (TPM) with the formula (number of reads per miRNA alignment) / (number of reads from the total sample alignment) × 10 6 [64]. The differentially expressed (DE) miRNAs in each sample were calculated with DEseq R package (1.18.0), with p ≤ 0.05 and fold change ≥2 as the threshold.

mRNA Analysis and Data Process
The 2100 RNA Integrity Number (RIN) and 28S/18S values were detected to evaluate the quality of RNAs. The GeneSpring software (version 12.5, Agilent Technologies, Santa Clara, CA, USA) was utilized to evaluate the coefficient of variation (CV) of each sample.
Feature Extraction software (version 10.7.1.1, Agilent Technologies Santa Clara, CA, USA) was employed to extract and analyze raw data from array images. Briefly, the raw data was normalized with the quantile algorithm, and the resultant flag value of any probe was assigned as "Detected" only if there were no "Compromised" or "Not Detected". DEGs were identified with p ≤ 0.05 and |log 2 FC| ≥ 1 as the threshold.

miRNA-mRNA Interaction Network Construction
With the online software TargetScan (www.targetscan.org, accessed on 6 November 2020), the potential target genes of DE miRNAs with more significant expression levels (p ≤ 0.05 and |log 2 FC| ≥ 2) were predicted and intersected, with the DEGs identified by microarray test (p ≤ 0.05 and |log 2 FC| ≥ 2). Then, the miRNA-mRNA interaction networks were constructed and visualized with the DE miRNAs and screened genes by Cytoscape (v3.7.2) [65].

Functional Analysis of Differentially Expressed Genes
DEGs regulated by DE miRNAs were screened to further understand their biological and metabolic pathways. Gene ontology (GO) annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis were respectively performed with the DAVID 6.8 (https: //david.ncifcrf.gov/, accessed on 6 November 2020) and KOBAS 3.0 (http://kobas.cbi.pku. edu.cn/index.php, accessed on 6 November 2020) using R based on the hypergeometric distribution [65]. Then, the GO terms and KEGG pathways with adjusted p ≤ 0.05 were significantly enriched in DEGs or the miRNA target genes.

Statistical Analysis
Data were analyzed using GraphPad Prism 8 (GraphPad, San Diego, CA, USA) with Student's t-test and presented as mean ± standard deviation (SD). The resulting p-values were adjusted using the Benjamini and Hochberg's approach for controlling the false discovery rate (FDR). Adjusted p < 0.05 indicated a significant difference.

Conclusions
In the present study, we comprehensively analyzed the changes in miRNA and mRNA profiles of the mammary gland of dairy cattle under S. aureus inoculation. Overall, 77 DE miRNAs and 1625 DEGs were identified in the S. aureus-challenged quarters. Among them, the predicted integrated regulatory network was constructed with the miRNAs (miR-664b, miR-23b-3p, miR-331-5p, miR-19b and miR-2431-3p) and the mRNAs (CD14, GNG2, COL4A1, MAPRE2, RAP1B, IL17A, LDOC1, LDLR and S100A9), which were significantly associated with inflammation and immunity. These findings could enhance the understanding of underlying immune response in bovine mammary glands against S. aureus infection and provide a useful foundation for the future application of the miRNA-mRNA-based genetic regulatory network in the breeding of cows resistant to S. aureus.

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/ 10.3390/pathogens10050506/s1. Table S1: Statistics of miRNA sequencing. Table S2: The quality control of mRNAs. Table S3: The variation coefficient of samples used for microarray test. Table S4: Comparison of the expression levels of seven miRNAs detected by transcriptome sequencing and qRT-PCR. Table S5: Comparison of the expression levels of eight mRNAs detected by microarray and qRT-PCR. Table S6: Functional annotations of key DEGs and their potential target miRNAs. Table S7: The primers used for qRT-PCR to validate the small RNA sequencing. Table S8: The primers used for qRT-PCR to validate the microarray test. Figure  Institutional Review Board Statement: All experimental protocols in this study were reviewed and approved by the Institutional Animal Care and Use Committee of Yangzhou University (ZZCX2019-SYXY-056). All methods in this study were carried out according in accordance with the Administration of Affairs Concerning Experimental Animals published by the Ministry of Science and Technology of China.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available in the main text and supplementary material of this article.

Conflicts of Interest:
The authors declare no conflict of interest.