The gut microbiome–Does stool represent right?

Abstract

Many stool-based gut microbiome studies have highlighted the importance of the microbiome. However, we hypothesized that stool is a poor proxy for the inner-colonic microbiome and that studying stool samples may be inadequate to capture the true inner-colonic microbiome. To test this hypothesis, we conducted prospective clinical studies with up to 20 patients undergoing an FDA-cleared gravity-fed colonic lavage without oral purgative pre-consumption. The objective of this study was to present the analysis of inner-colonic microbiota obtained non-invasively during the lavage and how these results differ from stool samples. The inner-colonic samples represented the descending, transverse, and ascending colon. All samples were analyzed for 16S rRNA and shotgun metagenomic sequences. The taxonomic, phylogenetic, and biosynthetic gene cluster analyses showed a distinctive biogeographic gradient and revealed differences between the sample types, especially in the proximal colon. The high percentage of unique information found only in the inner-colonic effluent highlights the importance of these samples and likewise the importance of collecting them using a method that can preserve these distinctive signatures. We proposed that these samples are imperative for developing future biomarkers, targeted therapeutics, and personalized medicine.

1. Introduction

Our current understanding of the gut microbiome places it at the center of multiple physiological processes and establishes its relevance to many facets of health and disease. Manipulating the microbiome holds the promise of novel solutions for both physiological and psychological disorders [[1], [2], [3], [4], [5]]. For example, the gut microbiota is a key factor in host immunity, and the majority of known diseases are either caused by, or reflected, in the microbiome's composition [2]. Studies have demonstrated longitudinal gradients in the composition of the microbiota at various locations along the gastroenterological (GI) tract [[6], [7], [8], [9]] as well as spatial changes between the lumen and mucosal wall of the GI tract [6,10]. Unfortunately, research concerning the inner-colonic microbiome relies on samples obtained via invasive biopsies and brushings during colonoscopies, which are often procured after the bowel has suffered the ill effects of an oral purgative. These methods are neither ideal nor scalable for comprehensive sampling of the inner-colonic microbiome. Additionally, stool, used as a proxy for the intestinal microbiota in most studies due to its relative ease of access, does not fully reflect the entire colon's microbiota and therefore provides only a partial view of the colon's microbial diversity [11].

The rapidly growing technological and high throughput capabilities of next-generation sequencing (NGS) have made microbiome studies from various natural environments relatively straightforward and approachable. The two typical sequencing strategies for bacterial microbiome analysis are 16S ribosomal RNA (rRNA) amplification and whole-genome shotgun sequencing (WGS), with each method possessing distinct advantages and disadvantages [[12], [13], [14], [15]]. The 16S rRNA sequencing focuses on amplifying selected variable regions within the bacterial 16S rRNA. While 16S rRNA sequencing is generally limited to identifying bacterial communities, it is technologically easier and cheaper than alternative methods. On the other hand, the more expensive WGS identifies bacteria and other microorganisms down to the strain level and can provide data for identifying the functions and composition of microbiome communities.

Living organisms from all kingdoms of life encode a large and diverse set of enzymatic pathways that produce specialized metabolites, which can be used as natural products in medicine [16]. Often, these metabolites are a product of enzymatic processes from different organisms that inhabit the same microenvironment. The use of WGS enables the detection of genomic features called Biosynthetic Gene Clusters (BGCs), known to produce biologically active secondary metabolites. Computational identification of BGCs in genome sequences of control and targeted populations [17] can allow systematically exploration and prioritization of specific genes, gene networks, and metabolites for their role in clinical outcomes.

Due to the complexity of capturing inner-colonic microbial samples, their uniqueness and phylogenetic diversity are consistently underexplored. The objectives of this study were to test the hypothesis that, in humans, stool is a poor proxy for the inner-colonic microbiome and to present data supporting this hypothesis. To this end, we conducted prospective clinical studies where stool and inner-colonic samples (HygiSample™) were collected from up to 20 patients. The inner-colonic samples were collected non-invasively during an FDA-cleared, prescription-only procedure to purge the colon by high-volume water irrigation (HygiPrep^Ⓡ). During the procedure, the colon effluent was excreted by the patient and collected by the attending clinical team. Due to the nature of the peristaltic movement of the gut, the colon effluent excreted over time likely reflects the different ecological niches of the colon, from the distal (sigmoid and left colon) to the proximal (right) colon moving towards the cecum. The procedures were performed by trained personnel under stringent standard operating procedures (SOPs) when medically indicated, such as before colonoscopy. The method has been shown to be safe, effective, and well tolerated in patients, with an adequacy of 97% in over 12,000 colonoscopy preparations [[18], [19], [20], [21]]. Clinical adequacy remains high even in patients with predictors of poor bowel preparation such as age, male sex, and co-morbidities [18,22,23]. The studies were designed to examine genomic-related characteristics of the collected colon effluent samples and to compare them to the microbiome of standard home-collected stool samples.

2. Methods

2.1. Clinical trial design

Samples were prospectively collected from 11 to 20 healthy participants undergoing a gravity-fed high-volume (>35 L) colonic lavage procedure with induced defecation (HygiPrep^Ⓡ) [[18], [19], [20]] as their bowel preparation for a screening colonoscopy at the Hygieacare center in Norfolk, VA. The studies were done under IRB (WIRB WCG IRB) approvals NCT04264923 and NCT04106232 (control arm) (clinicaltrial.gov). The study complied with all regulations, and informed consent was obtained from all participants. The inclusion and exclusion criteria and the number of study participants are described in the Supplemental Information (SI).

2.2. Sample collection

Study participants collected home stool samples using a standardized DNA Genotek OMNIgene® stool microbiome collection kit (Ontario, Canada). Colon effluent samples were collected continuously during the high-volume colonic lavage at the beginning, middle, and end of the procedure (approximately 1 h) to allow for a representation of the descending (left), transverse, and ascending (right) colon (Fig. 1). As expected from the right colon effluent, the last sample collected was characterized by a diluted mucoid consistency and white-yellowish color. The participants collected the home-collected stool samples approximately 48 h before their high-volume colonic lavage bowel prep. All samples were immediately preserved in DNA Genotek OMNIGene tubes with a preservation buffer. Overall, 20 stool samples and 62 colon effluent samples were collected and sent for 16S rRNA amplicon sequencing at Microbiome Insights Inc. (Vancouver, Canada) and whole genome sequencing (WGS) at SeqMatic (Fremont, CA, USA). More details about sample collection, preservation, and sequencing can be found in the SI.

Fig. 1.

Unique and shared microbial strains in inner-colonic effluent and stool samples, as detected by 16S rRNA sequencing and Amplicon Sequencing Variants (ASV) analysis. A. A Venn diagram representing the number (and percentage) of microbial strains that are unique and shared between stool (red) and the three inner-colonic effluent sample types pooled together (blue). The pie charts represent the microbial strains distribution for each group (unique to stool, unique to effluent, shared) by phylum taxonomic rank. For stool n = 11; n = 35 for inner-colonic effluent samples; and n = 46 for shared microbial strains. B. Number (and percentage) of microbial strains that are unique and shared between stool and the different inner-colonic effluent sample types “Effluent 1–3” representing the different colonic segments. For stool n = 11; Descending (left) colon, “Effluent-1”, n = 13; transverse colon “Effluent-2”, n = 11; and ascending (right) colon, “Effluent-3”, n = 11. . (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

2.3. 16S rRNA sequencing- DNA extraction, sample processing, and taxonomic analysis

Specimens were added to a MoBio PowerMag Soil DNA Isolation Bead Plate. DNA was extracted following MoBio's instructions on a KingFisher robot. Bacterial 16S rRNA genes were PCR-amplified with dual-barcoded primers targeting the V4 region. All extractions, sample processing, and sequencing for 16S rRNA were performed at Microbiome Insights Inc. (Vancouver, Canada). For further details, see S1.

Sequence diversity was estimated using Amplicon Sequencing Variants (ASVs), similar to Single-nucleotide polymorphism (SNP) analysis for functional genes. Analysis was performed using dada2 following the standard procedure for variant inference and phyloseq for downstream diversity analyses [24,25]. Following amplicon variant inference and chimera removal, the mate-paired reads were assigned against the ARB Silva database (v132) release [26, 27]. The presence and absence of ASVs, regardless of their abundance, were analyzed using Venn diagrams. Comparison of the taxonomic diversity within the different sample types was done by alpha diversity using the Shannon diversity index [28]. For the analysis of beta diversity between sample types (a proxy of the biogeographic location of the colon), both weighted and unweighted Unifrac distances were generated [28]. For the purpose of this report, we will refer to the ASVs as microbial strains. More details regarding the ASV analysis can be found in the Supplementary Methods.

2.4. DNA extraction, sample processing, and quality control for whole genome sequencing (WGS)

Metagenomic sequencing was done using NovaSeq 6000. The read outputs were 150 M ± 98.4 M (min-2.7 M, max-396 M) 2 × 150-bp paired-end reads. A modified version of the QIAamp PowerfecalDNA Pro Extraction Kit (Qiagen, USA) protocol was used for DNA extraction. DNA libraries were prepared, and samples were indexed, barcoded, and pooled for high-throughput sequencing. DNA fragment size analysis was performed using the DNA Tapestation (Agilent, country) with a target insert size of 300–500 bp. WGS sequencing libraries were constructed using the Nextera DNA FLEX Library Preparation Kit (Illumina, USA). MiSeq runs were used as quality control (QC) before full high throughput NovaSeq runs in order to confirm adequate sequencing yield and sample representation across the library. All extractions and WGS were performed at SeqMatic LLC. (Fermont, CA, USA). Duplicate reads and human contaminants were filtered out, adapters and low-quality bases were trimmed, and the resulting clean reads were used for taxonomic and metabolic function analyses. For further details, see Supplementary Methods.

2.5. Taxonomy, abundance, and differential expression analysis of whole genome sequencing data

Taxonomic classification of the WGS 150 bp paired-end reads was performed using Kraken2 (v2.0.9) followed by the Bracken (v2.6.0) package for correction and re-estimation of taxonomic abundance counts. Taxonomic analysis was done in R (version 4.0.2) using the phyloseq package (v1.32.0) with vegan (v2.5.6) and DESeq2 (v1.28.1) utilizing taxonomic abundance counts; see Supplementary Methods for additional details.

2.6. Biosynthetic gene cluster analysis

Biosynthetic gene clusters (BGC) were detected using the VBx1 analysis pipeline (Pragma Bio, South San Francisco, CA), which identifies BGCs using two parallel approaches. Following gene assembly, contigs >20 kb underwent ORF, gene, and metabolic pathway annotations. Groups of adjacent genes were flagged as probable BGCs, which serve as the core feature of this analysis. BGC annotations were performed using antiSMASH 5.0 to produce gene-annotated BGC sequences with putative molecular product classes, where applicable. BGCs were then stratified according to their source: ascending colon, transverse colon, and descending colon. Finally, duplicate BGCs between locations were identified based on Jaccard similarity using the PFAM domains annotated within a BGC. If two BGCs shared the exact same PFAM domains, they were deemed duplicates.

3. Results

The results of the clinical trials are represented for up to 20 patients. Healthy patients were routinely prepped for a colorectal screening colonoscopy by a gravity-fed high-volume and defecation-inducing colonic lavage bowel prep (BP). For the cohort of 11 healthy patients on which the 16S rRNA analysis was performed, the average age was 67.5 (min-41, max-78), with a BMI of 28 ± 6 (min–22, max-43), and 72% of the study participants were women. For the cohort of 20 healthy patients on which the WGS was performed, the average age was 65 ± 10 (min–41, max-78), 60% female (n = 12) and 40% male (n = 8), with a BMI of 28 ± 6 (min–19, max-43).

3.1. Microbial composition and diversity of microbial strains in stool and inner colonic samples

Bacterial communities detected in inner colonic samples differed significantly from those detected in stool samples. Analysis of the 16S rRNA Amplicon Sequence Variants (ASVs) and corresponding taxonomy recovered a total of 1099 microbial strains from all samples, of which 53% were shared between inner-colonic and stool samples (Fig. 1A). The stool samples had 176 unique microbial strains (16%), and 31% of the total ASVs were detected only in colon effluent samples (n = 344) (Fig. 1A), with a unique contribution from the different biogeographic locations of the colon (Fig. 1B). Specifically, we detected a decrease in the abundance of microbial strains belonging to the Firmicutes phylum and an increase in microbial strains belonging to the Proteobacteria phylum in inner-colonic samples compared to stool, without a significant change in the number of microbial strains belonging to the Bacteroidetes phylum (Fig. 1A). Microbial strains of additional, less common phyla were differentially found in the inner-colonic samples, accounting for 81 microbial strains in inner-colonic samples versus 44 strains in stool samples. Of these, we detected microbial strains of the phyla Actinobacteria, Fusobacteria, Tenericutes, Verrucomicrobia, Spirochaetes, Epsilonbacteraeota, Acidobacteria, Synergistetes, Cyanobacteria, Lentisphaerae, Planctomycetes, and Euryarchaeota (this list is presented in descending order of the number of microbial strains per phylum). No microbial strains from the three phyla, Acidobacteria, Spirochaetes, and Planctomycetes, were represented in stool samples.

The biodiversity of each sample type (inner-colonic effluent samples and stool samples) by alpha and beta diversities showed markedly different microbiome communities, both for the 16S rRNA (Fig. 2A, C, and D) and the whole genome sequencing (WGS, Fig. 2B, and E) datasets. For alpha diversity, as measured by the Shannon Index, there was a clear biogeographic gradient of bacterial diversity proceeding from stool and the right colon, where the average alpha diversity decreased towards the small intestine. This was observed for the 16S rRNA and WGS datasets, both when effluent samples were combined or separated into colon segments (Fig. 2A and B). The most pronounced difference was observed between stool and right colon (Effluent-3) samples. For the beta diversity, describing the difference in phylogenetic microbial strain diversity between samples, we found high variability, both for the pooled and segmented inner-colonic effluent samples, while the stool microbiome was relatively similar between the samples (Fig. 2C–E). When examining our 16S rRNA dataset, the first two axes of the weighted Unifrac Principal Coordinates Analysis (PCoA) analysis, which considers the abundance of microbial strains, explain half of the system variance. When using WGS to calculate the PCoA with the Bray-Curtis distance matrix, the taxonomic abundance input captured almost 60% of the variation of the microbial community with stool samples clustered further from the proximal inner-colonic samples (Fig. 2E). PERMANOVA analysis demonstrated significant differences in the microbial community structure between the inner colonic samples (Effluent-1, Effluent-2, Effluent-3) and stool samples (R2 = 0.21804, F = 8.6717, P = 0.001). Furthermore, the beta diversity, especially as reflected in the WGS analysis, shows that the inner-colonic samples can be roughly divided into two groups, one of the distal (left) colon (Effluent-1) and another that includes the more proximal microbiome communities, namely the transverse colon and the right colon (Effluent-2 and -3) (Fig. 2).

Fig. 2.

Comparison of the microbial community diversity of inner-colonic effluent samples collected during high-volume colon irrigation and home-collected stool samples. A. Alpha diversity of microbial strains based on 16S rRNA sequencing and ASV analysis. Left: Stool samples (n = 11) compared to combined data from all effluent samples (n = 35). Right: Comparison between stool (n = 11) and the three different effluent sample types, where: descending (left) colon, “Effluent-1”, n = 13; transverse colon, “Effluent-2”, n = 11; and, ascending (right) colon “Effluent-3”, n = 11. B. Alpha diversity of bacterial strains detected in whole genome sequencing. Sample types are: Stool, n = 20; Descending (left) colon, “Effluent-1”, n = 22; transverse colon, “Effluent-2”, n = 20; and ascending (right) colon (“Effluent-3”) n = 20. For both A and B, Results are presented as Shannon index. NS indicates p > 0.05; *p ≤ 0.05, **p ≤ 0.01; ***p ≤ 0.001, and ****p ≤ 0.0001. C and D. Beta diversity based on 16S rRNA sequencing and ASV analysis and represented by PCoA weighted Unifrac analysis. Inner-colonic effluent samples are represented in the different panels both as combined (C, n = 35) and separated into three inner-colonic sample types (D), representing the ascending (left) colon, “Effluent-1”, n = 13; transverse colon, “Effluent-2”, n = 11; and descending (right) colon, “Effluent-3”, n = 11. For stool samples, n = 11. E. Beta diversity based on whole genome sequencing and represented by PCoA with Bray-Curtis distances where: descending (left) colon, “Effluent-1”, n = 22; transverse colon, “Effluent-2” n = 20; and ascending (right) colon, “Effluent-3”, n = 20. For stool samples, n = 20.

3.2. Differential abundance of species in inner colonic effluent samples and stool samples

We performed differential abundance analysis between stool and effluent samples to examine the differential abundance of specific species between the stool and the three groups of colon effluent samples using our WGS dataset. Comparing stool to the left colon (Effluent-1) showed that 22 species were significantly enriched in Effluent-1 compared to stool (red dots), while only five species were significantly more abundant in the stool (blue dots) (Fig. 3, Table S1). This trend of enriched vs. depleted species continued along the colon, and more Effluent-enriched species were detected as inner-colonic sampling proceeded proximally. A comparison between stool and the transverse colon (Effluent-2) revealed 76 species that were significantly more differentially abundant in comparison to stool, while the stool samples had 10 differentially abundant species (Fig. 3, Table S2). The most significant differentially abundant species were found by comparing the right colon (Effluent-3) to stool samples. In this comparison, there were 20 species significantly enriched in stool samples and 96 species differently enriched in the right colon samples, which are the closest to the small intestine (Fig. 3, Table S3).

Fig. 3.

Differential abundance analysis comparing stool samples (n = 20) to the inner-colonic effluent samples, from the distal colon, “Effluent-1” representing the descending (left) colon (n = 22) (panel A), to “Effluent-2” representing the transverse colon (n = 22) (panel B), and the proximal “Effluent-3” representing the ascending (right) colon (n = 20) (panel C). The red dots indicate species that are significantly enriched in Effluent samples and whose log₂fold change was bigger than 2. Blue dots indicate species that are significantly enriched in stool and whose log₂fold change was bigger than 2. The grey dots indicate species whose padj was bigger than 0.05 or the log₂fold change was not bigger than 2. The horizontal red dash line indicates the padj equals to 0.05; The two vertical dashed lines indicate the absolute values of log₂fold change equal to 2. . (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.3. Metagenomic analysis of recovered BGCs

Biosynthetic gene clusters (BGCs) are locally clustered groups of two or more genes that encode a biosynthetic pathway that produces a secondary metabolite. Analysis of BGCs in inner-colonic effluent and stool samples showed significant differences between sample types (Fig. 4). Only 6% of identified BGCs were common to stool and pooled inner-colonic effluent samples (Effluent-1, -2, and -3), 25% were expressed only in stool, and 69% were unique to effluent samples (Fig. 4). When the effluent-specific BGCs were divided according to colon areas, 25% were found in Effluent-1 (left colon), 21% in Effluent-2 (transverse colon), and 11% in Effluent-3 (right colon).

Fig. 4.

Biosynthetic Genes Cluster (BGC) analysis of inner-colonic effluent and stool samples. The Venn Diagram on the left part represents the number and percentage of BGCs found only in stool samples (red, n = 20), the number and percentage of BGCs shared between stool and inner-colonic effluent samples (purple, n = 82), and the number and percentage of identified BGCs found only in the inner-colonic effluent sample (all effluent samples combined, blue, n = 62). To the right, the buddying circles from the blue part of the Venn diagram represent the number and percentage of unique BGCs detected in the different biogeographic parts of the colon, represented by green for descending (left) colon, “Effluent-1”, n = 22; Orange for transverse colon, “Effluent-2”, n = 20; and plum for ascending (right) colon, “Effluent-3”, n = 20. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

4. Discussion

Understanding the microbiome holds promise in developing novel biomarkers and therapeutics to treat and manage various health conditions [2,5]. Currently, microbiome databases are based upon the examination of stool samples or invasively-acquired colon samples obtained during procedures such as colonoscopy. Here, we present data from two prospective clinical studies describing the significant differences between the stool microbiome and inner-colonic microbiome collected during FDA-cleared defecation-inducing, gravity-fed, and high-volume colonic lavage. To explore microbial community differences between inner-colon effluent and stool samples, we examined several microbiome characteristics, including microbial diversity, community differential abundance, and the composition of biosynthetic gene clusters (BGCs).

The composition of the gut microbiota is driven my multiple factors, including motility and the nutrient landscape that characterizes specific metabolic niches and which is dictated by diet and host factors. Other factors affecting the microbiota are the presence of competing or cooperating species and microbial strains and the niches resulting from those interactions. In our study, the taxonomic and phylogenetic differences between inner-colonic effluent and stool samples increased gradually when approaching the proximal colon and the small intestine (Fig. 1, Fig. 2, Fig. 3, Tables S1–S3). The alpha diversity indicates a clear biogeographic gradient, with the overall microbial diversity decreasing towards the ecologically restrictive proximal colon (Fig. 2). The standard deviation of the alpha diversities revealed that different individuals are far more distinct in their inner-colonic microbial community than in their stool samples. This was also reflected in the beta diversity, where both 16S rRNA and WGS analyses showed the inner-colonic samples to be spread more widely across the multivariate space and further from the clustered stool samples, with an apparent phylogenetic gradient from the stool to the right colon (Effluent-3) (Fig. 2).

Our results indicate that the abundant microbial strains in stool samples are phylogenetically similar, but these samples differ in their rare microbial strains composition. This is also demonstrated in the analysis of present and absent microbial strains (Fig. 1), where several bacterial phyla are underrepresented in stool samples compared to inner-colonic samples (e.g., Proteobacteria and more exotic phyla), while Firmicutes, and specifically Clostridiales, are overrepresented in stool samples (Fig. 1). Thus, the abundant microbiota are relatively similar across patients when examining stool samples, while the expression of rare microbial strains is more specific to each individual. Our WGS dataset also identified differentially abundant species between the different inner-colonic effluent samples and stool samples, where the number of significantly-enriched species increased towards the small intestine (Effluent-3/stool comparison, Fig. 3, Table S3). Our taxonomic and phylogenetic analyses suggest that in order to get a complete picture of the colon microbiome, both stool and effluent samples should be analyzed. The BGC analysis (Fig. 4) revealed that 69% of the detected BGCs were unique to the inner-colonic effluent, yet 25% of the BGCs were unique to stool samples. This further substantiated the claim that analyzing both stool and inner-colonic effluents can provide the most complete information on the gut microbiome, from the rectum to the cecum.

The human gut microbiota has been in the spotlight for drug design and metabolism [29,30], and the importance of the gut microbiome for stratifying patients for drug efficacy has been widely described in recent years [[31], [32], [33]]. Examples exist for different medications, from the common painkiller Acetaminophen [31,34] to anti-cancer treatments wherein adjusting treatment based on microbiome composition may modulate drug efficacy [28,35,36]. It is clear that understanding how an individual will respond to a given drug is essential in developing effective treatment plans for unmet clinical needs. A clinical trial's probability of success (POS) is critical for drug development. Traditionally, from Phase I to approval, the success rates of FDA drug approval range from 9% to 11%, and a recent study suggested a 13.8% success rate [37]. Since the cost of developing new pharmaceutical agents can reach $1.8 billion [38], a success rate of ∼10% is overwhelmingly low. The use of biomarkers to select patients becomes more common as it enhances safety and serves as a surrogate clinical endpoint [39], and can lead to considerably more successful trials. This points to the growing role of individualized diagnostics in improving clinical trial success rates and further highlights the importance of the gut as an accessible compartment for the identification of new biomarkers.

In our study, both stool and inner-colonic effluent samples exhibit specific contributions to the microbiome, with more unique and rare information being detected in the inner-colonic effluent (Fig. 1, Fig. 2, Fig. 3, Fig. 4). This suggests that both stool and inner-colonic effluent samples should be analyzed to encapsulate the full spectrum of the individual's gut microbiome. Appropriate testing to accurately define the gut microbiome stands as an important aim of the research community as microbially-directed therapies are developed. Such information will move the field further toward advanced microbiome-based personalized medicine and will provide a variety of benefits, from the bench to the clinic.

Author contribution statement

Orly Levitan; David Johnson: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper.

Lanying Ma; Donato Giovannelli: Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper.

Dawn Burleson: Conceived and designed the experiments; Performed the experiments; Wrote the paper.

Peter McCaffrey; Ayin Vala: Conceived and designed the experiments; Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper.

Funding statement

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability statement

The data that has been used is confidential.

Declaration of interest’s statement

The authors declare no competing interests.

Appendix A. Supplementary data

The following are the Supplementary data to this article: