Pcyt2 deficiency causes age-dependant development of nonalcoholic steatohepatitis and insulin resistance that could be attenuated with phosphonoethylamine

The mechanisms of NASH development in the context of age and genetics are not fully elucidated. This study investigates the age-dependent liver defects during NASH development in mice with heterozygous deletion of Pcyt2 (Pcyt2+/−), the rate limiting enzyme in phosphatidylethanolamine (PE) synthesis. Further, the therapeutic potential of the artificial Pcyt2 substrate, phosphonoethylamine (PEA), is examined. Pcyt2+/− were investigated at 2 and 6–8 months (mo) of age and in addition, 6-mo old Pcyt2+/− with developed NASH were supplemented with PEA for 8 weeks and glucose and fatty acid metabolism, insulin signaling, and inflammation were examined. Heterozygous ablation of Pcyt2 causes changes in liver metabolic regulators from young age, prior to the development of liver disease which does not occur until adulthood. Only older Pcyt2+/− experiences perturbed glucose and fatty acid metabolism. Older Pcyt2+/− liver develops NASH characterized by increased glucose production, accumulation of TAG and glycogen, and increased inflammation. Supplementation with PEA reverses Pcyt2+/− steatosis, inflammation, and other aspects of NASH, showing that was directly caused by Pcyt2 deficiency. Pcyt2 deficiency is a novel mechanism of metabolic dysregulation due to reduced membrane ethanolamine phospholipid synthesis, and the artificial Pcyt2 substrate PEA offers therapeutic potential for NASH reversion.


Methods
Animals and treatments. Heterozygous Pcyt2 mice (Pcyt2 +/− ) were generated and genotyped as previously described 20 . All procedures were approved by the University of Guelph's Animal Care Committee and were in accordance with guidelines of the Canadian Council on Animal Care (CCAC). We also followed the ARRIVE guidelines for reporting results. Mice were housed in a temperature-controlled facticity and exposed to a 12 h light/12 h dark cycle beginning with light at 7:00 a.m. Mice were fed a standardized chow diet (Harlan Teklad S-2335) and had free access to water. Mice that were supplemented with PEtn (PEA, Sigma-Aldrich 268674) were done so through free access to water containing 1 mg/mL of PEtn. Dosage of PEtn was calculated based on physiological levels of Etn (10-75 µM) 25 and previously determined water intake 26 . Wild type littermates (Pcyt2 +/+ ) and Pcyt2 +/− mice (n = 6-10) were euthanized under both fasted and fed conditions at 2 and 6-8 mo. Pcyt2 +/− mice that were supplemented with PEtn (Pcyt2 +/− + PEtn) were euthanized under fed conditions at 6-8 mo (n = 6-12 per group). The treatment period lasted 8 weeks. No differences were observed between agematched males and females and thus, both sexes were used for final analysis.
Glucokinase and glucose-6 phosphatase activities. Hepatic glucokinase (Gk) activity was measured as previously described 28 with some modification. Liver samples (100 mg) were homogenized in 50 mM HEPES, 100 mM KCl, 1 mM EDTA, 5 mM MgCl 2 , and 2.5 mM dithioerythritol. Homogenates were briefly centrifuged and incubated for 10 min on ice with 25% PEG. The microsomal fraction/glucose 6-phosphatase activity was then removed by ultracentrifugation (100,000×g; 30 min; 4 °C). Glucose phosphorylating activities were measured by the production of NADPH from NADP + in the presence of glucose 6-phosphate dehydrogenase (G6PDH) and either 100 mM glucose or 0.5 mM glucose, to distinguish hexokinase (Hk) from glucokinase (Gk) activity. The activity obtained at 0.5 mM glucose is considered the HK activity. The subtraction of the activity measured at 100 mM glucose from the activity measured at 0.5 mM glucose is considered the Gk activity.
The glucose-6 phosphatase (G6Pase) assay was based on the hydrolysis of glucose-6-phosphate to Pi by the microsomal fractions isolated above. The microsomal fractions were incubated with 10 mM glucose-6-phosphate at 37 °C, and the reaction was stopped after 20 min with acid molybdate containing 2/9 volume of 10% SDS and 1/9 volume iof 10% ascorbic acid. The mixture was then incubated for 20 min at 45 °C and the Pi-molybdature3e complex absorbance read at 820 nm. Protein concentration was measured using BCA assay (Thermo Fisher Sci).
Glycogen content. The glycogen content was determined as previously described 29 . In brief, livers (50 mg; n = 12 per group) were immersed in 500 μl 30% potassium hydroxide saturated with sodium sulfate and boiled for 20-30 min until homogenous solution was obtained. Glycogen was precipitated with cold 95% ethanol, separated by centrifugation, and dissolved in distilled water. Glycogen content was determined at 490 nm after the addition of 5% phenol and 95% sulfuric acid. Standard curve was generated using pure glycogen (Roche).
Liver histology and immunohistochemistry. Livers (n = 4 per group) were fixed in 10% formalin in PBS at room temperature for 12-16 h and embedded in paraffin until histopathologic examination. Sections were de-waxed in xylene and rehydrated in a series of ethanol washes. Sections of 10 μm were stained with hematoxylin and eosin (H&E) to examine lipid droplets, periodic acid-Schiff reagent (PAS) for glycogen, 0.1% Picrosirius red (Sigma) for collagen, and F4/80 antibody (Abcam) for macrophages and were visualized with light The enrichment analysis tool Enrichr (https:// maaya nlab. cloud/ Enric hr/#) was used for the analysis of the microarray data (GEO microarrays GSE55617) and RT-PCR array data [30][31][32] . The bar charts of the top 10 enriched terms from the selected libraries and clustergrams of the input genes vs. the enriched terms were produced separately for upregulated and down-regulated genes. The Manhattan and Volcano plots that establish the significance of each gene set vs. its odds ratio were visualized with Appyter (https:// appyt ers. maaya nlab. cloud/#/ Enric hment_ Analy sis_ Visua lizer).
Blood biochemistry. For analysis of liver enzymes and albumin (n = 12), blood was collected after 12 h of fasting. Serum was separated immediately through centrifugation and sent to Animal Health Laboratory (University of Guelph) for biochemical analyses.
Statistical analysis. Data was analyzed using two-tailed unpaired t-test, and for differences between more than 2 groups one-way ANOVA with Tukey's post hoc test was performed. Significance was rejected at p ≥ 0.05. Results are represented as mean ± SD. All statistical tests were performed with Graphpad Prism 6 software.

Results
Liver glucose contributes to increased glucose tolerance in older Pcyt2 +/− . Pcyt2 +/− mice are similar weights at 2-mo compared to control littermates but progressively gain more weight than controls as they age despite consuming equal amounts of food 21 . To help elucidate the mechanism by which heterozygous ablation of Pcyt2 affects glucose metabolism by age, we performed glucose (GTT) and pyruvate (PTT) and tolerance tests on 2-mo and 8-mo mice. Pcyt2 +/− fasting glucose levels are unaltered at 2-mo of age, but by 8-mo are 20% higher than age-matched Pcyt2 +/+ littermates (Fig. 1A). Two-mo Pcyt2 +/− maintain normal glucose levels in response to the GTT while 8-mo Pcyt2 +/− are hyperglycemic compared to Pcyt2 +/+ littermates ( Fig. 1B and C). GTT area under the curve (AUC) was elevated by 38% in 8-mo Pcyt2 +/− showing age-dependent and reduced glucose clearance from plasma (Fig. 1C). This adds to our previous findings of elevated insulin levels in response to a glucose challenge in 9-mo but not 2-mo Pcyt2 +/−21 , showing that impaired glucose metabolism is a consequence, not a cause of the Pcyt2 +/− phenotype.
We determined the ability of the Pcyt2 +/− liver to utilize pyruvate for glucose production through an intraperitoneal injection of sodium pyruvate and measurement of the subsequent rise in plasma glucose levels. Glucose production is unchanged in 2-mo Pcyt2 +/− but increased in 8-mo Pcyt2 +/− with a 50% elevation in PTT AUC, relative to Pcyt2 +/+ littermates ( Fig. 1D and E). To reinforce the concept that Pcyt2 deficiency augments liver glucose production we determined glucose release from primary hepatocytes isolated from 8-mo mice. Indeed, primary Pcyt2 +/− hepatocytes exhibit a 64% increased glucose output compared to Pcyt2 +/+ hepatocytes (Fig. 1F). This completements our previous evidence showing increased formation of both DAG and TAG that could be normalized with overexpression of Pcyt2 complementary DNA in Pcyt2 deficient primary hepatocytes 13 , establishing perturbed glucose and lipid homeostasis in Pcyt2 +/− primary hepatocytes is a result of Pcyt2 deficiency.
Further indicators of elevated glucose production are shown in the increased expression of the key liver enzymes in the gluconeogenic pathway. In fasted 8-mo Pcyt2 +/− mice, mRNA levels of Pepck and G6Pase are increased by 2.36-and 2.21-fold, respectively, along with a 2.37-fold increase in G6Pase enzyme activity ( Fig. 1G and H). The expression of glycolytic L-Pk was modestly reduced by 31% but Gk expression was reduced 3.58-fold and Gk activity by 46% showing reduced glucose utilization by glycolysis in old Pcyt2 +/− liver ( Fig. 1I and J). Together these data show that liver glucose production by gluconeogenesis was normal at younger age, however, significantly increased and contributed to the elevated plasma glucose in older Pcyt2 +/− .  . Band intensities were measured using ImageJ. Data are presented as mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. H]bromopalmitate is increased by 49% demonstrating that both glucose and FA are more readily available in fasted Pcyt2 +/− than in fasted Pcyt2 +/+ mice ( Fig. 1K and L). Liver staining displayed altered glycogen storage and quantitative analysis revealed a 75% increase in glycogen content in older Pcyt2 +/− liver ( Fig. 1M and N).
Young Pcyt2 +/− exhibit defects in fatty acid metabolism that remain impaired with aging. Because young Pcyt2 +/− had normal GTT and PTT tests ( Fig. 1B and D) yet the microarray analysis indicated an early impairment in fatty acid metabolism ( Fig. 2A) we next checked the activity of mitochondrial and fatty acid metabolic pathways. Fasted 2-mo Pcyt2 +/− showed reduced levels and activity of the mitochondrial activators Sirt1 and Ampkα, 34% and 49%. Lipogenic ACC, and Srepb1c both in the precursor and active form were increased by 87%, 46% and 84%, respectively. Pkcα and Pkcβ1/2, well known DAG regulated kinases, increased by 52% and 72% while Pkcβ2 underwent a modest increase of 11%. In addition, the important angiogenic factor and inhibitor of lipolysis Angptl4 exhibited a 43% elevation in 2-mo Pcyt2 +/− (Fig. 2D). Importantly, Angpt4 and Angpt2 gene expression and signaling via STAT and FOXO1 pathways were also upregulated in 2-mo Pcyt2 +/− (Fig. 2B, Supplementary Table 3). In the fed state, only p-AMPK and Pkcβ1 were increased by 177% and 144%.
To determine if these defects persist into adulthood, we measured these proteins in fasted 6-mo Pcyt2 +/− (Supplementary Fig. 1). Indeed, Sirt1 is reduced by 31%, total Ampkα and p-Ampkα: Ampkα ratio decreased 56% and 43% and highly increased Srebp1c in both the precursor form (2.95-fold) and the active form (2.40-fold), along with a substantial (6.40-fold) increase in Acc. DAG regulated Pkcα and Pkcβ1/2, are also drastically elevated by  4). Band intensities were measured using ImageJ. Data are presented as mean ± SD. *p < 0.05; **p < 0.01. Together these data show that an early reduction in mitochondria energy production and increased fatty acid synthesis by lipogenesis 34 preceded development of adult Pcyt2 +/− liver steatosis, and became even more impaired with ageing.
Older Pcyt2 +/− liver develops impaired Pi3k/Akt signalling and steatohepatitis.. We next examined the activity of the insulin signaling pathway in 6-mo Pcyt2 +/− liver. Consistent with the microarray data, older Pcyt2 +/− shows severe impairment in the Irs1/Pi3K/Akt pathway in fasted state (Fig. 3C). Insulin receptor IR was not significantly modified, however, total Irs1 protein was reduced by 68% and pTyr-Irs1 was diminished by 60%. Pcyt2 deficiency caused a dramatic 88% reduction in p85-Pi3k and a 45% decrease in Akt1. Pdk mediated activation at pThr 308 -Akt is not affected however mTorc2 meditated pSer 473 -Akt activation was diminished by 56%. Pcyt2 +/− liver did not show impairments in the insulin signaling in fed state where phosphorylated and total content of Irs1, p85PI3K and Akt1/2 were unchanged.
Based on the pathway analysis indicating an upregulation of genes involved in lipogenesis and fatty liver disease and proinflammatory pathways, we examined serum biomarkers of liver disease/dysfunction and the primary histological features of NASH. We have previously shown that 2-mo Pcyt2 +/− have normal plasma TAG levels but by 8-mo Pcyt2 +/− plasma TAG content is elevated due to an age-dependent upregulation of very low density lipoprotein (VLDL) secretion and liver microsomal triglyceride transfer protein activity 21,34 . Hepatic enzymes alkaline phosphatase (ALP), alanine aminotransferase (ALT) and aspartate aminotransferase (AST), are normal at 2-mo, however, 6-mo Pcyt2 +/− exhibit elevations in ALP, ALT and AST by 70%, 72% and 66%, respectively ( Fig. 3D and E). Serum albumin level is unaffected in 2-mo Pcyt2 +/− but is reduced by 50% 6-mo Pcyt2 +/− (Fig. 3F and G).
PEtn modifies phospholipid and fatty acid metabolic genes and cell signaling proteins. We next supplemented mice with the Pcyt2 substrate PEtn at physiological levels 25 through drinking water and sacrificed mice at 8-mo as only older Pcyt2 +/− develop NASH. It was previously determined that supplementation does not influence water intake 26 . Because PEtn was constantly supplemented, we evaluated the molecular effect of PEtn in the fed state. As expected for single-allele deletion, in addition to the reduced mRNA ( Fig. 2A-b), www.nature.com/scientificreports/ Pcyt2 protein was also reduced in Pcyt2 +/− and was unaltered with its PEtn supplementation (Fig. 4A). The Kennedy pathway transporter Ctl1 35 was increased by PEtn at the mRNA level but protein content was not changed. PEtn increased Pss1 by 32% in Pcyt2 +/− , suggesting that PS synthesis from PC readily occurs in Pcyt2 deficiency. PEtn also increased PS decarboxylase (Psd) and Pss2 expression by 52% and 41%, respectively, indicating that PEtn stimulated an increase in the conversions of PE to PS by Pss2 and Psd decarboxylation of PS to PE (Fig. 4B). Because Pss2 uses PE at the level of the ER, this indicates that PEtn stimulation of the CDP-Etn Kennedy pathway was balanced by an increase in both Pss2 and Psd pathways, i.e., increased PE degradation to PS in the ER mitochondria associated membranes (MAM) by Pss2 occurs simultaneously with increased PS degradation to PE by Psd in the mitochondria. Taken together, such specific stimulatory effect of PEtn on Pss1, Pss2 and Psd genes that control the PC-PS-PE cycle showed that PEtn was readily metabolized to PE by the CDP-Etn Kennedy pathway. In addition, PEtn caused small but significant increase (15-17%) in the mRNA expression of the fatty acid and triglyceride metabolic regulators Ppar ∝ , Pparγ and Atgl (Fig. 4C).
The working model for the signaling perturbations in Pcyt2 +/− NASH is illustrated in Fig. 7. It shows how reduced de novo synthesis of the membrane PE phospholipid results in metabolic and genetic adaptations to maintain membrane bilayers and accommodate unused metabolic intermediates, resulting in changes in glucose and FA metabolism and inflammation that contribute to NASH development.

Discussion
Various mouse models of NASH have been reported, however, few recapitulate both the metabolic and histopathological features and often require special diets. For example, the most widely used diet to induce NASH is a choline/methionine deficient diet. However, this diet is criticized because it causes weight loss and does not induce insulin resistance, an important risk factor for NASH 24 . Here, we show that Pcyt2 +/− mice are an ideal translation model for the human disease because they develop NASH over time and within the context of key risk factors for the human condition (obesity and metabolic syndrome). In this study we focus on the mechanisms of age-related development of NASH in Pcyt2 +/− mice. At young age (2-mo) Pcyt2 +/− have no clinical symptoms of NAFLD, However, 2-mo Pcyt2 +/− exhibit early defects in fatty acid metabolism that favour FA synthesis and persist into adulthood, Adult (6-8mo) Pcyt2 +/− exhibit a fasting-specific deficit in Pi3k/Akt signalling with a shift to increased glucose production by gluconeogenesis, and reduced lipolysis and FA oxidation 20 . Together these impairments cause an increase in liver glycogen and lipid content, leading to steatosis and metabolic syndrome. In humans, NASH is diagnosed only by the hepatic histological findings. Adult Pcyt2 +/− exhibit all of the criteria for biopsy proven NASH 36 : steatosis, hepatocyte ballooning degeneration with Mallory bodies, inflammatory infiltration of macrophages, and fibrosis. Hepatic inflammation is the critical factor distinguishing NASH from simple steatosis and is further demonstrated in adult Pcyt2 +/− in the enrichment of proinflammatory pathways and increased mRNA/protein expression of proinflammatory modulators. Adult Pcyt2 +/− exhibit elevated serum ALP, AST and ALT and decreased albumin, which is suggestive of hepatocellular damage and progressive liver functional impairment 37,38 . Therefore, this firmly establishes the adult Pcyt2 +/− liver pathology as NASH.
A major paradox of type 2 diabetes is the selective impairment in the insulin mediated liver processes. The adult Pcyt2 +/− model recapitulates this by exemplifying the insulin resistant condition at older age with concomitantly increased gluconeogenesis and lipogenesis. In fasting, the contribution of glucose to the liver energy production is low, as shown by reduced glucose uptake and expression/activity of the glycolytic genes in older Pcyt2 +/− . Previous metabolic profiling demonstrated increased plasma glucose, reduced plasma glycerol (reduced lipolysis) and increased plasma acyl carnitines (reduced FA oxidation), and an altered amino acid metabolism in older Pcyt2 +/− , with the notable increase in major anaplerotic precursor, glutamine 20,39 . A suppression of glucose and FA utilization limits the available pathways for energy production and thus, forces anaplerosis to replenish TCA cycle intermediates and permit its continued function. Subsequent obligate cataplerosis is linked to glucose and lipid synthesis in the liver 40 and thus, an active contribution of glutamine/amino acids to TCA reconciles the contradiction of simultaneously increased hepatic glucose and FA production in Pcyt2 +/− . This mechanism is corroborated in mice 41 and human subjects with NAFLD 42 , showing the connection between increased anaplerosis and intrahepatic TAG accumulation, gluconeogenesis and insulin resistance.
We established that the mitochondria regulators Sirt1/p-Ampkα are inhibited and the main regulator of FA synthesis Srebp1c is upregulated early in asymptomatic young Pcyt2 +/− , as direct consequence of Pcyt2 gene deletion, not because of liver steatosis or insulin resistance, which develops at adult age. Pcyt2 heterozygosity leads to reduced flux by the Kennedy pathway and an accumulation of unused intermediates (DAG, ATP) that need to be accommodated 21 . An early increase in Srebp1 and FA synthesis is necessary to form TAG from DAG to reduce DAG levels and thus, even at young age, glucose and FA usage for energy are suppressed 21 . The function of conventional, DAG dependant PKCα/β1/2 is tightly linked to phospholipid homeostasis and stimulates ET/ Pcyt2 activity 3,8 and not surprisingly they were also constitutively upregulated in young Pcyt2 +/− . PKCα and PKC β1/2 among many other functions, are well-known inhibitors of insulin signaling 43 . As DAG/TAG accumulate, constitutively increased PKC, which occurs prior to NASH development, could progressively diminish insulin signaling in older Pcyt2 +/− . These findings are in line with previously established reduced FA oxidation and increased DAG in asymptomatic young Pcyt2 +/−21,44 , indicating that these are the main drivers of impaired insulin signaling and adult-onset NASH developemnt even on a regular chow diet. Together the inherent transcriptional and metabolic adaptations to reduced Pcyt2 expression and activity cause a progressive metabolic dysfunction, culminating in obesity, insulin resistance, and hypertriglyceridemia 21,34 .
Pcyt2 +/− liver shows heavily diminished Irs1, p-Irs1, PI3K and Akt1/2 and pSer 473 -Akt activation by Torc2 specifically in the fasted state, suggesting an increased protein degradation which agree with higher anaplerotic demands for amino acids, described above. There is also an increased gluconeogenesis in fasted state, showing a failed blockade of glucose production by insulin and increased Pepck and G6Pase expression and activity, controlled by gluconeogenic transcription factor Foxo1 45 . Indeed, Pcyt2 +/− phosphorylation of Foxo1 is decreased, leading to increased transcriptional activity which is in congruance with observed increase in gluconeogenesis. Supplementation of substrate PEtn is able to ameliorate Pcyt2 +/− NASH. PEtn supplementation increases total mTorc1 and mTorc1 substrate p70S6K levels, indicting a stimulation of protein synthesis, which may help rectify the abnormal amino acid metabolism shown in 6-mo Pcyt2 +/− microarray. PEtn increases the activation of Stat3 and restores the diminished phosphorylation of Foxo1, allowing for the improvement in NASH 46,47 . Notably, DAG-regulated Pkcα is reduced by PEtn supplementation, suggesting an attenuation of DAG accumulation and improvement in lipid metabolism. PEtn supplementation marginally increased Atgl and Pparα and parallels the effects of over expression of ATGL, an important lipase that governs hepatic TAG turnover and PPARα. Overexpression of ATGL increases FA oxidation, preventing hepatic lipid accumulation and increases PPARα activity, whereas knockdown causes steatosis 48,49 . Thus, it is conceivable that PEtn is able to improve Pcyt2 +/− steatosis through stimulating metabolic flux by the Kennedy pathway, restoring membrane phospholipid homeostasis and increasing the utilization of DAG to reestablish energy balance and improve lipid and glucose metabolism.
The beneficial effect of PEtn was also evident on several characteristics of the inflammatory response associated with NASH. Pcyt2 +/− livers exhibit an elevated content inflammatory factors Il-6, Tnfα, Socs3, Traf6, and Nfκb, demonstrating the chronic inflammation characteristic of NASH. PEtn significantly attenuated several Scientific Reports | (2022) 12:1048 | https://doi.org/10.1038/s41598-022-05140-y www.nature.com/scientificreports/ proinflammatory cytokines and hepatic fibrosis but was not able to reduce the nuclear content of Nfκb. Traf6 activates Nfκb through PI3k signaling 50 and Pcyt2 +/− Pi3k was not modified by PEtn, leading to the consistent activation of NFκb 51 . However, PEtn appears to alleviate oxidative stress as p38 Mapk undergoes dephosphorylation in Pcyt2 +/− that is partially restored with PEtn supplementation. While p38 activity is typically associated with inflammation, dephosphorylation of p38 Mapk occurs in response to increased Pi3k activity and oxidative stress and is shown to be regulated independently from Erk/Jnk 52,53 . Moreover, p-Eif2α was reduced by PEtn, indicating an improvement of Pcyt2 +/− oxidative stress. In mice with NAFLD induced with high fructose diet, phosphorylation of Eif2α is increased to protect hepatocytes from oxidative stress, fibrosis and death 54 . Nuclear content of Stat3 is dramatically increased with PEtn. Stat3 plays a complex role in liver inflammation, having both pro-and anti-inflammatory functions, but importantly STAT3 has been shown protect against hepatocellular damage and attenuate the inflammatory response in models of liver injury 55,56 . An important mechanism in fibrogenesis is the generation of mesenchymal cells through epithelial-tomesenchymal transition, which deposit extracellular matrix once activated 57 . PEtn most likely attenuated this process through stimulation of the Kennedy pathway, known to be a regulator of the reverse process, i.e., the mesenchymal-to-epithelial transition 58 and therefore inhibition of extracellular matrix deposition. The reduction in fibrosis is in line with an improvement in hepatic reduction of Tnfα and Il-6 59 .
In summary, this work emphasizes the importance of membrane phospholipid homeostasis in metabolic disease development and progression. Our work established for the first time that a negative metabolic energy balance, due to reduced membrane lipid synthesis, that results in excessive production of FA to accommodate unused intermediates could lead to all known features of NASH, including steatosis and inflammation. PEtn supplementation was able to reverse Pcyt2 +/− hepatic steatosis and inflammation. These effects indicate the CDP-Etn Kennedy pathway as a target in fatty liver disease and the therapeutic potential of PEtn. Given the increasing global presence of obesity and type 2 diabetes and lack of drug therapy for NASH, identifying new treatment options is critical.

Data and resource availability
The datasets generated during the current study are available from the corresponding author upon request. The resources and suppliers used in this study have been provided above.