Inositol pyrophosphate profiling reveals regulatory roles of IP6K2-dependent enhanced IP7 metabolism in the enteric nervous system

Inositol pyrophosphates regulate diverse physiological processes; to better understand their functional roles, assessing their tissue-specific distribution is important. Here, we profiled inositol pyrophosphate levels in mammalian organs using an originally designed liquid chromatography–mass spectrometry (LC-MS) protocol and discovered that the gastrointestinal tract (GIT) contained the highest levels of diphosphoinositol pentakisphosphate (IP7) and its precursor inositol hexakisphosphate (IP6). Although their absolute levels in the GIT are diet dependent, elevated IP7 metabolism still exists under dietary regimens devoid of exogenous IP7. Of the major GIT cells, enteric neurons selectively express the IP7-synthesizing enzyme IP6K2. We found that IP6K2-knockout mice exhibited significantly impaired IP7 metabolism in the various organs including the proximal GIT. In addition, our LC-MS analysis displayed that genetic ablation of IP6K2 significantly impaired IP7 metabolism in the gut and duodenal muscularis externa containing myenteric plexus. Whole transcriptome analysis of duodenal muscularis externa further suggested that IP6K2 inhibition significantly altered expression levels of the gene sets associated with mature neurons, neural progenitor/stem cells, and glial cells, as well as of certain genes modulating neuronal differentiation and functioning, implying critical roles of the IP6K2-IP7 axis in developmental and functional regulation of the enteric nervous system. These results collectively reveal an unexpected role of mammalian IP7—a highly active IP6K2-IP7 pathway is conducive to the enteric nervous system.

Inositol pyrophosphates regulate diverse physiological processes; to better understand their functional roles, assessing their tissue-specific distribution is important. Here, we profiled inositol pyrophosphate levels in mammalian organs using an originally designed liquid chromatography-mass spectrometry (LC-MS) protocol and discovered that the gastrointestinal tract (GIT) contained the highest levels of diphosphoinositol pentakisphosphate (IP 7 ) and its precursor inositol hexakisphosphate (IP 6 ). Although their absolute levels in the GIT are diet dependent, elevated IP 7 metabolism still exists under dietary regimens devoid of exogenous IP 7 . Of the major GIT cells, enteric neurons selectively express the IP 7 -synthesizing enzyme IP6K2. We found that IP6K2-knockout mice exhibited significantly impaired IP 7 metabolism in the various organs including the proximal GIT. In addition, our LC-MS analysis displayed that genetic ablation of IP6K2 significantly impaired IP 7 metabolism in the gut and duodenal muscularis externa containing myenteric plexus. Whole transcriptome analysis of duodenal muscularis externa further suggested that IP6K2 inhibition significantly altered expression levels of the gene sets associated with mature neurons, neural progenitor/stem cells, and glial cells, as well as of certain genes modulating neuronal differentiation and functioning, implying critical roles of the IP6K2-IP 7 axis in developmental and functional regulation of the enteric nervous system. These results collectively reveal an unexpected role of mammalian IP 7 -a highly active IP6K2-IP 7 pathway is conducive to the enteric nervous system.
In vivo studies using IP6K1-or IP6K2-knockout mice suggest that PP-IPs contribute to the development and maintenance of neuronal cells (15)(16)(17). In addition to these in vivo mice studies, our as well as other research groups have shown that PP-IPs are pathophysiologically involved in the progression of obesity (18,19), in cancer (20) and in neurodegenerative disorders such as Huntington's disease (21), amyotrophic lateral sclerosis (22), and Alzheimer's disease (23). Therefore, PP-IPs are currently being considered as potential therapeutic targets for several diverse human disorders (24,25). However, we are unaware of any systematic studies that have directly and comprehensively analyzed PP-IP distribution in mammalian tissues, which could provide valuable insights into the effects of pharmacological interventions on the PP-IP system.
Over the past decade, extensive efforts have been made to develop analytical methods for detecting PP-IPs. Traditionally, PP-IPs have been studied using radioisotopic 3 H-inositol labeling coupled with anion exchange chromatography (26), which allows sensitive detection of metabolically labeled PP-IPs from cultured cells. Electrophoretic separation and colorimetric visualization of PP-IPs (27) have also become alternative standard methods for distinguishing PP-IPs. However, PP-IPs in mammalian tissues can neither be radioisotopically labeled nor explicitly detected using colorimetric visualization. A mass spectrometric method coupled with capillary electrophoretic separation (capillary electrophoresismass spectrometry) (28) was recently reported for sensitive analysis of PP-IPs in biological samples at the isomer level. However, the instrument setup involved is complex and requires skillful handling, and this method is therefore rarely available in research institutes.
We recently developed an analytical method that directly detects mammalian-derived IP 7 and its precursor IP 6 using conventional liquid chromatography-tandem mass spectrometry (MS/MS) coupled with hydrophilic interaction liquid chromatography (HILIC) (29), enabling the previously impossible quantitation of PP-IPs in mammalian tissues. In this study, we analyzed PP-IP and their precursor IP 6 levels in mammalian organs using a refined HILIC-MS/MS protocol. We found that IP 7 was present at explicit levels in the mammalian central nervous system (CNS), where IP6Ks are highly expressed. Surprisingly, we also discovered that the highest IP 7 production was observed in the gastrointestinal tract (GIT), even after depletion of dietary derived IP 7 . Of the major GIT cells, enteric neurons selectively expressed IP 7 -synthesizing enzyme IP6K2, which was revealed by assessment of single-cell RNA sequencing (scRNA-seq) data sets and confirmed by immunohistochemical detection. Our HILIC-MS/MS survey using IP6K2-knockout (IP6K2 −/− ) mice exhibited that IP6K2-dependent enhanced IP 7 metabolism exists in the gut and duodenal muscularis externa where the myenteric plexus is located. We further performed whole transcriptome analysis of IP6K2-deficient and wildtype (WT) duodenal muscularis externa to define a physiological role of IP6K2-IP 7 pathway in the enteric nervous system (ENS).

Refinement of HILIC-MS/MS protocol for PP-IP analysis
Before investigating PP-IP metabolism in mammalian tissues, we improved our HILIC-MS/MS analysis protocol for unequivocal detection and more precise quantitation of PP-IPs. Medronic acid was shown to improve the chromatographic detection of phosphorylated compounds (30). Consistently, a form of this solvent additive optimized for HILIC analysis (InfinityLab deactivator additive, Agilent Technologies) significantly improved the chromatographic peak shapes of IP 6 and IP 7 (Fig. S1A) and achieved clear detection of low abundance (10 pmol) of IP 7 (Fig. S1B). We also determined the selected reaction monitoring (SRM) conditions for IP 8 (Table S1) by learning its fragmentation pattern (Fig. S1D) to quantitate IP 8 simultaneously with IP 6 and IP 7 (Fig. S1E). To benchmark this method for the detection of endogenous PP-IPs, we analyzed HCT116 cells treated with NaF, which is known to increase intracellular PP-IPs level (31). Our HILIC-MS/MS analysis detected explicit IP 7 and IP 8 SRM peaks in NaF-treated cells (Fig. S1F). We also observed a dose-dependent reduction in IP 7 level and the IP 7 /IP 6 ratio in HCT116 cells treated with the IP6K inhibitor TNP (Fig. S2). Thus, our refined HILIC-MS/MS protocol achieved robust, sensitive, and reliable detection of endogenous IP 6 , IP 7 , and IP 8 in biological samples.

The mammalian gastrointestinal tract contains high levels of PP-IPs
Using the newly developed HILIC-MS/MS protocol, we investigated the distribution of PP-IPs in experimental model rodents fed with a standard plant-based diet (CE-2; Clea, Japan). Fifteen organs, including the CNS and GIT, were harvested from standard diet-fed C57BL/6J male mice. Surprisingly, HILIC-MS/MS analysis showed that the GIT had the highest levels of IP 6 and IP 7 , even after extensive rinsing of the organs with phosphate-buffered saline (PBS) to wash out the digested contents ( Fig. 1, A and B andTable S2). Importantly, the IP 7 /IP 6 ratio in the GIT was remarkably high, by far the highest in all organs examined ( Fig. 1C and Table S2). A subtle IP 8 SRM peak was detected in stomach and small intestine samples, wherein IP 7 was abundant ( Fig. 1D) but was not detected in other organs. While IP 7 SRM peaks were clearly detected in CNS samples (Fig. 1E), IP 7 levels in the CNS were modest compared with those in the GIT. Moreover, the IP 7 /IP 6 ratio in the spinal cord appeared to be higher than that in the cerebrum (Fig. 1C).
Several reports have shown that IPs (mainly IP 6 , known as phytic acid) are present in a variety of crop seeds (32)(33)(34)(35); moreover, plants also generate PP-IPs, which are crucial for phosphorus-starvation responses (36)(37)(38). Therefore, we assumed that the plant-based CE-2 diet contains IP 6 and PP-IPs, and explicit chromatographic peaks of IP 6 , IP 7 , and IP 8 were observed in CE-2 samples ( Fig. 2A, upper panel). We next investigated their concentrations in purified diets with minimal levels of plant-derived components ( Fig. 2A, middle and lower panels). The two purified diets examined (iVid-neo and 70% casein) contained low amounts of IP 6 and negligible amounts of IP 7 and IP 8 . Quantitative analysis revealed that the levels of all PP-IPs in both purified diets were less than 2% of those in CE-2 (Fig. 2B). The concentrations of IP 6 and IP 7 and the relative level of IP 8 in CE-2 and two purified diets were summarized in Table S3. Thus, the abundances of IP 6 and PP-IPs differ among different mouse diets, and standard mouse diet CE-2 substantially contains IP 6 and PP-IPs, suggesting that dietary IP 6 and IP 7 affect the total levels of these molecules in the GIT of standard diet-fed mice.
Two possibilities were considered to explain the high IP 7 /IP 6 ratios in the stomach and duodenum (Fig. 1C): increased absorption of dietary IP 7 or the presence of endogenous IP 7 in these two GIT organs. To verify the former possibility (selective intestinal absorption of IP 7 ), we analyzed the feces of mice fed on CE-2 and estimated the loss of IP 6 and IP 7 in the digestive system. Similar to those for CE-2 samples, IP 6 and IP 7 SRM peaks were clearly observed in mouse feces samples (Fig. 2C), suggesting that considerable amounts of IP 6 and IP 7 still remain in the feces. Quantitative analysis showed that approximately 50% of IP 6 and IP 7 in ingested food remained in the feces (Fig. 2D). Since the IP 7 /IP 6 ratio remained unchanged between undigested CE-2 and feces ( Fig. 2E), dietary IP 6 and IP 7 might not be disproportionately degraded and absorbed throughout the digestive system. Therefore, we could exclude the possibility that dietary IP 7 is preferentially absorbed in the GIT, further highlighting that the high IP 7 /IP 6 ratios of the stomach and duodenum are probably attributable to cellular IP 7 metabolism.
Enhanced IP 7 metabolism is retained in the proximal GIT of rodents under conditions of depleted dietary IP 6 and PP-IP supply Dietary IP 6 and PP-IPs present in the diet blur the direct detection of mammalian-derived IP 6 and PP-IPs in the GIT of standard diet-fed mice. To more precisely validate the presence of endogenously synthesized IP 6 and PP-IPs in the GIT, we prepared C57BL/6J mice maintained under two different conditions both of which remove dietary PP-IPs, namely, 2 months feeding of purified diet iVid-neo containing negligible amounts of PP-IPs (Fig. 2, A and B and Table S3) and fasting for 48 h, and compared their tissue concentrations of IP 6 and PP-IPs with those from standard CE-2 diet-fed counterparts (Fig. 3A). The GIT of both purified diet-fed and fasted mice showed a reduction in IP 6 and IP 7 levels compared with those of standard diet-fed mice; however, the levels were still close to (in the case of IP 6 ) or far greater than (in the case of IP 7 ) the CNS levels (Fig. 3, B and C).    ( Fig. 3D). Importantly, the stomach and duodenum of these mice showed prominently higher IP 7 /IP 6 ratios than those of their standard diet-fed counterparts (Fig. 3E), implying further enhanced IP 7 metabolism compensated for the overall reduced IP 7 level. On the other hand, both purified diet-fed and fasted mice did not show any changes in the IP 6 and IP 7 levels as well as the IP 7 /IP 6 ratio in the CNS and testis compared with those of mice fed a standard diet. However, as with standard diet-fed mice, both purified diet-fed and fasted mice showed higher IP 7 /IP 6 ratios in the spinal cord than in the cerebrum, implying heterogeneous IP 7 metabolic activity in the rostral and caudal CNS. We also investigated IP 7 levels in the GIT of purified diet (70% casein)-fed Sprague-Dawley rats (Fig. S3A). Analogous to the results observed in the mouse model, both IP 6 and IP 7 levels in the GIT of these rats were drastically reduced compared with those in the standard diet-fed GIT and comparable with those in the CNS (Fig. S3, B and C). In addition, the IP 7 /IP 6 ratio was higher in the stomach and duodenum of purified diet-fed rats compared with that of standard diet-fed rats (Fig. S3D), further demonstrating very active IP 7 metabolism in the mammalian proximal GIT.

Enteric neurons highly express IP6K2 in the mammalian GIT
To investigate the expression levels of the three IP6Ks in each GIT cell type, we used scRNA-seq datasets and compared the expression levels of IP6Ks among GIT cell types. Quantitative analysis using a human embryonic intestinal cell scRNAseq dataset (39) showed that enteric neural cells expressed the highest levels of IP6K2 among different intestinal cells (Fig. 4A, upper panel). In enteric neural cells, IP6K2 was selectively expressed across enteric neuron subsets, such as motor    A, photographs (leftmost) and SRM chromatograms of IP 6 , IP 7 , and IP 8 (right panels) of standard diet (CE-2) and the two different purified diets (iVid-neo and 70% casein). B, relative concentrations of IP 6 , IP 7 , and IP 8 in the standard and purified diets. The values shown represent the mean ± SD of three independent experiments and are expressed relative to those of the standard diet. Asterisks indicate statistical significance (p < 0.05, Student's t test) compared with the standard diet. C, representative SRM chromatograms of IP 6 and IP 7 in the feces of standard diet-fed C57BL/6J mice. D, relative concentrations of IP 6 and IP 7 in the feces of standard diet-fed male C57BL/6J mice. The values shown are expressed relative to those of the standard diet (n = 3). Asterisks indicate statistical significance (p < 0.05, Student's t test) compared with the standard diet. E. the IP 7 /IP 6 ratio in standard diet and feces of standard diet-fed male C57BL/6J mice (n = 3). n.s., not significant (Student's t test); SRM, selected reaction monitoring.
IP6K2-IP 7 axis regulates the enteric nervous system neurons, interneurons, and neuroendocrine cells but not in glial cells (Fig. 4A, lower panel). This analysis was further supported by the IP6K quantitation using both E15.5 ( Fig. S4A) and E18.5 ( Fig. 4B) mouse embryonic ENS scRNAseq datasets (40). As in humans, IP6K2 isoform expression level in mouse enteric neurons was higher than in other neural cells such as neuroblasts, progenitors, glial cells, and Schwann cells. Moreover, the transcriptional analysis-based data were verified using immunohistochemical analyses. IP6K2 colocalized with the neuronal marker HuC/D in the mouse duodenal muscle layer, suggesting IP6K2 was expressed in the myenteric plexus (Fig. 4C). Other than enteric neurons, several cell types, including secretory progenitor cells, also expressed relatively high levels of IP6K2 (Fig. S4B). In addition, mouse enteric epithelial cell scRNA-seq data (41) showed that IP6K2 is expressed in mouse enteroendocrine cells (Fig. S4C). Expression levels of IP6K1 and IP6K3 in entire embryonic intestinal cells were low and negligible, respectively (Fig. 4, A and B). These results suggest that IP6K2 is highly expressed in mammalian enteric neurons.
IP6K2 −/− mice show significant impairment of IP 7 metabolism in the proximal GIT To estimate the importance of IP6K2 in endogenous IP 7 synthesis in the mammalian organs including GIT, we employed a genetically modified mouse in which IP6K2 exon 6, encoding the kinase domain, was specifically deleted (Fig. 5A, left panel) (42). To avoid any contamination of dietary-derived IPs in our analysis, IP6K2-knockout (IP6K2 −/− ) or wildtype (WT) mice raised on the standard CE-2 diet were switched to a purified diet (iVid-neo) for 1 week and then fasted for 48 h before sacrifice (Fig. 5A, right panel). In WT mice, IP6K2 mRNA containing the exon 6 sequence was expressed in the proximal GIT but only marginally compared with the expression in the CNS (Fig. S5A). As expected, the IP6K2 transcript was absent in IP6K2 −/− mouse organs. We  IP6K2-IP 7 axis regulates the enteric nervous system confirmed the loss of IP6K2 expression in IP6K2 −/− mice at the protein level using cerebrum lysate (Fig. S5B), because it has high IP6K2 protein expression and thus was useful for clearly validating the loss of IP6K2 in IP6K2 −/− mice. HILIC-MS/MS analysis showed that IP6K2 −/− mice had significantly lower levels of IP 7 in various organs, including the stomach and duodenum, compared with those in their WT counterparts, while IP 6 levels in each organ were almost the same between IP6K2 −/− and WT mice (Fig. 5, B and C). As previously observed (Figs. 3E and S3D), the IP 7 /IP 6 ratios in the stomach and duodenum of WT mice were much higher than those in the other organs examined (Fig. 5D). These two organs of IP6K2 −/− mice exhibited significant reduction in the IP 7 /IP 6 ratios compared with those of WT mice, suggesting active IP6K2 pathway works in stomach and duodenum. Consistently, the IP 7 SRM peaks for the IP6K2 −/− mouse stomach and duodenum were smaller compared with those of WT mice, while IP 6 levels were unchanged (Fig. S5, C and D). Collectively, these data demonstrate that IP6K2 is required for enhanced IP 7 metabolism in the mammalian proximal GIT.
IP6K2-dependent enhanced IP 7 metabolism exists in the gut and duodenal muscularis externa where the myenteric plexus is located Since IP 7 -synthesizing kinase IP6K2 is selectively expressed in enteric neurons (Fig. 4), we next sought to investigate IP 7 metabolism in the mammalian ENS. We collected the stomach and the consecutive 5-cm segments of duodenum, jejunum, and ileum from standard diet-fed (dietary PP-IPs provided) or fasted (dietary PP-IPs depleted) mice to more precisely survey the PP-IP profile in the proximal GIT. Some of these organs collected were subsequently used to isolate the muscularis externa where the myenteric plexus is located. These total GIT tissues and their muscularis externa were subjected to HILIC-MS/MS analysis to compare their IP 7 metabolism (Fig. 6A). Similar to the results shown in Figure 3, 48 h fasting of mice rendered drastic reduction of IP 6 and IP 7 levels with concomitant increase of the IP 7 /IP 6 ratio in total GIT tissues (Figs. 6B and S6). Although the muscularis externa contained less IP 6 and IP 7 than total GIT tissues, the muscle layer exhibited a higher IP 7 /IP 6 ratio than total GIT tissues, which IP6K2-IP 7 axis regulates the enteric nervous system was less dependent on dietary conditions. The IP 7 /IP 6 ratio of the duodenal muscularis externa was highest among the corresponding muscle layers of the neighboring GITs, implying highly active IP 7 metabolism in the duodenal ENS. To verify the relationship between IP6K2-IP 7 axis and the ENS, we first attempted to visualize the duodenal myenteric plexus of IP6K2 −/− mice by whole mount immunostaining (Fig. S7A). We found that IP6K2 deletion largely affected neither the morphological features nor the neuronal cell density in the duodenal myenteric plexus (Fig. S7B). We next prepared the muscularis externa from the stomach to the ileum of WT and IP6K2 −/− mice, first depleting dietary IP 7 in the GIT by 48 h fasting and performed HILIC-MS/MS analysis to evaluate IP 7 metabolism in the ENS of IP6K2 −/− proximal GITs (Fig. 6C). While IP 6 levels in the muscularis externa were almost equivalent between WT and IP6K2 −/− mice, IP 7 levels and IP 7 / IP 6 ratios were significantly reduced in the gut and duodenal muscularis externa of IP6K2 −/− mice (Figs. 6D and S8). These results suggest that IP6K2 actively produces IP 7 in the gut and duodenal muscularis externa where enteric neurons are concentrated.
The IP6K2-IP 7 axis is crucial for certain neurotranscriptome profiles associated with ENS development and functioning Considering the active IP6K2-IP 7 axis in the ENS, we assumed that alteration of IP 7 metabolism by IP6K2 deletion might affect the neuronal status in the proximal GIT. Thus, we randomly selected two neuronal genes expressed in the GIT as well as the CNS (43), namely, dopamine receptor D5 (Drd5) and cholecystokinin B receptor (Cckbr), and investigated their mRNA levels in both the CNS and GIT by quantitative PCR (qPCR) (Fig. 7A). Compared with those in WT mice, these mRNA levels were explicitly increased from the stomach through the small intestine of IP6K2 −/− mice, especially in the duodenum, but not the colon and CNS. To comprehensively appreciate the role of IP6K2-dependent IP 7 metabolism in neuronal gene expression in the mammalian ENS, we isolated the duodenal muscularis externa from WT and IP6K2 −/− mice and performed whole transcriptome analysis by RNA sequencing (RNA-Seq) (Fig. 7B). Gene set enrichment analysis showed that IP6K2 deletion suppressed certain gene sets associated with neural stem/progenitor cells, oligodendrocyte progenitor cells, and glial cells, concomitantly with the induction of those of mature neurons such as inhibitory, dopaminergic, or GABAergic neurons (Figs. 7C and S9A), implying that inhibition of the IP6K2-IP 7 pathway triggers neurodevelopmental imbalance in the mammalian ENS. The RNA-Seq analysis also exhibited that 107 and 134 of 23,405 genes were more than 1.5-fold increased or decreased in IP6K2 −/− with p-value less than 0.05, respectively (Fig. S9B). Pathway enrichment analysis of these genes showed that transcripts increased more than 1.5-fold in IP6K2 −/− were significantly enriched for proteins involved in neuronal signaling (neuroactive ligand-receptor interaction of Kyoto Encyclopedia of Genes and Genomes Annotation) (Fig. S9C). In these transcripts, we observed that seven genes associated with neuronal function (Nckipsd, and Hrh4) or development (Noto, Tbx1, Tbx18, Pax7, and Mycn) were prominently altered in their transcript levels between WT and IP6K2 −/− (Fig. S9D). To validate our RNA-Seq results, differential expression of these seven neuronal genes were further assessed by qPCR and all of these candidate genes exhibited similar significant or prominent changes in transcript levels as observed in RNA-Seq results (Fig. 7D). qPCR analysis also showed that expression of other neuronal genes, including Drd5 and Cckbr, explicitly increased in IP6K2 −/− duodenal muscularis externa (Fig. 7E). These changes were not observed in the RNA-Seq analysis possibly because they were below the lower detection limit and/or because of quantitation error (44,45  IP6K2-IP 7 axis regulates the enteric nervous system IP6K2-IP 7 axis contributes to certain neurotranscriptome profiles involved in ENS development and functioning.

Discussion
Mammalian PP-IPs have been implicated in obesity and diseases such as cancer and neurodegenerative disorders, and thus, their metabolism is a promising drug target (24,5). For this reason, in vivo PP-IP profiling of mammalian tissues is an important subject of research. However, this objective has been thwarted by various technical difficulties. Recently, we developed an HILIC-MS/MS analysis protocol for the sensitive and specific detection of IP 7 and its precursor IP 6 (29). In this study, we quantified in vivo PP-IP levels in mammalian organs using a refined HILIC-MS/MS protocol and evaluated the contribution of IP6K2 to PP-IP metabolism by analyzing mice lacking this IP 7 -synthesizing kinase.
Surprisingly, we found that standard diet-fed rodents possess far more IP 7 level in the GIT than in the CNS where IP 7 -synthesizing kinases IP6K1 and IP6K2 are most abundantly expressed among the mammalian organs (Fig. 1B).
Since standard diet CE-2 contains abundant IP 6 and PP-IPs (Fig. 2, A and B and Table S3), we deemed that dietary IPs should be taken away to precisely detect endogenous IP 7 in the mammalian GIT. The rodent GIT in the two different conditions depleting dietary IP 7 supply (i.e., feeding purified diets negligibly containing PP-IPs and fasting) still contained more abundant IP 7 than that in the CNS (Figs. 3, B and C and S3, B and C). Furthermore, the IP 7 /IP 6 ratio, an indicator of PP-IPs metabolism, was far higher in the proximal GIT than in the CNS irrespective of dietary condition (Figs. 3E and S3D), implying that endogenous IP 7 synthesis is enhanced in the proximal GIT. In line with this result, IP 7 levels in the stomach and duodenum were significantly diminished in IP6K2 −/− mice under dietary PP-IP-depleted conditions (Fig. 5D). Therefore, our HILIC-MS/MS analysis unexpectedly revealed enhanced IP 7 metabolism in the mammalian GIT.
The GIT consists of several histological layers including the muscularis externa that contains the myenteric plexus, a collection of large neuronal assemblies in the GIT. Our HILIC-MS/MS survey of the proximal GIT clarified that the muscularis externa has a higher IP 7 /IP 6 ratio than whole GIT tissues  IP6K2-IP 7 axis regulates the enteric nervous system and the duodenal muscle layer has a much higher IP 7 /IP 6 ratio than those of neighboring GIT segments (Fig. 6B). Considering the expression of the IP 7 -synthesizing enzyme IP6K2 in the myenteric plexus (Fig. 4) and the significant decrease of IP 7 /IP 6 ratio in IP6K2-deficient gut and duodenal muscle layers (Fig. 6D), these observations lead to the idea that IP6K2 actively synthesizes endogenous IP 7 in the ENS of the proximal GIT. Our results also implied the presence of endogenous IP 7 in other GIT layers because total GIT tissues of dietary IP 7 -depleted (fasted) mice contained a greater amount of IP 7 than their corresponding muscle layers (Fig. S6). Since another major nerve plexus exists in the submucosal layer (i.e., submucosal plexus), the submucosal layer may contain endogenous IP 7 to some extent. This PP-IP might also exist in mucosal epithelium because certain enteroendocrine cells, including tuft cells, express IP6Ks at the relatively high level (Fig. S4, B and C). Tuft cell is one of rare cell types present in intestinal epithelium. Park et al. recently showed that tuft cell development is controlled by inositol polyphosphate multikinase, an enzyme responsible for driving IP metabolic pathway leading to IP 7 synthesis (46). This fact and our data (Fig. S4, B and C) encourage to think that IP6K2 and IP 7 might underlie tuft cell physiology as well. Future study is required for assessing cell type-specific IP 7 metabolism to more precisely characterize IP 7 metabolism in the GIT.
Although IP6K2 was initially cloned as a Pi uptake stimulator from a rabbit duodenum complementary DNA library (47) and was annotated soon after as encoding an IP 7 -synthesizing enzyme (48,49), the role of IP6K2 in the GIT has not been investigated until now. In this study, we demonstrated the presence of an active IP6K2-IP 7  IP6K2-IP 7 axis regulates the enteric nervous system duodenal muscularis externa indicated that genetic ablation of IP6K2 causes certain gene products associated with mature neurons to accumulate concomitantly with the reduction of those of neural progenitor/stem cells and glial cells (Figs. 7C and S9A). Given that the developmental lineage of enteric neurons comprises several differentiation points such as neural crest cell migration, neuron-glia bifurcation, and neural stem/ progenitor cell differentiation into mature enteric neurons (50), inhibition of the IP6K2-IP 7 axis possibly causes developmental imbalances of the ENS at the several differentiation points including the maturation of both enteric neurons and glial cells. This idea is also supported by our findings that IP6K2 inhibition significantly altered the expression levels of several transcription factors regulating neural crest cell differentiation (Fig. 7D) (51)(52)(53)(54)(55)(56). In fact, IP6K2 activity was shown to be required for normal migration and development of neural crest cells in zebrafish (57). Besides, genetic inhibition of IP6K2 in the duodenal muscularis externa significantly or prominently changed mRNA levels of several genes modulating neuronal functions (Fig. 7, D and E). Notably, Nckipsd transcript, one of the transcripts most significantly induced in IP6K2 −/− duodenal muscularis externa, contributes to the formation of neural dendrites (58,59) and intracellular neuronal signaling (60). These pieces of knowledge lead to the hypothesis that the IP6K2-IP 7 axis contributes to development and functions of enteric neurons, even though the axis does not largely affect the entire morphological output of the ENS (Fig. S7). The molecular mechanism whereby the IP6K2-IP 7 axis regulates certain neurotranscriptome profile still remains elusive and will be clarified in future studies. Although developmental and functional ENS defects often result in fatal congenital disorders (61,62), IP6K2 −/− mice do not show such severe phenotypic defects: IP6K2 −/− mice are born at Mendelian ratio and grow normally, similar to WT mice (42). Thus, the IP6K2-IP 7 axis might serve as a fine-tuning factor for the developmental and functional regulation of the ENS, although we could not exclude the possibility that IP6K1, another major IP6K isoform, compensates for the loss of IP6K2. It will be meaningful to see whether ENS-specific inhibition of IP6K2 and/or IP6K1 influences gastrointestinal pathophysiologies and development of CNS diseases. Taken together, our observations provide valuable insights into the field of PP-IP biology and neurogastroenterology. Since dysregulation of IP 7 metabolism links to various human diseases including neurodegenerative diseases, studying IP 7 metabolism in human organs provides essential knowledge from the clinical point of view. The refined HILIC-MS/MS protocol we described in this study is capable of detecting IP 6 and IP 7 not only in rodent organs but also in human postmortem organs dissected after forensic intervention (Fig. S10). Unlike in rodents, IP 7 level and IP 7 /IP 6 ratio in the human proximal GITs (esophagus, greater curvature and lesser curvature of the stomach) were less abundant compared with those in human CNS. This is probably due to the high turnover rate of PP-IPs and the delay in dissecting human postmortem organs. Forensic intervention and subsequent organ dissection take hours after death. The presence of the intestinal flora may also facilitate the decomposition of these molecules in the GITs (63). Thus, care should be taken to assess IP 7 metabolism in human GITs. Although the refined HILIC-MS/MS protocol can detect both IP 7 and IP 8 , this protocol failed to detect endogenous IP 8 in all rodent and human organs examined in this study, even in mouse GIT where IP 7 was explicitly abundant. This fact suggested that mammalian-derived IP 8 is far less abundant than IP 7 and its quantitative evaluation requires sample pooling or a more sensitive analytical protocol such as capillary electrophoresismass spectrometry (28). In any case, we demonstrated that our novel protocol was able to evaluate IP 7 metabolism in human organs. Therefore, we foresee the diagnostic potential of our new analytical technique for analyzing IP 6 and IP 7 levels in clinical biopsy.
In conclusion, we investigated the distribution of PP-IPs in mammalian organs using a refined HILIC-MS/MS protocol and demonstrated that IP6K2-dependent IP 7 metabolism was enhanced in the ENS of the proximal GIT. This finding was corroborated by the observation that impairment of IP6K2-dependent IP 7 metabolism significantly altered certain neurotranscriptome profiles involved in ENS development (Fig. 8). Further studies are needed to dissect the role of IP 7 in neurogastroenterology and processes involving the gut-brain axis. We believe that these findings shed new light on the physiological significance of the mammalian PP-IP pathway as well as the regulatory mechanisms of ENS development, which might contribute to a better understanding of human diseases associated with altered PP-IP metabolism.

Cell culture
HCT116 cells (RIKEN, RCB2979, RRID: CVCL_0291) were cultured in Dulbecco's modified Eagle's medium (Nacalai Tesque) supplemented with 10% fetal bovine serum in 5% CO 2 . Cells prepared at 60% confluence in 10-cm dishes were incubated for 1 h with or without 50 mM sodium fluoride (Sigma-Aldrich). After washing twice with PBS, the cells were lysed in cell lysis buffer (0.01% Triton X-100, 1 mM EDTA, 20 mM Tris-HCl). A small aliquot was set aside for protein quantitation, and the rest was used for purification of IPs.

Mouse organs
All experiments involving animals were performed in accordance with protocols approved by institutional animal care guidelines (Tokai University School of Medicine). C57BL/ 6J mice and Sprague-Dawley rats obtained from Clea Japan (Tokyo, Japan) were maintained on a standard diet (CE-2; Clea Japan) or purified diet (iVid-neo; Oriental Kobo, 70% casein; Clea Japan). Some mice fed a standard diet or purified diet were fasted for 48 h before sacrifice. IP6K2 −/− (RRID: IMSR_JAX:036426) and WT mice maintained on a standard diet were switched to the purified diet for a week and subsequently fasted for 48 h before sacrifice. During fasting, mouse cages were changed to clean ones with new bedding every 24 h to reduce coprophagy. Mice and rats were anesthetized using isoflurane and then sacrificed by whole blood withdrawal from the left atrium. Before dissection of the organs, the animals were perfused transcardially with ice-cold PBS to wash out the residual blood and prevent the detection of IPs derived from blood cells. GIT organs (stomach, duodenum, small intestine, and colon) were cut open to remove feces and then extensively rinsed with PBS to wash out any dietary residuals. The duodenum, small intestine, and colon were harvested by cutting a 5-cm (mouse) or 10-cm (rat) segment from the distal end of stomach, between 10 and 15 cm (mouse) or 20 and 30 cm (rat) away from the duodenum and from the anus, respectively. The harvested organs were frozen until further use.

Isolation of muscularis externa from mouse GITs
The muscularis externa containing myenteric plexuses was prepared from mouse GITs as previously described with some modification (65,66). For HILIC-MS/MS analysis, mouse GIT segments were cut open along the attachment line of the mesentery and then placed onto a cold surface with the muscularis externa facing up. The muscularis externa of the GIT segments was isolated by gently scraping the outer layer with watchmaker tweezers under a binocular stereomicroscope. For whole-mount immunostaining, the mouse duodenum was cut open along the mesentery line, pinned onto a rubber plate, and then fixed with 4% paraformaldehyde overnight at 4 C. The muscularis layer was then gently separated from the GIT segment using watchmaker tweezers and a cotton swab under a binocular stereomicroscope. For RNA extraction, mouse GITs were immersed in saturated ammonium sulfate solution containing 20 mM EDTA and 25 mM sodium citrate (pH5.2) to inhibit RNA degradation. The muscularis externa of the segments was placed over a glass rod and then peeled away using a cotton swab along the attachment line of the mesentery under a binocular stereomicroscope as described (67). Isolated muscularis externa was stored in saturated ammonium sulfate solution and then frozen until further use.   Figure 8. Graphical summary of this study. Among the major mammalian organs, stomach and duodenum possess the highest level of IP 7 /IP 6 ratio, suggesting enhanced IP 7 metabolism in the proximal gastrointestinal tract. This active metabolism retains in myenteric plexus wherein IP 7 -synthesizing enzyme IP6K2 is highly expressed. Ablation of IP6K2 affects the configuration of neural cell types (i.e., inhibition of neural stem/progenitor cells and glial cells with concomitant induction of mature neurons) as well as expression levels of certain genes regulating enteric neuron development, demonstrating regulatory roles of active IP6K2-IP 7 axis in the development of the enteric nervous system.

Human postmortem organs
The human study was approved by the Ethics Committee of Tokai University (institutional review board number: 20I-02), and the study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki (68). Written informed consent, allowing the experimental use of the organ samples, was obtained from the relatives of all subjects. Human postmortem organs were obtained at autopsies from three donated bodies (two men and one woman; mean age, 62.3 ± 22.8 years; average body mass index, 26.4 ± 6.6). Some anomalies, such as cardiac hypertrophy, were observed in their bodies by a forensic pathologist. To minimize organ decomposition, organ sampling was confined to cases where the death date and ambient temperature were explicit and the accumulated degree-days (environmental temperature [ C] × postmortem interval [day]) value-an index for evaluating the quality of forensic samples (69)-of all three bodies were very low (close to or less than 20). The harvested organ samples (approximately 400 mg) were frozen until further use.

Gel electrophoresis of synthetic PP-IPs
The synthetic PP-IPs were validated using polyacrylamide gel electrophoresis as described (27). Briefly, synthetic PP-IPs samples mixed with orange G and bromophenol blue loading buffer were applied onto 35% polyacrylamide/Tris-borate-EDTA gel. The samples were electrophoresed overnight at 4 C at 600 V and 6 mA until the orange G and bromophenol blue had run through two-thirds of the gel. Gels were stained with toluidine blue and scanned using a computer scanner.

Purification of IPs
IPs in biological samples were purified as described (70), with some modification. Frozen organs, diets, and feces samples were homogenized using a Shake Master Neo (Bio Medical Science) in 500 μl of ultrapure water. Feces samples were air dried overnight before homogenization for accurate comparison of IP 6 and IP 7 concentrations with those in the diet. Crude lysate was mixed with an equal volume of 2 M perchloric acid, incubated on ice for 30 min, and centrifuged to remove tissue debris. After spiking with 3 nmol of ITPP-d 6 as an internal control, 5 mg of titanium dioxide beads (GL Sciences, Cat# 5020-75000) were added to each sample. The beads were incubated at 4 C for 30 min and washed twice with 1 M perchloric acid, and then 200 μl of 10% ammonium hydroxide was added for IP elution. The elution step was repeated to maximize recovery. The total eluate was dried using a SpeedVac concentrator (Thermo Fisher Scientific) and reconstituted in 125 μl of 100 mM ammonium carbonate/40% acetonitrile buffer, 50 μl of which was used for LC-MS.

HILIC-MS/MS analysis for PP-IPs
Chromatographic experiments were performed using a Nexera UHPLC instrument (Shimadzu). HILIC-based chromatographic separation of IP 6 , IP 7 , IP 8 , and the internal control ITPP-d 6 was achieved using a modified version of a previously described procedure (29). The mobile phase was composed of 300 mM ammonium bicarbonate buffer (pH 10.5) containing 0.1% InfinityLab deactivator additive (Agilent Technologies) as the aqueous mobile phase (eluent A), and 90% acetonitrile containing 10 mM ammonium bicarbonate buffer (pH 10.5) and 0.1% InfinityLab deactivator additive as the organic mobile phase (eluent B). Eluent B included more than 10% aqueous solvent to prevent polymeric aggregation of the major constituent (medronic acid) in the additive. In the entire LC system, chromatographic stainless steel tube was treated with 0.5% phosphoric acid in 90% acetonitrile overnight before analysis to block undesirable adsorption of analytes on the surface of the inner wall of the tube, while paying attention not to run the solvent into the mass spectrometer. The total flow rate of the mobile phase was 0.4 ml/min. Linear gradient separation was achieved as follows: 0 to 2 min, 75% B; 2 to 12 min, 75%-2% B; 12 to 15 min, 2% B.

RNA extraction and quantitative PCR analysis
GIT segments and their muscularis externa were carefully collected and subjected to RNA extraction, as described (71). Total RNA was extracted using TRIzol reagent (Invitrogen, Cat# 15596026). RNA concentration and quality were determined using a NanoDrop 8000 spectrophotometer (Thermo Fisher Scientific) and the 4150 TapeStation system (Agilent Technologies), respectively. Complementary DNA was generated using the High-Capacity Reverse Transcription Kit (Applied Biosystems). qPCR was performed using the KAPA SYBR Fast qPCR kit (Kapa Biosystems, Cat# KK4602) and a StepOne Plus Real-Time PCR system (Applied Biosystems). The primer sequences used in this study are listed in Table S4.

Computational analysis of scRNA-seq datasets
Publicly available human embryonic intestine scRNA-seq processed data (39) and mouse embryonic ENS matrix data (40) were downloaded from the Human Fetal Gut Atlas (https://simmonslab.shinyapps.io/FetalAtlasDataPortal/) and the GEO database (identifier: GSE149524), respectively. Mouse embryonic intestinal epithelial cell data (41) were obtained from the GEO database (identifier: GSE92332). The above datasets were analyzed using the R package Seurat version 4.0.0 (RRID: SCR_007322) (78) to perform dimensionality reduction by uniform manifold approximation and projection and/or generate dot plots showing the relative expression of IP6Ks across different clusters.

Statistical analysis
Data are expressed as the mean ± SD. Differences between two or more groups were analyzed using two-tailed Student's t test or one-way analysis of variance (ANOVA) followed by Bonferroni-type post hoc test, respectively. In RNA-Seq analyses, p values were determined using the corresponding analytical tools. Statistical significance was set at p < 0.05.

Data availability
The raw RNA-Seq data have been deposited at the DNA Data Bank of Japan (DDBJ) and are publicly available with accession number DRA014733. Other individual datasets and corresponding files generated in this study are available upon reasonable request from the corresponding authors.
Supporting information-This article contains supporting information.