Abstract
The analysis of myo-inositol phosphates (InsPs) and myo-inositol pyrophosphates (PP-InsPs) is a daunting challenge due to the large number of possible isomers, the absence of a chromophore, the high charge density, the low abundance, and the instability of the esters and anhydrides. Given their importance in biology, an analytical approach to follow and understand this complex signaling hub is highly desirable. Here, capillary electrophoresis (CE) coupled to electrospray ionization mass spectrometry (ESI-MS) is implemented to analyze complex mixtures of InsPs and PP-InsPs with high sensitivity. Stable isotope labeled (SIL) internal standards allow for matrix-independent quantitative assignment. The method is validated in wild-type and knockout mammalian cell lines and in model organisms. SIL-CE-ESI-MS enables for the first time the accurate monitoring of InsPs and PP-InsPs arising from compartmentalized cellular synthesis pathways, by feeding cells with either [13C6]-myo-inositol or [13C6]-D-glucose. In doing so, we uncover that there must be unknown inositol synthesis pathways in mammals, highlighting the unique potential of this method to dissect inositol phosphate metabolism and signalling.
Introduction
myo-Inositol (inositol hereafter) phosphates (InsPs) are second messengers involved in signaling processes in eukaryotes1. In principle, 63 phosphorylated InsPs can be generated by sequential phosphorylation of the hydroxy groups of inositol, resulting in significant analytical ramifications. Moreover, the fully phosphorylated inositol hexakisphosphate InsP6, the usually most abundant species, can be further phosphorylated to diphospho-inositol polyphosphates (PP-InsPs), called inositol pyrophosphates (Figure 1a and Supplementary Figure 1 for the mammalian PP-InsPs pathway)2,3. These structures contain one (PP-InsP5) or two ((PP)2-InsP4) phosphoric anhydride (P-anhydride) bonds in addition to the phosphate esters (Figure1a). The current model for biologically relevant isomers places the P-anhydrides in defined positions. For example, mammals, yeast, and plants produce 5-PP-InsP5 as the main isomer, with the P-anhydride residing in the plane of symmetry at the 5-position. The second, lower abundance, isomer is 1-PP-InsP5, which has a biologically likely irrelevant enantiomer, its mirror-image 3-PP-InsP5. Further phosphorylation of 5-PP-InsP5 leads to 1,5-(PP)2-InsP4. PP-InsPs are considered metabolic messengers, whose functions have recently become the focus of intense research4. Across species, they are signals in diverse processes including the regulation of energy metabolism and phosphate homeostasis5–7. Other organisms, e.g. the social amoeba Dictyostelium discoideum, produce distinct PP-InsPs isomers (Figure1a) whose functions remain elusive8,9.
The metabolic complexity of InsP turnover and their low abundance, in combination with the absence of a chromophore, and high charge density, has hampered research into these signaling molecules. The most widely applied quantification technology relies on metabolic labeling of cells using tritiated [3H]-inositol, followed by acidic extraction, strong anion exchange high performance liquid chromatography (SAX-HPLC), and manual scintillation counting of individual fractions10. While this approach is sensitive, it requires a dedicated radioactive suite, is expensive, and labor and time-consuming. Moreover, it is blind to inositol generated endogenously from D-glucose-6-phosphate by inositol-3-phosphate synthase 1 (ISYNA1). Therefore, postcolumn derivatization UV-detection approaches have been developed to avoid radiolabeling11,12. Recently, it was shown that inositol tetrakisphosphate 1-kinase 1 (ITPK1), which is present in Asgard archaea, social amoeba, plants, and animals, sequentially phosphorylates Ins(3)P ultimately leading to InsP6 and PP-InsPs11,13,14. Different InsP pools generated from exogenously-acquired or endogenously-synthesized inositol can potentially be monitored using chromatography coupled to mass spectrometry (LC-MS) after feeding cells with ‘heavier’ species such as [13C6]-inositol. However, standard SAX-HPLC using water-salt-based gradients is incompatible with MS detection and MS-compatible volatile buffers do not currently enable isomer assignment15. HPLC-MS/MS and hydrophilic interaction liquid chromatography coupled to MS (HILIC-MS/MS) unfortunately result in a suboptimal separation of the analytes, obliterating InsP isomer identity16.
The development of resolving methods using an electric field to separate the differentially charged InsPs has been pursued. High-voltage paper chromatography17 was instrumental in the discovery and establishment of Ins(1,4,5)P3 as the Ca2+ release factor18. The separation of higher InsPs by gel electrophoresis (PAGE) is another possibility; it, however, does not have the resolving power to distinguish PP-InsP regioisomers and does not detect lower InsPs due to staining inefficiency19.
A capillary electrophoresis (CE) mass spectrometry (MS) method is described herein, that complements and significantly improves analytical approaches in the field. It does not require derivatization, benefits from the separation efficiency of charged analytes by CE, and enables accurate isomer identification and quantification using stable isotope labeled (SIL) reference compounds, even in complex matrices. This new setup also enables stable isotope pulse labeling experiments to analyze the amount of endogenously synthesized inositol over time.
Results
Development of CE-ESI-MS for the analysis of InsPs and PP-InsPs
CE is known as an effective separation tool for phosphate-containing molecules. An early attempt to implement CE was made in the study of Ins(1,4,5)P3, but the method was not developed further20. We now introduce a CE-ESI-MS method, using a bare fused silica capillary and a simple background electrolyte (BGE), for parallel analyses of PP-InsPs and InsPs.
A set of PP-InsP and InsP standards (Figure 1b), representing mammalian metabolites (Figure 1a, Supplementary Figure 1), including Ins(1,4,5)P3, Ins(1,3,4,6)P4, Ins(1,4,5,6)P4, Ins(1,3,4,5)P4, Ins(2,3,4,5,6)P5, Ins(1,3,4,5,6)P5, Ins(1,2,3,4,6)P5, Ins(1,2,3,4,5)P5, InsP6, 5-PP-InsP5, 1-PP-InsP5, and 1,5-(PP)2-InsP4, was resolved with a BGE (35 mM ammonium acetate adjusted to pH 9.7 with NH4OH) by applying a 30 kV voltage over a regular bare fused silica capillary with a length of 100 cm. Detection of analytes was achieved with an ESI-TOF-MS instrument in the negative ionization mode. A stable CE separation current (23 μA) and ESI spray current (2.1 μA) were maintained with a sheath flow CE-ESI-MS interface. The limits of quantification (LOQs) for different InsPs were 250 nM, with 10 nL sample injection volume, i.e. 2.5 fmol of analyte. As baseline separation for the analytes is achieved, no issues with the inevitable in-source fragmentation with neutral loss of phosphate are encountered for accurate quantification. Generally, less than 10% in-source fragmentation products were produced from doubly charged anionic forms of InsP5 to InsP8 with neutral loss of phosphate (79.97 Da). Next, we validated the CE-ESI-MS protocol in analyzing InsPs from biological samples. Initially, we focused on two HCT116 cell lines (HCT116UCL and HCT116NIH) that have been shown to possess different PP-InsP levels21. The CE-ESI-MS method was fully compatible with current state-of-the-art InsP extraction by perchloric acid followed by enrichment with TiO2 (Figure 1c). We introduced and resolved in parallel stable-isotope labelled (SIL) internal standards of [13C6]1,5-(PP)2-InsP4, [13C6]5-PP-InsP5, [13C6]1-PP-InsP5, [13C6]InsP6 and [13C6]Ins(1,3,4,5,6)P522. Application of SIL standards is crucial, as the assignment of InsPs, particularly regioisomers, now becomes unambiguous. Spiking with precise amounts of SIL standards into a biological extract also enables a reliable quantitative assessment, since they compensate for matrix effects and analyte loss. Ins(1,3,4,5,6)P5, InsP6, and 5-PP-InsP5 were assigned by their isotopic pattern, accurate mass, and identical migration time with spiked SIL standards. Excellent resolution and column efficiency were obtained: 1.5×104, 4.6×104, 3.0×104 theoretical plates per meter for 5-PP-InsP5, InsP6 and Ins(1,3,4,5,6)P5, isolated from HCT116NIH extract, respectively (Figure 1c). Analysis of HCT116UCL extract found 1,5-(PP)2-InsP4 (Figure 2aI), a signal which was generally under the LOQ (250 nM) but within LOD in HCT116NIH extract (Figure 2aII), consistent with previous observations21. Furthermore, the analysis of HCT116UCL confirmed the presence of the less abundant 1-PP-InsP5 isomer, representing less than 10% of the cellular PP-InsP pool (Figure 2aIII). This is an important step forward for characterizing PP-InsP metabolism using mass spectrometry. The TiO2 extraction method was previously reported to fully recover PP-InsPs and InsP6 from mammalian cell extracts23. We confirmed this observation by spiking the HCT116NIH samples with SIL internal standards both before extraction (pre-spiking) and before measurement (post-spiking) (Supplementary Figure 2).
The CE-ESI-MS protocols are not limited to analysis of InsP6 and PP-InsPs. Using the described conditions, Ins(1/3,2,4,5,6)P5 and Ins(1,2,3,4,6)P5 or Ins(1,2,3,4/6,5)P5 were readily distinguished from Ins(1,3,4,5,6)P5 (Supplementary Figure 3a). Similarly, identification of InsP4 positional isomers was achieved by measuring the accurate mass in combination with spiking of InsP4 standards Ins(1,3,4,6)P4, Ins(1,4,5,6)P4 and Ins(1,3,4,5)P4 (Figure 2b).
Owing to minimal sample consumption (10 nL) and rapid analysis time (30 min) per run, measurement of technical replicates is feasible. The intra- and inter-day repeatabilities of the method analyzing HCT116NIH extract were evaluated, with mean RSDs of 3.0% (Day 1), 3.1% (Day 2) and 6.3% (Day 3) for three technical replicates, and a mean RSD of 3.6% for technical replicates from five individual days (Figure 1d). A comparison with the repeatabiliy of SAX-HPLC studies is difficult: details of technical replicates are often not provided, since the analysis of one sample takes about 6-8 hours with hands-on processing24,25.
CE-ESI-MS analysis of mammalian PP-InsP metabolism
The CE-ESI-MS method was also applied to monitor changes in PP-InsPs metabolism in mammalian cells in which their synthetic enzymes have been knocked out or perturbed using inhibitors. In mammals, PP-InsPs are synthesized by two different classes of enzymes (Supplementary Figure 1). The IP6Ks, by phosphorylating position 5 of InsP6, synthesize 5-PP-InsP5. The PPIP5Ks are bifunctional (kinase/phosphatase) enzymes that, by acting on position 1, mainly convert 5-PP-InsP5 into 1,5-(PP)2-InsP4. CE-ESI-MS analysis of HCT116UCLIP6K1,2−/− confirms prior observations that PP-InsPs are absent (Figure 2aV)26. Levels of PP-InsPs can be increased by blocking their dephosphorylation using sodium fluoride (NaF, 10 mM)27. CE-ESI-MS analysis of NaF-treated HCT116UCL (Supplementary Figure 3) or HCT116NIH cells (Figure 2c) demonstrated the expected 5-PP-InsP5 elevation (7.3-fold in HCT116UCL cells and 6.2-fold in HCT116NIH cells) with concomitant reduction in InsP6. We observed a reduction in Ins(1,3,4,5,6)P5 levels and appearance of 5-PP-InsP4 (Figure 2aIV), changes also observable by SAX-HPLC analysis of [3H]-inositol labeled HCT116UCL (Supplementary Figure 3). The synthesis of 5-PP-InsP4 is dependent on IP6Ks acting on Ins(1,3,4,5,6)P5: consistent with this, CE-ESI-MS analysis of NaF-treated HCT116UCLIP6K1,2−/− also failed to detect any 5-PP-InsP4. However, confirming a previous observation26, 1-PP-InsP5 became detectable in HCT116UCLIP6K1,2−/− after NaF treatment (Figure 2aVI). This is explained by PPIP5Ks’ capacity to use InsP6 as a substrate, particularly when its preferred substrate, 5-PP-InsP5, is absent. We also observed an increase in Ins(1/3,4,5,6)P4 levels in NaF-treated HCT116NIH cells (Figure 2b) as a result of PLC activation27. Analysis of HCT116NIHPPIP5K1,2−/− in comparison to WT cells showed a small increase of the non-metabolized substrate 5-PP-InsP5, the levels of which are 3.6-fold enhanced by NaF treatment (Figure 2c). We additionally analyzed the effect of a recently identified IP6K inhibitor: quercetin (Q)28 reduced 5-PP-InsP5 levels in both HCT116NIH and HCT116NIHPPIP5K1,2−/− cells by 50-60% (Figure 2c).
These results validate CE-ESI-MS as a technique to monitor with unprecedented accuracy changes in cellular InsPs and PP-InsPs metabolism in response to different stressors or genetic alterations.
Analysis of InsPs in mammalian cell lines and tissues
We determined the concentrations of Ins(1,3,4,5,6)P5, InsP6, and 5-PP-InsP5, in different mammalian cell lines, including HCT116UCL, HeLa, HT29, PC3, 293T, and MCF7 (Figure 2d, Supplementary Figure 4). The detected InsP6 and 5-PP-InsP5 cellular concentrations as well as their relative ratio are in accordance with earlier results obtained by [13C]-NMR29. The InsP6 cellular concentration, for example, was in the range of 24-47 μM (300-500 pmol/mg protein). However, Ins(1,3,4,5,6)P5 levels were surprisingly variable, potentially reflecting different functional roles (Supplementary Figure 5) across different cell lines. Therefore, the CE-MS method will be instrumental to uncover dynamics and physiological roles of InsP5 in mammalian cells.
We quantified the amount of InsPs and PP-InsPs in mouse organs, including liver, brain, muscle, kidney, and spleen (Supplementary Figure 6). Again, Ins(1,3,4,5,6)P5, InsP6, and 5-PP-InsP5 were the main InsP species. Comparably low InsP levels were detected in muscle. Analysis of InsP6 and 5-PP-InsP5 extracted from tissues performed by PAGE, lack the sensitivity, resolution and dynamic range, evidenced by CE-ESI-MS. The analysis of InsPs from animal organs or tissues cannot be performed by SAX-HPLC, prohibited by cost, feasibility, and ethical considerations, since it requires feeding a mouse with [3H]-inositol.
InsPs and PP-InsPs in Saccharomyces cerevisiae and Arabidopsis thaliana
We next analyzed InsPs and PP-InsPs in yeast and plant samples, to explore the applicability of CE-ESI-MS across experimental models. The [3H]-inositol SAX-HPLC method has been extensively employed to study yeast and plant InsP metabolism. PAGE methods conversely cannot be used to analyze InsPs from yeast, due to the abundant inorganic polyphosphate (polyP) that suppresses the InsP signals (Supplementary Figure 7a), and while PAGE has been applied to study InsPs including PP-InsPs in plant extracts14,30,31, the same limitations described above apply. Using SIL-CE-ESI-MS, profiling of InsPs and PP-InsPs was readily achieved for both Saccharomyces cerevisiae and Arabidopsis thaliana.
In wild type yeast extracts, InsP6, 5-PP-InsP5 and 1/3,5-(PP)2-InsP4 were detectable. In agreement with the literature32, 5-PP-InsP5 and 1/3,5-(PP)2-InsP4 were around 3% and 1% of the InsP6 level, respectively (Supplementary Figure 7).
We analyzed the InsPs present in shoots of A. thaliana wild type (Col-0) and in plants defective in Inositol Pentakisphosphate 2-Kinase (AtIPK1) or the ATP-binding cassette (ABC) transporter 5 (AtMRP5)33 that transports InsP6 into the vacuole (Figure 3): Ins(1/3,2,4,5,6)P5, InsP6, 5-PP-InsP5, 1/3-PP-InsP5 and 1/3,5-(PP)2-InsP4 were readily detected in shoot extracts of A. thaliana Col-0 seedlings. Surprisingly, comparable levels of 5-PP-InsP5 and 1/3-PP-InsP5 were observed. This represents a significant deviation from the general notion, derived from mammalian cell analysis, that 1-PP-InsP5 represents the minor isomeric species. The atipk1 mutant displayed decreased levels of PP-InsPs and InsP6, and a robust increase in Ins(1,3,4,5,6)P5 s. In atipk1 plants, 5-PP-InsP4 was also detected by CE-ESI-MS, which was previously unsuccessfully tracked by [3H]-inositol labelling and SAX-HPLC analysis34. Analysis of atmrp5 shoots revealed the expected elevated levels of 5-PP-InsP5 and 1/3,5-(PP)2-InsP4, but not of 1/3-PP-InsP5. This analysis represents the first detailed elucidation of single isomer PP-InsP alterations in plants, underlining the value of the SIL-CE-ESI-MS method for dissecting PP-InsP isomers and relative abundances.
CE-ESI-MS to analyze Dictyostelium discoideum PP-InsP metabolism
The social amoeba Dictyostelium discoideum possesses large amounts of PP-InsPs. However, this model organism contains different PP-InsPs isomers from those in mammals, yeast and plants, such as 6-PP-InsP5 and 5,6-(PP)2-InsP49,35. This complexity is a challenge for ideal CE separation conditions. Employing a different BGE with decreased ionic strength and pH (30 mM ammonium acetate adjusted to pH 9.0 with NH4OH) led to enhanced resolution of InsP6 and more anionic species. Under these conditions, all possible PP-InsP5 isomers, including non-natural ones obtained by chemical synthesis36–38, were separated and could thus be assigned in complex matrices (Figure 4a). This BGE also enabled the separation of all available InsP5 isomers. We then performed CE-ESI-MS analyses of D. discoideum extracts (Figure 4b). As expected, 4/6-PP-InsP5 was about twice as abundant as 5-PP-InsP5, while 1/3-PP-InsP5 was present at around 5% of the whole PP-InsP5 pool. We identified two (PP)2-InsP4 isomers, 5,4/6-(PP)2-InsP4, and 1,5-(PP)2-InsP4. Peak assignment and quantification were conducted in a single run (30 minutes), and accurate mass information was provided simultaneously (Figure 4c).
Endogenous inositol synthesis contributes to the InsP pools
The ability of MS to capture isotopic mass differences is a significant advantage of the CE-ESI-MS protocol. For example, by using isotopically labeled metabolic precursors, the contribution to the InsP pool of both the inositol acquired from the milieu and from endogenous synthesis from glucose can now be assessed (Figure 1a). In SAX-HPLC analysis, inositol-free medium is used to improve [3H]-inositol labeling, with the tracer used at 0.5-1 μM concentration25. This method does not detect endogenously generated inositol. To assess the contribution of endogenous synthesis of inositol to InsP cellular pools, we performed an inositol titration curve. Wild type HCT116UCL were incubated for five days, the incubation time used for [3H]-inositol labeling to reach metabolic equilibrium (7-8 cell division cycles), in inositol-free DMEM in the presence of 1, 10 and 100 μM [13C6]-inositol. We observed a dose-dependent incorporation of [13C6]-inositol into the [13C6]InsP5 and [13C6]InsP6 pools (Figure 5a). Using 1 μM [13C6]-inositol, less than 20% of the InsP5 and InsP6 pools were synthesized from exogenously acquired inositol; this value increased to 90% using 100 μM. A reference value for inositol concentration in human serum is 30 μM39, available only to cells in direct contact with serum. We next performed a time course, incubating cells with [13C6]-inositol (10 μM) in inositol-free DMEM for 1, 3 and 5 days (Figure 5b). Initially, all InsP5 and InsP6 species were constituted of [12C]-inositol. By 24 h [13C6]InsP5 and [13C6]InsP6 became detectable, representing a small fraction of their respective pools. After five days, [13C6]InsP5 and [13C6]InsP6 represented two thirds of the InsP5 and InsP6 pools. These results indicate either a sluggish InsP5 and InsP6 turnover, even more lethargic than previously thought40, or an endogenous synthesis of [12C]-inositol that substantially contributes to the InsP5 and InsP6 pools.
The inositol phosphate synthase (IPS or MIPS) called ISYNA1 in mammals converts glucose-6-phosphate to Ins(3)P41. To study the contribution of endogenous inositol synthesis to InsP pools, we generated two independent HCT116UCL ISYNA1−/− clones using CRISPR. The two knockout clones, KO1 and KO2, were verified by sequencing (Supplementary Figure 8) and western blot analysis (Figure 6a). The InsP6 level as analyzed by PAGE was unaffected in the ISYNA1−/− clones (Figure 6b), and there was no growth defect in normal medium (Figure 6c). Titrating the medium with different inositol concentrations revealed 10 μM inositol as sufficient to guarantee wild type growth rate, while the absence of inositol dramatically reduced the growth of both clones (Figure 6d-f). To study the contribution of ISYNA1 to the InsP pools, we incubated wild type HCT116UCL and ISYNA1−/− clones for 5 days in 25 mM [13C6]-glucose with 1 or 10 μM inositol, and then extracted and analyzed the samples by CE-ESI-MS (Figure 6g). In wild type cells with only 1 μM inositol, roughly 60% of the InsP5 and InsP6 pools were generated by the endogenous conversion of [13C6]-glucose-6-phosphate to Ins(3)P, detected as [13C6]InsP5 and [13C6]InsP6. Note that often 1 μM or less of inositol is used in [3H]-inositol labeling for SAX-HPLC experiments26,42. Our analysis, therefore, reveals that up to 60% of InsPs are not detected by traditional methods if inositol concentration is kept low to improve [3H]-inositol labeling efficiency. In the presence of a tenfold higher exogenous inositol concentration (10 μM), endogenously generated inositol contributed 15% to the InsP5 and InsP6 pools. Strikingly, we detected [13C6]InsP5 and [13C6]InsP6 in both ISYNA1−/−KO1 and ISYNA1−/−KO2, although to a lesser extent than in wild type cells (Figure 6h-j). This highly unexpected result indicates the existence of an alternative uncharacterized enzymology for inositol synthesis, underscoring the enormous potential of the CE-ESI-MS technique. This approach will be instrumental not only in elucidating the novel inositol synthetic route we have discovered, but also in any future assessment of inositol phosphate physiological functions across the kingdoms of life.
Discussion
The analysis of InsP and PP-InsP turnover in cells and tissues is a daunting challenge. Radiolabeling followed by SAX-HPLC has so far been the method of choice, as it provides the sensitivity needed for meaningful analyses of the less abundant species, with the advantage of selectively visualizing only InsPs and PP-InsPs when [3H]-labeled inositol is fed to cells. This approach, however, misses inositol endogenously synthesized from glucose, is restricted to specialized laboratories, is expensive, and time-consuming. While other approaches, such as [13C] labeling for NMR, PAGE, and HILIC-MS, have recently been developed to provide alternatives, a transformative approach is still missing.
Here, we have demonstrated that CE is an outstanding separation platform for InsPs and PP-InsPs. Moreover, CE coupling to an ESI-Q-TOF mass spectrometer facilitates parallel analyses of a multitude of analytes in a single run, requiring only 30 minutes and nL sample injection. According to the accurate mass information and identical mobility with (isotopic) standards, these densely charged species can now be readily assigned even in complex matrices. Additionally, the introduction of stable isotope labeled (SIL) reference compounds allows for quantification and correction of matrix effects in samples such as those obtained from yeast extracts rich in polyphosphates, where drifts in migration times of several minutes were observed. Using this approach, we were able to quantify InsPs and PP-InsPs in different species, and in wild type and knock-out cell lines additionally treated with inhibitors of several enzymes. Given the rapidity of the analysis, measurement of technical replicates becomes possible, underlining the robustness and fidelity of the method. Using CE-ESI-MS, we have been able to extract essential new information from several samples. For example, Ins(1,3,4,5,6)P5 concentrations are highly variable across different mammalian cell lines and whole organs. We also show that 1/3-PP-InsP5 is not always the minor isomer present, as exemplified by our analysis of A. thaliana seedlings, raising questions concerning its potential regulatory effects. Moreover, SIL inositol and SIL D-glucose were used in pulse labeling experiments, demonstrating that SIL-CE-ESI-MS can be used to monitor and quantify inositol isomer turnover originating from different sources. At low (1 μM) exogenous inositol concentration, around 60% of cellular inositol was derived from glucose after 5 days of labeling. Yet, knockout of the only known glucose-6-phosphate utilizing inositol synthase in mammalian cells (ISYNA1) did not lead to cells incompetent in transforming D-glucose to inositol. This finding reveals that there must exist a yet uncharacterized biochemical pathway for the synthesis of inositol deriving its carbon skeleton from glucose. We thus conclude that SIL-CE-ESI-MS will open our eyes to cellular pathways we have previously been blind to.
Methods
Materials and Reagents
InsP6, Ins(1,3,4,5,6)P5, Ins(2,3,4,5,6)P5, Ins(1,2,3,4,6)P5, Ins(1,2,3,4,5)P5, Ins(1,3,4,5)P4 and Ins(1,4,5)P3 with purity more than 95-97% ([31P] NMR) were purchased from Sichem. Ins(1,3,4,6)P4 and Ins(1,4,5,6)P4 were obtained from Cayman. 1-PP-InsP5, 5-PP-InsP5, 6-PP-InsP5 and 2-PP-InsP5 were synthesized in-house (Jessen)36–38. Stable-isotope labelled (SIL) internal standards [13C6]1,5-(PP)2-InsP4, [13C6]5-PP-InsP5, [13C6]1-PP-InsP5, [13C6] InsP6, and [13C6]Ins(1,3,4,5,6)P5 with purities higher than 96% were synthesized in-house (Fiedler)22,29. Concentrations of stock solutions of InsP and PP-InsP standards for quantification were determined by [1H] NMR and/or [31P] NMR as described below. Fused silica capillaries were obtained from CS-Chromatographie.
CE-ESI-MS Analysis
All experiments were performed on a bare-fused silica capillary with a length of 100 cm (50 μm internal diameter and 365 μm outer diameter) on an Agilent 7100 capillary electrophoresis system coupled to a Q-TOF (6520, Agilent) equipped with a commercial CE-MS adapter and sprayer kit from Agilent. Prior to use, the capillary was flushed for 10 min with 1N NaOH, followed by water for 10 min, and background electrolyte (BGE) for 15 min. BGE A (35 mM ammonium acetate titrated by ammonia solution to pH 9.7) was employed for the analysis of all mammalian cell and tissue extracts, as well as Saccharomyces cerevisiae and Arabidopsis thaliana extracts. BGE B (30 mM ammonium acetate titrated by ammonia solution to pH 9.0) was used for Dictyostelium discoideum. Samples were injected by applying 50 mbar pressure for 10 s, corresponding to 0.5% of the total capillary volume (10 nL). In the study of the endogenous inositol synthesis, samples were injected with 100 mbar pressure for 10 s (20 nL). After sample injection, a BGE post-plug was introduced by applying 50 mbar for 2 s. For each analysis, a constant CE current of either 23 μA (BGE A) or 19 μA (BGE B) was established by applying 30 kV over the capillary, which was kept at a constant temperature of 25 °C.
The sheath liquid was composed of a water-isopropanol (1:1) mixture spiked with mass references. It was introduced at a constant flow rate of 1.5 μL/min. ESI-TOF-MS was conducted in the negative ionization mode; the capillary voltage was set to −3000 V and stable ESI spray current at 2.1 μA. For TOF-MS, the fragmentor, skimmer, and Oct RFV voltage was set to 140, 60, and 750 V, respectively. The temperature and flow rate of drying gas was 250 °C and 8 L/min, respectively. Nebulizer gas pressure was 8 psi. Automatic recalibration of each acquired spectrum was performed using reference masses of reference standards (TFA anion, [M-H]-, 112.9855), and (HP-0921, [M-H+CH3COOH]−, 980.0163). Exact mass data were acquired at a rate of 1.5 spectra/s over a 60−1000 m/z range. Extracted ion electropherograms (EIEs) were created using a 10 ppm mass tolerance window for theoretical masses corresponding to the targeted inositol pyrophosphates and inositol phosphates.
Peak assignment of 1/3,5-(PP)2-InsP4, 5-PP-InsP5, 1/3-PP-InsP5, InsP6 and Ins(1,3,4,5,6)P5 in biological samples was achieved by accurate mass, isotopic pattern, and identical migration time. Ins(1,3,4,6)P4, Ins(1/3,4,5,6)P4 and 4/6-PP-InsP5 in biological samples were assigned by accurate mass and identical migration time with spiked standards. 5-PP-InsP4 and 4/6,5-(PP)2-InsP4 in biological samples were assigned by accurate mass and based on previous research9,26.
Quantification of PP-InsP and InsP in mammalian cells was performed with known amounts of individual isotopic standards spiked into the samples. The amount of isotopic internal standards (IS) that had to be added was delineated from the concentration of the respective analyte in the sample. Ratio of analyte peak area (Area)12C/ IS peak area (Area)13C is less than 5 to ensure a linear relationship. IS stock solutions of 125 μM [13C6]5-PP-InsP5, 500 μM [13C6]InsP6 and 500 μM [13C6]Ins(1,3,4,5,6)P5 were used for the spiking experiments. 0.4 μL IS stock solution were added into 10 μL samples before measurement. 5 μM [13C6]5-PP-InsP5, 20 μM [13C6]InsP6 and 20 μM [13C6]Ins(1,3,4,5,6)P5 were the final concentrations inside klsamples. The calibration curve for each analyte was constructed at eight levels by regression of nominal analyte concentration against a ratio of analyte peak area (Area)12C/ IS peak area (Area)13C (Supplementary Figure 9). The calibration curves were linear and provided a coefficient of determination ̎0.997 over the investigated range of concentrations (0.25-25 μM for 5-PP-InsP5, 1.0-100 μM for InsP6 and Ins(1,3,4,5,6)P5). For quantification, two technical replicates were conducted for each sample.
Quantification of InsP and PP-InsP in tissue extracts, Saccharomyces cerevisiae, Arabidopsis thaliana, and Dictyostelium discoideum extracts was performed by comparing analyte peak areas with the respective peak areas of SIL internal standards with known concentrations. Concentrations of SIL internal standard solutions were determined by quantitative 31P and 1H NMR against a certified reference standard (phosphoacetic acid, TraceCERT® Merk 96708-1G).
Maintenance and manipulation of mammalian cell lines
HCT116NIH and HCT116NIH PPIP5K−/− cells43 were grown in DMEM (Gibco) supplemented with 4.5 g/L glucose and 10 % heat inactivated FBS (Gibco). HCT116UCL (obtained from European Collection of Authenticated Cell Cultures [ECACC]) and HCT116UCL IP6K1,2−/− cells44 were cultured in DMEM supplemented with 10% FBS (Sigma) and 4.5 g/L glucose. All cells were grown in a humidified atmosphere with 5% CO2. InsP levels were modulated by incubating the cells for 60 min with 10 mM NaF or for 30 min with 2.5 μM quercetin28 prior to harvesting. For inositol limitation or [13C6]-inositol labelling experiments, inositol-free DMEM (MP Biomedicals) with 10% dialyzed FBS (Sigma) was used. Normal inositol (Sigma) or [13C6]-inositol29 were supplemented as appropriate. Cells were acclimatized to 10 μM inositol in inositol-free DMEM for one week before starting labelling experiments. For [13C6]-glucose (Sigma) labelling experiments, DMEM lacking both inositol and glucose (Thermo Fisher) was used, using 10% dialyzed FBS. Cells were washed twice in the relevant starvation medium before incubation.
Mammalian cell growth assay
To measure cell growth, the sulforhodamine B (SRB) assay was performed45. Cells were seeded into 96 well plates. After 24 h, the medium was removed, wells were washed, then 100 μL treatment medium was added. At each timepoint, cells were fixed in 10% trichloroacetic acid. Fixed plates were stained with 0.05% sulforhodamine B (Sigma) in 1% acetic acid, and fixed dye was solubilised in 10 mM Tris base before reading absorbance at 500 nm using a spectrophotometer.
To measure cell volume, 80-90% confluent cells were trypsinised and resuspended in growth medium. A Multisizer 4 (Beckman Coulter) machine was used, following the manufacturer instruction.
Generation of ISYNA1 KO cell lines
The human colon carcinoma cell line HCT116 was used to generate knockouts as it is pseudo-diploid, and has easily detectable amounts of InsP6 and InsP723. The Alt-R CRISPR-Cas9 (Integrated DNA Technologies) system was used, with guide sequence 5’-CCAAUCGACUGCGUU-3’. CRISPR components were introduced into the cells using a Neon electroporator (Thermo Fisher) and cells plated into 96 well plates using limiting dilution. Colonies were screened by western blotting using anti-ISYNA1 antibody (Santa Cruz sc-271830). Positive knockout clones were further confirmed by PCR and Sanger sequencing-based analysis (Genewiz CRISPR Analysis Package).
Purification of inositol phosphates by titanium dioxide pulldown
Extraction of inositol phosphates was performed according to the literature44. Briefly, 80-90% confluent cells were extracted using 1 M perchloric acid as described below. Titanium dioxide beads (Titansphere TiO 5 μm; GL Sciences) were used to pull down inositol phosphates, which were eluted using 3% ammonium hydroxide. The ammonia was eliminated and the samples concentrated using a speedvac evaporator for 1-3 h at 40°C or 60°C. For InsPs analysis by PAGE, the extracts were normalized to protein concentration and resolved using 35% PAGE gels, as previously described19. Inositol phosphates were visualized by Toluidine blue (Sigma) staining. A desktop scanner (Epson) was used to record the PAGE result.
Preparing cell extracts for CE-ESI-MS
Mammalian cells
Cells (8 million HCT116UCL cells, HCT116UCL IP6K1,2−/− cells and HCT116NIH PPIP5K−/− cells; 6 million HCT116NIH cells) were seeded into 15 cm dishes and allowed to grow for 48 hours (HCT116UCL and HCT116UCL IP6K1,2−/− cells) or 72 hours (HCT116NIH and HCT116NIH PPIP5K−/− cells) until 80-90% confluent. To harvest, dishes were quickly washed twice with cold PBS, then incubated with 1-5 mL cold 1 M perchloric acid on ice for 10 min. Acidic extracts were then collected from the plates, and inositol phosphates and other small polar molecules extracted using titanium dioxide beads23,44. To determine protein concentrations, post-extraction dishes were washed twice in PBS and proteins were solubilized via addition of 1.5 mL cell lysis buffer (0.1 % SDS in 0.1 M NaOH) followed by incubation for 15 min at room temperature. Cell extracts were then pelleted. Protein contents of cell lysates were determined using the DC protein assay (Biorad) with BSA as calibration standard. To provide cell volume values and cell counts for normalization, parallel dishes were prepared and trypsinized. For SIL-CE-ESI-MS, cells were seeded into 6 well plates. After 24 h, the medium was removed, cells were washed, then 2 mL treatment medium was added. Cells were harvested by trypsinization to maximize yield, before extraction with perchloric acid and TiO2 beads. Parallel dishes were prepared to provide cell counts and protein concentration values for normalization.
Plants
Arabidopsis seeds were surface sterilized and sown onto half-strength Murashige and Skoog (MS) medium supplemented with 1% succrose34. Plants were grown under 16h/8h day/night conditions at 22°C/20°C for 14 days. Light was provided by white LEDs (“True daylight”, Polyklima). Shoots (150-180 mg, fresh weight) were shock-frozen in liquid nitrogen, homogenized and immediately resuspended in 1 M perchloric acid (SigmaAldrich). Titanium dioxide purification of inositol phosphates was carried out as described above.
D. discoideum
Wild type amoeba D. discoideum AX2, obtained from the Dicty Stock Center (Northwestern University, Chicago, USA), was grown in SIH defined minimal media (Formedium) at 20°C in a flask with moderate shaking to a cell density of 3-4×106 cells/ml. Twenty million cells were harvested by centrifugation (1 000 g; 5 min), washed in KK2 buffer (20 mM K-Phosphate buffer pH 6.8), resuspended in 500 μL ice cold perchloric acid solution (1 M perchloric acid, 5 mM EDTA) and incubated on ice for 10 min, gently mixing the cell suspension every two min. The cell suspension was centrifuged (15 000 g; 5 min at 4°C) and the supernatant subject to TiO2 purification as described.
S. cerevisiae
Wild type yeast (BY4741) was grown in Complete Supplement Mixture (SCM) media (Formedium) overnight with shaking at 30°C to logarithmic phase (OD600=1-3). Forty OD600 units were harvested by centrifugation (1 000 g; 5 min), washed with ice cold water and resuspended in 500 μL ice cold perchloric acid solution (1 M perchloric acid, 5 mM EDTA). After adding ~300 μL acid-washed glass beads (Sigma Aldrich) yeast were vigorously vortexed for 5 min at 4°C. The lysate was centrifuged (15 000 g; 5 min at 4°C) and the supernatant (acid extract) subject to TiO2 purification as described.
Inositol phosphate analysis by SAX-HPLC
Analysis of InsP pathways after [3H]-inositol radiolabeling was carried out as previously described25. Briefly, cells were seeded into 6 well plates and grown in the presence of [3H]-inositol for 5 days to ~80% confluence. Treatment with NaF (10 mM) was for 1 hour. Cells were then washed with ice-cold PBS and extracted with perchloric acid, and after neutralization processed for SAX-HPLC analysis.
Western blotting
Cells were lysed in TX buffer (50 mM HEPES pH 7.4, 1 mM EDTA, 10% glycerol, 1% Triton X-100, 50 mM sodium fluoride, 5 mM sodium pyrophosphate) supplemented with protease and phosphatase inhibitor cocktails (Sigma). Lysates were cleared by centrifugation at 18,000 rpm for 5 min at 4°C, and protein concentrations measured by DC Protein Assay (Bio-Rad). Lysates were resolved using NuPAGE 4-12% bis-tris gels (Life Technologies) and proteins transferred to nitrocellulose membranes. Ponceau S solution (0.1% Ponceau S [Sigma] in 1% acetic acid) was used to confirm equal loading. Membranes were blocked for 1 hour in 5% non-fat milk in TBS-T (10 mM Tris base, 140 mM NaCl, 0.05% Tween) then blotted for the following primary antibodies at 1:100-1:1000 overnight in 3% milk: IP6K1 (HPA040825), IP6K2 (HPA070811), IPPK (HPA020603; Sigma), ISYNA1 (sc 271830), actin (sc-1616; Santa Cruz), histone H3 (ab1791; Abcam). Secondary HRP-conjugated antibodies (Sigma) were diluted in 3% milk. Signal was detected using Luminata Crescendo Western Substrate (Merck Millipore) and Amersham Hyperfilm (VWR) and a film developer.
Footnotes
↵* danyeqiu{at}gmail.com; a.saiardi{at}ucl.ac.uk; henning.jessen{at}oc.uni-freiburg.de