Labile Metabolite Profiling in Human Blood using Phosphorus NMR Spectroscopy

Abstract

Phosphorus metabolites occupy a unique place in cellular function as critical intermediates and products of cellular metabolism. Human blood is the most widely used biospecimen in the clinic and in the metabolomics field and hence an ability to profile phosphorus metabolites in blood, quantitatively, would benefit a wide variety of investigations of cellular functions in health and diseases. Mass spectrometry (MS) and NMR spectroscopy are the two premier analytical platforms used in the metabolomics field. However, detection and quantitation of phosphorus metabolites by MS can be challenging due to their lability, high polarity, structural isomerism, and interaction with chromatographic columns. The conventionally used ¹H NMR, on the other hand, suffers from poor resolution of these compounds. As a remedy, ³¹P NMR promises an important alternative to both MS and ¹H NMR. However, numerous challenges including the instability of phosphorus metabolites, their chemical shift sensitivity to solvent composition, pH, salt, and temperature, and the lack of identified metabolites have so far restricted the scope of ³¹P NMR. In the current study, we describe a method to analyze nearly 25 phosphorus metabolites in blood using a simple 1D NMR spectrum. Establishment of the identity of unknown metabolites involved a combination of (a) comprehensively analyzing an array of 1D and 2D ¹H/³¹P homonuclear and heteronuclear NMR spectra of blood; (b) mapping the central carbon metabolic pathway; (c) developing and using ¹H and ³¹P spectral and chemical shift databases; and finally (d) confirming the putative metabolite peaks with spiking using authentic compounds. The resulting simple 1D ³¹P NMR-based method offers an ability to visualize and quantify the levels of intermediates and products of multiple metabolic pathways including central carbon metabolism, in one step. Overall, the findings represent a new dimension for blood metabolite analysis and are anticipated to greatly impact the blood metabolomics field.

Graphical Abstract

INTRODUCTION

Among different classes of cellular metabolites, phosphorus metabolites occupy a unique place in cellular metabolism. They are ubiquitous metabolites fundamental to a variety of cellular functions. They represent critical intermediates and products of central carbon metabolism including glycolysis, the pentose phosphate pathway (PPP), and the tricarboxylic acid (TCA) cycle. For example, in glycolysis and the PPP, virtually all the intermediates are phosphorus compounds. The energy coenzymes (ATP, ADP, and AMP) and redox coenzymes (NAD⁺, NADH, NADP⁺, and NADPH), which are fundamental to cellular energy production are also phosphorus compounds. The selection of phosphorus compounds for cellular metabolism is intriguing; a major reason why they serve such an important role is potentially because they possess the much-needed (negative) charge to carry out enzymatic reactions. Moreover, their high polarity helps them confine themselves within the cell.^1,2 Generally, phosphorus metabolites are phosphates, phosphate esters, or phosphate anhydrides,¹ all of which possess multiple negative charges. Given their value for probing cellular functions in health and numerous diseases including cancer, obesity, heart and muscle diseases, reliable measurement of phosphorus metabolites in biological specimens continues to be of immense interest.

Human blood is the most widely used biospecimen in the clinic and in the metabolomics field and hence an ability to profile phosphorus metabolites in blood will benefit metabolomics applications. Mass spectrometry (MS) and NMR spectroscopy are the two premier analytical platforms used in the metabolomics field. However, MS is generally met with challenges for analysis of phosphorus metabolites due to their instability, high polarity, and multiple negative charges.³ In addition, the charged phosphate metabolites stick to glassware and chromatography columns and often cause poor peak line shapes, low recovery, and carry-over effects. In addition, often MS cannot distinguish among different structural isomers. For example, the ubiquitous cellular metabolites, fructose 1,6-bisphosphate, fructose 2,6-bisphosphate, and glucose 1,6-bisphosphate, are structural isomers that cannot be distinguished easily by MS. Similarly, the isomers, 2-phosphoglycerate and 3-phosphoglycerate, as well as dihydroxyacetone phosphate and glyceraldehyde-3-phosphate are also very challenging to distinguish by MS. There have been numerous efforts to overcome these MS-based detection challenges, which include, for example, the use of hydrophilic interaction chromatography (HILIC),⁴ ion pairing (IP),⁵ anion exchange chromatography,^6,7 mixed mode chromatography,^8,9 combination of IP and HILIC,¹⁰ and chemical derivatization.³

NMR spectroscopy, with an ability to identify unknown metabolites and its quantitative nature, is an important alternative to MS for analysis of phosphorus metabolites.^11–15 However, the widely used ¹H NMR suffers from the poor resolution arising from the narrow dispersion of signals (<10 ppm), which in particular has restricted the number of quantifiable phosphorus metabolites.^13–15 As a remedy, ³¹P NMR, with its 100% natural abundance and wide chemical shift dispersion offers an important alternative to MS as well as ¹H NMR. Recently, several studies using 2D NMR have demonstrated benefits of using ³¹P NMR for metabolite analysis.^16–18 Unlike 2D NMR, once phosphorus metabolites are positively identified, 1D ³¹P NMR promises several advantages for routine applications, which include the ease of performing experiments, data analysis, quantitation, and high throughput. Further, 1D NMR spectra are devoid of the interference from non-phosphorous compounds including the solvent (water) peak, which is generally present in all samples. Since the identification of separate signals for phosphorus compounds in blood cells nearly half century ago,¹⁹ 1D ³¹P NMR has been applied in in vivo, in vitro as well as ex vivo studies. In blood and blood cells, a few phosphorus metabolites have been detected using 1D ³¹P NMR.^20–22 However, the instability of phosphorus metabolites,^13,15,23 their chemical shift sensitivity to solvent, pH, salt, and temperature,^{16, 19} and the lack of identified metabolites have so far restricted the scope of ³¹P NMR for wider applications in the metabolomics field.

With the focus on addressing these issues, in the current study, we have investigated human blood using a combination of NMR techniques. A ³¹P and ¹H spectral database was first developed for authentic compounds under the same conditions as used for blood samples. Unknown phosphorus metabolites in blood were identified based on the comprehensive analysis of 1D and 2D ³¹P/¹H NMR spectra, in combination with the spectral database and spiking with authentic compounds. The study has led to the development of a ³¹P NMR-based method for analysis of nearly 25 phosphorus metabolites in blood in one-step, which is the highest number that has been analyzed using a simple 1D ³¹P NMR spectrum, to date. The identified metabolites represent intermediates and products of multiple metabolic pathways including central carbon metabolism. Hence, this ³¹P NMR-based method represents a new approach to visualize multiple metabolic pathways in one-step and is anticipated to greatly impact the blood metabolomics field.

MATERIALS AND METHODS

Chemicals and Solvents:

Methanol, chloroform, sodium phosphate (monobasic; NaH₂PO₄), sodium phosphate (dibasic; Na₂HPO₄), and 3-(trimethylsilyl)propionic acid-2,2,3,3-d₄ sodium salt (TSP) were obtained from Sigma-Aldrich (St. Louis, MO). Standard compounds (see Table S1) used for developing the ³¹P and ¹H NMR spectral database were obtained from Sigma-Aldrich (St. Louis, MO). Deuterium oxide (D₂O) was obtained from Cambridge Isotope Laboratories, Inc. (Andover, MA). Deionized water was purified using an in-house Synergy Ultrapure Water System from Millipore (Billerica, MA). All chemicals were used with no further purification.

Biospecimens:

Whole blood specimens from healthy individuals were purchased from Solomon Park Research Laboratories, Inc. (Burien, WA). In addition, fresh blood samples were procured from healthy individuals following the protocol approved by the IRB from the University of Washington. Blood samples were collected in heparinized BD Vacutainer tubes (BioVision, CA).

Ammonium bicarbonate and phosphate buffers:

For ³¹P NMR analysis, ammonium bicarbonate buffer (50 mM; pH =7.8) was prepared by dissolving 4 mg/mL ammonium bicarbonate in D₂O. Separately, for ¹H NMR analysis of standard phosphorus compounds, phosphate buffer (100 mM; pH=7.4) was prepared by dissolving 1124 mg anhydrous Na₂HPO₄ and 250 mg anhydrous NaH₂PO₄ in 100 g D₂O. Sodium phosphate (1.3 mM) or TSP (50 or 120 μM) were added to the buffer solutions to serve as a reference for ³¹P or ¹H NMR chemical shifts, respectively.

Standard compounds for database and spiking:

One mL stock solutions of standard compounds (1.0 and 50.0 mM) were prepared by dissolving appropriate amounts of the compounds in D₂O. From these solutions, 200 μL of 1.0 to 1.25 mM solutions were prepared using ammonium bicarbonate buffer to obtain ³¹P and ¹H NMR spectra at 600 MHz. Separately, 0.1 to 0.3 mM solutions of the same standards were prepared in 200 μL phosphate buffer to obtain ¹H NMR spectra at 800 MHz. The solutions were transferred to 3 mm NMR tubes for analysis.

Blood metabolite extraction:

Frozen human blood samples (300 to 350 μL) were mixed with a mixture of methanol and chloroform in a 1:2:2 ratio (v/v/v), vortexed for 2 min or until a homogenous mixture was formed, sonicated for 20 min at 4 °C, and vortexed again for 30 s. The mixtures were centrifuged at 13,400 × g for 30 min to pellet proteins/cell debris. The top aqueous layer was transferred to fresh vials and dried using nitrogen gas. The dried samples were mixed with 200 μL ammonium bicarbonate buffer containing TSP and transferred to 3 mm NMR tubes. Prior to sample preparation, the buffer was degassed using helium and the NMR tubes were flushed using the gas to remove dissolved oxygen, before and after transferring the solutions.

NMR Spectroscopy:

¹H and ³¹P 1D and 2D NMR experiments for standard compounds as well as blood samples were performed on a Bruker Avance III 600 MHz spectrometer equipped with a wide bore magnet and a 3 mm BBO probe with Z-gradients capable of ³¹P detection. Separately, ¹H 1D and 2D NMR experiments for the standard compounds were performed on a Bruker Avance III 800 MHz spectrometer equipped with a 5 mm cryogenically cooled ¹H {¹³C,¹⁵N} TXI probe with Z-gradients.

For ¹H 1D NMR spectra, the one-pulse or 1D NOESY pulse sequence (‘noesypr1d’, for standard compounds) along with the CPMG (Carr-Purcell-Meiboom-Gill) pulse sequence (‘cpmgpr1d’, for blood samples) with residual water signal suppression using presaturation, were used. Parameters used for the one pulse/1D NOESY experiments were, 6009 or 10,000 Hz (at 600 MHz) or 9615 Hz (at 800 MHz) spectral width, 3 s (at 600 MHz) or 25 s (at 800 MHz) recycle delay, 128 scans, and 32,768 time-domain points. Parameters used for the CPMG experiments were: 6493 Hz spectral width, 3 s recycle delay, 64 to 128 scans, and 32,768 time-domain points. The duration of the 180° pulse train length was 256 ms. For 1D ³¹P spectra, the one pulse sequence with no proton decoupling (“zg”) and with inverse gated proton decoupling (zgig) were used. Parameters used for ³¹P spectra were, 20,000 Hz spectral width, 32,768 time-domain points, 2 to 30 s relaxation delay, and 100 to 20,000 transients. The raw data (FIDs) were Fourier transformed after zero filling by a factor of two and multiplied using an exponential window function with a line broadening (LB) of 0.3 (for ¹H spectra at 800 MHz), 0.5 (for ¹H spectra at 600 MHz), 0.5 (for ³¹P spectra of blood at 600 MHz) or 1.0 Hz (for ³¹P spectra of standard compounds at 600 MHz). Separately, ³¹P data of blood were processed after applying a Gaussian broadening function with a Gaussian broadening (GB) = 0.35 and LB = −1.0 or −1.5.

To confirm the newly identified metabolite peaks in blood spectra, ¹H and ³¹P spectra were obtained after the addition of the stock solutions (1 to 10 μL; 1 or 50 mM) containing the authentic compounds, individually.

Homonuclear two-dimensional (2D) experiments such as ¹H-¹H correlation spectroscopy (COSY) using the pulse sequence ‘cosygpqf’ or ‘cosyprgpqf’, ¹H-¹H total correlation spectroscopy (TOCSY) experiments using the pulse sequence ‘mlevphpr’, and ¹H-¹H exchange spectroscopy (EXSY) using the pulse sequence ‘noesyphpr’, were performed for blood as well as some standard compounds. 2D experiments were performed with or without suppression of the residual water signal by presaturation during the relaxation delay. A sweep width of 6009 Hz (at 600 MHz) or 9615 Hz (at 800 MHz) was used in both dimensions; 450 or 512 FIDs were obtained with t₁ increments, each with 2048 complex data points. The number of transients used varied from 16 to 128 and the relaxation delay varied from 1.0 to 2.0 s.

2D ¹H-³¹P heteronuclear single quantum correlation (HSQC) experiment using the pulse sequence ‘hsqcetgpsisp2.2’ was performed for standards as well as blood samples. In addition, the 2D ¹H-³¹P HSQC-TOCSY experiment using the pulse sequence ‘hsqcgpmlph’ and ³¹P-³¹P COSY experiment using the pulse sequence ‘cosygpqf’ were performed for blood samples. Spectral widths 6009 Hz (¹H) and 8741 Hz (³¹P) for HSQC and 6493 Hz (¹H) and 4800 Hz (³¹P) for HSQC-TOCSY, 256 t₁ increments each with 2048 complex data points, 32 to 256 transients, and 1.0 or 1.5 s relaxation delay were used. Parameters used for the ³¹P-³¹P COSY experiment were as follows: 5081 Hz or 8741 Hz spectral widths in both dimensions; 512 t₁ increments each with 4096 complex data points; 128 or 256 transients and 1.0 s relaxation delay.

2D data were zero-filled to 4096 points in the t₂ dimension and 1024 points in the t₁ dimension. A 90° shifted squared sine-bell window function was applied to both dimensions except for the COSY, before Fourier transformation. For the COSY experiments, the data were multiplied by an exponential window function with a line broadening of 0.3 Hz in t₂ and unshifted sine-bell window function in t_1, and processed in a magnitude mode. The chemical shift scale was referenced to the internal inorganic phosphate peak (for ³¹P) or TSP peak (for ¹H) for both 1D and 2D spectra. Bruker Topspin version 4.1.4 or 3.6.5 software package was used for NMR data acquisition, processing, and analyses.

Unknown identification and metabolite quantification:

Unknown phosphorus metabolite peak identification in 1D ³¹P spectra of blood used the strategy that we have used for expanding the limit of blood metabolite analysis using 1D ¹H NMR spectroscopy.^11–15 Briefly, it involved (a) developing and using a combined ¹H and ³¹P 1D and 2D spectral database of authentic compounds obtained under conditions similar to blood samples; (b) mapping phosphorus metabolites in the central carbon metabolism pathway; (c) comprehensively analyzing a variety of 1D and 2D ¹H/³¹P NMR spectra of blood; and (d) obtaining blood spectra before and after spiking with authentic compounds. The identified metabolites were quantified using the PULCON (pulse length-based concentration determination) method, which does not require an internal standard.²⁴ PULCON works based on the principle of reciprocity^25–27 and allows the correlation of signal strength from a reference spectrum with the spectrum of interest. The 1D ³¹P spectrum of sodium phosphate obtained after probe tuning, matching, and 90-degree pulse calibration was used as an external reference for the quantitation.

RESULTS AND DISCUSSION

This study describes development of a 1D ³¹P NMR-based method for analysis of phosphorus metabolites in human blood. The establishment of the method involved a comprehensive analysis of an array of 1D and 2D ¹H/³¹P homonuclear and heteronuclear NMR spectra obtained at 600 and 800 MHz in combination with a spectral database of standard compounds. A one-time comprehensive analysis of human blood spectra resulted in the establishment of the identity for nearly 25 phosphorus metabolites from a single 1D NMR spectrum (Figure 1, Table 1 and Table S2). Metabolites such as F16BP and G16BP as well as 2-PG and 3-PG that are indistinguishable by MS can be identified unambiguously using ³¹P NMR. The results demonstrate that a simple 1D ³¹P NMR spectrum offers an ability to visualize the status of intermediates and products of multiple metabolic pathways including glycolysis and the PPP, in one step. The method represents a new dimension in blood metabolite analysis.

Figure 1.

1D ³¹P NMR spectrum (¹H-decoupled; 600 MHz) of human blood obtained after metabolite extraction (see Figure S48 for expanded regions with peak annotations for all 24 identified metabolites and Tables 1 and S2 for full names and chemical shifts of the metabolites).

Table 1:

Phosphorus metabolites in human blood analysed using 1D ³¹P NMR spectroscopy^*

Glycolysis Pathway	Energy Coenzymes/Metabolites	Redox Coenzymes	Others
α-D-Glucose 6-phosphate (G6P)	Adenosine triphosphate (ATP)	Nicotinamide adenine dinucleotide, reduced (NADH)	Glycero-3-phosphocholine (GPC)
Fructose 1,6-bisphosphate (F16BP)	Adenosine diphosphate (ADP)	Nicotinamide adenine dinucleotide, oxidized (NAD+)	Phosphorylethanolamine (PE)
Dihydroxyacetone phosphate (DHAP)	Adenosine monophosphate (AMP)	Nicotinamide adenine dinucleotide phosphate, reduced (NADPH)	Phosphocholine (PC)
Phosphoenolpyruvate (PEP)	Phosphocreatine (PCr)	Nicotinamide adenine dinucleotide phosphate, oxidized (NADP+)	Uridine diphosphate glucose (UDP-Glu)
3-phosphoglycerate (3-PG)			Uridine diphosphate galactose (UDP-Gal)
2-phosphoglycerate (2-PG)			Inosine 5’-monophosphate (IMP)
2, 3-bisphosphoglycerate (2,3-BPG)			Uridine monophosphate (UMP)
Inorganic phosphate (Pi)			α-glucose 1,6-bisphosphate (G16BP)

^*see Table S2 for ³¹P NMR chemical shifts in human blood

The 1D ³¹P NMR spectrum of blood is complex due to peaks from a large number of hitherto unknown metabolites with concentrations ranging over three orders of magnitude (~ 1 to >1000 μM). While the establishment of peak identities for a few high concentration metabolites such as ATP, 2,3-BPG, and inorganic phosphate was relatively straightforward based on the literature,^20–22 for others, a more comprehensive analysis was needed. To achieve the goal, we first enlisted phosphorus metabolites based on their association with cellular metabolism including central carbon metabolism: glycolysis, the PPP, and TCA cycle. We then established a combined ³¹P and ¹H 1D/2D NMR spectral database of individual metabolites using the commercially procured standard compounds (Table S1) and acquiring spectra under the same conditions as used for blood specimens. ³¹P and ¹H NMR spectra were obtained at 600 MHz using sodium phosphate and TSP as internal chemical shift references, respectively. Ammonium bicarbonate buffer was used as the solvent to avoid solvent interference especially for the ³¹P spectra. ³¹P NMR spectra were obtained both without and with ¹H decoupling (inverse-gated) to aid peak assignments. All ³¹P 1D and 2D NMR spectra were obtained at 600 MHz since the 800 MHz instrument did not have a ³¹P probe. Separately, ¹H 1D and 2D NMR spectra for standard compounds were obtained at 800 MHz to aid peak assignments, in part, based on our recent efforts in blood metabolite analysis that have provided the identification of ~90 metabolites including several phosphorus metabolites.¹⁵

Broadly, phosphorus metabolites can be grouped into four categories (Table S1): (1) metabolites containing one phosphate group (mono phosphates); (2) metabolites containing two phosphate groups that are not attached consecutively (bisphosphates); (3) metabolites with two or three phosphate groups that are attached consecutively (di or triphosphates); and (4) metabolites that contain both mono and diphosphate groups. Figures S1 to S43 show ³¹P spectra (¹H coupled and ¹H decoupled; 600 MHz) as well as ¹H spectra (800 MHz) for standard compounds. In ¹H decoupled 1D ³¹P NMR spectra, monophosphates showed a single peak; bisphosphates showed two peaks; di or triphosphates showed two or three peaks, respectively, with each peak split due to ³¹P-³¹P J coupling, with the exception of NADH and NADPH, wherein the two consecutively connected phosphates showed a single peak (see Figures S29 and S31); and metabolites with both mono and diphosphate groups showed separate peaks, one peak for the monophosphate and one or more peaks for the diphosphate (eg. NADPH) (see Figures S30 and S31). Four compounds, DHAP, G6P, glucosamine 6-phosphate, and ribulose 5-phosphate contain a single phosphate group, however, all four showed two peaks, and are the exceptions (Figures S15, S19, S22, and S36). In aqueous solution at 20 °C, DHAP is reported to exist as a mixture of keto, gem-diol, and enolic forms in the ratio 55:44:1. The three forms freely interconvert and the ratio between the keto and gem-diol forms is temperature dependent, with increased proportion of keto form forming at higher temperature.²⁸ At 25 °C, we detected two forms, keto and gem-diol, in a 47:53 ratio in a 3.0 mM solution in ammonium bicarbonate buffer (pH = 7.8) and in a 57:43 ratio in a 0.3 mM solution in phosphate buffer (pH= 7.4) (Figure S44). It is the keto form that is enzymatically active in glycolysis.²⁹ The keto and gem-diol forms undergo interconversion as is evident from the chemical exchange between the two forms in the 2D exchange spectroscopy (Figure 2), which is in agreement with an earlier report.²⁸ The two peaks observed for the hexose compounds, G6P and glucosamine 6-phosphate, potentially arise from the two anomeric forms, α and β (Figures S19 and S22). It is, however, unclear why two peaks were observed for the pentose compound, ribulose 5-phosphate (Figure S36).

Figure 2.

(a) Keto and gem-diol forms of dihydroxyacetone phosphate (DHAP) and portions of (b) ¹H³¹P 2D HSQC (at 600 MHz) and (c) ¹H¹H 2D EXSY (at 800 MHz) spectra of DHAP highlighting the keto and gem-diol forms. The 2D NMR cross peaks in (c) indicate chemical exchange between keto and gem-diol forms.

Figure 3 and Figures S45–S47 highlight the comprehensive analysis of 1D and 2D ¹H/³¹P homonuclear and heteronuclear spectra of blood in combination with the spectral database of standard compounds, employed for unknown metabolite identification. Unknown peak identification in 1D ³¹P NMR spectra was partly aided by the recent advances in metabolite identification using ¹H NMR spectra, which included many phosphorus metabolites.¹⁵ The analyses led to the identification of most of the peaks in 1D ³¹P NMR spectrum of blood. The putative metabolite peaks were further confirmed by spiking experiments using the authentic compounds. Twenty-four phosphorus metabolites were thus positively identified. Figure S48 shows expanded regions of 1D ³¹P NMR spectrum of a typical human blood sample with peak annotations for all 24 metabolites. Given the high reproducibility of NMR, the peak annotated spectrum serves as a useful template for the routine analysis of phosphorus metabolites in blood using ³¹P NMR. Concentrations of the identified metabolites were measured using the PULCON method²⁴ and the external reference, sodium phosphate. Figure 4 and Table S3 show the metabolite concentrations thus obtained for a typical human blood.

Figure 3.

Portions of a typical 600 MHz ¹H³¹P 2D HSQC spectrum of human blood with highlighting of some of the identified metabolites. PEP: phosphoenolpyruvate; G16BP: α-glucose 1,6-bisphosphate; 2,3-BPG: 2,3-bisphosphoglycerate; 3-PG: 3-phosphoglycerate. ATP: adenosine triphosphate; ADP: adenosine diphosphate.

Figure 4.

Concentrations of phosphorus metabolites in a typical healthy adult human blood measured using 1D ³¹P NMR (at 600 MHz) (see also Table S3).

In NMR spectra, generally, peak positions are sensitive to numerous factors including the solvent, pH, temperature, metal ions, and sample concentration. The sensitivity is more pronounced for ³¹P NMR spectra compared to other commonly used nuclei such as ¹H and ¹³C. The higher sensitivity of ³¹P NMR peak positions potentially arises from the negatively charged phosphate groups. For example, a small change in sample concentration alone causes significant peak shifts and peak overlap (Figure S49). Further, binding of the inherent metal ions such as Ca²⁺ and Mg²⁺ to metabolites such as ATP causes peak to shift and broaden significantly (Figure S50). Addition of metal ion chelating agents such as EDTA (ethylenediaminetetraacetic acid) to blood sample or collecting blood in EDTA tubes, however, alleviates the metal ion induced peak shifts and broadening (Figure S50).^30,31 A challenge unconnected with any analytical method is the sensitivity of many metabolites to enzyme activity, oxidation or hydrolysis. To date, we have addressed several factors that affect the stability of such metabolites and provided optimized protocols for analysis of blood as well as tissue using ¹H NMR,^13,15,23 In the current study, we have employed this protocol for sample harvesting, extraction, and measurement to analyze the labile phosphorus metabolites. In view of the sensitivity of the peak positions to altered sample conditions, we have provided annotations for all identified metabolite peaks, apart from the chemical shifts (Table S2), to simplify peak identification for routine applications (Figures 1 and S48).

Phosphorus metabolites identified in human blood, in this study, represent intermediates and products of numerous metabolic pathways (Table 1). These pathways include glycolysis, the PP and TCA cycle, nucleotide metabolism, purine and pyrimidine metabolism, glycerolipid metabolism, and sucrose metabolism. Major redox coenzymes (NAD⁺, NADH, NADP⁺, NADPH) and energy coenzymes (ATP, ADP, AMP), which are fundamental to cellular function, are also part of the identified metabolites. Red blood cells (RBCs) constitute >99% of the cells present in blood and RBCs are devoid of mitochondria (and nuclei). Hence, glycolysis is the major pathway operative in RBCs. Interestingly, in the current study, our results highlight the detection of mainly the glycolysis pathway metabolites in central carbon metabolism (Table 1) and almost all the phosphorus metabolites that represent glycolysis pathway are detected (Figure 5). Overall, the ability to visualize the levels of phosphorus metabolites that represent the status of multiple metabolic pathways in one step using a simple 1D ³¹P NMR spectrum is significant considering the current challenges to analyze them using other methods.

Figure 5.

Central carbon metabolism highlighting phosphorus metabolites (green and red). Those highlighted in green are detected in blood by ³¹P NMR spectroscopy. In blood, red blood cells lack mitochondria and hence glycolysis is the major pathway operative in RBCs, which produces 2,3-BPG. 2,3-BPG plays a critical role in the transport of oxygen to the human body cells.

2,3-BPG is the most highly concentrated phosphorus metabolite in human blood. It is produced through the Luebering–Rapoport pathway of glycolysis,³² which is also known as the bisphosphoglycerate shunt. It is synthesized inside the RBCs from 1,3-BPG (Figure 5) and it plays a vital role in the respiratory system as it is critically required to regulate oxygen release from hemoglobin and delivery to all cells in the human body. The binding of 2,3-BPG to hemoglobin causes decreased hemoglobin affinity to oxygen and it facilitates the oxygen release at the cellular level. The next most highly concentrated phosphorus metabolite in blood is ATP. ATP is an energy coenzyme produced inside the RBCs by glycolysis pathway through the Embden–Meyerhof pathway.³³ ATP is required for RBC function in the transport and release of oxygen. Such vital roles of 2,3-BPG and ATP in the oxygen supply to all cells in the human body account for their high concentrations in blood. Their altered levels therefore have implications in human health and diseases.

Many of the phosphorus metabolites detected in blood in this study are highly unstable. Using NMR, it is possible to monitor whether the sample is deteriorating or has deteriorated since NMR detects both metabolites and their oxidized/hydrolyzed products, simultaneously and quantitatively. For example, a recent study from our laboratory describes the dynamics of some of these labile metabolites.³⁴ The study monitored conversion of one labile metabolite to the other through oxidation or hydrolysis, under a variety of preanalytical conditions, as a function of time. Importantly, the results provided compelling evidence that warrants a paradigm shift in the sample collection protocol for reliable analysis of labile metabolites. Sample harvesting, extraction, and analysis conditions are very stringent for analysis of labile metabolites. However, such protocols are not strictly followed in conventional metabolomics studies, which leads to incorrect metabolite levels and study outcomes. As a part of addressing this challenge, we have recently provided a simple strategy to gain insight into the levels of labile metabolites even when stringent protocols are not followed; we showed that the total pool of each group of unstable metabolites better represents their levels in biological specimens. Using this strategy, we have determined the whole-body distribution of some these labile metabolites in a mouse model.³⁵

In conclusion, in this study we describe the development of a simple 1D ³¹P NMR-based method for analysis of phosphorus metabolites in human blood. The method development involved a comprehensive analysis of human blood using a variety of homonuclear and heteronuclear 1D/2D NMR techniques in combination with the development of a spectral database of standard compounds. This study demonstrates the one-step analysis, including identification and quantitation, of nearly 25 metabolites, which is the highest number that has been analyzed to date using a simple 1D ³¹P NMR spectrum. The ability to analyze such a large number of phosphorus metabolites is significant considering the current challenges for their analysis using the conventional methods including highly sensitive MS. Importantly, metabolites such as F16BP and G16BP as well as 2-PG and 3-PG that are generally indistinguishable by MS could be analyzed individually using ³¹P NMR. Phosphorus metabolites occupy a unique position in cellular metabolism as they represent the intermediates and products of numerous metabolic pathways including central carbon metabolism. The metabolites analyzed in this study include redox coenzymes, energy coenzymes and represent pathways including glycolysis, the PPP, and TCA cycle that are fundamental to cellular function. Such an ability to visualize multiple pathways in one-step using a simple ³¹P NMR method opens a new dimension in blood metabolite analysis and is anticipated to greatly impact the blood metabolomics field.