A pioneer study on human 3‐nitropropionic acid intoxication: Contributions from metabolomics

The neurotoxin 3‐nitropropionic acid (3‐NPA) is an inhibitor of succinate dehydrogenase, an enzyme participating both in the citric acid cycle and the mitochondrial respiratory chain. In human intoxications, it produces symptoms such as vomiting and stomach ache in mild cases, and dystonia, coma, and sometimes death in severe cases. We report the results from a liquid chromatography‐Orbitrap mass spectrometry metabolomics study mapping the metabolic impacts of 3‐NPA intoxication in plasma, urine, and cerebrospinal fluid (CSF) samples of a Norwegian boy initially suspected to suffer from a mitochondrial disease. In addition to the identification of 3‐NPA, our findings included a large number of annotated/identified altered metabolites (80, 160, and 62 in plasma, urine, and CSF samples, respectively) belonging to different compound classes, for example, amino acids, fatty acids, and purines and pyrimidines. Our findings indicated protective mechanisms to attenuate the toxic effects of 3‐NPA (e.g., decreased oleamide), occurrence of increased oxidative stress in the patient (such as increased free fatty acids and hypoxanthine) and energy turbulence caused by the intoxication (e.g., increased succinate). To our knowledge, this is the first case of 3‐NPA intoxication reported in Norway and the first published metabolomics study of human 3‐NPA intoxication worldwide. The unexpected identification of 3‐NPA illustrates the importance for health care providers to consider intake‐related intoxications during diagnostic evaluations, treatment and follow‐up examinations for neurotoxicity and a wide range of metabolic derangements.

Typically, patients with mild cases of intoxication vomit, feel nauseous and exhausted, and have diarrhea, stomach ache and headache, and most recover after a few days (Ming, 1995). Severe cases have more serious symptoms, usually developing dystonia and coma (Ming, 1995). No known antidote for 3-NPA intoxication is available (Ming, 1995;Xin Chen et al., 2018). The toxin is widely used in phenotypic animal models of Huntington's disease (HD), as it produces neurological derangements and mitochondrial damage and alterations in rodents similar to that of patients with HD (Ayala-Peña, 2013; Brouillet et al., 2005;Cano et al., 2021;Wiprich et al., 2020). 3-NPA has also been used in a model of mitochondrial dysfunction and oxidative stress (Lahiani-Cohen et al., 2019).
Human intoxications have mostly been reported from China following ingestion of moldy sugarcanes infected by Arthrinium spp., producing 3-NPA in toxic amounts (Liu et al., 1992;Ming, 1995;Xin Chen et al., 2018). From its first report (1972), the cases roughly reached 884 in 17 years, with 88 fatalities (Liu et al., 1992). Outside of Asia, 3-NPA has been detected in trees, plants and fruit juice in, for example, New Zealand (MacAskill et al., 2015), Australia (Ossedryver et al., 2013), Serbia (Jani c Hajnal et al., 2020), Nigeria (Ayeni et al., 2020) and North America (Anderson et al., 2005;Darby et al., 2020;Liu et al., 2017), and has been reported to cause intoxication in animals. Contaminated stale coconut water with 3-NPA was identified as the cause of death in a Danish man (Birkelund et al., 2021). Individual variation in susceptibility to 3-NPA toxicity may be due to differences in age, genetics, environment, epigenetics, and overall health (Aldridge et al., 2003). For example, only one of two children who consumed the same sugarcane in 1995 manifested typical symptoms and clinical signs (Ming, 1995). The fact that different parts of a plant can be unequally affected by fungi and thereby have varying amounts of 3-NPA, may have contributed to this difference.
The research to date has tended to focus on biochemical changes and alterations in metabolite levels in exposed organisms using targeted approaches (Binienda & Kim, 1997;Olsen et al., 1999;Tadros et al., 2005). These hypothesis-testing procedures may provide analysis of a small number of specific metabolites, but fail to provide a more complete presentation of all the metabolic derangements (Kell & Oliver, 2004). This is particularly relevant in clinical settings when differences between two study groups (e.g., case and control) are only partly known or the cause of disease, and thereby the affected metabolic pathway(s) and metabolites, are unclear. Thus, untargeted approaches (so-called metabolomics, metabonomics, metabolic profiling, etc.) can better provide vital information to health care providers for diagnosis, prognosis, and the choice and efficacy of therapy (Ashrafian et al., 2020;Gowda et al., 2008). For regulatory toxicology and risk assessment, the utility of metabolomics has been discussed as a novel and efficient tool (Olesti et al., 2021).
Metabolomics is the study of low-molecular weight (typically <1500 Da) compounds present in a biological sample (Viant et al., 2017;Wishart, 2013). Liquid chromatography-mass spectrometry (LC-MS) is one of the most widely used analysis methods for metabolomics (Patti et al., 2012). Compared with nuclear magnetic resonance (NMR) spectroscopy, LC-MS provides more sensitive analyses (Allwood & Goodacre, 2010) as low-concentration compounds are more easily detected, especially when using high resolution, high sensitivity MS instruments (e.g., Oribtraps (Ghaste et al., 2016)).
So far, metabolic profiling has only been applied to three nonhuman models exposed to this neurotoxin (Chang et al., 2011;Novoselov et al., 2015;Tsang et al., 2009). Therefore, for the first time, we report the results from an LC-Orbitrap MS metabolomics study of a 14-year-old Norwegian boy severely intoxicated with 3-NPA.

| Patient, clinical details and ethics statement
The patient was a 14-year-old Caucasian boy with a previous history of unilateral iridocyclitis, eczema and bilateral Achilles tendinitis, and several years of mild abdominal pain and diarrhea. His abdominal symptoms had worsened for 4-5 months ahead of an acute illness episode with 24 h of progressive abdominal pain, vomiting and confusion. He was admitted to the emergency department with reduced consciousness and dystonic movements, but with stable respiration and circulation. Blood tests showed elevated lactate (14 mmol/L, reference range 0.5-2.2 mmol/L) and ammonia (97 μmol/L, reference range 10-50 μmol/L) and compensated metabolic acidosis with increased anion gap of 29.6 mmol/L. Cerebral and medullary MRI revealed symmetric acute infarctions in the putamen without other findings. The changes were suspicious of methanol intoxication, but no methanol was detected in his blood, and an organic acidemia or mitochondriopathy was thus considered. Treatment with high caloric glucose infusion, biotin, thiamine, riboflavin, carnitine, antibiotics, and antiviral treatment was initiated. The lactic acidosis reversed and ammonia normalized within a few hours. The subsequent days, his level of consciousness and dystonia were fluctuating, and periodically he could follow instructions with latency. On Day 4, he deteriorated to a comatose state. Cerebral MRI at this stage showed progression of T 2 weighted high signal intensity and lactate peaks in the dorsal striatum, as well as lesions in the brainstem, consistent with Leigh syndrome. Mitochondriopathy remained a relevant diagnosis, but thiamine transporter two deficiency and intoxication with metronidazole were also suggested.
Cerebral MRI on Day 6 showed no additional progression. Clinically, he gradually recovered from Day 10. He developed axonal polyneuropathy, possibly a critical illness neuropathy. Four weeks after hospital admittance, he could start rehabilitation at a specialist hospital for physical medicine and rehabilitation. After four more weeks, he was discharged to his home. After several months of fatigability and minor residual symptoms from his neuropathy, the patient has since been stable (for more than 2 years); that is, no clinical neurological sequelae and no similar episodes of acute illness.
As one of the tentative diagnoses was a mitochondrial disease, his samples were analyzed at our routine diagnostics laboratory and our laboratory for inborn errors of metabolism. Except from com- of some kind was also one of the initial tentative diagnoses, a full forensic toxicology screening of samples from the patient was also performed, but with no findings. Figure 1 gives an overview of the identification process performed in this study.

The Regional Ethics Committee assessed the project in line with
Norwegian laws and legislations, and concluded in decision 108825/2020 that the project did not need their approval. The Data Protection Officer at Oslo University Hospital (GDPR art. 37) assessed the project (20/20905) and confirmed that the processing of personal data was in line with GDPR art. 6 (1) (a) and art. 9 (2) (a), since written informed consent from the patient, now aged 17 years, was obtained.
The use of a urine sample from a healthy volunteer (written, informed consent) was approved by the Regional Committee for Medical and Health Research Ethics (case number: 173346).

| Samples, preparations and LC-MS analysis
Heparin plasma, urine and cerebrospinal fluid (CSF) samples were collected from the patient for analysis. Table 1 provides an overview of the specimens with sampling time points. Collected samples were stored at À20 C prior to preparation and analysis. A urine sample F I G U R E 1 Flow chart describing the process leading to the identification of 3-nitropropionic acid in the acute phase patient urine sample. Note: Created with BioRender.com from a healthy volunteer (control) was taken right before analysis. For establishment of retention time, accurate mass and fragmentation pattern, an aqueous 3-NPA standard solution was analyzed using the same methods (described below) as the patient and control samples.
For peak area comparison, two plasma samples taken the same day ("acute phase" Day 2) were compared with two plasma samples taken 9 and 12 months later ("follow-up phase"), respectively. Regarding urine samples, one from the acute phase (Day 2) was compared with a sample taken 12 months later (follow-up phase). For CSF, a sample taken during the acute phase (Day 2) was compared with a sample taken two days (45.5 h) later (Day 4) (subacute phase, during the second round of deterioration), since no CSF sample from the follow-up phase is available. The length of storage before analysis and the number of freeze-thaw cycles for each sample varied in this study. In short, samples were stored about 7, 10.5, or 19.5 months before metabolomics mapping and the number of freeze-thaw cycles was two for most samples.

| RESULTS
3.1 | Annotated, identified and altered metabolites 3-NPA was annotated in the patient urine sample from the acute phase using an untargeted analysis (metabolomics). This finding was verified by analysis of a 3-NPA reference standard, using the same metabolomics method, and supported by the acquired spectrum of the unknown peak in the GC-MS trace for organic acids. The level of metabolites detected in samples collected during three phases (Table 1) was compared to identify the metabolic changes.
Tables S1-S3 illustrate for each material the annotated or identified metabolites with a ratio ≤0.50 or ≥2.0 between samples from two different clinical phases. The numbers of such comparison were as follows: 80 (plasma), 160 (urine), and 62 (CSF). In plasma, most of these metabolites belonged to the groups of amino acids and related compounds, fatty acids, and glycolysis and citric acid cycle. In urine and CSF, the majority belonged to the group of amino acids and related compounds, glycolysis and citric acid cycle, and purines and pyrimidines and related compounds. Some compounds were altered in available phases for more than one sample material (Table 2). Among them, the six metabolites including 3-hydroxybutyrate, hypoxanthine, lactate, succinic anhydride, succinate and methylmalonate presented high ratio of acute as well as subacute to follow-up phase (i.e., 2.0-9.9 or ≥10) for all materials. However, the first three metabolites showed decreasing pattern (from ≥10 to 2.0-9.9) when considering subacute to follow-up phase. In contrast, theobromine had ≥10 times lower T A B L E 1 Overview of patient samples analyzed, with dates and sampling time points T A B L E 2 Metabolites with altered amounts between acute phase and subacute phase (a/S in plasma and CSF), acute phase and follow-up phase (a/F in plasma and urine) as well as subacute phase and follow-up phase (S/F in plasma) sam
As shown in Figure 2, several compounds in the glycolysis pathway and citric acid cycle were altered in amounts in the acute phase compared with the subacute phase in CSF samples and with the follow-up phase in both plasma and urine samples. Figure 3 depicts how the amount of 3-NPA rapidly decreased (86% decrease in peak area) from the first to the second plasma sample obtained as little as 10 h later during the acute phase (see Table 1 for sample ID). In the third sample, 35 h later (and in the following three samples), 3-NPA was no longer detectable. This indicates an intoxication in which the toxin is rapidly degraded and eliminated while the metabolic and toxic effects and damages linger on for a long time before biochemical and clinical recovery and normalization can occur. A similar pattern of rapidly decreasing amounts as that of 3-NPA was observed for, for example, propionate and lactate in plasma samples (Figure 3).

| Source of 3-NPA
A thorough investigation was done to find the source of 3-NPA intoxication, including interviews with the patient and family and observations in the home and surrounding neighborhoods. There were no positive results with respect to identification of the origin of the toxin.
The family lives in a relatively modern house and did not report any problems with moisture, and so forth. The neighborhood is typically suburban, with relatively modern detached houses. The family lives a typical active Norwegian life. Neither the environment nor the activities of the family gave any clues to an exposure to 3-NPA. The patient had not traveled outside Norway but had stayed at home. The patient's diet was found to be normal and similar to that of his family with no exotic ingredients or food imported from China or other Eastern regions. The family mostly bought their food in the local food store, with one exception: they frequently bought a type of flour in Denmark, but there was none left in the house or shop to analyze.
The patient took a nutritional supplement during the last months before he became ill. This supplement was analyzed with our metabolomics platform and there was no 3-NPA detected.

| DISCUSSION
This study mapped the metabolic impacts of 3-NPA intoxication in plasma, urine and CSF samples of a single patient. Prior to our analyses, a full forensic toxicology screening of collected samples during the acute phase revealed no indication of any exogenous toxin. This F I G U R E 2 Simplified representations of the glycolysis pathway and citric acid cycle, showing altered metabolites in plasma, urine and cerebrospinal fluid (CSF) samples. The subacute phase CSF sample was taken 2 days later than the acute phase. The follow-up samples were taken 9 and 12 months later than the acute phase samples for plasma, and 12 months later for urine samples illustrates how untargeted approaches enable unexpected findings when targeted ones with a pre-defined list of analytes fail to detect relevant metabolites that are not included in the panel.

| Comparative metabolomics studies of 3-NPA intoxication
To our knowledge, no other publications on metabolomics of 3-NPA toxicity in humans exist to compare our results with, and available reports on biochemical alterations are very few and limited. Reported analysis of blood, urine, and CSF specimens generally fell within the normal range (Liu et al., 1992;Ming, 1995;Xin Chen et al., 2018). One exception (Birkelund et al., 2021) highlighted metabolic acidosis along with elevated lactate, 3-hydroxybutyrate (as we, too, observed in plasma) and glucose in blood. An alternative source for comparison is patients with complex II deficiency, but our literature search resulted in no metabolomics research similar to ours.
These four amino acid amides were absent in our samples. In metabolic profiling of exposed rats, Chang et al. (2011) found 16 and 3 alterations in brain and plasma samples, respectively.

| Protective responses to 3-NPA intoxication
The level of 3-NPA rapidly decreased from plasma sample 1 to 2, indicating a quick excretion and/or metabolism/detoxification of the toxin. During and after this process, some protective mechanisms are naturally coupled to attenuate the toxic effects of 3-NPA. For example, 3-hydroxybutyrate was demonstrated to be neuroprotective in a 3-NPA HD mouse model, likely due to its bioenergetic effects (Lim et al., 2011). In the present study, 3-hydroxybutyrate was increased in all acute phase samples compared with the subacute phase (CSF) and the follow up-phase (plasma and urine). We also observed decreased C0-carnitine in the acute phase CSF sample versus the subacute phase sample. This might imply increased carnitine consumption due to compensatory mechanisms to maintain normal mitochondrial metabolism. The higher carnitine level in the subacute phase sample may represent a partial biochemical normalization, but is probably also explained by carnitine supplementation in-between the two sample time-points. In a plasma sample taken day two of the acute phase, prior to the carnitine treatment, the level of C0-carnitine was reported to be in the lower area of the reference range (13 μmol/L, reference range 10-49 μmol/L). In rat models, pre-treatment with carnitine provided a protective effect through enhanced mitochondrial metabolism of long chain free fatty acids, thus preventing 3-NPA induced inhibition of mitochondrial function and the resulting brain temperature decrease (Binienda, 2003), reduction of mortality and neuronal degeneration (Binienda et al., 2004), as well as protection against oxidative stress (Binienda & Ali, 2001)  Oxidative stress plays an essential role in the pathogenesis of many diseases, and it occurs when there is an imbalance between the generation of reactive oxygen (and nitrogen) species (ROS) and the antioxidant defense systems (Halliwell & Gutteridge, 2015). The administration of 3-NPA gave rise to formation of ROS (Olsen et al., 1999), contributing to 3-NPA's neurotoxicity. Increased amounts of total free fatty acids was seen in a rat model investigating different brain regions after 3-NPA injection (Binienda & Kim, 1997).
Elevated free fatty acids in plasma were shown to be associated with increased amounts of free radicals in plasma (Paolisso et al., 1996).
The current study found free fatty acids such as oleic acid, linoleic acid, and arachidonic acid increased in acute phase plasma samples, which could indicate increased oxidative stress in our patient. Moreover, taurine was increased in plasma and decreased in urine when comparing acute phase versus follow-up phase samples. The antioxidant effect of taurine was reported in pre-treated rats before 3-NPA exposure (Tadros et al., 2005). The occurrence of oxidative stress was further supported in our patient by increased amounts of hypoxanthine in all acute phase sample materials versus samples from the subacute (CSF) and follow-up (plasma and urine) phases. Elevated hypoxanthine is an indicator of hypoxia (Saugstad, 1988), and 3-NPA has been shown to cause histotoxic hypoxia (Hamilton & Gould, 1987). Hypoxia is known to result in oxidative stress by generating ROS (Abramov et al., 2007;Hernansanz-Agustín et al., 2014;Pialoux & Mounier, 2012).

| Energy turbulence caused by 3-NPA intoxication
Complex II inhibition, where 3-NPA directly targets, is biochemically characterized by grossly elevated succinate, often in combination with accumulation of other citric acid cycle metabolites (Bourgeois et al., 1992;Zieli nski et al., 2016). Hence, a large amount of this dicarboxylic acid was detected in all our sample materials taken during the acute phase. In addition to the presence of or elevated succinate in urine and brain (Brockmann et al., 2002;Rustin et al., 1997), respectively, high levels of lactate in plasma (Birch-Machin et al., 1996)  Repeatability using the same LC-MS method as in this study was mapped for a wide range of metabolites in our recent study (Skogvold et al., 2021), with retention time and peak area relative standard deviations of 0.1%-0.4% and 2%-10%, respectively. As such, we are confident that the highlighted differences in amounts of the metabolites reported in this work are not due to technical/ instrumental variation, but represent real, biological differences at the time of sampling.
Compared with other 3-NPA studies, the list of altered metabolites in all three sample materials of ours was more comprehensive (Chang et al., 2011;Novoselov et al., 2015;Tsang et al., 2009). Having an individual patient as the sole unit of observation for investigating the changes in the metabolome during different clinical phases, limits the statistical strength. However, the fact that the patient was his own control can also be considered a strength since it neutralizes inter-individual differences that would otherwise be found even in the best of matched controls and that would have introduced random differences and statistical noise.
Further research should map the global metabolic alterations in both complex II deficiency and HD patients compared with controls to investigate the compliance with findings reported in this study.
That will also increase the knowledge of the biochemical impacts of and possible treatment strategies for the two diseases.

| CONCLUSION
This is the first case of 3-NPA intoxication reported in Norway, and the first report of a full metabolomics mapping of human 3-NPA intoxication worldwide. Following the unexpected identification of 3-NPA in the urine of an acutely, seriously ill patient suspect of intoxication or mitochondrial disease, we performed metabolomics LC-Orbitrap MS studies of plasma, urine, and CSF samples to compare levels of metabolites during the acute phase and in the subacute phase with a second clinical deterioration (CSF) and in the long-term follow-up phase (plasma and urine). Our results revealed many changes not previously published. This is one of only a handful of cases of 3-NPA intoxications reported worldwide outside of China.
The unexpected identification of 3-NPA illustrates the importance for health care providers to consider intake-related intoxications during diagnostic evaluations, treatment and follow-up examinations for neurotoxicity and a wide range of metabolic derangements. As 3-NPA has been found to cause intoxications in both Norway and Denmark, the possibility of including 3-NPA in forensic toxicology screenings in Europe should be considered.

FUNDING INFORMATION
This research received no external funding.

DATA AVAILABILITY STATEMENT
The data underlying this article will be shared on reasonable request to the corresponding author.