Origin of breath isoprene in humans is revealed via multi-omic investigations

Plants, animals and humans metabolically produce volatile isoprene (C5H8). Humans continuously exhale isoprene and exhaled concentrations differ under various physio-metabolic and pathophysiological conditions. Yet unknown metabolic origin hinders isoprene to reach clinical practice as a biomarker. Screening 2000 individuals from consecutive mass-spectrometric studies, we herein identify five healthy German adults without exhaled isoprene. Whole exome sequencing in these adults reveals only one shared homozygous (European prevalence: <1%) IDI2 stop-gain mutation, which causes losses of enzyme active site and Mg2+–cofactor binding sites. Consequently, the conversion of isopentenyl diphosphate to dimethylallyl diphosphate (DMAPP) as part of the cholesterol metabolism is prevented in these adults. Targeted sequencing depicts that the IDI2 rs1044261 variant (p.Trp144Stop) is heterozygous in isoprene deficient blood-relatives and absent in unrelated isoprene normal adults. Wild-type IDI1 and cholesterol metabolism related serological parameters are normal in all adults. IDI2 determines isoprene production as only DMAPP sources isoprene and unlike plants, humans lack isoprene synthase and its enzyme homologue. Human IDI2 is expressed only in skeletal-myocellular peroxisomes and instant spikes in isoprene exhalation during muscle activity underpins its origin from muscular lipolytic cholesterol metabolism. Our findings translate isoprene as a clinically interpretable breath biomarker towards potential applications in human medicine.


Introduction
Isoprene (2-methyl 1,3-butadiene; C 5 H 8 ) is an omnipresent and the most abundant hemiterpene in our planet's atmosphere 1,2 , Isoprene is globally produced by vegetation and along with its chiral monoterpenes (e.g.alpha pinene) acts as predictor for ecosystem uxes 3 , forest emissions and drought response 4 , environmental pollution 5 , cloud chemistry 6 and climate change 7 .Besides being abundantly emitted via the de novo emission pathway in plants 8 , isoprene is also endogenously produced by animals and humans.
The C 5 isoprene unit is the basic building block for terpenoid including steroid hormones biosynthesis that has been biologically essential for ubiquitous terrestrial life forms since billions of years in the course of evolution 9 .It is the second most abundant and endogenous volatile organic compound (VOC) in our exhaled breath 10 .Exhaled concentrations range between 100-300 ppbV in healthy adults.Though the evolutionary signi cance of isoprene production (and biological function) is not well understood in humans.differences/dynamics in its exhaled alveolar concentrations are well reported as potential markers for various physiological, metabolic and pathophysiological effects.While any kind of muscle activity/exercise immediately spikes its expression 11,12 , studies have reported differential breath pro les and signi cant concentration changes under conditions such as cardio-respiratory diseases [13][14][15] , hypercholesteremia 16 , oxidative stress 17 , cancers 18,19 , sexual arousal 20 and ageing [21][22][23] .Despite such interesting observations, the clinical translation of isoprene as a routine biomarker is hindered due to the uncertainty upon its exact metabolic origin.Surprisingly, while signi cant de ciency or absence of non-sterol C 5 H 8 is reported in cases with inherited diseases like mevalonic aciduria, hyper immunoglobulinemia D syndrome, autoin ammatory periodic fever [24][25][26] and Duchenne muscle dystrophy 27 , there are rare (< 0.3% population prevalence) adults living healthy lives without (absence) exhaling any traceable isoprene and exhaling very low (de ciency) concentrations of isoprene.Such breath isoprene aberrated healthy adults may hold the fundamental key to its true origin in human breath.
In 2005 Turner et al found one isoprene de cient healthy adult 28 .While compared to isoprene normal adults, no correlations were seen between isoprene exhalation and fasting blood cholesterol pro les.In 2021, we executed breathomics, lipid pro ling and gene expression analyses in a isoprene absent rare German adult and her isoprene de cient parents and sibling sister.Outcomes depicted no aberration in cholesterol levels and/or in gene expression of the mevalonate pathway enzymes and indicated a recessive inheritance of this healthy character 29 .Therefore, we questioned the putative human origin (hepatic cholesterogenesis) of exhaled isoprene that was proposed (in 1984 by Deneris et al) based on in vitro experiments in rat liver cells 30 .Nevertheless, a single rare case insu cient for detailed down-stream multi-omic analysis to determine the exact source.
In a recent case study on one isoprene de cient American adult male and his blood-relatives by Harshman et al 31 and in another study on an isoprene de cient Italian adult female and her bloodrelatives by Biagini et al 32 also demonstrated no relation of exhaled isoprene pro les to plasma cholesterol levels.They neither nd any isoprene absent rare adult nor investigated the human cholesterol metabolism related gene expressions.
Recently, by conducting untargeted breathomics in clinical screening scenarios, we discovered four more isoprene absent healthy German adults.Thus, to determine the exact metabolic origin, we have now performed multi-omic (whole exome sequencing, breathomics and relevant serological analyses) investigations in these ve (amongst 2000 recruited subjects) rare adults along with targeted sequencing of lead variants in blood-related (isoprene de cient) and unrelated (isoprene normal) healthy German adults.Our ndings will be able to denominate the principal human origin of isoprene and translate this VOC as a breath biomarker towards various clinical applications.
Exhaled alveolar isoprene concentrations (corresponding room air subtracted) in rare adult-1's father (aged 60 years, German), mother (aged 60 years, German) and sibling sister (aged 30 years, German) were 15.86 ppbV, 17.54 ppbV and 27.24 ppbV, respectively.These three adults were isoprene de cient and healthy.Previous serological investigations showed normal lipid pro les in them 29 .Demographic data (age, gender, ethnic origin, health status and lifestyle habits), relevant serological parameters (viz.plasma lipid pro le, bile metabolites and sex hormones) and exhaled concentrations of prime endogenous and exogenous VOCs from the ve isoprene absent adults are presented in Table 1.
No considerable serological aberration was found in the rare adults.Exhaled endogenous VOCs (except isoprene) were also within the expected normal ranges.Exhaled exogenous VOCs were mainly related to lifestyle habits.
Figure 2 represents the ltering strategies for the identi cation of candidate mutations following exome sequencing of isoprene-absent healthy German adults (n = 5).The ltering strategy for the detection of rare homozygous deleterious variants shared by all individuals are presented in 2 (A).Over 63,000 variants passed the upstream bioinformatics pipeline to secure su cient data quality and were present in at least one of the ve isoprene-absent adults.As a recessive mode of inheritance was suspected, we next ltered for homozygous variants, resulting in roughly 25,000 remaining candidate variants.To justify biological signi cance, we only included mutations that resulted in changes on amino acid level, including frameshift, missense, nonsense or in-del variants, lowering the number of potential candidates to around 9000.Knowing roughly the frequency of the investigated character (5 homozygotes out of 2000), we aimed to exclude variants with a reported population frequency of 15% or higher, being tantamount to roughly 1% homozygotes.Finally, out of those 556 rare homozygous variants, only one (IDI2, c.431G > A) was shared by all ve isoprene-absent adults.The detected mutation is listed in dbSNP (rs1044261) but no association towards isoprene metabolism was mentioned so far.
We further investigated if there are any heterozygous variants located within the mevalonate arm of cholesterol biosynthesis and steroid hormone metabolism pathway that might be related to the character (Fig. 2B).Again, out of the roughly 63,000 variants that passed upstream quality assessment, over 25,000 mutations resulted in frameshift, in-del, nonsense or missense mutations.We then analyzed the pathway genes ACAT2, HMGCS1, HMGCR, MVK, PMVK, MVD, IDI1, IDI2, FDPS1, GGPS1, FDFT1, SQLE, LSS and DHCR7, resulting in 12 variants that are present in at least one of the isoprene-absent adults.However, no additional mutation was shared by all cases, except for an FDFT1 variant that is very common in the European population and can therefore be excluded.Corresponding amino acids, variants, impact, variant allele frequencies and prevalence (%) of homozygous and heterozygous variants in the European population (based on gnomAD) are presented in Table 1.
We next aimed to investigate the inheritance pattern of the IDI2 c.431G > A variant.Therefore, con rmatory targeted bidirectional Sanger sequencing was performed for all isoprene absent adults (Fig. 3).As expected, the results demonstrated a homozygous mutation in the respective position in all specimens, con rming the previous whole exome sequencing ndings.Family members of isoprene absent adult-1, who demonstrated < 30 ppbV of isoprene in exhaled breath, were also checked, revealing a heterozygous genotype in her mother, father and sibling sister.Two unrelated healthy adults with normal isoprene pro les were sequenced as controls.Those control participants did not show the IDI2 c.431G > A variant, neither homozygous nor heterozygous.
We nally interrogated the UCSC Genome Browser and UniProt databases to elucidate the biological signi cance of the detected IDI2 c.431G > A variant.The IDI2 gene is located on the short arm of chromosome 10 and consists of ve exons, with coding sequences in exons two to ve (Fig. 4A).The mutation of interest is located at the beginning of exon ve and results in a stop gain at p.W144*, accounting for a truncated mRNA (Fig. 4B).This genomic area is highly conserved, indicating a crucial and ubiquitous role of the transcriptional site.
The wildtype IDI2 protein spans 227 amino acids (aa), with a large hydrolase domain spanning aa 49 to 199 (Fig. 4C).There are two active sites of the enzyme, one at the N-terminus of the hydrolase domain and the other one at aa 148, only four aa downstream from the detected p.W144* variant (Fig. 4D).Due to the truncated mRNA transcript, the second enzyme active site is deleted in the ve individuals sharing the c.431G > A mutation, likely resulting in impaired or absent protein function.In addition, the loss of the C-terminal part of IDI2 exon 5 also results in the deletion of two out of four magnesium binding sites which is a required cofactor for the enzyme's function.

Discussion
For the last 39 years, we have erroneously regarded hepatic cholesterogenesis (producing > 90% of human cholesterol) as the prime origin of human exhaled isoprene.This believe was based on in vitro synthesis of isoprene from DL-mevalonate by utilizing a rat liver cytosolic fraction 30 .Consequently, various physio-metabolic and clinical conditions driven interesting differences (cross-sectional) and/or changes (longitudinal) in isoprene exhalation could not be explained via the well-known/established effects of those conditions on hepatic cholesterogenesis.As a result, breath isoprene could not step into routine clinical practice as a noninvasive biomarker.In 2021, we challenged the putative origin of breath isoprene 29 and here, we have discovered the actual origin of human exhaled C 5 H 8 by multi-omic analysis of genes and metabolites.
Distribution of exhaled isoprene concentrations from 2000 screened subjects recon rmed previously reported age dependency [21][22][23]29 of its exhalation. Compete absence of exhaled isoprene in the rare adults is caused by the shared homozygous IDI2 variant (stop-gain mutation at c.431 position) driven functional aberrations of enzyme active site and metal-cofactor binding sites.Looking at the isoprene exhalation in our previous study 29 , we assumed that the inheritance of the character (isoprene absence) has a recessive trait.Here, the heterozygous presence of the IDI2 c.431G > A variant in the isoprene de cient healthy parents and sibling sister of rare adult-1 and absence of this IDI2 mutation in unrelated healthy adults (isoprene normal) genetically con rmed our previous assumptions.
In this study, the prevalence (< 0.25%) of isoprene absent adults closely mirrored the actual homozygous prevalence (0.23%) of the IDI2 variant in the EU population whereas, the occurrence of isoprene de cient adults (64/1318 i.e. 4.8% among adults) did not closely mirror the actual heterozygous prevalence (6.95%) of the IDI2 variant in the EU population.This is mainly due to the fact that our clinical screening was not restricted to European subjects.The overall expression (homozygous and heterozygous) of the mutated IDI2 is different in other ethnic origins (Table S2) and therefore, the observed age distributions of isoprene, de ciency and/or absence may differ amongst another ethnicity/population. Besides, the cutoff limit of isoprene de ciency in adults was set by us to < 50 ppbV in this study.Exhaled isoprene concentrations can uctuate by 5-25 ppbV in an individual simply due to his/her normal physiological variations in respiratory and hemodynamic parameters, natural menstrual rhythms and/or oral contraception and menopause 21,[33][34][35] etc.
While the expression of human IDI1 is conserved in various tissues and high within the mitochondria and proteasome of the hepatocytes, its divergent isoform IDI2 is highly expressed only within the peroxisome of the skeletal myocytes 36,37 .Human IDI1 is poorly expressed in skeletal muscle.IDI1 and/or IDI2 catalyze the isomerization of isopentenyl diphosphate [(C 14 )IPP] to the highly nucleophilic dimethylallyl diphosphate [(C 14 )DMAPP].In humans, the conversion of IPP to DMAPP takes place in two metabolic paths -during cholesterol biosynthesis in the endoplasmic reticulum of hepatocytes and during lipid catabolism (involving cholesterol metabolism) in the peroxisome of skeletal myocytes 38,39 .Only DMAPP (not IPP) is converted to C 5 H 8 via isoprene synthase (IspS) enzyme in plants [40][41][42] .As humans do not have isoprene synthase and bioinformatic sequence alignment (whole exome/functional domain based) via BLAST search tool 43 did not locate human enzyme homologs of isoprene synthase, the wild type IDI1 and IDI2 genes (and related proteins) should serve as the determinant for human isoprene production from those two aforementioned metabolic paths.
Besides humans, while looking at other terrestrial and marine mammals, we observed interesting facts upon breath isoprene.In a recent pre-clinical study, mass-spectrometry based untargeted pro ling of exhaled VOCs in spontaneously breathing awake healthy and/or in uenza A virus infected pigs, we could not detect breath isoprene 44 .On the other hand, in pre-clinical breathomic studies on goats and on cattle, we observed signi cant concentrations of breath isoprene from both ruminants 45,46 .Via massspectrometry based comprehensive screening of exhaled metabolites from bottlenose dolphins, Aksenoy et al did not detect any trace of isoprene 47 .Our present search in the Ensembl genome database 48 and EMBL-EBI resource 49 showed that IDI2 is not at all expressed in pigs and in bottlenose dolphins but is well expressed in goats and cattle, underlining functional IDI2 as discriminator between isoprene presence and absence.IDI1 is ubiquitously expressed in many tissues in all these animals.Previously, via microextraction-coupled mass-spectrometric measurements of headspace of arterial and venous blood samples collected from mechanically ventilated humans and pigs, we observed extremely low (up to 10-fold lower than in human) isoprene concentrations within the portal and mixed venous blood of pigs 50 .Such tiny fraction may be washed-out (i.e.stored previously) and/or produced via minimal IDI1 activity in the peripheral compartments.Nevertheless, as soon as the blood crossed the hepatic circulation, isoprene concentrations were diminished in hepatic venous samples -most likely due to a high rate of isoprene metabolism in pig liver.Similar to pigs and dolphins, only the IDI1 is expressed in rats and mainly within the liver.Most likely, due to a low (compared to higher mammals) isoprene oxidation rate in rat liver 51 , Deneris et al had detected a certain fraction of isoprene in rat liver cytosol in vitro.They suggested that isoprene could be produced in rat liver via non-enzymatic degradation of IPP and/or DMAPP and postulated in general that breath isoprene is linked to hepatic cholesterogenesis 30 .
While the pre-clinical nding of Deneris et al was correct, the general inference drawn on the origin of human breath isoprene based on those outcomes from rats was wrong.In human hepatocellular microsomes, the complex cytochrome P450 enzyme system immediately oxidizes isoprene and isoprene monoepoxides to avoid hemiterpene toxicity 52 .The oxidation rate in human liver microsomes is magnitudes higher compared to rats 51 .Due to such high isoprene oxidation rate, hepatic cholesterogenesis is insu cient to contribute any considerable concentration of isoprene to human exhalation.
In the present study, despite normal plasma lipid pro les, bile substrates, sex-hormones and wild type IDI1 in rare adults and blood-related family members, signi cant aberrations in their isoprene exhalations con rmed our previously suggested 29 independence of human exhaled isoprene from hepatic cholesterogenesis and related principal pathways.Therefore, due to the absence of functional IDI2 expressions in these isoprene aberrated adults, the source of human breath isoprene should mainly be directly attributed to muscular metabolic activity and not to hepatic cholesterol biosynthesis.
Skeletal muscles represent around 40% of adult human body mass and predominantly utilize glucose and lipids to produce energy, regulate intramyocellular signaling and integrity 53,54 .Insulin governs the balance between glucose and fatty acid metabolism in muscle 55 and peroxisomal beta-oxidation senses intracellular fatty acids and regulates lipolysis 56 .Besides mitochondrial oxidation, peroxisomal betaoxidation of very long-chain fatty acids, long-chain fatty acids and dicarboxylic acids produces acetyl-CoA.Acetyl-CoA is channeled towards farnesyl diphosphate (farnesyl-PP) production inside the peroxisomes 39 .All enzymes (except 3-hydroxy-3-methylglutaryl-CoA reductase/HMGCR) step-wise converting acetyl-CoA to farnesyl-PP contain functional peroxisomal targeting signals and at the second last step of this pathway, IPP is converted to DMAPP via the IDI2 enzyme as only IDI2 is highly expressed here.Farnesyl-PP exits the peroxisomes to execute various metabolic processes in other cellular organelles.
Any kind of muscle movement/activity immediately gives rise to breath isoprene 11,57 .Due to its low aqueous solubility and high volatility, isoprene is positively related to cardiac output and negatively related to minute ventilation 58,59 .Both low-intensity and exhaustive exercise demonstrated an instant and profound increase in exhaled alveolar isoprene concentrations at the initial warm-up phase (that increases muscle perfusion) followed by gradual decrease with increasing work-load, which indicates its possible production and washout from the active muscle compartments 12,57,60 .Exercise immediately increases skeletal muscular lipolysis, fatty acid transport from plasma to sarcoplasm and triglyceride hydrolysis to compensate energy demand.Thus, our present ndings ascertain that isoprene is potentially originating from lipolysis in the skeletal muscle and wild type IDI2 denominates the presence of isoprene in exhaled human breath and also acts as the rate limiting factor for endogenous isoprene production.
Although we were able to detect the IDI2 mutation in PBMC, IDI2 gene and protein expression are skeletal muscle speci c.Previously we observed IDI1 but not IDI2 gene expression in PBMC of healthy adults with absence, de ciency and normal breath isoprene 29 .To assess the biological consequences of the IDI2 variant, gene and protein expression studies would require the collection of muscle biopsies of the affected adults.Those investigations are, however, behind the scope of the present study and limited by ethical considerations.
We discovered the genetic origin of human breath isoprene production and related biochemical routes.
The rare character of isoprene absence/de ciency in healthy human adults is autosomal (10th chromosome, locus: 10p15.3)recessive.We translated isoprene as the rst breath VOC biomarker with well-de ned down-stream endogenous origin and metabolic pathways.This knowledge will rede ne the clinical interpretations of this noninvasive biomarker for various physio-metabolic, pathophysiological and inherited conditions.We assume that the presence of DMAPP may not be essential for healthy human life as all principal pathways converting acetyl-CoA to farnesyl-PP are associated with lipid and cholesterol metabolism can utilize IPP to execute them normally.Already reported endocrine regulation 35 and age dependency 21,22,29 of isoprene exhalation indicates new research scopes of IDI2 activity in human aging, muscle mass development and related conditions.Similarly, further investigation of IDI2 gene and protein expressions in skeletal muscle tissue along with breath isoprene expressions under various exercise trainings and in individuals with muscle dystrophy and risk of rhabdomyolysis e.g.under statin interventions or injury may reveal unexplored frontiers in sports/ tness and musculoskeletal medicine and inter-organ metabolic cross-talk.
In order to understand the evolutionary signi cance of human IDI2 gene and the presence/rationales of isoprene production pathway in human, we need further multi-omic based system-wide evaluation from the genome to up-stream cascades.Tissue speci c gene and protein expression followed by transcriptomics and proteomics may lead us to the actual enzymatic and metabolic signi cances (DMAPP to isoprene) that are linked to the last IDI2 exaptation taken place ~ 70 million years ago for an unknown function.Inducible IDI2 de ciency in animal models could also shed light on the kinetics and physiological background of isoprene metabolism.Further investigations should screen a large number of isoprene de cient subjects for IDI2 allele frequencies to realize its exact genetic correlation(/predisposition) factor with breath isoprene expressions.In order to apply isoprene or any other endogenous biomarker to routine clinical practice, well-de ned down-stream origins and pathways should be addressed.

Experimental Design
Based on our previously observed rare character of exhaled isoprene aberrations (absence and de ciency) in healthy adults 29 , we aimed to further investigate the actual reason(s) at the very downstream level and thereby, to discover the principal human origin of this endogenous hemiterpene.
In adults, exhaled alveolar isoprene concentrations of 100-300 ppbV is regarded as the normal range and ± 50 ppbV is regarded as the limit of normal physio-metabolic uctuations.Exhaled alveolar concentration of 00 ppbV is regarded as isoprene absence and < 50 ppbV is considered as signi cant isoprene de ciency.In adults, exhaled isoprene concentrations between.
In order to nd adults with isoprene aberrations, we re-evaluated the isoprene exhalations in 2000 human subjects (aged between < 1-100 years) from 15 consecutive clinical breath screening studies by applying real-time mass-spectrometry (proton transfer reaction -time of ight -mass spectrometry / PTR-ToF-MS).All studies were conducted at the University Medicine Rostock, 18057 Rostock, Germany in accordance with the amended Declaration of Helsinki guidelines.Ethical approvals from the Institutional Ethics Committee (IEC, University Medicine Rostock, Rostock, Germany) and signed informed consents from all participants were obtained prior to participation.In case of infants, both parents provided signed consents.Within these studies, healthy volunteers were recruited during physiological, metabolic, exercise and dietary/nutritional monitoring (Ethical approval numbers: A2011-67, A2012-0103, A2014-0037, A2015-0008, A2015-0076, A2018-0025 and A2020-0300) and both healthy and sick subjects were recruited during pathophysiological and/or therapeutic monitoring (Ethical approval numbers: A2012-0103, A2012-0071, A2015-0043, A2017-0106, A2018-0097, A2019-0040, A2020-0085 and A2021-0012).In the present study, ethical approval no.A2021-0012 is assigned to the investigations involving venous blood collection, peripheral blood mononuclear cells isolation, DNA isolation, multi-omics (breathomics, untargeted and targeted genomics) and serological metabolites analysis in healthy isoprene aberrated and isoprene normal adults.
After nding the isoprene absent rare adults, we executed multi-omic investigations of shared downstream aberrations in genes and relevant metabolites among such adults, in blood-relatives and in unrelated healthy controls.Therefore, at rst, we conducted whole exome sequencing to identify unknown homozygous variants shared by the rare adults.Then the shared mutations were checked via bidirectional Sanger sequencing in blood-related (isoprene de cient) and unrelated (isoprene normal) healthy adults.Plasma lipids, metabolites and hormones related to cholesterol metabolism were also checked serologically in all these subjects.Due to ethical reasons, we were allowed to conduct genomics only in adults.Amongst the isoprene absent rare adults, we could only access the blood-related adult family members (parents and sibling sister) of the rare adult-1.

Breath Sampling, Voc Data Analysis And Quanti cation
All spontaneously breathing subjects maintained de ned posture 34 and performed oral breathing 33 via customized mouthpiece 61 or mask by following our state-of-the-are sampling protocol 59 .Continuous side-stream sampling ( ow: 20-100 ml/min) from the mouthpiece or mask were performed via the heated (75-100°C) transfer-line of a PTR-ToF-MS-8000 or a PTR-ToF-MS-1000 (Ionicon Analytik GmbH, Innsbruck, Austria) under pre-optimized experimental conditions 62,63 .Most importantly, PTR timeresolution of 200 ms, drift-tube temperature of 75°C, voltage of 610 V and pressure of 2.3 mbar were used to reach the optimal E/N ratio of 139 Td 21,29,64 .After automatic recording of a data le/min the mass scale was recalibrated based on masses namely, 21.0226 (H 3 O + -isotope), 29.998 (NO + ) and 59.049 We used a PTR-MS viewer software (version 3.228) to process raw data.VOC data were measured continuously in counts per second (cps).Measured VOC counts were normalized to primary ion (H 3 O + ) counts.Breath-resolved assignment of expiratory (alveolar/end-tidal) and inspiratory (room air) phases were executed via custom-made 'breath tracker' algorithm 65,66 .Here, we used an endogenous VOC (e.g.acetone) with orders of magnitudes higher concentration in exhalation than in room air as the tracker mass.
Measured VOCs were quanti ed either via reaction rate coe cients (k-rates) between VOC and primary ion (at the E/N ratio of 140 Th) or via multi-component VOC standard mixture under matrix adapted conditions (breath humidity) by using a liquid calibration unit (LCU, Ionicon Analytik GmbH, Innsbruck, Austria) 64,67 .Isoprene was quanti ed via LCU based calibrations.

Venous Blood Sampling
A total of 50 ml of antecubital venous blood was collected from each subject by a skilled and licensed physician for serological and genetic analysis.

Plasma Metabolite Analysis
Plasma lipids (total cholesterol, lipoproteins a, high density lipoproteins (HDL), low density lipoproteins (LDL) and triglycerides), bile substrates (total-, direct-and indirect bilirubin) and sex-hormones (estrogen, progesterone and testosterone) were analyzed at the central laboratory (Institute of Clinical Chemistry and Laboratory Medicine, University Medicine Rostock) via conventional methods (listed in Table 1).

Isolation Of Peripheral Blood Mononuclear Cells
A total of 45 µl of venous blood was collected from all participants and mixed 1:1 with pre-warmed PBS (PANbiotech, Aidenbach, Germany).Density gradient centrifugation (1,200 x g, 12 min, 4°C, brake 0) using PAN-coll separation solution (PANbiotech) was carried out to separate peripheral blood mononuclear cells (PBMC).Isolated PBMC populations were washed twice (180 x g, 10 min) in PBS and cell numbers were determined.Cell pellets were stored at -80°C for subsequent analyses.

Exome Sequencing
Cell pellets were thawed and genomic DNA was isolated using the NucleoSpin® Tissue kit (Macherey-Nagel, Düren, Germany) according to the manufacturer's instructions.Samples were sequenced via INVIEW Human Exome sequencing service (Euro ns Genomics, Ebersberg, Germany).The QIAGEN Clinical Insight Interpret platform (Qiagen, Hilden, Germany) was used to identify rare and unknown variants, using the following ltering criteria: only variants in exonic regions or variants that are ± 20 bp anking those regions; variants pass upstream pipeline ltering; variants with call quality ≥ 20; variants with allele fraction ≥ 90; variants that result in frameshift or in-frame indel or start/stop codon change or missense or nullizygous; variants with homozygous population frequency < 1% in Europeans or no population data available.

Bidirectional Sanger Sequencing
The PCR was carried in a nal volume of 25 µl containing 100 ng of genomic DNA, 0.4 µM of each primer (Euro ns Genomics), 250 µM of each dNTP (Agilent Technologies, Santa Clara, USA), 5 µl 5x Hi-Fi Reaction buffer (Meridian Bioscience, Cincinnati, OH, USA) and 0.5 µl VELOCITY™ DNA polymerase (Meridian Bioscience) using the following primers: F, AATTCTGTGTTTTACATTAGCGTTG; R, CTGGGACAGGTAGAGGATGCT.The PCR conditions were 2 min at 98°C followed by 35 cycles of 30 sec at 98°C, 30 sec at 57°C and 15 sec at 72°C with a nal step of 5 min at 72°C.The products were further processed via QIAquick PCR Puri cation Kit (Qiagen) according to the manufacturer's instructions.Puri ed PCR products were sequenced using the Mix2Seq sequencing service (Euro ns Genomics).

Statistical Analysis
Exhaled alveolar inspired room air isoprene concentrations from a minute of steady spontaneous breathing (with normal respiratory rate of 10-14 breaths/min) were considered for quanti cation.
Measurement was repeated three times in each subject.Due to non-parametric distributions, median values were used for statistical analysis.
Statistical signi cances of differences in isoprene concentrations between different age groups/subgroups were tested by means of Kruskal-Wallis one-way ANOVA on ranks (Kolmogorov-Smirnov test for normality followed by the multiple comparisons via post-hoc Dunn's method at p-value ≤ 0.005) in SigmaPlot version 14.From all pairwise-multiple comparisons, statistically signi cant differences with respect to 'isoprene normal are of study importance (please see Fig. 1).Detailed data on each group size (N), group median, difference in ranks and corresponding p-values are presented in Table S1.In order to avoid overlaps of exhaled isoprene concentrations from different age groups and/or subgroups, the 'total recruited subjects' was not compared statistically (please see Fig. 1).

Figure 1
Figure 1 represents the distribution of exhaled alveolar and inhaled room air concentrations (ppbV) of isoprene measured via real-time PTR-ToF-MS from 2000 recruited subjects.Exhaled isoprene concentrations are also presented within different age groups and subgroups.Statistically signi cant (p-

Figures
Figures

Figure 1 Distribution
Figure 1

Table 1 is
available in the Supplementary Files section.