Volatile organic compound profiling for tracking clinical Pseudomonas aeruginosa isolates Authors: M. Trinidad García-Barceló, Biel Martorell, Alexandre González-Alsina, Antonio Doménech-Sánchez, Rosa M. Gomila, Sebastián Albertí, Margalida Mateu-Borràs Published: 2026 ABSTRACT Pseudomonas aeruginosa is an opportunistic respiratory pathogen frequently affecting hospitalized patients and individuals with chronic lung diseases such as cystic fibrosis (CF), bronchiectasis, and chronic obstructive pulmonary disease (COPD). Current diagnostic methods for airway infection are often invasive or lack sensitivity, highlighting the need for non-invasive approaches. Volatile organic compounds (VOCs) released by microbes represent promising candidates for non-invasive detection of respiratory pathogens. In this study we characterized the volatilome of clinical P. aeruginosa isolates and identified candidate VOC markers associated with bacterial metabolism. An untargeted volatilomics workflow based on gas chromatography–mass spectrometry (GC–MS) was applied to profile the in vitro VOCs produced by 48 clinical P. aeruginosa isolates associated with acute and chronic respiratory infections. We identified three previously unrecognized VOCs consistently produced by all isolates, suggesting conserved metabolic features of P. aeruginosa. Comparative analysis revealed marked metabolic differences between infection phenotypes: chronic isolates produced higher levels of aliphatic volatiles, whereas acute isolates were enriched in aromatic compounds, particularly acetophenone. Longitudinally sampled clonally related isolates from CF patients exhibited distinct and evolving volatilomic signatures, indicating metabolic adaptation over the course of chronic infection. Because LasR dysfunction is common during chronic airway colonization, we further assessed its impact on VOC production. LasR-functional isolates produced elevated levels of sulfur-containing VOCs and specific ketones, while LasR-deficient isolates displayed increased alcohol production. Overall, our findings show that isolates from chronic infections are associated with a shift from aromatic to aliphatic VOCs compared to acute isolates, and highlight volatilomics as a powerful tool for probing metabolic adaptation in P. aeruginosa and for identifying candidate VOC markers associated with infection-related phenotypes.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1038/s41598-026-58199-2. Introduction Pseudomonas aeruginosa is a ubiquitous environmental and opportunistic pathogen, and a major cause of nosocomial respiratory infections. It frequently affects immunocompromised individuals and patients with underlying chronic respiratory diseases such as cystic fibrosis (CF), bronchiectasis3 or chronic obstructive pulmonary disease (COPD)1–4. In CF, P. aeruginosa establishes long-term airway infections, remaining one of the organisms most frequently isolated over the course of disease2. Persistent colonization by P. aeruginosa is a key factor driving CF progression, often resulting in a faster decline in lung function, more frequent hospitalizations, and increased mortality. Initially the infection is typically intermittent, but once the bacterium becomes established, eradication is extremely challenging2. During this transition from acute to chronic infection, P. aeruginosa undergoes extensive genotypic and phenotypic, as well as metabolic adaptation5–8. Traits commonly associated with chronic infection include modifications in lipopolysaccharide, reduced motility, mucoid conversion, enhanced antimicrobial resistance, and biofilm formation – many of which correlate with poorer clinical outcomes5,6. These adaptations presumably facilitate bacterial persistence in the CF airway, help evade immune clearance, and contribute to progressive tissue damage. Given the diversity of adaptive phenotypes, these traits may provide valuable information about the stage of the infection, disease progression, and patient prognosis. However, directly sampling bacteria from the lower airways is difficult, which limits the feasibility and accuracy of using phenotypic analysis to monitor infection dynamics. Currently, P. aeruginosa detection in CF patients primarily relies on sputum culture; however, sputum samples are often difficult to obtain, especially in young children or patients with advanced disease9. Alternative diagnostic approaches remain suboptimal: bronchoalveolar lavage is invasive, while upper airway swabs have limited ability to predict lower airway colonization and may lack sufficient sensitivity10–12. Moreover, widespread use of CFTR modulator therapies has reduced mucus production in many patients, further limiting the availability of sputum samples despite persistent infection with common CF pathogens13,14. Therefore, a non-invasive, sensitive method for early detection of P. aeruginosa respiratory infection is urgently needed. Volatile organic compounds (VOCs) in exhaled breath have emerged as promising candidates for non-invasive assessment of respiratory infections. These small molecules are produced by pathogenic microorganisms as metabolic products or secondary metabolites involved in defense, competition, and intracellular signaling9,15,16. Since VOCs may reflect microbial metabolic activity in vivo, they can act as chemical signatures reflecting species- and strain- associated metabolic differences9. Human breath contains thousands of VOCs that convey valuable information about the physiological and pathological status of the patient17. Analysis of breath volatiles has demonstrated potential for diseases such as malaria18 and pulmonary tuberculosis19 and it is currently used in clinical practice for Helicobacter pylori detection20. These findings support the feasibility of breath-based volatilomic approaches. In the context of chronic lung infections, breath-based diagnostics offer several advantages. They provide a non-invasive way of sampling the entire ventilated lung, overcoming the spatial limitations of sputum and bronchoalveolar lavage sampling. Furthermore, breath analysis has the potential to reflect real-time metabolic activity of pathogens in situ, potentially reducing diagnostic time. In addition to early detection, such analyses can provide insight into the metabolic status of pathogens within the lung environment. For P. aeruginosa respiratory infections in CF patients, breath-based volatilomic profiling may help assess and monitor infection dynamics by linking microbial phenotypic and genotypic adaptations to corresponding changes in volatile metabolite production17. To develop reliable VOC-based approaches for assessing and monitoring P. aeruginosa respiratory infections in CF, it is essential to consider the genomic, phenotypic, and metabolic diversity of this species and how it evolves within the CF lung17. During chronic infection, P. aeruginosa populations diversify into clonal variants, many of which acquire loss-of-function mutations in global regulatory genes such as lasR. This gene encodes a major transcriptional regulator that modulates multiple virulence pathways through quorum sensing. The emergence of lasR-deficient variants in CF patients has been associated with enhanced inflammation and disease progression21. These mutants exacerbate neutrophil recruitment and inflammation by disrupting the activity of LasR-regulated proteases- such as elastase LasB and alkaline protease AprA- which can degrade inflammatory mediators including ICAM-122 and C5a23. The resulting hyperinflammatory environment contributes to progressive lung injury characteristic of chronic CF infections. In a previous study we found a clear relationship between LasR expression and the global exometabolomic profile of P. aeruginosa clinical strains8. Here, we investigated the volatilome profiles of 48 clinical P. aeruginosa isolates, comprising 24 strains from acute respiratory infections and 24 from chronic respiratory infections in CF patients. Moreover, we analyzed the contribution of LasR regulator in shaping the volatilome of P. aeruginosa. By comparing these volatilomes, we aimed to identify candidate metabolic markers associated with acute and chronic infection states and provide insights into pathogen adaptation within the CF lung. Results In vitro volatilome profiling of P. aeruginosa respiratory infection isolates An untargeted metabolomics workflow was applied to characterize the in vitro volatilomes of 48 clinical P. aeruginosa isolates grown in artificial sputum media (ASM) using gas chromatography-mass spectrometry (GC-MS). After comprehensive data processing, 54 unique volatile metabolites were associated to P. aeruginosa growth and metabolic activity. Of these, 34 compounds were present at significantly higher levels in the headspace of bacterial cultures than in control samples (log₂ fold change ≥ 1.5, p < 0.05), consistent with active production by the bacteria (Table 1). In contrast, 20 compounds were detected at lower levels in the headspace of bacterial cultures than in control samples, indicating possible consumption or degradation by the microorganisms (Supplementary Table 1). Ten metabolites were produced by ≥ 95% of the isolates (n ≥ 46) and, following Davis et al. criteria17, were defined as the core volatilome; however, only eight of these compounds could be assigned putative identities. The non-core VOCs consisted of 24 compounds, of which 16 were assigned putative identities (Table 1). Many of these compounds have previously been reported as P. aeruginosa–associated volatiles; however, several volatile compounds not previously linked to P. aeruginosa were detected across multiple chemical classes. Newly identified produced compounds included a polysulfide (dimethyl tetrasulfide), aromatic compounds (2-[(methylthio)methyl]furan, xylene isomer and prop-1-en-2-ylbenzene (α-methylstyrene), alcohols (thietane-3-ol and 4-methyl-2-pentanol), alkanes (1-methyl-2-pentylcyclopropane and 4-methyloctane), a terpene (2,6,6-trimethylbicyclo[3.1.1]hept-2-ene (α-pinene)), a phthalate (2-ethylhexyl hydrogen benzene-1,2-dicarboxylate (mono(2-ethylhexyl)phthalate)), and a thioester (S-methyl 3-methylbutanethioate) (Table 1). Table 1Volatile compounds produced by P. aeruginosa respiratory infection isolates.Compound nameaFormulaChemical classNo. of samples (n = 48)ReferencesbCore volatile compoundsDimethyl disulfideC2H6S2Polysulfide47 9,16,17,24–39 Dimethyl trisulfideC2H6S3Polysulfide46 24,38 Dimethyl tetrasulfideC2H6S4Polysulfide48Cyclopent-2 -en-1-one (2-cyclopenten-1-one)C5H6OKetone48 38 CyclopentanoneC5H8OKetone48 38 CyclopentanolC5H10OAlcohol47 38 2-(methylthio)methylfuranC6H8OSAromatic481-methyl-2-pentylcyclopropaneC9H18Cycloalkane48Unknown 6Unknown48Unknown 10Unknown47Noncore volatile compoundsThiocyanic acid, methyl esterC2H3NSEster10 17,25,35,38 Thietan-3-olC3H6OSAlcohol-thioether22-pentanoneC5H10OKetone44 24,34,35,37 3-methyl-3-buten-1-olC5H10OAlcohol10 24 3-methyl-1-butanolC5H12OAlcohol32 16,17,35,37,40,41 3-methyl-2-pentanoneC6H12OKetone9 24,34,35 S-methyl 3-methylbutanethioateC6H12OSThioester442-methyl-butanoic acid, methyl esterC6H12O2Ester43 17 4-methyl-2-pentanolC6H14OAlcohol362-heptanoneC7H14OKetone37 24,35,37 AcetophenoneC8H8OAromatic Ketone2 35 Xylene (isomer)C8H10Aromatic2Prop-1-en-2-ylbenzene (α-methylstyrene)C9H10Aromatic24-methyloctaneC9H20Alkane102,6,6-trimethylbicyclo[3.1.1]hept-2-ene (α-pinene)C10H16Terpene42-ethylhexyl hydrogen benzene-1,2-dicarboxylate (mono(2-ethylhexyl)phthalate)C16H22O4Phthalate2Hydrocarbon 1Hydrocarbon3Unknown 1Unknown27Unknown 2Unknown2Unknown 4Unknown2Unknown 5Unknown37Unknown 11Unknown2Unknown 12Unknown1Unknown 14Unknown9a Compounds shown in bold represent volatile molecules that have not been previously reported in P. aeruginosa.b References that document these volatiles in P. aeruginosa cultures. Table 1Volatile compounds produced by P. aeruginosa respiratory infection isolates.Compound nameaFormulaChemical classNo. of samples (n = 48)ReferencesbCore volatile compoundsDimethyl disulfideC2H6S2Polysulfide47 9,16,17,24–39 Dimethyl trisulfideC2H6S3Polysulfide46 24,38 Dimethyl tetrasulfideC2H6S4Polysulfide48Cyclopent-2 -en-1-one (2-cyclopenten-1-one)C5H6OKetone48 38 CyclopentanoneC5H8OKetone48 38 CyclopentanolC5H10OAlcohol47 38 2-(methylthio)methylfuranC6H8OSAromatic481-methyl-2-pentylcyclopropaneC9H18Cycloalkane48Unknown 6Unknown48Unknown 10Unknown47Noncore volatile compoundsThiocyanic acid, methyl esterC2H3NSEster10 17,25,35,38 Thietan-3-olC3H6OSAlcohol-thioether22-pentanoneC5H10OKetone44 24,34,35,37 3-methyl-3-buten-1-olC5H10OAlcohol10 24 3-methyl-1-butanolC5H12OAlcohol32 16,17,35,37,40,41 3-methyl-2-pentanoneC6H12OKetone9 24,34,35 S-methyl 3-methylbutanethioateC6H12OSThioester442-methyl-butanoic acid, methyl esterC6H12O2Ester43 17 4-methyl-2-pentanolC6H14OAlcohol362-heptanoneC7H14OKetone37 24,35,37 AcetophenoneC8H8OAromatic Ketone2 35 Xylene (isomer)C8H10Aromatic2Prop-1-en-2-ylbenzene (α-methylstyrene)C9H10Aromatic24-methyloctaneC9H20Alkane102,6,6-trimethylbicyclo[3.1.1]hept-2-ene (α-pinene)C10H16Terpene42-ethylhexyl hydrogen benzene-1,2-dicarboxylate (mono(2-ethylhexyl)phthalate)C16H22O4Phthalate2Hydrocarbon 1Hydrocarbon3Unknown 1Unknown27Unknown 2Unknown2Unknown 4Unknown2Unknown 5Unknown37Unknown 11Unknown2Unknown 12Unknown1Unknown 14Unknown9a Compounds shown in bold represent volatile molecules that have not been previously reported in P. aeruginosa.b References that document these volatiles in P. aeruginosa cultures. Volatile compounds produced by P. aeruginosa respiratory infection isolates. a Compounds shown in bold represent volatile molecules that have not been previously reported in P. aeruginosa. b References that document these volatiles in P. aeruginosa cultures. Chemical classification of the produced core volatilome revealed that polysulfides were the most abundant class, representing 3 of the 8 identified compounds (37.5%) (Fig. 1), while ketones were the second largest group (25%). The remaining classes- including alcohols, esters, aromatics, alkanes and miscellaneous compounds- collectively accounted for 37.5%. Among non-core volatiles, ketones and alcohols were the most prevalent classes (25% each), followed by esters (18.75%), aromatic compounds (12.5%) and alkanes (6.25%) (Fig. 1). Fig. 1Chemical classification and counts of volatile compounds that were produced to levels significantly different from the medium alone by 48 P. aeruginosa respiratory-infection isolates. (A) identified compounds; (B) all detected compounds, including identified and unassigned compounds; (C) proportion of identified compounds; (D) proportion including both identified and unassigned compounds. C, Core volatilome; NC, Noncore volatilome. Chemical classification and counts of volatile compounds that were produced to levels significantly different from the medium alone by 48 P. aeruginosa respiratory-infection isolates. (A) identified compounds; (B) all detected compounds, including identified and unassigned compounds; (C) proportion of identified compounds; (D) proportion including both identified and unassigned compounds. C, Core volatilome; NC, Noncore volatilome. A total of 20 compounds with significantly lower abundances in the headspace of bacterial cultures compared to controls (log2 fold change ≤ -1.5, p-value < 0.05) were interpreted as being consumed by the bacteria (Supplemental Table 1). Of these, five were metabolized by at least 95% of the strains analyzed, constituting the core volatilome of consumed compounds. Among metabolites not previously associated with P. aeruginosa metabolism we found several aldehydes (2-methylbutanal, pentanal, 2-phenylacetaldehyde (benzeneacetaldehyde) and octanal) (Supplemental Table 1). In fact, aldehydes were the only chemical class among the consumed core volatilome that were consistently consumed by P. aeruginosa (log2 fold change ≤ -1.5, p-value < 0.05). Among non-core consumed volatiles, aldehydes were also the most frequently metabolized chemical class (45.5%), followed by alcohols (18.2%) and ketones (18.2%) (Supplemental Fig. 1). Together, these findings highlight the chemical diversity of the P. aeruginosa volatilome and reveal distinct patterns of volatile production and consumption that may reflect strain-specific metabolic activity. Candidate VOC markers associated with P. aeruginosa metabolic profiles Identified VOCs detected at significantly higher levels in the headspace of bacterial cultures than in medium controls are shown in Fig. 2. Notably, several VOCs were completely absent from the uninoculated medium, suggesting that they are likely derived from bacterial metabolism. These compounds are highlighted in red. Among these, several cyclic compounds, including cyclopent-2-en-1-one (2-cyclopenten-1-one), cyclopentanone, and cyclopentanol, were identified as components of the P. aeruginosa volatilome. Simple ketones, such as 2- pentanone, as well as branched-chain derivatives such as 3-methyl-2-pentanone, or compounds such as 2-methyl butanoic acid methyl ester, were also detected in the headspace of P. aeruginosa clinical isolates. These compounds are commonly associated with P aeruginosa15,17,24,34 and are thought to arise as metabolic intermediates or by-products of fatty acid metabolism42–44. In addition, S-methyl-3-methylbutanethioate and thiocyanic acid methyl ester, volatiles previously suggested to play roles in microbial interactions or to exert antimicrobial effects45,46, were detected, further contributing to the volatile signature of P. aeruginosa. Alcohols and allylic alcohols, including 3-methyl-1-butanol and 3-methyl-3-buten-1-ol, which have been previously reported as P. aeruginosa VOCs (16, 17, 24; 25, 41), were detected and are likely derived from amino acid or isoprenoid metabolism47,48. In contrast, hydrocarbon derivatives such as 1-methyl-2-pentylcyclopropane, represent less common volatiles not previously attributed to the P. aeruginosa volatilome. Metabolites detected in both blank medium and bacterial headspace but significantly enriched in the headspace of bacterial cultures (log2 fold change ≥ 1.5, FDR corrected p < 0.05), are highlighted in black in Fig. 2. Among these, sulfur-containing compounds, including dimethyl disulfide, dimethyl trisulfide, and dimethyl tetrasulfide were prominent. These classic microbial volatiles, typically derived from methionine and cysteine metabolism, are frequently associated with P. aeruginosa37. Several ketones and alcohols, such as 2-heptanone, acetophenone and 4-methyl-2-pentanol, likely arising from fatty acid degradation, were also consistently detected across clinical isolates. Hydrocarbons such as 4-methyloctane were also identified. In addition, heterocyclic VOCs such as 2-(methylthio)methylfuran and thietane-3-ol, typically originating from lipid or amino-acid transformations49, were detected across various clinical isolates. The aromatic compounds xylene and prop-1-en-2-ylbenzene (α-methylstyrene), were identified for the first time as constituents of the P. aeruginosa volatilome in this study. Collectively, these produced VOCs reflect the metabolic diversity of P. aeruginosa and may serve as candidate markers associated with its metabolic profiles. Notably, compounds produced at significantly higher levels than the medium by more than 95% of the isolates- including the polysulfides (dimethyl disulfide, dimethyl trisulfide and dimethyl tetrasulfide), as well as cyclopent-2-ene-1-one (2-cyclopenten-1-one), cyclopentanone, cyclopentanol, 2-methylthio-methylfuran and 1-methyl-2-pentyl cyclopropane- represent particularly consistent candidate VOC markers associated with P. aeruginosa metabolism (Fig. 2). Compounds detected at significantly lower levels in the headspace of bacterial cultures relative to the medium alone are shown in Supplemental Fig. 2 and were interpreted as being consumed by P. aeruginosa. These included 2-methylpropanal, 3-methylbutanal, hexanal, benzaldehyde, 3-methylbutanal, and other compounds. This inverse abundance pattern suggests that P. aeruginosa either consumes, transforms, or otherwise reduces the abundance of these VOCs during growth. The consumed metabolites represent a chemically diverse group, including short chain branched aldehydes, aromatic aldehydes, heterocyclic alcohols, and aliphatic hydrocarbons, highlighting extensive interactions between bacterial metabolism and the surrounding chemical environment. Interestingly, aldehydes were the only chemical class consistently represented in the core volatilome of consumed compounds, suggesting that aldehyde metabolism is a conserved metabolic feature of P. aeruginosa. Although VOCs that decrease in abundance may be less suitable as direct candidate markers, they nonetheless contribute to the distinct volatilomic signature of P. aeruginosa. Fig. 2Relative abundances of identified volatile compounds detected at significantly higher levels in at least one P. aeruginosa isolate compared to the control medium. Only compounds with log2 fold change ≥ 1.5, p < 0.05 are shown. Compounds exclusively detected in the headspace of bacterial cultures and absent from the medium are shown in red, whereas compounds also present in the medium but at significantly lower levels are shown in black. Each dot represents an isolate and corresponds to the mean of three experimental replicates. Box plots display metabolite abundance as peak areas in arbitrary units. For each compound, isolates were classified as producers when VOC levels were significantly different from the corresponding blank medium control (t-test, p < 0.05). Only producing isolates are displayed, whereas non-producing isolates were omitted from the plots for visualization purposes. The number of producing isolates (n) is indicated below each VOC name. a2,6,6-trimethylbicyclo[3.1.1]hept-2-ene (IUPAC name). b2-ethylhexyl hydrogen benzene-1,2-dicarboxylate (IUPAC name). Relative abundances of identified volatile compounds detected at significantly higher levels in at least one P. aeruginosa isolate compared to the control medium. Only compounds with log2 fold change ≥ 1.5, p < 0.05 are shown. Compounds exclusively detected in the headspace of bacterial cultures and absent from the medium are shown in red, whereas compounds also present in the medium but at significantly lower levels are shown in black. Each dot represents an isolate and corresponds to the mean of three experimental replicates. Box plots display metabolite abundance as peak areas in arbitrary units. For each compound, isolates were classified as producers when VOC levels were significantly different from the corresponding blank medium control (t-test, p < 0.05). Only producing isolates are displayed, whereas non-producing isolates were omitted from the plots for visualization purposes. The number of producing isolates (n) is indicated below each VOC name. a2,6,6-trimethylbicyclo[3.1.1]hept-2-ene (IUPAC name). b2-ethylhexyl hydrogen benzene-1,2-dicarboxylate (IUPAC name). Volatilome profiling reveals distinct VOC signatures associated with acute and chronic P. aeruginosarespiratory isolates A hallmark of P. aeruginosa infection in the CF lung is its progressive adaptation to the host environment. We therefore hypothesized that changes in volatile metabolite profiles could reflect the transition from acute infection to chronic colonization. To test this, we investigated whether volatilome profiles could discriminate between acute and chronic P. aeruginosa isolates. Comparative volatilome analysis of respiratory P. aeruginosa isolates revealed metabolic differences in specific VOCs between acute and chronic infection phenotypes. Specifically, the aliphatic compounds 1-methyl-2-pentyl-cyclopropane and 4-methyloctane were produced at significantly higher levels by chronic respiratory infection isolates compared to acute respiratory infection isolates (Fig. 3A). Conversely, acetophenone was produced at higher levels by the acute respiratory infection isolates (Fig. 3B). Volatile compounds differentially consumed or transformed by acute and chronic respiratory P. aeruginosa isolates are shown in Supplementary Fig. 3. Chronic isolates exhibited greater depletion of several aldehydes, including hexanal, pentanal, 2-phenylacetaldehyde (benzeneacetaldehyde) and octanal (Supplemental Fig. 3A). Conversely, acute isolates showed stronger reductions in benzaldehyde and 3-hexen-2-one (Supplemental Fig. 3B). This divergence in compound consumption suggests that the two clinical phenotypes preferentially metabolize different subsets of VOCs, resulting in distinct volatilome signatures associated with each phenotype. Taken together, these results suggest that the volatilome of P. aeruginosa clinical isolates is influenced by strain and infection phenotype. Despite identical growth conditions, acute and chronic isolates displayed different volatile profiles, particularly in the relative abundance of ketones, aldehydes and alkanes. Notably, reduced production of aromatic, biologically active VOCs such as acetophenone, together with increased abundance of aliphatic compounds, may reflect metabolic adaptations associated with chronic P. aeruginosa colonization. Fig. 3Relative production of VOCs differed significantly between acute and chronic respiratory infection P. aeruginosa isolates (A) Compounds detected at higher abundance in chronic isolates. (B) Compounds enriched in acute isolates. Each dot represents an isolate, plotted as the mean of three experimental replicates. Each box plot shows the metabolite levels represented by peak areas in arbitrary units. Data were normalized to the medium. Statistical significance was evaluated using an unpaired t-test with Welch’s correction. *p < 0.05; *** p < 0.001; **** p < 0.0001. Relative production of VOCs differed significantly between acute and chronic respiratory infection P. aeruginosa isolates (A) Compounds detected at higher abundance in chronic isolates. (B) Compounds enriched in acute isolates. Each dot represents an isolate, plotted as the mean of three experimental replicates. Each box plot shows the metabolite levels represented by peak areas in arbitrary units. Data were normalized to the medium. Statistical significance was evaluated using an unpaired t-test with Welch’s correction. *p < 0.05; *** p < 0.001; **** p < 0.0001. Volatilome profiles reveal adaptation over time in chronic infection. To further assess whether volatilome profiles capture temporal adaptation during chronic infection, we examined a collection of 12 pairs of clonally related P. aeruginosa chronic infection isolates collected longitudinally from 10 individuals with CF. Two patients each contributed two isolate pairs, resulting in four pairs derived from two patients. Early and late isolates were defined based on their time of collection within each patient. This approach allowed us to examine volatilome changes associated with within-host evolution. Principal Component Analysis (PCA) of all 24 chronic isolates (12 early-late pairs) did not reveal a clear global separation between early and late isolates (Fig. 4A). This result is not surprising, as isolates were obtained from different patients and represent distinct evolutionary trajectories, with variable timing and degrees of adaptation to the host environment. However, when early and late isolates of each patient were analyzed as clonal pairs, PCA revealed changes in the volatilome (Fig. 4B). In most cases, late isolates displayed distinct positions in PCA space relative to their corresponding early isolates, indicating measurable changes in volatile profiles over time. These within-patient separation suggest that volatilome signatures evolve during chronic infection in a host-specific manner, reflecting progressive metabolic adaptation rather than a uniform path shared across patients. Together, these results indicate that while early and late chronic isolates cannot be distinguished globally across patients, volatilome profiling can detect temporal metabolic changes within clonally related isolates. This supports the potential of volatilomics to track within-host adaptation of P. aeruginosa during chronic infection. Fig. 4Principal component analysis (PCA) of the discovery data set of 12 pairs of clonally related P. aeruginosa isolates from individuals with CF. (A) PCA scores of VOC profiles obtained from triplicate experimental replicates for each of the 24 isolates. (B) PCA showing clustering of each pair of clonally related isolates. Samples were plotted using principal components analysis 1 and 2 as coordinates. Each dot represents a single experimental replicate. Orange symbols correspond to “early” isolates and blue symbols correspond to “late” isolates. Principal component analysis (PCA) of the discovery data set of 12 pairs of clonally related P. aeruginosa isolates from individuals with CF. (A) PCA scores of VOC profiles obtained from triplicate experimental replicates for each of the 24 isolates. (B) PCA showing clustering of each pair of clonally related isolates. Samples were plotted using principal components analysis 1 and 2 as coordinates. Each dot represents a single experimental replicate. Orange symbols correspond to “early” isolates and blue symbols correspond to “late” isolates. Loss of LasR function rewires the volatilome of P. aeruginosa clinical strains Among the adaptive variants that emerge during P. aeruginosa respiratory infection in CF patients are mutations in the regulatory gene lasR. While the clinical significance of LasR deficiency is well established16, its metabolic consequences- particularly regarding VOC production and consumption- remain poorly defined. In a previous study, we identified a significant association between the LasR phenotype and the global exometabolomic profile of clinical isolates, including several metabolites whose production appeared to be linked to LasR expression8. To determine whether LasR also influences VOC production, we analyzed the volatilomes of 48 P. aeruginosa strains isolated from acute and chronic respiratory infections. Strains were classified as LasR⁺ (functional LasR) or LasR⁻ (nonfunctional LasR). As shown in Fig. 5A, LasR+ isolates produced significantly higher levels of dimethyl disulfide, dimethyl trisulfide, 2-(methylthio)methylfuran, thiocyanic acid methyl ester, 3-methyl-2-pentanone and 2,6,6-trimethylbicyclo[3.1.1]hept-2-ene (α-pinene). In contrast, LasR− isolates produced higher levels of cyclopentanol and 3-methyl-1-butanol (Fig. 5B). Regarding consumption, LasR+ strains preferentially depleted 2-phenylacetaldehyde (benzeneacetaldehyde) and 4-methylpentan-2-one (methyl isobutyl ketone), whereas LasR− strains consumed or transformed 2-methylbutanal and 3-methylbutanal (Supplemental Fig. 4). Overall, these results indicate that LasR contributes to the VOC profile of clinical P. aeruginosa isolates. Fig. 5Relative abundances of volatile organic compounds differentially produced by P. aeruginosa respiratory isolates classified as LasR+ (functional lasR gene) and LasR− (non-functional lasR gene) groups. (A) Compounds detected at higher abundance in LasR+ isolates. (B) Compounds enriched in LasR− isolates. Each dot represents an isolate, plotted as the mean of three experimental replicates. Each box plot shows the metabolite levels represented by peak areas in arbitrary units. Data were normalized to the medium control. Statistical significance was determined using an unpaired t-test with Welch’s correction. *p < 0.05; ** p < 0.01. Relative abundances of volatile organic compounds differentially produced by P. aeruginosa respiratory isolates classified as LasR+ (functional lasR gene) and LasR− (non-functional lasR gene) groups. (A) Compounds detected at higher abundance in LasR+ isolates. (B) Compounds enriched in LasR− isolates. Each dot represents an isolate, plotted as the mean of three experimental replicates. Each box plot shows the metabolite levels represented by peak areas in arbitrary units. Data were normalized to the medium control. Statistical significance was determined using an unpaired t-test with Welch’s correction. *p < 0.05; ** p < 0.01. Discussion Volatile organic compounds in exhaled breath have been proposed as a potential noninvasive approach for respiratory infections. Given that P. aeruginosa is one of the most important human respiratory pathogens, numerous studies have attempted to characterize the VOCs produced by this microorganism as candidate markers associated with P. aeruginosa infection9,16,17,24–38. In the present study, a substantial proportion of the VOCs we detected overlapped with previously reported compounds. Specifically, when considering both produced and consumed compounds, 8 of the 13 identified core volatiles (61.5%) and 15 of the 27 identified non-core volatiles (55.6%) identified here have been described in earlier P. aeruginosa volatilomic studies (Table 1 and Supplementary Table 1). This concordance underscores the reproducibility of volatilomic analyses across independent studies and further supports their association with P. aeruginosa. Among these compounds, dimethyl disulfide consistently emerged as a prominent VOC associated with P. aeruginosa metabolism. Its frequent detection in in vitro studies9,16,17,24–36, together with its presence in clinical samples such as sinus mucus from infected patients30, highlights its translational relevance and supports its potential relevance as a candidate VOC marker. Similarly, thiocyanic acid methyl ester, although detected in only a subset of isolates in this study, has previously been identified in the breath of CF patients39. This finding supports its production in vivo and suggests that its presence may be influenced by strain-specific factors. A considerable fraction of VOCs identified in this study have not been previously associated with the P. aeruginosa volatilome. As we aimed to identify candidate VOC markers associated with P. aeruginosa metabolism and infection phenotypes, only produced compounds were considered. Specifically, 3 of the 8 core volatiles (37.5%) and 8 of the 16 non-core volatiles (50%) were reported here for the first time (Table 1). Most of the newly identified non-core compounds were detected in only a limited number of strains, reflecting intraspecies metabolic variability. In contrast, dimethyl tetrasulfide, 2-(methylthio)methylfuran, and 1-methyl-2-pentylcyclopropane were produced by all isolates analyzed and are newly identified compounds of the P. aeruginosa volatilome. Their consistent production across genetically diverse isolates suggests that they may represent conserved metabolic features and therefore represent previously unrecognized candidate VOCs associated with P. aeruginosa metabolism. Additionally, S-methyl-3-methylbutanethioate and 4-methyl-2-pentanol, while not part of the core volatilome, were observed in most isolates, further supporting their potential translational relevance. It is important to note that some compounds were detected exclusively in the bacterial cultures and not in the medium; however, this may reflect a sensitivity limitation, and we cannot exclude the possibility that these VOCs were present in the medium at levels below our detection threshold. We examined temporal changes by comparing 12 clonally related pairs of P. aeruginosa isolates from 10 CF patients, designated as early and late. When analyzed at the population level, early versus late isolates could not be distinguished based on their volatilome profiles. However, pairwise comparisons within each patient revealed clear differences between early and late isolates. These findings align with Davis et al.17, who reported that intrapatient early and late isolates are often no more similar to each other than to isolates from other patients, highlighting the strong influence of posttranscriptional and phenotypic diversity that arises during chronic infection. Our results extend this observation by demonstrating that, despite high variability among isolates, temporal changes within a single patient’s volatilome can be detected, supporting the potential for personalized monitoring. Breath-based diagnostics for P. aeruginosa infections may therefore need to be individualized, using each patient’s initial breathprint as a baseline for detecting alterations that could indicate shifts in population phenotype and an increased risk of exacerbations or treatment failure. Comparisons between acute and chronic isolates revealed distinct volatilomic signatures. Chronic isolates produced higher levels of aliphatic hydrocarbons and ketones, including 1-methyl-2-pentylcyclopropane and 4-methyloctane, whereas acute isolates produced higher levels of acetophenone. These patterns are consistent with known metabolic adaptations during long-term infection, such as slower growth rate, biofilm formation, oxygen limitation, and nutrient restriction, which promote anaerobic and microaerophilic metabolism and therefore metabolic remodeling50,51. Chronic isolates therefore exhibit a highly diverse volatilome enriched in hydrocarbons and ketones17. Acute isolates, by contrast, maintain metabolism typical of oxygen-rich, planktonic growth and show stronger expression of quorum sensing pathways, including the expression of the master quorum sensing regulator LasR, likely explaining higher acetophenone production. While the role of acetophenone in P. aeruginosa has not been clarified, its analog 2-aminoacetophenone (2-AA) is a known quorum sensing-regulated molecule52. We further investigated the impact of LasR mutations on P. aeruginosa volatilome. LasR mutations frequently arise during chronic CF infections and are associated with disease progression and hyperinflammatory host responses in part due to altered protease activity (elastase LasB and alkaline protease AprA) affecting ICAM-122 and C5a23 degradation. In our study, LasR+ isolates produced higher levels of sulfur- and ketone-containing VOCs, such as dimethyl disulfide, dimethyl trisulfide, and 3-methyl-2-pentanone, whereas LasR− isolates produced more alcohols (e.g., cyclopentanol, 3-methyl-1-butanol). Additional VOC differences, including 2-(methylthio)methylfuran, thiocyanic acid methyl ester, and α-pinene, were also linked to LasR status, consistent with its influence on global metabolic pathways8,53. Despite these insights, our study has some limitations. First, we analyzed a limited number of isolates under standardized laboratory conditions, which may not fully capture the in vivo diversity of P. aeruginosa metabolism. Second, the artificial sputum medium (ASM) mimics aspects of the CF airway environment but does not replicate host-pathogen interactions or the complex spatial and chemical heterogeneity of the lung. Third, we focused exclusively on P. aeruginosa and did not evaluate the potential influence of co-infecting respiratory pathogens. Given that polymicrobial infections are common in CF, future studies should address how microbial interactions may influence VOC profiles and candidate marker signatures. In addition, although this study demonstrates that VOC profiles reflect phenotypic and metabolic differences among P. aeruginosa isolates, several aspects relevant to clinical translation remain unaddressed. We did not assess VOC persistence following bacterial death or lysis and therefore cannot determine whether the detected compounds primarily reflect metabolically active cells. Moreover, while VOC profiling may be useful for infection stratification, it does not provide information on antibiotic susceptibility and thus remains complementary to conventional culture-based diagnostics. Nevertheless, this work provides a comprehensive analysis of volatilome differences between acute and chronic isolates and highlights candidate VOC markers associated with distinct infection phenotypes. Importantly, our results are consistent with the concept mentioned by Davis et al.17 that chronic P. aeruginosa adaptation alters its metabolic flux rather than completely activating or suppressing specific pathways, supporting the idea that future VOC-based diagnostic strategies should consider relative abundance rather than simple presence or absence. Materials and methods Bacterial strains Two sets of P. aeruginosa clinical isolates from respiratory infections were analyzed. The first set included 24 strains from ten CF patients, selected from a previously described longitudinal cohort54(Table 2). For each patient, the earliest isolate (designated as the early isolate) and the most recent clonally related isolate (designated as the late isolate), separated by at least three years, were examined. Chronic isolates were clonally related, as determined by pulsed-field gel electrophoresis (PFGE) and multilocus sequencing typing (MLST)54. In the isolate designation, the first two digits indicate the patient number, and the subsequent four digits indicate the date of isolation (MM/YY). The second set comprised 24 epidemiologically unrelated isolates obtained from patients presenting with acute respiratory infections in a hospital setting, without evidence of established chronic airway colonization at the time of sampling55(Table 2). Acute and chronic isolates therefore represent distinct clinical contexts, with chronic isolates predominantly derived from CF patients and acute isolates originating from sporadic respiratory infections. Each isolate was previously classified as LasR functional (LasR+) or non-functional (LasR−) based on C5a-cleavage activity (a phenotype known to depend on LasR) together with sequence analysis of the lasR gene8,23. Table 2P. aeruginosa clinical isolates used in this study.Chronic infection isolatesAcute infection isolatesIsolate IDaLasRphenotypebIsolate IDcLasRphenotypebFQSE05-0403+A1+FQSE05-0111+A2-FQSE06-0403-A3+FQSE06-0610-A4-FQSE10-0503+A5-FQSE10-0111+A6-FQSE11-0603+A7+FQSE11-1110-A8+FQSE12-0603-A9-FQSE12-1206-A10+FQSE12-1007-A11-FQSE12-1110-A12+FQSE15-0803-A13+FQSE15-0111-A14+FQSE16-0803-A15+FQSE16-0910-A16+FQSE21-1003-A17+FQSE21-0410-A18-FQSE21-0505-A19-FQSE21-1110-A20+FQSE24-0304-A21+FQSE24-1010-A22+FQSE28-1006+A23+FQSE28-1110+A24+a) Chronic isolates are designated as FQSE followed by the patient number, and the date of the isolation (MM/YY). b) LasR phenotype was determined by C5a cleavage ability (8,23).c) Acute isolates are designated by the letter A followed by the strain number. Table 2P. aeruginosa clinical isolates used in this study.Chronic infection isolatesAcute infection isolatesIsolate IDaLasRphenotypebIsolate IDcLasRphenotypebFQSE05-0403+A1+FQSE05-0111+A2-FQSE06-0403-A3+FQSE06-0610-A4-FQSE10-0503+A5-FQSE10-0111+A6-FQSE11-0603+A7+FQSE11-1110-A8+FQSE12-0603-A9-FQSE12-1206-A10+FQSE12-1007-A11-FQSE12-1110-A12+FQSE15-0803-A13+FQSE15-0111-A14+FQSE16-0803-A15+FQSE16-0910-A16+FQSE21-1003-A17+FQSE21-0410-A18-FQSE21-0505-A19-FQSE21-1110-A20+FQSE24-0304-A21+FQSE24-1010-A22+FQSE28-1006+A23+FQSE28-1110+A24+a) Chronic isolates are designated as FQSE followed by the patient number, and the date of the isolation (MM/YY). b) LasR phenotype was determined by C5a cleavage ability (8,23).c) Acute isolates are designated by the letter A followed by the strain number. P. aeruginosa clinical isolates used in this study. a) Chronic isolates are designated as FQSE followed by the patient number, and the date of the isolation (MM/YY). b) LasR phenotype was determined by C5a cleavage ability (8,23). c) Acute isolates are designated by the letter A followed by the strain number. Sample preparation Bacterial isolates were first grown overnight in Luria-Bertani broth at 37 °C. Five hundred microliters of overnight cultures were transferred into 250 mL Erlenmeyer flasks containing 10 mL of artificial sputum medium (ASM) – prepared as previously described56. The flasks were fitted 30.7 mm Versilic ® silicone caps to allow VOC accumulation and incubated at 37 °C with agitation at 180 rpm. In parallel, one flask containing ASM without bacterial inoculation was included in each experiment as negative control. The bacterial cultures were incubated for 24 h under the experimental conditions used in this study. Final bacterial loads were quantified after incubation, and no significant differences were observed between isolates. The Erlenmeyer flasks containing bacterial cultures were then purged with dry nitrogen delivered through the silicone caps via a needle at a flow rate of 180 mL/min. The headspace was then withdrawn through a separate outlet and directed into thermal desorption tubes containing 150 mg of Tenax GR using a sampling pump operating at 150 mL/min for 30 min. Analytical instrumentation VOCs were analyzed using an Ultra-xr autosampler coupled to a thermal desorption (TD) unit (Unity 2, Markes International, Bridgend, UK), which was connected to a gas chromatograph (Agilent 6890 N, Agilent Technologies, Santa Clara, CA, USA) equipped with a mass selective detector (5975 MSD, Agilent Technologies). The TD sequence consisted of an initial pre-desorption step (0.1 min, 20 mL/min), followed by primary desorption at 300 °C for 10 min with a 10 mL/min split. During primary desorption, analytes were cryofocused in a trap containing 30 mg of Tenax TA maintained at − 30 °C. The trap was then rapidly heated from − 30 °C to 300 °C and held for 10 min to achieve secondary desorption, after which the analytes were transferred to the gas chromatograph. Compound separation was performed on an HP-5MS capillary column (30 m × 0.25 mm × 0.25 μm, Agilent Technologies) using helium as the carrier gas at a constant flow rate of 1.6 mL/min. Samples were introduced into the GC in splitless mode, and the transfer line was maintained at 210 °C. The oven temperature program was as follows: 30 °C (3.5 min), ramp to 50 °C at 5 °C/min, to 90 °C at 10 °C/min, to 130 °C at 15 °C/min, to 180 °C at 30 °C/min, and finally to 280 °C at 40 °C/min with a final hold of 10 min. Mass spectra were recorded in electron ionization mode (70 eV) over an m/z range of 45–600 Da. Data processing and analysis Chromatographic data was processed using Agilent MSD ChemStation software. Automated peak detection and integration were performed in ChemStation and subsequently reviewed manually to correct baseline deviations or split peaks. Since the analysis was untargeted and no internal standards were used, peak areas were interpreted as relative semi-quantitative indicators of VOC abundance. Tentative VOC identification was carried out by matching experimental mass spectra to the NIST05 Mass Spectral Library (National Institute of Standards and Technology, 2005). Only matches with spectral similarity scores ≥ 85% and consistent fragmentation patterns were accepted as putative identifications. Compound identities were not confirmed using authentic analytical standards or retention index. Compounds lacking satisfactory matches were designated as unknown VOCs. For statistical analysis, each detected VOC was measured in three independent biological replicates. Compounds were classified as produced if their log2 fold change relative to the medium control was ≥ 1.5 with p < 0.05, and as consumed if the log2 fold change was ≤ -1.5 with p < 0.05. Multiple t-tests with false discovery rate (FDR) correction or unpaired t-test with Welch’s correction were applied as appropriate. Principal component analysis (PCA) was performed with Perseus version 1.6.1557 to visualize global volatilome differences between isolates and to assess specific clustering. All other figures were generated using GraphPad Prism version 10.6.1. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1