A total of 41 dengue patients and 23 healthy volunteers were recruited at the First Affiliated Hospital of Guangzhou University of Chinese Medicine in 2023. According to the Guidelines for the diagnosis and treatment of dengue in China (2018) published by the Chinese Society of Infectious Diseases, Chinese Medical Association, patients were diagnosed with dengue fever if they met the following criteria: 1) epidemiological history of recent travel to dengue endemic areas; 2) fever, weak, muscle or joint pain, rash, and even bleeding tendency; 3) decreased peripheral white blood cells or platelets; 4) DENV IgM, NS1 antigenemia, or DENV nucleic acid positivity. Demographic and clinical data of dengue patients in the study were obtained from electronic medical records and case report forms.

The blood samples were collected in the morning on an empty stomach using standard venipuncture procedures. To isolate plasma, heparin tubes were used for anticoagulants. Plasma samples were separated by centrifuging at 3,000 g for 10 min, collected into new centrifuge tubes, transferred to a Thermo Scientific 70U Series freezer, and stored at -80°C. This study was approved by the Ethical Committee of First Affiliated Hospital of Guangzhou University of Chinese Medicine (NO. K-2022-066). All examinees in both dengue fever and healthy groups voluntarily joined this study with informed consent signed. Metabolomic data analysis was entrusted to be performed by Shanghai Luming Biological Technology Co., LTD (Shanghai, China).

150 μL of the sample and 600 μL protein precipitant (a mixture of L-2-chlorophenylalanine, succinic acid-d4, L-Valine-d8, and bile acid-D4 dissolved in methanol-acetonitrile as internal standard, with a volume ratio of the two solvents of 2:1, 4 μg/mL) were mixed and vortexed for 10 s. Subsequently, the whole samples were extracted by ultrasonic for 10 min in an ice-water bath, and stored at -40°C overnight. The next day, the samples were centrifuged at 4°C (12000 rpm) for 10 min. The supernatant (150 µL) was removed from each tube, filtered through a 0.22 µm microfilter, and transferred to an LC vial. Samples were stored at -80°C until LC-MS analysis.

150 μL of supernatant was transferred to a glass sampling vial and dried under vacuum at room temperature using a centrifugal concentration dryer. Then 80 μL of methoxylamine hydrochloride in pyridine (15 mg/mL) was added and oxidized for 60 min at 37°C in a steam bath shaker. After removing the sample, 50 μL of BSTFA and 20 μL n-hexane were added to the mixture, with 10 μL of the internal standard mixtures, which was vortexed vigorously and derivatized for 60 min at 70°C. The mixture of internal standards contained the following reagents, indicated by Cat Number: G162300, N-9M-AU4-B, N-10M-A18-D, N-12M-AU15-D, N-14M-A24-E, G161798, N-18M-O9-C, N-20M-J27-E, N-22M-JY30-E, and N-24M-S6-A. Finally, the samples were left at room temperature for 30 min followed by GC-MS analysis.

The quality control (QC) sample was prepared by mixing equal volumes of extracts from all samples to evaluate the sample preparation, derivatization, sample loading, and the stability of the mass spectrometry system during sample testing.

An Ultra High-Performance Liquid Tandem High-Resolution Mass Spectrometer (UHPLC-HRMS) (Waters ACQUITY UPLC I-Class plus/Thermo QE) was used to analyze the metabolic profiling in both positive and negative ion modes. An ACQUITY UPLC HSS T3 Chromatography Column (1.8 μm, 100 mm×2.1 mm) was used in both positive and negative modes. The binary gradient elution system consisting of (A) water (containing 0.1% formic acid) and (B) acetonitrile was used to achieve the separation: 0 min, 5% B; 2 min at 5%B; 4 min at 30%B; 8 min at 50%B; 10 min at 80%B; 14 min at 100%B; 15 min at 100%B; 15.1 min at 5%B and 16 min, 5%B. The flow rate was 0.35 mL/min and the column temperature was 45°C. All samples were stored at 10°C during analysis, and the injection volume was 3 μL.

The mass spectrometer was operated as follows: Spray voltage, 3800 V (+) and 3000 V (−); Capillary temperature, 320°C; Aux gas heater temperature, 350°C; Sheath gas flow rate, 35 arbitrary units; Auxiliary gas flow rate, 8 arbitrary units; S-lens RF level, 50. The mass range was from m/z 70 to 1,050. The resolution was set at 70,000 for the full MS scans and 17500 for MS/MS. The NCE/stepped NCE was set at 10, 20 and 40 eV.

The derivatized samples were analyzed on an Agilent 7890B-5977A gas chromatography system. Derivatives were separated using a DB-5MS fused silica capillary column (30 m × 0.25 mm × 0.25 μm, Agilent J & W Scientific, Folsom, CA, USA). The carrier gas was helium (> 99.999%), and the flow rate was 1.0 mL/min. The syringe temperature was maintained at 260 °C. A splitless mode was used with an injection volume of 1 μL. The oven was held at an initial temperature of 60°C for 0.5 min, then increased to 125°C at a rate of 8°C/min, 210°C at a rate of 8°C/min, 270°C at a rate of 15°C/min, 305°C at a rate of 20°C/min, and finally held at 305°C for 5 min. The temperature of the ion source (electron impact, EI) and the MS quadrupole was set to 230°C and 150°C, respectively. The collision energy was 70 eV, and the mass spectrometric data was acquired in a full-scan mode (m/z 50-500).

Raw LC-MS data were baseline filtered, peaks identified, integrated, retention time corrected, peak aligned and normalized using Progenesis QI v3.0 software (Nonlinear, Dynamics, Newcastle, UK). The main parameters of precursor tolerance 5 ppm and product tolerance 10 ppm were based on The Human Metabolome Database (HMDB) and Lipidmaps (v2.3), while 10 ppm precursor tolerance and 20 ppm product tolerance were based on LuMet-Animal v3.0 and METLIN databases. The compounds were identified based on the above databases according to several dimensions of retention time, the precise mass-to-charge ratio (M/z), secondary fragments, and isotopic distribution. The extracted data were then further processed by removing all peaks in the group with more than 50% missing values (ion intensity = 0), replacing the zero values with half of the minimum value and screening compounds based on qualitative results. Compounds scoring less than 36 points out of 80 were also considered inaccurate and removed.

The raw D-format GC/MS data obtained were converted to abf format for quick data retrieval using Analysis Base File Converter software. The data were then imported into MS-DIAL v4.24 software for peak detection, peak identification, MS2Dec deconvolution, characterization, peak alignment, wave filtering, and missing value interpolation, and finally the data matrix. Metabolite characterization was based on the LuMet-GC 5.0 database and NIST Chemistry Web Book. All internal standard peaks and pseudo-positive peaks were then excluded and further processed to exclude all peaks in the group with more than 50% missing values (ionic intensity = 0) and replace the zero value with half of the minimum value. The signal intensities of all peaks in each sample were split and normalized based on internal standards with a filtered RSD greater than 0.1. Redundancy removal and peak merging were then performed, and compounds with scores below 70 out of 100 were also considered inaccurate and removed. In each sample, all peak signal intensities were segmented and normalized according to the internal standards with RSD greater than 0.1 after screening. After the data was normalized, redundancy removal and peak merging were conducted to obtain the data matrix. Compounds scoring less than 70 points out of 100 were also considered inaccurate and removed.

The positive and negative ion data from the LC-MS/MS and the data obtained from the GC-MS were combined into a data matrix that contained all the information extracted from the raw data and could be used for subsequent analysis.

Principle Component Analysis (PCA) was used to observe the overall distribution of the samples and the stability of the whole analytical process. To distinguish the different metabolites between groups, orthogonal partial least squares discriminant analysis (OPLS-DA) was used. To prevent overfitting, the quality of the model was evaluated using 7-fold cross-validation and 200 Response Permutation Testings (RPTs). The Variable important in projection (VIP) values from the OPLS-DA model were used to rank the contribution of each variable to group discrimination. Additionally, a two-tailed Student’s T-test was used to verify the significance of the differences in metabolites between groups. We selected differential metabolites based on VIP values greater than 1.0 and P-values less than 0.05.

The RNA-seq data were downloaded from Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/) with GEO series accession numbers GSE51808, GSE96656, and GSE206829, the differentially expressed genes (DEGs) were obtained using the R programming language. The populations for these 3 datasets included dengue patients and healthy volunteers (Rao et al., [[NoYear]]; Kwissa et al., 2014; Popper et al., 2018). Further, targets in the ABC transporters pathway were also obtained from GeneCards (https://www.genecards.org/) and the Comparative Toxicogenomics Database (CTD, https://ctdbase.org/). The intersection of dengue fever and ABC transporters was obtained via Venn to focus on the potential core targets. The protein-protein interaction (PPI) network was constructed by STRING 12.0 (https://string-db.org/). CytoHubba, a Cytoscape plugin (version 3.10.1), was used to find the hub genes. A further whole peripheral blood was collected in part of the subjects. The TRIzol reagent (Invitrogen, CA, USA) was used to extract the total RNA. Then cDNA was generated from the extracted RNA using the HiScript II Reverse Transcriptase. The primer Ct value was detected by real-time fluorescence quantitative PCR(RT-qPCR) three times. The non-parameter tests were performed to calculate the difference in expressed genes of sequencing data and RT-qPCR analyses.

The population characteristics of the dengue fever patients and healthy volunteers were presented in Table 1. There were no significant differences in the sex ratio between the two groups (P < 0.05). There was a statistical difference in age between the two groups, with the patient group being slightly older than the volunteer group. Nonetheless, it is unlikely that subtle age differences would cause any bias in the results of metabolomics studies. The patients were all seen within 1 to 5 days of the acute disorder. Among them, 7 dengue patients were infected with DENV-1 and 34 with the DENV-2 serotype.

PCA and OPLS-DA were used to analyze the LC/MS and GC/MS data obtained from two sets of plasma metabolome sequencing. On the PCA score plot, samples in the same group were clustered closer to each other, while the coordinate points of different samples with significant differences were relatively far away. The OPLS-DA is corrected from PLS-DA (Partial least squares discriminant analysis) by filtering out noise not related to categorical information. In Figures 1A–D, PCA and OPLS-DA analyses clearly distinguished the dengue fever group (orange dots) and the healthy group (blue dots), indicating significant differences in plasma metabolites between the two groups.

Splot diagrams were drawn to represent the eigenvalues and correlations of the metabolites, with metabolites closer to the upper right and lower left corners indicating more significant differences. Both LC/MS (Figure 1E, R2 = 0.462, Q2 = −0.427< 0) and GC/MS (Figure 1F, R2 = 0.645, Q2 = −0.517< 0) obtained good predictions and avoided overfitting.

A total of 2,759 metabolites were identified using an integrated method of LC-MS and GC-MS-based untargeted metabolomics. The criteria for screening differentially expressed metabolites was VIP>1 based on OPLS-DA model analysis and significant inter-group differences based on the t-test (P<0.05). There were 197 metabolites significantly changed in the dengue group compared to the healthy group, of which 61 were up-regulated and 136 were down-regulated (Figure 2A). The volcano plots showing the differential metabolites were shown in Figure 2B.

To better point out the relationship between the samples and the up-and down-regulation differences in metabolite expression between the two groups, heatmap analysis and plotting were performed for VIP Top10 up- and down-regulated metabolites. As shown in Figure 2E, compared to healthy control volunteers, the metabolites in plasma samples significantly altered in dengue fever patients, as indicated by increased levels of trans-cinnamic acid, L-acetylcarnitine, SM(d17:1/17:0), 1,2,4,5-Cyclohexanetetrol, 5-(hydroxymethyl)pyrrolidin-2-One, 1,2,3,4-tetrahydro-6-propanoylpyridine, 2-C-methyl-D-erythritol-4-phosphate, physalolactone, S-japonin, and 9-tridecynoic acid, as well as decreased levels of Pc(16:0/0:0)[Rac], 2-lysophosphatidylcholine, lysoPC (0:0/18:2(9Z,12Z)), PC (0:0/16:0), GM4 (d18:1/16:0), PC (18:0/0:0), PC (16:0/18:1(6E)), PC (18:1(11Z)/0:0), aglepristone, and PC (O-16:0/2:0). Furthermore, 2 up-regulated and 5 down-regulated shared metabolites were identified compared with previous related studies (Figures 2C, D).

Figure 3A showed the distribution of the top 10 up- and down-regulated differential metabolites between the two groups. In addition, the top 10 up- and down-regulated metabolites were also correlated, and there was a negative correlation between the up- and down-regulated metabolites (Figure 3B). The power calculations of top 10 up-regulated metabolites were listed (Table 2).

Receiver operating characteristic curves (ROCs) were constructed using the top 10 up-regulated metabolites of VIP to differentiate dengue from healthy controls. The AUC values of each metabolite were 0.843, 0.824, 0.800, 0.971, 1.000, 1.000, 0.721, 0.948, 0.812 and 0.999 respectively (Figure 3C).

Metabolites in plasma from both groups were mapped to metabolic pathways using the Kyoto Encyclopedia of the Genome (KEGG). ABC transporters, as well as Protein digestion and absorption, aminoacyl-tRNA biosynthesis, Mineral absorption, and D-Amino acid metabolism, were matched with a higher number of metabolites that were significantly enriched at the same time (P < 0.05). The top 20 significantly enriched metabolite pathways were shown in Figures 4A–C. In addition, the mainly perturbed metabolic pathways along with the metabolite data were shown in Table 3, Figure 4D.

We performed pathway enrichment analyses for both up- and down-regulated metabolites (Figure 5A). Notably, ferroptosis pathway was enriched in up-regulated metabolites. The ABC transporters pathway was upregulated in some metabolites and downregulated in others. Thus, the results from the metabolite/gene set enrichment analysis for disease (GSEA) were performed to further identify the activation/suppression status of ABC transporters pathway. A graphical representation of the GSEA results in Figure 5B will contribute to understanding how the pathways were under the suppression status (NES= -1.47< -1, P-value= 0.027<0.05, FDR = 0.199<0.25). Meanwhile, the heat map was used to visualize the trend of expression inhibition of the relevant metabolites (Figure 5C). The box plots for each significantly down-regulated metabolite in ABC transporters were constructed to visualize the differences (Figures 6A–N).

A total of 3401 genes were differentially expressed in datasets of dengue fever (specific screening conditions are |Log FC| > 1, p-value < 0.05, Figures 7A-C). GeneCards and CTDs contained 110 and 44 ABC transporters pathway targets, respectively, for a total of 131 targets included in the final file (Figure 7D). Finally, 16 cross-targets of dengue with ABC transporters were obtained. Figure 7E showed the correlation network and then, using the CytoHubba plug-in, 3 central genes (ABCC5, ABCB1, and ABCG5) were identified via the different algorithms, and showed a down-regulation trend in the transcriptome data (Figures 8A–E). Then the RT-qPCR analyses were conducted to validate the RNA-seq experiments (Figures 8F–H). The primer sequences were shown as Table 4. At last, a flow chart of the study is shown in Figure 9.

Compared with previous studies (Cui et al., 2013), the shared differences in metabolism in dengue patients included 2 up- and 5 down-regulation. Oleic acid, one of the up-regulated metabolites, has proved to be a substance known to increase DENV amplification in natural infection, but not increase the production of dengue virus-like particles (VLPs), that restrict the development of live attenuated vaccines (Ramphan et al., 2017). Docosahexaenoic acid (DHA), another up-regulated metabolic, was positively related to the progression to severe dengue and had the action of regulating inflammatory reactions (Villamor et al., 2018). The shared 5 down-regulation metabolite obtained PC (14:0/0:0), PC (16:1(9Z)/0:0), PC (0:0/16:0), LysoPC (0:0/18:2(9Z,12Z)), PC (18:0/0:0), are all involved in the pathway of phospholipids catabolism. Likewise, the phospholipids showed a decreased and reversible-change trend in DENV-infected humanized mice models (Cui et al., 2017). In the present study, the circulating levels in dengue fever patients of Trans-Cinnamic Acid, L-Acetylcarnitine, SM(d17:1/17:0), 1,2,4,5-Cyclohexanetetrol, 5-(Hydroxymethyl) Pyrrolidin-2-One, 1,2,3,4-Tetrahydro-6-Propanoylpyridine, 2-C-Methyl-D-Erythritol-4-Phosphate, Physalolactone, S-Japonin, and 9-Tridecynoic Acid were higher than those in the control group, and with a high AUC value as well, may function as diagnosis biomarkers to distinguish patients infected with DENV.

The previous study (Cui et al., 2013) revealed that major perturbed metabolic pathways included fatty acid biosynthesis and β-oxidation, phospholipid catabolism, steroid hormone pathway, etc. Relative to the previous research, the Biosynthesis of unsaturated fatty acids was identified as one of the enrichment metabolic pathways in this study. Actually, there was a relative study to highlight both saturated fatty acids and unsaturated fatty acids showed a decreasing trend in the dengue fever patients versus healthy volunteers by GC/MS (Khedr et al., 2015). A previous study had reported that stearoyl-CoA desaturase-1 (SCD1) inhibitor showed potent suppression of DENV replication in a dose-dependent manner, however, exogenous supplementation of unsaturated fatty acids resulted in the reversal of the effect of SCD1, suggesting that biosynthesis of fatty acids contributes to viral replication capacity (Hishiki et al., 2019).

In addition, the following metabolic pathway is enriched as follows: Protein digestion and absorption, aminoacyl-tRNA biosynthesis, Mineral absorption, and D-Amino acid metabolism. The pathway of Protein digestion and absorption was found to be suppressed upon downregulation changes in metabolites with DENV infection. Aminoacyl-tRNAs are the essential substrates for translation depending on the catalyzed reaction of synthetase, which is an essential component involved in protein synthesis and may be potent drug targets to anti-viral (Chakraborti et al., 2021). Furthermore, the inhibition of D-Amino acid metabolism was observed in the present study. Furthermore, the inhibition of D-Amino acid metabolism was observed in the present study, there are no reports about these metabolite features in dengue patients. Research has shown that the level of D-amino acid was dynamically changed in COVID-19 patients, low in the severe and increasing before the recovery (Kimura-Ohba et al., 2023).

ABC transporters, consisting of a ubiquitous superfamily of integral membrane proteins, characteristic of two nucleotide-binding domains (NBDs) and two transmembrane domains (TMDs), have the function of the ATP-powered translocation of many substrates across membranes (Rees et al., 2009). ABC transporters are relevant to the pathological process in broad human diseases and thus may be promising targets for drug discovery (Moore et al., 2023). The ABC transporters may play a key role in arboviral infection and immunity, previous research had reported that most transporters of the ABC subfamily were downregulated via the transcriptomics data (Kumar et al., 2021). Likewise, ABC transporters were altered in the mice’s urine with DENV infection (Zheng and Wang, 2022). However, those metabolic items research of ABC transporters have not been sufficiently differentiated in human dengue patients. We provided the first in vivo evidence that dengue virus infection disrupts the function of ABC transporters in human metabolism. This present study revealed that the ABC transporters pathway was under suppression status. We intended to describe the major disturbance metabolism pathway of ABC transporters in patients with dengue fever, and thus be viewed as the foundation, with more detailed and in-depth studies on the impact to follow. ABC transporters have the function of mediating the uptake of various nutrients, our findings indicated that Protein digestion and absorption, aminoacyl-tRNA biosynthesis, Mineral absorption, and D-Amino acid metabolism were all affected while DENV-infected. From this, the disturbance of ABC transporters is not an isolated event, but rather a key mechanism that causes or exacerbates abnormal distribution of critical intracellular metabolites, thereby impacting viral replication or immunopathology.

ABCC5, ABCB1, and ABCG5 are distributed in distinct families of human ABC transporters and have the function of encoding for membrane transporters, and the dysfunctions are involved in many disease pathways (Alam and Locher, 2023). ABCC5 was proved as a general glutamate conjugate and analog transporter, with the effect of limiting brain levels of endogenous metabolites, drugs, and toxins (Jansen et al., 2015). ABCB1, member 1 in the ATP-binding cassette subfamily B, protects vital organs from outside chemicals and expels medications from malignant cells. The inhibitor molecular to ABCB1 would likely be prescribed as a potential anti-cancer drug (Patel et al., 2024). Moreover, ABCB1 had set complex and diverse regulatory mechanisms in oxidative stress (OS), which mainly exerts a protective effect by preventing the penetration and removing the production of OS, however, it also decreased under the instances with severe OS (Shchulkin et al., 2024). ABCG5 transporters can keep xenosterols away from accumulating in the human body (Patel et al., 2018), the gene variation of ABCG5 would lead to balance disruption and subsequently cause bleeding abnormality and macrothrombocytopenia (Kanaji et al., 2013). Interestingly, thrombocytopenia is a common factor that increases the risk of bleeding in patients with dengue fever, but platelet transfusions are unlikely to improve the prognosis (Archuleta et al., 2020). Based on the analyses above, we propose the potential mechanism that DENV may interfere with the function of the ABC transporter pathway in the host, specifically targeting ABCC5, ABCB1, and ABCG5, and thus enhance disease.