Individual written informed consent was obtained from all study participants before the collection of samples, after receiving oral and written complete information about study objectives and procedures. The project was approved by the Ethics Committee of the Medical Research Institute of the University of Antioquia (committee 010 of May 14, 2019). The confidentiality of the information was guaranteed by using an alphanumeric code to identify the samples of each participant. The protocol for sample collection and research procedures was governed by the provisions of Resolution No. 008430 of 1993 of the Ministry of Health of Colombia.

Descriptive, prospective and cross-sectional study, with convenience sampling, of pregnant women residing in Puerto Libertador, Tierralta or Quibdó. Five study groups were defined according to the presence of pregnancy and Plasmodium and STH infection:

The women were ascribed to the research between 2016-2022 in the obstetric units of the hospital establishments in the municipalities of Puerto Libertador and Tierralta in the south of the department of Córdoba, and Quibdó, in the department of Chocó, which are endemic for malaria and STH (Carmona-Fonseca, 2020; Ministerio de Salud y Protección Social, U. de A, 2015).

The women were permanent residents of the study area at the time of delivery, with prenatal care and delivery attended at the respective local hospital. In addition, they did not present other infections according to the results of the tests for TORCHS and HIV, nor diseases such as high blood pressure, gestational diabetes, preeclampsia-eclampsia, kidney failure, asthma, and they reported a negative result in a serological test for acute dengue virus infection.

The only exclusion criteria were withdrawal of informed consent and failure to obtain samples to diagnose the infectious status.

After inclusion, a questionnaire was completed for each woman to record demographic, clinical, and epidemiological data.

At the time of delivery, peripheral blood samples from the pregnant woman and the placenta were collected in tubes with ethylenediaminetetraacetic acid (EDTA) to make the microscopic diagnosis of plasmodial infection using thick blood smears (TBS). In parallel, fractions of the blood samples were placed on Whatman #3 filter paper for the molecular diagnosis of Plasmodium infection by real-time quantitative PCR (qPCR) (Arango et al., 2013). Additionally, participants were asked for a single sample of approximately 3 g of feces. A fraction of this sample was used to determine the presence of STH (Ascaris lumbricoides, Trichiuris trichiura, and Necator americanus/Ancylostoma duodenale-hookworms) through direct and concentration fecal examinations. The other fraction was used to validate the absence of STH by PCR in the negative samples by microscopy.

The microscopic diagnosis of plasmodial infection was performed by an experienced microscopist in each local hospital, following the criteria suggested by the WHO (World Health Organization, 2023). A thick smear was considered negative when no parasitic form was observed in a minimum of 200 microscopic fields.

The extraction of parasite DNA for the molecular diagnosis of plasmodial infection was carried out with the QIAamp^® DNA Mini kit (QIAGEN) following the manufacturer’s instructions. The diagnosis was made from blood samples collected on Whatmann #3 filter paper. For this, a 6 mm diameter circle was used (~25 µL of blood) and the DNA obtained was resuspended in 50 µL of molecular grade water. qPCR was performed following a previously described procedure (Alvarez-Larrotta et al., 2018; Gavina et al., 2017). Briefly, samples were first analyzed for Plasmodium DNA using genus-specific primers and a hydrolysis probe (Plasprobe). Samples with a cycle threshold (Ct) <45 were analyzed in two species-specific reactions for P. falciparum and P. vivax. Only those samples positive in both genus and species were considered positive. P. vivax (DNA from the Salvador I strain) and P. falciparum (DNA from the 3D7 strain) were used as positive controls. Cycling and detection were performed in the Applied Biosystems StepOnePlus™ real-time PCR system under universal amplification conditions.

Direct fecal smears and concentration examinations (Ritchie-Frick method) for STH diagnosis were carried out by expert professionals from the Parasitology group of the Faculty of Medicine of the University of Antioquia. The number of eggs per gram of feces was calculated to estimate the intensity of infection (Barreto et al., 2017). For molecular diagnosis of STH, DNA was extracted from 200 mg of feces using the Stool DNA Isolation kit (Norgen) and following the manufacturer’s instructions; subsequently a qPCR-Taqman assay was performed according to other authors (Basuni et al., 2011; Sow et al., 2017), in order to validate the absence of A. lumbricoides, T. trichiura, N. americanus/A. duodenale. Cycling and detection were performed in the Applied Biosystems StepOnePlus™ real-time PCR system under universal amplification conditions.

From each participant, 6 mL of whole blood was collected by venipuncture, deposited in Vacutainer tubes (BD Biocienses) with EDTA. The tubes were centrifuged at 400 × g for 10 minutes at room temperature; the entire volume of plasma obtained was collected (approx. 2 mL) and fractionated into 2 mL aliquots to be stored and subsequently frozen at -80°C. The study and characterization of EVs was carried out in the Plasmodium vivax and Exosome Research Group (PvREX) (https://www.pvrex.org/) of the Barcelona Institute for Global Health (ISGlobal) and the Germans Trias i Pujol Research Institute (IGTP) (Badalona, Barcelona, Spain), following sequential steps ( Supplementary Figure 1 ).

EVs were isolated and purified from plasma samples under sterile conditions using class II biosafety hoods. In brief, the plasma was thawed on ice and then centrifuged twice for 10 minutes at 4 °C, first at 2000 × g, and then at 8000 × g, to eliminate cellular debris and excess lipids present in the sample. Sepharose columns were prepared according to previously established protocols (de Menezes-Neto et al., 2015). Briefly: i) a nylon mesh of approximately 1 cm² (filter) was introduced into the needle pivot of sterile syringes; ii) Sepharose CL-2B (Sigma–Aldrich) was packed into the syringes to a final volume of 10 mL by plugging the needle pivot with a three-way valve; iii) it was left to decant overnight at 4°C; iv) then, the columns were equilibrated by adding 10 mL of sterile 1X Phosphate Buffered Saline (PBS) used as elution buffer. Once all the elution buffer passed through the column by gravitation, without allowing the Sepharose column to dry, approximately 1 mL of the plasma, previously centrifuged, was loaded onto each column and fourteen fractions of 500 µL each were collected immediately using sterile 1X PBS, for that, the column was topped up with PBS intermittently during the acquisition of the fourteen fractions of 500 µL. Concentration of soluble proteins, mostly contaminants, contained in the fractions collected after chromatography was measured using the Pierce™ BCA Protein Assay Kit (Thermo Scientific) following the manufacturer’s instructions. The fractions were aliquoted and frozen at -80°C until use in subsequent assays.

We used a bead-based flow cytometry assay (BBA) to identify SEC fractions enriched in the classic EV marker, CD9 (Martin-Jaular et al., 2011; de Menezes-Neto et al., 2015). Briefly, 45 µL of each fraction was mixed with 5 µL of 4 µm aldehyde/sulfate-latex beads (Invitrogen) prediluted 1:10 in 1X PBS, and incubated for 15 minutes at room temperature. As a negative control, 45 µL of 1X PBS was added instead of samples. Coupled beads were blocked by incubation overnight with 1 mL of bead-coupling buffer (BCB), consisting of PBS with 0.1% bovine serum albumin and 0.01% NaN₃, at room temperature with rotation. EV-coupled beads were then centrifuged at 5000 × g for 10 minutes at room temperature, washed once with BCB, and resuspended in 250 μL of BCB before incubation with the primary antibody.

For primary staining of EVs, 5 µL of antibodies against CD9 tetraspanin (Mouse-anti-CD9 human; clone VJ1/20 (Immunostep S.L-9PU-01MG) 1:500 dilution was mixed with 45 µL of sample (EV + beads) in 96-well round bottom dishes and incubated for 30 minutes at 4°C. After incubation and washing, plates were centrifuged at 2500 × g for 10 minutes. Immediately, EV-coupled beads were resuspended in 50 µL of Goat F(ab)2 anti-Mouse IgG (H+L) Human ads-FITC polyclonal secondary antibody (Southern Biotech-1032-02) and incubated in the dark for 30 minutes at 4°C. Subsequently, plates were washed twice with 150 µL of BCB, and centrifuged at 2500 × g for 10 minutes. As a negative staining control (Control), the EV-coupled beads of each fraction were incubated only with secondary antibody. Finally, stained beads were resuspended in 100 µL of 1X PBS and acquisition and reading were performed on the BD FACSLyricTM flow cytometer (BD Biosciences). One hundred beads for each sample were obtained in the cytometer. Analyzes were performed using FlowJo software version 10.6.1 to compare the median fluorescence intensity (MFI) of the EV-coupled beads, thereby measuring the amount of EVs.

The NanoSight LM10 (Malvern Instruments Ltd) equipped with a 638 nm laser and a coupled camera (model F-033) was used. Data were analyzed with NTA software version 3.1, setting the detection threshold to 5, and the blur and maximum jump distance to automatic. To carry out the measurements, fraction obtained by SEC from each plasma sample, with the highest EV enrichment observed in the BBA, was diluted 1:10 with sterile 1X PBS. Readings were taken with single capture for 60 seconds at 30 frames per second (fps), with a camera level set to 16 and with manual temperature monitoring.

For the proteomic analysis of the EVs of each study group, the procedure was as follows: first, 5 samples were randomly selected from the total samples of each study group and 200 µL of the SEC fraction were taken from each of the 5 selected samples more enriched with EVs according to the BBA assay for CD9 (faction with the highest MFI peak), to make a final mixture with a volume of 1 mL for each study group ( Table 1 ). The proteins were then precipitated in the 5 mixtures, for which 250 µL of cold acetone (-20°C) was added to each mixture and incubated for 1 hour at -20°C, and then the proteins were recovered by centrifugation at 16000 × g for 15 minutes at 4°C. The precipitated proteins were processed to carry out the digestion and purification of peptides with the commercial iST kit (PreOmics), according to the manufacturer’s instructions. Finally, they were sent for analysis to the Proteomics Unit of the Center for Genomic Regulation (CRG) of Barcelona-Spain. One µg of each sample was analyzed by LC-MS using a 90-minute gradient on the Orbitrap Eclipse mass spectrometer (Thermo Fisher Scientific). As a quality control, bovine serum albumin samples were analyzed between each sample to avoid carryover and evaluate instrument performance. Data are available via ProteomeXchange with identifier PXD051270.

The variables median fluorescence intensity, EVs size, EVs concentration, soluble protein concentration and others do not have a Gaussian distribution according to the Kolmogorov-Smirnov test, therefore, the non-parametric Kruskal-Wallis (KW) ANOVA test was used to compare the medians of the five study groups. If this test showed a significant difference (p<0.05), Dunn’s test for multiple comparisons was used to identify pairs of groups that differ statistically. The analyzes were done with the GraphPad Prism 8.0.1 program.

Young women (15 to 42 years and median 20 to 25.8 years), height (median between 156 and 158 cm), weight (median between 58 and 61 kg among four groups and 71.8 kg in the fifth, pregnant women without infection), with hemoglobin with median values between 10.8 and 12.6 g/dL, but which is lower in women with an infection compared to those without infection. None of these four variables has a significant difference between the five study groups, but it is clear that infected pregnant women have less than 12 g/dL of hemoglobin, while non-infected women have 12 or more g/dL of hemoglobin, a situation that is clinically important ( Table 2 ).

To determine the SEC fractions in which EVs were enriched, a bead-linked flow cytometry assay or BBA was performed using a classic EV biomarker such as tetraspanin CD9; median fluorescence intensity (MFI) was measured. An illustrative result of the determination of MFI for each of the study groups is shown in Supplementary Figure 2A . The set of results obtained demonstrated the presence, with variable intensity, of tetraspanin CD9 in all samples in each of the study groups. It was also observed that in more than 90% of the samples evaluated, the highest peak of CD9 expression was obtained in SEC fractions 6 or 7, which demonstrates consistency and homogeneity in the preparation of columns and the elution of fractions ensuring the reproducibility and homogeneity of the EVs isolated in the different samples. In all of the samples evaluated, an abundant content of soluble proteins was not found in the fractions where the highest peak MFI was obtained for CD9 (indicator of the amount of EVs). We determined if there was any difference in the expression of CD9 between the five groups evaluated and, although there were no significant differences (p=0.732), there was a tendency to show a higher value of MFI in the expression of CD9 in pregnant women with Plasmodium infection (P) followed by STH infection (G), compared to the other groups ( Supplementary Figure 2A ).

Nanoparticle tracking analysis (NTA) measures the number of particles per mL and the variation in their size. To apply the NTA, the fractions with the highest expression of CD9 in each sample, that is, those richest in EVs, were chosen ( Supplementary Figure 2B ). In general, a more heterogeneous size distribution of the nanoparticles was observed in the EV fractions of the groups of pregnant women with some infection (G, P or PG) compared to the control groups (C and NG). Furthermore, it could be seen that most of the EVs in the C and NG control groups are concentrated in a size range between 50 and 300 nm, while in the groups with some infection the size range was much more variable, between 100 and 700 nm ( Supplementary Figure 2B ).

To determine these possible differences in quantity and size, the groups were compared according to the concentration of particles per mL ( Figure 1A ) and their median size ( Figure 1B ) in each of the groups. When looking into the concentration of particles per mL, the lowest concentration is in the control group NG followed very closely by control group C; the three infected groups (P, G, PG) have values that are 2.5 times or more than the controls and are very similar between them, Interestingly, an slightly higher concentration of particles have been detected in PG coinfection ( Figure 1A ), but there is no statistically significant differences between the groups (p=0.105), probably due to their small sample size. Then, analyzing the median size of the particles in each sample, particles detected ranged between ≈120 nm and ≈200 nm. The lowest values correspond to NG group and the highest to G group. Among the infected, the largest size is for G with no difference between P and PG (p=0.118) ( Figure 1B ). Though, we were able to observe that the particles purified by the SEC, independently from the sample analyzed, have a median size range corresponding to exosomes.

A total of 823 quantifiable proteins with measurable abundance values were identified within the five study groups. The group of non-pregnant women (NG) with 553 proteins and pregnant women infected with Plasmodium (P) with 522 proteins were the ones that showed the highest number of proteins associated with EVs, compared to the other three: healthy pregnant women (C) 449, pregnant women with STH (G) 423 and pregnant women with coinfection (PG) 328 ( Supplementary Figure 3 ).

Of the total quantifiable proteins, 758 were identified as human. After filtering, 11 keratins were excluded as contaminants, resulting in 561 human proteins with at least two unique peptides (UP) per protein group being included in the final EV proteome dataset ( Figure 2A ). Comparison with the top 100 most commonly identified EV proteins in Vesiclepedia revealed that 77% were present in our dataset, demonstrating strong concordance between our proteomic findings and previously published EV studies ( Figure 2B ). EV markers included CD9, Flotillin-1, Annexin A2, and Enolase 1, among others. The heatmap in Figure 2C illustrates the abundance of the top 15 EV markers in our dataset, with NG and P samples exhibiting the broadest range of EV markers compared to C, G and PG samples. Interestingly, hierarchical clustering revealed a separation in protein profiles, distinguishing non-pregnant samples from the others ( Figure 2C ).

We then examined each group individually. A Venn diagram comparing the five groups revealed that leukocyte immunoglobulin-like receptor subfamily B member 2 (LILRB2) was uniquely present in the co-infected women sample. Additionally, 34 proteins were exclusively identified in the P group, while 10 proteins were found only in the G group ( Figure 3 ). Two proteins, a Z-dependent protease inhibitor (SERPINA10) and the C-reactive protein (CRP), were present in all infected groups, corresponding to P, G, and PG groups. Furthermore, 8 proteins were common to both G and P, 7 were shared between G and PG, and 7 were found in both P and PG. In total, 69 human proteins were uniquely identified in the infected groups, encompassing Plasmodium infection, STH infection, and co-infection. Moreover, several proteins were detected only in healthy donor samples. Details on these proteins can be found in Supplementary Table 1 .

To further explore the functional implications of the proteins only identified in the infected samples and not found in the control samples, as well as their combinations, we conducted a GO annotation and a KEGG pathway enrichment analysis. These analyses aimed to elucidate the biological processes, molecular functions, cellular components and pathways associated with the proteome dataset. Regarding enriched biological processes, we identified proteins associated with female pregnancy, heterophilic cell-cell adhesion via plasma membrane, cell adhesion molecules, proteolysis, blood coagulation and regulation of the immune response system, among others ( Figures 4A–C ). Moreover, the analysis revealed significant enrichment in cellular components related to extracellular exosomes, extracellular region, extracellular space, cell surface, and plasma membrane. This suggests a strong presence of proteins involved in extracellular communication through extracellular vesicles. The molecular function enrichment highlighted that many proteins in this dataset are involved in protein tyrosine kinase binding, serine-type endopeptidase activity and calcium ion binding. Finally, the two most prominent KEGG pathways identified were the complement and coagulation cascades and the renin-angiotensin system ( Figure 4D ). Details on detected proteins in patients with Plasmodium and/or STH infections are shown in Tables 3 and 4 .

Finally, the analysis of the data obtained from LC-MS with the Proteome Discoverer v2.5 software, resulted in the identification, with high confidence, of six proteins of P. vivax in the EVs from pregnant women infected only with malaria parasites (group P); 15 proteins of T. trichiura, and one protein of N. americanus ( Tables 3 and 4 ) in the EVs from pregnant women infected only with STH (group G). Unexpectedly, parasite proteins were not identified in the group of coinfected pregnant women (group PG).