The Proteome of Plasma Extracellular Vesicles in Latent and Active Tuberculosis Reveals Phase‐Specific Composition and Immunometabolic Insights

ABSTRACT

Tuberculosis (TB) is an infectious disease that affects millions of people worldwide. Extracellular vesicles (EVs) are mediators of intercellular communication that participate in the immunopathogenesis of TB. Here, we characterized plasma EVs from healthy individuals and from individuals with latent (LTB) or active TB (ATB). We found increased total protein concentrations in EVs from LTB and ATB individuals compared to healthy controls, without significant differences in EV concentrations, but reduced exosome abundance in LTB. Proteomic and bioinformatic analysis showed that LTB EVs were enriched in proteins associated with the regulation of inflammation, tissue remodeling, and lipid metabolism, including TGFβ and apolipoproteins (APO) A1 and E. In contrast, ATB EVs were enriched in mycobacterial antigens and host proteins associated with acute inflammation, including serum amyloid A1 and A2 (SAA1/A2) and the calgranulin complex proteins S100A8 and S100A9. These host proteins were central and/or cohesive nodes in the protein–protein interaction networks of these EVs. Dot blot analysis confirmed the increased levels of these EV proteins in a second group of individuals with LTB or ATB. The EV levels of mycobacterial antigens correlated positively with the EV levels of SAA1/A2 and S100A8/A9, but negatively with their levels of TGFβ and APOA1. Principal component analysis and hierarchical clustering suggested that the simultaneous analysis of mycobacterial antigens and these EV proteins could discriminate between healthy, LTB, and ATB individuals. Our findings confirm that plasma EVs provide insights into the host immunometabolic response to TB and may be a source of clinically useful biomarkers.

Plasma extracellular vesicles (EVs) from individuals with latent (LTB) or active tuberculosis (ATB) exhibit distinct proteomic compositions, reflecting the underlying immunometabolic states of each phase. In LTB, EVs are enriched in proteins associated with tissue remodeling and lipid metabolism, including TGFβ, APOA1, and APOE, which may participate in granuloma formation and immune regulation. In ATB, EVs are enriched in proteins associated with acute inflammation, including SAA1/SAA2 and S100A8/S100A9. These proteomic differences could lead to the identification of clinically relevant biomarkers.

Abbreviations

APOA1

apolipoprotein A1

APOE

apolipoprotein E

ATB

active tuberculosis

CD

cluster of differentiation

EDTA

ethylenediaminetetraacetic acid

EV

extracellular vesicle

FDR

false discovery rate

HC

healthy controls

IGRA

Interferon‐Gamma Release Assay

LprG

lipoprotein G (Mtb antigen)

LTB

latent tuberculosis

MS/MS

tandem mass spectrometry

Mtb

Mycobacterium tuberculosis

NTA

nanoparticle tracking analysis

PBS

phosphate‐buffered saline

PCA

principal component analysis

S100A8/S100A9

calgranulin complex proteins

SAA1/SAA2

serum amyloid A1/A2

STRING

search tool for the retrieval of interacting genes/proteins

TB

tuberculosis

TGFβ

transforming growth factor beta

TLR

toll‐like receptor

1. Introduction

Tuberculosis (TB), a disease mainly caused by Mycobacterium tuberculosis (Mtb), remains a major global health challenge. Every year, more than 10 million new cases and over one million deaths are attributed to this disease worldwide, establishing it as the world's deadliest infection caused by a single microorganism [1]. Mtb is an intracellular pathogen that can establish a latent infection or cause an active disease. Active Mtb infection primarily causes a pulmonary disease, but systemic dissemination can occur. The onset of active TB (ATB), or the progression of latent TB (LTB) to ATB, depends on the host immunometabolic status and on host‐pathogen interactions [2]. LTB is characterized by the containment of Mtb‐infected cells within granulomas, in which chronic inflammation limits bacterial dissemination. This clinical form can persist for several years in an asymptomatic and non‐transmissible stage [3]. ATB involves an acute inflammatory response mediated by innate and adaptive mechanisms, and the patient presents with clinical symptoms and has the potential to transmit Mtb to susceptible individuals [3]. LTB is detected through interferon‐gamma release assays (IGRAs), while the diagnosis of ATB relies on clinical symptoms, imaging studies, and microbiological tests [4]. Effective antibiotic regimens for TB exist, but access to and adherence to treatment remain challenging due to limited drug availability, side effects, and long treatment durations [1].

Extracellular vesicles (EVs) are non‐replicative, nano‐sized, bilayer membrane‐enclosed structures that are released by all cells. The EVs released by human cells are classified into apoptotic bodies, microvesicles, and exosomes based on their biogenesis, molecular composition, and size. EVs are involved in intercellular communication and in waste removal [5, 6]. The EVs content is dynamically shaped by the host's physiological state and can include both host‐ and pathogen‐derived molecules [5, 6]. LTB and ATB represent distinct immunometabolic states, despite sharing a common infectious etiology, so the content of plasma EVs may reflect the host immune response in each condition. A comprehensive characterization of these EVs, including their abundance, molecular cargo, corona‐associated proteins, and functional effects on target cells, could lead to a better understanding of the immune and metabolic processes that are differentially modulated in LTB and in ATB. However, a definitive characterization of plasma EVs in LTB and in ATB has not yet been achieved.

Increasing evidence shows that EVs released during Mtb infection participate in host–pathogen interactions. For example, EVs secreted by Mtb‐infected macrophages carry mycobacterial RNA, which activates cytosolic RNA‐sensing pathways in recipient cells, leading to type I interferon production, enhanced phagosome maturation, and increased bacterial killing [7]. EVs derived from infected macrophages also transport mycobacterial proteins, such as Ag85 and ESAT‐6, which can be delivered to uninfected cells and influence both innate and adaptive immune responses, including antigen presentation and T cell activation [8]. Moreover, Mtb‐derived EVs contain lipid and lipoprotein pathogen‐associated molecular patterns that engage TLR2 on host cells, inducing proinflammatory cytokine release and modulating immune responses in the lung microenvironment [9]. So, the characteristics of the EVs released during LTB or ATB may provide meaningful information for disease characterization and for biomarker discovery. In this study, we investigated the phenotypic features and molecular composition of plasma EVs from individuals with LTB and ATB to explore their relevance to the biological processes underlying TB pathophysiology.

2. Materials and Methods

2.1. Ethical Considerations and Inclusion Criteria

This study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics in Research Committee of the General Hospital of Mexico (protocol code: DI/21/503/05/21; date of approval: May 3, 2021). Written informed consent was obtained from all subjects involved in the study. Healthy individuals tested negative for the IGRA QuantiFERON‐TB Gold Plus (Qiagen, Germantown, MD, USA) and had no acute or chronic illnesses. Individuals with LTB had a positive IGRA result but no other comorbidities. Patients with ATB were recruited from the Pneumology Service of the General Hospital of Mexico at the time of their initial diagnosis. Early‐stage pulmonary ATB was confirmed by clinical evaluation, chest imaging, Xpert MTB/RIF (Berkeley, CA, USA), positive culture, and presence of acid‐fast bacilli in sputum smears. All patients were infected with rifampicin‐sensitive Mtb strains and presented no additional comorbidities. Healthy individuals (n = 15) included 6 men and 9 women, with ages from 25 to 59 years. Individuals with LTB (n = 10) included 7 men and 3 women, with ages from 29 to 60 years. ATB patients (n = 15) included 11 men and 4 women, with ages from 28 to 62 years.

2.2. Isolation of Plasma Extracellular Vesicles

Peripheral blood was collected in two 10 mL EDTA tubes per individual and processed within the first hour after sampling. Blood samples with visible hemolysis were excluded. Plasma was obtained by centrifugation at 1200 × g for 10 min at room temperature, followed by centrifugation at 2000 × g for 15 min to obtain platelet‐poor plasma. Platelet‐poor plasma was filtered through a sterile 0.8 μm membrane to remove cell debris. EVs were isolated from 4 mL of filtered platelet‐poor plasma using an exoEasy Maxi kit (Qiagen). EVs were eluted in 2 mL of XE buffer (provided in the kit), aliquoted and used immediately, or stored at −80°C. Aliquots were thawed only once prior to use. All the experimental procedures were performed on these exoEasy‐isolated plasma EVs.

In each group (healthy, LTB, and ATB individuals), the EVs isolated from 10 individuals were used for EV characterization by total protein quantification and nanoparticle tracking analysis. The EVs from 6 of these healthy individuals, from 7 of these LTB individuals, and from 5 of these ATB patients were also analyzed by nano‐flow cytometry. The EVs from 4 of these healthy, LTB or ATB individuals were used for proteomic analysis. The EVs from 5 of these LTB individuals were used for the dot blot assay; these individuals were different from those that were used for proteomic analysis. The EVs isolated from 5 additional healthy individuals and from 5 additional ATB patients were used for the dot blot assay. The EV samples were allocated randomly to each experiment. No samples were excluded from the analyses.

2.3. Characterization of the Extracellular Vesicles

EVs were characterized according to MISEV 2023 recommendations [10]. The characterization included (1) indirect quantification of their molecular content by total protein quantification, (2) determination of their concentration and size distribution with nanoparticle tracking analysis, (3) direct visualization by transmission electron microscopy, and (4) marker identification with nano‐flow cytometry.

2.3.1. Bicinchoninic Acid Assay

Total protein concentration was quantified using the bicinchoninic acid assay (Thermo Fisher Scientific, Waltham, MA, USA), with a bovine serum albumin standard curve. EV samples were lysed in RIPA buffer (Merck, Darmstadt, Germany) prior to quantification. Absorbance at 570 nm was measured in a Multiskan FC (Thermo Fisher Scientific), and data analysis was performed with the SkanIt software v.6.1.1 (Thermo Fisher Scientific). Total protein concentration was used to normalize sample input in all further analytical procedures.

2.3.2. Nanoparticle Tracking Analysis

Particle concentration and size distribution were determined by nanoparticle tracking analysis using a NanoSight NS300 system (Malvern Panalytical, Malvern, Worcestershire, UK) equipped with a 405 nm laser. EV suspensions were diluted in particle‐free PBS to reach the optimal dynamic range. Particle‐free PBS and XE buffer were used as negative controls. For each sample, three 40 s videos were recorded in randomly selected fields, and measurements were processed with Nanoparticle tracking analysis software v.3.44 (Malvern Panalytical). Results were reported as particle size (mean and mode diameter) and particle concentration (particles/mL).

2.3.3. Transmission Electron Microscopy

EV morphology was assessed by transmission electron microscopy using negative staining with uranyl acetate on formvar/carbon‐coated copper grids. Samples were allowed to partially dry prior to visualization. Images were acquired in a JEM‐1400 transmission electron microscope (JEOL Ltd., Tokyo, Japan).

2.3.4. Nano‐Flow Cytometry

EVs were stained with CellVue Claret (Molecular Targeting Technologies, West Chester, PA, USA), anti‐CD35‐PE (RRID:AB_2085021, BioLegend, San Diego, CA, USA), and Annexin V‐APC (BioLegend), or with CellVue Claret, anti‐CD63‐PE (RRID:AB_10896786, BioLegend), and anti‐CD81‐APC (RRID:AB_2564020, BioLegend). EVs were first identified as lipid nanoparticles positive for CellVue Claret. Within this gate, microvesicles were defined as CD35‐PE+/Annexin V‐APC+ events, and exosomes were defined as CD63‐PE+/CD81‐APC+ events. Samples were acquired in a CytoFLEX S flow cytometer (Beckman Coulter, Brea, CA, USA) at a flow rate of 5000 events per second. A total of 500,000 events were captured per sample. Data were analyzed using FlowJo software (RRID:SCR_008520) v.10.8.1 (LCC, Ashland, OR, USA).

2.4. Proteomic Analysis of the Extracellular Vesicles

Four EV samples per group (healthy, LTB, and ATB individuals) were used for proteomic analysis. 500 μL of each EV suspension were dehydrated in a Vacufuge plus (Eppendorf, Hamburg, Germany) under vacuum for 3.5 h at 30°C. Dried samples were stored at −80°C until proteomic analysis, which was performed as previously described [11]. MS/MS spectra were analyzed using Spectronaut software v.12.1 (Biognosys, Zurich, Switzerland) for protein identification. Volcano plots and linear abundance profiles were used to identify the proteins with differential concentrations between the groups. The proteins that were increased in LTB individuals, compared to healthy controls (fold‐change ≥ 0.8, p ≤ 0.01), and the proteins that were increased in LTB individuals, compared to ATB patients (fold‐change ≥ 0.8, p ≤ 0.01), were selected. The proteins that were increased in ATB patients, compared to healthy controls (fold‐change ≥ 2.0, p ≤ 0.01), and the proteins that were increased in ATB patients, compared to LTB individuals (fold‐change ≥ 2.0, p ≤ 0.01), were also selected. Different fold‐change thresholds were applied for LTB and ATB comparisons because these clinical conditions are characterized by distinct inflammatory profiles. ATB is associated with extensive systemic inflammation and tissue damage, which typically results in broader proteomic alterations with higher fold‐change values. In contrast, LTB is associated with more limited inflammation. Therefore, applying the same fold‐change threshold to both conditions could have excluded proteins with potential functional relevance in LTB [12, 13].

The metabolic and functional pathways associated with the selected proteins were determined with the STRING database (RRID:SCR_005223) v.12 [14], using the Gene Ontology Biological Process category and Homo sapiens as the reference species [network type: full STRING network; interaction basis: confidence; minimum required interaction score: high confidence (0.7); disconnected nodes: hidden; clustering method: Markov Cluster Algorithm with inflation value = 3]. Only Gene Ontology terms with a false discovery rate (FDR) < 0.001 were retained. The −log(FDR) values for each term were determined, and the terms were grouped based on their similarity (Jaccard index = 0.4).

From the proteins that were increased in LTB individuals, compared to healthy controls (fold‐change ≥ 0.8, p ≤ 0.01), and from the proteins that were increased in ATB patients, compared to healthy controls (fold‐change ≥ 2.0, p ≤ 0.01), we selected those that belonged to the five Gene Ontology terms with the highest −log(FDR). These proteins were imported into Cytoscape (RRID:SCR_003032) [15], and protein–protein interaction networks were constructed. The maximal clique centrality algorithm was used to identify the proteins with structural centrality in the networks, and the clustering coefficient algorithm was used to identify the proteins that provide cohesion to the networks. With this strategy, we identified enriched and highly connected proteins in the proteome of plasma EVs from LTB and ATB individuals.

2.5. Dot Blot Assay

The dot blot assay was performed as previously reported [16]. Nitrocellulose membranes with a pore size of 0.1 μm were marked with a 1‐cm grid. 5 μg of control proteins or EV samples were applied to each position of the grid. Bovine serum albumin (from the bicinchoninic acid assay kit, Thermo Fisher Scientific) was used as a negative control, while recombinant TGFβ, APOA1, APOE, SAA1/A2, and S100A8/A9 (Thermo Fisher Scientific), and total Mtb extract (obtained from a culture of Mtb H37Rv in Middlebrook medium) were used as positive controls. Membranes were blocked for 1 h at room temperature in PBS containing 0.05% Tween‐20 and 4% skim milk, with gentle agitation. The following primary antibodies were used: anti‐CD63 (RRID:AB_2806456), anti‐CD81 (RRID:AB_2809614), anti‐APOA1 (RRID:AB_10987487), anti‐APOE (RRID:AB_2544564), and anti‐SAA1/A2 (RRID:AB_2815921) from Thermo Fisher Scientific; polyclonal anti‐total Mtb antigens (Cat#MBS534825) and monoclonal anti‐Mtb LprG (Rv1411c, Cat#MBS7162603) from MyBioSource (San Diego, CA, USA); and anti‐TGFβ (RRID:AB_296604) and anti‐S100A8/A9 (RRID:AB_3750749) from Abcam (Waltham, MA, USA). The primary antibodies were diluted from 1:200 to 1:7000 in PBS with 0.05% Tween‐20 and 1% skim milk. The diluted antibodies were added to the membranes and incubated overnight at 4°C. After this incubation, the membranes were washed five times in PBS containing 0.05% Tween‐20. The secondary antibody (anti‐rabbit IgG‐HRP, RRID:AB_2533967, Thermo Fisher Scientific) was diluted 1:2000 in PBS containing 0.05% Tween‐20 and 1% skim milk and incubated for 1 h at 37°C. Membranes were then washed five times with PBS, and the Clarity Western ECL Substrate (Bio‐Rad, Hercules, CA, USA) was added. After 3 min, images were acquired with a ChemiDoc Imaging System (Bio‐Rad). Dot blot images were analyzed with ImageJ software (RRID:SCR_003070) [17] to quantify the intensity of each spot. For each spot, the background intensity was subtracted from the spot intensity. In addition, all measurements were normalized to their respective positive controls (the spot intensity of each recombinant protein was set to 1). The normalized spot intensities were analyzed with Python (RRID:SCR_024202) v.3.13 (Python Software Foundation) to determine Pearson correlation coefficients and to perform principal component analysis (PCA) and hierarchical clustering.

2.6. Statistical Analysis

Statistical analysis was performed with GraphPad Prism software (RRID:SCR_002798) v.10.2.0 (Dotmatics, Boston, MA, USA). The data distribution of each dataset was evaluated using the Kolmogorov–Smirnov test. Comparisons were performed with the Kruskal–Wallis test, followed by Dunn's multiple comparisons test, at a 95% confidence level. Significance thresholds were set at p < 0.05.

3. Results

3.1. Latent Tuberculosis Reduces the Abundance of Exosomes in Plasma Extracellular Vesicles

EVs were isolated from the plasma of healthy, LTB, and ATB individuals. Across the three groups, the EVs had diameters between 100 and 400 nm, a size range consistent with microvesicles (200–1000 nm) and exosomes (50–200 nm) (Figure 1a). A similar vesicular size profile was observed by transmission electron microscopy (Figure 1b). To assess whether Mtb infection influenced the size of plasma EVs, the mean and modal diameters were compared between the three groups. No significant differences were observed in the mean diameters (Figure 2a), but the plasma EVs from individuals with ATB had a significantly smaller modal diameter (164.97 nm) than the plasma EVs from individuals with LTB (222 nm), suggesting a shift in the predominant plasma EV subset in LTB (Figure 2b). No significant differences were observed in total EV concentration between the three groups, with counts ranging from 3 × 10¹⁰ to 1 × 10¹¹ particles/mL in all EV samples (Figure 2c). However, total EV protein concentration was higher in both LTB and ATB compared with healthy controls (Figure 2d,e). These results suggest that, although overall EV abundance is preserved, TB is associated with alterations in EV size distribution and protein content.

FIGURE 1.

Plasma extracellular vesicles in latent and active tuberculosis have heterogeneous sizes. (a) Nanoparticle tracking analysis of plasma EVs from healthy, LTB, and ATB individuals. The graph represents particle concentrations for each particle diameter. Each line represents the average data for 10 healthy, 10 LTB, and 10 ATB individuals. (b) Transmission electron microscopy microphotographs of plasma EVs from a healthy individual. The sample was negatively stained with uranyl acetate and observed in a JEM‐1400 transmission electron microscope. The left panel shows two large vesicles. The right panel shows a small vesicle (red arrow), an intermediate vesicle (blue arrow), and a larger vesicle (green arrow).

FIGURE 2.

Tuberculosis modulates extracellular vesicle size and total protein concentration without altering the total vesicle concentration. (a) Mean particle diameters and (b) modal particle diameters of plasma EVs from healthy controls (HC) and from individuals with LTB or ATB were assessed by nanoparticle tracking analysis. (c) EV concentrations (particles/mL) were measured by nanoparticle tracking analysis. (d) Total EV protein concentration was quantified with the bicinchoninic acid assay. (e) The ratio of total protein concentration to particle concentration was calculated for each EV sample. Individual data points (10 healthy, 10 LTB, and 10 ATB individuals) and median. Kruskal–Wallis test followed by Dunn's multiple comparisons test. *p < 0.05, **p < 0.01, ****p < 0.0001.

To further characterize the plasma EVs from the three groups, we used nano‐flow cytometry, with the membrane‐labeling fluorescent dye CellVue Claret to identify total EVs, followed by staining with anti‐CD35/Annexin V to detect microvesicles and anti‐CD63/anti‐CD81 to detect exosomes (Figure 3a). The proportion of microvesicles remained unchanged in the three groups (Figure 3b), but EVs from LTB and ATB individuals showed a higher dispersion in their microvesicle proportions, compared to healthy controls. A significant reduction in the percentage of exosomes was observed in LTB individuals compared to healthy controls, with a nonsignificant decreasing trend relative to ATB (Figure 3c). This reduction in exosome abundance is consistent with the previously described differences in EV size (plasma EVs from LTB individuals had fewer small vesicles) and suggests that LTB selectively affects the relative contribution of exosomes to the total EV pool.

FIGURE 3.

Latent tuberculosis reduces exosome abundance in plasma extracellular vesicles. (a) Nano‐flow cytometry strategy for microvesicle and exosome identification in plasma EVs. exoEasy‐isolated plasma EVs were stained with CellVue Claret, anti‐CD35‐PE, and Annexin V‐APC, or with CellVue Claret, anti‐CD63‐PE, and anti‐CD81‐APC. EVs were first identified as lipid nanoparticles positive for CellVue Claret. Within this gate, microvesicles were defined as CD35‐PE+/Annexin V‐APC+ events, and exosomes were defined as CD63‐PE+/CD81‐APC+ events. EVs from a healthy control were used to define this gating strategy. (b) Percentage of microvesicles in the plasma EVs of each group. (c) Percentage of exosomes in the plasma EVs of each group. Individual data points (6 healthy, 7 LTB, and 5 ATB individuals) and median. Kruskal–Wallis test followed by Dunn's multiple comparisons test. *p < 0.05.

3.2. Latent and Active Tuberculosis Change the Host Protein Content of Plasma Extracellular Vesicles

To identify the proteins that are present in the plasma EVs of healthy, LTB, and ATB individuals, proteomic characterization of the EVs was performed. Between 500 and 1000 proteins were identified in each EV sample. We confirmed the presence of the tetraspanins CD9, CD63, and CD81, which are commonly used as canonical exosomal markers, in all EV samples [10, 18]. Other EV‐associated proteins previously reported in exosomes (CD59 and CD82) were also detected in all the EV samples. Calnexin, which is indicative of contamination of the EV preparations [10], was below the quantification threshold in all the EV samples. Intracellular or granule‐derived proteins (such as myeloperoxidase, proteinase 3, elastase, and the NADPH oxidase subunits p40^phox, gp91^phox, p22^phox, p67^phox, and p47^phox), which could suggest contamination of the EV preparations [19], were also below the quantification threshold in all the EV samples (Table S1).

To explore the biological significance of the plasma EV content, we performed Gene Ontology enrichment analyses (Figure S1 and Tables S2–S5). EVs from LTB individuals were enriched in proteins associated with the regulation of inflammation, with tissue remodeling, and with lipid metabolism, when compared to EVs from healthy controls. EVs from LTB individuals were also enriched in proteins related to tissue remodeling and complement activation, when compared to EVs from ATB patients. In contrast, EVs from ATB patients were enriched in proteins associated with inflammation and with acute‐phase responses to pathogens, when compared to EVs from healthy controls. EVs from ATB patients were also enriched in some proteins related to inflammation, when compared to EVs from LTB individuals.

We then focused on the proteins that belonged to the five most significantly enriched biological processes [highest −log(FDR)] for the LTB vs. healthy controls and the ATB versus healthy controls comparisons (Tables S2 and S4), and we used them to construct protein–protein interaction networks. The maximal clique centrality algorithm was used to identify the proteins with structural centrality in the networks, and the clustering coefficient algorithm was used to identify the proteins that provide cohesion to the networks (Figure 4). Transforming growth factor β (TGFβ) was the most central node in the LTB network, with the highest maximal clique centrality score. The clustering coefficient analysis identified apolipoprotein A1 (APOA1) and apolipoprotein E (APOE) as highly cohesive nodes (Figure 4a). Maximal clique centrality analysis identified serum amyloid A1 and A2 (SAA1/A2), C‐reactive protein, and haptoglobin as the most central nodes in the ATB network. The clustering coefficient analysis identified SAA1 and SAA2 as highly cohesive nodes, while C‐reactive protein and haptoglobin had lower clustering scores. Instead, the calgranulin complex proteins S100A8 and S100A9 were identified as highly cohesive nodes, forming a module consistent with neutrophil‐mediated inflammatory responses (Figure 4b).

FIGURE 4.

Central and cohesive proteins in the protein networks of LTB and ATB plasma extracellular vesicles. The proteome of plasma EVs from 4 healthy, 4 LTB, and 4 ATB individuals was determined. The proteins that were increased in LTB individuals, compared to healthy controls (fold‐change ≥ 0.8, p ≤ 0.01), and the proteins that were increased in ATB patients, compared to healthy controls (fold‐change ≥ 2.0, p ≤ 0.01), were selected. Then, the proteins that belonged to the five biological processes with the highest −log(FDR) values for each comparison (LTB vs. healthy controls, and ATB vs. healthy controls), were selected. Cytoscape was used to construct protein–protein interaction networks of these proteins for (a) LTB and (b) ATB, and to analyze the networks using the maximal clique centrality and clustering coefficient algorithms. Red circles denote the most central or cohesive nodes identified through maximal clique centrality and clustering coefficient analyses, respectively. Yellow circles denote the least central or cohesive nodes. The networks show the ten most central or cohesive proteins in each analysis. APOA1, apolipoprotein A1; APOE, apolipoprotein E; APOC1, apolipoprotein C1; APP, amyloid precursor protein; CD14 (lipopolysaccharide co‐receptor); CRP, C‐reactive protein; FCGR3A, Fc gamma receptor IIIa (CD16a); HAMP, hepcidin antimicrobial peptide; HP, haptoglobin; ICAM1, intercellular adhesion molecule 1 (CD64); LAMC1, laminin subunit gamma 1; LBP, lipopolysaccharide‐binding protein; LTBP1, latent transforming growth factor β‐binding protein 1; LUM, lumican; PF4, platelet factor 4; TGFβ, transforming growth factor β; THBS1, thrombospondin 1; SAA1, serum amyloid A1; SAA2, serum amyloid A2; SERPINA1, Serpin family A member 1; SPARC, secreted protein acidic and cysteine rich; S100A8 and S100A9 (calgranulin complex proteins); VTN, vitronectin.

3.3. Extracellular Vesicle TGFβ, APOA1, and APOE Are Increased in Latent Tuberculosis, While SAA1/A2 and S100A8/A9 Are Increased in Active Tuberculosis

To confirm the increased levels of TGFβ, APOA1, and APOE in plasma EVs from LTB individuals, and the increased levels of SAA1/A2 and S100A8/A9 in plasma EVs from ATB patients, we performed a dot blot assay with plasma EVs from a new set of healthy, LTB, and ATB individuals. The tetraspanins CD63 and CD81, which are commonly used as canonical markers of exosomes [10, 18], were detected in all the EV samples. EVs from LTB individuals and from ATB patients contained total Mtb antigens, including LprG. The content of these mycobacterial antigens was higher in EVs from ATB patients than in EVs from healthy or LTB individuals. EVs from LTB individuals had higher levels of TGFβ and APOE than EVs from healthy or ATB individuals, and higher levels of APOA1 than EVs from ATB individuals. EVs from ATB patients had higher contents of SAA1/A2 and S100A8/A9 than EVs from LTB individuals (Figure 5a,b).

FIGURE 5.

Plasma extracellular vesicles from LTB individuals have increased contents of TGFβ, APOA1, and APOE, while extracellular vesicles from ATB patients have increased contents of SAA1/A2 and S100A8/A9. (a) Plasma EVs were obtained from healthy, LTB, and ATB individuals (n = 5 per group), and analyzed with dot blot assays. 5 μg of protein were loaded per spot. Albumin was used as a negative control, while Mtb total extract and Mtb LprG, and recombinant TGFβ, APOA1, APOE, SAA1/A2, and S100A8/A9, were used as positive controls. The healthy (HC), LTB, and ATB individuals that were analyzed are listed in each row, and the corresponding primary antibodies (anti‐CD81, anti‐CD63, polyclonal anti‐total Mtb antigens, anti‐Mtb LprG, anti‐TGFβ, anti‐APOA1, anti‐APOE, anti‐SAA1/A2, and anti‐S100A8/A9) are shown above each column. (b) Semi‐quantitative analysis of the proteins based on the normalized spot intensity for each sample. Individual data points and the mean. Kruskal–Wallis test followed by Dunn's multiple comparisons test. *p < 0.05, **p < 0.01.

PCA showed that EV total Mtb antigens and LprG contributed to the separation of the ATB cluster, as did SAA1/A2 and S100A8/A9. TGFβ and APOE contributed to the separation of the LTB cluster, whereas APOA1 was more closely associated with the healthy controls cluster (Figure 6a). Total Mtb antigens and LprG showed strong positive correlations with SAA1/A2 and S100A8/A9 but negative correlations with TGFβ and APOA1. SAA1/A2 and S100A8/A9 also showed negative correlations with TGFβ and APOA1 (Figure 6b). Hierarchical clustering showed that the simultaneous analysis of total Mtb antigens, LprG, SAA1/A2, S100A8/A9, TGFβ, APOA1, and APOE can discriminate between healthy individuals, LTB individuals, and ATB patients (Figure 6c).

FIGURE 6.

SAA1/A2, S100A8/A9, TGFβ, APOA1, and APOE are differentially expressed in plasma extracellular vesicles from healthy, LTB, and ATB individuals. (a) Principal component analysis (PCA) of the proteins in EVs from 5 healthy controls (HC), 5 LTB individuals, and 5 ATB patients. Each point represents an individual EV sample. Vector orientation and length represent the direction and contribution of each protein to the separation of the groups. (b) Pearson correlation matrix showing pairwise comparisons of protein levels across all groups. (c) Hierarchical clustering heatmap grouping EV samples (n = 5 for each group) according to their similarity.

4. Discussion

In this study, we analyzed plasma EVs from individuals with LTB and ATB. For EV isolation, we used the exoEasy Maxi kit, a membrane‐affinity spin‐column method that yields a high recovery of intact EVs. exoEasy‐isolated EVs span a heterogeneous size range, including exosomes and microvesicles [20, 21]. Microvesicles predominantly contain cytoplasmic and organellar proteins, whereas exosomes contain cytoplasmic, organellar, and in some cases, nuclear proteins [22]. We report that the plasma EVs from healthy, LTB, and ATB individuals had diameters between 100 and 400 nm, which is consistent with microvesicles and exosomes. In addition, the three groups had comparable EV protein‐to‐particle ratios.

Protein quantification is standard in EV characterization. Kruh‐García et al. analyzed serum exosomes from patients with ATB, with or without HIV coinfection, and found a higher protein content in the exosomes from these ATB patients, compared to healthy controls and to HIV‐positive individuals [23]. Zhang et al. also reported increased protein levels in serum exosomes from ATB patients compared to healthy controls [24]. In accordance with these findings, we report that LTB and ATB individuals had plasma EVs with higher protein content than the EVs from healthy controls. We also report a decreased exosome production in LTB individuals, as assessed by nano‐flow cytometry, and an increased number of smaller‐diameter vesicles in ATB patients, with total EV concentrations (particles/mL) remaining similar among healthy, LTB, and ATB individuals. The reduced exosome production in LTB individuals could be caused by Mtb antigens, which can modulate host vesicular trafficking pathways [25]. We did not find significant differences in the proportion of microvesicles (CD35+/Annexin V+ EVs) among the three groups, but EVs from LTB and ATB individuals showed a greater variability in microvesicle proportions than EVs from healthy controls. This variability may reflect the biological heterogeneity of Mtb infection, characterized by varying inflammatory, immunological, hematological, and metabolic states among individuals. In addition, phosphatidylserine, which is detected by Annexin V, is associated with cellular activation, membrane remodeling, apoptosis, and procoagulant states, all of which may vary substantially among TB patients [26, 27].

We then characterized the proteome of plasma EVs from healthy, LTB, and ATB individuals. We found that EVs from LTB individuals were enriched in proteins involved in the regulation of inflammation and in tissue remodeling, which may be related to granuloma formation. In contrast, EVs from ATB patients were enriched in proteins associated with inflammation and acute‐phase responses to pathogens, which correlates with the presence of an acute inflammatory response against Mtb in these patients. To analyze the functional significance of EV protein content in LTB and ATB, we selected the proteins that (1) were enriched in plasma EVs from each condition relative to EVs from healthy controls and (2) were central and/or cohesive nodes in the EV protein–protein interaction networks. With these criteria, we identified TGFβ, APOA1, and APOE as characteristic proteins of LTB EVs, and SAA1/A2 and S100A8/A9 as characteristic proteins of ATB EVs.

TGFβ is a key immunoregulatory cytokine associated with granuloma formation and with immune tolerance [28, 29]. APOA1 and APOE bind to and neutralize lipopolysaccharides, thereby reducing neutrophil‐driven inflammation and exerting a protective role in the lungs [30, 31]. APOA1 enhances macrophage activation via TLRs [32], but it also limits chronic inflammation and modulates macrophage lipid metabolism [33, 34]. A previous study found reduced apolipoprotein levels in exosomes from ATB patients compared with healthy controls [35]. Exosomes from patients with drug‐resistant TB have lower APOA1 content than those from patients with non‐drug‐resistant TB, and this decrease in APOA1 may be due to increased cholesterol utilization by resistant Mtb strains [36]. APOE facilitates the uptake of mycobacterial lipid antigens [37, 38] and promotes the establishment of tolerogenic environments [39, 40]. APOE is also involved in cholesterol transport, so APOE deficiency increases cholesterol availability for Mtb, making APOE‐deficient mice more susceptible to ATB [41]. APOE‐deficient mice also overproduce leukotriene B4 and 12‐hydroxyeicosatetraenoic acid, which leads to excessive neutrophil recruitment and severe inflammation [42]. All these studies suggest that TGFβ, APOA1, and APOE are functionally relevant during LTB.

SAA1/A2 are acute‐phase proteins that contribute to inflammation and correlate with innate immune activation [43, 44]. These proteins have been implicated in Mtb pathogenicity. Kawka et al. report that SAA1 is an opsonin that promotes Mtb phagocytosis and antigen presentation by macrophages [45]. However, in human macrophages, treatment of Mtb with SAA1 leads to increased bacterial internalization and survival in a dose‐dependent manner [46]. SAA1 and SAA2 have been linked to TB‐induced bone loss in mice [47]. Pierneef et al. reported that serum SAA1/A2 levels are increased in ATB patients, compared to LTB individuals. Interestingly, serum SAA1/A2 levels are higher in COVID‐19 patients than in those with ATB [48]. Gopal et al. demonstrated that S100A8/A9 are important drivers of lung inflammation during ATB and that neutrophil recruitment depends on these proteins. In a nonhuman primate model of ATB, they demonstrated that increased S100A8/A9 levels were associated with enhanced lung cellular infiltration and tissue damage [49]. Scott et al. found that serum S100A8/A9 levels are increased in ATB patients compared with LTB individuals [50]. These studies suggest that SAA1/A2 and S100A8/A9 are functionally relevant during ATB. However, SAA1/A2 and S100A8/A9 are not specific to TB‐associated inflammation [43, 44, 51]. So, in addition to SAA1/A2 and S100A8/A9, we analyzed the EV content of total Mtb antigens and of LprG, a mycobacterial lipoprotein that participates in mycobacterial lipid transport and in host–pathogen interactions, and that is consistently enriched in EVs derived from Mtb cultures and from infected macrophages [26, 52]. We found that the EV levels of total Mtb antigens and LprG showed strong positive correlations with the levels of SAA1/A2 and S100A8/A9, but negative correlations with the levels of TGFβ and APOA1, suggesting that, in ATB, plasma EVs contain both Mtb antigens and host inflammatory markers.

For our proteome characterization, we analyzed total exoEasy‐isolated EVs, so we did not determine the specific contribution of each EV subset (microvesicles and exosomes) to the EV proteome. exoEasy‐isolated EVs retain their corona, which refers to a set of plasma proteins that are stably adsorbed on EV surfaces. These proteins can modulate EV biodistribution, targeting, and biological activity [53, 54]. The inner corona contains proteins with strong affinity for lipids, while additional proteins form the more dynamic outer corona. Proteins that have been identified in the inner corona of EVs include plasminogen, β‐globulin, serine proteases, apolipoproteins (APOA1, APOB, and APOC3), complement and coagulation proteins, and immunoglobulin chains [53, 54, 55]. Another characteristic of exoEasy‐isolated EVs is that they are co‐isolated with small amounts of soluble plasma proteins and lipoproteins [20, 21]. This limitation is not exclusive to exoEasy, since it is also observed to varying degrees with other EV isolation methods [21, 56, 57]. A limitation of our study is that we cannot determine whether the identified proteins are true EV cargo proteins, corona proteins, or co‐isolated plasma proteins. However, previous studies [35, 36] report changes in apolipoprotein levels in EVs during TB, indicating that differential enrichment of apolipoproteins in EVs is reproducible across independent TB cohorts, so it is unlikely to result exclusively from random EV contamination with plasma proteins. In addition, our bioinformatics analysis showed that, unlike APOA1 and APOE, other plasma proteins were not enriched in EVs from LTB individuals. These results suggest that, even if APOA1 and APOE are not EV cargo proteins, they are still biologically relevant EV‐associated proteins whose differential enrichment reflects host metabolic and immunological responses during LTB.

Other authors have analyzed the proteome of serum or plasma EVs in TB, reporting both Mtb and host proteins that are enriched in ATB patients [23, 24, 58, 59]. The proteins identified in this study (TGFβ, APOA1, and APOE for LTB EVs; and SAA1/A2 and S100A8/A9 for ATB EVs) provide insights into the role of EVs during TB immunopathogenesis. Some of these proteins had been previously analyzed in EVs from TB patients [36], while others had been previously analyzed in the serum of TB patients, or in mouse models of TB, as indicated above. Our study suggests that these proteins are all present in EVs during Mtb infection. The strengths of our study include the analysis of LTB individuals in addition to ATB patients; the use of network analysis to select proteins that are not only enriched in the EVs but that are also biologically relevant for the disease; and the confirmation of the increased expression of the identified proteins in a second group of individuals. However, our study has some limitations. It has been reported that alterations in plasma lipid concentrations may influence the association of apolipoproteins with circulating EVs, and that conditions associated with low‐grade chronic inflammation can alter EV composition [60, 61]. In our study, these factors could have contributed to interindividual variability in EV composition, limiting our ability to attribute the observed proteomic differences to LTB or ATB. Further studies with larger cohorts of individuals are required to determine whether the proteomic differences that we report have a diagnostic utility for LTB or ATB, and whether these proteins could be clinically relevant biomarkers even in the presence of comorbidities or variable plasma lipid concentrations.

5. Conclusion

This study demonstrates that plasma EVs from individuals with LTB and ATB exhibit distinct proteomic compositions, reflecting the underlying immunometabolic states of each clinical phase. LTB and ATB individuals had higher total EV protein concentrations than healthy controls, while LTB was additionally characterized by reduced exosome abundance. In LTB, EVs were enriched in proteins associated with tissue remodeling and lipid metabolism, including TGFβ, APOA1, and APOE; these proteins may be involved in granuloma formation and immune regulation. In contrast, ATB‐derived EVs were enriched in proteins associated with acute inflammation, including SAA1/SAA2 and S100A8/S100A9, and they also contained Mtb antigens such as LprG. Our findings provide insights into the role of EVs during the host immunometabolic response to LTB and ATB.

Author Contributions

Conceptualization, E.S.‐P. and I.W.‐B.; Methodology, J.C.‐C., B.E.G.‐P., C.W.‐B. and V.H.R.‐G.; Software, A.M.‐T.; Validation, L.V.‐C.; Formal analysis, L.V.‐C., A.M.‐T., R.C.‐S., J.S.‐L., C.W.‐B. and A.H.‐S.; Investigation, E.S.‐P., J.C.‐C., M.S.‐F., G.M.O.‐D., K.I.M.‐L.; Data curation, E.S.‐P.; Writing – original draft preparation, E.S.‐P.; Writing – review and editing, I.W.‐B.; Visualization, A.M.‐T., C.W.‐B. and V.H.R.‐G.; Supervision, A.H.‐S. and I.W.‐B.; Project administration, I.W.‐B.; Funding acquisition, R.C.‐S., A.H.‐S. and I.W.‐B. All authors reviewed, revised, and approved the final version of the manuscript and agreed to be accountable for all aspects of the work.

Funding

This research was funded by Secretaría de Ciencia, Humanidades, Tecnología e Innovación (SECIHTI, Mexico, grant CF‐2019/217572 to A.H.‐S. and I.W.‐B.), and by Secretaría de Investigación y Posgrado, Instituto Politécnico Nacional, Mexico (grants 20240231 and 20253674 to I.W.‐B.). E.S.‐P. and J.C.‐C. were recipients of SECIHTI fellowships.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Quality control of plasma extracellular vesicles.

Figure S1: Biological processes associated with the protein content of plasma extracellular vesicles in latent and active tuberculosis. The proteome of plasma EVs from 4 healthy, 4 LTB, and 4 ATB individuals was determined. (a and b) The proteins that were increased in LTB individuals, compared to healthy controls (fold‐change ≥ 0.8, p ≤ 0.01, Table S2), and the proteins that were increased in LTB individuals, compared to ATB patients (fold‐change ≥ 0.8, p ≤ 0.01, Table S3), were selected. (c and d) The proteins that were increased in ATB patients, compared to healthy controls (fold‐change ≥ 2.0, p ≤ 0.01, Table S4), and the proteins that were increased in ATB patients, compared to LTB individuals (fold‐change ≥ 2.0, p ≤ 0.01, Table S5), were also selected. The metabolic and functional pathways associated with these proteins were determined with the STRING database, using the Gene Ontology Biological Process category and the Markov Cluster Algorithm. The graphs show the −log(FDR) values for the 10 most significantly enriched Gene Ontology terms in each comparison, with the terms grouped according to their similarity.

Table S2: EV proteins increased in LTB individuals, compared to healthy controls.

Table S3: EV proteins increased in LTB individuals, compared to ATB individuals.

Table S4: EV proteins increased in ATB individuals, compared to healthy controls.

Table S5: EV proteins increased in ATB individuals, compared to LTB individuals.

Data Availability Statement

The data that support the findings of this study are available in the Materials and Methods, Results, and/or Appendix S1 of this article.