Cano et al. Journal of Nanobiotechnology (2023) 21:54 Journal of Nanobiotechnology https://doi.org/10.1186/s12951-023-01793-7 RESEARCH Open Access Plasma extracellular vesicles reveal early molecular differences in amyloid positive patients with early‑onset mild cognitive impairment Amanda Cano1,2*, Ester Esteban‑de‑Antonio1, Mireia Bernuz3, Raquel Puerta1, Pablo García‑González1,2, Itziar de Rojas1,2, Claudia Olivé1, Alba Pérez‑Cordón1, Laura Montrreal1, Raúl Núñez‑Llaves1, Óscar Sotolongo‑Grau1, Emilio Alarcón‑Martín1, Sergi Valero1,2, Montserrat Alegret1,2, Elvira Martín1, Pamela V. Martino‑Adami4, Miren Ettcheto2,5,6, Antonio Camins2,5,6, Assumpta Vivas7, Marta Gomez‑Chiari7, Miguel Ángel Tejero7, Adelina Orellana1,2, Lluís Tárraga1,2, Marta Marquié1,2, Alfredo Ramírez4,8,9,10,11, Mercè Martí3, María Isabel Pividori3,12, Mercè Boada1,2† and Agustín Ruíz1,2*† Abstract In the clinical course of Alzheimer’s disease (AD) development, the dementia phase is commonly preceded by a prodromal AD phase, which is mainly characterized by reaching the highest levels of Aβ and p-tau-mediated neu‑ ronal injury and a mild cognitive impairment (MCI) clinical status. Because of that, most AD cases are diagnosed when neuronal damage is already established and irreversible. Therefore, a differential diagnosis of MCI causes in these prodromal stages is one of the greatest challenges for clinicians. Blood biomarkers are emerging as desirable tools for pre-screening purposes, but the current results are still being analyzed and much more data is needed to be implemented in clinical practice. Because of that, plasma extracellular vesicles (pEVs) are gaining popularity as a new source of biomarkers for the early stages of AD development. To identify an exosome proteomics signature linked to prodromal AD, we performed a cross-sectional study in a cohort of early-onset MCI (EOMCI) patients in which 184 biomarkers were measured in pEVs, cerebrospinal fluid (CSF), and plasma samples using multiplex PEA technology of 5 † Mercè Boada and Agustín Ruíz Senior co-authors have contributed equally Department of Pharmacology, Toxicology and Therapeutic Chemistry, Faculty of Pharmacy and Food Sciences, University of Barcelona, *Correspondence: 08028 Barcelona, Spain Amanda Cano 6 Institute of Neurosciences, University of Barcelona, Barcelona, Spain acano@fundacioace.org 7 Departament de Diagnòstic Per La Imatge, Clínica Corachan, Barcelona, Agustín Ruíz Spain aruiz@fundacioace.org 8 Department of Neurodegenerative Diseases and Geriatric Psychiatry, 1 Ace Alzheimer Center Barcelona – International University of Catalunya University Hospital Bonn, Medical Faculty, 53127 Bonn, Germany (UIC), C/Marquès de Sentmenat, 57, 08029 Barcelona, Spain 9 German Center for Neurodegenerative Diseases (DZNE), 53127 Bonn, 2 Biomedical Research Networking Centre in Neurodegenerative Diseases Germany (CIBERNED), Madrid, Spain 10 Department of Psychiatry and Glenn, Biggs Institute for Alzheimer’s 3 Grup de Sensors I Biosensors, Departament de Química, Universitat and Neurodegenerative Diseases, San Antonio, TX 78229, USA Autònoma de Barcelona, 08193 Bellaterra, Spain 11 Cluster of Excellence Cellular Stress Responses in Aging-Associated 4 Division of Neurogenetics and Molecular Psychiatry, Department Diseases (CECAD), University of Cologne, 50931 Cologne, Germany of Psychiatry and Psychotherapy, Faculty of Medicine and University 12 Biosensing and Bioanalysis Group, Institut de Biotecnologia I de Hospital Cologne, University of Cologne, 50937 Cologne, Germany Biomedicina (IBB-UAB), Mòdul B Parc de Recerca UAB, Campus Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain © The Author(s) 2023. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. The Creative Commons Public Domain Dedication waiver (http://​creat​iveco​ mmons.​org/​publi​cdoma​in/​zero/1.​0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 2 of 19 ­ link© proteomics. The obtained results showed that proteins measured in pEVs from EOMCI patients with established O amyloidosis correlated with CSF p-tau181 levels, brain ventricle volume changes, brain hyperintensities, and MMSE scores. In addition, the correlations of pEVs proteins with different parameters distinguished between EOMCI Aβ( +) and Aβ(-) patients, whereas the CSF or plasma proteome did not. In conclusion, our findings suggest that pEVs may be able to provide information regarding the initial amyloidotic changes of AD. Circulating exosomes may acquire a pathological protein signature of AD before raw plasma, becoming potential biomarkers for identifying subjects at the earliest stages of AD development. Highlights • The pEVs proteome revealed early molecular differences between EOMCI Aβ( +) and Aβ(-) subjects. • Olink© neurology proteins from pEVs of EOMCI Aβ( +) patients exhibited a strong negative correlation with CSF p-tau181. • The levels of pEVs and CSF O­ link© proteins correlated with CSF p-tau181 levels and age. • Brain MRI characteristics of EOMCI Aβ( +) patients correlated with pEVs protein levels. • pEVs biomarkers only linked early signs of inflammation to brain atrophy in EOMCI Aβ( +) individuals. • The correlation of cognitive status and pEVs biomarkers differed between EOMCI Aβ( +) and Aβ(-) patients. Keywords Plasma exosomes, Extracellular vesicles, Cerebrospinal fluid, Proteomics, Alzheimer’s disease, Mild cognitive impairment Graphical Abstract Introduction for 60–80% of total dementia cases. AD is a neurode- The prevalence of dementia is rapidly growing due to generative condition that progressively and irreversibly social and demographic changes, mainly due to the impairs cognitive functions, resulting in a complete loss increase in the population’s longevity [1]. Consequently, of autonomy [3]. AD is the only condition among the 10 the economic and social impact of the dementia epidemic principal mortality causes worldwide that is still without threatens the sustainability of healthcare systems world- a preventative treatment or cure. Currently, most AD wide [2]. Alzheimer’s disease (AD) is the leading under- cases are diagnosed after irreversible neuronal damage lying cause of dementia in the elderly and is responsible has occurred [4]. Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 3 of 19 The pathophysiological course of AD begins with the overproduction of EVs has been described [11, 12]. This formation of the first senile plaques, composed of extra- process is hypothesized to be a result of the increased neuronal deposits of amyloid-β peptide (Aβ) and amyloid need for cellular communication. It is also supposed vascular deposits, and neurofibrillary tangles, composed to function as a signal for activating immune and non- of intra-neuronal deposits of hyperphosphorylated tau immune processes, regenerating tissues, and recovering (p-tau). These pathological changes lead to synaptic dys- physiological homeostasis or as a spreading mechanism function and, ultimately, neuronal loss, brain atrophy, of disease hallmarks [12, 13]. However, the role of EVs and dementia It is accepted that AD pathogenesis has in AD development is mostly unknown. A possible a very extensive preclinical stage in which the protein pathological signature of AD in circulating exosomes deposits occur silently and the manifestation of symp- might be instrumental for early AD detection and toms is not clinically appreciable. After this silent period, could also provide further knowledge of the underly- the cognitive performance of the individuals begins to ing molecular mechanisms. Many efforts are currently decline [5]. Generally, memory and executive cognitive being made to elucidate the role of EVs in AD, both as alterations appear first. This period corresponds to the pathology spreaders and as diagnostic tools through the mild cognitive impairment (MCI) phase of the disease. early detection of their biomolecule profiles [15]. The Notably, the MCI phase is mainly characterized by reach- leading research groups in the field of biomarkers have ing the highest levels of Aβ and p-tau-mediated neuronal already reported that EVs can propagate Aβ pathology injury and pathological changes in the volumes of differ- in cell cultures [16], are able to create clusters around ent brain regions. However, in this step, MCI subjects the Aβ plaques [17], have reported that pEVs of AD still retain the ability to perform daily life activities inde- patients show altered expression of proteins involved in pendently. Finally, when patients develop dementia, all AD pathogenesis [18–22], have identified abnormal lev- these parameters reach their maximum levels and coex- els of proteins in pEVs in MCI patients who converted ist simultaneously in the already irreversible stages of the to AD dementia [23]. disease [5]. At this point, neurological damage prevents To further investigate the role of circulating EVs in AD, the patient from having normal functionality and per- we decided to initiate our own exosome research pro- sonal autonomy, and behavioral, cognitive, and memory gram. Here, we present a cross-sectional study on two alterations ultimately result in the patient’s death within groups of patients with early-onset MCI (EOMCI) (Aβ ±) approximately 10 years of diagnosis [2, 6]. from the BIOFACE cohort [24] in which we explored the Therefore, since many neurodegenerative diseases, potential of pEVs as an early diagnostic tool for AD by including AD and non-neurodegenerative diseases, may comparing the pEVs proteome profile to the paired CSF present an MCI phase [7], a proper differential diagnosis and plasma proteome of the same individuals. in the prodromal stages of the disease is one of the great- est challenges in clinical practice. Apart from neurologi- Materials and methods cal and neuropsychological evaluations, AD diagnosis is Standard protocol approvals, registrations and patient based on neuroimaging, and cerebrospinal fluid (CSF) consents biomarkers [4]. However, the invasive nature and high All protocols of the BIOFACE study have been approved cost of PET/CSF-biomarkers have promoted the grow- by the Clinical Research Ethics Commission of the Hos- ing scientific interest in peripheral biomarkers, includ- pital Clinic (Barcelona, Spain) in accordance with the ing those derived from plasma and serum [8]. Although current Spanish regulations in the field of biomedical plasma biomarkers are showing very promising results, research and the Declaration of Helsinki. Likewise, in they have not yet been implemented in routine clinical accordance with Spain’s Data Protection Law, all par- practice and are still being studied [9]. Thus, given this ticipants were informed about the study’s goals and scientific interest in plasma biomarkers, recent studies procedures by a neurologist before signing an informed have focused on the potential utility of circulating plasma consent form. Patients’ privacy and data confidentiality extracellular vesicles (pEVs) [10]. were protected in accordance with applicable laws. EVs are nanometric vesicles released by most cell Study participants and selection criteria types, including neurons, that contain proteins, lipids, A total of n = 80 patients diagnosed at Ace Alzheimer metabolites, or RNA [11]. EVs play an important role Center Barcelona with EOMCI (under 65 years old) in communication between neighboring cells and those were included in the BIOFACE study [24, 25]. Accord- of other tissues [12, 13]. Moreover, recent evidence has ing to the International Working Group 2 criteria, sub- shown that EVs can also cross the blood–brain barrier jects with altered biomarkers (a decrease in Aβ1-42 and (BBB) bidirectionally, thus enabling central and periph- an increase in total tau (t-tau) and phosphorylated tau eral communication [14]. In pathological conditions, an proteins at threonine residue 181 (p-tau181) in CSF) were Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 4 of 19 diagnosed with prodromal AD. The inclusion criteria [29]. In brief, 3.5 mL of plasma samples were centrifuged were as follows: (i) age of onset between 50 and 65 years (10,000 g, 30 min, 4 °C) to remove cellular debris, and old; (ii) MMSE score ≥ 26; (iii) Clinical Dementia Rating the supernatant was ultra-centrifuged twice (100,000 g, (CDR) = 0.5; (iv) minimum elementary school level of 60 min, 4 °C) to remove microvesicles and other debris, education (≥ 6 years); (v) willingness to undergo a lum- and then pellet the EVs. All centrifugations were done bar puncture; (vi) capacity to provide written informed with a Sorvall Discovery M150 SE (Thermo Scientific) consent; and (vii) fluency in Spanish. The exclusion cri- Ultracentrifuge using an S100AT6 rotor. The nanoparti- teria included: (i) contraindication for performing brain cle tracking analysis (NTA), measured using a NanoSight magnetic resonance imaging (MRI); (ii) active consump- LM10-HS system with a tuned 405 nm laser (NanoSight tion of alcohol or drugs; (iii) known neurological diseases Ltd., UK), was used to analyze the concentration and associated with cognitive impairment, such as Hunting- particle size of the pEVs. Cryogenic transmission elec- ton’s disease, multiple sclerosis, or large vessel stroke; and tron microscopy (Cryo-TEM) was used to determine the (iv) limited capacity to provide informed consent. morphology of exosomes. Images were collected by a Jeol JEM 2011 (JEOL USA Inc., USA) using an accelerating obtained EVs was measured using the Pierce™ BCA Pro- voltage of 200 kV. The total protein concentration of the Brain MRI images All BIOFACE patients underwent a brain MRI with a Sie- tein Assay Kit (Thermo Fisher). mens VIDA 3 T at Clínica Corachán’s Radiology Depart- ment, (Barcelona, Spain) at the baseline visit as described Proteomics elsewhere [24]. MRI studies were examined by a group of To prepare the pEVs samples for ­Olink© proteomics and experienced neuroradiologists and reported according Pierce™ BCA total protein quantification, pEVs pellets to standard practice. The images were processed at Fun- were lysed with 40 µL of lysis buffer (50 mM TRIS pH dacio ACE ´ Neuroimaging Laboratory. All images were 7.4, 150 mM NaCl, 1 mM EDTA pH 8, 1% Triton × 100, processed with Free surfer 6.0.1 (https://​surfer.​nmr.​mgh.​ 0.01% sodium deoxycholate). Protein concentrations in harva​rd.​edu/). CSF, plasma, and pEVs samples were quantified using the validated, highly sensitive, and specific ProSeek Multiplex Plasma and CSF sample collection immunoassay developed by O ­ link© Proteomics (Uppsala, All samples were collected at the baseline of the study. Sweden) as described elsewhere [30]. Biomarker measure- On the same day, plasma and CSF samples were collected ments were conducted using multiplex Proximity Exten- from each patient. Blood samples were collected in poly- sion Assay (PEA) technology, following the manufacturer’s propylene tubes with EDTA (BD Vacutainer). Plasma was two different commercially available ­ProSeek® Multiplex protocol [31]. In brief, 1 µl of samples were analyzed using separated by centrifugation (2000xg, 10 min, 4 °C), ali- quoted, and stored at -80 °C until use. CSF was obtained panels, Inflammation (code 95302) and Neurology (code by LP. An expert neurologist at Ace Alzheimer Center 95801), which allow the detection of 92 proteins from Barcelona performed LPs in accordance with established each panel simultaneously (Additional file 1: Figures S1 consensus recommendations [26]. The patient was fasted, and S2). Antigens were incubated with pairs of antibod- placed in a sitting position, and anesthetized with 1% sub- ies containing DNA oligonucleotides bound to each of the cutaneous mepivacaine. In polypropylene tubes, 13 mL 184 proteins to be measured [32, 33]. Oligonucleotides in of CSF were collected (Sarstedt Ref 62.610.018). CSF was close proximity produced a template for hybridization and centrifuged for common AD biomarker determination extension. Pre-amplification was performed using a poly- (2000xg, 10 min, 4 °C), and the supernatant was aliquoted merase chain reaction method (PCR). Following digestion and stored at − 80 °C until use. The collection protocol of residual primers, specific primers were digested on a followed the recommendations of the Alzheimer’s Bio- quantitative real-time PCR chip (Dynamic Array IFC; Flui- markers Standardization Initiative [27]. On the day of the digm Biomark) using a Biomark HD Instrument. Protein analysis, an aliquot was thawed and used for the deter- quantities were expressed as normalized protein expression mination of Aβ1-40, Aβ1-42, t-tau, and p-tau181 proteins. (NPX) values on the ­log2 scale. Proteomic measurements of Quantification was performed using a chemiluminescence all the samples were carried out at the same time to avoid enzyme immunoassay (CLEIA) with the Lumipulse G 600 intra- and inter-assay variability. II automatic platform (Fujirebio Inc.) [28]. Statistical analysis Isolation and characterization of pEVs To perform a risk stratification study, subjects from each The gold standard ultra-centrifugation technique was study group were classified into the [A/T/(N)] scheme by used to isolate and purify pEVs from plasma samples converting the CSF levels of Aβ1-42 (A), p-Tau181 (T), and Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 5 of 19 t-Tau (N) into binary variables (abnormal, + ; normal, -) using in house proxy cut-off values [34]. Proxy cut-offs were obtained by plotting receiver operating character- istic (ROC) curves (CSF biomarker level as predictor and conversion as outcome) and calculating the Youden index, that is, the threshold value that provided the best tradeoff between sensitivity and specificity, using the roc and coords functions from the R package pROC. Since positive amyloidosis in CSF is a well-established indicator of an increased risk of phenoconversion to AD dementia [35], subjects were divided into Amyloid A( +) and A(-) groups for comparison. The NPX values for each protein in each sample were calculated as previously described [31] and were used as input for the data analysis pipeline. Values below the LOD and proteins with over 25% missing values were excluded from the analysis. The quality control was performed on both the sample and protein levels. For sample QC, four internal controls were added to each sample to monitor the quality of assay performance as well as the quality of individual sam- ples. The QC was performed in two steps: (1) only data with a standard deviation lower than that of the internal controls (0.2 NPX) were reported; and (2) only samples that deviated less than 0.3 NPX from the median value of the controls were considered to have passed the QC. Regarding protein QC, the detection limit was estimated from negative controls present on every plate, plus three standard deviations. The proteins that showed a missing data frequency of over 25% were also excluded (Addi- tional file 1: Table S1). Statistical analyses were conducted using GraphPad Prism 8.0 and R Studio. CSF levels of Aβ1-42, p-tau181, and t-tau were ­log2-transformed to normalize the data. Dif- ferences between the sexes were also analyzed. The Pear- son correlation between age, sex, Qalb, MMSE, and CSF levels of ­log2-transformed Aβ1-42 and p-tau181 was cal- culated. After the correlation test, all values were Fisher Z-transformed to allow comparison of estimates: Fig. 1 Characterization of pEVs. A Concentration of pEVs samples measured by NTA analysis. Concentration values are expressed Z = 0, 5 ∗ (ln(1 + r) − ln(1 − r)) (1) as value^11. Measurements run by triplicate. B cryo-TEM image of isolated pEVs. Scale bar 200 nm. C Histogram shows the total where r is the Pearson r correlation value. protein concentration of the pEVs samples of EOMCI Aβ( +)/ (-) patients. Statistical analysis was performed with an unpaired Results t test with Welch’s correction. p = 0.0397; Difference between Characterization of pEVs and proteomics means ± SEM = 314.0 ± 143.0; CI (95%) = 611.8 to 16.21 The demographics and basic biochemistry of the BIO- FACE cohort are display in Additional file 1: Table S2. NTA analysis revealed that pEVs samples had a homog- enous particle population, with an average size of statistically significant differences between pEVs samples 98.3 ± 3.7 nm and a concentration of 1.01811 ± 3.7829 of EOMCI Aβ( +) and EOMCI Aβ(-) patients (Fig. 1C). particles/ml (Fig. 1A). Cryo-TEM images showed isolated ­Olink© technology detected 85 proteins in CSF, 91 in pEVs with a spherical shape, smooth surface particle and plasma and 77 in pEVs using the neurology panel, and 61 particle size in accordance with NTA analysis (Fig. 1B). in CSF, 76 in plasma and 57 in pEVs using the inflamma- Interestingly, the concentration of total protein showed tion panel (Additional file 1: Table S1). Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 6 of 19 The pEVs proteome reveals early molecular differences regions were not discernible. However, multiple neurology between EOMCI Aβ( +) and Aβ(‑) subjects proteins demonstrated a clearly polarized correlation with Pearson correlations were performed between O ­ link© pro- WMH in CSF and pEVs (positive and negative, respec- teins measured in three compartments (CSF, plasma, and tively) in EOMCI Aβ( +) patients, whereas EOMCI Aβ(-) pEVs), two patient groups (EOMCI Aβ( +) and EOMCI Aβ(- patients exhibited no correlation (Fig. 5, Table 1). Plasma )), and key AD endophenotypes (Additional file 1: Tables exhibited a similar but less pronounced pattern as pEVs. S3-S5). In the neurology panel, when comparing EOMCI Moreover, both CSF and plasma did not show statistical Aβ( +) and EOMCI Aβ(-) groups, pEVs proteins exhibited differences in the correlation of their protein signatures and statistically significant differences in their correlation with brain region volumes between EOMCI Aβ( +) and EOMCI p-tau181, whereas CSF and plasma compartments did not Aβ(-) patients. Instead, several biomarkers from both the (Fig. 2A). CSF only exhibited statistically significant differ- neurology and inflammation panels showed a significant ences in Aβ1-42 with neurology proteins. In the same way, correlation with the volume of certain brain regions in the when analyzing the O ­ link© inflammation panel, pEVs were pEVs samples of EOMCI Aβ( +) patients. Moreover, some not able to show statistically significant differences in any of proteins such as TRAIL, NTRK2, or PDGFR alpha exhib- the key AD endophenotypes analyzed between Aβ( +) and ited correlations with different brain areas, whereas pEVs Aβ(-) groups. However, some non-significant differences in from EOMCI Aβ(-) patients did not (Table 1). the correlations were observed (Fig. 2B). CSF showed statis- tically significant differences in correlations with p-tau181, pEVs biomarkers only link early signs of inflammation Qalb, and age, whereas plasma showed statistically significant to brain atrophy in EOMCI Aβ( +) individuals differences only with p-tau181 and its inflammation proteins. Several proteins of the inflammatory cascade showed statistically significant differences between EOMCI The levels of pEVs and CSF Olink© proteins correlate Aβ( +) and Aβ(-) patients in pEVs samples. Specifi- with CSF p‑tau181 levels and age cally, IL12B and CXCL11 were significant in CSF and When the magnitude of the effect and the degree of cor- pEVs compartments but not in plasma. Furthermore, relation of the biomarkers with p-tau181 were compared in CXCL5 and CX3CL1 only showed statistical differ- the ­Olink© neurology protein panels, the protein measured ences in pEVs but not in CSF or plasma (Fig. 6). Ven- in pEVs of EOMCI Aβ( +) patients exhibited a strong nega- tricular enlargement is strongly correlated with a tive correlation. In contrast, pEVs of EOMCI Aβ(-) patients decline in cognitive performance, CSF, and pathologic did not exhibit any clear correlation with CSF p-tau181 markers of AD [36]. When comparing the correlation (Additional file 1: Table S3). Intriguingly, the direct meas- of inflammation protein with ventricle volumes, CSF urement of protein levels in CSF did not reveal any differ- and plasma did not show any significant correlation or ences in correlations between EOMCI Aβ( +) and EOMCI difference between EOMCI Aβ( +) and Aβ(-) patients. Aβ(-) patients, with the same proteins exhibiting greater In contrast, pEVs showed a clear difference, with a effects in both patient groups (Additional file 1: Table S4). strong positive correlation of multiple proteins with Plasma exhibited no significant correlation, failing to dis- brain ventricular volumes (Table 1) exclusively in the tinguish between both EOMCI Aβ( +) and EOMCI Aβ(-) samples of EOMCI Aβ( +) patients (Fig. 7). samples (Fig. 3). Of note, when comparing the global degree of correlation between O ­ link© proteins with p-tau181 and The correlation of cognitive status and pEVs biomarkers age simultaneously, pEVs and CSF protein levels exhibited differs between EOMCI Aβ( +) and Aβ(‑) patients strong co-correlations (R2 > 0.5), whereas plasma could not Neurology and inflammation biomarkers did not show exhibit this characteristic. Interestingly, pEVs of EOMCI a specific correlation with MMSE performance in both Aβ( +) patients showed the highest concordance between CSF and plasma samples. Moreover, these correlations protein level correlations with p-tau181 and age (R2 = 0.6287) were not able to differentiate between EOMCI Aβ( +) (Fig. 4A). Moreover, when comparing the co-correlation and Aβ(-) patients. In contrast, although the pEVs did of biomarkers to p-tau181 and white matter hypointensities not show a high degree of correlation, proteins from both (WMH), pEVs of EOMCI Aβ( +) patients also exhibited the the neurology and inflammation panels showed a nega- strongest degree of co-correlation (R2 = 0.5194), whereas nei- tive correlation with MMSE in EOMCI Aβ( +) patients, ther CSF nor plasma could show it (Fig. 4B). which was absent in EOMCI Aβ(-) patients (Fig. 8). Brain MRI characteristics of EOMCI Aβ( +) patients Discussion correlate with pEVs protein levels Although it is well known that the first molecular altera- When comparing EOMCI Aβ( +) and EOMCI Aβ(-) tions of AD pathogenesis can occur up to 15 years before patients’ MRI, differences in the volumes of several brain the onset of clinical symptoms [37–39], a differential Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 7 of 19 Fig. 2 Heatmaps of the Pearson correlation’s of CSF p-Tau181, CSF Aβ1-42, Qalb, Age, Sex and MMSE levels vs A neurology biomarkers and B inflammation biomarkers in CSF, plasma and pEVs samples of Aβ( +) and Aβ(-) EOMCI patients. Statistical analysis was performed with Fisher’s exact test. Baptista-Pike method was used to compute CI’s. Correlations of interest were set at Pearson r > 0.5 and r < − 0.5. p < 0.05 (*). pEVs neurology— CSF pTau181: p = 0.0007; CI (95%) = 0.000 to 0.3355/CSF neurology—CSF Aβ1− 42: p =  < 0.0001; CI (95%) = 5.090 to Infinity/CSF inflammation— CSF pTau181: p =  < 0.0001; CI (95%) = 0.06201 to 0.3390/CSF inflammation—Qalb: p = 0.0275; CI (95%) = 0.000 to 0.6870/CSF inflammation—Age: p =  < 0.0001; CI (95%) = 0.000 to 0.1544/Plasma inflammation—CSF pTau181: p = 0.0135; CI (95%) = 0.000 to 0.6799 Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 8 of 19 Fig. 3 Volcano plots show the significance, expressed by the effect of correlation (­ R2) vs fold-change, expressed by Pearson r, of the correlation of neurology biomarkers vs CSF p-Tau181 levels diagnosis in the early stage of AD development is still yielding EVs with the same size, concentration, and shape one of the greatest challenges nowadays. In this sense, as those described in previous studies [43–47]. Total non-invasive, cost-effective, population-wide pre-screen- protein content was higher in pEVs of EOMCI Aβ( +) ing techniques based on the analysis of AD molecular patients than in Aβ(-) patients. MCI patients with Aβ( +) hallmarks associated with early stages, including pre- status are more likely to develop dementia than MCI amyloidotic stages, are necessary. Numerous efforts are Aβ(-) patients [48]. These results are in agreement with currently being made to investigate the diagnostic poten- those described by Goetzl et al., who found that the lev- tial of plasma biomarkers. Unfortunately, in these early els of numerous classical and alternative pathway com- stages, it is difficult to detect molecular alterations at the plement effector proteins in astrocyte-derived plasma central level in plasma using current analysis methods. exosomes were significantly higher in patients with AD In addition, conventional plasma biomarkers currently than in healthy controls [49]. Curiously, this is not lim- available might not fully reflect AD status and are not ited to neurodegenerative diseases. Sharma et al. also Pierce™ BCA, were higher in patients with melanoma very specific to AD pathophysiology [40–42]. For these found that total exosome protein levels, as measured by reasons, EVs have garnered much interest as potential biomarkers since they are essential communication tools than in healthy controls [46]. This could be related to the between neighboring cells and the periphery. Their pro- increase in cellular communication in pathological con- tein, lipid, or mRNA content is directly related to the ditions. Overall, our findings support previous research cellular processes between the cells involved and their that found increased EVs biogenesis in pathological con- environment. However, EVs profiles and their relation- ditions [11, 12]. ship with early AD processes are understudied, which In relation to AD endophenotypes and covariates, prompted the development and conceptualization of our results showed a correlation between the p-tau181 this work. Thus, we performed a cross-sectional study in levels in CSF and pEVs biomarkers. When analyzing 80 patients with EOMCI Aβ42( +) and EOMCI Aβ42(- the shape of the correlations between the protein lev- ), analyzing 184 protein biomarkers of neurology and els of both panels and the main AD parameters, the inflammation origin in paired samples of CSF, plasma, pEVs showed clear differences between EOMCI Aβ( +) and pEVs. and Aβ(-) patients, highlighting p-tau181. When com- Regarding the characterization assays, the results paring the correlation between neurology biomark- showed that the isolation process of pEVs was successful, ers and p-tau181, plasma showed no clear correlation or Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 9 of 19 Fig. 4 Graphs show the linear regression and equation parameters of the Pearson´s co-correlation of A the correlation of biomarkers vs CSF p-Tau181 and the correlation of biomarkers age; and B the correlation of biomarkers vs CSF p-Tau181 and the correlation of biomarkers white matter hypointensities in CSF, plasma and pEVs samples of EOMCI Aβ ( +)/(-) patients. Dots represent every single protein difference between both subgroups, and CSF showed a Intriguingly, EOMCI subjects with positive amyloidosis clear positive correlation in both cases; pEVs were able showed a fraction of proteins displaying negative corre- to show a clear difference between the pattern of cor- lations between neurology biomarkers and CSF levels of relation between EOMCI Aβ( +) and Aβ(-) patients. p-tau181 in pEVs, suggesting an AD-specific signature of Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 10 of 19 Fig. 5 Volcano plots show the significance, expressed by the effect of correlation (­ R2) vs fold-change, expressed by Pearson r, of the correlation of neurology biomarkers vs white matter hypointensities neurodegeneration in the plasmatic EVs compartment. Together with other pathophysiological factors, tau This negative correlation suggests the hypothesis that the phosphorylation, which is primarily induced by the for- emerging amyloidosis state contributes in some way to mation and accumulation of Aβ oligomers, initiates the a blackout in the neuron-derived EVs proteome. Impor- neurodegenerative process. These alterations also lead to tantly, these results are somehow in agreement with the deregulation of other proteins and cell types in the those found by Goetzl et al., who identified four proteins brain [52]. In this context of amyloidosis and tau phos- of neuronal origin (GluA4-containing glutamate recep- phorylation, our results showed that several neurology tor, neuronal pentraxin 2, neurogranin, and neurexin 2α) biomarkers in pEVs samples were highly negatively corre- that were significantly decreased in the neuronal-derived lated with CSF p-tau181 only in EOMCI Aβ( +) patients. pEVs of AD patients compared to those of healthy con- Most of them are involved in cell growth, synaptic plas- trols [50]. Researchers at the Boston University School of ticity, and neuron-extracellular matrix communication. Medicine showed that AD brain-derived exosomes could Specifically, among the proteins related to also spread tau pathology in healthy mouse interneurons growth factors, we found that GDNFR alpha [51]. Their study proposed a novel mechanism for the 3 (r = − 0.662/R2 = 0.4382) and PDGFR alpha spread of tau in hippocampal GABAergic interneurons (r = − 0.5947/R2 = 0.3319) decreased in pEVs of EOMCI through brain-derived extracellular vesicles and their Aβ( +) patients with the highest levels of CSF p-tau181. In subsequent neuronal dysfunction. Our results showed no both cases, previous studies have pointed out that both significant differences in microtubule-associated protein protein levels are significantly altered in MCI and AD tau (MAPT) levels between EOMCI Aβ( +) and EOMCI patients [53], even in early-stage AD patients, suggest- Aβ(-) patients. In plasma samples, MAPT protein was ing an adaptive process of the impaired brain [54]. Our undetectable by the O ­ link© panel. Our results showed results also highlight the importance of both pathways. © that ­Olink technology was likely insufficiently sensi- Interestingly, pEVs PDGFR alpha not only showed a tive for this purpose at these early stages. These findings strong negative correlation with p-tau181 levels in EOMCI could be related to the previously described dissemina- Aβ ( +) patients, but also with WMH (r = − 0.6340/ tive role of EVs and could be one of the distinctive hall- R2 = 0.4019), the main hallmark of brain vascular impair- marks of the beginning of AD molecular alterations at ment. Likewise, pEVs PDGFR alpha also appeared to be these early stages. strongly correlated with different ventricle volumes only in EOMCI Aβ ( +) patients (see Table 1). Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 11 of 19 Table 1 Correlation between brain volumes and biomarkers that overcame FDR-correction from the BIOFACE cohort. p<0.001 (***), p<0.01 (**) Brain region Biomarker Panel Statistics CSF_ PLASMA_ pEVs _ CSF_ PLASMA_ pEVs _ Aβ42( +) Aβ42( +) Aβ42( +) Aβ42(−) Aβ42(−) Aβ42(−) WM-hypo DDR1 Neurology R2 0.013 0.091 0.581 0.036 0.021 0.007 Pearson r 0.116 − 0.302 − 0.7625 − 0.190898 − 0.144 − 0.086 P value 0.648 0.239 0.000 0.143 0.278 0.540 P summary ns ns *** ns ns ns Left Lat Ventr PRTG​ Neurology R2 0.101 0.001 0.564 0.005 0.008 0.006 Pearson r − 0.317 0.026 0.751 − 0.068 0.087 − 0.079 P value 0.200 0.921 0.001 0.600 0.514 0.562 P summary ns ns *** ns ns ns CPA2 Neurology R2 0.015 0.001 0.532 0.004 0.005 0.048 Pearson r − 0.122 0.030 0.730 − 0.067 − 0.068 − 0.219 P value 0.629 0.909 0.001 0.610 0.609 0.098 P summary ns ns *** ns ns ns PDGF-R- Neurology R2 0.003 0.005 0.466 0.041 0.000 0.032 alpha Pearson r − 0.050 − 0.073 0.682 − 0.203 0.015 − 0.179 P value 0.844 0.781 0.003 0.115 0.910 0.187 P summary ns ns ** ns ns ns TRAIL Inflammatory R2 0.133 0.026 0.466 0.029 0.001 0.002 Pearson r − 0.364 − 0.161 0.682 − 0.169 0.026 0.041 P value 0.137 0.524 0.002 0.193 0.841 0.758 P summary ns ns ** ns ns ns CX3CL1 Inflammatory R2 0.060 0.105 0.483 0.001 0.014 0.000 Pearson r − 0.245 − 0.324 0.695 − 0.035 0.120 − 0.004 P value 0.328 0.190 0.002 0.791 0.362 0.974 P summary ns ns ** ns ns ns Right Lat TRAIL Inflammatory R2 0.166 0.023 0.485 0.018 0.000 0.000 Ventr Pearson r − 0.407 − 0.153 0.696 − 0.133 0.018 0.011 P value 0.094 0.545 0.001 0.307 0.892 0.931 P summary ns ns ** ns ns ns Left Inf Lat NTRK2 Neurology R2 0.137 0.006 0.649 0.003 0.018 0.000 Ventr Pearson r − 0.370 0.079 0.787 − 0.051 − 0.132 0.003 P value 0.131 0.764 0.000 0.698 0.318 0.984 P summary ns ns *** ns ns ns 4th-Ventricle PDGF-R- Neurology R2 0.070 0.060 0.596 0.002 0.000 0.007 alpha Pearson r − 0.265 − 0.246 0.772 0.047 − 0.010 0.081 P value 0.288 0.342 0.000 0.716 0.939 0.551 P summary ns ns *** ns ns ns MSR1 Neurology R2 0.002 0.110 0.560 0.005 0.019 0.011 Pearson r 0.044 0.332 0.749 0.070 0.137 0.105 P value 0.862 0.193 0.000 0.593 0.301 0.433 P summary ns ns *** ns ns ns GZMA Neurology R2 0.011 0.023 0.508 0.007 0.038 0.033 Pearson r 0.106 0.152 0.713 0.084 0.196 0.180 P value 0.675 0.561 0.001 0.520 0.137 0.175 P summary ns ns *** ns ns ns Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 12 of 19 Table 1 (continued) Brain region Biomarker Panel Statistics CSF_ PLASMA_ pEVs _ CSF_ PLASMA_ pEVs _ Aβ42( +) Aβ42( +) Aβ42( +) Aβ42(−) Aβ42(−) Aβ42(−) SPOCK1 Neurology R2 0.073 0.156 0.584 0.026 0.038 0.023 Pearson r − 0.270 0.395 0.764 − 0.161 − 0.195 − 0.153 P value 0.279 0.116 0.001 0.215 0.139 0.284 P summary ns ns *** ns ns ns 2 NTRK2 Neurology R 0.131 0.055 0.572 0.007 0.044 0.004 Pearson r − 0.362 0.234 0.756 − 0.086 − 0.210 − 0.060 P value 0.140 0.365 0.001 0.509 0.110 0.674 P summary ns ns ** ns ns ns 2 NTRK3 Neurology R 0.092 0.064 0.511 0.031 0.033 0.012 Pearson r − 0.303 0.252 0.715 − 0.176 − 0.182 − 0.111 P value 0.222 0.329 0.001 0.174 0.167 0.418 P summary ns ns ** ns ns ns 2 IL8 Inflammatory R 0.007 0.073 0.464 0.001 0.004 0.005 Pearson r − 0.086 0.271 0.681 0.028 0.065 0.069 P value 0.735 0.278 0.002 0.830 0.622 0.605 P summary ns ns ** ns ns ns 2 TRAIL Inflammatory R 0.003 0.023 0.435 0.003 0.004 0.001 Pearson r − 0.057 0.150 0.660 − 0.055 − 0.066 − 0.029 P value 0.822 0.552 0.003 0.676 0.616 0.827 P summary ns ns ** ns ns ns Regarding the proteins involved in the commu- vascular dementia patients by measuring the BCAN con- nication of neurons with the extracellular matrix, centration in CSF [57]. Since EVs s are a means of com- DDR1 (r = − 06.179/R2 = 0.3818) and BCAN munication for neuronal cells, the impairment in the 2 (r = − 0.6196/R = 0.3840) were the most highly corre- synaptic transmission induced by Aβ and p-tau could be lated in the Aβ( +) group. Previous studies have identified related to the reduced levels of BCAN found in pEVs of co-correlated peptide clusters associated with varying patients with already established amyloidosis. levels of p-tau. Many of these are involved in plasticity SPOCK1, also called Testican-1, has also been and extracellular matrix remodeling, including DDR1, related to synaptic function and cell communica- suggesting that they could be involved in the tissue reac- tion and is mainly expressed in neurons and oligo- tion around Aβ plaques [55]. This could be related to the dendrocytes, but its function is still unknown. Our reduced levels found in pEVs of EOMCI Aβ( +) patients results show decreased levels of SPOCK1 in pEVs of but not in Aβ(-) patients. In addition, our results showed EOMCI Aβ ( +) patients with increased CSF p-Tau181 a strong negative correlation between DDR1 levels and (r = − 0.5532/R2 = 0.3060). Another altered protein in WMH (r = − 0.7625/R2 = 0.5813) only in EOMCI Aβ( +) pEVs of EOMCI Aβ( +) patients was sphingomyelin patients. Related to Brevican (BCAN), Jonesco et al. phosphodiesterase 1 (SMPD1), a lysosomal acid sphin- demonstrated in a cross-sectional study that this protein gomyelinase that converts sphingomyelin to ceramide. exhibits differential serological levels in AD, other types Our results demonstrated a highly negative correla- of dementia, and non-dementia patients [56]. Similarly, tion of SMPD1 with p-tau181 in the pEVs compartment Minta et al. were able to discriminate between AD and of EOMCI Aβ( +) patients (r = − 0.6178/R2 = 0.3817). Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 13 of 19 Fig. 6 Histograms show the levels of several inflammatory biomarkers in CSF, plasma and pEVs samples of EOMCI Aβ ( +)/(-) patients. Statistical analysis was performed with an unpaired t test with Welch’s correction. p < 0.05 (*). IL12B pEVs: p = 0.0465; CI (95%) = 0.0051 to 0.6309/IL12B CSF: p = 0.005; CI (95%) = 0.2104 to 0.6800/CXCL11 pEVs: p = 0.0201; CI (95%) = 0.1278 to 1.406/CXCL11 CSF: p = 0.0371; CI (95%) = 0.0287 to 0.8745/ CXCL5 pEVs: p = 0.0486; CI (95%) = 0.0030 to 0,9610/CX3CL1 pEVs: p = 0.0369; CI (95%) = 0.0257 to 0.7659 Finally, contactin-5 protein (CNTN5) also appeared to hypothesis plays a significant role in the development have a strong inverse correlation with CSF p-tau181 lev- of AD [59]. Our results showed that all fluids correlated els (r = − 0.5761/R2 = 0.3319) in CSF and pEVs but not in with WMH and neurology biomarkers in EOMCI Aβ( +) plasma. Tedeschi Dauar et al. evaluated the association patients, being negative in the case of pEVs and plasma of CNTN5 genotype and protein levels with pathological and positive in CSF, whereas EOMCI Aβ(-) patients did hallmarks and clinical manifestations of AD. Their results not exhibit any significant correlation in any fluid. The highlighted that the rs146168 variant of CNTN5 plays a inverse correlation between the central fluid and pEVs role in the risk of developing AD, and that CNTN5 CSF is remarkable. Furthermore, when comparing the cor- levels are strongly correlated with AD pathology, espe- relations of WMH and p-tau181 to neurology biomark- cially in the pre-symptomatic phase of the disease [58]. ers, only pEVs from EOMCI Aβ( +) patients exhibited a Increased tau phosphorylation and resulting axonal dam- strong co-correlation. All of these findings support the age could be behind this reduction in CNTN5 levels hypothesis that pEVs might contain information about a observed in pEVs. specific signature of AD neurodegeneration, which is also The relationship between EVs biomarkers and vascu- related to early vascular alterations in the initial stages of lar impairment was another fascinating finding. Grow- AD [60]. ing scientific evidence suggests that the vascular health Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 14 of 19 Fig. 7 Volcano plots show the significance, expressed by the effect of correlation (­ R2) vs fold-change, expressed by Pearson r, of the correlation of inflammation biomarkers vs 4th ventricle volume The inflammation process is one of the typical hall- and could reflect the ability of EVs to demonstrate the marks of these early stages of AD pathology. Intrigu- well-described inflammation process that occurs in these ingly, pEVs also showed some differences that CSF and early stages. plasma did not. Firstly, pEVs of EOMCI Aβ( +) and Aβ(-) In relation to brain volumes, whereas CSF and plasma subjects exhibited different levels of chemokines and were unable to demonstrate a correlation between interleukins previously implicated in AD-like pathology inflammatory biomarkers and ventricle volumes in and cognitive decline, whereas plasma and some CSF EOMCI Aβ( +) and Aβ(-) patients, pEVs of EOMCI did not. These proteins were CXCL5, CX3CL1, IL12B Aβ( +) patients exhibited a positive correlation, as evi- and CXCL11. Curiously, using the same technology and denced by an enlargement of the 4th ventricle and an neurology/inflammation panels of ­ Olink© proteom- increase in the levels of inflammatory biomarkers. It is ics, Nielsen et al. recently evaluated the protein cargo tempting to speculate that pEVs could provide informa- of plasma-derived EVs in patients with AD, MCI, and tion about brain atrophy processes even before structural healthy controls [61]. Their results demonstrated that MRI could detect the differences in volume measure- CCL11 showed diagnostic capabilities between healthy ments at this early stage. These results are in agreement controls and AD in EV samples. Regarding IL12B, with previous findings that highlighted ventricular many authors have highlighted the association between enlargement as an objective and sensitive measure of abnormal IL12 levels and AD. Vom Berg et al. reported neuropathological changes associated with MCI and AD a decade ago that the inhibition of IL-12/IL-23 signaling progression [66, 67]. Likewise, the relationship between reduced AD-like pathology and cognitive decline [62]. inflammation biomarkers and ventricle volumes has An outstanding study by Pedrini et al. revealed that IL-10 also been described. Thus, Walker et al. evaluated the and IL-12/23p40 were jointly associated as predictors of relationship between systemic inflammation and neu- amyloid-β load in AD patients [63]. Similarly, Johansson rodegeneration by measuring circulating inflammatory et al. showed that CSF IL-12/23 p40 concentration was markers and different brain volumes and found that an decreased in AD and MCI patients compared to healthy increased inflammation composite score was associated controls [64]. Lin et al. found that polymorphisms in with 1788 ­mm3 greater ventricular volume [68]. IL-12-associated genes (including the rs730691 variant The MMSE is a well-known and used short screening in the IL12B gene) were associated with cognitive aging tool for providing an overall measure of cognitive impair- [65]. Our results were in agreement with those findings ment in clinical settings. However, the MMSE cannot Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 15 of 19 Fig. 8 Volcano plots show the significance, expressed by the effect of correlation (­ R2) vs fold-change, expressed by Pearson r, of the correlation of A neurology biomarkers and B inflammation biomarkers vs MMSE be used as a standalone test to identify MCI patients at show statistically significant differences between EOMCI risk of developing dementia [69]. The combination of a Aβ( +) and Aβ(-) patients. However, its correlations with detailed clinical history, cognitive tests, and biomarker both neurology and inflammation biomarkers in pEVs analysis provides a comprehensive diagnosis for patients differed between patients with established amyloidosis in the MCI phase. In our study, the MMSE was unable to and those who did not have the disease. This could be a Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 16 of 19 combined indicator of the cognitive decline that precedes biomarkers for the early diagnosis of AD. However, addi- the advanced phases of the disease’s development, but tional research is required to comprehend the molecu- its relationship is not entirely clear and requires further lar pathways underlying these findings and validate the investigation. obtained results in an independent cohort. However, our study has some limitations. The restricted sample volume and concentration of the pEVs samples Abbreviations significantly conditioned the development of proteomic Aβ Amyloid β assays. Due to the volume/concentration requirements of AD Alzheimer’s disease the proteomics platforms, the full sample was intended BBB Blood-brain barrier CLEIA Chemiluminescense enzyme immunoassay for the ­Olink© proteomics screening assays. Likewise, the CSF Cerebrospinal fluid sample size was restricted to 80 individuals due to the DALYs Disability-adjusted life years complexity of recruiting patients for the study. Although EOMCI Early-onset mild cognitive impairment EVs Extracellular vesicles the EVs isolation process is one of the gold standards LP Lumbar puncture (ultracentrifugation), this technique would not be feasible MCI Mild cognitive impairment for large-scale studies due to the high volume of sample MRI Magnetic resonance imaging NPX Normalized protein expression required (3.5 ml). Due to the exploratory nature of this NTA Nanoparticle Tracking Analysis study, which aimed to evaluate a general AD signature pEVs Plasma extracellular vesicles in circulating pEVs, total EVs were selected. However, PCR Polymerase chain reaction p-tau Hyperphosphorylated tau the use of specific cell-derived EVs would be of interest t-Tau Total Tau for future assays to investigate more specific molecular WMH White matter hypointensities pathways related to the physiopathological processes of the disease development. However, despite being pre- Supplementary Information liminary, our results connect Aβ( +) AD status with some The online version contains supplementary material available at https://​doi.​ key endophenotypes, suggesting that pEVs protein con- org/​10.​1186/​s12951-​023-​01793-7. tent could be related explicitly to early AD processes. Additional file 1: Figure S1. Neurology panel of Olink Proteomics (code Although further molecules must be analyzed in search 95801). Distribution of analytical measuring range, defined by the lower of a specific “EVs AD signature,” we found intriguing evi- and upper limits of quantification (LLOQ-ULOQ), and normal plasma levels dence that EVs may reflect early molecular alterations of where data is available (dark blue bars) for 92 analytes. Figure S2. Inflam‑ mation panel of Olink Proteomics (code 95302). Distribution of analytical the neurodegenerative process. In this regard, additional measuring range, defined by the limits of quantification LLOQ-ULOQ, research is required to validate our findings in independ- for 90 out of 92 analytes. Table S1. List of proteins excluded/included in ent series and identify direct relationships between spe- the study after the quality control analysis. Table S2. Demographics and biochemistry of the BIOFACE cohort. Table S3. Pearson correlations and cific pEVs proteins and AD endophenotypes, including effect between biomarkers in pEVs and most common parameters of AD disease progression. analysis. Data ordered by R2. Displayed proteins above R2 > 0.3. Table S4. Pearson correlations and effect between biomarkers in CSF and most common parameters of AD analysis. Data ordered by R2. Displayed pro‑ Conclusions teins above R2 > 0.3. Table S5. Pearson correlations and effect between At the earliest stages of AD development, molecular biomarkers in plasma and most common parameters of AD analysis. Data alterations are nearly undetectable in complex matrices ordered by R2. Displayed proteins above R2 > 0.3. such as CSF or plasma. Our findings suggest that EVs Acknowledgements may contain very specific information about the molec- Authors acknowledge the support to the BIOFACE study group: Núria Aguil‑ ular processes occurring in their originating cells and era, Diana M. Ariton, Anna Calvet, Urszula Bojarin, Pilar Cañabate, Laia Cañada, microenvironment. These vesicles are capable of provid- María José Castillón, Ana Espinosa, Ainhoa García-Sánchez, Núria Lleonart, Mariola Moreno, Nathalia Muñoz,Ana Belén Nogales, Gemma Ortega, Silvia ing this molecular information, whereas CSF and plasma Preckler, Vanesa Pytel, Mario Ricciardi, Ángela Sanabria, Susana Seguer, María are not. In summary, our results suggest that pEVs have Isabel Ramis, Maitee Rosende-Roca, Juan Pablo Tartari and Liliana Vargas. the potential to reveal the molecular events preceding Author contributions clinical decline. At this early stage, our results may indi- AC performed the experimental assays, designed and conceptualized the cate that their biomarker profile is associated with the study, analyzed and interpreted the data, wrote and revised the manuscript. AD-specific neurodegeneration process governed by EEdA performed the clinical evaluation and patient’s recruitment. MB contrib‑ uted to the experimental assays. RP and PG-G contributed to data acquisition p-tau and Aβ, premature neuroinflammation processes, and harmonization. IdR, CO, AP-C, LM, RN-L, ÓS-G, EA-M, SV, MA, EM, PV. M-A, and brain atrophy. In addition, their peripheral accessi- ME, AC, AV, MG-C, MÁT and AO contributed to data acquisition. LT, MM, AR, bility and low invasiveness make pEVs a potential source MM and MI P supervised the study and contributed to data interpretation. MB conceptualized the study and obtained the financing support. AR conceptual‑ of biomarkers for screening purposes. Thus, the cur- ized, supervised, analyzed and interpreted the data, and wrote the manuscript. rent work sheds light on the search for new peripheral All authors read and approved the final manuscript. Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 17 of 19 Funding A. Cano received support from the Spanish Ministry of Science, Innovation 6. Joling KJ, Janssen O, Francke AL, Verheij RA, Lissenberg-Witte BI, Pieter- and Universities under the grant Juan de la Cierva (FJC2018-036012-I) and Jelle V, et al. Time from diagnosis to institutionalization and death in Instituto de Salud Carlos III (ISCIII) under the grant Sara Borrell (CD22/00125). people with dementia. Alzheimers Dement. 2020;16(4):662–71. P. García-González is supported by CIBERNED employment plan (CNV- 7. Alzheimer’s Association. More than normal aging: understanding mild 304-PRF-866). I. de Rojas is supported by the ISCIII under the grant cognitive impairment. 2022. https://​www.​alz.​org/​media/​Docum​ents/​ FI20/00215. Authors acknowledge the support of the Spanish Ministry of alzhe​imers-​facts-​and-​figur​es-​speci​al-​report.​pdf Science and Innovation, Proyectos de Generación de Conocimiento grants 8. Zetterberg H. Blood-based biomarkers for Alzheimer’ s disease — An PID2021-122473OA-I00, PID2021-123462OB-I00 and PID2019-106625RB-I00. update. J Neurosci Methods. 2019;319:2–6. ISCIII, Acción Estratégica en Salud, integrated in the Spanish National R + D + I 9. Teunissen CE, Verberk IMW, Thijssen EH, Vermunt L, Hansson O, Zet‑ Plan and financed by ISCIII Subdirección General de Evaluación and the Fondo terberg H, et al. Blood-based biomarkers for Alzheimer’s disease: towards Europeo de Desarrollo Regional (FEDER “Una manera de hacer Europa") grants clinical implementation. Lancet Neurol. 2021;S1474–4422(21):00361–6. PI17/01474, PI19/00335, PI22/01403 and PI22/00258. The support of CIBERNED 10. Schöll M, Maass A, Mattsson N, Ashton NJ, Blennow K, Zetterberg H, et al. (ISCIII) under the grants CB06/05/2004 and CB18/05/00010. The support from Biomarkers for tau pathology. Mol Cell Neurosci. 2019;97:18–33. PREADAPT project, Joint Program for Neurodegenerative Diseases (JPND) 11. de Toro J, Herschlik L, Mongini C, Waldner C. Emerging roles of exosomes grant Nº AC19/00097, from HARPONE project, Agency for Innovation and in normal and pathological conditions: new insights for diagnosis and Entrepreneurship (VLAIO) grant Nº PR067/21, from ADAPTED project, the EU/ therapeutic applications. Front Immunol. 2015;6:203. EFPIA Innovative Medicines Initiative Joint Undertaking Grant Nº 115975, and 12. Kalluri R, LeBleu VS. The biology, function, and biomedical applications of from DESCARTES project, German Research Foundation (DFG). The support of exosomes. Science. 2020;367:eaau6977. Fundación bancaria “La Caixa” and Grífols SA (GR@ACE project). 13. Cano A, Ettcheto M, Bernuz M, Puerta R, Esteban de Antonio E, Souto EB, et al. Extracellular vesicles, the emerging mirrors of brain physiopathol‑ Availability of data and materials ogy. Int J Biol Sci. 2023;19(3):721–43. The data that support the findings of this study are available from the cor‑ 14. Wood MJ, O’Loughlin AJ, Lakhal S. Exosomes and the blood–brain barrier: responding authors upon reasonable request. implications for neurological diseases. Ther Deliv. 2011;2(9):1095–9. 15. Zhang T, Ma S, Lv J, Wang X. The emerging role of exosomes in Alzheimer’ s disease. Ageing Res Rev. 2021;68: 101321. Declarations 16. Sardar M, Anna S, Schultz A, Civitelli L, Hildesjö C, Larsson M, et al. Alz‑ heimer’ s disease pathology propagation by exosomes containing toxic Ethics approval and consent to participate amyloid - beta oligomers. Acta Neuropathol. 2018;136(1):41–56. All protocols of the BIOFACE study have been approved by the Clinical 17. Zheng T, Pu J, Chen Y, Mao Y, Guo Z, Pan H. Plasma exosomes spread and Research Ethics Commission of the Hospital Clinic (Barcelona, Spain) in cluster around β -amyloid plaques in an animal model of Alzheimer’ s accordance with the current Spanish regulations in the field of biomedical disease. Front Aging Neurosci. 2017;9:12. research and the Declaration of Helsinki. Likewise, in accordance with Spain’s 18. Fiandaca MS, Kapogiannis D, Mapstone M, Boxer A, Eitan E, Schwartz Data Protection Law, all participants were informed about the study’s goals JB, et al. Identification of preclinical Alzheimer’s disease by a profile of and procedures by a neurologist before signing an informed consent form. pathogenic proteins in neurally derived blood exosomes: a case-control Patients’ privacy and data confidentiality were protected in accordance with study. Alzheimers Dement. 2015;11(6):600-607.e1. applicable laws. 19. Kapogiannis D, Boxer A, Schwartz JB, Abner EL, Biragyn A, Masharani U, et al. Dysfunctionally phosphorylated type 1 insulin receptor substrate in Consent for publication neural-derived blood exosomes of preclinical Alzheimer’s disease. FASEB We give our consent for the manuscript to be published in Journal of J. 2015;29(2):589–96. Nanobiotechnology. 20. Mullins RJ, Mustapic M, Goetzl EJ, Kapogiannis D. Exosomal biomarkers of brain insulin resistance associated with regional atrophy in Alzheimer’s Competing interests disease. Hum Brain Mapp. 2017;38(4):1933–40. All authors have no conflict of interest to state. 21. Goetzl EJ, Boxer A, Schwartz JB, Abner EL, Petersen RC, Miller BL, et al. Altered lysosomal proteins in neural-derived plasma exosomes in pre‑ clinical Alzheimer disease. Neurology. 2015;85(1):40–7. Received: 19 October 2022 Accepted: 24 January 2023 22. Goetzl EJ, Kapogiannis D, Schwartz JB, Lobach IV, Goetzl L, Abner EL, et al. Decreased synaptic proteins in neuronal exosomes of frontotemporal dementia and Alzheimer’s disease. FASEB J. 2016;30(2):4141–8. 23. Winston CN, Goetzl EJ, Schwartz JB, Elahi FM, Rissman RA. Complement protein levels in plasma astrocyte-derived exosomes are abnormal References in conversion from mild cognitive impairment to Alzheimer’s disease 1. Nichols E, Szoeke CEI, Vollset SE, Abbasi N, Abd-Allah F, Abdela J, et al. dementia. Alzheimers Dement (Amst). 2019;11:61–6. Global, regional, and national burden of Alzheimer ’ s disease and other 24. Esteban de Antonio E, Pérez-Cordón A, Gil S, Orellana A, Cano A, Alegret dementias, 1990–2016: a systematic analysis for the Global Burden of M, et al. BIOFACE: a prospective study of risk factors, cognition, and Disease Study 2016. Lancet Neurol. 2019;18:88–106. biomarkers in a cohort of individuals with early-onset mild cognitive 2. Alzheimer’s Association. Alzheimer’s disease facts and figures. Alzheimers impairment. study rationale and research protocols. J Alzheimers Dis. Dement. 2021;17(3):1–103. 2021;83(3):1233–49. A, et al. Automatized FACEmemory® scoring is related to Alzheimer’ 3. Jack CR Jr, Bennett DA, Blennow K, Carrillo MC, Dunn B, Haeberlein S, 25. Alegret M, Sotolongo Grau O, De AEE, Cordón AP, Orellana A, Espinosa Holtzman DM, et al. Toward a biological definition of Alzheimer’s disease. Alzheimers Dement. 2018;14(4):535–62. s disease phenotype and biomarkers in early - onset mild cognitive 4. Cummings J, Isaacson R, Schmitt F, Velting D. A practical algorithm for impairment : the BIOFACE cohort Open Access. Alzheimer’s Res Ther. managing Alzheimer’ s disease: what, when, and why ? Ann Clin Transl 2022;14(1):43. Neurol. 2015;2(3):307–23. 26. Vanderstichele H, Bibl M, Engelborghs S, Le Bastard N, Lewczuk P, 5. Sperling RA, Aisen PS, Beckett LA, Bennett DA, Craft S, Fagan AM, et al. Molinuevo JL, et al. Standardization of preanalytical aspects of cerebro‑ Toward defining the preclinical stages of Alzheimer’s disease: recom‑ spinal fluid biomarker testing for Alzheimer’s disease diagnosis: a con‑ mendations from the National Institute on Aging-Alzheimer’s Association sensus paper from the Alzheimer’s biomarkers Standardization Initiative. workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Alzheimers Dement. 2012;8:65–73. Dement. 2011;7(3):280–92. Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 18 of 19 27. Molinuevo JL, Blennow K, Dubois B, Engelborghs S, Lewczuk P, Perret- 48. Okello A, Koivunen J, Edison P, Archer HA, Turkheimer FE, Någren K, et al. Liaudet A, et al. The clinical use of cerebrospinal fluid biomarker Conversion of amyloid positive and negative MCI to AD over 3 years An testing for Alzheimer’s disease diagnosis: a consensus paper from the 11C-PIB PET study. Neurology. 2009;73(10):754–60. Alzheimer’s biomarkers standardization initiative. Alzheimers Dement. 49. Goetzl EJ, Schwartz JB, Abner EL, Jicha GA, Kapogiannis D. High comple‑ 2014;10(6):808–17. ment levels in astrocyte-derived exosomes of Alzheimer disease. Ann 28. Leitão MJ, Silva-Spínola A, Santana I, Olmedo V, Nadal A, Le BN, et al. Clini‑ Neurol. 2018;83(3):544–52. cal validation of the Lumipulse G cerebrospinal fluid assays for routine 50. Goetzl EJ, Abner EL, Jicha GA, Kapogiannis D, Schwartz JB. Declin‑ diagnosis of Alzheimer’s disease. Alzheimers Res Ther. 2019;11(9):91. ing levels of functionally specialized synaptic proteins in plasma 29. Zhang Y, Bi J, Huang J, Tang Y, Du S, Li P. Exosome: a review of its clas‑ neuronal exosomes with progression of Alzheimer’ s disease. FASEB J. sification, isolation techniques, storage, diagnostic and targeted therapy 2018;32(2):888–93. applications. Int J Nanomed. 2020;15:6917–34. 51. Ruan Z, Pathak D, Kalavai SV, Yoshii-kitahara A, Bhatt N, Takamatsu-yukawa 30. Whelan CD, Mattsson N, Nagle MW, Vijayaraghavan S, Hyde C, Janelidze S, K, et al. Alzheimer’ s disease brain-derived extracellular vesicles spread tau et al. Multiplex proteomics identifies novel CSF and plasma biomarkers of pathology in interneurons. Brain. 2021;144:288–309. early Alzheimer’ s disease. Acta Neuropathol Commun. 2019;7(1):169. 52. Brinkmalm G, Zetterberg H. The phosphorylation cascade hypothesis of 31. Assarsson E, Lundberg M, Holmquist G, Björkesten J, Thorsen SB, Ekman Alzheimer’s disease. Nature Aging. 2021;1:498–9. D, et al. Homogenous 96-plex PEA immunoassay exhibiting high sensitiv‑ 53. Forlenza OV, Miranda AS, Guimar I, Talib LL, Diniz BS, Gattaz WF, et al. ity, specificity, and excellent scalability. PLoS ONE. 2014;9(4): e95192. Decreased neurotrophic support is associated with cognitive decline in 32. Olink©Proteomics. Olink Target 96 Neurology. 2021. https://​olink.​com/​ non-demented subjects. J Alzheimers Dis. 2015;46(2):423–9. produ​cts-​servi​ces/​target/​neuro​logy-​panel/ 54. Straten G, Saur R, Laske C, Gasser T, Annas P, Basun H, et al. Influence of 33. Olink©Proteomics. Olink Target 96 Inflammation. 2021. https://​olink.​com/​ lithium treatment on GDNF serum and CSF concentrations in patients produ​cts-​servi​ces/​target/​infla​mmati​on/ with early Alzheimer’s disease. Curr Alzheimer Res. 2011;8(8):853–9. 34. Orellana A, Garc P, Valero S, Montrreal L, De RI, Hern I, et al. Establishing 55. Morshed N, Lee MJ, Rodriguez FH, Lauffenburger DA, Mastroeni D, White In-House Cutoffs of CSF Alzheimer’ s Disease Biomarkers for the AT (N) FM. Quantitative phosphoproteomics uncovers dysregulated kinase stratification of the Alzheimer Center Barcelona Cohort. Int J Mol Sci. networks in Alzheimer’s disease. Nature Aging. 2021;1:550–65. 2022;23:6891. 56. Jonesco DS, Karsdal MA, Henriksen K. The CNS-specific proteogly‑ 35. Sörensen A, Blazhenets G, Schiller F, Meyer PT, Frings L. Amyloid biomark‑ can, brevican, and its ADAMTS4-cleaved fragment show differential ers as predictors of conversion from mild cognitive impairment to serological levels in Alzheimer’s disease, other types of dementia and Alzheimer’s dementia: a comparison of methods. Alzheimers Res Ther. non-demented controls: a cross-sectional study. PLoS ONE. 2020;15(6): 2020;12(1):155. e0234632. 36. Apostolova LG, Green AE, Babakchanian S, Hwang KS, Chou Y-Y, Toga AW, 57. Minta K, Brinkmalm G, Portelius E, Johansson P, Svensson J, Kettunen P, et al. Hippocampal atrophy and ventricular enlargement in normal aging, et al. Brevican and neurocan peptides as potential cerebrospinal fluid mild cognitive impairment (MCI), and Alzheimer Disease. Alzheimer Dis biomarkers for differentiation between vascular dementia and Alzhei‑ Assoc Disord. 2012;26(1):17–27. mer’s disease. J Alzheimers Dis. 2021;79(2):729–41. 37. Dubois B, Hampel H, Feldman HH, Scheltens P, Aisen P, Andrieu S, et al. 58. Tedeschi Dauar M, Picard C, Rosa-Neto P, Villeneuve S, Poirier J. CNTN5 is Preclinical Alzheimer’s disease: definition, natural history, and diagnostic associated with disease risk and pathology throughout the Alzheimer’s criteria. Alzheimers Dement. 2016;12(3):292–323. disease continuum. Alzheimers Dement. 2021;17(S3): e052359. 38. DeTure MA, Dickson DW. The neuropathological diagnosis of Alzheimer’s 59. Mortamais M, Artero S, Ritchie K. White matter hyperintensities as early disease. Mol Neurodegener. 2019;14(1):32. and independent predictors of Alzheimer’s disease risk. J Alzheimers Dis. 39. Hampel H, Hardy J, Blennow K, Chen C, Perry G, Kim SH, et al. 2014;42:393–400. The amyloid-β pathway in Alzheimer’s disease. Mol Psychiatry. 60. Moreno-Grau S, de Rojas I, Hernández I, Quintela I, Montrreal L, Alegret 2021;26(10):5481–503. M, et al. Genome-wide association analysis of dementia and its clinical 40. Frontera JA, Boutajangout A, Masurkar AV, Betensky RA, Ge Y, Vedvyas endophenotypes reveal novel loci associated with Alzheimer’s disease A, et al. Comparison of serum neurodegenerative biomarkers among and three causality networks: the GR@ACE project. Alzheimers Dement. hospitalized COVID-19 patients versus non-COVID subjects with normal 2019;15(10):1333–47. cognition, mild cognitive impairment, or Alzheimer’s dementia. Alzhei‑ 61. Nielsen JE, Pedersen KS, Vestergård K, Maltesen RG, Christiansen G, mers Dement. 2022;18(5):899–910. Lundbye-christensen S. Novel blood-derived extracellular vesicle-based 41. Pase MP, Himali JJ, Aparicio HJ, Romero JR, Satizabal CL, Maillard P, et al. proximity extension assay. Biomedicines. 2020;8:199. Plasma total-tau as a biomarker of stroke risk in the community. Ann 62. Vom Berg J, Prokop S, Miller KR, Obst J, Kälin RE, Lopategui-Cabezas I, Neurol. 2019;86(3):463–7. et al. Inhibition of IL-12/IL-23 signaling reduces Alzheimer’s disease-like 42. Aparicio H, Himali J, Himali D, Romero J, Lioutas V-A, Pase M, et al. pathology and cognitive decline. Nat Med. 2012;18(12):1812–9. Association of plasma nfl levels with risk of cardiovascular disease in the 63. Pedrini S, Gupta VB, Hone E, Doecke J, O’Bryant S, James I, et al. A framingham heart study (S33.005). Neurology. 2022;98((18 Suppl)):1718. blood-based biomarker panel indicates IL-10 and IL-12/23p40 are jointly 43. Mustapic M, Eitan E, Werner JK Jr, Berkowitz ST, Lazaropoulos MP, Tran J, associated as predictors of β-amyloid load in an AD cohort. Sci Rep. et al. Plasma extracellular vesicles enriched for neuronal origin: a poten‑ 2017;7:14057. tial window into brain pathologic processes. Front Neurosci. 2017;11:278. 64. Johansson P, Almqvist EG, Wallin A, Johansson J-O, Andreasson U, Blen‑ 44. Garcia Cumba LM, Peterson TE, Cepeda MA, Johnson AJ, Parney IF. Isola‑ now K, et al. Reduced cerebrospinal fluid concentration of interleu‑ tion and analysis of plasma-derived exosomes in patients with glioma. kin-12/23 subunit p40 in patients with cognitive impairment. PLoS ONE. Front Oncol. 2019;9:651. 2017;12(5): e0176760. 45. Kumar A, Kim S, Su Y, Sharma M, Kumar P, Singh S, et al. Brain cell-derived 65. Lin E, Kuo P-H, Liu Y-L, Yang AC, Tsai S-J. Association and interaction exosomes in plasma serve as neurodegeneration biomarkers in male effects of interleukin-12 related genes and physical activity on cogni‑ cynomolgus monkeys self-administrating oxycodone. EBioMedicine. tive aging in old adults in the Taiwanese population. Front Neurol. 2021;63: 103192. 2019;10:1065. 46. Sharma P, Diergaarde B, Ferrone S, Kirkwood JM, Whiteside TL. Melanoma 66. Nestor SM, Rupsingh R, Borrie M, Smith M, Accomazzi V, Wells JL, et al. cell-derived exosomes in plasma of melanoma patients suppress func‑ Ventricular enlargement as a possible measure of Alzheimer’s disease tions of immune effector cells. Sci Rep. 2020;10:92. progression validated using the Alzheimer’s disease neuroimaging initia‑ 47. Muller L, Hong C-S, Stolz DB, Watkins SC, Whiteside TL. Isolation of tive database. Brain. 2008;131(9):2443–54. biologically-active exosomes from human plasma. J Immunol Methods. 2014;411:55–65. Cano et al. Journal of Nanobiotechnology (2023) 21:54 Page 19 of 19 67. Ertekin T, Acer N, Köseoğlu E, Zararsız G, Sönmez A, Gümüş K, et al. Total intracranial and lateral ventricle volumes measurement in Alzheimer’s disease: a methodological study. J Clin Neurosci. 2016;34:133–9. 68. Walker KA, Hoogeveen RC, Folsom AR, Ballantyne CM, Knopman DS, Windham BG, et al. Midlife systemic inflammatory markers are associated with late-life brain volume. Neurology. 2017;89(22):2262–70. 69. Arevalo-Rodriguez I, Smailagic N, RoquéiFiguls M, Ciapponi A, Sanchez- Perez E, Giannakou A, et al. Mini-Mental State Examination (MMSE) for the detection of Alzheimer’s disease and other dementias in people with mild cognitive impairment (MCI). Cochrane Database Syst Rev. 2015;2015(3):10783. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub‑ lished maps and institutional affiliations. Ready to submit your research ? Choose BMC and benefit from: • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations • maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions