The association of high-density lipoprotein cargo proteins with brain volume in older adults in the Atherosclerosis Risk in Communities (ARIC)

Abstract

Background

High-density lipoprotein (HDL) modulates the blood-brain barrier and cerebrovascular integrity, likely influencing the risk of Alzheimer's disease (AD), neurodegeneration, and cognitive decline.

Objective

This study aims to identify HDL protein cargo associated with brain amyloid deposition and brain volume in regions vulnerable to AD pathology in older adults.

Methods

HDL was separated from the plasma of 65 non-demented participants of the Atherosclerosis Risk in Communities (ARIC) study using a fast protein liquid chromatography method. HDL cargo proteins were measured using a label-free, untargeted proteomic method based on mass spectrometry and data-independent acquisition. Linear regression with multiple imputations assessed the associations between each HDL cargo protein (log2-transformed) and brain amyloid deposition or temporal-parietal meta-ROI volume, adjusting for covariates.

Results

The mean (SD) age of the participants was 76.3 (5.4) years old, 53.8% (35/65) female, 30.8% (20/65) black, and 28.1% (18/64, 1 missing) APOE4 carriers. We found few HDL cargo proteins associated with brain amyloid deposition and considerably more HDL cargo proteins associated with temporal-parietal meta-ROI volume. Two HDL cargo proteins mostly associated with temporoparietal meta-ROI volume were fibrinogen B (FGB) and plasminogen (PLG). A doubling of FGB in HDL was associated with a greater temporoparietal meta-ROI volume of 1638 mm³ (95% CI [688, 2589]). In comparison, a doubling of PLG in HDL was associated with a lower temporoparietal meta-ROI of 2025 mm³ (95% CI [-3669, −1034]).

Conclusions

This study suggests that HDL cargo proteins associated with temporal-parietal meta-ROI volume are involved in complement and coagulation pathways.

Keywords

Alzheimer's disease brain amyloid deposition brain volume HDL protein cargo high-density lipoprotein (HDL)

Introduction

High-density lipoprotein (HDL) modulates the blood-brain barrier (BBB) and maintains cerebrovascular integrity, which is essential for preventing cerebral small vessel diseases and Alzheimer's disease (AD).^1,2 HDL's lipid cargo, such as cholesterol, has been well-examined in dementia with conflicting results. Recent studies suggest that higher HDL-C increases the risk of dementia, and specifically AD,^3–5 while other studies argue that small HDL, a subtype enhancing cholesterol efflux capacity, is beneficial for cognition and is reduced in participants with AD.^6,7

HDL are heterogeneous in size and cargo. Although HDL's lipid cargo has been extensively studied, HDL's protein cargo has been studied to a lesser extent and is limited to a few apolipoproteins in AD. For example, the genes for apolipoproteins E and J (apoE and apoJ) are well-established genetic risk factors for AD. Previous studies of apoE and apoJ only used unfractionated plasma and did not physically isolate HDL.^8–13 Furthermore, previous studies failed to consider the hundreds of diverse proteins that plasma lipoproteins carry with physiological functions beyond lipid transport (e.g., apoE and apoJ), such as complement and coagulation.¹⁴

Investigating the functionally diverse proteins that plasma lipoproteins carry requires a new methodology. In the last few years, we have developed an anion exchange fast protein liquid chromatography (AEX-FPLC) method to separate HDL. Furthermore, we established a label-free quantitative mass spectrometry method to identify proteins in fractionated HDL and their relative abundances. Making use of these novel methods for protein quantification, this study aimed to identify HDL cargo proteins associated with two AD-related outcomes: brain amyloid deposition and brain volume in temporal-parietal regions vulnerable to AD-related atrophy (i.e., temporal-parietal meta-region of interest [ROI] volume)¹⁵ in older adults enrolled in the Atherosclerosis Risk in Communities (ARIC) study.

Methods

Study population

The ARIC Study is a prospective epidemiological study conducted in four U.S. communities initially designed to investigate the causes of atherosclerosis and cardiovascular risk disease in a predominantly White and Black population. A detailed study design was published.¹⁶ The ARIC-NCS study is an ancillary study that evaluates the midlife risk factors of dementia.¹⁷ It recruited 6538 participants in 2011–2013. Participants underwent a physical exam, a detailed neurocognitive assessment, and, in selected individuals, a brain MRI. Blood and urine samples were collected, processed, and stored at the time of the exam. For this particular study, the inclusion criteria were: 1) the availability of “frozen never-thawed” fasting plasma samples available at the ARIC-NCS baseline; 2) adjudicated as cognitively normal or mild cognitive impairment (MCI) at the time of the ARIC-NCS baseline blood collection; 3) undergone brain MRI scans at the ARIC-NCS baseline; and 4) undergone PET scans with ¹⁸F-AV-45 (florbetapir) in the ARIC-PET study in 2012–15 (baseline). Of 340 eligible participants, 65 ARIC participants were included in this exploratory study to examine the association of HDL cargo proteins and AD-related outcomes.

Assessment of proteins and relative abundance in fractionated plasma lipoproteins

We isolated plasma HDL using an AEX-FPLC technique¹⁸ on a NGC^TM Quest 10 Fast Protein Liquid Chromatography system equipped with a binary pump, manual injector, a single wavelength (280 nm) ultraviolet (UV) detector, a conductivity monitor and a BioFracTM fraction collector (Bio-Rad Laboratories, Hercules, CA, USA). First, 100 µL of pooled plasma samples were diluted 2X using 100 µL PBS 1X and filtered through a 0.22 µm PES syringe filter to remove particulate material, which helped keep column pressure stable. 100 µL of the diluted plasma sample were injected and plasma lipoprotein isolation was performed on a TSKgel DEAE-NPR anion exchange column (2.5 µm particle size, 3.5 cm×4.6 i.d.) coupled with a TSKgel DEAE-NPR anion exchange guard column (5 µm particle size, 0.5 cm×4.6 i.d.) and a 1 mm anion exchange Opti-guard^® guard column (Supelco). Chromatographic conditions were adapted from Hirowatari et al.¹⁹ and Ji et al..²⁰ Mobile phase buffer A was 50 mM Tris-HCl with 1 mM EDTA, pH 7.5 and buffer B was 50 mM Tris-HCl with 1 mM EDTA and 1 M NaClO₄, pH 7.5 at room temperature. Step gradient for isolating plasma lipoprotein consisted of 5% B from 0–5 min, 10% B from 5–10 min, 12% B from 10–15 min, 14% B from 15–20 min, 16% B from 20–25 min and 50% B from 25–27 min at 0.8 mL/min flow rate. The anion exchange column was equilibrated at 5% B for 30 min before the first injection and re-equilibrated at 5% B for 10 min between sample injections. The chromatographic separation of plasma lipoproteins yielded six fractions (see Supplemental Figure 1). Fractions 2–5 correspond to HDL, LDL, IDL and VLDL, respectively. Each fraction was collected directly into 4 mL Amicon^® Ultra 3 K NMWL centrifugal filters within each respective collection window as follows: Fraction 1 (0.51 −2.36 min), Fraction 2 (6.7–9.1 min); Fraction 3 (11.68 −15.00 min); Fraction 4 (16.66–20.2 min); Fraction 5 (21.83–25.23 min); and Fraction 6 (26.46–28.1 min). Each fraction was collected directly into a 4 mL Amicon^® Ultra 3 K NMWL centrifugal filter and concentrated at 3760 g for 30 min. Resulted volume was transferred to a 0.5 mL Amicon^® Ultra 3 K NMWL centrifugal filter and further concentrated at 14,000 g for another 30 min. Protein concentration of each concentrated fraction was measured by UV at 280 nm (NanoDrop 2000c Spectrophotometer, Thermo Scientific NanoDrop Instruments).

This protocol was applied to frozen-never-thawed plasma samples. Fractions 2–5 were then concentrated using centrifugation filters and protein concentration was measured using a NanoDrop Spectrophotometer. We then used 10 μg of protein from HDL (fraction 2) for proteomics analysis. Briefly, proteins were reduced and alkylated with dithiothreitol (DTT) and iodoacetamide before digestion with Trypsin/ Lys-C. The digested peptides were then desalted and analyzed using a label-free liquid chromatography-mass spectrometry (MS) method based on data-independent acquisition (DIA). Spectronaut^TM (Biognosis, Switzerland), a proprietary software for DIA proteomics analysis, was used to analyze the DIA proteomic data and establish protein identification and relative abundance. For relative protein abundances, protein fragments were first quantified to the peptide level using the mean precursor peak area of the top three most abundant precursors, normalized using the median peptide abundance across all samples, and then log₂-transformed. Peptides are quantified to the protein level based on the mean peptide abundance of the top 3 most abundant peptides. Any missing values were assumed to be missing at random and imputed using multiple imputations and a random-forest algorithm.²¹

Brain amyloid deposition and temporal-parietal meta-ROI volume

Neuroimaging measures were collected as a part of the ARIC-PET Study at the ARIC Visit 5/ ARIC-NCS baseline. Global standard uptake value ratio (SUVR) from [¹⁸F] AV-45 (Florbetapir F18) PET Scans in the ARIC-PET study were used to calculate amyloid deposition in the brain.²² A scan was positive if the global cortical SUVR was more than 1.2 (the sample median).²² A standard set of MRI sequences was performed at each ARIC site and analyzed at the ARIC MRI Reading Center (Mayo Clinic) using methods described in detail previously.²³ Magnetization-prepared rapid acquisition gradient echo (MP-RAGE), axial T2* gradient echo and axial T2 fluid-attenuated inversion recovery (FLAIR) sequences were obtained. Brain volume was calculated using Freesurfer (v.5.1.0; http://surfer.nmr.mgh.harvard.edu) on MP-RAGE sequences. The temporal-parietal meta-ROI volume combines the brain volumes of precuneus, hippocampus, parahippocampal gyrus, entorhinal cortex, and inferior parietal lobules based on associations of this group of regional volumes with AD pathology.¹⁵

Somascan

The SomaScan version 4 platform uses multiplexed modified DNA-based aptamer (Somalogic)²⁴ to measure plasma protein concentrations. SOMAmers are short single-stranded DNA with modified nucleotides and are used as protein-binding reagents. The relative concentrations of 5284 aptamers were quantified on plates at Somalogic using processes previously described with relative fluorescence intensity calibrated using standards on each plate and normalized for plate variation,^24–26 and overall concentration of all proteins in a sample are reported using adaptive normalization by maximum likelihood (ANML).

Other covariates

At ARIC visit 5, as part of the ARIC-NCS, participants underwent a detailed neurocognitive battery, a neurological examination, and a sample of them also had a brain MRI. Using information collected during the ARIC-NCS as well as prior cognitive testing, a committee of experts adjudicated cases of MCI and dementia as previously described.¹⁷

Education level, sex, and race were self-reported by the participant at the study baseline. Information on smoking and medication use was self-reported at all study visits. Body mass index was defined as weight in kilograms divided by the square of height in meters measured with the participant wearing light clothing. Prevalent coronary heart disease, stroke, and heart failure were defined according to published criteria.^27–29 Prevalent diabetes was defined as a self-reported physician diagnosis of diabetes or use of antidiabetic medication. Total cholesterol, HDL cholesterol, and triglycerides were measured in blood samples at the ARIC visit 5. APOE4 genotype was determined using the TaqMan assay (Applied Biosystems, Foster City, CA).³⁰

Statistical analysis

Before analysis, proteins quantified in the HDL with > 40% missingness across the samples were excluded. Among the 125 remaining proteins, the highest missing rate was 38.45%, with an average of 13.5%, consistent with other DIA studies.³¹ Additionally, we excluded the remaining proteins that were not designated as likely HDL proteins in HDL Proteome Watch, a consensus HDL protein database managed by Dr Sean Davidson's group,³² leaving 96 HDL cargo proteins for data analysis. Of the 96 HDL cargo proteins, 77 were measured in plasma using SomaScan (SomaLogic, Boulder, CO).

Linear regression with multiple imputations with chained equations (MICE) was employed to impute missing protein values and assess each protein's association with brain amyloid deposition or temporal-parietal meta-ROI volume. We set the number of imputed datasets to 40 to ensure precise estimates and used 30 iterations for each imputed data to reach convergence. We wanted to determine whether HDL cargo proteins have added values to existing covariates associated with each outcome. Therefore, we first used lasso regressions to select covariates for each outcome (without the inclusion of HDL cargo proteins) from the list of 17 variables, including age, sex, race (black or white), APOE4 status (yes or no), BMI, education, center, cognitive status (cognitive normal or MCI), diabetes (yes or no), history of coronary heart disease (yes or no), heart failure (yes or no), or stroke (yes or no), use of antihypertensive medication (yes or no), use of statin medication (yes or no), HDL-C, triglycerides, and total intracranial volume. For brain amyloid deposition, we selected APOE4 status, education, and use of statin medications; for temporal-parietal meta-ROI volume, age, cognitive status, diabetes, education, prevalence of heart disease, prevalence of heart failure, triglycerides, and total intracranial volume. We then included these selected covariates with each HDL cargo protein in the linear regression models. The p-value threshold for the statistical significance of HDL cargo protein associated with an outcome is less than 0.05.

Results

Characteristics of the study participants

The study included 65 non-demented ARIC participants. The mean (SD) age was 76.3 (5.4) years, 54% (or 35/65) were female, 31% (or 20/65) were Black individuals; 72% (or 47/65) were cognitively normal (CN), and 28% (or 18/65) had MCI. Table 1 details the participants’ characteristics. Of note, the study participants’ characteristics were not significantly different from those eligible but not selected for the study (Supplemental Table 1).

Table 1.

Characteristics of the study participants.

	Total
N	65
Age (y)	76.3 (5.4)
Gender
Female (n [%])	35 (54%)
Male (n [%])	30 (46%)
Race
Black (n [%])	20 (31%)
White (n [%])	45 (69%)
BMI (kg/m²)	29.7 (5.2)
Cognitive Status Diagnosis
Normal (n [%])	47 (72%)
MCI (n [%])	18 (28%)
APOE4 Carrier Status
No (n [%])	47 (72%)
Yes (n [%])	18 (28%)
Education Level
Basic-less than completed high school (n [%])	13 (20%)
Intermediate-completed high school or equivalent (n [%])	26 (40%)
High-completed at least some college (n [%])	26 (40%)
ARIC Field Center
Forsyth County, NC (n [%])	19 (29%)
Jackson City, MS (n [%])	19 (29%)
Washington County, MD (n [%])	27 (42%)
Current Cigarette Smoking Status
Non-smoker (n [%])	59 (91%)
Smoker (n [%])	5 (8%)
Missing (n [%])	1 (2%)
Diabetes Prevalence
No (n [%])	43 (66%)
Yes (n [%])	21 (32%)
Missing (n [%])	1 (2%)
Hypertension Prevalence
No (n [%])	20 (31%)
Yes (n [%])	45 (69%)
Missing (n [%])	0 (0%)
Coronary Heart Disease Prevalence
No (n [%])	56 (86%)
Yes (n [%])	9 (14%)
Missing (n [%])	0 (0%)
Stroke Prevalence
No (n [%])	63 (97%)
Yes (n [%])	2 (3%)
Missing (n [%])	0 (0%)
Definite or Possible Heart Failure Prevalence
No (n [%])	58 (89%)
Yes (n [%])	7 (11%)
Cholesterol Lowering Medication Use
No (n [%])	31 (48%)
Yes (n [%])	34 (52%)
Antihypertensive Medication Use
No (n [%])	13 (20%)
Yes (n [%])	52 (80%)
Statin Use
No (n [%])	32 (49%)
Yes (n [%])	33 (51%)
HDL-Cholesterol (mg/dL)	48.8 (11.6)
LDL-Cholesterol (mg/dL)	103.8 (36.4)
Total Cholesterol (mg/dL)	179.2 (43.7)
Triglycerides (mg/dL)	133.2 (54.0)
Systolic Blood Pressure (mmHg)	126.8 (11.3)
Diastolic Blood Pressure (mmHg)	64.8 (10.3)
Amyloid-β Load on PET (SUVR)	1.3 (0.3)
Total Intracranial Volume (mm³)	1387917.2 (160915.9)
Temporal-Parietal Meta-ROI Volume (mm³)	60444.2 (7068.2)

BMI: body mass index; MCI: mild cognitive impairment; HDL: high-density lipoprotein; LDL: low-density lipoprotein; PET: positron emission tomography; SUVR: standardized uptake value ratio; ROI: region of interest.

HDL cargo proteins

The 96 HDL cargo proteins’ main functions are lipid transport, blood coagulation, and complement activation and regulation (Figure 1A). Figure 1B illustrates clusters of HDL cargo proteins. Many cargo proteins had positive correlations within HDL, allowing for their clustering into groups. However, we also observed negative correlations between HDL protein clusters. These observations are consistent with the knowledge that HDL function is a balance of their protein cargos with opposing properties (e.g., pro- and anti-inflammatory). The ARIC study measured 77 of the 96 HDL cargo proteins in plasma using SomaScan, which allowed us to examine how these 77 plasma proteins correlated with their HDL counterparts. Pearson correlation coefficients suggested that some of these plasma proteins were positively correlated with their HDL counterparts, but some were negatively correlated. Overall, more proteins were positively and negatively correlated (Supplemental Table 2). Figure 1C illustrates clusters of these plasma proteins when measured in plasma. Comparison between Figure 1B and 1C suggests that these proteins had very different clustering patterns when measured in HDL (i.e., HDL cargo proteins) than in plasma. The stronger Pearson correlation coefficients between these proteins when measured in HDL than in plasma suggest physical interactions of these proteins within HDL.

Figure 1.

HDL cargo protein functions and networks. (A) HDL cargo proteins are functionally enriched for lipid transport, platelet degranulation, blood coagulation, and complement activation and regulation. The proteins identified were mapped using the String database (v11.5) and had a confidence score of 0.6 or greater. Cytoscape was used to create the protein interaction networks. (B) Protein clusters of HDL cargo proteins. (C) Protein clusters of HDL cargo proteins measured in plasma. Protein correlation matrix was hierarchically clustered using the average linkage method.

Associations of HDL cargo proteins with brain amyloid deposition

We found two HDL cargo proteins (complement component 4 binding protein beta [C4BPB] and alpha-1-microglobulin/bikunin precursor [AMBP]), measured in fractionated HDL using the label-free DIA untargeted mass spectrometry method, were associated with brain amyloid deposition, after adjusting for covariates (APOE ɛ4 carrier status [yes or no], education, and statin medication) (Figure 2A and Supplemental Table 3). Each unit increase of C4BPB and AMBP in HDL (log₂-transformed) was associated with less brain amyloid of 0.179 SUVR (95% CI [-0.308, −0.050]) and 0.050 SURV (95% CI [-0.099, −0.001]), respectively.

Figure 2.

Association of HDL cargo proteins (A) and plasma proteins (B) with brain amyloid deposition. (A) 96 HDL cargo proteins were assessed by FPLC separation of HDL followed by label-free data-independent acquisition mass spectrometry; (B) 77 plasma proteins were assessed by SomaScan. Linear regression was used to assess the associations between each HDL cargo protein (log₂-transformed) or each plasma protein (log₂-transformed) and brain amyloid deposition, adjusting for covariates (APOE4 status, education, and use of statin medications). The beta coefficient represents the change in brain amyloid deposition (SUVR) for a one-unit change in each HDL cargo protein (log₂-transformed) (A) or each plasma protein (log₂-transformed) (B). p value (unadjusted) represents the statistical significance of the beta coefficient. A p-value less than 0.05 suggests a one-unit change in each protein (log₂-transformed) likely has a non-zero effect on the brain amyloid deposition (SUVR).

Meanwhile, we examined how these same HDL cargo proteins, when measured in plasma using SomaScan, were associated with brain amyloid deposition. One of these cargo proteins, lumican (LUM), was associated with greater brain amyloid deposition of 0.337 SUVR (95% CI [0.073, 0.601]) when measured in plasma (Figure 2B and Supplemental Table 4).

Associations of HDL cargo proteins with temporal-parietal meta-ROI volume

We found some HDL cargo proteins were associated with temporal-parietal meta-ROI volume after adjusting for covariates (age, cognitive status, diabetes, education, prevalence of heart disease, prevalence of heart failure, triglycerides, and total intracranial volume). Each unit increase of 9 proteins in HDL, fibrinogen beta chain (FGB), fibrinogen alpha chain (FGA), inter-alpha-trypsin inhibitor heavy chain 3 (ITIH3), fibulin (FBLN1), complement component C8A, complement component 1 s (C1 s), paraoxonase 1 (PON1), apolipoprotein B (APOB), vitronectin (VTN), were associated with a greater temporal-parietal meta-ROI volume of 1638 mm³ (95% CI [688, 2589]), 1910 mm³ (95% CI [650, 3169]), 1240 mm³ (95% CI [411, 2068]), 1530 mm³ (95% CI [421, 2639]), 1506 mm³ (95% CI [357, 2654]), 1131 mm³ (95% CI [234, 2029]), 1404 mm³ (95% CI [222, 2586], 653 mm³ (95% CI [44, 1262]), and 947 mm³ (95% CI [12, 1881]), respectively (Figure 3A and Supplemental Table 5). Each unit increase of 4 proteins in HDL, plasminogen (PLG), C1 inhibitor (SERPING1), retinol-binding protein 4 (RBP4), complement component 5 (C5), were associated with less temporal-parietal meta-ROI volume of 2351 mm³ (95% CI [-3669, −1034]), 1203 mm³ (95% CI [-2108, −298]), 1085 mm³ (95% CI [-1941, −230]), 1423 mm³ (95% CI [-2738, −108]), respectively (Figure 3A and Supplemental Table 6).

Figure 3.

Association of HDL cargo proteins (A) and plasma proteins (B) with temporal-parietal meta-ROI volume. (A) 96 HDL cargo proteins were assessed by FPLC separation of HDL followed by label-free data-independent acquisition mass spectrometry. (B) 77 plasma proteins were assessed by SomaScan. Linear regression was used to assess the associations between each HDL cargo protein (log₂-transformed) or each plasma protein (log₂-transformed) and temporal-parietal meta-ROI volume, adjusting for covariates (age, cognitive status, diabetes, education, prevalence of heart disease, prevalence of heart failure, triglycerides, and total intracranial volume). The beta coefficient represents the change in temporal-parietal meta-ROI volume (mm³) for a one-unit change in each HDL cargo protein (log₂-transformed) (A) or each plasma protein (log₂-transformed) (B). p value (unadjusted) represents the statistical significance of the beta coefficient. A p-value less than 0.05 suggests a one-unit change in each protein (log₂-transformed) likely has a non-zero effect on the temporal-parietal meta-ROI volume (mm³).

Meanwhile, we examined how these HDL cargo proteins, when measured in plasma using SomaScan, were associated with temporal-parietal meta-ROI volume. Several of these proteins, such as higher haptoglobin (HP), complement factor H (CFH), complement factor H related 1 (CFHR1), histidine-rich glycoprotein (HRG), coagulation factor X (F10), and lower C4BPB, apolipoprotein F (APOF), coagulation factor IX (F9), were associated with greater temporal-parietal meta-ROI volume after adjusting for covariates (Figure 3B and Supplemental Table 6).

Discussion

We examined the association of HDL cargo proteins with brain amyloid deposition and brain volume in regions affected by AD pathology (i.e., temporal-parietal meta-ROI volume). The study revealed interesting findings about HDL cargo proteins and these AD-related outcomes. First, more HDL cargo proteins were associated with temporal-parietal meta-ROI volume than brain amyloid deposition. The few associations between HDL cargo proteins and brain amyloid deposition were surprising because of the hypothesis that APOA-1-enriched HDL may boost amyloid-beta efflux out of the brain and into the blood.² It is plausible that HDL-mediated amyloid-beta efflux removes amyloid-beta peptide built up in the cerebrovascular walls, a pathological condition known as cerebral amyloid angiopathy, and does not affect amyloid-peptide built-up in the brain parenchyma, which is the amyloid deposition measured by amyloid PET.^33,34

Second, many HDL cargo proteins associated with temporoparietal meta-ROI volume are related to coagulation and complement pathways. These pathways are known to be crucial to the body's defense systems but were recently discovered to be critical in neuronal development and synaptic plasticity.³⁵ Two HDL cargo proteins mostly associated with temporal-parietal meta-ROI are FGA and PLG, which are related to coagulation. FGB is a subunit of fibrinogen that forms fibrin clots when cleaved by thrombin, and PLG is a precursor of plasmin that lyses fibrin clots. Although how HDL interacts with these proteins and why these interactions may affect brain volume remains unclear, we do find some evidence for a detrimental role of plasma fibrinogen on brain volumes and a protective role of HDL in preventing fibrinogen from causing harm. Plasma fibrinogen, as part of a midlife inflammation composite score, was associated with smaller brain volume later in life.³⁶ Mechanistically, upon cerebrovascular damage, plasma fibrinogen forms proinflammatory fibrin deposits in the CNS, which induces microglial activation, eliminating synaptic spines and promoting cognitive deficits in an AD model.³⁷ Plasma HDL is known to interact with plasma fibrinogen and influence its concentration. For example, fibrinogen αC-region may contain Apo-binding sites,³⁸ and apo A-I is found in fibrin clots.³⁹ An increase in HDL-C may decrease plasma fibrinogen over time, as suggested by a longitudinal epidemiological study from ARIC that a 15 mg/dL increase in HDL-C was associated with an 11.3 mg/dL decrease in the adjusted mean of plasma fibrinogen (adjusted for baseline fibrinogen).⁴⁰ The impact of plasma HDL on plasma fibrinogen concentration is relatively small because plasma fibrinogen concentrations are generally 200–400 mg/dL. That being said, increased plasma levels of HDL-C and apoA-1 can also modify plasma fibrin clot functions by improving plasma fibrin clot permeability and susceptibility to lysis.⁴¹ Taken together, we postulate that the association of higher FGB in HDL with greater temporal-parietal meta-ROI could be explained by HDL's abilities to reduce the harmful effects of fibrin clots⁴¹ in a way that attenuates cerebral hypoperfusion and brain atrophy.⁴²

We also observed lower PLG in HDL was associated with greater volume in temporal-parietal meta-ROI. Because high fibrinogen and low plasminogen are prothrombotic, higher fibrinogen and lower plasminogen in HDL in association with greater temporal-parietal meta-ROI are consistent with anti-thrombotic properties of HDL, as HDL contains many proteases that can inhibit activation of plasminogen.⁴³ Nevertheless, it is worth noting that HDL also contains proteases that can activate plasminogen,⁴³ and HDL's pro- and anti-thrombotic properties are based on its protein cargos with opposing functions. We also found fibrinogen subunit alpha (FGA) associated with temporoparietal meta-ROI. Fibulin 1 (FBLIN1), another HDL cargo protein we identified to be associated with temporoparietal meta-ROI, was also reported to bind fibrinogen and incorporate it into fibrin clots.⁴⁴ Overall, the study suggests that HDL may play an important role in regulating coagulation, and such interactions might be important for maintaining BBB integrity and brain volume.

In addition, HDL cargo proteins associated with temporal-parietal meta-ROI belong to complement cascade (C8A, C1S, C5) and protease inhibitors (SERPING1 and ITIH3). There is a limited understanding of how HDL interacts with the complement cascade and why these interactions are associated with brain volume. The complement system is a proteolytic cascade where serine proteases activate and amplify complement components in an ordered manner. It is known that the complement system regulates neuroinflammation and BBB integrity. For example, a recent study suggests that the gamma subunit of complement component 8 is a neuroinflammation inhibitor,⁴⁵ and astrocytic complement C8 gamma protects BBB integrity.⁴⁶ It is plausible that HDL modulates the complement cascade through the activity of protease inhibitors, which in turn regulate BBB and neuroinflammation and influence brain volume.

Other HDL cargo proteins associated with temporal-parietal meta-ROI volume include APOB and PON1. The HDL proteome watch suggested that APOB is a likely HDL protein.³² Two forms of APOB are typically known in human plasma: the full-length apoB100 consisting of 4536 amino acids and apoB48 representing the protein's N-terminal 2152 amino acids (48%).⁴⁷ However, APOBs other than B48 and B100 can also exist in HDL.⁴⁸ It is unclear how APOB in HDL might be related to brain volume. PON1 is an antioxidative enzyme that hydrolyzes oxidized LDL cholesterol. The association of PON1 in HDL with a greater temporal-parietal meta-ROI volume is consistent with its known antioxidative effects.

We also examined how HDL cargo proteins, when measured in plasma, were associated with temporal-parietal meta-ROI volume. We found a different set of proteins (CFH, CFHR1, HP, HRG, F10, C4BPB, APOF, F9) belonging to complement and coagulation pathways. These observations suggest that proteins in complement and coagulation are associated with brain volume, both through HDL-dependent and HDL-independent pathways.

The strengths of this study include the physical isolation of HDL in a well-characterized study cohort and the comparison of the associations between HDL cargo proteins and the same proteins when measured in plasma (i.e., SomaScan) to two AD-related outcomes. Because ARIC participants were well-characterized regarding comorbidities and medications, we evaluated them as covariates and adjusted for them when appropriate. This study's limitations are several. First, it included a cross-sectional design, which does not allow for assessing a causal relationship between these HDL cargo proteins and brain volume. Second, this study is exploratory and limited by sample size. We reported effect sizes (i.e., beta coefficients) and p-values without adjustment for multiple comparisons, although we reported Benjamini-Hochberg (BH)-adjusted p-values in Supplemental Tables 3–6. The study results warrant replication in a larger cohort to assess the generalizability. Third, we used 10μg of HDL proteins for the mass spectrometry analysis. Because the amount of HDL proteins in a given volume of plasma varies from person to person, using the same amount of protein (i.e., 10 μg) does not consider inter-individual total HDL protein differences. Future studies should consider this variability. Fourth, it is plausible that non-HDL-associated proteins coeluting with the HDL fraction are identified as HDL cargo proteins. However, to ensure the HDL cargo proteins we identified in our study are consistent with others, our analysis only included “likely HDL proteins” based on the HDL Proteome Watch.³² These likely proteins were identified only when three independent proteomics studies reported their presence in isolated HDL. Fifth, we acknowledge the methodological difference between SomaScan and mass spectrometry. We intended to compare how HDL cargo proteins when measured in HDL and when measured in plasma. However, because different techniques measured HDL proteins and plasma proteins, we cannot exclude the possibility that the differences are due to technical rather than biological.

Conclusions

We identified HDL cargo proteins with roles in coagulation and complement associated with temporal-parietal meta-ROI volume. These results suggest that HDL may help maintain brain volume in older adults by regulating coagulation and complement proteins.

Supplemental Material

sj-docx-2-alz-10.1177_13872877241305806 - Supplemental material for The association of high-density lipoprotein cargo proteins with brain volume in older adults in the Atherosclerosis Risk in Communities (ARIC)

Supplemental material, sj-docx-2-alz-10.1177_13872877241305806 for The association of high-density lipoprotein cargo proteins with brain volume in older adults in the Atherosclerosis Risk in Communities (ARIC) by Danni Li, Binchong An, Lu Men, Matthew Glittenberg, Pamela L Lutsey, Michelle M Mielke, Fang Yu, Ron C Hoogeveen, Rebecca Gottesman, Lin Zhang, Michelle Meyer, Kevin Sullivan, Nicole Zantek, Alvaro Alonso and Keenan A Walker in Journal of Alzheimer's Disease

Supplemental Material

sj-xlsx-3-alz-10.1177_13872877241305806 - Supplemental material for The association of high-density lipoprotein cargo proteins with brain volume in older adults in the Atherosclerosis Risk in Communities (ARIC)

Supplemental material, sj-xlsx-3-alz-10.1177_13872877241305806 for The association of high-density lipoprotein cargo proteins with brain volume in older adults in the Atherosclerosis Risk in Communities (ARIC) by Danni Li, Binchong An, Lu Men, Matthew Glittenberg, Pamela L Lutsey, Michelle M Mielke, Fang Yu, Ron C Hoogeveen, Rebecca Gottesman, Lin Zhang, Michelle Meyer, Kevin Sullivan, Nicole Zantek, Alvaro Alonso and Keenan A Walker in Journal of Alzheimer's Disease

Footnotes

Acknowledgments

The Atherosclerosis Risk in Communities study has been funded in whole or in part with Federal funds from the National Heart, Lung, and Blood Institute, National Institutes of Health, Department of Health and Human Services, under Contract nos. (75N92022D00001, 75N92022D00002, 75N92022D00003, 75N92022D00004, 75N92022D00005). The ARIC Neurocognitive Study is supported by U01HL096812, U01HL096814, U01HL096899, U01HL096902, and U01HL096917 from the NIH (NHLBI, NINDS, NIA and NIDCD). SomaLogic Inc. conducted the SomaScan assays in exchange for use of ARIC data. This work was supported in part by NIH/NHLBI grant R01 HL134320. The ARIC-PET (Positron Emission Tomography) study is funded by the NIA (R01AG040282 to Dr Gottesman). The authors thank the staff and participants of the ARIC study for their important contributions.

ORCID iDs

Danni Li

Binchong An

Lu Men

Matthew Glittenberg

Pamela L Lutsey

Michelle M Mielke

Fang Yu

Ron C Hoogeveen

Rebecca Gottesman

Lin Zhang

Kevin Sullivan

Nicole Zantek

Alvaro Alonso

Keenan A Walker

Author contributions

Danni Li (Conceptualization; Funding acquisition; Investigation; Methodology; Project administration; Resources; Supervision; Writing – original draft; Writing – review & editing) Binchong An (Formal analysis) Lu Men (Formal analysis) Matthew Glittenberg (Formal analysis) Pamela L. Lutsey (Resources; Writing – review & editing) Michelle M. Mielke (Writing – review & editing) Fang Yu (Writing – review & editing) Ron C. Hoogeveen (Resources; Writing – review & editing) Rebecca Gottesman (Resources) Lin Zhang (Writing – review & editing) Michelle Meyer (Resources) Kevin Sullivan (Resources; Writing – review & editing) Nicole Zantek (Result interpretation); Alvaro Alonso (Resources; Supervision; Writing – review & editing); Keenan A. Walker (Supervision; Writing – review & editing)

Funding

This work was partly supported by the NIH/NIA grants R21AG061372 and 1RF1AG079100-01A1 to DL. KW is funded by the National Institute on Aging (NIA) Intramural Research Program.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability

The data supporting this study's findings are available on request from the corresponding author. However, due to privacy or ethical restrictions, the data are not publicly available.

Supplemental material

Supplemental material for this article is available online.

References

Stukas

Robert

Wellington

. High-density lipoproteins and cerebrovascular integrity in Alzheimer’s disease. Cell Metab 2014; 19: 574–591.

Chernick

Zhong

. The role of HDL and HDL mimetic peptides as potential therapeutics for Alzheimer's disease. Biomolecules 2020; 10: 1276.

Kjeldsen

Thomassen

Juul Rasmussen

, et al. Plasma high-density lipoprotein cholesterol and risk of dementia: observational and genetic studies. Cardiovasc Res 2022; 118: 1330–1343.

Luo

Thomassen

Bellenguez

, et al. Genetic associations between modifiable risk factors and Alzheimer disease. JAMA Netw Open 2023; 6: e2313734–e2313734.

Hussain

Robb

Tonkin

, et al. Association of plasma high-density lipoprotein cholesterol level with risk of incident dementia: a cohort study of healthy older adults. Lancet Reg Health West Pac 2023; 43: 100963.

Pedrini

Hone

Gupta

, et al. Plasma high density lipoprotein small subclass is reduced in Alzheimer’s disease patients and correlates with cognitive performance. J Alzheimers Dis 2020; 77: 733–744.

Martinez

Weissberger

Kuklenyik

, et al. The small HDL particle hypothesis of Alzheimer's disease. Alzheimers Dement 2023; 19: 391–404.

Song

Poljak

Crawford

, et al. Plasma apolipoprotein levels are associated with cognitive status and decline in a community cohort of older individuals. PLoS One 2012; 7: e34078.

Thambisetty

Kinsey

, et al. Plasma clusterin concentration is associated with longitudinal brain atrophy in mild cognitive impairment. Neuroimage 2012; 59: 212–217.

10.

Gupta

Doecke

Hone

, et al. Plasma apolipoprotein J as a potential biomarker for Alzheimer's disease: australian Imaging, Biomarkers and Lifestyle study of aging. Alzheimers Dement 2016; 3: 18–26.

11.

Gupta

Laws

Villemagne

, et al. Plasma apolipoprotein E and Alzheimer disease risk: the AIBL study of aging. Neurology 2011; 76: 1091–1098.

12.

Gupta

Wilson

Burnham

, et al. Follow-up plasma apolipoprotein E levels in the Australian Imaging, Biomarkers and Lifestyle Flagship Study of Ageing (AIBL) cohort. Alzheimers Res Ther 2015; 7: 16.

13.

Rasmussen

Tybjaerg-Hansen

Nordestgaard

, et al. Plasma levels of apolipoprotein E and risk of dementia in the general population. Ann Neurol 2015; 77: 301–311.

14.

Gordon

Deng

, et al. Proteomic characterization of human plasma high density lipoprotein fractionated by gel filtration chromatography. J Proteome Res 2010; 9: 5239–5249.

15.

Dickerson

Stoub

Shah

, et al. Alzheimer-signature MRI biomarker predicts AD dementia in cognitively normal adults. Neurology 2011; 76: 1395–1402.

16.

Wright

Folsom

Coresh

, et al. The ARIC (atherosclerosis risk in communities) study: JACC focus seminar 3/8. J Am Coll Cardiol 2021; 77: 2939–2959.

17.

Knopman

Gottesman

Sharrett

, et al. Mild cognitive impairment and dementia prevalence: the Atherosclerosis Risk in Communities Neurocognitive Study (ARIC-NCS). Alzheimers Dement (Amst) 2016; 2: 1–11.

18.

Seidl

Flaten

. Characterization of human plasma lipoproteins using anion exchange fast protein liquid chromatography and targeted mass spectrometry assay. Proteomics 2021; 21: 2000224.

19.

Hirowatari

Yoshida

Kurosawa

, et al. Measurement of cholesterol of major serum lipoprotein classes by anion-exchange HPLC with perchlorate ion-containing eluent. J Lipid Res 2003; 44: 1404–1412.

20.

Zhang

, et al. Anion exchange HPLC isolation of high-density lipoprotein (HDL) and on-line estimation of proinflammatory HDL. PLoS One 2014; 9: e91089.

21.

Van Buuren

Groothuis-Oudshoorn

. Mice: multivariate imputation by chained equations in R. J Stat Softw 2011; 45: 1–67.

22.

Gottesman

Schneider

Zhou

, et al. Association between midlife vascular risk factors and estimated brain amyloid deposition. JAMA 2017; 317: 1443–1450.

23.

Knopman

Griswold

Lirette

, et al. Vascular imaging abnormalities and cognition: mediation by cortical volume in nondemented individuals: atherosclerosis risk in communities-neurocognitive study. Stroke 2015; 46: 433–440.

24.

Gold

Ayers

Bertino

, et al. Aptamer-based multiplexed proteomic technology for biomarker discovery. PLoS One 2010; 5: e15004.

25.

Walker

Chen

Zhang

, et al. Large-scale plasma proteomic analysis identifies proteins and pathways associated with dementia risk. Nat Aging 2021; 1: 473–489.

26.

Candia

Cheung

Kotliarov

, et al. Assessment of variability in the SOMAscan assay. Sci Rep 2017; 7: 14248.

27.

Loehr

Rosamond

Chang

, et al. Heart failure incidence and survival (from the Atherosclerosis Risk in Communities study). Am J Cardiol 2008; 101: 1016–1022.

28.

Rosamond

Folsom

Chambless

, et al. Stroke incidence and survival among middle-aged adults: 9-year follow-up of the Atherosclerosis Risk in Communities (ARIC) cohort. Stroke 1999; 30: 736–743.

29.

Folsom

Aleksic

Catellier

, et al. C-reactive protein and incident coronary heart disease in the Atherosclerosis Risk In Communities (ARIC) study. Am Heart J 2002; 144: 233–238.

30.

Knopman

Mosley

Catellier

, et al. Fourteen-year longitudinal study of vascular risk factors, APOE genotype, and cognition: the ARIC MRI Study. Alzheimers Dement 2009; 5: 207–214.

31.

Muntel

Kirkpatrick

Bruderer

, et al. Comparison of protein quantification in a complex background by DIA and TMT workflows with fixed instrument time. J Proteome Res 2019; 18: 1340–1351.

32.

Davidson

Shah

Sexmith

, et al. The HDL Proteome Watch: compilation of studies leads to new insights on HDL function. Biochim Biophys Acta Mol Cell Biol Lipids 2022; 1867: 159072.

33.

Lewis

Cao

, et al. Overexpression of human apolipoprotein A-I preserves cognitive function and attenuates neuroinflammation and cerebral amyloid angiopathy in a mouse model of Alzheimer disease. J Biol Chem 2010; 285: 36958–36968.

34.

Robert

Button

Martin

, et al. Cerebrovascular amyloid Angiopathy in bioengineered vessels is reduced by high-density lipoprotein particles enriched in Apolipoprotein E. Mol Neurodegener 2020; 15: 23.

35.

Berkowitz

Chapman

Dori

, et al. Complement and coagulation system crosstalk in synaptic and neural conduction in the central and peripheral nervous systems. Biomedicines 2021; 9: 1950.

36.

Walker

Hoogeveen

Folsom

, et al. Midlife systemic inflammatory markers are associated with late-life brain volume: the ARIC study. Neurology 2017; 89: 2262–2270.

37.

Merlini

Rafalski

Coronado

PER

, et al. Fibrinogen induces microglia-mediated spine elimination and cognitive impairment in an Alzheimer’s disease model. Neuron 2019; 101: 1099–1108.e6.

38.

Tsurupa

Ho-Tin-Noé

Anglés-Cano

, et al. Identification and characterization of novel lysine-independent apolipoprotein (a)-binding sites in fibrin (ogen) αC-domains. J Biol Chem 2003; 278: 37154–37159.

39.

Talens

Leebeek

Demmers

, et al. Identification of fibrin clot-bound plasma proteins. PLoS One 2012; 7: e41966.

40.

Folsom

Rasmussen

, et al. Determinants of population changes in fibrinogen and factor VII over 6 years: the Atherosclerosis Risk in Communities (ARIC) Study. Arterioscler Thromb Vasc Biol 2000; 20: 601–606.

41.

Zabczyk

Hondo

Krzek

, et al. High-density cholesterol and apolipoprotein AI as modifiers of plasma fibrin clot properties in apparently healthy individuals. Blood Coagul Fibrinolysis 2013; 24: 50–54.

42.

Strickland

. Blood will out: vascular contributions to Alzheimer’s disease. J Clin Invest 2018; 128: 556–563.

43.

Gordon

Remaley

. High density lipoproteins are modulators of protease activity: implications in inflammation, complement activation, and atherothrombosis. Atherosclerosis 2017; 259: 104–113.

44.

Tran

Tanaka

Litvinovich

, et al. The interaction of fibulin-1 with fibrinogen: a potential role in hemostasis and thrombosis. J Biol Chem 1995; 270: 19458–19464.

45.

Kim

J-H

Afridi

Han

, et al. Gamma subunit of complement component 8 is a neuroinflammation inhibitor. Brain 2021; 144: 528–552.

46.

Kim

J-H

Han

Suk

. Protective effects of complement component 8 gamma against blood-brain barrier breakdown. Front Physiol 2021; 12: 671250.

47.

McLeod

Wang

, et al. Apolipoprotein B sequence requirements for hepatic very low density lipoprotein assembly: evidence that hydrophobic sequences within apolipoprotein B48 mediate lipid recruitment. J Biol Chem 1996; 271: 18445–18455.

48.

Linton

Farese Jr

Young

. Familial hypobetalipoproteinemia. J Lipid Res 1993; 34: 521–541.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.43 MB

0.07 MB