Changes in the plasma proteome at asymptomatic and symptomatic stages of autosomal dominant Alzheimer’s disease

The autosomal dominant form of Alzheimer’s disease (ADAD) is far less prevalent than late onset Alzheimer’s disease (LOAD), but enables well-informed prospective studies, since symptom onset is near certain and age of onset is predictable. Our aim was to discover plasma proteins associated with early AD pathology by investigating plasma protein changes at the asymptomatic and symptomatic stages of ADAD. Eighty-one proteins were compared across asymptomatic mutation carriers (aMC, n = 15), symptomatic mutation carriers (sMC, n = 8) and related noncarriers (NC, n = 12). Proteins were also tested for associations with cognitive measures, brain amyloid deposition and glucose metabolism. Fewer changes were observed at the asymptomatic than symptomatic stage with seven and 16 proteins altered significantly in aMC and sMC, respectively. This included complement components C3, C5, C6, apolipoproteins A-I, A-IV, C-I and M, histidine-rich glycoprotein, heparin cofactor II and attractin, which are involved in inflammation, lipid metabolism and vascular health. Proteins involved in lipid metabolism differed only at the symptomatic stage, whereas changes in inflammation and vascular health were evident at asymptomatic and symptomatic stages. Due to increasing evidence supporting the usefulness of ADAD as a model for LOAD, these proteins warrant further investigation into their potential association with early stages of LOAD.


Supplementary Table S8. Plasma levels of heparin cofactor II (HCII) for non-carriers (NC), asymptomatic (aMC) and symptomatic mutation carriers (sMC).
HCII was quantified using ELISA in the low abundance protein fractions derived from plasma and used in iTRAQ experiments. Protein levels are expressed as µg protein of interest per mg total protein in the sample. HCII NC (n = 12), µg per mg protein (SD) 16

Plasma immunodepletion and sample preparation
The six most abundant plasma proteins (albumin, transferrin, immunoglobulins G and A, haptoglobin and antitrypsin) were immunodepleted from plasma samples (20 µl) using the Agilent (Santa Clara, USA) Multiple Affinity Removal System Hu6 column and buffer kit on a HP 1090 HPLC system (Agilent, Santa Clara, USA) according to manufacturer's instructions. We previously verified that this method only removes the six targeted proteins 1 . The depleted fractions were buffer exchanged and concentrated into 20 mM NaHCO 3 using Amicon 3 kDa centrifugal devices (Millipore, Billerica, USA). Total protein was quantified by absorbance measurements (280 nm) with a ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, USA), using the relationship; A 280 of 1 = 1 mg mL -1 . Low abundance protein fractions were analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS PAGE) (NuPAGE 4-12% gradient Bis-Tris gels, Life Technologies, Carlsbad, USA) to verify consistent depletion of the six most abundant proteins across all samples ( Supplementary Fig. S3 and S4). Samples were stored at -80 °C until further use.

iTRAQ-labelling and purification of labelled peptides
Labelling of tryptic peptides with iTRAQ 8-plex reagents (Sciex, Framingham, USA) was carried out according to manufacturer's instructions with slight modifications as outlined in Muenchhoff, et al. 1 . Briefly, 50 µg of the low abundance proteins from each plasma sample were reduced with tris (2carboxyethyl) phosphine, alkylated with iodoacetamide and digested with trypsin overnight at 37°C. Tryptic peptides were combined with iTRAQ reagents and the pH-adjusted labelling reaction was allowed to proceed for 2 hrs at room temperature. iTRAQ-labelled peptides were combined and passed through a cation exchange cartridge to remove excess reagents. The eluent was evaporated to dryness, resuspended in 500 µl 0.2% heptafluorobutyric acid (HFBA) and purified using a C18 macrotrap (Michrom Bioresources, Auburn, USA). To ensure maximal recovery of labelled peptides, the C18 flow through was passed through an Oasis cartridge to capture any peptides not bound on the macrotrap. The eluents from both cartridges were combined, evaporated to dryness and resuspended in 100 µl 0.05% HFBA, 1% formic acid for liquid chromatography tandem mass spectrometry (LC-MSMS) analysis.

LC-MSMS
iTRAQ-labelled peptides were analysed by LC-MSMS. LC was carried out on a LC Packings capillary HPLC system, comprised of a Dionex UltiMate 3000 RSLCnano pump system, Switchos valve unit and Famos autosampler (Thermo Scientific Dionex, Waltham, USA). The resuspended iTRAQ labelled peptides were injected onto a C18 precolumn cartridge (Acclaim PepMap 100, 5 µm 100 Å, Thermo Scientific Dionex, Waltham, USA), which was washed for 10 min prior to switching inline to a capillary column (10 cm) containing C18 reverse phase packing material (Reprosil-Pur, 1.9 µ, 200 Å, Dr. Maisch GmbH, Ammerbuch-Entringen, Germany). Peptides were eluted using a 240 min gradient of buffer A (H 2 O:CH 3 CN of 98:2 containing 0.1% formic acid) to buffer B (H 2 O:CH 3 CN of 20:80 containing 0.1% formic acid) at 200 nL/min. High voltage (2300 V) was applied through a low volume tee (Upchurch Scientific, Oak Harbor, USA) at the column inlet and the outlet positioned ~1 cm from the orifice of a TripleTOF 5600 + hybrid tandem mass spectrometer (ABSciex, Foster City, USA). Positive ions were generated by electrospray ionization and the TripleTOF 5600 + system operated in information-dependent acquisition mode. A time of flight MS survey scan was acquired (m/z 375-1600, 0.4 s) and up to ten multiply charged ions (m/z 375-1250, counts > 200, charge state ≥ 2 + and ≤ 5 + ) sequentially selected by Q1 for MSMS analysis. Nitrogen was used as collision gas and an optimum collision energy automatically chosen (based on charge state and mass). Tandem mass spectra were accumulated for 0.3 s. LC-MSMS was performed twice for analytical replicates.

Proteins in inflammation
Four components of the complement system were found differentially abundant in NC, aMC and sMC, namely, C3, C5, C6 and C4b-binding protein α chain. There is evidence to indicate that the complement system might play a crucial role in AD pathology. Aβ interacts with components of the classical and alternate pathway, activating both in the AD brain 2,3 . Activation of the complement system in AD affords some protection in the form of clearance of Aβ, but also causes harm through chronic inflammation leading to tissue damage and lysis of neurons 3 . Many studies have previously suggested complement components as biomarkers for AD (e. g. 1,4-7 ).
ACT, one of three serine protease inhibitors (SERPIN) found differentially abundant, is an acute phase protein, regulating protease activity of neutrophil cathepsin G, mast cell chymase and others during inflammation, thereby preventing tissue damage 8 . It might also be involved in atherosclerotic processes and stabilisation of the aorta as indicated by its differential vascular expression in atherosclerotic lesions and abdominal aortic aneurysms 9 . Its expression is elevated in the AD brain, it co-localises with amyloid plaques and might induce tau hyperphosphorylation 10,11 . In vivo studies in transgenic mouse models report accelerated amyloid plaque formation upon expression of human ACT 10,[12][13][14][15][16] , which could potentially be due to ACT inhibiting a serine protease involved in Aβ degradation 17 . Although controversial, variation in the SERPINA3 promoter and ACT protein sequence have also been associated with increased risk of LOAD [18][19][20] . A number of reports suggested ACT levels in plasma/serum as a biomarker for LOAD (see 21 for a review).
AHSG also known as fetuin-A is a multifunctional negative acute phase protein that can bind cations, such as calcium. It is involved in regulation of mineralization and ostoegenesis, insulin signaling as well as inflammation 22 . AHSG serum/plasma levels are known to predict incident type 2 diabetes and vascular disease risk due to its inhibitory action on vascular calcification 23 . In a rodent model, peripherally administered bovine AHSG protected against early cerebral ischemic injury, likely due to its anti-inflammatory properties 24 . Due to its links to vascular disease and neuroinflammation, both components of AD pathology, AHSG has been proposed as a biomarker for LOAD in plasma and CSF, and was shown to associate with severity of cognitive decline 23,[25][26][27] . Individuals homozygous for the AHSG 1 allele in an Italian population were found to be at nearly four times greater risk of developing LOAD 28 .
Protein α-1-microglobulin/bikunin precursor is the precursor for the structurally and functionally unrelated proteins α-1-microglobulin and bikunin. α-1-microglobulin is a member of the lipocalin superfamily, a group of proteins that carry lipophilic ligands. It may protect cells from oxidative stress by transporting heme groups and free radicals released from hemoglobin from cytosols and extravascular fluids to the kidneys 29 . It also negatively regulates the immune response of lymphocytes, T-cells and granulocytes [30][31][32] . α-1-microglobulin levels were found to differ in the plasma from AD and control patients 33 and associated with brain atrophy in AD patients 34 . Bikunin, also known as inter-α-trypsin inhibitor light chain, is a Kunitz-type protease inhibitor possibly involved in endothelial cell growth and extracellular matrix stabilisation 35,36 . Proteins in the inter-αtrypsin inhibitor family consist of the common light chain protein bikunin linked to one or two of various heavy chains (H1-4) by a chondroitin sulphate chain. Interestingly, significant differences in ITIH2 were also observed here. The protease inhibitory activity of bikunin might prevent inflammation-related proteolytic activity. The inter-α-trypsin inhibitor heavy chains can be transferred to hyaluronan (a major component of the pericellular matrix), resulting in formation of a heavy chainhyaluronan complex and the release of bikunin, which is then excreted into urine. Heavy chains also interact with components of the complement system to lower levels of the powerful mediator C5a 37 . Changes in abundance of plasma inter-α-trypsin inhibitor heavy chain 2 were previously reported in MCI 1 and LOAD 21,38 .
ATRN is the only protein found differentially abundant in the early asymptomatic stage of ADAD but not in the later stage. ATRN is expressed as three isoforms, with isoform 1 having a C-terminal extension that anchors the protein in the membrane, whereas isoforms 2 and 3 are secreted. No peptides specific to any of the three isoforms were detected; hence, no conclusions can be drawn on presence or quantity of the individual isoforms. Expression of the secreted isoforms is down-regulated in the human brain and these isoforms have been shown to disrupt neurite formation in vitro 39 . By contrast, the membrane-bound isoform is expressed in the CNS, where it is critical for myelination 39,40 . Rats with loss of function mutations in glycosylated transmembrane ATRN have age-dependent spongiform degeneration, hypomyelination and abnormal ROS metabolism in the brain [41][42][43] . ATRN is also involved in the regulation of skin pigmentation, energy control and immunity 44,45 . In the immune system, ATRN regulatory activity allows immune cells to interact to form regulatory clusters. This regulatory activity of ATRN might be affected by the balance of membrane-bound and soluble isoforms 46 . The mechanism of ATRN function is not well understood with suggestions of DP4 protease activity for ATRN being disputed 47 .

Proteins in hemostasis and vascular health
Two proteins with functions in hemostasis and vascular health were found differentially abundant in the NC, aMC and sMC groups. Both proteins are also able to modulate the immune response, reflecting the close connection of these systems.
HRG modulates a variety of biological processes, including blood coagulation, fibrinolysis, angiogenesis, complement activation and aggregation of immune complexes 48,49 . It exerts its influence via binding of various ligands, e.g. heparin, heparin sulphate, plasminogen and plasmin, fibrinogen, complement component C1q, IgG, Fc γ receptor and divalent cations 50 . Similar to AHSG mentioned above, HRG also belongs to the cystatin type 3 family of proteins, and is able to inhibit the formation of spontaneous apatite calcifications formed from calcium and phosphate ions (both divalent cations) in vitro. Hence, it possibly prevents ectopic calcification 50 . Levels of HRG in blood were reported to differ between MCI 5 and AD 51 subjects and healthy controls.
HCII is a SERPIN that in the presence of glucosaminoglycans (e.g. heparin or dermatan sulphate) inhibits thrombin. HCII is particularly relevant in the intima and media of the vascular wall, which is rich in dermatan sulfate 52 . As such, HCII protects against thrombin-induced vascular remodelling and consequently atherosclerosis 53 . It may also promote angiogenesis via an AMP-activated protein kinase-endothelial nitric oxide synthase pathway 54 . Hence, HCII has been suggested as a therapeutic target and potential biomarker for arterial disease [53][54][55] .

Proteins in lipid metabolism
Four apolipoproteins, ApoA1, ApoA4, ApoC1 and ApoM, differed in abundance in NC, aMC and sMC groups. Apolipoproteins are constituents of lipoprotein particles, such as chylomicrons, VLDL, LDL and HDL, which transport lipids between tissues for fuel and cholesterol metabolism. The apolipoproteins serve as carrier, receptor-binding and regulatory proteins in these particles. As such, they are crucial components in lipid metabolism with implications for cardiovascular disease, obesity and diabetes mellitus (for a review see 56 ). Recently, the apolipoproteins have also emerged as a protein family of particular interest in AD, since alterations in lipid metabolism have been associated with AD pathology 57,58 and the APOE ε4 allele is the most significant genetic risk factor for LOAD. Furthermore, plasma levels of clusterin (also known as apolipoprotein J) are emerging as a meaningful biomarker for MCI 59 and LOAD 60 , and carriers of CLU risk alleles show faster rates of cognitive decline 61,62 .