Quantification of N-terminal amyloid-β isoforms reveals isomers are the most abundant form of the amyloid-β peptide in sporadic Alzheimer’s disease

Abstract Plaques that characterize Alzheimer’s disease accumulate over 20 years as a result of decreased clearance of amyloid-β peptides. Such long-lived peptides are subjected to multiple post-translational modifications, in particular isomerization. Using liquid chromatography ion mobility separations mass spectrometry, we characterized the most common isomerized amyloid-β peptides present in the temporal cortex of sporadic Alzheimer’s disease brains. Quantitative assessment of amyloid-β N-terminus revealed that > 80% of aspartates (Asp-1 and Asp-7) in the N-terminus was isomerized, making isomerization the most dominant post-translational modification of amyloid-β in Alzheimer’s disease brain. Total amyloid-β1–15 was ∼85% isomerized at Asp-1 and/or Asp-7 residues, with only 15% unmodified amyloid-β1–15 left in Alzheimer’s disease. While amyloid-β4–15 the next most abundant N-terminus found in Alzheimer’s disease brain, was only ∼50% isomerized at Asp-7 in Alzheimer’s disease. Further investigations into different biochemically defined amyloid-β-pools indicated a distinct pattern of accumulation of extensively isomerized amyloid-β in the insoluble fibrillar plaque and membrane-associated pools, while the extent of isomerization was lower in peripheral membrane/vesicular and soluble pools. This pattern correlated with the accumulation of aggregation-prone amyloid-β42 in Alzheimer’s disease brains. Isomerization significantly alters the structure of the amyloid-β peptide, which not only has implications for its degradation, but also for oligomer assembly, and the binding of therapeutic antibodies that directly target the N-terminus, where these modifications are located.


Introduction
Neuropathology and amyloid-b (Ab) positron emission tomography (PET) studies indicate that the accumulation of Ab in sporadic Alzheimer's disease brain begins more than 20 years before the onset of clinical symptoms. 1,2 Evidence supports that this accumulation is a result of decreased clearance and not a change in the production of Ab in sporadic Alzheimer's disease. 3,4 The subtle 2-5% decrease in its clearance results in total accumulation of $6.5 mg Ab in the brain over the 20 year time span 2,5 compared to 1.7 mg in age-matched control tissue. However, several questions regarding the Ab-amyloid hypothesis 6 remain unanswered, including what leads to the decrease in clearance and what triggers the aggregation of Ab into extracellular plaques 7 along with intracellular tau-reactive neurofibrillary tangles. 8 The impairment in the clearance increases the half-life of the Ab polypeptide and the process of amyloidosis in Alzheimer's disease entombs the peptide for decades, making it a long-lived peptide. The prolonged time frame of amyloidosis is a common feature across multiple neurodegenerative diseases, 9,10 predisposing the polypeptide chains to undergo multiple spontaneous non-enzymatic post-translational modifications (PTMs), which can render them resistant to normal cellular proteolysis mechanisms. 11,12 The earliest Edman sequencing and more recent mass spectrometry-based analyses have shown that there is a diverse population of N-terminally truncated species of Ab 42 (e.g. Ab 1-42 , Ab 2-42 , Ab 4-42 ). 7,13 Moreover, multiple PTMs of Ab have been described and include nitration, 14 pyroglutamate formation, 15,16 phosphorylation, 17 methionine oxidation, 18 dityrosine cross-linking 19 and structural changes of the polypeptide backbone. Structural changes, in particular, occur on the amino acid level via non-enzymatic, spontaneous processes and facilitated by the amino acids with asymmetric central carbon atom. The most common structural protein modification associated with aging is stereoisomerization of Asp/Asn (aspartate/asparagine) residues and have been particularly useful for protein dating. 20,21 Deamidation of L-Asn residue to L-Asp as well as racemization/isomerization to D-Asp and D/L-iso-Asp via succinimide intermediate 22 (Fig. 1A) potentially should provide the information of the age of the Ab plaques. 23 In Alzheimer's disease brain, the striking feature of the fibrillar Ab is its sequential N-terminal truncation along with Asp and Ser (serine) epimerization. 13,15,24,25 Qualitative estimates from the plaque-derived Ab indicate almost 25% of Asp-1 and 75% Asp-7 are isomerized in Alzheimer's disease brains. 9,[26][27][28][29] The antibodies currently in clinical trials target multiple forms (soluble oligomeric and insoluble fibrils) of Ab due to their potential roles in the pathogenesis and disease progression. 30,31 Other than the mid-domain and C-terminus Ab, the other most common target epitope of these antibodies is the PTM-prone N-terminus of Ab. [32][33][34] Indepth understanding of the target engagement warrants detailed analysis of the PTMs (especially isomerization) associated with these epitopes. However, comprehensive characterization of these isomers/epimers along with their quantitative estimation is yet to be done in Alzheimer's disease brain compared to age-matched control brains. 29 Identification and quantification of the most relevant stereoisomers/structural isomers of Ab is challenging. These isomers are structurally similar, which increases the difficulty of chromatographic separation and are indistinguishable to single-stage mass spectrometers (MS) due to their identical mass-to-charge (m/z) ratios. Analytical chromatographic separation of N-terminal isomers and epimers of Ab and their simultaneous characterization using MS/MS fragmentation techniques have been investigated. [35][36][37][38][39] Chiral chromatography was shown to separate synthetic Ab epimers and isomers containing Asp and Ser residues. 40 Ion mobility separation-mass spectrometry (IMS-MS) is a powerful tool for the analysis and characterization of isomerized and epimerized peptides in the gas phase. [41][42][43][44][45] Recently, synthetic tryptic Ab peptide isomers have been shown to resolve in IMS-MS using structures for lossless ion manipulations (SLIM). 46,47 Coupling of online LC to SLIM-IMS demonstrated the potential of LC-IMS-MS in resolving challenging peptide isomers. 47 In this article, we have identified, characterized and quantified the most common isomers of Ab isoforms  (dashed blue line) and c-secretase (dashed green lines) leads to generation of canonical Ab 1-40 and Ab 1-42 peptides. Proteolytic digestion using LysN enzyme (dashed red line) of the Ab peptide generates N-terminal, mid-domain and C-terminal fragments that were probed for quantitative evaluation in this study. (C) Quantitative proteomics workflow for the estimation of total Ab in the amyloid rich biochemical pools of the brain after digestion with LysN enzyme and spiking of respective stable isotope standard (SIS) peptides without any enrichment strategy (red asterisk). SISCAPA strategy was used for the sparsely enriched peripheral/vascular (Na 2 CO 3 ) pool and soluble pool (TBS) of Ab after enzymatic digestion with LysN and spiking with SIS peptides (indicated by blue asterisk). extracted from the temporal cortex of Alzheimer's disease brains by using liquid chromatography (LC) coupled to drift tube IMS-QTOF MS. In particular, we determined Alzheimer's disease-specific changes of the Ab N-terminal pool in comparison to age-matched control brains; we report the total levels of Ab 1-15 and its associated modified isomers. We also quantified the total levels of the most abundant isomers of Ab [4][5][6][7][8][9][10][11][12][13][14][15] . As an extension of the previously reported biochemical method, 5 we have further quantified the amount of the two classical C-terminal isoforms of Ab, i.e. Ab 42 and Ab 40 13 in the most Ab-enriched brain compartments. More than 92% of Ab in post-mortem brains is partitioned in the insoluble/ fibrillar and detergent soluble phase, while only <7% is extractable as vesicular and soluble. 5 For quantitative estimation of Ab peptides in these moderately/sparsely enriched pools, we developed stable isotope standards and capture by anti-peptide antibodies (SISCAPA) 48,49 of Ab with polyclonal antibodies. The distribution of the isomer ratios indicated a pattern of compartmentalization of highly isomerized Ab 1-15 and Ab 4-15 in the insoluble/ fibrillar and membrane pool, with a comparatively lower extent of isomerization in the vesicular and soluble pools. This data allowed us to estimate the accurate biochemical identity and distribution of the spontaneously isomerized Ab in post-mortem brain. This PTM is intricately associated with slow turnover rates and degradation of Ab which accumulates over decades in sporadic Alzheimer's disease.

Brain tissue
Twenty post-mortem temporal cortex tissue samples were obtained from the Victorian Brain Bank (Australia). In detail, the cohort consisted of age-matched healthy control brains (n ¼ 9), where the number of plaques and tangles were histopathologically analyzed and well below the cut-off values for Alzheimer's disease. No other major neuropathological disease was present. Alzheimer's disease brains (n ¼ 11) met the standard criteria for Alzheimer's disease neuropathological diagnosis (Demographic summary Supplementary Table 1). The study followed the ethics committees of the University of Melbourne (Ethics 1750801.3).

Immunohistochemistry
Segments of frontal cortex from the same cases were fixed in 10% neutral buffered formalin and processed by standard histological methods for paraffin embedding and sectioning (8 mm). Sections were deparaffinised, endogenous peroxidase blocked with 5% aqueous hydrogen peroxide (5 min), treated (5 min) with 98-100% FA (Scharlau AC10852500), rinsed and immersed in Tris buffer (0.5 M pH 7.6). Sections were incubated in a 1/100 dilution of Dako anti-amyloid antibody (MO872-clone 6 F/3D) for 60 min at room temperature. Positively labelled Ab was detected with a peroxidase labelled streptavidin/biotin system (Dako K0675) with a diaminobenzidine chromogen. Sections were counterstained with Harris's haematoxylin, dehydrated and cover-slipped for imaging. Low and high magnification images were obtained with a Leica ICC50 HD camera on a Leica DM 750 binocular microscope.

Tissue homogenization protocol and fractionation
Hemisected freshly frozen postmortem brain tissue was processed as previously described with some modifications. 5 Briefly, the frozen brains at À80 C were warmed to À20 C on ice and the leptomeningeal vessels were removed. The grey matter was dissected into $0.25 g aliquots from temporal cortex (Brodmann's area 21). During dissection process, care was taken to keep the tissues frozen. The tissue was weighed out and was first bio-mashed through the Biomasher (Omni International) by centrifugation at 14 000 g for 1 min at room temperature. To the biomashed tissue, Tris-buffered saline (TBS, 50 mM Tris-HCl, 150 mM NaCl, pH 8.5) containing EDTA-free protease inhibitors (Roche) was added at a ratio of 1:4 (tissue: buffer, w/v). This solution was transferred to ultracentrifuge tubes (Beckman Coulter) and centrifuged (Optima MAX-XP from Beckman Coulter) at 100 000 g for 30 min at 4 C. The supernatant was collected, referred to as TBS fraction henceforth, and stored on ice until freezing.
The resulting pellet was then resuspended in 100 mM Na 2 CO 3 pH 11 (1:4, tissue: buffer) and incubated for 20 min on ice before another ultracentrifugation step at 100 000 g was carried out for 30 min at 4 C. The supernatant containing peripheral membrane and vesicular material was recovered into an Eppendorf tube, referred to as Na 2 CO 3 fraction.
The pellet resulting from Na 2 CO 3 fractionation was resuspended with urea 2 detergent buffer (7 M urea, 2 M thiourea, 4% CHAPS, 30 mM bicine, pH 8.5) and spun at 100 000 g for 30 min at 4 C. The supernatant was aspirated out, referred to as urea 2 detergent fraction. These three biochemical fractions were then snap frozen in liq. N 2 and stored at À80 C until further processing.
The residual pellet was finally incubated in 200 mL 70% glass-distilled FA (GDFA) for 2 h at room temperature in a fume hood. The FA fractions were spun at 13 200 g for 15 min at 4 C and supernatant was collected. The FA fractions (fourth biochemical fraction) were aliquoted into 10 mL portions and snap frozen in liquid N 2 , freeze dried in a lyophilizer and stored at À80 C. A summary of the biochemical fractionation procedure can be found in Fig. 1C.
In-solution LysN digestion of formic acid, urea-detergent, Na 2 CO 3 and TBS fractions A total of 10 mL of both lyophilized FA and urea 2 detergent fractions were re 2 suspended/diluted to 100 mL in 100 mM tri-ethyl ammonium bicarbonate buffer (TEAB), pH 8.5. Next, the samples were reduced by incubating with dithiothreitol (DTT) to a final concentration of 20 mM at 37 C for 30 min, followed by alkylation using 25 mM iodoacetamide (IAA) in the dark for another 30 min. The samples were then diluted to 200 mL with 100 mM TEAB buffer, pH 8.5 and digested overnight by incubation at 37 C after adding LysN metalloprotease at enzyme: protein ratio of 1:100. The same in-solution digestion process was performed with 50 mL for the Na 2 CO 3 and 100 mL for the TBS fractions. The Na 2 CO 3 fraction was diluted to 100 mL and the TBS fraction to 170 mL with 8 M urea, 100 mM TEAB buffer (pH 8.5), respectively. Sample reduction and alkylation were carried out as described above. Finally, the two fractions were diluted to 250 mL for proteolytic digestion with LysN. All the proteomic sample processing was performed at pH 8.5. The digestion reaction was quenched by adding 10% FA to a final concentration of 0.1%. The FA and urea 2 detergent samples were then spiked with 10 mL of SIS Ab peptides mixture (200 fmol/mL of Ab NEP peptides), while only 5 mL was spiked into the Na 2 CO 3 and TBS samples. The acidified samples were finally loaded onto an Oasis HLB mElution 96 well-plate (Waters). The wells were washed with 250 mL of 0.1% FA, followed by 250 mL of 5% methanol, 0.1% FA. The peptides were finally eluted with two sequential washes of 25 mL of 75% ACN, 0.1% FA. The eluent was lyophilized and stored at À20 C until further processing. The FA and urea 2 detergent samples were re-constituted in 25 mL of 2% ACN, 0.05% TFA, vortexed for 30 min on ice and sonicated for 2 min. The re-constituted samples were centrifuged at 10 000 g for 5 min and the supernatant was aliquoted in MS vials (Agilent Technologies) for analysis.
Ab SISCAPA (stable isotope standards and capture by antipeptide antibodies) Enrichment experiments were performed in a round-bottom 96-well polypropylene plates using the magnetic bead protocol. The NEP Ab anti-peptide antibodies were coupled to PureProteome NHS FlexiBind Magnetic Beads (Millipore) according to the manufacture's protocol. At first, the capture efficiencies of the anti-peptide antibodies were determined in a complex background. Lyophilized LysN-digested pooled brain homogenate (10 mg total digested protein) was resuspended in 200 mL with PBS, 0.03% CHAPS pH 7.5 buffer along with the 500 fmol of respective SIS Ab peptides and 1 mg of specific antibody ( Supplementary Fig. 1). These antibodies specifically captured LysN-cleaved versions of Ab peptides with no cross-reactivity for tryptic-cleaved versions.
For the multiplexed experiment, 1 mg of each antibody was added to the sample mixture and 1 M Tris-HCl pH 7.5 to a final concentration of 0.2 mM. To this mixture, 500 fmol SIS Ab peptides were added. The mixture was incubated overnight at 4 C with shaking at 800 rpm.
After overnight incubation the magnetic beads were magnetized, and the supernatant was discarded. Next, the magnetic beads were manually washed three times with 0.1 M ammonium acetate, 0.5 M NaCl, 0.03% CHAPS (pH 7.5) followed by another three washes with 0.1 M ammonium acetate, 15% ACN, pH 7.5. Finally, the captured peptides were eluted from the magnetic beads with 25 mL of 5% acetic acid, 15% ACN with shaking at 600 rpm and 2 min incubation.
This SISCAPA process was used only on the LysNdigested lyophilized Na 2 CO 3 /TBS brain fractions for Ab enrichment.

LC-drift tube ion mobility mass spectrometry
An Agilent 1290 Infinity series UHPLC system coupled to Agilent 6560 Drift Tube Ion Mobility QToF high-resolution MS (Agilent Technologies, Santa Clara, USA) was used for UHPLC-ESI-IM-MS separations. 0.1% FA in water (mobile phase A) and 0.1% FA in 100% ACN (mobile phase B) were used as a solvent system. Samples were loaded onto an Agilent Advanced Bio Peptide Mapping C 18 Column (2.1 Â 150 mm, 2.7 mm) through ultra-low dispersion kit (5067-5963 Agilent Technologies), maintained at 60 C in thermostatted column compartment (TCC) and eluted at 0.4 mL/min flow rate with the following linear gradient: t (min), % B: 0, 2.5; 5,6; 64, 22; 85, 29; 90, 34; 95, 81; 97, 81; 97, 2.5; stop time, 99 min. The ESI source parameters operating in positive ion mode were as follows; gas temp., 300 C; drying gas, 6 L/min; nebulizer, 35 psi; sheath gas temp., 275 C, sheath gas flow, 12 L/min; Vcap, 4500 V. The peptides were analyzed in the positive 4-bit multiplexing IM-QTOF mode in the m/z range of 290-1700 with a maximum drift time of 50 ms using nitrogen as drift gas, trap fill time of 3.2 ms; trap release time of 0.3 ms, and acquisition rate of 1 IM frame/s. The drift tube was operated with an absolute entrance voltage of 1700 V and an exit voltage of 250 V (drift field 18.529 V/cm) and the trapping funnel RF was set at 150 V. An Agilent ESI-Low Calibration mixture was injected both before the analysis to tune the instrument in the m/z range of 100-1700 and at the start of the worklist to perform singlefield Collisional Cross Section ( DT CCSN 2 ) recalibration. The drift gas upgrade kit maintained both the drift tube and trap funnel pressure at constant 3.94 6 0.01 and 3.80 6 0.02 Torr, respectively, while the drift tube ambient temperature was stable at 23.5 6 0.3 C across all the acquisition runs.  Supplementary Table 2. The source ESI parameters as well the collision energies were optimized for these peptides in the positive ion mode. The typical parameters were: gas temperature 200 C, gas flow 15 L/min, nebulizer 40 psi, sheath gas temperature 250 C and sheath gas flow 11 L/min. The capillary voltage was 4500 V and the nozzle voltage was set at 1000 V. The optimized iFunnel parameters were 150 and 60 V for high-and low-pressure RF, respectively. A total of 20 mL of LysN digested Na 2 CO 3 SISCAPA samples were injected on to the columns.

Data processing and statistical analyses
The IMS-MS data files collected using 4-bit multiplexing mode were first de-multiplexed using vendor-supplied software without any smoothing applied. 50 Data postprocessing, including DT CCSN 2 calibration and feature finding was carried out using IM-MS browser and Mass Profiler from MassHunter Suite (B.08.00, Agilent Technologies, Santa Clara, USA). Following post-processing, the raw data were imported into Skyline (v4.2) with formula annotations of the targeted peptides added to the method. Data for each peptide was extracted in the software in a MS1 filtering mode 51 using the accurate mass of the top three isotopic peaks, drift time and RT for the precursor list workflow. The peak abundance for the Ab 1-15 , Ab 4-15 , Ab 16-27 , Ab 28-40 and Ab [28][29][30][31][32][33][34][35][36][37][38][39][40][41][42] peptides in this study were performed on the first three peaks of the isotope cluster. The peak areas of the endogenous peptides and their heavy analogues (R ¼ 13 C 6 , 15 N 4 , K ¼ 13 C 6 , 15 N 2 ) were extracted to derive the light-to-heavy ratios. The absolute quantification was determined by comparing the peak areas of the SIS peptides.
For the drift tube IMS, the resolving power R and resolution r are defined as R ¼ t d /w and r ¼ 1.18 * (t d1 À t d2 )/(w 1 þ w 2 ), where t d is the drift time of the ion and w is the full peak width at half-maximum (FWHM). To determine statistically significant differences between potential biomarkers (healthy versus Alzheimer's disease brain tissue), the unpaired independent sample t-test was used, while Pearson's correlation was used to assess correlation between different cell fractions. For the total amount of isomers and their normalized ratio, adjusted P values were calculated with one-way analysis of variance (ANOVA), corrected for multiple comparison false discovery rate (P < 0.05) with Benjamini-Hochberg correction. The means of most common isomers of Ab [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15] and Ab 4-15 were summarized as pie charts for Alzheimer's disease and control brains, respectively, obtained from different biochemical fractions.
The degree of amyloid pathology was assessed in the post-mortem temporal cortex tissue using anti-Ab immunohistochemistry (IHC) and semi-quantitatively scored by an independent assessor (CAM) after anti-Ab (aa 8-17; 6 F/3D) staining. The scoring system comprised of; the following four categories: À, absent or not discernible, þ, slight; þþ, moderate; þþþ, severe. The semi-quantitative IHC scores were compared with the quantitative results obtained using IMS-MS in this study.

Data availability
Patient and post-mortem brain tissue demographics, experimental details of nano-LC-ESI-MS/MS for ETD-PRM, MRM transition list for Ab SISCAPA on QQQ, age-ofdeath correlation with absolute quantity of Ab peptides and the respective isomers are provided in the supporting information. Additional data related to this article may be requested from the authors.

Characterization of epimerization of Asp residues in brain-derived Ab
Qualitative bottom-up proteomic identification of Ab peptides in Alzheimer's disease brain tissue demonstrated a range of N-terminal truncations including Ab 1-15, Ab 2-15 , Ab 3Glu-15 , Ab 4-15 and the two canonical C-terminal peptides, Ab 28-40 and Ab [28][29][30][31][32][33][34][35][36][37][38][39][40][41][42] (Supplementary Table 3). 13 Previous reports characterizing stereoisomers of synthetic Ab have particularly demonstrated that isomerization at position Asp-1 and Asp-7 is frequent. 23,38,52 However, neither the extent of isomerization of Asp-1 and Asp-7 residues in Ab 1-15, Ab 2-15 , Ab 3Glu-15 and Ab 4-15 have been directly measured in human brain nor has a systematic study been conducted using ion mobility to determine the effect of isomerization on the structure of these peptides in the gas phase. We postulated that the orthogonality of the online LC and drift tube ion mobility separations (DT-IMS) would provide the required analytical resolution, even if at modest R $ 50, to resolve the N-terminal Ab isomers/racemers derived from Alzheimer's disease brains and improve the detection/quantification limits of these complex biological samples. We applied DT-IMS in combination with chromatography and synthetic heavy labelled Ab standards to characterize the identity of isomerized Ab 1-15 , Ab 2-15 , Ab 3Glu-15 and Ab 4-15 from Alzheimer's disease brain.

Absolute quantitation of N-terminal Ab peptides and estimation of Asp-1 and Asp-7 isomerization
In order to understand how the isomerization of Asp-1 and Asp-7 was associated with Alzheimer's disease, we investigated the changes in the total percentage of each isomer/epimer across the biochemical fractions ( Fig. 3C  and D). The percentage of isomerized to unmodified Ab 1-15 indicated significant decrease of the native 1-L, 7-L-Asp (1) peptide with the concomitant statistically significant increase in 1-iso-L-Asp, 7-iso-L-Asp Ab 1-15 (10) isomer in all the biochemical fractions in Alzheimer's disease (Fig. 3C). Overall, $85% of Ab 1-15 was detected in its isomerized form in the amyloid-rich fractions of Alzheimer's disease, while controls showed up to 50% isomerization depending on the pathology (Fig. 3C,  Supplementary Fig. 10). Quantitatively, $50% isomer 10 in the most amyloid-rich fractions of Alzheimer's disease brains compared to 20-27% in controls (Fig. 3C) was documented. Furthermore, we observed $21-30% Ab 1-15 with either Asp-1 or Asp-7 isomerized in Alzheimer's disease. In contrast, in controls singly isomerized Asp-1 or Asp-7 Ab 1-15 are the predominant species in the FA fractions ($37%, P ¼ 0.006) (Fig. 3C). These data indicated an increased isomerization event of Ab 1-15 in Alzheimer's disease brain for an extended period of time. Strikingly, even the TBS soluble Ab 1-15 present in the Alzheimer's disease brains demonstrated $70% isomerization ($37% isomer 10, P ¼ 0.0059) compared to only $30% in controls (Fig. 3C).
The potential influence of age at death on the accumulation of isomerized Ab in the Alzheimer's disease brains was evaluated. As expected, the total levels of Ab [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15] and Ab 4-15 were positively correlated with the age at death in the Alzheimer's disease brains ( Supplementary  Fig. 11) but not in control brains ( Supplementary Fig.  12). However, no correlations with the isomer ratios of Ab 1-15 or Ab 4-15 were observed in Alzheimer's disease or controls ( Supplementary Figs. 11 and 12). These results corroborate the spontaneous non-enzymatic reaction as the primary mechanism for the generation of these isomers on long-lived Ab in the brains (not artefact of sample preparation).

Discussion
The slowly progressive nature of Alzheimer's disease with almost 20 years of Ab accumulation from threshold to the onset of dementia predisposes the depositing Ab peptide to undergo multiple biochemical changes at the molecular level. We used a quantitative proteomics approach coupled with ion mobility mass spectrometry to unravel the diversity of isomerized Ab N-termini found in the Alzheimer's disease brain. Our major findings include (i) characterization of isomerization of the Asp residues (Asp-1 and Asp-7) in four common sequentially truncated N-termini of Ab found in Alzheimer's disease brain tissues; (ii) quantitative estimation of the level of Ab [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15] and Ab [4][5][6][7][8][9][10][11][12][13][14][15] in the biochemical pools with significant elevation in the Alzheimer's disease brain tissue; (iii) evaluation of the isomer ratios of Ab 1-15 and Ab 4-15 , with significant elevation in doubly isomerized Ab 1-15 and isomerized Ab 4-15 levels in the insoluble/fibrillar and membrane pools, while the sparsely populated vesicular and soluble Ab pools have lower proportion of these PTMs; (iv) brain-derived Ab primarily has Ala-42 as the C-terminus which is significantly increased in Alzheimer's disease, while Ab with Val-40 C-terminus is increased in Alzheimer's disease but does not reach statistical significance compared to control brains.
Iso-aspartate formation is one of the most common modifications associated with long-lived proteins/peptides. 22,58 The rate of iso-Asp formation in model peptides is considerably slower (half-life t 1/2 , 53-266 days depending on the sequence) compared to asparagine deamidation/isomerization (t 1/2 , 1.4days). In vitro, the N-terminus of Ab has been documented to undergo such spontaneous isomerization at Asp1 and Asp7 residues. 23 Slow reaction rates (t 1/2 $ 231 days, Asp-1; t 1/2 $ 462 days, Asp-7 for Ab 1-40 ) 12 along with decreased fractional clearance rates in the CNS of Alzheimer's disease (28 ng/h Ab 1-42 deposition) 5,59 can be used to estimate the age of the depositing isomerized peptides. Pathological controls with $50% isomerized Ab 1-15 (one t 1/2 ) indicate that the observed mild Ab deposit (diffuse plaques) is nearly 8 months old. In contrast, data from the $85% isomerized Ab 1-15 (three t 1/2 ) in Alzheimer's disease (Fig. 3C), indicate that the age of this peptide is at least 4 years. Similarly, the age of Ab 4-15 in Alzheimer's disease is nearly 1.2 years compared to 6 months in control brains. Interestingly, the rate of racemization of L-Asp to D-Asp was originally used to estimate 30 years for plaques formation. 60 Further investigation using better modelling of Ab biogenesis and altered clearance rates observed in Alzheimer's disease patients would yield better estimates for these long-lived PTMs.
Along with N-terminal Asp isomerization, sequential truncated isoforms of Ab such as Ab pGlu have been well documented from different biochemical pools of Alzheimer's disease brain. 61 All of these PTMs have been linked to the hypothesis of how Ab is toxic to neurons. However, they do not completely address the underlying feature of how or what causes the accumulation of Ab to occur. Structural reorganization of the peptide chain due to Asp isomerization leads to alteration in the biochemical and physical properties of the peptide. Our data indicate that internal Asp residue isomerization reorients the peptide backbone, leading to changes in the shape and size of these Ab peptides (Fig. 2) compared to unmodified ones in the gas phase. One of the possible links between the more stable long-lived isomerized/epimerized Ab 62 and neurotoxicity could stem from their inherent resistance to enzymatic degradation by primary cathepsin found in the lysosomes. 12 Ab residues 1-11 are predicted to play a critical role in the antigen recognition by antibodies targeting the N-terminus of Ab peptide. [63][64][65] It has been suggested that N-terminus of Ab is the dominant epitope, exposed on the surface of aggregated fibrillary deposits, while Ab mid-domain drives oligomerization and toxicity. 66 Antibodies that target N-terminus are considered competent in reducing Ab deposits, while antibodies to mid-domain epitopes 67,68 should abrogate the toxic oligomers. Despite considerable reduction in Ab (lowering of Ab-PET signal) by monoclonal antibodies primarily to the Ab N-terminus, 33,34,69 active and passive immunotherapy trials have largely failed to reach their primary end points. 70 Our data suggest, antibodies targeting the mid-domain Ab might prove efficacious as it has very little PTM, while specifically targeting the older isomerized Ab N-terminus for clearance will be better strategy for immunotherapy. Designing better therapeutic antibodies against modified Ab would need further investigation into the structural properties of these PTMs and their influence on the antibody-mediated target engagement.
It has been postulated that the hydrophobic C-terminus of Ab is responsible for inducing membrane permeability, 71 while the N-terminal domain induces innate immune responses from the microglia. Interestingly, it has been found that iso-Asp-7 Ab 42 compared to wild-type Ab 42 led to significantly increased phosphorylation of proteins, including tau (MAPT) from SH-SY5Y neuroblastoma cell-culture models. 72 Accumulation of iso-aspartate in proteins is known to be lethal in the PIMT (protein iso-aspartate methyltransferase) deficient mouse, suffering from progressive epileptic seizures. 73,74 Soluble Ab oligomers isolated from Alzheimer's disease brains have been shown to induce hyperexcitability in individual neurons and neuronal circuits [75][76][77] Induction of hyperexcitability has been invoked to explain the clinical observation that there is a significantly higher incidence of epilepsy in Alzheimer's disease patients compared to agematched controls. 78,79 Our results indicate that soluble Ab 1-15 derived from Alzheimer's disease brain is significantly isomerized ($50% doubly isomerized, 20% singly isomerized) compared to soluble Ab 1-15 ($20% doubly isomerized, 17% singly isomerized) in age-matched control brains (Fig. 3C). It would be interesting to quantitatively estimate how much of these Alzheimer's disease brain-derived soluble Ab oligomers are isomerized at the N-terminus.
While we documented abundant N-terminal Asp-1 and Asp-7 isomerization/racemization in all the four different biochemical pools in both Alzheimer's disease and control brains (Fig. 3, Supplementary Fig. 10), surprisingly no modified Ab [16][17][18][19][20][21][22][23][24][25][26][27] was observed ( Supplementary Fig. 7). The presence of unique Ab 16-27 species points to two major revelations: (i) N-terminus of Ab is conformationally flexible allowing spontaneous reactions to occur and (ii) contrary to previous reports, 9,80 Asp-23 is unmodified in sporadic Alzheimer's disease. This indicates that this residue could either be solvent in-accessible or involved in H-bonding interactions precluding it from succinimidemediated isomerization. With the current resolution of R $ 50 for our DT-IMS-MS method, it is not possible to rule out any other amino acid (such as Ser) isomerization on this peptide. Future investigations with techniques like SLIM-IMS providing higher resolution (R > 300) 46,47 will lead to better understanding and characterization of other low abundant structural PTMs of Ab in Alzheimer's disease brains.
The data presented here and by others 13,59,81-83 is consistent with Ab 42 being the dominant neuronal peptide form accumulating in Alzheimer's disease brain with Ab 40 levels increasing with perivascular amyloidosis. 54 Label-free intact MS has estimated that $70% of Ab depositing in the Alzheimer's disease brain has Ala-42 as the C-terminus compared to $10% terminating at Val-40. 13 Historically, the majority of the peptide originally sequenced from the plaque-derived amyloid was Ab 4-42 . 7 Our results indicate that Ab peptides depositing specifically in the insoluble pools of Alzheimer's disease brain have approximately equal amounts of BACE-1 cleaved Ab (Asp-1 as the N-terminus) and ragged N-terminus peptide (Phe-4 residue) (Table 1, Fig. 3). Interestingly, recent MALDI-MS imaging of post-mortem Alzheimer's disease tissues with congophilic amyloid angiopathy (CAA) provided a distinct qualitative pattern of N-and C-terminal variations of deposited Ab-extracellular plaques in the cerebral parenchyma were enriched with Ab 42 while the vessels had less aggregation prone Ab 40 . [83][84][85] Quantitative estimation of Ab 42 and Ab 40 in biochemically defined pools from the temporal cortex of sporadic Alzheimer's disease revealed less than 0.01% of total Ab 42 and 0.3% of total Ab 40 are in the soluble cytosolic TBS fraction with the rest being distributed in either the vesicular (1.1% Ab 42 , 1.7% Ab 40 ), membranous (28.3% Ab 42 , 6.8% Ab 40 ) and/or insoluble fully polymerized fibrillar phase (70.7% Ab 42 , 91.2% Ab 40 ). In contrast, total Ab 42 in control brain tissue (0.5 6 1.8 pmol/mg brain) is much lower concentrated compared to Alzheimer's disease brain (3.2 6 1.8 pmol/mg brain). Again, most of the Ab 42 is still partitioned in the membranous (57.2%) and insoluble/fibrillar fraction (41.2%). Drugs that can target the C-terminus, 86 specifically Ab 42 for clearance have much better chance to exploit the equilibrium of amyloid deposition in Alzheimer's disease brain.
Development of therapeutic drugs and interventions for ameliorating or decreasing the progress of Alzheimer's disease requires techniques that can accurately and quantitatively monitor the changes in the amyloid biomarkers in the CSF and/or the blood along with PET imaging. Our results show that isomerized Ab is intricately associated with the accumulation of Ab 42 in the braina key distinguishing signature from freshly generated Ab. Future studies will be required to understand the role of these isomers in the disease, but this is clearly an important question to answer due to the >80% abundance of isomerized Ab in Alzheimer's disease brain.

Conclusion
In summary, in this study we have shown that different biochemical pools of Ab has different amounts of N-terminus isomerization. Insoluble plaques and membrane fractions in Alzheimer's disease brains have $85% isomerized Ab 1-15 , while vesicular and soluble fractions have lower percentage of isomerization. The extent of isomerization on Ab extracted from Alzheimer's disease brains is 3 years older than the Ab found in age-matched control brains. Quantitatively, BACE-1-cleaved Asp-1 Nterminus is present in almost equimolar amounts with Phe-4 truncated N-terminus, an interesting Ab metabolic by-product of unclear origin. Our data provide the link between older isomerized Ab and the consequences it might have in the disease aetiology, such as oligomers that diffuse out of these plaques into soluble pool will be neurotoxic due to their inherent resistance to lysosomal degradation. Strategies in designing better immunotherapeutic must take into consideration of the extensive PTMs of the N-terminus of Ab and specifically target older isomerized Ab species for better target engagement and clearance.

Supplementary material
Supplementary material is available at Brain Communications online.