Quantitative Mass Spectrometric Assay of Whole and CNBr-Cleaved Amyloid-β Peptides in Human Brain

Abstract

Various amyloid-β (Aβ) peptides accumulate in brain in Alzheimer’s disease, and the amounts of specific peptide variants may have pathological significance. The quantitative determination of these variants is challenging because losses inevitably occur during tissue processing and analysis. This report describes the use of stable-isotope-labeled Aβ peptides as internal standards for quantitative mass spectrometric assays, and the use of cyanogen bromide (CNBr) to remove C-terminal residues beyond Met35. The removal of residues beyond Met35 reduces losses due to aggregation, and facilitates the detection of post-translationally modified Aβ peptides. Results from 8 human brain samples suggest that the tissue concentrations of the 42-residue Aβ peptide tend to be similar in different patients. Concentrations of the 40-residue Aβ peptide are more variable, and may be greater or lesser than the 42-residue peptide. The concentration of the CNBr cleavage product closely matches the sum of the 40-residue and 42-residue peptide concentrations, indicating that these two Aβ peptides account for most of the C-terminal variants in these patients. CNBr treatment facilitated the detection of post-translational modifications such as pyroglutamyl and hexose-modified Aβ peptides.

Keywords

Immunoprecipitation internal standards isotope labeling post-translational modification

INTRODUCTION

The accumulation of aggregated amyloid-β (Aβ) peptides in the brain is one of the hallmarks of Alzheimer’s disease (AD). Several variant peptides with N- and C-terminal truncations are commonly found in both human disease and animal models [1 –3], and it has been suggested that the ratio of some variants may have pathological significance [4]. Therefore, a method for quantifying low nanomolar concentrations of unambiguously identified Aβ peptide variants in brain tissue is of interest.

ELISAs are most commonly used for the quantification of Aβ peptides in tissue extracts, although their specificity for variant forms may be unclear, and losses of unknown magnitude may occur during tissue processing. For these reasons, quantification by mass spectrometry with internal stable-isotope-labeled (SIL) standards is advantageous. Because they are subject to the same extraction, ionization, and fragmentation efficiencies as the analytes of interest, the addition of SIL standards in known amounts to samples early in the extraction process helps avoid the underestimation of analyte concentrations due to losses during sample processing.

Mass spectrometry is commonly used without internal standards to identify the presence of Aβ peptide variants in tissues [1–3 , 5–7]. For example, Wildburger et al. recently documented the presence of 26 unique “proteoforms” by mass spectrometry using methods that could only determine the relative amounts of any given proteoform between samples [1]. However, Pannee et al. used U-¹⁵N-labeled forms of Aβ³⁸, Aβ₄₀, and Aβ₄₂ as internal SIL standards to assay these peptides in cerebrospinal fluid (CSF) and plasma [8, 9]. Mawuenyega et al. [10] and Ovod et al. [11] made use of Aβ peptides containing ¹³C₆-Leu as well as U-¹⁵N-labeled forms of Aβ₃₈, Aβ₄₀, and Aβ₄₂ to quantify these peptides in cell cultures, CSF, and plasma. Following Kaneko et al. [12], Nakamura et al. [4] used Aβ₃₈ containing two ¹³C₉-Phe and two ¹³C₆-Ile residues as internal SIL standards to assay Aβ₃₈, Aβ₄₀, and Aβ₄₂ in the same samples. They cited analytical simplicity, and its lower tendency to aggregate and adhere to the walls of storage tubes, to justify the use of SIL-Aβ₃₈ as a standard for all three peptides, instead of also using SIL-Aβ₄₀ and SIL-Aβ₄₂.

The use of internal SIL standards for the quantification of Aβ peptides in brain tissue by mass spectrometry has not been reported, and is more challenging than assays in CSF and plasma because the additional steps involved in extracting peptides from solid tissues and disaggregating them for analysis greatly increases the potential for quantitatively large peptide losses. This report describes the use of multiple-reaction monitoring (MRM) mass spectrometry with internal SIL standards to determine the absolute concentrations of unambiguously-identified Aβ peptide variants in brain tissue. Cyanogen bromide (CNBr) was used to cleave residues beyond Met35, eliminating a series of aggregation-promoting hydrophobic C-terminal residues, and enabling the collective determination of C-terminal variants as a single species.

Results indicate that either the 40-residue variant (Aβ₄₀) or the 42-residue variant (Aβ₄₂) may predominate in AD brain tissue. The concentrations of Aβ₄₂ tend to be similar among AD patients, whereas the concentrations of Aβ₄₀ are more variable. Concentrations of the CNBr cleavage product (Aβ_35hsl) are equal to the sum of Aβ₄₀ and Aβ₄₂ concentrations, indicating that quantitatively significant amounts of other C-terminal Aβ peptide variants are not present. The collective determination of C-terminal variants facilitated the detection of post-translationally modified peptides, including substantial amounts of N-terminal variants such as pyroglutamyl Aβ₄₂ (Aβ_p3-42) and variants with hexose-modified side chains.

MATERIALS AND METHODS

Reagents

4G8 and 6E10 antibodies were purchased from BioLegend (San Diego, CA). ¹³C₆-Leu was purchased from Cambridge Stable Isotopes (Tewksbury, MA). Aβ₄₀ and Aβ₄₂ were synthesized with ¹³C₆-Leu at positions 17 and 34 (to yield ¹³C₁₂-Aβ₄₀ and ¹³C₁₂-Aβ₄₂) by Life Tein (Somerset, NJ). Pierce Protease and Phosphatase Inhibitor Mini Tablets and Dynabeads™ M-280 Sheep Anti-Mouse IgG were purchased from Thermofisher (Pittsburgh, PA). All other chemicals were purchased from Fisher Scientific (Pittsburgh, PA).

Immunobead preparation

Dynabeads™ (M-280 Sheep Anti-Mouse IgG) were washed 3 times with 600 μl PBS (NaCl 137 mM, KCl 2.7 mM Na₂HPO₄ 10 mM, KH₂PO₄ 1.8 mM) and incubated overnight with 16 μg each of 4G8 and 6E10 antibodies per 100 μl beads at 4°C in PBSI (PBS buffer with protease inhibitor, one tablet per 10 ml PBS). Following incubation beads were washed 4 times with PBS and stored in PBSI at 4°C for up to one week.

Human brain tissues

Frontal cortex from 8 human patients with documented severe AD histopathology, and 3 healthy human brains, were obtained from the Center for Neurodegenerative Disease Research at the University of Pennsylvania with the characteristics listed in Table 1. All brains were designated Braak stage V or VI. 50 mg portions of these tissues were dissected, weighed and homogenized in 200 μl PBSI with 10 μl of 250 nM ¹³C₁₂-Aβ₄₀ and 10 μl of 430 nM ¹³C₁₂-Aβ₄₂ using a tip sonicator (F60 Sonic Dismembrator, Fisher Scientific, Pittsburgh, PA). 660 μl of formic acid was added to the homogenate (to yield 75% formic acid), which was then sonicated for another 30 s. The resulting suspension was clear. After 30 min at room temperature it was dried under nitrogen gas. Dried samples were resuspended in 800 μl PBSI and adjusted to pH 7.4 by adding 35 μl 10M NaOH and either additional NaOH or HCl as required. This resuspended extract was incubated with 50 μl Dynabeads with bound antibodies for 1 h at room temperature. The beads were retained with a magnet during 4 washes with PBS, and bound peptide was released by adding 60 μl of 70% formic acid.

Table 1

Characteristics of the human brain samples analyzed in this work

#	Sex	Age	Pathology	PMI (h)	ApoE isoform
AD brains
1	M	69	AD	22	3/3
2	F	66	AD	19	3/3
3	M	79	AD	17	3/3
4	M	86	AD	17	3/3
5	M	82	AD	3	3/4
6	M	74	AD	18	4/4
7	M	67	AD	10	4/4
8	M	81	AD	4	4/4
NL brains
1	F	67	None	24	3/3
2	M	68	None	17	2/3
3	M	81	None	19	3/3

CNBr cleavage

Peptide cleavage was done by adding CNBr crystals (∼3 mg) to the sample after release from the beads and incubating at room temperature. Cleavage of Aβ₄₀ and Aβ₄₂ to a peptide truncated beyond residue Met35 and terminating with a homoserine lactone (Aβ_35hsl) was monitored by mass spectrometry and determined to be complete after 15 min.

LC/MS analyses

10 μl of the solution containing released peptides in 70% formic acid were injected onto a 1×50 mm Eclipse XDB, 3.5 μm C18 column (Agilent, Santa Clara, CA). Solvent A was 5% acetonitrile, 95% water, and 0.01% trifluoroacetic acid. Solvent B was 100% acetonitrile and 0.01% trifluoroacetic acid. The mobile phase was pumped at 100 μl/min as the composition was changed linearly from 10% to 80% solvent B over 8 min, and then held at 100% B for 3 additional minutes. The eluent was introduced via electrospray ionization (ESI) into an ABI 4000 mass spectrometer (Sciex, Toronto, Canada) operating in positive ion mode with a declustering potential of 125 V, ionization energy of 5500 V, drying gas at 300°C, and nitrogen CAD gas at 5 psi. Collision energies and MRM transitions were listed in Table 2. The integrated areas from MRM signals were divided by signals from the corresponding internal standard; i.e., Aβ₄₀/¹³C-Aβ₄₀, Aβ₄₂/¹³C-Aβ₄₂, and Aβ_35hsl/¹³C-Aβ_35hsl using Analyst software (Sciex, Toronto, Canada). Peptide concentrations per gram of tissue were determined from the integrated areas and tissue weights according to $\frac{^{12} C analyte signal area \times^{13} C internal standard (moles)}{^{13} C internal standard signal area \times tissue sample mass (grams)}$

Table 2

MRM transitions for Aβ peptides, variants, and ¹³C₁₂-labeled internal standards

Molecular species	m/z of most abundant parent ion isotopomer at z = 4	Fragmention	m/z of most abundant fragment isotopomer	m/z of internal standard parent ion	m/z of internal standard daughter ion	CE
Aβ₄₀	1083.2949	b39, 4+	1054.0246	1086.2949	1057.0246	32
Aβ₄₂	1129.3252	b40, 4+	1078.7917	1132.3252	1081.7917	37
Aβ_35hsl	968.4821	b32, 3+	1200.5885	971.4821	1203.5885	35
Aβ_p3-42	1078.5526	b39, 4+	1027.2901			37
Aβ_p3-35hsl	917.4634	b30, 3+	1132.5637			35
Aβ₄₀+hexose	1123.8081	b39, 4+	1094.5378			37
Aβ_35hsl+hexose	1008.9953	b32, 3+	1254.6061			35
Aβ₃₈	1033.7607	b37, 4+	1015.0022			37
Aβ₃₉	1058.5278	b38, 4+	1029.2575			37
Aβ_3-39	1012.0118	b36, 4+	982.7415			37
Aβ_2-40	1054.5382	b38, 4+	1025.2679			37
Aβ_3-40	1036.7789	b37, 4+	1007.5086			37
Aβ_4-40	1004.5183	b36, 4+	975.2480			37
Aβ_5-40	967.7512	b35, 4+	938.4809			37
Aβ₄₁	1111.5660	b40, 4+	1078.7917			37
Aβ_2-42	1100.5685	b39, 4+	1050.0350			37
Aβ_3-42	1082.8092	b38, 4+	1032.2757			37
Aβ_4-42	1050.5486	b37, 4+	1000.0151			37
Aβ_5-42	1013.7815	b36, 4+	963.2480			37

CE, collision energy (volts).

The concentrations of the ¹³C-labeled internal standards were determined by integrated absorbance at 280 nm, using an in-line absorbance detector. For the Aβ_35hsl internal standard, the sum of the Aβ₄₀ and Aβ₄₂ internal standards was used.

Limit of detection (LOD) determination

Synthetic Aβ₄₀ and Aβ₄₂ were dissolved in a mixture of 40% hexafluoroisopropanol (HFIP) and 60% 5 mM HCl and then lyophilized overnight. The dissolution and lyophilization steps were repeated a second time before finally dissolving the lyophilized peptide in HFIP at 20°C. Peptide concentrations were determined in two ways. One based on the absorbance at 280 nm with an extinction coefficient of 1490 M^–1cm^–1 for the single Tyr residue present in each peptide. The other was a BCA assay (ThermoFisher) with bovine serum albumin a reference standard. These assays invariably agreed to within 20%. Synthetic Aβ₄₀ and Aβ₄₂ peptides in HFIP were evaporated, redissolved in PBSI to make stock solutions of each, and the concentrations of these stock solutions were determined using the BCA assay. Increasing amounts of these stock solutions were added to 50 mg samples of normal brain tissue homogenized in 75% formic acid. Samples without the addition of synthetic Aβ yielded no measurable signals for Aβ₄₀ or Aβ₄₂. LODs were the minimal amounts of added Aβ₄₀ and Aβ₄₂ that yielded clear peaks, at the expected retention times, with peak heights 3-fold greater than the mean background.

Western blot

Human brain samples from normal and AD brains, were processed as described above. Samples in formic acid were dried, reconstituted in 60 μl PBSI and neutralized to pH 7.4 using NaOH. Samples were electrophoresed together with PageRulerTM Plus pre-stained protein ladder (ThermoFisher) on a Tricine 10–20% gel (500 ng of protein per well) and were transferred to a 0.2 micron nitrocellulose membrane. The membrane was removed and blocked with 2% milk (Bio-Rad) in phosphate-buffered saline with Tween (PBS-T: 0.1% Tween 20 in 137 mM NaCl, 10 mM phosphate, 2.7 mM KCl, pH 7.4) at room temperature. The membrane was then immunoblotted overnight in 4°C with 6E10 and 4G8 antibodies diluted 1 : 200 in PBS-T, washed with PBS-T, and treated with goat anti-mouse HRP-conjugated secondary antibody (1 : 1000) for 1 h at room temperature. The membrane was then washed, incubated with the SuperSignal^TM West Femto (ThermoScientific), and visualized using ChemiDoc^TM imaging system (Bio-Rad).

Peptide sequencing

Beads from two 60 mg brain tissue samples were combined and washed as described above. Bound peptides were released by suspending the beads in 60 μl 70% formic acid, and cleaved by CNBr treatment. This solution was passed through a C18 column and injected onto a Q Exactive system by the Protein core facility for sequence analysis using the Xcalibur software.

RESULTS

Western blot analysis of immunoprecipitated material revealed two major bands, one at 58 kDa in both normal and AD brains, and another at 4 kDa that only appeared in AD brain (Fig. 1). While the latter undoubtedly represents monomeric Aβ peptide, the identity of the 58 kDa band is unclear, but may be the heavy chain of mouse IgG. This band is reminiscent of a previously reported 56 kDa oligomeric form of the Aβ peptide identified in transgenic mice [13]. In this case, it is abundant in normal human brain. Proteomic analysis of this material did not yield interpretable results.

Fig.1

Western blot of immunoprecipitated material from two AD brains and one healthy brain. There are 4–5 kDa bands in material from both AD brains, but not the healthy (NL) brain. Faint bands at 9 kDa and 13 kDa may represent oligomeric Aβ peptides. This immunoprecipitated material is treated with 70% formic acid prior to quantitative analysis, so that any fibrils and oligomers present would be dissociated and quantified as monomers. Bands at 58 kDa in all three samples are most likely the heavy chains of 4G8 and 6E10.

Synthetic Aβ₄₀ and Aβ₄₂, along with the corresponding internal SIL standards, eluted as narrow peaks when analyzed by ESI-MRM-LC/MS with transitions corresponding to the +4 charge state (Table 2). Aβ₄₀ eluted at 3.97 min, slightly ahead of Aβ₄₂ at 3.98 min (Supplementary Figure 1). LODs were determined to be 500 pg for Aβ₄₀, and 2000 pg for Aβ₄₂ in 50 mg tissue samples (10 ng/gm and 40 ng/gm, respectively).

Peptide concentrations in human frontal cortex samples are illustrated in Fig. 2. Average Aβ₄₀ and Aβ₄₂ concentrations in the 8 AD samples were similar at 37 and 50 μg/gm tissue respectively. The individual concentrations of Aβ₄₀ varied widely (s.d. = 62 μg/gm), whereas concentrations of Aβ₄₂ were more similar between patients (s.d. = 20 μg/gm). Neither Aβ₄₀ nor Aβ₄₂ could be detected in 50 mg of normal brain tissue, but small amounts of Aβ₄₀ could be detected in 300 mg samples. Aβ₄₂ could not be detected in 300 mg samples in normal brain tissue.

Fig.2

Aβ₄₀ and Aβ₄₂ concentrations in tissue from human frontal cortex tissue. Left panels: AD brains. Right panels: healthy brains. Three portions from each brain (designated by □, ○, and ▴) were assayed independently (except for Aβ₄₂ from patient 7), and all of the individual results are plotted. The ApoE genotype of each patient is indicated and the sample numbers correspond to the patient numbers in Table 1. Note the use of a vertical log scale. For samples in which no Aβ peptide was detected, the limit of detection was plotted. Symbols representing the average concentrations (designated by •) are equal to or larger than the standard error of the mean. Sample numbers correspond to the AD and NL patient numbers in Table 1.

To assess the contribution of C-terminal variants to the accumulated Aβ peptide load, samples were treated with CNBr after release from the beads. CNBr cleaves after methionine residues, leaving a homoserine lactone ring (hsl) at the C-terminal. Aβ peptides contain a methionine residue at position 35 and therefore, Aβ₄₀, Aβ₄₂, and all other C-terminal variants beyond residue 35 will be cleaved into Aβ_35hsl. There is also a methionine residue in amyloid-β protein precursor (AβPP) immediately preceding the β-secretase cleavage site. Therefore, CNBr cleavage of AβPP and C99 (the product of β-secretase activity on AβPP) could both yield the same CNBr cleavage product as Aβ₄₀ and Aβ₄₂. However, neither AβPP nor C99 were detected by western blot analysis and therefore we assume that their contributions are negligible. The LOD for Aβ_35hsl was 200 pg (4 ng/gm), less than half the LOD of Aβ₄₀ and tenth of the LOD of Aβ₄₂. Aβ_35hsl concentrations in AD brains were similar to the sum of the Aβ₄₀ and Aβ₄₂ concentrations, suggesting that no other C-terminal truncation forms made substantial contribution variants to the accumulated Aβ peptide load (Fig. 3).

Fig.3

The sum of Aβ₄₀ and Aβ₄₂ concentrations compared to the Aβ_35hsl concentration in 8 AD brains. Results are the mean and standard error of three independent determinations from each patient. Sample numbers correspond to the AD patient numbers in Table 1.

To assess the presence of other variant forms among the accumulated Aβ peptide, two different approaches were taken. The first was scanning for the neutral losses of the C-terminal valine of Aβ₄₀, and the C-terminal isoleucine+alanine of Aβ₄₂, in the +4 charge state. Scans performed on samples from four of the AD brains revealed signals arising from one peptide that was 40.5 m/z larger than Aβ₄₀, which eluted at 3.9 min, and a second peptide that was 25.2 m/z smaller than Aβ₄₂, which eluted at 4.1 min. High resolution sequencing revealed that the first peptide was Aβ₄₀ with single hexose modifications on Lys16, Lys28, or Arg 5. The second peptide was Aβ_p3-42, bearing a pyroglutamyl N-terminus. Absolute quantitation of hexose-modified Aβ₄₀ and Aβ_p3-42 variants was not possible because they were not detected by ESI-MRM-LC/MS and, even if they were, calibrated internal standards for these peptides were not available. However, they were detected after CNBr treatment; MRM signals from Aβ_35hsl with hexose modifications were 10% and 2% of the Aβ_35hsl signals in 2 of 4 AD brains tested, and not detectable in the other 2. MRM signals for Aβ_p3-35hsl ranged between 8 to 40% of Aβ_35hsl in all 4 brains tested (Fig. 4).

Fig.4

Signal magnitudes of hexose-modified Aβ_35hsl and pyroGlu-terminated Aβ_p3-35hsl as a percentage of Aβ_35hsl signal magnitudes in 4 AD brains. Sample numbers correspond to the AD patient numbers in Table 1.

The second approach to assessing the contribution of other variant forms was examining for the presence of Aβ peptide variants using the transitions listed in Table 2. These transitions monitor b-ion fragments in the 4+ charge state derived from naturally-occurring N-terminal and C-terminal truncated forms. Aside from Aβ₄₀ and Aβ₄₂, the largest signals corresponded to Aβ_2-40, Aβ₃₉, Aβ_3-42, and Aβ_4-42, although wide variability between brains was observed (Fig. 5). Absolute quantitation of these truncation variants was not possible because calibrated internal standards for these peptides were not available. However, the signals arising from Aβ₄₀ and Aβ₄₂ together accounted for ∼70% of the total signals for all peptide variants.

Fig.5

Signal magnitudes for the unmodified Aβ variants listed in Table 2 as a percentage of total signal magnitude, for 3 AD brains. Sample numbers correspond to the AD patient numbers in Table 1.

DISCUSSION

These results reveal that Aβ₄₀ concentrations in the samples studied were broadly distributed from 1.4–170 μg/gm brain, whereas Aβ₄₂ concentrations were more narrowly distributed from 20–56 μg/gm. A similar pattern is apparent in at least 3 independent sets of previously published data, although it has not received comment [2 , 14]. Despite significant differences in total Aβ concentration reported in these studies, Aβ₄₀ concentrations within each study varied greatly between patients whereas Aβ₄₂ concentrations were similar.

It is widely appreciated that Aβ₄₂ has a greater tendency to aggregate than Aβ₄₀, yet Aβ₄₀ concentrations exceeded Aβ₄₂ concentrations in 25% of the 22 samples from 8 AD brains. Since the net accumulation of any peptide is the difference between production and clearance, an as-yet-unidentified biochemical process may either curtail the production, or increase the clearance, of Aβ₄₂ when a threshold amount of Aβ₄₂ has accumulated, regardless of patient age, sex, ApoE phenotype, or extent of Aβ₄₀ accumulation. Pivtoraikoa et al. observed that most Aβ₄₂ in the brain is located in plaques and they reasoned that this location is due to the expression and aggregation of Aβ₄₂ at early stages of the disease [15]. The production of Aβ₄₂, in turn, may be regulated by changes in the cleavage selectivity of γ-secretase at different disease stages, or changes in the clearance of Aβ₄₂ over the course of the disease.

In healthy controls, Aβ₄₀ concentrations were dramatically lower (∼0.1 μg/gm) than in AD patients while Aβ₄₂ was not detected. The LOD for Aβ₄₂ is higher than Aβ₄₀ in our system (4000 versus 500 pg, respectively), which may have precluded detection of the former. However, these results are similar to previous reports for healthy controls [3, 15]. Our results from AD patients, on the other hand, are consistently higher than in previous reports. Total Aβ concentrations in prior reports range between 0.02 to 10 μg/gm tissue [2 , 14–19], whereas they average around 80 μg/gm in this work. These differences may be due simply to differences between the patients studied in each case, or differences between brain regions. However, another possible reason is that the use of a simplified peptide extraction protocol, and the addition of internal SIL standards to the initial tissue homogenate, corrected for losses that would otherwise cause an underestimate. The use of internal SIL standards eliminates many other potential sources of error including uncertainties about the efficiency of the extraction procedure, oxidation or formylation of the peptides during extraction, and affinity or specificity of the antibodies. One potential source of error that remains is incomplete disaggregation of amyloid fibrils. However, the conditions applied to disaggregate fibrillar Aβ in this work are at least as vigorous as that described in previously published studies, and the concentrations measured are as high or higher than previously published studies.

The determination of Aβ peptide concentrations by ELISA [20, 21] is widely practiced, but is subject to error from interference with antibody binding affinity, losses during extraction, and cross-reactivity with other peptides. Other antibody-based quantification methods such as immunomagnetic reduction [22] are vulnerable to the same errors. As cited above, quantitative mass spectrometric methods have been developed for Aβ peptides in CSF and plasma. All approaches, however, require chemically aggressive disaggregation using high concentrations of guanidine hydrochloride or formic acid, and are vulnerable to aggregation following extraction, or adsorption to the walls of tubing and sample containers.

To optimize sample preparation for mass spectrometric analysis in these investigations, three adaptations were made. The first is the simplification of the extraction method. Instead of including steps such as ultracentrifugation and filtration, the procedure described herein is composed of only two major steps, defibrillization of plaques in formic acid and immunoprecipitation. Thus, steps that might lead to losses (e.g., sample transfer between vials) are omitted, allowing a better and more consistent detection of Aβ concentrations in tissues. The elimination of various intermediate steps entails a cost; e.g., the approach does not separate water soluble, detergent-soluble, and formic acid-soluble forms of Aβ peptides. However, most of the Aβ in the brain is likely to be in formic acid-soluble plaques.

The second adaptation is the use of internal SIL standards. Known amounts of these standards (quantified by absorbance at 280 nm) were added to each tissue sample prior to extraction, thereby controlling for losses during tissue processing and sample analysis.

The third adaptation is the use of CNBr for the cleavage of Aβ. The cleavage product is much less likely to aggregate, and the LOD of the Aβ_35hsl is lower than the LOD of Aβ₄₀ and Aβ₄₂. Most importantly, the quantification of Aβ_35hsl represents all C-terminal Aβ peptide variants in a single measurement. Results suggest that C-terminal variants other than Aβ₄₀ and Aβ₄₂ make a quantitatively negligible contribution to the accumulated Aβ peptide in AD brain.

One possible limitation with the use of CNBr is that both AβPP and C99 (the product of a single cleavage event by the β-secretase) are also susceptible to CNBr cleavage, and there is a Met residue in AβPP immediately preceding the Asp residue that becomes the N-terminus of Aβ_40/42. Therefore, CNBr treatment could produce Aβ_35hsl from AβPP and C99, as well as from Aβ₄₀ and Aβ₄₂. A similar issue might exist with ELISAs where both AβPP and C99 interact with common Aβ antibodies (e.g., 6E10 and 4G8). However, neither AβPP nor C99 were detected in western blot analysis of material immunoprecipitated with these antibodies, making it unlikely that they contribute significantly to the Aβ_35hsl signal in this analysis. In addition, Fig. 3 shows that the concentrations of Aβ_35hsl are similar to the sum of Aβ₄₀ and Aβ₄₂ concentrations in the brain, indicating that these two peptides are the only two significant sources of Aβ_35hsl in our analyses.

In contrast to C-terminal variants, N-terminal variants may make a quantitatively significant contribution to the accumulated Aβ peptide in AD brain. There were substantial signals corresponding to Aβ_p3-35hsl (Fig. 4), as well as Aβ_2-40, Aβ_3-42, and Aβ_4-42 (Fig. 5). Pyroglutamate modification is the most commonly reported N-terminal modification [2 , 23–26], and it is associated with increased lipid peroxidation, increased membrane permeabilization, and calcium influx in neurons [27], increased peptide aggregation [28], and accelerated neuronal loss in the CA1 pyramidal layer of young transgenic mice [15 , 30]. We observed pyroglutamate modification of Glu3 in all 4 of the AD brains tested, with strong signals ranging from 10–40% of the signal from Aβ_35hsl. The large number of N-terminal variants that have been described precluded the inclusion of all potentially important N-terminal variants.

Post translational modification of Aβ peptides can alter their propensity to form fibril and oligomer formation in vitro, and to exert cytotoxicity [19]. In particular, peptide glycation has been implicated in AD, and advanced glycation end products have been reported in close proximity to amyloid plaques in human brains [20]. We detected hexose-modified Aβ peptides in 2 out of 4 brains tested, with the modification located mainly on Lys16 or Lys28, but with small amounts on Arg5. There was no information available about whether or not these patients had diabetes.

In summary, a mass spectrometric approach to determining the absolute concentrations of Aβ peptides in brain tissue is described. This approach unambiguously identifies peptide variants, and controls for losses during sample preparation. CNBr treatment facilitates the collective determination of C-terminal variants and aggregation-prone peptides, as well as posttranslational modifications.

Footnotes

ACKNOWLEDGMENTS

This work was supported by grants from the NIH-NIA, and the Alzheimer’s Association. Proteomic analysis was performed in the Quantitative Proteomics Resource Core in Perelman School of Medicine, University of Pennsylvania. Human brain samples were obtained from the Center for Neurodegenerative Disease Research Brain Tissue Bank.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Wildburger

, Esparza

, LeDuc

, Fellers

, Thomas

, Cairns

, Kelleher

, Bateman

, Brody

(2017) Diversity of amyloid-beta proteoforms in the Alzheimer’s disease brain. Sci Rep 7, 9520.

Shinohara

, Koga

, Konno

, Nix

, Shinohara

, Aoki

, Das

, Parisi

, Petersen

, Rosenberry

, Dickson

, Bu

(2017) Distinct spatiotemporal accumulation of N-truncated and full-length amyloid-beta(42) in Alzheimer’s disease. Brain 140, 3301–3316.

Moore

, Chakrabarty

, Levites

, Kukar

, Baine

, Moroni

, Ladd

, Das

, Dickson

, Golde

(2012) Overlapping profiles of A beta peptides in the Alzheimer’s disease and pathological aging brains. Alzheimers Res Ther 4, 18–32.

Nakamura

, Kaneko

, Villemagne

, Kato

, Doecke

, Dore

, Fowler

, Li

, Martins

, Rowe

, Tomita

, Matsuzaki

, Ishii

, Arahata

, Iwamoto

, Ito

, Tanaka

, Masters

, Yanagisawa

(2018) High performance plasma amyloid-beta biomarkers for Alzheimer’s disease. Nature 554, 249–254.

Kelley

, Perry

, Bach

SBH

(2018) Characterization of proteins present in isolated senile plaques from Alzheimer’s diseased brains by MALDI-TOF MS with MS/MS. ACS Chem Neurosci 9, 708–714.

Allue

, Sarasa

, Izco

, Perez-Grijalba

, Fandos

, Pascual-Lucas

, Ogueta

, Pesini

, Sarasa

(2016) Outstanding phenotypic differences in the profile of amyloid-beta between Tg2576 and APPswe/PS1dE9 transgenic mouse models of Alzheimer’s disease. J Alzheimers Dis 53, 773–785.

Portelius

, Bogdanovic

, Gustavsson

, Volkmann

, Brinkmalm

, Zetterberg

, Winblad

, Blennow

(2010) Mass spectrometric characterization of brain amyloid beta isoform signatures in familial and sporadic Alzheimer’s disease. Acta Neuropathol 120, 185–193.

Pannee

, Portelius

, Oppermann

, Atkins

, Hornshaw

, Zegers

, Hojrup

, Minthon

, Hansson

, Zetterberg

, Blennow

, Gobom

(2013) A selected reaction monitoring (SRM)-based method for absolute quantification of Aβ(38), Aβ(40), and Aβ(42) in cerebrospinal fluid of Alzheimer’s disease patients and healthy controls. J Alzheimers Dis 33, 1021–1032.

Pannee

, Gobom

, Shaw

, Korecka

, Chambers

, Lame

, Jenkins

, Mylott

, Carrillo

, Zegers

, Zetterberg

, Blennow

, Portelius

(2016) Round robin test on quantification of amyloid-beta 1-42 in cerebrospinal fluid by mass spectrometry. Alzheimers Dement 12, 55–59.

10.

Mawuenyega

, Kasten

, Sigurdson

, Bateman

(2013) Amyloid-beta isoform metabolism quantitation by stable isotope-labeled kinetics. Anal Biochem 440, 56–62.

11.

Ovod

, Ramsey

, Mawuenyega

, Bollinger

, Hicks

, Schneider

, Sullivan

, Paumier

, Holtzman

, Morris

, Benzinger

, Fagan

, Patterson

, Bateman

(2017) Amyloid beta concentrations and stable isotope labeling kinetics of human plasma specific to central nervous system amyloidosis. Alzheimers Dement 13, 841–849.

12.

Kaneko

, Nakamura

, Washimi

, Kato

, Sakurai

, Arahata

, Bundo

, Takeda

, Niida

, Ito

, Toba

, Tanaka

, Yanagisawa

(2014) Novel plasma biomarker surrogating cerebral amyloid deposition. Proc Jpn Acad Ser B Phys Biol Sci 90, 353–364.

13.

Lesne

, Koh

, Kotilinek

, Kayed

, Glabe

, Yang

, Gallagher

, Ashe

(2006) A specific amyloid-beta protein assembly in the brain impairs memory. Nature 440, 352–357.

14.

Tamaoka

, Kondo

, Odaka

, Sahara

, Sawamura

, Ozawa

, Suzuki

, Shoji

, Mori

(1994) Biochemical-evidence for the long-tail form (A-Beta-1-42-43) of amyloid-beta protein as a seed molecule in cerebral deposits of Alzheimer’s disease. Biochem Biophys Res Commun 205, 834–842.

15.

Pivtoraiko

, Abrahamson

, Leurgans

, DeKosky

, Mufson

, Ikonomovic

(2015) Cortical pyroglutamate amyloid-beta levels and cognitive decline in Alzheimer’s disease. Neurobiol Aging 36, 12–19.

16.

Izco

, Pesini

, Perez-Grijalba

, Fandos

, Sarasa

(2013) Optimized protocol for amyloid-β extraction from the brain. J Alzheimers Dis 34, 835–839.

17.

Ikonomovic

, Wecker

, Abrahamson

, Wuu

, Counts

, Ginsberg

, Mufson

, DeKosky

(2009) Cortical alpha 7 nicotinic acetylcholine receptor and beta-amyloid levels in early Alzheimer disease. Arch Neurol 66, 646–651.

18.

Paravastu

, Qahwash

, Leapman

, Meredith

, Tycko

(2009) Seeded growth of beta-amyloid fibrils from Alzheimer’s brain-derived fibrils produces a distinct fibril structure. Proc Natl Acad Sci U S A 106, 7443–7448.

19.

Barykin

, Mitkevich

, Kozin

, Makarov

(2017) Amyloid beta modification: A key to the sporadic Alzheimer’s disease? Front Genet 8, 58.

20.

Sasaki

, Fukatsu

, Tsuzuki

, Hayashi

, Yoshida

, Fujii

, Koike

, Wakayama

, Yanagihara

, Garruto

, Amano

, Makita

(1998) Advanced glycation end products in Alzheimer’s disease and other neurodegenerative diseases. Am J Pathol 153, 1149–1155.

21.

Prior

, Monning

, Schreitergasser

, Weidemann

, Blennow

, Gottfries

, Masters

, Beyreuther

(1991) Quantitative changes in the amyloid beta-A4 precursor protein in Alzheimer cerebrospinal-fluid. Neurosci Lett 124, 69–73.

22.

Teunissen

, Chiu

, Yang

, Scheltens

, Zetterberg

, Blennow

(2018) Plasma amyloid-β (Aβ42) correlates with cerebrospinal fluid Aβ42 in Alzheimer’s disease. J Alzheimers Dis 62, 1857–1863.

23.

Mori

, Takio

, Ogawara

, Selkoe

(1992) Mass spectrometry of purified amyloid beta protein in Alzheimer’s disease. J Biol Chem 267, 17082–17086.

24.

Iwatsubo

, Saido

, Mann

, Lee

VMY

, Trojanowski

(1996) Full-length amyloid-beta (1-42(43)) and amino-terminally modified and truncated amyloid-beta 42(43) deposit in diffuse plaques. Am J Pathol 149, 1823–1830.

25.

Harigaya

, Saido

, Eckman

, Prada

, Shoji

, Younkin

(2000) Amyloid beta protein starting pyroglutamate at position 3 is a major component of the amyloid deposits in the Alzheimer’s disease brain. Biochem Biophys Res Commun 276, 422–427.

26.

Jawhar

, Wirths

, Bayer

(2011) Pyroglutamate Amyloid-β (Aβ): A hatchet man in Alzheimer disease. J Biol Chem 286, 38825–38832.

27.

Gunn

, Wong

, Johanssen

, Griffith

, Masters

, Bush

, Barnham

, Duce

, Cherny

(2016) Amyloid-β peptide Ab3pE-42 induces lipid peroxidation, membrane permeabilization, and calcium influx in neurons. J Biol Chem 291, 6134–6145.

28.

Dammers

, Schwarten

, Buell

, Willbold

(2017) Pyroglutamate-modified A beta(3-42) affects aggregation kinetics of A beta(1-42) by accelerating primary and secondary pathways. Chem Sci 8, 4996–5004.

29.

Lopez-Noguerola

, Giessen

NME

, Ueberuck

, Meissner

, Pelgrim

, Adams

, Wirths

, Bouter

, Bayer

(2018) Synergistic effect on neurodegeneration by N-truncated A beta(4-42) and pyroglutamate A beta(3-42) in a mouse model of Alzheimer’s disease. Front Aging Neurosci 10, 64.

30.

Saido

, Iwatsubo

, Mann

DMA

, Shimada

, Ihara

, Kawashima

(1995) Dominant and differential deposition of distinct beta-amyloid peptide species, A-Beta(N3(Pe)), in senile plaques. Neuron 14, 457–466.