Background
The significance of proteolytic degradation products of proteins and peptides in the central nervous system (CNS) relate not only to their potential use as biomarkers of disease activity, but also to the understanding of disease mechanisms. Amyloid β (Aβ) peptides are intensely studied due to their central role in Alzheimer’s disease (AD) [
18,
19,
26]. Aβ is produced from neurons by sequential cleavage of amyloid precursor protein (APP) by the β- and γ-secretases [
5], and released into the interstitial fluid of the brain, which communicates freely with the cerebrospinal fluid (CSF). Aβ is cleared from the brain by several pathways including both extra- and intra- cellular proteolytic degradation, resulting in the production of a large number of different peptides found in brain tissue and CSF [
37]. Aβ is also cleared across the blood-brain barrier into the blood, and by CSF absorption into the circulatory or lymphatic system (for review see [
49]). While Aβ is effectively degraded in the blood, it is not clear to which extent it is degraded in the interstitial fluid or CSF [
24].
AD patients have reduced CSF levels of the 42 amino acid long peptide Aβ42 [
6,
36], but low CSF Aβ concentration has also been found in infectious or inflammatory CNS disorders, for example in patients suffering from bacterial meningitis, neuroborreliosis, multiple sclerosis and HIV with CNS engagement [
17,
27,
28,
33,
46,
50]. In bacterial meningitis, low CSF Aβ42 normalizes after proper antibiotic treatment, and it has been hypothesized that the decrease in CSF Aβ42 during the acute infection may be caused by impaired clearance of Aβ from the brain, and not related to the inflammation in itself [
46]. Another explanation may be increased activities of APP and Aβ metabolizing enzymes with inflammation.
In the current work, we explore for the first time the use of proteolytic 18O labeling and mass spectrometry (MS) to detect endogenous proteolytic activity in human CSF. CSF was collected from patients with bacterial meningitis before and after successful treatment with antibiotics. The samples were incubated with 18O-enriched water and subjected to immunoaffinity purification of Aβ followed by MS to measure the degree of incorporation of 18O into proteolytically produced Aβ peptides, enabling determination of the relative amount of peptides formed by endogenous proteolytic activity in the CSF. Using this approach, we identified products of a specific Aβ-degrading activity in CSF from meningitis patients in the acute phase. This activity, likely derived from blood cell components in the CSF, was identified to arise from insulin-degrading enzyme (IDE), possibly representing a major degradation pathway of Aβ. We also show that this proteolytic activity can be blocked by a recombinant version of the mid-domain anti-Aβ antibody solanezumab that has an epitope covering the discovered cleavage site on Aβ.
Methods
Study subjects
CSF was provided from the Clinical Neurochemistry Laboratory, Mölndal, Sweden. The samples were surplus from the clinical routine, used after de-identification. The samples were taken from patients with neurochemical signs of meningitis including increased CSF cell counts and CSF/serum albumin ratio (see Table
1 for demographics). Blood samples were obtained from healthy volunteers. Erythrocyte concentrate (225–340 ml prepared from 400 to 450 ml whole blood) was provided by the Blood Center at the Sahlgrenska University Hospital. The study was approved by the local Institutional Review Boards and was performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki.
Table 1
Clinical characteristics in the study groups. Cell count is n/μl CSF (reference range < 4 /μl). Albumin is mg/l (reference range < 400 mg/l)
1 | Acute phase | 31 | 43 | 560 |
After treatment | 0 | 2 | 259 |
2 | Acute phase | 445 | 80 | 3132 |
After treatment | 0 | 15 | 215 |
3 | Acute phase | 364 | 80 | 2546 |
After treatment | 2 | 37 | 143 |
4 | Acute phase | 310 | 12 | 665 |
After treatment | 46 | 4 | 211 |
5 | Acute phase | 9000 | 900 | 1650 |
After treatment | missing | missing | 1710 |
6 | Acute phase | 3400 | 744 | missing |
After treatment | 17 | 60 | 361 |
Proteolytic 18O-labeling in CSF
The CSF samples (1 ml) were pH-stabilized by addition of 1 M HEPES buffer, pH 7.4 to a final concentration of 100 mM. H2
18O (Sigma) were added to the samples at a 1:1 ratio (v/v), and the samples were incubated for 24 h at room temperature. The samples were stored at −80 °C pending analysis.
Aβ immunoaffinity purification
Aβ peptides were immunoprecipitated using Aβ-specific antibodies coupled to magnetic beads as described previously [
38]. Briefly, 4 μg of the anti-Aβ antibodies 6E10 and 4G8 (Signet Laboratories, Dedham, MA, USA) were separately added to 50 μL each of magnetic Dynabeads M-280 Sheep Anti-Mouse IgG (Invitrogen, Carlsbad, CA, USA). The 6E10 and 4G8 antibody-coated beads were mixed and added to the CSF samples to which 0.025% Tween20 in phosphate-buffered saline (pH 7.4) had been added. After washing, the Aβ peptides were eluted using 100 μL 0.5% formic acid.
Mass spectrometry
Mass spectrometry was performed using a matrix-assisted-laser-desorption/ionizationtime-of-flight/time-of-flight (MALDI TOF/TOF) instrument (UltraFleXtreme, Bruker Daltonics, Bremen, Germany). Samples were prepared as described previously [
38].
Test of Aβ degrading activity in CSF containing trace amounts of blood
The isotope labeled Aβ peptide Aβ1-40 Arg13C15N (Anaspec, Inc., San Jose, USA) was dissolved in dimethylsulfoxide (DMSO, Aldrich) to a concentration of 1 mg/mL. The stock solution was aliquoted and immediately stored at −20 °C. Before use the labeled peptide was diluted and mixed to a final concentration of 0.8 μM. A 10-μL aliquot of the peptide solution was added to 930 μL CSF to which blood was added at different final concentrations (0, 0.05, 0.5, 5% (v/v)) followed by incubation in room temperature overnight at room temperature. IP-MS was conducted as described above.
In vitro test of Aβ1-40 degradation by IDE
An aliquot (10 μL) of a 0.8 μM solution of the isotopic labeled Aβ1-40 (Arg13C15N) was added to 980 μL CSF and incubated 5 min in room temperature. Recombinant human IDE (R&D systems) was diluted in 50 mM ammonium bicabonate to 23.6 ng/mL and 5.9 ng/mL. 10 μL of each solution was added to two separate pre-incubated CSF samples and left over night in room temperature. Immunoprecipitation and mass spectrometry were conducted as described above.
Inhibition of IDE activity by insulin
The inhibition of IDE activity was addressed by adding isotopic labeled Aβ1-40 (Arg13C15N, end concentration 10 ng/mL) to CSF followed by addition of recombinant human IDE (0.1, 1 or 10 μM, Sigma) and 0.5% human blood. The samples were incubated overnight at room temperature. Immunoprecipitation and mass spectrometry were conducted as described above.
Identification of Aβ degrading activity in leukocytes, thrombocytes, and erythrocytes
Leukocytes and thrombocytes were isolated from 10 mL blood as described previously [
13,
35]. The cells were resuspended in 500 μL ultra-pure water, lysed by four freeze/thaw cycles and finally centrifuged 10 min (+4 °C, 31,000 x g). Erythrocytes were prepared using a Reveos automated blood component extractor.
To 100 μL ammonium bicarbonate, spiked with 5 pmol/μL isotopic labeled Aβ1-40, 5 μL leukocytes, thrombocytes, or erythrocytes were added, followed by incubation overnight at room temperature. Immunoprecipitation and mass spectrometry were conducted as described above.
The effect of antibodies on Aβ degradation
To 1 ml PBS containing isotopic labeled Aβ1-40 (0.016 μM) 4 μg of each of the recombinant antibodies solanezumab, bapineuzumab and crenezumab were added [
8,
53], followed by incubation for 1 h at room temperature. The antibodies were expressed and purified as previously described [
53]. Each sample was spiked with serum (5%, end concentration) followed by incubation overnight at room temperature. The samples were placed in a 100 °C heating block for five minutes to disrupt potential antibody/Aβ complexes, and subsequently centrifuged for 10 min at 4 °C (31.000 x g). The supernatants were transferred to new tubes and Aβ peptides were immunoprecipitated and analyzed by mass spectrometry as described above.
Isotope distribution calculations
Theoretical peptide isotope distributions with and without incorporation of one
18O atom at 50% abundance were calculated using the software Isotope Distribution Calculator v 0.3 [
21]. The fractional abundance of Aβ peptides, produced by proteolysis during the incubation period, was calculated by linear regression analysis according to Mirgorodskaya et al. [
31], based on the observed isotope distributions of the detected peptides.
Discussion
We report on an enzymatic
18O labeling-mass spectrometry assay for determination of the relative amount of the peptides formed by endogenous proteolytic activity.
18O-labeling has previously been used for determination of proteins’ C-termini [
41], for the identification of cross-linked peptides in proteolytic peptide mixtures [
1], to assist the interpretation of fragment ion mass spectra [
43], for generating calibrators for mass spectrometric protein quantification [
31], and for differential analysis of protein mixtures [
2,
30,
52,
55]. In a recent paper,
18O-labeling was used to monitor proteolytic degradation during sample preparation of mouse brain [
48]. Here we show for the first time that
18O-labeling can also be used to detect endogenous proteolytic activity associated with disease. While the current study focuses on the processing of Aβ, the method may be applied to detect proteolytic processing events leading to the formation of any peptide and could thus be a universal tool to assess the integrity of CSF peptides over time; an important aspect in the development of biomarkers.
In bacterial meningitis, activated neutrophils are activated and released. Using
18O-labeling, we found that proteolytic activity in the CSF from patients with bacterial meningitis in the acute phase degrades full-length Aβ peptides into shorter forms, with Aβ1-19, Aβ1-20 and Aβ1-24 being prominent cleavage products. In CSF samples from bacterial meningitis patients taken after antibiotic treatment, the activity was abrogated, also concurring previous data showing that Aβ1-42 in CSF is normally stable over time, even at room temperature [
4].
Spiking CSF and solutions containing Aβ with blood produced the similar Aβ fragments as in CSF from patients with bacterial meningitis, suggesting a blood component as the likely source of the observed activity. Similar activity was observed when Aβ1-40 was incubated with serum. The lesser activity observed in plasma likely reflects the presence of EDTA in the collection tubes, which previously has been shown to inhibit IDE proteolytic activity [
25]. Further spiking experiments showed that the proteolytic activity resided in leukocytes as well as erythrocytes. IDE activity has previously been associated with leukocytes [
42] and erythrocytes [
44]. In-vitro experiments have shown that IDE can cleave Aβ between position 19 and 20, and 20 and 21 in the amino acid sequence [
29], suggesting it as a possible mediator of the observed activity. We also detected prominent Aβ1-19 and Aβ1-20 signals in CSF samples spiked with IDE (Fig.
5). There are some differences between the results obtained with PBS spiked with blood. For example, a prominent Aβ1-28 signal is present in CSF incubated with IDE but absent in Aβ1-40 incubated with blood. This may be attributed to other proteolytic activities in blood that further degrade Aβ1-28 to shorter Aβ species. Similarly, Aβ1-24 in an Aβ1-40 sample spiked with blood may be the product of other blood-derived proteases.
IDE is a zinc-metallopeptidase which has been implicated in several prevalent diseases including Type 2 diabetes mellitus and AD [
12]. IDE has been found to degrade and influence brain Aβ in experimental animal models [
3,
23,
32,
54] and it has also been shown to be present in serum [
25] and CSF [
40]. In the present experiments, no formation of Aβ1-19 or Aβ1-20 was observed in CSF samples from patients taken after antibiotics treatment with normalized leukocyte counts. Furthermore, it has previously been shown that Aβ1-42 in CSF is normally stable over time, even at room temperature [
4].
Our results indicate that cleavages at positions Aβ19 and Aβ20 represents a significant pathway for physiological Aβ degradation in blood, and that these cleavages may be mediated by IDE. Since IDE-deficient mice show accumulation of Aβ protein [
15], hyperinsulinemia and glucose intolerance enhancing or emulating the activity of IDE may lower the Aβ burden and may be of interest to pursue as an AD-treatment.
The dramatic increase observed in plasma Aβ, in response to treatment with Aβ-specific monoclonal antibodies [
14,
45] has been attributed to increased clearance of soluble Aβ from the brain into the periphery. According to the peripheral sink hypothesis, the therapeutic antibody changes the equilibrium for Aβ between the brain and plasma, with increased clearance from the brain and a resulting increase in plasma Aβ levels [
11]. However, studies in mouse models, in which the peripheral level of Aβ was decreased by either by reducing production or increasing degradation of Aβ in the periphery, failed to show any effect on Aβ levels in the brain [
16,
20,
51]. An alternative hypothesis is that some anti-Aβ antibodies cause a prolonged half-life of Aβ peptides in the bloodstream by blocking protease cleavage sites [
7].
Here we show that treatment with a recombinant version of solanezumab protects Aβ from degradation in serum. In contrast, bapineuzumab or crenezumab did not affect the Aβ peptide profiles, with the exception that Aβ1-13 seems to increase. This difference may be due to the much lower affinity of crenezumab [
10] and bapineuzumab than solanezumab for Aβ. The pathophysiological relevance of the observations needs to be determined in future studies.
Conclusions
In the present study, we show that a technique based on enzymatic 18O labeling-mass spectrometry is useful for identifying and determining the relative amount of peptides formed by endogenous proteolytic activity in human CSF. Using this technique we found an enzymatic activity in blood leukocytes and erythrocytes that was identified as IDE that cleaves Aβ in the mid-domain of the peptide, and could be inhibited by a recombinant version of the mid-domain anti-Aβ antibody solanezumab. If, as these results suggest, the increase in plasma Aβ upon treatment with some Aβ-specific antibodies is caused by blocking the protease cleavage site, that is that the higher affinity antibody (solanezumab) reduces IDE processing of the peptide, it should be considered that therapeutic antibodies may in fact interfere with Aβ clearance by stabilizing Aβ peptides.
Acknowledgements
The authors would like to thank Anni Westerlund and Kristin Augutis for technical assistance.