Background
The neuropathological process of Alzheimer’s disease (AD) is characterized by accumulation of plaques composed of aggregated amyloid-β (Aβ) protein, neurofibrillary tangles consisting of hyperphosphorylated tau, and neuronal degeneration and loss [
1]. The cerebrospinal fluid (CSF) biomarkers of these processes—amyloid-β 1–42 (Aβ
42), total tau, and tau phosphorylated at threonine 181 (p-tau)—show very consistent changes in AD dementia and prodromal AD [
2], and they have been included as evidence for the presence of AD pathology in research diagnostic criteria for AD [
3,
4]. However, in addition to amyloid and tau pathology, processes such as inflammation and synaptic dysfunction also play an important role and may correlate more directly with cognitive decline [
5‐
7]. Hence, biomarkers for these processes may be valuable for disease monitoring and to predict prognosis or rate of cognitive decline.
On the basis of in-house data derived from unbiased CSF proteomic studies, we selected a panel of synaptic and other proteins for analysis with parallel reaction monitoring (PRM) MS. Explorative proteomics provides a hypothesis-free approach for detecting a large number of proteins and peptides in human body fluids such as plasma and CSF [
8]; however, the disadvantage of this technique is the relatively poor reproducibility [
9]. PRM is an MS method that focuses on predefined sets of proteins, or peptide peaks from proteolytically digested proteins, thereby allowing higher analytical sensitivity and higher throughput of samples [
10,
11]. By adding known amounts of isotope-labeled internal standards to the sample, quantification of the actual concentration can be achieved [
11]. An advantage over immunoassays is that there is no need for antibodies. Hence, PRM is well suited for validation of explorative proteomic studies [
9].
The selection of proteins for the present study was based on a literature review for their possible involvement in AD pathophysiology. The PRM panel consisted of neurosecretory protein VGF (VGF), chromogranin A (CHGA), and secretogranin 2 (SCG2), granins that are presumed to be involved in axonal or synaptic vesicle transport (the granins VGF, CHGA, and SCG2) [
12]; cystatin C (CysC), a protease involved in Aβ degradation [
13,
14]; β
2-microglobulin (β
2M) and lysozyme C (LysC), proteins involved in the innate immune system [
15,
16]; and neurexins (NRXNs) NRXN-1, NRXN-2, and NRXN-3 as well as neuronal pentraxin 1 (NPTX1), neurofascin (NFASC), and neurocan core protein (NCANP), proteins involved in synapse formation and stabilization [
17‐
21]. Several of these proteins, including VGF, CHGA, SCG2, CysC, and β
2M, have been suggested in previous studies to be involved in AD pathology [
12‐
14,
22‐
28]. A pilot study that we performed after developing the PRM panel showed promising results, with lower levels for several proteins in patients with AD dementia compared with control subjects [
29]. The aims of the present study were to validate this panel of proteins in a larger and independent cohort, as well as to investigate whether the panel has potential for distinguishing patients with prodromal AD and patients with AD dementia from cognitively healthy subjects.
Discussion
In this study, we aimed to validate a PRM-MS assay consisting of a panel of proteins presumed to be involved in processes regarding secretory vesicle functioning, synaptic functioning, and innate immunity [
12‐
21]. Having performed a pilot study [
29], in the present study, we investigated this panel in a larger and independent cohort that included subjects with MCI in addition to patients with AD dementia and control subjects.
Contrary to our expectations, levels of most proteins were higher in patients with MCI than in the other groups, a finding attributable mainly to those patients who at follow-up had progressed to AD dementia. Stratified analyses showed that these differences were driven mainly by proteins involved in secretory vesicle function (SCG2, CHGA) or synapse formation and stabilization (NRXNs, NPTX1, NCANP), whereas CysC (a cysteine protease inhibitor) was similar between groups and LysC and β
2M (both involved in the immune system) showed only small, borderline significant differences. In addition, we found that correlations of most PRM peptides with Aβ
42 differed markedly between groups, with positive correlations in control subjects and subjects with sMCI but negative correlations in subjects with MCI-AD. In contrast to the findings of our pilot study [
29], there were no differences between patients with AD dementia and control subjects, except for VGF. Importantly, however, the proteins showing differences between groups in the present study were the same proteins as in the pilot study. Hence, the difference between the two studies may be due to differences in the selection of patients or control subjects.
The results regarding VGF in the present study are in line with our pilot study, with clearly lower values in AD dementia than in the other groups. In the biomarker panel in the present study, VGF has been studied most extensively in AD research. Lower levels in AD dementia than in control subjects have been described before [
13,
24,
28,
36]. In experiments with hippocampal neurons from rats, VGF peptides increased synaptic plasticity and enhanced proliferation of cells [
37,
38]. Moreover, in another study, VGF-knockout mice showed memory impairment [
39]. Our findings of lower levels in patients with AD dementia are hence concordant with these previous studies.
To our knowledge, increased levels of synaptic proteins in patients with MCI due to AD have not been described before. However, on one hand, a recent study showed that higher levels of CHGA in CSF predict future decreases in Aβ
42 in healthy elderly subjects [
25]. On the other hand, a decreased level in brain tissue and CSF of patients with AD compared with control subjects has also been described [
27,
28,
40]. Studies on the role of NRXNs, NPTX1, NFASC, and NCANP in AD are rare, although the proteins have all been suggested to have a role in synapse formation, plasticity, and stability [
17‐
21]. On the basis of using MS-based proteomics, CSF NRXN-1 has been found to be slightly decreased in patients with AD compared with control subjects in a small cohort of 16 subjects [
36]. A recent study in mice suggested that NRXN-2 interacts with Aβ oligomers, resulting in loss of synapses, whereas blocking of this interaction prevented Aβ-induced memory impairment [
41]. NPTX1 has been found to increase in cell cultures after treatment with Aβ, to be present in dystrophic neurites and around Aβ plaques in AD brains, and to modulate synaptic transmission [
42,
43]. A recent study showed that NCANP expression in astrocytes was increased after incubation with Aβ [
44]. It has been established that synaptic damage and loss are early events in AD that are already evident in the MCI stage [
5]. Moreover, studies in the early 1990s showed that although the number of synapses declined in early stages of AD, this was first compensated by an increase in synapse size of the remaining synapses and only in later stages was overcome by further synapse loss [
45,
46]. Our finding that the correlations between PRM peptides and Aβ
42 differed between disease stages also suggests a dynamic pathophysiological trajectory with progression of the disease. The positive correlations in control subjects and subjects with stable MCI suggest an association with normal aging, whereas this association is reversed in cases of AD pathology. It can be hypothesized that a compensatory mechanism in patients in an early stage of AD pathology results in upregulation of proteins involved in synaptic plasticity and stabilization, whereas concentrations decline in later stages, associated with further synapse loss. Studies finding increased connectivity using functional MRI in amyloid-positive elderly subjects may also point in this direction [
47,
48]. Because our AD dementia group consisted of patients still in a relatively early stage of the disease, concentrations at the time of measurement may have been between this elevated level and a subsequent decrease. This was a cross-sectional study, however, and longitudinal studies with extensive clinical follow-up and repeated lumbar punctures are needed to assess changes of these proteins within individuals over the course of the disease. Subjects in the earliest disease stages (SCD and MCI at baseline) will be most interesting to investigate for further validation of the present study.
Among the strengths of our study are the relatively large number of patients for this type of experiment and the fact that the cohort was matched for sex and age. The small size of the group with MCI due to AD was a limitation, however, and the results have to be replicated in independent cohorts to be able to draw definitive conclusions. The use of patients with subjective cognitive decline as control subjects in our study is a possible limitation. The advantage, however, is that these subjects received the same cognitive screening as the other patients, ruling out other diagnoses with as much certainty as possible. In addition, the aim was to find CSF biomarkers useful for routine clinical practice; therefore, using perfectly healthy control subjects instead of subjects with SCD might overestimate differences. Strengths of the PRM technique are that it allows for investigation of proteins for which there are no antibodies available and that the specificity is very high. This is important in the search for new biomarkers and for verification of shotgun proteomic findings. A disadvantage, however, is the time needed for analysis, which makes high throughput of a large volume of samples, as would be needed in daily practice, unfeasible.
Acknowledgements
This study was supported by a grant From Alzheimer Nederland (WE.15-2013-08), the Swedish Research Council, Åhlén-stiftelsen, the European Research Council, Swedish State Support for Clinical Research (ALFGBG), the Swedish Brain Foundation, and the Swedish Alzheimer Foundation, and it is part of the BIOMARKAPD Project within the framework of the European Joint Programme – Neurodegenerative Disease Research (JPND), the Knut and Alice Wallenberg Foundation, the Torsten Söderberg Foundation, Frimurarestiftelsen, Stiftelsen för Gamla Tjänarinnor, Gun and Bertil Stohnes Stiftelse, and Magnus Bergvalls Stiftelse. Research done at the VUmc Alzheimer Center is part of the neurodegeneration research program of Amsterdam Neuroscience. The VUmc Alzheimer Center Amsterdam is supported by Alzheimer Nederland and Stichting VUmc fonds. The clinical database structure was developed with funding from Stichting Dioraphte.