Transferrin Biosynthesized in the Brain Is a Novel Biomarker for Alzheimer's Disease

Abstract

Glycosylation is a cell type-specific post-translational modification that can be used for biomarker identification in various diseases. Aim of this study is to explore glycan-biomarkers on transferrin (Tf) for Alzheimer's disease (AD) in cerebrospinal fluid (CSF). Glycan structures of CSF Tf were analyzed by ultra-performance liquid chromatography followed by mass spectrometry. We found that a unique mannosylated-glycan is carried by a Tf isoform in CSF (Man-Tf). The cerebral cortex contained Man-Tf as a major isofom, suggesting that CSF Man-Tf is, at least partly, derived from the cortex. Man-Tf levels were analyzed in CSF of patients with neurological diseases. Concentrations of Man-Tf were significantly increased in AD and mild cognitive impairment (MCI) comparing with other neurological diseases, and the levels correlated well with those of phosphorylated-tau (p-tau), a representative AD marker. Consistent with the observation, p-tau and Tf were co-expressed in hippocampal neurons of AD, leading to the notion that a combined p-tau and Man-Tf measure could be a biomarker for AD. Indeed, levels of p-tau x Man-Tf showed high diagnostic accuracy for MCI and AD; 84% sensitivities and 90% specificities for MCI and 94% sensitivities and 89% specificities for AD. Thus Man-Tf could be a new biomarker for AD.

Keywords: Alzheimer’s disease; biomarker; cerebrospinal fluid; glycan-isoforms; transferrin.

Figure 1

N-glycan structures of transferrin in human cerebrospinal fluid. Monosaccharide symbols are according to Consortium for Functional Glycomics (www.functionalglycomics.org/) (*1). Minor components (<0.6%) are not listed (*2). Glycan structure is the same as that of serum transferrin (*3). Position(s) of additional sugar(s) out of left curly brancket is not identified (*4).

Figure 2

Human CSF was subjected to immunoprecipitation using anti-Tf antibody. Immunoprecipitated fraction (ppt fr) and supernatant (sup) were subjected to SDS-PAGE (A) followed by immuno- (B) and lectin-blotting (C). Solid and dotted lines indicate migration positions of non-Sia-Tf and Sia-Tf, respectively. Blots are probed with anti-Tf antibody (αTf-Ab) (B) or UDA lectin (C). Migration positions of immunogloblin heavy and light chains are indicated with closed and open triangles, respectively.

Figure 3

Brain CT image of a hydranencephaly patient (A). Control CSF and CSF-like fluid of hydranencephaly patients (Case 1 and 2) were subjected to SDS-PAGE (B) followed by immuno- (C) and lectin-blotting (D,E). Migration positions of serum Tf, albumin (Alb), immunoglobulin heavy (Ig-H) and light chains (Ig-L) are indicated with open triangles (B). Serum specimen is applied to a separated lane on the same SDS-PAGE gel. Solid and dotted lines indicate migration positions of non-Sia-Tf and Sia-Tf, respectively. Blots are probed with anti-Tf antibody (αTf-Ab) (C), BC2L-A (D) and UDA (E). Figure 3A is cited from our previous paper, J. Biochem., 2018, 164, 205–13 [17].

Figure 4

Tf mRNA expression was examined by in situ hybridization. The lateral lobe sections were hybridized with anti-sense (A) and sense (B) probes, and then visualized with a mixture of 4-nitroblue tetrazolium chloride and 5-bromo-4-chloro-3-indoyl-phosphate mixture (NBT/BCIP). In a high-power field, Tf mRNA-positive cells are indicated with red triangles (C1). The corresponding area on the mirror image section was stained with anti-NeuN antibody (αNeuN-Ab), and the antigens were visualized with the Immpact DAB Substrate (DAB). NeuN-positive neuron-like cells are indicated with blue triangles (C2). Tf protein expression was examined by immunohistochemistry using anti-Tf antibody (αTf-Ab). In a high-power field of the lateral cortex, many neuron-like cells are stained by anti-NeuN antibody, visualizing with NBT/BCIP (D1, blue triangles and arrows). The corresponding area on the mirror image section is stained by anti-Tf antibody, visualizing with DAB. Tf-positive cells are indicated with red triangles, while NeuN-positive, but Tf-negative, cells are indicated with arrows (D2, right panel). The cellular nuclei are stained with methyl green.

Figure 5

Tf was purified from detergent extracts of human occipital cortex by immunoaffinity column chromatography and subjected to SDS-PAGE (A). Glycans were liberated from Tf by in-gel digestion with PNGase F. Liberated glycans are reduced, permethylated and then subjected to matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MS) (B). Secondary fragment ions derived from the ion at m/z 1595.8 were analyzed on tandem mass spectrometry (MS/MS) (C).

Figure 6

Tf glycan isoforms were quantified by anti-Tf antibody/rBC2L-A lectin-sandwich ELISA. The isoforms are captured on a microtiter plate by using anti-Tf antibody (Tf-Ab) (A). The mannose-terminated isoform (Man’-Tf) is quantified with rBC2L-A lectin (B). No significant signal is detected with Sia-Tf and GlcNAc’-Tf.

Figure 7

Man-Tf levels are quantified in CSF of neurological disease: iNPH, idiopathic normal pressure hydrocephalus; MCI, mild cognitive impairment; AD, Alzheimer’s disease; PSP, progressive supranuclear palsy; FTD, frontotemporal degeneration; PD, Parkinson’s disease; DLB, dementia with Lewy bodies. Man-Tf data show normal distribution in iNPH, AD, PSP+FTD groups, but not in others. Multiple comparisons were assessed by Kruskal-Wallis method followed by Bonferroni correction. Significant differences are indicated with p values.

Figure 8

Correlation coefficients for p-tau and Man-Tf; MCI+AD (A), MCI (B) and AD (C).

Figure 9

Human hippocampal sections from control (normal; A,B) and AD (C,D) brains was co-stained with anti-Tf antibody (αTf-Ab; red) and anti-p-tau antibody (αp-tau-Ab; green). Images were scanned by NanoZoomer and analyzed by NDP.view2 Plus Image viewing software U12388-01.

Figure 10

Levels of p-tau (A) and p-tau x Man-Tf (B) are indicated. P-tau data show normal distribution in iNPH group, but not in others. Data of p-tau x Man-Tf do not show normal distribution in all groups. Multiple comparisons are assessed by Kruskal-Wallis method followed by Bonferroni correction. Sensitivities, specificities, AUCs, and cutoff values of markers are indicated (C).