The Distribution and Expression of Picalm in Alzheimer Disease

Abstract

PICALM, the gene encoding phosphatidylinositol-binding clathrin assembly (picalm) protein, was recently shown to be associated with risk of Alzheimer disease (AD). Picalm is a key component of clathrin-mediated endocytosis. It recruits clathrin and adaptor protein (AP)-2 to the plasma membrane and along with AP-2 recognizes target proteins. The attached clathrin triskelions cause membrane deformation around the target proteins enclosing them within clathrin-coated vesicles to be processed in lysosomes or endosomes. We examined the distribution of picalm in control and AD brain tissue and measured levels of picalm mRNA by real-time polymerase chain reaction. Immunolabeling of brain tissue showed that picalm is predominately present in endothelial cells. This was further supported by the demonstration of picalm in human cerebral microvascular cells grown in culture. Picalm mRNA was elevated in relation to glyceraldehyde-3-phosphate dehydrogenase but not factor VIII-related antigen or CD31 mRNA in the frontal cortex in AD. No change was seen in the temporal cortex or thalamus. The transport of Aβ across vessel walls and into the bloodstream is a major pathway of Aβ removal from the brain and picalm is ideally situated within endothelial cells to participate in this process. Further research is needed to determine whether PICALM expression is influenced by Aβ levels and whether it affects Aβ uptake and transport by endothelial cells.

INTRODUCTION

Two recent genome-wide association studies of Alzheimer disease (AD) demonstrated an association with PICALM variants (1, 2). PICALM was originally identified as the clathrin assembly lymphoid myeloid leukemia (CALM) gene in studies of a rare translocation present in lymphoid and myeloid acute leukemias (3). The gene product phosphatidylinositol-binding clathrin assembly (picalm) protein plays a key role in endocytosis (4), which is important in processes such as the regulation of receptors, synaptic transmission and the removal of apoptotic cells (5). Processing of the amyloid precursor protein (APP) following endocytosis suggests involvement of this process in the regulation of β-amyloid (Aβ) levels (6-9). The transport of endocytosed APP into lysosomes is reported to result in APP degradation entry into endosomes and Aβ production (9).

One mechanism of endocytosis is dependent on clathrin-coated vesicles (CCVs), which are involved in receptor-mediated endocytosis at the plasma membrane and in the intracellular movement of macromolecules (10, 11). CCVs are formed on the cytoplasmic side of the plasma membrane or at the trans-Golgi network following the attachment of clathrin triskelions. The polyhedral lattices formed by these triskelions lead to the deformation of the membrane and vesicle formation (4, 5). CCV subsequently fuse with either endosomes, in which internalized target proteins are modified or recycled back to the plasma membrane, or with lysosomes, in which case the target protein is degraded (5, 12).

The attachment of clathrin triskelions to the membrane is by means of APs, of which there are 2 types: tetrameric (AP-1 to -4) and monomeric (AP180 and picalm) (13). In addition to binding to clathrin, APs interact with membrane proteins containing specific sequences that mark them as proteins for internalization (14, 15). There are 3 forms of picalm (i.e. full length [652 residues] and 2 shorter forms [632 and 610 residues]), which is an important component of clathrin-mediated endocytosis. Its N-terminal binds to phosphatidylinositol-4,5-biphosphate present in the plasma membrane and its C-terminal to clathrin and AP-2, recruiting both to the membrane (4). Picalm is reported to be present within neurons, astrocytes and oligodendrocytes (4, 10, 11, 16), and alterations in its levels affect clathrin-mediated endocytosis. Endocytosis can be blocked by picalm overexpression (11) and by degradation of picalm and AP-2 by calpain and caspase proteases, the levels of which are increased in AD (13, 17).

Here, we examined the expression and distribution of picalm in control and AD brains. Our findings suggest that it may contribute to Aβ clearance from the brain parenchyma into the bloodstream.

MATERIALS AND METHODS

Tissue from control and AD brains was selected from the South West Dementia Brain Bank. The brains had been divided mid-sagittally at autopsy: the left half had been sliced and frozen at −80°C for biochemical studies and the right half fixed in formalin for detailed neuropathological assessment. Diagnosis of AD was made according to Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) criteria (18). According to National Institute on Aging-Reagan criteria (19), 17/19 of these cases had a high likelihood that the dementia was due to AD. The remaining 2 cases (ID numbers 23 and 29, Table 1) had an intermediate likelihood that dementia was due to AD. Aβ load and cerebral amyloid angiopathy (CAA) scores were determined as previously described (20, 21) (Table 1). CAA was graded in arteriolar vessels in the cortex and in the leptomeninges according to the Olichney method (21).

Table 1.

Profiles of Cases Studied

Case number	PM delay (hour)	Age (year)	Braak stage	Aβ load (%)*	CAA score†
Control cases
1	46	95	0	0.00	0
2	24	78	II	0.00	0
3	25	71	I	0.00	0
4	35	82	III	0.00	0
5	42	72	I	0.00	0
6	45	90	II	0.00	0
7	12	78	II	0.00	0
8	36	73	II	0.01	0
9	18	93	II	0.00	0
10	30	82	II	0.00	0
11	15	89	II	0.49	0
12	24	83	II	0.00	0
13	3	82	II	0.19	0
14	24	79	#	#	#
15	37	82	II	0.00	0
16	48	78	I	0.00	0
17	23	76	II	0.00	0
18	35	73	III	0.13	0
19	5.5	90	II	0.00	0
20	37.75	93	III	0.07	0
AD cases
21	9	78	V	1.09	0
22	42	81	VI	7.50	1
23	43	77	IV	0.79	0
24	35	78	VI	7.03	0
25	24	82	VI	0.65	2
26	25	70	VI	1.35	0
27	12	74	VI	3.37	3
28	28	79	VI	2.10	1
29	21	90	IV	1.62	2
30	20	93	VI	7.93	0
31	5	83	V	1.69	0
32	26	77	VI	3.10	3
33	24	74	V	3.33	0
34	21	78	VI	2.23	1
35	39	89	V	1.67	3
36	14	77	VI	5.65	0
37	48	83	V	0.23	0
38	48	74	V	1.53	0
39	36	87	VI	5.65	0

AD = Alzheimer disease; CAA = cerebral amyloid angiopathy; PM = post-mortem; SEM = standard error of mean.

^*Mean Aβ load (% area of cortex immunopositive for Aβ), measured as described in (20).

^†Median CAA score, graded according to the Olichney method (21), where 0 = no CAA, 1 = mild CAA, 2 = moderate CAA, 3 = severe CAA.

^#No remaining formalin-fixed tissue for Braak staging, Aβ load measurements or CAA scoring.

Western Blot

To determine the specificity of the picalm antibody, brain tissue homogenates were prepared in sodium-dodecyl sulphate buffer as previously described (22) and were run alongside GST-tagged recombinant picalm protein (AbNova, Taipei, Taiwan) on a 4% to 20% Tris-HCl precast gel (Bio-Rad, Hertfordshire, UK. 150V, 1 hour). Protein was transferred to nitrocellulose membrane (Anachem, Luton, UK. 20V, overnight [ON] at 4°C), and detected using the Millipore SNAP i.d. protein detection system (Millipore, Billerica, MA). The membrane was blocked in 0.5% milk powder in Tris-buffered saline/0.05% Tween-20, incubated with picalm (1:50, 10 minutes; Calm C-18, Santa Cruz Biotechnology, Heidelberg, Germany) and with peroxidise anti-goat (1:10,000, 10 minutes; Vector Laboratories, Peterborough, UK). Visualization of the target protein was by enhanced chemiluminescence using the Amersham ECL detection reagents (GE Healthcare, Buckinghamshire, UK).

Immunohistochemistry

Seven-μm-thick sections from formalin-fixed, paraffin-embedded blocks of frontal and temporal lobe from 7 control and 7 AD brains were immunostained for picalm. The sections of frontal lobe included the superior and middle frontal gyri (Brodmann area 6), cingulate cortex (area 24), paracentral lobule (area 32), and underlying white matter. Those of temporal lobe included the parahippocampal gyrus (area 36), and inferior (area 20), middle (area 21) and superior temporal (area 22) gyri and the hippocampus. Dewaxed and hydrated sections were immersed in methanol containing 3% H₂O₂ (30 minutes), boiled in sodium citrate buffer (pH 6), blocked in 10% normal rabbit serum (20 minutes), and incubated with anti-picalm antibody (1:800, ON; Calm C-18, Santa Cruz Biotechnology). This was followed by incubations with biotinylated anti-goat secondary antibody (1:100, 20 minutes; Vector Laboratories), with avidin-biotin horseradish peroxidase complex (20 minutes; VectaElite ABC; Vector Laboratories) and with 3,3′-diaminobenzidine (DAB, 10 minutes; Vector Laboratories). All incubations were conducted at room temperature. Sections were counterstained with hematoxylin, dehydrated, cleared and mounted.

Double Immunofluorescent Labeling of Brain Tissue

Fluorescein tyramide-labeled factor VIII-related antigen/von Willebrand factor (FVIIIRA/VWF) was combined with streptavidin Alexa Fluor 555 labeling of biotinylated picalm. Sections were initially treated as above. Following treatment with sodium citrate buffer, sections were incubated in 0.12% potassium permanganate in phosphate-buffered saline (PBS; 25 minutes) and in 1% oxalic acid and 1% potassium pyrosulphate in PBS (15 minutes) to reduce autofluorescence. Sections were washed and labeled for FVIIIRA/VWF (VWF; Abcam, Cambridge, UK) using the tyramide signal amplification (TSA) fluorescence system (Perkin Elmer, Milano, Italy). This involved initial immersion in Tris-sodium chloride buffer containing 0.5% TSA blocking reagent for 30 minutes and subsequent incubation in primary antibody (1:800; ON, room temperature), peroxidise anti-rabbit antibody (1:200; 30 min) and fluorescein-labeled tyramide (1:50; 10 minutes). This was followed by ON incubation with anti-picalm (1:100), 20-minute incubation with biotinylated anti-goat (1:100) and 1-hour incubation with streptavidin Alexa Fluor 555 (Invitrogen, Paisley, UK). Sections were mounted in Vectashield mounting medium (Vector Laboratories) and images acquired using a confocal laser-scanning microscope (Nikon Instruments, The Netherlands).

The TSA biotin system (Perkin Elmer) was used for double immunofluorescent labeling of picalm and Aβ (4G8; Covance Cambridge Bioscience, Cambridge, UK). Following treatment with sodium citrate buffer as above, sections were blocked in TSA blocking buffer (30 minutes), incubated with anti-picalm (1:800, ON), biotinylated anti-goat (30 minutes), streptavidin-HRP (30 minutes), biotinyl tyramide (10 minutes) and streptavidin Alexa Fluor 555 (1 hour). Sections were washed and incubated ON with 4G8 (1:2000) and with donkey anti-mouse Alexa Fluor 488 (1 hour).

Double Immunofluorescent Labeling of Cerebrovascular Endothelial Cells

Human brain microvascular endothelial cells (TCS Cellworks, Buckingham, UK) were grown on fibronectin-coated coverslips in endothelial cell medium supplemented with 5% fetal bovine serum and 1% growth supplement (TCS Cellworks) at 37°C in 5% CO₂. Cells were fixed in 4% paraformaldehyde (10 minutes; Alfa Aesar, Lancashire, UK), blocked and permeabilized in a solution containing 5% normal donkey serum and 0.1% triton X-100 in PBS (20 minutes) then incubated ON at 4°C with FVIIIRA/VWF (1:100) and picalm (1:50) antibodies. Cells were subsequently incubated with donkey anti-rabbit Alexa Fluor 488 (1 hour), with biotinylated anti-goat antibody (1:100, 20 minutes) and with streptavidin Alexa Fluor 555 (1 hour). Cells were mounted in Vectashield mounting medium and images acquired using a confocal laser-scanning microscope.

RNA Extraction and Reverse Transcription

RNA extraction from the frontal and temporal cortex and from the thalamus of 20 controls and 19 AD cases was performed as previously described (23). Profiles of the cases are listed in Table 1. There were no significant differences in postmortem delay (p = 0.832) or age (p = 0.433) between the 2 groups. RNA concentration was determined using a Ribogreen RNA quantification kit (Invitrogen). RNA was reverse transcribed using the High Capacity c-DNA Archive Kit (Applied Biosystems, Foster City, CA), and the concentration of the resulting cDNA determined using the Picogreen DNA quantification kit (Invitrogen). These kits were used according to the manufacturers’ instructions.

Real-Time Polymerase Chain Reaction

The expression of PICALM and the calibrator genes glyceraldehyde-3-phosphate dehydrogenase (GAPDH), FVIIIRA/VWF and CD31/platelet endothelial cell adhesion molecule were measured using the ABI 7000 sequence detection system (Applied Biosystems, Hs00200318_m1, Hs99999905_m1, Hs01109438_m1 and Hs01065279_m1 respectively). The 20-μl reaction mixture comprised SensiMix dT (Quantace, London, UK), TaqMan gene expression assay (Applied Biosystems) and 10 ng cDNA. Each sample was analyzed in triplicate using the following program settings: 50°C (2 minutes), 95°C (10 minutes), and 40 cycles of 95°C (15 seconds) and 60°C (1 minute). PICALM expression in relation to each calibrator gene was calculated (ΔCt) and used to determine the fold change in PICALM expression (2^−ΔΔCt) in each case relative to the mean expression in controls.

Statistical analysis

The Mann-Whitney U test was used to assess the significance of changes in PICALM expression between AD and control groups. P values < 0.05 were regarded as significant.

RESULTS

Western Blot Analysis and Immunolabeling

The specificity of the picalm antibody was confirmed by Western blot. The antibody recognized the GST-tagged recombinant protein (~93 kDa) and picalm in brain tissue homogenates (~62-72 kDa) (Fig. 1). The 652 and 632 amino acid forms of picalm have very similar molecular weights (~69 kDa and ~71 kDa), and could not be resolved by Western blot.

Figure 1.

The specificity of the picalm antibody was tested by Western blot. Anti-picalm recognizes GST-tagged recombinant picalm (~93 kDa) in lane 1 and picalm in brain tissue homogenates in lanes 2 and 3 (~62–72 kDa). Both full-length picalm (71/72 kDa) and the shorter isoform (62–66 kDa) are detected. The third isoform (~69 kDa) could not be separated from the 71/72-kDa full-length picalm band.

Immunolabeling of human brain tissue showed that picalm was largely restricted to blood vessel walls (Fig. 2). Picalm antibody labeled vessels throughout the grey and white matter and leptomeninges. There was only weak neuronal or glial labeling (Fig. 2A) and there was no evidence of picalm in Aβ plaques or neurofibrillary tangles.

Figure 2.

Sections of human brain tissue immunolabeled with anti-picalm. (A) Picalm is present in blood vessel walls in a section of normal frontal cortex. Neurons are only very weakly labeled (arrows). (B, C) The predominant localization of picalm in the walls of blood vessels is also evident in (A) normal temporal cortex (B) and Alzheimer disease temporal cortex (C). Magnifications: x50.

The location of picalm within the vessel walls was shown by double immunofluorescent labeling to be associated with endothelial cells: the distribution of picalm closely followed that of the endothelial cell marker FVIIIRA/VWF (Fig. 3A-F). The expression of picalm by cerebrovascular endothelial cells was confirmed by immunofluorescent labeling of human brain microvascular endothelial cells grown in culture (Fig. 3G-I). Double immunofluorescent labeling of picalm and Aβ confirmed the absence of picalm from plaques and showed good preservation of picalm even in vessels with extensive CAA (Fig. 4).

Figure 3.

Double immunofluorescent labeling shows the endothelial localization of picalm. A-F: The distribution of picalm in blood vessel walls closely follows that of the endothelial marker, factor VIII-related antigen/von Willebrand factor (FVIII) in sections of Alzheimer disease frontal cortex. G-I: Human brain microvascular endothelial cells grown in culture express picalm and confirm the endothelial localization of this protein. Magnifications: x350.

Figure 4.

(A, B) Double immunofluorescent labeling for β-amyloid (Aβ) (A) and picalm (B) demonstrates good preservation of picalm even in vessels with extensive deposition of Aβ in cerebral amyloid angiopathy. (C) The merged image demonstrates different distribution of these 2 antigens. Magnifications: x300.

Real-Time Polymerase Chain Reaction

Because picalm was found to be predominantly associated with endothelial cells, we quantified its gene expression in relation to 2 calibrator genes that encode for the endothelial cell markers FVIIIRA/VWF and CD31, and in relation to the “housekeeping” gene GAPDH. PICALM expression was significantly increased in the frontal cortex in AD in relation to all 3 calibrator genes (Fig. 5A), although the increase reached significance only in relation to GAPDH (Table 2, p = 0.047). There was no significant change in PICALM expression in AD compared to control temporal cortex or thalamus (Fig. 5B, C; Table 2).

Figure 5.

Levels of picalm mRNA in the frontal and temporal cortex and thalamus of Alzheimer disease (AD) and normal control brains was measured by RT-PCR. Fold change in PICALM expression in relation to GAPDH (●), FVIIIRA (■) and CD31 (◆) was calculated using the 2^−ΔΔCt method. Expression is significantly increased in the AD frontal cortex in relation to GAPDH (p = 0.047) (A) but remains unchanged in the AD temporal cortex (B) and thalamus (C) vs. normal controls. Graphs show geometric means and 95% confidence intervals.

Table 2.

Statistical Analysis of PICALM Expression in Alzheimer Disease and Control Groups In Relation to Calibrator Genes

Anatomic Area Analyzed	GAPDH	FVIIIRA	CD31
	(n, control/AD)	(n, control/AD)	(n, control/AD)
Frontal lobe	0.047 (20/19)	0.311 (19/19)	0.404 (20/16)
Temporal lobe	0.325 (19/19)	0.296 (20/19)	0.718 (20/16)
Thalamus	0.893 (19/18)	0.817 (19/19)	0.165 (19/17)

Data are p values (Mann Whitney U test) of comparisons of PICALM expression to glyceraldehyde-3-phosphate dehydrogenase (GAPDH), factor VIII-related antigen (FVIIIRA) and CD31. Significant value is in underlined.

DISCUSSION

Picalm plays an important role in clathrin-mediated endocytosis, which is important in a number of processes (4, 5). Previous studies in cell lines and rat brain tissue reported its presence in neurons, astrocytes and oligodendrocytes (4, 10-11, 16). In addition, both overexpression and degradation of picalm disrupt the endocytic pathway (11, 13, 17). Our present studies show that picalm in human brain tissue is largely restricted to endothelial cells, with relatively little in neurons or glial cells. These ex vivo findings were supported by the demonstration of picalm in human cerebrovascular endothelial cells in vitro.

The transport of Aβ across the vessel wall and into the bloodstream is a major pathway of Aβ removal from the brain, impairment of which is thought to be important in the development of AD. The endothelial cell localization of picalm would be optimal for its participation in this process. In the small cohort we examined, PICALM expression was greater in the AD frontal cortex in comparison to all 3 genes evaluated but the increase was significant only in relation to GAPDH (in relation to all cells), and not just endothelial cell markers. Because the analysis generated correlated rather than independent sets of data (as indicated by the similar alteration patterns in relation to the 3 calibrator genes), conventional correction for multiple testing (e.g. Bonferroni correction) would be excessively conservative, but the modest size of the increase should nonetheless be noted. Although our in vitro studies indicate that cerebrovascular endothelial cells are capable of producing picalm, it remains to be determined whether all of the picalm present in endothelial cells in the brain is synthesized within these cells.

The explanation for the elevated PICALM expression in the frontal cortex in AD is unclear. One possibility is that this is a response to elevated levels of Aβ. However, if that were the case, increased expression would also be expected in the temporal cortex. We speculate that other region-specific or biochemical changes related to the hyperphosphorylation of tau may modulate the influence of Aβ on PICALM expression. Alternatively the timing and distribution of any change in PICALM expression may relate to the stage of spread of disease. PICALM overexpression can inhibit the formation of CCV (11); thus, too large an increase in expression in AD could be detrimental to the removal of Aβ by endocytosis. If the elevation in PICALM expression is a response to Aβ, a change in the ratio of picalm to Aβ beyond a certain limit could result in a negative feedback signal that reduces picalm levels.

PICALM variants are associated with late-onset AD risk (1, 2), but whether these variants influence the production of picalm protein and Aβ load has not yet been determined. Vessel wall uptake is not the only mechanism by which variations in PICALM might influence Aβ levels. Picalm may also affect the processing of APP. APP was found in CCV isolated from PC12 cells (6) and a reduction in APP internalization and in Aβ release were demonstrated in vitro in CHO cells (7), and in vivo in mouse brains following the disruption of clathrin-mediated endocytosis (8).

This is the first study to show the presence of picalm in endothelial cells of human brain tissue and it suggests an increase in PICALM expression in AD. It remains unclear whether this increase is a reaction to the elevated level of Aβ or actually impairs Aβ uptake by endothelial cells and subsequent transport into the bloodstream. The elucidation of the role of picalm in AD may provide important insights into the pathogenesis of the disease.