It is established that apoE3 forms disulphide-linked homodimers and apoE3-apoA-II heterodimers in human plasma and CSF [30,33]. As apoE4 lacks Cys it cannot form disulphide bonds. Whether apoE3 exists in a dimeric state in human brain tissue is unknown and we therefore focused on this issue. Western blot analysis of TBS-soluble brain homogenates derived from the hippocampus of a control apoE3/3 subject indicated a clear apoE3 homodimer when the sample was run under non-reducing conditions (Fig 1A). The homodimer was detected at ~95 kDa (as opposed to the predicted ~68 kDa) which is consistent with previous data [30,33]. ApoE contains a major cut site for thrombin in the linker region [39]. Due to the high proteolytic specificity of thrombin [40,41], it was used here to further confirm the identification of the ~95 kDa band as an apoE homodimer. The putative apoE3 homodimer was removed after incubation with thrombin, as predicted, further confirming that the ~95 kDa band is not likely to be due to non-specific binding of the antibody used.

Incubation of the homogenate in the presence of heat-inactivated thrombin resulted in a partial loss of the apoE dimer which suggests that endogenous proteases may also degrade apoE (Fig 1A). A band of relatively lower intensity was also observed at ~43 kDa in the non-incubated apoE3 control condition. This is consistent with the apoE-apoA-II heterodimer previously detected in human plasma and CSF [30,33]. A series of apoE fragments was also detected with a major band at ~24 kDa. This is in close agreement to our previous observations [29].

Additional hippocampal and frontal cortex homogenates from control apoE3/3 donors were analysed and this revealed that the apoE3 homodimer was present in all samples and accounted for 8.3 ± 0.9% (mean ± SE, n = 6) of the total apoE present in the hippocampus and 16.5% ± 4.1% (mean ± SE, n = 7) of the total apoE present in the frontal cortex (Fig 2). Although the percentage of apoE present as the homodimer was on average increased in the frontal cortex, this difference did not reach statistical significance. We also analysed frontal cortex and hippocampus derived from apoE3/3 AD samples and found that apoE3 dimers were detected in all AD samples and were identical to those in the control samples (Fig 1B). There was no significant difference regarding the proportion of apoE that was present in the dimerised form when frontal cortex from the two groups were compared (Control 16.5 ± 4.1%, mean ± SE, n = 7; AD 11.2 ± 2.4%, mean ± SE, n = 5).

We also used an additional rigorous extraction protocol employing extraction buffer that contained the detergent Triton-X100. This was done in order to maximise recovery of apoE that may be associated with TBS-insoluble material. Both the control and AD samples were found to contain apoE homodimers when samples were extracted in detergent-containing buffer (Fig 3). These data are very similar to the results obtained with the TBS-soluble homogenates (Fig 1).

As predicted, apoE dimers were not detected in any of the 7 apoE4/4 AD samples analyzed (Fig 4, see also Fig 1C). Analysis of heterozygous apoE3/4 AD samples revealed a significant 61% reduction in the percentage of apoE3 present as homodimer as compared to apoE3/3 homozygous AD samples (AD 3/3 11.2 ± 2.4%, mean ± SE, n = 5; AD 3/4 4.4% ± 0.6, mean ± SE, n = 5; p < 0.03). Representative blots from heterozygous apoE3/4 AD samples are shown (Fig 5).

One potential issue that arises in the analysis of protein modifications in human post-mortem material is the potential for artifactual changes to be induced by post-mortem interval (PMI). Even though clear ~95 kDa and ~43 kDa apoE dimers were detected in apoE3 AD homozygous samples with a post-mortem delay of as short as 1 h (and PMI for the AD sample shown in Fig 1B was only 7 h), we also probed for apoE dimers in human neuroblastoma cell lysates and freshly harvested rabbit brain. The human SK-N-SH neuroblastoma cell line expresses the APOE ε3/ε3 genotype [42,43] and synthesizes large amounts of apoE under serum starved conditions [11], whereas rabbits are one of the few non-human species known to contain an apoE Cys¹¹² residue [31].

Analysis of SK-N-SH cell lysates under non-reducing conditions revealed the presence of the ~95 kDa apoE homodimer and a more prominent (than human brain) ~43 kDa heterodimer (Fig 6A). Since the ~43 kDa dimer could theoretically represent a disulphide-linked apoE N-terminal domain homodimer (a predicted MW of ~44 kDa), we used a C-terminal specific monoclonal antibody in this experiment and the detection of the ~43 kDa band indicates an intact C-terminus. This rules out the presence of disulphide-linked N-terminal domain homodimer. The ~43 kDa band may represent an apoE3-apoA-II heterodimer as has been previously described in human plasma and CSF [30,33].

In the analysis of rabbit frontal cortex, the brain was dissected and processed immediately at the time of death to eliminate post-mortem delay and all measures were taken to prevent serum and CSF contamination (see Materials and Methods). An apoE band at ~95 kDa was also detected in rabbit brain when samples were run in the non-reduced state (Fig 6B), again indicating that apoE containing Cys¹¹² does form a homodimer in the brain.

To address the possibility that the observed apoE dimerisation may occur during tissue homogenization and processing for electrophoresis, freshly prepared rabbit brain and frozen human frontal cortex tissue (AD apoE3/3) were homogenized in buffer containing the thiol trapping agent iodoacetamide. Homogenization was also performed using a detergent-rich lysis buffer to delipidate apoE-containing lipoproteins and help prevent the possibility of dimers forming spontaneously on lipoprotein particles during homogenization and processing. The presence of 100 mM iodoacetamide did not result in a decrease in apoE homodimer levels in either rabbit or human brain tissue (Fig 7A). The thiol-trapping efficiency of 100 mM iodoacetamide was confirmed by spectrophotometric analysis of the total thiol concentration of the homogenates using Ellman's reagent (5,5'-Dithio-bis(2-nitrobenzoic acid)). The total thiol concentration was significantly reduced in both rabbit and human brain homogenates (84% and 96%, respectively) treated with iodoacetamide (Fig 7B). This data demonstrates that apoE dimers are not artificially formed during tissue homogenization and processing, and further indicates that apoE dimers are generated in the brain.

Brain tissue samples were obtained through the Australian Brain Donor Program with ethics approval from the University of New South Wales Human Research Ethics Committee (approval No. HREC03322). The research was carried out in compliance with the Helsinki Declaration. Cortical neuritic plaques and neurofibrillary tangles were assessed according to current international standards in order to pathologically confirm the diagnosis of AD post-mortem [48,49]. Sample details are provided in Table 1.

Samples were taken from the frontal cortex or hippocampus; both areas that are affected by AD [50]. The homogenisation protocol was previously described in detail [29], and was used for all samples unless stated otherwise. In brief, between 60-90 mg of brain tissue was homogenized with a pre-chilled 1 mL glass dounce homogenizer, using 15 volumes of ice-cold tris-buffered saline (TBS) pH 7.4. Protease and phosphatase inhibitors (Calbiochem, San Diego, USA) were added to all samples except those to be used in experiments requiring enzyme addition (samples C6, AD5 and AD6). After centrifugation at 16,000 g for 25 minutes at 4°C the TBS-soluble supernatant fraction was collected. Where specified, the pellet was then washed with 50 μL of TBS and centrifuged again for 5 min, the supernatant was then discarded and pellet was resuspended in 15 volumes of TBS containing protease and phosphatase inhibitors and 1% (w/v) Triton X-100 (TBS-X) by pipetting up and down. Samples were then mixed by rotation for 30 min at 4°C, followed by centrifugation at 16,000 g for 25 min at 4°C and collection of the TBS-X soluble supernatant fraction.

A second protocol was also used, where specified, to homogenise brain tissue in a detergent-rich lysis buffer. In brief, brain tissue was homogenized with a pre-chilled 1 mL glass dounce homogenizer, using 10 volumes of DRLB (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.1% SDS, 0.5% IGEPAL CA-630, 0.5% sodium deoxycholate) with protease and phosphatase inhibitors and either with or without 100 mM iodoacetamide (Sigma) to trap thiol groups and thus prevent changes in disulphide bond status during homogenization and sample processing [51]. Homogenates were then centrifuged at 16,000 g for 10 min at 4°C in an Eppendorf 5417-R refrigerated Microfuge and the supernatants collected.

An adult male Watanabe rabbit was euthanased via cardiac puncture using 5 mL Lethabarb (1 mL per 2 kg body weight, Virbac, Sydney, Australia) and the brain surgically removed to dry ice and processed immediately in order to eliminate post-mortem delay. Approximately 100 mg of tissue was removed from the cerebral cortex, all visible vasculature was removed and the sample was rinsed three times in ice-cold phosphate-buffered saline (PBS). The sample was then homogenized following the protocol used above for the human samples.

Genomic DNA was extracted from brain tissue and APOE amplified by PCR. Briefly, each reaction (50 μL) contained 200 nM of each primer (Invitrogen, Carlsbad, CA) 5'-TCCAAGGAGCTGCAGGCGGCGCA-3' (forward) and 5'-ACAGAATTCGCCCCGGCCTGGTACACTGCCA-3' (reverse), 2 mM dNTPs, 2 mM MgCl2, 2 U Taq polymerase (PCR reagents supplied by Promega, Madison, WI) and 400 ng DNA, all combined in nuclease free H₂O. Amplification was carried out with 38 cycles of denaturation (95°C, 30 sec), annealing (60°C, 30 sec) and extension (70°C, 30 sec). The 244 bp PCR product was purified using the QIAquick PCR purification kit (Qiagen, Venlo, Netherlands), following the manufacturer's protocol, and eluted in 40 μL H₂O. Endonuclease restriction digests (25 μL) were performed on 15 μL of eluted DNA using either AflIII (5 U) or HaeII (20 U) in the presence of BSA (100 μg/mL) and the supplied buffer (New England Biolabs, Ipswich, MA) at 37°C for 16 h. The ε3 allele is resistant to both enzymes while ε4 is cleaved by AflIII (producing a 190 bp product) and ε2 is cleaved by HaeII (producing a 191 bp product) assessed using ethidium bromide stained 8% polyacrylamide gels.

Cell culture media and additives were from Invitrogen (Melbourne, Australia). Human neuronal SK-N-SH cells were routinely grown in DMEM, 10% (v/v) fetal calf serum (FCS), 2 mM glutamine, and 100 IU/ml penicillin and 100 μg/ml streptomycin. Cultures were grown in 75 cm² flasks at 37°C in 5% CO₂ and plated into 6-well plates for use in experiments. To induce apoE expression, SK-N-SH cells were cultured under serum starved conditions (5 days of culture without media replenishment) and harvested in cell lysis buffer (10 mM Tris-HCl, 10 mM Na₂PO₄/NaHPO₄, pH 7.5, 130 mM NaCl, 1% Triton-X-100, 10 mM NaPP_i) as described previously [11]. SK-N-SH cells have been shown to be apoE3/3 homozygous [42,43].

Bicinchoninic acid protein assays were performed on brain homogenate samples and equal amounts of protein were separated on 12% SDS-PAGE gels and transferred onto 0.45 μm nitrocellulose membranes at 100 V for 30 min. Membranes were Ponceau-stained and scanned before blocking overnight at 4°C in PBS containing 5% (w/v) non-fat dry milk. The membranes were then probed with the relevant antibodies at 22°C for 1 h to reveal the bands of interest. Concentrations of antibodies were: goat polyclonal anti-human apoE 1/5000 (Calbiochem) or mouse monoclonal anti-human apoE 21-F3-D2 1/1000 (Biogenesis, Poole, UK). The membranes were washed three times in PBS containing 0.1% (w/v) Tween-20 and then incubated with horseradish peroxidase-conjugated rabbit anti-goat (Dako, 1/2500) or rabbit anti-mouse (Dako, 1/1000) secondary antibody for 1 h. The proteins of interest were detected using enhanced chemiluminescence (ECL, Amersham Biosciences) and X-ray film. Signal intensity was quantified using Image-J software. Specifically, a fixed area was used to separately measure signal intensity from i) the region encompassing intact ~34 kDa and fragmented ~24 kDa apoE, ii) apoE homodimer at ~95 kDa, and iii) an adjacent blank region to serve as a background control. The background value (iii) was subtracted from the apoE measurements (i and ii) and the homodimer quantification was expressed as a percentage of total apoE. Where possible, relative differences between samples were assessed on the same blots or using simultaneously processed gels with identical film exposure times.

Thrombin digestion was performed by incubating brain homogenates (30 μg of protein) prepared in the absence of protease inhibitors with 4.5 U of thrombin (Sigma, St. Louis, MO) in PBS at 37°C for 16 hours. Two control conditions were also analysed: homogenates either stored at -80°C for the incubation period or incubated with thrombin that was heat-inactivated at 95°C for 15 minutes.

The concentration of total thiol groups, both protein-bound and free, in specified brain homogenates was determined using 5,5'-dithio-bis(2-nitrobenzoic acid) (DTNB, Sigma) also known as 'Ellman's Reagent' as described previously [52]. Briefly, a 30 μL volume of brain homogenate (either undiluted or diluted in homogenisation buffer) was combined with 75 μL of dilution buffer (30 mM Tris HCl, 3 mM EDTA pH 8.2), 25 μL of DTNB reagent (1.19 mg/mL DTNB in methanol) and 400 μL methanol. Samples were centrifuged at 3000 g for 5 min at room temperature. 3 × 90 μL aliquots of supernatant were collected and absorbance was measured at 415 nm. Thiol concentration was determined by reference to a cysteine (Sigma) standard curve.