Article Text


Is there a relation between APOE expression and brain amyloid load in Alzheimer’s disease?
  1. J-C Lambert1,
  2. D Mann2,
  3. F Richard1,
  4. J Tian2,
  5. J Shi2,
  6. U Thaker2,
  7. S Merrot3,
  8. J Harris4,
  9. B Frigard3,
  10. T Iwatsubo5,
  11. C Lendon4,
  12. P Amouyel1
  1. 1Unité INSERM 508, Institut Pasteur de Lille, Lille cédex, France
  2. 2Neuroscience Research Group, University of Manchester, Hope Hospital, Manchester, UK
  3. 3Centre Hospitalier Gériatrique, Wasquehal, France
  4. 4Department of Psychiatry, University of Birmingham, Birmingham, UK
  5. 5Department of Neuropathology and Neuroscience, University of Tokyo, Tokyo, Japan
  1. Correspondence to:
 Dr Jean-Charles Lambert
 unité INSERM 508, Institut Pasteur de Lille, BP 245,1 rue du professeur Calmette, 59019 Lille cédex, France;


Background: It has been proposed that, independent of the ε4 allele, APOE promoter polymorphisms (−491 A/T and −219 G/T) may be risks factor for Alzheimer’s disease by modulating APOE expression.

Objective: To measure the level of APOE expression in Alzheimer’s disease.

Methods: Brains were obtained at necropsy from 114 patients with early and late onset sporadic Alzheimer’s disease in Greater Manchester (UK) during years 1986 to 2001. Total RNA was extracted from 84 brains. Purified lymphocytes were obtained from fresh blood from 16 probable Alzheimer cases from Lille (France). APOE and β-actin gene expression was determined by reverse transcriptase polymerase chain reaction in brain and lymphocytes.

Results: An inverse correlation between APOE expression level and Aβ loads was observed. As previously described and extended to 114 cases here, an association between the −219 TT genotype and a higher level of parenchymal Aβ deposition was found, irrespective of APOE ε4 allele status. This effect was more pronounced in older individuals, whereas higher Aβ load appeared more closely related to ε4 in the younger age group (cut off point at the median age at death (72.5 years)). The −219 TT genotype was associated with a decrease in APOE expression. There was a 60% decrease in APOE expression in lymphocytes from probable Alzheimer cases v controls (p = 0.01).

Conclusions: In the oldest individuals, reduced APOE expression, modulated in part by −219 G/T polymorphism, may influence risk and constitute a determinant Aβ load in Alzheimer’s disease.

  • apoE, apolipoprotein E protein
  • APP, amyloid precursor protein
  • CAA, cerebral amyloid angiopathy
  • CERAD, Consortium to Establish a Registry for Alzheimer’s Disease
  • DSM-III-R, Diagnostic and Statistical Manual of Mental Disorders, third edition, revised
  • NINCDS-ADRDA, National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer’s Disease and Related Disorders Association
  • Alzheimer’s disease
  • amyloid
  • APOE expression

Statistics from

Epidemiological and molecular studies suggest that multiple genes and environmental factors underlie the aetiology of Alzheimer’s disease. To date, four genes have been implicated. The apolipoprotein E (APOE) gene is recognised as a major risk factor for sporadic forms of Alzheimer’s disease with non-Mendelian patterns of inheritance,1 while pathogenic mutations in the amyloid precursor protein (APP), presenilin-1 (PS1), and presenilin-2 (PS2) genes are responsible for some rare early onset autosomal dominant forms.2

In vitro and in vivo studies show that pathogenic mutations in APP, PS1, and PS2 favour Aβ peptide production, particularly Aβn-42(43), the species suspected of initiating the formation of amyloid plaques. These latter observations support the amyloid cascade hypothesis, namely that of overproduction of or failure to degrade the Aβ peptide, which leads to amyloid deposition and perhaps provokes neurofibrillary degeneration and finally neuronal death.2

The exact role of apolipoprotein E protein (apoE) in the development of Alzheimer’s disease is still not understood. It has been suggested that apoE, and in particular the E4 isoform, may influence or provoke impaired dendritic outgrowth and arborisation,3 or enhance oxidative neuronal damage.4 Most interest has been focused on the amyloid cascade hypothesis whereby apoE may induce cholesterol dependent modifications of APP metabolism,5 enhanced Aβ aggregation and toxicity,6 or reduced Aβ clearance.7

Recent work suggests that, besides exerting a qualitative effect on the occurrence of Alzheimer’s disease through variations in APOE ε2/ε3/ε4 coding polymorphisms, the quantitative expression of APOE may also be a key disease determinant, for the following reasons:

  • although contradictory, numerous studies have reported a potential disruption of apoE levels in the brain,8,9,10,11,12,13 CSF, or plasma of Alzheimer cases compared with controls14–24;

  • Aβ load in transgenic murine brain seems to be dependent on the extent of APOE expression. In these transgenic models, mouse apoE appears to increase amyloid deposition whereas human apoE decreases it25,26;

  • four polymorphisms have been discovered in potential regulatory regions of the APOE gene.27–30 Two of them, which lie within the promoter, have received much attention (−491 A/T and −219 G/T). Both may be functional—that is, they appear to modify gene expression in vivo and in vitro29–36 and have been associated with an increased risk of developing Alzheimer’s disease30,31,35–37 and modulation of Aβ load.38–41

However, not all studies have found an association between these promoter polymorphisms and Alzheimer’s disease,42–44 so a better characterisation of their association with Alzheimer pathology and in particular with the hallmarks of Alzheimer’s disease will contribute to our understanding of their true role.

To examine further the possible role of apoE in the pathology of Alzheimer’s disease, we measured APOE mRNA levels in Alzheimer brains and assessed the association between APOE expression and Aβ loads. We also studied the relation between APOE promoter polymorphisms and both APOE expression and Aβ load. Finally, we measured the APOE mRNA levels in lymphocytes from probable Alzheimer cases and controls and related these to genotype.


Brain samples

Alzheimer’s disease brains were obtained at necropsy from 114 patients with early and late onset sporadic Alzheimer’s disease accessioned from the Greater Manchester region of the United Kingdom during years 1986 to 2001 (mean (SD) age at death, 73.1 (9.1) years; mean age at onset, 65.9 (10.3) years; 51% male). All patients were white.

Pathological diagnoses were made in accordance with CERAD neuropathological criteria for Alzheimer’s disease.45 All patients were at Braak stages 5 or 6 at time of death.46 The amount of plaque associated Aβn-40, Aβn-42(43), and total Aβ (Aβn-40+Aβn-42(43)) was quantified by image analysis in the frontal cortex (Brodmann areas 8 and 9) as previously described.47 The extent of cerebral amyloid angiopathy (CAA) in leptomeningeal and intraparenchymal arteries was rated semiquantitatively in 91 of the 114 patients (80%).48 The severity of CAA was assessed semiquantitatively on a five point scale (grades 0 to 4). Briefly, CAA was rated as follows:

  • 0, no blood vessels (small arteries, arterioles, and capillaries) are stained;

  • 1, a few leptomeningeal vessels only are involved;

  • 2, a few leptomeningeal vessels only are affected; mild intracortical vascular involvement;

  • 3, many leptomeningeal and intracortical vessels affected;

  • 4, many leptomingeal and intracortical vessels affected, with dyshoric angiopathy associated with intracortical vessels.

Tau load in the frontal cortex was determined by image analysis in 86 of the patients after immunostaining for phosphorylated tau, using a standard procedure employing monoclonal antibody AT8 (Innogenetics, Gent, Belgium) as primary antibody.49

Lymphocyte samples

Purified lymphocytes from fresh blood of 16 probable cases of Alzheimer’s disease (mean (SD) age, 81.8 (5.8) years, 25% male) and 36 controls (80.3 (6.5) years, 29.3% male) were cultured for 72 hours in the presence of 0.1% phytohaemagglutinin. These individuals were recruited from the same geographical area (Greater Lille) and all were also white. The diagnosis of probable Alzheimer’s disease was established according to the DSM-III-R and NINCDS-ADRDA criteria. The white controls were defined as subjects without DMS-III-R dementia criteria and with full integrity of their cognitive functions. Each individual or next of kin gave informed consent.


DNA was extracted from frozen brain tissue and used to genotype APOE coding and promoter polymorphisms.36

Real time reverse transcriptase polymerase chain reaction

Total RNA from brains of 84 of the 114 patients with Alzheimer’s disease (74%) was extracted from frozen brain tissue using a phenol/chloroform protocol (TRIzol® reagent, Invitrogen, San Diego, California, USA). Following harvesting, total RNA from lymphocytes was extracted using the RNeasy mini kit (Qiagen, Stanford, California, USA) and DNase treatment. The quality of total RNA was assessed using an Agilent 2100 bionalyser, and the ratio of ribosomal RNA 28S/18S systematically estimated employing Agilent 2100 biosizing software (range from 0.0 to 1.8 in brains or more than 1.7 in lymphocytes).50–52

Reverse transcription was done using 50 ng of total RNA, and expression was measured by real time polymerase chain reaction (PCR), using Taqman technology to coamplify cDNA from the APOE and β-actin genes, as described by the supplier (Applied Biosystems, Foster City, California, USA).

Statistical analyses

Statistical analysis was carried out using SAS statistical software, v.7.0 (SAS Institute Inc, Cary, North Carolina, USA). As the level of total RNA degradation (estimated by rRNA 28S/18S ratio) was identified as a potential confounder, this degradation level was subsequently included in a multivariate analysis of covariance, using a general linear model for comparison of the amount of mRNA between Alzheimer’s disease and APOE promoter genotypes. Standardised residuals of linear regression analysis between degradation and APOE/β-actin ratio were withheld for further correlations between Αβ loads and mRNA amount. The relations between Aβ loads and standardised residuals were tested using the non-parametric Spearman test.

Aβ load data were compared between −219 TT and GG+GT genotypes because homozygosity for the T allele has been associated with an increased risk of developing Alzheimer’s disease. Aβ load data were log transformed to normalise distributions. The effect of the coding and promoter polymorphisms on plaque associated Aβ load or the degree of CAA was tested with a multivariate analysis of covariance using a general linear model, or with Wilcoxon non-parametric test according to the number of individuals in each category.


RNA is notoriously labile and highly sensitive to RNAse. As a result, postmortem delay can strongly affect the results when studying mRNA expression. However, to use only postmortem delay as a potential confounder does not allow one to correct for storage, transportation, or extraction, which may have lead to further autolysis. In order to take these biases into account, the quality of total RNA and its degradation were assessed using the 28S/18S ratio of ribosomal RNA, measured by the Agilent 2100 bionalyser and biosizing software, a new technology allowing estimation of RNA degradation (and particularly RNAse action).50–52 In Alzheimer’s disease and control brains, the level of total RNA degradation was inversely correlated with the APOE/β-actin ratio by linear regression (p<0.0001, r2 = 0.164). Adjustments for this variable were thus made systematically.

We measured the level of APOE mRNA in the brain of 84 of the Alzheimer cases (fig 1) and observed that the APOE mRNA level correlated inversely with the level of Aβn-42 (p = 0.015) and with total Aβ load (p = 0.05) using the non-parametric Spearman test.

Figure 1

 Correlation between Aβ load and standardised residuals of the amount of APOE mRNA (that is, the level of APOE expression adjusted according to the level of total RNA degradation). The level of significance was estimated using the non-parametric Spearman test.

We next studied the association of the −491 A/T and −219 G/T polymorphisms in the APOE promoter that have been associated with levels of APOE expression. As the TT genotype of the −219 G/T polymorphism and the AA genotype of the −491 A/T polymorphism were reported potentially to increase the risk of developing Alzheimer’s disease in a study of 1732 Alzheimer cases and 1926 controls from six independent populations,36 we assessed whether these genotypes were associated with the neuropathological hallmarks and level of APOE expression in Alzheimer brains. The distributions of the APOE coding and −491 A/T and −219 G/T polymorphisms were similar to those previously reported in other Alzheimer populations.36 Our initial study on just 74 Alzheimer cases was increased here by 40 brains to 114. In this extended population, we still observed a significant effect of the −219 G/T polymorphism on Aβ loads.38n-40 and total Aβ load were increased by 124% and 35%, respectively, in individuals bearing the −219 TT genotype, compared with GT and GG (table 1; p = 0.007 and p = 0.032, respectively) while a non-significant borderline increase of 25% was observed for Aβn-42 deposition (p = 0.08). The amount of Aβn-40 was also significantly increased (p = 0.014) in Alzheimer patients carrying at least one ε4 allele (table 1). However, this latter increase was no longer significant when adjusted for the effect of the −219 G/T polymorphism (table 1). Finally, adjustment for a putative effect of the ε2 allele did not modify the association of the −219 G/T polymorphism with Aβ burden.

Table 1

 Effect of the −219 G/T and coding APOE polymorphisms on amyloid loads and cerebral amyloid angiopathy in the frontal cortex of 114 patients with Alzheimer’s disease

The effect of the APOE −219 G/T polymorphism may be specific to amyloid deposition in the form of plaques, as no effect of this polymorphism on the severity of CAA was detected, neither was ε4 detected, even though Aβn-40 is the major Aβ peptide species associated with this particular pathology (table 1). We may hypothesise we were not able to detect an effect of the APOE polymorphisms on CAA because the frontal region was not the area most affected by such lesions. However, the APOE polymorphisms were not associated with the CAA severity even in the occipital region which is the most affected (data not shown).53 Furthermore, no effect of the APOE promoter polymorphisms on tau load was detected (data not shown). No effect of the −491 A/T polymorphism on Aβ deposition was observed (table 1) and neither of the APOE promoter polymorphisms was associated with differential disease duration (data not shown).

Previous reports have described an increased risk of developing Alzheimer’s disease associated with the APOE −219 G/T polymorphism, but only in the oldest individuals.36,37 We divided our population into two subgroups, using median age at death (72.5 years) as the cut off point. In the older subgroup, Aβn-40, Aβn-42, and total Aβ loads were significantly increased, by 257% (p = 0.007), 31% (p = 0.035), and 45% (p = 0.016), respectively, in individuals bearing the APOE –219 TT genotype (table 2). This increase of Aβn-40, Aβn-42, and total Aβ loads may be independent of the coding APOE polymorphisms (+204%, p = 0.07; +35%, p = 0.21; and +54%, p = 0.05, respectively, in ε3/ε3 bearers; +98%, p = 0.05; +45%, p = 0.06; and +55%, p = 0.04 in ε3/ε4 bearers). Conversely, the APOE ε4 allele was only associated with an increase in Aβn-40 load in the younger subgroup (+238%, p = 0.006).

Table 2

 Effect of the −219 G/T polymorphism on amyloid load according to age in the brains of 114 patients with Alzheimer’s disease

It is important to note that when the population was divided using other ages at death as the cut off, determined by tertile or quartile design (oldest age class above 76 years or above 79 years, respectively), the results were clearer in term of Aβ loads. For instance, in the highest quartile class of age (⩾79 years), the Aβn-40, Aβn-42, and total Aβ loads were increased by, respectively, +274% (p = 0.05), +41% (p = 0.05), and +53% (p = 0.02) in individuals bearing the APOE −219 TT genotype.

We did not observe any significant differences in the amount of APOE mRNA between any of the polymorphic variants for −219 G/T, −491 A/T or coding polymorphisms in Alzheimer brains (data not shown). However, as the association of the −219 G/T polymorphism on Aβ load was strongest in the oldest individuals, we assessed whether the amount of APOE mRNA was dependent on age according to the APOE −219 G/T polymorphism. We observed that the −219 TT genotype was associated with an age dependent decrease in the amount of APOE mRNA (p = 0.02, non-parametric Spearman test). This correlation was not observed for cases bearing at least one −219 G allele (p = 0.77, non-parametric Spearman test). In the oldest Alzheimer subgroup (cut off at 72.5 years), the level of APOE mRNA was decreased by 65% in individuals bearing the −219 TT genotype compared with younger individuals of the same genotype (respectively, 0.23 (0.11) and 0.15 (0.06); p = 0.02). Such a decrease was not detectable in the oldest Alzheimer subjects bearing at least a G allele when compared with younger individuals (respectively, 0.20 (0.11) and 0.20 (0.11); NS).

Finally, we assessed whether the level of APOE expression was different in lymphocytes between Alzheimer cases and controls. We found a 60% decrease in APOE mRNA in the Alzheimer cases compared with the controls (p<0.01; fig 2). This may indicate that a decrease in APOE expression is a marker of Alzheimer’s disease. No significant effect of the APOE promoter or coding polymorphisms was detected (data not shown).

Figure 2

 Comparison of the level of APOE expression in Alzheimer’s disease and control lymphocytes by real time reverse transcriptase polymerase chain reaction. Bold lines indicate the mean values. AD, Alzheimer’s disease.


In this study we found that a decrease of APOE mRNA in human Alzheimer’s disease brains was correlated with an increase in Aβ loads. In the oldest cases, the APOE −219 TT genotype was associated with a decrease in APOE expression and an increase in Aβ loads. These data may indicate that variations in APOE expression are important for the development of amyloid plaques in the brain in Alzheimer’s disease, independent of the APOE genotype. This observation is of interest for the following reasons. First, although the finding is controversial, a majority of studies reported decreases in apoE levels in the CSF of Alzheimer patients14–20 while other found an increase21,22 or no change.23,24 Second, in brain tissues, both decreases8,9,10,11 and increases12,13 of APOE mRNA and proteins have been reported, but despite these discrepancies, overall it has been proposed that levels of apoE are reduced in Alzheimer’s disease.54 Our observations in human lymphocytes support this, even though they are peripheral cells.

Some points need further discussion. For instance, we previously described a specific ε4 allele overexpression in Alzheimer’s disease brains carrying the ε3/ε4 genotype compared with controls of the same genotype.55 At that time, we did not measure the global level of expression of the APOE gene, and our previous data cannot be compared directly with the results reported here. However, even though our previous observation may in part have resulted from a technical bias, restricting their significance,56 this observation raises an important issue—that there may be a contradiction between the consequences of a decrease in APOE expression and the deleterious effect of the apoE4 isoform (generally assumed to be an increase in negative function compared with the apoE3 isoform).57 Specific age dependent mechanisms may nevertheless help us to gain a better understanding of this potential contradiction. For instance, it is well documented that the apoE4 isoform facilitates Aβn-40 deposition47 and this relation may underlie the pathogenesis of Alzheimer’s disease in younger persons. It is well established that the risk of Alzheimer’s disease associated with the ε4 allele decreases in patients who develop the disease after 65 to 70 years.1,58 Our present findings may indicate that possession of the APOE −219 TT genotype is associated with a reduced APOE expression and a greater Aβ accumulation in such older people, consistent with the increased risk for Alzheimer’s disease conferred by this genotype in older subjects.36,37 We may thus hypothesise that it is an insidious overall lack of brain apoE, rather than any particular isoform or allele specific process that promotes Aβ formation and the development of Alzheimer’s disease in such older people.

In respect of our results, there are some further caveats that need to be considered. First, there may be systematic bias in the measurement of APOE expression resulting from, for example, postmortem autolysis, storage, and transportation. As a result, we systematically corrected our results for degradation of extracted total RNA by the use of the Agilent technology, which, to date, is likely to be the most powerful way of estimating total RNA quality.50–52 However, we are aware that this correction may be insufficient for highly degraded RNA samples, and replication of our observations in other large sets of brain tissue is necessary. Furthermore, measurement of apoE protein level is pertinent, as this may be less susceptible to postmortem autolysis, and APOE protein synthesis is highly dependent on regulation at post-transcriptional levels. Second, we are also aware that quantification of low abundance mRNAs is more sensitive to problems derived from mRNA degradation, although APOE is highly expressed in brain.59 Finally, multiple stratifications may lead to false positive results, and corrections for multiple testing could be applicable. However, it is important to note that the polymorphisms are not completely independent, nor are the tested variables, and the results need to be coherent at both epidemiological and biological level.

Our present data substantiate our initial preliminary report on the effects of −219 G/T polymorphism on Aβ load.18 However, in this study we were not able to detect any effect of the −491 A/T polymorphism on the pathological variables studied. Our findings do not confirm our initial report of an association of the −491 AA genotype on Aβn-40 load in a subset of non-ε4 Alzheimer’s disease bearers22; in the present study, we observed a non-significant (p = 0.27) increase of 78% in Aβn-40 load (+78%) in non-ε4 bearing Alzheimer cases with the −491 AA genotype.

It is not clear why these two polymorphisms—both capable of modulating APOE expression12–18—do not apparently act in a similar way in the aetiology of Alzheimer’s disease. However, several observations may explain this apparent discrepancy. As the decrease in APOE expression only appears in Alzheimer’s disease carriers of the −219 TT genotype in an age dependent manner, we may hypothesise that the cellular mechanism leading to this decrease causes the development of Alzheimer’s disease only in the oldest individuals, thereby explaining why the association of the −219 G/T and −491 A/T polymorphisms with dementia differs according to age.36,37 Consequently, it is likely that transcriptional factors able to bind the sequences containing the −491 A/T and −219 G/T polymorphisms are different. For example, we have observed that the −219 G/T, but not the −491 A/T polymorphism, modulates the binding of oestrogen receptors.60 Similarly, Campillos et al described a specific interaction of the heterogeneous nuclear ribonucleoprotein A1 protein with the −219T allele, leading to the modulation of the APOE promoter activity.61 Furthermore, there may be a tissue dependent effect of the APOE promoter polymorphisms on expression, leading to a local modulation of this expression in the brain. Regulation of the transcription of APOE is likely to be highly complex and is influenced not only by functional proximal polymorphisms (−491 A/T and −219 G/T) but also by distal regulatory regions. Environmental factors are also likely to influence this complex genetic regulatory mechanism.

In conclusion, numerous coherent observations indicate that a decrease in APOE expression may be important for the development of Alzheimer’s disease. Indeed, in mice a decrease in amyloid deposition is inversely associated with the copy number of the human APOE gene, this effect being the stronger when the human clusterin (apoJ) is present.62 Likewise, in our study, the highest level of APOE mRNA found correlated with lowest amyloid depositions. We also showed that the −219 TT genotype is associated with both an increased risk of developing Alzheimer’s disease and an increased amount of amyloid deposition in brains in the oldest subjects. These associations might be explained by the fact that the −219 T allele is correlated with a decrease in gene reporter activity in vitro,29 with low APOE mRNA levels in the brain of old cases as well as with a low plasma level in young people.32 Finally, several reports have described an increased risk of myocardial infarction associated with the −219 TT genotype.32,63,64 This is of particular interest as it has been suggested that apoE may participate at a local level in the degradation of arteriosclerotic plaques.32 We hypothesise that apoE participates in the degradation of amyloid deposits, as suggested by our data and by recent transgenic mouse results.65


We thank Miss Alexia Houzet from Genoscreen Society for the RT-PCR work.


View Abstract


  • Competing interests: none declared

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.