Localization and quantitation of 68 kDa neurofilament and superoxide dismutase-1 mRNA in alzheimer brains

Molecular Brain Research, 9 (1991) 1-8 Elsevier BRESM 70241

1

Research Reports

Localization and quantitation of 68 kDa neurofilament and superoxide dismutase-1 mRNA in Alzheimer brains Martin J. Somerville 1-3, Maire E. Percy 2'3, Catherine Bergeron 4, Larry K.K. Yoong 4, Etienne A. Grima 5 and Donald R.C. McLachlan I 1Department of Physiology, University of Toronto, Toronto, Ont. (Canada), 2Neurogenetics Laboratory, Surrey Place Centre, Toronto, Ont. (Canada), 3Department of Obstetrics and Gynaecology, University of Toronto, Toronto, Ont. (Canada), *Department of Pathology (Neuropathology), Toronto General Hospital, Toronto, Ont. (Canada) and 5Division of Cardiology, Cardiology Research, St. Michael's Hospital, Toronto, Ont. (Canada) (Accepted 22 May 1990)

Key words: Neurofilament; Superoxide dismutase-1; Alzheimer's disease; In situ hybridization; Cerebellum; Hippocampus

The technique of in situ hybridization with tritiated RNA probes was used to study the expression of the 68 kDa neurofilament (NF68) gene and the superoxide dismutase-1 (SOD-l) gene in the brains of Alzheimer's disease (AD) patients. Messenger RNA (mRNA) for these proteins was localized and quantified in single cells of formalin-fixed, paraffin-embedded sections of 4 pairs of AD and Huntington's disease (HD) brains from patients matched for age at death and autopsy interval. The cerebellar cortex and hippocampal CA1 and CA2 regions were compared in these two groups of subjects, since in AD the CA2 region of the hippocampus and the cerebellum have been found to be relatively unaffected by the Alzheimer process in comparison to the hippocampal CA1 region. The amount of NF68 mRNA was reduced by approximately 50% in pyramidal cells of both the CA1 and CA2 of AD hippocampus (P < 0.001), and by 15% in the Purkinje cells of AD cerebellum (P < 0.05) relative to that of the HD individuals. SOD-1 mRNA was reduced by about 22% in the CA1 of AD brains (P < 0.001) with no corresponding reduction in the CA2, and by only 5% in the AD cerebellum (P > 0.5). The paired design of the study suggests that these results are not simply attributable to the effects of autopsy interval or the agonal process in each patient's death. INTRODUCTION Alzheimer's disease ( A D ) is an age-related disorder that is characterized by selective neuronal death, neurofibrillary degeneration and neuritic plaque formation 17'34. The selectivity of the neuronal involvement is particularly striking in the hippocampus, where different regions within this structure appear to have differing degrees of susceptibility to the disease process. The hippocampus contains a band of large pyramidal neurons that are of relatively consistent size and shape, a portion of which is apparently affected, the other not. Both neuronal loss and neurofibrillary degeneration are most conspicuous in the cornu ammonis 1 region (CA1), or Sommer's sector, whereas the neighbouring CA2 region and end plate are relatively spared 1,6. In contrast, other nerve cell populations, such as the Purkinje cells in the cerebellum of A D patients are largely free of these degenerative features 1°. Our group has previously shown, using Northern and quantitative dot-blot analyses of extracts of neocortex, that there are non-random alterations of gene expression in A D 25. Of particular interest was the finding that

message for one of the neurofilament proteins, the 68 kDa neurofilament (NF68), considered to be the most essential subunit for normal neurofilament assembly 4, was reduced to about 30% of control levels, an effect that was not entirely the result of neuron loss 21. This observation has been independently confirmed 5. Message for the Cu/Zn superoxide dismutase ( S O D - l ) , an enzyme that converts superoxide radicals to hydrogen peroxide and that has been directly implicated in the process of aging and cell death 14, was reduced to 60% of control levels (unpublished results). In addition, the amount of SOD-1 protein has been found to be slightly reduced in A D hippocampus 23. Both NF684'19 and SOD-13'7 have been shown to have a high level of expression in neurons. SOD-1 is encoded on the long arm of chromosome 2118. There is general agreement that the triple gene dosage of SOD-1 which accompanies D o w n syndrome (DS; trisomy 21) causes some of the pathological features of this disease 2,26,28,36. Because early onset A D is c o m m o n in adult DS patients, SOD-1 expression is also being studied in the Alzheimer process in DS 28 and in the general population 37. Theoretically, a relative excess or deficit of SOD-1 activity c o u l d be physiologically dele-

Correspondence: M.J. Somerville, Neurogenetics Laboratory, Room 423, Surrey Place Centre, 2 Surrey Place, Toronto, Ont., Canada M5S 2C2. 0169-328X/91/$03.50 © 1991 Elsevier Science Publishers B.V. (Biomedical Division)

terious as either situation promotes the formation of hydroxyl radicals and leads to unscheduled oxidation 28. Although genetic linkage studies appeared to rule out NF682°'33 and SOD-129 as candidate AD genes in familial AD, the above studies have suggested that altered levels of NF68 and SOD-1 are associated with the Alzheimer process. Northern and quantitative dot-blot analyses, however, do not allow a precise determination of whether mRNA reduction is restricted to neurons with degenerative features. In situ hybridization permits the localization and quantitation of specific mRNAs in individual cells. To further investigate the role of altered NF68 and SOD-1 expression in AD, we compared expression of these genes in the Purkinje cells (neurons) of the cerebellum and the pyramidal cells (neurons) of the CA1 and CA2 regions of the hippocampus, in 4 pairs of AD patients and control individuals using in situ hybridization. The control brains were obtained from patients with Huntington's disease (HD) and dementia to exclude the non-specific effects of a long chronic illness, drugs, and the prolonged agonal process. These HD brains were matched with the AD brains for age at death and autopsy interval, and were largely devoid of Alzheimer features. The work described in this paper has been presented in abstract f o r m 31'32.

MATERIALS AND METHODS

Brain tissue Formalin-fixed tissue was obtained from 4 autopsy-confirmed cases of AD and 4 cases of HD. All cases were originally submitted by the Canadian Brain Tissue Bank to the division of Neuropathology at Toronto General Hospital for the purposes of diagnosis. A large number of AD and HD brains were available, allowing the selection of pairs of cases closely matched for age at death and post-mortem interval (Table I). Portions of the cerebellum and hippocampus from each brain were paraffin-embedded; sections 5 /~m in thickness were mounted onto Denhardt's/poly-L-lysinetreated slides under RNase-free conditions 35.

proteinase K for 10 min at 37 °C. The hybridization was done at high stringency using 50% formamide, 10% dextran sulfate, 5 mM EDTA, 300 mM NaCI and 20 mM Tris-HC1 (pH 8.0) at 50 °C for 12-14 h. Sections were then RNase-treated, washed in 0.1 x SSC and dehydrated in an ethanol series. Slides were coated with Kodak NTB-2 nuclear track emulsion (Eastman-Kodak, Rochester, NY) and exposed in the dark at 5 °C for 2-10 weeks. After this period, the autoradiograms were developed and stained with hematoxylin and eosin.

Quantitation All grain counting on human tissue was done manually using a coding system for each case that left the observer unaware of the patients' identity or diagnosis. In each of the CA1 and CA2 regions of the hippocampus, grain counts were determined for 50 pyramidal cells. Similarly, the grains from 50 Purkinje cells of the cerebellum were counted for each patient. Autoradiographic grains had to be clearly visible over the nucleus and/or cytoplasm or in direct association with the cell membrane in order to be counted. The extent of neurofibrillary degeneration was determined in the CAI and CA2 regions using a Bielschowsky stained section adjacent to that used for each hybridization. The boundary between CA1 and CA2 was first identified, and marked with ink. Neurons with and without neurofibrillary tangles (NFTs) were counted in 4 adjacent 0.625 mm 2 fields in each region using an occular graticule.

Cell size analysis A Leco 2001 image analysis system was used to determine the cross-sectional perikaryal area of individual cells. In each of the CA1 and CA2 regions of the hippocampus, the areas were determined for 25 pyramidal cells; in the cerebellum, the areas of 25 Purkinje cells were recorded.

Statistical analysis The pooled data obtained from determinations of grain counts and cell sizes were subjected to a Student's paired t-test to determine the significance between differences in each area. In addition, regression analyses were used to examine the effects of perikaryai cross-sectional area, autopsy interval, age at death, cause of death, brain weight and presence or absence of A D pathology on the amount (expression) of NF68 or SOD-1 mRNA, These analyses were performed not only between CA1, CA2 and Purkinje cells of AD (affected) and HD (control) brain pairs, but also between the CA1 (affected) and CA2 and cerebellar (control) neurons of the AD and HD brains.

TABLE I

Characteristics of patient brain tissue Probes Murine NF68 cDNA coding for the second a-helical coil and carboxy tail region of the neurofilament 19 and human SOD-1 cDNA coding for the entire SOD-1 protein 12 were provided in plasmid form by Drs. S.A. Lewis and Y. Groner, respectively. These sequences were excised and reinserted into pT3/T7 expression vectors (Bethesda Research Labs. (BRL)). The sense and antisense strands for each insert were transcribed (following the BRL protocol) in vitro using T3 and T7 polymerases with [3H]UTP and [3H]CTP. Transcripts were isolated by phenol extraction and repeated ethanol precipitations, and reduced to a length of 50-150 bases by limited alkaline hydrolysis 27.

Hybridization Throughout this procedure, each AD sample and its paired HD control were treated simultaneously. Brain sections were deparaffinized in a xylene/ethanol series. Subsequent RNA/RNA hybridization followed the protocol of Pardue 27, with minor modifications. Pretreatment of tissue sections was carried out using 10 ~g/ml of

Patients are listed in the pairs that were used for analysis.

Patient number

Cause of death

Neurological status

Wet brain weight

Age

Autopsy interval

(h)

(g) 184/82 111/84

Sudden death Pneumonia

Huntington Alzheimer

975 1220

61 63

6 6

255/85 11/83

Sudden death Pneumonia

Huntington Alzheimer

890 650

60 63

6.5 10

413/83 373/85

Not known Pneumonia

Huntington Alzheimer

1200 1190

61 65

2 3

298/84 83/85

Pneumonia Not known

Huntington Alzheimer

1280 890

61 64

2 10

Autopsy interval simulation

N F 6 8 quantitation

The brain of an atravet/sodium pentabarbitol-anaesthetized longtailed macaque (Macaca fascicularis) was removed directly after cardiac saline perfusion and the hippocampal and cerebellar tissues were dissected free. Portions (approx. 1 x 1 x 1.5 cm) of these tissues were stored in a humidified chamber at 21 °C for a period of up to 24 h. At regular intervals over the 24 h period, portions of cerebellum and hippocampus were immersed in 4% buffered formalin (pH 6.8) and processed for RNase-free histological sectioning. These tissue sections were hybridized with tritiated NF68 or SOD-1 RNA sequences and stained with hematoxylin/eosin as described earlier. Grain counts and cell sizes were determined with the Leco image analysis system.

To allow a direct determination of cell grain density (i.e., relative amount of R N A ) between A D and n o n - A D brains, the pyramidal cells of the CA1 and C A 2 regions of hippocampus and the Purkinje cells of cerebellum were used for quantitation. The hybridization signal was strong when the antisense probe was used. A very weak signal was obtained from the sense probe and no signal was detected when sections were treated with RNase prior to the antisense hybridization (Fig. 1). Within the A D hippocampus, a striking reduction in the amount of NF68 m R N A was obvious from visual inspection in both CA1 and CA2 when compared to H D sections. A consistent reduction in hybridization signal in A D was obtained in all 4 pairs of samples. A D cerebellar Purkinje cells did not show a consistent reduction in grain counts (results not shown). W h e n the data were pooled (Table II) the grain count was reduced by 15% in A D cerebellum (P < 0.05), and by over 50% in both CA1 and CA2 of A D hippocampus (P < 0.001) (Fig. 2).

RESULTS N F 6 8 localization

Hybridization grains were localized exclusively to neuronal cells within hippocampus (pyramidal cells and granule cells) and the cerebellum (Purkinje cells, Golgi cells and basket cells). There was no difference in this overall pattern of localization between the A D and H D brains.

i,¸

:~/~lit

Fig. 1. Photomicrographs of in situ hybridizations using antisense probes for NF68 mRNA on: Alzheimer cerebellum (A); Huntington cerebellum (B); Alzheimer CA1 (C); Huntington CA1 (D); Alzheimer CA2 (E); Huntington CA2 (F). Huntington CA1 hybridized with sense probe (G). Control CA1 hybridized with antisense probe after RNase pretreatment (H). Bar = 10 ~m.

TABLE II Pooled grain counts from Huntington and Alzheimer brain sections hybridized with probes for NF68 and SO D-I

Each value represents data pooled from 4 patients. Values in parentheses represent mean count//~m2, adjusted using mean values for cell areas from Table III. Source of tissue

Probe

Counts for each brain region (mean value for200 cells ± S.E.M.) Cerebellum

CA1

CA2

Huntington

NF68

33.84_+ 1.6 (0.0459)

48.03 + 1.4

(0.1447)

46.72 ± 1.7

Alzheimer

NF68

28.31 _+1.5" (0.0406)

21.16 ± 0.7**

(0.0706)

20.07 ± 0.7** (0.0527)

Huntington

SOD-1

64.81 ± 2.5 (0.0880)

34.13 ± 1.1

(0.1028)

32.58 ± 1.1

(0.0798)

Alzheimer

SOD-1

61.41 ± 2.1 (0.0881)

26.72 + 1.0"*'*** (0.0892)

35.68 ± 1.3

(0.0937)

(0.1144)

*P < 0,05 relative to Huntington count. P < 0.001 relative to Huntington count. ***P < 0.001 relative to Alzheimer and Huntington CA2 counts. **

There was no significant difference within the A D group between CA1 and CA2 counts. Similarly, the H D group showed no discernible difference in grain counts between the CA1 and CA2 (Table II and Fig. 2).

pyramidal cells of CA2 (mean = 394.5/~m2). Although A D neurons in cerebellum, CA1 and CA2 were all smaller than H D neurons in the same areas, the differences were not statistically significant.

SOD-1 localization Grains were found primarily over the Purkinje cells of cerebellum and the pyramidal cells of hippocampus with a weak signal (above background) over the granule cell layer of the dentate gyrus. The localization of autoradiographic grains did not differ between A D and H D brains.

Neurofibrillary degeneration NFTs in A D hippocampus were found to be more numerous in CA1 than in CA2 (Table III). Both regions of A D hippocampus had more NFTs than the corresponding regions of H D brains.

SOD-1 quantitation Pyramidal cells of CA1 and CA2 and Purkinje cells of the cerebellum were analyzed. As with NF68, the use of the sense probe or an RNase pretreatment resulted in little or no signal (Fig. 3). The A D hippocampal CA1 had a significantly lower grain count than the H D CA1 and the H D CA2 (P < 0.001). Grain counts in the A D and H D C A 2 regions, however, did not differ significantly (Table II). Similarly, grain counts in cerebellar Purkinje cells did not differ significantly between A D and H D cases (Table II and Fig. 4). Within the A D group, the pyramidal cells of CA1 had a significantly lower grain count than those of the CA2 (P < 0.001). The H D group, however, showed no discernible difference in grain counts between the CA1 and CA2 (Table II and Fig. 4). Cell size comparison For each brain the cross-sectional areas of 25 cells were compiled in each region. As shown in Table III, Purkinje cells (mean = 716.7/zm 2) are approx, twice as large as pyramidal cells (mean = 355.2/~m2), and CA1 pyramidal cells (mean = 315.9/zm 2) are smaller (P = 0.02) than the

Grain count versus cell size When mean NF68 grain counts (Table II) were divided by the mean corresponding cell areas (Table III) the count//zm 2 was 50% lower in A D CA1 and CA2, and 11% lower in A D cerebellum than in H D tissue. When

G

90

r a i

80

n

60

C

50

o u n

40

t

2O

70

30

10

.....

[]

Huntington's

•

Alzhelmer's

CA1

CA,?.

Fig. 2. Histogram showing the pooled grain counts for NF68 in the areas indicated of Alzheimer and Huntington brains. Each bar shows the mean + 1 S.E.M. for 4 brains (200 cells).

ii

i¸ . . . . . . . . .

~

i~i:i~i~i ¸~ Fig. 3. Photomicrographs of in situ hybridizations using antisense probes for SOD-1 mRNA on: Alzheimer cerebellum (A); Huntington cerebellum (B); Alzheimer CA1 (C); Huntington CA1 (D); Alzheimer CA2 (E); Huntington CA2 (F). Huntington CA1 hybridized with sense probe (G). Control CA1 hybridized with antisense probe after RNase pretreatment (H). Bar = 10/~m.

mean SOD-1 grain counts in AD tissue were divided by the mean of corresponding cell areas the count//tm 2 was 13% lower in CA1, 17% higher in CA2, and not significantly altered in cerebellum relative to the HD tissue (Table II). The SOD-1 grain count//~m2 was 5% lower in AD CA1 than in AD CA2, and 29% higher in HD CA1 than in HD CA2.

Autopsy interval effect In macaque cerebellum and hippocampus the grain counts obtained for NF68 and SOD-1 decreased as a reciprocal of time post mortem using monotone smoothing (obtained by fitting running least squares lines and then using isotonic regression as implemented in the 'Pool Adjacent Violators' (PAV) algorithm) 9 for best fit (Fig. 5). Cerebellar grain counts for both probes showed greater variability than hippocampal grain counts. Both probes showed a significant reduction in grain count over the 24 h post mortem interval using a 2-sided t-test of the 4 gradients of the regression lines (mean slope = -0.523,

lOO G r

90 8O

a

i

70

n

6o

C

so

0 u

40

n t

3o 2o I0 Cerebellum

[]

Huntington's

•

Alzheim~r's

CA1

CA2

Fig. 4. Histogram showing the pooled grain counts for SOD-1 in the areas indicated of Alzheimer and Huntington brains. Each bar shows the mean ± 1 S.E.M. for 4 brains (200 cells).

TABLE III

Pooled cell size determinations (cross-sectional area) Each value represents data pooled from 4 patients. Values in parentheses represent the pooled mean % of cells with NFT, counted in each region.

Source of

Area for each brain region in (mean value for 100 cells + S. E. M.)

tissue

Huntington

Alzheimer

Cerebellum

CA1

CA2

736.7 + 19.4

332.0 + 17.0"* (0.60)

408.3 + 17.5 (0.54)

299.7 + 13.0"* (9.54)

380.7 + 17.2 (7.55)

696.7 + 21.3

**P < 0.05 relative to CA2.

P < 0.002, 95% confidence interval does not include 0), but this reduction was not significant between the 2 and 10 h period (mean slope = 0.497, P > 0.2, 95%

i 1 Hippocampus Grain Count

20~-o

o o

101 .----.--_---I

0

I

5

I

10

15

I

20

I

25

Autopsy interval (hrs)

40,

Cerebellum

30, Grain Count

O

O

O

20, og g 0

100

I

5

I

10

I

15

I

20

I

25

Autopsy interval (hrs)

Fig. 5. The amount of mRNA (grain count) for ~ NF68 and O - - - O SOD-1 in each of the two brain regions indicated over time post mortem (autopsy interval) in a long-tailed macaque. Each value shown is the mean grain count for 25 Purkinje cells (cerebellum) or 25 pyramidal cells (hippocampus).

confidence interval includes 0). No significant change in cell size was observed over the 24 h period (results not shown). A regression of grain count on autopsy interval of the human brain tissue used (2-10 h post mortem) showed no significant effect (P > 0,1, with the 95% confidence interval for mean A D and H D slopes both including 0). DISCUSSION

The elucidation of genetic activity within the brain is fundamental to the understanding of AD. By determining the relative abundance of the m R N A for NF68 and SOD-1 in situ it is possible to obtain some insight into the association between these genes and AD. Although some studies have reported that m R N A pools in A D brains are not affected by the length of the post mortem interval 13'24, the thawing of brain tissue has been shown to reduce total R N A levels 24. Formalin fixation of tissue, followed by in situ analysis, presumably circumvents this problem by fixing ribonucleases intracellularly and preventing redistribution of these enzymes. In the autopsy interval simulation with monkey brain, we could find no significant effect of autopsy interval on the relative abundance of m R N A for either NF68 or SOD-1 between the 2 and 10 h period (P > 0.2) (Fig. 5). In addition, regression analysis shows no significant effect (P > 0.1) in the human brains of autopsy interval (2-10 h) on NF68 or SOD-1 counts in A D or HD. Our results show a substantial reduction (approx. 50%) in the amount of neurofilament m R N A in A D CA1 and CA2 regions of hippocampus, relative to H D CA1 and CA2. This finding confirms previous studies in which total RNA was found to be reduced in Alzheimer CA1 and CA28, and indicates that the reduction does not result from neuron loss. We also found some reduction (approx. 16%) in the amount of NF68 m R N A in AD cerebellum relative to H D cerebellum. These reductions are unrelated to perikaryal size, as confirmed by regression analysis (P > 0.75). Also, the NF68 grain count/cell area still showed a reduction of greater than 50% in A D hippocampus relative to HD. The weak signals obtained from the sense probe for NF68 and SOD-1 show that non-specific RNA binding is minimal, and the lack of signals following an RNase pretreatment indicates that cellular RNA is the template for hybridization. When SOD-1 m R N A was compared in sections from AD and H D brains, a reduction (approx. 22%) could be found only in the CA1 of AD hippocampus. Although the CA1 cells were found on average to be about 20% smaller than those of the CA2 in both A D and HD, there was a striking reduction of the SOD-1 grain count/cell area in the A D CA1 versus CA2 in comparison to that

in the H D tissue (Table II). This shows that in comparison to H D hippocampus, A D CA1 has a selective reduction of SOD-1 m R N A content that is not simply a function of cell size. H D brains were used as controls for A D because of the large number of cases available for study with a relatively short post mortem interval (less than 12 h), which allowed the matching of several parameters simultaneously. Furthermore, the agonal circumstances are usually similar in both groups including the degree of hypoxia and the protracted terminal course. Although the deaths of 3 of the 4 A D cases and one H D case were attributed to pneumonia, and 2 of the 4 H D cases were sudden (one choked on gastric contents and one had a myocardial infarction) (Table I), regression analysis of mode of death on SOD-1 expression showed no significant effect (P > 0.1). With the provision that our sample size was small, the reduction of SOD-1 m R N A in A D CA1 may be a change that is specific to this disorder and unrelated to ischemic or hypoxic damage. When cell size is compared to the amount of NF68 m R N A in each H D and A D brain there is no apparent correlation between the perikaryal area and the amount of this message, and no significance was observed (P > 0.75) using regression analysis of NF68 count vs. cerebellar and pyramidal cell size. What may be of greater relevance to the expression of this gene, however, is the size (volume) of the axons Purkinje cells have in comparison to pyramidal cells, since axonal caliber has been shown to be dependent on neurofilament expression 16'3°. The amount of SOD-1 m R N A that was determined within Purkinje cells and pyramidal cells, however, more closely parallels the perikaryal cross-sectional areas of these cells (P < 0.00002 by linear regression analysis from individual case data) and is presumably more directly associated with perikaryal volume. There is general agreement in the literature that the CA1 region of the hippocampus contains more NFTs in A D than the CA21'6, although there is controversy regarding the degree of this difference 6. In our study, brains with NFTs in the CA1 always had NFTs in the CA2 (data not shown). Although our series is small, it clearly indicates that the reduction of NF68 m R N A in the pyramidal cells of CA1 and CA2 regions of AD hippocampus greatly exceeds the percentage of such cells with NFTs. NF68 m R N A was reduced in A D by greater than 50% in both CA1 and CA2 (Table II) whereas 9.5% of neurons in CA1 and 7.5% of neurons in CA2 were found to contain NFTs (Table III). This is in keeping with REFERENCES 1 Ball, M.J., Neuronal loss, neurofibrillary tangles and granulo-

previous reports that total protein production was diminished relatively equally in NFT- and non-NFT-containing neurons 22, and that total R N A was reduced in affected and unaffected areas 8. Similarly, there is a significant reduction of NF68 message in A D Purkinje cells of cerebellum in the apparent absence of neurofibrillary degeneration. Therefore, it appears that a more general reduction in transcription of the NF68 gene occurs than previously thought, and that NF68 expression is a more sensitive indicator of the disease process than standard histopathological techniques. Recent investigations have shown in neocortex that a reduction in the expression of NF68 is associated with condensation of chromatin in the 5' regulator region of the gene 21. Axotomy and regeneration of axons are also accompanied by reduced NF68 pool size 11'15. Whether the reduction in NF68 m R N A in CA1 and CA2 neurons of the A D brains that we analyzed is related to axonal damage, regeneration, or a mechanism unique to A D is unknown. The expression of NF68 contrasts with that of SOD-1 which was reduced only in the CA1 hippocampal area. For individual AD patients, there was no correlation between the level of m R N A for SOD-1 and the percentage of cells with NFTs (P > 0.8). As the pyramidal neuron density in the A D CA1 region is reported to be reduced in comparison to that of the CA26, the reduced m R N A level for SOD-1 may, therefore, correlate with the process of neuron death rather than NF1~ formation. In conclusion, our results suggest that the expression of both NF68 and SOD-1 in individual neurons is affected in A D relative to H D patients. A further investigation of the expression of these genes in a larger series of A D patients and control individuals, and a study of the significance of their altered expression in A D is warranted.

Acknowledgements. This investigation would not have been possible without the tissue that was provided by the Canadian Brain Tissue Bank. Partial support was obtained by grants made available to MEP from the Ontario Mental Health Foundation (No. 93485/87), and the Ontario Ministry of Community and Social Services Lottery Grants Programme. M.J.S. was the recipient of a Natural Sciences and Engineering Research Council (NSERC) post-graduate scholarship, a University of Toronto Open Fellowship and an advanced student bursary from the Gerontology Research Council of Ontario. M.E.P. was a National Health Research Scholar (NHRDP). We are grateful to Dan Williams, LECO Inc., for providing a model 2001 image analyzer for this study, and to Dr. D.E Andrews and Marguerite Ennis (the Clinical Research Support Unit, University of Toronto), for assistance with the statistical analysis.

vacuolar degeneration in the hippocampus with ageing and dementia, Acta Neuropathol., 37 (1977) 111-118. 2 Burger, P.C. and Vogel, ES., The development of the pathologic

changes of Alzheimer's disease and senile dementia in patients with Down's syndrome, Am. J. Pathol., 73 (1973) 457-476. 3 Cebailos, I., Agid, F.J-., Hirsch, E.C., Dumas, S., Kamoun, P.P., Sinet, P.M. and Agid, Y., Localization of copper-zinc superoxide dismutase mRNA in human hippocampus by in situ hybridization, Neurosci. Lett., 105 (1989) 41-46. 4 Chin, S.S.M. and Liem, R.K.H., Neurofilaments: a review and update. In G. Perry (Ed.), Alterations in the Neuronal Cytoskeleton in Alzheimer Disease, Plenum, New York, 1987, pp. 1-24. 5 Clark, A.W., Krekoski, C.A., Parhad, I.M., Liston, D., Julien, J.-P. and Hoar, D.I., Altered expression of genes for amyloid and cytoskeletal proteins in Alzheimer cortex, Ann. Neurol., 25 (1989) 331-339. 6 Davies, D.C., Hippocampal pathology and memory impairment in Alzheimer's disease. In D.C. Davies (Ed.), Alzheimer's Disease: Towards an Understanding of the Aetiology and Pathogenesis, Libbey, London, 1989, pp. 89-105. 7 Delacourte, A., Defossez, A., Caballos, I., Nicole, A. and Sinet, P.M., Preferential localization of copper zinc superoxide dismutase in the vulnerable cortical neurons in Alzheimer's disease, Neurosci. Lett., 92 (1988) 247-253. 8 Doebler, J.A., Markesbury, W.R., Anthony, A. and Rhoads, R.E., Neuronal RNA in relation to neuronal loss and neurofibriUary pathology in the hippocampus in Alzheimer's disease, J. Neuropathol. Exp. Neurol., 46 (1987) 28-39. 9 Friedman, J. ahd Tibshirani, R., The monotone smoothing of scatterplots, Technomet., 26 (1984) 243-250. 10 Giaccone, G., Tagliavini, E, Ghetti, B., Frangione, B. and Bugiani, O., Alzheimer's disease: Widespread distribution of preamyloid deposits in the grey matter, Alzheimer Dis. Ass. Disord., 3 (1989) 24. 11 Goldstein, M.E., Weiss, S.R., Lazzarini, R.A., Shneidman, P.S., L~es, J.F. and Schlaepfer, W.W., mRNA levels of all three neurofilament proteins decline following nerve transection, Mol. Brain Res., 3 (1988) 287-292. 12 Groner, Y., Lieman-Hurwitz, J., Dafni, N., Sherman, L., Levanon, D., Bernstein, Y., Danciger, E. and Elroy-Stein, O., Molecular structure and expression of the gene locus on chromosome 21 encoding the Cu/Zn superoxide dismutase and its relevance to Down syndrome. In G.F. Smith (Ed.), Molecular Structure of the Number 21 Chromosome and Down Syndrome, Ann. N.Y. Acad. Sci., New York, 450, 1985, pp. 133-156. 13 Guillemette, J.G., Wong, L., McLachlan, D.R.C. and Lewis P.N., Characterization of messenger RNA from the cerebral cortex of control and Alzheimer afflicted brain, J. Neurochem., 47 (1987) 987-997. 14 Halliwell, B., Free radicals, oxygen toxicity and aging. In R.S. Sohal (Ed.), Age Pigments, Elsevier/North-Holland, Amsterdam, 1982, pp. 1-62. 15 Hoffman, P.N. and Cleveland, D.W., Neurofilament and tubulin expression recapitulates the developmental program during axonal regeneration: induction of a specific fl-tubulin isotype, Proc. Natl. Acad. Sci. U.S.A., 85 (1988) 4530-4533. 16 Hoffman, P.N., Cleveland, D.W., Griffin, J.W., Landes, P.W., Cowan N.J. and Price, D.L., Neurofilament gene expression: a major determinant of axonal caliber, Proc. Natl. Acad. Sci. U.S.A., 84 (1987) 3472-3476. 17 Hirano, A., Certain aspects of the pathology of degenerative diseases of the central nervous system. In R. Strong (Ed.), Central Nervous System Disorders of Aging, Raven, New York, 33, 1988, pp. 17-24. 18 Huret, J.L., Delabar, J.M., Marlhens, E, Aurais, A., Nicloe, A., Berthier, M., Tanzer, J. and Sinet, P.M., Down syndrome with duplication of a region of chromosome 21 containing the CuZn superoxide dismutase gene with detectable karyotypic abnormality, Hum. Genet., 75 (1987) 251-257. 19 Lewis, S.A. and Cowan, N.J., Genetics, evolution, and expres-

sion of the 68,000-mol-wt neurofilament protein: isolation of a cloned cDNA probe, J. Cell Biol., 100 (1985) 843-850. 20 Lucotte, G. and David F., No cosegregation between the neurofilament (NF-L) gene marker and the Alzheimer disease susceptibilty gene (FAD) in a great familial pedigree, Alzheimer Dis. Ass. Disord., 3 (1989) 19. 21 Lukiw, W.J. and McLachlan, D.R.C., Chromatin structure and gene expression in Alzheimer's disease, Mol. Brain Res., 7 (1990) 227-233. 22 Mann, D.M.A., Changes in protein synthesis. In B. Reisberg (Ed.), Alzheimer's Disease: The Standard Reference, Free Press-Macmillan, New York, 1983, pp. 107-115. 23 Marklund, S.L., Adolfson, R., Gottfries, C.G. and Winblad B., Superoxide dismutase isoenzymes in normal brains and in brains from patients with dementia of Alzheimer type, J. Neurol. Sci., 67 (1985) 319-325. 24 Maschhoff, K., White, C.L. III, Jennings, L.W. and MorrisonBogorad, M.R., Ribonuclease activities and distribution in Alzheimer's and control brains, J. Neurochem., 52 (1989) 1071-1078. 25 McLachlan, D.R.C., Lukiw, W.J., Wong, L., Bergeron, C. and Bech-Hansen, T., Selective mRNA reduction in Alzheimer's disease, Mol. Brain Res., 3 (1988) 255-262. 26 Oliver, C. and Holland, A.J., Down's syndrome and Alzheimer's disease: a review, Psych. Med., 16 (1986) 773-787. 27 Pardue, M.L., In situ hybridization. In B.D. Hames and S.J. Higgins (Eds.), Nucleic Acid Hybridisation: a Practical Approach, IRL Press, Oxford, 1985, pp. 179-202. 28 Percy, M.E., Dalton, A.J., Markovic, V.D., McLachlan, D.R.C., Hummel, J.T., Rusk, A.C.M. and Andrews, D.F., Red cell superoxide dismutase, glutathione peroxidase and catalase in Down syndrome patients with and without manifestations of Alzheimer disease, Am. J. Med. Genet., 35 (1990) 459-467. 29 St. George-Hyslop, P.H., Tanzi, R.E., Polinsky, R.J., Haines, J.L., Nee, L., Watkins, P.C., Myers, R.H., Feldman, R.G., Pollen, D., Drachman, D., Growdon, J., Bruni, A., Foncin, J.-F., Salmon, D., Frommelt, P., Amaducci, L., Sorbi, S., Piacentini, S., Stewart, G.D., Hobbs, W.J., Conneally, P.M. and Gusella, J.E, The genetic defect causing familial Alzheimer's disease maps on chromosome 21, Science, 235 (1987) 885-890. 30 Schlaepfer, W.W., Neurofilaments: structure, metabolism and implications in disease, J. Neuropath. Exp. Neurol., 46 (1987) 117-129. 31 Somerville, M.J., Bergeron, C., Grima, E.A., McLachlan, D.R., Yoong, L.K.K. and Percy, M.E., Selective expression of the superoxide dismutase-1 gene in Alzheimer brains, Genet. Soc. Can. Bull., 20 (1989) 43. 32 Somerville, M.J., Bergeron, C., Grima, E.A., McLachlan, D.R., Yoong, L.K.K. and Percy, M.E., Reduction of 68 kDa neurofilament mRNA in the hippocampal pyramidal cells of Alzheimer brains, Alzheimer Dis. Ass. Disord., 3 (1989) 31. 33 Somerville, M.J., McLachlan, D.R. and Percy, M.E., Localization of the 68,000 dalton human neurofilament gene (NF68) using a murine cDNA probe, Genome, 30 (1988) 499-500. 34 Terry, R.D. and Hansen, L.A., Some morphometric aspects of Alzheimer disease and of normal aging. In R.D. Terry (Ed.), Aging and the Brain, Raven, New York, 1988, pp. 109- 114. 35 Uhl, G.R. and Gendelman, H.E., Appendix C. Sectioning and slide treatment. In G.R. Uhl (Ed.), In Situ Hybridization in Brain, Plenum, New York, 1986, pp. 264-268. 36 Wisniewski, K.E., Dalton, A.J., McLachlan, D.R.C., Wen, G.Y. and Wisniewski, H.M., Alzheimer's disease in Down's syndrome: clinicopathologic studies, Neurology, 35 (1985) 957961. 37 Zemlan, EP., Thienhaus, O.J. and Bosmann, H.B., Superoxide dismutase activity in Alzheimer's disease: possible mechanism for paired helical filament formation, Brain Res., 476 (1989) 160-162.