Addiction B iolog y (1997) 2, 337 ± 348
R ESEA R CH A RT IC LE
Concentrations of transferrin and carbohydrate-de® cient transferrin in postm ortem hum an brain from alcoholics PETER R. DODD, ALLISO N L. ECKERT, LINDA M . FLETCH ER, 2 2 1 JILLIAN J. KR IL, CLIVE G. HARPER & JUNE W. H ALLIDAY
1
1
Clinical Research Laborator y, Royal B risbane Hospital Research Foundation, Liver Unit, 2 Queensland Institute of M edical Research, D epartment of Patholog y, University of Sydney and Departm ent of Anatom ical Patholog y, Royal Prince Alfred Hospital, Sydney, Australia
Abstract Transfer rin (T f ) and its carbohydrate-de® cient isofor m (CDT) were m easured by radioimm unoassay in phosphate-bu Œered saline extracts of two infor mative areas of cerebral cortex tissue obtained at autopsy from alcoholics without other associated disease ( n 5 4); alcoholics with cir rhosis of the liver ( n 5 4) and agematched controls ( n 5 4). Tota l T f was also m easured in two infor m ative cortical areas from ® ve dem entia cases. All cases were m ale. Total im m unoreactive T f was assayed directly in the extract, CDT im m unorea ctivity in the concentr ated eluate after the sialylated form was removed by passing through DE AE-S ephacel at pH 5´65. B rain CDT averaged 10% of total T f overall. Although replicate extractions of individual sam ples gave consiste nt assays for both substances, there was wide variation both between di Œerent cortica l areas from a given case and between cases within groups. There were no signi® cant di Œerences between total T f levels in uncom plicated alcoholics, dem entia cases and controls, but cir rhotic alcoholics gave signi® cantly higher values. The CDT : T f ratio was not increased in the brains of either group of alcoholics com pared to controls. W hereas the ser um CDT : T f ratio is an excellen t marker of recent alcohol consum ption, brain T f and CDT concentrations do not m ark alcoholism nor dem entia, and their biolog ical variability diminishes their usefuln ess as disease indices. H owever, brain T f m ay be a m arker of cir rhosis-ind uced changes.
Introdu ction Markers of excessive alcohol consumption have been developed to augment the diagnosis of chronic alcoholism, which is based mainly on clinical history and responses to questionnaires combined with laboratory investigations of limited selectivity and sensitivity (Schellenberg et al. , 1989; Stibler, 1993; Lieber et al. , 1993; Carlsson
et al., 1993). Alcoholic status is particularly di cult to determine in cases where only autopsy tissue is available. Estimates of the average antemortem daily alcohol consumption of deceased cases m ust be derived from an examination of medical records together with responses to questionnaires by relatives of the deceased. A reliable, objective index of alcoholism would be of value
Correspondence to: Dr Peter R. Dodd, Clinical Research Laboratory, Royal Brisbane Hospital Research Foundation, Br isbane 4029, Australia. Fax: + 61-73-362 0108; e-mail: peterD@ qimr.edu.au Received for publication 22nd July 1996. Accepted 20th October 1996. 1355 ± 6215/97/030337 ± 12 Society for the Study of Addiction to Alcohol and Other Drugs
338
Peter R. Dodd et al.
for classifying such m aterial, especially if it varied between brain areas according to their susceptibility to damage (Dodd et al., 1992). A marker which has become widely accepted for studies of living cases is the carbohydratede® cient form (CDT ) of the iron-binding protein transferrin (T f ). CD T is a desialylated transferrin with an isoelectric point between 5.7 and 5.9. It has been found in the serum of alcoholic subjects and has been shown to be raised after the longterm ingestion of alcohol (Stibler, 1991). The marker allows alcoholic and non-alcoholic liver disease cases to be distinguished, since it is not raised in serum from the latter group (KwohGain et al., 1990; Fletcher et al. , 1991), except in a few cases of primary biliary cirrhosis (Bell et al. , 1993; Bean et al., 1995). Transferrin is also a carrier for aluminium, which has been thought to accumulate in senile plaques and thus contribute to the ñ tiology of Alzheimer-type dementia (DAT) (Roskams & Connor, 1990; Edwardson et al. , 1992). High levels of transferrin have been reported to be present in the human brain (Morris et al., 1987). However, although free-radicalmediated excitotoxicity, which would be promoted by reactive iron species, has frequently been invoked as a potential pathogenic mechanism underlying the brain damage associated with alcohol abuse, and in neurodegenerative diseases such as Parkinson disease or DAT (Dodd et al. , 1994), evidence on the concentrations of brain iron and T f in degenerative brain diseases has been con¯ icting (M orris et al. , 1987; Thompson et al. , 1988; Connor et al. , 1992a,b; Loeç er et al. , 1995). We studied whether the concentration of either form of T f was altered in cerebral cortex tissue obtained at autopsy from well-characterized chronic alcoholic cases with and without associated liver disease. Key neurotransmitter receptor sites have been shown to correlate with histopathological indices of localized cortical damage. We were interested to assess whether the densities of these sites might also correlate with C DT or T f concentrations, or with CD T : T f ratios, which would hence be of value as pathogenic m arkers. To provide a positive disease control for the study, we also ascertained whether T f concentrations were altered in the brains of dementia cases. Since some of the disparities in earlier work may have derived from methodological diŒerences, we used a highly speci® c assay procedure to quantify the T f species (Kwoh-Gain et al., 1990; Fletcher
et al. , 1991). Unfortunately, due to di culties in obtaining post-mortem tissue and because of the complexity of the analytical technique, only a limited number of cases were available for study. A preliminary report of this work has appeared (Eckert et al. , 1991). To the best of our knowledge, this is the ® rst report on the concentration of C DT, or on its regional distribution, in human brain.
M ethods C ase selectio n Tissue was obtained from alcoholics with or without pathologically con® rmed cirrhosis, and from matched controls (Table 1). A further ® ve cases had clinically diagnosed dem entia (Table 2). At autopsy, the latter cases were classi® ed under criteria which have been given elsewhere (Scott, Tannenberg & Dodd, 1995). These cases were somewhat older than the ® rst set (m edian age 70 years vs. 57 years for the cases in Table 1), and the post-mortem delay was somewhat longer (median 48 hours vs. 23 hours for Table 1 cases). Three of the cases (13 ± 15) were diagnosed as Alzheimer’ s disease, with abundant senile plaques and neuro® brillary tangles combined with neuronal loss (Scott et al. , 1995). Case 16 also had abundant senile plaques, but tangles were generally absent and numerous cortical Lewy bodies were seen: this is a case of diŒuse Lew y body disease . Case 17 had neither plaques, nor tangles, nor Lewy bodies, nor neuronal loss: it is an exam ple of fronta l lobe dem entia lacking distinctive histologic al features , and m ay be regarded as a disease control for excessive plaque deposition. Inform ed consent for autopsy was obtained from the next of kin in all cases; ethical clearance for the project is current under RBH protocol No. 1992 /22. In all cases, tissue was obtained both from a cortical region which is susceptible to damage in the disease (for alcoholics, superior frontal cortex; for dementia cases, m id-temporal cortex), and from a relatively spared cortical region ( primary motor cortex in all cases) (Dodd et al., 1994). Tissue processing and storage for both neurochemical studies and histochemical analyses was carried out as previously described (Scott et al. , 1995). Tissue storage time at 2 70ë C was well within the range over which transferrin has been shown to be stable (Stibler, Borg & Joustra, 1986, and unpublished observations), and there was no
B rain transferrin in alcoholics
339
Table 1. Details of alcoholic cases and controls Age, years
pm delay, hours
Alcohol intake
Brain pathology
Liver pathology
58 44
15 24
< 20 g /day < 20 g /day
NAD NAD
Normal U nknown
3
62
8
nil
NAD
4
53
8
2 ± 3 g /day
Old infarct in globus pallidus
NAD Steatosis, congestion M ild steatosis, congestion Congestion
Alcoholics 5 58
80
80 g /day
Mild steatosis
Normal
Congestion
Normal
Case no. Controls 1 2
Liver function test
Normal Normal
Cause of death Acute peritonitis Acute m yocardial infarct Adenocarcinoma of ú sophagus Ischñ m ic heart disease
6
56
48
80 g /day
Ventricular dilatation NAD
7
58
29
> 100 g /day
NAD
Congestion
Normal
8
38
22
> 80 g /day
NAD
Congestion, steatosis
Normal
200 g /day
NAD
Bilirubin 32, albumin 31
Intraperitoneal hñ m orrhage
Ventricular dilatation Vascular m alformation Ventricular dilatation
Micronodular cirrhosis, steatosis, cholestasis Cirrhosis, cholestasis Micronodular cirrhosis Micronodular cirrhosis
Bilirubin 66, albumin 30 Bilirubin 23, albumin 43 Unavailable
Septic shock
Alcoholics with Cirrhosis 9 59 46
10
58
17
> 80 g /day
11
56
80
80 g /day
12
44
15
> 80 g /day,?¯
Bronchopneumonia Chronic renal failure Acute m yocardial infarct Status postoperative
Bronchopneumonia Chronic liver disease
All cases were m ale. NAD , no abnormality detected. For case 9, alcohol consumption was stated as ``approaching 200 g /day’’ in the questionnaire response; for case 12, there was evidence that consumption had dropped in the 2 ± 3 weeks immediately preceding the patient’s death. U nits for serum bilirubin were l m ol /L, and the normal reference range was 0 ± 18; units for serum albumin were g /L, and the normal reference range was 40 ± 52.
discernible correlation between storage time and T f concentration. No serum samples were available for the cases studied.
Transfer rin assay T f and its carbohydrate-m odi® ed congener were assayed by the method developed by Stibler et al. (1986), as modi® ed by Kwoh-Gain et al. (1990). As this method had been optimized for the analysis of serum samples, minor adaptation was required for the analysis of brain samples. Samples (0.1 ± 0.3 g) of tissue were chipped from the blocks of frozen, un® xed tissue (Scott et al. , 1995) and rapidly thawed by immersion in a large volume of warm 0.32 M sucrose, then immediately brought to 4ë C in an ice-bath. The samples were homogenized at room temperature in 1 vol of 50 m M HEPES buŒer pH 7.4 contain-
ing KCl and ovalbumin, and the extracts clari® ed by centrifugation. Disposable plastic m icrocolumns were packed with 200 l L of DEAESephacel and equilibrated with 2- N -morpholinoethanesulphonic acid (M ES) buŒer, pH 5.65. To eliminate the eŒect of partial iron saturation on isoelectric pro® le (Stibler et al. , 1986), 100 l L samples of homogenate were saturated with iron by the addition of 10 l L of 25 m M ferric nitrilotriacetate (FeNTA). They were diluted with 890 l L of MES buŒer and dialysed against M ES buŒer at room temperature for 2 hours. To bring the T f concentration into the most sensitive range of the standard curve, the dialysate volume was reduced to 400 l L with the aid of Aquacide. The carbohydrate-de® cient form (CD T) was isolated from the extract by applying it to a DEAESephacel column and eluting with 1 mL of M ES buŒer. Norm al transferrin is retained in the
340
Peter R. Dodd et al.
Table 2. B rain total transfer rin in dem entia cases
programs exactly as described previously (Dodd et al. , 1992).
Assay Case no. 13 14 15 16 17
Block 1 2 1 2 1 1 1
1
2
3
4
27 51 124 90 141 245 99
17 37 106 28 63 241 111
36
20
Tf assays were carried out and analysed as described in Methods. For each case, block 1 was taken from the mid-temporal gyrus and block 2 from the pre-central gyrus of the m otor cortex, from clinically diagnosed DAT cases w hose diagnosis was ultimately established at autopsy. See Methods section for case details. The multiple assays indicated were each carried out on entirely independent tissue extracts, i.e. a separate tissue sample was cut, thawed and taken through the full extraction and analytic process. See Methods. Values quoted are in l g of immunoreactive Tf per g of tissue.
column because of negatively charged sialate groups, while the carbohydrate-de® cient form passes through unhindered. Immunoreactive T f was assayed in both this eluate (CDT) and the original extract (total T f ) by radioimm unoassay (R IA) against [ 1 2 5I]T f and unlabelled T f standards. The method was based on that of Anderson et al. (1986), with minor modi® cations. The procedure uses anti-Tf immunoglobulin insolubilized on Matrix-Pel 102. In contrast to the published method (Anderson et al., 1986), the initial Sephadex G-25M column was eluted with borate buŒer rather than water, and the HEPE S-KC l buŒer did not contain detergent. To prepare [ 1 2 5I]T f , 100 l g of human T f was iodinated by the Bolton± Hunter technique 125 (Bolton & Hunter, 1973). Free I and unreacted reagent were removed by chromatography on Sephadex G-25M in PD10 columns. A typical displacement curve is shown in Fig. 1, which also illustrates that such curves were amenable to more accurate analysis with the aid of probit transforms of the values (see below).
[ 3 H ]-D iazepam recep tor binding site assay In some cases measurements of benzodiazepine binding site B M A X values were m ade in the same tissue samples from the alcoholic cases used to assay T f and C DT. The assays were performed and analysed with the aid of the EBDA and LI GAND
Protein Total protein in the samples was assayed by the method of Lowry et al. (1951) against bovine serum albumin standards.
M aterials Bolton± Hunter labelling kit was obtained from Amersham (Sydney, NSW, Australia). DEAESephacel, Sephadex G-25M , PD 10 columns and plastic microcolumns were purchased from Pharmacia (Uppsala, Sweden). M atrix-Pel 102 was purchased from Amicon Corp (Lexington, MA, USA). Puri® ed human T f and Aquacide were obtained from Calbiochem-Behring P /L (Sydney), and anti-Tf immunoglobulin from Silenius Laboratories P /L (Dandenong, Vic, Australia). All other chemicals were obtained from Sigma Chemicals (St Louis, MO, USA) or BDH Chemicals (Kilsyth, Vic) and were of the highest grade of purity available. A stock FeNTA solution was prepared by m ixing 25 m M ferric citrate with 50 m M nitrilotriacetic acid in 200 mL of water. The pH was adjusted to 4.0 with Na 2 CO 3 and the mixture heated until dissolved. Tris(hydroxymethyl)methylamine (Tris base) was then added to bring the pH to 6.5 and the solution stored at 0 ± 4ë C until use.
Data analysis The values from the T f R IA were transformed into probits to linearize the standard curve (Fig. 1). The line of best ® t was obtained by least squares analysis with the aid of the Cricket Graph TM III program (Computer Associates International, Islandia, NY, USA), and the values for unknowns obtained by back-substitution of their probit values into the regression equation. Parametric statistical tests (Winer, Brown & Michels, 1991) were performed with the aid of the SystatT M program (Systat Inc., Evanston, IL, USA).
R esults and discussion In con® rm ation of earlier work (M orris et al., 1992b), human cerebral cortex tissue obtained at autopsy was found to contain high concentrations
B rain transferrin in alcoholics
341
1 25 Figure 1. Transferrin RIA standard cur ve. The T f assay was car ried out as described in the Methods section using [ I]T f displaced by a range of concentrations of unlabelled T f. Unknowns were diluted so as to fall as near as possible to the central region of the cur ve. For quantitation, percentage binding values were converted to probits as shown in the inset.
of T f (Fig. 2, Table 2). Measurem ents m ade in slow-frozen samples by an R IA based on a highly speci® c antibody were somewhat lower than previously reported values based on nephelometry (M orris et al., 1987) or ELISA (Loeç er et al. , 1995) in snap-frozen material. This is consistent with the improved speci® city of the present method (Kwoh-Gain et al. , 1990). The values for CDT : T f ratios in brain (Fig. 4) were some 10fold higher than we have previously observed in human serum samples (Kwoh-Gain et al., 1990; Fletcher et al., 1991), i.e. a much higher proportion of brain tissue T f was of the carbohydratede® cient form (Brunngraber & Webster, 1986; HoŒmann et al. , 1995), irrespective of the disease group to which the case belonged (Fig. 4). As Fig. 2 shows, the tissue T f concentration showed no correlation, nor any detectable tendency to decline, with either the age of the patient at death or the delay between death and autopsy.
When independent extracts were made of the same tissue sample, the T f concentrations were generally in good agreement, with some exceptions (Table 2). However, total T f varied over a more than 14-fold range when diŒerent cortical areas from the same cases, or the same cortical region from diŒerent cases, were compared (Table 2). Thus, the between-subject percentage coe cient of variation was some 2.3-fold greater than the equivalent within-sample value. Figure 2 shows the extent of the between-subject variation. In contrast to some previous studies, we did not ® nd a signi® cant change, either a decrease (Morris et al. , 1987; Connor et al. , 1992b) or an increase (Loeç er et al. , 1995), in either form of T f in dementia cases. The average for dementia cases (Table 2) was 114 6 35 l g of total immunoreactive T f per g of tissue, slightly higher than but not signi® cantly diŒerent from the average value in controls (Fig. 3): t 7 5 0.88, p 5 0.41.
342
Peter R. Dodd et al.
Figure 2. EŒect of age and post-m ortem delay on total T f concentrations. Total immunoreactive T f was extracted and assayed from samples of cerebral cortex taken from the cases listed in Tables 1 and 2. Plotted values are the mean T f value calculated across all brain regions assayed for each case. (A) eŒect of post-m ortem delay ( pmd); (B ) eŒect of age at death. Neither correlation approached statistical signi® cance ( pm d: r15 5 0.05, NS; age: r15 5 0.07, NS): the regression lines are shown only as a guide.
B rain transferrin in alcoholics
343
Figure 3. Total T f concentrations in the brains of alcoholics. Total immunoreactive T f was extracted and assayed from sam ples of super ior frontal and primar y motor cortex taken from the cases set out in Table 1. The mean 6 SEM values shown were calculated from the means of the values in the two cortical regions from each case, of the num bers of cases shown in Table 1. *, signi® cantly diŒerent from the average across the other two groups, t 10 5 2.649, p 5 0.025.
Note that the highest cortical T f concentration occurred in the single diŒuse Lewy body disease case (Table 2): Lewy bodies are also a feature of Parkinson disease, where subcortical T f abnormalities have been postulated to play a pathogenic role (Loeç er et al. , 1995). If this case were removed from the analysis, the average total immunoreactive T f value in the other four dementia cases was 82 6 16 l g per g of tissue, comparable to the average value in controls of 79 6 10 l g. It is also noteworthy that the m ean value in case 17 (Table 2), which lacked senile plaques (as well as lacking tangles and Lewy bodies, see M ethods) was close to the average value in normal controls. The reproducibility of results usually observed when independent extracts were made from the same sample suggests that the high degree of variability in T f concentration is a biological phenomenon, and may help explain the con¯ icting results which have been reported, including data from a single
research group (Connor et al. , 1992b; Loeç er et al. , 1995). Detailed analysis of the autopsy reports and histological material gave no indication of any reason for the variations observed across cases 13 ± 15 (Table 2). Mean total T f was found to be slightly, but nonsigni® cantly, increased in the brains of chronic alcoholics compared with non-alcoholic controls, and this increase was more marked in alcoholics with liver disease (Fig. 3). An analysis of variance based on the three groups and two brain areas lacked power because of the small group sizes (Table 1), although it has been noted that it is di cult to assess statistical power in m ultifactorial designs (W iner, Brown & Michels, 1991). However, mean total T f concentration for the combined cortical areas in the brains of cirrhotic alcoholics was signi® cantly diŒerent from the average concentration in non-cirrhotic (alcoholic or non-alcoholic) cases, which did not diŒer one from another (Fig. 3). Calculation
344
Peter R. Dodd et al.
(Winer et al., 1991) showed that the t test used to compare cirrhotic and non-cirrhotic cases had a power of approximately 0.8 to detect a diŒerence at least this large at p < 0.05 with groups of the size used (Fig. 3). Although it has been argued that T f concentration m ay be raised in Alzheimer’ s or Parkinson’s disease as part of a selective excitotoxic mechanism, the result with cirrhotic alcoholics suggests that this may not be so, since excitotoxicity is not generally thought to be involved in the brain damage associated with cirrhosis. Rather, raised total T f m ay be a m ore general indication of tissue damage. One possibility is that it re¯ ects the astrocytosis concomitant to alcoholism (Dodd et al., 1992), since reactive astrocytes are also commonly found in neurodegenerative disease. Astrocytosis is m ore m arked and widespread in cirrhotic than in non-cirrhotic alcoholics, and this m ay be combined with a redistribution of T f from its near-exclusive localization in oligodendrocytes in normal brain (Gerber & C onnor, 1989; M artin et al. , 1991; Morris et al. , 1992a), to a pathological appearance in other cells, especially astrocytes, as has been reported in DAT (Connor et al. , 1992a). Whether increased total T f speci® cally re¯ ects the appearance of the Alzheimer Type II astrocytes pathognomonic of hepatic encephalopathy remains to be determined. It would be especially valuable if future studies could also exam ine nonalcoholic cirrhotic cases, to help clarify this point. However, it should be noted that a proper quantitation of Alzheimer Type II astrocytes is di cult to achieve, because these cells occur sparsely in most brain areas even in severe cases of hepatic encephalopathy; because they are di cult to identify with conventional histopathological stains; and because there appears to be no truly speci® c histological or immunocytochemical marker for them. Detailed examination of the data revealed that large biological variability was not con® ned to dementia cases: two of the cirrhotic alcoholic cases had mean T f levels which overlapped with those of non-cirrhotic cases, while the other two showed markedly higher values. Thus, even within this closely matched set of cases, brain total T f concentrations varied over an almost 9-fold range (data not shown). Further work will be required to con® rm the robustness of the ® nding shown in Fig. 3. There was no indication that brain areas which show a greater degree of pathological damage in chronic alcoholism were endowed with a variant
T f concentration (Main EŒect for Area in the analysis of variance: F 1 ,9 5 0.65, p 5 0.44). There was a non-signi® cant decrease in the CDT : T f ratio in alcoholics, which appeared more marked if liver disease was present (Fig. 4). However, the Groups term fell well short of statistical signi® cance (F 2 ,9 5 0.50, P 5 0.62). There was no evidence that the changes were greater in the pathologically susceptible region (superior frontal) than in the spared region ( primary motor) of the cerebral cortex (Fig. 4): M ain EŒect for Area, F 1 ,9 5 0.57, p 5 0.47; Group 3 Area Interaction, F 2 ,9 5 0.06, p 5 0.94. It should be noted that the direction of change in CDT : T f ratio is opposite to that predicted from studies of CDT : T f ratios in the serum of chronic alcoholics (Kwoh-Gain et al. , 1990; Fletcher et al., 1991; Stibler, 1993): that is, there was no increase in CD T : T f ratio in the brains of alcoholics. Hence, it would appear to be unlikely that either brain C DT or brain T f are derived exogenously to any substantial extent ( JeŒeries et al. , 1984; Pardridge, Eisenberg & Yang, 1987; Loeç er et al., 1995; Recht et al. , 1990), at least in pathological conditions. It should also be noted that, whereas total imm unoreactive T f is little altered in the serum of chronic alcoholics, in contrast to the marked changes in C DT : T f ratio (KwohGain et al., 1990; Fletcher et al. , 1991), the reverse was true for brain total T f and CD T (Figs 3, 4). A further test on the hypothesis that brain total T f , or brain CDT : T f ratio, might re¯ ect the localized eŒect of alcoholic status ante m ortem was provided by comparing these values with measures of GABA A-benzodiazepine receptor binding sites, since previous work has shown that the densities of these sites are altered in parallel with the extent of tissue damage assessed morphometrically (Dodd et al. , 1992, 1996). Neither parameter correlated with the density of ``centraltype’’ benzodiazepine receptor sites labelled with [ 3 H]-diazepam (Brain C DT : T f ratio is shown in Fig. 5). Indeed, there was no correlation between either T f measure and any index of GABA Abenzodiazepine receptor site density (data not shown): sites were labelled with [ 3 H]-muscimol, [ 3 H]-diazepam or [3 H]-¯ unitrazepam (Dodd et al., 1992). It should be emphasized that all the cited assays, including those using [ 3 H]diazepam, were carried out under conditions which eliminated binding to the ``peripheraltype’’ benzodiazepine binding site (Dodd et al.,
B rain transferrin in alcoholics
345
Figure 4. CDT:T f ratios in the brains of alcoholics. Total immunoreactive T f and CDT were extracted and assayed from sam ples of super ior frontal and primar y motor cortex taken from the cases set out in Table 1. Values are means 6 SEM of the numbers of cases shown in Table 1; u , super ior frontal cortex; j , primar y motor cortex. Statistical analysis of this data is given in the text.
1992). It would be of interest if future studies could exam ine the relationship between brain T f or CD T concentration and the densities of ``peripheral-type’ ’ sites, which are thought to be con® ned to astrocytes. Brain T f has been widely studied because of its potential role in the pathogenesis of neurodegenerative diseases. Support for the idea that T f may moderate the Al3 + -mediated toxicity which has been proposed to underlie the development of senile plaques comes from the ® nding that brain areas accumulating the highest concentrations of aluminium in experimental m odels of Al toxicity have the highest concentrations of T f receptors (M orris et al. , 1989; Edwardson et al., 1992; Florence et al., 1994). However, it has been reported that T f receptors are reduced, rather than increased, in pathologically signi® cant brain areas from DAT cases (Kalaria et al., 1992; Morris et al., 1994a); and recent work has suggested that brain Al concentration is not, in fact, raised in DAT (Edwardson et al. , 1992). Similarly, a role for T f -mediated toxicity in Parkinson disease is at variance with the tissue distri-
butions of T f receptors, and the direction of change of T f receptor density, in relation to pathological damage (M ash et al., 1991; Morris et al., 1994b). The data reported here provide some support for the notion that increased T f concentration may be a weak marker of reactive astrocytosis; however, there is little support for the idea that human brain T f is derived primarily from serum, since the relative concentrations of T f subtypes (Fig. 4) did not re¯ ect known variations in the serum levels of these substances in alcoholic vs. non-alcoholic cases (Stibler, 1993). This is in line with work showing a wide species variation in T f gene expression in brain, including an apparent lack of expression in human brain (Tu et al. , 1991). It is concluded that whereas serum CD T : T f ratio is now the marker of choice for alcoholic status, this is not re¯ ected in the brain CDT : T f ratio. Similarly, total brain T f concentration appears not to be useful as a marker of dementia or alcoholism. However, total imm unoreactive T f was strongly increased in cirrhotic alcoholic cases, probably as a consequence of glial proliferation,
346
Peter R. Dodd et al.
3
Figure 5. R elationship between CDT:T f ratios and [ H ]-diazepam binding site densities in the brains of alcoholics. The respective parameter s were assayed on the same tissue sam ples, each der ived from the super ior frontal and primar y motor cortex taken from the cases set out in Table 1. The regression lines are shown for comparison only, since neither correlation was signi® cant: ± Ð u , super ior frontal cortex, r 10 5 0.33, NS; ± ± ± s , primar y motor cortex, r 10 5 0.22, NS.
and should be a valuable index in studies of the eŒect of cirrhosis of the liver on brain structure and function.
Acknow ledgem ents This work was supported ® nancially by grants from the NHMRC and the AAB. PRD is an NHM RC Senior Research Fellow. We are indebted to Dr Tony Tannenberg and the pathology registrars at the Royal Brisbane and M ater Misericordiñ hospitals for performing the autopsies and making the histopathological assessments of the dementia cases. We wish to thank Dr Dean 3 Moss for helpful discussion. The data on [ H]diazepam binding site density was kindly provided by M r Greg Thomas.
References A NDERSON, G. J., M ACKERRAS, A., M ACK, U., P OWELL, L. W. & H ALLIDAY, J. W. (1986) Simple radioimm unoassay for transferrin using insolubulized antitransferrin antibodies: its application to cultured cells, Journal of Laborator y Clinical Medicine, 107, 59 ± 65. B EAN, P., S UTP HIN, M . S., L IU, Y. S., A NTON, R., R EYNOLDS, T. B., S HOENFELD, Y. & P ETER, J. B.
(1995) Carbohydrate-de® cient transferrin and falsepositive results for alcohol abuse in primary biliary cirrhosis: diŒerential diagnosis by detection of mitochondrial autoantibodies, Clinical Chemistr y, 41, 858 ± 861. B ELL, H ., TALLAKSEN, C., S JAÊ HEIM , T. et al. (1993) Serum carbohydrate-de® cient transferrin as a m arker of alcohol consumption in patients with chronic liver diseases. Alcoholism: Clinical and Experim ental R esearch, 17, 246 ± 252. B OLTON, A. E. & H UNTER, W. M. (1973) The labelling of proteins to high speci® c activities by conjugation to a 1 2 5I-containing acylating agent. B iochemical Journal, 133, 529 ± 539. B RUNNGRABER, E. G. & W EBSTER, J. C. (1986) Changes in glycoprotein carbohydrate content in the aging hum an brain, Neurochem ical Research , 11, 579 ± 588. C ARLSSON, A. V., H ILTUNEN, A. J., B ECK, O., S TIBLER, H . & B ORG, S. (1993) Detection of relapses in alcohol-dependent patients: comparison of carbohydrate-de® cient transferrin in serum, 5-hydroxytryptophol in urine, and self-reports, Alcoholism: Clinical and Experim ental R esearch, 17, 703 ± 708. C ONNOR, J. R., M ENZIES, S. L., S T M ARTIN, S. M . & M UFSON, E. J. (1992a) A histochemical study of iron, transferrin, and ferritin in Alzheim er’s diseased brains, Journal of Neuroscience Research , 31, 75 ± 83. C ONNOR, J. R., S NYDER, B. S., B EARD, J. L., F INE, R. E. & M UFSON, E. J. (1992b) Reg ional distribution of iron and iron-regulatory proteins in the brain in aging and Alzheim er’s disease, Journal of Neuroscience R esearch, 31, 327 ± 335. D ODD, P. R., T HOMAS, G. J., H ARPER, C. G. & K RIL,
B rain transferrin in alcoholics J. J. (1992) Am ino acid neutrotransmitter receptor changes in cerebral cortex in alcoholism : eŒect of cirrhosis of the liver, Journal of Neurochemistry, 59, 1506 ± 1515. D ODD, P. R., S COTT, H . L. & W ESTPHALEN, R. I. (1994) Excitotoxic mechanism s in the pathogenesis of dementia, Neurochem istry Inter national, 25, 203 ± 219. D ODD, P. R., K RIL, J. J., T HOMAS, G. J., W ATSON, W. E. J., J OHNSTON, G. A. R. & H ARPER, C. G. (1996) Receptor binding sites and uptake activities m ediating GABA neurotransmission in chronic alcoholics with Wernicke encephalopathy, B rain Research, 710, 215 ± 228. E CKERT, A. L., D ODD, P. R., C OOKE, R. W., K RIL, J. J. & H ARPER, C. G. (1991) Transferrin and desialyltransferrin in post m ortem human brain from Alzheim er disease and chronic alcoholic cases, Journal of Neurochem istry, 57 Suppl, S88B (Abstract). E DWARDSON, J. A., C ANDY, J. M., I NCE, P. G. et al. (1992) Alum inium accumulation, b -amyloid deposition and neuro® brillary changes in the central nervous system, Ciba Foundation Symposia, 169, 165 ± 179. F LETCHER, L. M., K WOH -G AIN, I., P OWELL, E. E., P OW ELL, L. W. & H ALLIDAY, J. W. (1991) M arkers of chronic alcohol ingestion in patients with nonalcoholic steatohepatitis: an aid to diagnosis, H epatology, 13, 455 ± 459. F LORENCE, A. L., G AUTHIER, A., P ONSAR, C., VAN DEN B OSCH - DE A GUILAR, P. & C RICHTON, R. R. (1994) An experimental animal m odel of aluminium overload, Neurodegeneration, 3, 315 ± 323. G ERBER, M. R. & C ONNOR, J. R. (1989) Do oligodendrocytes m ediate iron regulation in the hum an brain? Annals of Neurology, 26, 95 ± 98. H OFFMANN, A., N IMTZ, M ., G ETZLAFF, R. & C ONRADT, H. S. (1995) `Brain-type’ N-glycosylation of asialotransferrin from human cerebrospinal ¯ uid, FEB S Letters, 359, 164 ± 168. JEFFERIES, W. A., B RANDON, M . R., H UNT, S. V., W ILLIAMS, A. F., G ATTER, K. C., & M ASON, D. Y. (1984) Transferrin receptor on endothelium of brain capillaries, Nature, 313, 162 ± 163. K ALARIA, R. N., S ROMEK, S. M., G RAHOVAC, I. & H ARIK, S. I. (1992) Transferrin receptors of rat and hum an brain and cerebral m icrovessels and their status in Alzheim er’s disease, B rain Research, 585, 87 ± 93. K WOH-G AIN, I., F LETCHER, L. M., P RICE, J., P OWELL, L. W. & H ALLIDAY, J. W. (1990) Desialylated transferrin and mitochondrial aspartate aminotransferase compared as laboratory m arkers of excessive alcohol consumption, Clinical Chemistr y, 36, 841 ± 845. L IEBER, C. S., X IN, Y., L ASKER, J. M. & R OSMAN, A. S. (1993) Comparison of new m ethods for measuring carbohydrate-de® cient transferrin (C DT): application to a public health approach for the prevention of alcoholic cirrhosis, A lcohol and Alcoholism Supplement, 2, 111 ± 116. L OEFFLER, D. A., C ONNOR, J. R., JUNEAU, P. L. et al. (1995) Transferrin and iron in normal, Alzheim er’s disease, and Parkinson’ s disease brain regions, Journal of Neurochem istry, 65, 710 ± 724. L OWRY, O. H ., R OSENBROUGH, N. J., F ARR, A. L. &
347
R ANDALL, R. J. (1951) Protein m easurement with the Folin± phenol reagent, Journal of B iological Chemistr y, 193, 265 ± 275. M ARTIN, S. M., L ANDEL, H . B., L ANSING, A. J. & VIJAYAN, V. K. (1991) Immunocytochemical double labelling of glial ® brillary acidic protein and transferrin permits the identi® cation of astrocytes and oligodendrocytes in the rat brain, Journal of Neuropathology and Exper imental Neurology, 50, 161 ± 170. M ASH, D. C., P ABLO, J., B UCK, B. E., S ANCHEZ-R AMOS, J. & W EINER, W. J. (1991) Distribution and number of transferrin receptors in Parkinson’s disease and in M PTP-treated m ice, Exper imental Neurology, 114, 73 ± 81. M ORRIS, C. M., C OURT, J. A., M OSHTAGHIE, A. A. et al. (1987) Transferrin and transferrin receptors in norm al brain and Alzheim er’ s disease, B iochem ical Society Transactions, 15, 891 ± 892. M ORRIS, C. M., C ANDY, J. M., O AKLEY, A. E. et al. (1989) Comparison of the regional distribution of transferrin receptors and aluminium in the forebrain of chronic renal dialysis patients, Journal of the Neurological Sciences, 94, 295 ± 306. M ORRIS, C. M ., C ANDY, J. M ., B LOXHAM, C. A. & E DWARDSON, J. A. (1992a) Immunocytochemical localization of transferrin in the hum an brain, Acta A natomica (B asel), 143, 14 ± 18. M ORRIS, C. M., C ANDY, J. M ., K EITH, A. B. et al. (1992b) Brain iron hom eostasis, Journal of Inorganic B iochem istr y, 47, 257 ± 265. M ORRIS, C. M., C ANDY, J. M ., K ERWIN, J. M. & E DWARDSON, J. A. (1994a) Transferrin receptors in the normal human hippocam pus and in Alzheimer’ s disease, Neuropathology and Applied Neurobiology, 20, 473 ± 477. M ORRIS, C. M ., C ANDY, J. M ., O MAR, S., B LOXHAM, C. A. & E DWARDSON, J. A. (1994b) Transferrin receptors in the parkinsonian midbrain, Neuropathology and A pplied Neurobiology, 20, 468 ± 472. P ARDRIDGE, W. M., E ISENBERG, J. & YANG, J. (1987) H uman blood± brain barrier transferrin receptor, M etabolism , 36, 892 ± 895. R ECH T, L., T ORRES, C. O., S M ITH, T. W., R ASO, V. & G RIFFIN , T. W. (1990) Transferrin receptor in norm al and neoplastic brain tissue: implications for brain-tum our immunotherapy, Journal of Neurosurgery, 72, 941 ± 945. R OSKAMS, A. J. & C ONNOR, J. R. (1990) Aluminium access to the brain: a role for transferrin and its receptor, Proceedings of the National Academy of Sciences of the USA, 87, 9024 ± 9027. S CHELLENBERG, F., B ENARD, J. Y., L E G OFF, A. M ., B OURDIN, C. & W EILL, J. (1989) Evaluation of carbohydrate-de® cient transferrin compared with Tf index and other m arkers of alcohol abuse, A lcoholism: Clinical and Experim ental R esearch, 13, 605 ± 610. S COTT, H. L., TANNENBERG, A. E. G. & D ODD, P. R. (1995) Variant forms of neuronal glutamate transporter sites in Alzheim er’s disease cerebral cortex, Journal of Neurochemistr y, 64, 2193 ± 2202. S TIBLER, H., B ORG, S. & J OUSTRA, M. (1986) Micro anion exchange chromatography of carbohydratede® cient transferrin in serum in relation to alcohol consumption, Alcoholism: Clinical and Experim ental R esearch, 10, 535 ± 544.
348
Peter R. Dodd et al.
S TIBLER, H. (1991) Carbohydrate-de® cient transferrin in serum: a new marker of potentially harmful alcohol consumption reviewed, Clinical Chem istr y, 37, 2029 ± 2037. S TIBLER, H. (1993) Diagnosis of alcohol-related neurological diseases by analysis of carbohydrate-de® cient transferrin in serum, Acta Neurologica Scandinav ica, 88, 279 ± 283. T HOMPSON, C. M ., M ARKESBERY, W. R., E HMANN, W. D., M AO, Y. X. & VANCE, D. E. (1988) Reg ional
brain trace-element studies in Alzheim er’s disease, Neurotoxicology, 9, 1 ± 7. T U, G. F., A CHEN, M. G., A LDRED, A. R., S OUTHWELL, B. R. & S CHREIBER, G. (1991) T he distribution of cerebral expression of the transferrin gene is species speci® c, Journal of B iological Chem istry, 266, 6201 ± 6208. W INER, B. J., B ROWN, D. R. & M ICHELS, K. M. (1991) Statistical Principles in Exper imental Design, pp. 497 ± 582 (Sydney, NSW, M cGraw-H ill).
R ESEA R CH A RT IC LE
Concentrations of transferrin and carbohydrate-de® cient transferrin in postm ortem hum an brain from alcoholics PETER R. DODD, ALLISO N L. ECKERT, LINDA M . FLETCH ER, 2 2 1 JILLIAN J. KR IL, CLIVE G. HARPER & JUNE W. H ALLIDAY
1
1
Clinical Research Laborator y, Royal B risbane Hospital Research Foundation, Liver Unit, 2 Queensland Institute of M edical Research, D epartment of Patholog y, University of Sydney and Departm ent of Anatom ical Patholog y, Royal Prince Alfred Hospital, Sydney, Australia
Abstract Transfer rin (T f ) and its carbohydrate-de® cient isofor m (CDT) were m easured by radioimm unoassay in phosphate-bu Œered saline extracts of two infor mative areas of cerebral cortex tissue obtained at autopsy from alcoholics without other associated disease ( n 5 4); alcoholics with cir rhosis of the liver ( n 5 4) and agematched controls ( n 5 4). Tota l T f was also m easured in two infor m ative cortical areas from ® ve dem entia cases. All cases were m ale. Total im m unoreactive T f was assayed directly in the extract, CDT im m unorea ctivity in the concentr ated eluate after the sialylated form was removed by passing through DE AE-S ephacel at pH 5´65. B rain CDT averaged 10% of total T f overall. Although replicate extractions of individual sam ples gave consiste nt assays for both substances, there was wide variation both between di Œerent cortica l areas from a given case and between cases within groups. There were no signi® cant di Œerences between total T f levels in uncom plicated alcoholics, dem entia cases and controls, but cir rhotic alcoholics gave signi® cantly higher values. The CDT : T f ratio was not increased in the brains of either group of alcoholics com pared to controls. W hereas the ser um CDT : T f ratio is an excellen t marker of recent alcohol consum ption, brain T f and CDT concentrations do not m ark alcoholism nor dem entia, and their biolog ical variability diminishes their usefuln ess as disease indices. H owever, brain T f m ay be a m arker of cir rhosis-ind uced changes.
Introdu ction Markers of excessive alcohol consumption have been developed to augment the diagnosis of chronic alcoholism, which is based mainly on clinical history and responses to questionnaires combined with laboratory investigations of limited selectivity and sensitivity (Schellenberg et al. , 1989; Stibler, 1993; Lieber et al. , 1993; Carlsson
et al., 1993). Alcoholic status is particularly di cult to determine in cases where only autopsy tissue is available. Estimates of the average antemortem daily alcohol consumption of deceased cases m ust be derived from an examination of medical records together with responses to questionnaires by relatives of the deceased. A reliable, objective index of alcoholism would be of value
Correspondence to: Dr Peter R. Dodd, Clinical Research Laboratory, Royal Brisbane Hospital Research Foundation, Br isbane 4029, Australia. Fax: + 61-73-362 0108; e-mail: peterD@ qimr.edu.au Received for publication 22nd July 1996. Accepted 20th October 1996. 1355 ± 6215/97/030337 ± 12 Society for the Study of Addiction to Alcohol and Other Drugs
338
Peter R. Dodd et al.
for classifying such m aterial, especially if it varied between brain areas according to their susceptibility to damage (Dodd et al., 1992). A marker which has become widely accepted for studies of living cases is the carbohydratede® cient form (CDT ) of the iron-binding protein transferrin (T f ). CD T is a desialylated transferrin with an isoelectric point between 5.7 and 5.9. It has been found in the serum of alcoholic subjects and has been shown to be raised after the longterm ingestion of alcohol (Stibler, 1991). The marker allows alcoholic and non-alcoholic liver disease cases to be distinguished, since it is not raised in serum from the latter group (KwohGain et al., 1990; Fletcher et al. , 1991), except in a few cases of primary biliary cirrhosis (Bell et al. , 1993; Bean et al., 1995). Transferrin is also a carrier for aluminium, which has been thought to accumulate in senile plaques and thus contribute to the ñ tiology of Alzheimer-type dementia (DAT) (Roskams & Connor, 1990; Edwardson et al. , 1992). High levels of transferrin have been reported to be present in the human brain (Morris et al., 1987). However, although free-radicalmediated excitotoxicity, which would be promoted by reactive iron species, has frequently been invoked as a potential pathogenic mechanism underlying the brain damage associated with alcohol abuse, and in neurodegenerative diseases such as Parkinson disease or DAT (Dodd et al. , 1994), evidence on the concentrations of brain iron and T f in degenerative brain diseases has been con¯ icting (M orris et al. , 1987; Thompson et al. , 1988; Connor et al. , 1992a,b; Loeç er et al. , 1995). We studied whether the concentration of either form of T f was altered in cerebral cortex tissue obtained at autopsy from well-characterized chronic alcoholic cases with and without associated liver disease. Key neurotransmitter receptor sites have been shown to correlate with histopathological indices of localized cortical damage. We were interested to assess whether the densities of these sites might also correlate with C DT or T f concentrations, or with CD T : T f ratios, which would hence be of value as pathogenic m arkers. To provide a positive disease control for the study, we also ascertained whether T f concentrations were altered in the brains of dementia cases. Since some of the disparities in earlier work may have derived from methodological diŒerences, we used a highly speci® c assay procedure to quantify the T f species (Kwoh-Gain et al., 1990; Fletcher
et al. , 1991). Unfortunately, due to di culties in obtaining post-mortem tissue and because of the complexity of the analytical technique, only a limited number of cases were available for study. A preliminary report of this work has appeared (Eckert et al. , 1991). To the best of our knowledge, this is the ® rst report on the concentration of C DT, or on its regional distribution, in human brain.
M ethods C ase selectio n Tissue was obtained from alcoholics with or without pathologically con® rmed cirrhosis, and from matched controls (Table 1). A further ® ve cases had clinically diagnosed dem entia (Table 2). At autopsy, the latter cases were classi® ed under criteria which have been given elsewhere (Scott, Tannenberg & Dodd, 1995). These cases were somewhat older than the ® rst set (m edian age 70 years vs. 57 years for the cases in Table 1), and the post-mortem delay was somewhat longer (median 48 hours vs. 23 hours for Table 1 cases). Three of the cases (13 ± 15) were diagnosed as Alzheimer’ s disease, with abundant senile plaques and neuro® brillary tangles combined with neuronal loss (Scott et al. , 1995). Case 16 also had abundant senile plaques, but tangles were generally absent and numerous cortical Lewy bodies were seen: this is a case of diŒuse Lew y body disease . Case 17 had neither plaques, nor tangles, nor Lewy bodies, nor neuronal loss: it is an exam ple of fronta l lobe dem entia lacking distinctive histologic al features , and m ay be regarded as a disease control for excessive plaque deposition. Inform ed consent for autopsy was obtained from the next of kin in all cases; ethical clearance for the project is current under RBH protocol No. 1992 /22. In all cases, tissue was obtained both from a cortical region which is susceptible to damage in the disease (for alcoholics, superior frontal cortex; for dementia cases, m id-temporal cortex), and from a relatively spared cortical region ( primary motor cortex in all cases) (Dodd et al., 1994). Tissue processing and storage for both neurochemical studies and histochemical analyses was carried out as previously described (Scott et al. , 1995). Tissue storage time at 2 70ë C was well within the range over which transferrin has been shown to be stable (Stibler, Borg & Joustra, 1986, and unpublished observations), and there was no
B rain transferrin in alcoholics
339
Table 1. Details of alcoholic cases and controls Age, years
pm delay, hours
Alcohol intake
Brain pathology
Liver pathology
58 44
15 24
< 20 g /day < 20 g /day
NAD NAD
Normal U nknown
3
62
8
nil
NAD
4
53
8
2 ± 3 g /day
Old infarct in globus pallidus
NAD Steatosis, congestion M ild steatosis, congestion Congestion
Alcoholics 5 58
80
80 g /day
Mild steatosis
Normal
Congestion
Normal
Case no. Controls 1 2
Liver function test
Normal Normal
Cause of death Acute peritonitis Acute m yocardial infarct Adenocarcinoma of ú sophagus Ischñ m ic heart disease
6
56
48
80 g /day
Ventricular dilatation NAD
7
58
29
> 100 g /day
NAD
Congestion
Normal
8
38
22
> 80 g /day
NAD
Congestion, steatosis
Normal
200 g /day
NAD
Bilirubin 32, albumin 31
Intraperitoneal hñ m orrhage
Ventricular dilatation Vascular m alformation Ventricular dilatation
Micronodular cirrhosis, steatosis, cholestasis Cirrhosis, cholestasis Micronodular cirrhosis Micronodular cirrhosis
Bilirubin 66, albumin 30 Bilirubin 23, albumin 43 Unavailable
Septic shock
Alcoholics with Cirrhosis 9 59 46
10
58
17
> 80 g /day
11
56
80
80 g /day
12
44
15
> 80 g /day,?¯
Bronchopneumonia Chronic renal failure Acute m yocardial infarct Status postoperative
Bronchopneumonia Chronic liver disease
All cases were m ale. NAD , no abnormality detected. For case 9, alcohol consumption was stated as ``approaching 200 g /day’’ in the questionnaire response; for case 12, there was evidence that consumption had dropped in the 2 ± 3 weeks immediately preceding the patient’s death. U nits for serum bilirubin were l m ol /L, and the normal reference range was 0 ± 18; units for serum albumin were g /L, and the normal reference range was 40 ± 52.
discernible correlation between storage time and T f concentration. No serum samples were available for the cases studied.
Transfer rin assay T f and its carbohydrate-m odi® ed congener were assayed by the method developed by Stibler et al. (1986), as modi® ed by Kwoh-Gain et al. (1990). As this method had been optimized for the analysis of serum samples, minor adaptation was required for the analysis of brain samples. Samples (0.1 ± 0.3 g) of tissue were chipped from the blocks of frozen, un® xed tissue (Scott et al. , 1995) and rapidly thawed by immersion in a large volume of warm 0.32 M sucrose, then immediately brought to 4ë C in an ice-bath. The samples were homogenized at room temperature in 1 vol of 50 m M HEPES buŒer pH 7.4 contain-
ing KCl and ovalbumin, and the extracts clari® ed by centrifugation. Disposable plastic m icrocolumns were packed with 200 l L of DEAESephacel and equilibrated with 2- N -morpholinoethanesulphonic acid (M ES) buŒer, pH 5.65. To eliminate the eŒect of partial iron saturation on isoelectric pro® le (Stibler et al. , 1986), 100 l L samples of homogenate were saturated with iron by the addition of 10 l L of 25 m M ferric nitrilotriacetate (FeNTA). They were diluted with 890 l L of MES buŒer and dialysed against M ES buŒer at room temperature for 2 hours. To bring the T f concentration into the most sensitive range of the standard curve, the dialysate volume was reduced to 400 l L with the aid of Aquacide. The carbohydrate-de® cient form (CD T) was isolated from the extract by applying it to a DEAESephacel column and eluting with 1 mL of M ES buŒer. Norm al transferrin is retained in the
340
Peter R. Dodd et al.
Table 2. B rain total transfer rin in dem entia cases
programs exactly as described previously (Dodd et al. , 1992).
Assay Case no. 13 14 15 16 17
Block 1 2 1 2 1 1 1
1
2
3
4
27 51 124 90 141 245 99
17 37 106 28 63 241 111
36
20
Tf assays were carried out and analysed as described in Methods. For each case, block 1 was taken from the mid-temporal gyrus and block 2 from the pre-central gyrus of the m otor cortex, from clinically diagnosed DAT cases w hose diagnosis was ultimately established at autopsy. See Methods section for case details. The multiple assays indicated were each carried out on entirely independent tissue extracts, i.e. a separate tissue sample was cut, thawed and taken through the full extraction and analytic process. See Methods. Values quoted are in l g of immunoreactive Tf per g of tissue.
column because of negatively charged sialate groups, while the carbohydrate-de® cient form passes through unhindered. Immunoreactive T f was assayed in both this eluate (CDT) and the original extract (total T f ) by radioimm unoassay (R IA) against [ 1 2 5I]T f and unlabelled T f standards. The method was based on that of Anderson et al. (1986), with minor modi® cations. The procedure uses anti-Tf immunoglobulin insolubilized on Matrix-Pel 102. In contrast to the published method (Anderson et al., 1986), the initial Sephadex G-25M column was eluted with borate buŒer rather than water, and the HEPE S-KC l buŒer did not contain detergent. To prepare [ 1 2 5I]T f , 100 l g of human T f was iodinated by the Bolton± Hunter technique 125 (Bolton & Hunter, 1973). Free I and unreacted reagent were removed by chromatography on Sephadex G-25M in PD10 columns. A typical displacement curve is shown in Fig. 1, which also illustrates that such curves were amenable to more accurate analysis with the aid of probit transforms of the values (see below).
[ 3 H ]-D iazepam recep tor binding site assay In some cases measurements of benzodiazepine binding site B M A X values were m ade in the same tissue samples from the alcoholic cases used to assay T f and C DT. The assays were performed and analysed with the aid of the EBDA and LI GAND
Protein Total protein in the samples was assayed by the method of Lowry et al. (1951) against bovine serum albumin standards.
M aterials Bolton± Hunter labelling kit was obtained from Amersham (Sydney, NSW, Australia). DEAESephacel, Sephadex G-25M , PD 10 columns and plastic microcolumns were purchased from Pharmacia (Uppsala, Sweden). M atrix-Pel 102 was purchased from Amicon Corp (Lexington, MA, USA). Puri® ed human T f and Aquacide were obtained from Calbiochem-Behring P /L (Sydney), and anti-Tf immunoglobulin from Silenius Laboratories P /L (Dandenong, Vic, Australia). All other chemicals were obtained from Sigma Chemicals (St Louis, MO, USA) or BDH Chemicals (Kilsyth, Vic) and were of the highest grade of purity available. A stock FeNTA solution was prepared by m ixing 25 m M ferric citrate with 50 m M nitrilotriacetic acid in 200 mL of water. The pH was adjusted to 4.0 with Na 2 CO 3 and the mixture heated until dissolved. Tris(hydroxymethyl)methylamine (Tris base) was then added to bring the pH to 6.5 and the solution stored at 0 ± 4ë C until use.
Data analysis The values from the T f R IA were transformed into probits to linearize the standard curve (Fig. 1). The line of best ® t was obtained by least squares analysis with the aid of the Cricket Graph TM III program (Computer Associates International, Islandia, NY, USA), and the values for unknowns obtained by back-substitution of their probit values into the regression equation. Parametric statistical tests (Winer, Brown & Michels, 1991) were performed with the aid of the SystatT M program (Systat Inc., Evanston, IL, USA).
R esults and discussion In con® rm ation of earlier work (M orris et al., 1992b), human cerebral cortex tissue obtained at autopsy was found to contain high concentrations
B rain transferrin in alcoholics
341
1 25 Figure 1. Transferrin RIA standard cur ve. The T f assay was car ried out as described in the Methods section using [ I]T f displaced by a range of concentrations of unlabelled T f. Unknowns were diluted so as to fall as near as possible to the central region of the cur ve. For quantitation, percentage binding values were converted to probits as shown in the inset.
of T f (Fig. 2, Table 2). Measurem ents m ade in slow-frozen samples by an R IA based on a highly speci® c antibody were somewhat lower than previously reported values based on nephelometry (M orris et al., 1987) or ELISA (Loeç er et al. , 1995) in snap-frozen material. This is consistent with the improved speci® city of the present method (Kwoh-Gain et al. , 1990). The values for CDT : T f ratios in brain (Fig. 4) were some 10fold higher than we have previously observed in human serum samples (Kwoh-Gain et al., 1990; Fletcher et al., 1991), i.e. a much higher proportion of brain tissue T f was of the carbohydratede® cient form (Brunngraber & Webster, 1986; HoŒmann et al. , 1995), irrespective of the disease group to which the case belonged (Fig. 4). As Fig. 2 shows, the tissue T f concentration showed no correlation, nor any detectable tendency to decline, with either the age of the patient at death or the delay between death and autopsy.
When independent extracts were made of the same tissue sample, the T f concentrations were generally in good agreement, with some exceptions (Table 2). However, total T f varied over a more than 14-fold range when diŒerent cortical areas from the same cases, or the same cortical region from diŒerent cases, were compared (Table 2). Thus, the between-subject percentage coe cient of variation was some 2.3-fold greater than the equivalent within-sample value. Figure 2 shows the extent of the between-subject variation. In contrast to some previous studies, we did not ® nd a signi® cant change, either a decrease (Morris et al. , 1987; Connor et al. , 1992b) or an increase (Loeç er et al. , 1995), in either form of T f in dementia cases. The average for dementia cases (Table 2) was 114 6 35 l g of total immunoreactive T f per g of tissue, slightly higher than but not signi® cantly diŒerent from the average value in controls (Fig. 3): t 7 5 0.88, p 5 0.41.
342
Peter R. Dodd et al.
Figure 2. EŒect of age and post-m ortem delay on total T f concentrations. Total immunoreactive T f was extracted and assayed from samples of cerebral cortex taken from the cases listed in Tables 1 and 2. Plotted values are the mean T f value calculated across all brain regions assayed for each case. (A) eŒect of post-m ortem delay ( pmd); (B ) eŒect of age at death. Neither correlation approached statistical signi® cance ( pm d: r15 5 0.05, NS; age: r15 5 0.07, NS): the regression lines are shown only as a guide.
B rain transferrin in alcoholics
343
Figure 3. Total T f concentrations in the brains of alcoholics. Total immunoreactive T f was extracted and assayed from sam ples of super ior frontal and primar y motor cortex taken from the cases set out in Table 1. The mean 6 SEM values shown were calculated from the means of the values in the two cortical regions from each case, of the num bers of cases shown in Table 1. *, signi® cantly diŒerent from the average across the other two groups, t 10 5 2.649, p 5 0.025.
Note that the highest cortical T f concentration occurred in the single diŒuse Lewy body disease case (Table 2): Lewy bodies are also a feature of Parkinson disease, where subcortical T f abnormalities have been postulated to play a pathogenic role (Loeç er et al. , 1995). If this case were removed from the analysis, the average total immunoreactive T f value in the other four dementia cases was 82 6 16 l g per g of tissue, comparable to the average value in controls of 79 6 10 l g. It is also noteworthy that the m ean value in case 17 (Table 2), which lacked senile plaques (as well as lacking tangles and Lewy bodies, see M ethods) was close to the average value in normal controls. The reproducibility of results usually observed when independent extracts were made from the same sample suggests that the high degree of variability in T f concentration is a biological phenomenon, and may help explain the con¯ icting results which have been reported, including data from a single
research group (Connor et al. , 1992b; Loeç er et al. , 1995). Detailed analysis of the autopsy reports and histological material gave no indication of any reason for the variations observed across cases 13 ± 15 (Table 2). Mean total T f was found to be slightly, but nonsigni® cantly, increased in the brains of chronic alcoholics compared with non-alcoholic controls, and this increase was more marked in alcoholics with liver disease (Fig. 3). An analysis of variance based on the three groups and two brain areas lacked power because of the small group sizes (Table 1), although it has been noted that it is di cult to assess statistical power in m ultifactorial designs (W iner, Brown & Michels, 1991). However, mean total T f concentration for the combined cortical areas in the brains of cirrhotic alcoholics was signi® cantly diŒerent from the average concentration in non-cirrhotic (alcoholic or non-alcoholic) cases, which did not diŒer one from another (Fig. 3). Calculation
344
Peter R. Dodd et al.
(Winer et al., 1991) showed that the t test used to compare cirrhotic and non-cirrhotic cases had a power of approximately 0.8 to detect a diŒerence at least this large at p < 0.05 with groups of the size used (Fig. 3). Although it has been argued that T f concentration m ay be raised in Alzheimer’ s or Parkinson’s disease as part of a selective excitotoxic mechanism, the result with cirrhotic alcoholics suggests that this may not be so, since excitotoxicity is not generally thought to be involved in the brain damage associated with cirrhosis. Rather, raised total T f m ay be a m ore general indication of tissue damage. One possibility is that it re¯ ects the astrocytosis concomitant to alcoholism (Dodd et al., 1992), since reactive astrocytes are also commonly found in neurodegenerative disease. Astrocytosis is m ore m arked and widespread in cirrhotic than in non-cirrhotic alcoholics, and this m ay be combined with a redistribution of T f from its near-exclusive localization in oligodendrocytes in normal brain (Gerber & C onnor, 1989; M artin et al. , 1991; Morris et al. , 1992a), to a pathological appearance in other cells, especially astrocytes, as has been reported in DAT (Connor et al. , 1992a). Whether increased total T f speci® cally re¯ ects the appearance of the Alzheimer Type II astrocytes pathognomonic of hepatic encephalopathy remains to be determined. It would be especially valuable if future studies could also exam ine nonalcoholic cirrhotic cases, to help clarify this point. However, it should be noted that a proper quantitation of Alzheimer Type II astrocytes is di cult to achieve, because these cells occur sparsely in most brain areas even in severe cases of hepatic encephalopathy; because they are di cult to identify with conventional histopathological stains; and because there appears to be no truly speci® c histological or immunocytochemical marker for them. Detailed examination of the data revealed that large biological variability was not con® ned to dementia cases: two of the cirrhotic alcoholic cases had mean T f levels which overlapped with those of non-cirrhotic cases, while the other two showed markedly higher values. Thus, even within this closely matched set of cases, brain total T f concentrations varied over an almost 9-fold range (data not shown). Further work will be required to con® rm the robustness of the ® nding shown in Fig. 3. There was no indication that brain areas which show a greater degree of pathological damage in chronic alcoholism were endowed with a variant
T f concentration (Main EŒect for Area in the analysis of variance: F 1 ,9 5 0.65, p 5 0.44). There was a non-signi® cant decrease in the CDT : T f ratio in alcoholics, which appeared more marked if liver disease was present (Fig. 4). However, the Groups term fell well short of statistical signi® cance (F 2 ,9 5 0.50, P 5 0.62). There was no evidence that the changes were greater in the pathologically susceptible region (superior frontal) than in the spared region ( primary motor) of the cerebral cortex (Fig. 4): M ain EŒect for Area, F 1 ,9 5 0.57, p 5 0.47; Group 3 Area Interaction, F 2 ,9 5 0.06, p 5 0.94. It should be noted that the direction of change in CDT : T f ratio is opposite to that predicted from studies of CDT : T f ratios in the serum of chronic alcoholics (Kwoh-Gain et al. , 1990; Fletcher et al., 1991; Stibler, 1993): that is, there was no increase in CD T : T f ratio in the brains of alcoholics. Hence, it would appear to be unlikely that either brain C DT or brain T f are derived exogenously to any substantial extent ( JeŒeries et al. , 1984; Pardridge, Eisenberg & Yang, 1987; Loeç er et al., 1995; Recht et al. , 1990), at least in pathological conditions. It should also be noted that, whereas total imm unoreactive T f is little altered in the serum of chronic alcoholics, in contrast to the marked changes in C DT : T f ratio (KwohGain et al., 1990; Fletcher et al. , 1991), the reverse was true for brain total T f and CD T (Figs 3, 4). A further test on the hypothesis that brain total T f , or brain CDT : T f ratio, might re¯ ect the localized eŒect of alcoholic status ante m ortem was provided by comparing these values with measures of GABA A-benzodiazepine receptor binding sites, since previous work has shown that the densities of these sites are altered in parallel with the extent of tissue damage assessed morphometrically (Dodd et al. , 1992, 1996). Neither parameter correlated with the density of ``centraltype’’ benzodiazepine receptor sites labelled with [ 3 H]-diazepam (Brain C DT : T f ratio is shown in Fig. 5). Indeed, there was no correlation between either T f measure and any index of GABA Abenzodiazepine receptor site density (data not shown): sites were labelled with [ 3 H]-muscimol, [ 3 H]-diazepam or [3 H]-¯ unitrazepam (Dodd et al., 1992). It should be emphasized that all the cited assays, including those using [ 3 H]diazepam, were carried out under conditions which eliminated binding to the ``peripheraltype’’ benzodiazepine binding site (Dodd et al.,
B rain transferrin in alcoholics
345
Figure 4. CDT:T f ratios in the brains of alcoholics. Total immunoreactive T f and CDT were extracted and assayed from sam ples of super ior frontal and primar y motor cortex taken from the cases set out in Table 1. Values are means 6 SEM of the numbers of cases shown in Table 1; u , super ior frontal cortex; j , primar y motor cortex. Statistical analysis of this data is given in the text.
1992). It would be of interest if future studies could exam ine the relationship between brain T f or CD T concentration and the densities of ``peripheral-type’ ’ sites, which are thought to be con® ned to astrocytes. Brain T f has been widely studied because of its potential role in the pathogenesis of neurodegenerative diseases. Support for the idea that T f may moderate the Al3 + -mediated toxicity which has been proposed to underlie the development of senile plaques comes from the ® nding that brain areas accumulating the highest concentrations of aluminium in experimental m odels of Al toxicity have the highest concentrations of T f receptors (M orris et al. , 1989; Edwardson et al., 1992; Florence et al., 1994). However, it has been reported that T f receptors are reduced, rather than increased, in pathologically signi® cant brain areas from DAT cases (Kalaria et al., 1992; Morris et al., 1994a); and recent work has suggested that brain Al concentration is not, in fact, raised in DAT (Edwardson et al. , 1992). Similarly, a role for T f -mediated toxicity in Parkinson disease is at variance with the tissue distri-
butions of T f receptors, and the direction of change of T f receptor density, in relation to pathological damage (M ash et al., 1991; Morris et al., 1994b). The data reported here provide some support for the notion that increased T f concentration may be a weak marker of reactive astrocytosis; however, there is little support for the idea that human brain T f is derived primarily from serum, since the relative concentrations of T f subtypes (Fig. 4) did not re¯ ect known variations in the serum levels of these substances in alcoholic vs. non-alcoholic cases (Stibler, 1993). This is in line with work showing a wide species variation in T f gene expression in brain, including an apparent lack of expression in human brain (Tu et al. , 1991). It is concluded that whereas serum CD T : T f ratio is now the marker of choice for alcoholic status, this is not re¯ ected in the brain CDT : T f ratio. Similarly, total brain T f concentration appears not to be useful as a marker of dementia or alcoholism. However, total imm unoreactive T f was strongly increased in cirrhotic alcoholic cases, probably as a consequence of glial proliferation,
346
Peter R. Dodd et al.
3
Figure 5. R elationship between CDT:T f ratios and [ H ]-diazepam binding site densities in the brains of alcoholics. The respective parameter s were assayed on the same tissue sam ples, each der ived from the super ior frontal and primar y motor cortex taken from the cases set out in Table 1. The regression lines are shown for comparison only, since neither correlation was signi® cant: ± Ð u , super ior frontal cortex, r 10 5 0.33, NS; ± ± ± s , primar y motor cortex, r 10 5 0.22, NS.
and should be a valuable index in studies of the eŒect of cirrhosis of the liver on brain structure and function.
Acknow ledgem ents This work was supported ® nancially by grants from the NHMRC and the AAB. PRD is an NHM RC Senior Research Fellow. We are indebted to Dr Tony Tannenberg and the pathology registrars at the Royal Brisbane and M ater Misericordiñ hospitals for performing the autopsies and making the histopathological assessments of the dementia cases. We wish to thank Dr Dean 3 Moss for helpful discussion. The data on [ H]diazepam binding site density was kindly provided by M r Greg Thomas.
References A NDERSON, G. J., M ACKERRAS, A., M ACK, U., P OWELL, L. W. & H ALLIDAY, J. W. (1986) Simple radioimm unoassay for transferrin using insolubulized antitransferrin antibodies: its application to cultured cells, Journal of Laborator y Clinical Medicine, 107, 59 ± 65. B EAN, P., S UTP HIN, M . S., L IU, Y. S., A NTON, R., R EYNOLDS, T. B., S HOENFELD, Y. & P ETER, J. B.
(1995) Carbohydrate-de® cient transferrin and falsepositive results for alcohol abuse in primary biliary cirrhosis: diŒerential diagnosis by detection of mitochondrial autoantibodies, Clinical Chemistr y, 41, 858 ± 861. B ELL, H ., TALLAKSEN, C., S JAÊ HEIM , T. et al. (1993) Serum carbohydrate-de® cient transferrin as a m arker of alcohol consumption in patients with chronic liver diseases. Alcoholism: Clinical and Experim ental R esearch, 17, 246 ± 252. B OLTON, A. E. & H UNTER, W. M. (1973) The labelling of proteins to high speci® c activities by conjugation to a 1 2 5I-containing acylating agent. B iochemical Journal, 133, 529 ± 539. B RUNNGRABER, E. G. & W EBSTER, J. C. (1986) Changes in glycoprotein carbohydrate content in the aging hum an brain, Neurochem ical Research , 11, 579 ± 588. C ARLSSON, A. V., H ILTUNEN, A. J., B ECK, O., S TIBLER, H . & B ORG, S. (1993) Detection of relapses in alcohol-dependent patients: comparison of carbohydrate-de® cient transferrin in serum, 5-hydroxytryptophol in urine, and self-reports, Alcoholism: Clinical and Experim ental R esearch, 17, 703 ± 708. C ONNOR, J. R., M ENZIES, S. L., S T M ARTIN, S. M . & M UFSON, E. J. (1992a) A histochemical study of iron, transferrin, and ferritin in Alzheim er’s diseased brains, Journal of Neuroscience Research , 31, 75 ± 83. C ONNOR, J. R., S NYDER, B. S., B EARD, J. L., F INE, R. E. & M UFSON, E. J. (1992b) Reg ional distribution of iron and iron-regulatory proteins in the brain in aging and Alzheim er’s disease, Journal of Neuroscience R esearch, 31, 327 ± 335. D ODD, P. R., T HOMAS, G. J., H ARPER, C. G. & K RIL,
B rain transferrin in alcoholics J. J. (1992) Am ino acid neutrotransmitter receptor changes in cerebral cortex in alcoholism : eŒect of cirrhosis of the liver, Journal of Neurochemistry, 59, 1506 ± 1515. D ODD, P. R., S COTT, H . L. & W ESTPHALEN, R. I. (1994) Excitotoxic mechanism s in the pathogenesis of dementia, Neurochem istry Inter national, 25, 203 ± 219. D ODD, P. R., K RIL, J. J., T HOMAS, G. J., W ATSON, W. E. J., J OHNSTON, G. A. R. & H ARPER, C. G. (1996) Receptor binding sites and uptake activities m ediating GABA neurotransmission in chronic alcoholics with Wernicke encephalopathy, B rain Research, 710, 215 ± 228. E CKERT, A. L., D ODD, P. R., C OOKE, R. W., K RIL, J. J. & H ARPER, C. G. (1991) Transferrin and desialyltransferrin in post m ortem human brain from Alzheim er disease and chronic alcoholic cases, Journal of Neurochem istry, 57 Suppl, S88B (Abstract). E DWARDSON, J. A., C ANDY, J. M., I NCE, P. G. et al. (1992) Alum inium accumulation, b -amyloid deposition and neuro® brillary changes in the central nervous system, Ciba Foundation Symposia, 169, 165 ± 179. F LETCHER, L. M., K WOH -G AIN, I., P OWELL, E. E., P OW ELL, L. W. & H ALLIDAY, J. W. (1991) M arkers of chronic alcohol ingestion in patients with nonalcoholic steatohepatitis: an aid to diagnosis, H epatology, 13, 455 ± 459. F LORENCE, A. L., G AUTHIER, A., P ONSAR, C., VAN DEN B OSCH - DE A GUILAR, P. & C RICHTON, R. R. (1994) An experimental animal m odel of aluminium overload, Neurodegeneration, 3, 315 ± 323. G ERBER, M. R. & C ONNOR, J. R. (1989) Do oligodendrocytes m ediate iron regulation in the hum an brain? Annals of Neurology, 26, 95 ± 98. H OFFMANN, A., N IMTZ, M ., G ETZLAFF, R. & C ONRADT, H. S. (1995) `Brain-type’ N-glycosylation of asialotransferrin from human cerebrospinal ¯ uid, FEB S Letters, 359, 164 ± 168. JEFFERIES, W. A., B RANDON, M . R., H UNT, S. V., W ILLIAMS, A. F., G ATTER, K. C., & M ASON, D. Y. (1984) Transferrin receptor on endothelium of brain capillaries, Nature, 313, 162 ± 163. K ALARIA, R. N., S ROMEK, S. M., G RAHOVAC, I. & H ARIK, S. I. (1992) Transferrin receptors of rat and hum an brain and cerebral m icrovessels and their status in Alzheim er’s disease, B rain Research, 585, 87 ± 93. K WOH-G AIN, I., F LETCHER, L. M., P RICE, J., P OWELL, L. W. & H ALLIDAY, J. W. (1990) Desialylated transferrin and mitochondrial aspartate aminotransferase compared as laboratory m arkers of excessive alcohol consumption, Clinical Chemistr y, 36, 841 ± 845. L IEBER, C. S., X IN, Y., L ASKER, J. M. & R OSMAN, A. S. (1993) Comparison of new m ethods for measuring carbohydrate-de® cient transferrin (C DT): application to a public health approach for the prevention of alcoholic cirrhosis, A lcohol and Alcoholism Supplement, 2, 111 ± 116. L OEFFLER, D. A., C ONNOR, J. R., JUNEAU, P. L. et al. (1995) Transferrin and iron in normal, Alzheim er’s disease, and Parkinson’ s disease brain regions, Journal of Neurochem istry, 65, 710 ± 724. L OWRY, O. H ., R OSENBROUGH, N. J., F ARR, A. L. &
347
R ANDALL, R. J. (1951) Protein m easurement with the Folin± phenol reagent, Journal of B iological Chemistr y, 193, 265 ± 275. M ARTIN, S. M., L ANDEL, H . B., L ANSING, A. J. & VIJAYAN, V. K. (1991) Immunocytochemical double labelling of glial ® brillary acidic protein and transferrin permits the identi® cation of astrocytes and oligodendrocytes in the rat brain, Journal of Neuropathology and Exper imental Neurology, 50, 161 ± 170. M ASH, D. C., P ABLO, J., B UCK, B. E., S ANCHEZ-R AMOS, J. & W EINER, W. J. (1991) Distribution and number of transferrin receptors in Parkinson’s disease and in M PTP-treated m ice, Exper imental Neurology, 114, 73 ± 81. M ORRIS, C. M., C OURT, J. A., M OSHTAGHIE, A. A. et al. (1987) Transferrin and transferrin receptors in norm al brain and Alzheim er’ s disease, B iochem ical Society Transactions, 15, 891 ± 892. M ORRIS, C. M., C ANDY, J. M., O AKLEY, A. E. et al. (1989) Comparison of the regional distribution of transferrin receptors and aluminium in the forebrain of chronic renal dialysis patients, Journal of the Neurological Sciences, 94, 295 ± 306. M ORRIS, C. M ., C ANDY, J. M ., B LOXHAM, C. A. & E DWARDSON, J. A. (1992a) Immunocytochemical localization of transferrin in the hum an brain, Acta A natomica (B asel), 143, 14 ± 18. M ORRIS, C. M., C ANDY, J. M ., K EITH, A. B. et al. (1992b) Brain iron hom eostasis, Journal of Inorganic B iochem istr y, 47, 257 ± 265. M ORRIS, C. M., C ANDY, J. M ., K ERWIN, J. M. & E DWARDSON, J. A. (1994a) Transferrin receptors in the normal human hippocam pus and in Alzheimer’ s disease, Neuropathology and Applied Neurobiology, 20, 473 ± 477. M ORRIS, C. M ., C ANDY, J. M ., O MAR, S., B LOXHAM, C. A. & E DWARDSON, J. A. (1994b) Transferrin receptors in the parkinsonian midbrain, Neuropathology and A pplied Neurobiology, 20, 468 ± 472. P ARDRIDGE, W. M., E ISENBERG, J. & YANG, J. (1987) H uman blood± brain barrier transferrin receptor, M etabolism , 36, 892 ± 895. R ECH T, L., T ORRES, C. O., S M ITH, T. W., R ASO, V. & G RIFFIN , T. W. (1990) Transferrin receptor in norm al and neoplastic brain tissue: implications for brain-tum our immunotherapy, Journal of Neurosurgery, 72, 941 ± 945. R OSKAMS, A. J. & C ONNOR, J. R. (1990) Aluminium access to the brain: a role for transferrin and its receptor, Proceedings of the National Academy of Sciences of the USA, 87, 9024 ± 9027. S CHELLENBERG, F., B ENARD, J. Y., L E G OFF, A. M ., B OURDIN, C. & W EILL, J. (1989) Evaluation of carbohydrate-de® cient transferrin compared with Tf index and other m arkers of alcohol abuse, A lcoholism: Clinical and Experim ental R esearch, 13, 605 ± 610. S COTT, H. L., TANNENBERG, A. E. G. & D ODD, P. R. (1995) Variant forms of neuronal glutamate transporter sites in Alzheim er’s disease cerebral cortex, Journal of Neurochemistr y, 64, 2193 ± 2202. S TIBLER, H., B ORG, S. & J OUSTRA, M. (1986) Micro anion exchange chromatography of carbohydratede® cient transferrin in serum in relation to alcohol consumption, Alcoholism: Clinical and Experim ental R esearch, 10, 535 ± 544.
348
Peter R. Dodd et al.
S TIBLER, H. (1991) Carbohydrate-de® cient transferrin in serum: a new marker of potentially harmful alcohol consumption reviewed, Clinical Chem istr y, 37, 2029 ± 2037. S TIBLER, H. (1993) Diagnosis of alcohol-related neurological diseases by analysis of carbohydrate-de® cient transferrin in serum, Acta Neurologica Scandinav ica, 88, 279 ± 283. T HOMPSON, C. M ., M ARKESBERY, W. R., E HMANN, W. D., M AO, Y. X. & VANCE, D. E. (1988) Reg ional
brain trace-element studies in Alzheim er’s disease, Neurotoxicology, 9, 1 ± 7. T U, G. F., A CHEN, M. G., A LDRED, A. R., S OUTHWELL, B. R. & S CHREIBER, G. (1991) T he distribution of cerebral expression of the transferrin gene is species speci® c, Journal of B iological Chem istry, 266, 6201 ± 6208. W INER, B. J., B ROWN, D. R. & M ICHELS, K. M. (1991) Statistical Principles in Exper imental Design, pp. 497 ± 582 (Sydney, NSW, M cGraw-H ill).
