8
Brain Research, 592 (1992) 8-16 © 1992 Elsevier Science Publishers B.V. All rights reserved 0006-8993/92/$05.00
BRESI~80
Iron and transferrin uptake by brain and cerebrospinal fluid in the rat A n d r e w C r o w e a n d E v a n H. M o r g a n Department of Physiolo~', The Unicersityof Western Australia, Nedlands, WA (Australia) (Accepted 14 April 1992)
Key words: Iron transport; Transferrin; Albumin; Blood-brain barrier; Choroid plexus; Cerebrospinal fluid
Iron and transferrin uptake into the brain, CSF and choroid plexus, and albumin uptake into the CSF and choroid plexus, were determined after the intravenous injection of [59Fe-t~I]transferrin and [13+I]albumin into control rats aged 15, 21 and 63 days and 21-day iron-deficient rats. Iron uptake by the brain was unidirectional, greatly exceeded that of transferrin and was equivalent to 39 and 36% of the plasma iron pool per day in the 15-day control and 21-day iron-deficient rats. The rate of transferrin catabolism in the rats was only about 20% of the plasma pool per day. Iron and transferrin uptake into the brain and CSF decreased with increasing age and was greater in the iron-deficient than in the control 21-day rats. The quantity of =2Sl-transferrin recovered in the CSF could account for only a small proportion of the iron taken up by the brain. Albumin transfer to the CSF also decreased with age but was lower than that of transferrin and was not affected by iron deficiency. Similarly, the plasma: CSF concentration ratios of transferrin and albumin, as determined immunologically, decreased with age and were greater for transferrin than albumin, It is concluded that iron uptake by the brain is dependent on iron release from transferrin at the cerebral capillary endothelial cells with recycling of transferrin to the plasma and transfer of the iron into the brain interstitium. Only a small fraction of the transferrin bound by brain capillari~:s is transcytosed into the brain and CSF, this being one source of CSF transferrin while other sources are local synthesis and transfer
from the plasma by the choroid plexuses,
INTRODUCTION The mechanism by which plasma iron, which is bound to transferrin, is transferred to the brain has been partially elucidated by the observations that brain capillary endothelial cells possess transferrin receptors t6'2s and can endocytose transferrin u. This has led to the hypothesis that iron is transported into the brain by transcytosis of the iron-transferrin complex across the capillaries 't,2',2~. If this is the case the concentrations of m l and 5'~Fe in the brain should increase in proportion to their plasma concentrations after intravenous injection of transfcrrin labelled with t2Sl and 5~Fe. However, this does not occur; S~Fe accumulates in the brain relatively more quickly than 1251(ref. 28). These results have several possible explanations, One is that binding of iron-transferrin to endothelial cells is followed by release of the iron and its transfer to the brain, and return of the apotransferrin to the plasma. Such a mechanism would probably involve endocytosis and recycling of transferrin as occurs in other types of
cells 14. Alternative possibilities are that transcytosis of iron-transferrin into the brain interstitium is followed by release of its iron and then by removal of transferrin by one or more of at least three processes. These are transcytosis across the capillary endothelium to the plasma, return to the plasma via the cerebrospinal fluid, and degradation within the brain. The aim of the experiments described in this paper was to investigate the above possibilities by studying the uptake of plasma iron and transferrin into the brain, cerebrospinai fluid and choroid plexuses, and the rate of transferrin catabolism, in rats with varying rates of brain iron uptake. The experiments were performed in control rats aged 15, 21 and 63 days and in iron-deficient rats aged 21 days. These ages were cho. sen because brain uptake is maximal at 15 days and has fallen to low levels by 63 days 2~, while at 21 days iron deficiency has its maximal effect, relative to control rats, of increasing iron uptake by the brain 2~. The uptake of iron and transferrin were measured at varying times after the intravenous injection of rat transfer-
Correspondence: E.H. Morgan, Department of Physiolo~, The University of Western Australia, Nedlands 6009, W.A. Australia. Fax: (61) 380 1025.
counted in samples of the pl'.,~ma, brain homogenate and cerebrospinal fluid, and in all of the cl~oroid plexuses after precipitation with 10% trichloracetic acid in ordeL- to separate protein-bound from free t2Sl and 13tI. Only protein-bound radioactive iodine values were used to calculate protein uptake by the brain, cerebrospinal fluid and choroid plexuses. The experimental procedure described above was varied in three types of experiments. In one, 15-day rats were perfused through the left ventricle with approx. 20 ml 0.15 M NaCI prior to removing the brain in order to reduce the content of plasma in the brain and choroid plexuses. This procedure was not used in the other experiments so that the blood in the choroid plexuses could be used as an aid in their dissection and removal. In the second, the long-term retention of SgFe in the brain was determined at times up to 56 days after injection of labelled transferrin. These rats were injected with [13tl]alhumin 3 rain before the time of sacrifice, for measurement of plasma volume and to allow correction of the brain 59Fe values for SgFe present in the blood. The third experiment was aimed at measuring the rates of catabolism of plasma transferrin and albumin. For this purpose daily 50-pl samples of blood were collected into microhaematocrit tubes from the ventral tail vein for 7-8 days after intravenous injection of the radiolabelled proteins. The plasma was separated and counted for radioactivity.
rin labelled with 1251 and SgFe. Rat [tall]albumin was injected simultaneously with the transferrin to enable corrections to be made for the transferrin contained within the blood in the brain, and for comparison with transferrin transfer to the cerebrospinal fluid and transferrin catabolism. MATERIALS AND METHODS
Materials Radioactive iron (SgFeCI3) and iodine (Na125I and Nat31l) were purchased from Amersham International, Amersham, U.K. Rat transferrin and rat albumin were isolated from plasma and labelled with 59Fe and 12Sl, and :31I, respectively as described earlier 12'3t. The transferrin was used in the experiments described below while in the diferric state.
Anhnals Rats of a Wistar strain were used. Control rats were fed a standard laboratory rat diet which contains 170 mg Fe/kg. Iron deficiency was produced as described earlier z7 by feeding a low iron diet, commencing during pregnancy and continuing during suckling until the rats were aged 21 days. The presence of iron deficiency was confirmed by measurement of haematocrit which averaged 21% for the rats fed the low iron diet and 31% for the controls.
Analytical methods The haematocrit was measured by the microhaematocrit method. Plasma and CSF transferrin and albumin concentrations were assayed by radial immunodiffusion using specific antisera produced in rabbits, The specificity of the antisera was demonstrated by OuchterIony double immunodiffusion against rat serum and purified rat transferrin and albumin. Radioactivity was counted in a three-channel y-scintillation counter (LKB-Wallac 1281 Compu-gamma) with appropriate corrections for spill-over between the three channels. Statistical evaluation of the results was performed by analysis of variance. Evidence of a statistically significant difference between mean values was considered to be P value of less than 0.05.
Experimenta! pexTcedures For the measurement of iron and transferrin uptake by the brain, cerebrospinal fluid and ehoroid plexuses, rats were injected with 100 pA of a mixture of [S')Fe-Z~-~l]transferrin (25-50 /~g protein) and [13t I]albumin (5(I-100/tg protein) via a lateral tail vein while under light ether anaesthesia. A 50-ktl sample of blood was collected into a heparinized microhaematocrit tube from the ventral tail vein approx. 3 rain later and was used for measurement of haematocrit and plasma radioactivity in order to calculal¢ plasma volume. At times varying from 0.5 Io 6 h after the injections the rats were anaesthetized with intraperitoneal sodium pel',tobarbiton¢, a sample of blood was take. from the heart with a needle and syringe, cerebrospinal fluid was collected from rite cislerna magna using a glass microcapillary tube, tile brain was removed and the choroid plexuses dissected from the lateral and fourth ventricles. The brain was then weighed and homogenized in 0.15 M NaCI using a Dounce homogenizer (Kontes Glass Co., Vineland, N J, U.S,A,). Radioactivity was 15 I)ay
Calc, lations The values for [IZ'~l]transferrin and '~')Feuptake by the brain were corrected for the brain content of blood, calculated from the brain and blood [t'~tl]albl~min results, and the S')Fc values were ;dso corrected for the amount of isotope which was attributable to the uptake of transferrin, as previously described 2s, Hence, the results for the brain uptake of transferrin represent tr;msferrin present in plasma-free brain tissue (including the microvessel walls) while those for iron represent the amount that has been released front transfer-
21 Day
200'
63 Day
200"
200"
"~ 150
150
150
-=" ItH)'
100
tO0
511'
50
50
0
0
0
2
4
6
2
4
6
0
2
4
6
Time (h)
Fig. 1. Uptake of iron ( i El) and transferrin (e o ) into the brain and transferrin into the CSF ( • A) in rats of differing ages after the intravenous injection of [1251-59Fe]labelled transferrin. Both control (['1 0 A ) and iron-deficient ( • • • ) 21-day rats were studied. The results were calculated as described in the text using the values for brain (corrected for plasma content) and CSF uptake of the radiolabels and for their mean activities in the plasma from the time of injection until sampling. Each value is the mean + S.E.M. from 4-6 animals.
10 rin and has entered the capillary walls and extravascular compartments of the brain. These types of corrections were not used for the CSF and choroid plexuses where the results there, 3re refer to total uptake of the labelled proteins and iron. Thus, , , r the choroid plexuses the values include the [l'51]transferrin and ~Fe present in the contained blood. In the case of CSF each sample ,,as carefully examined for contamination with blood and coloured samples were discarded. The amount of iron transferred from the plasma to the brain was calculated in terms of plasma volume equivalent (V), the volume of plasma which contained the amount of iron transferred to the brain:
v
0 =
(1)
=
C
TABLE ! Uptake of plasma iron and transferrin into the brain and transferrin i, to the CSF in rats of differing aTes The rates of iron uptake in/zl/day were obtained from the slopes of the regression lines in Fig. 1 and were also calculated as fractions of the plasma iron pool using these values and the plasma volumes, and in absolute units (p,g/day) using the mean plasma iron values from earlier work 2~. Also shown are the rates of increase in brain nonhome iron as determined by chemical measurement previously 27. The results for brain and CSF transferrin uptake are the plateau values in Fig. 1 at 4-6 h after injection of the labelled proteins. The results for 21-day rats are for control (C) and iron-deficient (ID) animals
Age(days) fl'CO)d t
l'
(2)
Where V = plasma volume equivalent of iron transferred to brain (ml), Q = brain 5~Fe radioactivity corrected for plasma and brain transferrin-bound 5~Fe (cpm), C = mean plasma S~Fe concentration from time of injection until time of sampling (cpm/ml), t' = sampling time (h). and C(t) = plasma S'~Fe concentration at any time (cpm/ml). This calculation is valid if iron transfer to the brain is a unidirectional process, as was shown to be the case (see Results). The values for C(t) used to determine C-"were the initial plasma radioactivity (3 rain) for each rat and the radioactivity at the time of sacrifice and tissue sampling (0.5-6 h). Hence the values for C were mean values from 4-6 animals at each time point. Similar calculations were applied to the transfer of transferrin into the brain, and transferrin and albumin into the CSF, Since these processes are not unidirectional the results could not be used to calculate the rate of transfer of the proteins to brain and CSF. However, they did give a measure of the steady state levels of plasma-derived transferrin in the brain, and transferrin and albumin in the CSF in lerms of ~1 plasma.
RESULTS
Iron and tran~ferrin uptake by the brahz The uptake of iron and transferrin by the brain calculated as described above is illustrated in Fig. 1. In each group of animals iron uptake was linear with respect to time and, in the control rats, the rate of uptake decreased as the rats aged from 15 to 63 days. However, in the 21-day iron-deficient rats iron uptake in terms of ttl plasma was significantly greater than in the 15-day controls. The mean values for the rates of iron uptake are presented in Table I, along with the values in terms of percent of plasma pool of iron per day and as /zg Fe/day calculated using the mean plasma iron concentrations previously determined for rats of the same age and treatment 27. In the 15-day and 21-day iron-deficient rats more than 30% of the plasma iron pool was transferred to the brain, However, due to the low plasma iron concentration of the iron-deficient rats the absolute rate of iron uptake by the brain was lower than in their controls or the 15-day rats. For the above calculations all of the results of the 6-h period of study were used to calculate brain iron
Brain Fe uptake (/~l/day) (plasma pool/day) (/zg/day) Brain non-heine Fe increase (p,g/day) Brain transferrin uptake (/zD CSF transferrin uptake (ttl)
15
21(C)
21 (ID)
63
600 0.39 0.66
330 0.12 0.50
770 0.36 0.23
60 0.007 0.11
0,45
0.32
0.13
0.10
7.2
5.3
8.9
3.7
3.2
1.2
2.3
0.6
uptake in the control rats but in the iron-deficient rats only data from rats sampled after the first half hour were used. The reason for doing this is that 59Fe injected as diferric transferrin is cleared from the plasma of iron-deficient rats very quickly during the first 10-15 min and thereafter more slowly 27, probably because the diferric transferrin has a competitive advantage for binding to transferrin receptors over the animals' own transferrin t3,~7 which, in iron-deficient animals, is mainly in the iron-free and monoferric forms. However, by 30 min when approximately 85% of the 59Fe had been cleared from the plasma the radiolabelled plasma would be more nearly representative of the native transferrin so that 59Fe uptake by the brain from then on can be used to calculate the rate of iron uptake. The value calculated in this way was 35% less than if all of the data from 0 to 6 h had been used. In the case of the control rats of all three age groups very similar values for brain iron uptake were obtained whether or not the first half-hour results were used in the calculations. This indicates that a similar problem to that in iron-deficient animals does not arise in these animals in which there is a much higher concentration of circulating endogenous diferric transferrin. The calculated uptake of transferrin by the brain was much lower than that of iron and not linear with time, but increased during the first 2 h and thereafter showed little further change (Fig. 1). However, like iron, the steady state level of transferrin uptake at 4-6 h after injection (Table I) decreased with age in the
11 control rats and was significantly higher in the iron-deficient than in the control 21-day animals. In another set of rats an assessment was made as to whether or not the [t2Sl]transferrin in the brain was bound to cellular membranes. At times varying from 10 min to 4 h after injecting rats with [SZSl-59Fe]transferrin and [t311]albumin the brains were removed, homogenized in 4 vols. ice-cold 0.15 M NaC! and samples of the homogenates centrifuged at 10,000 × g for 60 rain at 4°C. The protein-bound t2Sl in the supernatants and pellets was then corrected for contaminating plasmaderived [tzsI]transferrin using the albumin values, and the proportion of the labelled transferrin which was in the supernatant was calculated. In the 15-day rats this value increased from 17% at 10 min to 52% 4 h after the injections (Fig. 2). Similar measurements were made on 21- and 63-day control rats at 15 min and 4 h after injection, the results showing no significant difference from those found in the 15-day animals (Fig. 2). The linear uptake of iron calculated from the SgFe results suggested that the process was unidirectional, from blood to brain, at least over the 6-h period of study. In order to verify this 18-day control and irondeficient rats and 63-day control rats were injected with [SgFe-SZSl]transferrin and the amount of SgFe in the brain determined 0.25, 1, 28 and 56 days later. As shown in Fig. 3, the "SgFewas retained in the brain for 56 days in the older rats and for at least 28 days in the younger ones, after which it declined to significantly lower levels.
Iron, transferrin and albumin transfer to CSF The concentration of SgFe in the CSF was greatest during the first 0.5 h after injection and thereafter fell to low values in all groups of rats (Fig. 4). When expressed as percent of injected dose of radioactivity the early SgFe values were higher than the corresponding ones for transferrin and albumin which increased during the first 2-4 h in the 15- and 21-day rats and were relatively constant throughout the 6-h period in the 63-day animals. All of the values were significantly higher in the 15-day rats and decreased with the age of the animals. The values for SgFe and [~2Sl]transferrin, but not those for [tat l]albumin, were significantly higher in the iron-deficient 21-day rats than in their controls. In all groups of rats the values for radiolabelled transferrin were significantly greater than for albumin. The passage of transferrin into the CSF was calculated in a similar manner to iron and transferrin uptake by the brain, assuming that the total volume of CSF in all rats was 0.25 ml 3 and that cisternal CSF is representative of all of the CSF. The values increased during the first 2-4 h to reach plateau levels which
$0
$0
40
U
30'
Q,
.E 10 7;
!i!
00.1
O.S
i:i
1
2
4
Time (h}
Fig. 2. Percentage of [x2sI]transferrin present in the supcrnatant fraction obtained by homogenization and centrifugation of brains from 15 (stippled columns), 21 (hatched columns) and 63 (black columns) day rats at differing times after intravenous injection of the labelled protein. Each value is the mean + S.E.M. from 3-4 animals. Measurements were also made 10 rain after injection of 63-day rats but the supernatant radioactivity values did not differ significantly from the background.
were maintained until the end of the experiment (Fig. 1). In all cases these plateau levels were less than those for the brain (Table I). In order to allow a further comparison to be made between the concentrations of ['25I]transferrin in the brain and [I2"Sl]transferrin and [lalI]albumin in the CSF the concentration ratios for brain extracellular fluid and plasma, and CSF and plasma, were calculated (Fig. 5). For this purpose it was assumed that brain extracellular fluid is 20% of brain weight at 15 and 21 days and
2,S
•
181).Co,
[ ] IS-D.I:cdot 2,0'
~
1.s'
~
1,0'
0.5'
0.0
0.25
1
28
56
Time (Days)
Fig. 3. Brain 59Fe at varying times after the intravenous injection of l1251-59Feltransferrin into 18-day control, 18-day iron-deficient and 63-day control rats. Each value is the mean :i: S.E.M. from 4 rats. The results were corrected for the 59Fe present in the residual blood which was present in the brains.
12 21 Day Control
IS Day
0.6"
63 Day
21 Day Fe Deficient
0.3]
O.OS
0.3"
0,05
0.S' T = 0.4"
0.2'
0.04'
0.2
"~ o.3-
0.03
~ 0.2"
0 , 1 ' ~
OA
0.02
OA" 00
O,fll
2
4
0,0 . . . . . . . . . 2
6
4
0.0
6
2 4 6
0.00
2
4
6
Time(h) Fig. 4. Uptake of "~'~Fe ( • ). [ L"Sl]transfe,rrin (e) and [l~ I]albpmin (11) into the CSF after injection of the labelled proteins into rats of differing ages. Each value is the mean _+S.E.M, of 4 - 6 animals.
15% at 63 days, respectively :(''33. In all groups of animals these values for brain transferrin were much higher than for CSF transferrin, and the CSF values significantly higher for transferrin than albumin. When compared with controls, iron deficiency resulted in significantly higher ratios for transferrin in brain and CSF but not for albumin. As noted above, only about 50% of brain [~Z~l]transferrin is in a soluble form. Hence, the results for brain transferrin shown in Fig, 5 are probably twice as great as the true value for brain extracellular fluid, but even if this is taken into ~,ccount the results for brain transfcrrin are still higher than for CSF, The transferrin and albumin concentrations of CSF and plasma were determined for each group of animals (Fig. 6). The values for both proteins decreased with the age of the rats. The mean CSF: plasma concentration ratio for transferrin was significantly greater than that of albumin for all groups of rats, and for both
proteins, this ratio decreased with age. In the case of albumin all of the concentration ratios were almost the same as those for the [13~l]labelled protein (Fig. 5). With transferrin, similar CSF:plasma ratios were obtained from the immunological and radioisotope determinations in the 15- and 21-day rats, while at 63 days the latter was about twice as great as the former. These results suggest that a steady state in the relative distribution of the radiolabelled proteins between plasma
and CSF had been reached by 4-6 h after their injection, the difference between the ratios for transferrin in the 63-day animals being explained by local transferrin synthesis which is greater in adult than young ratslS,2~.
Iron, transferrin and albumin uptake by the choroid plexuses The amounts of 5~Fe, [t2Sl]transferrin and [t311]albumin in the choroid plexuses of the lateral and fourth
iS Day
21 Day
63 Day S
4 3'
3
2,
2.
I 0
t,
ii . . . . . . . . . . . 2
4
6
2
4
6
0
2
4
6
Time (h)
Fig, 5, Brain extracellular fluid/plasma ratio expressed as a percentage for [IZSl]transferrin (o o) and CSF/plasma ratios for [12"~l]transferrin ( • & } and [L~=i]albumin ( • D ) in rats after the intravenous injection of the labelled proteins. Both control (e • • ) and iron-deficient (o [] A ) 21-day rats were studied, Each value is the mean of:I:S,E,M, of 4-6 animals, The brain values were corrected for residual plasma and were calculated on the basis of extracellular fluid being 20% of brain weight at 15 and 21 days and 15% at 63 days of age.
13
day animals (Fig. 7). Also, in all groups of rats the mean values for transferrin uptake were higher than those for albumin although it was only in the 21-day animals that these differences were statistically significant. This suggests that there was a small degree of specific uptake of transferrin. A large proportion of the radioactivity present in the choroid plexuses in these experiments would have been due to the blood plasma which they contained. In order to reduce this, and hence obtain a better assessment of the amounts of iron and protein which had entered the ceils a n d / o r interstitial space of the ehoroid plexuses, a group of 15-day rats were perfused with 0.15 M NaCI through the heart 2 h after injecting the labelled proteins, immediately before removing the CSF, brain and ehoroid plexuses. As shown in Fig. 7 approximately 90% ef the radiolabelled transferrin and albumin was removed by this procedure but amounts of both proteins which were significantly above zero persisted, while the perfusion produced no significant change in the amount of transferrin-free 59Fe. The results for transferrin and iron uptake by the brain, and for the entry of the labelled proteins and 59Fe into the CSF did not differ from those obtained with unperfused rats (results not shown).
160
-[
[
Q.
120
80 40 @ 15 D Con
21 D Con 21 D Fe Del 63 D Con
15DCon
21DCon 21DFeOef
30 25 20 15
P
10 5
E 0
63DCon
1.2
~
0.4 0.0
150Con
~-
-
21DCon 21DFeDef
63DCon
Fig. 6. CSF and plasma concentrations and CSF/plasma concentration ratios (as percent) of transferrin and albumin in rats of different ages. The proteins were assayed by radial immunodiffusion. Each value is the mean+ S.E.M. of 6-8 measurements.
Turnover of plasma transferrin and albumin Iron uptake by the brain was greater in iron-deficient than control 21-day rats (Fig. 1). The possibility that this difference would be reflected by changes in the rate of transferrin catabolism was therefore investigated. Albumin turnover was also measured, for comparison with transferrin. Control and iron-deficient rats were injected intravenously with [t2Sl]transferrin and [13tl]albumin at the age of 18 days and daily blood samples collected for the next 7 days. After the first two days the plasma concentrations of [~251]transferrin and [~31l]albumin decreased in an exponential manner.
ventricles showed no significant trends with respect to time after injection. Hence, they were pooled for each group of rats. The results for transferrin-free "~Fe uptake were significantly lower at 63 days than in the 15- and 21-day animals but there were no significant differences between the 15- and 21-day groups or be. tween control and iron-deficient 21-day rats. The values for transferrin and albumin decreased with increasing age of the rats, but there were no significant differences between the control and iron-deficient 210.020]
Non-Pcrfuscd
Perfuscd
00tt111 .Tfo |
~
o.olo]
D ~,~
0.005 0,000
h
15 day Con
.
21 Day Con
.
. 21 Day
. 63 day
Fe Def
Con
. .........
15 clay Con
Fig. 7. Uptake of [tZSl]transferrin, [131l]albumin, and "~Fe into choroid plexuses of the lateral and fourth ventricles of rats of different ages. The transferrin and albumin values have not been corrected for plasma but the 59Fe values have been corrected for transferrin-bound 5'JFe and represent 59Fe which has been taken up in excess of plasma and tissue-bound transferrin. The results are from non-perfused rals aged 15, 21 (control and iron-deficient) and 63 days and from 15-day rats which were perfused with ice-cold 0.15 M NaCI immediately before removing the brain. The values for the non-perfused rats are the means of 30 rats sampled at 0.5-6 h after injection of the labelled proteins, and for the perfused animals are the means of 6 rats 2 h after the injections.
14
01 Tf
--
AIb
Tf
AIb
0.3
_I 0.1
Control
FO Deficient
Fig, 8. Rates of transferrin and albumin catabolism calculated as described in the text from the rates of disappearance of [IZSl]transferrin and [l~ll]aibumin from the plasma at 2-7 days after the intravenous injection of the labelled protein into control and iron-deficient IS-day old rats. Each value is the mean + S.E.M. of 6 rats.
The slopes of these exponentials, after correction for growth of the animals and the daily loss of plasma due to sampling, represent a measure of the rates of catabolism of the proteins. Iron deficiency was accompanied by a small, but significant increase in the rate of catabolism of transferrin from 19.1-23.3% of the plasma pool per day, but a decrease in that of albumin (Fig. 8). DISCUSSION The results of these experiments confirm earlier observations that the rate of iron uptake by the brain in the rat decreases as the animals mature from 15 days to adulthood 2s, that the development of iron deficiency is accompanied by an increase in the fractional uptake of plasma iron by the brain 27, and that the process of iron uptake by the brain is essentially unidirectional with only a very slow rate of turnover of brain iron~, even in the presence of iron deficiency. Moreover, the calculated rates of iron uptake per day were reasonably close to the daily increment of nonhaem iron in the brain determined chemically 27, especially when one considers that some of the 5~Fe taken up by the brain will be incorporated into haem as well as into non-haem compounds, especially in the younger animals in which the brain was still growing. This validates the method of measuring iron uptake by the brain and is also in keeping with a very slow rate of brain iron turnover. In all groups of rats the rate of iron uptake by the brain greatly exceeded that of transferrin. This difference was clearly apparent within the first 30 rnin after injecting the labelled transferrin. Hence, the protein must either be rapidly degraded in the brain or recycled back to the plasma. However, degradation of the
transferrin with return of [t2Sl]labelled catabolic products to the circulation can be excluded for the following reasons. In the 15- and 21-day control rats the rates of iron uptake as/~1 plasma per day were equivalent to 39 and 12% of the plasma volume, respectively, while the rate of transferrin catabolism in rats of about this age was approximately 20% of the plasma pool per day. Hence all, or a large fraction, of the transferrin catabolism would need to occur in the brain if this were to occur as part of the process of iron donation. Moreover, in the iron-deficient 21-day animals daily brain iron uptake in terms of percentage of the plasma pool (36%) was three times as high as in the controls yet the rate of transferrin catabolism increased only by about 20% relative to the controls. Since the quantitative difference between iron and transferrin uptake by the brain cannot be explained by protein degradation it is clear that the transferrin must recycle to the plasma after delivery of its iron. One possible pathway for this would be transcytosis across the cerebral capillary endothelia followed by passage through the brain interstitium into the CSF from which reabsorption to the blood could occur, probably via the subarachnoid villi. Evidence for the mechanisms required to transcytose transferrin across the endothelium is available. Brain endothelial cells possess transferrin receptors 16'zs and can endocytose the protein and transfer at least some of it into the extrava~ular compartment II. The progressive transfer of [SZSl]transferrin from a bound state (presumably mainly or entirely by capillary endothelial cells) to an unbound state (presumably in brain interstitial fluid and CSF) over a 2-4 h period which was observed in the present work is in keeping with this concept. Moreover, the labelled transferrin appeared in CSF where its concentration increased over a comparable time period, in addition, the amount of [~2Sl]transferrin entering the CSF varied according to the age and iron status of the rats in a manner similar to the variations in transferrin and iron uptake by the brain, suggesting that passage of transferrin from plasma to CSF is related to its function in iron delivery to the brain. Radiolabelled albumin also appeared in the CSF but at lower concentrations than transferrin, relative to their plasma concentrations. Since brain capillary endothelium has been shown to lack receptors for albumin 24 and is extremely impermeable with respect to diffusion of proteins it is likely that the albumin entered the CSF via the choroid plexuses. This probably occurs by passive diffusion and, if so, a similar mechanism of transfer would also apply to transferrin. This protein has been shown to leave the total vascular space at approximately the same rate as albumin m9and would therefore be expected to cross
15 the choroid plexuses by passive processes equally as well. Thus, the plasma-derived transferrin in CSF probably arises from two sources, cerebral capillaries and the choroid plexuses, and the difference between the relative concentrations of radiolabelled transferrin and albumin in the CSF may indicate the amount of [lZ~l]transferrin derived from the general capillaries of the brain. That such is the case is supported by the observation that iron deficiency in the 21-day rats was associated with higher [12sI]transferrin levels in the CSF and in the brain than in control rats, but there were no differences in the [13mI]albumin concentrations in the CSF between the two groups of animals. Qualitatively, as discussed above, the movement of transferrin from brain to CSF and then back to the blood could account for the recycling of that protein. However, quantitatively it is quite inadequate. Based on the rates of iron uptake by the brain and the concentrations of [12sI]transferrin found in the CSF it can be calculated that the CSF would have to turnover between 4 and 14 times per h, in order to account for transferrin recycling in the different groups of rats. This is far greater than the measured rate of turnover which has been estimated to be about 0.5 times per h 3'7. Hence, another route of recycling in addition to CSF, which accounts for the majority of the recycling, must be postulated. Theoretically, this could be retro-transeytosis across the cerebral capillarie~ from brain extravascular space to plasma. Some evidence for such a process has been published a2 but has been discounted by other workers who believe that the evidence for passage of proteins right across the endothelium in this direction is inadequate 4,s. Indeed, they claim that cerebral capillaries are highly polarized with respect to protein transfer, allowing this to occur only in the direction blood to brain. If so, the situation with respect to iron and transferrin would be very similar to that in the haemochorial placenta where transfer occurs only in the direction from maternal to fetal plasma, and iron transfer far exceeds that of transferrin 2°. Another argument against transcytosis of transferrin having an important role in delivery of iron to the brain or in transferrin recycling to the plasma is the observed rates of these processes as measured using radio-opaque markers and electronmieroscopys. The times involved are measured in hours, not minutes, as would be required for iron deliver,. If transcytosis of transferrin, although probably occurring at a slow rate, is not a quantitatively important process in the supply of iron to the brain what mechanisms operate? At least two should be considered. One is receptor-mediated endocytosis of transferrin fol-
lowed by release of iron from the transferrin in the endocytotic vesicles, transfer of the iron to the brain interstitium and recycling of the iron-depleted transferrin to the blood plasma. This chain of events has been shown to be operative in the process of iron acquisition of many types of cells ~4 and is also believed to be responsible for iron transfer across the haemochorial placenta 20. The second possible mechanism is that iron is released from transferrin on the luminal plasma membrane of the endothelial cells after binding to the receptor, possibly by the action of a transmembrane ferroreductase, after which the iron is transported into the brain and the transferrin released back to the plasma. This type of mechanism has been proposed for iron uptake by hepatocytes 3° and some other types of cells6. An interesting feature of the CSF results was the change in relative concentration of [~25I]traasferrin and 59Fe with time after injection. At 0.5-1 h the S9Fe values were higher than those of transferrin but after that 59Fe concentration decreased while [l:51]transferrin concentration increased or remained steady. The mechanism responsible for this is uncertain but it could be that the choroid plexus can remove iron from plasma transferrin and transport it into the CSF at a rate faster than that at which the protein itself is transferred. Within the CSF the iron may bind to endogenous unlabelled transferrin synthesized in the choroid plexus and other sites in the brain t. If the process of iron transfer to the CSF from the ehoroid is rapid the concentration of S'JFe in the CSF would increase soon after injection and would then fall as the plasma concentration of S9Fe decreased and the CSF was reabsorbed. The results for the choroid plexuses support this suggestion since they showed evidence of a small but significant specific binding of transferrin (i.e. in excess of albumin) and some uptake of SgFe which had been released from transferrin. Relatively high concentrations of proteins in the CSF of fetal and young animals, decreasing as they mature, has been reported previously by many authors. Hence, the present results are not surprising, although the concentration of transferrin decreased relatively less than that of albumin. Probably this difference between the two proteins results from local synthesis of transferrin, which increases as the rat matures ts'z9, while CSF albumin is derived largely or completely from the plasma. In addition, transferrin synthesis in the brain, which is not affected by iron deficiency is, may have masked any difference in CSF concentration of the protein attributable to different rates of transfer of the protein from plasma to CSF in the control and iron-deficient 21-day animals.
16 There have been few studies of the transfer of radiolabelled proteins from plasma to CSF in the rat. Amtorp 2 reported that [t~l]albumin transfer diminished as rats aged from birth to 30 days, which is in keeping with the present work, although his values for the CSF:blood [12Sl]albumin ratio were somewhat higher, possibly because he measured total ~2Sl, not protein-bound =~I. The transfer of transferrin from plasma to CSF has been reported only for fetal sheep in which it was found to be transferred more efficiently than albumin re. The concentration of transferrin in CSF of adult rats has also been reported previously 2t, the values being comparable to those reported herein. In conclusion, the results of this and previous investigations support the following hypothesis. Iron uptake by the brain is dependent on iron release from transferrin at the cerebral capillary endothelial cells, with recycling of the transferrin to the plasma and transport of the iron into the brain ¢xtravascular space. There the iron is probably bound by t=ansferrin which has been synthesized locally and transported to the cells of the brain, although this has yet to be proven. A small fraction of the transferrin bound by brain capillary cells is transcytosed into the brain, thence into the CSF, representing one source of CSF transferrin. Other sources are locally synthesized transferrin and transferfin derived from the plasma via the choroid plexuses. Acktwwh, dgentents, This work wt~s supported by a grant from the National Health and Medical Research Council of Australia.
REFERENCES 1 Aired, A,R., Dickson, P,W,, Morley, P.D. and Schr~iber, O,, Distribution of tran~ferrin synthesis in brain and other tissues in the rat, J, Biol, Chem,, 262 (1987)5293=5297. 2 Amtorp, O,, Transfer of It:S.albumin from blood into brain and ¢¢rebrospinal fluid in the newborn and juvenile rats, Acta PhysioL Scand., 9 r" (1976)399-406, 3 Bass, N,H. and Lundberg, P,, Postnatal development of bulk flow in the cerebrospinal fluid system of the albino rat: clearance of carboxyl.[14c] inulin after intrathecal infusion, Brai~ Re,v,, 52 (1973) 323-332. 4 Broadwell, R,D., Balin, B.J., Salcman, M. and Kaplan, R.S., Brain-blood barrier? Yes and no, Proc. Natl. Acad. SoL U.S,A,, 80 (1983) 7352-7356. 5 Broadwell, R.D.. Transcytosis of macromolecules through the blood-brain barrier: a cell biological perspective and critical appraisal. Acta Neuropathol=, 79 (1989) I 17-128, Crane. F,L., Sun, I,L,, Clark, M.G., Grebing, C. and Low, H., Transplasma-membrane rcdox systems in growth and develop. ment, Biochim. Biophys. Acre, 811 (1985) 233-264, 7 Cserr, H,, Potassium exchange between cerebrospinal fluid, plasma and brain, Am, I, Physiol,, 209 (1965) 1219-1226, 8 Cscrr, H,F., Cooper, D,N,, Suri, P,K. and Patlak, C.S,, Effiux of radiolabelled polyethylene glycols and albumin from rat brain, Am. J, Physiol., 240 (1981) F319-328, 9 Dallman, P,R, and Spirito, R.A,, Brain iron in the rat: extremely slow turnover in normal rats may explain long-lasting effects of early iron deficiency, J, Nutrit,, 107 (1977) 1075-1081, l0 Dziesgielewska, K,M,, Evans, C,A,N,, Malinowska, D,H,, Moll-
gard, K., Reynolds, M.L. and Saunders, N.R., Blood-cerebrospinal fluid transfer of plasma proteins during fetal development in the sheep, J. Physiol., 300 (1980) 457-465. I 1 Fishman, J.B., Rubin, J.B., Handrahan, J.V., Conner, .I.R. and Fine, R.E., Receptor-mediated transcytosis of transferrin across the blood-brain barrier, J. Neurosci. Re,v, 18 (1987) 299-304. 12 Gardiner, M.E. and Morgan, E.II., Transfcrrin and iron uptake by the liver in the rat, Aust. J. Exp. Biol. Med. Sci., 52 (1974) 723-736. 13 Huebers, H., Csiba, E., Heubers, E. and Finch, C.A., Molecular advantage of diferric transferrin in delivering iron to reticulocytes: a comparative study, Prec. Soc. Exp. Biol. Med., 179 (1985) 222-226. 14 Huebers, H.A. and Finch, C.A., The physiologyof Ir~nsferrin and transferrin receptors, PhysioL Rev., 67 (1987) 520-582. 15 Idzerda, R., Huebers, H., Finch, C.A. and McKnight, G.S., Rat transferrin gene expression, tissue-specific regulation by iron deficiency, Prec. Natl. Acad. SoL USA, 83 (1986) 3723-3727. 16 Jefferies, W.A., Brandon, M.R., Hunt, S,V., Williams, A,F., Getter, K.C. and Mason, D.Y., Transferrin receptor on endothelium of brain capillalies, Nature, 312 (1984) 161-163. 17 Kornfeld, S.. The effect of metal attachment to human apotransferrin on its binding to reliculocytes, Biochim. Biophys. Acre, 194 (1969) 25-33. 18 Levin, M., Tuil, D,, Uzaa, G., Dreyfus, J.-C., and Kahn, A., Exwession of the transferrin gene during development of nonhepatic tissues: high level of transferrin mRNA in fetal muscle and adult brain, Biochem. Biophys. Res. Cerumen., 122 (1984) 212-217. 19 Morgan, E.H., Transferrin and albumin distribution and turnover in the rabbit, Aust. J. Exp. Biol. Med. Sci., 47 (1969) 361-366, 20 Morgan, E.H., Placental transfer of iron, Prec. Aust. Physiol. Pharmacol. Soc., 13, 1 (1982) 11-1% 21 New, H.. Dziegielewska, K.M. and Saunders, N,R., Transferrin in fetal rat brain and cerehrospinal fluid, Int. J. Der. Neurosci., I (1983) 369-373. 22 Pardridge, W.M., Receptor-mediated peptide transport through the blood-brain barrier, Emlocr. Ret'., 7 (1986) 314-33{I. 23 Pardridge, W.M., Recent advances in blood-brain barrier transport, Aml. Ret'. Pharmacol. ToxicoL, 28 (1988) 25-39. 24 Pardridge, E,M., Eisenberg, J. and Cefalu, W,T., Absence of albumin receptor in brain capillaries in rive or in vitro, Am. J, Phy,ffol, 249 (1985) E264=267. 25 Pardridge, W,M., Eisenberg, J, and Yang, J,, Human blood-brain barrier Iransferrin receptor, Metabolism, 26 (1987) 892-895. 26 Smith, Q,R,, Woodbury, D.M, and Johanson, C,E., Kinetic analysis of ['~*CI]-[:"Na]- and ['~Hlmannitol uptake into the in rive choroid plexus-cerebrospinal fluid brain system: ontogeny of the blood-brain barrier and blood-CSF barriers, Der, Bra#, Res., 3 (1982) 181-198. 27 Taylor, E,M,, Crowe, A, and Morgan, E,H,, Transferrin and iron uptake by the brain: effects of altered iron status, J, Neurochem,, 57 (1991) 1584-1592. 28 Taylor, E,M, and Morgan, E.H,, Developmental changes in transferrin and iron uptake by the brain in the rat, Def. Brain Res., 55 (1990) 35-42, 29 Thomas, T., Schreiber, G, and Jaworowski, A., Developmental patterns of gene expression of secreted proteins in brain and choroid plexus, Dee. Biol,, 134 (1989) 38-47. 30 Thorstensen, K. and Romslo, 1,, Uptake of iron from transferrin by isolated rat hepatocytes, A redox-mediated plasma membrane process?, J. Biol, Chem,, 263 (1988) 8844-8850. 31 Trinder, D,, Morgan, E,H. and Baker, E., The effects of an antibody to the rat transferrin receptor and of rat serum of albumin on the uptake of diferric transferrin by rat hepatocytes, Biochim, Biophys, Acre, 943 (1988) 440-446, 32 Van Deurs, B., Vesicular transport of horseradish pcroxidase from brain to blood in segments of the cerebral microvasculature in adult mice, Brain Res., 124 (1977) I-8. 33 Woodward, D,L,, Reed, DJ. and Woodbury, D.M,, Extracellular space of the rat cerebral cortex, Am. J. PhysioL, 212 (1967) 367-370,