Iron deficiency and anemia are major nutritional concerns throughout the world (Baynes and Bothwell 1990). Iron deficiency has been associated with hematological changes (red blood cell deformation), stunted growth, altered thermoregulatory function and decreased cognitive function (Ashkenazi et al. 1982, Brigham and Beard 1996, Lozoff and Brittenham 1986). Iron is essential for proper central nervous system metabolism through its role in the synthesis of neurotransmitters, myelin formation and brain growth (reviewed by Beard et al. 1993). In iron-deficient rats, there is a significant decrease in brain iron, increase in brain transferrin and a slight decrease in brain ferritin concentration (Chen et al. 1995b). It is unclear whether these changes in iron metabolism are due to altered acquisition or slow turnover rate of brain iron (Dallman and Spirito 1977). By examining changes in brain iron and iron regulatory proteins due to iron repletion in rats, we sought to gain a clearer sense of how the brain regulates changes in iron concentration due to iron deficiency.

Iron is heterogeneously distributed in the brain of adults, with the highest concentration being in the basal ganglia, substantia nigra and deep cerebellar nuclei (Hill 1988). It has recently been shown that some of the areas of the brain that are iron rich in adult rat brains are not iron rich for the first 60 d of life (Benkovic and Connor 1993). Because the concentration of iron is highest in the brain at birth, decreases through weaning and then begins to increase during critical periods of development, e.g., myelination (Roskams and Connor 1994), it is of great interest to examine the regional responses of the brain to dietary iron deficiency during growth. Developmental iron deficiency clearly alters the functioning of the brain in rats with functional sequelae (Felt and Lozoff 1996).

Iron is stored in the brain primarily as ferritin, whereas transferrin is responsible for transporting iron in cerebral spinal fluid and in plasma (Connor et al. 1992, Octave et al. 1983). Both ferritin and transferrin levels are highest at birth in newborn rats and decline thereafter (Roskams and Connor 1994). Overall, very little is known about the regional response of transferrin and ferritin to iron deficiency and iron repletion during development.

The reversibility of the observed alterations in brain iron metabolism due to iron deficiency has yet to be elucidated (Dallman and Spirito 1977). Although young iron-deficient rats respond rapidly to iron repletion therapy with respect to hematology and storage of non-heme liver iron, it is not clear if the brain transferrin and ferritin concentrations respond to iron repletion in a rapid fashion. Further, it is unclear if the brain responds heterogeneously with respect to iron repletion, i.e., does an supposedly iron-rich region such as the substantia nigra acquire the same amount of iron, transferrin and ferritin as regions, such as the cortex, that have a lower concentration of iron?

Table 1. Hematological variables and liver non-heme iron concentration in iron-deficient rats (ID), control rats (CN) and rats fed iron-deficient diet followed by an iron-adequate diet (REPL) in Experiment 11 [View Table]

Thus, this study consisted of two experiments in post-weaning rats; the first experiment addressed the effect of iron repletion on brain iron, transferrin and ferritin levels, and the second experiment expanded on this by examining changes in different brain regions due to systemic iron deficiency and then repletion. Our hypothesis was that the brain would respond to iron depletion and to repletion in a heterogeneous fashion. This hypothesis was tested at only one developmental stage, but clearly, other developmental periods require careful examination.

METHODS

RESULTS

Cytological analysis of brain iron did not reveal clear quantitative differences among groups. It is important to note that at this young age, little iron accumulation has yet occurred. Thus dramatically different iron content regions, as seen in adult rat brains, have not yet been generated. At the cellular level, in all three experimental groups, oligodendrocytes were the predominant iron-containing cell type (Fig. 1). However, there were more cells that resembled immature oligodendrocytes in the white matter of the ID group compared with the CN group. The REPL group contained a mixture of morphologically immature and mature oligodendrocytes. These observations are consistent with hypomyelination during iron deficiency, though this was not measured directly.

Brain and liver cytosolic iron concentrations were significantly lower in ID rats than in CN rats (Fig. 2A, P < 0.05 and P < 0.01 respectively). Similar results were obtained in the microsomal fractions (data not shown). In the REPL rats, brain cytosolic iron concentration was significantly higher than in the ID rats (P < 0.05) and not different from the CN rats. In contrast, liver cytosolic iron concentration in REPL rats was 90% greater than in ID rats (P < 0.01) but was still 30% lower than in CN rats (P < 0.05), indicating incomplete restoration of liver iron stores within the time frame of this experiment (Fig. 2A).

Two weeks of repletion did not normalize brain cytosolic transferrin, despite brain iron concentrations being normal (Fig. 2B). Cytosolic transferrin concentrations in the brains and livers of REPL rats were not different from those of ID rats but were significantly greater than those of CN rats by 30 and 49% respectively (P < 0.05). Liver cytosolic ferritin concentrations were 20% lower in ID rats than in CN and REPL rats (P < 0.05, Fig. 2C). Brain cytosolic ferritin was 20% lower in ID rats than in CN rats (Fig. 2C, P = 0.15) and 40% lower than in REPL rats (P < 0.05). Iron therapy for 2 wk in the REPL rats led to a significant increase in ferritin concentration in both organs (P < 0.05), suggesting a prompt upregulation of ferritin during repletion.

Table 2. Hematological variables and liver non-heme iron concentrations in iron deficient rats (ID), control rats (CN) and rats fed iron-deficient diet followed by an iron-adequate diet (REPL) in Experiment 21 [View Table]

Table 3. Growth and food efficiency in iron-deficient rats (ID), control rats (CN) and rats fed iron-deficient diet followed by an iron-adequate diet (REPL) in Experiment 2 over the last 4 wk of the protocol1 [View Table]

Brain regions differed significantly in iron concentrations regardless of dietary treatment groups (P < 0.0001, Fig. 3). Dietary treatment also had a significant independent effect on brain region iron concentrations (P = 0.03).⁴ Interestingly, in the CN rats, hippocampal iron levels were significantly higher than in all other regions, a pattern not observed in our other studies of adult rodent brains (Focht et al. 1997, Roskams and Connor 1994). Hippocampus and cortex were the only two regions with significantly lower levels of iron in response to dietary ID (P < 0.05). When regional iron concentration in ID rats was expressed as a percentage of that in CN rats, the hippocampus and cortex were 57 and 64% of CN (P < 0.05) compared with ~80 to >100% of CN in all other regions.

Brain transferrin concentration (Fig. 4) also varied significantly across brain regions (P < 0.0001), though the variation within just the CN group was marginally significant (P = 0.067). Dietary treatment had an independent and significant effect (P = 0.045). The interaction of diet and brain region was significant (P = 0.035), indicating that not all brain regions responded similarly to dietary treatments. Hippocampal transferrin concentration was higher in ID rats than in CN or REPL rats (P < 0.05. Transferrin concentrations in deep cerebellar nuclei, pons, superficial cerebellum and substantia nigra tended to be greater for ID rats than CN rats (P = 0.13). Transferrin concentrations did not increase in response to iron deficiency in the other regions. Four weeks of iron repletion led to a normalization of regional transferrin concentrations in REPL rats with the exception of the striatum, in which the striatal transferrin concentration was lower than that observed in CN rats (P < 0.05) and similar to the striatal transferrin concentration observed in ID rats.

Ferritin concentrations also varied significantly across the eight brain regions examined (P = 0.004, Fig. 5). Dietary iron deficiency significantly affected ferritin concentrations (P = 0.05), but no particular brain region appeared more susceptible than other brain regions.

The relationship of Hb concentration, an index of systemic iron deficiency, to regional brain iron concentration was investigated. Hemoglobin concentration and substantia nigra iron concentration were not significantly correlated (Fig. 6, bottom panel), whereas there was a strong positive and significant relationship between Hb and cerebral cortex iron concentrations (r = 0.883, P < 0.001, Fig. 6, top panel). Similarly, the relationship between liver iron, another index of systemic iron status, and substantia nigra iron was not significant, whereas there was again a strong positive curvilinear relationship between liver iron and cortex iron (data not shown).

DISCUSSION

Normalization of brain iron, in terms of overall concentrations in either microsomal or cytosolic fractions, extends our previous observations that rat brain iron can be reduced rapidly with post-weaning iron deficiency (Chen et al. 1995b). Repletion studies also demonstrate that the rat brain is fully capable of regaining iron in a rapid fashion, despite early observations that brain turnover (or more exactly, brain iron loss from the body) was a very slow process (Dallman and Spirito 1977). Others have shown an increase in iron uptake into the rat brain during iron deficiency, with transcytosis of iron and transferrin across the blood brain barrier (Crowe and Morgan 1992). The brain obtains iron via regulation of the transferrin receptor on the surface of endothelial cells on the brain microvasculature (Kalaria et al. 1992). The increase in uptake of iron observed in iron deficiency is not reflective of overall changes in blood brain barrier permeability and is highly selective for iron and transferrin. Although speculative at this time, it is reasonable to assume that upregulation of transferrin receptors is partially responsible for this alteration in uptake, because they also have a heterogeneous distribution (Hill et al. 1985, J. R. Connor, unpublished observation).

The regional specificity for brain iron acquisition is of great interest, because the distribution of iron in the developing brain is not the same as that in the adult brain (Connor 1994, Focht et al. 1997, Hill 1988, Roskams and Connor 1994). In a recent quantitative study using brains of 10-wk-old Fisher 344 rats, the striatum, hippocampus, thalamus and frontal cortex had equivalent iron concentrations of approximately 820 nmol/g, followed by brain stem, medial cortex, pons and cerebellum with 625-715 nmol/g (Focht et al. 1997). This is about double the amount measured here, where iron intake was much more highly regulated. Although there is a rough similarity in the distribution of iron, the absolute differences are more pronounced. In the same study, older (24-mo-old) rats had further iron accumulations to at least 890 nmol/g. Systematic evaluations of strain differences in brain iron distribution in rats have not been conducted. The substantia nigra, an iron-rich region in the adult rat brain as measured by histochemical methods (Hill 1988), is not particularly high in iron concentration in young rats. This region did, however, exhibit a dramatic increase in iron concentration during iron repletion, suggesting that iron was somehow "targeted" to go to the substantia nigra. It is interesting to recall that substantia nigra, globus pallidus and caudate putamen accumulate iron in elderly humans, particularly those with Alzheimer's disease and Parkinson's disease (Connor 1992, Good et al. 1992, Loeffler et al. 1995). Direct determinations of the regional responses of the transferrin receptor in future studies will test the possibility that it is upregulation of the transferrin receptor that is responsible for this targeted uptake of iron.

Iron is important for normal dopamine functioning in regions connected to the substantia nigra (Beard et al. 1994, Chen et al. 1995a, Youdim et al. 1989). The substantia nigra is rich in both iron and dopaminergic neurons (Bjorklund and Lindvall 1984, Hill 1988). The nigro-striatal dopaminergic pathway in the striatum has been found to be sensitive to iron deficiency with both decreases in dopamine D₂ receptors and increases in extracellular dopamine observed (Ashkenazi et al. 1982, Chen et al. 1995a, Youdim et al. 1989). Recent studies from our laboratory demonstrate that extracellular dopamine concentrations in the striatum return to normal within this same time frame for restoration of ventral mid-brain iron content (Nelson, C., Erikson, K. and Beard, J. L., unpublished observations). In addition, radioligand binding studies demonstrate alterations in dopamine transporters in striatal pathways in iron deficiency (Morse, A., Jones, B., and Beard, J. L., unpublished observations).

The predominant cell type containing iron in the young brain is the oligodendrocyte, with rapid accumulation of iron in these cells at the onset of myelination (Connor 1994). Cytological examination in the current study revealed "immature" appearing oligodendrocytes as a result of iron deficiency, which do not immediately revert to normal appearance within several weeks of iron therapy. These cells actively express transferrin receptor during development; when their function is compromised, the receptor expression drops dramatically (Roskams and Connor 1992). In addition, transferrin mRNA production and oligodendrocyte maturation are tightly coupled processes, suggesting that iron availability to these cells has a direct influence on their functioning (Bartlett et al. 1991).

Brain ferritin concentrations paralleled those of iron concentrations in some cases (e.g., no effect of iron deficiency in the substantia nigra), but in many brain regions there was a poor relationship between the two. This cannot be explained with our current data, although it is possible that the H and L chain isomers of ferritin are responding differently as the brain is richer in the H than the L isomer (Arosio et al. 1991). Current studies with more refined antibodies for specific chains of ferritin will provide an answer to whether specific H:L ratio changes occur in response to development and iron status.

This study also indicated that brain, and certain brain regions, respond to iron deficiency and iron repletion much more aggressively than was expected based on previous data on brain iron turnover (Dallman and Spirito 1977). One earlier study noted that caudate putamen iron concentration could be restored within several months of iron therapy in the post-weaning rats (Ben-Shachar et al. 1986). In contrast, pre-weaning or lactational exposure to low iron leads to irreversible changes in brain iron even after repletion for long periods of time (Dallman and Spirito 1977, Felt and Lozoff 1996, Youdim et al. 1989). Because peak myelination occurs around postnatal d 10 in rats, it is reasonable to assume this is a critical period, during which oligodendrocyte requirements for iron are quite high. Our current protocol did not address this question but the observed changes in oligodendrocyte appearance in iron deficiency after postnatal d 21 demonstrate a continued sensitivity to iron status well beyond the early and peak portions of the myelination period.

The higher transferrin concentration in the brain of REPL rats compared CN with rats indicates a biological adaptation that is still not complete after 2-4 wk of iron repletion. The source of this transferrin is not clear because transferrin can be derived from the choroid plexus, oligodendrocytes, as well as transcytosed across the blood brain barrier (Connor and Benkovic 1992, Crowe and Morgan 1992). The regional variation suggests, however, the possibility that transferrin plays a role in some sort of distribution system that likely requires the expression of transferrin and transferrin receptor.

In summary, these studies indicate there is a heterogeneous distribution and response of brain iron and iron regulatory proteins due to iron deficiency and iron repletion, respectively. Further investigations using early developmental periods as elegantly defined by Felt and Lozoff (1996) will help us to identify the critical periods of development during which "reversible" damage becomes "irreversible" damage.