Gene Expression of Iron-Related Proteins during Iron Deficiency Caused by Scurvy in Guinea Pigs

ARCHIVES OF BIOCHEMISTRY AND BIOPHYSICS

Vol. 325, No. 2, January 15, pp. 295–303, 1996 Article No. 0037

Gene Expression of Iron-Related Proteins during Iron Deficiency Caused by Scurvy in Guinea Pigs Anna Gosiewska,1 Fatemeh Mahmoodian, and Beverly Peterkofsky2 Laboratory of Biochemistry, National Cancer Institute, Bethesda, Maryland 20892-4255

Received June 29, 1995, and in revised form November 1, 1995

The regulation of expression of hepatic iron-related proteins was examined during iron deficiency caused by scurvy in guinea pigs. Previous studies showed that some effects of scurvy, such as suppression of collagen gene expression, result from events associated with weight loss. During the initial phase of scurvy when vitamin C is depleted but animals grow normally, serum iron levels decreased to 50% of normal. During the second phase of scurvy when animals lose weight, there was a further decrease in iron levels to 10–15% of normal. Serum transferrin levels increased during scurvy, but this increase was related neither to the rate of weight loss nor to hepatic transferrin mRNA expression, which decreased. Serum ferritin levels diminished early in scurvy with a preferential loss of the L subunit. In liver, however, both ferritin subunits were almost undetectable even in scorbutic animals gaining weight. Ferritin gene expression during vitamin C deficiency was correlated with serum ferritin levels in that the level of mRNA for the H subunit remained relatively constant while that of the L subunit decreased early. Transferrin receptor mRNA expression in liver was induced as soon as iron levels decreased early in scurvy, which is similar to results reported for iron-depleted cultured cells. In contrast to results in cell culture, expression of iron regulatory protein 1 mRNA was decreased to approximately 50% of normal early in scurvy with a concomitant decrease in hepatic cytosolic aconitase activity. Our data indicate that iron deficiency occurs early during vitamin C deficiency and leads to changes in expression of ironrelated proteins that differ in some aspects from regulation by iron in cell culture. Other events associated with weight loss in late scurvy may play a further role in this regulation. q 1996 Academic Press, Inc. Key Words: vitamin C; iron; transferrin; ferritin; 1 Present address: Johnson and Johnson, Wound Healing Technology Resource Center, Skillman, NJ 08558. 2 To whom correspondence should be addressed. Fax: (301) 4023095.

transferrin receptor; iron regulatory protein; iron-responsive element.

One of the consequences of vitamin C deficiency in guinea pigs is abnormal iron metabolism (1) including a reduction in serum iron concentrations (2). Serum iron concentrations also are decreased in vitamin Cdeficient humans (3). A relationship between iron and ascorbate also is observed in humans with iron-overload diseases (4). Patients are generally treated with the chelator desferrioxamine to remove excess iron, and concomitant ascorbate administration enhances removal (4), presumably because ascorbate releases iron from its storage form. In cell culture, ascorbate appears to enhance the amount of iron stored in ferritin by preventing the lysosomal degradation of ferritin to nonutilizable hemosiderin (5). Ascorbate stimulates iron absorption from the gastrointestinal tract (1) while iron stimulates uptake of ascorbate into isolated microsomes (6). Many of these interactions are related to the redox system between ferric iron and ascorbate. When iron availability is changed in cell cultures, the synthesis of ferritin and the TfR3 are coordinately regulated in opposite directions through posttranscriptional mechanisms. In iron-deficient cells, interaction of cytosolic IRP1 with IREs leads to inhibition of the rate of translation of ferritin mRNAs, with no change in their concentration and with stabilization of TfR mRNA (7, 8). IREs are stem–loop structures located in the untranslated regions of the mRNAs for ferritin and TfR (7, 8). IRPs have been cloned from several species 3 Abbreviations used: Tf, transferrin; TfR, transferrin receptor; IRP1, iron regulatory protein 1; IRE, iron-responsive element; DTT, dithiothreitol; SDS–PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; IGFBP, insulin-like growth factor binding protein; BSA, bovine serum albumin; DAB, diaminobenzidene; FH, ferritin H subunit; FL, ferritin L subunit; UIBC, unsaturated iron-binding capacity; RT–PCR, reverse transcriptase–PCR.

295

0003-9861/96 $12.00 Copyright q 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.

/ m4949$9264

12-12-95 03:13:34

arca

AP: Archives

296

GOSIEWSKA, MAHMOODIAN, AND PETERKOFSKY

and the sequence analysis indicates identity with cytosolic aconitase (7, 9). The protein is a fully active aconitase without RNA binding activity only when an iron– sulfur cluster is ligated to the three cysteines of the protein (7, 10). There is a second IRP that does not have aconitase activity (11, 12). To date there are few in vivo studies on the relationship between serum iron availability and the proteins involved in iron homeostasis. Since vitamin C deficiency in guinea pigs leads to iron deficiency, it was used as an in vivo model to study the regulation of expression of genes for proteins involved in iron metabolism. Guinea pigs that have been on a vitamin C-free diet for 2 weeks grow normally although tissue vitamin C levels are rapidly depleted, a period designated as phase I of scurvy (13). During the next 2 weeks they become anorexic and lose weight, a period designated as phase II of scurvy. During phase II there is inhibition of collagen gene expression and proteoglycan synthesis in connective tissues (13). Inhibition appears to result from induction of IGFBPs that inhibit IGF-Idependent functions (14, 15). These events are duplicated in fasted guinea pigs receiving vitamin C, so that phase II of scurvy is equivalent to fasting, at least with respect to regulation of IGF-I-dependent functions such as collagen gene expression by IGFBPs (13–15). In contrast, defective wound healing is observed even in phase I of scurvy and is not related to induction of IGFBPs (16). Therefore, it was of interest to determine at what stage of scurvy iron deficiency occurred and whether any observed changes in the expression of iron-related proteins during scurvy resulted directly from iron depletion or from systemic changes associated with weight loss. EXPERIMENTAL PROCEDURES Vitamin C-deficient guinea pigs. The procedures for obtaining vitamin C-deficient guinea pigs as well as tissue collection and preparation of sera from these animals and normal controls have been described previously (15). All animals received a vitamin C-free guinea pig diet but normal controls were supplemented orally with vitamin C (25 mg/100 g body wt) daily. Scorbutic guinea pigs were kept on the ascorbate-free diet for up to 4 weeks to obtain a wide range of weight loss. The main group of scorbutic animals was described previously (15) and their weight loss during phase II (3rd and 4th weeks) was 5.1, 9.2, 18.6, 24, and 28%. An additional group of scorbutic guinea pigs was obtained with 4.8, 7.7, 13.4, 20, 20.7, 25, and 28% weight loss. For both groups, there were animals from phase I of scurvy that had been on the vitamin C-free diet for up to 21 days but were still gaining weight, and they are designated as scorbutic with 0% weight loss. Normal animals were gaining weight at approximately 7.8 g/day. Animal experimental procedures were carried out in accordance with National Institutes of Health guidelines. Iron assay. Serum iron and total iron binding capacity were measured by a colorimetric assay using a kit from Sigma (St. Louis, MO). Liver extracts. Approximately 20–50 mg of frozen liver samples was immersed in 500 ml of 20 mM Tris–HCl, pH 7.4, containing 2% Triton X-100 at 07C and minced with scissors. Minced tissue was

/ m4949$9264

12-12-95 03:13:34

arca

transferred to a glass homogenizer and homogenized with 20 strokes, and the homogenates were centrifuged for 15 min at 16,000g at 47C. Protein content of the supernate (extract) was determined using the BCA protein assay (Pierce Chemical Co., Rockford, IL). Extracts were stored in small aliquots at 0207C. Hepatic aconitase activity. Aconitase activity in liver extracts was followed spectrophotometrically by measuring the disappearance of cis-aconitate at 240 nm as a function of time for up to 20 min (17). The reaction mixture contained 0.1 M Tris–HCl, pH 7.4, 0.1 mM cisaconitate, and either liver extract (approximately 40 mg of protein) or purified aconitase from porcine heart (Sigma) in a total volume of 600 ml. One unit of aconitase activity was defined as the amount of enzyme which causes the disappearance of 1 mmol of cis-aconitate at 247C in 1 min under the conditions described. The enzyme activity also was measured after activation with 25 mM ferrous ammonium sulfate and 2.5 mM L-cysteine–HCl at 07C for 1 h. In some cases, 5 mM DTT and 25 mM sodium sulfide also were added (10). Ligand blotting for serum transferrin. Serum samples (1 ml) or 1 mg of human transferrin (Sigma) were diluted in SDS–PAGE sample buffer (0.05 M Hepes, pH 7.2, 20% sucrose, 2% SDS, 0.002% bromphenol blue, 0.5 M urea) without heating, and proteins were separated on 10% SDS–PAGE under nonreducing conditions (14). Proteins were transferred to an Immobilon-P membrane (Millipore), as described previously (14). Specific binding of 0.1 mCi of 59FeCl3 (Amersham) to transferrin was carried out in 10 ml of alkaline buffer (0.1 M Tris– HCl, pH 8.1, 1% BSA, and 0.1% Tween 20) overnight at 47C. Nonspecific binding was determined by incubation of duplicate membranes with 8.9 mM unlabeled iron for 2 h at 47C before adding 59Fe. After incubation, membranes were washed twice with 0.1 M NaCl/0.05 M Tris–HCl, pH 7.4, containing 0.2% Tween 20, and three times with 0.11 M NaCl/0.05 M Tris, pH 7.4. Membranes were dried and exposed to Kodak X-OMAT film with an intensifying screen at 0807C. Intensity of bands was quantitated by densitometry using the Image program on a Macintosh computer. Bovine holotransferrin (Sigma) was used to determine iron release in sample buffer (without bromphenol blue) by measuring absorption at 465 and 280 nm (18). Controls consisted of holotransferrin dissolved in Tris buffer, pH 8.3, where release does not occur, and acetate buffer, pH 4.1, where complete release occurs. Western blotting for serum ferritin. Serum samples (5 ml) or liver extracts (30 mg of protein) were fractionated on 12% SDS–PAGE under nonreducing conditions (14). Immune-reactive ferritin was detected with anti-human ferritin IgG fraction (The Binding Site, San Diego, CA) at a 1:50 dilution and with a Vectastain kit with biotinylated anti-sheep IgG antibody at a 1:500 dilution (Vector Laboratories, Burlingame, CA). Horse spleen apoferritin (0.1 mg) was used as a standard (Calbiochem). The blots obtained after staining with DAB (Vector Laboratories) were photographed. Total and poly(A)/ RNA preparation. Total RNA from liver was extracted and poly(A)/ RNA was prepared as described previously (15). Quantitation of mRNAs for Tf, FH, FL, TfR, and IRP1. Reverse transcriptase–polymerase chain reaction (RT–PCR) was used to quantitate the expression of genes of iron hemostasis using specific sense and antisense oligonucleotides (Table I). Primers were designed based on conserved sequences in human, rat, or mouse mRNAs (19–26). The PCR products were identified by formation of a cDNA fragment of the expected size and sequence analysis with a Sequenase Version 2 T7 DNA polymerase sequencing kit (USB). For sequencing IRP1, an additional primer (1A) was used with primer 2 to generate a 375-bp cDNA fragment. Preliminary experiments were carried out to optimize conditions for each set of oligonucleotides and to determine a range of RNA that would yield sufficient cDNA to visualize and exhibit proportionality to cDNA. Total RNA was used to minimize the differences in the quality and recovery of the samples (27). PCR reactions were carried out for 25–30 cycles consisting of 30 s at 957C, 45 s at the annealing temperature specific for each pair of oligonucleotides, and 1 min at 727C. The final step was for 10 min

AP: Archives

297

IRON-RELATED GENE REGULATION DURING SCURVY TABLE I

Oligos for Reverse Transcriptase–PCR 1. 5* Sense 2. 5* Antisense

Probe Tf

1. 2. 1. 2. 1. 2. 1. 2. 1A. 1. 2.

FH FL IRP1

TfR

5*CTGTGCCCTGATGGTACC 3* 5*GTCATCTCTGAACAGAAG 3* 5*GCCATCAACCGCCAGATC 3* 5*AGTAGTGACTGATTCACA 3* 5*ATTATTCCACCGACGTG 3* 5*GTTTTACCCCACTCATCTT 3* 5*GTCGGTGGTATTGAAGCA 3* 5*GTTAGCAATCGTAGCTCG 3* 5*TCAGGAATCATCCACCAG 3* 5*GTGAATGGATCTATAGTGATT 3* 5*ATTGCTCACAGTGAGCTT 3*

at 727C. Aliquots of 10 ml from each PCR reaction were withdrawn from under the mineral oil and electrophoresed on 2% agarose containing ethidium bromide. The gels were photographed and negatives were analyzed by densitometry.

RESULTS

Serum Iron and Unsaturated Iron-Binding Capacity (UIBC) To determine if decreased serum iron was an early event or was related to the later weight loss of scurvy, analysis (Table II) was performed in phases I (7–21 days, 0% weight loss) and II (20–30 days, weight loss). The time when weight loss began varied with individual animals, as observed previously (13, 15). Serum iron decreased to 75–54% of the normal level in phase I (0% weight loss) and additional decreases occurred

Size (bp)

Annealing temperature (7C)

219

52

288

47

256

52

234

52

375 390

47

after weight loss commenced (Table II, columns 3 and 4). The animal with 18.6% weight loss showed a smaller decrease in its iron level, although it was 40% of normal. UIBC represents serum transferrin with unsaturated iron-binding sites. This parameter increased during phase I (Table II, columns 5 and 6), confirming decreased serum iron levels and suggesting an increase in transferrin. After 20 days on the vitamin C-free diet, values remained at almost four times normal, regardless of whether animals were losing or gaining weight (Table II, columns 5 and 6). Serum Ferritin Levels Anti-human ferritin antibody was used to detect guinea pig ferritin. With the horse ferritin standard,

TABLE II

Serum Iron and UIBC during Scurvy a Days on vitamin C-free diet

Wt loss (%)

Iron (mg/dl)

Ironb (S/N, %)

UIBC (mg/dl)

UIBCc (S/N, %)

7 14 21* 23 20 22 22 25 26 26 30* 30*

0 0 0 4.8 5.1 7.7 9.2 13.4 18.6 20.4 24.5 28.0

195.4 192.5 140.7 133.6 88.6 63.2 43.6 27.3 104.9 37.3 14.1 27.1

75.9 74.7 54.6 51.9 34.4 24.5 16.9 10.6 40.7 14.5 5.5 10.6

110.0 138.0 397.8 432.0 340.6 342.6 479.6 387.2 362.4 472.8 443.3 496.7

1.1 1.4 3.9 4.2 3.3 3.4 4.7 3.8 3.6 4.7 4.5 4.9

a Results represent assays of serum from single animals except for those marked *, where results represent the means from two animals with similar amounts of weight loss. b Iron, normal levels 257.6 { 8.8 mg/dl (nÅ 10). c UIBC, normal levels 101.8 { 7.7 mg/dl (n Å 10).

/ m4949$9264

12-12-95 03:13:34

arca

AP: Archives

298

GOSIEWSKA, MAHMOODIAN, AND PETERKOFSKY

to measure total Tf. Human transferrin was used as a standard and specificity was established by incubating blots in the absence (Fig. 2A, top) or presence (Fig. 2A, bottom) of unlabeled iron. Quantitation of binding by densitometry is shown in Fig. 2B. There was an early increase in binding to about 1.8 times normal in serum of scorbutic guinea pigs that were still gaining weight (0% weight loss) and a further increase during the period of weight loss. Hepatic Cytosolic Aconitase Activity Since IRP1 loses or gains cytosolic aconitase activity in cultured cells dependent on iron availability, we determined whether there were changes in aconitase activity in liver during scurvy. High-speed centrifugation of liver homogenates assured separation of cytosolic aconitase from the mitochondrial form (17). The assay was validated with porcine aconitase. The decrease in A240 was proportional to time, substrate-dependent,

FIG. 1. Western blot analysis of serum and hepatic ferritin. (A) Serum or (B) liver extracts from normal (N) or scorbutic (S) guinea pigs either gaining weight (0%) or with varying extents of weight loss (5.1–28%) were electrophoresed, blotted, and incubated with or without anti-ferritin antibody. Horse apoferritin was used as a standard (STD) and the molecular weights of the H and L subunits in kDa are indicated on the right. Blots in A, bottom, and in B, S0 (scorbutic, 28% weight loss), were not treated with primary antibody to control for nonspecific binding.

the antibody reacted with two proteins corresponding to the 22- and 19-kDa ferritin subunits (Fig. 1A, top). There was no reaction with ferritin in the absence of primary antibody (Fig. 1A, bottom, and Fig. 1B, S0). Ferritin H and L subunits were detected in serum from two normal guinea pigs, although the levels varied (Fig. 1A, top, N). Levels of the L subunit decreased in scorbutic guinea pig serum (Fig. 1A, top), while the H subunit was still detectable in animals with 28% weight loss. Similar results were obtained on other Western blots. An even more striking decrease in ferritin protein levels was observed in the liver (Fig. 1B). In this case, however, both subunits were coordinately decreased to almost undetectable levels even in scorbutic guinea pigs still gaining weight. Serum Tf Levels Preliminary experiments with bovine holotransferrin showed that SDS–PAGE sample buffer was as efficient as acidic conditions at completely releasing iron from the protein (data not shown). Therefore, binding of 59Fe on membrane blots of serum proteins was used

/ m4949$9264

12-12-95 03:13:34

arca

FIG. 2. Serum transferrin levels during scurvy. Serum from normal (N) or scorbutic (S) guinea pigs either gaining weight (0%) or with varying extents of weight loss (5.1–28%) and transferrin standard (STD) were electrophoresed and ligand blotted as described under Experimental Procedures. (A) Autoradiograms after 1 day of exposure to X-ray film. The bottom blot was preincubated with unlabeled iron as a control for nonspecific binding. (B) Transferrin concentrations were quantitated by densitometry of ligand blots and values are expressed as values { SE for scorbutic serum relative to the mean value for normal serum. Error bars are not visible for low SE.

AP: Archives

IRON-RELATED GENE REGULATION DURING SCURVY

FIG. 3. Rates of hepatic aconitase activity. Liver extracts from a normal (N) or a scorbutic (S) guinea pigs with 28% weight loss were incubated without or with Fe/2 plus cysteine (Fe/SH) and then assayed for aconitase activity. Specific activities (units (U)/mg of protein) are indicated in parentheses.

and stimulated with Fe/2 plus cysteine (data not shown). Aconitase activity in guinea pig liver extracts also was substrate-dependent (data not shown). Activity in liver extracts from normal and scorbutic (28% weight loss) guinea pigs was stimulated by approximately 25–40% with Fe/2 plus cysteine (Fig. 3), in agreement with reports that cytosolic aconitase is approximately 80% active upon isolation (9). Further addition of dithiothreitol plus sulfide did not lead to greater activation (data not shown). The specific activity in the extract from the scorbutic guinea pigs was 28.0 and 24.5% of the activity in the normal extract, whether measured with or without Fe/2 plus cysteine (Fig. 3), respectively. A decrease in the activity during scurvy was confirmed after assaying liver extracts from the entire series of scorbutic guinea pigs with varying weight loss. A representative analysis is shown in Fig. 4. The specific activity in scorbutic liver, expressed as a percentage of normal, decreased to 43%, even in scorbutic guinea pigs that were still gaining weight (0% weight loss) after 3 weeks on the vitamin C-free diet. When animals entered phase II of scurvy and lost weight, however, there was a further decrease to approximately 28% of the normal activity (Fig. 4). Results were almost identical with or without preactivation. Expression of Hepatic mRNAs for Iron-Related Proteins cDNAs for iron-related proteins were obtained by RT–PCR using liver RNA from normal guinea pigs and primers based on sequences from cDNAs of other species (19–26). The products were sequenced to establish

/ m4949$9264

12-12-95 03:13:34

arca

299

their identity (Fig. 5). Partial sequences from the guinea pig were compared with human and rat cDNA sequences using the FastA component of the GCG computer program and a high degree of homology was found for the three species; the differences in bases and amino acids are shown in Table III. The cDNAs were used as probes with Northern blots but because of the relatively low expression of the ironrelated genes (data not shown), the results could not be used for quantitation. Therefore, RT–PCR was used to measure any changes in expression of these genes during vitamin C deficiency. Reactions were carried out with several concentrations of total liver RNA from normal and scorbutic guinea pigs and optimal conditions for each set of specific oligonucleotide primers. The cDNA products obtained are shown in Fig. 6. Negatives from photographs of the ethidium bromidestained gels were used to quantitate the results by densitometry (Table IV). Expression of four of the five iron-related genes examined was decreased during vitamin C deficiency. The relative amount of hepatic Tf mRNA (Table IV, column 3) was not affected during the first 2 weeks of scurvy (7–14 days) but it decreased to 67.1% of normal after 21 days, when scorbutic animals were still gaining weight (0% weight loss). At that point, serum iron had decreased to 55% of normal. After weight loss commenced (phase II, 20–30 days), levels of Tf mRNA fluctuated but did not decrease much below 50% of normal. The relative amounts of mRNAs for the ferritin subunits were differentially affected by vitamin C deficiency. The ferritin L subunit mRNA concentration (Table IV, column 4) was significantly decreased by 2

FIG. 4. Hepatic aconitase activity as a function of weight loss during scurvy. Enzyme activity was measured either after activation with Fe/2 plus cysteine (/Fe/SH) or in untreated extracts as described under Experimental Procedures and specific activities were calculated from five timepoints. Results are expressed as the percentage of activity in scorbutic extracts { SE relative to the mean of normal activities. Error bars are not visible for low SE. The normal values (n Å 8) in units/mg were 199.2 { 20 for unactivated and 278.4 { 32.5 for activated samples. Two separate analyses of extracts gave similar results and the data from one are presented.

AP: Archives

300

GOSIEWSKA, MAHMOODIAN, AND PETERKOFSKY

FIG. 5. Amino acid and base sequences of guinea pig cDNAs. The sequences for Tf, FL, FH, IRP1, and TfR were obtained as described under Experimental Procedures. Only partial sequences could be obtained since PCR products were not cloned. The amino acid sequences were determined using the DNA Strider computer program.

weeks, and in the growing scorbutic animal at 21 days (0% weight loss) the level was 37% of normal. The level remained low throughout phase II of scurvy. In con-

/ m4949$9264

12-12-95 03:13:34

arca

trast, the ferritin H subunit mRNA (Table IV, column 5) was not affected significantly until scorbutic animals had lost 24–28% body weight.

AP: Archives

301

IRON-RELATED GENE REGULATION DURING SCURVY TABLE III

Sequence Differences between Guinea Pig, Human, and Rat cDNAs for Iron-Related Genes Differences compared to Guinea Pig Human

Rat

cDNA

Length (bp)

Bases

aa

Bases

aa

Tf FL FH IRP1 TfR

159 147 255 345 189

37 15 24 28 26

17 6 4 3 9

55 24 NA 35 32

22 9 NA 6 11

Note. Guinea pig sequences were obtained from RT–PCR products as described under Experimental Procedures. The FastA computer program was used for sequence comparisons between different species. aa, amino acids; NA, not available in the program.

Although IRP1 mRNA is not affected by iron depletion in cultured cells (7), its level was altered during vitamin C deficiency (Table IV, column 6). As in the case of FL and Tf mRNAs, there was a significant decrease in IRP1 mRNA levels even in guinea pigs that were still gaining weight (14 and 21 days). Levels remained relatively constant throughout the later stage of scurvy. Contrary to effects on the other iron-related genes, relative TfR mRNA concentrations were significantly increased after only 7 days of vitamin C deficiency (Table IV, column 7). Concentrations continued to increase during phase I and II of scurvy and reached a maximum level almost five times normal.

FIG. 6. Changes in Tf, FL, FH, IRP1, and TfR mRNAs during scurvy as measured by RT–PCR. Three different RNA concentrations were used, except for TfR, to prepare and amplify cDNA which was then electrophoresed. Gels were stained with ethidium bromide and photographed. The sizes of PCR products in base pairs (bp) are indicated.

/ m4949$9264

12-12-95 03:13:34

arca

DISCUSSION

The present study confirms that depletion of vitamin C in guinea pigs leads to iron deficiency and, in addition, it defines the time of onset of iron deficiency as early in phase I of scurvy. The results also demonstrate changes in the expression of genes for proteins involved in iron homeostasis early in scurvy, when serum iron levels were already decreased and weight loss had not yet occurred. Therefore, the regulation of these proteins during scurvy appears to be related mainly to iron deficiency rather than to induction of insulin-like growth factor binding proteins or changes in other circulating hormones that result from the onset of anorexia (13–15, 28). Iron deficiency during the early phase of scurvy may be caused by decreased absorption of iron from the gastrointestinal tract (1) as a direct result of ascorbate depletion. Ascorbate levels in guinea pig tissues are decreased by about 90% after 1 week on a vitamin C-free diet and reach 4–5% of normal levels after 2 weeks (13, 16), and blood levels closely parallel those in tissues (2). At later stages, however, decreased iron intake caused by anorexia, as well as iron loss related to peripheral blood loss during hemorrhaging, may contribute to the problem. Changes in the relative abundance of ferritin mRNAs during scurvy-induced iron deficiency in vivo suggest that their regulation may not be solely through IREs, as appears to be the case in cultured cells (7, 8). In iron-depleted cultured cells, ferritin translation is inhibited by binding of IRP1 to IREs in H and L mRNAs, but the levels of the mRNAs do not change (7). In hepatoma cell cultures, ascorbate appears to enhance this type of regulation of ferritin by iron and both ferritin subunits are equally affected (29). Regulation of ferritin levels in scorbutic guinea pigs, however, was quite different. There was a preferential decrease in the concentration of the L subunit in serum although both subunits decreased simultaneously in liver. There also

AP: Archives

302

GOSIEWSKA, MAHMOODIAN, AND PETERKOFSKY TABLE IV

Hepatic Gene Expression during Scurvya Days on vitamin C-free diet

Wt lossb (%)

7 14 21 20 22 26 30 30

0 0 0 5.1 9.2 18.6 24.0 28.0

Tf 104.7 106.2 67.1 52.0 78.8 86.3 88.8 48.4

{ { { { { { { {

FL 5.0 7.2 2.3 4.4 2.5 3.2 5.9 3.3

94.0 81.9 36.9 54.1 67.9 22.7 43.3 29.1

{ { { { { { { {

FH 8.8 1.6 4.5 8.4 4.8 2.1 1.1 7.1

90.3 86.5 108.4 133.7 118.3 95.2 81.0 70.1

{ { { { { { { {

IRP1 4.4 4.4 9.4 9.1 13.8 5.9 7.3 4.9

94.8 88.3 47.3 57.7 49.9 57.1 64.1 66.9

{ { { { { { { {

4.1 4.4 6.3 4.3 4.3 1.2 2.6 1.5

TfR 122.5 166.0 222.5 243.0 348.0 477.5 446.7 235.5

{ { { { { { { {

6.3 10.3 6.2 24.2 18.3 40.2 42.0 10.6

a Concentrations of mRNAs were determined by densitometry of RT–PCR products derived from three different total RNA concentrations for each sample. Results are expressed as the means { SE for each scorbutic animal as a percentage of the mean from two normal animals (S/N 1 100). b Scorbutic animals with 0% weight loss were gaining weight normally.

was differential regulation of the ferritin mRNAs. Hepatic mRNA for the L subunit decreased during phase I of vitamin C deficiency, while the concentration of the H subunit mRNA was not reduced until late in phase II. This pattern of mRNA regulation could explain differential changes in the ferritin subunits in serum, but it does not account for the coordinate disappearance of the subunits in liver. One possibility is that cytoplasmic stores of ferritin in the liver are degraded in the absence of ascorbate while secretion of newly synthesized ferritin into the circulation reflects hepatic mRNA expression. It has been postulated that ascorbic acid blocks the lysosomal uptake of ferritin and subsequent conversion of ferritin to nonutilizable hemosiderin (5), so increased degradation in the absence of ascorbate might be expected. Differential regulation of ferritin subunits has been observed in other in vivo systems. Synthesis of the L subunit in liver was preferentially induced by iron administered to rats (30) with concomitant increases in the rate of transcription of the L subunit gene and the steady-state level of ferritin L mRNA, while expression of the H subunit gene did not increase significantly (31). Similarly, in patients with iron-overload diseases, only the level of mRNA for the ferritin L subunit was increased in liver compared to normal controls (32). Regulation of TfR and ferritin subunit expression in cultured cells by changes in iron levels appears to be determined by whether or not iron is bound to IRP1. In iron-depleted cells, IRP1 cannot bind iron to form an iron–sulfur cluster and thus it shifts from aconitase activity to IRE binding activity (7, 8). This results in stabilization and an increase in the level of TfR mRNA and TfR (7, 8). Regulation of TfR mRNA during iron deficiency associated with scurvy in guinea pigs appeared to be similar to that observed in cultured cells in that there was an increase in TfR mRNA concentra-

/ m4949$9264

12-12-95 03:13:34

arca

tions. This increase, however, may not be due solely to the interaction of IRP1 with TfR RNA, since the level of IRP1/aconitase was significantly reduced at the mRNA and protein levels. The decrease in cytosolic aconitase activity in liver of scorbutic guinea pigs probably represents a decrease in enzyme protein rather than conversion to apoenzyme with RNA binding activity. This conclusion is based on the observation that there was relatively little activation of enzyme by Fe/2 and reducing agents in normal or scorbutic liver extracts so that the activity in scorbutic liver relative to normal was low regardless of activation. Therefore, our results suggest that in vivo there may be a second mechanism by which iron levels regulate the TfR mRNA which does not involve IRP binding to IREs. Another interesting observation was the discordant regulation between serum Tf protein and hepatic Tf mRNA expression. There was an increase in total serum Tf protein levels, as measured by 59Fe ligand blots and as UIBC, which represents serum transferrin with unoccupied iron-binding sites. In contrast, there was a decrease in hepatic Tf mRNA. The reverse of this effect was recently observed in patients with iron-overload diseases (32); serum Tf levels were decreased while hepatic mRNA expression for Tf was increased. These results suggest that there are additional mechanisms for regulating Tf in the liver. Nutritional iron deficiency in newly hatched chicks also led to increased levels of circulating Tf but, in contrast to our results, this was accompanied by an increase in Tf synthesis and an increase in Tf mRNA levels in the liver (33). In summary, our results suggest that iron deficiency associated with scurvy is induced early and leads to changes in expression of iron-related genes that do not involve the induction of IGFBPs, which occurs at a later stage. In contrast to iron-depleted cell cultures, however, regulation of ferritin H and L and TfR genes dur-

AP: Archives

IRON-RELATED GENE REGULATION DURING SCURVY

ing iron deficiency in vivo does not appear to occur solely through IRP1 binding to IREs. Because of the early onset of iron deficiency, the function of proteins other than those involved in iron homeostasis may be impaired and cause other abnormalities observed during vitamin C deficiency, such as anemia (1). We recently found that defective wound healing during phase I of scurvy is not related to the role of ascorbate in proline hydroxylation or to the induction of IGFBPs (16). It is possible that iron deficiency could contribute to this defect. REFERENCES 1. Chatterjee, C. (1967) in The Vitamins (Sebrell, W. H., Jr., and Harris, R. S., Eds.), pp. 407–457, Academic Press, New York. 2. Veen-Baigent, M. J., Ten Cate, A. R., Bright-See, E., and Rao, A. V. (1975) in Annals of the New York Academy of Science (King, C. G., and Burns, J. J., Eds.), pp. 339–354, N. Y. Acad. Sci., New York. 3. Cacciola, E., Consoli, U., Giustolisi, R., and Cacciola, E. (1994) Haematologica 79, 96–97. 4. Nienhuis, A. W. (1981) N. Engl. J. Med. 304, 170–171. 5. Hoffman, K. E., Yanelli, K., and Bridges, K. R. (1991) Am. J. Clin. Nutr. 54, 1188S–1192S. 6. Peterkofsky, B., Tschank, G., and Luedke, C. (1987) Arch. Biochem. Biophys. 254, 282–289. 7. Klausner, R. D., Rouault, T. A., and Harford, J. B. (1993) Cell 72, 19–28. 8. Theil, E. C. (1994) Biochem. J. 304, 1–11. 9. Kennedy, M. C., Mende-Mueller, L., Blondin, G. A., and Beinert, H. (1992) Proc. Natl. Acad. Sci. USA 89, 11730–11734. 10. Basilion, J. P., Kennedy, M. C., Beinert, H., Massinople, C. M., Klausner, R. D., and Rouault, T. A. (1994) Arch. Biochem. Biophys. 311, 517–522. 11. Henderson, B. R., Seiser, C., and Ku¨hn, L. C. (1993) J. Biol. Chem. 268, 27327–27334. 12. Samaniego, F., Chin, J., Iwai, K., Rouault, T. A., and Klausner, R. D. (1994) J. Biol. Chem. 269, 30904–30910. 13. Peterkofsky, B. (1991) Am. J. Clin. Nutr. 54, 1135S–1140S. 14. Peterkofsky, B., Gosiewska, A., Kipp, D. E., Shah, V., and Wilson, S. (1994) Growth Factors 10, 229–241.

/ m4949$9264

12-12-95 03:13:34

arca

303

15. Gosiewska, A., Wilson, S., Kwon, D., and Peterkofsky, B. (1994) Endocrinology 134, 1329–1339. 16. Kipp, D., Wilson, S., Gosiewska, A., and Peterkofsky, B. (1995) Wound Repair Regener. 3, 192–203. 17. Henson, C. P., and Cleland, W. W. (1967) J. Biol. Chem. 242, 3833–3838. 18. Van Renswoude, J., Bridges, K. R., Harford, J. B., and Klausner, R. D. (1982) Proc. Natl. Acad. Sci. USA 79, 6186–6190. 19. Yang, F., Lum, J. B., McGill, J. R., Moore, C. M., Naylor, S. L., van Bragt, P. H., Baldwin, W. D., and Bowman, B. H. (1984) Proc. Natl. Acad. Sci. USA 81, 2752–2756. 20. Huggenvik, J. I., Idzerda, R. L., Haywood, L., Lee, D. C., McKnight, G. S., and Griswold, M. D. (1987) Endocrinology 120, 332–340. 21. Boyd, D., Vecoli, C., Belcher, D. M., Jain, S. K., and Drysdale, J. W. (1985) J. Biol. Chem. 260, 11755–11761. 22. Torti, S. V., Kwak, E. L., Miller, S. C., Miller, L. L., Ringold, G. M., Myambo, K. B., Young, A. P., and Torti, F. M. (1988) J. Biol. Chem. 263, 12638–12644. 23. Rouault, T. A., Tang, C. K., Kaptain, S., Burgess, W. H., Haile, D. J., Samaniego, F., McBride, O. W., Harford, J. B., and Klausner, R. D. (1990) Proc. Natl. Acad. Sci. USA 87, 7958– 7962. 24. Yu, Y., Radisky, E., and Leibold, E. A. (1992) J. Biol. Chem. 267, 19005–19010. 25. McClelland, A., Ku¨hn, L. C., and Ruddle, F. H. (1984) Cell 39, 267–274. 26. Roberts, K. P., and Griswold, M. D. (1990) Mol. Cell. Endocrinol. 14, 531–542. 27. Foley, K. P., Leonard, M. W., and Engel, J. D. (1993) Trends Genet. 9, 380–385. 28. Palka, J., Bird, T. A., Oyamada, I., and Peterkofsky, B. (1989) Growth Factors 1, 147–156. 29. Toth, I., Rogers, F. T., McPhee, J. A., Elliott, S. M., Abramson, S. L., and Bridges, K. R. (1995) J. Biol. Chem. 270, 2846–2852. 30. Bomford, A., Conlon-Hollingshead, C., and Munro, H. N. (1981) J. Biol. Chem. 256, 948–955. 31. White, K., and Munro, H. N. (1988) J. Biol. Chem. 263, 8938– 8942. 32. Pietrangelo, A., Rocchi, E., Ferrari, A., Ventura, E., and Cairo, G. (1991) Hepatology 14, 1083–1089. 33. McKnight, G. S., Lee, D. C., Hemmaplardh, D., Finch, C. A., and Palmiter, R. D. (1980) J. Biol. Chem. 255, 144–147.

AP: Archives