Identification of differentially expressed genes in response to dietary iron deprivation in rat duodenum

James F. Collins, Christina A. Franck, Kris V. Kowdley, Fayez K. Ghishan


We sought to identify novel genes involved in intestinal iron absorption by inducing iron deficiency in rats during postnatal development from the suckling period through adulthood. We then performed comparative gene chip analyses (RAE230A and RAE230B chips; Affymetrix) with cRNA derived from duodenal mucosa. Real-time PCR was used to confirm changes in gene expression. Genes encoding the apical iron transport-related proteins [divalent metal transporter 1 (DMT1) and duodenal cytochrome b] were strongly induced at all ages studied, whereas increases in mRNA encoding the basolateral proteins iron-regulated gene 1 and hephaestin were observed only by real-time PCR. In addition, transferrin receptor 1 and heme oxygenase 1 were induced. We also identified induction of novel genes not previously associated with intestinal iron transport. The Menkes copper ATPase (ATP7a) and metallothionein were strongly induced at all ages studied, suggesting increased copper absorption by enterocytes during iron deficiency. We also found significantly increased liver copper levels in 7- to 12-wk-old iron-deficient rats. Also upregulated at most ages examined were the sodium-dependent vitamin C transporter, tripartite motif protein 27, aquaporin 4, lipocalin-interacting membrane receptor, and the breast cancer-resistance protein (ABCG2). Some genes also showed decreased expression with iron deprivation, including several membrane transporters, metabolic enzymes, and genes involved in the oxidative stress response. We speculate that dietary iron deprivation leads to increased intestinal copper absorption via DMT1 on the brush-border membrane and the Menkes copper ATPase on the basolateral membrane. These findings may thus explain copper loading in the iron-deficient state. We also demonstrate that many other novel genes may be differentially regulated in the setting of iron deprivation.

  • iron deficiency anemia
  • ATP7a
  • copper transport
  • intestine
  • microarray
  • gene chip

iron is a critical element for many metabolic processes (4). However, excess iron stores can lead to oxidative damage, as free iron readily participates in redox reactions within cells that can lead to the production of reactive oxygen species and associated cellular and molecular damage. Thus body iron levels must be tightly controlled. Body iron stores are normally regulated at the level of intestinal absorption (7). Most dietary iron is absorbed in the proximal small intestine, with the greatest overall absorption rates and adaptive response observed in the duodenum (42). Several physiological effectors are known to modulate dietary iron absorption in mammals, as reviewed recently by Hentze et al. (18). These putative modulators of intestinal iron absorption, which include hepatic stores, erythroid, hypoxia, and inflammatory mediators, act directly on the intestinal epithelium to control iron absorption. A recently identified antimicrobial peptide, hepcidin, is recognized as the hormone responsible for regulating intestinal iron absorption in response to body iron status. Hepcidin, which is synthesized in the liver and secreted into the circulation in response to increased body iron stores, has been shown to decrease intestinal iron transport in mice (24). In addition, hepcidin gene expression is related to body iron stores in normal humans (13). Furthermore, iron feeding in rats results in a prompt increase in hepcidin levels, which inhibits iron absorption (12).

Several proteins involved in duodenal iron absorption have recently been identified with the use of iron deprivation in murine models that leads to increased expression of genes related to iron transport. Duodenal cytochrome b (Dcytb) is a ferric reductase that reduces iron to Fe2+ (27), which is then transported into epithelial cells by the divalent metal transporter 1 [DMT1; also called DCT1 (16) and nRAMP2 (39)]. Within enterocytes, iron is either sequestered within protein complexes of ferritin or trafficked to the basolateral membrane for export into the circulation. Extrusion across the basolateral membrane is likely accomplished by the coordinated action of the basolateral iron transport protein iron-regulated gene 1 [IREG1 (28); also called MTP1 (1) and ferroportin (8)] and hephaestin, which oxidizes iron for binding to transferrin and distribution throughout the body in the circulation (40). Despite the identification of these described genes over the past several years, a complete understanding of the molecular events associated with intestinal iron absorption has not yet been achieved. This fact is exemplified by the microcytic anemia mice (11) and Belgrade rats (10), which are able to absorb substantial amounts of dietary iron despite the lack of normal DMT1. Additionally, sex-linked anemia (sla) mice, which have a deletion in the hephaestin gene that eliminates 194 amino acids in the putative protein, have substantial accumulation of iron within enterocytes and moderate to severe hypochromic, microcytic anemia (40). These mice nevertheless have the ability to absorb some dietary iron despite the possible mistargeting of the hephaestin protein to a supranuclear compartment (rather than the basolateral membrane) (23).

The goal of the current study was to use microarray techniques to examine changes in gene expression in the rat duodenum associated with iron deprivation. We accomplished this by depriving rats of dietary iron at different stages of development and then performing comparative gene chip analyses with cRNA derived from duodenal mucosa of groups of rats killed at 8 days, 21 days, 6 wk, 12 wk, and 36 wk of age. This novel approach allowed us to track the relative expression of iron-responsive genes longitudinally over the entire course of postnatal development. We also examined the effect of iron deficiency on known iron transport genes with gene chip and quantitative real-time PCR methods.


Experimental animals.

Sprague-Dawley rats were obtained from Harlan (Madison, WI) and were housed in the University of Arizona Animal Care facility. Modified AIN-93G rodent diets were obtained from Dyets (Bethlehem, PA), which contained either 198 parts per million (ppm) Fe (DYET no. 115135; high-Fe diet; same Fe content as standard rat chow) or 3 ppm Fe (DYET no. 115102; low-Fe diet). The diets were identical except for the addition of pure ferric citrate to the high-Fe diet. Tap water was initially tested for iron content, and it did not show any detectable iron. Rats were supplied with food and tap water ad libitum. To produce 8-day-old (e.g., suckling) and 21-day-old (e.g., weanling) iron-deprived rats, 8-wk-old female rats were placed on either the high- or low-Fe diet for 2 wk and were then paired with 8-wk-old male rats. Females remained on the same diet throughout gestation and lactation. Pups were cross-fostered to different dams on the same diet to maintain a constant dam-to-pup ratio. To produce 6-wk-old iron-deprived rats, 8-wk-old female rats were placed on either the high- or low-Fe diet and immediately paired with 8-wk-old male rats. The dams remained on either the high- or low-Fe diet throughout gestation and lactation, and once the pups were weaned they remained on the same diet that the dam was fed. For 12-wk-old rats, 3-wk-old rats were placed on high- or low-Fe diets for 9 wk and then killed. To produce 36-wk-old iron-deprived rats, 8-wk-old rats were placed on high- or low-Fe diets for 28 wk and then killed.

For all studies, only male rats were used and groups of three to five animals were considered as one group (n = 1). Each experiment was repeated three times with samples derived from different groups of iron-deficient or iron-replete rats. Rats were anesthetized by CO2 narcosis, and blood was obtained by cardiac puncture. Blood was sent to the University of Arizona Animal Care Pathology Services laboratory for complete blood cell counts (CBC) with differential analysis of blood samples. Rats were then killed by cervical dislocation, and 2–5 in. of the small intestine (depending on age) just distal to the pyloric sphincter were removed. The intestinal segment was flushed with PBS and opened lengthwise, and light mucosal scrapes were taken. Approximately equal amounts of mucosal tissue were mixed in the same tube from all the rats in that group, with each individual sample being immediately frozen in liquid nitrogen. All samples were stored at −80°C until use. All animal procedures were approved by the University of Arizona Institutional Animal Care and Use Committee.

RNA purification.

RNA was purified from mucosal tissue with TRIzol reagent (Invitrogen) as previously described (20). Total RNA (100 μg) was then further purified with the RNeasy Mini Kit (Qiagen) according to the manufacturer's suggested protocol. The RNA was eluted at the final step twice with the same 30 μl of RNAse-free water and quantified by ultraviolet (UV) spectrophotometry. RNA was then visualized by denaturing agarose gel electrophoresis, and RNA concentrations were adjusted by densitometry of the gel. Only high-quality RNA, as judged by intactness of the ribosomal bands, was used for gene chip analyses.

Preparation of samples for gene chip analyses.

cRNA was produced from duodenal mucosa RNA samples essentially according to the manufacturer's instructions (Affymetrix; Expression Analysis Technical Manual). Experimental repetitions done in triplicate at each age were performed at the same time with cRNA samples derived from different groups of experimental rats that were either iron deficient or iron replete. RNA was purified from all six groups at each age simultaneously, followed by cRNA production, and then 1 μl of each cRNA sample was analyzed by gel electrophoresis. After gel electrophoresis, densitometry was performed and the most concentrated cRNA sample was quantified by UV spectrophotometry. Subsequently, the relative concentration of all other cRNA samples from that age group was calculated according to optical density of the most concentrated sample. Only cRNA samples that showed a smear of material from high to low molecular weight (e.g., significantly above and below the ribosomal RNA bands) were used for gene chip analyses. By these procedures, we ensured that equal amounts of high-quality cRNA at each age were hybridized with the gene chips.

Gene chip analyses.

cRNA was fractionated, hybridization cocktails were prepared, and then rat genome RAE230A and RAE230B chips were hybridized with 10 μg of cRNA by standard Affymetrix protocols. Hyb cocktails were hybridized to only one chip and were then discarded. Chips were immediately washed and stained with the GeneChip Fluidics Station 400 (Affymetrix) utilizing the EukGE-WS2v4 fluidics protocol. After chips had been washed and stained, they were scanned twice with the Agilent Gene Array Scanner (Affymetrix).

Gene chip data analysis and reduction.

After scanning the chips, absolute CHiP (.CHP) files were generated by Microarray Suite software (MAS, version; Affymetrix) with scaling set for all probe sets with a target value of 500 and normalization set for all probe sets. Other parameters were set at default values. Comparison CHiP files were then generated that compared all three high-Fe data sets individually with all three low-Fe data sets at each age, with all analysis settings remaining constant. All subsequent data analysis and reduction was performed with Data Mining Tool software (DMT, version 3.0; Affymetrix).

Within the DMT software, first Detection and Change was pivoted for all nine comparisons. To identify genes that showed increased expression with iron deprivation at each age, the Count and Percent function was then used with Present, Marginal, Increasing, and Marginally Increasing selected. Probe sets were then selected that showed increases in six to nine of the nine comparisons, and the probe set was saved. This new probe set was then selected, and signal log ratio (SLR) data were pivoted, followed by averaging the SLRs from the nine comparisons. Probes sets were then selected that had average SLRs of 0.6 (i.e., 1.5-fold increase) and higher. A similar strategy was used to identify genes that were downregulated with iron deficiency.

Probe sets from these analyses at each age were then uploaded into the Expression Queries folder on the Affymetrix web site through the NetAffx Analysis Center (, using the Batch Query function under the Query-Expression label. The Intersection tool was then used to identify probe sets that behaved similarly at the different ages. Subsequently, lists of probe set IDs of genes that increased or decreased across four or five ages were saved. These lists were then imported back into the DMT software, and SLR and signal intensity data from the high-Fe and low-Fe groups were pivoted and averaged. This overall data analysis and reduction strategy allowed us to come up with short lists of candidate duodenal genes that were regulated by body or enterocyte iron status.

Final gene chip data are presented in tables that show genes that increased or decreased expression across four or five ages studied (genes that were regulated in four of the ages are shown in Supplemental Tables 16, available online at Shown are gene name, gene symbol, GenBank accession number for the Affymetrix target sequence, and aliases/biological function. If the gene symbol is not known, “??” was placed in the table at that position. If the gene name is listed as “similar” to a gene, this is the name assigned by Affymetrix for that individual probe set. For some genes, the cDNA has not been cloned from rats, and if this is the case, percent homology to known mouse or human cDNA clones is shown. Furthermore, in some cases, homology was only found to mouse or human chromosomal regions, so 10–20 kb of these regions were searched against DNA sequence databases to see what gene(s) was present in this region. If a known gene was present there, we listed these genes as “on the same chromosomal region.” Other tables show gene symbol, GenBank accession number, the average fold change from the nine comparisons done with DMT software at each age, and the average expression levels from the three high-Fe groups and the three low-Fe groups at each age. All gene chip data can be found in the GEO repository under accession no. GSE1892 (

View this table:
Table 1.

CBC analysis of rat blood taken at death

Quantitative real-time PCR.

Total RNA from the three high-Fe or three-low Fe groups at each age was combined in equal amounts (33.3 μg/group) and was further purified with the RNeasy Mini Kit (Qiagen). RNA samples were treated for DNA contamination with the DNA-Free kit (Ambion) by the manufacturer's suggested protocol. RNA was then visualized by denaturing agarose gel electrophoresis, and concentrations were adjusted by densitometric analysis and UV spectrophotometry. First-strand cDNA was then produced from 250 ng of total RNA with random primers and the SuperScript II Reverse Transcription kit (Invitrogen) by standard methods. Two microliters of the cDNA reaction were then used as a template for RT-PCR with a SYBRgreen Master Mix (Applied Biosystems) and primers designed with the Primer Designer software (version 4). PCR amplification was performed with the Bio-Rad iCycler with the following cycling parameters: 95°C for 10 min, followed by 45 cycles of 95°C for 15 s and then 60°C for 1 min. Taqman 18S rRNA primers (Applied Biosystems) were also used with 2 μl of RT reaction and Taqman PCR Master Mix (Applied Biosystems) under identical conditions to amplify 18S, which was used as an internal standard. Data from each gene of interest were normalized for 18S within each individual RT reaction. Each amplification was repeated three times from different RT reactions, and the data were averaged. The following primers were utilized for PCR amplification: for the Menkes Copper ATPase (APT7a; accession no. NM_052803): forward 5′-AAGTGGATGTGGAACTTGTA-3′ and reverse 5′-CTGGAACGACCTTAATGATA-3′; for DMT1 [+/−iron-responsive element (IRE) transcript; accession no. NM_013173]: forward 5′-GTGGAGTTGGCAATCATTGG-3′ and reverse 5′-ATCTGCGATGGTGATGAGGA-3′; for Dctyb (accession no. XM_215749): forward 5′-GCTGCAGACGCAGAGTTAAG-3′ and reverse 5′-ATATACATCGGCCTATGGCT-3′; for IREG1 (accession no. AF394785): forward 5′-GTGGATAAGAATGCCAGACT-3′ and reverse 5′-CGCAGAGAATGACTGATACA-3′; for hephaestin (accession no. NM_133304): forward 5′-GACTGTGGTGTTCAAGAATA-3′ and reverse 5′-GAATGTTCCACTGGTAAGTA-3′; and for the type IIb sodium-phosphate cotransporter (NaPi-IIb; accession no. AF157026): forward 5′-GGCAACACATTGAGGAGTTC-3′ and reverse 5′-CAGCGATACTTGGCAGAGAT-3′.

Statistical analysis.

Data from blood analysis of rats, animal weights, gene chip quality control parameters, and real-time PCR data were analyzed by Student's t-test, and P values <0.05 were considered significant.


Rat health status.

Blood analysis of the rats revealed that iron-deprived rats at 8 days, 21 days, 6 wk, and 12 wk of age showed signs of hypochromic, microcytic anemia. These signs included decreased red blood cell counts, hemoglobin levels, hematocrit, mean corpuscular volume, and mean corpuscular hemoglobin (Table 1). Furthermore, red blood cell distribution width was significantly increased in iron-deficient rats at these ages. The 36-wk-old rats, despite being deprived of dietary iron for 28 wk, showed no differences between control and iron-deprived groups in any parameters measured with the CBC blood test (Table 1).

Rats were weighed before death, and data from each age and dietary group were averaged. Iron-deprived rats at 8 days (sucklings; 18.38 ± 1.25 g high Fe vs. 17.14 ± 1.05 g low Fe; P = 0.004), 21 days (weanlings; 59.98 ± 4.64 g high Fe vs. 43.81 ± 5.51 g low Fe; P < 0.0001), and 6 wk (168.33 ± 14.79 g high Fe vs. 134.33 ± 14.38 g low Fe; P < 0.0001) of age weighed less than their control groups that had been fed a high-Fe diet. However, 12-wk-old [375.29 ± 27.2 g high Fe vs. 361.63 ± 33.33 g low Fe; not significant (NS)] and 36-wk-old (512.90 ± 37.81 g high Fe vs. 544.63 ± 37.29 g low Fe; NS) rats showed no differences in weights between the iron-replete and iron-deficient groups.

Gene chip control parameters.

The key quality control parameters for the gene chip experiments presented in this manuscript are background, raw Q, scale factor, β-actin 3′-to-5′ ratio, GAPDH 3′-to-5′ ratio, and percent present calls (Affymetrix). Data from three 230A and 230B chips at each age and for either high- or low-Fe diet were averaged, and the means ± SD were determined. Average backgrounds were all below the threshold of 100, and average raw Q values were all below the 3.5 threshold (with the exception of the 12-wk low-Fe group, which was slightly above 3.5; however, all other control parameters were within acceptable ranges for this group). Scale factor was less than threefold different between data sets that were compared with one another, and this parameter was also within the manufacturer's suggested guidelines. Finally, β-actin and GAPDH 3′-to-5′ ratios were significantly below the 3.0 threshold, and percent present calls were similar between experiments.

Gene chip data.

Genes expressed differentially in iron-replete vs. iron-deficient states were examined and were classified as either increased or decreased. We found that some genes increased or decreased at all five ages studied and also that some other genes increased or decreased in four of the five ages studied. Table 2 shows genes that increased in all five ages studied. These include DMT1, Dcytb, transferrin receptor 1, metallothionein, tripartite motif protein 27, the Menkes copper ATPase (ATP7a), glycerol-3-phosphate acyltransferase, factor-responsive smooth muscle protein, pyruvate carboxylase, phosphoglucomutase-related protein, acidic calponin 3, glutathione peroxidase 3, lipocalin interacting membrane receptor, gap junction protein β2, retinoblastoma binding protein 7, progressive ankylosis, and integrin α6. Genes that showed increased expression in all ages except sucklings (e.g., 8 days of age) and genes that were upregulated at all ages except in 36-wk-old rats are shown in Supplemental Tables 1 and 3. The corresponding average fold increases from the nine comparisons at each age and average expression levels in the high- and low-Fe groups for genes that increased expression at all five ages are presented in Table 3, and the same data for genes that increased in four out of five ages is presented in Supplemental Tables 2 and 4. Genes that showed decreased expression at all five ages are shown in Table 4, and average fold decreases and expression levels for these genes are presented in Table 5. These genes were aminopeptidase A, death-associated protein (similar to), monoamine oxidase A, and lysosomal apyrase-like 2. Genes that decreased at all ages except in sucklings are shown in Supplemental Table 5, and the average fold decrease and expression levels for these genes are presented in Supplemental Table 6.

View this table:
Table 2.

Genes that increased in all 5 ages

View this table:
Table 3.

Fold change and expression data for genes that increased in all 5 ages

View this table:
Table 4.

Genes that decreased in all 5 ages

View this table:
Table 5.

Average fold changes and expression levels for genes that decreased in all 5 ages

View this table:
Table 6.

Confirmation of gene chip data by quantitative real-time PCR

Real-time PCR confirmation of gene chip data.

To demonstrate the validity of our data and the overall reliability of gene chip analyses to quantify relative expression levels of various genes, we performed extensive real-time PCR assays for genes of interest (e.g., those known to be iron responsive and other novel genes that are potentially important for intestinal iron absorption). All primers used for real-time PCR analysis were from within the open reading frame of the transcripts (unlike the Affymetrix target sequences, which are typically within the 3′ untranslated region of transcripts). The results of these studies are presented in Table 6, which shows the average fold change from three independent assays along with statistical analysis of the data.


We used a novel microarray approach to examine the effect of iron deprivation on duodenal gene expression during various stages of postnatal development in the rat. We chose to study rats at certain ages that coincided with important developmental steps. Rats at 8 days of age are sucklings, rats at 21 days of age are at the suckling/weaning transition, and 6-wk-old rats are considered adolescents (just before sexual maturation). We also chose to study rats at two ages during adulthood: the first group at 12 wk of age after iron deprivation since the time of weaning and the other group at 36 wk of age that were not deprived of iron until they were sexually mature adults (e.g., starting at 8 wk of age). Using this approach, we were able to assess the effect of iron deficiency and iron deprivation on duodenal gene expression across key developmental stages.

The key finding of our studies was that the brush-border iron transport machinery (e.g., DMT1 and Dcytb) was significantly induced with iron deprivation throughout postnatal development, whereas the genes encoding the basolateral proteins thought to be necessary for iron extrusion from intestinal epithelial cells (e.g., IREG and hephaestin) were not as significantly or consistently upregulated. However, these lesser changes in expression of IREG1 and hephaestin could translate into significant changes in protein activity and may thus have true physiological importance. Other genes known to be involved in iron metabolism were also induced with iron deprivation [e.g., heme oxygenase 1 (HO1) and transferrin receptor 1 (TFR1)]. Furthermore, we identified iron-dependent regulation of many genes that have not previously been associated with iron metabolism. Some of these genes, such as the Menkes copper ATPase (ATP7a), the sodium-dependent vitamin C transporter, and others, may be directly involved in intestinal iron absorption, as they were induced not only in iron-deficient, anemic rats but also in iron-deprived rats that did not display a noticeable iron-deficient phenotype (e.g., the 36-wk-old group).

To ensure that our studies accurately duplicated iron deficiency in humans, we developed a natural feeding strategy with diets that contained high (198 ppm; actually levels identical to commercial rodent chows) and low (3 ppm) Fe levels. This approach has been used extensively by many investigators to decipher various aspects of intestinal iron absorption, as many of the key genes involved in this process are regulated by cellular or body iron status (12, 29, 37).

Several genes were induced at all five ages studied (Table 2). Some of these genes have been shown previously to be iron responsive; however, many are novel. DMT1 and Dcytb showed strong induction at all ages studied, which was confirmed by real-time PCR (Table 6). Our data thus strongly suggest that at least some functional aspects of intestinal iron transport and its upregulation in the iron-deficient state are conserved among sucklings, weanlings, and older rats. Furthermore, our findings confirm previous studies showing increased DMT1 and Dcytb mRNA expression levels in adult, iron-deficient rodents (5, 16, 27, 46).

DMT1 is known to have two 3′ splice variants, one containing an IRE (+IRE) and the other not containing this element (−IRE) (15). Our data showed stronger induction of DMT1 mRNA expression at all ages studied, as detected with an Affymetrix probe set that recognizes both 3′ splice variants (+/− IRE) as opposed to a probe set that only recognizes the +IRE transcript (Table 1). Both of these probe sets recognize both 5′ splice variants that have been recently described (21). These data suggest that iron deficiency in rats leads to transcriptional induction of DMT1, which induces expression of both 3′ splice variants. This induction would be independent of the possible regulation of DMT1 mediated through the iron regulatory protein (IRP) pathway, which may involve +IRE transcript stabilization via the IRE (15). Additionally, transcriptional induction during iron deficiency may be related to the 5′ slice variants (21).

Although expression of DMT1 and Dcytb was strongly induced, other genes encoding basolaterally expressed proteins involved in iron transport (e.g., IREG1 and hephaestin) did not show induction by the microarray studies. Previous studies have yielded conflicting results in iron-deficient rodents (6, 28, 33, 40). Therefore, we sought to confirm this observation by real-time PCR (Table 6). IREG1 and hephaestin mRNA expression by real-time PCR was increased at some of the ages studied, although induction was inconsistent and significantly lower than the induction of the genes encoding the brush-border iron transport proteins. Overall, these data lend support to prior observations suggesting that the mucosal transfer step is more highly regulated in intestinal iron absorption and that efflux across the basolateral membrane of enterocytes may be less sensitive to iron deficiency (29, 34).

Other known iron-responsive genes, namely HO1 and TFR1, were also induced in iron-deficient rats at all ages studied (for TFR1) or at all ages except in sucklings (for HO1). TFR1 mRNA levels are known to be increased in iron deficiency in rat duodenum (25). Increased TfR1 mRNA expression in the iron-deficient state is thought to be mediated by the IRP/IRE system (17). HO1 encodes an intracellular protein that is involved in releasing iron from the heme molecule, once heme has been endocytosed by enterocytes (30). The heme absorption pathway represents a distinct mechanism for intestinal iron absorption. It is thus noteworthy that HO1 is induced in iron-deficient rats at several ages, suggesting that this pathway is also of physiological importance in rodents and that heme iron absorption may also be induced by iron deficiency.

Another gene of potential interest that was induced at all five ages was tripartite motif protein 27 (TRIM27; also called Ret finger protein). TRIM27 is of particular interest, because it was one of only a few genes that were shown to be induced across all five ages when data were analyzed with the most stringent parameters. Furthermore, a second probe set that recognizes this gene showed induction at four of the five ages (see Supplemental Tables 2 and 4). TRIM27 has not been previously shown to be expressed in the duodenal mucosa; therefore, its physiological role in the intestine is unknown.

We also noted strong induction of basolateral membrane copper ATPase (ATP7a) gene expression in iron-deprived rats of all ages examined, which is a novel and potentially important finding. ATP7a induction was seen at all ages, and results were similar to those seen for other iron-responsive genes such as DMT1 and Dcytb, which showed not only strong induction but also very high expression levels (Tables 1 and 2). Furthermore, we detected an approximately four to fivefold increase in liver copper levels in 7- to 12-wk-old iron-deficient rats (data not shown), and previous observations have also shown increased body copper levels in iron-deficient rats (32, 36) and humans (9). The current study thus provides intriguing evidence that may explain copper loading in the iron-deficient state.

We speculate that under iron-deficient conditions, more copper is transported into enterocytes by DMT1, leading to induction of ATP7a, which responds in turn to increased intracellular copper levels. DMT1 has been shown in vitro to transport copper in colonic carcinoma (Caco-2) cells (2), and it is also known to transport a number of heavy metal ions in various model transport systems (16). Additional evidence suggesting that DMT1 is responsible for copper absorption during iron deficiency comes from the fact that a known, high-affinity brush-border copper transporter [CTR1 (35)] did not show induction by the gene chip analyses at any age studied. Furthermore, the strong induction of metallothionein, which can bind copper intracellularly and has been shown to be important for intestinal copper homeostasis (22), also supports the concept that enterocytes absorb more copper during iron deficiency.

Another gene that was induced across all five ages by iron-deprivation is the lipocalin interacting membrane receptor (LIMR). This observation is potentially significant, as a recent study suggested that lipocalin [24p3/Ngal (neutrophil gelatinase-associated lipocalin)] may be able to mediate iron absorption by some cell types (45). Another recent report has suggested that LIMR may be a prototype of a new family of endocytic receptors (43). Additionally, Goetz et al. (14) proposed that NGAL (e.g., lipocalin) participates in the antibacterial iron-depletion strategy of the innate immune system. Thus it is possible that lipocalin plays a role in iron homeostasis in the gut.

Also of potential physiological relevance was the induction of the sodium-dependent vitamin C transporter (SVCT) in the four older age groups. Induction of this gene was modest (between 1.6- and 2.7-fold), as detected by two distinct probe sets on the chips. Although both probe sets showed homology to both SVCT1 and SVCT2, we surmise that we were likely detecting SVCT1, because only SVCT1 (and not SVCT2) has been shown to be expressed in the intestinal epithelium (38). Furthermore, a recent study showed increased intracellular ascorbic acid levels in intestinal epithelial cells of iron-deficient rats (3). This is of significance because the Dcytb protein is thought to bind ascorbic acid intracellularly, which likely serves as an electron donor to facilitate iron reduction for transport across the apical membranes of enterocytes via DMT1. Our data are thus consistent with the hypothesis that increased absorption of vitamin C may be important for enhanced iron absorption during iron deficiency.

In addition to induction of a large number of genes, iron deficiency led to decreased expression of some genes, although the number of genes that were downregulated was smaller than the number of genes that were upregulated. Two such genes decreased by iron deficiency were monoamine oxidases A and B, which have recently been localized to multiple cell types within the human duodenal mucosa (31). The monoamine oxidases (MAO) are known to be involved in oxidation of neurotransmitters and hormones, and interestingly, both of these proteins have been previously associated with dopamine metabolism and oxidative stress. In fact, a recent paper has described the MAO-mediated metabolism of dopamine as a potential cause of oxidative stress and has discussed the role of ferrous and ferric ions in relation to Parkinson disease (19). Our data have demonstrated a strong reduction in mRNA expression of these genes during dietary iron deprivation, and thus the role of these proteins in the duodenum during iron deficiency warrants further study.

Several genes of potential interest were downregulated at four of the five ages studied, including Na-Pi-IIb (SLC34A2; up to 19-fold at 6- and 12 wk of age). This enterocyte-specific, apically expressed protein is thought to be responsible for intestinal Pi absorption (44), and the mammalian duodenum is thought to have the greatest ability to absorb dietary Pi (41). This finding may be significant, as decreased Pi levels could have detrimental effects on many metabolic processes. Several other transporters showed downregulation with iron deprivation including GLUT 5 (SLC2A5; a facilitated glucose/fructose transporter), a zinc transporter (SLC39A8), a sulfate transporter (SLC26A2), a sodium-dependent vitamin transporter (SLC5A6), and a monocarboxylic acid transporter (SLC16A1). These data suggest that iron deficiency may have an effect on membrane transport processes in the small intestinal epithelium.

Another recent study by Marzullo et al. (26) has utilized a differential display, reverse-transcription approach to identify iron-regulated genes in the rat small intestine. The authors studied 9- to 10-wk-old rats that had been fed either a diet containing 38.4 ppm Fe (control) or an iron-deficient diet containing 14.7 ppm Fe for 32 days. Interestingly, these authors found that cytochrome-c oxidase subunit II mitochondrial and serum- and glucocorticoid-regulated kinase genes were regulated by iron status. These genes were not regulated by iron status in the current studies. However, results between the two studies are hard to compare because the following experimental parameters were different: age of the animals, Fe content of the diets, length of feeding the diets, the specific gut segment used (small intestine vs. duodenum), and preparation of the tissue (mucosal scrape vs. whole intestine).

In conclusion, we used microarray techniques to discover novel genes associated with intestinal iron absorption. Comparative gene chip analysis is a plausible approach to this end, as many genes known to be involved in iron transport are induced at the mRNA level by iron deprivation. In some cases this increase in mRNA levels is due to well-described alterations in transcript stability through the IRP/IRE regulatory system (i.e., ferritin, transferrin receptor, possibly DMT1 and IREG1), but transcriptional events must also be important. We identified novel genes that are very likely to be critical for iron absorption. These include the copper ATPase (ATP7a), lipocalin interacting membrane receptor, the sodium-dependent vitamin C transporter (SVCT1), tripartite motif protein 27, and the ABC transporter ABCG2 (breast cancer-resistance protein), all of which were strongly induced in the setting of iron deficiency. A range of other genes, whose roles in gut physiology and iron transport are unknown, also appear to be regulated by body or enterocyte iron status. Our findings thus demonstrate that iron deprivation results in a large spectrum of differentially expressed genes in the duodenal epithelium, and identification of these genetic changes is likely to increase our understanding of the complex physiology of intestinal iron homeostasis.


These studies were supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants 1 R21 DK-068349 (J. F. Collins), K24 DK-02957 (K. V. Kowdley), 2R01-DK-412174 (F. K. Ghishan), and R37-DK-33209 (F. K. Ghishan) and Health Resources and Services Administration Grant 1 C76 HR 00432-01.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract