Physiol. Genomics Fuel your research with LabChart
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Physiol. Genomics 26: 55-67, 2006. First published April 25, 2006; doi:10.1152/physiolgenomics.00297.2005
1094-8341/06 $8.00
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
26/1/55    most recent
00297.2005v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via ISI Web of Science (1)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Chicault, C.
Right arrow Articles by Mosser, J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Chicault, C.
Right arrow Articles by Mosser, J.
Received 1 December 2005; accepted in final form 17 April 2006.
Physiological Genomics 26:55-67 (2006)
1094-8341/06 $8.00 © 2006 American Physiological Society

Iron-related transcriptomic variations in CaCo-2 cells, an in vitro model of intestinal absorptive cells

Céline Chicault1, Bertrand Toutain1, Annabelle Monnier2, Marc Aubry1, Patricia Fergelot1, Andre Le Treut1, Marie-Dominique Galibert1 and Jean Mosser1,2

1 Centre National de la Recherche Scientifique UMR 6061 Génétique et Développement, Université de Rennes 1, Groupe Oncogénomique, IFR140 GFAS, Faculté de médecine, Rennes, France
2 OUEST-genopole, transcriptomic platform, IFR140, Rennes, France


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 
Regulation of iron absorption by duodenal enterocytes is essential for the maintenance of homeostasis by preventing iron deficiency or overload. Despite the identification of a number of genes implicated in iron absorption and its regulation, it is likely that further factors remain to be identified. For that purpose, we used a global transcriptomic approach, using the CaCo-2 cell line as an in vitro model of intestinal absorptive cells. Pangenomic screening for variations in gene expression correlating with intracellular iron content allowed us to identify 171 genes. One hundred nine of these genes are clustered into five types of expression profile. This is the first time that most of these genes have been associated with iron metabolism. Functional annotation of these five clusters indicates potential links between the immune response, proteolysis processes, and iron depletion. In contrast, iron overload is associated with cellular metabolism, especially that of lipids and glutathione involving redox function and electron transfer.

enterocytes; microarray analysis of gene expression; hemin overload; iron deficiency


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 
IRON IS ESSENTIAL for all living organisms because it participates in many reactions involved in key processes including oxygen transport, electron transfer, and DNA synthesis. However, iron can be toxic, if present in excess, by generating oxidative radicals. Iron homeostasis therefore needs to be tightly controlled to prevent oxidative stress that affects cell integrity.

Iron is absorbed in the duodenum by absorptive enterocytes located in the villus of the intestinal epithelium. Iron is available in two forms: heme-bound iron and nonheme iron. Heme-bound iron is more efficiently absorbed than nonheme iron and enters enterocytes by a specific pathway (39) that involves an intestinal heme transporter (HCP1) (35). Nonheme iron is first reduced by CYBRD1 (DCYTB) (26) and carried out by the DMT1 transporter (SLC11A2/Nramp2) (16). Once intracellular, iron can either be stored in ferritin or exported by the Ferroportin-1 transporter (SLC40A1/IREG1) (8). Exported ferrous iron is then oxidized by hephaestin (12) before readily binding to circulating transferrin.

Body iron homeostasis is tightly regulated, with three described regulators controlling iron absorption, since excretion is not regulated (32). A dietary regulator has been proposed, implying a local iron absorption regulated by enterocytes according to the intracellular iron level (13). A second mechanism, still uncharacterized, the erythropoietic regulator, adjusts intestinal iron absorption in response to erythropoiesis needs regardless of body iron stores. The store regulator has been described to explain iron absorption regulation according to iron stores. It links the liver, the reticulo-endothelial system, and the duodenal absorptive epithelium. Recent work, using HeLa and HEK293 cells, has demonstrated that hepcidin binds to Ferroportin-1, leading to its internalization (31). Because Ferroportin-1 is expressed in macrophages, hepatocytes, and enterocytes, hepcidin is expected to block iron export form these cells. Hepcidin gene expression is induced by iron overload in mice and is facilitated by the HFE protein (1). The inhibitory function of the HFE protein in the duodenum is unclear. It may interact with the transferrin receptor protein (TfR) (10) within the duodenum crypt cells, where it may act as a body iron sensor (40).

Misregulation of iron absorption can lead to iron deficiency or iron overload, and hemochromatosis is one of the most common diseases. This primary iron overload is characterized by an excessive iron absorption and appears to be genetically heterogeneous: at least five genes have already been implicated (33). The most common form is type-1 hemochromatosis due to mutations in the HFE gene. The C282Y mutation has been identified as the major causative mutation (9), but the clinical penetrance of C282Y homozygosity appears very low, suggesting that other factors might modulate the clinical expression of type-1 hemochromatosis (4).

To identify new genes whose expression varies according to iron level, we used a global transcriptomic approach. Microarray technology allows the characterization of genes or gene networks according to their transcriptional profiles, observed in different biological conditions. The use of dedicated microarrays for iron metabolism has been reported recently for a restricted set of potentially interesting genes (29). The aim of our study is to identify genes potentially involved in novel mechanisms of iron metabolism regulation. We used CaCo-2 cells treated either with an iron-free medium or with a hemin-supplemented medium, mimicking iron overload. The screening of 18,000 human genes, using pangenomic microarrays, led to the identification of unexpected iron-dependent genes.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 
CaCo-2 cell culture.
CaCo-2 cells were grown in DMEM containing 10% heat-inactivated fetal bovine serum (FBS), 1x nonessential amino acids, 2 mM L-glutamine, and 100 IU/ml antibiotics (penicillin and streptomycin) (all reagents were purchased from Gibco BRL, Carlsbad, CA) in a humidified atmosphere of 95% air and 5% CO2. Cells at passage 36 were seeded onto 75-cm2-area flasks and were allowed to differentiate for 18 days postconfluence before iron treatment. The culture medium was changed every 2 days.

Fully differentiated CaCo-2 cells were subjected to various treatments by the use of different culture media: 1) DMEM containing 10% FBS (DMEM-FBS); 2) DMEM containing 10% FBS and 100 µM hemin (DMEM-Hemin) (Sigma, Munich, Germany); 3) Iscove's modified Dulbecco medium (IMDM; Gibco BRL) supplemented with insulin (5 mg/l), apotransferrin (5 mg/l), sodium selenite (0.005 mg/l), triiodothyronine (0.034 mg/l), epidermal growth factor (0.020 mg/l), hydrocortisone (4 mg/l), 1x nonessential amino acids, 2 mM L-glutamine, and 100 IU/ml antibiotics (penicillin and streptomycin) (14) (SF-0), and 4) IMDM supplemented as described above, with 4.4 µM ferric ammonium citrate (FAC 28%; Merck) (SF-FAC). Cells treated with hemin for 24 h (DMEM-Hemin) were compared with cells treated with the same medium without hemin (DMEM-FBS). Cells cultivated in the iron-free medium for 48 h (SF-0) were compared with those treated with iron-free medium supplemented with FAC (SF-FAC). Three independent culture experiments were performed for each culture condition. The nonheme iron concentration in each medium was determined by a colorimetric assay using TPTZ chromogen (Olympus, Olympus Diagnostica). The results are expressed in micromolar. The concentrations were as follows: DMEM-FBS [nonheme iron] = 3.5 µM, DMEM-Hemin [nonheme iron] = 3.1 µM, SF-0 [nonheme iron] = 0.1 µM, and SF-FAC [nonheme iron]= 3.0 µM.

Determination of intracellular ferritin levels.
A chemiluminescent assay was used to determine ferritin levels. Briefly, pellets containing one million cells were lysed in 20 mM Tris (pH 7.4) containing 0.5% Triton and supplemented with 1x antiprotease solution (protease inhibitor cocktail; Roche, Penzberg, Germany), and the samples were processed for chemiluminescence assay using the ADVIA Centaur System (Bayer Diagnostics) according to the manufacturer's instructions. The total protein concentration, against which the ferritin level is reported, was evaluated using the Bradford method (Bio-Rad protein assay, Munich, Germany). Ferritin levels are expressed as nanograms of ferritin per milligram of total proteins.

Target preparation and microarray hybridization.
Total RNA from CaCo-2 cells was isolated using the TRI Reagent kit (Sigma, St. Louis, MO) according to the manufacturer's instructions. RNA was quantified with the Nanodrop ND-1000 spectrophotometer (NanoDrop Technologies, Rockland, DE), and RNA integrity was assessed with RNA 6000 Nano LabChip kit using the Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). Only RNA with an RNA integrity number (RIN) >9.6 was processed for labeling (2100 Expert software, Agilent Technologies). Each total RNA extract constituted a test sample, and an equimolar pool of all RNAs was used as the reference sample.

RNA samples (tests and reference) were labeled using the Agilent Low RNA Input Fluorescent Linear Amplification kit (5184-3523) according to the manufacturer's instructions. Briefly, hybridized targets consisting of amplified and fluorescent-labeled cRNA were obtained by reverse transcription, starting with 500 ng of total RNA, and then concomitant amplification and labeling involving T7-polymerase in vitro transcription. Test samples and reference samples were labeled with cyanine-5 (Cy5)- and cyanine-3 (Cy3)-labeled CTP [10 mM; Perkin-Elmer/NEN Life Science (NEL 580, 581), Boston, MA], respectively. Cy5 and Cy3 labeling was monitored with the Nanodrop ND-1000 spectrophotometer and was in all cases between 1.0 and 1.4 pmol/µl.

Hybridization was performed using the Agilent oligonucleotide microarray in situ Hybridization Plus kit (5184-3568), following the manufacturer's recommendations. Briefly, 750 ng of test sample cRNA were mixed with 750 ng of reference sample cRNA in the presence of a control target. The final solution was subjected to fragmentation (30 min at 60°C in the dark) and hybridized onto each 22K 60-mer human oligonucleotide microarray (G4110-60520, Agilent Technologies) in a rotation oven (60°C, 17 h and 4 rpm) in the dark. Slides were disassembled and washed in solutions I and II and dried using a nitrogen-filled air gun. The slides were then scanned according to Agilent's instructions.

Data acquisition and processing.
Microarrays were scanned with a dynamic autofocus microarray scanner [Agilent dual-laser DNA microarray scanner (G2566AA)], using Agilent parameters. Feature Extraction software 7.5 (G2567AA) was used to extract data and to analyze the signal. Agilent settings were applied except that local background subtraction and global background adjustment were performed. Poor-quality features were flagged and ratio values were normalized, using the Linear&Lowess method. Only features showing a signal-to-noise ratio (SNR) in at least one channel ≥2.6 were used for further analysis. Saturated features (50% of pixels > saturation threshold, 65,502) and nonuniform features were discarded. The mean signal ratio of the two fluorescent intensities (Cy5 cRNA test/Cy3 pooled cRNA) is expressed in logarithm (base 2).

The microarray data have been uploaded to the Gene Expression Omnibus (GEO) database (series no. GSE3573 and sample nos. GSM81961–GSM81972).

Selection of differentially expressed genes.
The significance analysis microarrays (SAM) method (37) was used to select differentially expressed genes by comparing triplicate data obtained with one treatment to a second set of triplicate data obtained with another treatment (2-class unpaired data). Genes with two missing values per triplicate for each treatment were discarded for these analyses; otherwise, missing values were replaced using the K-nearest neighbors calculation method with k = 10. Genes were identified that were differentially expressed with a false discovery rate (FDR) <0.08, and only genes showing a difference of >1.5-fold were used for subsequent clustering analysis.

Hierarchical clustering.
Hierarchical clustering was performed using the GeneSight 3.0 software (Biodiscovery, Marina Del Rey, CA) (average linkage clustering using Pearson's correlation as similarity metric). Only genes selected by SAM as described above were used for hierarchical clustering.

Functional annotation.
The WebGestalt toolkit (web-based gene set analysis toolkit; http://genereg.ornl.gov/webgestalt) was used for functional annotation of defined clusters. This toolkit includes information from Gene Ontology Tree Machine software (GOTM) (44). WebGestalt queries were performed with lists of official gene symbols [approved by the Human Genome Organisation (HUGO) Nomenclature Committee] corresponding to the clusters of genes studied. For identification of enriched Gene Ontology (GO) terms (i.e., GO terms with a number of associated genes significantly higher than expected), the whole Agilent array was used as a reference list. The hypergeometric statistical method test was used, and only statistically enriched terms (P < 0.05) with at least two genes were selected.

Real-time quantitative RT-PCR assay.
Total RNAs were used for quantitative RT-PCR. The High-Capacity cDNA Archive kit (Applied Biosystems, Foster City, CA) was used for reverse transcription and the SYBR Green PCR master kit (Applied Biosystems) for quantitative PCR. Gene expression was analyzed with the ABI Prism 7000 sequence detection system, and results were handled with the associated software (version 1.2; Applied Biosystems). The 18S ribosomal RNA subunit was used as the internal reference for all analyses. Differences in transcript level were determined using the cycle threshold method as described by the manufacturer. mRNA levels are expressed as the log (base 2) of the ratio of the studied condition to the DMEM-FBS condition. Pairs of treatments (SF-0 vs. SF-FAC, DMEM-FBS vs. DMEM-Hemin) were compared with the Wilcoxon-Mann-Whitney test, and P < 0.05 was considered to be statistically significant.

The following forward (F) and reverse (R) primers were designed using the Primer Express software (version 2.0-PE Applied Biosystems). Primer specificity was assessed from the monophase dissociation curves, and all were similarly efficient (data not shown) (Table 1).


View this table:
[in this window]
[in a new window]
  		Table 1. Primers used for quantitative RT-PCR

 

   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 GRANTS
 REFERENCES
 
To investigate the transcriptional response of CaCo-2 cells to iron treatments, differentiated cells were grown, for 24 or 48 h, in media containing various iron contents. Hemin (100 µM) in DMEM-10% FBS medium (DMEM-FBS and DMEM-Hemin) was used to study response to heme iron overload; CaCo-2 cell culture in an iron-free medium (SF-0) was compared with culture in the same medium supplemented with 4.4 µM of FAC (SF-FAC) to study the effects of iron deficiency.

Iron status of cells.
Intracellular ferritin was much more abundant (90-fold) when hemin was present in the basal medium (DMEM-Hemin vs. DMEM-FBS). Ferritin was ninefold less abundant in iron-free medium than in the FAC-supplemented medium (SF-0 vs. SF-FAC). Ferritin abundance was higher in SF-FAC medium than in DMEM-FBS medium despite the two having the same iron concentration. This suggests greater import when iron is provided to the cells as ferric iron citrate (Fig. 1).


Figure 1
View larger version (10K):
[in this window]
[in a new window]
  		Fig. 1. Intracellular ferritin levels in the different cell culture conditions were measured with a chemiluminescent assay and are expressed as nanograms of ferritin per milligram of total proteins. Addition of hemin (2) to the DMEM-FBS (1) medium induced a 90-fold increase in ferritin level. Cells grown in iron-free medium (3) contained 9 times less ferritin than those in the same medium with 4.4 µM ferric ammonium citrate (FAC; 4). All measurements were performed in triplicate. Nos. at top of bars correspond to mean values of ferritin concentrations derived from the triplicate experiments.

 
Data analysis and statistical methods.
We computed the Pearson correlation coefficient for expression ratios between all the arrays to check that differences between biological replicates of the same condition were smaller than those between different conditions. The Pearson correlation coefficient for replicate arrays (i.e., targets from 3 independent culture experiments) was ≥0.89 in all but one case (0.87). The exception was one of the DMEM-FBS culture condition triplicates, and considering this result less reproducible, we excluded this array from subsequent analyses.

After slide normalization by Linear&Lowess and features filtering, Cy5/Cy3 log ratios for the 12 arrays displayed an identical distribution. The box plots are centered on zero and have fairly similar spreads (data not shown), indicating that relative expression levels are similar in each array. We therefore did not normalize the arrays for scale (i.e., mean center and normalization genes and/or arrays) (42).

Microarray results.
To select genes of interest, we performed three different SAM two-class analyses: we compared pairs of different iron-related culture conditions (Table 2). Two hundred sixty-one genes showed a significant change in expression in response to hemin treatment (Table 2, SAM 1). Among the 215 overexpressed genes, the gene coding for the heme-oxygenase enzyme displayed the highest change. One hundred sixty-eight genes were differently expressed in cells cultivated in iron-free medium and FAC-supplemented medium. Most, including the transferrin receptor-encoding gene (TFRC), were overexpressed (128) in iron-free medium (Table 2, SAM 2).


View this table:
[in this window]
[in a new window]
  		Table 2. Selection of differentially expressed genes by using a 2-class unpaired method by SAM software

 
We classified culture conditions into two groups according to intracellular ferritin levels: 1) the high-ferritin situation consisting of DMEM-Hemin and SF-FAC culture conditions and 2) the low-ferritin situation including DMEM-FBS and SF-0 culture conditions (Fig. 1). The effects of intracellular iron, assessed by ferritin concentration in the cells, were studied by comparing high-ferritin and low-ferritin situations (Table 2, SAM 3). SAM analysis identified 74 genes displaying a significant difference of expression in these two situations. They included several genes for members of the metallothionein family which were overexpressed in high-ferritin situation.

In total, the three SAM two-class analyses highlighted 461 genes with a significant difference in their expression according to iron status within differentiated CaCo-2 cells. However, only 171 genes displayed differences in expression between two situations ≥1.5-fold (Table 2, column b).

Clustering analyses.
Hierarchical clustering was performed using the 171 genes selected by SAM and with an expression change >1.5-fold. The resulting hierarchical clustering tree indicated that biological replicates grouped together (Fig. 2A). Furthermore, conditions with similar iron content (i.e., SF-FAC and DMEM-FBS) clustered in a same subbranch. In contrast, the hemin iron overload condition clustered independently, indicating that it produced characteristic transcriptome modifications. It is interesting to note that SF-FAC and DMEM-FBS conditions cluster in the same branch although they are associated with high (SF-FAC) or low (DMEM-FBS) intracellular ferritin concentrations. It is likely due to a higher number of differentially expressed genes selected by SAM 1 analysis. Thus, from a biological point of view, our data suggest a higher contribution of hemin overload in the transcriptome variations of CaCo-2 cells.


Figure 2
View larger version (32K):
[in this window]
[in a new window]
  		Fig. 2. Hierarchical clustering was performed with GeneSight 3.0 software (Biodiscovery,) (average linkage clustering using Pearson's correlation as similarity metric). Rows represent the 171 genes selected by significance analysis microarrays (SAM) and with differences ≥1.5-fold. Columns represent the 11 arrays corresponding to 4 cell culture conditions (DMEM-FBS, DMEM-Hemin, SF-FAC, SF-0; see MATERIALS AND METHODS for details) (each cell culture condition was hybridized on triplicate slides except the DMEM-FBS condition, for which 1 slide was discarded). Red and green indicate up- and downregulated expression, respectively. A: vertical clustering reveals triplicate arrays (representing the same cell culture condition) clustered together. SF-FAC and DMEM-FBS conditions (same iron content) clustered together, and the hemin condition was separate from the others. B: horizontal clustering reveals expression profiles from which 5 were selected according to cell culture conditions. Bars at right indicate the location gene clusters of interest (clusters numbered from 1 to 5).

 
We focused on branches of the hierarchical clustering tree, each of which corresponds to a different profile of expression behavior toward iron cell content. Five gene clusters were defined (Fig. 2B). Cluster 1 corresponds to the 10 genes of a single branch with changes in expression correlating with intracellular ferritin levels. Most of them are metallothionein (MT) family genes (Table 3). Cluster 2 corresponds to the 60 genes overexpressed in the hemin condition. Cluster 3 consists of six genes with lower expression in the hemin condition. Cluster 4 includes 15 genes whose expression is inversely correlated with the intracellular ferritin content. Cluster 5 corresponds to 18 genes overexpressed in iron-free medium (Fig. 3A and Table 3). Considering iron metabolism, genes are found in different clusters: the transferrin receptor gene is included in cluster 4, heme oxygenase 1 and haptoglobin genes are found in cluster 2 and are overexpressed in the hemin condition, and the aminolevulinate delta synthase 1 gene is in cluster 3 with genes that are downregulated in the hemin condition.


View this table:
[in this window]
[in a new window]
  		Table 3. List of genes included in the 5 clusters*

 

Figure 3
View larger version (45K):
[in this window]
[in a new window]
  		Fig. 3. A: expanded view of selected gene clusters numbered from 1 to 5. Cluster 1, 10 genes upregulated in the high-ferritin condition; cluster 2, 60 genes upregulated in the presence of hemin; cluster 3, 6 genes downregulated in the hemin condition; cluster 4, 15 genes upregulated in the low-ferritin condition; cluster 5, 18 genes upregulated in the absence of iron. B: validation of microarrays results by quantitative RT-PCR. Each graph represents genes tested by RT-PCR for each cluster. Nos. at right of clusters in A correspond to those indicated in insets in B. PCR was performed in triplicate for each RNA sample with SYBR Green and specific primers for each gene. We performed all analyses using 18S ribosomal RNA as a reference. Results are expressed as log (base 2) of the ratio of condition tested to the DMEM-FBS condition. aMT1n corresponds to primers amplifying a sequence common to MT1K, MT1J, and MT1E. bGSTAn corresponds to primers amplifying a sequence common to GSTA1, GSTA2, and GSTA5. *Significant with P < 0.05 (Wilcoxon-Mann-Whitney test).

 
Quantitative RT-PCR.
To confirm the gene expression variations identified by the microarray experiments, selected genes were tested by quantitative RT-PCR. The genes were selected according to their potential biological pertinence to iron metabolism and size of the differences in expression. At least 13% of the genes in each cluster were tested (Fig. 3B). Cluster 1 validation included the analysis of the expression of NQO1 and the MT family genes. The NQO1 gene was overexpressed in the hemin condition, but no significant (P > 0.05) overexpression in the SF-FAC condition was detected. Although the metallothionein family 1 (MT1) and 2 (MT2) genes are addressed with 16 distinct 60-mer probes, the overall high sequence similarity (>85% identity) of the 11 MT1 genes and the MT2 gene does not allow gene-specific hybridization. We therefore used primers named MT1n for quantitative RT-PCR. These primers amplify several members of the MT1 family. MT1n genes were overexpressed in SF-FAC and DMEM-Hemin conditions (high intracellular levels of ferritin) relative to the SF-0 and DMEM-FBS conditions (low intracellular levels of ferritin). We then used primers specific for the genes for the MT2A, MT1E, MT1X (27), and MT1K isoforms. All MT genes were overexpressed in the high-ferritin situation, except that MT1X expression was not significantly higher due to a high expression level in the iron depletion condition. Moreover, MT1X and MT2A did not display the lowest expression observed for the other tested MT genes (Fig. 3B, cluster 1). The eight genes tested in cluster 2 showed an overexpression in the hemin condition (Fig. 3B, cluster 2). We used GSTAn primers, which amplify concomitantly various genes for members of the glutathione-S-transferase A family. Three of the four genes tested (PCK1, DDIT4, ALAS1) in cluster 3 were underexpressed in the hemin condition; the difference in STC2 expression was not significant despite lower expression in this condition (Fig. 3B, cluster 3). Genes in clusters 4 and 5 displayed similar gene expression differences as assessed by real-time PCR (Fig. 3B, clusters 4 and 5). To conclude, quantitative RT-PCR results confirmed the microarray data. Although some of the variations were small, the tendency was respected, and most of the results (87.5%) were statistically significant (P < 0.05).

Functional annotation.
Functional annotation was performed to identify GO terms associated with each of the five gene clusters related to iron conditions. For each cluster, annotations are displayed as a directed acyclic graph (DAG) of GO terms retrieved from GOTM. This annotation was limited to the biological process of the GO hierarchy, named GO Process terms. Enriched GO Process terms are those with significantly enriched gene number retrieved from gene clusters by comparison with a reference gene set, leading to the highlight of terms that may be more relevant for the previous gene set (44).

Of the 18 genes in cluster 5, 17 were associated with a GO Process term. The corresponding DAG displayed the 12 enriched GO Process terms annotating 14 genes that are spread in two main branches (Fig. 4C).


Figure 4
View larger version (22K):
[in this window]
[in a new window]
  		Fig. 4. Directed acyclic graph (DAG) of Gene Ontology (GO) terms using Gene Ontology Tree Machine software (GOTM) for the biological process hierarchy and for cluster 2 (A), cluster 4 (B), cluster 5 (C), and cluster 3 (D). Hypergeometric test was used as the statistical method to select enriched GO Process terms with at least 2 genes and with a significant level of P < 0.05 (enriched GO Process terms are in gray). Only genes associated with the more informative enriched GO Process terms are indicated.

 
1) The immune response pathway included eight enriched GO Process terms associated with two to five genes. These five genes are biologically linked to the immune response, and two of them are also annotated with children terms specifying the "humoral defense mechanism" and "antimicrobial humoral response."

2) The second main branch led to the "proteolysis and peptidolysis" enriched GO Process term. Two enriched GO Process terms were associated with three genes coding for proteins with protease functions (MMP7, TMPRSS4) or involved in protein ubiquitination, targeting protein-to-proteasome degradation (UBD). Interestingly, two genes appeared indirectly linked to the immune response: UBD is encoded in the major histocompatibility complex and is synergistically inducible by tumor necrosis factor-{alpha} (TNF{alpha}) and interferon-{gamma} (18); the MMP7 gene encodes a metalloproteinase implicated in intestinal mucosal defense by cleaving defensin precursors into a mature form and thereby regulating defensin protein (antimicrobial peptides) activity (41).

In addition to the two main branches, cluster 5 is also linked to more general enriched GO Process terms: "cell communication" and "growth" linked to 10 and 2 genes, respectively.

Cluster 4 contains 15 genes overexpressed in the low-ferritin situation. Nine of these fifteen genes are associated with a process in the GO hierarchy. Six genes participated to the enrichment of eleven enriched GO Process terms, and the resulting DAG (Fig. 4B) displayed two main branches:

1) Again, the proteolysis and peptidolysis pathway included eight enriched GO Process terms that were associated with three genes and a branch highlighting the "ubiquitin cycle" enriched GO Process term that was associated with two genes. The TFRC gene is indirectly associated with proteolysis and peptidolysis, describing the proteolysis process of TfR in soluble TfR (TfRs). FBXL4 encodes a member of the F-box protein family, which is one of the four subunits of the ubiquitin protein ligase complex called SCFs (SKP1-cullin-F-box) involved in phosphorylation-dependent ubiquitination. Association of the C9orf43 and RNF186 genes with proteolysis and peptidolysis results from electronic inference.

2) The second main branch displayed "ion transport" and particularly "metal ion transport" enriched GO Process terms associated with three and two genes, respectively: SCN10A encodes a protein that mediates the voltage-dependent sodium ion permeability of excitable membranes, CLIC5 encodes a intracellular chloride channel, and the protein encoded by the TFRC gene mediates the cellular uptake of iron via receptor-mediated endocytosis of its ligand transferrin.

The six genes in cluster 3 retrieved only two general enriched GO Process terms: "biosynthesis" and "cellular biosynthesis" associated with two genes, PCK1 and ALAS1 (Fig. 4D).

Cluster 1 is mostly composed of the MT family genes and consequently not functionally annotated. The MT family encodes proteins with a high content of cysteine residues involved in binding various heavy metals. The two other genes in cluster 1 encode the following: NQO1 and PLXNA2. NQO1 is an enzyme with quinone reductase activity involved in conjugation reactions of hydroquinone compounds in detoxification pathways. PLXNA2 is a member of a large family of receptors that recognize secreted semaphorin signaling molecules, involved in axon guidance and growth cone collapse in the central nervous system.

Forty-nine of the sixty genes overexpressed in the hemin condition (cluster 2) are associated with a GO Process term. The corresponding DAG displayed the 15 enriched GO Process terms annotating 38 genes which are spread in two main branches (Fig. 4A).

1) Two genes (AKR1C2 and TFF3) are linked to the "digestion" process. The AKR1C2 gene encodes a member of the aldo/keto reductase superfamily and binds bile acids. The TFF3 gene encodes a member of the trefoil family, which are secretory proteins expressed in the gastrointestinal mucosa and involved in epithelial restitution.

2) Two subbranches derived from the "metabolism" enriched GO Process term. One branch stopped at the process term "cellular lipid metabolism," which is associated with five genes. Four of these five encode reductase, hydrolase, or lipid desaturase (bile acids, glucocerebrosides, steroids, fatty acids) enzymes, and one encodes a nuclear receptor family member, activated by bile acids, steroids, and drugs, and induces human CYP3A4 and CYP7A1. The second subbranch, "cellular metabolism," includes 11 enriched GO Process terms associated with 9 genes and stopped at the GO Process term "cysteine metabolism." Seven of these eleven terms are linked to the GCLM and GCLC genes, coding for the two subunits of the first rate-limiting enzyme of glutathione synthesis, {gamma}-glutamylcysteine synthetase (GCL). The four other terms are "organic acid metabolism," "carboxylic acid metabolism," "cofactor metabolism," and "glutamine family amino acid metabolism." The molecular functions of the nine annotated genes are mostly associated with redox phenomena and electron transfer, strongly suggesting that iron excess leads to electron transfer within the cells.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
GRANTS
 REFERENCES
 
For our analysis of variations of intestinal transcripts in response to iron, we used the CaCo-2 human enterocyte-like cell line, a widely used model of intestinal epithelial cell differentiation and absorption phenomena (17). CaCo-2 cells contain all the proteins implicated in iron uptake, trans-epithelial transport, and basolateral transfer (3). Moreover, their iron absorption capacity responds to variations in the cellular iron content by adapting gene expression at the transcriptional or translational levels (15, 25).

We used an iron- and serum-free medium rather than an iron chelator for iron depletion in CaCo-2 cells to avoid any potential nonspecific transcriptomic effect of the chelator. The effects of iron depletion were compared with those of serum-free medium supplemented with FAC. To study iron overload in cells, we preferred to use the serum-containing medium previously used during the differentiation process. Measurements of the intracellular ferritin concentration correlated with the expected intracellular iron content and were also in agreement with the various culture conditions. Nevertheless, despite an identical iron concentration in culture media, the ferritin level in SF-FAC was seven times higher than that in DMEM-FBS, which in turn was slightly higher than that in SF-0. The ferritin level was higher in the hemin condition than in FAC (Fig. 1). This suggests that hemin enters cells more readily than nonheme iron, leading to a significant iron overload as assessed from the intracellular ferritin level.

Analyses of microarrays revealed that <1% (171 genes) of the 18,000 transcripts tested were affected by iron status (SAM analyses with an FDR <0.08 and a fold change >1.5). We distinguished five gene clusters with distinct expression profiles according to iron status. Quantitative RT-PCR validated the microarray findings for these transcripts, and consequently these clusters appear to be experimentally robust. Indeed, correlation between microarray and RT-PCR data was excellent (0.96), confirming the accuracy of the microarray technology used. The five identified clusters can be considered to be biologically significant because they all (except for cluster 5) included genes whose expression is in complete agreement with the literature. Indeed, the increase of TFRC (29) expression in the condition of iron depletion and the increase of HMOX1 (38) and the decrease of ALAS1 (30) in the hemin overload condition have already been described elsewhere. However, the upregulation in the iron depletion condition previously reported for the hephaestin and Ferroportin-1 genes (5) was not found to be significant in our experiments, although small variations were observed. In our experimental conditions, DMT1 expression was unchanging, but the corresponding probe hybridized all splice variant forms of DMT1 mRNA. However, RT-PCR analysis using DMT1 splice variant-specific primers confirmed expression data previously reported for DMT1 (data not shown) (19). Unfortunately, the CYBRD1 gene could not be analyzed, since no corresponding probe was present in the 22K microarrays.

By inducing hypoxia, iron depletion is believed to activate the transcription factor hypoxia-inducible factor-1 (HIF-1) in cells. Only 5 of the 60 putative HIF-1 target genes described by J. W. Lee et al. (24) are found in our four clusters (cluster 2: TFF3, HMOX1; cluster 3: DDIT4; cluster 4: TFRC; cluster 5: ADM). This suggests that, in our experiments, HIF-1 did not act as a key regulator.

Annotation of cluster 5 (i.e., genes overexpressed in iron-free medium) highlighted two pathways: the immune response and proteolysis phenomena processes. This suggests that the immune response is related to intracellular iron depletion. This result is interesting because the literature describes a link between iron metabolism and the immune system: iron homeostasis is a key factor for the optimal functioning of the immune system (34). Furthermore, several molecules involved in iron metabolism are also linked to the immune system. For example, the HFE protein is a member of the MHC class I-like proteins, and hepcidin factor is an antimicrobial peptide induced by iron overload and LPS treatment (LPS from gram-negative bacteria). A recent study has provided evidence that the bacterial siderophore deferoxamine, implicated in iron chelation, specifically stimulates CCL20 chemokine expression and secretion in human intestinal epithelial cells (IECs), resulting in chemotaxis of peripheral blood mononuclear cells. Thus direct chelation of host iron by infected bacteria may contribute to the immune response in the intestinal mucosa (23). Accordingly, the five identified genes, linked to the immune response process, are overexpressed in iron-depleted CaCo-2 cells, mimicking the iron chelation condition occurring during bacterial infection. The explanation of the association of proteolysis with iron depletion (cluster 4 and cluster 5) is not obvious. Nevertheless, a link between proteolysis and the immune response is highlighted by two genes (MMP7 and UBD), reinforcing the biological relation between iron metabolism and immunity.

For the cluster of 60 genes overexpressed in the hemin-overloaded cells, functional annotation provided only 15 enriched GO Process terms. The corresponding DAG of the enriched GO Process terms shows that many of the genes are significantly associated with redox and electron transfer phenomena. The GCLM and GCLC genes encode the two subunits of human GCL (first rate-limiting enzyme in glutathione biosynthesis) and are upregulated in the hemin condition to compensate for the decrease of reduced glutathione observed with iron overload (6). Glutathione (GSH) is a tripeptide that participates in many important cellular functions (metabolism of endogenous and exogenous compounds, cell proliferation, regulation of gene expression). However, GSH is a key element in the detoxification of reactive oxygen species (ROS) such as H2O2 and lipid peroxides (20). The glutathione-S-transferase enzyme, also found in cluster 2, catalyzes glutathione conjugation. Interestingly, the "lipid metabolism" process is associated with hemin overload. Indeed, iron induces lipid peroxidation, resulting in their conjugation with GSH. Finally, overexpression of the GCLC and GCLM genes might be essential to compensate for the loss of GSH, dedicated to the detoxification of iron-overloaded cells (20).

Cluster 1 contains several members of the metallothionein superfamily (MT) of genes, which are overexpressed as the ferritin concentration increases, although we were unable to identify differences between these family members because of their high degree of sequence identity. Several roles have been proposed for the MT members, including detoxification of nonessential heavy metals and excess essential metals, homeostatic regulation of essential metals, modulation of intracellular metal transport, and scavenging free radicals and ROS (36). MT members do not bind iron (22), although MT gene expression is induced by iron in chick liver (11) and by heme in mouse hepatoma cells (2). MT promoters include a series of metal regulatory elements (MREs) involved in gene expression, through their interaction with the metal regulatory transcription factor (MTF)-1 zinc-finger transcription factor, the activity of which is dependent on the local zinc concentration. Heavy metal or hydrogen peroxide binds to MT proteins, inducing the release of zinc. Free zinc activates MTF-1 and subsequently induces MT gene expression (43). Presumably, iron mediates an increase of MT gene expression by an indirect effect, possibly through the production of iron-related ROS (28). This regulation process, which involves zinc activation of the MTF-1 transcription factor, suggests the presence of MRE enriched promoters for genes in clusters 1 and 2. However, no such enrichment was observed other than for the MT family genes (21). The absence of direct correlation between iron-induced gene expression and the presence of MRE motifs suggests a much more complex regulation process. Effectively, Daniels and Andrews (7) proposed that MT gene regulation in response to metals involves a dynamic transcription factor complex.

This pangenomic screening allowed us to identify 109 genes differentially expressed according to iron intracellular content. Functional annotation linked discriminative iron deficiency genes to immune response. This is of particular interest, since genes known to be implicated in iron metabolism are also linked to immunity (hepcidin, HFE, and so forth). Also, functional annotation highlighted that, in the hemin overloaded medium condition, cells protected themselves from iron toxic effects.

On the basis of current biological knowledge, functional annotation did not emphasized iron-related process or function for the new discriminating genes we found associated with iron on the basis of their expression. This lack of information could be a consequence of using a pruned version of the GO DAG (GO slim), as is the case with GOTM. Indeed, granular annotations are mapped to general categories to help identify GO categories with enriched gene numbers. As a result, a fine-grained annotated GO term could be missed but replaced by an enriched course-grained GO term. To improve functional annotation accuracy, two strategies are presently initiated. We developed an annotation tool that extracts cluster-associated terms from the GO Annotation database (GOA; evidence, http://www.ebi.ac.uk/GOA) and from the literature (term-gene co-occurrences in PubMed, http://www.pubgene.org). The goal of this new approach is to enrich the annotation with information retrieved from the literature. Previous results clearly indicate a more detailed and deeper functional annotation on our gene clusters (Aubry M et al., unpublished observations). The second strategy consists of focusing on function and process of unenriched GO terms, for those relevant in our biological context. This GO term analysis also needs an inventory of all the genes (human or not) associated with iron-related terms in GOA. In addition, a careful and systematic inspection of cis-regulatory elements of the co-regulated promoter genes may contribute to the inference of a common cis-element with putative transcriptomic regulation mechanisms involving transcription factors, some of which are expected to participate in iron absorption regulation.

New genes included in cluster 2 (genes overexpressed in the hemin condition) are likely also to be involved in the iron-induced stress response. These genes have yet to be characterized by biological experiments at the mRNA and protein levels.

We expected from these analyses to give new insights into the understanding of iron homeostasis and the iron regulation network and to identify new putative target genes implicated in iron metabolism regulation.


   GRANTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
REFERENCES
 
This work was supported by grants from the Centre National de la Recherche Scientifique and the Conseil Régional de Bretagne.


   ACKNOWLEDGMENTS
 
We thank Hélène Bedrine-Ferran for carefully reading and revising this manuscript and Elisabeth Denis, Stéphanie Mottier, and Isabelle Gicquel for technical assistance and helpful discussion.


   FOOTNOTES
 
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).

Address for reprint requests and other correspondence: J. Mosser, CNRS UMR 6061 Génétique et Développement, Université de Rennes 1, 2 av du Pr Léon Bernard, CS 34317, 35043 Rennes cedex, France (e-mail: jean.mosser{at}univ-rennes1.fr).


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 

  1. Ahmad KA, Ahmann JR, Migas MC, Waheed A, Britton RS, Bacon BR, Sly WS, and Fleming RE. Decreased liver hepcidin expression in the Hfe knockout mouse. Blood Cells Mol Dis 29: 361–366, 2002.[CrossRef][ISI][Medline]
  2. Alam J and Smith A. Heme-hemopexin-mediated induction of metallothionein gene expression. J Biol Chem 267: 16379–16384, 1992.[Abstract/Free Full Text]
  3. Alvarez-Hernandez X, Nichols GM, and Glass J. Caco-2 cell line: a system for studying intestinal iron transport across epithelial cell monolayers. Biochim Biophys Acta 1070: 205–208, 1991.[Medline]
  4. Beutler E. The HFE Cys282Tyr mutation as a necessary but not sufficient cause of clinical hereditary hemochromatosis. Blood 101: 3347–3350, 2003.[Free Full Text]
  5. Chen H, Su T, Attieh ZK, Fox TC, McKie AT, Anderson GJ, and Vulpe CD. Systemic regulation of Hephaestin and Ireg1 revealed in studies of genetic and nutritional iron deficiency. Blood 102: 1893–1899, 2003.[Abstract/Free Full Text]
  6. Cornejo P, Varela P, Videla LA, and Fernandez V. Chronic iron overload enhances inducible nitric oxide synthase expression in rat liver. Nitric Oxide 13: 54–61, 2005.[CrossRef][ISI][Medline]
  7. Daniels PJ and Andrews GK. Dynamics of the metal-dependent transcription factor complex in vivo at the mouse metallothionein-I promoter. Nucleic Acids Res 31: 6710–6721, 2003.[Abstract/Free Full Text]
  8. Donovan A, Brownlie A, Zhou Y, Shepard J, Pratt SJ, Moynihan J, Paw BH, Drejer A, Barut B, Zapata A, Law TC, Brugnara C, Lux SE, Pinkus GS, Pinkus JL, Kingsley PD, Palis J, Fleming MD, Andrews NC, and Zon LI. Positional cloning of zebrafish ferroportin1 identifies a conserved vertebrate iron exporter. Nature 403: 776–781, 2000.[CrossRef][Medline]
  9. Feder JN, Gnirke A, Thomas W, Tsuchihashi Z, Ruddy DA, Basava A, Dormishian F, Domingo R Jr, Ellis MC, Fullan A, Hinton LM, Jones NL, Kimmel BE, Kronmal GS, Lauer P, Lee VK, Loeb DB, Mapa FA, McClelland E, Meyer NC, Mintier GA, Moeller N, Moore T, Morikang E, Prass CE, Quintana L, Starnes SM, Schatzman RC, Brunke KJ, Drayna DT, Risch NJ, Bacon BR, and Wolff RK. A novel MHC class I-like gene is mutated in patients with hereditary haemochromatosis. Nat Genet 13: 399–408, 1996.[CrossRef][ISI][Medline]
  10. Feder JN, Penny DM, Irrinki A, Lee VK, Lebron JA, Watson N, Tsuchihashi Z, Sigal E, Bjorkman PJ, and Schatzman RC. The hemochromatosis gene product complexes with the transferrin receptor and lowers its affinity for ligand binding. Proc Natl Acad Sci USA 95: 1472–1477, 1998.[Abstract/Free Full Text]
  11. Fleet JC, Andrews GK, and McCormick CC. Iron-induced metallothionein in chick liver: a rapid, route-dependent effect independent of zinc status. J Nutr 120: 1214–1222, 1990.[Abstract/Free Full Text]
  12. Frazer DM, Vulpe CD, McKie AT, Wilkins SJ, Trinder D, Cleghorn GJ, and Anderson GJ. Cloning and gastrointestinal expression of rat hephaestin: relationship to other iron transport proteins. Am J Physiol Gastrointest Liver Physiol 281: G931–G939, 2001.[Abstract/Free Full Text]
  13. Frazer DM, Wilkins SJ, Becker EM, Murphy TL, Vulpe CD, McKie AT, and Anderson GJ. A rapid decrease in the expression of DMT1 and Dcytb but not Ireg1 or hephaestin explains the mucosal block phenomenon of iron absorption. Gut 52: 340–346, 2003.[Abstract/Free Full Text]
  14. Gangloff MB, Lai C, Van Campen DR, Miller DD, Norvell WA, and Glahn RP. Ferrous iron uptake but not transfer is down-regulated in Caco-2 cells grown in high iron serum-free medium. J Nutr 126: 3118–3127, 1996.[Abstract/Free Full Text]
  15. Gunshin H, Allerson CR, Polycarpou-Schwarz M, Rofts A, Rogers JT, Kishi F, Hentze MW, Rouault TA, Andrews NC, and Hediger MA. Iron-dependent regulation of the divalent metal ion transporter. FEBS Lett 509: 309–316, 2001.[CrossRef][ISI][Medline]
  16. Gunshin H, Mackenzie B, Berger UV, Gunshin Y, Romero MF, Boron WF, Nussberger S, Gollan JL, and Hediger MA. Cloning and characterization of a mammalian proton-coupled metal-ion transporter. Nature 388: 482–488, 1997.[CrossRef][Medline]
  17. Hidalgo IJ, Raub TJ, and Borchardt RT. Characterization of the human colon carcinoma cell line (Caco-2) as a model system for intestinal epithelial permeability. Gastroenterology 96: 736–749, 1989.[ISI][Medline]
  18. Hipp MS, Kalveram B, Raasi S, Groettrup M, and Schmidtke G. FAT10, a ubiquitin-independent signal for proteasomal degradation. Mol Cell Biol 25: 3483–3491, 2005.[Abstract/Free Full Text]
  19. Hubert N and Hentze MW. Previously uncharacterized isoforms of divalent metal transporter (DMT)-1: implications for regulation and cellular function. Proc Natl Acad Sci USA 99: 12345–12350, 2002.[Abstract/Free Full Text]
  20. Iles KE and Liu RM. Mechanisms of glutamate cysteine ligase (GCL) induction by 4-hydroxynonenal. Free Radic Biol Med 38: 547–556, 2005.[CrossRef][ISI][Medline]
  21. Kindermann B, Doring F, Budczies J, and Daniel H. Zinc-sensitive genes as potential new target genes of the metal transcription factor-1 (MTF-1). Biochem Cell Biol 83: 221–229, 2005.[CrossRef][ISI][Medline]
  22. Kojima N, Young CR, and Bates GW. Failure of metallothionein to bind iron or act as an iron mobilizing agent. Biochim Biophys Acta 716: 273–275, 1982.[Medline]
  23. Lee HJ, Choi SC, Choi EY, Lee MH, Seo GS, Kim EC, Yang BJ, Lee MS, Shin YI, Park KI, and Jun CD. Iron chelator induces MIP-alpha/CCL20 in human intestinal epithelial cells: implication for triggering mucosal adaptive immunity. Exp Mol Med 37: 297–310, 2005.[ISI][Medline]
  24. Lee JW, Bae SH, Jeong JW, Kim SH, and Kim KW. Hypoxia-inducible factor (HIF-1)alpha: its protein stability and biological functions. Exp Mol Med 36: 1–12, 2004.[ISI][Medline]
  25. Martini LA, Tchack L, and Wood RJ. Iron treatment downregulates DMT1 and IREG1 mRNA expression in Caco-2 cells. J Nutr 132: 693–696, 2002.[Abstract/Free Full Text]
  26. McKie AT, Barrow D, Latunde-Dada GO, Rolfs A, Sager G, Mudaly E, Mudaly M, Richardson C, Barlow D, Bomford A, Peters TJ, Raja KB, Shirali S, Hediger MA, Farzaneh F, and Simpson RJ. An iron-regulated ferric reductase associated with the absorption of dietary iron. Science 291: 1755–1759, 2001.[Abstract/Free Full Text]
  27. Mididoddi S, McGuirt JP, Sens MA, Todd JH, and Sens DA. Isoform-specific expression of metallothionein mRNA in the developing and adult human kidney. Toxicol Lett 85: 17–27, 1996.[CrossRef][ISI][Medline]
  28. Min KS, Tanaka N, Horie T, Kawano H, Tetsuchikawahara N, and Onosaka S. Metallothionein-enriched hepatocytes are resistant to ferric nitriloacetate toxicity during conditions of glutathione depletion. Toxicol Lett 158: 108–115, 2005.[CrossRef][ISI][Medline]
  29. Muckenthaler M, Richter A, Gunkel N, Riedel D, Polycarpou-Schwarz M, Hentze S, Falkenhahn M, Stremmel W, Ansorge W, and Hentze MW. Relationships and distinctions in iron-regulatory networks responding to interrelated signals. Blood 101: 3690–3698, 2003.[Abstract/Free Full Text]
  30. Munakata H, Sun JY, Yoshida K, Nakatani T, Honda E, Hayakawa S, Furuyama K, and Hayashi N. Role of the heme regulatory motif in the heme-mediated inhibition of mitochondrial import of 5-aminolevulinate synthase. J Biochem (Tokyo) 136: 233–238, 2004.[Abstract/Free Full Text]
  31. Nemeth E, Tuttle MS, Powelson J, Vaughn MB, Donovan A, Ward DM, Ganz T, and Kaplan J. Hepcidin regulates cellular iron efflux by binding to ferroportin and inducing its internalization. Science 306: 2090–2093, 2004.[Abstract/Free Full Text]
  32. Parkkila S, Niemela O, Britton RS, Fleming RE, Waheed A, Bacon BR, and Sly WS. Molecular aspects of iron absorption and HFE expression. Gastroenterology 121: 1489–1496, 2001.[ISI][Medline]
  33. Roetto A and Camaschella C. New insights into iron homeostasis through the study of non-HFE hereditary haemochromatosis. Best Pract Res Clin Haematol 18: 235–250, 2005.[Medline]
  34. Schaible UE and Kaufmann SH. Iron and microbial infection. Nat Rev Microbiol 2: 946–953, 2004.[CrossRef][ISI][Medline]
  35. Shayeghi M, Latunde-Dada GO, Oakhill JS, Laftah AH, Takeuchi K, Halliday N, Khan Y, Warley A, McCann FE, Hider RC, Frazer DM, Anderson GJ, Vulpe CD, Simpson RJ, and McKie AT. Identification of an intestinal heme transporter. Cell 122: 789–801, 2005.[CrossRef][ISI][Medline]
  36. Takahashi Y, Ogra Y, and Suzuki KT. Nuclear trafficking of metallothionein requires oxidation of a cytosolic partner. J Cell Physiol 202: 563–569, 2005.[CrossRef][ISI][Medline]
  37. Tusher VG, Tibshirani R, and Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 98: 5116–5121, 2001.[Abstract/Free Full Text]
  38. Uc A and Britigan BE. Does heme oxygenase-1 have a role in Caco-2 cell cycle progression? Exp Biol Med (Maywood) 228: 590–595, 2003.[Abstract/Free Full Text]
  39. Uc A, Stokes JB, and Britigan BE. Heme transport exhibits polarity in Caco-2 cells: evidence for an active and membrane protein-mediated process. Am J Physiol Gastrointest Liver Physiol 287: G1150–G1157, 2004.[Abstract/Free Full Text]
  40. Waheed A, Parkkila S, Saarnio J, Fleming RE, Zhou XY, Tomatsu S, Britton RS, Bacon BR, and Sly WS. Association of HFE protein with transferrin receptor in crypt enterocytes of human duodenum. Proc Natl Acad Sci USA 96: 1579–1584, 1999.[Abstract/Free Full Text]
  41. Wilson CL, Ouellette AJ, Satchell DP, Ayabe T, Lopez-Boado YS, Stratman JL, Hultgren SJ, Matrisian LM, and Parks WC. Regulation of intestinal alpha-defensin activation by the metalloproteinase matrilysin in innate host defense. Science 286: 113–117, 1999.[Abstract/Free Full Text]
  42. Yang YH, Dudoit S, Luu P, Lin DM, Peng V, Ngai J, and Speed TP. Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation. Nucleic Acids Res 30: e15, 2002.[Abstract/Free Full Text]
  43. Zhang B, Georgiev O, Hagmann M, Gunes C, Cramer M, Faller P, Vasak M, and Schaffner W. Activity of metal-responsive transcription factor 1 by toxic heavy metals and H2O2 in vitro is modulated by metallothionein. Mol Cell Biol 23: 8471–8485, 2003.[Abstract/Free Full Text]
  44. Zhang B, Schmoyer D, Kirov S, and Snoddy J. GOTree Machine (GOTM): a web-based platform for interpreting sets of interesting genes using Gene Ontology hierarchies. BMC Bioinformatics 5: 16, 2004.[CrossRef][Medline]




This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
26/1/55    most recent
00297.2005v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via ISI Web of Science (1)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Chicault, C.
Right arrow Articles by Mosser, J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Chicault, C.
Right arrow Articles by Mosser, J.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Visit Other APS Journals Online
Copyright © 2006 by the American Physiological Society.