Iron is a critical element required for normal homeostasis. This fact is best exemplified by the role of iron in diverse physiological processes such as DNA synthesis, respiration and oxygen transport. Body iron levels must be precisely maintained within certain limits as iron-deficiency and iron-overload result in perturbations of normal metabolism. To achieve appropriate levels of cellular iron and to avoid iron-loading, several mechanisms have evolved; these include the control of iron transport, storage and recycling by regulatory proteins. A key step in controlling overall body iron levels is absorption of dietary iron in the duodenum. Iron absorption is a highly regulated process, as mammals have no controlled means to excrete excess iron. Recently, several key proteins involved in intestinal iron transport have been identified. These include the brush-border membrane associated proteins duodenal cytochrome b (Cybrd1; a ferrireductase), and divalent metal transporter 1 (Dmt1 or Slc11a2), which together allow dietary iron to be reduced and transported into enterocytes. The combined activity of hephaestin (Hp; a ferroxidase) and ferroportin (Fpn; the basolateral iron exporter) subsequently allows oxidized iron to exit the cell and bind to transferrin for distribution throughout the body. Intestinal iron transport is in part regulated by the liver-derived, antimicrobial peptide, hepcidin. Under conditions of iron excess, elevated hepcidin acts to remove Fpn from the basolateral surface of enterocytes effectively creating a mucosal block to iron [1].

Iron-deprivation increases expression of genes involved in intestinal iron-transport in laboratory rodents [2, 3] and in humans [4]. Some of these induced genes are regulated post-transcriptionally via the iron regulatory protein (IRP)/iron-response element (IRE) system. These genes include ferritin, transferrin receptor (Tfr) and possibly Dmt1 [5] and Fpn [6]. Despite the large body of work describing these regulatory events, many transcripts encoding proteins involved in iron homeostasis do not have IREs and it is thus very likely that transcriptional regulation may also be important [7]. Despite the wealth of knowledge regarding transcriptional regulation of gene expression in response to iron-deprivation in lower species such as yeast [8], no such regulatory networks have been identified to date in mammals.

Our previous studies were the first that utilized a genome wide approach to identify genes regulated during iron-deficiency in the mammalian duodenum [2, 3]. We identified a large number of differentially expressed genes (DEGs), many of which had never been described to be regulated by iron or body iron status. In the current manuscript, we utilized computational and bioinformatics approaches to identify regulatory mechanisms that may mediate the genetic response to iron deprivation. We thus first used clustering algorithms to group genes from our previous GeneChip^® studies to identify groups of co-regulated genes. Of particular interest was a group of 228 probe sets (representing ~168 genes) that were up-regulated across several different stages of postnatal development. Interestingly, of these genes, only the Dmt1 and Tfr genes contained 3' IREs, suggesting that analysis of transcriptional regulatory regions could be a fruitful approach to identify common regulatory mechanisms. Our subsequent analyses found that SP1-like binding sites were statistically enriched in promoters from these genes in rat and were conserved across 3 mammalian species. Additional studies suggested that SP and FOX transcription factors may work synergistically to regulate some genes during iron-deprivation; in fact, we found conserved FOX binding sites in many genes with conserved SP1 binding sites. We thus hypothesize that SP and FOX family transcription factors are involved in regulating expression of a subset of intestinal genes induced during iron-deficiency. We also found strong induction (~30-fold) of specificity factor 6 (SP6) by qRT-PCR, suggesting that it may be the SP family member that mediates the genetic response to iron-deprivation. Overall, these data suggest the existence of a transcriptional regulatory network(s) which responds to iron intake levels or other physiological signals associated with iron-deficiency in the mammalian intestine.

A well-characterized finding during iron-deficiency in mammals is that genes involved in iron-transport or iron-homeostasis are up-regulated. In some cases, increased gene expression may be mediated post-transcriptionally (via the IRE/IRP system), but transcriptional mechanisms are also likely important [7]. This latter supposition is supported by the identification of iron and metal-responsive transcription factors from a number of lower species that mediate the cellular response to increased or decreased metal ion levels [8, 11].

Analysis of a novel GeneChip^® dataset resulted in the identification of 1484 DEGs from the duodenum of iron-deficient rats at several postnatal ages, providing us with a unique opportunity to identify groups of induced genes that were potentially coordinately regulated and identify potential mechanisms that could be responsible for this common regulation. Only 2 genes in our up-regulated cluster contained IREs that could potentially mediate increases in steady state mRNA levels. It thus emerged that other levels of regulation (i.e. most likely transcriptional) were likely responsible for the alterations in the expression of this group of genes. We therefore focused on promoter analysis to identify potential candidate regulatory mechanisms.

Gene promoters were identified by a combination of approaches, including manual analysis of probe set IDs from the 228 probe sets in the large up-regulated cluster and BLAST searches that allowed us to analyse alternative GenBank accession numbers for the corresponding cDNAs, which we could link to a promoter sequence in the Cold Spring Harbor Database. Background sequences were assembled as described in the Methods section. Transcription factor binding site (TFBS) analysis utilized the Clover program [9] and the TRANSFAC professional database. Results revealed several enriched TF binding sites, but SP1 was the only significant TFBS identified across 3 mammalian species. Several putative SP1 matrices were identified and all had very high raw scores, indicating high similarity to consensus SP1 binding sites. We then employed additional approaches to consider whether SP1 or a related TF may be an important regulator of duodenal genes involved in the response to iron-deprivation.

Testing for evolutionary conservation of TFBSs often improves the ability to predict true binding sites from false positives, given that promoters in general show low nucleotide sequence homology across species except in regions that interact with gene regulatory proteins. We therefore determined the locations and specific sequences of the putative SP1 sites from the genes used for the enrichment analysis and compared them across rat, mouse and human. Strikingly, the nucleotide sequences of the SP1 sites were highly conserved and often, multiple putative binding sites existed with the distances between them also conserved across species. This conservation of the SP1 sites was identified in ~40 genes from the large cluster; the actual number of genes could be significantly higher though, as the curators of the Cold Spring Harbor Gene Promoter Database acknowledge that perhaps only 60% of their promoter sequences may be accurate.

SP1 is one of the first transcription factors to be purified, cloned and characterized from mammalian cells [12], and is one of the most potent transcriptional activators identified to date [13, 14]. SP1 plays important roles in a wide variety of physiological processes such as cell cycle regulation and apoptosis, hormonal activation, and angiogenesis, among others. SP1 has been shown to interact with GC- rich sites in various gene promoters via three Cys₂His₂ zinc-finger motifs [15]. It is now clear that SP1 is a part of a large family of similar zinc-finger transcription factors (also called Krüppel-like factors) that includes 21 identified members in humans, 17 in mouse and 11 in rat [10]. Like other trans-acting factors, SP1 has three functional domains: 1) a DNA-binding domain, 2) a nuclear localization signal and 3) a transactivation domain that mediates protein-protein interactions that control transcriptional initiation. Among SP1 family members, the DNA binding domain is highly conserved [16, 17], while the transactivation domain varies widely between family members [10]. This high conservation in the DNA binding domains suggests that many of the SP/KLF family members bind highly similar DNA sequences. Our data thus suggests that SP1 or a related TF could be involved in regulating some of the genes in the 228 up-regulated cluster during iron-deficiency.

Comparison of GeneChip^® data from the experimental animal groups considered in this manuscript along with another series of GeneChip^® experiments performed in the Belgrade rat model of genetic iron-deficiency (i.e. Dmt1-deficient; not shown), revealed that SP6/KLF14 was induced during both dietary and genetic iron-deficiency. In fact, SP6 was one of < 30 genes that was induced in both models, utilizing stringent statistical analysis approaches. This intriguing observation was confirmed by qRT-PCR analyses that revealed a ~30-fold induction in adult, iron-deficient rats, including experiments performed with 2 different sets of SP6-specific primers. SP6 is a member of the SP/KLF family that was originally identified by a bioinformatics scan of the human genome using the conserved zinc finger DNA binding domains [18]. A subsequent study demonstrated the role of SP6 (also called epiprofin) in dental development in mice [19]. These authors suggested that SP6 binds highly GC-rich regions, similar to other members of the SP/KLF family of DNA binding proteins. Besides this study however, very little is known about SP6 and its target genes. From the current studies we conclude that SP6 is highly induced during iron-deficiency, thus making it a strong candidate for being involved in the genetic response to iron-deficiency.

Further analyses suggested that an SP1-like factor and FOX TFs may function synergistically to regulate some genes during iron-deficiency. The presence of several genes in the up-regulated cluster with SP1-like and FOX binding sites strengthens this prediction. Like the SP/KLF transcription factors, the FOX family of DNA binding protein has multiple members [20].

In summary, utilizing a unique GeneChip^® dataset and a variety of computational approaches has allowed us to predict that some genes important for the response to iron-deficiency in the rodent intestine are transcriptionally regulated by SP1 or a related TF. This prediction combined with the experimental observation that SP6/KLF14 is strongly induced by iron-deficiency suggests that SP6 could be involved in this genetic response. Further computational analysis allows us to hypothesize that a FOX family DNA binding protein may also be involved in this regulation.