There is a growing appreciation for the diverse roles of zinc as a signalling substance in biological systems. Zinc signalling is brought about by changes in intracellular concentrations of labile Zn2+, resulting in both genomic and non-genomic effects. The genomic responses are largely mediated by MTF1 (metal-regulatory transcription factor 1), which binds to MREs (metal-response elements) in the 5′ regulatory region of genes in response to zinc. Treatment of cultured zebrafish ZF4 cells with siRNA (small interfering RNA) to MTF1 changed the transcriptional response to zinc for over 1000 genes, as assessed using an oligonucleotide microarray. From this primary list of MTF1-dependent genes, we identified a relatively small cohort that showed a configuration of MREs in their 5′ regulatory regions similar to known MTF1 targets. This group showed a remarkable dominance of nucleic acid-binding proteins and other proteins involved in embryological development, implicating MTF1 as a master regulator of gene expression during development.
MTF1 (metal-regulatory transcription factor 1) was identified as the protein that bound the previously discovered MREs (metal-response elements) (core consensus 5′-TGCRCNC-3′) which had been established as being responsible for metal-mediated transcriptional induction of MT (metallothionein) [1,2]. Owing to the pivotal function played by MT in metal detoxification, MTF1 was heralded as the molecular switch that co-ordinated the cellular defence against excess metal ions. Elegant mechanistic characterization of the functional motifs of the MTF1, revealing six zinc fingers involved in DNA binding and a transactivation domain incorporating acidic, proline-rich and serine/threonine-rich regions, has demonstrated that this molecule does contain a metal switch; however, direct activation is the sole preserve of zinc .
The multifaceted role played by MTF1 in metal homoeostasis became apparent when it was found to up-regulate ZnT1 (zinc transporter 1), also known as SLC (solute carrier) 30A1, [4–6], the major zinc exporter, in response to excess zinc as well as being involved in inducing expression of proteins that protected against reactive oxygen species, one of the major mechanism by which metals cause toxicosis . Therefore MTF1 co-ordinates transcriptional control of the major metallochaperone, the mechanism by which excess metals are removed, and the cellular defences against any damage caused by excess metals. Any initial assumptions that the extent of its functional footprint was limited to responses to metal toxicology were soon questioned with the identification of a wider role in co-ordinating system-level stress responses ranging from inducing generic heat-shock to moderating specific signalling pathways . Indeed, when MT-null mice were generated and showed only a subtle cadmium-sensitive phenotype , people questioned whether its major regulator may also be redundant or at least have limited importance. The revelation that MTF1-null mice were embryonically lethal was direct evidence that this transcription factor was core to more than metal toxicology, playing an essential role in differentiation and development . Since this observation, several groups have striven to define the extent of the network of proteins controlled by MTF1 and through this to understand its many biological roles. We have been attempting to dissect out the complex role played by zinc in mediating transcriptional control in zebrafish by probing specific and global transcript changes during manipulation of zinc status to manifest conditions of zinc deficiency or overload . Intriguingly, these animal experiments have revealed a substantive bias towards alteration in expression of transcription factors and other genes of key importance for development. We therefore set out to examine the role played by MTF1 in zebrafish due to its adoption as a key developmental model.
Defining MTF1 targets
A major challenge in identifying targets of MTF1, either from the literature or through experimentation, is in separating the primary effect of any prescribed manipulation from secondary responses. A simple example is provided by those experiments that have associated zinc-responsive genes with MTF1 regulation using single time-point data. This approach, although informative in relation to zinc, makes the following assumptions: (i) no secondary zinc-mediated transcriptional regulators exists (a shaky assumption where the majority of transcription control exploits zinc-finger proteins); (ii) you can distinguish between zinc-mediated MTF1 transcriptional control and any toxicological response to zinc; and (iii) the observed response is mediated directly by MTF1 rather than being effects that are downstream in the response cascade and therefore only controlled indirectly by MTF1. Direct in situ detection of transcription factor binding, using techniques such as ChIP (chromatin immunoprecipitation)-on-chip, is ideal for locating binding sites, but does not provide information relating to the functional action of the detected association and has significant technical limitations, such as availability of both highly specific antibodies against the transcription factor of interest and promoter or tiling microarrays.
For many years, MTF1 was considered solely as a positive regulator of transcription, but this assumption was proved incorrect by the discovery that it can repress expression of the zinc importer, SLC 39A10 [ZIP10 (ZRT/IRT-like protein 10)] [12,13]. We found that expression of the zebrafish SLC39A10 in response to zinc is tissue-dependent  and that the gene has two alternative transcripts . These transcripts have separate promoters with opposing actions providing positive and negative MTF1/MRE zinc-mediated transcriptional control respectively, apparently depending on tissue type [11,14]. The two promoters were cloned independently, and site-directed mutagenesis was exploited in combination with in vitro reporter assays in HepG2 cells to confirm the involvement of the MRE-binding sites in both repression and activation. These results suggest further that the opposing actions by the two promoters in response to zinc do result primarily from the context of the MRE itself and may be moderated by co-activators and co-repressors. In two recent papers, it was elegantly shown that both mouse and Drosophila MTF1 bind different co-activator subunits depending on the core promoter context [15,16], so this may extend to formation of multiprotein complexes with complementary co-repressors.
Because of the diverse effects of MTF1 on gene expression, we set about designing an experiment that would reveal candidate MTF1 target genes and identify the directionality of the MTF1-mediated zinc response. First, we would exploit zinc-based gene induction, but couple it with siRNA (small interfering RNA) knockdown technology to explore the population of transcripts that were regulated, both positively and negatively, by zinc where MTF1 was present, but where the observed regulation was abolished when MTF1 knockdown was achieved by siRNA treatment. This was performed in the zebrafish cell line ZF4 using a zinc concentration (10 μM) well within the homoeostatic range where no adverse effects, i.e. cell death or cell detachment, were observed. The effectiveness of our approach was confirmed by exploiting a QPCR (quantitative reverse transcription–PCR) assessment of the elimination of MT mRNA by the siRNA. Treatment of ZF4 cells with siRNA to MTF1 reduced the abundance of MT2 mRNA by 78 and 98% in the presence of 10 μM zinc and with no added zinc respectively (Figure 1A). Following this positive validation, we performed global analysis of transcript level changes using an oligonucleotide-based 17000 element zebrafish microarray . Examination of the data showed that ZnT1 and Zip10 exhibited the expected up- and down-regulation respectively in response to zinc in the presence of MTF1, but, upon siRNA knockdown, these responses were absent (Figure 1B).
Suppression of zinc effects on expression of MT and zinc transporters ZnT1 and ZIP10 by MTF1 siRNA
Functional footprint of MTF1
Walter Schaffner's group made the first comprehensive effort to map targets for MTF1 by synthesis of experimental data, bioinformatics and literature . This approach provided substantive evidence relating to MTF1 regulation of genes of diverse functions, including ZnT1, G6PD (glucose-6-phosphate dehydrogenase), CEBPA (CCAAT/enhancer-binding protein α), NOTCH1, AFP (α-fetoprotein) and LCN1 (lipocalin 1), many of which seem to display MTF1-dependent expression across most vertebrates and Drosophila . Further incisive work exploiting mouse MTF1-knockout strains and siRNA approaches has contributed substantially to the understanding of this field [8,10,12,13,18]. Our own work on zebrafish indicates that there may be as many genes that are negatively regulated by MTF1 as there are genes that are induced by MTF1 binding to their promoters. Separating out those zebrafish genes whose response to zinc was substantially reduced in the presence of MTF1 siRNA revealed 413 where a positive zinc regulation was suppressed and 599 which exhibited suppression of negative zinc regulation (for a full list with annotation see Supplementary Table S1 at http://www.biochemsoctrans.org/bst/036/bst0361252add.htm). Ontological bias analysis of these combined gene lists provides a direct insight into the processes influenced by MTF1 (Figure 2). These data clearly illustrate the connection between MTF1 and the processes of development, biogenesis and morphogenesis, providing mechanistic clues to the embryonic lethality of the MTF1-knockout mouse. Interestingly, analysis of the distribution of Interpro protein domains among proteins encoded by MTF1-dependent genes revealed a bias for a group of 18 proteins with homeobox domains overrepresented at a significance of P=1.17×10−5, also linking to this key development role. The high prevalence of genes with molecular function definitions pertaining to regulation of gene expression (e.g. transcription factor activity or DNA binding) shows that MTF1 is central to the biological co-ordination of significant transcriptional cascades. The enrichments of transcripts related to liver and placenta may be related to the critical point in development when the MTF1-null mice cease to develop. Intriguingly, transcripts associated with development of the eye and particularly the retina also appear to be overrepresented; the connection to this site is reinforced by the identification of four genes within the visual signal transduction pathway (bias P=0.014). This result is not highly surprising in the organ where the highest concentration of zinc in the body is found. However, it is of particular interest that the severely zinc-deficient Zip4−/− or Zip−/+ mice embryos develop anophthalmia and are often born lacking one or both eyes . Finally, the identification of nine MTF1-dependent genes in the insulin signalling pathway (KEGG hsa04910; P=0.09) resonates with the recently revealed genetic zinc link to Type 2 diabetes 
Functional annotation analysis of putative MTF1-responsive genes
Finding the first ripples in the MTF1 cascade
A complicating factor in identifying MTF1 targets from knockdown experiments is that it is difficult to separate those genes that are controlled directly by MTF1 from those that are further downstream in the cascade. However, the prediction can be improved by analysis of the promoter sequences of the putative MTF1-responsive genes. Classical approaches to transcription factor binding, such as combined site-directed mutagenesis and reporter assays, have defined a consensus MRE sequence with which MTF1 associates. Unfortunately, the core sequence 5′-TGCRCNC-3′, being only 7 bp and containing an undefined residue and a position which can accommodate one of two bases, has a probable random occurrence of 1 in 2048 bp. This restricts the effectiveness of using its sole presence as a good indicator of MTF1 regulation. However, further comprehensive analysis of many native promoters, manly those of MT, have produced evidence showing that a number of extended binding features (normally involving 3′ extension of the core sequence) are more effective at transcriptional activation and demonstrate a substantial increased efficacy if a pair of MREs is present proximal to the start of transcription . This evidence can be used to score (or weight) the probability that a promoter may respond to MTF1. Therefore the 1012 genes shown to respond to zinc in an MTF1-dependent manner were mapped on to their respective loci using the Ensembl Zebrafish genome (Zv7) and sequences upstream of the coding regions were downloaded. Using this procedure, we were able to retrieve 482 upstream sequences. These sequences were searched for the MRE core consensus, 5′-TGCRCNC-3′, as well as all other more constrained MRE motifs contained within the Transfac database (2007) using GCG version 11.1 (Accelrys) software. A total of 297 sequences contained at least one MRE within 2 kb upstream of the start codon. By using a scoring system that took into account the total number of MRE motifs, their specificity (e.g. consensus or exact), whether they occurred in pairs (e.g. <50 bp apart) and whether the first pair was proximal (within <500 bp of the start codon), we filtered the MTF1-responsive genes to identify putative MTF1 targets (scoring results are shown in Supplementary Table S2 at http://www.biochemsoctrans.org/bst/036/bst0361252add.htm). The final shortlist of candidate MTF1 targets consisted of 43 genes. A staggering 44% of these are involved in developmental processes, with ‘multicellular organismal development’ being the largest significantly enriched GO (Gene Ontology) Biological Processes term (P=1.2×10−5). The largest GO Molecular Function category among the candidate MTF1 targets was ‘nucleic acid binding’, which comprised 48% of the genes in the list (P=5.6×10−6). This result is completely in line with the fact that MTF1 is essential for embryonic organ development, whereas targeted deletion post-partum is not lethal [10,13]. Another interesting finding is that, among the candidate MTF1 targets, there were more genes that were down-regulated in the presence of zinc and MTF1 than there were genes that were up-regulated by this condition. The meta-relationship between these elements could be observed by constructing a network that included known interactions between all MTF1-dependent genes from our experiment in addition to the postulated effects of MTF1 on the 43 candidate target genes (Figure 3 and Supplementary Table S2). The network, which contains 59 objects, shares key elements with similar networks derived for genes observed as regulated using in vivo experiments with zinc excess (17 common features) and zinc deficiency (12 common features), confirming the validity of the approach. Importantly, the signature features of MTF1 induction on MT2 and ZnT1 (SLC39A1) and repression of ZIP10 (SLC39A10) expression in zebrafish were captured in the network . Candidate MTF1 targets also included the predicted zebrafish NDRG4 (N-myc downstream regulated 4) gene (zgc:136766). A paralogue to this gene, NDRG1, has already been described as a likely MTF1 target in mouse . Among the genes that were up-regulated by zinc in an MTF1-dependent manner, A2CET1 (similar to radial spokehead-like 1) (seven MREs), IMPDH2 (IMP dehydrogenase 2) (five MREs), LRP12 (low-density-lipoprotein-related protein 12) (four MREs), ZIC1 (zic family member 1) (four MREs) and SNX4 (sorting nexin 4) (three MREs) scored highly in our ranking system described above (Figure 3 and Supplementary Table S1). Top scoring genes in the population that showed zinc- and MTF1-dependent down-regulation included FOXO5 (forkhead box O5) (five MREs), PIK3R3 [phosphoinositide 3-kinase, regulatory subunit, polypeptide 3 (p55γ)] (four MREs), MEIS2.2 (myeloid ecotropic viral integration site 2.2) (eight MREs), zgc:113054 (four MREs), HER7 (hairy and Enhancer of split-related 7) (six MREs), SP9 (sp9 transcription factor) (seven MREs), H2AFY (histone H2A family member Y) (six MREs), ATP1A3B (ATPase, Na+/K+-transporting, α3b polypeptide) (four MREs), HAND2 (heart and neural crest derivatives-expressed transcript 2) (five MREs), NEUROD (neurogenic differentiation) and PCDH10A (protocadherin 10a) (four MREs) (Figure 3 and Supplementary Table S2).
Direct interaction network for the putative MTF1 target genes and their interacting partners among genes that were regulated by zinc in an MTF1-dependent fashion
MTF1 is required for embryonic development and for metal and oxidative stress responses after birth . This is reflected in the functional classifications of genes that showed MTF1-dependent zinc induction in zebrafish ZF4 cells. Through bioinformatic filtering of these genes, by rules applied to their upstream MRE landscape, we were able to reduce and refine the number of candidate MTF1 targets down to 43 genes. Nearly half of these code for transcription factors and other proteins involved in development, thus clearly linking to the essentiality of MTF1 in embryogenesis. Although the genes identified here are putative MTF1 targets, which require secondary functional validation, they allow us to formulate discrete testable hypotheses about the biological functions of MTF1 in health and disease. For example, there are persistent hints in the literature of links between zinc signalling, MTF1 and cancer [8,10,12,22–24]. In our MTF1 target gene search, we found five genes [e.g. LRP12, PAX7 (paired box 7), CBFB (core-binding factor, β subunit), IMPDH2 and SLC39A10] with known connections to cancer, adding to the emerging picture of the involvement of zinc in this disease.
Increasing evidence is accruing supporting the role for zinc as a significant signalling molecule somewhat analogous to calcium , and it was shown recently to act as a second messenger, transducing FcεRI (high-affinity IgE receptor) stimulation to intracellular events . Zinc signalling is mediated by zinc transporters of the SLC30 and SLC39 families that carry zinc across biological membranes [14,27]. The mechanisms by which the zinc ‘signal’ can evoke activation of complex and diverse pathways remain for the most part unresolved. Specificity may in part be conferred by tissue-specific expression of zinc transporters. However, the evidence shown in the present article brings together a weight of evidence highly suggestive that MTF1 may act as a key node in transmitting the zinc signalling process through transcriptional control. By controlling a complex network of other transcription factors, MTF1 sits at the centre of a web that penetrates many essential biological processes.
Metal Metabolism: Transport, Development and Neurodegeneration: A Biochemical Society Focused Meeting held at Imperial College London, U.K., 9–10 July 2008. Organized and Edited by David Allsop (Lancaster, U.K.) and Harry McArdle (Rowett Research Institute, Aberdeen, U.K.).