The Zinc-Sensing Mechanism of Mouse MTF-1 Involves Linker Peptides between the Zinc Fingers

MATERIALS AND METHODS

RESULTS

Sequence alignment comparisons of MTF-1 zinc finger peptides from several species revealed that the finger linker, i.e., the four or five amino acids found adjacent to the second histidine of each zinc finger, was highly conserved among these vertebrate peptides, but much less so in Drosophila MTF-1 (Table 1). Interestingly, the amino acid sequence of each linker was unique to each finger pair, and only the linkers between zinc fingers 4 and 5 (L4-5) and fingers 5 and 6 (L5-6) conformed to the Cys₂His₂ zinc finger consensus sequence TGEKP (36, 37). In humans and mice, there is a conservative change with arginine replacing the lysine in L5-6. L2-3 differed from the consensus peptide by a single amino acid (TKEKP), whereas the other linker peptides were much less similar.

The combination of unique amino acid sequences and high evolutionary conservation of linker peptides between the first four zinc fingers of MTF-1 suggests that they may play a role in its metalloregulatory function. To examine this possibility, each of the linker peptides between these zinc fingers was individually mutated into the canonical TGEKP linker peptide. These mutations were made in an otherwise intact full-length mouse MTF-1 that was Flag tagged at the C terminus (MTF-1_Flag), and these proteins were expressed in vitro in a TnT lysate system (Fig. 1A) or in vivo in transfected mouse cells that lack endogenous MTF-1 (Fig. 1B). The ability of MTF-1 and its finger linker mutants to bind in vitro to a metal-response element (MRE-d) binding site was then assayed using an EMSA (Fig. 1). In the absence of the addition of exogenous zinc to the binding reaction, little specific MTF-1-binding activity was detected in the TnT lysate or in whole-cell extracts from transfected cells expressing native MTF-1_Flag, as reported previously (8, 17), and a similar result was obtained after mutation of L2-3. In sharp contrast, replacing the unusual RGEYT linker L1-2 resulted in constitutive DNA-binding activity that is independent of exogenous zinc, whereas replacing the short TGKT linker L3-4 resulted in diminished DNA-binding activity in response to zinc. That these results did not reflect the levels of expression of these proteins was confirmed by Western blotting (Fig. 1).

The biological relevance of the results given above was examined by monitoring the ability of MTF-1_Flag and the corresponding finger linker mutants to activate transcription of the endogenous chromatin packaged MT-I gene (Fig. 2). This was accomplished by electroporation of mutant and wild-type MTF-1_Flag expression vectors into mouse embryo fibroblasts which lack endogenous MTF-1. Transfected cells were then incubated with zinc or cadmium for 6 h and changes in MT-I mRNA were measured by Northern blotting. Very low levels of MT-I mRNA were present in the nontransfected cells, and this gene was not responsive to zinc or cadmium, but metal responsiveness was restored by native MTF-1_Flag and by the protein with the TGEKP replacement mutation of L2-3 (TKEKP). Consistent with the in vitro DNA-binding results, the linker 1 and 2 replacement mutation (L1-2; RGEYT to TGEKP) resulted in constitutive activation of the MT-I gene, whereas the linker 3 and 4 replacement mutation (L3-4, TGKT to TGEKP) was inactive under these conditions (Fig. 2). These results did not reflect apparent differences in MTF-1_Flag expression levels among the transfected cells.

The mechanisms of action of MTF-1 include metal-induced nuclear translocation and formation of an MTF-1-chromatin complex on the MT-I promoter. The nuclear translocation of MTF-1_Flag and the finger linker mutants in transfected cells was examined by Western blot analysis of cytoplasmic and nuclear extracts from control and metal-treated cells (Fig. 3), and the formation of the MTF-1-chromatin complex was monitored using a ChIP assay (Fig. 4). Native MTF-1_Flag, as well as this protein in which L2-3 or L3-4 was replaced by the canonical linker peptide, was detected in both the cytoplasmic and nuclear extracts in untreated cells. It should be noted that there is about three times more total protein per cell in the cytoplasmic extract than in the nuclear extract, but an equal amount of protein for each extract was analyzed by Western blotting. Therefore, the vast majority of MTF-1 is in the cytoplasm in these cells before treatment, and in all three cases, zinc and cadmium caused an almost complete loss of detectable MTF-1 in cytoplasmic extracts and an increased signal in the nuclear extracts (Fig. 3). Thus, metal-induced nuclear translocation of MTF-1_Flag was unimpaired by these finger linker replacement mutations, even though induction of MT-I gene expression was abolished by the canonical replacement of the shorter TGKT L3-4, under these conditions (Fig. 2). Interestingly, the L1-2 canonical replacement mutation resulted in constitutive nuclear localization of MTF-1_Flag (Fig. 3), consistent with the constitutive DNA-binding activity and activation of the MT-I gene (Fig. 1 and 2).

Interactions between MTF-1_Flag and the endogenous MT-I promoter were examined by ChIP analysis of fixed and sheared chromatin extracted from the transfected and control MEF cells as described above. Immunoprecipitation was carried out using an anti-Flag antibody. The MT-I promoter and exon I were amplified by PCR (Fig. 4). Amplification of input DNA was used to normalize between samples, and PCRs were performed in the linear range of cycle numbers (28 cycles) to assure quantitative comparisons between control and metal-treated samples. In the absence or presence of wild-type MTF-1_Flag, basal amounts of the MT-I promoter were precipitated by the anti-FLAG antibody. The background in the ChIP assay was higher than that seen using stably transfected cells due to transfection efficiency. Nonetheless, wild-type MTF-1_Flag was clearly recruited to the MT-I promoter in response to zinc and high levels of cadmium, as anticipated. The effects of the finger linker replacement mutations in MTF-1_Flag on metal-induced binding to the MT-I promoter mimicked those on MT-I mRNA induction, as described above (Fig. 2). That is, the L1-2 canonical replacement mutation lead to constitutive binding of MTF-1_Flag to the MT-I promoter and metals had little effect on the amount of binding. In contrast, the L2-3 replacement mutation resulted in slightly increased constitutive binding but a clear metal-induction of additional binding, whereas the L3-4 replacement mutation severely diminished/abolished binding to the MT-I promoter.

The results presented above suggest that L1-2 is a particularly important structural component of the zinc-sensing mechanism of MTF-1. This observation was explored in more detail by systematically mutating the amino acids in the native linker peptide RGEYT and then determining their effects on zinc-induced DNA-binding activity of MTF-1_Flag synthesized in vitro in the TnT lysate (Fig. 5) or in whole-cell extracts from transfected MEF cells (Fig. 6). The approach was to change residues in the native linker into those found in the canonical linker TGEKP. A single amino acid change of tyrosine to lysine at the penultimate residue or threonine to proline at the C-terminal end of the linker resulted in a reduced dependence on exogenous zinc for DNA-binding activity (Fig. 5A and 6A). Further mutation of the first amino acid arginine to threonine did not exacerbate the reduced dependence on exogenous zinc (Fig. 5B and 6B). In contrast, mutating both the penultimate tyrosine and C-terminal threonine residues of the native linker to lysine and proline, respectively, resulted in constitutive DNA-binding activity of MTF-1_Flag (Fig. 5B and 6B). Western blotting was used to confirm that nearly equal amounts of each MTF-1_Flag peptide were present in the TnT lysates (Fig. 5C) and the whole-cell extracts (Fig. 6C). Thus, two amino acids in L1-2 play a key role in the zinc-dependent DNA-binding activity of MTF-1.

DISCUSSION

The finding that the peptide sequences between adjacent zinc fingers play key roles in the zinc-sensing mechanisms of MTF-1 is consistent with the remarkable conservation of all five finger linker sequences among vertebrate MTF-1 orthologs. Our studies also suggest that the finger linker sequence between zinc fingers 1 and 2 confers unique properties to the zinc finger domain that allows it to sense changes in available intracellular zinc and to regulate MT-I gene expression.

In culture medium, the majority of MTF-1 is located in the cytosol, where it must sense changes in zinc availability (Fig. 3A). Nuclear translocation of wild-type MTF-1 is rapidly stimulated by the addition of low micromolar concentrations of exogenous zinc to the medium (Fig. 3A). Once bound to DNA, MTF-1 is significantly more resistant to EDTA inactivation and the zinc finger domain becomes resistant to proteolytic digestion (17, 18). We noted previously that zinc activation of MTF-1 does not occur efficiently at a low temperature (4°C), whereas once activated with zinc, its DNA binding was not diminished at a lower temperature (17, 18). Therefore, it seems likely that replacing L1-2 in MTF-1 with the canonical TGEKP peptide increases intrinsic zinc affinities for a least a subset of the zinc fingers. Our observation of constitutive nuclear translocation of the L1-2 canonical replacement mutant in the absence of exogenous zinc (Fig. 3B) strongly suggests that the well-established structural ordering of the MTF-1 fingers upon zinc binding (50) triggers this translocation process. The apparent lower zinc affinity for at least a subset of MTF-1 fingers suggests that the unusual RGEYT L1-2 linker “detunes” the finger affinities to a range of intracellular zinc that is optimal for biological functions.

Ultimately, the ternary complex of zinc-MTF-1-DNA in the nucleus regulates MT-I gene expression. Therefore the greater zinc affinity of at least some of the MTF-1 L1-2 mutants we examined is expected to shift the equilibrium toward formation of the ternary complex at lower zinc concentrations (i.e., to become a constitutively active transcription factor), as was observed (Fig. 1 and 2). As discussed in detail below, compared with similar zinc finger-DNA complexes in which adjacent fingers are separated by canonical TGEKP linkers, differences in finger-DNA interactions may also play a role in mediating the apparent detuning of the MTF-1 complex with DNA (Fig. 1). We note that our studies do not eliminate the possibility that the MTF-1 finger linker peptides are involved in other intermolecular or intramolecular interactions that govern the zinc-sensing mechanism, although no such interactions have been demonstrated. Based on nuclear magnetic resonance structural studies of the zinc finger domain, we conclude that these interactions do not include zinc binding to the linker peptides (50).

Previous finger deletion and swapping studies of MTF-1 suggested a role for zinc finger 1 in the metal-sensing function (9), but studies of the reactivity of cysteine thiol groups in the zinc fingers resulted in data that were inconsistent with this concept (4). Zinc finger 5 is hypersensitive to cysteine thiol oxidation, which indicates that the metal-binding properties of that finger may be unusual (50), and zinc fingers 5 and 6 have been suggested to play a role in the metal-sensing mechanism of MTF-1. They were reported to stabilize MTF-1-DNA complexes in vitro (12), and they are critical for MTF-1 to interact with and activate the endogenous MT-I gene (29). However, mutation of these fingers of MTF-1 does not alter its zinc dependence for MRE-binding activity measured in vitro using an EMSA (9, 29). Biochemical studies suggest that zinc fingers 1 through 4 play critical roles in high-affinity DNA binding (12). It is interesting that zinc fingers 1, 3, and 6 appear to have lower zinc-binding affinities relative to the other zinc fingers of MTF-1 (26, 50).

Evidence presented in the studies cited above is consistent with earlier suggestions that multiple zinc fingers participate in the zinc-sensing mechanism of MTF-1 (50). The quasi-ordered nature of the ensemble of solution states of the core DNA-binding zinc fingers of MTF-1 is further stabilized by DNA-binding (51; B. Potter, L. Feng, V. Matskevich, G. K. Andrews, and J. H. Laity, unpublished data). The mutational studies reported here demonstrate that finger-finger interactions in MTF-1 mediated by the RGEYT peptide connecting fingers 1 and 2 influence the equilibrium concentrations of free and stable chromatin-DNA-bound states, thus at least partially conferring zinc-sensing properties to the domain. Consistent with the notion of interacting fingers, recent studies of MTF-1 zinc fingers also reported cooperativity in metal binding (4, 26, 50). Finger-finger interactions also occur in the Zap1 transcription factor, which functions as a zinc-sensor in Saccharomyces cerevisiae (7). Zinc binding to zinc fingers 1 and 2 of Zap1 initiates conformational changes that are stabilized by interactions between these fingers (61). In contrast, earlier studies of tandem arrays of nonmetalloregulatory zinc fingers have demonstrated that they can bind zinc and adopt canonical ββα structures independently (6, 35). Thus, zinc fingers in Zap and MTF-1, the only two eukaryotic zinc sensors described to date, are unusual in their zinc-binding and interacting properties.

Although intrinsic zinc binding by MTF-1 appears to be detuned by the RGEYT linker between fingers 1 and 2, several additional factors relating to linker ordering and concomitant finger orientations upon DNA binding may also contribute to the metal-sensing function of the protein (36). The glycine at position 2 of the canonical TGEKP linker is involved in DNA-induced capping by forming hydrogen bonds with the C-terminal region of the α-helix in the preceding finger (37). All of the zinc finger linkers in MTF-1, except for the finger 2 and 3 linker, also have glycine at position 2. Moreover, lysine has been proposed as a possible glycine substitution at this linker position to still facilitate α-helix capping (5). In this context, it is not surprising that the canonical swap mutant of L2-3 had little effect on MTF-1 zinc and DNA-binding properties (Fig. 1 and 2). The lysine at position 4 can interact with the phosphate backbone of the DNA (49), but tyrosine is found in that position in the RGEYT linker connecting fingers 1 and 2 of MTF-1. The C-terminal proline of the canonical TGEKP linker orients adjacent fingers along the major groove of DNA (49). Threonine is found in this position in the linkers between fingers 1 and 2 and fingers 3 and 4. Our site-directed mutagenesis of L1-2 suggested that the two C-terminal positions of the linker are particularly important. Altering potential interactions with the phosphate backbone by replacing tyrosine with lysine and affecting the orientation of the adjacent finger by replacing the threonine with proline in zinc finger 1 of MTF-1 abolished the dependence on exogenous zinc for activity. This observation may suggest that these tyrosine and threonine residues at the C-terminal positions of the native RGEYT linker in MTF-1 contribute to finger-finger interactions which destabilize zinc binding in the zinc finger domain. In contrast, similar changes in L3-4 had the opposite effect on the dependence on exogenous zinc for activity and reduced the sensitivity of MTF-1 to zinc in vivo and in vitro. In that case, the native linker is unique among the MTF-1 finger linkers and consists of only four residues (TGKT). Replacement of that linker with the canonical linker peptide TGEKP lengthened the native linker by a residue, which also altered the net charge of the linker, and added a C-terminal proline, which is predicted to affect the orientation of the next finger (49).

An issue that requires consideration is whether the concentrations of zinc required to activate the DNA-binding activity of MTF-1 in vitro are physiologically relevant. Since the unusual requirement for exogenous zinc for wild-type MTF-1-finger-DNA complex formation in vitro strongly suggests a lower affinity for zinc and, thus, greater zinc ion concentration requirement for folding, it is reassuring that the biological activity of MTF-1 in vivo unerringly reflected the zinc dependence for DNA-binding of MTF-1 that was measured in vitro. However, the actual amount of zinc available to interact with MTF-1 in the TnT lysates and whole-cell extracts is not known because many other proteins and potential zinc-binding sites are present. Recent calorimetric studies indicate an apparent dissociation constant for zinc for the purified MTF-1 zinc finger domain in the high nanomolar to low micromolar range (L.S. Feng, G. K. Andrews, and J. H. Laity, unpublished observation). Similar studies of zinc fingers 1 and 2 in Zap1 identified a functional zinc-responsive concentration in the low-nanomolar range (apparent K_d, ∼5 nM) (61). In contrast, canonical Cys₂His₂ zinc fingers fold independently and bind zinc with high affinity (10⁻⁹ to 10⁻¹² M⁻¹) (34, 35). These observations are consistent with the notions presented here and suggest that Zap1 zinc fingers, like at least some MTF-1 fingers, are detuned such that their activity and structure are dependent on higher concentrations of zinc than are other zinc fingers. It was estimated that at least 100 labile atoms of zinc per yeast cell is required to occupy zinc fingers 1 and 2 of Zap1 (61).

Several recent studies have reported the uptake kinetics for ⁶⁵Zn in several different types of cultured mammalian cells (15, 22, 23, 59). In the HEK cells, the initial rate of uptake of labeled zinc from the culture medium was ∼4 pmol zinc/min/mg protein with an accumulation of 25 pmol zinc/mg protein within 10 to 15 min of exposure to elevated concentrations of zinc (59). For the sake of this discussion, a value of 300 pg of protein per cell and a value of 3 × 10⁻¹² liters for the volume of a “typical” cell were used (41) to calculate that ∼4 × 10⁶ to 8 × 10⁶ atoms of zinc can be taken up per HEK cell in these experiments. This would increase the concentration of zinc inside the cell by ∼2 to 4 μM. Similar studies of zinc uptake in K562 cells reported the accumulation per cell of an additional 0.7 μM zinc (in complete medium) to 27 μM zinc (in minimal medium) during a short exposure to elevated zinc (22, 23). Studies of zinc uptake in PC3 and LNCaP cells report the rapid accumulation per cell of an additional 15 μM zinc and 200 μM zinc, respectively, after a short (5- to 10-min) incubation in medium containing 15 μM zinc (15). Clearly these are rough and static estimates that do not address the availability of the rapidly accumulated zinc, and the volume of a eukaryotic cell is highly dependent on the cell type and varies over 4 orders of magnitude. Nonetheless, these zinc uptake data support the concept that the concentrations of zinc required to induce the DNA-binding activity of MTF-1 in vitro fall within a range of available zinc concentrations that may occur, at least transiently, in mammalian cells acutely exposed to high levels of zinc.