Femtomolar Sensitivity of Metalloregulatory Proteins Controlling Zinc Homeostasis

See allHide authors and affiliations

Science  29 Jun 2001:
Vol. 292, Issue 5526, pp. 2488-2492
DOI: 10.1126/science.1060331


Intracellular zinc is thought to be available in a cytosolic pool of free or loosely bound Zn(II) ions in the micromolar to picomolar range. To test this, we determined the mechanism of zinc sensors that control metal uptake or export in Escherichia coli and calibrated their response against the thermodynamically defined free zinc concentration. Whereas the cellular zinc quota is millimolar, free Zn(II) concentrations that trigger transcription of zinc uptake or efflux machinery are femtomolar, or six orders of magnitude less than one atom per cell. This is not consistent with a cytosolic pool of free Zn(II) and suggests an extraordinary intracellular zinc-binding capacity. Thus, cells exert tight control over cytosolic metal concentrations, even for relatively low-toxicity metals such as zinc.

Zinc is an essential element for living organisms (1) and is the second most abundant transition metal in seawater and in humans. It is considerably less toxic than redox-active metals such as copper and is more soluble in oxygenated buffers than iron. Zinc serves as a cofactor in all six classes of enzymes as well as several classes of regulatory proteins (2, 3). Several families of integral membrane proteins transport Zn(II), moving it across membranes into and out of cells (4, 5). Less is known about the intracellular chemistry and mechanisms by which Zn(II) is sensed, stored, or incorporated as a cofactor. A common assumption is that Zn(II)-requiring enzymes and transcription factors passively acquire this essential cofactor from a cytosolic pool thought to be 10−5 to 10−12 M in free Zn(II) (4, 6–9). Direct measurements of cytosolic zinc pools have proved difficult because fractionation can lead to cross contamination between intracellular sites.

Several Zn(II)-responsive transcription factors are known to mediate zinc homeostasis in vivo (10–15) and are thought to do so by monitoring changes in this hypothetical pool of free zinc. The mammalian MTF1 sensor is estimated to have a dissociation constant K d below 90 μM (4). Estimates of the zinc sensitivity of theSynechococcus PCC7942 SmtB protein vary from <0.01 nM to 3.5 μM (7, 8). Expression of E. colizinc uptake and export genes is regulated by Zur and ZntR metalloregulatory proteins, respectively (16–20). We report here the mechanism of Zur and the calibration of both of these zinc-sensing metalloregulatory proteins to directly establish their functional “set point” relative to [Zn(II)]free. The femtomolar sensitivity of the pair indicates that intracellular fluxes of zinc between metalloenzymes and metal sensor, storage, and transport proteins are likely to involve direct transfer of the ion between proteins in kinetically controlled substitution reactions.

The total zinc content of E. coli, also known as the zinc quota, was established by inductively coupled plasma mass spectrometry (ICP-MS) analysis of whole cell lysate. Cells accumulate each transition element to a different extent, but zinc is concentrated by the greatest factor (Fig. 1). Growth in a metal-depleted medium establishes the minimal quota for this element, or 2 × 105 atoms of zinc per cell [determined here as a colony-forming unit (CFU)] (21, 22) (Fig. 1A). These quotas can be interpreted as total cellular concentrations by dividing the moles per cell by a maximum volume for a cell grown in this medium (Fig. 1, B and C). In the case of minimal medium, the minimal [Zn(II)]total corresponds to 0.2 mM. This value is ∼2000 times the ambient total zinc concentration in this depleted medium (Fig. 1B). The ability of microbes to accumulate metals such as iron to such high concentrations under starvation conditions is well established, but not documented for zinc. Cells grown in a medium replete with metals accumulate twice as much zinc per cell; however, the cell volume also doubles (23, 24), leaving the total zinc concentration unchanged (Fig. 1C). Thus, the total concentration of Zn(II) is tightly controlled in bacteria cells. We note that the iron and calcium quotas measured for each E. coli cell are the same as that of zinc. This indicates similar cellular requirements for these metal ions under these growth conditions. Far lower requirements are observed for other essential metals such as manganese and copper.

Figure 1

Metal content of E. coli cells grown in LB and glucose minimal medium as determined by ICP-MS. (A) Atoms per cell (determined as CFU) for each metal ion in minimal (MM) and LB medium. The reported values are the mean of three independent measurements; error bars are SDs. (B andC) The E. coli metallome, i.e., the total metal content of the cell, is represented in terms of moles per cellular volume for cells grown in minimal medium (B) or LB medium (C) and compared with the total metal concentrations in the relevant growth medium (22). The unfilled columns represent detection limits for low-abundance elements under these experimental conditions.

To gauge how much of the total cellular zinc is free in the cytosol, we determined the in vitro zinc responses of Zur and ZntR under conditions in which the Zn(II) concentration can be precisely controlled (25). Zur binds DNA in the presence of Zn(II), and excess EDTA inhibits this binding (17); however, the molecular mechanism of transcriptional control has not been established. DNA footprinting (Fig. 2A) demonstrates that metal occupancy of Zur modulates its DNA affinity; this directly controls the binding and activity of RNA polymerase (RNAP) at this promoter (26). The Zn1Zur form was stable as isolated and did not bind DNA when contaminating Zn(II) in the buffer was sequestered by a stringent chelator (lane 2). In the absence of added Zn(II), Zur did not compete for DNA binding, allowing RNAP to bind (Fig. 2A, lane 5) and form an open complex. The addition of excess Zn(II) allowed Zur to bind to the znuCpromoter (PznuC), and this prevented RNAP binding (lanes 3 and 6). This presumably involves the Zn2Zurform; however, the metal protein stoichiometry of the DNA binding form has yet to be established. The orientation of RNAP was determined by examining the location of deoxyribonuclease (DNase I) and KMnO4hypersensitive bands in the RNAP footprint (Fig. 2C). The DNase I hypersensitivities are usually found upstream of the transcription start site (–20 through –60 in Fig. 2A) (19). KMnO4 hypersensitivity outlines the transcription bubble that extends from the –10 box to the transcription start site, establishing that RNAP forms an open complex poised to transcribe in the znuC direction of the promoter. The znuCpromoter has an extended –10 (TGnTATTAT) with no clear –35 hexamer (Fig. 2C). Zur binding completely blocks the extended –10, thereby sterically hindering RNAP binding. These results directly establish a metal-induced repression mechanism for Zur.

Figure 2

Zur footprinting and znuCpromoter structure. (A) Zur DNase I (left, lanes 1 to 7), and KMnO4 (right, lanes 1 to 6) footprinting with the nontemplate strand of PznuC. Zur = 50 nM; RNAP = 50 nM; DNA = 1 nM; Zn(II) = 25 μM; TPEN = 10 μM. Lanes G are guanine-specific sequence ladders. (B) Mechanism for Zur regulation [extended (ext) –10]. The promoter location and protein-binding sites are scaled to match the gels shown in (A). (C) Summary of footprinting results from (A) outlining the Zur- and RNAP-binding sites and the znuC promoter structure. The extended –10 is outlined in black and the areas protected by Zur or RNAP from DNase I cleavage are highlighted in gray. Vertical arrows (↑↓) indicate DNase I–hypersensitive sites and asterisks (*) indicate KMnO4-hypersensitive sites. The transcription start site for the znuC RNA transcript (+1) was determined by primer extension (39). Dashed horizontal arrows indicate the location of the imperfect palindrome (17).

Zur-DNA interaction correlated with the concentration of Zn(II) in the assay (Fig. 3). WithN,N,N′,N′-tetrakis(2-pyridylmethyl)ethylenediamine (TPEN) present as a zinc buffer, the free zinc concentration at half-maximal Zur binding to the DNA is 9.6 (±3.0) × 10−17 M. The [Zn(II)]free was calculated with [TPEN]total,[Zn(II)]total, andK′, the apparent binding constant at a given pH and ionic strength:Embedded Image(1)This value was calculated from the absolute binding constants for Zn-TPEN (25, 27). As shown in Eq. 2, log KZn-TPEN = 15.2 at pH 7.6, 0.1 M ionic strength:Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image(2)where K a1,K a2, K a3, andK a4 are successive protonation constants for TPEN.

Figure 3

DNase I footprinting of Zur and the template strand of PznuC with varying [Zn(II)]. Zur = 50 nM; DNA = 1 nM; Zn(II) = 0 to 30 μM; TPEN = 27.5 μM; [Zn(II)]free was calculated with [Zn(II)]total, [TPEN]total, and K′Zn-TPEN. A mixture of Zur and a 340-bp DNA fragment containing the Zur operator was titrated with ZnSO4 to a final concentration of 30 μM total Zn(II). The extent of projection of the footprinted region [see (19) for methods] in the left panel is plotted against [Zn(II)] free in the right panel.

To determine whether the zinc-dependent binding of Zur to DNA correlated with repression of the znu genes, we conducted in vitro run-off transcription assays with Zur and the znuZn(II) uptake system (28). Amounts of znuC RNA transcript correlate with [Zn(II)]free (Fig. 4). With 25 μM TPEN in the buffer, half-maximal repression by Zur occurs when the total Zn(II) concentration is 6.4 (±0.4) μM, which corresponds to [Zn(II)]free= 2.0 (±0.1) × 10−16 M. This value is very close to the [Zn(II)]free value from the independent Zur–DNA-binding assay, indicating that RNAP does not substantially affect the metal response of this Zn(II) receptor. Zur competes with TPEN for Zn(II), denoting a high Zn(II) affinity of this metalloprotein.

Figure 4

Zur and ZntR transcription assay results as a function of [Zn]free. [RNAP] = [Zur] = [ZntR] = 50 nM, [DNA template] = 4 nM, [TPEN] = 25 μM. The dotted line and the solid line represent the fit of the Zur/PznuC (open triangles) and ZntR/PzntA (closed circles) data points, respectively, to a sigmoidal function. The area highlighted in gray is the range of [Zn]free between the half-maximal induction point on the two curves.

To establish the relative sensitivity of ZntR for Zn(II), we also conducted transcription experiments under identical conditions with ZntR and the zntA promoter (Fig. 4). Half-maximal activation by ZntR occurred at 16.6 (±0.7) μM Zn(II) with 25 μM TPEN, which corresponds to [Zn]free = 11.5 (±1.3) × 10−16 M. Independent measurements of ZntR affinity for Zn(II) reveal a similar value for the zinc-protein dissociation constant (29).

The transcriptional response profiles of the Zur-PznuC and ZntR-PzntA systems reveal that these proteins work in series to control zinc homeostasis. The precise zinc concentrations at which these systems switch on and off define the physiological onset of zinc starvation and zinc toxicity. As the cell acquires sufficient zinc for growth and cell division, expression of the znu uptake system is decreased. Once the minimal quota is slightly exceeded and the zinc burden of the cell begins to increase, the zntA efflux system is expressed. Comparison of the profiles provides a gauge of the optimal free zinc concentrations in the cells. The narrowness of this window reveals the delicate balance between Zn(II) deficiency and overload, suggesting a remarkably small tolerance range for cytoplasmic zinc.

The magnitudes of these Zn(II) response thresholds are surprisingly small, especially when the volume of the cell is considered. In E. coli, cell volumes typically vary over a factor of 7 depending on the stage in the cell cycle or upon the nutrient content of the growth medium (23, 24, 30, 31). With 1.8 × 10−15 liter as the maximum volume of a typicalE. coli cell in exponential phase in minimal growth medium, the lowest possible concentration of free zinc, corresponding to one zinc atom per cell, would be 1 × 10−9 M. Given that these key metalloregulatory proteins are saturated at a free zinc concentration of about 10−15 M, at least six orders of magnitude smaller than the absolute minimal value, we conclude that under normal growth conditions, there is no persistent pool of free Zn(II) ions in the cytoplasm. Initial reports for the SmtB metalloregulatory protein suggested a Zn(II) affinity in the 10−6 M range (7), but a more comprehensive study suggests a lower dissociation constant of <10−11 M (8). The latter value is consistent with the proposal that free zinc in the cytosol is not physiologically available under standard growth conditions.

These results establish several key boundary conditions for the intracellular chemistry of zinc and suggest a reevaluation of its cell biology. Considering that zinc is accumulated to a total concentration of 0.2 mM, the term “trace element” is not a useful description of the zinc status from a cell's point of view. Extraordinary requirements for iron, an element not typically referred to as a trace substance in cells, are well established (32).

Despite the abundance of zinc in the cell, the extraordinary affinity of the primary zinc sensor/regulator proteins argues against a role for free Zn(II) ions in cytosolic cellular transactions. As first suggested for copper (33), the intracellular milieu apparently consists of a myriad of tight metal ion-binding sites, which greatly outnumber the full metal ion content of the cell. Comparison of the data in Fig. 1A with the limited E. coli proteome data available to date suggests a similar situation for Zn(II): At least 12% of the zinc quota (2 × 105 atoms per cell grown in minimal medium) is tied up by only eight proteins, the major known component being RNAP (5000 copies, two Zn per copy), and five tRNA synthetases (∼2000 to 3000 copies each) (24). More than 40 additional E. coli proteins are known to require a tightly bound zinc; however, the copy numbers of these proteins are not known. A minimal estimate still accommodates the majority of atoms reported per cell in Fig. 1A. When one also considers the modest affinity of Zn(II) for nonspecific sites in the 106proteins, 107 amino acids, and 108 nucleotide equivalents in DNA and RNA per cell, an overcapacity for binding of Zn(II) and other transition metals is not unreasonable. Mammalian cells in culture maintain a total zinc quota within a narrow range around 0.4 fmol per cell, corresponding to a total cellular Zn(II) concentration in the millimolar range (4, 34). It remains to be seen whether eukaryotic cells function without free zinc in the cytosol as well. Reports of high concentrations of free zinc in mammalian cells (4, 6, 7,9) may reflect vesicular sites where free Zn(II) is maintained at high concentrations, as has been proposed elsewhere (35).

It is not clear how proteins that specifically require Zn(II) obtain this cofactor in the face of such an overcapacity for zinc sequestration; however, it appears unlikely that the cellular inorganic chemistry of zinc is under simple thermodynamic control. The most likely explanation may involve cytoplasmic zinc trafficking factors that control the kinetics of metal ion exchange between proteins, as has been proposed for the copper metallochaperone Atx1 (36).

  • * To whom correspondence should be addressed. E-mail: t-ohalloran{at}


View Abstract

Stay Connected to Science

Navigate This Article