Measuring steady-state and dynamic endoplasmic reticulum and Golgi Zn²⁺ with genetically encoded sensors

Abstract

Zn²⁺ plays essential roles in biology, and cells have adopted exquisite mechanisms for regulating steady-state Zn²⁺ levels. Although much is known about total Zn²⁺ in cells, very little is known about its subcellular distribution. Yet defining the location of Zn²⁺ and how it changes with signaling events is essential for elucidating how cells regulate this essential ion. Here we create fluorescent sensors genetically targeted to the endoplasmic reticulum (ER) and Golgi to monitor steady-state Zn²⁺ levels as well as flux of Zn²⁺ into and out of these organelles. These studies reveal that ER and Golgi contain a concentration of free Zn²⁺ that is 100 times lower than the cytosol. Both organelles take up Zn²⁺ when cytosolic levels are elevated, suggesting that the ER and Golgi can sequester elevated cytosolic Zn²⁺ and thus have the potential to play a role in influencing Zn²⁺ toxicity. ER Zn²⁺ homeostasis is perturbed by small molecule antagonists of Ca²⁺ homeostasis and ER Zn²⁺ is released upon elevation of cytosolic Ca²⁺ pointing to potential exchange of these two ions across the ER. This study provides direct evidence that Ca²⁺ signaling can influence Zn²⁺ homeostasis and vice versa, that Zn²⁺ dynamics may modulate Ca²⁺ signaling.

Zinc is the second most abundant transition metal in biological organisms, with mammalian cells containing an estimated 100–500 μM total Zn²⁺ (1, 2). The majority of this Zn²⁺ is bound to proteins and enzymes and indeed it has been estimated that approximately 10% of the human proteome or 3,200 proteins require zinc for their structure and function (3). It is now well established that mammalian cells contain a small but measurable pool of free or labile Zn²⁺ in the cytosol that is buffered in the picomolar range (4–6). Fluctuations in cytosolic Zn²⁺ have been shown to influence signaling cascades such as the mitogen activated protein kinase pathway (7), as well as cellular processes such as mitochondrial function (8, 9), apoptosis (10, 11), and dendritic cell maturation (12). Although many studies have focused on cytosolic Zn²⁺, the overall distribution of zinc at the subcellular level is not well defined. In particular, it is not clear whether organelles contain labile Zn²⁺, if so how these Zn²⁺ pools influence organelle function, and whether organelles modulate fluctuations in cytosolic Zn²⁺. Yet defining where Zn²⁺ is located, whether it translocates from one place to another in response to stimuli, and how Zn²⁺ flux influences cellular processes is a fundamental part of elucidating how cells regulate and are influenced by this essential metal ion.

Mammalian cells contain an elaborate network of Zn²⁺ transporters, including 10 ZnT/cation diffusion facilitator (SLC30) family members and 14 Zrt-, Irt-like protein (SLC39) family members (13). These transporters help mediate zinc flux into and out of the cell and intracellular organelles. It is well established that Zn²⁺ can be concentrated into secretory vesicles (14), and recent studies have demonstrated a labile pool of Zn²⁺ in mitochondria (9, 15, 16), yet little is known about other intracellular organelles such as the endoplasmic reticulum (ER) and Golgi. Numerous proteins found within the secretory pathway require Zn²⁺ for their function, including resident ER chaperones such as calnexin and calreticulin, as well as secreted proteins such as metalloproteases and alkaline phosphatases (2). Moreover, deletion of an ER-localized ZnT in both yeast and higher eukaryotic cells led to activation of the unfolded protein response and general ER dysfunction (17, 18), suggesting Zn²⁺ plays an essential role in normal ER function.

In this work, we develop high-affinity genetically encoded Zn²⁺ sensors targeted to organelles to allow direct monitoring of ER and Golgi Zn²⁺ levels. These sensors reveal free Zn²⁺ in both the ER and Golgi at a concentration just under 1 pM. We demonstrate that the ER and Golgi sequester excess cytosolic Zn²⁺ indicating that organelles have the potential to play a role in modulating bioavailability of cytosolic Zn²⁺. Moreover, using a combination of Ca²⁺ and Zn²⁺ sensors along with small molecule antagonists of calcium homeostasis, we discover a connection between ER Ca²⁺ stores and ER Zn²⁺ dynamics, providing evidence of interplay between Zn²⁺ homeostasis and Ca²⁺ signaling.

Results

Generation of High-Affinity Zn²⁺ Sensors Comprising Zap1, Cyan, and Yellow Fluorescent Proteins (ZapCY1 and ZapCY2).

In this work, we optimized a sensor developed by Qiao et al. (19) which is comprised of the first and second zinc fingers of Saccharomyces cerevisiae Zap1 sandwiched between two fluorescent proteins [enhanced cyan fluorescent protein (CFP) and enhanced yellow fluorescent protein (EYFP)]. Zn²⁺ binding induces a conformational change in the pair of zinc fingers leading to an increase in FRET from CFP to YFP (Fig. 1A). In yeast suspensions of the original sensor, Zn²⁺ led to a 1.3-fold increase in the FRET ratio (R), defined as the FRET emission intensity divided by the CFP emission intensity (19). To optimize the sensor, we truncated CFP, replaced EYFP with the more pH stable citrine (20), and changed the amino acids in the linker regions to those found in Ca²⁺ sensors. Such changes have previously been shown to optimize the FRET response of genetically encoded calcium sensors (21). To characterize the sensor, which we named ZapCY1, we measured the apparent dissociation constant () of purified sensor protein in vitro. The Zn²⁺ binding curve was fit to a one-site saturation model yielding a (Fig. 1B). Titration with a series of metal ions revealed that ZapCY1 was selective for Zn²⁺ over other metal ions (Ca²⁺, Mg²⁺, Cu¹⁺, Cu²⁺, Mn²⁺, Co²⁺, Ni²⁺, Fe²⁺) (Fig. 1C and SI Appendix, Fig. S1A). Only Zn²⁺ gave rise to a FRET ratio change when ZapCY1 protein was treated with excess metal ions. Because Ca²⁺ and Mg²⁺ are present at high concentrations in cells, we verified that micromolar concentrations of these ions do not perturb the sensor in vitro (SI Appendix, Fig. S1C) or in HeLa cells (SI Appendix, Fig. S2E). Treatment with neocuproine, a Cu¹⁺ chelator, and 2,2-bipyridine, an Fe²⁺ chelator, had no effect on the sensor in HeLa cells, indicating that physiological Cu¹⁺ and Fe²⁺ do not interfere with the sensor in cells under resting conditions (SI Appendix, Fig. S2). Fig. 1D demonstrates that ZapCY1 exhibits a large decrease in the FRET ratio when treated with a membrane permeable metal chelator N,N,N^′,N^′-tetrakis-(2-pyridylmethyl)-ethylenediamine (TPEN), followed by an increase upon saturation with Zn²⁺, yielding the minimum and maximum ratio (R_min and R_max), respectively. In cells, ZapCY1 exhibits a 4.15-fold dynamic range (R_max/R_min) (Fig. 1D).

Fig. 1.

Generation of high-affinity Zn²⁺ sensor ZapCY. (A) Schematic of the ZapCY sensor. (B) In vitro titration of ZapCY1 (●, K_d = 2.53 pM) and ZapCY2 (▪, K_d = 811 pM, n = 0.44) sensor. (C) Metal selectivity of ZapCY1 and ZapCY2 sensor. Two micromolar ZapCY1 solutions were combined with 15 μM EDTA and 20 μM indicated metals, to yield 2∶5 molar ratio of protein:metal. For ZapCY2, 1 μM metal was added to 2 μM sensor solutions to yield 2∶1 solutions. Data represent the FRET ratio upon addition of a given metal minus the R_min. (D) Representative experiment (n = 5 cells) demonstrating changes in the FRET ratio of ZapCY1 in the cytosol of HeLa cells. (E) Representative experiment (n = 6 cells) demonstrating changes in the FRET ratio of ZapCY2 in the cytosol of HeLa cells. For D and E, FRET ratios differ from in vitro ratios due to different instrumentation and data processing. Microscope filter combinations for FRET and CFP: 430/24 excitation filter, 455 dichroic, 535/25, and 470/24 emission filters, respectively.

Mutation of two cysteines in the binding domain of ZapCY1 to histidine lowered the , generating another sensor ZapCY2. In vitro Zn²⁺ titration data fit equally well to an empirical Hill expression ( of 811 pM, n = 0.44, Fig. 1B) and a two-site saturation model (SI Appendix, Methods). This sensor yielded a 1.4-fold dynamic range in HeLa cells (Fig. 1E). Pseudocolored FRET ratio images of a representative titration are presented in SI Appendix, Fig. S3. ZapCY2 was less selective than ZapCY1 at a 2∶5 molar ratio of protein:metal (SI Appendix, Fig. S1D), but was reasonably selective for Zn²⁺ at a 2∶1 molar ratio (Fig. 1C). However, ZapCY2 was perturbed by Cu¹⁺, even at very low molar excesses in vitro (SI Appendix, Fig. S1A). Therefore, we tested whether Cu¹⁺ was capable of influencing measurements of resting cytosolic Zn²⁺; neocuproine had no effect on the resting FRET ratio, indicating that physiological Cu¹⁺ does not perturb the sensor under resting conditions (SI Appendix, Fig. S2D). In addition, ZapCY2 was not perturbed by endogenous Fe²⁺ or micromolar levels of Ca²⁺ and Mg²⁺ in cells (SI Appendix, Figs. S1 and S2).

Fig. 1 demonstrates that ZapCY1 was fully saturated in the cytosol under resting conditions (i.e., R_resting = R_max), whereas ZapCY2 was only partially saturated. Using ZapCY2, we measured R_resting, R_min, and R_max in 16 individual cells and used these parameters to estimate the concentration of free Zn²⁺ in the cytosol to be 80 ± 7 pM (SI Appendix, Table S1). This estimate is consistent with previous reports that resting cytosolic concentrations in human colon adenocarcinoma HT-29, HEK 293, and rat insulinoma INS-1(832/13) cells are in the hundreds of picomolar range (5, 22).

Measurement of Zn²⁺ in the ER and Golgi.

Because we were unsure whether ER and Golgi Zn²⁺ levels would be higher or lower than the cytosol, the high-affinity ZapCY1 and a low-affinity Zn²⁺ sensor ( 1 μM), ZifCY1, previously developed in our lab (15) were targeted to lumen of the ER and the inner surface of the Golgi membrane (Fig. 2 A and D). Colocalization with established ER and Golgi markers confirmed targeting to the desired organelles (SI Appendix, Figs. S4 and S5). Treatment with TPEN and Zn²⁺ revealed that both sensors were functional in each of the targeted locations (Fig. 2). TPEN led to a decrease in the FRET ratio for both ER-ZapCY1 and Golgi-ZapCY1 (Fig. 2 B and E), indicating the presence of free Zn²⁺ in the ER and Golgi. Under the same conditions, no decrease in the FRET ratio was observed for the low-affinity ER-ZifCY1 and Golgi-ZifCY1, likely because the level of Zn²⁺ in ER and Golgi was below the detection limit. Under resting conditions, ZapCY1 was 100% saturated in cytosol, but only 26% saturated in ER and 18% in Golgi, revealing that the labile Zn²⁺ level in ER and Golgi was lower than in cytosol (SI Appendix, Fig. S6A). To calculate resting Zn²⁺ concentrations in the ER and Golgi, HeLa cells were treated with TPEN to obtain R_min and perfused with Zn²⁺ and pyrithione to saturate the sensor and obtain R_max (Fig. 2 B and E). For the ER, thapsigargin was used in combination with TPEN because this aided depletion of the ER. Pseudocolored FRET ratio images of a representative titration are presented in SI Appendix, Fig. S3. The free Zn²⁺ in ER ([Zn²⁺]_ER) was estimated to be 0.9 ± 0.1 pM and 0.6 ± 0.1 pM in Golgi ([Zn²⁺]_Golgi) (SI Appendix, Table S1). The dynamic range of the targeted sensors were comparable [ER-ZapCY1, 2.16 ± 0.06 (n = 13 cells, three independent experiments); Golgi-ZapCY1, 2.09 ± 0.07 (n = 13 cells, three independent experiments)], but were lower than that observed in the cytosol. These data demonstrate that the ER and Golgi contain free Zn²⁺ that is buffered at a concentration significantly lower than the cytosol.

Fig. 2.

Targeting Zn²⁺ sensors to ER and Golgi in HeLa cells. (A) Fluorescence image of ER-ZapCY1. (B) Representative traces (n = 6 cells) for ER-ZapCY1. The FRET ratio decreased with 150 μM TPEN/10 μM thapsigargin and increased with addition of 5 μM pyrithione and 10 nM Zn²⁺. (C) Representative traces (n = 6 cells) for ER-ZifCY1. The FRET ratio did not change with TPEN but increased with 25 μM pyrithione, 0.001 mg/mL saponin, and 500 μM Zn²⁺. (D) Fluorescence image of Golgi-ZapCY1. (E) Representative traces (n = 4 cells) for Golgi-ZapCY1. The FRET ratio decreased with 150 μM TPEN and increased with addition of 5 μM pyrithione and 10 nM Zn²⁺. (F) Representative traces for Golgi-ZifCY2 (n = 7 cells). The FRET ratio did not change with TPEN but increased with addition of 25 μM pyrithione, 0.001 mg/mL saponin, and 500 μM Zn²⁺.

Zn²⁺ Sequestration into ER and Golgi.

Intracellular Zn²⁺ levels are tightly regulated by transporters and buffering by cytosolic proteins such as metallothionein (2, 22, 23). A fundamental unanswered question is whether organelles sequester zinc, particularly given that there is an excess of high-affinity ligands in the cytosol (22). To address this question the plasma membrane was permeabilized with saponin and buffered Zn²⁺ solutions were added to cells. Zn²⁺ uptake was observed with both the high-affinity ER-ZapCY1 and low-affinity ER-ZifCY1, demonstrating that the ER can concentrate high levels of Zn²⁺ (Fig. 3 A and B). In the Golgi, only Golgi-ZapCY1 detected Zn²⁺ influx, suggesting that the Golgi can concentrate cytosolic Zn²⁺ but to a lower level than the ER (Fig. 3). Next, we explored whether changes in the extracellular environment could affect cytosolic Zn²⁺ levels and organelle sequestration. Fig. 3C demonstrates that when 100 μM Zn²⁺ is added to cell media, in the absence of permeabilizing agents, Zn²⁺ is transported into the cytosol, reaching levels of approximately 5.5 nM (SI Appendix, Fig. S7). In this experiment, excess Zn²⁺ that accumulates in the cytosol is sequestered into both the ER and Golgi, providing direct evidence that Zn²⁺ is transported into organelles when cytosolic levels reach the nanomolar range. SI Appendix, Fig. S11 presents pseudocolor ratio images of the changes in [Zn²⁺]_ER over the time course presented in Fig. 3C.

Fig. 3.

Comparison of Zn²⁺ uptake in ER and Golgi. (A) HeLa cells expressing ZifCY1 targeted to ER and Golgi were permeabilized with saponin for 30 min and then incubated with ZnCl₂ (ER: black square, 100 μM, n = 10 cells; red circle, 12 μM, n = 4 cells; green diamond, 0.58 μM, n = 4 cells; violet triangle, 0.044 μM, n = 7 cells; and blue square, Golgi, 100 μM, n = 9 cells). (B) HeLa cells expressing ER-ZapCY1 were permeabilized with saponin for 30 min and then incubated with 193 pM (violet triangle, n = 4 cells), 2 nM (green circle, n = 7 cells), and 44 nM (red square, n = 4 cells) Zn²⁺. Uptake in ER was recorded as low as 2 nM Zn²⁺. (C) HeLa cells expressing high-affinity Zn²⁺ sensor ER-ZapCY1 (n = 6 cells), Golgi-ZapCY1 (n = 4 cells), and ZapCY2 were incubated with 100 μM ZnCl₂. All Zn²⁺ concentrations below 100 μM were buffered solutions. The left axis represents the cytosolic concentrations whereas the right axis represents the FRET changes in the ER and Golgi.

Regulation of ER Zn²⁺ Uptake.

To examine the mechanism of Zn²⁺ uptake into the ER, we explored whether the sarco/endoplasmic-reticulum-Ca²⁺-ATPase (SERCA) was responsible for Zn²⁺ uptake. SERCA is the primary means by which Ca²⁺ is pumped into the ER and Zn²⁺ has been shown to penetrate many Ca²⁺ channels (24), leading us to question whether SERCA could also pass Zn²⁺. Surprisingly, treatment with a SERCA inhibitor (100 nM thapsigargin) did not inhibit Zn²⁺ uptake, but instead enhanced uptake (Fig. 4A). The ER is the main storehouse of Ca²⁺ within the cell and ER Ca²⁺ levels are regulated by SERCA, as well as Ca²⁺ release channels such as the inositol-1,4,5-trisphosphate receptor (IP₃R) and ryanodine receptor (RyR). Inhibition of SERCA causes Ca²⁺ to slowly leak out of the ER, leading to ER Ca²⁺ depletion, elevation of cytosolic Ca²⁺, and activation of store-operated Ca²⁺ channels to promote influx of Ca²⁺ across the plasma membrane. Thus we next explored whether other agents that perturb ER Ca²⁺ impact ER Zn²⁺ uptake. Cells were treated with histamine which leads to ER Ca²⁺ release through the IP₃R, 2-aminoethoxydiphenyl borate (2-APB) an IP₃R inhibitor, and diltiazem, an L-type Ca²⁺ antagonist and inhibitor of RyR. All reagents were used at standard concentrations known to perturb Ca²⁺ homeostasis (25–27). Both agents (thapsigargin and histamine), which induce release of ER Ca²⁺ into the cytosol, led to enhanced Zn²⁺ uptake, whereas 2-APB which inhibits Ca²⁺ release through the IP₃R did not have a statistically significant effect on Zn²⁺ uptake (Fig. 4B). Diltiazem, a RyR inhibitor (27), also gave rise to enhanced Zn²⁺ uptake. At present, we do not have a cohesive explanation for why pharmacological agents with widely different direct impacts on ER Ca²⁺ (two agents promote release, whereas one inhibits release) all enhance Zn²⁺ uptake. However, it is clear that maintenance of ER Ca²⁺ involves a delicately balanced network of processes and perturbation of one component often results in multiple changes in ER Ca²⁺ handling. The pharmacological perturbations presented in this paper do suggest that disruption of the ER Ca²⁺ regulatory network influences Zn²⁺ uptake.

Fig. 4.

Regulation of Zn²⁺ uptake in ER. (A) Thapsigargin enhanced Zn²⁺ uptake in ER. HeLa cells expressing ER-ZapCY1 sensor were treated with DMSO (n = 17 cells, three experiments) or 100 nM thapsigargin (n = 17 cells, three experiments) for 5 min (R₀), and then incubated with 100 μM ZnCl₂ (R). The relative ratio (R/R₀) indicates the changes of Zn²⁺ in ER. (B) Effect of different drugs on Zn²⁺ uptake in ER. HeLa cells expressing ER-ZapCY1 sensors were treated with DMSO (n = 17 cells from three experiments), 100 nM thapsigargin (n = 17 cells, three experiments), 50 μM 2-APB (n = 23 cells, three experiments), 100 μM histamine (n = 18 cells from three experiments), or 50 μM diltizem (n = 17 cells, three experiments) for 5 min followed by addition of 100 μM ZnCl₂ to activate ER Zn²⁺ uptake for 30 min. Compared with DMSO, thapsigargin, histamine, and diltiazem induced greater Zn²⁺ uptake (**p < 0.01, ANOVA, Tukey honestly significant difference post hoc test).

To further characterize the connection between Ca²⁺ dynamics and ER Zn²⁺ homeostasis, we explored whether [Ca²⁺]_cyto impacts [Zn²⁺]_ER. For these experiments, genetically targeted Ca²⁺ sensors D1ER (28) and D3cpv (29) were used to measure [Ca²⁺]_ER and [Ca²⁺]_cyto, respectively. Treatment of HeLa cells with thapsigargin in either the presence or absence of extracellular Ca²⁺ led to Ca²⁺ efflux from ER where reduction of [Ca²⁺]_ER was independent of the extracellular media employed in the experiment (Fig. 5A). Calcium efflux from the ER led to rapid elevation of [Ca²⁺]_cyto and a decrease in [Zn²⁺]_ER (Fig. 5 B and C). ER Ca²⁺ depletion upon thapsigargin treatment activates store-opened Ca²⁺ channels on plasma membrane, which, in the presence of extracellular Ca²⁺, led to sustained elevation of [Ca²⁺]_cyto and continued decrease in [Zn²⁺]_ER (Fig. 5B). On the other hand, in the absence of extracellular Ca²⁺, [Ca²⁺]_cyto returned back to resting levels and coincident with this [Zn²⁺]_ER increased, possibly as a result of influx from the cytosol (Fig. 5C). These perturbations suggest that Ca²⁺ levels in the cytosol influence whether Zn²⁺ is taken up into or released from the ER.

Fig. 5.

Ca²⁺ dynamics affect Zn²⁺ homeostasis in the ER. (A) Effect of thapsigargin on [Ca²⁺]_ER. Treatment with 100 nM thapsigargin, caused [Ca²⁺]_ER to decrease in the presence (n = 6 cells) and absence of (n = 6 cells) extracellular Ca²⁺. (B) Increasing [Ca²⁺]_cyto reduced [Zn²⁺]_ER. Treatment with 100 nM thapsigargin caused [Ca²⁺]_cyto (n = 5 cells) to increase; [Zn²⁺]_ER decreased (n = 6 cells) concomitantly. (C) Thapsigargin increased ER Zn²⁺ when extracellular calcium was absent. [Ca²⁺]_cyto increased (n = 4 cells) and then decreased to resting level upon treatment with 100 nM thapsigargin in calcium free Hepes-buffered Hanks Balanced Salt Solution. [Zn²⁺]_ER slightly decreased (n = 5 cells) when [Ca²⁺]_cyto increased and then increased when [Ca²⁺]_cyto returned to the resting level. (D) Increasing [Ca²⁺]_cyto by activating TrpA1 reduced [Zn²⁺]_ER. HeLa cells expressing TrpA1 were treated with AITC, causing [Ca²⁺]_cyto (n = 4 cells) to increase and [Zn²⁺]_ER (n = 4 cells) to decrease. (E) Effects of extracellular Ca²⁺ on [Zn²⁺]_ER (n = 6 cells). HeLa cells were preincubated with 100 μM ZnCl₂ for 30 min, then increasing extracellular Ca²⁺ reduced [Zn²⁺]_ER. (F) Effects of extracellular Zn²⁺ on [Ca²⁺]_ER (n = 5 cells). HeLa cells were treated with thapsigargin for 30 min to block the SERCA pump. Increasing extracellular Zn²⁺ reduced [Ca²⁺]_ER.

The above experiments all utilized thapsigargin which raises cytosolic Ca²⁺ by inducing release from the ER. To elevate cytosolic Ca²⁺ by an independent means, cells were transfected with the transient receptor potential channel TrpA1. TrpA1 is a cation permeable channel that can be activated by phytochemicals such as mustard oil and cinnemaldehyde (30), and has been shown to transport both Ca²⁺ and Zn²⁺ (31). Treatment of cells expressing TrpA1 with the active ingredient in mustard oil, allyl-isothiocyanate (AITC), led to Ca²⁺ influx across the plasma membrane and concomitant decrease in [Zn²⁺]_ER in accord with the rising [Ca²⁺]_cyto (Fig. 5D). Zinc release required the presence of extracellular Ca²⁺ because it did not occur when extracellular Ca²⁺ was absent (SI Appendix, Fig. S8). Because Zn²⁺ also acts as an activator for TrpA1 (31), addition of 100 μM extracellular Zn²⁺ in the presence of extracellular Ca²⁺ to HeLa cells expressing TrpA1 also increased [Ca²⁺]_cyto and decreased [Zn²⁺]_ER (SI Appendix, Fig. S8). These experiments demonstrate that cytosolic Ca²⁺ fluctuations that arise from ER release or influx across the plasma membrane directly impact ER Zn²⁺ homeostasis.

The connection between Ca²⁺ and Zn²⁺ was further explored by changing extracellular media conditions. HeLa cells were pretreated with 100 μM Zn²⁺ for 30 min to increase the [Zn²⁺]_ER level, and subsequent addition of increasing extracellular Ca²⁺ (10, 50, and 100 mM) caused [Zn²⁺]_ER to decrease in a stepwise fashion (Fig. 5E), further demonstrating that cytosolic Ca²⁺ alters [Zn²⁺]_ER. To explore whether the cytosolic Zn²⁺ could in turn impact [Ca²⁺]_ER, we examined ER Ca²⁺ upon elevation of cytosolic Zn²⁺. HeLa cells were treated with thapsigargin for 30 min to lower ER Ca²⁺ and prevent refilling of the organelle. Addition of 100 μM ZnCl₂ to the extracellular media results in an increase in cytosolic Zn²⁺ to nanomolar levels (Fig. 3C), and this stimulus induced release of Ca²⁺ from the ER (Fig. 5F). These experiments indicate that elevated cytosolic Ca²⁺ can induce ER Zn²⁺ release and, conversely, elevated cytosolic Zn²⁺ can induce ER Ca²⁺ release.

Discussion

With the growing recognition that labile Zn²⁺ plays important roles in influencing signaling pathways and cellular functions, there has been extensive effort over the past few years to develop fluorescent sensors for monitoring Zn²⁺ fluxes in cells (5, 15, 32, 33). Previously we developed two zinc sensors that utilized a single zinc finger from Zif268 as the sensing module (ZifCY1 , ZifCY2 ) (15). A complementary design based on the copper binding proteins Atox1 and WD4 has also been developed (called eCALWY1-6) with ranging from 2 pM to 2.9 nM (5). Although the high-affinity sensor ( 2 pM) was sensitive to Cu¹⁺, lower-affinity versions were specific for Zn²⁺. In this work, we develop ZapCY1, which exhibits high-affinity () and selectivity (Fig. 1C), and a 4.15-fold dynamic range in the cytosol (approximately twofold in the ER and Golgi); but the dissociation kinetics are slow (Fig. 1D). On the other hand, ZapCY2 exhibits faster dissociation kinetics (Fig. 1E), a lower Zn²⁺ affinity (), reduced selectivity, and a smaller dynamic range (approximately 1.4-fold). The fact that these sensors differ by only two amino acids highlights how difficult it is to predict how sensor properties will be altered by changes in the primary sequence.

Numerous studies have reported measurement of cytosolic free Zn²⁺ in a variety of cell types, including red blood cells (34), leukemic cells (35), splenocytes and thymocytes (36), PC-12 (6), HT-29 (22), HEK 293 (5), INS-1 (832/13) (5), hepatocytes (37), and neurons (38), yielding estimates for resting free Zn²⁺ concentrations from 5 to 1,000 pM. In this paper, we report estimates of free Zn²⁺ in ER and Golgi, which are 0.9 and 0.6 pM, respectively, at least 100 times lower than our estimate of the cytosolic Zn²⁺ concentration (80 pM) in HeLa cells (SI Appendix, Table S1).

As suggested by the wide range of reported Zn²⁺ levels, estimates of absolute Zn²⁺ concentration should be regarded with caution. Some discrepancies in Zn²⁺ measurements may reflect true variation in Zn²⁺ levels in different cell types and/or cell state. Indeed in HT29 cells it has been shown that Zn²⁺ levels change with cell state (614 pM in resting vs. 1.25 nM in differentiated cells) (22). However, it is also likely that both the method and nature of the probe influence the absolute estimate. Important considerations include the concentration of the sensor in the cell, potential buffering of endogenous Zn²⁺, reliability of in situ calibration, and the accuracy of intensity-based vs. ratiometric measurements. For example, the concentration of some small molecule Zn²⁺ sensors has been shown to perturb the sensitivity of the sensor, thus influencing calculation of Zn²⁺ concentrations (39). The most reliable estimates will likely result from measurements in which the sensor minimally perturbs the cellular environment, sensor concentration is minimized, and the of the sensor is close to the free Zn²⁺ concentration or, better yet, multiple probes with slightly different values are utilized to constrain the measurements.

In the present study, we explored whether sensor expression level influenced estimates of cytosolic free Zn²⁺ levels by examining the calculated Zn²⁺ concentration as a function of YFP fluorescence intensity. The direct YFP signal does not depend on the presence of Zn²⁺ and hence is an indicator of protein concentration. Over the intensity range examined, we did not see a correlation suggesting that, at the expression levels used in this study, the sensor concentration did not significantly impact our estimates of free Zn²⁺ (SI Appendix, Fig. S6). Still, we acknowledge that each sensor has attributes that limit the accuracy of absolute estimates of Zn²⁺. However, our data clearly provide a reasonable estimate of the relative levels of Zn²⁺ in the cytosol, ER, and Golgi. This relative difference is evident using the same sensor (ZapCY1) targeted to different locations where the sensor is 100% saturated in the cytosol but only partially occupied in the ER and Golgi.

In this work, we sought to characterize the role of the ER and Golgi in transporting cytosolic Zn²⁺. Zn²⁺ uptake into the ER is particularly intriguing because ER function has been shown to depend on Zn²⁺ (17, 18), and the ER plays a central role in integrating cellular signaling events, serving as the hub of Ca²⁺ signaling. Surprisingly we found that Ca²⁺ dynamics influenced ER Zn²⁺ homeostasis and, conversely, that Zn²⁺ could influence ER Ca²⁺ homeostasis. These data suggest a critical link between Ca²⁺ signaling and metal homeostasis. Elevation of cytosolic Ca²⁺ led to ER Zn²⁺ release, whereas elevation of cytosolic Zn²⁺ led to ER Ca²⁺ release, suggesting possible exchange of these two ions across the ER membrane. Given that Ca²⁺ dynamics can be initiated by numerous stimuli and affect a wide range of downstream cellular processes, it will be intriguing to explore whether different Ca²⁺ signaling pathways result in concomitant changes in Zn²⁺ homeostasis.

The electrochemical driving force for movement of ions across the ER membrane can be estimated by considering the difference between the ER membrane potential (V_ER) and the equilibrium potential for a given ion (e.g., E_Zn²⁺). The equilibrium potential exactly balances the chemical driving force caused by a concentration gradient and can be calculated using the Nernst equation:

[1]

With our values of [Zn²⁺]_cyto =  ∼ 80 pM and [Zn²⁺]_ER ∼ 0.9 pM, the E_Zn²⁺ for movement of Zn²⁺ from the ER to the cytosol would correspond to E_Zn²⁺ = 58 mV, whereas movement from the cytosol to the ER would correspond to -58 mV. Currently, there is a dearth of direct experimental evidence for V_ER. However, some studies have used theoretical arguments combined with experimental data on Ca²⁺ fluxes from the ER into the cytosol to estimate V_ER of neurons (-95 mV) and pancreatic acinar cells (-74 mV) (40). If the V_ER of HeLa cells is in a similar range, movement of Zn²⁺ ions from the cytosol into the ER would yield V_ER - E_Zn²⁺ = +74 mV - (-58 mV) = 132 mV. Because the difference in the membrane and equilibrium potential yields a positive value, movement of Zn²⁺ into the ER would be an energetically downhill process (ΔG < 0). However, even if ion movement is energetically favorable, membrane permeability is controlled by channels and pumps, which may need to be activated by a cellular signal (such as increased cytosolic Zn²⁺). Indeed, our experiments demonstrate that elevation of cytosolic Zn²⁺ above 1 nM induces transport of Zn²⁺ into the ER. Treatment with thapsigargin in the absence of extracellular Ca²⁺ led to more rapid Zn²⁺ uptake, perhaps due to alteration of V_ER or enhanced membrane permeability to Zn²⁺(activation of a Zn²⁺ channel). Future studies will be aimed at using the genetically encoded sensors developed in this study to identify the protein(s) responsible for this uptake.

Unexpectedly, elevation of cytosolic Ca²⁺ led to release of Zn²⁺ from the ER. Using estimates similar to above, Zn²⁺ release from the ER into the cytosol would correspond to V_ER - E_Zn²⁺ = -74 mV - (58 mV) = -132 mV, an energetically uphill process. Therefore, it is possible this transport is energy dependent. Although, the membrane potential of the ER under this experimental paradigm is likely to be complicated by Ca²⁺ and counterion flux across the ER membrane and is likely changing over the course of the experiment (40, 41).

In conclusion, we have generated two high-affinity genetically encoded sensors for Zn²⁺ and demonstrated that these sensors enable measurement of steady-state Zn²⁺ levels within the ER and Golgi as well as flux of Zn²⁺ into and out of these organelles. Our study reveals a surprising correlation between Zn²⁺ and Ca²⁺ regulation in the ER that suggests potential exchange of these ions across the ER membrane. We suspect that, as the tools for monitoring cellular metals continue to grow, they will help uncover unique connections between metal ions and cellular signaling pathways.

Methods

SI Appendix, Methods includes the design and in vitro characterization of sensors, colocalization protocol, comparison of in vitro and cellular FRET ratios, and methodology for conversion of FRET ratios into Zn²⁺ concentrations.

Cellular Imaging.

Sensor constructs were transiently transfected into HeLa cells and imaged in phosphate-free Hepes-buffered Hanks Balanced Salt Solution, pH 7.4, 48 h after transfection. Imaging experiments were carried out on an Axiovert 200 M inverted fluorescence microscope (Zeiss) with a Cascade 512B CCD camera (Roper Scientific), equipped with a Xenon arc lamp (XBO75), and data were collected using Metafluor software (Universal Imaging). Details regarding data collection and processing are presented in SI Appendix, Methods.

Statistical Analysis.

Statistical analysis was performed using ANOVA with a post hoc test in KaleidaGraph program. Error bars indicate SEM.