ZIGIR, a Granule-Specific Zn²⁺ Indicator, Reveals Human Islet α Cell Heterogeneity

SUMMARY

Numerous mammalian cells contain abundant Zn²⁺ in their secretory granules, yet available Zn²⁺ sensors lack the desired specificity and sensitivity for imaging granular Zn²⁺. We developed a fluorescent zinc granule indicator, ZIGIR, that possesses numerous desired properties for live cell imaging, including >100-fold fluorescence enhancement, membrane permeability, and selective enrichment to acidic granules. The combined advantages endow ZIGIR with superior sensitivity and specificity for imaging granular Zn²⁺. ZIGIR enables separation of heterogenous β cells based on their insulin content and sorting of mouse islets into pure α cells and β cells. In human islets, ZIGIR facilitates sorting of endocrine cells into highly enriched α cells and β cells, reveals unexpectedly high Zn²⁺ activity in the somatostatin granule of some δ cells, and uncovers variation in the glucagon content among human α cells. We expect broad applications of ZIGIR for studying Zn²⁺ biology and Zn²⁺-rich secretory granules and for engineering β cells with high insulin content for treating diabetes.

Graphical Abstract

In Brief

Ghazvini Zadeh et al. develop a fluorescent zinc indictor, ZIGIR, that labels Zn²⁺-rich secretory granules with high specificity and sensitivity. ZIGIR tracks trafficking and exocytosis of native granules, enables sorting of islet α cells and β cells with high purity, and reveals human α cell heterogeneity and high Zn²⁺ activity in the human somatostatin granule.

INTRODUCTION

Zn²⁺ is an important metal ion that plays numerous roles in biochemistry, cell biology, and animal physiology. Among ~30,000 proteins identified in the human proteome, ~10% of these proteins have been identified as potential zinc (Zn) binding proteins (Andreini et al., 2006). Through coordination with specific amino acids of a polypeptide chain, Zn²⁺ supports the folding, structure, and enzymatic activity of a large array of proteins. The proper regulation and handling of Zn²⁺ activity are vital for maintaining cell function and fitness, and malfunction of Zn²⁺ homeostasis or aberrant Zn²⁺ signaling has been associated with various human diseases (Rink, 2011). Pancreatic islet cells, β cells in particular, contain a high level of intracellular Zn²⁺, a fair portion of which is stored in their secretory granules. ZnT8 (encoded by SLC30A8 gene), a granule-specific Zn²⁺ transporter, is abundantly expressed in pancreatic islet cells and plays a major role in Zn²⁺ uptake into the secretory granule. During stimulated secretion, Zn²⁺ is co-released with other granular content into the extracellular medium (Dodson and Steiner, 1998; Li et al., 2011). Once released, Zn²⁺ can affect the secretory cells from which Zn²⁺ is released or nearby cells through an autocrine or paracrine mechanism, respectively (Bloc et al., 2000; Hardy et al., 2011; Ishihara and Wollheim, 2016; Popovics and Stewart, 2011). Furthermore, the released Zn²⁺ may travel to distant cells through the circulation to modulate the biochemistry of other tissues or organs by acting as an endocrine signal (Tamaki et al., 2013).

The importance of understanding Zn²⁺ regulation and Zn²⁺ signaling in islet cells is highlighted by the association of the SLC30A8 gene with type 2 diabetes (T2D) from genome-wide association studies (GWASs). These studies have uncovered that specific single-nucleotide polymorphisms (SNPs) of the SLC30A8 gene can either increase or reduce the risk of T2D (Davidson et al., 2014; Rutter and Chimienti, 2015). Haploinsufficiency of the SLC30A8 gene can have a strong protective effect (odds ratio 0.4 for p.Arg138* carriers), reducing T2D risk in humans (Dwivedi et al., 2019; Flannick et al., 2014). These findings raise the interesting possibility of targeting Zn²⁺ transporting pathways in islet cells as a potential therapeutic strategy for treating diabetes.

To track cellular Zn²⁺ levels and to investigate Zn²⁺ regulation at specific cellular compartments, fluorescent Zn²⁺ indicators are invaluable tools: they enable imaging of Zn²⁺ dynamics because of their high sensitivity and compatibility with live cell imaging (Chen et al., 2015; Hessels and Merkx, 2015; Li, 2015). It remains challenging to track Zn²⁺ activity (labile or readily exchangeable Zn²⁺) in cells with high specificity and sensitivity. A few fluorescent Zn²⁺ sensors, including Zinquin and Newport green (NPG) PDX, have been reported for imaging granular Zn²⁺ (Lukowiak et al., 2001; Zalewski et al., 1994). However, these sensors are limited by their non-specific cellular distribution, pH sensitivity, and in the case of Zinquin, requirement for UV excitation. Furthermore, quinoline-based Zn²⁺ sensors, including Zinquin and TSQ (6-methoxy-8-p-toluenesulfonamido-quinoline) are known to bind to Zn proteins by forming the ternary complex comprising of sensor-Zn²⁺ protein (Meeusen et al., 2011). This complicates the interpretation of the source of the observed fluorescence signal. More recent fluorescent Zn²⁺ sensors that have been applied to imaging of granular or vesicular Zn²⁺ include FluoZin-3/AM (Gee et al., 2002), ZP4 (Burdette et al., 2003), ZincBY-1 (Que et al., 2015), and SpiroZin2 (Rivera-Fuentes et al., 2015). Because these Zn²⁺ sensors bind Zn²⁺ with nanomolar (nM) affinity, they are not suitable for distinguishing Zn²⁺ activities between subcellular compartments containing abundant Zn²⁺ (≥100 nM). Most cellular organelles, including endoplasmic reticulum (ER), mitochondria, Golgi, and cytosol, have Zn²⁺ activity in the sub-nM range (Kambe et al., 2015). In contrast, in the insulin granule of islet β cells, six insulin molecules coordinate with two Zn ions to form the insulin₆-Zn₂ complex (Dodson and Steiner, 1998; Emdin et al., 1980). Because the granular insulin content is more than 70 mM (Huang et al., 1995; Matthews et al., 1982), total Zn²⁺ in the insulin granule may reach or even exceed 20 mM. The exact free-Zn²⁺ activity in the insulin granule is not known but was thought to be above 1 μM (Chabosseau and Rutter, 2016; Kambe et al., 2015). Other limitations of previously reported granular Zn²⁺ sensors, besides the undesirable high Zn²⁺ affinity (in nM), include modest Zn²⁺ responsivity (~5-fold enhancement) and promiscuous cellular distribution. FluoZin-3, for example, has been commonly used for imaging cytosolic Zn²⁺, yet several studies have documented the localization of FluoZin-3 to other cellular compartments, including vesicles (McCormick et al., 2010; Wellenreuther et al., 2009), lysosome (Aydemir et al., 2009; Hwang et al., 2008; Kaltenberg et al., 2010; Roh et al., 2012), and Golgi (Qin et al., 2013). To overcome these limitations, we have developed a fluorescent Zn granule indicator, ZIGIR. ZIGIR binds Zn²⁺ with submicromolar affinity and displays more than 100-fold fluorescence enhancement upon Zn²⁺ complexation. In cells, ZIGIR is found to accumulate in acidic granules, including secretory granules, and importantly, is refractory to cellular pH fluctuations to exhibit robust Zn²⁺ responsivity at both neutral and acidic pH. The pH resistance makes ZIGIR ideally suitable for imaging Zn²⁺ activity in acidic compartments such as secretory granules. Because of its low Zn²⁺ affinity (K_D(Zn²⁺) = 0.4 μM), ZIGIR fluorescence is only observed in Zn²⁺-rich secretory granules, not in other cellular compartments. Besides fluorescence microscopy, ZIGIR is compatible with flow cytometry and fluorescence-activated cell sorting (FACS) to enable the separation of islet α cells, β cells, and δ cells of both mouse and human islets. Flow cytometry analysis of ZIGIR-labeled human islet cells reveals the hitherto unknown high Zn²⁺ activity in the secretory granule of human δ cells and the heterogeneity of human α cells that vary in glucagon content over a broad range. Moreover, we demonstrate that in living β cells, ZIGIR labeling can serve as a surrogate marker of insulin granule content to enable the separation of β cell subsets with distinct insulin content by FACS. This advancement should facilitate future engineering of β cells with high insulin content for β cell replacement therapy for type 1 diabetes.

RESULTS

Design, Synthesis, and In Vitro Characterization of ZIGIR

The lumen of secretory granules is normally acidic with an intraluminal pH of 5–6 (De Young et al., 1987; Stiernet et al., 2006). To impart pH resistance to the granular Zn²⁺ sensor, we chose carboxyrhodamine as the fluorophore. Fluorescence of carboxyrhodamine is insensitive to pH changes between 4 and 9. Moreover, compared with other commonly used fluorophores, including fluorescein, rhodamine dyes are more photostable (Beija et al., 2009). In ZIGIR, a Zn²⁺ binding motif consisting of a 2-pyridylmethyl-[2-(2-pyridyl)ethyl]amine is linked to carboxyrhodamine through its 5-amino substituent (Figure 1A). Because Zn²⁺-rich granules are thought to have Zn²⁺ activity in the micromolar range, and because Zn²⁺ levels among other cellular compartments, including the cytosol, nucleus, ER, and mitochondrion, are at least three orders of magnitude lower (Kambe et al., 2015), we reasoned that Zn²⁺ sensors with approximately micromolar affinity would be appropriate for sensing Zn²⁺ activity in the Zn²⁺-rich granules.

Figure 1. Design and Characterization of ZIGIR.

(A) Chemical structure of ZIGIR and its mode of action for sensing Zn²⁺.

(B) Absorption spectra of ZIGIR in the presence or absence of Zn²⁺.

(C) Fluorescence emission spectra of ZIGIR (Exitation = 545 nm) at increasing Zn²⁺ concentrations: 0.63, 40, 160, 400, 630, and 6,300 nM (from bottom to top).

(D) Zn²⁺ titration of ZIGIR as measured from its emission at 572 nm. The solid line represents the least-squares exponential fit.

(E) ZIGIR is refractory to physiological pH fluctuation and maintains its Zn²⁺ responsivity between pH 5 and pH 9. ZIGIR emission was measured in either nominally Zn²⁺-free solutions or in 10 μM Zn²⁺ solutions. The insert shows the effect of pH on ZIGIR fluorescence from pH 3.5 to pH 9.3 in either the presence (filled square) or the absence of Zn²⁺ (open triangle).

(F) Summary of photophysical properties and Zn²⁺ responses. ϕ_fl is the fluorescence quantum yield, ε is the extinction coefficient, N_h is the Hill coefficient, and FC is the fold change in fluorescence brightness (ϕ_fl × ε) from the Zn²⁺-free to the Zn²⁺-bound state.

We synthesized ZIGIR from common starting materials in five steps (Figure S1A). ZIGIR absorbed maximally near 550 nm, with a minimal bathochromic shift (3 nm) in its absorption maximum going from the Zn²⁺-free to the Zn²⁺-bound state (Figure 1B). In contrast, ZIGIR fluorescence intensity increased more than 100-fold upon binding to Zn²⁺, reaching a fluorescence quantum yield of 23% (Figures 1C and 1F). Zn²⁺ titration of ZIGIR yielded a K_D(Zn²⁺) of 0.4 μM and a Hill coefficient of 1.03, confirming 1:1 stoichiometry of Zn²⁺ binding (Figure 1D). To assess the sensitivity of ZIGIR to physiological pH changes, we compared its Zn²⁺ responsivity at two pH values: pH 7.5 and pH 5.5. These pH values were chosen to mimic the neutral environment in the cytosol and the acidic milieu of the secretory granule, respectively. At both pH values, ZIGIR displayed about the same fluorescence signal as long as the Zn²⁺ level was kept the same (Figure 1E). Expanding the pH titration from pH 3.5 to pH 9.3 revealed that fluorescence of ZIGIR was largely refractory to pH changes from pH 5 to pH 9.3 (Figure 1E, insert).

The fluorescence response of ZIGIR was selective for Zn²⁺ against other ions, including Na⁺, K⁺, Ca²⁺, Mg²⁺, Fe²⁺, Ni²⁺, Co²⁺, Cu²⁺, and Mn²⁺ (Figure S1B). Among the metals tested, Cd²⁺ was the only other ion besides Zn²⁺ that enhanced ZIGIR intensity appreciably. Because the cellular Cd²⁺ content is several orders of magnitude lower than the cellular Zn²⁺ content (Wong et al., 2017), Cd²⁺ is not expected to interfere with Zn²⁺ sensing by ZIGIR in cells.

There are some structural similarities between ZIGIR and the previously reported ZnAF family of Zn²⁺ sensors (Komatsu et al., 2005). ZnAF3, for example, contains the same Zn²⁺ binding motif as ZIGIR, and it binds Zn²⁺ with a K_D(Zn²⁺) of 0.79 μM (Figure S1C). However, several important differences exist between these two classes of Zn²⁺ sensors, and these differences endow ZIGIR with crucial advantages for imaging Zn²⁺ activity in acidic secretory granules. First, ZnAF3 is built on fluorescein dye whose fluorescence intensity is highly sensitive to physiological pH fluctuations. At pH 6, fluorescence intensity of ZnAF3 is only ~20% of that at neutral pH, whereas at pH 5, ZnAF3 is practically non-fluorescent (Komatsu et al., 2005). In contrast, ZIGIR maintains fairly high fluorescence intensities from pH 5 to pH 9 (Figure 1E). This puts ZIGIR in a better position for imaging Zn²⁺ of acidic granules. Second, ZIGIR dominates in fluorescence enhancement upon Zn²⁺ binding by an order of magnitude over ZnAF3 (104-fold versus 11-fold enhancement) (Figure S1C). This translates to a larger dynamic range and a lower background signal for live cell imaging. Third, ZIGIR is cell membrane permeable and can enter cells and subcellular compartments by passive diffusion (vide infra). However, ZnAF sensors are membrane impermeable, so they need to be converted into acetyl esters for cell loading and live cell imaging (Komatsu et al., 2005). Once inside cells, the acetyl ester is rapidly hydrolyzed by cellular esterases to regenerate ZnAF dyes, which are trapped in the cytosol. This ester hydrolysis in the cytosol may deter further diffusion of ZnAF dyes across intracellular membranes and limit ZnAFs from entering membrane-bound organelles, including the secretory granule (vide infra).

ZIGIR Fluorescence Imaging and FACS of Cultured Cells

To assess ZIGIR’s ability for imaging granular Zn²⁺ in living cells, we loaded a MIN6 β cell with ZIGIR (1 0μM for 15 min) and followed its cellular uptake by time-lapse imaging. ZIGIR rapidly entered cells and accumulated in the granule-like compartments (Video S1). After dye loading, cells were washed once and examined by confocal fluorescence microscopy. ZIGIR imaging revealed numerous fluorescent puncta throughout the cytoplasm (Figure 2A), consistent with the known abundance of Zn²⁺-rich insulin granules in this cell line (Miyazaki et al., 1990). To confirm the specificity of ZIGIR labeling to the insulin granule, we subsequently fixed and permeabilized the cells and stained them with an insulin antibody. At the cellular level, there was a good correlation between ZIGIR fluorescence and insulin immunofluorescence (Figures 2A and 2B), consistent with the selective labeling of the insulin granule by ZIGIR. In addition, ZIGIR staining showed little overlap with cell organelles involved in secretion or membrane cycling, including the late endosome/lysosome (marked by the LAMP2 protein) (Carlsson et al., 1988) (Figure 2C) and Golgi (marked by the GM130 protein) (Nakamura et al., 1995) (Figure 2D). We further characterized ZIGIR-positive puncta by quantifying their sizes in MIN6 cells (Figure 2E). The histogram of the measured diameters showed a peak value at 200 nm (Figure 2F), with an average diameter of 270 ± 100 nm (mean ± SD, >250 puncta analyzed). This result is remarkably consistent with the measurement of insulin granule size by transmission electron microscopy using the technique of high-pressure freezing, which yielded a mean diameter of 243 ± 73 nm for the rat insulin granule (Fava et al., 2012). The agreement on granule size supports ZIGIR labeling specificity for the insulin secretory granule. To examine Zn²⁺ responsivity of ZIGIR in live cells, we raised the cellular Zn²⁺ level by adding a Zn²⁺ ionophore (pyrithione, 10 μM) and high Zn²⁺ (20 μM). Zn/pyrithione was expected to raise Zn²⁺ levels throughout cells (Li, 2015; Zalewski et al., 1991), yet we found that the granular ZIGIR signal still dominated cellular fluorescence after elevating the Zn²⁺ concentration globally with Zn/pyrithione (Figure 2G). The average granular ZIGIR intensity was more than 4,600 times or 350 times higher than the cytoplasmic signal in the basal medium or in Zn/pyrithione, respectively (Figure 2H). These results support highly selective enrichment of ZIGIR in the granular compartments (vide infra). Besides responding to Zn²⁺ elevation, ZIGIR sensed changes in Zn²⁺ drop. Treating cells with a membrane-permeable Zn²⁺ chelator, TPEN, reduced the cellular free-Zn²⁺ concentration. Consistently, TPEN dose-dependently dampened ZIGIR intensity, and it decreased cellular ZIGIR fluorescence by more than 70-fold at 100 μM (Figures 2I and 2J). Altogether, the data showed that ZIGIR is membrane permeable, accumulates in the insulin granule once inside cells, and maintains good Zn²⁺ responsivity to report Zn²⁺ fluctuations in living cells.

Figure 2. ZIGIR Selectively Labels the Insulin Granule and Maintains Zn²⁺ Responsivity in Living β Cells.

(A-D) Confocal imaging of MIN6 cells labeled with ZIGIR (live cells) and subsequently fixed and permeabilized for immunofluorescence using antibodies against marker proteins of cellular organelles. The scatterplot (B) of the cellular ZIGIR intensity and the corresponding insulin immunofluorescence showed a Pearson’s R value of 0.80 ± 0.06 (mean ± SD, n = 116 cells). Nuclei are shown in blue with Hoechst 33342.

(E and F) Measurement of granule size by ZIGIR imaging. Confocal images of ZIGIR-stained MIN6 cells (E) were analyzed by ImageJ (analyze particles plugin). ZIGIR-positive puncta from 0.01 to 0.2 μm² (particle diameter of 100–500 nm) were analyzed to yield the granule diameter distribution (F) with a mean diameter of 270 ± 100 nm (SD, 252 granules).

(G and H) ZIGIR was selectively enriched in the granule and responded to Zn²⁺ elevation. Confocal images of MIN6 cells labeled with ZIGIR in the basal secretion assay buffer (SAB) or after adding Zn/pyrithione (20/10 μM). Quantification of the granular and bulk cytoplasm ZIGIR signal is shown in (H) (mean ± SEM, >600 granules analyzed for each condition; the ZIGIR intensity ratio between granule and cytosol is shown above the bar).

(I and J) ZIGIR signal decreased with reducing cellular Zn²⁺ concentration. Confocal images of MIN6 cells labeled with ZIGIR in SAB or after adding TPEN (100 μM). Quantification of the ZIGIR signal at different TPEN concentrations (J; mean ± SEM, >200 granules analyzed for each condition). TPEN, at 25, 50, or 100 μM, dampened the ZIGIR signal by 11-, 37-, or 76-fold, respectively. Scale bar, 5 μm. ZIGIR was loaded to cells at 1 μM for 15 min at 37°C for all experiments.

In contrast with ZIGIR, several Zn²⁺ sensors that were reported for monitoring granular Zn²⁺, including FluoZin-3, NPG, TSQ, and ZP4, exhibited either promiscuous or cytoplasmic labeling in MIN6 cells (Figures 3A–3C; Figure S2). FluoZin-3, for example, has been used as a granule Zn²⁺ sensor (McCormick et al., 2010; Wellenreuther et al., 2009), yet in MIN6 cells, FluoZin-3 labeling showed a non-specific cellular distribution with both cytoplasmic and punctate appearance that poorly overlapped with the ZIGIR signal (Figures 3A and 3B). Moreover, raising the cellular Zn²⁺ concentration with Zn²⁺/pyrithione selectively increased the ZIGIR signal only in the granular compartments, but it elevated the FluoZin-3 signal drastically throughout entire cells (Figure 3C; Video S2), confirming the promiscuous cellular distribution of FluoZin-3 as reported by others (Aydemir et al., 2009; Hwang et al., 2008; Kaltenberg et al., 2010; McCormick et al., 2010; Qin et al., 2013; Roh et al., 2012; Wellenreuther et al., 2009). We also compared ZIGIR with its structural analog ZnAF3, which possesses the same Zn²⁺ binding pocket and micromolar Zn²⁺ binding affinity as ZIGIR (Figure S1C). Despite the structural similarity, ZnAF3 displayed uniform cytoplasmic distribution both in the basal medium and in the high Zn²⁺ solution (Zn²⁺/pyrithione) (Figures 3D–3F). We attributed the difference in their cellular distribution to several factors, including pH sensitivity, membrane permeability, and granule uptake, as mentioned earlier. ZnAF3 binds Zn²⁺ with lower affinity than ZIGIR (K_D(Zn²⁺) 0.79 μM for ZnAF3 versus 0.4 μM for ZIGIR) (Figure S1C), yet only ZnAF3 lit up throughout cells after adding Zn²⁺/pyrithione, whereas the ZIGIR signal remained restricted to the granule. Had ZIGIR been present in the cytosol, it would have lit up everywhere just like ZnAF3. The ZIGIR, but not ZnAF3, was found exclusively in granules both before and after adding Zn/pyrithione, which demonstrated the selective accumulation of ZIGIR in the acidic granules. Our conclusion of selective accumulation of ZIGIR in the acidic granules was solely based on the fluorescence readout of ZIGIR. More direct confirmation of this result would require measurement of the amount of ZIGIR (by quantitative mass spectrometry, for example) in the granule and in the cytosol after cellular fractionation.

Figure 3. A Unique Combination of Properties Endows ZIGIR with Unprecedented Specificity for Labeling Zn²⁺-Rich Secretory Granules.

(A-F) Only ZIGIR, not other Zn²⁺ sensors, selectively labels insulin granules in β cells. MIN6 cells were co-loaded with ZIGIR (1 μM) and FluoZin-3/AM (5 μM, A-C) or with ZIGIR and ZnAF3-Ac (2 μM, D-F) and imaged by confocal microscopy first in the basal SAB solution and subsequently in a high Zn²⁺ solution (Zn/pyrithione, 20/10 μM). Cell nuclei are shown in blue with Hoechst 33342. Scale bar, 5 μm.

(G-L) ZIGIR is acidotropic and accumulates in acidic granules. Confocal images of H1299 cells (G-I) or U2OS cells (J-L) labeled with ZIGIR in the basal SAB (H and K) or after adding Zn/pyrithione (I and L). LTG (0.4 μM) was added to H1299 cells in the high Zn²⁺ buffer (I, middle image). After ZIGIR imaging, U2OS cells were fixed and stained with antibodies against LAMP2 and GM130 (L, middle image). Scale bar, 10 μm.

To better understand the granular labeling specificity of ZIGIR, we applied ZIGIR to mammalian cells that are not known to contain Zn²⁺-rich secretory granules. In the human lung carcinoma cell line H1299, we could barely detect ZIGIR fluorescence under the same dye loading and imaging conditions used for MIN6 cells (Figures 3G and 3H). After adding Zn/pyrithione to raise the cellular Zn²⁺ concentration, distinct ZIGIR-positive spots emerged in these cells (Figure 3I). To interrogate the cellular identity of these puncta, we subsequently added LysoTracker green (LTG), an acidotropic fluorescent tracer for labeling acidic granules. Imaging of LTG in the green channel revealed numerous intensely fluorescent dots in H1299 cells. The LTG-marked granules overlapped well with ZIGIR-positive puncta (Figure 3I), suggesting that ZIGIR was enriched in the acidic granules. Consistent with the observation in H1299 cells, in U2OS human bone osteosarcoma epithelial cells, we observed little cellular ZIGIR signal in the basal medium (Figures 3J and 3K) but distinct ZIGIR-positive dots after raising the Zn²⁺ concentration by adding Zn/pyrithione (Figure 3L). To confirm the identity of these ZIGIR-positive puncta, we fixed cells and stained them with antibodies against LAMP2 (lysosome) and GM130 (Golgi). The ZIGIR-positive dots showed striking overlap with LAMP2 but little overlap with GM130 (Figure 3L), demonstrating that ZIGIR selectively targeted the lysosome.

Similar to LTG, ZIGIR is neutral overall and contains a weakly basic tertiary amine (Figure 1). Small neutral molecules containing a weakly basic amine easily diffuse across the cell membrane and tend to accumulate in cellular compartments with low internal pH (Anderson and Orci, 1988). Because of its low Zn²⁺ affinity (K_D (Zn²⁺) = 0.4 μM), ZIGIR, once taken up by the acidic granules, remains weakly or non-fluorescent unless Zn²⁺ activity far exceeds nMs. Because most cellular compartments have Zn²⁺ activities in the low-nM or sub-nM range (Kambe et al., 2015), ZIGIR taken up by acidic granules remained undetected (with the exception of Zn²⁺-rich secretory granules) until we artificially inflated cellular Zn²⁺ activity with Zn/pyrithione. Hence, a unique combination of several desirable properties endowed ZIGIR with the unprecedented specificity and sensitivity for imaging Zn²⁺-rich secretory granules in living cells. These properties include membrane permeability, acidotropic propensity, approximately micromolar Zn²⁺ affinity, pH insensitivity, and >100-fold Zn²⁺-dependent fluorescence enhancement.

To examine the photostability of ZIGIR in cells, we performed time-lapse confocal imaging of MIN6 cells. Over the course of fluorescence imaging, the intensity of ZIGIR-labeled granules remained brightly fluorescent (Figures S3A and S3B; Video S3), demonstrating its satisfactory photostability for live cell imaging. Cells loaded with ZIGIR showed no sign of cytotoxicity, and they maintained the same viability and growth rate as the control unlabeled cells (Figures S3C and S3D). Moreover, labeling islet cells with ZIGIR did not affect the secretory function, and mouse islets stained with ZIGIR responded to glucose and released insulin, as similarly seen in the control islets (Figure S3E). To track individual insulin granules with high spatial resolution, we resorted to total internal reflection fluorescence (TIRF) microscopy. TIRF imaging of ZIGIR-labeled INS-1 β cells revealed numerous granules near the plasma membrane (Video S4). These insulin granules displayed distinct ZIGIR fluorescence well above the background signal, and they exhibited diverse dynamic behaviors along the plasma membrane or moved into or out of the evanescent illumination field. KCl (40 mM) stimulation depolarized the cell membrane to induce Ca²⁺ entry and insulin release. During exocytosis, a ZIGIR-labeled granule fused with the plasma membrane and released its ZIGIR as a cloud dissipation into the medium before it disappeared (Figure 4A; Video S5).

Figure 4. ZIGIR Is a Versatile Probe for Imaging Granule Movement and Exocytosis and for Sorting β Cells Based on Their Insulin Content.

(A) Tracking insulin granule exocytosis by TIRF imaging and ZIGIR labeling. Consecutive TIRF images of INS-1 β cells post-KCl stimulation. Images were taken at 1 frame/s. A granule (highlighted by arrows) undergoing exocytosis released ZIGIR into the medium as a cloud (frame 4) and disappeared afterward (cf. Video S5). Scale bar, 1.5 μm.

(B-F) ZIGIR as a surrogate marker of insulin granule content for sorting β cells. ZIGIR-high and ZIGIR-low MIN6 β cells were isolated by FACS from the ZIGIR histogram (B) and assayed for their insulin content (C, mean ± SEM of 3 replicates, *p < 0.05). (D-F) Quantification of immunofluorescence of NPC-1 (lysosome), GM130 (Golgi), or insulin of sorted ZIGIR-high or ZIGIR-low MIN6 cells (D, mean ± SEM of 50 cells, ****p < 0.001). Representative confocal immunofluorescence images of sorted MIN6 cells stained with GM130 and insulin (E) or with NPC-1 and insulin (F). Nuclei are marked by DAPI in blue. Scale bar, 10 μm.

ZIGIR imaging of MIN6 cells revealed cell-to-cell variations in the abundance of ZIGIR-positive puncta among individual cells (Figures 2A and 2C; Figure S2). The MIN6 β cell is a transformed β cell line derived from a mouse insulinoma. This cell line was reported to consist of heterogenous β cells differing in their insulin level and glucose response (Minami et al., 2000; Yamato et al., 2013). Because ZIGIR is an intensity-based fluorescent indicator, its cellular fluorescence signal can be affected by both granule Zn²⁺ activity and dye loading. Because ZIGIR accumulates in the acidic secretory granule, its cellular uptake is controlled by the total number of secretory granules or the hormone content (Figures 2A and 2B). We thus explored the possibility of using ZIGIR as a surrogate marker of insulin content for sorting β cells that differed in insulin level by flow cytometry. Consistent with fluorescence microscopy, flow cytometry analysis of the ZIGIR signal revealed a wide spread of ZIGIR intensity among MIN6 cells (Figure 4B). We separated the cells into ZIGIR-high and ZIGIR-low subsets by FACS and compared their insulin content. The results showed that the ZIGIR-high subset contained appreciably more insulin than the ZIGIR-low cells (Figure 4C). Analysis of sorted cells by immunofluorescence confirmed a higher insulin level in ZIGIR-high cells yet found no difference in the amount of lysosome or Golgi between ZIGIR-high and ZIGIR-low MIN6 cells (Figures 4D–4F). The results validated the specificity of ZIGIR staining as a surrogate marker of the hormone content of Zn²⁺-rich secretory granules such as the insulin granule and demonstrated the versatility of ZIGIR as a label for both fluorescence microscopy and flow cytometry.

Analyzing Granular Zn²⁺ Activity and Sorting Mouse Islet Endocrine Cells

Among islet endocrine cells, β cells have long been known to contain a high level of total Zn²⁺ in their secretory granules. However, it remains unclear how Zn²⁺ activities differ among different types of secretory granules, including the glucagon granule and the somatostatin granule. To address the question, we combined ZIGIR and flow cytometry to analyze the granular Zn²⁺ activity in primary mouse islet cells. To facilitate distinguishing among endocrine cells, we labeled mouse islets with a β cell marker, a Cy5 dye conjugate of the Exendin-4 peptide (Ex4) (Figure S4). Ex4 is a high-affinity ligand of the glucagon-like peptide 1 receptor (GLP-1R). Because GLP-1R is highly expressed in mouse islet β cells (Tornehave et al., 2008), fluorescently labeled conjugates of Ex4 are rapidly internalized into β cells through receptor-mediated endocytosis to mark β cells (Kim et al., 2017; Li et al., 2015a). After labeling mouse islets with Ex4-Cy5, we dispersed the islets into single cells and labeled them with ZIGIR (Figure 5A). Flow cytometry analysis of the labeled islet cells revealed four distinct subsets of cells, P1–P4, on the two-dimensional scatterplot (Figure 5B). To identify each subset of cells, we isolated them by FACS and analyzed the sorted cells by immunofluorescence using antibodies against insulin, glucagon, and somatostatin. The immunofluorescence results confirmed that the P1 and P2 subsets were essentially pure β cells (99% ± 0.3%, n = 3) and α cells (98% ± 0.5%, n = 3), respectively (Figure 5C). The P1 subset showed the highest level of Ex4-Cy5 and the ZIGIR signal, consistent with the abundant GLP-1R expression on the β cell and high Zn²⁺ content in the insulin granule. Ex4-Cy5 uptake in the P2 subset (α cells) was about 1% of the level seen in β cells, which could be accounted for by the low yet detectable GLP-1R expression in some mouse α cells (De Marinis et al., 2010). The P3 subset was highly enriched with δ cells (81% ± 4.8%, n = 3) and exhibited a ZIGIR signal lower than that of both P1 and P2. The remaining ~20% of cells in P3 and most cells in P4 were not stained by any of the three hormone antibodies. These cells likely represented a rare PP cell, islet endothelial cells, lymphoid cells, or exocrine cells.

Figure 5. Flow Cytometry Analysis of Granular Zn²⁺ Activity and Sorting of Mouse Islet Cells with ZIGIR.

(A) Workflow of cell labeling, FACS, and post-sorting analysis.

(B) Flow cytometry histogram of ZIGIR (top) and the corresponding 2D scatterplot (bottom) of mouse islet cells labeled with ZIGIR and Ex4-Cy5.

(C) Confocal immunofluorescence images (left column) of sorted islet cells using antibodies against three islet hormones. Cell-type distributions in each subset of sorted cells are shown to the right (mean ± SEM for 3 replicates; >200 cells were analyzed for P1 or P2 and >60 cells were analyzed for P3 or P4 in each replicate). Us, cells unstained by any of the three hormone antibodies.

(D) Confocal immunofluorescence images of a mouse pancreas section stained with antibodies against islet hormones and ZnT8. The enlarged images of the area highlighted by the dashed box are shown in the bottom row, with ZnT8 pseudo-colored in red and individual hormones in green to aid visualization of expression overlap.

(E) Only ZIGIR, not other Zn²⁺ sensors, could resolve distinct islet endocrine cells according to their granular Zn²⁺ levels. Flow cytometry histograms of mouse islet cells labeled with ZIGIR and three other fluorescent Zn²⁺ sensors.

Also see Figures S5B–S5E. Scale bar, 20 μm.

To our knowledge, this flow cytometry analysis of the cellular ZIGIR signal represented the only systematic effort to compare Zn²⁺ activity among three types of secretory granules of islet cells. The results showed that the median fluorescence intensity (MFI) of ZIGIR in α cells was nearly an order of magnitude stronger than that of δ cells but was about 5 times weaker than that of β cells (Figure 5B). To confirm that variations in ZIGIR intensities among three endocrine cells reflected differences in granule Zn²⁺ activity instead of dye loading, we treated ZIGIR-labeled islet cells with TPEN to dampen the cellular Zn²⁺ activity. TPEN treatment effectively reduced the ZIGIR signal of both β cells and α cells to about the same level as that of δ cells, such that the ZIGIR fluorescence intensity of TPEN-treated islet cells collapsed into one peak, as shown by flow cytometry (Figure S5A). The measurements confirmed the very high Zn²⁺ activity in the insulin granule and suggested an appreciable amount of free Zn²⁺ in the glucagon granule that was substantially higher than that of the somatostatin granule. The relatively high Zn²⁺ concentration in the α cell granule resonates with a previous study that used X-ray microanalysis to suggest that some granules in rodent α cells contain a high level of Zn²⁺ (Foster et al., 1993). Our results of granular Zn²⁺ levels are also in line with the pattern of ZnT8 expression in mouse islet cells. ZnT8 is highly expressed in pancreatic islets and is the major transporter responsible for importing Zn²⁺ into the dense core granules (Davidson et al., 2014). Consistent with the earlier reports (Artner et al., 2010; Murgia et al., 2009; Solomou et al., 2015), immunofluorescence of ZnT8 in mouse pancreatic sections confirmed ZnT8 expression in both α cells and β cells, but not δ cells (Figure 5D). This provided a molecular basis to account for the low level of labile Zn²⁺ in the somatostatin granule as measured by ZIGIR. Besides ZnT8 expression, another factor that could contribute to the very high concentration of labile Zn²⁺ in the insulin granule is the Zn²⁺ chelating property of insulin (Dodson and Steiner, 1998). The high abundance of insulin in the secretory granule of the β cell effectively turns the lumen of the insulin granule into a high-capacity Zn²⁺ sponge.

Fluorescent Zn²⁺ indicators that have been previously reported for imaging granular Zn²⁺ activity include FluoZin-3/AM (Gee et al., 2002), NPG (Lukowiak et al., 2001), ZP4 (Burdette et al., 2003; Solomou et al., 2015), ZincBY-1 (Que et al., 2015), and SpiroZin2 (Rivera-Fuentes et al., 2015). We compared ZIGIR with several commercially available Zn²⁺ sensors using flow cytometry in mouse islet cells. Only ZIGIR, not FluoZin-3, NPG, or ZP4, was able to resolve islet cells into distinct subsets according to their granular Zn²⁺ activity (Figure 5E; Figures S5B–S5E), demonstrating the superior sensitivity, specificity, and dynamic range of ZIGIR as a granular Zn²⁺ probe.

Analyzing and Sorting Human Islet Endocrine Cells with ZIGIR

Thus far, most studies on ZnT8 and Zn²⁺ signaling have been carried out in mouse islet cells, and less is known about the expression of ZnT8 and Zn²⁺ distribution in human islet endocrine cells. To analyze granular Zn²⁺ activity in human islets by flow cytometry, we dispersed human islets into single cells and labeled them sequentially with ZIGIR and cell surface markers. In contrast with mouse islet cells, we found that labeling of human islet cells with Ex4-Cy5 was rather inefficient (Figure S6). We therefore applied antibodies reactive toward human pancreatic endocrine cells (Figure 6A), including the HPi2 antibody (HIC1–2B4) (Dorrell et al., 2008) and an antibody of a tetraspanin protein, TM4SF4 (Muraro et al., 2016). The islet endocrine cells (HPi2+) (Figure S7A) contained a subset of cells showing a high ZIGIR signal but low TM4SF4 expression (P1 subset, Figure 6B). Immunofluorescence analysis of the sorted cells confirmed that the P1 subset was highly enriched with human β cells, together with a small percentage of δ cells (Figure 6C). Most remaining cells displayed high TM4SF4 expression yet a wide spread of the ZIGIR signal. We divided them into two subsets, P2a or P2b, that showed a high or low ZIGIR signal, respectively (Figure 6B). Both P2a and P2b subsets consisted of α cells of high purity (≥95%), consistent with the reported abundant expression of TM4SF4 in human α cells (Muraro et al., 2016). We obtained similar flow cytometry results from different human donors and confirmed high enrichment of β cells or α cells in the P1 or P2 subset (including P2a and P2b), respectively (Figures S7B–S7E). The distinct ZIGIR signals of P2a and P2b α cells suggested varied glucagon granule abundances, different labile Zn²⁺ activities in the glucagon granule, or both. In addition, the sigmoid Zn²⁺ response of ZIGIR (cf. Figure 1D) could enhance the spread the ZIGIR staining intensity within a relatively narrow range of Zn²⁺ concentrations. To examine these possibilities, we quantified glucagon immunofluorescence intensities of these two subsets of α cells and found that the glucagon signal of P2a was at least four times as high as that of P2b α cells (Figures 6D and 6E; Figure S7B), confirming substantially higher glucagon content in P2a cells that corroborated their stronger ZIGIR staining. In addition to containing more glucagon granules, P2a cells might express more ZnT8 protein than P2b cells. Future studies using more quantitative techniques such as western blot should help address the issue. Regardless of the exact mechanism underlying distribution of ZIGIR labeling of human α cells, we observed this phenomenon repeatedly in isolated islets from several human donors (Figure 6; Figures S7B–S7E), suggesting a general phenomenon of human α cell heterogeneity defined by their distinct glucagon content.

Figure 6. Flow Cytometry Analysis and Sorting of Human Islet Cells with ZIGIR.

(A) Workflow of human islet labeling and analysis.

(B) 2D scatterplot of human islet endocrine cells (donor SAMN10737781) by ZIGIR and TM4SF4 labeling. Islet endocrine cells were sorted into three subsets.

(C) Cell composition of the sorted P1, P2a, and P2b subsets analyzed by immunofluorescence. Cells that were negatively stained for all three hormones (Ins, Gcg, and Sst) were designated as unstained (us). N is the total number of cells that were imaged and analyzed.

(D) Confocal images of Gcg immunofluorescence of sorted P2a and P2b subsets. Scale bar, 50 μm.

(E) Quantification of Gcg immunofluorescence of sorted cells. (mean ± SEM, >50 cells for each subset, ****p < 0.0001).

Besides α cell heterogeneity, flow cytometry analysis of ZIGIR labeling revealed another major difference between human and mouse islet cells. In contrast to the mouse islet δ cell that showed a very low ZIGIR signal (Figures 5B and 5C), the human δ cell displayed strong ZIGIR staining that was comparable to the human β cell (Figures 6B and 6C; Figures S7B and S7C), suggesting high Zn²⁺ activity in the human somatostatin granule. To investigate the molecular basis for this phenomenon, we examined ZnT8 expression in human islets by multi-color immunofluorescence and found that ZnT8 was expressed in all three major islet endocrine cells, including the δ cell (Figure 7). The expression pattern of the ZnT8 protein in human islet cells was consistent with published RNA sequencing (RNA-seq) data documenting abundant expression of the SLC30A8 gene in human α cells, β cells, and δ cells (Segerstolpe et al., 2016), and it accounted for the strong ZIGIR signal and high granular Zn²⁺ activity in at least some human δ cells.

Figure 7. ZnT8 Is Expressed in Three Major Endocrine Cells of Human Pancreatic Islets.

Confocal immunofluorescence images of a human pancreas section stained with antibodies against three islet hormones and ZnT8. The enlarged images of the area highlighted by the dashed box are shown in the panels below, with ZnT8 pseudo-colored in red and individual hormones in green to aid visualization of the expression overlap. Scale bar, 20 μm.

DISCUSSION

Available Zn²⁺ sensors lack the required specificity and sensitivity for imaging Zn²⁺-rich secretory granules. Building upon our success in engineering a class of fluorescent Zn²⁺ indicators, ZIMIRs, for monitoring Zn²⁺ release at the cell surface (Li et al., 2011, 2015a, 2015b), we developed a specific granule Zn²⁺ indicator, ZIGIR, to fill this technology gap. ZIGIR possesses a set of desired properties, including Zn²⁺ selectivity, submicromolar Zn²⁺ affinity, >100-fold fluorescence enhancement, pH insensitivity, membrane permeability, and acidotropic propensity (Figures 1, 2, and 3). These combined advantages enable ZIGIR to mark Zn²⁺-rich secretory granules with unprecedented selectivity and brightness and to detect granular Zn²⁺ activity with high sensitivity and a large dynamic range. As a testimony to ZIGIR’s selectivity and dynamic range, only ZIGIR, not other Zn²⁺ sensors, including FluoZin-3, NPG, TSQ, and ZP4, was able to resolve granular Zn²⁺ activity of mouse α cells, β cells, and δ cells by flow cytometry (Figure 5E), and only ZIGIR showed exquisite granule staining in live cells by confocal imaging (Figures 2 and 3; Figure S2). A detailed comparison of ZIGIR with previously reported Zn²⁺ sensors, including several pH-insensitive ones (Que et al., 2015; Taki et al., 2004; Wu et al., 2005), is provided in Table S1. With respect to the photochemical mechanism of Zn²⁺ sensing by ZIGIR, we invoke photoinduced electron transfer from the lone pair of electrons of the nitrogen atom of the bottom aniline ring to the top xanthene fluorophore (Komatsu et al., 2005). Importantly, we propose that the aniline nitrogen functions predominantly as a Lewis base instead of a Bronsted-Lowry base (Figure S8). This scheme provides an explanation of why ZIGIR fluorescence is refractory to acidification down to pH 3.5 as long as Zn²⁺ is absent (Figure 1E).

In recent years, several genetically encoded Zn²⁺ sensors have been developed (Hessels and Merkx, 2015; Qin et al., 2013). Although these protein-based Zn²⁺ indicators can be conveniently targeted to subcellular organelles by fusing with the appropriate localization sequences (Hessels et al., 2015), their Zn²⁺ responsivity and dynamic range are generally below those of small synthetic probes, such as ZIGIR described herein. Furthermore, compared with the genetically encoded Zn²⁺ sensors, a distinct advantage of ZIGIR is that it can be easily applied to freshly isolated primary cells without the requirement of cell infection or protein expression, a property that can be particularly valuable for studying Zn²⁺ in primary cells of primates, including human.

Combining ZIGIR and Ex4-Cy5, we developed a simple procedure of sorting mouse islet α cells, β cells, and δ cells simultaneously (Figure 5). This sorting procedure should considerably facilitate molecular and functional analyses of primary mouse islet endocrine cells. Previous methods for sorting islet endocrine cells largely relied on genetic approaches by expressing fluorescent proteins under the control of cell-specific promoters, such as the insulin promoter, preproglucagon promoter, or somatostatin promoter (Egerod et al., 2015; Hara et al., 2003; Quoix et al., 2007). However, such methods only label one cell type at a time and can be further limited by incomplete cell labeling because of partial penetrance of the transgene reporter. Sorting different types of islet endocrine cells concurrently using this approach would require breeding double- or triple-transgenic mice by expressing different fluorescent proteins in each class of endocrine cells. Breeding such mice can become cumbersome, especially when additional genetic mutations need to be introduced. In contrast, our method only involves a simple labeling procedure using ZIGIR and Ex4-Cy5 to yield essentially pure α cells, β cells, and highly enriched δ cells in one step. The procedure is rapid and ought to be applicable to both wild-type and mutant mice.

In β cells, insulin granules labeled with ZIGIR were distinct and remained brightly fluorescent over the course of time-lapse imaging (Figure 2; Figure S3; Videos S1 and S2). This highlights the potential of applying ZIGIR to tag and to image native secretory granules in live cells. Tracking the spatial and temporal dynamics of secretory granules has important implications in understanding the mechanisms regulating granule distribution, trafficking, and docking near the plasma membrane, and defects in these cellular processes have been observed in islet cells from diabetic donors (Fu et al., 2019; Gandasi et al., 2018; Rorsman and Renström, 2003; Rosengren et al., 2012). Traditionally, tagging and imaging of dense core granules employ fluorescent proteins fused with a granule marker such as phogrin, syncollin, pro-insulin, islet amyloid polypeptide, or neuropeptide Y (Barg et al., 2002; Ferri et al., 2019; Michael et al., 2004; Rutter et al., 2006; Tabei et al., 2013). Limitations of these genetically encoded granule reporters include the requirement for viral infection, protein overexpression, and its potential perturbation on granule biogenesis, movement, or glucose response (Ferri et al., 2019; Michael et al., 2004). In contrast, ZIGIR labels the native dense core granule within minutes in freshly isolated islets, and the cells can be imaged immediately afterward. As we showed by TIRF imaging, ZIGIR labeling not only enabled tracking of individual granules and their dynamic movements near the surface but also could be exploited as an optical marker for exocytosis (Figure 4A; Videos S4 and S5). We therefore anticipate broad applications of ZIGIR in studying the biology of Zn²⁺-rich secretory granules in primary living cells, including islet α cells and β cells, when combined with the advanced optical microscopy.

The association of SNPs of the SLC30A8 gene with T2D has motivated numerous studies of analyzing ZnT8 transporter and Zn²⁺ signaling in pancreatic islets. Most of these studies have focused on mouse islet β cells (Davidson et al., 2014), with fewer works performed on mouse α cells (Le Marchand and Piston, 2010; Solomou et al., 2015). Because ZnT8 protein was not detected in mouse δ cells (Figure 5D) (Artner et al., 2010; Murgia et al., 2009; Solomou et al., 2015), studies on islet Zn²⁺ signaling or regulation thus far have neglected this important type of endocrine cell. In contrast with the mouse δ cell, we unexpectedly observed a high ZIGIR signal in at least some human δ cells, as similarly seen in human β cells (Figure 6; Figures S7B and S7C), and confirmed ZnT8 protein expression in the human δ cell (Figure 7). Altogether, these data suggested high Zn²⁺ activity in the somatostatin granule that is unique to the human δ cell. Islet δ cells, human δ cells in particular, make extensive contacts with surrounding α cells and β cells by projecting long filopodia-like processes (Arrojo E Drigo et al., 2019). The juxtaposition of δ cells with α cells and β cells facilitates paracrine signaling among the islet endocrine cells, and δ cells can exert a potent inhibitory effect on glucagon or insulin secretion during somatostatin release (Rorsman and Huising, 2018). Our data raise the possibility of a signaling role of granular Zn²⁺ from human δ cells. How SLC30A8 SNPs may affect ZnT8 expression or the granular Zn²⁺ level in human δ cells, and whether this pool of granular Zn²⁺ is co-released with somatostatin to modulate paracrine signaling in the islet, should become an interesting topic for future studies.

In addition to the high granular Zn²⁺ activity in the human somatostatin granule, ZIGIR reveals wide variations in glucagon content that is unique to human α cells. Of the five human donors examined by us, the ZIGIR signal in α cells from each donor spanned nearly two orders of magnitude (Figure 6; Figures S7B–S7E). Because ZIGIR is an intensity-based Zn²⁺ indicator instead of a ratiometric one, its cellular signal is affected both by granule Zn²⁺ activity and by the amount of dye loaded into cells. ZIGIR accumulates in the secretory granule, so the amount of secretory granule in a cell would affect the amount of ZIGIR uptake, which in turn contributes to the total ZIGIR fluorescence intensity emitting from a cell. Indeed, we have confirmed that ZIGIR-high α cells contained about four times as much glucagon as ZIGIR-low α cells (Figure 6; Figure S7B). To our knowledge, this α cell heterogeneity defined by the wide spread of glucagon content has not been reported before, and it appears to be a general phenomenon unique to human islets. Little is known about the regulation of glucagon synthesis in human α cells. In rodents, there have only been a few studies addressing how glucagon synthesis is modulated: glucagon was reported to have a positive autocrine effect to enhance glucagon synthesis (Leibiger et al., 2012), and the interaction between Pax6 and Alx3 transcription factors can suppress glucagon gene expression (Mirasierra and Vallejo, 2016). In addition, disruption of glucagon signaling can lead to hyperglucagonemia via hyperaminoacidemia (Gelling et al., 2003; Kim et al., 2017). How human α cells regulate their glucagon content under physiological conditions and how these processes lead to the formation of subsets of α cells with distinct glucagon content are exciting topics for future studies. Besides variations in hormone content, another important aspect of α cell heterogeneity relates to their functional properties. Previous imaging and electrophysiological studies have hinted at functional heterogeneity, as shown by the variation in amplitude, shape, and duration of nutrient-stimulated Ca²⁺ activity, ion current, or membrane capacitance among α cells (Huang et al., 2011; Le Marchand and Piston, 2010; Manning Fox et al., 2006; Rorsman et al., 2012; Zhang et al., 2013). Whether these biochemical or electrophysiological variations correlate with glucagon content in human α cells is not known. With ZIGIR labeling and imaging, we are poised to perform functional analysis on live human α cells with distinct glucagon content and to investigate the molecular underpinnings for these differences that are unique to human islets. Over the past decade, there has been increasing appreciation for the role of α cells and glucagon secretion in diabetes research (Unger and Cherrington, 2012). Investigating human α cell heterogeneity and how subpopulations of α cells may contribute to the glucose homeostasis in physiology or to the glucose volatility in diabetes presents exciting opportunities and challenges.

Besides islet endocrine cells, several mammalian cells, including prostate epithelial cells, excitatory neurons in the central nervous system, and mast cells, contain a high level of Zn²⁺ in their secretory granules (Frederickson et al., 2005). Given its superb specificity and sensitivity for reporting Zn²⁺-rich secretory granules, we anticipate broad applications of ZIGIR in diverse biological imaging studies, such as Zn²⁺ regulation, tracking of granule dynamics in living cells, and isolation of cell clones with high hormone content to facilitate cell engineering for cell replacement therapy.

STAR★METHODS

Detailed methods are provided in the online version of this paper and include the following:

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Wen-hong Li (wen-hong.li@utsouthwestern.edu).

Materials Availability

All unique reagents generated in this study are available on request from the Lead Contact with a completed Materials Transfer Agreement.

Data and Code Availability

This study did not generate any unique datasets or code.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Use of primary human pancreatic islets from deidentified organ donors was approved by the Institutional Review Board of University of Texas Southwestern Medical Center.

Primary Human donor islets information

Entry	Donor ID	Islet Source	Age (year)	Gender	BMI	HbA1c	Cause of death
1	SAMN10737781	IIDP	66	M	27.2	NA	Stroke
2	R299	ADI	44	F	25.4	4.7	Neurological
3	SAMN11046361	IIDP	57	M	35.9	NA	Stroke
4	SAMN11633049	IIDP	48	M	38.8	NA	Cerebrovascular/stroke
5	R309	ADI	47	F	27.4	5.5	Neurological
6	SAMN09479728	IIDP	41	F	25	NA	Cerebrovascular/stroke
7	R272	ADI	56	M	26.8	5.6	Neurological

IIDP: Integrated Islet Distribution Program; ADI: Alberta Diabetes Institute; NA: Not available.

METHOD DETAILS

Chemical Synthesis of ZIGIR

All reagents were purchased from Aldrich or VWR. Anhydrous solvents were stored over activated molecular sieves (3Å or 4Å). TLC was performed on precoated silica gel 60F-254 glass plates (EM Science). Reaction products were purified by low-pressure flash chromatography (FC) using silica gel 60 (63 – 200 μm; EM Science). ¹H-NMR spectra were acquired on a Varian 400-MHz or 500-MHz spectrometer. Chemical shifts (δ, ppm) were reported against tetramethylsilane (0 ppm). MALDI-TOF MS was performed on a Voyager-DE PRO biospectrometry workstation (Applied Biosystems) using 2,5-dihydroxy benzoic acid as the matrix.

5-Nitro-3-oxo-3H-spiro[isobenzofuran-1,9’-xanthene]-3′,6’-diyl bis(trifluoromethanesulfonate) (1)

4-Nitrophthalic acid anhydride (0.97 g, 5.0 mmol) and resorcinol (1.30 g, 11.8 mmol) were dissolved in 100 mL of methanesulfonic acid. The mixture was stirred at 80°C for 12 h. After cooling, the reaction was quenched in 100 mL of ice water and the mixture was filtered through a sintered glass filter. The retentate was dried under vacuum at 50°C for 8 h. The resulting dark red solid was then suspended in anhydrous pyridine (20 mL). Tf₂O (2.2 equiv.) was added dropwise at 0 °C. The reaction mixture was stirred at room temperature (r.t.) until the complete consumption of the starting material. Pyridine was evaporated under a reduced pressure and the crude residue was suspended in CHCI₃ (50 mL) and washed with saturated NaCl. The organic layer was dried over Na₂SO₄, and the concentrated crude product was purified by FC (hexane/EtOAc, 9:1 → 2:1) to provide the product (0.676 g, 21.1%), the 6-nitro isomer (0.632 g, 19.7%), and the mixture of both 5- and 6-isomers (0.295 g, 9.2%) as white solids. ¹H NMR (400 MHz, DMSO-d₆) δ 8.75 (d, J = 2.1 Hz, 1H), 8.63 (dd, J = 8.6, 2.1 Hz, 1H), 7.86 (d, J = 8.4 Hz, 1H), 7.81 (d, J = 2.4 Hz, 2H), 7.34 (dd, J = 8.8, 2.5 Hz, 3H), 7.26 (d, J = 8.9 Hz, 3H). MS: [M + H]⁺ calcd for C₂₂H₁₀F₆NO₁₁S₂⁺ 641.96; found: 642.27.

2-(6-(Dimethylamino)-3-(dimethyliminio)-3H-xanthen-9-yl)-5-nitrobenzoate (2)

DME (1.5 mL) was added to a mixture of compound 1 (100 mg, 0.16 mmol), Me₂NH.HCl (40 mg, 0.47 mmol), Pd₂dpa₃ (30 mg, 0.03 mmol), XPhos (38 mg, 0.08 mmol) and Cs₂CO₃ (300 mg, 0.96 mmol) in a pressure tube. The tube was degassed and purged with Argon three times. The tube was then sealed and heated at 70°C for 18 h. After cooling, the mixture was diluted with MeOH and silica (~1 g) was added to the mixture. The dried mixture was then purified by FC using a gradient of 5 → 15% MeOH in DCM to afford the product as a purple solid (45 mg, 67%). ¹H NMR (CD₃OD, 400 MHz) δ 8.97 (d, J = 1.9 Hz, 1H), 8.48 (dd, J = 8.3, 2.0 Hz, 1H), 7.56 (d, J = 8.3 Hz, 1H), 7.19 (d, J = 9.5 Hz, 2H), 7.07 – 7.00 (m, 2H), 6.96 (d, J = 2.4 Hz, 2H), 3.29 (s, 12H). MS: [M + H]⁺ calcd for C₂₄H₂₂N₃O₅⁺ 432.1554; found: 432.6610.

5-Amino-2-(6-(dimethylamino)-3-(dimethyliminio)-3H-xanthen-9-yl)benzoate, sodium salt (3)

Under an argon atmosphere, compound 2 (55 mg) was treated with CH₂Cl₂/TFA (1/1, 2 mL) at r.t. overnight. The reaction mixture was dried under vacuum and used for the next step without further purification. The red solid was dissolved in MeOH/H₂O (1/1, 5 mL) and refluxed with NaSH (0.5 g) for 1 h. The reaction mixture was concentrated and purified by FC (10% → 40% MeOH in CH₂Cl₂) to give the target compound as a red solid (32 mg, 92%). ¹H NMR (CD₃OD, 400 MHz) δ 7.48 (d, J = 2.1 Hz, 1H), 7.38 (dd, J = 9.5, 0.6 Hz, 2H), 7.09 – 6.95 (m, 4H), 6.90 (d, J = 2.4 Hz, 2H), 3.27 (s, 12H). MS: [M + H]⁺ calcd for C₂₄H₂₄N₃O₃⁺ 402.1812; found: 402.5546.

ZIGIR

NaCNBH₃ (63 mg, 0.3 mmol, 10 equiv.) was added to a solution of compound 3 (10 mg, 0.03 mmol) and compound 4 (60 mg, 0.18 mmol, 6 equiv; Li et al., 2011) in anhydrous MeOH containing dried Na₂SO₄ (200 mg, 100 eq). The mixture was stirred at r.t. overnight and filtered. The filtrate was concentrated under vacuum, and the resulting residue was purified by reversed phase column chromatography (LiChroprep RP-18) to afford ZIGIR as a red film in 15% yield. ¹H NMR (CD₃OD, 400 MHz) δ 8.44 (td, J = 0.8, 4.6 Hz, 2H), 7.83 – 7.63 (m, 1H), 7.48 (d, J = 9.5 Hz, 2H), 7.30 (dd, J = 0.8, 7.2 Hz, 1H), 7.20–7.27 (m, 4H), 6.93 (d, J = 8.4 Hz, 1H), 6.74 (dd, J = 2.4, 9.2 Hz, 2H), 6.65–6.70 (m, 2H), 3.85 (s, 2H), 3.28 (s, 12H), 3.26 (t, J = 6.0 Hz, 2H), 2.96–3.02 (m, 4H), 2.87 (t, J = 6.1 Hz, 2H). MS: [M + H]⁺ calcd for C₃₉H₄₁N₆O₃⁺ 641.3235; found: 641.8668.

Characterization of ZIGIR in vitro

Details of ZIGIR synthesis are described in the Supplemental Information. UV Vis spectra of ZIGIR were recorded in a 1-cm path quartz cell on a Shimadzu 2401 PC spectrometer. Fluorescence excitation and emission spectra were recorded on a Fluorolog 3 spectrometer (Jobin-Yvon Horiba, Edison, NJ). Zinc titration was performed by adding ZIGIR (1.0 μM final concentration) to buffered Zn²⁺ solutions containing 100 mM HEPES (pH 7.4). Nitrilotriacetic acid (NTA, 10 mM) and varying concentrations of ZnSO₄ (0–9 mM) were mixed to reach free Zn²⁺ concentrations between 0.1 nM and 40 nM (Li et al., 2011). Zn²⁺ concentrations above 40 nM were controlled by iminodiacetate (IDA, 10 mM) and varying amounts of ZnSO₄ (0 – 9.7 mM) in 100 mM HEPES (pH 7.4) (Sasaki et al., 2011). To determine Zn²⁺ binding dissociation constants (K_d(Zn²⁺)), the Zn²⁺ titration data were fitted to the least square exponential equation (Prism 7). Fluorescence quantum yields of ZIGIR at pH 7.4 were determined using rhodamine 6G as the reference (Ф_fl = 0.94 in MeOH) (Magde et al., 2002).

To examine the pH sensitivity of ZIGIR, we recorded the fluorescence emission spectra from pH 3 – 9.4 in either nominally Zn²⁺-free solutions containing 5 mM iminodiacetate or in 25 μM ZnSO₄ solutions. The pH was controlled with 10 mM pH buffers including chloroacetic acid (pH 3.1), acetate (pH 4.1 and 5.0), 2-morpholinoethanesulfonic acid (MES, pH 6.14), HEPES (pH 7.5) and N-cyclohexyl-2-aminoethanesulfonic acid (CHES, pH 9.4). To study the metal selectivity, the fluorescence of ZIGIR (1 μM) was measured in the presence of 10 μM TPEN and an excess of metal ion including 1 mM KCl, 1 mM NaCl, 1 mM CaSO₄, 1 mM MgSO₄, 15 μM MnSO₄, 15 μM FeSO₄, 15 μM NiSO₄, 15 μM CoCl₂, 15 μM CuSO₄, 15 μM CdSO₄ or 15 μM ZnSO₄. The emission intensity was normalized to that of 15 μM ZnSO₄ (100%).

Cell culture and imaging

MIN6 and INS-1 beta cells were cultured as previously described (Li et al., 2011). To correlate the intensity of ZIGIR labeling and protein expression by immunofluorescence in the same cells (Figures 2, 3, and 4), cells were cultured on glass coverslips (18 mm in diameter, #1.5 glass) for ~36 – 48 hours to reach ~50% confluence. Prior to labeling, cells were washed with a secretion assay buffer (SAB) containing 114 mM NaCl, 4.7 mM KCl, 1.2 mM KH₂PO₄, 2.5 mM CaCl₂, 1.16 mM MgSO₄, 3 mM glucose, and 20 mM HEPES (pH 7.4). Cells were then incubated with ZIGIR (1 μM) in SAB at 37°C for 15 min and washed with SAB. The coverslip was then transferred into a Ludin imaging chamber and imaged on a Zeiss LSM 780 confocal microscope using 561 nm laser for exciting ZIGIR (Em 568 – 610 nm). After ZIGIR imaging, the cells were fixed with 4% PFA (15 min, RT), permeabilized with PBST (PBS with 0.15% Triton X-100) for 10 min, washed with PBS, and incubated with a blocking buffer (10% vol/vol donkey serum in PBS) for 30 min at 37°C. The cells were then treated with primary antibody in the blocking buffer for 30 min at 37°C, washed with PBS (3 × 10 min), and treated with dye-labeled secondary antibodies in the blocking buffer for 30 min at 37°C. After PBS washing for 10 min, cells were stained with DAPI (300 nM) for 5 min, washed with PBS and imaged on the same microscope. Image analysis and correlation of ZIGIR signal and protein immunofluorescence was performed with Fiji software and JACoP plugin. The primary and secondary antibodies include: Guinea pig anti-insulin and AF488-conjugated donkey anti-guinea pig secondary antibody (1:200); Rabbit anti-NPC1 or Rabbit anti-GM130 and Cy3-conjugated donkey anti-rabbit (1:200); Mouse anti-LAMP2 and AF647-conjugation anti-mouse (1:200), Rat anti-LAMP2 and Cy5-conjugation anti-rat (1:200)

For other imaging experiments, cells were seeded on 35-mm Petri dishes with glass bottoms (MatTek) and cultured for ~24–48 hours to reach ~50% confluence. ZIGIR labeling and imaging followed the same conditions described before. Other Zn²⁺ sensor including FluoZin-3/AM (5 μM) (Han et al., 2018; Hwang et al., 2008; Kaltenberg et al., 2010; Qin et al., 2013; Roh et al., 2012), ZnAF3/Ac (2 μM) (Komatsu et al., 2005), NPG/AM (5 μM) (Parnaud et al., 2008), TSQ (30 μM) (Frederickson et al., 1987; Karim and Petering, 2017; Meeusen et al., 2011), or ZP4 (5 μM) (Nolan et al., 2004) was co-loaded with ZIGIR (1 μM) in a SAB solution (3 mM glucose) at 37°C for 20 min. Cells were then treated with either Zn/pyrithione (10 μM) or TPEN (25, 50 or 100 μM) in SAB, then re-imaged under the same setting. The granular ZIGIR intensity was quantified with “Analyze Particles” function of ImageJ, and the cytoplasm ZIGIR signal intensity was quantified by drawing ROIs of similar granular size in regions free of granular staining. To analyze granule size, we focused on puncta in the range of 0.01 μm² to 0.2 μm² (corresponding to particle diameter 100 – 500 nm) since we reasoned that larger puncta likely corresponded to closely clustered granules that could not be resolved by optical imaging. To examine ZIGIR accumulation in acidic granules, LysoTracker Green DND-26 was added to H1299 cells at 0.4 μ for ~5 min before imaging (Ex 488 nm, Em 498 – 552 nm).

TIRF imaging of ZIGIR-labeled INS-1 cells was carried out on a Deltavision™ OMX SR system (GE Healthcare Life Sciences) using a 60x objective (APON,1.49 NA) and 568 nm laser excitation. ZIGIR emission was filtered through a bandpass emission filter (591–626 nm) and detected with a CMOS camera (PCO).

Glucose-Stimulated Insulin Secretion

Hand-picked mouse (C57BL/6J, male, 22 weeks-old) islets were allowed to recover overnight in RPMI supplemented with 10% FBS, Pen/Strep (100 U/mL and 100 μg/mL), 2 mM L-glutamine. Islets were washed twice with pre-warmed KRBH solution (2.8 mM glucose, 102 mM NaCl, 5 mM KCl, 1.2 mM MgCl₂, 2.7 mM CaCl₂, 20 mM HEPES, 5 mM NaHCO₃, and 10 mg/ml BSA, pH 7.4). 70 Islets were then incubated in KRBH (37°C) for 45 min, then with ZIGIR (0.5 μM) for 15 min, washed with pre-warmed KRBH once again. Another 70 islets were incubated in KRBH (37°C) for 1 hr as a control. 10 islets were then transferred into each well of 12-well plates with 0.5 mL pre-warmed KRBH to start the secretion assays. Insulin was assayed at 2.8 mM and 16.7 mM glucose for 30 min (37°C, 5% CO2) with CISBIO Ultra-sensitive Insulin Assay following the manufacturer’s protocol on a PHERAstar FS microplate HTRF reader (BMG Labtech). For assaying total insulin quantification, islets were directly extracted with ethanol-HCl (0.5 mL per 10 islets).

FACS of MIN6 β-cells or mouse islet cells

MIN6 cells cultured on 60 mm Petri dish were labeled with ZIGIR, washed and trypsinized. The cells were resuspended in the sorting buffer (SAB Buffer containing 3 mM glucose, 0.5% BSA and 0.1 mg/mL DNase I) and sorted on a FACSAria II SORP cell sorter (BD Biosciences). DAPI (0.5 μM) was added to the cells just before sorting to gate dead/live cells. Live MIN6 cells were sorted into ZIGIR-High and ZIGIR-Low subsets. The sorted cells were cultured for 1 week, then seeded in 35-mm imaging dishes and cultured for ~24 hours to reach ~50% confluence. The cells were then fixed with 4% PFA (15 min at RT), immunostained for insulin, imaged by confocal microscopy and quantified with Fiji as previously described. To measure the insulin contents of the sorted ZIGIR-High and ZIGIR-Low subsets, the sorted cells were washed twice with PBS, resuspended (1 × 10⁶ cells/mL) in an ice-cold acid-ethanol solution (2% by vol. of conc. HCl (37% in water) in anhydrous EtOH) and sonicated for 2 min at high power (Bioruptor, UCD-200, Diagenode). The acid solution was neutralized with an equal volume of 1M Tris buffer (pH 7.4) and assayed for total insulin using the HTFR Insulin High Range Assay kit following vendor’s protocol.

C57BL/6J mice were maintained in 12-h dark/light cycles, with ad libitum access to diet (Teklad 2016) and water. All protocols for mouse use and euthanasia were reviewed and approved by the Institutional Animal Care and Use Committee of the University of Texas Southwestern Medical Center. Mouse islets were isolated as previously described by perfusing Collagenase P (Roche, 1.4 mg/mL in HBSS with 5 mM glucose) through the common bile duct (Li et al., 2011). Following a 15 min digestion at 37°C, the pancreas digestion was washed twice with HBSS by centrifugation and the islets were handpicked under a dissection scope. Isolated islets were cultured for 2 hours or more in RPMI-1640 medium supplemented with 10% (v/v) FBS, 2.0 mM sodium-pyruvate, 100 U/mL penicillin and 100 mg/mL streptomycin at 37°C in 5% CO₂. For flow cytometry analysis or FACS, we incubated islets with Ex4-Cy5 (50 nM) for ~1 h prior to islet dispersion. Mouse islets were dispersed with 0.05% Trypsin solution in DPBS at 37°C for 15 min. The dispersed islet cells were filtered through a 70 μm cell strainer and incubated with 0.5 μM ZIGIR at 37°C for 15 min in the cell sorting buffer. For comparison, dispersed mouse islet cells were labeled in parallel with a previously reported Zn²⁺ sensor including FluoZin3/AM (2 μM) (Han et al., 2018; Hwang et al., 2008; Kaltenberg et al., 2010; Qin et al., 2013; Roh et al., 2012), NPG/AM (25 μM) (Lukowiak et al., 2001), or TSQ (30 μM) (Frederickson et al., 1987; Karim and Petering, 2017; Meeusen et al., 2011) at 37°C for 15 min. The labeled cells were then washed once with SAB and sorted on a FACSAria Fusion or a FACSAria II SORP cell sorter (BD Biosciences). DAPI (0.5 μM) was added to the cell suspension just prior to the sorting. The sorted live cells (in 0.3 mL sorting buffer) were adhered to the polylysine coated (0.1 mg/mL for 15 min) glass slides using a Cytospin 4 Cytocentrifuge (ThermoFisher; 300 rpm for 3 min). The attached cells were then fixed with 4% PFA (15 min at RT) and immunostained for insulin (Guinea Pig Anti-Insulin), glucagon (Mouse Anti-Glucagon) and somatostatin (Rabbit Anti-Sst). After washing and labeling with the corresponding secondary antibodies (1:200; AF488-conjugated donkey anti-guinea pig; Cy3-conjugated donkey anti-rabbit; AF647-conjugated donkey anti-mouse), cells were imaged by the confocal microscopy to score for α-cells, β-cells and δ-cells in each subsets of sorted cells, P1–P4.

Immunofluorescence of mouse pancreas cryosections

C57BL/6J mouse pancreata were harvested and placed in a Histosette II Tissue Cassette and fixed with 4% PFA for 2 hours at 4°C. After washing with PBS (30 min × 3), the fixed pancreas was immersed in a sucrose solution (30% wt./vol. in PBS) overnight at 4°C. The pancreas was blotted to remove excess sucrose and immersed in Cryogel at RT for at least 2 hours. The Cryogel mold was frozen in acetone/dry ice bath and cut into 10 μm cryo-sections onto Superfrost Plus Slides using a Leica CM1950 Cryostat. For immunofluorescence, cryosections were warmed to RT and rehydrated in PBS for 10 min, permeabilized with PBST (0.3%) for 30 min, washed with PBST (0.1%) and treated with a blocking buffer (10% donkey serum in PBST (0.1%)) for 1 hour. The tissue sections were then labeled with a rabbit anti-ZnT8 antibody (HI-C, 1:50), a guinea pig anti-insulin antibody (1:500), a mouse anti-glucagon antibody (1:2000) and a sheep anti-somatostatin antibody at 4°C overnight. The tissue was then washed with PBS (10 min × 3) and stained with the corresponding secondary antibodies for 1 hour. After washing with PBS (3 × 5 min), the labeled cryo-section was shielded with a glass coverslip using Fluoromount-G and imaged by confocal microscopy.

Deidentified human FFPE pancreatic sections were obtained from nPOD. After deparaffinization, rehydration and antigen retrieval, the tissue sections were permeabilized, blocked and stained with antibodies against ZnT8, insulin, glucagon and somatostatin similarly as described above.

Flow analysis and FACS of human islet cells

Human islets were purchased from IIDP or Alberta Diabetes Institute and cultured in CMRL-1066 medium with 10% FBS. Human islets were dispersed with 0.05% Trypsin solution in DPBS at 37°C for 15 min and filtered through a 70 μm cell strainer. The dispersed cells were incubated with 1 μM ZIGIR at 37°C for 15 min in the sorting buffer. The cells were then washed once and incubated on ice for 20 min with two antibodies, APC-conjugated HIC1 and Alexa Fluor 488-conjugated anti-TM4SF4. The cells were then washed twice with cold sorting buffer and labeled with DAPI before sorting. The sorted live cells (P1, P2a and P2b subsets) were analyzed by immunofluorescence using the same method described above for the sorted mouse islet cells.

QUANTIFICATION AND STATISTICAL ANALYSIS

All statistical analyses were performed using GraphPad Prism 8. Statistics were performed by unpaired t test or one-way ANOVA. Quantification data were presented as (mean ± SEM). Significance is indicated in each figure with asterisks (*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001).

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Guinea pig polyclonal anti-insulin	Agilent	Cat# A0564; RRID: AB_10013624
Rabbit polyclonal anti-NPC1	Abcam	Cat# ab134113; RRID: AB_2734695
Rat anti-LAMP2 antibody [GL2A7]	Abcam	Cat# ab13524; RRID: AB_213473
Rabbit polyclonal anti-GM 130	Sigma-Aldrich	Cat# G7295; RRID: AB_532244
Mouse monoclonal anti-LAMP2	BD Biosciences	Cat# 555803; RRID: AB_39613
Mouse monoclonal APC-conjugated HIC1	Novus Biologicals	NBP1–18946APC
Mouse monoclonal anti-TM4SF4 AF488-conjugated	R&D Systems	FAB7998G
Mouse monoclonal anti-Glucagon	Sigma-Aldrich	Cat# G2654; RRID: AB_259852
Sheep polyclonal anti-somatostatin	American Research Products	Cat# 13–2366; RRID: AB_1542966
Alexa Fluor® 488 AffiniPure Donkey Anti-Guinea Pig IgG (H+L)	Jackson Immunoresearch	Cat# 706-545-148; RRID: AB_2340472
Alexa Fluor® 647 AffiniPure Donkey Anti-Mouse IgG (H+L)	Jackson Immunoresearch	Cat# 715-605-151; RRID: AB_2340863
Cy3 AffiniPure Donkey Anti-Rabbit IgG (H+L)	Jackson Immunoresearch	Cat# 711-165-152; RRID: AB_2307443
Cy5 AffiniPure Donkey Anti-Rat IgG (H+L)	Jackson Immunoresearch	Cat# 712-175-153; RRID: AB_2340672
Cy3 AffiniPure Donkey Anti-Sheep IgG (H+L)	Jackson Immunoresearch	Cat# 713-165-147; RRID: AB_2315778
Rabbit anti-ZnT8	University of Denver, Howard Davidson Group	N/A
Bacterial and Virus Strains
Deidentified human FFPE pancreatic sections	nPOD	Donor# 6004
Chemicals, Peptides, and Recombinant Proteins
LysoTracker Green DND-26	ThermoFisher Scientific	Cat# L7526
2-Morpholinoethanesulfonic Acid	TCI America	Cat# 50-014-40052
ZIGIR	This paper	N/A
HEPES Buffer	Sigma-Aldrich	Cat# H3575
N-Cyclohexyl-2-aminoethanesulfonic acid	Frontier Scientific	Cat# 50-144-8697
TPEN	Sigma-Aldrich	Cat# P4413
DNase I recombinant, RNase-free	Roche	Cat# 4716728001
Cryogel	EMS	Cat# 62806–01
Collagenase from Clostridium histolyticum	Sigma-Aldrich	Cat# C7657
Normal Donkey Serum	Jackson Immuno Research Labs	Cat# NC9624464
RPMI-1640 medium	GIBCO	Cat# 11875–093
TSQ	Enzo Life Sciences	Cat# ENZ-52153
FluoZin-3 AM	ThermoFisher Scientific	Cat# F24195
Newport Green PDX acetoxymethyl ether	Invitrogen	Cat#N24191
Zinpyr-4	Strem Chemicals Inc	Cat# 07–0312
1-Hydroxypyridine-2-thione zinc salt	Sigma-Aldrich	Cat# H6377
Critical Commercial Assays
HTFR Insulin High Range Assay kit	CisBio	Cat# 62IN1PEG
Experimental Models: Cell Lines
Mouse insulinoma beta-cells MIN6	Miyazaki et al., 1990	N/A
INS-1 832/13 Rat Insulinoma Cell Line	EMD Millipore	SCC207
Human Bone Osteosarcoma Epithelial Cells (U2OS Line)	Joachim Seemann Lab	N/A
Experimental Models: Organisms/Strains
C57BL/6J mice	Jackson Laboratory	Stock No:000664
Software and Algorithms
ImageJ	https://imagej.nih.gov/ij/	N/A
ChemDraw 19.0.22	PerkinElmer Informatics	N/A
Graphpad Prism 8.0.3	GraphPad Software	N/A
Other
Histosette II Tissue Cassette	Fisher Scientific	Cat# 15-182-701c
Superfrost Plus Slides	Fisher Scientific	Cat# 12-550-15
Fluoromount-G	SouthernBiotech	Cat# 0100-01
Ludin imaging chamber	Life Imaging Services	https://www.lis.ch/thechamber.html

Highlights.

• ZIGIR is a fluorescent Zn²⁺ indicator that selectively labels Zn²⁺-rich granules

• ZIGIR imaging tracks granule dynamics and exocytosis with high sensitivity

• ZIGIR enables sorting of human or mouse islet α cells and β cells with high purity

• ZIGIR reveals human α cell heterogeneity defined by distinct glucagon content