Localization of all four ZnT zinc transporters in Dictyostelium and impact of ZntA and ZntB knockout on bacteria killing

Transition metals, such as iron (Fe), zinc (Zn), manganese (Mn), cobalt (Co) and copper (Cu), are essential for the survival of all living organisms (Weiss and Carver, 2018). These metals are incorporated into active sites of metalloenzymes, and organize secondary structures such as the Zn²⁺ finger domains of many proteins, including transcription factors. They are therefore implicated in a wide range of crucial biological processes. However, excess of these metals is toxic for living organisms, partially because they compete for metal-binding sites in enzymes. In this context, recent studies have revealed transition metals as a ‘double-edged sword’ during infection of phagocytic cells. Upon phagocytosis, the innate immune phagocyte restricts intravacuolar bacterial growth either by depleting essential metal ions (e.g. Fe²⁺ and Mn²⁺) or by accumulating others, such as Cu²⁺ and Zn²⁺, to intoxicating concentrations (Flannagan et al., 2015; Lopez and Skaar, 2018). On the pathogen side, bacteria have evolved several strategies to survive excess of metal ions, both in the environment (Ducret et al., 2016; Gonzalez et al., 2018) and in contact with phagocytes (Botella et al., 2011). For instance, metal efflux transporters, such as cation diffusion facilitators and P-type ATPases, remove excess ions from the bacterial cytoplasm (Chan et al., 2010; Kolaj-Robin et al., 2015).

Recently, Zn²⁺ poisoning was shown to be among the killing strategies of macrophages (Botella et al., 2011). Inside macrophages, free Zn²⁺ mainly localizes to late endosomes and lysosomes, and to a smaller extent to early endosomes (Botella et al., 2011). Zn²⁺ presumably enters cytoplasmic organelles by fusion with Zn²⁺-containing endosomes, called zincosomes, or by the direct action of one or several Zn²⁺ transporters located at the membrane of the organelle. For example, Zn²⁺ has been observed to accumulate in phagolysosomes containing the nonpathogenic bacteria Escherichia coli, contributing to their killing. In addition, Zn²⁺ transporters are upregulated in macrophages infected with Mycobacterium tuberculosis and, after a cytosolic burst, free Zn²⁺ is delivered and accumulates into the Mycobacterium-containing vacuole (MCV) 24 h after infection (Botella et al., 2011; Pyle et al., 2017; Wagner et al., 2005). Interestingly, vice versa, the expression of the P₁-type Zn²⁺-exporting ATPase CtpC of M. tuberculosis, the Zn²⁺-exporting P_1B-type ATPase ZntA of E. coli and Salmonella enterica serovar typhimurium, and the Zn²⁺ cation diffusion facilitator of Streptococcus pyogenes increases during infection of human macrophages and neutrophils (Botella et al., 2011; Kapetanovic et al., 2016; Ong et al., 2014).

However, the import mechanism and temporal dynamics of Zn²⁺ inside phagosomes containing nonvirulent and virulent bacteria are still poorly understood. Two families of Zn²⁺ transporters, ZnT (Zn²⁺ transporter Slc30a) and ZIP (Zrt/Irt-like protein Slc39a), have been described in metazoans and in the amoeba Dictyostelium discoideum (hereafter referred to as Dictyostelium) (Dunn et al., 2018; Kambe et al., 2015). In metazoans, the ZnT family consists of nine proteins (i.e. ZNT1–8 and ZNT10) that decrease cytosolic Zn²⁺ levels by exporting it to the extracellular space or sequestering it into the lumen of organelles. In contrast, the 14 members of the ZIP family (i.e. ZIP1–14) catalyse the transport of free Zn²⁺ from the extracellular space and organelles into the cytosol (Kambe et al., 2015). In Dictyostelium, four putative ZnT proteins and seven ZIP transporters have been identified, but the directionality of transport has not yet been experimentally confirmed (Dunn et al., 2018; Sunaga et al., 2008).

In recent years, Dictyostelium has evolved as a powerful model system to study phagocytosis and host-pathogen interactions, including cell autonomous defences and bacterial killing (Bozzaro et al., 2008; Dunn et al., 2018). The core machinery of the phagocytic pathway is highly conserved between human phagocytes and Dictyostelium, which digests and kills nonpathogenic bacteria for nutrition. First, bacteria are recognized by receptors at the plasma membrane, leading to an actin-dependent deformation of the membrane and the closure of the phagocytic cup (Neuhaus et al., 2002). A few minutes after uptake, lysosomes fuse with the phagosome, delivering the vacuolar H⁺-ATPase (vATPase), which acidifies the phagosomal lumen (pH <4) and contributes to the digestion of bacterial components in concert with various sets of lysosomal enzymes that are transferred to the phagosome by fusion with endosomes and lysosomes (Gotthardt et al., 2002). These enzymes are then retrieved from the phagosomes together with the vATPase starting 45 min after uptake, leading to a postlysosomal compartment with neutral lumenal pH (Gotthardt et al., 2002). Undigested material is then expelled from the cell by exocytosis. In Dictyostelium, free Zn²⁺ has been recently localized in the contractile vacuole (CV), an organelle crucial for osmoregulation, and in the endolysosomal system, including phagosomes containing nonpathogenic E. coli (Buracco et al., 2018).

Here, we monitor free Zn²⁺ in the endophagocytic pathways of Dictyostelium, to compare with knowledge acquired in macrophages, and to understand the evolutionary origin of the metal poisoning strategy. Moreover, so far it remains unclear whether ZnTs are involved in the Zn²⁺ accumulation in the endosomal-lysosomal pathway and, consequently, we extended our analysis to various znt knockouts.

To decipher in more detail the recently reported subcellular localization of free Zn²⁺ in Dictyostelium (Buracco et al., 2018), cells were pre-incubated overnight with the fluid-phase tracer tetramethylrhodamine (TRITC)-dextran [which accumulates in all endosomes and lysosomes (Hacker et al., 1997)]. The cells were stained with either Fluozin-3 AM (FZ-3; Fig. 1A) or N1-(7-Nitro-2,1,3-benzoxadiazol-4-yl)-N1,N2,N2-tris(2-pyridinylmethyl)-1,2-ethane-diamine (NBD-TPEA; Fig. 1B), two fluorescent probes selective for Zn²⁺, and fed with 3 µm latex beads. Importantly, using these two fluorescent probes, we were able to localize Zn²⁺ inside the different cellular compartments, but not in the cytosol. As revealed by FZ-3 and NBD-TPEA staining (Fig. 1A,B), Zn²⁺ was clearly detectable inside the bead-containing phagosomes (BCPs, asterisks) and zincosomes that colocalized with the endosomal marker TRITC-dextran (arrowheads). To define more precisely the identity of zincosomes, Dictyostelium expressing VatB-RFP, a marker of lysosomes and the CV (Bracco et al., 1997), or RFP-VacA, a marker of postlysosomes (Wienke et al., 2006), were stained with both Zn²⁺ probes and fed with latex beads (Fig. 1C–I). Using these markers, Zn²⁺ was detected inside bladders of the CV before discharge (Fig. 1C, arrows; Movie 1), in VatB-RFP-positive lysosomes (Fig. 1F, arrowheads) and in VatB-RFP-negative, but RFP-VacA-positive, postlysosomes (Fig. 1D,F–I, arrowheads). Besides its osmoregulatory function and role in Ca²⁺ sequestration, the CV is proposed to serve as a transient sink for divalent metals (Bozzaro et al., 2013). Consequently, our observations are in line with the hypothesis that toxic metals and other ions are expelled from the cell via the CV (Bozzaro et al., 2013; Heuser et al., 1993). Importantly, we observed that lysosomes were not labelled by FZ-3 (Fig. 1D; Movie 1), and there was no NBD-TPEA signal inside the CV (Fig. 1E).

Taken together, our data support the localization of Zn²⁺ inside zincosomes that are of lysosomal and postlysosomal nature, as well as inside BCPs and in the CV (Fig. 1J).

We precisely monitored the dynamics of appearance and accumulation of Zn²⁺ in the BCP by time-lapse microscopy (Fig. 2A–D). Strikingly, the NBD-TPEA signal became visible inside BCPs very early after uptake, peaked at 15–20 min and then disappeared until the beads were exocytosed (Fig. 2A,C; Movie 2), whereas the FZ-3 signal was first observed only after ∼20 min postuptake, peaked at 35–40 min and persisted until exocytosis (Fig. 2B,D; Movie 3). We verified that the two probes were specific for Zn²⁺, because the signals of both NBD-TPEA and FZ-3 were eliminated by TPEN [N,N,Nʹ,N′-tetrakis-(2-pyridylmethyl)-ethylenediamine], a cell-permeant Zn²⁺ chelator, but not by the nonpermeant Zn²⁺ chelator diethylenetriaminepentaacetic acid (DTPA; Fig. S1A,B).

It is known that a pH lower than 5 quenches by ∼50% the FZ-3 signal, whereas a pH lower than 4 leads to a complete extinction of the FZ-3 signal (Gee et al., 2002). On the contrary, NBD-TPEA is not affected by low pH, but quenched by Cu²⁺ concentrations of ∼30 µM (Xu et al., 2009). The phagosomes of Dictyostelium, more acidic than those in macrophages (Yates et al., 2005), reach a pH lower than 3.5 a few minutes after particle uptake (Marchetti et al., 2009; Sattler et al., 2013), which could explain the fact that, in wild-type cells, most of the BCPs became FZ-3 fluorescent only after ∼30 min of bead uptake (Fig. 2B,D). To demonstrate that the absence of FZ-3 signal in endosomes and lysosomes is due to low pH, cells stained with FZ-3 and 10 kDa dextran were incubated with the lysosome-disrupting agent L-leucyl-L-leucine methyl ester (LLOMe) as described previously (Lopez-Jimenez et al., 2018). LLOMe creates small pores, leading to proton leakage and re-neutralization of the compartments (Repnik et al., 2017). This can be monitored directly with the help of the pH-dependent dye 8-hydroxypyrene-1,3,6-trisulfonic acid (HPTS) (Fig. 2E, left). In the case of FZ-3, addition of LLOMe resulted in an acute and drastic increase of the FZ-3 signal in most vesicles (Fig. 2E, right), confirming our hypothesis. To further assess whether low pH is also responsible for the absence of FZ-3 signal in the early BCPs of wild-type cells, we fed latex beads to PIKfyve knockout (KO) cells, which are strongly inhibited in phagosomal acidification (Buckley et al., 2018). As shown in Fig. 2F, BCPs became FZ-3 positive only few minutes after uptake, and remained brightly fluorescent until exocytosis. In sharp contrast, the BCPs of wshA mutant cells, which are incapacitated in re-neutralizing phagosomes due to a defect in the retrieval of the vATPase (Carnell et al., 2011), did not show any FZ-3 fluorescence (Fig. 2G). These results confirm that FZ-3 is quenched in early BCPs, owing to the low lumenal pHs, and becomes fluorescent only during re-neutralization, when the vATPase is retrieved (Fig. 1D). Regarding the sensitivity of NBD-TPEA to Cu²⁺, cells were pre-treated with different concentrations of CuSO₄ before incubation with the probe and beads. Five µM CuSO₄ was sufficient to completely quench the signal of NBD-TPEA inside BCPs (Fig. S1C), indicating that its fluorescence is likely quenched by the import of Cu²⁺ into maturing BCPs. Altogether, integrating the information acquired with both dyes (Figs 1 and 2; Fig. S1), we conclude that free Zn²⁺ is in fact present inside BCPs from early stages after bead uptake and until exocytosis of the particle. In addition, FZ-3 and NBD-TPEA can be used in combination with fluorescent organelle reporters to decipher the localization and function of Zn²⁺ during grazing on bacteria.

In macrophages, Zn²⁺ can be delivered to phagosomes via fusion with endosomes (Botella et al., 2011). We wondered whether this was also the case in Dictyostelium. Careful examination of amoebae during the phagocytosis assays mentioned above (i.e. pre-incubation with TRITC-dextran, staining with FZ-3 or NBD-TPEA, and feeding with 3 µm latex beads) revealed that zincosomes observed in the vicinity of the BCP (asterisks) can sometimes be captured in the process of fusion, during which they deliver their content into the BCP lumen, leading to a crescent shape that increased and spread over time (Fig. 3A; Movie 4, arrows). These observations suggest that Zn²⁺ is delivered to maturing BCPs via zincosome-phagosome fusion. Interestingly, a large fraction of the delivering zincosomes was surrounded by AmtA, an ammonium transporter present in endosomes and phagosomes (Kirsten et al., 2008; Uchikawa et al., 2011), confirming their endosomal nature (Fig. 3B). We also observed that Zn²⁺ was present in zincosomes formed by scission of a vesicle from NBD-TPEA-positive BCPs (Fig. 3C; Movie 5, arrows). In summary, the dynamic relationship between zincosomes and BCPs is complex. Our experiments clearly show that Zn²⁺ can be delivered by zincosome-BCP fusion, although it has not yet been possible to quantitate the contribution of this pathway to the total phagosomal Zn²⁺ concentration.

Zn²⁺ is sequestered into cellular organelles by Zn²⁺ transporters of the ZnT family (Kambe et al., 2015). In order to determine the localization of the Dictyostelium Zn²⁺ transporters, cells expressing fluorescent chimeras of each of the four ZnT proteins were fixed and stained with antibodies against several CV and endosomal markers (Figs 4 and 5). ZntA-mCherry partially colocalized with VatA, a subunit of the vATPase present in both lysosomes and membranes of the CV network (Neuhaus et al., 1998), and perfectly colocalized with Rhesus50, a protein that localizes exclusively to the CV (Benghezal et al., 2001) (Fig. 4A).

To monitor whether ZntA-mCherry might additionally localize to phagosomes at some stage of maturation, cells were fed with latex beads, fixed and stained with an antibody against p80 (Fig. 4B), a predicted Cu transporter that is located throughout the endosomal pathway and concentrates at postlysosomes (Ravanel et al., 2001). These experiments excluded colocalization of ZntA and p80 (Fig. 4B), and highlighted that ZntA-mCherry was, in fact, not detected in BCPs (Fig. S2, asterisks). To confirm the exclusive presence of ZntA at CV membranes, live microscopy was performed on ZntA-mCherry-expressing cells incubated with FZ-3. The fluorescent probe accumulated in ZntA-positive bladders of the CV that frequently discharged their content, expelling FZ-3-chelated Zn²⁺ to the extracellular milieu (Fig. 4C, arrowheads). Therefore, we conclude that ZntA is located exclusively to the CV bladders and network.

Using the same approach, cells expressing ZntB-mCherry and incubated with FZ-3 were fed latex beads, revealing the presence of ZntB-mCherry at BCPs (Fig. 4D, asterisks), starting from 6 min to until 35 min after bead uptake (Fig. 4D, arrows; Movie 6). In addition, ZntB-mCherry was also observed around FZ-3-positive zincosomes (Fig. 4D, arrowheads). The ZntB-mCherry-labelled zincosomes (Fig. 4E, arrowheads) and BCPs (Fig. S3A, asterisks) were also positive for NBD-TPEA, indicating that the ZntB-positive compartments can be acidic and likely contain a relatively low Cu²⁺ concentration. In addition, ZntB-mCherry partially colocalized with p80 at BCPs (Fig. 4F,G), indicating their lysosomal or early postlysosomal nature. This is in line with the observation that ZntB-mCherry also colocalized with the vATPase (Fig. S3B), but not with Rhesus50 or the endoplasmic reticulum marker PDI (Fig. S3C,D). Similar to its human homologue ZNT10 (also known as SLC30A10) (Bosomworth et al., 2012; Dunn et al., 2018), ZntB was also located at recycling endosomes and/or the Golgi complex, as determined by its juxtanuclear colocalization with p25, a marker for recycling endosomes (Charette et al., 2006) (Fig. 4H). Note that, in Dictyostelium, recycling endosomes concentrate around the microtubule organizing centre (MTOC), a region in which, typically, the Golgi complex is also located. These results led us to conclude that ZntB locates mainly to organelles of the endosomal pathway.

To decipher the subcellular localization of ZntC and ZntD, homologues of human ZNT6 (also known as SLC30A6) and ZNT7 (also known as SLC30A7), respectively (Dunn et al., 2018), cells expressing ZntC- and ZntD-mCherry were fixed and stained with antibodies against p80 and p25. ZntC- and ZntD-mCherry both colocalized with p25 in the juxtanuclear region (Fig. 5A,B), and did not colocalize significantly with p80 (Fig. 5C,D), suggesting that these transporters either locate in recycling endosomes that are usually low in p80 (Charette et al., 2006), or in the Golgi complex. Importantly, ZntC- and ZntD-mCherry were not observed at BCPs at any stage of maturation (Fig. 5E,F). The locations of the various Dictyostelium ZnTs are schematized in Fig. 5G.

Zn²⁺ accumulates inside the CV and is expelled from the cell when the CV discharges (Figs 1C and 4C). Because ZntA was the only ZnT localized at the CV membrane (Fig. 4A–C; Fig. S2), we wondered whether ZntA was involved in the import of Zn²⁺ into the CV lumen. To test this hypothesis, a zntA KO was created (see Materials and Methods and Fig. S4A,B), and wild-type and zntA KO cells expressing VatB-RFP were incubated with 3 µm latex beads and FZ-3 (Fig. 6A,B). As expected, the FZ-3 signal was visible in BCPs of wild-type and mutant cells (Fig. 6A,B); however, no signal was observed in the VatB-RFP-positive CV of zntA KO cells (Fig. 6B). This suggests that ZntA mediates the main route of Zn²⁺ delivery into the CV. Overexpressing mCherry-ZntA in zntA KO cells rescued the transport of Zn²⁺ inside the CV network (Fig. 6C). Interestingly, when cells expressing AmtA-mCherry were stained with NBD-TPEA, the intensity of the signal in zincosomes of the zntA KO was more intense than in wild-type cells (Fig. 6D,E). Importantly, knocking out zntA did not alter the number of zincosomes (Fig. 6F,G). By quantifying the integrated signal intensity inside BCPs, we observed a higher level of zinc inside the phagosomes of zntA KO cells compared with wild type (Fig. 6H,I).

In conclusion, these data indicate that ZntA is the main Zn²⁺ transporter of the CV system, and that the absence of ZntA leads to an increased concentration of Zn²⁺ in the endosomal system (Fig. 6J).

As mentioned above, ZntB localized at zincosomes and phagosomes (Fig. 4D–H; Fig. S3). To test whether ZntB mediates the transport of Zn²⁺ into these compartments, a zntB KO, generated within the Genome Wide Dictyostelium Insertion (GWDI) project, was used (Fig. S4C). The insertion site of the blasticidin cassette was confirmed by genomic PCR (Fig. S4D). Wild-type and zntB KO cells were incubated overnight with TRITC-dextran, stained with FZ-3 or NBD-TPEA, and fed with 3 µm latex beads (Fig. 7A–D). Strikingly, both FZ-3 and NBD-TPEA signals in the BCPs of zntB KO cells appeared strongly reduced (Fig. 7A,D). By adjusting the image settings, and by quantification of the integrated signal intensity inside BCPs, the signal in zntB KO cells was shown to be decreased by ∼60% compared with wild type (Fig. 7B,C). The residual amount of Zn²⁺ detected might be delivered to BCPs by fusion with zincosomes that are also present in the zntB KO (Fig. 7D, arrows). Importantly, overexpression of ZntB-mCherry in the zntB KO rescued the defect in Zn²⁺ content (Fig. 7E).

We propose that ZntB is the main Zn²⁺ transporter in lysosomes and postlysosomes (Fig. 7F), and that, in its absence, residual levels of Zn²⁺ are reached within these compartments by fusion with zincosomes or trafficking from recycling endosomes, where Zn²⁺ is transported by ZntC or ZntD.

As ZntB is located at BCPs, a generic type of phagosomes that have a relatively transient nature, leading to particle exocytosis after about 60 min, we wondered whether ZntB is also present at compartments containing the nonpathogenic Mycobacterium smegmatis, which are documented to have phagolysosomal identity but are more persistent, releasing killed bacteria after a couple of hours. ZntB-mCherry localized at the membrane of the MCV immediately after bacteria uptake, and Zn²⁺, detected by NBD-TPEA, also accumulated inside the MCV at the same time (Fig. S5). The concentration of intraphagosomal Zn²⁺ appeared to increase during the early stages of infection (from 1 min to 33 min postuptake), which agrees with the hypothesis of ZntB being the main lysosomal Zn²⁺ transporter. Strikingly, as we observed before for BCPs (Fig. 7D), Zn²⁺ could also be delivered to the MCV by fusion of ZntB-mCherry-decorated zincosomes (Fig. S5, 33 min).

In macrophages, infection with E. coli leads to an increase in the cytosolic level of Zn²⁺, followed by its transport inside phagosomes (Botella et al., 2011). When Dictyostelium was infected with E. coli, we could not observe a cytosolic burst of Zn²⁺, but Zn²⁺ appeared in the phagosomes rapidly after uptake (Fig. 8A). This was due, at least in part, to fusion of the phagosomes with zincosomes (Fig. 8A, arrowheads) in line with our previous observations (Fig. 3A; Fig. S5) and the findings in macrophages (Botella et al., 2011).

To determine whether Zn²⁺ contributes to intraphagosomal bacteria killing by Dictyostelium, a Zn²⁺-hypersensitive E. coli mutant with an inactivated P_1B-type Zn²⁺ efflux ATPase [ΔzntA (Rensing et al., 1997)] was used. We first confirmed the hypersensitivity of this mutant to Zn²⁺ (Fig. 8B); 0.2 mM Zn²⁺ was enough to strongly inhibit its growth in vitro, whereas the proliferation of wild-type bacteria was only inhibited by concentrations of Zn²⁺ above 1.25 mM (Fig. 8B). This differential and dose-dependent inhibition of bacteria growth was not observed for Fe²⁺ and Mn²⁺. The response to Cu²⁺ was biphasic. Low concentrations led to a weak dose-dependent effect, and higher concentrations fully inhibited growth. Nevertheless, both wild type and ΔzntA behaved similarly, and no differential effect of Cu²⁺ was observed (Fig. S6A–C). In line with observations in macrophages (Botella et al., 2011), Dictyostelium killed the ΔzntA mutant bacteria faster than the wild-type E. coli (Fig. 8C,D). Interestingly, both E. coli strains were killed faster by the Dictyostelium zntA KO than by wild-type Dictyostelium (Fig. 8C), whereas no significant difference in killing was observed between Dictyostelium wild-type and zntB KO cells (Fig. 8D). This suggests that accumulation of Zn²⁺ in the phagolysosomes contributes to bacterial killing and that these compartments harbour a higher Zn²⁺ concentration in the zntA KO than in wild-type Dictyostelium.

Intracellular bacteria and pathogens are limited to nutrients that are available inside the host cell. Professional phagocytes are able to exploit this dependence and have developed strategies to restrict intracellular bacteria growth or killing. For instance, essential nutrients such as Fe and Zn²⁺ are either withheld from the pathogen-containing vacuole or pumped into the phagosomal lumen to ensure bacteria killing in concert with other immunity factors. Nramp1 mediates Fe²⁺ sequestration from the phagosome and leads consequently to the starving of the phagosomal bacteria (Bozzaro et al., 2013; Peracino et al., 2006). Surprisingly, both nutrient deprivation (Djoko et al., 2015; Kehl-Fie and Skaar, 2010; Subramanian Vignesh et al., 2013) and metal poisoning have been reported in the case of Zn²⁺ (Botella et al., 2011; McDevitt et al., 2011; Soldati and Neyrolles, 2012). Here, we investigated the subcellular localization and role of Zn²⁺ in phagocytosis and killing using Dictyostelium as a model professional phagocyte.

Inside cells, Zn²⁺ is either bound to proteins or sequestered as free Zn²⁺ into the various cellular compartments or vesicles. In mammals, 50% of the total cellular Zn²⁺ is present in the cytoplasm, 30–40% is in the nucleus and 10% is located at the plasma membrane (Kambe et al., 2015). The cytosolic Zn²⁺ concentration ranges from pM to low nM (Vinkenborg et al., 2009).

By using two fluorescent probes, FZ-3 and NBD-TPEA, we investigated the cellular distribution of compartmentalized free Zn²⁺ in the professional phagocyte Dictyostelium. In line with previous observations from Buracco and colleagues (Buracco et al., 2018), Zn²⁺ located inside the endolysosomal system and, more precisely, inside zincosomes of lysosomal and postlysosomal nature (Fig. 1E–J), as well as inside the CV network (Figs 1C and 4C). In addition, Zn²⁺ was present in the lumen of phagosomes soon after bead uptake and until exocytosis (Fig. 2A–D). Free Zn²⁺ is delivered to maturing phagosomes both via direct pumping and via zincosome fusion (Fig. 3). Our data indicate that ZntB is a major endosomal ZnT [the signal is lower in phagosomes of zntB KO cells (Fig. 7A–C)], but membrane trafficking from other endosomes, such as recycling endosomes, or from the biosynthetic pathway (Golgi), which are positive for ZntC and ZntD, might contribute to endosomal Zn²⁺ levels. Unfortunately, it is not yet possible to measure precisely the contributions of each pathway and transporter to the Zn²⁺ concentration in the endosomal-phagosomal compartments. Nevertheless, we clearly document that zincosomes are of endosomal origin (lysosomal and postlysosomal, Fig. 1).

The seminal elemental analysis study of Wagner and colleagues revealed that the Zn²⁺ concentration inside the phagosomes containing various mycobacteria could reach high µM to mM levels (Wagner et al., 2005). Because the differential growth inhibition of wild type and the ΔzntA E. coli mutant was observed at Zn²⁺ concentrations between 0.20 mM and 1.25 mM (Fig. 8B), we conclude that the concentration of Zn²⁺ inside the phagosome might be in the low mM range. Because of the limitations of the fluorescent probes available, an absolute quantitation of the Zn²⁺ concentration inside various organelles is likely not yet achievable.

It was suggested that the concentration of cytosolic Zn²⁺ fluctuates in response to various stimuli (Kambe et al., 2015). For instance, after incubation of macrophages with E. coli, Zn²⁺ is released from storage complexes, followed by pumping and sequestration into phagosomes. However, this was not observed in Dictyostelium by feeding cells with beads or with bacteria (Figs 2A,B and 8A; Fig. S5). We reason that, because the concentration of cytosolic Zn²⁺ is very low [between pM and low nM (Kambe et al., 2015)], it is consequently under the detection limit of FZ-3 and NBD-TPEA, explaining why we could not monitor cytosolic Zn²⁺.

The subcellular homeostasis of Zn²⁺ is tightly regulated through uptake, storage, re-distribution and efflux mechanisms that are, among others, mediated by Zn²⁺ transporters of the ZnT and ZIP family (Bird, 2015). Seven members of the ZIP family and four members of the ZnT family have been identified in Dictyostelium (Dunn et al., 2018; Sunaga et al., 2008). ZntC and ZntD are the Dictyostelium homologues of human ZNT6 (Huang et al., 2002) and ZNT7 (Kirschke and Huang, 2003), respectively. As their mammalian counterparts (Kambe et al., 2015), they localized at recycling endosomes and/or in the Golgi complex of Dictyostelium (Fig. 5A–G). ZntA, which is not closely related to any specific human Zn²⁺ transporter (Dunn et al., 2018), was located at the membrane of the CV (Figs 4A–C and 5G; Fig. S2A). Dictyostelium ZntB is the closest homologue of human ZNT1 (also known as SLC30A1) and ZNT10 (Dunn et al., 2018). Whereas ZnT1 is a plasma membrane protein (Palmiter and Findley, 1995), ZnT10 locates at the Golgi complex and at early and/or recycling endosomes (Patrushev et al., 2012), a location similar to the one of ZntB, described here (Fig. 4H). In addition, ZntB was observed at BCPs at the phagolysosomal stage (Fig. 4D–G; Fig. S3A,B). Loss of ZntB led to a 60% reduction of the FZ-3 fluorescence inside BCPs (Fig. 7A–E), leading to the conclusion that ZntB is the main endolysosomal Zn²⁺ transporter in Dictyostelium (Fig. 7F).

Interestingly, Dictyostelium growth was only inhibited by Zn²⁺ or Cu²⁺ concentrations at 50- or 500-fold the physiological levels, suggesting a very efficient control of Zn²⁺ and Cu²⁺ homeostasis (Buracco et al., 2018). In Dictyostelium cells, a classical metallothionein activity is not detected (Burlando et al., 2002), and consequently the CV was proposed to serve as a detoxification system for metal ions such as Fe²⁺, Zn²⁺ and Cu²⁺ (Bozzaro et al., 2013; Buracco et al., 2018; Peracino et al., 2013).

Changes in the cytosolic Zn²⁺ levels under non-steady-state conditions are remedied in a process described as ‘muffling’ by diverse mechanisms such as the cytosolic buffering by metallothioneins, the extrusion of Zn²⁺ from the cell and the sequestration of Zn²⁺ into organelles (Colvin et al., 2010). In Dictyostelium, Zn²⁺ is sequestered into the CV thanks to ZntA-mediated transport (Fig. 6A,B). We think that the increased Zn²⁺ levels found in lysosomes (Fig. 6D–G) and BCPs (Fig. 6H,I) of zntA KO cells indicate an unexpected crosstalk between the endosome-related water-pumping CV and the endosomal-phagosomal maturation pathway. Our data indicate that the CV plays a major role in regulating Zn²⁺ in Dictyostelium, and that incapacitating a CV Zn²⁺ transporter leads to a domino effect and increases the phagosomal-endosomal Zn²⁺ concentration, thereby leading to a more bactericidal environment (Fig. 8C).

Botella and colleagues revealed metal poisoning of ingested microbes as a novel killing strategy of macrophages (Botella et al., 2011). Besides Zn²⁺ poisoning, one can speculate that bacteria also have to face Cu poisoning inside the phagosome. In line with that hypothesis, NBD-TPEA was quenched at later phagocytic stages (Fig. 2A,C; Fig. S1C), a plausible sign of Cu²⁺ accumulation. Interestingly, the expression of the Dictyostelium homologue of the phagosomal Cu transporter ATP7A is upregulated upon feeding with bacteria, consistent with a possible role of Cu poisoning during phagocytosis (Hao et al., 2016; White et al., 2009).

Excess Zn²⁺ levels might inhibit bacterial ATP production by impairing the activity of cytochromes (Beard et al., 1995). Additionally, excess Zn²⁺ might replace other metals in the active site of various enzymes or occupy nonspecific binding sites (Nies, 1999). Here, we investigated the role of Zn²⁺ during phagocytosis and killing of bacteria by Dictyostelium. Zn²⁺ was observed inside phagosomes containing E. coli (Fig. 8A) and M. smegmatis (Fig. S5). Similar to the situation described in macrophages, an E. coli strain deficient in the Zn²⁺ efflux P_1B-type ATPase ZntA was killed faster than the wild type (Fig. 8C,D), leading to the conclusion that Zn²⁺ poisoning belongs to the killing repertoire of Dictyostelium. Although the accumulation of Zn²⁺ inside lysosomes and BCPs of the zntA KO led to a better killing capacity of Dictyostelium (Fig. 8C), bacteria killing in the zntB KO was unaffected (Fig. 8D). This suggests that Zn²⁺ poisoning is an evolutionarily conserved process and might act in concert with other killing factors, such as phagosomal acidification, reactive oxygen species production, and deprivation or poisoning by other metals, which would compensate for the loss of ZntB.

All the Dictyostelium material used in the study is listed below (Table S1). Dictyostelium Ax2(Ka) and AX4 cells were cultured axenically at 22°C in HL5-C medium (Foremedium) supplemented with 100 U/ml penicillin and 100 μg/ml streptomycin to avoid contamination. Cell lines expressing fluorescent reporters and KO cell lines were cultured in the presence of selective antibiotics [hygromycin (50 µg/ml), neomycin (5 µg/ml) or blasticidin (5 µg/ml)]. To monitor the localization of ZnTs, Dictyostelium was transformed with plasmids carrying the ZntA-, ZntB-, ZntC- or ZntD-mCherry constructs [pDM1044 backbone (Veltman et al., 2009)]. The zntA KO was generated in the Ax2(Ka) background by homologous recombination following the one-step cloning protocol previously described (Wiegand et al., 2011). In brief, left and right arms of zntA were amplified using the primers 5′-AGCGCGTCTCCAATGCTGCAGGGAAGTGAGGGTGTG-3′ (forward) and 5′-AGCGCGTCTCCGTTGGTTTATGTTCGTGTTCATG-3′ (reverse), and 5′-AGCGCGTCTCCCTTCCAACAATAGATCCCGAAG-3′ (forward) and 5'-AGCGCGTCTCCTCCCCTGCAGGTGGATGTGCACTTC-3′ (reverse), and cloned into the StarGate^® Acceptor Vector pKOS-IBA-Dicty1 using the StarGate cloning kit. The resulting plasmid was transformed into Dictyostelium by electroporation, and positive clones were selected with blasticidin (Fig. S4A). Correct integration into the genome was tested by PCR using different combinations of primers: zntA flanking forward 5′-CGATTTGTTGTTACCTAAATATTCGTG-3′ and zntA flanking reverse 5′-CACCCAATTTACACTAGTTTCACC-3′, zntA inside forward 5′-GTGGTGAAGATGGTAGTAGTAGTG-3′ and zntA inside reverse 5′-CATGA GTACACCTAAACTTTCACG-3′, Bsr forward 5′-AGATCTTGTTGAGAAATGTTAAATTGATC-3′ and Bsr reverse 5′-TTGAAGAACTCATTCCACTCAAATATAC-3′ (Fig. S4B).

The zntB KO (AX4 background) was obtained as part of the GWDI Project (https://remi-seq.org) and was generously provided by Prof. Christopher Thompson. The individual mutant was obtained from the grid. To confirm the insertion site of the blasticidin cassette into the zntB gene, genomic DNA was isolated from wild type and zntB KO using the High Pure PCR Template Preparation Kit (Roche), and diagnostic PCR was performed according to the recommendations on the GWDI website. Primer combinations with the two zntB specific primers, zntB forward 5′-GGCAATTCCACGTTTCATCAG-3′ and zntB reverse 5′-GTAACGAATTGAATCCAAATCG-3′, binding ∼400 bp up- or downstream the insertion sides, and the two primers specific for the blasticidin cassette pGWDI1, 5′-GTTGAGAAATGTTAAATTGATCC-3′ and pGWDI2 5′-ATAGAAATGAATGGCAAGTTAG-3′, were used to confirm the insertion (Fig. S4C,D).

E. coli wild type and ΔzntA were kindly provided by Prof. Christopher Rensing (Rensing et al., 1997; Chinese Academy of Sciences Beijing, China), and cultured in LB medium. M. smegmatis (Hagedorn and Soldati, 2007) was cultured in 7H9 medium supplemented with 10% OADC, 0.05% Tween 80 and 0.2% glycerol at 32°C in shaking. Erlenmeyer flasks containing 5 mm glass beads were used to minimize clumping of bacteria. Vybrant DyeCycle Ruby Stain (Thermo Fisher Scientific) was used to stain intracellular M. smegmatis before live imaging was performed.

The day before imaging, Dictyostelium was plated on two- or four-well ibidi dishes. Three hours before imaging, HL5-C was changed to SIH [Formedium, full synthetic medium with low Zn²⁺ concentration (2.3 mg/l ZnSO₄)]. After 2 h in SIH, cells were washed three times in Soerensen buffer and stained with 2 µM FZ-3 (Thermo Fisher Scientific, F-24195; 400 µM stock in DMSO; Gee et al., 2002) or 5 µM NBD-TPEA (Sigma-Aldrich, N1040; 1 mM stock in DMSO; Qian et al., 2009) for 30 min in the dark. In order to synchronize phagocytosis, cells were cooled on a cold metal plate for 10 min before adding the 3 µm latex beads (Sigma-Aldrich, LB30). Beads were mixed with Soerensen buffer and added to the cells. After centrifugation at 500 g for 2 min at 4°C, the medium was carefully aspirated from the dish and an agarose overlay was placed on top of the cells, as described before (Barisch et al., 2015). Cells were imaged on an inverted 3i Marianas spinning disc confocal microscope using the 63× glycerol or 100× oil objectives. Where indicated, cells were treated for 3 h with different chelators: TPEN (Sigma-Aldrich, 87641), DTPA (Sigma-Aldrich, D6518) or CuSO₄×5H₂O (Sigma-Aldrich, C3036) before and throughout staining with the Zn²⁺ probes. To label all endosomes, TRITC-dextran (70 kDa; Sigma-Aldrich, T1162); 1 mg/ml; stock 10 mg/ml in ddH₂O) or Alexa Fluor 647-dextran (10 kDa; Molecular Probes, D22914; 30 µg/ml; 2 mg/ml stock in water) was added overnight and throughout FZ-3 and NBD-TPEA staining. To induce damage in lysosomes, cells were incubated with 7 mM LLOMe (Bachem, G-2550; 1 M stock in water). As a positive control for lysosomal damage, cells were treated with 0.2 mM of 524 Da HPTS (Molecular Probes, H348; 0.2 M stock in water).

To quantify the number of zincosomes inside wild type and zntA KO, cells were plated overnight on two-well ibidi slides. After staining with NBD-TPEA, images were taken using an ImageXpress spinning disc confocal microscope (Molecular Devices) and the number of zincosomes in wild type and zntA KO was assessed using MetaXpress (Molecular Devices).

The temporal and special dynamics of Zn²⁺ inside BCPs was quantified using ImageJ and the ‘CenterOnClick’ plugin (N. Roggli, University of Geneva, unpublished) that automatically centres the ‘clicked’ particle of interest in a recalculated image for further visualization and analysis. The integrated density inside a ‘donut’ that was drawn around the FZ-3 signal was measured using ‘plot Z-axis profile’.

Antibodies against p80 (Ravanel et al., 2001) were purchased from the Geneva antibody platform (University of Geneva, Switzerland). An anti-RFP antibody (Chromotek) was used to increase the fluorescence of mCherry-expressing fusion proteins. As secondary antibodies, goat anti-mouse, anti-rabbit and anti-rat IgG coupled to Alexa Fluor 488, Alexa Fluor 546 (Thermo Fisher Scientific) or CF640R (Biotium) were used. For immunofluorescence, Dictyostelium cells were fixed with cold MeOH or 4% paraformaldehyde, as described previously (Hagedorn et al., 2006). Images were recorded with a Zeiss LSM700 confocal microscope using a 63×/1.4 NA or a 100×/1.4 NA oil-immersion objective.

E. coli wild type and ΔzntA were grown in LB medium overnight at 37°C with shaking at 150 rpm. Bacteria were diluted to an optical density at a wavelength of 600 nm (OD₆₀₀) of 0.1 and plated in 96-well plates containing LB medium with different concentrations of heavy metals (i.e. ZnSO₄, CuSO₄, FeCl₃, MnCl₂ from 0.05 mM to 2.5 mM). The OD₆₀₀ was measured every hour using a 96-well plate reader (SpectraMax i3, Molecular Devices).

Intracellular killing of E. coli wild type and ΔzntA carrying a GFP-harbouring plasmid (Valdivia and Falkow, 1997) was monitored as described previously (Leiba et al., 2017). A 1:10 dilution of overnight E. coli cultures was centrifuged for 4 min at 18,000 g and bacteria were resuspended in 300 µl filtered HL5-C. The bacterial suspension (10 µl) was plated on each well of a four-well ibidi slide and centrifuged for 10 min at 500 g. Then, 300 µl of a 1×10⁶ cells/ml Dictyostelium culture in LoFlo (synthetic low-fluorescent medium, Foremedium) was overlaid on the bacteria, and images were recorded at 22°C with a Leica AF6000 LX wide-field microscope using the 40× dry objective at 30 s intervals. For each phagocytosed bacterium, the time between phagocytosis and fluorescence extinction (killing) was determined manually using ImageJ, and the probability of bacterial survival was represented as a Kaplan–Meyer estimator. The data of three independent experiments were pooled and statistical comparisons between Kaplan–Meier curves were calculated using the log-rank test.

To monitor the involvement of Zn²⁺ in the killing of E. coli, Dictyostelium cells were stained with NBD-TPEA, as mentioned above, and bacteria were labelled using CF594 succinimidyl ester (SE) (Sigma-Aldrich, SCJ4600031). Briefly, an overnight culture of bacteria was diluted 1:10 in Soerensen buffer and incubated with 2 µl of a 10 mM CF594 SE stock solution [in dimethyl sulfoxide (DMSO)] for 1 h in the dark. After two washes with Soerensen buffer, bacteria were resuspended in 1 ml filtered HL5-C. Subsequently, 10 µl of the bacteria suspension was added to the pre-cooled cells on an eight-well ibidi slide and centrifuged onto cells for 1 min at 500 g at 4°C. Images were taken at 90 s intervals using a spinning disc confocal microscope with a 63× objective.

We gratefully acknowledge the imaging platform of the University of Geneva for their expert and friendly support. We thank the GWDI Project (https://remi-seq.org) for the zntB REMI cell line. We are grateful to Prof. Michael Rensing for sharing the ΔzntA E. coli KO. We thank Dr Elena Cardenal-Muñoz for careful reading and editing of the manuscript and thoughtful suggestions; Dr Olivier Schaad for helping with the statistics; Dr Monica Hagedorn and Prof. Markus Maniak for comments on the manuscript; and Dr Olivier Neyrolles for inspiring the project. T.S. is a member of iGE3 (www.ige3.unige.ch).