ZnT9 is an evolutionarily conserved zinc transporter with a mitochondrial presequence

There are ten ZnTs in mammals, but in D. melanogaster, only seven. They, thus, do not all correspond to each other. Previous phylogenetic analyses between the flies and humans showed that several ZnTs have clear counterparts (orthologues); they are notably ZnT1, ZnT7 and ZnT9 [24]. Alignment of presumed ZnT9 orthologues from C. elegans, D. melanogaster, D. rerio, M. musculus and H. sapiens revealed that ZnT9 is highly conserved evolutionarily (Fig. 1A). All of them contain a potential mitochondrial presequence, suggesting the mitochondrial localization. The mitochondrial presence of the C. elegans ZnT9 and human ZnT9 has indeed been established [17,18,19]. Since the previous publications used GFP fusion to localize ZnT9 to the mitochondrion, to minimize the tag effect, we used the small tag HA to attach to mZnT9 to confirm its mitochondrial residence. mZnT9-HA was expressed in 4T-1 (mouse breast cancer cell line) cells. Mitochondria were isolated and expression of mZnT9-HA was examined by western blotting using anti-HA antibody. The results indicated that mZnT9-HA is predominantly in the mitochondrial fraction and very little in the post-mitochondrial fraction (Fig. 1B), consistent with previous reports.

Fig. 1

SLC30A9 (ZnT9) is an evolutionarily conserved mitochondrion-targeted zinc transporter. (A) Shown here is a comparison of ZnT9 from C. elegans, D. melanogaster, D. rerio, M. musculus and H. sapiens. Potential mitochondrial signals are colored by purple. Highly conserved residues are colored by yellow. Mutations in human SLC30A9 (pAla350del and p.Gly418Val, marked with a red and blue arrow, respectively) causing a novel cerebro-renal syndrome were previously reported [20, 23]. The six potential transmembrane domains are labelled above with lines. Notably, ZnT9 contains a set of highly conserved regions outside of the transmembrane domains. Also, two highly conserved His residues, indicated with * in the figure, exist between TM III and TM IV. Another distinctive feature is that the motif HXXXD occurring in TM V of most ZnTs is changed to V/IXXXD in ZnT9. This HXXXD motif, together with another HXXXD sequence in TM II, are believed to coordinate for zinc-binding in ZnTs. Accession numbers for depicted sequences are as follows: C. elegans ZnT9, NP_497603.3; D. melanogaster ZnT9, NP_725207.1; D. rerio ZnT9, NP_001008575.1; M. musculus ZnT9, NP_848766.2; H. sapiens ZnT9, NP_006336.3. (B) mZnT9 localizes to the mitochondrion. mZnT9-HA was expressed in 4T-1 (mouse breast cancer cell line) cells. Mitochondria were isolated and were marked by TOM20. The expression of mZnT9-HA was examined by western blotting using anti-HA antibody. T: total cell; C: cytosol; M: mitochondria

As anticipated, there are six putative transmembrane domains (TM) in ZnT9, as in other ZnT proteins [25]. Unsurprisingly, the six putative transmembrane domains are highly conserved during evolution. Within the region between TM III and TM IV of ZnT9, two conserved histidine residues notably exist (labelled with * in Fig. 1A). Another outstanding feature is the motif V/IXXXD (X stands for any amino acid) in TM V of ZnT9, which is HXXXD in many other ZnTs and considered to bind zinc in coordination with the HXXXD motif in TM II [26], which is preserved in ZnT9. Besides the transmembrane domains, the C terminus of ZnT9 and some inter-transmembrane regions also exhibit highly conserved blocks, implicating important, albeit unknown biological functions.

ZnT9 suppression results in mitochondrial defect and severe movement disorder in the fly

To elucidate the role of ZnT9 in vivo, we first undertook a functional analysis of the Drosophila ZnT9 orthologue (dZnT9). Two independent transgenic lines (1# and 2# lines) of dZnT9 RNAi and one mutant line were obtained. The expression of dZnT9 was first determined by qPCR. The transcript of dZnT9 was significantly decreased in both dZnT9 RNAi flies (directed by the ubiquitous Da-GAL4 driver) (Fig. S1A-B). Between the two RNAi lines, RNAi 2# expression suppression was more efficient. The dZnT9 mutant flies carry an insertion of MiMIC (Minos Mediated Integration Cassette) in the coding exon [27], resulting in virtually non-detectable ZnT9 expression (Fig. S1C).

dZnT9 knockdown or knockout appeared not to affect much the Drosophila at larval stages (Fig. S1D-F). They generally proceeded to pupal formation without much reduction in viability. However, dZnT9 interruption seriously affected the pupal stage (Fig. S1H-I). Compared to that of dZnT9 RNAi 2# flies, the eclosion (the process from pupae to adults) rate of dZnT9 RNAi 1# flies was less dramatically affected (Fig. S1G-H), corresponding to its higher remaining ZnT9 expression (Fig. S1A-B). In order to analyze the adult phenotype, we first chose the milder dZnT9 RNAi 1# line because the eclosion rate of dZnT9 RNAi 1# flies was not as much affected. About 50% of the dZnT9 RNAi 1# flies got stuck in the food after a few days of rearing (Fig. 2A), implying severe motion disability. Indeed, movement ability measurement confirmed that these alive dZnT9 RNAi 1# flies displayed apparent movement deficiency (Fig. 2B and Media File S1). Additionally, roughly 50% of them displayed abnormal wing posture, which means they could not close their wings (Fig. 2C). This wing defect could be due to muscle or hinge defects, or even the abnormality of nerve system. The more severe knockdown line, the dZnT9 RNAi 2# Drosophila, died at the pupal stage without eclosed flies (Fig. S1H). dZnT9-knockout Drosophila resembled largely the more severe form of knockdown: they all died as pupae (Fig. S1I).

Fig. 2

Drosophila ZnT9 knockdown led to mitochondrial defects, incapacitated movement, and mitochondrial zinc elevation. (A) The percentage of flies dying from food stickiness (n = 8 per group, each sample includes 60 flies). dZnT9 knockdown did not result in lethality per se, but some flies died from food-sticking due to motion disability. (B) Climbing performances of WT and dZnT9 RNAi 1# flies (n = 3 per group, each sample includes 20 flies). (C) The percentage of flies with abnormal wing posture (n = 3 per group, each sample includes 20 flies). (D) TEM images of mitochondria (marked by red arrows) in the thoracic muscles of WT and dZnT9 RNAi 1# flies. Scale bar = 500 nm. (E) The zinc level of mitochondria from WT and dZnT9 RNAi 1# flies (n = 3 per group)

During the implementation of the project, two ZnT9 papers came out reporting the functions of the C. elegans ZnT9 orthologue (cZnT9). It was shown that the mutant worm was viable with abnormal mitochondrial morphology. Further research demonstrated that cZnT9 influences the mitochondrial zinc level and suggested that cZnT9 effluxes zinc from mitochondria to the cytosol [17,18,19]. Based on these results, we analyzed the mitochondria in the thoracic muscles of dZnT9 RNAi 1# flies. The dZnT9 RNAi mitochondrion was grossly misshapen, and the crista was especially severely disrupted (Fig. 2D). The mitochondrial zinc level in dZnT9 RNAi 1# flies was then measured by inductively coupled plasma-mass spectrometry (ICP-MS). The zinc level after dZnT9 knockdown was increased significantly compared with the control (Fig. 2E). Taken together, our results confirm the requirement of ZnT9 in Drosophila development and show a unique role for ZnT9 in the homeostatic regulation of mitochondrial zinc.

Phenotypes of dZnT9 knockdown can be rescued by mouse ZnT9 or zinc chelation

Since ZnT9 was highly conserved among several species, we wanted to know whether mouse ZnT9 expression could rescue the defects of dZnT9 knockdown. The movement deficiency of dZnT9 RNAi 1# flies driven by 69B-GAL4, which expresses in embryonic epiderma and imaginal discs, could be partially rescued by mZnT9 expression (Fig. 3A). Moreover, the eclosion rate of dZnT9 RNAi 2# flies could also be rescued by mZnT9 (Fig. 3B). Previous findings indicated that SCaMC-2 (SLC-25A25) is a mitochondrial zinc transporter in C. elegans, moving zinc from the cytosol to mitochondria [18]. We asked whether the defects of dZnT9 loss of function could be rescued by SCaMC (the orthologue of SCaMC-2 in the fly) knockdown. The results indicated that the reduction of SCaMC could alleviate the movement and eclosion defects of dZnT9 knockdown flies (Fig. 3A-B).

Fig. 3

Defects of Drosophila ZnT9 knockdown could be complemented by mouse ZnT9 or zinc chelator TPEN treatment. (A) The movement deficiency of dZnT9 RNAi 1# flies was rescued by exogenously introduced mZnT9 or endogenous SCaMC knockdown. Expression or knockdown was by the UAS-GAL4 system using 69B-GAL4 as the driver (n = 3 per group, each sample includes 20 flies). (B) The eclosion rate of dZnT9 RNAi 2# flies was rescued by mZnT9 expression or SCaMC RNAi (n = 6 per group, each sample includes 60 flies). (C) The movement deficiency of dZnT9 RNAi 1# flies was partially rescued by TPEN (n = 3 per group, each sample includes 20 flies). (D) The abnormal wing posture of dZnT9 RNAi 1# flies was partially rescued by TPEN (n = 3 per group, each sample includes 20 flies). (E) Representative images of wing posture from WT and dZnT9 RNAi 1# flies reared on normal food (NF) or treated with TPEN (n = 6 per group). Scale bar = 0.5 mm

Given that zinc was accumulated in the mitochondria of dZnT9 knockdown flies (Fig. 2E), we wanted to know if zinc chelation could confer a similar rescue. TPEN, a permeable zinc chelator, reversed the movement deficiency and abnormal wing posture in dZnT9 knockdown flies (Fig. 3C-E). These results reaffirmed that the observed dZnT9 knockdown phenotypes are real (not off-target) and are a consequence of zinc dyshomeostasis (accumulation).

ZnT9 is indispensable to the early embryonic development of mice

The ZnT9 function in mammals was addressed by analyses of ZnT9 in the mouse. Global Znt9 knockout mice were created, but homozygous knockout mice could not be identified in more than 100 pups from heterozygotes matings (Fig. 4A), strongly indicating embryonic lethality. To confirm this assertion and investigate the phenotype in more detail, early embryos were isolated from Znt9+/− crosses, and we found that the homozygous knockout embryos were lethal before E10.5. At E10.5, the Znt9−/− embryos were severely reduced in size and deformed in shape (Fig. 4B), indicating that ZnT9 is essential to the early embryonic development of mice.

Fig. 4

ZnT9 is essential for the early embryonic development of mice. (A) Genotype of offspring from heterozygotes matings. (B) Morphology of WT and KO embryos at E10.5 dissected free of the yolk sacs (n = 9 and 13 for WT and KO, respectively). Scale bar = 1 mm

The early lethality phenotype of Znt9 knockout embryos prompted us to investigate whether there is important function for ZnT9 in the adult. We decided to generate an inducible Znt9 knockout mice (referred to as Znt9 iKO mice) using a line of mice ubiquitously expressing the tamoxifen-inducible Cre recombinase under the Rosa26 promoter [28]. To inactivate Znt9 in adult mice, Znt9 iKO mice were treated with tamoxifen intraperitoneally at 6 to 8 weeks of age and euthanized 2 months after the treatment. Few abnormalities were noticed. Znt9 iKO mice were viable and healthy by this time. The peripheral blood counts and plasma-related indexes of Znt9 iKO mice were also examined. Again, the Znt9 iKO mice did not present much aberration (Fig. 5A-H). We further performed histological analysis and found no significant abnormalities in the liver, kidney and spleen from Znt9 iKO mice (Fig. 5I-K). To examine the expression of Znt9, i.e., the inducible knockout efficiency, we performed qPCR analysis using different tissues of Znt9 iKO mice. Znt9 expression levels in the liver, and spleen were drastically diminished, while in the lung and kidney tissues Znt9 expression was reduced but not as much affected. Znt9 expressions in the heart, brain, and muscle were insignificantly suppressed (Fig. 5L). Because the heart, brain, and muscle are primarily mitochondria-enriched tissues, inefficient gene removal in these tissues possibly underlies the lack of phenotypes in the Znt9 iKO mice. Thus, our inducible KO approach proved not an effective means to analyze ZnT9 function post-early embryonic development. Nevertheless, it shows that the liver and spleen, or more broadly, those tissues not mitochondrion-rich, are not vulnerable to Znt9 loss.

Fig. 5

Inducible Znt9 knockout mice present few overt abnormalities. (A-C) Peripheral blood counts of WT and iKO mice (n = 7 and 10 for WT and iKO, respectively). (D-H) Determination of plasma related indexes of WT and iKO mice (n = 6 and 4 for WT and iKO, respectively). (I) Representative histological sections of liver from WT and iKO mice (n = 5 mice per group). Scale bar = 0.05 mm. (J) Representative histological sections of kidney from WT and iKO mice (n = 5 mice per group). Scale bar = 0.1 mm. (K) Representative histological sections of spleen from WT and iKO mice (n = 3 mice per group). Scale bar = 0.2 mm. (L) Znt9 expression levels in different organs of WT and iKO mice (n = 7 and 4 for WT and iKO, respectively)

Targeted mutagenesis of Znt9 in the brain leads to dwarfism and lethality

The low efficiency of inducible loss of ZnT9 in mitochondrion-rich organs led us to try tissue-specific Znt9 knockout. Clinical features of Birk-Landau-Perez syndrome, caused by ZnT9 mutations, are movement disorder, intellectual disability, oculomotor apraxia, developmental regression, and renal insufficiency [20,21,22,23]. These characteristics suggested to us that ZnT9 might play a critical role in the nervous system, a mitochondrion-rich tissue. To delineate the role of ZnT9 in the mice brain, we used Nestin-Cre to excise the floxed-Znt9 allele in the whole brain of mice (referred to as Znt9 cKO mice). To validate the deletion efficiency of Znt9 cKO, RT-PCR and qPCR were performed, and a significant decrease of Znt9 expression levels was indeed observed in the Znt9 cKO brain (Fig. S2A). As an additional control, we also analyzed Znt9 expression in the liver and kidney tissues and found no significant changes in relative Znt9 RNA levels of Znt9 cKO mice (Fig. S2B-C).

Brain Znt9 cKO mice could be born without apparent external abnormalities. However, by 2 and 3 weeks of age, Znt9 cKO mice were much reduced in size and weight (Fig. 6A-C). Znt9 cKO mice also presented serious movement disorder and tremors, dying at 3 to 5 weeks after birth (Fig. 6D and Media File S2). These results indicate that Znt9 ablation in the mouse brain results in dwarfism, severe loss of motion ability and a much-shortened lifespan. Interestingly, in the medical field, it is known that severe mitochondrial deficiency sometimes correlates with stunted growth [29]. The exact mechanism still needs to be clarified.

Fig. 6

Targeted mutagenesis of Znt9 in the brain produced severely dwarf and shortly lived mice. (A) Body weight of WT and cKO mice at P14 (n = 23 and 13 for WT and cKO, respectively). (B) Body weight of WT and cKO mice at P21 (n = 20 and 18 for WT and cKO, respectively). (C) Representative images of WT and cKO mice at P21. Scale bar = 1 cm. (D) Kaplan–Meier survival curve of WT and cKO mice (n = 25 and 23 for WT and cKO, respectively)

The GH/IGF-1 axis is severely impaired in the Znt9 cKO mice

Since no reduction in growth was observed in worms and flies, we subsequently asked how ZnT9 affected the size of mammals. Is it due to a growth hormone (GH) shortage? Pituitary glands, a part of the brain, are composed of neurohypophysis or the posterior lobes, and adenohypophysis, including the anterior and intermediate lobes [30]. Each cell type of adenohypophysis is characterized by the hormone it secretes. There are seven different hormones in the adenohypophysis: growth hormone produced by somatotropes, prolactin (PRL) produced by lactotropes, thyroid-stimulating hormone (TSH) produced by thyrotropes, adrenocorticotropic hormone (ACTH) produced by corticotropes, luteinizing hormone (LH) and follicle-stimulating hormone (FSH) produced by gonadotropes, and melanocyte-stimulating hormone (MSH) produced by intermediate lobe melanotropes [31]. GH is a key regulator hormone that plays an important role in body growth [32]. The production of IGF-1 by hepatocytes is regulated by GH [33]. Previous studies have established that the GH/IGF-1 axis plays a major role in controlling postnatal growth [32].

To examine possible defects in the pituitary glands, we dissected the tissue. We found that the pituitary glands of Znt9 cKO mice were smaller compared to wildtype (Fig. 7A). However, the reduction in size was roughly proportionally to the overall body decrease. Again, the posterior and anterior lobes areas of Znt9 cKO mice were all in proportion smaller than the control (Fig. S3A-C). While the ratio of total area and anterior lobes area to body weight of Znt9 cKO mice showed no significant change, the ratio of posterior lobes area to body weight increased (Fig. S3D-F). Histological sections of Znt9 cKO pituitaries showed no apparent defects (Fig. 7B). However, the mitochondria from the Znt9 cKO pituitaries exhibited disrupted cristae in morphology (Fig. 7C). Since the GH/IGF-1 axis controlled the body growth, the Igf-1 and Gh expression levels were measured. We found that the expression levels of Igf-1 and Gh both decreased dramatically in Znt9 cKO mice to almost an undetectable level (Fig. 7D-F). How could GH be so much affected? PIT1 is a major transcription factor and plays an essential role in the differentiation of somatotropes, lactotropes, and thyrotropes [31]. We then measured Pit1 expression in Znt9 cKO mice. Pit1 expression was downregulated slightly but not dramatically (Fig. 7G), at least not affected to the extent as the Gh. These results indicate that the GH/IGF-1 axis of Znt9 cKO mice is severely impaired, possibly partially due to Pit1 decrease, but other factors may also be involved.

Fig. 7

Targeted mutagenesis of Znt9 in the brain gravely impaired the GH/IGF-1 axis. (A) Representative images of pituitaries from WT and cKO mice at P21 (n = 10 and 7 for WT and cKO, respectively). Scale bar = 0.5 mm. (B) Representative histological sections of pituitaries from WT and cKO mice (n = 3 mice per group). Scale bar = 0.05 mm. (C) TEM images of mitochondria from WT and cKO pituitaries (n = 3 mice per group). Red arrows mark defective mitochondria from cKO mice. Scale bar = 200 nm. (D) Elisa analysis of plasma IGF-1 levels (n = 6 and 3 for WT and cKO, respectively). (E) qPCR for Igf-1 in the liver of WT and cKO mice (n = 6 mice per group). (F-G) qPCR for Gh (F) and Pit1 (G) in the pituitaries of WT and cKO mice (n = 6 per group). A sample was pooled from three pituitaries of the same genotype

One previous work reported that Nestin-cre is not expressed in the pituitary glands [34]. Since the GH/IGF-1 axis was impaired in Znt9 cKO mice, we wanted to explore if the defect was due to the loss of Znt9 in the pituitary or the hypothalamus. qPCR and genomic PCR were performed to measure the knockout efficiency in the pituitary. The result indicated that Znt9 was not eliminated effectively in the pituitary of Znt9 cKO mice; most animals did not carry significant Znt9 deletion (Fig. S2D-E). The genomic PCR for hypothalamus from Znt9fl/fl, cKO/+ and cKO mice was also performed. Znt9 was deleted efficiently in the hypothalamus (Fig. S2F). Therefore, it seems the loss of GH/IGF signaling was not due to the pituitary per se, but possibly a result of the hypothalamus-pituitary signaling.

Znt9 cKO is not associated with dopamine deficiency but results in a decrease in multiple neuroactive ligand − receptor signalings

In addition to being small in size, Znt9 cKO mice also displayed movement disorder and tremors, especially in the four limbs (Media File S2). These behaviors are reminiscent of the features of Parkinson’s disease. The main cause of Parkinson’s disease is impairment of dopaminergic neurons in the substantia nigra, due to the formation of protein inclusions named Lewy bodies (LBs) there, leading to a decrease of the dopamine level in the striatum [35]. We have previously shown that cytosolic zinc levels could affect tyrosine hydroxylase (TH) activity, leading to dopamine level change, through zinc-iron competition for TH [36]. To address whether the motion abnormality and tremors in Znt9 cKO mice arise from dopamine deficiency, we measured the dopamine level in the striatum of Znt9 cKO mice. No significant change was found (Fig. S4A). The expression levels of tyrosine hydroxylase in the substantia nigra of Znt9 cKO mice also presented no significant difference (Fig. S4B-C). Consistently, Nissl staining of Znt9 cKO striatum did not show obvious defects (Fig. S4D). These data suggest that the tremors of Znt9 cKO mice are not due to a decrease of dopamine in the striatum.

The observation that Znt9 mutation translates to GH signaling loss but the upstream Pit1 is not as affected indicates that other upstream components may be involved. RNA-seq data analysis of the Znt9 cKO mice brain revealed a striking feature: the downregulation of neuroactive ligand − receptor interaction stands out (Fig. S5). In addition, abundant signaling pathways changed in Znt9 cKO mice brain (Fig. S6). This suggests that Znt9-loss-resulted mitochondrial dysfunction is associated with multiple neuro-signaling defects.

Mitochondrial ETC activity is impaired by zinc accumulation

How does ZnT9 loss affect mitochondrial functions? Does zinc accumulation inhibit the respiratory complex (the electron transport or the ETC) activity? To answer these questions, we first knocked down ZnT9 in 293T cells by transfecting three separate shRNAs. ZnT9 expression analyses by qPCR revealed that ZnT9 was downregulated significantly in 293T cells after shRNA transfection either separately or together (Fig. 8A). The mitochondrial complex activity was then determined by transfecting the more effective shRNA 3# or the three shRNAs together. Complex I and II activities decreased dramatically after ZnT9 knockdown in 293T cells (Fig. 8B-C). Furthermore, ATP contents and mitochondrial membrane potential (MMP) were also measured. Loss of ZnT9 led to lower ATP contents and mitochondrial membrane potential (Fig. 8D-E).

Fig. 8

Mitochondrial ETC activity is vulnerable to zinc accumulation. (A) Transcripts for ZnT9 from 293T cells were quantified by qPCR 72 h after transfection as indicated. (B) The relative activity of complex I in 293T cells was examined 72 h after transfection as indicated. (C) The relative activity of complex II in 293T cells was examined 72 h after transfection as indicated. (D) The relative ATP levels in 293T cells were measured 72 h after transfection as indicated. (E) The relative mitochondrial membrane potential levels in 293T cells were measured 72 h after transfection as indicated. (F) The mitochondria from WT and Znt9 cKO mice brain were isolated and the labile zinc content was determined using TSQ. (G) The relative activity of complex I after the mitochondria isolated from 293T cells were incubated with different concentrations of ZnSO4 for 2 h. (H) The relative activity of complex II after the mitochondria purified from 293T cells were incubated with different concentrations of ZnSO4 for 2 h. (I) 293T cells were incubated with different concentrations of ZnSO4 for 24 h, and the relative ATP levels were examined using the isolated mitochondria. (J) 293T cells were incubated with different concentrations of ZnSO4 for 24 h, and then the relative mitochondrial membrane potential levels were determined. (K) Expression of NDUFS1, SDHB, and UQCRFS1 was examined by western blotting after the mitochondria were incubated with different concentrations of ZnSO4 for 2 h

To confirm that the mitochondrial labile zinc in Znt9 cKO mice brain is elevated, we isolated mitochondria and measured the free zinc level with TSQ, a cell-permeable zinc probe. Loss of Znt9 led to mitochondrial zinc increase (Fig. 8F). This is consistent with previous publications showing increased mitochondrial zinc levels [17,18,19]. The impact of zinc on mitochondrial ETC activity could be direct or indirect. For example, zinc accumulation could affect iron-sulfur (Fe-S) formation or ETC complex assembly. We therefore analyzed the consequence of mitochondria, isolated from 293T cells, after a short-time zinc treatment. Short-time treatment of mitochondria with zinc similarly led to suppression of complex I and II activities (Fig. 8G-H). 293T cells were treated with zinc for 24 h and then the mitochondria were isolated. Zinc supplementation resulted in decreased ATP contents and mitochondrial membrane potential (Fig. 8I-J). However, the complex expression levels and Fe-S synthesis appeared not significantly changed (Fig. 8K). These data indicate that zinc could exert an immediate direct inhibition on ETC activity.