Neuron-specific modulation of SLC30A10 identifies dopaminergic and glutamatergic neurons as targets of manganese-induced motor disease

Significance

At elevated levels, essential metals accumulate in the brain and induce motor disease. But, the neuronal targets of metals are unknown, limiting therapeutic progress. Because metal efflux transporters have high metal specificity, we hypothesized that neuron-specific modulation of efflux transporters may alter metal levels in targeted neurons and isolate the neuronal targets of metal toxicity. We tested this by focusing on manganese-induced motor disease due to its public health relevance. We generated several mouse strains with neuron-specific knockout or overexpression of the manganese efflux transporter SLC30A10. We establish that 1) elevated manganese targets dopaminergic and glutamatergic neurons to induce motor disease, and 2) neuron-specific efflux transporter modulation is an effective strategy to define the neuronal mechanisms of metal-induced disease.

Abstract

Essential metals accumulate in the basal ganglia at elevated levels and induce incurable motor disease. But, unlike other motor diseases, the neuronal targets of essential metals are unknown, and this fundamental knowledge gap has limited therapeutic progress. Because metal efflux transporters have high specificity, we hypothesized that neuron-specific knockout or knockin (i.e., overexpression) of efflux transporters may alter metal levels in targeted neurons and define the neuronal targets of metal-induced disease. To test this, we focused on manganese (Mn)-induced motor disease, which is a public health problem. We generated six neuron-specific Slc30a10 mouse strains with knockout or knockin of the Mn efflux transporter Slc30a10 in dopaminergic, GABAergic, or glutamatergic neurons. In the knockout strains, SLC30A10 was depleted and Mn levels were elevated in targeted brain regions. However, only dopaminergic- or glutamatergic-, but not GABAergic-, specific knockouts developed motor deficits without Mn exposure. Conversely, in the knockins, SLC30A10 was elevated and the increase in Mn levels after Mn exposure was attenuated in targeted regions. However, only dopaminergic- or glutamatergic-, but not GABAergic-, specific knockins were protected against Mn-induced motor deficits. Dopaminergic-specific Slc30a10 knockouts also exhibited deficits in dopaminergic neurotransmission that were consistent with their motor phenotype. Overall, 1) elevated Mn targets dopaminergic and glutamatergic neurons to induce motor disease, and 2) neuron-specific knockout/knockin of efflux transporters is an effective strategy to isolate the neuronal targets and underlying mechanisms of metal-induced neurological disease.

Levels of the essential metals manganese (Mn), copper (Cu), and iron (Fe) must be maintained within a narrow physiological range in the brain to support neurological function without neurotoxicity. At elevated levels, Mn, Cu, and Fe primarily accumulate in the basal ganglia in the brain and induce motor disease (1–5). Parkinsonism, dystonia, and/or other motor function deficits are well-established features of Wilson’s disease (a genetic disease of Cu toxicity), genetic or environmental Mn toxicity, and neurodegeneration with brain iron accumulation diseases (1, 3–5). Notably, a well-defined circuitry of GABAergic, glutamatergic, and dopaminergic neurons in the basal ganglia controls motor function (6–8). But, the primary basal ganglia neuronal types targeted by elevated levels of essential metals to induce motor deficits are unclear. This represents a fundamental gap in knowledge because 1) for most other motor diseases, the neuronal targets of disease are well established [e.g., dopaminergic neurons in idiopathic Parkinson’s disease (7, 8)], and 2) a clear understanding of underlying neuronal targets is necessary to develop effective treatments for neurological diseases [e.g., levo-dopa for dopamine replacement in Parkinson’s disease (9)]. The lack of information about the neuronal targets of essential metals is a major reason that there are no available treatments for metal-induced motor disease.

The inability to identify the neuronal targets of essential metal-induced motor disease has primarily been a limitation of available experimental models. Current strategies generally rely on exogenous metal exposure in wild-type animal models, but this approach induces metal accumulation in all basal ganglia neurons in an untargeted manner and precludes identification of the contributions of specific neuronal types to the disease process. Thus, development of model systems in which essential metal levels can be altered in some, but not all, basal ganglia neurons is a prerequisite for defining the neuronal targets of motor disease induced by elevated levels of essential metals.

Homeostatic control of essential metals depends on the activity of influx and efflux transporters. Notably, while mammalian metal influx transporters transport a variety of cations [e.g., DMT1 (10)], efflux transporters are specific for the transported metal under physiologic conditions [e.g., SLC30A10 for Mn (11), ATP7A or ATP7B for Cu (12), and Fpn for Fe (13)]. Genetic diseases highlight the specificity of metal efflux transporters. Homozygous loss-of-function mutations in ATP7B or SLC30A10, respectively, induce Cu overload and Wilson’s disease (1, 5, 12) or Mn retention and hereditary Mn-induced parkinsonism (4, 14, 15), and mutations in Fpn cause a form of hemochromatosis, an iron metabolism disease (16). The transport specificity of metal efflux transporters led us to hypothesize that neuron-specific knockout or knockin (i.e., overexpression) of an efflux transporter in mice may selectively alter levels of only the transported metal in only the targeted neurons and thereby provide a model system to isolate the role of the targeted neurons in metal-induced motor disease.

To test our hypothesis, we focused on Mn because of the public health relevance of Mn-induced neurotoxicity and motor disease. Elevations in brain Mn levels and subsequent motor disease may occur due to elevated Mn exposure, genetic mutations, or liver disease (4, 14, 15). Individuals may be exposed to elevated Mn from occupational (e.g., welding, mining, steel, and battery manufacture, etc.) or environmental (e.g., drinking water) sources (4). Mn is widely used in modern industry, and estimates suggest that >100,000 welders in North America are routinely exposed to elevated Mn (4, 17, 18). Additionally, environmental Mn exposure from drinking water is a major global health concern, particularly in the developmentally sensitive early-life periods of infancy, childhood, and adolescence (4). In the United States alone, ~2.6 million people consume drinking water with elevated Mn levels (i.e., with Mn levels that exceed the current Environmental Protection Agency health reference level of 0.3 mg Mn/L) (19). In addition to elevated exposure, homozygous loss-of-function mutations in genes coding for the Mn transporters SLC30A10 or SLC39A14 induce hereditary Mn-induced motor disease (4, 14, 15). SLC30A10 controls brain Mn via Mn efflux activity in the liver and intestines that mediates Mn excretion and direct Mn efflux activity in the brain (4, 14, 15, 20–22). SLC39A14 is a Mn influx transporter that imports Mn from the blood into the liver and intestines for subsequent excretion by SLC30A10 (4, 14, 15, 23, 24). Loss-of-function of either transporter leads to the retention of Mn in the body and the brain and the onset of Mn-induced motor disease (4, 14, 15, 20–24). Finally, the liver is a major organ that excretes Mn, and individuals with liver dysfunction (e.g., due to cirrhosis) may also retain excess Mn and develop Mn-induced motor disease without exposure to elevated Mn (14). Overall, Mn-induced motor disease is a significant biomedical problem that requires the development of effective treatment strategies.

In this study, we used neuron-specific Slc30a10 knockout or knockin mouse strains to isolate the neuronal targets of Mn-induced motor disease. Our results substantially advance understanding of the pathobiology of Mn-induced motor dysfunction and establish the feasibility of using efflux transporter modulation in mice as a general strategy to define the mechanisms of neurological disease induced by elevated metal levels.

Results

Generation and Validation of Neuron-Specific Slc30a10 Knockout Mouse Strains.

To define the neuronal targets of Mn-induced motor disease, our overall approach was to identify neurons in which an increase in Mn levels was required or sufficient to induce motor disease. We used neuron-specific Slc30a10 knockout or knockin mice as experimental models to respectively address the sufficiency and required aspects of this study. Results from the neuron-specific Slc30a10 knockout strains are presented first followed by those from the knockin mice.

To identify neurons in which an increase in Mn levels was sufficient to induce motor disease, we used neuron-specific Slc30a10 knockout mice as the experimental model because 1) neuron-specific knockout of Slc30a10 was expected to increase Mn levels in targeted neurons, and 2) motor function tests were then expected to unambiguously determine whether an increase in Mn in the targeted neurons was sufficient to induce motor deficits. Support for the above rationale came from our prior work establishing that 1) at the cellular level, the function of SLC30A10 is to mediate Mn efflux, reduce cellular Mn levels, and protect against Mn toxicity (11); 2) full-body Slc30a10 knockout mice develop Mn-dependent motor deficits (22); 3) SLC30A10 is widely expressed in the brain, including in all basal ganglia regions (striatum, globus pallidus, and substantia nigra) and the thalamus (22), which receives the basal ganglia output (6–8); and 4) pan-neuronal/glial Slc30a10 knockout mice, lacking SLC30A10 expression in all neurons, astrocytes, and oligodendrocytes, retain excess Mn in the brain and develop motor deficits under physiological conditions (i.e., in the absence of exposure to elevated Mn), implying that expression of SLC30A10 in the brain is necessary to control physiological brain Mn levels and motor function (21).

We generated dopaminergic-, GABAergic-, or glutamatergic-specific Slc30a10 knockout mice to target all major neuronal types in the basal ganglia and thalamus (6–8) by crossing floxed Slc30a10 mice with neuron-specific Cre mice [TH-Cre, GAD2-Cre, or Vglut2-Cre to respectively target dopaminergic, GABAergic, or glutamatergic neurons (25–29)]. We have extensively described the floxed Slc30a10 strain and used it previously to generate full-body and other cell-specific Slc30a10 knockout strains (20–23, 30). The neuron-specificity of the Cre strains we used is also well-established (25–29, 31). To validate region-specific depletion of SLC30A10 in the neuron-specific knockout strains, we first used qRT-PCR from brain punch tissue. In the dopaminergic-specific knockouts, SLC30A10 was depleted in the substantia nigra, which is enriched in dopaminergic neurons (6–8), but not the striatum, which is enriched in GABAergic neurons (6–8) (Fig. 1A). Conversely, in the GABAergic-specific strain, SLC30A10 levels were reduced in the striatum, but not the substantia nigra (Fig. 1B). The small effect of SLC30A10 depletion in the substantia nigra of the dopaminergic-specific knockouts is likely because the brain punch technique used to isolate brain regions cannot separate the substantia nigra pars compacta, which contains dopaminergic neurons (6–8), from the substantia nigra pars reticulata, which contains GABAergic neurons (6, 8). In the glutamatergic-specific knockouts, SLC30A10 was depleted in the thalamus, which contains glutamatergic neurons (6), but not the striatum (Fig. 1C). Depletion of SLC30A10 was also evident in the substantia nigra of the glutamatergic-specific strain (Fig. 1C), but this was expected because glutamatergic neurons are also present in this brain region (32). For additional validation, we assayed for the presence of the recombined (i.e., knockout) Slc30a10 allele in the above brain regions of the neuron-specific knockout strains and their respective controls. As expected, the recombined allele was absent in all control strains (SI Appendix, Fig. S1 A–C) and detected in the targeted regions of each knockout strain—i.e., the striatum of the GABAergic-specific; substantia nigra of the dopaminergic-specific; and thalamus of the glutamatergic-specific strain (SI Appendix, Fig. S1 A–C). The recombined allele was also detected in the substantia nigra of the GABAergic- or glutamatergic-specific knockout strain (SI Appendix, Fig. S1 B and C), but as described for the qRT-PCR results, this is consistent with the presence of GABAergic and glutamatergic neurons in substantia nigra brain punches (6, 8, 32). We could not use antibody-based approaches for further validation of SLC30A10 depletion in the knockout strains because, as we reported previously (22, 30), there are no commercial or custom-generated antibodies that specifically detect SLC30A10 in mouse tissue. However, 1) the results of the qRT-PCR and genomic DNA PCR assays were clear, rigorous, and internally consistent; 2) we have used similar techniques to characterize other Slc30a10 knockout strains previously (20–23, 30); and 3) the neuron-specificity of the Cre-expressing strains we utilized is very well-established (25–29, 31). Thus, SLC30A10 is specifically depleted in targeted regions of the neuron-specific knockout strains.

Fig. 1.

SLC30A10 expression is depleted in targeted regions of neuron-specific Slc30a10 knockout mice. (A–C) QRT-PCR analyses from indicated regions and genotypes of 2- to 4-mo-old mice. Mean expression in the control genotype for each region was normalized to 1. N = 3 per group. Mean ± SE. *P < 0.05 and n.s., not significant by two-way ANOVA and Sidak’s post hoc test.

Mn Levels Are Elevated in Targeted Brain Regions of Neuron-Specific Slc30a10 Knockout Mice.

Next, we assayed for changes in Mn levels in the neuron-specific Slc30a10 knockout strains. We performed these and subsequent behavior assays in the knockout mice under physiological conditions (i.e., without Mn exposure) because 1) the goal of the experiments with the knockout mice was to identify the neurons in which an increase in Mn levels was sufficient to induce motor disease, and 2) Mn exposure would increase Mn levels in an untargeted manner (i.e., in neurons with or without Slc30a10 knockout) and thereby confound the sufficiency test. Initial experiments using bulk brain punch tissue from dopaminergic- or GABAergic-specific knockouts and their controls did not reveal genotype-specific differences in levels of Mn (or other metals) in the basal ganglia regions or thalamus (SI Appendix, Figs. S2 and S3). This outcome was not surprising because 1) increases in Mn levels in targeted brain regions of the neuron-specific knockouts were expected to be modest, based on our prior observation that Mn levels are only modestly elevated in pan-neuronal/glial Slc30a10 knockouts under physiological conditions (21), and 2) the presence of untargeted cells (i.e., neurons or glia without Slc30a10 knockout) combined with the low absolute amount of Mn in the brain under physiological conditions (<2 ng Mn per brain punch; SI Appendix, Figs. S2 and S3) was expected to make the detection of modest group differences in bulk brain punch tissue unlikely. To address these challenges, we utilized laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS), which enhanced the spatial resolution of our assay by ~50-fold, from 500 µm in the brain punches to 10 µm with LA-ICP-MS. Notably, the LA-ICP-MS assays revealed that, compared with controls, there was a significant increase in the distribution of cells with higher Mn content in the substantia nigra of the dopaminergic-, striatum of the GABAergic-, and thalamus of the glutamatergic-specific knockout (Fig. 2 A–I). Thus, results of the LA-ICP-MS assays indicate that Mn levels are elevated in the targeted regions of the neuron-specific Slc30a10 knockout strains.

Fig. 2.

Mn levels are elevated in targeted regions of neuron-specific Slc30a10 knockout mice. (A–I) LA-ICP-MS was performed on brain sections prepared from PND 21 to 22 mice of indicated genotypes and processed for LA-ICP-MS. N=3 per group. Mn/K data were Z-score standardized within each animal, and then data from all animals per genotype combined for genotype comparison; the density (panels B, E, and H) and density proportion (panels C, F, and I) of Mn/K data are shown for genotype comparison. Kolmogorov–Smirnov test of difference of distribution between control and knockout, P < 0.0001 for A–C (19,200 to 22,500 data points per genotype); P = 0.0176 for D–F (17,400 data points per genotype); and P = 0.0478 for G–I (25,500 data points per genotype).

Dopaminergic- or Glutamatergic-, But Not GABAergic-, Specific Slc30a10 Knockouts Exhibit Motor Deficits.

Having established that SLC30A10 was depleted and Mn levels were elevated in targeted regions of the neuron-specific Slc30a10 knockout mice, we assayed for motor function of the knockouts and their controls using the balance beam and open-field tests at ~postnatal days (PND) 28 or 56, respectively. Our choice of the behavior tests and age of performance was based on our extensive prior analyses indicating that performance on these tests at these ages is a sensitive indicator of Mn-induced motor deficits (20–22, 33, 34). Mn-induced neurotoxicity increases missteps in the balance beam test and reduces movement in the open-field test (20–22, 33, 34). In the balance beam test, only the dopaminergic-specific knockout strain exhibited an increase in missteps, indicative of motor deficits (Fig. 3 A and B). Performance of the GABAergic- or glutamatergic-specific knockouts was comparable to littermates (Fig. 3 C–F). The dopaminergic-specific Slc30a10 knockouts also exhibited a hypolocomotor phenotype in multiple parameters of the open-field test (Fig. 4 A–D). Movement of the GABAergic-specific Slc30a10 knockouts was unaffected in the open-field test (Fig. 4 E–H). Additionally, the glutamatergic-specific Slc30a10 knockouts exhibited a hypolocomotor phenotype in some but not all parameters of the test (Fig. 4 I-L). The results of the behavior tests (Figs. 3 and 4), combined with those of the gene expression and metal analyses assays (Figs. 1 and 2), lead to the following conclusions: 1) Because the dopaminergic-specific Slc30a10 knockouts exhibited performance deficits in both motor tests (Figs. 3 and 4), an increase in Mn levels in dopaminergic neurons is sufficient to induce a profound deficit in motor function; 2) the motor deficits of the glutamatergic-specific knockouts in the open-field but not the balance beam test (Figs. 3 and 4) imply that loss-of-function of Slc30a10 and increase in Mn levels in glutamatergic neurons is also sufficient to induce motor deficits, but the more subtle phenotype of the glutamatergic-specific Slc30a10 knockouts suggests that glutamatergic Mn toxicity may induce motor dysfunction via a secondary dopaminergic insult (Discussion); and 3) the lack of a motor phenotype in the GABAergic-specific Slc30a10 knockouts (Figs. 3 and 4) indicates that an increase in Mn levels in GABAergic neurons is not sufficient for the onset of Mn-induced motor disease. In totality, the assays with the neuron-specific Slc30a10 knockout strains indicate that loss-of-function of Slc30a10 and an increase in Mn levels in dopaminergic or glutamatergic, but not GABAergic, neurons is sufficient to induce motor deficits, and motor defects are more pronounced with Slc30a10 depletion and Mn level elevations in dopaminergic neurons than in glutamatergic neurons.

Fig. 3.

Dopaminergic-, but not GABAergic- or glutamatergic-, specific Slc30a10 knockouts exhibit performance deficits in the balance beam test. (A–F) Average or total missteps in the balance beam test in PND 28 to 29 mice of indicated genotypes. For A and B, N = 11 controls (6 males and 5 females) and 12 dopaminergic-specific Slc30a10 knockouts (7 males and 5 females). For C and D, N = 9 controls (3 males and 6 females) and 9 GABAergic-specific Slc30a10 knockouts (4 males and 5 females). For E and F, N = 10 controls (6 males and 4 females) and 10 glutamatergic-specific Slc30a10 knockouts (4 males and 6 females). Mean ± SE. *P < 0.05 and n.s., not significant by the t test.

Fig. 4.

Dopaminergic- or glutamatergic-, but not GABAergic-, specific Slc30a10 knockouts exhibit a hypolocomotor phenotype on the open-field test. (A–L) Open-field activity data in the 6 to 15 min interval in ~PND 56 mice of indicated genotypes. For A–D, N= 15 controls (9 males and 6 females) and 15 dopaminergic-specific Slc30a10 knockouts (8 males and 7 females). For E–H, N= 15 controls (7 males and 8 females) and 15 GABAergic-specific Slc30a10 knockouts (9 males and 6 females). For I–L, N= 15 controls (9 males and 6 females) and 15 glutamatergic-specific Slc30a10 knockouts (8 males and 7 females). Mean ± SE. *P < 0.05 and n.s., not significant by the t test.

Development of a Cell-Specific Slc30a10 Knockin Strategy in Mice.

To complement the sufficiency experiments performed using the neuron-specific Slc30a10 knockout mice, we sought to identify the neurons in which an increase in Mn levels was necessary to induce motor disease. For this, we hypothesized that neuron-specific knockin (i.e., overexpression) of Slc30a10 1) would attenuate the increase in Mn levels after Mn exposure in targeted neurons, and 2) this attenuation may rescue Mn-induced motor deficits. To test this hypothesis, we developed a conditional Slc30a10 knockin mouse strain in which the open reading frame of mouse Slc30a10 gene was inserted into the ROSA 26 locus downstream of the constitutive EF1a promoter (Fig. 5A). A floxed stop codon was placed between the promoter and the Slc30a10 open reading frame so that knockin gene expression occurred only after Cre-mediated recombination (Fig. 5A). Before generating neuron-specific Slc30a10 knockin mice, it was necessary to validate the specificity of the Slc30a10 knockin model at the organ-level. Therefore, we generated pan-neuronal/glial Slc30a10 knockin mice by crossing the floxed Slc30a10 knockin strain with Nestin-Cre mice, which induces recombination in all neurons, astrocytes, and oligodendrocytes (35, 36). QRT-PCR analyses revealed that SLC30A10 was overexpressed in the brain, but not the liver, of the pan-neuronal/glial Slc30a10 knockins (Fig. 5B). Furthermore, in the absence of treatment with elevated Mn, brain Mn levels of the pan-neuronal/glial knockin mice were comparable to littermate controls (Fig. 5C). But, after acute exposure to Mn, the increase in brain Mn levels of the pan-neuronal/glial knockins was attenuated (Fig. 5C). In contrast, there was no effect of Slc30a10 knockin genotype on blood Mn levels (Fig. 5D). Thus, knockin of Slc30a10 in the whole brain specifically increases brain SLC30A10 expression and reduces brain Mn accumulation after Mn exposure, providing the justification to proceed with generation and analyses of neuron-specific Slc30a10 knockin strains.

Fig. 5.

Pan-neuronal/glial Slc30a10 knockin mice exhibit a brain-specific increase in SLC30A10 expression and reduction in Mn levels after Mn exposure. (A) Gene targeting strategy for development of conditional Slc30a10 knockin mice. (B) QRT-PCR from the midbrain or liver of 6-wk-old mice of indicated genotypes. Mean expression in control mice for each organ was separately normalized to 1. N = 6 per group. Mean ± SE. *P < 0.05 and n.s., not significant by two-way ANOVA and Sidak’s post hoc test. (C and D) ICP-MS from the midbrain and blood of mice of indicated genotypes treated with 0 or ~4 mg Mn/kg subcutaneously 3-times a week starting at 3-wk of age until analyses at ~6-wk of age. Mean ± SE. N = 8 vehicle-treated controls, 6 vehicle-treated knockins, 8 Mn-treated controls, and 4 Mn-treated knockins. *P < 0.05 and n.s., not significant by two-way ANOVA and Sidak’s post hoc test.

SLC30A10 Is Overexpressed in Targeted Regions of Neuron-Specific Slc30a10 Knockin Mice.

To target all major neuronal types in the basal ganglia and thalamus (6–8) and complement the studies with the knockout strains, we crossed the floxed Slc30a10 knockin mice with TH-Cre, GAD2-Cre, or Vglut2-Cre strains and generated dopaminergic-, GABAergic-, or glutamatergic-specific Slc30a10 knockin mice. QRT-PCR analyses confirmed that, compared with littermate controls, SLC30A10 expression was elevated in the targeted regions of the neuron-specific knockin strains—i.e., in the substantia nigra, but not the striatum of the dopaminergic-specific Slc30a10 knockins (Fig. 6A); in the striatum, but not the substantia nigra of the GABAergic-specific Slc30a10 knockins (Fig. 6B); and in the thalamus, but not the striatum of the glutamatergic-specific Slc30a10 knockins (Fig. 6C). The relatively modest effect of SLC30A10 elevation in the substantia nigra of the dopaminergic-specific knockins (Fig. 6A) is likely a consequence of the presence of untargeted cells in the substantia nigra brain punch that includes both the pars compacta and pars reticulata, as described for the knockouts. Furthermore, an increase in SLC30A10 expression was evident in the substantia nigra of the glutamatergic-specific Slc30a10 knockins (Fig. 6C), but also as described for the knockouts, this is consistent with the presence of glutamatergic neurons in this region (32). Collectively, the expression studies confirm the specificity of SLC30A10 overexpression in the targeted regions of the neuron-specific Slc30a10 knockin strains.

Fig. 6.

SLC30A10 expression is elevated in targeted regions of neuron-specific Slc30a10 knockin mice. (A–C) QRT-PCR analyses from indicated regions and genotypes of 3- to 4-wk-old mice. Mean expression in the control genotype for each region was normalized to 1. For A, N = 3 per group. For B and C, N = 4 per group. Mean ± SE. *P < 0.05 and n.s., not significant by two-way ANOVA and Sidak’s post hoc test.

Increases in Mn Levels After Mn Exposure Are Attenuated in Targeted Regions of Neuron-Specific Slc30a10 Knockin Mice.

After this, we assayed for changes in Mn levels in the basal ganglia and thalamus of the neuron-specific Slc30a10 knockin mice with or without exposure to an extensively characterized oral Mn regimen that models environmental Mn exposure in humans (4, 20, 21, 33, 34). We performed initial experiments using bulk brain punch tissue and subsequently utilized LA-ICP-MS, similar to the approach for the neuron-specific knockout mice. The rationale for starting with the brain punch samples, despite the likelihood that the presence of untargeted cells may make detection of group differences challenging, was that 1) the increase in tissue Mn levels that occurs after Mn treatment may amplify and thus enable detection of at least some group differences, and 2) outcomes may help focus the challenging LA-ICP-MS studies. In the absence of Mn exposure, there were no genotype differences in Mn levels in the basal ganglia or thalamus (SI Appendix, Fig. S4 A–L). After Mn exposure, the GABAergic-specific Slc30a10 knockins exhibited a trending attenuation of Mn accumulation in the targeted, but not untargeted, regions (i.e., in the striatum and globus pallidus, but not thalamus or substantia nigra) (SI Appendix, Fig. S4 E–H). We did not detect other trending or significant genotype differences in levels of Mn in the other neuron-specific knockin strains (SI Appendix, Fig. S4 A–D and I–L) or significant changes in the levels of other metals in all three strains (SI Appendix, Figs. S5–S7). Subsequently, we performed LA-ICP-MS assays in Mn-exposed neuron-specific knockin or control mice (we limited the LA-ICP-MS analyses to the Mn exposed condition because the bulk tissue assays suggested that genotype differences were unlikely to be detected in the absence of Mn exposure) (Fig. 7 A–I). Compared with controls, there was a significant increase in the distribution of cells with lower Mn content in the substantia nigra of the dopaminergic-specific knockins or thalamus of the glutamatergic-specific knockins (Fig. 7 A–C, and G–I). Additionally, although the overall distribution of Mn content of cells in the striatum of control and GABAergic-specific Slc30a10 knockins did not statistically differ, a subset of striatal cells in the GABAergic-specific knockins still had a lower Mn content than controls (Fig. 7 D–F). Overall, the results of the LA-ICP-MS and bulk tissue assays imply that, after Mn exposure, the accumulation of Mn in the targeted regions of the neuron-specific Slc30a10 knockin mice is attenuated.

Fig. 7.

Accumulation of Mn after Mn exposure in targeted regions of neuron-specific Slc30a10 knockin mice is attenuated. (A–I) LA-ICP-MS was performed on brain sections prepared from PND 20 to 21 mice of indicated genotypes and processed for LA-ICP-MS. N = 3 per group. Mn/K data were Z-score standardized within each animal, and then data from all animals per genotype combined for genotype comparison; the density (panels B, E, and H) and density proportion (panels C, F, and I) of Mn/K data are shown for genotype comparison. Kolmogorov–Smirnov test of difference of distribution between control and knockin, P = 0.0001 for A–C (14,400 data points per genotype); P > 0.05 for D–F (17,400 data points per genotype); and P < 0.0001 for G–I (25,500 data points per genotype).

Dopaminergic- or Glutamatergic-, But Not GABAergic-, Specific Slc30a10 Knockin Mice Are Protected Against Mn-Induced Motor Deficits.

We then assayed for motor function of the vehicle- or Mn-exposed neuron-specific Slc30a10 knockin mice and their littermate controls. As expected, on the balance beam test, Mn exposure increased missteps in the littermate controls of all the knockin strains (Fig. 8 A–F). Notably, the dopaminergic- or glutamatergic-specific Slc30a10 knockin strains were completely protected against the Mn-induced increase in missteps while missteps in Mn-exposed GABAergic-specific Slc30a10 knockins were comparable to Mn-treated littermates (Fig. 8 A–F). Mn treatment did not produce striking deficits in the open-field test in the littermate controls (SI Appendix, Figs. S8 and S9), likely because the dose of Mn used in this study in the preweaning period was low, which precluded further analyses of rescue effects in the neuron-specific Slc30a10 knockin mice. Despite this, the results of the balance beam test were unambiguous and indicated that dopaminergic- or glutamatergic-, but not GABAergic-, specific Slc30a10 knockins were protected against Mn-induced motor disease. Put together, the results of the gene expression, metal analyses, and behavior assay indicate that overexpression of SLC30A10 and attenuation of the Mn-exposure induced increase in Mn levels in dopaminergic or glutamatergic, but not in GABAergic, neurons protects against Mn-induced motor deficits. Thus, after Mn exposure, an increase in Mn levels in dopaminergic and glutamatergic, but not GABAergic, neurons is required for Mn-induced motor disease.

Fig. 8.

Dopaminergic- or glutamatergic-, but not GABAergic-, specific Slc30a10 knockins are protected against Mn-induced deficits in the balance beam test. (A–F) Average or total missteps on the balance beam test in PND 28 mice of indicated genotypes and Mn treatment conditions. For A and B, N=13 vehicle-treated controls (6 males and 7 females), 11 vehicle-treated knockins (8 males and 3 females), 11 Mn-treated controls (6 males and 5 females), and 10 Mn-treated knockins (2 males and 8 females). For C and D, N = 8 vehicle-treated controls (5 males and 3 females), 11 vehicle-treated knockins (6 males and 5 females), 8 Mn-treated controls (2 males and 6 females), and 10 Mn-treated knockins (7 males and 3 females). For E and F, N = 9 vehicle-treated controls (4 males and 5 females), 9 vehicle-treated knockins (4 males and 5 females), 9 Mn-treated controls (6 males and 3 females), and 9 Mn-treated knockins (4 males and 5 females). Mean ± SE. *P < 0.05 and n.s., not significant by two-way ANOVA and Tukey’s post hoc test.

Critical Aspects of Dopaminergic Neurotransmission Are Altered in Dopaminergic-Specific Slc30a10 Knockout Mice.

The analyses of the neuron-specific Slc30a10 knockout or knockin mice revealed that an increase in Mn levels in dopaminergic and/or glutamatergic neurons is necessary and sufficient for the onset of Mn-induced motor disease. As the final part of this study, we sought to determine whether the increase in Mn levels in the target neurons was sufficient to produce molecular and neurochemical changes that related to the motor deficits. We focused these studies on the dopaminergic, instead of the glutamatergic, system because of the centrality of altered dopaminergic neurotransmission to Parkinson’s disease (6–8); prior findings indicative of dopaminergic dysfunction in Mn-induced motor disease (21, 37–39) (these prior studies did not determine whether the dopaminergic dysfunction was caused by direct toxicity of Mn in dopaminergic neurons or reflective of Mn effects in other cell types); the likelihood that glutamatergic toxicity may impact motor function through effects on dopaminergic neurons (e.g., excitotoxicity) (40); and the expectation that analyses of the direct effects of elevated dopaminergic Mn on dopaminergic neurotransmission would be more interpretable than of the indirect effects of elevated glutamatergic Mn on the dopaminergic system. We used dopaminergic-specific Slc30a10 knockout mice and their littermate controls, instead of the corresponding Slc30a10 knockin strain, so that we could determine the minimal effects of increasing dopaminergic neuronal Mn levels on dopamine dynamics and perform the experiment without Mn exposure, which would be necessary if the knockin mice were used and which could confound interpretation due to effects on nondopaminergic neurons.

First, we assayed for changes in the expression or localization of critical modulators of dopaminergic neurotransmission in the dopaminergic-specific Slc30a10 knockouts using immunofluorescence and qRT-PCR. Microscopy assays revealed that, compared with littermate controls, in the dopaminergic-specific Slc30a10 knockouts there was a significant decrease in VMAT2 levels in the substantia nigra pars compacta, but not the striatum, and no change in DAT or TH levels in either brain region (SI Appendix, Fig. S10 A–L). We verified that VMAT2 and TH signals overlapped in both brain regions (SI Appendix, Figs. S10A and S11). There were no genotype differences in mRNA levels of assayed genes (SI Appendix, Fig. S12 A–G). Prior work, including in pan-neuronal/glial Slc30a10 knockout mice, has established that Mn-induced motor disease is characterized by a decrease in stimulated extracellular striatal dopamine levels without degeneration of dopaminergic neurons (21, 37–39, 41, 42) (as mentioned in the preceding paragraph, the prior studies did not identify the neuronal type targeted by Mn to induce the dopamine release deficits). Results of the current microscopy assays were essentially consistent with the prior studies—in the dopaminergic-specific Slc30a10 knockouts, the decrease in VMAT2 levels raised the possibility of alterations in dopamine release/cycling while the unchanged TH levels implied the absence of dopaminergic neurodegeneration. Therefore, as the next step, we directly assayed for baseline and stimulated extracellular dopamine levels in the dorsal striatum of the dopaminergic-specific Slc30a10 knockouts or controls using in vivo microdialysis with probe placement in the dorsal striatum (Fig. 9A and SI Appendix, Fig. S13 A–D). However, we did not detect genotype differences in baseline or high K⁺-evoked extracellular dopamine levels using 50 or 100 mM K⁺ for stimulation (Fig. 9 B and C). As a control, we performed the microdialysis assay in the GABAergic-specific Slc30a10 knockout strain or its littermates and did not identify any genotype differences (Fig. 9D). We also used a no-net-flux approach to more accurately measure baseline dopamine levels in the dopaminergic-specific Slc30a10 knockouts or littermates, but again did not identify any genotype differences (Fig. 9 E and F). Because the lack of a genotype effect in the microdialysis experiment could be due to the low temporal resolution of the approach (samples are collected over 20-min intervals) or overstimulation of the dopaminergic terminals (the high K⁺ stimulation occurs for 10 min), we used optogenetics as a more sensitive measure of dopamine release. In dopaminergic-specific Slc30a10 knockouts or Cre-expressing controls, we stereotactically injected an adeno-associated virus (AAV1) containing a Cre-dependent expression cassette for the red light-sensitive ion channel, Chrimson (AAV1-CAG-FLEX-Chrimson-tdTomato) into the substantia nigra pars compacta (Fig. 10 A and B) and an AAV containing the nonconditional expression cassette for the dopamine sensor GRAB-DA (AAV1-hSyn-GRAB-gDA3m) into the dorsal striatum (Fig. 10 A and B and SI Appendix, Fig. S14). Chrimson stimulation of substantia nigra pars compacta cell bodies evoked dopamine release in the dorsal striatum that increased with stimulus duration and stimulation frequency in control mice, but that was not observed in dopaminergic-specific Slc30a10 knockout mice (Fig. 10 C–F). Overall, in the dopaminergic-specific Slc30a10 knockouts, protein expression of a critical regulator of dopamine cycling was reduced and a deficit in dopamine release was evident by optogenetics, indicating that an increase in Mn levels in dopaminergic neurons is sufficient to induce deficits in dopamine neurotransmission that are consistent with the motor phenotype of Mn neurotoxicity.

Fig. 9.

Stimulated and baseline striatal dopamine levels of dopaminergic-specific Slc30a10 knockouts assayed by in vivo microdialysis are comparable to controls. (A) Schematic of probe placement. DS—dorsal striatum. (B–D) In vivo microdialysis to measure stimulated dopamine levels in the dorsal striatum of 2- to 4-mo-old mice of indicated genotypes using indicated KCl concentrations for the high K⁺ pulse. For B, N = 3 controls and 5 dopaminergic-specific Slc30a10 knockouts. For C, N = 3 per group. For D, N = 6 controls and 5 GABAergic-specific Slc30a10 knockouts. There were no differences between groups using repeated measures two-way ANOVA and Sidak’s post hoc test. (E and F). No-net-flux in vivo microdialysis to measure baseline dopamine levels of 3- to 4-mo-old dopaminergic-specific Slc30a10 knockouts or controls with regression lines (E) and x-intercept of regression lines (F). N = 3 per group. Mean ± SE. For F, n.s., not significant by the t test.

Fig. 10.

Dopaminergic-specific Slc30a10 knockouts exhibit a profound deficit in stimulated striatal dopamine release assayed by optogenetics. (A) Schematic of viral injection and fiber placements for AAV1-GRAB-DA and AAV1-FLEX-Chrimson-tdTomato in dopaminergic-specific Slc30a10 knockout or control (TH-Cre) mice. DS—dorsal striatum; SNc—substantia nigra pars compacta. (B) Example histology for Chrimson expression in the SNc and GRAB-DA expression in the DS of control or knockout mice. Fiber placements are highlighted by dashed lines. Scale bar, 100 µm. (C and D) GRAB-DA fluorescence following Chrimson stimulation of SNc dopamine neurons for 1s (C) or 3s (D) with increased stimulation frequency. (E and F) Area under curve analysis of GRAB-DA fluorescence following Chrimson stimulation for 1s (E) or 3s (F) with increased stimulation frequency. N = 4 control and 5 knockout mice for C–F. Statistical analyses were as follows: E, two-way repeated measures ANOVA; stimulus × genotype interaction F_(3,21) = 4.003, P = 0.0212; Bonferroni post hoc **P = 0.0098. F, two-way repeated measures ANOVA; stimulus × genotype interaction F_(3,21) = 12.17, P < 0.0001; Bonferroni post hoc ****P < 0.0001. Data are presented as mean ± SE.

Discussion

The major outcome of this study is to identify the neuronal targets of Mn-induced motor disease. Prior studies variously attributed the disease to the effects of elevated Mn on dopaminergic, GABAergic, or glutamatergic neurons in the basal ganglia (15, 41, 43). But, an inability to selectively increase Mn levels in specific neurons and assay for neurological sequelae precluded definitive identification of the neurons targeted by elevated Mn to induce motor disease. We generated neuron-specific Slc30a10 knockout or knockin mice as models to selectively modulate Mn levels in target neurons and identify the neuronal targets of elevated Mn. Despite expected changes in SLC30A10 expression and Mn levels in targeted regions of the neuron-specific Slc30a10 strains, there were striking strain-specific differences in the motor phenotypes. In the absence of Mn exposure, motor deficits were evident in the dopaminergic- or glutamatergic-specific Slc30a10 knockouts (albeit to varying degrees), but not in the GABAergic-specific Slc30a10 knockouts, indicating that an increase in Mn levels in dopaminergic or glutamatergic, but not GABAergic, neurons is sufficient to induce motor deficits. Conversely, after Mn exposure, dopaminergic- or glutamatergic-, but not GABAergic-, specific Slc30a10 knockins were protected against Mn-induced motor deficits, indicating that an increase in Mn levels in dopaminergic and glutamatergic, but not GABAergic, neurons is necessary to induce motor deficits. Thus, dopaminergic and glutamatergic neurons are the neuronal targets of elevated Mn responsible for Mn-induced motor disease.

There is an important difference in the sufficiency and requirement outcomes of our study—to induce motor disease, an increase in Mn levels 1) in either dopaminergic or glutamatergic neurons is sufficient (results of the neuron-specific Slc30a10 knockout mice); but 2) in both dopaminergic and glutamatergic neurons is required (results of the neuron-specific Slc30a10 knockin mice). This difference is likely reflective of the differing modes of brain and neuronal Mn accumulation in the two experimental models. In the neuron-specific Slc30a10 knockouts, SLC30A10 depletion and subsequent targeted elevation of neuronal Mn levels likely begin during embryonic development and persist thereafter. The effects of elevated Mn in either glutamatergic or dopaminergic neurons (i.e., in only one neuronal type) in this model are likely sufficiently pronounced to produce motor deficits, although with varying degrees of severity—i.e., more pronounced deficits with elevation of Mn in dopaminergic neurons than in glutamatergic neurons. In contrast, the postnatal Mn exposure in the neuron-specific Slc30a10 knockin mice may induce more modest downstream toxicity in dopaminergic and glutamatergic neurons such that elevations of Mn in both dopaminergic and glutamatergic neurons are necessary and equally important for the onset of motor deficits. Thus, the relative contribution of dopaminergic and glutamatergic Mn toxicity to the motor phenotype may depend on the context of the Mn exposure and brain Mn accumulation.

In addition to identifying the neuronal targets of Mn-induced motor disease, our results establish the neuron-specific Slc30a10 mouse strains as ideal model systems for defining underlying molecular mechanisms of disease. This conclusion emerges from the observation that dopaminergic-specific Slc30a10 knockouts exhibited a reduction in VMAT2 expression and an inhibition of light-stimulated striatal dopamine release, indicating that an increase in Mn levels in dopaminergic neurons is sufficient to alter critical aspects of dopaminergic neurotransmission. While our experiments in isolation cannot attribute causality to these changes for the motor phenotype of the dopaminergic-specific Slc30a10 knockouts, available evidence suggests that the deficits we observed are likely important for the onset of motor dysfunction. Reduced function of VMAT2 is causally linked to parkinsonism (VMAT2 transports dopamine into vesicles and reduced VMAT2 function may increase cytosolic dopamine, which may induce neurotoxicity by producing reactive intermediates) (44, 45). A Mn-induced decrease in VMAT2 protein levels was also recently observed in a dopaminergic cell model (46). Furthermore, reductions in striatal dopamine levels are the predominant driver of the parkinsonian phenotype (6–8); deficits in stimulated striatal dopamine release have been previously reported in other models of Mn-induced motor deficits (e.g., pan-neuronal/glial Slc30a10 knockouts, full-body Slc39a14 knockouts, and Mn-treated rats) (21, 37–39); and treatment with the dopamine reuptake inhibitor methylphenidate, which increases extracellular dopamine levels in the striatum (47), has been reported to rescue fine motor deficits of Mn-exposed rats (39). To obtain a comprehensive mechanistic understanding of Mn-induced motor disease, critical next steps are 1) establishing whether there is a causal relationship between the decrease in VMAT2 expression, inhibition of dopamine release and motor dysfunction in dopaminergic-specific Slc30a10 knockouts; 2) defining the molecular mechanisms of these changes—possibilities include specific changes induced by elevated Mn in protein trafficking (48–50), transcription and metabolism (33, 34, 51–53), and as described above, increased cytosolic dopamine accumulation and subsequent toxic sequelae (44, 45); 3) more broadly characterizing the dopaminergic dysfunction of the dopaminergic-specific Slc30a10 strains; and 4) using similar approaches to analyze crosstalk between glutamatergic and dopaminergic neurotransmission in glutamatergic-specific Slc30a10 strains—e.g., excitotoxicity and glutamate control of striatal dopamine release (40, 54, 55). The neuron-specific Slc30a10 strains developed in this study are essential for this entire line of future investigation because, in these models, detected molecular and neurochemical changes will be directly attributable to Mn level alterations in targeted neurons and not confounded by Mn effects in nontargeted neurons.

There is a notable difference in the dopamine release deficit in the dopaminergic-specific Slc30a10 knockouts and prior work on other model systems of Mn neurotoxicity. Microdialysis assays with direct depolarization of the striatum were sufficient to detect deficits in striatal dopamine release in pan-neuronal/glial Slc30a10 knockouts, full body Slc39a14 knockouts, and Mn-treated rats (21, 37–39). In contrast, in the dopaminergic-specific Slc30a10 knockouts, the deficit in dopamine release was only evident using optogenetics and stimulation of the dopaminergic cell bodies but not using microdialysis and striatal depolarization. The difference between the prior work (21, 37–39) and our current findings may be a consequence of the increase in Mn levels in nondopaminergic cells (e.g., glutamatergic neurons or glial cells) in the experimental models used in the previous studies that may have exacerbated the dopamine release deficits. Thus, it is possible that elevated Mn induces a dose-dependent continuum of dopamine release deficits with lower levels of Mn elevation inducing subtle deficits in dopamine release that require techniques more sensitive than microdialysis for detection and higher Mn levels producing more profound defects detectable by microdialysis. In this context, resolving whether the inhibition of the light-stimulated dopamine release in the dopaminergic-specific Slc30a10 knockouts is reflective of a deficit in the ability of action potentials initiating in the dopaminergic cell-body to induce neurotransmitter release in the nerve terminal or an inability of the dopaminergic neurons to respond to light stimulation itself is an important issue for future investigation that should provide substantial insights into the nature of the dopaminergic dysfunction induced by elevated Mn and its relationship to motor disease.

In a wider context, our study also establishes the feasibility of using Cre/LoxP-mediated neuron-specific efflux transporter knockout or knockin as a strategy to selectively alter metal levels in targeted neurons and define the contributions of specific neuronal populations to metal-induced neurological disease. This approach should be applicable to studying motor diseases caused by elevated Cu or Fe because the Cu efflux transporters ATP7A and ATP7B and the Fe efflux transporter Fpn1 are expressed in neurons in the brain (56, 57). Because Cre-mediated recombination will not be limited to only brain regions of interest and may also occur in neurons that are not necessarily being targeted (e.g., the TH-Cre line we used to target dopaminergic neurons also induces recombination in noradrenergic neurons) (25), careful experimental design and data interpretation will be integral to the general use of this approach. For our study, it is highly unlikely that any recombination of Slc30a10 outside of the targeted neurons in the basal ganglia/thalamus that may have occurred influenced our conclusions because 1) the basal ganglia and thalamus play a predominant role in controlling motor function (6–8); 2) Mn has well-established effects on these brain regions (4, 15, 21, 22, 41, 43); and 3) the neuron-specific Slc30a10 knockouts or knockins exhibited changes in gene expression, metal levels, and dopamine dynamics in the basal ganglia and thalamus that directly relate to motor activity. In conclusion, using neuron-specific knockout or knockin of Slc30a10 in mice, we identify the neuronal targets of Mn-induced motor disease. These results provide fundamental insights into the biology of Mn-induced motor dysfunction and a widely applicable approach to study the mechanisms of neurological disease caused by elevated levels of essential metals in the brain.

Methods

Methodological details are provided in the SI appendix. We have previously described the Slc30a10 floxed knockout strain (20–23, 30). We generated a new conditional floxed Slc30a10 knockin line for this study in which the open reading frame of mouse Slc30a10 was integrated into the Rosa 26 locus downstream of the EF1a promoter and a floxed stop codon. Neuron-specific strains were generated by crossing the floxed knockout or knockin strains with TH-Cre, GAD2-Cre, or Vglut2-Cre strains obtained from The Jackson Laboratories (25–29). We reported preliminary phenotypic analyses of some of the neuron-specific Slc30a10 knockout strains in a preprint (58). The Nestin-Cre strain (35, 36), which we have previously described (21, 22), was used to generate the pan-neuronal/glial Slc30a10 knockins. QRT-PCR (primers in SI Appendix, Table S1), genomic DNA PCR, bulk tissue ICP-MS from brain punches, balance beam and open-field tests, in vivo microdialysis, and immunofluorescence analyses were performed essentially as described previously (20–23, 30, 33, 34, 59). Mn exposure was using a subcutaneous regimen in the pan-neuronal/glial Slc30a10 knockins and littermates (22) or a well-characterized oral regimen that models human environmental Mn exposure in the neuron-specific Slc30a10 knockins and littermates (4, 20, 21, 33, 34, 59). Detailed methodology for LA-ICPMS is in the SI appendix. Virus production, injection, fiber photometry, and data analyses for optogenetics were performed using standard techniques (60, 61) described in the SI appendix. Statistical approaches are indicated in the respective figure legends. Both sexes were included, and sexes were combined for analyses because of the lack of obvious sex-specific differences. P < 0.05 was considered statistically significant.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.