Conditional disruption of AMP kinase in dopaminergic neurons promotes 1 Parkinson ’ s disease-associated phenotypes in mice 2 3


 Emerging studies implicate energy dysregulation as an underlying trigger for Parkinson’s disease (PD), suggesting that a better understanding of the molecular pathways governing energy homeostasis could help elucidate therapeutic targets for the disease. A critical cellular energy regulator is AMP kinase (AMPK), which we have previously shown to be protective in PD. However, precisely how AMPK function impacts on dopaminergic neuronal survival and disease pathogenesis remains elusive. Here, we created a tissue-specific AMPK-knockout mouse model where the catalytic subunits of AMPK are ablated in nigral dopaminergic neurons. Using this model, we demonstrated that loss of AMPK function promotes dopaminergic neurodegeneration and associated locomotor aberrations. Accompanying this is a substantial reduction in the number of mitochondria in the surviving AMPK-deficient nigral dopaminergic neurons, suggesting that an impairment in mitochondrial biogenesis may underlie the observed PD-associated phenotypes. Importantly, the loss of AMPK function enhances the susceptibility of nigral dopaminergic neurons in these mice to 6-hydroxydopamine-induced toxicity. Taken together, these findings highlight the importance of neuronal energy homeostasis by AMPK in PD and position AMPK pathway as an attractive target for future therapeutic exploitation.

Mice behavioral assessment. All animals were transported to the animal behavioral room for 149 habituation to the novel room at least 30 minutes prior to experiments. Locomotor activity 150 recordings were carried out in an opaque open field arena (45 x 45 cm). The testing apparatus 151 was enclosed in a ventilated, quiet and dim procedure room to minimize animals' stress levels 152 while allowing normal exploratory behavior. Animals were introduced into the arena and 153 allowed to explore freely for 10 minutes. Exploratory behavior was recorded using a 154 monochrome industrial camera (DMK 22AUC03, The Imaging Source, Taiwan) and 155 locomotor activity data was collected using the ANY-maze video tracking system. Motor 156 performance assays were also carried out using a rotarod as previously described (7). 157 158 Short Report 7 Generation and characterization of 6-OHDA lesion PD model and apomorphine-induced 159 rotation assay. 7 AMPK-WT or 13 AMPK-cKO mice were lesioned with 1.5 µl of 5 µg/µL 160 6-hydroxydopamine (6-OHDA) using a method previously described (7). For every week post-161 surgery, up to five weeks, the apomorphine-induced rotation assay was carried out and the 162 number of rotations made in one minute was counted and recorded. 4 to 5-month-old wild-type 163 C57B6 male mice injected with 6-OHDA were sacrificed via cervical dislocation. Mice brains 164 were harvested and processed for western blot analysis as previously described (4). 165 166 Immunohistochemistry, Stereological analysis of TH-positive neurons and mitochondrial 167 analysis. Mice brains were processed for immunohistochemistry and stereological analysis as 168 previously described (7). For mitochondrial analysis, mice brains were sectioned in the coronal 169 plane at 10 µm thickness. Brain sections containing the SNpc were washed with washing buffer 170 (0.1% Triton X-100/PBS) before they were blocked with blocking buffer (5% NGS in 0.1% 171 Triton X-100/PBS) for one hour at room temperature. Brain sections were then incubated in 172 anti-TH and anti-TOM20 antibodies in blocking buffer overnight at 4°C and washed with 173 washing buffer the next day. Following which, brains sections were incubated in secondary 174 fluorescent antibodies at room temperature for four hours and washed with washing buffer 175 before being mounted on glass slides with Vectorshield mounting media containing DAPI. 176 Slides were viewed under the Olympus Fluoview FV1000 Laser Scanning Confocal 177 Microscope. Confocal images of 20 stained mice brain sections (~1000 TH-positive cells) per 178 group were analysed using a MATLAB algorithm that was modified from (8). 179 180 Short Report 8 Statistical analysis. Statistical analyses were performed using the student's two-tailed 181 unpaired t-test (*p<0.05, **p<0.001 or unless otherwise stated). All data was expressed as 182 mean (S.D.) generated from at least three independent experiments unless otherwise stated.

AMPKa expression is abolished in TH-expressing cells of AMPK-cKO mice brains 206
We used the Cre/loxP-mediated recombination strategy to conditionally target the 207 Prkaa1/2 loci in the mouse brain. AMPK floxed mice (Prkaa1/2 f/f ) in which the sequence 208 encoding the catalytic site of both of the α subunits was flanked by loxP sequences has 209 previously been described (9). To obtain DA neuronal-specific disruption of AMPK activity, 210 Prkaa1/2 f/f mice were crossed with TH-Cre transgenic mice expressing Cre recombinase under 211 the control of the tyrosine hydroxylase promoter, to create TH-Cre; Prkaa1/2 f/f (Prkaa1/2 f/f -212 TH-cKO) mice (herein refers to as AMPK cKO mice). Using PCR, we verified the presence of 213 Prkaa1/2 f/f in AMPK-WT mice and presence of both TH-Cre and Prkaa1/2 f/f in AMPK-cKO 214 mice (Fig. 1A). We further ascertained the deletion of AMPKa subunits in TH-expressing cells 215 via immunofluorescence staining of AMPK-WT and AMPK-cKO mice brain sections using 216 antibodies specific to AMPK-α1 and AMPK-α2 subunits. In both the nigral and striatal regions 217 of AMPK-cKO mice, the expression of AMPK-α1 is significantly reduced (p=0.0000222) (Fig. 218 1B-C and E). Unlike the predominantly cytoplasmic AMPK-α1 species, AMPK-α2 is normally 219 localized to the nucleus (10). Thus, we examined AMPK-α2 expression mainly in the SNpc 220 DA neurons where the cell bodies reside. We similarly recorded a significant reduction in 221 AMPK-α2 expression in AMPK-cKO compared to AMPK-WT mice (p=0.00161) ( Fig. 1D-E). 222 Whereas dual knockout of the genes encoding AMPK-α1 and AMPK-α2 in the germline is 223 embryonic lethal (11), AMPK-cKO mice do not exhibit developmental defects, gross 224 abnormalities, or accelerated mortality (not shown). 225

AMPK-cKO mice have reduced dopaminergic neuronal count and exhibit locomotor 228 deficits compared to age-matched AMPK-WT mice 229
Working on the premise that AMPK is critical for DA neuronal survival, we anticipate 230 that AMPK deficiency predisposes DA neurons toward degeneration. As early as three-231 months-old, AMPK-cKO mice exhibit reduced TH-positive staining of the neurites in the 232 striatum and the cell bodies in the SNpc ( Fig. 2A). This difference is significant when 233 quantified using an unbiased stereology-based approach (p=0.00369) and persists even after 234 aging to 22 months old (p=0.00641) (Fig. 2B). As SNpc DA neurons are predominantly 235 responsible for modulating locomotion, we anticipate that the loss of these neurons in the 236 AMPK-cKO mice will result in locomotor deficits, which are characteristic of mice with PD 237 (12). The locomotor function of the AMPK-cKO mice and their wild-type counterparts were 238 assessed using the open field and rotarod tests at various experimental time points (3, 8, 12 and 239 22 months old) in order to account for age-associated changes. In general, AMPK-cKO mice 240 travel much less than AMPK-WT mice in the open field, as evident from the representative 241 track plots (Fig. 2C). At three months old, the total distance travelled horizontally by the 242 AMPK-cKO mice within the arena is significantly reduced when compared to age-matched 243 AMPK-WT mice (p=0.00129) (Fig. 2C). Likewise, vertical rearing activity in the arena is also 244 markedly reduced in the AMPK-cKO mice when compared to AMPK-WT mice and this 245 difference persists through age (p=0.0253 at 3 months; p=0.00108 at 8 months; p=0.0101 at 12 246 months; p=0.0229 at 22 months) (Fig. 2D). This is further supported by observations from the 247 rotarod assay, a commonly used behavioural assessment for motor coordination, where 248 AMPK-cKO mice falls off the accelerating rod more readily than AMPK-WT mice from eight 249 to 22 months old (p=0.0000422 at 8 months; p=0.0120 at 12 months; p=0.0408 at 22 months) 250 ( Fig. 2E), suggesting that motor coordination is impaired in the AMPK-cKO mice, especially 251 as they age. Taken together, these findings indicate that loss of AMPK function predisposes SNpc DA neurons to earlier degeneration, which manifests phenotypically in the form of 253 locomotor deficits. 254

AMPK-cKO mice display aberrations in mitochondrial homeostasis 256
To better understand the molecular underpinnings of our observations, we examined 257 for any changes in terms of mitochondrial homeostasis, considering that the protective effects 258 of AMPK in ameliorating PD-related phenotypes is mediated by its downstream effector and 259 mitochondrial biogenesis regulator PGC-1a (6) Next, we wondered if these mice would be more susceptible to mitochondrial toxins 278 given the changes in mitochondrial homeostasis. To study this, we subjected them to a PD-279 related toxin 6-hydroxydopamine (6-OHDA) that is known to form free radicals and potently 280 inhibit mitochondrial complexes I and IV. In wild-type mice injected with 6-OHDA, levels of 281 the phosphorylated and active form of AMPK (pAMPK) is upregulated in the ventral midbrain 282 region (containing SNpc) of the injected hemisphere (Fig. 4A), possibly a result of 283 mitochondrial dysfunction that triggers a consequent reduction in ATP production (13). With 284 the knowledge that functional AMPK signalling is required for prompt cellular response to 285 metabolic changes and stresses (14), we speculate that this upregulation of AMPK activity may 286 mediate a protective response against 6-OHDA-mediated toxicity and accordingly, loss of 287 AMPK function may enhance susceptibility towards this toxin. Indeed, there is a significant 288 reduction in the number of TH-positive nigral DA neurons on the injected (ipsilateral) side of 289 AMPK-cKO mice compared to that of AMPK-WT mice (p=0.0117) (Fig. 4C). This can be 290 visualized rather apparently from the representative images of mice brain sections stained with 291 TH (Fig. 4D). Unlike the AMPK-WT mice, AMPK-cKO mice have complete lesioning of the 292 striatum at five weeks post-lesion accompanying the dramatic reduction in SNpc TH-staining 293 ( Fig. 4C-D). Loss of DA neurons unilaterally gives rise to a rotational phenotype in the 294 contralateral direction when these mice are injected with apomorphine, a DA agonist (15), 295 which can be quantified to reflect the severity of the lesion. From two to five weeks post-296 surgery, AMPK-cKO mice exhibit more pronounced rotational behaviour compared to AMPK-297 WT mice (p=0.0225 at week 2; p=0.00126 at week 3; p=0.00621 at week 4; p=0.0141 at week 298 5) (Fig. 4B). Collectively, these results show that the ablation of AMPK catalytic subunits 299 aggravates 6-OHDA-induced DA neuronal loss.

Discussion 301
To study the role of AMPK in DA neurons and PD pathogenesis, we generated a 302 conditional AMPK-KO mouse model where the catalytic alpha subunits of AMPK are ablated 303 in TH-expressing DA neurons using the Cre/loxP-mediated recombination strategy. AMPK-304 cKO mice exhibit accelerated DA neuronal loss and associated locomotor deficits. 305 Accompanying this is a reduction in the number of mitochondria in the surviving AMPK-306 deficient nigral DA neurons, suggesting that aberrant mitochondrial biogenesis may underlie 307 the PD-associated phenotypes observed during AMPK deficiency. Importantly, the AMPK-308 cKO mice are more susceptible to DA neurotoxicity mediated by a mitochondrial and PD-309 related toxin, 6-OHDA. Taken together, our findings highlight the importance of AMPK in DA 310 neuronal homeostasis and thereby its therapeutic potential for PD. 311 To date, studies that support a neuroprotective role of AMPK in PD mostly employed 312 an overexpression or activation strategy, including our previous study in the Drosophila (5). 313 To confirm these findings, it is necessary to evaluate the effects of AMPK ablation, which is 314 anticipated to promote DA neurodegeneration. However, in vivo AMPK-KO models designed 315 to study AMPK's function in the brain are few and far between, much less in neurodegenerative 316 diseases like PD (Viollet et al., 2009) showed here that loss of AMPK function in mice promotes nigral DA neurodegeneration and 324 gives rise to locomotor deficits. Supporting this, several studies in Drosophila showed that 325 Short Report 14 genetic loss-of-function of AMPK regulatory subunits (β or γ), or the expression of a dominant-326 negative form of AMPK results in overt neurodegeneration, reduction in locomotor activity 327 and overall survival (17)(18)(19). In mice, ablating the catalytic subunits of AMPK in 328 photoreceptors leads to early retinal defects including axonal retraction and synaptic alterations 329 that is normally associated with aging (20). Although these findings are not specific to the 330 context of nigral DA neurons, they collectively highlight the importance of an intact AMPK 331 heterotrimer in neuronal survival. Our study concurs with these by further demonstrating that 332 AMPK's catalytic function is critical to nigral DA neuronal homeostasis. 333 At the molecular level, we showed that the PD-associated phenotypes in AMPK-cKO 334 mice occur concurrently with a reduction in the number of mitochondria, suggesting that 335 aberrant mitochondrial biogenesis likely underlies the degeneration of AMPK-deficient DA 336 neurons. As the energy powerhouse, mitochondria play an important role in energy metabolism 337 in neurons, which are especially dependent on oxidative phosphorylation. Our previous work 338 demonstrated that AMPK-mediated rescue of mitochondrial abnormalities in PD mutant flies 339 is dependent on the mitochondrial biogenesis regulator PGC-1α (6). Conversely, when AMPK 340 activity is blocked in rat cortical neurons, the regulation of downstream PGC-1α-NRF-1 341 pathway is interrupted and neuronal activity becomes uncoupled from mitochondrial energy 342 metabolism (21). In line with this, we found that the mitochondria numbers in nigral DA 343 neurons of AMPK-deficient mice are reduced, reflecting a dysfunctional AMPK-PGC-1α axis. 344 Since the ability of the cell to ensure a steady supply of healthy mitochondria is impaired, 345 mitochondrial reactive oxygen species (mROS) may be elevated, as reported in AMPK-KO 346 mouse embryonic fibroblasts that undergo premature senescence due to elevated mROS (22). 347 Given this, it seems logical that the AMPK-cKO mice are more vulnerable to 6-OHDA, whose 348 toxicity is largely accepted to be mediated through oxidative stress (23). We (i.e. current as an attempt to cope with 6-OHDA-mediated toxicity (13) The datasets used and/or analysed during the current study are available from the