Genetic and pharmacological reduction of CDK14 mitigates α-synuclein pathology in human neurons and in rodent models of Parkinson’s disease

Parkinsona’s disease (PD) is a debilitating neurodegenerative disease characterized by the loss of midbrain dopaminergic neurons (DaNs) and the abnormal accumulation of α-Synuclein (α-Syn) protein. Currently, no treatment can slow nor halt the progression of PD. Multiplications and mutations of the α-Syn gene (SNCA) cause PD-associated syndromes and animal models that overexpress α-Syn replicate several features of PD. Decreasing total α-Syn levels, therefore, is an attractive approach to slow down neurodegeneration in patients with synucleinopathy. We previously performed a genetic screen for modifiers of α-Syn levels and identified CDK14, a kinase of largely unknown function as a regulator of α-Syn. To test the potential therapeutic effects of CDK14 reduction in PD, we ablated Cdk14 in the α-Syn preformed fibrils (PFF)-induced PD mouse model. We found that loss of Cdk14 mitigates the grip strength deficit of PFF-treated mice and ameliorates PFF-induced cortical α-Syn pathology, indicated by reduced numbers of pS129 α-Syn-containing cells. In primary neurons, we found that Cdk14 depletion protects against the propagation of toxic α-Syn species. We further validated these findings on pS129 α-Syn levels in PD patient neurons. Finally, we leveraged the recent discovery of a covalent inhibitor of CDK14 to determine whether this target is pharmacologically tractable in vitro and in vivo. We found that CDK14 inhibition decreases total and pathologically aggregated α-Syn in human neurons, in PFF- challenged rat neurons and in the brains of α-Syn-humanized mice. In summary, we suggest that CDK14 represents a novel therapeutic target for PD-associated synucleinopathy.


INTRODUCTION
Scientific, BP231, Hampton, NH, USA), 2 % Tween 80 (Fisher Scientific, BP338-500) and 90 % 126 ddH2O) 16 hours before stereotactic surgeries. Brain infusion catheters with 3.5 cm long catheter 127 tubing (Alzet® Brain Infusion Kit) were attached as per manufacturer's instructions. Brain 128 infusion assemblies were incubated at 37 °C in sterile saline until implantation. 4-month-old PAC 129 α-Syn A53T TG were deeply anesthetized with isoflurane for the stereotactic implantation of brain 130 infusion assemblies. FMF-04-159-2 (release rate of 0.35 mg/kg/day) or its vehicle solution was 131 continuously administered into the cerebral ventricles (coordinates relative to bregma: -1.1 mm 132 medial-lateral; -0.5 mm antero-posterior and -3 mm dorso-ventral) for 28 days with the brain 133 infusion catheter attached to the skull and the connected pump in a subcutaneous pocket of the 134 mouse's back. The body weight of mice was measured and their activity, neurological signs, facial 135 grimace, coat condition and respiration were scored (from 0 to 3) within 28 days after the surgery. 136 Mice were sacrificed and organs were collected on the 28 th day of the administration period. turning was recorded. The mean time to turn was calculated from 5 consecutive trials for each 164 mouse. For the rotarod test, mice were placed on a rotating, textured rod (IITC Life Science, 165 Woodland Hills, CA, USA), with the speed gradually increasing from 4 to 40 rpm over 5 min. The 166 latency to fall from the rotating rod was recorded for every mouse. Four trials per day with 10 min 167 inter-trial intervals were performed for three consecutive days. homogenizer. Samples were further lysed 1:6 (w/v) using the tissue weight in TSS Buffer (140 176 mM NaCl, 5 mM Tris-HCl), then TXS Buffer (140 mM NaCl, 5 mM Tris-HCl, 0.5 % Triton X-177 100), and SDS Buffer (140 mM NaCl, 5 mM Tris-HCl, 1 % SDS), as previously described (32).

178
For immunohistology with paraffin sections, mice were anesthetized with 120 mg/kg Euthanyl 179 (DIN00141704) and intracardially perfused with 10 mL of PBS, followed by 20 mL of 10 %    The assay utilizes a 96 deep well plate with a filter plate attached on top. Samples were thawed on 213 ice, vortexed and centrifugated at 13 000 x g. 10 µL of sample was loaded in the center of the filter 214 and dried in a stream of nitrogen. Following derivatization by phenyl isothiocyanate and drying of 215 filter spots, dopamine content was extracted by adding 300 µL of extraction solvent, centrifugation 216 into the lower collection plate and dilution by MS running solvent. Mass spectrometric analysis 217 was performed with the ABSciex 4000 QTrap® tandem mass spectrometer in combination with 218 an Agilent 1260 series UHPLC system (Agilent Technologies, Palo Alto, CA, USA). Samples 219 were delivered by an LC method followed by a direct injection method. Data was analyzed using 220 Analyst 1.6.2 and expressed as dopamine concentration relative to tissue weight.    ~10 5 cells per coverslip and maintained in culture for 9 ( Fig. S1D and S2B) or 21 days (Fig. 2). 281 Neurons were treated with 2 µg/mL of sonicated α-Syn PFFs (StressMarq, Victoria, BC, Canada, 282 SPR-324) at 2 ( Fig. S1D and S2B) or 7 days in vitro (DIV) (Fig. 2). 14 days after adding PFFs to

447
injected mice 448 We first examined existing in situ hybridization data for the expression of Snca and Cdk14 in the 449 murine brain. We observed that both genes are expressed in similar brain regions, including the 450 hippocampus and the SN (Fig. S1A). Since PD is a chronic disease, inhibition of a candidate 451 modifier would have to be safe in the long term. We analyzed Cdk14 protein levels in different 452 mouse organs and tested the effects of Cdk14 depletion on survival, fertility, and organ 453 cytoarchitecture in vivo. We found that CDK14 protein is highly abundant in the brain, as well as 454 in the lung and the spleen (Fig. S1B). Cdk14 nullizygous mice are viable, fertile, and exhibited 455 normal brain morphology (Fig. S1C, (27)) and synaptic integrity (Fig. S1D). Furthermore, we did 456 not observe altered morphology of the lung and spleen by Cdk14 ablation (Fig. S1E). We next 457 asked whether silencing Cdk14 is sufficient to mitigate behavioral and histological phenotypes 458 observed in cultured neurons and in mice exposed to pathogenic α-Syn pre-formed fibrils (mouse 459 PFFs; Fig. 1A and B; Fig. S2A). 6 months following intrastriatal injection of α-Syn PFFs, there is 460 a stereotypic brain-wide accumulation of pS129 α-Syn -a marker of human synucleinopathies -461 in addition to SN DaNs loss and mild motor impairments (40-42). In our experimental paradigm, 462 6 months after PFF injection (at an age of 12 months), we found that PFF injected WT mice 463 exhibited reduced forelimb force generation in the grip strength test compared to their saline-464 treated counterparts (Fig. 1B), similar to what has been previously reported (41,42). In contrast, 465 the PFF-induced weakening of grip strength was not observed in Cdk14 +/or in Cdk14 -/mice. We 466 did not observe PFF-mediated changes (in any genotype tested) in the other 8 behavioral tests 467 21 conducted (including tests for cognitive and motor function). Importantly, we noted that saline-468 injected Cdk14 +/and Cdk14 -/mice consistently performed like their WT counterparts in each test, 469 suggesting that chronic Cdk14 reduction is not deleterious to the brain (Fig. S2C). 470 Next, we analyzed the relative pathological burden of accumulated α-Syn throughout the brain of Cdk14 -/mice compared to their WT littermates (Fig. 1C). This was particularly evident at PFFs ipsilateral to the injection, relative to their saline-treated counterparts (Fig. 1D). PFF-induced 488 loss of dopamine fibers was not accompanied by altered striatal dopamine content (Fig. S2D). TH Cdk14 -/mice in comparison to saline-injected controls (Fig. S2D) note that the effect of Cdk14 on α-Syn was selective to pathological forms of the protein, as partial 494 reduction or ablation of Cdk14 in Cdk14 +/or Cdk14 -/mice, respectively, did not change levels of 495 endogenous mouse α-Syn (Fig. S2E).  Fig. 2A). We applied this filtered, seed-competent media to naïve WT 505 cultures to test the seeding capacity of α-Syn and found that loss of Cdk14 dramatically reduced 506 α-Syn pathology (measured by pS129 α-Syn accumulation), 14 days following media application 507 (Fig. 2B).

510
Having observed the benefits of Cdk14 depletion in the PFF mouse model of PD, we next tested 511 whether this benefit translates to human neurons. We infected DaNs derived from a PD patient 512 carrying an A53T mutation in α-Syn (33) as well as its isogenic control with lentiviruses carrying 513 23 Cas9/sgRNAs against CDK14. Neurons infected with sgRNAs targeting either exon 3 (E3) or exon 514 8 (E8) of CDK14, exhibited approximately 50 % of the CDK14 levels of the control cultures (Fig.   515 3). We found that A53T mutant cells show a marked increase of pS129 α-Syn compared to isogenic 516 controls, and that CDK14 knockdown significantly lowers pS129 α-Syn levels (Fig. 3). by ELISA (Fig. 4A). We also tested whether the CDK14 inhibitor affects the α-Syn load in rat 526 primary neuronal cultures treated with α-Syn PFFs. Here, PFF treatment induced a spike in 527 insoluble (Urea buffer-soluble) α-Syn protein, which CDK14 inhibition markedly reduced (Fig.   528   4B). Interestingly, the PFF treatment also induced an increase of CDK14 insoluble protein, which 529 was not present in untreated neurons or in neurons treated with α-Syn monomers (Fig. 4B).

532
Since the treatment of human neurons and PFF-challenged rat neurons with the CDK14 inhibitor 533 showed a reduction in total and insoluble α-Syn protein, respectively, we next tested whether 534 pharmacological inhibition of CDK14 modifies α-Syn levels in vivo. We administered FMF-04- Syn A53T TG mice (Fig. 5A) which harbor the PD-associated A53T mutant human α-Syn gene in 537 the absence of mouse Snca (28). Administration of the CDK14 inhibitor did not modify body 538 weight development, nor induce any signs of distress or pain as indicated by alterations of 539 locomotion, facial expression, or coat condition of PAC α-Syn A53T TG mice in comparison to 540 vehicle-treated counterparts (Fig. S3A). Similarly, CDK14 inhibitor treatment did not induce 541 changes in the cytoarchitecture of the lung, spleen, and liver (Fig. S3B). To quantify levels of 542 pathogenic forms α-Syn, we collected brains after 1 month of CDK14 inhibitor treatment (Fig.   543 5A) and analyzed protein content of the TSS, TXS and SDS buffer-soluble fractions (Fig. 5B). We Together these results show that in vivo administration of the CDK14 inhibitor in the brain engages 551 its target and mitigates certain pathogenic forms of human α-Syn in the mouse brain in a protein 552 fraction-dependent manner without inducing obvious discomfort or pain. body of evidence suggests that α-Syn may play a role not only at the presynaptic space but also in 568 the immune system (37,46,47). Therefore, careful titration of its levels may be clinically crucial.

569
As a result, we asked whether candidates that are more amenable to traditional pharmacology (i.e. CDK14 is a pharmacologically tractable target for synucleinopathy. 586 We observed that loss of Cdk14 in PFF-treated mice reduced the level of α-Syn 587 histopathology in cortical areas, such as the somatomotor cortex. Surprisingly, we did not detect 588 changes in the load of pS129 α-Syn-positive cells by Cdk14 ablation in the striatum, the PFF 589 injection site (Fig. 1C). In line with this, we noticed a similar PFF-induced degeneration of the 590 dopaminergic nigrostriatal system in WT, Cdk14 +/and Cdk14 -/mice ( Fig. 1D and Fig. S2D).

591
These observations imply that loss of CDK14 reduces the degree of intercellular α-Syn spreading 592 rather than protecting neurons which are directly exposed to PFFs. Indeed, when we tested this in 593 a cultured neuron system, we found that genetic reduction of Cdk14 dramatically decreased the 594 spreading capacity of seed-competent α-Syn (Fig. 2). Therefore, it is plausible that CDK14 as measured by ELISA quantification (Fig. 4A). FMF-04-159-2 was recently designed to provide 606 an improved pharmacological tool for the inhibition of CDK14 as treatment for colorectal cancer 607 (26). Interestingly, FMF-04-159-2 was described to covalently bind and inhibit CDK14 at ~100 608 nM (IC50 = 86 nM (26)), a dosage which lowered α-Syn levels to ~12 % of vehicle-treated controls 609 in our in vitro experiments. Applying the CDK14 inhibitor to PFF-challenged rat cortical neurons 610 reduced the amount of aggregated (Urea buffer-soluble) α-Syn species (Fig. 4B), phenocopying 611 the low degree of pS129 pathology in cortical neurons of PFF-treated Cdk14 -/mice (Fig. 1C) or 612 cultures ( Fig. 2 and Fig. S2B). Interestingly, we found that Cdk14 accumulated in the urea-soluble  Fig. 5A and B). Reduction of total α-Syn was accompanied by lower amounts of CTT α-Syn, 620 indicating that this form of α-Syn, which increases α-Syn's propensity to aggregate and enhances inhibitor treatment, such as loss of motor activity or imbalance during locomotion (Fig. S3A), 629 implying that higher amounts of cytosolic pS129 α-Syn do not substantially promote cerebral and Syn pathology. Future studies will help refine the mechanism whereby CDK14 regulates α-Syn.

644
Our results examining the genetic and pharmacological reduction of CDK14 in PD models 645 set the stage for future pre-clinical studies. In all behavioral experiments conducted, Cdk14 -/mice 646 were indistinguishable from WT mice ( Fig. 1B and Fig. S2C). Furthermore, loss of Cdk14 did not 647 alter the architecture of brain tissue or peripheral organs (Fig. S1C to Fig. S1E) implying that 648 Cdk14 loss is not deleterious in vivo. This is supported by human genetics where the loss of CDK14 649 appears to be well tolerated. Additionally, Cdk14 inhibition in vivo did not induce any signs of 650 discomfort (Fig. S3A) (Fig 4A). Further preclinical experiments will test whether the drug rescues PD-like neuron loss 656 and behavioral phenotypes in models of synucleinopathies; potentially paving the way for its use 657 in humans.

658
In sum, we show that CDK14 inhibition causes a decrease of total α-Syn concentrations