Tau pathology reduction with SM07883, a novel, potent, and selective oral DYRK1A inhibitor: A potential therapeutic for Alzheimer's disease

Abstract Dual‐specificity tyrosine phosphorylation‐regulated kinase‐1A (DYRK1A) is known to phosphorylate the microtubule‐associated tau protein. Overexpression is correlated with tau hyperphosphorylation and neurofibrillary tangle (NFT) formation in Alzheimer's disease (AD). This study assessed the potential of SM07883, an oral DYRK1A inhibitor, to inhibit tau hyperphosphorylation, aggregation, NFT formation, and associated phenotypes in mouse models. Exploratory neuroinflammatory effects were also studied. SM07883 specificity was tested in a kinase panel screen and showed potent inhibition of DYRK1A (IC50 = 1.6 nM) and GSK‐3β (IC50 = 10.8 nM) kinase activity. Tau phosphorylation measured in cell‐based assays showed a reduction in phosphorylation of multiple tau epitopes, especially the threonine 212 site (EC50 = 16 nM). SM07883 showed good oral bioavailability in multiple species and demonstrated a dose‐dependent reduction of transient hypothermia‐induced phosphorylated tau in the brains of wild‐type mice compared to vehicle (47%, p < 0.001). Long‐term efficacy assessed in aged JNPL3 mice overexpressing the P301L human tau mutation (3 mg/kg, QD, for 3 months) exhibited significant reductions in tau hyperphosphorylation, oligomeric and aggregated tau, and tau‐positive inclusions compared to vehicle in brainstem and spinal cord samples. Reduced gliosis compared to vehicle was further confirmed by ELISA. SM07883 was well tolerated with improved general health, weight gain, and functional improvement in a wire‐hang test compared to vehicle‐treated mice (p = 0.048). SM07883, a potent, orally bioavailable, brain‐penetrant DYRK1A inhibitor, significantly reduced effects of pathological tau overexpression and neuroinflammation, while functional endpoints were improved compared to vehicle in animal models. This small molecule has potential as a treatment for AD.


| INTRODUC TI ON
There remains an urgent need for treatments to improve or slow progressive neurodegenerative diseases (Cummings, Lee, Ritter, & Zhong, 2018). Tau pathology is considered a key driver for a broad spectrum of neurodegenerative diseases, collectively known as tauopathies (Iqbal, Liu, & Gong, 2015). These include primary tau diseases such as frontotemporal lobar degeneration with tau inclusions (FTLD-tau), progressive supranuclear palsy (PSP), and Pick's disease, as well as multimodal diseases such as Alzheimer's disease (AD). In AD, tau aggregation and spreading appears to be enabled by amyloid-beta aggregation, but therapeutics aimed at regulating the amyloid cascade have failed to prevent disease progression in symptomatic patients (Giacobini & Gold, 2013;Holtzman et al., 2016). In the context of a rapidly growing AD population with no disease-modifying therapeutics, targeting tau appears to be a plausible approach; the spatiotemporal pattern of tau pathology in AD is highly correlated with brain atrophy and observed cognitive decline (Giannopoulos, Chiu, & Praticò, 2018). Tau is a multifunctional protein, which may become neurotoxic as a result of post-translational modifications including a high degree of phosphorylation that leads to its oligomerization, sequestration, and aggregation (Spillantini & Goedert, 2013). Tau can be excessively phosphorylated at multiple epitopes, and these toxic forms of hyperphosphorylated tau (pTau) are found in higher quantities in AD brains. The buildup of insoluble oligomeric forms and gradual deposition of filamentous aggregated tau protein into intraneuronal pretangles and neurofibrillary tangles (NFTs) characterize all tauopathies (Bodea, Eckert, Ittner, Piguet, & Götz, 2016;Morris, Maeda, Vossel, & Mucke, 2011). While a pleiotropy of therapeutics has recently been tested in clinical trials to prevent tau aggregation and spreading, the exact form(s) of tau responsible for its toxicity and spreading remain to be identified (Khanna, Kovalevich, Lee, Trojanowski, & Brunden, 2016). Therefore, regulating the tau cascade upstream at the phosphorylation level is an appealing strategy.
Over 80 epitopes can be phosphorylated on tau, and more than 40 have been identified to be phosphorylated specifically in AD brains, with each pTau site having some degree of specificity for various kinases (Martin et al., 2013;Polanco et al., 2017;Šimić et al., 2016). Drug development has targeted inhibition of downstream kinases such as the mitogen-activated protein kinase (MAPK) family or glycogen synthase kinase-3 beta (GSK-3β); both have shown benefit in preclinical models, however, thus far, without successful clinical outcomes (Le Corre et al., 2006;Lovestone et al., 2015;Onishi et al., 2011;Vogel et al., 2009). While multiple kinase inhibitors have recently been approved by the FDA, primarily in oncology, none have been approved for CNS disorders (Wu, Nielsen, & Clausen, 2015).
Targeting kinases remains a viable approach to reduce tau phosphorylation and requires careful development to ensure brain penetrance and therapeutic efficacy.
The dual-specificity tyrosine phosphorylation-regulated kinase-1A (DYRK1A) is a novel protein kinase that can directly and indirectly regulate phosphorylation of tau on numerous residues (Azorsa et al., 2010;Ryoo et al., 2007;Song et al., 2015;Walte et al., 2013). Overproduction of DYRK1A has been linked to increased tau phosphorylation and aggregation and is a likely contributor to higher incidence of AD in the Down syndrome (DS) population (Liu et al., 2008;Wegiel, Gong, & Hwang, 2011). Due to its localization on chromosome 21, the DYRK1A gene is 1.5-fold overexpressed in DS and increased activity of DYRK1A has been correlated with abnormal brain development, cognitive disabilities, and an early onset of AD in individuals with DS (Becker, Soppa, & Tejedor, 2014). Furthermore, DYRK1A-induced tau phosphorylation has been shown to disrupt the normal function of tau as a microtubule stabilizer and promote its aggregation (Liu et al., 2008). Mice overexpressing DYRK1A have increased pTau in the brain, while crossing DYRK1A heterozygous mice with DS mice regulated pTau, amyloid load, and symptoms (García-Cerro, Rueda, Vidal, Lantigua, & Martínez-Cué, 2017), implying DYRK1A is a highly dosage-sensitive gene. Finally, in AD patients, overexpression of DYRK1A was observed in postmortem brains, further suggesting its contribution to tau hyperphosphorylation and the accumulation of NFTs (Ferrer et al., 2005). Targeting DYRK1A appears to be a viable treatment approach for AD. It is hypothesized that inhibition of DYRK1A activity may reduce tau pathology and inflammation and thus prevent, slow, or reverse AD or other chronic tauopathies. SM07883 was developed by rational design as an orally available, brain-penetrant, small-molecule DYRK1A kinase inhibitor and showed regulation of multiple phospho-tau epitopes both in vitro and in vivo in mouse brains. Repeat dosing in transgenic mice expressing brainstem tau pathology demonstrated efficacy in reducing formation of insoluble tau fragments and the subsequent cascade leading to tau aggregation, formation of tau-positive inclusions, and associated neuroinflammation. Ultimately, loss of function and morbidity were found to be limited with SM07883 treatment compared to controls.

| Discovery and pharmacological properties of SM07883, a potent DYRK1A kinase inhibitor
SM07883 is a small-molecule 3-acylamino-isoquinoline analog ( Figure 1a), as referred in PCT patent application publication

K E Y W O R D S
Alzheimer's disease, DYRK1A, mouse models, neurodegenerative diseases, neuroinflammation, Tau number WO, 2017189823, that was rationally designed and optimized through iterative medicinal chemistry to achieve significant exposure of the drug in the brain with oral administration.
The ability of SM07883 to reduce tau phosphorylation was evaluated in 2 cell-based assays utilizing double transfection and overexpression of both human microtubule-associated protein tau (MAPT) and DYRK1A genes in HEK293T cells followed by Western blotting and confirmation by ELISA. In HEK293T cells, the dual transfection allowed for higher expression and detection of tau phosphorylation at multiple sites with pronounced effect at Thr212, Ser396, and Thr181 ( Figure 1c). Dual transfection also potentiated the phosphorylation signal with AT8 (an antibody which binds to the pSer202/Thr205 sites) and the Thr231 epitope. After SM07883 treatment, a dose-dependent decrease in phosphorylation was observed for all the tested epitopes with a calculated EC 50 ranging from 7 nM for Thr212 to 228 nM for Ser396 ( Figure S1b).
When evaluated with ELISA allowing multiple screens over time, SM07883 was shown to inhibit phosphorylation at Thr212 with an average EC 50 of 16 nM (±9 s.e.m.; Figure 1d) across tested lots.
SM07883 was tested in unstimulated, fast-dividing SH-SY5Y neuroblastoma cells, in which a high level of phosphorylated tau at Ser396 was shown to be reduced by overnight (16 hr) treatment when compared to vehicle (EC 50 of 200 nM; ±66 s.e.m. across tested lots; F I G U R E 1 Discovery of SM07883 as a potent and selective DYRK1A inhibitor and regulator of tau phosphorylation. (a) Markush structure of SM07883, a 3acylamino-isoquinoline analog, where R 1 is (un)substituted 5-membered heteroaryl, and R 2 is selected from either (un) substituted C1-4 alkylene-heterocyclyl, or (un)substituted heterocyclyl, or (un) substituted C1-4 alkylene-carbocyclyl, or (un)substituted carbocyclyl as referred in WO, A2. (b) SM07883 inhibition of DYRK1A kinase activity using the Z-Lyte™ platform. An IC 50 value of 1.6 nM was determined from this 11-point dose-response curve. (c) Western blot of HEK293T (co-transfected with human DYRK1A and MAPT genes) cell lysates collected after 16-hr treatment. SM07883 dose-dependently inhibited potentiated tau phosphorylation. EC 50 values with ratios over β-actin are provided in Figure  S1b. Asterisks indicate samples showing higher loading artifact. (d) An example of EC 50 of 13 nM generated for this lot of SM07883 using an ELISA format elaborated for screening purposes of phosphorylated tau at Thr212. (e) A secondary screen was performed in SH-S5Y5 human neuroblastoma cells. A dosedependent reduction of phosphorylated tau at Ser396 generated an EC 50 of 184 nM for this lot of SM07883. Note: All IC 50 /EC 50 curves are shown as mean of duplicates ± SEM (some SEM bars smaller than data point symbol) Figure 1e). Screening of published DYRK1A and GSK-3β inhibitors showed that tau phosphorylation at Ser396 was preferentially regulated by GSK-3β inhibitors, while DYRK1A inhibitors had higher inhibitory activity at the Thr212 site (Table S2). SM07883 was the only compound screened with the ability to inhibit tau phosphorylation at both the Thr212 and Ser396 epitopes with nanomolar potency.
Although the most common post-translational modification of tau proteins is phosphorylation, O-glycosylation and acetylation may also affect localization, degradation, and function and are closely associated with the level of phosphorylated tau (Carlomagno et al., 2017;Liu et al., 2002). Regulation of acetylated and glycosylated tau was investigated in unstimulated SH-SY5Y cells exposed to overnight treatment with SM07883. Western blot analysis showed only a marginal reduction at the highest tested dose when staining for glycosylated tau at serine 400 with N-acetyl-d-glucosamine (O-GlcNAc; Figure S2). A modest decrease in tau acetylation at K280 was observed but remained minimal compared to the reduction in tau phosphorylation at Ser396 observed at all tested doses.
Inhibition of CLK4 has been suggested to have a role on inhibition of SR protein phosphorylation with consequences on regulation of splicing and inclusion of tau exon 10 (Hartmann et al., 2001).
Since SM07883 inhibited CLK4 in vitro, the compound was tested overnight on SH-SY5Y cells and RNA was collected. The compound showed no regulation of the 3R/4R ratio of the MAPT gene encoding for tau using specific primers on flanking regions of exon 10 compared to control using RT-PCR. However, the use of a potent CLK2 inhibitor (Cmpd #79, USPTO patent application publication number US20160008365A1; CLK2 IC 50 : 9 nM) showed an evident reduction in the 4R form of tau indicative of splicing of exon 10 ( Figure   S3). Furthermore, no inhibition of SR phosphorylation was observed by Western blot analysis after 1-hr incubation of SM07883 on SH-SY5Y cells while the CLK2 inhibitor Cmpd #79 dramatically reduced phosphorylation of SRSF4 and SRSF6 at all tested doses ( Figure S4).

| Pharmacodynamic studies in wild-type mice
Following oral administration in mice, SM07883 was well absorbed with peak plasma concentration (Cmax) occurring 2 hr postdose and with an estimated bioavailability of 92% ( Figure 2a). Pharmacodynamic properties were consistent across rodents and higher species (Table   S3). The estimated terminal half-life of 3.3 hr was suitable for oncedaily administration in mice. Additionally, low efflux was determined in vitro in Caco-2 or MDR1-MDCK cells (Table S1) and significant brain penetration was observed with a brain-to-plasma (Kp) ratio of 1.9 in mice. SM07883 showed approximately dose-log-linear increases in exposures (plasma and brain) across the pharmacodynamics (single dose) and tau transgenic mouse (3-month daily doses) models. In vitro free fraction in rodent brain homogenate was 6% and correlated well with the in vivo distribution of SM07883 in plasma, brain, and CSF ( Figure 2b). An additional study in rats showed SM07883 to be homogenously distributed throughout brain regions ( Figure S5).

| SM07883 reduced tau phosphorylation in vivo
Inhibition of tau hyperphosphorylation (pTau) in vivo by SM07883 was tested in an anesthesia-induced transient tau hyperphosphorylation mouse hypothermia model (Bretteville et al., 2012).
Phosphorylation of tau at Ser202/Thr205 sites was significantly increased by anesthesia in vehicle-treated versus untreated control brain lysates when analyzed by AT8 Western blotting Phosphorylation of the Thr212 epitope on tau was also significantly affected in this model and followed the same inverted correlation with SM07883 brain exposure (Figure 3c,d, p < 0.005).
SM07883 had no effect on regulating anesthesia-induced decrease in body temperature ( Figure S6a), suggesting these effects were due to direct regulation of tau phosphorylation. A single oral dose of 2.5, 5, or 10 mg/kg of SM07883 showed that inhibition of the signal was the strongest at 4 hr after administration and the reduction in tau phosphorylation remained significant at 8 hr
Similarly, SM07883 significantly reduced sarkosyl-insolu- The total amount of human transgenic tau was also found increased and is a likely consequence of increased accumulation of tau inclusions ( Figure S8b). Compared to vehicle, significant reductions of pTau, aggregation of tau, and hindbrain AT8 staining were also observed with an alternative dose regimen of SM07883 (10 mg/kg daily or every other day; Figure S9).

| SM07883 reduced tau-associated glial activation
Markers of neuroinflammation associated with tau pathology were also evaluated in JNPL3 brains and spinal cord samples.
Representative micrographs in Figure 5c and Figure 5d show a reduction in gliosis (GFAP immunoreactivity) and the number of Iba1reactive microglial cells (p < 0.001, Figure 5g) following treatment with SM07883 compared with vehicle-treated animals. Reduction in GFAP expression was quantified by ELISA in spinal cord samples from mice treated with SM07883 compared to both vehicle-treated groups (n = 19 and n = 18, respectively; p < 0.001, Figure 5f).

| Improvement in weight loss, mortality, and morbidity in JNPL3 mice treated with SM07883
Over the course of the 3-month treatment, compared to vehicle, SM07883 showed a reduction in weight loss and mortality, which are inherent characteristics of the JNPL3 transgenic strain (Lewis et al., 2001). Initiation of daily dosing in all groups was accompanied by a noticeable drop in weight in all groups, most likely a consequence of the sudden daily manipulation of the mice (Figure 6a). While  (Table 2).
Tremors were often associated with rough coat and reduced mobility. SM07883-treated mice, however, showed a quick recovery in weight loss after initiation of the dosing (Figure 6a, triangles) and the group had a positive net gain over the course of the study (1.3 g ± 0.57, n = 19; Figure 6b, dark gray bar) as well as a reduction of the clinical signs (Table 2).

| Improvement of motor task performance with SM07883 treatment
As

| D ISCUSS I ON
There are no disease-modifying therapies for AD, and despite con- It is proposed that DYRK1A expression is regulated by a wide range of molecular changes and disease and exposure to stimuli such as the proinflammatory cytokine TNF-α, cellular stress signals, or soluble beta-amyloid (Aβ) was shown to induce DYRK1A expression, while in return, these increases in DYRK1A expression induced tau phosphorylation, Aβ production, and Aβ-induced tau aggregation (Choi & Chung, 2011;Coutadeur et al., 2015;Kang, Choi, Park, & Chung, 2005). The higher levels of DYRK1A expression found in F I G U R E 3 SM07883 reduced hypothermia-induced tau phosphorylation in a dose-dependent manner and correlated with increasing SM07883 brain exposure. BALB/c mice were administered a single oral dose of vehicle or SM07883 before hypothermia-induced anesthesia before brain collection and analysis by Western blotting for tau phosphorylation at AT8 (Ser202/Thr205) or Thr212; β-actin and ratio are displayed as bars (left y-axis), while SM07883 exposure is shown as lines (right y-axis). (a) Dose-response after 1 hr of anesthesia treatment (4 hr postvehicle (n = 3)) or SM07883 administration (n = 3-5). (b) Western blots used for densitometry results of AT8 (top) and β-actin (bottom) for each tested dose. (c-d) Dose-response with blotting for tau phosphorylation at Thr212 (n = 3-5). Asterisks represent a loading control sample loaded on both gels, L is the ladder mark showing the 60 kDa size band. (e) Time course of brain collection from 2 to 8 hr postadministration of 10 (left), 5 (middle), or 2.5 mg/kg (right) of SM07883, while hypothermia induction remained induced 1 hr prior to sacrifice [corresponding Western blots are shown in Figure S6b]. (f-g) Extended time course (10 mg/kg, left; and 5 mg/kg right) with hypothermia 1 hr prior to sacrifice showing clearance of SM07883 at 24 hr and reduction of the tau signal. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, one-way ANOVA the brains of people with AD and Pick's disease and the overlap with hyperphosphorylated tau suggest that the kinase is contributing to tau disease pathogenesis (Ferrer et al., 2005). Treatment of JNPL3 mice with SM07883 significantly reduced pTau at multiple AD critical epitopes in the brainstem compared to vehicle; these results were like the effects seen in vitro. Consequently, SM07883 significantly reduced tau aggregation, leading to significantly lower numbers of tau-positive inclusions. The reduction in brainstem pathology in JNPL3 mice treated with SM07883 was correlated with improved motor task performance, while vehicle-treated mice worsened. The effects on weight, general well-being, and survival rate indicated that SM07883 was well tolerated and prevented F I G U R E 4 Reduction of pathological tau signals with treatment by SM07883. Brainstems and spinal cords from JNPL3 mice were collected after 3 months of a once-daily dose of vehicle (n = 20, □), SM07883 (3 mg/kg, n = 19, ▼), or wild-type littermates administered the vehicle only (n = 9, •) with *p < 0.05 and ***p < 0.001 when compared to vehicle. (a) SM07883 significantly reduced pTau in the brainstem compared to JNPL3 mice treated with the vehicle. Lysates that were analyzed by Western blot for tau phosphorylation at Thr212/214, Thr231, AT8 sites (Thr202/Ser205), Thr181, and ratio over β-actin were determined by densitometry. The AT8 monomer band of 55 kDa was used for this calculation.
(b) SM07883 significantly reduced AT8 staining in the sarkosyl-insoluble fraction from the brainstem compared to vehicle-treated mice as analyzed by Western blotting (*p = 0.010). Only fragments ≥ 64 kDa were counted. Representatives blots can be found in Figure S7a  Sustained systemic and brain exposure after oral administration demonstrated good bioavailability and blood-brain barrier penetrance, and made SM07883 suitable for once-a-day dosing in these studies. With high exposure in CNS compartments across multiple species and significant pharmacodynamics response in mice, these results indicate that reducing DYRK1A activity may be a viable treatment in diseases such as AD, where an increase in enzyme activity may regulate the cascade of phosphorylation leading to tauopathies and neurodegeneration.

| Animals
Seven-week-old wild-type (WT) male BALB/c mice (Envigo) were used for pharmacokinetics and pharmacodynamics studies.
Sprague Dawley male rats (Charles River) were used for the compound brain distribution study. Six-month-old male transgenic Tg(PrnpMAPT*P301L) JNPL3Hlmc ("JNPL3") and age-matched wild-type mice were purchased from Taconic Farms and allowed to age in-house until they were 10 months of age. These mice were placed on a NIH #31M diet (Envigo) throughout the entire study.

| Cells and cell-based assays
Human embryonic kidney cells, HEK293T cells (ATCC), were cultured in Dulbecco's modified Eagle's medium [DMEM], 10% fetal bovine serum [FBS], and 1% penicillin/streptomycin), while human neuroblastoma SH-SY5Y cells (Sigma) were cultured in 1:1 DMEM/F-12 medium supplemented with 15% FBS and 1% nonessential amino acid and incubated at 37°C with 5% carbon dioxide. 1X neat F I G U R E 6 Improvement in weight loss, mortality, morbidity, and functional deficit in JNPL3 mice treated with SM07883. (a) Weekly body weight changes from 9 months of age until termination of the study at 13 months of age. JNPL3 mice were dosed once-daily with either SM07883 (3 mg/kg, n = 19; ▼) or vehicle (n = 20; □) starting at 10 months of age; referred to week 0. Age-matched wild-type received the vehicle (n = 9). Cross signs indicate early deaths in the vehicle group.

| Single-dose pharmacodynamics in mice
Wild-type BALB/c male mice (7 weeks old) were administered a single dose per oral gavage of vehicle or SM07883 (1.25, 2.5, 5, 10 or 25 mg/kg). Plasma and tissue processing for bioanalytical studies are described in the supplementary materials. For pharmacodynamics studies, one hour prior to brain and plasma collection mice were treated with a ketamine/xylazine (K/X) cocktail (150 µl of 100 mg/ kg) via intraperitoneal (I.P.) administration to trigger anesthesia-induced hypothermia and subsequent brain tau phosphorylation.
Exactly one hour postinduction, mice were decapitated, and tissues were collected, flash-frozen, and stored for Western blot analysis as described below. Untreated (nonanesthesia) control mice were sacrificed under CO 2 .

| Repeat-dose efficacy in tau transgenic mice and wire-hang test
Nine wild-type littermates and 39 tau JNPL3 male mice carrying the human P301L tau mutation form of tau from autosomal dominant tau FTD patients (Lewis et al., 2001) were randomized in groups after ini- and reducing agent (Catalog #1610792, Bio-Rad) prior to boiling.
Samples were further diluted in 1X LDS and loaded onto a NuPAGE protein electrophoresis gel, then transferred onto nitrocellulose blots before blocking and blotting with a primary antibody followed by antimouse HRP-conjugated antibody and revealed by bioluminescence.
Blots were probed for β-actin for control of total protein loading.
Primary antibodies used in this study can be found in Table S6. Western blot confirmation of the regulation of phosphorylated tau epitopes and other related DYRK1A/MAPT targets was analyzed through measurement of band densitometry using the NIH freeware ImageJ, and values were normalized as ratios over beta-actin or total tau (HT7).
Homogenized brainstem samples in RIPA were further sonicated and separated by extraction in 1% sarkosyl solution for 1 h on an orbital shaker and ultracentrifugated at 150,000 g for 1 hr.
After rinsing with RIPA and PBS, samples were air-dried and sarkosyl-insoluble pellets were resuspended in 1X LDS/10M urea and beta-mercaptoethanol. Samples were further denatured at 95°C for 5 min prior to dilution, loading onto NuPAGE gels and blotting with an AT8 antibody as described above. Unless noted, no band at the 55 kDa size, expected to be monomeric tau, was detected and bands of 64 kDa and above were quantified.
To confirm the reduction of tau fragments, specific size of aggregated forms of tau were evaluated in spinal cord lysates by HTRF using a tau aggregation assay combining the same total tau antibody conjugate with a donor and acceptor fluorochrome (6FTAUPEG, Cisbio). According to the manufacturer, specific wavelength (665 nm) fluorescence is emitted if both antibodies are within 9 nM TA B L E 2 Clinical observations at termination of the study in 13-month-old JNPL3 mice distance. Although the exact size range of tau aggregates detected in this assay is unknown, the assay does not recognize nonaggregated forms of tau. The plate was read with the Envision multiplate reader (PerkinElmer).

| Statistical analysis
For dose-response, analysis of variance (ANOVA) with Dunnett's multiple comparison correction was used; the nonparametric Kruskal-Wallis test with Dunn's multiple comparison correction was also used as a sensitivity analysis (Figures 3 and 6).
Due to the positively skewed and strictly non-negative response data, parametric generalized linear models assuming a gamma distribution (with reciprocal link function) were used to estimate differences between vehicle and treatment groups in the JNPL3 study (Figures 4 and 5). For severely skewed data where gamma distribution did not characterize the data well, rank-based Wilcoxon-Mann-Whitney/Kruskal-Wallis tests were employed. A summary of statistical tests used in JNPL3 mice analysis is provided in Table S6.

ACK N OWLED G M ENTS
We thank John Hood, PhD, Betty Tam