How early gut inflammation and a Parkinson's-related gene interact differently in male and female mice to cause Parkinson's-like traits

To further assess effects of prodromal intestinal inflammation, LRRK2 G2019S, and biological sex on PD-related neuropathology, 10–13 week old male and female hLRRK2^G2019S Tg mice and WT littermates were subjected to DSS treatment and assessed 30 weeks later (at 50 weeks of age) for phosphorylated α-synuclein and dopaminergic neuron loss in the substantia nigra (SN), two hallmark pathologies of PD²⁵. DSS-treated male hLRRK2^G2019S Tg mice exhibited elevated phosphorylated α-synuclein in the SN compared to DSS-treated male WT littermates (Fig. 2e, f), whereas DSS-treated female hLRRK2^G2019S Tg mice exhibited no difference relative to their female WT controls. Notably, approximately one-third of human LRRK2 mutation carriers are reported to be α-synuclein negative^26,27, suggesting that this mouse model has greater face validity for the major subset of LRRK2 mutation carriers who are α-synuclein positive. Consistent with the increases in phosphorylated α-synuclein, decreased numbers of tyrosine hydroxylase (TH⁺) neurons were detected in the SN of DSS-treated male hLRRK2^G2019S Tg mice compared to their treatment-matched WT controls, with no genotype-dependent differences in females (Fig. 2g). These neuropathological abnormalities were not seen in male hLRRK2^G2019S Tg mice treated with vehicle (water) instead of DSS (Fig. S4, S5), indicating that prodromal intestinal inflammation expedites the onset and severity of PD-related neuropathology in genetically predisposed males.

We observed that some microglia from DSS-treated hLRRK2^G2019S Tg mice contained phosphorylated α-synuclein deposits and were localized adjacent to TH⁺ axons (Fig. 2h, i). Given that microglia are able to internalize neuron-derived α-synuclein²⁸ and accumulate α-synuclein intracellularly to promote dopaminergic neuron loss²⁹, we quantified phosphorylated α-synuclein loads localized within IBA-1⁺ microglia. Mean fluorescence intensity (MFI) of phosphorylated α-synuclein within SN microglia was elevated in DSS-treated male hLRRK2^G2019S Tg mice compared to DSS-treated male WT littermates (Fig. 2j), with no differences in the absolute numbers of IBA-1⁺ microglia in the SN (Fig. 2k). Both male and female hLRRK2^G2019S Tg mice and WT littermates exhibited increased IBA-1 and CD68 intensity in the SN in response to DSS treatment (Fig. S6), indicating that prodromal intestinal inflammation broadly stimulates microglial activation in both biological sexes, independently of genotype. To further gain insight into whether the increases in phosphorylated α-synuclein are concomitant to increases in α-synuclein aggregation, we measured levels of α-synuclein following detergent extraction from the insoluble fraction of brain homogenates. DSS-treated male hLRRK2^G2019S mice exhibited modest increases in insoluble α-synuclein in striatum relative to wildtype littermate controls (p = 0.0556, Fig. 2l), suggesting elevations in aggregated α-synuclein. Taken together, these findings reveal that prodromal intestinal inflammation potentiates adverse effects of LRRK2 G2019S in a sex-dependent manner, to promote α-synuclein phosphorylation in the SN, α-synuclein accumulation within SN microglia, loss of dopaminergic neurons, and motor impairment, particularly in genetically predisposed males.

To further assess the specificity of this gene-environment interaction to the G2019S mutation in human LRRK2, we tested the effects of prodromal intestinal inflammation in two additional mouse lines—mice harboring a knock-in (KI) G2019S mutation in the endogenous mouse LRRK2 (mLRRK2^G2019S KI³⁰) and transgenic mice expressing human WT LRRK2 (hLRRK2^WT31,). In contrast to the increased sensitivity of hLRRK2^G2019S Tg mice to acute DSS treatment (Fig. 1), mLRRK2^G2019S KI mice exhibited DSS-induced body weight loss that was mostly comparable to that seen in WT littermate controls (Fig. S7a–c). Despite no overt differences in responsiveness to prodromal intestinal inflammation, both male and female mLRRK2^G2019S KI mice treated with DSS went on to develop motor deficits in the open field, rotarod, pole descent, and grip strength tests at 15 and 30 weeks after DSS treatment, as compared to treated WT littermate controls (Fig. S7d–h). These motor impairments were largely absent in male and female mLRRK2^G2019S KI mice that were treated with vehicle (water) instead of DSS (Fig. S8), indicating that prodromal intestinal inflammation expedites and/or exacerbates motor impairment induced by the G2019S mutation in mouse LRRK2.

Although overexpression of WT LRRK2 is used to mimic the elevated kinase activity induced by the G2019S mutation³¹, hLRRK2^G2019S, and hLRRK2^WT Tg mice demonstrated distinct responses to DSS-induced intestinal inflammation. While hLRRK2^G2019S mice displayed increased sensitivity to DSS treatment compared to treated WT controls (Fig. 1), hLRRK2^WT Tg mice exhibited modest reductions in body weight loss in response to acute DSS treatment, with males showing poor recovery relative to females and WT controls (Fig. S9a–c). Despite these differential acute responses to DSS treatment, both male and female hLRRK2^WT Tg mice developed motor impairments in response to prodromal intestinal inflammation that were detectable at 15 weeks and 30 weeks after DSS treatment (35 and 50 weeks of age), with male hLRRK2^WT Tg mice exhibiting more severe behavioral defects than females, relative to their respective sex- and treatment-matched WT controls (Fig. S9d–h). There were nuanced differences in the nature of motor impairments observed in hLRRK2^WT and hLRRK2^G2019S mice, which may point to effects of elevated LRRK2 kinase activity due to human LRRK2 overexpression (seen comparably in both lines) versus the added effect of the G2019S mutation on kinase hyperactivity (in the hLRRK2^G2019S line). In particular, male hLRRK2^WT Tg mice exhibited significantly reduced hindlimb grip strength, which was consistent with outcomes seen in hLRRK2^G2019S males (Fig. 2). However, hLRRK2^G2019S males also exhibited reduced forelimb grip strength, which was not seen in the hLRRK2^WT line, suggesting an effect of the G2019S mutation on severity of reduced strength. Moreover, male hLRRK2^G2019S mice exhibited hyperactivity in the open field, which contrasts the hypoactivity seen in hLRRK2^WT males, and male hLRRK2^WT mice also exhibited increased latency to descend in the pole test, which was not seen in the hLRRK2^G2019S line. Notably, both mLRRK2^G2019S KI and hLRRK2^WT Tg developed genotype- and DSS-dependent motor impairments despite no increases in sensitivity to DSS treatment, indicating that the severity of response to DSS-induced intestinal inflammation does not correlate with the severity of motor impairment; rather, DSS treatment may serve as an environmental trigger of mechanisms that expedite or enhance pathogenesis in the context of genetic predisposition. Moreover, the bias for genetically predisposed males to develop more severe and earlier onset motor impairment in response to prodromal intestinal inflammation was seen only in mice expressing human LRRK2, and human LRRK2 G2019S especially, which may be due to transgenic overexpression of the proteins or species-specific differences in the activity of human vs. mouse LRRK2³². Altogether, findings from these experiments reveal that early DSS treatment reproducibly promotes the onset and severity of motor impairment across three different LRRK2-based mouse models for PD. These results strongly suggest that gut-brain interactions between prodromal intestinal inflammation and LRRK2 G2019S elevate the risk for PD.

The heightened susceptibility of male hLRRK2^G2019S Tg mice to PD-related motor impairments and neuropathological abnormalities in response to DSS treatment (Figs. 1 and 2) presents the opportunity to identify mechanisms underlying the ability of biological sex to modify risk for PD. To gain insight, we first asked whether there are sex differences in the expression of endogenous mouse LRRK2 or human LRRK2 G2019S in male vs. female hLRRK2^G2019S Tg mice and WT littermate controls at baseline or 7 days after DSS treatment. In contrast to a prior report that LRRK2 expression is increased in inflamed tissues³³, we observed no change in the expression of endogenous mouse LRRK2 transcript in the colon and decreased expression of transgenic human LRRK2 transcript in the colon on day 7 of DSS treatment, with no sex differences at baseline or after DSS treatment (Fig. S10). As expected, we observe a genotype-dependent increase hLRRK2 mRNA expression and phosphorylated LRRK2 protein in striatal tissue of hLRRK2^G2019S Tg mice relative to wildtype littermates, but with no sex differences (Fig. S11). These findings suggest that the observed sex differences in response to prodromal intestinal inflammation are not due to differential expression of LRRK2 between males and females.

To evaluate roles for sex chromosomes in causing greater male risk for PD-related behavioral and neuropathological symptoms in response to prodromal intestinal inflammation³⁵, we mated XX female hLRRK2^G2019S Tg mice with XY^–Sry gonadal male mice in which the testis determining gene Sry was removed from the Y chromosome and inserted autosomally^35,36. In so doing, we generated XY and XX gonadal female carriers of the human LRRK2 G2019S transgene. XY and XX gonadal female hLRRK2^G2019S Tg mice were subjected to three rounds of DSS treatment (Fig. 3e). Both XX and XY gonadal female hLRRK2^G2019S Tg mice exhibited severe body weight loss during the acute phase of DSS treatment (Fig. S13a), which was comparable to the genotype-dependent enhancements in body weight loss seen in conventional female hLRRK2^G2019S Tg mice (Fig. 1b). Notably, compared to XX controls, XY gonadal female hLRRK2^G2019S Tg mice also exhibited delayed body weight loss (Fig. S13a), similar to that seen in male hLRRK2^G2019S Tg mice treated with DSS (Fig. 1b). This indicates that conferral of XY sex chromosomes to gonadal female hLRRK2^G2019S Tg mice phenocopies the body weight responses to DSS treatment seen in male hLRRK2^G2019S Tg mice.

To further assess the roles of sex chromosomes in regulating risk for motor impairment, DSS-treated XY and XX gonadal female hLRRK2^G2019S Tg mice were subjected to behavioral testing in a battery of motor tasks at 15 and 30 weeks after DSS treatment (35 and 50 weeks of age). DSS-treated XY gonadal female hLRRK2^G2019S Tg mice exhibited decreased moving distance in the open field, weaker grip strength, and longer latency to maneuver in the pole descent test, compared to treatment-matched XX hLRRK2^G2019S Tg littermate controls (Fig. 3d–h; Fig. S13b, c). We further examined the PD pathology in the brains of XY female hLRRK2^G2019S Tg mice after behavioral testing (52 weeks of age). DSS-treated XY female hLRRK2^G2019S Tg mice exhibited significantly increased levels of phosphorylated α-synuclein in the SN relative to DSS-treated XX hLRRK2^G2019S Tg controls (Fig. 3i, j). Increased burden of phosphorylated α-synuclein was also seen in SN microglia of DSS-treated XY female hLRRK2^G2019S Tg mice relative to treatment-matched XX controls (Fig. 3l, m), which phenocopies the sex-dependent neuropathological phenotypes seen in male hLRRK2^G2019S Tg mice exposed to prodromal intestinal inflammation (Fig. 2e–i). There were modest, but not statistically significant, decreases in TH⁺ neurons in DSS-treated XY female hLRRK2^G2019S Tg mice relative to XX controls (Fig. 3k). Overall, these results reveal that differences in effects of XY vs. XX sex chromosomes, not testicular hormones, likely cause sex differences in gene-environment interactions between prodromal intestinal inflammation and LRRK2 G2019S that increase risk for behavioral and neuropathological features of PD particularly in male hLRRK2^G2019S Tg mice.

Finally, we further considered the potential onset of features of neuroinflammation. Consistent with the elevated phosphorylated α-synuclein detected in SN microglia of DSS-treated male hLRRK2^G2019S Tg mice at symptomatic ages (52 weeks, Fig. 2h–j), increased phosphorylated α-synuclein levels were similarly detected in SN microglia of male, but not female, hLRRK2^G2019S Tg mice at a pre-symptomatic time point immediately after the third round of DSS treatment (~19 weeks, Fig. 4d–f; Fig. S14a). There were no differences in levels of CD4⁺ T cells, CD8⁺ T cells or B cells in the brain parenchyma, or in various pro-inflammatory cytokines in the cerebrospinal fluid (CSF) of DSS-treated hLRRK2^G2019S Tg mice relative to vehicle-treated controls (Fig. S14b–e). Consistent with alterations in serum cytokines detected on day 7 of DSS treatment (~11 weeks), serum IL-12p40 was similarly elevated in male and female hLRRK2^G2019S Tg mice after the 3 rounds of DSS treatment (~19 weeks) (Fig. S14f, h), suggesting chronic elevation of this cytokine in response to DSS-induced intestinal inflammation. Serum IL-17A was also elevated at this time point (Fig. S14g). However, there were no sex differences in the DSS-induced increases in select serum cytokines (Fig. S14h). Taken together, these data indicate that DSS-induced intestinal inflammation leads to a sex-dependent accumulation of total α-synuclein in colonic tissue, increase in α-synuclein⁺ macrophages in the colonic lamina propria, and elevation of phosphorylated α-synuclein loads within microglia of the SN, which all precede the sex-dependent onset of PD-related neuropathology and motor impairments in male hLRRK2^G2019S Tg mice. Overall, findings from this study indicate that early intestinal disruptions causally exacerbate pathology and motor dysfunction conferred by various genetic and environmental risk factors for PD.

hLRRK2^G2019S Tg (JAX #018785), hLRRK2^WT Tg (JAX #012445³¹,), and mLRRK2^G2019S KI (JAX #030961³⁰,) mice were purchased from Jackson Laboratories, and bred in at the UCLA Center for Health Sciences barrier facility. We confirmed expression of α-synuclein protein in the hLRRK2^G2019S Tg (JAX #018785) mouse line, which is consistent with previous reports⁶³. XY^–(Sry⁺) male mice, fathers of the “Four Core Genotypes” model (JAX #010905⁵¹,) were bred at UCLA. Animals were maintained on a 12-h light-dark cycle in a temperature-controlled environment, with autoclaved bedding. XY⁻(Sry⁺) males have been backcrossed to the C57Bl/6 J background for multiple generations to establish congenicity. This aligns with the hLRRK2^G2019S Tg mouse line, which is also on the C57Bl/6 J background. Sterile water and autoclaved standard chow (Lab Diet 5010) were provided ad libitum. For all experiments, wild-type littermates were used as controls. All experiments were performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals using protocols approved by the Institutional Animal Care and Use Committee at UCLA.

All surgical procedures were performed under sterile conditions. Prior to anesthesia, the mouse was injected with 0.1 ml carprofen and sterile lactated ringer’s (1:99). The mouse was then anesthetized with isoflurane, shaved at the site of incision (scrotum), administered purolube petroleum jelly on the eyes to prevent dehydration, and placed on a warm pad heated by circulating water bath. The scrotum was washed three times alternately with betadyne and 70% ethanol, and then a small cut was made in the scrotum and cremaster muscle. The testis was gently extruded through the incision site, clamped with a hemostat and ligated with an absorbable suture, then cutoff. The hemostat was removed, the cremaster was then sutured, and the scrotum skin was shut with wound clips. The procedure was then repeated on the other side. The incision sites were treated with betadyne and topical antibiotics (Taro: polymyxin B sulfate, bacitracin zinc, and neomycin sulfate in petrolatum). The wound clips were removed 7–10 days later under light isoflurane anesthesia.

10–13-week-old mice were given 2% DSS in autoclaved water for 1 week, followed by 2 weeks recovery, which was repeated an additional two times for a total of three rounds of DSS treatment to model prodromal ulcerative colitis²⁰. Body weight, stool consistency, and rectal bleeding were assessed daily during the 1 week of DSS treatment and 1 week following each DSS treatment. Clinical scores were calculated according to methods described in ref. ²¹.

Each round of DSS treatment involves 7 days of DSS administration followed by 7 days of recovery, which exceeds the length of the estrous cycle. Levels of the serum cytokines IL-10 and IFNg, but not IL-12, IL-9, or IL-33, are reported to vary depending on the stage of estrous⁶⁴. As such, we predict that the variation in IL-12, IL-9, IL-33 levels in female mice is not due to differences in estrous stage. For CSF collection, mice were injected intraperitoneally with ketamine (100 mg/kg) and xylazine (10 mg/kg), and the cisterna magna was punctured according to methods detailed in ref. ⁶⁵. For serum collection, whole blood was collected from the aorta abdominalis, left at room temperature for 15 min to coagulate, and centrifuged at 2000 × g for 15 min. Serum and CSF samples were assessed for the cytokines IL-22, IFNg, IL-1b, IL-18, IL-23, IL-6, IL-4, IL-10, TGFb, TNFa, IL-33, IL-17A, IL-12p40, and IL-13 using a custom LEGENDplex kit according to manufacturer’s instructions (Biolegend). Flow cytometry was performed using a BD Celesta flow cytometer in UCLA’s Broad Stem Cell Research Center Flow Cytometry Core facility.

Colon lamina propria leukocytes were isolated following procedures adapted from ref. ⁶⁶. Briefly, colons were aseptically dissected from mice anesthetized with isoflurane, rinsed twice in ice-cold PBS, and incubated while shaking for 20 min at 37 °C in HBSS buffer containing 5 mM EDTA and 10 mM HEPES. Following 20 s vortex, supernatant was discarded, and remaining tissues were incubated while shaking for 20 min at 37 °C in pre-warmed RPMI buffer containing 4% FCS, 0.5 U/ml dispase, 0.5 mg/ml collagenase D, 0.25 mg/ml DNaseI. Following 20 s vortex, supernatant was discarded, and tissues were minced in pre-warmed buffer and digested while shaking for an additional 45 min at 37 °C. Following 20 s vortex, digested suspension was passed through a 70 μm cell strainer, washed with ice-cold RPMI, and centrifuged at 2000 rpm for 10 min at 4 °C. Pelleted cells were resuspended in complete RPMI, stained with LIVE/DEAD Fixable Aqua Stain (Thermofisher) for 20 min following the manufacturer’s instructions, and then stained for subsets of markers at 5 μg/ml: Ly6G-BV421 (Biolegend, 127628), Ly6C-BV605 (Biolegend, 128036), CD4-BV650 (Biolegend, 100469), NK1.1-BV711 (Biolegend, 108745), B220-FITC (Biolgend, 103205), CD8a-PerCP-Cy5.5(Tonbo bioscience, 65-0081-U025), CD45-PE-Cy7 (Tonbo bioscience, 60-0451-U100), CD11c-APC-eFlour780 (Thermofisher, 47-0114-82), CD3-AF700 (Biolegend, 100215), CD11b-PE-eFlour610 (Thermofisher, 61-0112-80), MHCII-FITC (Thermofisher, 11-5321-82), CD64-PE-Cy7 (Biolegend, 139313). FMOs were used to determine the gating strategy for select markers without clear populations, such as B220-FITC, Ly6C-BV605, NK1.1-BV711, CD11b-PerCP, and CD64-PE.

For intracellular staining, cells were fixed and permeabilized using the Fixation/Permeabilization kit (BD Biosciences) following the manufacturer’s instructions, and then stained for markers at 5 μg/ml for IL-17A- FITC (Thermofisher, 11-7177-81) and IFNg-PE (Thermofisher, 12-7311-81), 10 μg/ml for TNFa-BV421 (Biolegend, 506327), IL-1b-PE (Thermofisher, 12-7114-82), and α-synuclein-APC (Novus Biologicals, NBP1-05194AF647). For intracellular alpha-synuclein staining, FMO controls for live-dead aqua, CD45-PeCy7 (Biolegend, 103113), CD11b-PerCPCy5.5 (Biolegend, 101227), and CD64-PE (Thermofisher, 12-0641-82), were used to determine the gating strategy for colonic macrophages. To gate on positive alpha-synuclein cells, mouse IgG1-AF647 was used as isotype control (Novus Biologicals, NBP1-97005AF647). Stained cells were resuspended in PBS, filtered through a 0.45 um strainer, and assessed via a BD Fortessa or Attune NxT flow cytometer and analyzed using FlowJo (Fig.). S15

Brain leukocytes were isolated following procedures adapted from⁶⁷. Briefly, anesthetized mice were transcardially perfused with PBS, intact brains were dissected, stored in Hibernate-A medium (Thermofisher), and then transferred into a dounce homogenizer containing HBSS supplemented with 10% FBS, glutamine, and pen/strep. Tissues were homogenized by passing with slide plunger 6–8 times, triturated using a 5 ml serological pipet 5–6 times, and then passed through a 70 um filter with an additional wash with buffer. Suspension was centrifuged for 5 min at 500 × g, and supernatant was discarded. Cells were resuspended in 1 ml 30% percoll, inverted 4–5 times, and then centrifuged for 7 min at 500 × g. Myelin layer was carefully decanted, pellet was resuspended in ice-cold PBS, and centrifuged for 5 min at 500 × g. Finally, cells were resuspended in PBS and stained for markers described above.

Behavioral testing for motor ability was conducted at the UCLA Behavioral Testing Core. On each testing day, mice were habituated to the testing room for at least 30 min before initiating. Behavioral equipments were cleaned with 70% ethanol and Accel disinfectant before and after each session. Each behavioral test was conducted with at least 3 days of recovery between different tests.

Colon and striatal tissue was lysed in RLE buffer, and RNA was extracted using the RNeasy Kit (Qiagen). cDNA was synthesized by qScript cDNA SuperMix kit (Quantabio). Each reaction was set up with 1 μL of DNA sample, 5 μL of SYBR green master mix (ThermoFisher Scientific). forward (human lrrk2: 5’- TGATTCTCGTTGGCACACAT-3’, mouse lrrk2:5’-GCACATGCTCTGTCCACTCT-3’) and reverse (human lrrk2:5’- GCCAAAGCATCAGATTCCTC-3’, mouse lrrk2: 5’-CATGGGCATGCTTCTGCATC-3’) primers (Integrated DNA Technologies) at a final concentration of 500 nM each, and ultrapure water (ThermoFisher Scientific) to the final volume of 20 μL. Quantitative PCR with reverse transcription (qPCR) was performed on a QuantStudio 5 thermocycler (ThermoFisher Scientific) with the following conditions: initial denaturation at 50 °C for 2 min and 95 °C for 10 min followed by 40 cycles each consisting of denaturation at 95 °C for 15 sec, annealing at 60 °C for 60 sec and maintain at 4 °C at the last step.

Proteins were extracted from frozen colon following beads beating in ice-cold lysis buffer (1%SDS, 1 mM Na₃V0₄,1 mM NaF, 1 tablet of Pierce Protease and Phosphatase Inhibitor Mini Tablets (Thermofisher, A32961) per 10 ml). Proteins were electrophoresed in 4–12% Bis-Tris gels (Thermofisher, NP0322BOX), transferred to Nitrocellulose membrane, and probed with the following antibodies. Primary antibodies: rabbit anti-LRRK2 Antibody (D18E12) (Cell Signaling #13046), rabbit anti-phospho-LRRK2 (Ser935) Antibody (Cell Signaling #59576), rabbit anti-beta Actin monoclonal antibody (79E2D7) (Thermofisher, A01865-40). Secondary antibody: Goat anti-Rabbit IgG (Thermofisher, 65-6120). HRP signal was detected by Clarity™ Western ECL Substrate (Bio-Rad, 1705060), and protein bands were visualized by ChemiDoc XRS + imager (Bio-Rad Laboratories) and quantified by Fiji ImageJ.

Snap frozen striatums from DSS-treated 52wks WT and LRRK2 males were homogenized in ice-cold 1× DPBS (Gibco) to produce 10% brain homogenates. Total protein concentration was determined using a BCA protein assay kit (Pierce). Protein concentration was standardized to 5 mg/mL using 1x DPBS. Brains were incubated on ice for 20 mins prior to centrifugation for 20 mins at 20,000 × g at 4 C. The supernatant containing the soluble-synuclein fraction was stored at −80 °C and used for ELISA. The insoluble fraction was obtained by resuspending the pellet in 10x Triton-X (Sigma), incubating on ice for 20 mins and then centrifuging for 20 mins at 20,000 × g at 4 C. Triton-X is a relatively mild detergent, typically used to separate monomeric and oligomeric forms of α-synuclein. We used it at 10× concentration in order to confer stronger detergent effects. The supernatant containing the detergent extracted insoluble-synuclein fraction was stored at −80 °C and used for ELISA. Soluble fractions were diluted using a 1:20,000 dilution and the detergent extracted insoluble fractions were diluted to 1:10,000 using the sample buffer provided by the manufacturer. The mouse α-synuclein ELISA kit (Abcam) was used to quantify levels of α-synuclein following the manufacturer instructions.

Statistics and Reproducibility was performed using GraphPad Prism. Significance was determined by two-way ANOVA followed by Sidak’s post hoc test for multiple group comparisons. Comparisons included analyses of effects for genotype and treatment, and genotype and sex. All data are represented as mean ± s.e.m. (standard error of the mean). *P < 0.05 was considered statistically significant. **P < 0.01. ***P < 0.001. Notable non-significant differences are indicated in the figures by “n.s.”.

Further information on research design is available in thelinked to this article. Nature Portfolio Reporting Summary

Supplementary Information Description of Supplementary Materials Supplementary Data 1 Supplementary Data 2 Supplementary Data 3 Supplementary Data 4 Supplementary Data 5 Reporting Summary