The Parkinson's disease–associated kinase LRRK2 regulates genes required for cell adhesion, polarization, and chemotaxis in activated murine macrophages

Leucine-rich repeat kinase 2 (LRRK2) encodes a complex protein that includes kinase and GTPase domains. Genome-wide association studies have identified dominant LRRK2 alleles that predispose their carriers to late-onset idiotypic Parkinson's disease (PD) and also to autoimmune disorders such as Crohn's disease. Considerable evidence indicates that PD initiation and progression involve activation of innate immune functions in microglia, which are brain-resident macrophages. Here we asked whether LRRK2 modifies inflammatory signaling and how this modification might contribute to PD and Crohn's disease. We used RNA-Seq–based high-resolution transcriptomics to compare gene expression in activated primary macrophages derived from WT and Lrrk2 knockout mice. Remarkably, expression of a single gene, Rap guanine nucleotide exchange factor 3 (Rapgef3), was strongly up-regulated in the absence of LRRK2 and down-regulated in its presence. We observed similar regulation of Rapgef3 expression in cells treated with a highly specific inhibitor of LRRK2 protein kinase activity. Rapgef3 encodes an exchange protein, activated by cAMP 1 (EPAC-1), a guanine nucleotide exchange factor that activates the small GTPase Rap-1. Rap-1 mediates cell adhesion, polarization, and directional motility, and our results indicate that LRRK2 modulates chemotaxis of microglia and macrophages. Dominant PD-associated LRRK2 alleles may suppress EPAC-1 activity, further restricting motility and preventing efficient migration of microglia to sites of neuronal damage. Functional analysis in vivo in a subclinical infection model also indicated that Lrrk2 subtly modifies the inflammatory response. These results indicate that LRRK2 modulates the expression of genes involved in murine immune cell chemotaxis.

identified many variants in this protein that predispose to late-onset Parkinson's disease. The most common mutation in the LRRK2 gene results in a change from glycine to serine at amino acid 2019 (G2019S). This SNP is the highest known risk factor for the development of Parkinson's disease, accounting for 5-7 % of autosomal-dominant familial cases (4, 5) and 1-2 % of sporadic cases in western populations (6). GWAS studies have also revealed a link with Crohn's disease and leprosy. Genetic links with these diseases demonstrate a non-neuronal, but clearly innate immune component to LRRK2 biology (7).
LRRK2 expression is enriched in macrophages, B-cells and dendritic cells (8).
Innate immune stimuli such as interferon gamma (IFN-γ) stimulate LRRK2 expression, revealing a responsiveness to the activation of innate immune signalling pathways mediated by pattern recognition receptors (9). Furthermore activation of Toll-like receptors (TLRs) by pathogen associated molecules such as bacterial lipopolysaccharide (LPS) leads to phosphorylation of LRRK2 by IkB kinase at two serine residues (Ser 910, Ser935) (10). The IkB kinase family is normally associated with the phosphorylation of IκB proteins that sequester NF-κB in the cytoplasm. Phosphorylation and ubiquitination of IκB proteins leads to proteolysis, and subsequent transfer of NF-κB into the nucleus (11). LRRK2 phosphorylation is dependent on the TLR adaptor (Myd88), an innate immune signal transducer that mediates signalling from cell surface TLRs, as well as TLR7, 8, and 9 which signal from the endosomal compartmen The role of inflammatory processes in the aetiology of Parkinson's is further illustrated by administration of LPS both systemically and directly into the substantia nigra. In the latter case LPS causes irreversible degeneration of dopaminergic neurons of the pars compacta, observed a week after injection (12,13). Notably non-dopaminergic neurons of the nigrostriatal system, as well as proximal dopaminergic neurons not associated with the nigrostriatal pathway, remain unaffected by direct LPS injection. Therefore LPS injection and the resulting inflammatory insult demonstrates remarkable sensitivity and specificity to the dopaminergic circuitry associated with Parkinson's disease. Another study demonstrated that the same pattern of Parkinson's disease-like microglial activation, followed by neurodegeneration over ten months was observed when LPS or TNFα were administered systemically in mice via intraperitoneal injection (14). In humans, a laboratory worker accidentally exposed to Salmonella derived LPS developed many symptoms of Parkinson's disease including bradykinesia, rigidity and tremor at rest, as well as other neurological problems resulting from damage to the substantia nigra as well as the cerebral cortex (15). Human Parkinsonism has been further linked to immune activation through the role of neurotrophic viral infection, and in particular, infection by the human influenza virus. Individual cases of viral infection leading to neuropathology and death have been reported, as well as increased incidence of Parkinson's disease following pandemic flu, such as experienced in 1918 (16).
At present little is known about how LRRK2 modifies the gene expression programme induced by innate signalling pathways and how this might contribute to Parkinson's disease initiation and progression. In this study we have used high-resolution transcriptomics to compare gene expression in primary macrophages derived from wild-type and LRRK2 deficient mice. This analysis reveals that LRRK2 modulates the expression of a small subset of genes that are involved in chemotaxis and membrane remodelling.

mRNA sequencing, quality control and read mapping
We have used RNA sequencing to determine how LRRK2 modifies gene expression in primary bone marrow derived macrophages from wild type and LRRK2 -/-mice. If valid comparisons are to be made it is necessary to confirm the similarity in nature and purity of macrophage cultures before RNA extraction and RNA sequencing. One day prior to RNA extraction, a portion of differentiated macrophages were prepared for flow cytometry analysis and stained for various cell surface markers: CD11b for cells of the myeloid lineage, F4/80 for mouse macrophages, and CD11c for monocytic-derived cells, including macrophages (17). These markers revealed no significant differences in the differentiation state of the cells, with uniform expression of CD11b, and highly similar expression levels of F4/80 and CD11c. CD11c surface expression in LRRK2 KO macrophages displayed a slightly higher level of variability between cultures than WT equivalent cells ( Fig S1). Overall, cultures were considered similar enough to proceed with differential gene expression analysis.
We then treated WT and LRRK2 -/macrophages with either LPS or muramyl dipeptide (MDP), activators of the TLR4 and NOD2 mediated innate responses respectively. After two hours of stimulation RNA was extracted for mRNA sequencing. A mean read depth of over 22.2x10 6 reads/sample was achieved with a range of 16.0x10 6 -24.9x10 6 reads/sample (Table S1). Reads were of high quality, requiring a mean of less than 0.1 % of reads to be trimmed during quality control. A mean of 87.5 % of reads could be unambiguously mapped to gene encoding regions of the genome. Therefore by comparing the frequency of reads per gene between samples, relative levels of expression can be determined. These datasets were then analysed for differential gene expression.

Differential gene expression
Datasets of mapped counts were interrogated for differences in each LRRK2 genotype upon innate immune stimulation, as well as for underlying differences between genotypes in unstimulated cells (Fig 1a). DEseq2 (18) determines a statistical model accounting for variance in counts per gene, and base mean of counts allowing the statistical significance of apparent differences in gene expression to be estimated. This analysis revealed LPS treatment caused differential expression of 4985 and 5354 genes in WT and LRRK2 -/-macrophages respectively. By contrast MDP treatment resulted in 1483 significantly differentially expressed genes in WT macrophages, compared to 1478 genes in LRRK2 KO macrophages (Fig 1b).
In the absence of stimulation only eight genes were significantly differentially expressed ( Table 1). One of these is LRRK2 a result that . is confirmed by qPCR. The lesion in the LRRK2 KO mouse deletes part of exon 1 and exon 2 leading to termination at an out of frame stop codon in exon 3. As the LRRK2 transcription unit has 51 exons this should lead to nonsense mediated decay (NMD) of the transcript. The level of transcript measured is about 50% of the wild-type which indicates that NMD is inefficient in this case (19) . Other genes regulated include Kif21a: a member of the kinesin family of motor proteins, Camk2b, a calcium/calmodulin responsive protein kinase, Cd59a: a regulator of the membrane attack complex in mice, and Nnt: a NAD(P) transhydrogenase with implications in defence against oxidative stress. Very little is known about the Lrmda gene except that it consists of a region of LRRs. The remaining results are not represented at the protein level and so are unlikely to have relevance to the current study. The gene detected as being of the highest significance, Gm14150, is described as a pseudogene, produced by the incorporation of reverse transcribed mRNA into the genome, while Gm44305 is a retained intron. These are likely not differentially expressed genes, but preexisting genomic difference between strains (20).
The broad characteristics of the MDP and LPS responses, as well as similarities and differences between WT and LRRK2 deficient macrophages in their response to innate immune activation were visualised with 'volcano' plots ( Fig. 2). In both LPS and MDP experiments, a greater number of genes were up-regulated than down-regulated. LPS treatment led to 3192 genes being up-regulated and 2696 down-regulated, MDP treatment caused 1020 genes to be upregulated and 676 down-regulated (Fig 1). Furthermore, quantification confirmed that a greater number of genes were differentially expressed upon LPS treatment in LRRK2 deficient macrophages than WT macrophages. Perhaps the most interesting observation from this analysis is that a single gene was found to be down-regulated in wild type macrophages and up-regulated in LRRK2 -/-macrophages upon treatment with LPS. Transcription of this gene, Rapgef3, is almost halved upon LPS treatment in WT macrophages while being increased just over 7-fold in LRRK2 null macrophages; a complete reversal in transcriptional regulation upon loss of LRRK2.

Two parameter analysis identifies a small subset of differentially responding genes that are involved in cell motility and chemotaxis
In contrast to conventional differential gene expression analysis, two-parameter analysis provides an alternative method to identify differentially responding genes between genotypes (21) (see http://bioconductor.org/packages/release/bioc/vi gnettes/edgeR/inst/doc/edgeRUsersGuide.pdf, section3.5). This method revealed eleven genes with significantly different responses to LPS as between wild type and LRRK2 -/-macrophages (Padj < 0.1) (Fig. 3, Table 2). All differentially responding genes showed an increased level of transcription upon LPS stimulation in LRRK2 KO cells compared to WT cells. Rapgef3 with a Padj value of 5 x 10 -37 and a fold change of about 11 was the gene most strongly regulated by the presence of LRRK2. Four other genes identified encode either chemokine ligands (Ccl3, 4 , 5) or receptors (Ccrl2) that mediate chemotactic responses, suggesting a common theme of regulated motility (see Discussion).
In order to confirm the results of the RNASeq experiments qRT-PCR was used to directly measure the level of 6 of the 11 genes identified in WT and LRRK2 -/-macrophages (Fig. 4A). This confirms the results of the RNASeq analysis for Rapgef3, chemokines and transcription factor Atf-3. We next asked whether chemical inhibition of the LRRK2 kinase also induces the expression of Rapgef3. We treated wild-type and LRRK2 -/-macrophages with the highly specific inhibitor GSK2578215A (22). Rapgef3 was induced about 6 fold in treated, wildtype macrophages as compared to untreated controls. By contrast no differences in expression level were detected when LRRK2 mutant macrophages were treated with GSK2578215A relative to the untreated controls (Fig. 4B). Thus, LRRK2 kinase activity is required for the observed regulation of Rapgef3 gene expression, consistent with the results of the RNASeq analysis.
In contrast to LPS, MDP treatment identified no significant differentially responding genes (Fig. 3c,d). This aligns with the less immunogenic nature of MDP compared to LPS stimulation and demonstrates the high stringency of the two-parameter method.

Elevated levels of EPAC-1 protein in activated macrophages lacking LRRK2.
To determine whether of Rapgef3 mRNA and EPAC-1 protein levels are correlated we stained LRRK2 deficient and wildtype macrophages with a fluorescent monoclonal antibody specific for EPAC-1. As shown in Fig.  5a EPAC-1 is ubiquitous and in many cells is distributed in the expected punctate, peri-nuclear pattern (Fig. S2). We then quantified protein levels and 6 hours after treatment with LPS the LRRK2 deficient cells have significantly higher levels of EPAC-1, consistent with the RNASeq and qPCR results (Fig. 5b). We were unable further validate these observations using western blot as the available antibodies are insufficiently specific or sensitive in this assay.

A sustained inflammatory response in LRRK2 knock out mice.
To explore whether LRRK2 affects innate immune function we have used a model of sub-clinical bacterial infection. Wild type and LRRK2 deficient mice were infected with Salmonella typhimurium and three markers of inflammation, interleukin 18 (IL-18), interferon g (IFNg) and splenomegaly were measured. As shown in Fig. 6 LRRK2 -/-mice have elevated levels of both IL-18 and IFNg compared to controls at 14 days post challenge. These results are statistically significant (p< 0.05, with the exception IFNg at 14 days p=0.06; see legend to Fig 6). This indicates that the absence of LRRK2 causes an enhanced inflammatory response.

Discussion
In this study we investigated how LRRK2 modifies inflammatory signalling mediated by LPS and MDP. We have identified a small subset of genes that are activated by LPS in macrophages that lack LRRK2 but not in wildtype control cells. About half of these molecules are involved in cell migration, motility and chemotaxis. Of particular note is the guanine nucleotide exchange factor (GEF) Rapgef3, by far the most strongly induced gene identified with an 11-15 fold increase in transcription in the absence of LRRK2. Rapgef3 is also the only gene that is down regulated in wild type macrophages but up regulated in LRRK2 -/-cells, as compared to unstimulated control cells. Rapgef 3 is also derepressed by treatment of macrophages with the highly specific kinase inhibitor GSK2578215A (22), indicating that inter-or intra-molecular phosphorylation is required. Rapgef3 encodes EPAC-1 a GEF that mediates cAMP dependent activation of the small G-protein Rap1. EPAC promotes Rap-1 GDP-GTP exchange leading to cell adhesion, cell polarization and enhanced leucocyte chemotaxis (23,24). On the other hand another study found that LPS treatment paralyzes monocyte chemotaxis, an effect that requires activated Rap-1 (25). It is thus likely that LRRK2 indirectly regulates the mobility of macrophages and microglia that have been activated by innate stimuli such as LPS. An attractive hypothesis is that dominant PD associated LRRK2 alleles such as G2019S that have higher constitutive kinase activity may regulate the motility of macrophages by further suppressing EPAC-1 levels. In that regard a recent study found that G2019S microglia and LRRK2 -/-cells when activated by ADP have retarded and enhanced motility respectively as compared to wild-type control cells. G2019S microglia also have an impaired ability to isolate brain injury (26). These authors present evidence for the involvement of focal adhesion kinases, however this may be indirect and a possible role of the LRRK2/EPAC-1/Rap1 axis should be investigated.
LRRK2 is part of an ancient and highly conserved pathway of directional motility (2). In the slime mould Dictyostelium gbpC is one of eleven paralogues of LRRK2. GbpC null cells are severely defective in chemotaxis because they cannot polarize cells effectively and have altered patterns of myosin phosphorylation that are probably mediated by activation of Rap1. It is interesting to note also that in Dictyostelium many LRRK2 paralogues encode GEFs within their modular structure.
As well as EPAC1 four other messages that function in chemotaxis are differentially expressed. Three are chemokines, and one is a chemokine receptor-like protein. CCL3 and CCL4 are also known as 'macrophage inflammatory protein 1'; MIP-1α and MIP-1β respectively. CCL5 is also known as 'regulated on activation, normal T cell expressed and secreted' (RANTES). These chemokines are all members of the 'CC Chemokine/Receptor family' and share a common receptor in CCR5. CCL3 and CCL5 may also bind CCR1, while CCL5 binds a further receptor, CCR3 (27). These chemokines are all classified as proinflammatory, meaning they are induced by inflammatory stimuli to recruit inflammatory cells to a site of inflammation, as opposed to homeostatic chemokines, which are constitutively expressed in certain tissues (28). CCRL2 is chemokine receptor-like protein, with over 40 % sequence identity to CCR1, CCR2, CCR3 and CCR5 (29) and highest amino acid sequence similarity to CCR1. Interestingly, CCRL2 has been reported as a non-canonical receptor for CCL5 (30), as well as CCL19 and chemerin (31). A microarray screen of unstimulated mouse microglia has shown that CX3CR1, a non-canonical chemokine receptor of fractalkine, expressed exclusively in microglia, is upregulated by knockout of LRRK2 (32). This reaffirms that chemokine responses may be involved in LRRK2 biology in diverse immunological contexts.
Message encoding the transcription factor ATF3 is induced by 3 fold in the absence of LRRK2 and is a negative regulator of proinflammatory TLR4 signalling, a part of the LPS induced negative-feedback loop (33,34). Thus it is possible that G2019S mutation of LRRK2 might down regulate ATF3, resulting in attenuated negative feedback of TLR4 signalling, enhanced inflammation, and greater neuronal stress. Other transcription factors messages identified encode MXD1 and CSRNP1. MXD1 acts in a network with MYC and MAX, forming the MYC/MAX/MXD1 axis (33). MXD1 is in competition with MYC for the binding of MAX leading to a mixture of MXD1/MAX and MYC/MAX dimers controlling transcriptional output. In the context of cancer, MYC signalling affects cell adhesion, cell shape, and reduces cell migration through modulation of the actin cytoskeleton (35). Finally, CSRNP1 is a transcription factor that is upregulated by Axin, as well as inflammatory stimuli in the form of IL-2 (36,37). Axin is a negative regulator of the Wnt signalling pathway, acting to sequester the transcription factor β-catenin to the cytoplasm (38).
Two transcripts identified have been linked directly to Parkinson's disease. Abtb2 encodes 'Ankyrin-rich BTB/POZ domain containing protein-2' (BPOZ-2). This protein causes inhibition of alpha-synuclein aggregation (39). Lentiviral delivery of the BPOZ-2 gene appears to stimulate autophagic clearance of alpha-synuclein, resulting in reduced alphasynuclein pathology in the basal ganglia. Another protein identified is the G protein-coupled receptor, HCAR2, otherwise known as 'niacin receptor 1'. Niacin has been proposed as a treatment for Parkinson's disease, although evidence of efficacy is lacking (40,41). Activation of HCAR2 in macrophages has an anti-inflammatory effect. Activation by niacin results in inhibition of CCL2 induced macrophage migration (42), as well as an inhibited response to LPS stimulation (43) or inflammatory cytokine release (44).
Previous studies of LRRK2 using kinase inhibitors have revealed immunological functions. However these results should be treated with caution owing to significant offtarget effects that act on the ERK pathway (45). We therefore used a sub-clinical infection model to test for inflammatory phenotypes in the LRRK2 -/-mice. Mice were infected with Salmonella typhimurium and this revealed that LRRK2 significantly dampens the inflammatory response. In this model IL-18 comes most likely from myeloid cells, particularly monocytes and the principal inflammatory activator of Salmonella is LPS. This suggests that the observed differences in gene expression reported here cause subtle but significant changes in the inflammatory response in vivo that may be caused by alterations in the chemotactic capacity of monocytes and macrophages It is interesting that only a small number of differentially expressed genes were observed between LRRK2 genotypes under resting conditions. This shows that LRRK2 only exerts an effect on the macrophage transcriptome under stimulated conditions. This finding also reflects work by another group that identified no changes in gene expression in unstimulated human fibroblasts or brain tissue between G2019S LRRK2 carriers and controls (46).
In conclusion the current study identifies the control of directional motility and chemotaxis of macrophages by LRRK2 as a potentially critical mechanism in the aetiology of PD. It suggests that the function and regulation of the LRRK2/EPAC-1/Rap1 axis and how this impacts on pro-and anti-inflammatory properties of microglia should be investigated. If active EPAC-1/Rap1 confers a neuroprotective phenotype on microglia, EPAC-1 specific agonists such as cAMP analog, 8-(4-chloro-phenylthio)-2′-Omethyladenosine-3′,5′-cyclic monophosphate (8-pCPT-2′-O-MecAMP) may have therapeutic value in PD (47).

Experimental procedures
Mice, genotyping and cell culture WT C57BL/6J mice were obtained from Charles River, UK. LRRK2 -/-B6.129X1(FVB)-Lrrk2 tm1.1Cai /J mice were obtained from The Jackson Laboratory, United States (48). All mice strains were bred independently. All work involving live animals complied with the University of Cambridge Ethics Committee regulations and was performed under the Home Office Project License number 80/2572. DNA from Ear snips of LRRK2 -/-B6.129X1(FVB)-Lrrk2 tm1.1Cai /J mice was isolated for genotyping using the Phire animal tissue digest PCR kit (Thermo Fisher Scientific, USA). Genotyping PCR was carried out in accordance with recommendations by the Jackson Laboratory.. Genotyping PCR products were run on a 1 % agarose gel.
For the differentiation and culture of 'primary bone marrow derived macrophages' (pBMDMs), mice were killed between 8 and 16 weeks of age by cervical dislocation, skin was sterilized with 70 % ethanol, and legs removed. Under sterile conditions, the tibia and femur were isolated, cleaned of muscle, and the proximal and distal epiphysis cut away. Bone marrow was flushed out of the bone using primary growth media (Dulbecco's modified Eagle's medium (DMEM) (Thermo Fisher Scientific, USA) supplemented with 10 % foetal calf serum' (FCS) (Thermo Fisher Scientific, USA), 20 % L929 conditioned media and 8 mM L-glutamine (Sigma-Aldrich, USA)). Isolated cells were centrifuged at 300 g for 10 min at 15 °C, and resuspended in 60 ml of growth media, and allowed to grow at 37 °C in 5 % CO2. Cells were supplemented with a further 60 ml of growth media after 2 days, and media replaced every 3 days. All experiments were performed on cells between 6 and 11 days after initial bone marrow isolation. Live cell counts were performed using a haemocytometer with trypan blue staining (Sigma Aldrich, USA).

Animal Infections and Data Collection S. Typhimurium strain M525P (49) a strain of intermediate virulence, was used to establish a subclinical infection in vivo.
In particular, overnight (stationary phase) bacterial cultures were first washed and resuspended in Dulbecco's PBS (D-PBS, Sigma) and then diluted to the desired dose. Wild-type and Lrrk2-/-mice were subsequently challenged intravenously with 0.2ml of the bacterial suspension. The exact dose, as determined by serial dilution and plating the inoculum on LB plates before and after infection, was 1.3x10 4 CFU/mouse. Mice were bled and then euthanised at certain intervals after the initial challenge, their spleens were aseptically removed and weighed. Mouse serum was analysed via ELISA for levels of IL-18 (MBL International) and IFN-g (DuoSet Development kit, R&D Systems).
Flow cytometry 1x10 6 cells/well were plated in 12-well tissue culture plates and left to adhere over night at 37 °C in 5 % CO2. Cells were then resuspended into MACS buffer (PBS supplemented with 2 % FCS, 1 mM EDTA (Merck and Co, USA)) and spun at 300 g for 6 minutes in a conical-bottom 96-well plate (Thermo Fisher Scientific, USA). To block Fc-mediated reactions, cells were re-suspended in MACS buffer supplemented with 1:100 rat antimouse CD16/CD32 functional grade purified (93 clone; eBioscience, USA) and incubated at 4 °C for 15 minutes. Cells were spun at 300 g for 6 minutes, then re-suspended in MACS buffer supplemented with rat anti-F4/80 conjugated with FITC, hamster anti-CD11c conjugated with phycoerythrin and rat anti-CD11b conjugated with PerCP-cyanine5.5. Staining was performed for 30 mins at 4 °C. Cells were then centrifuged at 300 g for 6 mins and re-suspended in MACS buffer 3 x to remove unbound antibody before finally being spun at 300 g for 6 minutes, and resuspended in MACS buffer supplemented with 2 % methanol-free formaldehyde (Thermo Fisher Scientific, USA) to fix. Fixed cells were analysed using an Attune NxT acoustic focusing cytometer (Life Technologies, USA) for triple labeling experiments.

RNA sequencing and transcriptomic data analysis
Bone marrow was isolated from 16-week old female mice housed in the same facility for this study. 3x10 6 cells/well were plated in Greiner 6well tissue culture plates (Sigma-Aldrich, USA) a day prior to RNA extraction and left to adhere over night at 37 °C in 5 % CO2. Where appropriate, cells were then treated with 100 ng/ml ultrapure LPS from E. coli O111:B4, or 10 µg/ml MDP. LPS was sonicated prior to application to cells. After 2 hours incubation at 37 °C in 5 % CO2, cells were washed in PBS, then scraped into PBS at 4 °C. RNA was isolated using RNeasy mini kit in combination with QIAshredder cell homogenization following the manufacturers instructions. To remove genomic DNA, extracted RNA was DNase treated using TURBO DNA-free kit. Resulting RNA was analysed using a Nanodrop 1000 spectrophotometer. Samples with A 260/230 < 1.8 were further purified with RNeasy MinElute Cleanup Kit (Qiagen, Germany). Samples were then flash frozen in liquid nitrogen, and RNA was quantified using a Qubit Fluorometer (Thermo Fisher Scientific, USA). RNA integrity was verified using 2100 Bioanlyser (Agilent Genomics, USA), and mRNA library preparation was performed using TruSeq Stranded mRNA Library Prep Kit (Illumina, USA) with quality control by 2200 Tapestation (Agilent Genomics, USA). High output sequencing runs of single-end 75 bp read length were performed on NextSeq500 (Illumina, USA) using NextSeq 500/550 High Output v2 Kit (75 cycles) (Illumina, USA). A minimum read depth of 18x10 6 reads per sample was achieved. Read pre-processing, mapping with quality control was performed using a standard pipeline. Ensembl Mus_musculus.GRCm38.dna.primary_assembly .fa (release 84) reference genome file was used to map reads, using the annotated transcripts from the ensembl Mus_musculus.GRCm38.84.gtf. Differential gene expression analysis was performed using DESeq2. Analysis was performed as a 'paired comparison experiment' for each treatment group as comparisons between genotype are made between different samples of different mice (unpaired), while comparisons of treated vs untreated samples are made using samples from the same mice (paired). A target frame and design matrix were adapted from an analogous scenario laid out in the EdgeR user guide section 3.5: "Comparisons both between and within subjects" (21). This analysis enabled simple differential gene expression analysis between genotypes, and 2-parameter analysis to compare responses of each genotype to innate immune stimuli by interrogation of a targets frame. This targets frame identifies each sample as belonging to a mouse (mouse.n), and each of these mice as being treated with LPS, MDP, or left untreated (media).

Quantitative reverse transcription PCR
qRT-PCR was performed using SensiFAST SYBR No-ROX One-Step Kit (Bioline, UK) following the manufacturers instructions and appropriate primers selected based on data submitted to the primer bank database (table 4)(50). qRT-PCR reactions were performed using a Rotor-Gene Q (Qiagen, Germany), and quantification of fold-changes of transcript were calculated using cycle threshold values accounting for reaction efficiency (51). For inhibitor studies, cells were pre-treated with 1.3 µM of GSK2578215A for four hours. Following this, they were treated with 100 ng/ml of LPS for a further two hours. Cells were then washed with 1X PBS and harvested. Total RNA was used for RT-qPCR using the Luna One step RT-qPCR kit (NEB) following the manufacturer's protocol.

Immunofluorescence microscopy
Cells were fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X100 (20min), and then blocked by washing with 0.05% Tween 20 in PBS, followed by blocking for 30min at 37ºC in 0.05% Tween 20 0.5% BSA. Cells were immune-stained overnight at 4ºC with Alexa-488 conjugated Anti-EPAC1 monoclonal antibody (Abcam ab201506). The Anti-Epac1 antibody is raised against human-EPAC but to an epitope that is conserved in the mouse homologue. After staining, slides were washed with blocking buffer, PBS, and then mounted in curing mountant with DAPI (Diamond antifade, Thermo Fisher P36966).
Fluorescence was detected under a FV1200 confocal microscope 60x oil immersion objective with integration as 512x512 pixels each 0.413x0.413µm with detector gain set for minimal saturated pixel count (4096 in the 13bit intensity Olympus format). Z-stacks of 5x1µm depth were collected upwards from the coverslip to collect the total cellular fluorescence. Projections summing each pixel in each z series were calculated in ImageJ and the image converted to 'FITS' image format to preserve the Olympus raw pixel intensity count data. Fields of view from media-treated and LPS-treated cells for each genotype were collected in batches of 10 images as close in time as possible. As a result intensity differences within a genotype are more reliable than comparison between genotypes. This approach was chosen so that the differing response of the cell types to LPS could be detected more accurately.
Projection image analysis was with Mathematica 11 -cell bodies were masked using the MorphologicalBinarize function extended by a border of 5 pixels (2µm) for edge inclusion. Cell number was obtained from the binarized DAPI channel using the MorphologicalComponent function to give a count of the nuclei. EPAC1 immuno-fluorescence from both cytoplasm and nuclei compartments were summed typically giving an intensity of millions of Olympus fluorescence units. Data from approximately 200-300 cells in 20 fields of view were analysed for each sample and reported mean intensity given in million units per cell. Cells cut by image borders were included in the pixel counts even if not containing the nucleus (as on average a cell should be cut by a border into two halves one with and one without a nucleus). The mean total intensity per cell was calculated as a measure of EPAC1 expression with a S.E.M. based on number of fields of view included (rather than the number of cells). Ratios of average fluorescence per cell in LPS treated vs medium were calculated at each time point for each genotype. In addition, the ratio of wild-type to knockout cells was calculated.       Wildtype and LRRK2 -/-(n=5) mice were infected with Salmonella typhimurium 525P. IL-18 (a), interferon g (b) and spleen weight (c) were measured. Significance was evaluated with the Man-Whitney test. Differences in IL-18 were significant (day1: P=0.0159 (*), day 14: P=0.0159 (*) day 35: P=0.0095 (**)), IFNg although elevated at day 14 was not significant at the 95% confidence level (p= 0.09), spleen weight was significant at day 14: P=0.0317 (*).