Research ArticleInfectious Disease

Lrrk2 alleles modulate inflammation during microbial infection of mice in a sex-dependent manner

See allHide authors and affiliations

Science Translational Medicine  25 Sep 2019:
Vol. 11, Issue 511, eaas9292
DOI: 10.1126/scitranslmed.aas9292

Lrrk2ing in the shadows

The leucine-rich repeat kinase-2 (LRRK2) gene has been associated with Parkinson’s disease, leprosy, and Crohn’s disease. We examined whether Lrrk2 plays a role in virulent infections. In mice carrying different versions of Lrrk2, we found that Lrrk2 altered the course of bacterial and viral infections by modulating inflammation. Animals expressing the Parkinson’s disease–linked p.G2019S Lrrk2 mutation showed reduced replication of a bacterial pathogen. During viral encephalitis, the p.G2019S Lrrk2 mutation worsened survival in mice, predominantly in females. Animals with viral encephalitis expressing a variant that blocked Lrrk2’s enzyme function showed improved survival. Lrrk2 may modulate the course of microbial infections in a manner that depends on mouse genotype, sex, and the type of pathogen.

Abstract

Variants in the leucine-rich repeat kinase-2 (LRRK2) gene are associated with Parkinson’s disease, leprosy, and Crohn’s disease, three disorders with inflammation as an important component. Because of its high expression in granulocytes and CD68-positive cells, LRRK2 may have a function in innate immunity. We tested this hypothesis in two ways. First, adult mice were intravenously inoculated with Salmonella typhimurium, resulting in sepsis. Second, newborn mouse pups were intranasally infected with reovirus (serotype 3 Dearing), which induced encephalitis. In both mouse models, wild-type Lrrk2 expression was protective and showed a sex effect, with female Lrrk2-deficient animals not controlling infection as well as males. Mice expressing Lrrk2 carrying the Parkinson’s disease–linked p.G2019S mutation controlled infection better, with reduced bacterial growth and longer animal survival during sepsis. This gain-of-function effect conferred by the p.G2019S mutation was mediated by myeloid cells and was abolished in animals expressing a kinase-dead Lrrk2 variant, p.D1994S. Mouse pups with reovirus-induced encephalitis that expressed the p.G2019S Lrrk2 mutation showed increased mortality despite lower viral titers. The p.G2019S mutant Lrrk2 augmented immune cell chemotaxis and generated more reactive oxygen species during virulent infection. Reovirus-infected brains from mice expressing the p.G2019S mutant Lrrk2 contained higher concentrations of α-synuclein. Animals expressing one or two p.D1994S Lrrk2 alleles showed lower mortality from reovirus-induced encephalitis. Thus, Lrrk2 alleles may alter the course of microbial infections by modulating inflammation, and this may be dependent on the sex and genotype of the host as well as the type of pathogen.

INTRODUCTION

Parkinson’s disease (PD) is characterized by progressive neurodegeneration and frequent inclusion formation in surviving dopaminergic neurons. Whereas the etiology of late-onset, “idiopathic” PD remains to be determined, both genetic and environmental factors are known to contribute (1, 2). Braak and Del Tredici (3) hypothesized that PD originates in the enteric nervous system and olfactory bulb, two sites in close proximity to the environment (46). We and others consider the etiology of late-onset PD to be the outcome of interactions among genetic susceptibility (7, 8), encounters with pathogens in the environment and associated tissue responses, with effects due to gender, and the passage of time (9). This concept of “complex disease” development, whereby multiple factors are essential to generate a disease phenotype, is similar to other late-onset diseases in humans (10).

Allelic changes in the gene encoding leucine-rich repeat kinase-2 protein (LRRK2) occur in >1% of typical PD cases. Sexual dimorphism with a female bias (rather than the male bias generally seen in PD) has been described in at least three cohorts of patients with PD carrying LRRK2 mutations (1115). Substitutions at more than six distinct residues in LRRK2 are associated with familial clusters of PD, often in a geographic region–specific manner (1618). The commonest mutation, p.G2019S, confers increased kinase activity on LRRK2 (1921). Thus, LRRK2 kinase inhibition has been actively pursued as a therapy for PD (2225). Nevertheless, it remains unknown how these LRRK2 mutations lead to neurodegeneration (26). Given the relatively low mean penetrance rate for carriers of an LRRK2 mutation across multiple ethnicities, it is plausible that either a genetic or environmental cofactor may be required to express the PD phenotype (27, 28).

Wild-type and mutant LRRK2 protein functions have been mostly explored in the context of the nervous system (20, 2935), where many putative roles for mammalian Lrrk2 have been described including modulation of neurotransmission, vesicular transport, neurite length, maintenance of mitochondrial health, the regulation of autophagy, formation (and processing) of inclusions, and the modulation of microglial responses (32, 3641). Therefore, different concepts related to PD pathogenesis downstream of mutant LRRK2 protein have emerged. The lack of a consensus for the pathogenesis of LRRK2-linked disease has been partially explained by modeling differences and the gene’s low expression rate in those areas of the brain invariably affected by PD (4246).

Allelic LRRK2 variants, of which some await confirmation, have also been associated with Crohn’s disease (CD) and an endophenotype of leprosy (4751). A recently identified variant, p.N2081D, was shown to increase LRRK2 kinase activity and to be associated with both CD and PD (in cohorts of two distinct ethnicities and with a larger effect size for the CD association) (52). A second variant, p.M2397T, was found to be linked to CD and leprosy (47) and thought to affect the turnover of LRRK2 protein (53). The mechanisms by which distinct LRRK2 alleles modulate the risk of developing three different diseases in three organs, i.e., the gut (CD), brain (PD), and peripheral nervous system (leprosy), remain unknown. Other unanswered questions include whether neighboring genes possibly modulate disease susceptibility, whether a shared function of LRRK2 (19, 21, 52) can explain the association of this gene with three very different disorders, or whether LRRK2 exerts distinct functions in different cell types.

Recently, several teams including ours have found higher expression of LRRK2 in immune cells than neurons, namely, in primary macrophages, monocytes, dendritic cells, natural killer (NK) cells, pneumocytes, microglia, and Epstein-Barr virus–transformed immune cells, from both primates and rodents (5461). Moreover, LRRK2 is up-regulated after exposure to microbial pathogens and downstream of interferon-γ (IFN-γ)– and nuclear factor κB (NF-κB)–mediated signaling pathways (54, 62). LRRK2 expression has also been detected in isolated human neutrophils (60, 63). In contrast, markedly lower amounts of LRRK2 have been found in cells of the adaptive immune system (55). In other immune system–related studies, a shared function for Lrrk2 and Nod2 (both linked to CD and leprosy) has recently been shown in intestinal Paneth cells (64). It was also reported that wild-type Lrrk2 regulates renal function and other immune functions such as antigen processing during phagosome maturation (29, 6568).

Some groups have demonstrated that the absence of endogenous Lrrk2 in rodents conferred neuroprotection after the administration of the bacterial product lipopolysaccharide (LPS) (57, 69), the neurotoxin paraquat, or overexpression of SNCA complementary DNA (cDNA) encoding α-synuclein. These effects were attributed to reduced inflammation (58, 70, 71). Asymptomatic human p.G2019S LRRK2 carriers were shown to have altered inflammatory markers in their biological fluids (72).

We hypothesized that a function for LRRK2 lies within the innate immune system to protect mammals against select xenobiotic threats. We also predicted that the outcomes of virulent viral and bacterial infections in mice carrying a mutant Lrrk2 genotype would be altered compared to wild-type animals. Last, we conjectured that any change in disease outcomes would be LRRK2 kinase dependent. We tested our hypothesis by examining the expression of LRRK2 in human leukocytes and tissues during inflammation, by monitoring the outcome of viral and bacterial infections in Lrrk2 mutant animals, and by exploring related signaling events in mouse primary immune cells as well as select tissues.

RESULTS

LRRK2 is highly expressed in human immune cells and tissues

We previously reported the relative abundance of murine Lrrk2 protein in fluorescence-activated cell sorting (FACS)–isolated cells from select organs including the spleen, ileal Peyer’s patches, and mesenteric lymph nodes. In wild-type mice, we had found the highest expression of Lrrk2 in granulocytes, followed by CD68-positive monocytes and dendritic cells (55). Here, we now demonstrate that among leukocyte subtypes isolated from venous blood of healthy human subjects, the highest expression of LRRK2 mRNA was detected in neutrophils (Fig. 1), followed by monocytes and then B cells (Fig. 1A).

Fig. 1 Expression of LRRK2 in human immune cells and tissues.

(A) LRRK2 mRNA expression in primary human immune cells including neutrophils, monocytes, γδ-type T cells, B cells, NK cells, CD4+ T cells, and CD8+ T cells isolated from healthy donors. LRRK2 mRNA expression is also demonstrated in unstimulated macrophages and LPS-stimulated macrophages derived from human monocytes. (B) Expression of LRRK2 mRNA in select human tissues including bone marrow, heart, skeletal muscle, uterus, liver, fetal liver, spleen, thymus, thyroid gland, prostate gland, brain, lung, small intestine, and colon. Fluorescence intensity values [arbitrary units (AU)] represent the mean and SD of three independent measurements. The reference sample is composed of a mixture of mRNAs isolated from 10 human tissue lysates collected from healthy donors. (C to E) Representative images of an abdominal lymph node (C); spleen, lung, and terminal ileum specimens collected at autopsy and surgery (D); and human brain regions and olfactory epithelium (E). Tissue sections (thickness, 10 μm) were stained by immunohistochemistry using antibodies to myeloperoxidase (MPO) or CD68, the polyclonal HL-2 antibody (Ab) against leucine-rich repeat kinase-2 (LRRK2), monoclonal antibody MJFF4 against LRRK2, or anti–α-synuclein (α-SYN) antibody. Antibody dilutions were 1:200 (anti-LRRK2, anti-MPO, and anti-CD68 antibodies) and 1:1000 (anti–α-SYN antibody). Positive immunoreactivity is indicated by the brown color, with a blue hematoxylin counterstain. Absorption experiments were carried out with excess nonspecific antigen [bovine serum albumin (BSA)] or cognate antigen (LRRK2 peptide) to demonstrate specificity of antigen detection by anti-LRRK2 antibody. Black arrows designate anti-LRRK2 reactive leukocytes, single black asterisks identify the lumen of blood vessels, and black arrowheads designate dopaminergic cells containing neuromelanin. Two black asterisks (**) designate the ethmoid sinus adjacent to the olfactory epithelium in the human nasal cavity.

We also found that human LRRK2 transcripts are abundant in immune function–related tissues including the bone marrow, lungs, spleen, and lymph nodes (Fig. 1B). We confirmed these findings in formalin-fixed sections of human organs using routine immunohistochemistry (Fig. 1, C and D) using a well characterized antibody to LRRK2 (55, 73). The cellular distribution for LRRK2 in serial sections of human lymph nodes (within the medullary sinus), the spleen (within splenocytes of the red pulp), and lungs (within pneumocytes) was similar to that of neutrophils and macrophages, as shown by anti-myeloperoxidase antibody staining and anti-CD68 antibody staining (Fig. 1, C and D). LRRK2 detection was shown to be specific, as determined by antigen pre-absorption of the polyclonal antibody and using a second monoclonal anti-LRRK2 antibody (Fig. 1C) (55). LRRK2 reactivity was less associated with typical markers of T cells and B cells (CD20 and CD3, respectively); LRRK2 expression was also detected in LPS-stimulated human macrophages (Fig. 1A) (54, 55).

To determine whether the absence of Lrrk2 expression altered the number of innate immune cells, in either circulating blood or the spleen, we used flow cytometry to select neutrophils (CD11b+Ly6G+), monocytes, and macrophages (CD11b+Ly6G) from wild-type, heterozygous (Lrrk2WT/KO), and Lrrk2-deficient mice (Lrrk2KO/KO) (29). Under these conditions, we found no discernible difference between the three genotypes and observed no sex difference. We also confirmed that murine Lrrk2 was present in circulating immune cells at birth (fig. S1, A to C).

LRRK2 is abundant in tissue-infiltrating leukocytes during inflammation

To probe for LRRK2 reactivity in human tissues during inflammation, we stained sections of gut from patients with CD, brainstem tissue infected by HIV, brain cortex infected by rabies virus, and virally infected peripheral nerve roots (Fig. 1D and fig. S1D). In all cases, we detected robust LRRK2 reactivity in infiltrating and intravascular leukocytes during acute illnesses (rabies encephalitis and peripheral nerve root infection) and chronic diseases (CD and HIV encephalopathy). In contrast, in sections of newborn human gut (before microbial colonization), no LRRK2 signal was detected (Fig. 1D).

We detected either no signal or faint reactivity for LRRK2 in dopaminergic neurons of the substantia nigra and other neurons of the human brainstem, as previously reported (4244, 46). However, select neurons of cerebellar nuclei (dentate nucleus) showed more robust LRRK2 reactivity. In human brainstem, we observed the strongest signals for LRRK2 in intravascular leukocytes including in sections of human midbrain collected at autopsy (Fig. 1E). This was also true for patients with late-onset PD, incidental Lewy body disease, and young-onset PARK2-linked and mutant SNCA-linked parkinsonism (fig. S1D). LRRK2 reactivity was also detected in the human olfactory epithelium collected at autopsy (Fig. 1E). We concluded that LRRK2 appears to be abundant in infiltrating leukocytes of human tissues during acute or chronic inflammation, including virally infected neural tissues. Under these conditions, we rarely detected any LRRK2 signal in glial cells.

Lrrk2 alleles modulate bacterial load in tissues of infected mice

We next sought to test if endogenous murine Lrrk2 altered the outcome of disease caused by an acute bacterial infection (Figs. 2 and 3). We chose a sepsis model induced by the intravenous inoculation of adult mice with Salmonella typhimurium, a facultative anaerobe and Gram-negative bacterium (74). We defined three endpoints in our comparisons of wild-type compared to Lrrk2 mutant mice including quantification of bacterial load in those organs that showed high LRRK2 expression, assessment of innate immune defenses in tissue lysates and primary cell cultures, and monitoring survival during sepsis (Fig. 2).

Fig. 2 Mutant Lrrk2 alleles alter growth rates of S. typhimurium during sepsis in mice.

Results for bacterial replication rates and flow cytometry analyses of select organs from adult mice of different Lrrk2 genotypes, which were inoculated with S. typhimurium (ST) or saline (mock-infected) by tail vein injection, are shown. (A to C) Colony-forming units (CFU) were counted in the spleens of ST-infected mice at day 5 post-infection (dpi). Spleens were collected from (A) wild-type (WT) mice, heterozygous (Lrrk2WT/KO) animals, and Lrrk2 knockout (KO) mice (Lrrk2KO/KO). (B) Spleens were also collected from WT mice, heterozygous mice (Lrrk2WT/GS), and p.G2019S mutant Lrrk2 knockin (KI) animals (Lrrk2GS/GS). (C) Spleens were also collected from WT mice, heterozygous mice (Lrrk2WT/DS), and p.D1994S mutant Lrrk2 KI mice (Lrrk2DS/DS). (D) Representative flow cytometry plots from spleens of ST-inoculated mice and mock-infected mice at 5 dpi, showing numbers of neutrophils, monocytes, and macrophages. (E) Graphic depiction of quantified flow cytometry results in (D). (F) The number of S. typhimurium bacteria were measured using a CFU assay of spleens from p.G2019S/WT heterozygous female mice at day 5 after depletion of neutrophils or neutrophils and monocytes. (G) Fold changes in bacterial CFU counts [shown in (F)] compared to animals injected with an antibody isotype control. *P ≤ 0.05, **P < 0.01, and ***P < 0.001. n.s., not significant. Tests used were one-way ANOVA with Tukey post-comparison for (A) to (C) and two-way ANOVA with Bonferroni post-comparison for (E).

Fig. 3 p.G2019S mutant Lrrk2 elevates ROS and prolongs mouse survival during sepsis.

(A) Reactive oxygen species (ROS) production measured by DCF (2′,7′-dichlorodihydrofluorescein diacetate) staining [relative fluorescence units (RFU)] in mouse bone marrow–derived macrophage (BMDM) cultures 24 hours after S. typhimurium (ST) inoculation. (B) No difference was seen in cell survival rates for BMDMs collected from three different genotypes: WT mice, heterozygous Lrrk2GS/WT mice, and p.G2019S homozygous mutant Lrrk2GS/GS mice. Cell survival was measured using a neutral red assay. (C) ROS production in BMDMs derived from mice infected with reovirus type 3 Dearing (REO) measured by DCF staining. (D and E) Quantification of hydrogen peroxide (H2O2) production in spleen (D) and brain (E) of S. typhimurium (ST)–infected mice (5 dpi) measured by the Amplex Red assay. (F) Pooled data from two biological replicates of the Amplex Red assay on mouse brains [carried out as in (E)] normalized to mock-infected WT. Each symbol denotes one animal. (G) In a separate experiment, perfused brains from ST-infected mice at 5 dpi were processed for flow cytometry. (H) Quantitative assessment of the number of brain-infiltrating CD45HIGH and CD11b+ myeloid cells in (G) (n = 5 mice per genotype). (I) Kaplan-Meier curves for survival rates of S. typhimurium (ST)–infected mice of the following genotypes: WT mice, heterozygous Lrrk2WT/GS animals, and homozygous p.G2019S mutant Lrrk2 KI mice (Lrrk2GS/GS). (J) Kaplan-Meier curves for survival rates of S. typhimurium (ST)–infected mice of the following genotypes: Lrrk2 KO mice (Lrrk2KO/KO), heterozygous Lrrk2DS/WT mice, and p.D1994S mutant Lrrk2 KI mice (Lrrk2DS/DS). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Tests used were one-way ANOVA with Tukey post-comparison for (D) to (F) and (H), two-way ANOVA with Bonferroni post-comparison for (A) to (C), and log-rank test for (I) and (J).

At 5 days post-infection (dpi) of adult mice (8 to 10 weeks old) with S. typhimurium, spleens from male and female wild-type and Lrrk2-deficient mice were collected to quantify the number of replication-competent bacteria as colony-forming units (CFU). We observed that heterozygous Lrrk2 and homozygous Lrrk2 knockout littermates showed a higher rate of bacterial replication (Fig. 2A), which was consistent with a previous report (75). There was a significant sex difference with the female mice showing an about fivefold increase in replication-competent bacteria (wild-type versus Lrrk2 knockout mice; P ≤ 0.05) compared to male mice.

We next assessed the effect of the PD-linked p.G2019S mutation in Lrrk2 using knockin mice. Under the same conditions, heterozygous and homozygous knockin animals showed a significant reduction in CFU numbers for S. typhimurium in their spleens versus wild-type littermates (P < 0.01; Fig. 2B). This difference showed allele number dependency (Fig. 2B), where a homozygous knockin genotype in females generated a ≤10-fold reduction in replication-competent bacteria. A similar trend was seen in liver specimens from a small number of p.G2019S Lrrk2 knockin animals (fig. S2A). We also assessed the effect of a second PD-linked variant, the p.R1441C Lrrk2 mutation that alters Lrrk2 guanosine triphosphatase (GTPase) activity (37, 76), in a smaller study. S. typhimurium CFU counts showed a similar trend in organs harvested from female heterozygous and homozygous mice, but the difference did not reach significance (fig. S2, B and C). Our results suggested that wild-type Lrrk2 expression conferred relative protection against Salmonella replication in infected tissues, which was enhanced by the PD-linked p.G2019S Lrrk2 mutation.

To test the role of enzyme activity in the Salmonella infection mouse model, we then examined kinase-dead Lrrk2 p.D1994S mutant knockin mice (56). Unexpectedly, the loss of Lrrk2 kinase activity, which we monitored by Lrrk2 autophosphorylation in vivo, yielded no measurable difference for CFU counts in the spleens of female wild-type compared to heterozygous and homozygous p.D1994S knockin littermates (Fig. 2C).

The p.G2019S Lrrk2 mutation enhances antimicrobial host immune responses in mice

Bacterial clearance requires the recruitment of and infiltration by activated immune cells. We therefore examined the relative number of myeloid cells present in the spleen by flow cytometry. S. typhimurium–infected spleens from p.G2019S Lrrk2-expressing mice, but not spleens from mock-infected mice, harbored significantly more monocytes, macrophages, and neutrophils than did spleens from wild-type littermates (P < 0.01; Fig. 2, D and E). This result suggested enhanced infiltration of the spleen by immune cells in mice expressing the mutant Lrrk2 (59).

To determine whether the antibacterial replication effect in p.G2019S mutant Lrrk2-expressing mice was dependent on myeloid cells, we used targeted antibody pretreatment to deplete either neutrophils alone or monocytes and neutrophils in combination from the circulation of S. typhimurium–infected heterozygousG2019S/WT mice (Fig. 2F and fig. S3, A and B) (77). The protective effect, as measured by CFU counts, was eliminated (Fig. 2F). The loss of neutrophils and monocytes led to a >700-fold increase in bacterial burden in the spleen, whereas the loss of neutrophils alone increased the CFU count by >30-fold (Fig. 2G). Notably, this analysis was performed in heterozygous female mice because of the more pronounced effect in bacterial clearance compared to wild-type animals (Fig. 2B). These results suggested that the augmented effect against S. typhimurium by the p.G2019S mutant Lrrk2 was mediated partly by myeloid cells.

Among the critical host responses to invading pathogens is the production and release of reactive oxygen species as well as of cytokines and chemokines by immune cells. Therefore, we measured reactive oxygen species (H2O2) production in cultures of primary bone marrow–derived macrophages from these Lrrk2 mutant mice. We detected a significant (P < 0.001) increase in H2O2 production after exposure to Salmonella in macrophage cultures derived from p.G2019S mutant Lrrk2-expressing mice (Fig. 3A). When examining cell viability in macrophage cultures by a neutral red assay, we found that infected wild-type and p.G2019S mutant Lrrk2-expressing macrophages showed the same survival rates (Fig. 3B). We observed a similar augmentation in H2O2 production in cultures of primary macrophages from p.G2019S mutant Lrrk2-expressing mice after infection with reovirus (serotype 3 Dearing; T3D) (Fig. 3C). These results were consistent with a previous report that Lrrk2 mediated reactive oxygen species production in mouse dopaminergic neurons of the substantia nigra (78). We did not, however, observe any p.G2019S mutant Lrrk2-dependent differences in the release of 32 cytokines and chemokines by bacterially stimulated primary macrophage cultures (fig. S4). As expected, lysates of S. typhimurium–infected spleens from adult mice expressing the p.G2019S mutant Lrrk2 showed higher H2O2 concentrations compared to spleen homogenates from wild-type animals (Fig. 3D). When we examined these S. typhimurium–infected mice further, we also recorded more H2O2 production in the brains of p.G2019S mutant Lrrk2-expressing animals (Fig. 3, E and F). The effective doubling of H2O2 concentrations in homozygous knockin mice occurred despite the fact that S. typhimurium did not infect the nervous system and, therefore, did not induce any detectable recruitment of leukocytes into the brain (Fig. 3, G and H).

The p.G2019S mutant Lrrk2 prolongs survival of mice with sepsis

We next examined whether the enhanced control of bacterial replication in tissues from p.G2019S mutant Lrrk2-expressing mice resulted in a survival benefit. Given the observed sex bias in our CFU assays, survival was assessed in female mice. In wild-type female mice, a moribund state from complications of S. typhimurium–induced sepsis invariably occurred between days 5 and 9 post-infection (Fig. 3I). The expression of one or two p.G2019S-encoding Lrrk2 alleles extended the survival of S. typhimurium–infected female mice by more than 7 and ~12 days, respectively (Fig. 3I). When assessing the outcome of lethal sepsis using logistic regression analysis, we calculated a reduced odds ratio (OR) for death by 8 dpi to 0.036 [confidence interval (CI), 0.003 to 0.484; P = 0.012] in animals carrying one p.G2019S allele and an OR for death of 0.057 (CI, 0.004 to 0.817; P = 0.035) at 8 dpi for homozygous knockin female mice (wild-type animals; OR, 1).

Unexpectedly, knockout mice that lacked Lrrk2 expression as well as animals expressing the kinase-dead p.D1994S mutant Lrrk2 did not show significant (P > 0.05) differences in their survival during sepsis compared to age-matched, wild-type littermates (Fig. 3J), regardless of sex. We concluded from these experiments that the p.G2019S mutant Lrrk2 conferred a protective gain-of-function effect in the context of Salmonella infection potentially through enhanced kinase activity.

Lrrk2 deficiency worsens survival in mice with reovirus-induced encephalitis

Given the association of LRRK2 variants with both PD and leprosy, we next questioned whether Lrrk2 also modulated disease outcomes in the context of microbial invasion of the nervous system. To this end, we used a previously characterized mouse model of viral infection (4, 79). Reovirus-T3D is a ubiquitous double-stranded RNA (dsRNA) virus, which, when administered nasally to newborn pups, leads to viremia with systemic distribution of virus (79). The illness is associated with pneumonitis and transient ileitis in newborn mice, which peaks at 3 dpi. Infection of the brain occurs through propagation from the nasal epithelia via olfactory and trigeminal nerves (4), although entry via the bloodstream may also be involved (Fig. 4A). In wild-type mouse pups, reovirus-T3D infection leads to lethal encephalitis between 9 and 24 dpi. In testing a role for Lrrk2 variants in the course of viral encephalitis, we chose three predetermined endpoints: quantification of viral titers in select organs, length of survival, and degree of neuropathology (Fig. 4). To this end, we first confirmed that Lrrk2 was already expressed in the brains of newborn mice (Fig. 4B).

Fig. 4 Lrrk2 deficiency reduces survival in mice infected with reovirus.

(A) Newborn WT mice were intranasally inoculated within 24 hours of birth with one LD50 dose of reovirus-T3D, leading to infection of the olfactory epithelium. Photomicrographs of sections from the olfactory epithelium and adjacent structures at days 3 and 4 post-infection (left two panels) and from neonatal mouse midbrain at day 11 post-infection (right panel) are shown. Sections were stained with anti-REO antibody (positive immunoreactivity in brown with a blue hematoxylin counterstain). Sagittal skull images were prepared by holocranohistochemistry, as described (4). ES, ethmoid sinus; LP, lamina propria; CP, cribriform plate; OE, olfactory epithelium. Black asterisk denotes infected olfactory neuron in the OE, and black arrows depict reovirus protein transport within axons of cranial nerve I. (B) Western blots of brain lysates from WT mouse pups that were collected on the postnatal days (p) indicated. (C and D) Separate cohorts of mice were analyzed for reovirus burden measured by the number of plaque-forming units (PFU) in lysates of whole organs. PFU counts in the lungs (C) and brains (D) of Lrrk2 KO mice (Lrrk2KO/KO) and WT littermates at 3 and 11 dpi are shown. Each symbol represents one animal. (E) Activation of protein kinase R (PKR) after reovirus-T3D infection, as shown by Western blotting of whole lung homogenates at 3 dpi. (F) Kaplan-Meier curves for survival rates of reovirus-infected mice. Survival rates for reovirus-infected male mice were 45% for WT mice, 43% for heterozygous animals (Lrrk2WT/KO), and 42% for Lrrk2 KO mice (Lrrk2KO/KO). Survival rates for reovirus-infected female mice were 50% for WT, 27% for heterozygous (Lrrk2WT/KO), and 29% for Lrrk2 KO mice (Lrrk2KO/KO). A minimum of n = 26 mice per genotype were used (male to female ratio, ~50:50). P = 0.047 between WT and Lrrk2KO/KO mice. n.s., not significant. Tests used were two-way ANOVA with Bonferroni post-comparison for (C) and (D) and log-rank test for (F).

We inoculated wild-type and Lrrk2 knockout mice (29) with the previously determined median lethal dose (LD50) dose of reovirus-T3D, and mice were sacrificed on days 3 and 11 post-infection for viral titer analyses of lung and brain tissues. As shown in Fig. 4 (C and D), we saw trends in higher viral titers in infected organs from Lrrk2-deficient mice but recorded no significant (P > 0.05) differences in lysates of lung and brain tissue between the two genotypes (wild-type versus knockout), which were bred in parallel. We also saw no viral titer differences in other organs (spleen, liver, ileum, and kidneys) collected on days 3, 5, 7, and 11 post-infection. Nevertheless, we confirmed the expected systemic tissue responses to dsRNA virus infection in these mice, such as induction of IFN-induced protein kinase R expression in the lungs (Fig. 4E).

For our second endpoint, we observed increased mortality beginning at 9 dpi in Lrrk2 knockout mice compared to wild-type mice (P < 0.047), with heterozygous animals showing an intermediate survival rate (Fig. 4F). This experiment included littermate controls and was performed blinded to genotype. When we analyzed all group comparisons of Lrrk2-deficient and wild-type animals (both sexes and with additional cohorts inoculated at higher doses), the OR for death from encephalitis was 3.45 in Lrrk2-deficient mice (P < 0.002). Strikingly, and consistent with the sex bias that was revealed in our S. typhimurium sepsis mouse model, the rise in mortality from encephalitis was greater in Lrrk2-deficient female mice. The survival rate of heterozygous females was nearly identical to that of female Lrrk2-deficient mice, whereas the survival rates for male Lrrk2-deficient mice and male heterozygous mice were essentially the same as for wild-type animals (Fig. 4F).

Lrrk2 deficiency worsens the pathological outcome of reovirus-induced encephalitis

To better understand the lower survival rate of Lrrk2 knockout mice, we next explored neuropathological outcomes at the height of reovirus infection in mouse brain (Fig. 5). For this, we analyzed additional mouse cohorts similarly infected with one LD50 dose of reovirus. At 11 dpi, mice were sacrificed at random (i.e., not as a function of signs of illness), and intact skulls (4) were processed by automated staining for microscopy. For the analysis of heterozygous × heterozygous (Lrrk2WT/KO) crosses, we compared mock-infected mice (either wild-type or Lrrk2-deficient) to reovirus-T3D–infected wild-type mice and heterozygous and Lrrk2-deficient littermates. Figure 5A shows representative images of staining by anti-reovirus antibody to detect viral proteins in infected mouse brains. Sections were also stained with anti-Iba1 antibody, a marker of microglia and invading macrophages, by anti-A60 (anti-NeuN) antibody to identify neurons and by anti-tyrosine hydroxylase antibody, a marker of dopamine producing neurons. Unbiased analyses revealed a significant increase in total anti-reovirus antibody immunoreactivity throughout serial brain sections from Lrrk2-deficient mice compared to wild-type animals in the most consistently infected regions of the brain, namely, midbrain and thalamus (P < 0.01; n = 27; 175 sections; Fig. 5, B to D). Other brain regions, such as the pons, medulla oblongata, and frontal cortex, also showed elevated reovirus protein signal (P < 0.03; fig. S5, A and B). As we had observed in the survival studies, the anti-reovirus antibody staining of these brain sections uncovered a female sex bias (Fig. 5E).

Fig. 5 Worse pathological outcomes of reovirus-induced encephalitis in mice with Lrrk2 deficiency.

Representative immunohistochemistry images of thalamus and midbrain (tectum; substantia nigra) from mock-infected and reovirus-T3D–infected WT and Lrrk2 KO mice. (A) Animals were sacrificed at 11 dpi. Intact skull sections (10 μm each) were stained with antibodies against reovirus protein (REO), microglia and macrophages (Iba1), and neuronal antigens A60/NeuN and tyrosine hydroxylase (TH). Positive immunoreactivity is indicated in brown, hematoxylin counterstain is indicated in blue, and cresyl violet staining is indicated in purple. (B to G and I) These panels summarize the quantification of immunoreactive signals in mock-infected (mock) animals (of any genotype) and of reovirus-T3D–infected WT, heterozygous (HET; Lrrk2WT/KO), and Lrrk2 KO mice. (H and J) Pearson correlation graphs for immunoreactive signals for anti-A60 antibody–stained (P ≤ 0.0001, R2 = 0.3594, Pearson r = −0.5995) or anti-Iba1 antibody–stained (P ≤ 0.001, R2 = 0.1363, Pearson r = 0.3692) sections compared to anti-REO antibody–stained sections. Number of animals, n = 27, >175 sections analyzed. *P < 0.05 and **P < 0.01. n.s., not significant. Tests used were one-way ANOVA with Bonferroni post-comparison for (B) to (D), (F), (G), and (I) and Mann-Whitney t test for (E).

We next investigated whether higher reovirus protein signals in Lrrk2-deficient mouse brains were due to a decreased rate of viral protein degradation. When we screened whole brain homogenates for changes in autophagy markers between the two genotypes (wild type versus Lrrk2 knockout) (29), we saw no consistent changes in LC3 isoforms and p62 (fig. S5C). However, it could be possible that the analysis of whole brain lysates was not sensitive enough to detect regional changes in classical autophagy markers after reovirus infection.

In agreement with the higher burden of viral proteins seen in the nervous system, staining with anti-A60 antibody revealed a reduction in neuronal signals in Lrrk2-deficient mice, as shown by a trend when all brain regions were combined and by a significant difference when midbrain sections were analyzed in isolation (wild type versus knockout; P < 0.01; Fig. 5, F and G). Anti-reovirus antibody reactivity and anti-A60 antibody staining showed a high degree of negative correlation (−0.599 by Pearson correlation; P < 0.0001; n = 26 animals; 61 sections; Fig. 5H). In contrast, a high degree of positive correlation was found between anti-reovirus antibody staining and microglial anti-Iba1 antibody reactivity throughout infected brain regions from all mice (0.3692 by Pearson correlation; P < 0.001; n = 25 animals; 90 sections; Fig. 5, I and J). Although we observed an apparent reduction in anti-tyrosine hydroxylase antibody staining in reovirus-T3D–infected animals, our sagittal sectioning technique of intact skulls—which we had chosen to monitor viral expression from the olfactory system (4) to the medulla oblongata—precluded stereotactic quantification of substantia nigra neurons (Fig. 5A). We concluded from these experiments that Lrrk2 deficiency conferred three negative outcomes on mice with reovirus-induced encephalitis: higher burden of viral protein in the brain, greater loss of neurons, and a higher mortality rate with a female sex bias.

The p.G2019S Lrrk2 mutation lowers viral burden during peak reovirus infection in mice

To assess the course of viral encephalitis in mice carrying the p.G2019S Lrrk2 mutation (Fig. 6), we examined multiple cohorts of reovirus-infected homozygous p.G2019S mutant Lrrk2-expressing mice and wild-type mice (56). Given the tropism of reovirus-T3D and the detection of anti-LRRK2 reactivity in the olfactory epithelium of adult humans (Fig. 1), we first assessed reovirus expression in the nasal epithelia of intact murine skull sections (4, 79). At 3 dpi, we detected no difference in anti-reovirus antibody staining between p.G2019S mutant Lrrk2-expressing mice and wild-type mice (n = 6 mice and n = 12 sections; fig. S6, A and B).

Fig. 6 p.G2019S mutant Lrrk2 lowers reovirus titers during peak infection and augments immune cell chemotaxis.

Newborn mice of distinct Lrrk2 genotypes were intranasally inoculated with one LD50 dose of reovirus-T3D within 24 hours of birth. Viral titers measured by PFU counts are shown for homogenates of lungs (A) and brains (B) from mice carrying two mutant p.G2019S alleles (Lrrk2GS/GS) and WT littermate controls. Tissues were collected at 3 and 11 dpi from the p.G2019S mutant animals and WT mice (A and B) and from animals carrying one or two kinase-dead p.D1994S mutant Lrrk2 KI alleles and their WT littermates (C). Each symbol represents one animal. (D) Representative flow cytometry plots of mock- and reovirus-infected mouse brains collected at 9 dpi, with gating for infiltrating myeloid cells identified as CD45HIGH and CD11b+. (E) Graphic summary of flow cytometry experiments carried out as in (D). Each symbol represents two pooled brain specimens (n = 8 to 10 mice). (F) Quantification of mRNAs encoding the chemokines CCL5 and CCL3 in bone marrow–derived macrophages from WT and p.G2019S mutant Lrrk2 KI mice (Lrrk2GS/GS) after reovirus infection. (G) Western blots of reovirus-T3D–infected mouse brain lysates probed for total STAT1 and Ser727-STAT1 phosphorylation (and two control proteins, β-actin and β-tubulin). *P < 0.05 and **P < 0.01. n.s., not significant. Tests used were one-way ANOVA with Bonferroni post-comparison for genotype comparisons.

However, similar to the Salmonella infection mouse model, p.G2019S mutant Lrrk2-expressing knockin mice showed a reduction in viral titers during peak infection of lung and brain at days 3 and 11 post-infection dpi, respectively (P < 0.05; n ≥ 10 mice per group; Fig. 6, A and B). The differences in viral load were independent of sex (fig. S6C). This effect of the p.G2019S Lrrk2 mutation was not seen before or after peak reovirus infection (Fig. 6, A and B). The recorded differences in viral titers were not only time point dependent but also organ specific and related to kinase activity. Under the same conditions, viral titers from mice carrying the p.D1994S Lrrk2 knockin mutation showed elevated virus plaque-forming units (PFUs) in the lungs of homozygous p.D1994S knockin mice at 3 dpi but not in the brains of these animals at 11 dpi (Fig. 6C).

The Lrrk2 p.G2019S mutation enhances leukocyte recruitment into reovirus-infected mouse brains

We next examined whether improved control of viral titer by the p.G2019S mutant Lrrk2-expressing mice was associated with enhanced leukocyte recruitment (59). In a separate cohort of wild-type mice, we examined homogenates of reovirus-T3D–infected mouse brains for the influx of neutrophils and macrophages at a time point during encephalitis (9 dpi) before the terminal stage was reached in most wild-type animals. Using flow cytometry analysis of brain homogenates, we recorded an approximately threefold rise in infiltrating CD11b+ and CD45HIGH leukocytes (Fig. 6, D and E) in p.G2019S mutant Lrrk2 knockin animals (n = 8 to 10 mice per group). The recruitment of myeloid cells into the nervous system was specific for reovirus infection, because flow cytometry of S. typhimurium–infected mouse brain lysates showed no difference in detectable leukocyte numbers (Fig. 3, G and H), consistent with the fact that Salmonella is not known to seed the nervous system.

In parallel, we screened by multiplex enzyme-linked immunosorbent assay (ELISA) homogenates of infected mouse brains and lungs compared to tissue lysates from mock-infected animals for 32 cytokines and chemokines that are associated with enhanced chemotaxis of immune cells. We observed no consistent pattern of dysregulated pro- or anti-inflammatory signaling between p.G2019S mutant Lrrk2-expressing mice and wild-type animals (n = 6 mice per organ; fig. S6D). We next addressed the question of chemotactic signaling in isolated macrophages. In mouse bone marrow–derived macrophage cultures, we observed an increase in mRNA for murine orthologs of CCL5 (Rantes) and CCL3 (MIP-1α) about 6 hours after initial reovirus infection in p.G2019S mutant Lrrk2-expressing mice compared to wild-type animals (Fig. 6F). Both chemokines are expressed downstream of IFN-1 (80).

To validate these findings in the context of immune signaling in vivo, we assessed signal transducer and activator of transcription 1 (STAT1) activation (81) in brain homogenates from reovirus-T3D–infected mice. We detected an increase in pSer727-STAT1 and total STAT1 in brain lysates from reovirus-infected p.G2019S mutant Lrrk2 knockin mice compared to brain homogenates from wild-type littermates (Fig. 6G). In parallel bone marrow–derived macrophage cultures, we discovered early and bidirectional changes in pSer727-STAT1 between wild-type, Lrrk2-deficient, and p.G2019S mutant Lrrk2-expressing mice after reovirus-T3D infection. These results suggested that expression of the p.G2019S mutant Lrrk2 increased both chemotactic signaling ex vivo and enhanced myeloid cell recruitment in vivo during reovirus-T3D infection in mice.

The p.G2019S Lrrk2 mutation leads to variable pathological outcomes in reovirus-induced encephalitis

We next examined the neuropathology of two additional cohorts of virally infected mice. First, sagittal brain sections from wild-type and homozygous p.G2019S mutant Lrrk2-expressing mice (n = 27 animals; 190 sections; Fig. 7) were processed for automated staining and quantified in a blinded fashion. In a second cohort of animals comparing the same two genotypes, we analyzed intact skulls (n = 8 mice; as in Fig. 4) from reovirus-infected mutant knockin mice and wild-type mice, which generated the same outcomes for the two brain processing techniques (representative images are shown in Fig. 7A). Using anti-reovirus antibody, we saw significant differences (P < 0.0001) between virally infected and mock-infected animals (Fig. 7B). The range of anti-reovirus antibody positive staining of brain sections from p.G2019S mutant Lrrk2 knockin mice varied greatly, from a very low number of infected neurons per brain section in some mice to infection of the majority of neurons in others (Fig. 7B). In both mouse cohorts, the mean signal for anti-reovirus antibody activity (when all brain regions were combined) showed a trend for reduction in infected p.G2019 mutant Lrrk2-expressing mice (Fig. 7B), which matched the reduction of viral titers shown in Fig. 6C. Nevertheless, when counts for immunoreactivity were logarithmically transformed and adjusted for sex, we recorded increased anti-reovirus antibody positivity in the midbrain and cortical brain sections from female compared to male Lrrk2 mutant–expressing animals (Fig. 7, C and D).

Fig. 7 p.G2019S mutant Lrrk2 generates variable, sex-dependent pathological outcomes in reovirus-induced encephalitis.

(A) Representative immunohistochemistry images of thalamus, midbrain, and frontal cortex sagittal sections (10 μm each) from mock-infected (mock) and reovirus-T3D (REO)–infected newborn mice. Mice were WT or had two p.G2019S mutant Lrrk2 alleles. Brains were collected at 11 dpi. Sections were stained with antibodies against reovirus protein (REO), microglia and macrophages (Iba1), and the neuronal antigens A60/NeuN and tyrosine hydroxylase (TH). Positive immunoreactivity is indicated in brown, hematoxylin counterstain is indicated in blue, and cresyl violet staining is indicated in purple. (B to E, G to I, and K) Box plots representing immunoreactive signal quantification after antibody staining by immunohistochemistry for reovirus protein, Iba1, and A60/NeuN. Box plots show staining results for sections from mouse midbrain (tectum), cortex, or all brain regions analyzed together. (F, J, and L) Pearson correlation graphs plotting anti-REO antibody positivity against positivity with antibodies against Iba1 (F; P ≤ 0.001, R2 = 0.2224, Pearson r = 0.4716) and A60/NeuN (J; P ≤ 0.001, R2 = 0.1286, Pearson r = −0.3587), and anti-REO antibody positivity against cresyl violet (CV) positivity (L; P = n.s. for WT and P < 0.01for Lrrk2GS/GS). (F and J) Data from pooled male and female mice. (L) Data from reovirus-infected females only. Total number of animals, n = 27, with >190 sections analyzed. n.s., nonsignificant. **P < 0.01, ***P < 0.001, and ****P < 0.0001. Tests used were one-way ANOVA with Bonferroni post-comparison for (B) to (E), (G) to (I), and (K).

We noted increased infection-dependent microglial anti-Iba1 antibody reactivity with a wide range of morphological changes in mouse brain (Fig. 7A). We recorded a reduction in total anti-Iba1 antibody signals in the brain sections from p.G2019S mutant Lrrk2-expressing animals compared to wild-type mice (Fig. 7E), which paralleled the reduction in viral titers in mouse brain lysates. Overall, the anti-Iba1 antibody signals showed a strong positive correlation with the presence of reovirus protein in each mouse brain region (0.4716 by Pearson correlation; P < 0.001; n = 27 mice; 207 sections; Fig. 7F). As anticipated, we noted a loss of total neuronal staining (anti-A60/NeuN antibody reactivity) stemming from reovirus-induced encephalitis compared to brains from mock-infected animals, which was significant (P < 0.01) in wild-type mice. There was a trend in the same direction for the p.G2019S mutant Lrrk2-expressing mice when compared to mock-infected animals. Female, but not male, mice accounted for the greatest differences seen in all brain regions except the midbrain (Fig. 7, G to I). The presence of reovirus protein showed a negative correlation with anti-A60 antibody reactivity (−0.3587 by Pearson correlation; P < 0.001; n = 27 mice; 193 sections; Fig. 7J).

Last, cresyl violet staining of all nucleated cells in mouse brain sections, including resident neurons and infiltrating leukocytes, showed greater variability in sections from female mouse brains, particularly from the cortex. Cresyl violet–positive counts from infected p.G2019S mutant Lrrk2-expressing mouse brains were significantly greater (P < 0.01) than those from wild-type animals (Fig. 7, A, K, and L). These collective findings suggested greater than expected variability in the pathological outcomes of reovirus-induced encephalitis in p.G2019S mutant Lrrk2-expressing mice (despite lower viral titers) compared to wild-type animals, with detectable microscopic changes recorded more frequently in females.

The p.G2019S Lrrk2 mutation promotes increased total α-synuclein in reovirus-infected mouse brains

In autopsy studies of human brains from patients with LRRK2-associated PD, pleomorphism in neuropathological outcomes has been observed, such as α-synuclein accumulation in some carriers of a p.G2019S (or p.R1441C) mutation and tau dysregulation in others carrying the same mutation (11, 17, 82). We and others have recently found that α-synuclein confers a systemic, antimicrobial effect in vivo; the gene encoding α-synuclein has also been found to be up-regulated after infection with select dsRNA viruses (4, 83). Because additional reports demonstrated that the p.G2019S Lrrk2 mutation can modify the accumulation of α-synuclein in cells (84) and in vivo (85), we first probed for total α-synuclein and tau signals in our microscopy studies. In a pilot study with automated quantification of brain sections from wild-type and p.G2019S mutant Lrrk2-expressing mice (carried out as in Fig. 7), we recorded no detectable differences in anti–α-synuclein and tau immunoreactivities. We questioned whether the lack of a detectable difference was related to the abundance and long half-life of both α-synuclein and tau proteins (4).

We therefore examined freshly dissected brains from additional cohorts of mice using Western blotting and ELISA for the quantification of total tau and α-synuclein proteins (Fig. 8) (86, 87). For mock-infected animals, there was no difference in murine tau and total α-synuclein in the brains of wild-type mice, heterozygous animals, and mice homozygous for the p.G2019S Lrrk2 allele (n ≥ 5 mice per group). When measuring tau signals by Western blotting (directed against residues 210 to 241 of the holoprotein) as well as by ELISA, we obtained inconclusive results (Fig. 8, A and B). This may reflect distinct antibody specificities among these two readouts and postnatal splicing changes that occurred in the tau-encoding Mapt gene (88).

Fig. 8 Lrrk2 kinase activity modulates α-synuclein concentrations and alters survival during reovirus infection of mice.

(A to E) Quantification of two PD-linked proteins, tau and α-synuclein, in brain lysates from reovirus (REO)-T3D–infected mouse pups. Mice were one of three distinct genotypes: WT mice, heterozygous p.G2019S/WT animals (Lrrk2WT/GS), or homozygous p.G2019S mutant KI mice (Lrrk2GS/GS). (A and B) Murine tau protein was measured by densitometry after Western blotting of brain lysates (see fig. S8C) (A) and by sandwich ELISA for total tau protein (B). (C and D) Quantification of total α-synuclein concentrations in brain lysates measured by sandwich ELISA in mock-infected (C) and reovirus-T3D–infected (REO) mice (D). Brains were collected at 10 to 11 dpi. Each symbol represents one animal, and sex is denoted by colors (male, blue; females, gray). (E) Results for α-synuclein concentrations as a ratio of ELISA data from reovirus-infected animals compared to mock-infected mice, where the mean of ratios calculated in WT animals served as the denominator (mean value of 1). (F) Female mice, but not male animals expressing the p.G2019S mutant Lrrk2 (Lrrk2GS/GS), showed worse survival rates from reovirus-T3D–induced encephalitis. The survival rates of male mice were 48% for WT and 46% for the p.G2019S mutant Lrrk2 (Lrrk2GS/GS) animals. Among female mice, survival rates from reovirus-T3D–induced encephalitis were 57% for WT mice and 33% for the p.G2019S mutant KI Lrrk2 (Lrrk2GS/GS) animals. (G) Genetically ablated kinase activity in the p.D1994S mutant KI Lrrk2-expressing animals (Lrrk2DS/DS) led to increased survival rates during reovirus-induced encephalitis in both male and female mice. Heterozygous KI animals (Lrrk2WT/DS) showed an intermediate survival rate with a female sex bias. In this experiment, the survival rates for females and males combined were 8% for WT, 21% for heterozygous animals (Lrrk2WT/DS), and 40% for the p.D1994S mutant Lrrk2-expressing animals (Lrrk2DS/DS). Tests used were one-way ANOVA with Bonferroni post-comparison for (A) to (E) and log-rank test for (F) and (G).

However, in the same mouse brain lysates, we detected a statistically significant (P < 0.05) increase in total α-synuclein in infected p.G2019S mutant Lrrk2-expressing mice compared to wild-type animals using a validated ELISA (Fig. 8, C and D). These results were recorded in two different cohorts (n = 56 mice; fig. S8, A and B), were independent of sex, and reflected a ≤50% increase during acute reovirus infection (Fig. 8E). These ELISA results were validated by Western blotting using an antibody specific for mammalian α-synuclein (fig. S8C). Under these conditions, we did not see changes in high–molecular weight species of α-synuclein. We concluded that during the course of acute encephalitis caused by a dsRNA reovirus in mice carrying the p.G2019S Lrrk2 mutation, there was an increase in total α-synuclein protein in mouse brains.

Kinase activity affects the survival of female Lrrk2 mutant mice with encephalitis

Although p.G2019S mutant Lrrk2-expressing mice showed better control of reovirus-T3D replication in target organs during peak infection, encephalitis in these animals was associated with a wider range of pathology. We therefore asked what would be the effect of the p.G2019S Lrrk2 mutation on survival. The death rate from encephalitis caused by reovirus-T3D infection was increased in homozygous p.G2019S mutant Lrrk2-expressing mice compared to wild-type animals beginning day 9 post-infection and had a female sex bias (P < 0.025; Fig. 8F). We speculated that the worse survival outcome because of encephalitis, which was shared between p.G2019S mutant Lrrk2-expressing and Lrrk2-deficient mice, could be explained by an augmented immune response in the former and reduced antiviral defenses in the latter.

To test the role of Lrrk2 kinase activity in the survival rates during reovirus-induced encephalitis, we examined kinase-dead p.D1994S mutant Lrrk2-expressing mice. Under the same conditions as used above, homozygous p.D1994S mutant Lrrk2-expressing mice showed a higher survival rate compared to wild-type littermates. The protective effect was greater in females than in males and showed an intermediate outcome for heterozygous animals (Fig. 8G). The outcome for the mice expressing the kinase-dead mutant was associated with reduced Lrrk2 autophosphorylation and, intriguingly, the mutant protein’s lower expression in the lungs but not in the brain (fig. S2D) (56).

Last, to compare mortality rates from reovirus-induced encephalitis across homozygous Lrrk2 genotypes tested, we calculated the related OR (fig. S9). For Lrrk2-deficient animals to die from reovirus-induced encephalitis after day 9 post-infection, the OR was elevated to 3.45 (P = 0.002). In p.G2019S mutant Lrrk2-expressing animals, the OR was 1.46 when both sexes were combined (P = 0.257). For female mice, the OR was 2.67 (P = 0.056). For p.D1994S mutant Lrrk2-expressing mice to die from reovirus-induced encephalitis after day 9 post-infection, the OR was significantly reduced to 0.12 (P = 0.038). The total number of animals examined in these genotypic comparisons, both with and without the correction for sex, and the corresponding CIs for the calculated ORs are listed in fig. S9.

DISCUSSION

Our results suggest that wild-type Lrrk2 expression is protective in controlling the growth of Salmonella in mice. This is independent of its kinase function as demonstrated by findings in mice expressing the p.D1994S kinase-dead mutant Lrrk2. In contrast, the constitutively kinase hyperactive p.G2019S mutant of Lrrk2 conferred a net gain-of-function effect resulting in improved control over microbial pathogens. This effect was mediated by augmentation of chemotaxis and reactive oxygen species production by myeloid cells. Because of the prolonged survival recorded in S. typhimurium–infected mice expressing the p.G2019S mutant Lrrk2, it is plausible that the pro-inflammatory response is evolutionarily advantageous to a host that has been invaded by potentially lethal bacteria (64, 89). Our findings are consistent with two recent reports. The first report identified a Nod2-type role for Lrrk2 in protecting the murine intestinal tract against Listeria monocytogenes infection (64). The second report used a Salmonella infection model similar to ours to highlight the role of Lrrk2-mediated Nlrc4 inflammasome activation in the control of infection in mice (75). The effects of the p.G2019S mutant Lrrk2 alleles on survival of mice after S. typhimurium infection are notable. Only a few other genes have been reported to modulate survival of S. typhimurium–infected mice akin to the p.G2019S mutation in Lrrk2. These include IL-10 deficiency, IL-35 deletion in B cells, ablation of IFN-1 signaling, and Skip deficiency (9093). In our study, the observed contribution of the p.G2019S mutant Lrrk2 to sepsis modulation was predominantly mediated by myeloid cells (Fig. 2).

Consistent with a sexual dimorphism reported in patients with PD who carry a heterozygous p.G2019S mutation in LRRK2 (1315, 94), each of the mutant Lrrk2 alleles we tested showed a female sex bias in the outcome of the two microbial infections examined. Although the p.G2019S Lrrk2 mutation led to lower replication of a dsRNA reovirus during peak infection (as was true for Salmonella), reovirus-infected mice expressing the kinase hyperactive p.G2019S mutant Lrrk2 showed a higher mortality from encephalitis in females. These animals also demonstrated a wider range of neuropathological outcomes when compared to males and wild-type mice (Figs. 6 to 8). We reasoned that the pro-inflammatory state conferred by the p.G2019S mutant Lrrk2 was systemically beneficial during sepsis but detrimental to brain health during encephalitis. Notably, in the context of Salmonella bacteremia, the aseptic brain of p.G2019S mutant Lrrk2-expressing mice also contained higher H2O2 concentrations than seen in S. typhimurium–infected wild-type littermates (Fig. 3).

Our findings in mice with encephalitis converged with our results from human tissues during inflammation: Tissues from subjects diagnosed with acute or chronic infection of the nervous system (Fig. 1 and fig. S1) showed LRRK2-expressing infiltrating immune cells. A possible concomitant role for murine Lrrk2 in microglia-mediated responses could contribute to the results described in our work, as several laboratories have recently demonstrated such an effect in different animal models (57, 70, 95). Mice carrying kinase-dead mutant Lrrk2 alleles were relatively protected in the context of reovirus-induced encephalitis. Whether this protective effect by genetically engineered kinase ablation was caused by a lesser immune response by infiltrating leukocytes or altered responses to pathogens by resident neurons and glia remains to be determined (96).

Several reports have identified a protective effect of Lrrk2 deficiency in the context of exposure to bacterial products such as LPS or select neurotoxins or after expression of exogenous SNCA cDNA (57, 69, 97), all of which has been attributed to reduced inflammation. However, as shown here, if the pathogen represents a replication-competent microbe, the absence of Lrrk2 protein could worsen the course of disease (Figs. 2, 4, and 5). The net outcome of interactions between host genotype and host environment appears to be dependent on the pathogen. A recent report suggested that Lrrk2 knockout mice control Mycobacterium tuberculosis better in early, but not later, stages of infection (68). Although we did not detect any cytokine changes in whole-brain (or lung) lysates, we recorded increased chemotaxis in reovirus-infected p.G2019S mutant Lrrk2-expressing mice. These results support the findings by Moehle et al. (59), who described enhanced myeloid cell recruitment into the midbrain of LPS-treated rats expressing the p.G2019S mutant Lrrk2.

With respect to PD-linked pathology, we noted an increase in total α-synuclein in p.G2019S mutant Lrrk2-expressing animals that was dependent on infection of the brain. The results for tau protein were inconclusive and require further studies. In mice, reovirus infection frequently leads to ubiquitin-positive inclusion formation in infected cells (4, 79); however, we saw no α-synuclein–positive aggregates or tau-positive aggregates in the brains of acutely infected mice. Models of chronic nervous system infection may provide greater insight into a possible link between Lrrk2-linked inflammation, misprocessing of amyloidogenic proteins, and altered neurobehavior (98).

Our findings of the pro-inflammatory effects of the p.G2019S mutant Lrrk2 on brain health during a systemic infection could provide new insights into our understanding of parkinsonism linked to LRRK2 mutations. The pleomorphic outcomes of neuropathology seen in humans with PD could be attributed to either heterogeneity in environmental triggers (9) or genetic modifiers (28) that serve as cofactors or both. Such a scenario for the expression of a LRRK2-linked phenotype in humans is now testable through environmental exposure history studies of identical twins and carefully genotyped pedigrees (9, 82).

An important question that remains is the identity of the most responsible cell type that promotes pathogenesis in LRRK2-linked parkinsonism (42, 43). Our findings support a growing body of evidence that the LRRK2 protein functions in immune cells both within the brain (microglia and infiltrating leukocytes) and the periphery (myeloid cells) (59). These results suggest that innate immune cells should be further investigated (57, 58) in studies of neurodegeneration (99, 100). Such a shift in research perspective will likely provide an answer to the question of whether LRRK2 alleles confer risk association for PD, leprosy, and CD mainly through one mechanism, such as modulation of inflammation, or through multiple distinct mechanisms in more than one cell type and more than one organ system. The answer promises to inform future therapeutic interventions for these three disabling disorders.

This study represents a multi-pronged approach to elucidating LRRK2’s role in immune functions, yet there are several limitations. We have not yet fully elucidated the molecular mechanisms by which Lrrk2 regulates host defenses within murine immune cells and, possibly, in nonimmune cells. Our collective results suggest that they are multimodal and highlight the importance of taking a systems-based approach. We found that wild-type Lrrk2 confers kinase-independent and kinase-dependent effects in modulating host responses, but through which binding partners and substrates, respectively, will require additional studies to elucidate (21, 95). A second limitation of the study is that we have not yet examined the effects of pharmacological inhibition of Lrrk2’s kinase activity in mice using our two microbial infection models. We posit that such studies will inform safety considerations regarding kinase inactivation as a therapeutic target. Furthermore, our approach focused on acute infection mouse models to test Lrrk2’s role in innate immunity, but chronic or recurrent infection animal models are required to better understand the function of LRRK2 in chronic conditions, such as PD, leprosy, and CD. Antimicrobial responses in the colonized gut and in directly inoculated adult nervous system structures should be investigated in future work. Last, our study does not provide an explanation for the consistently observed sexual dimorphism. Because the female sex bias was not seen in primary immune cells, it could be driven by hormonal regulation or specificity of the tissue environment. Our findings highlight the importance of sex-based analyses in LRRK2-related studies in animal models and in human subjects.

MATERIALS AND METHODS

Study design

The research objective of this study was to test the role of Lrrk2 in immune function in systemic and brain health using two animal models of virulent infection. We tested a role for distinct Lrrk2 alleles in the course of viral encephalitis induced by reovirus-T3D and bacterial sepsis induced by S. typhimurium. We chose three main endpoints: quantification of pathogen replication in key tissues, length of survival and degree of pathology in brains from mice with viral encephalitis, and changes in immune signaling in tissue lysates and primary cells from genotyped animals. Randomization and blinding of experimenters were done throughout the study with the exception of quantifying viral titers, which was done in a randomized, but not blinded, fashion. Experiments were carried out in at least three biological replicates, and each replicate was done with at least three technical replicates, unless specified otherwise in the figure legends. The work reported here followed the ARRIVE guidelines for animal studies. Animal experiments were performed in accordance with the guidelines of the Canadian Council on Animal Care and stipulations of the Ethics Board and the Animal Care Committee at the University of Ottawa.

Mouse colonies

129.Lrrk2KO/KO mice, which were described previously (29), were backcrossed onto BL6 mice from Charles River Laboratories (strain code 027) for at least 10 generations. Mutant Lrrk2GS/GS mice (56), Lrrk2D1994S (56), and Lrrk2R1441C mice (29) were all bred on a C57BL/6J background. Snca KO/KO C57BL/6J mice were obtained from M. Farrer. All mouse colonies were maintained as heterozygous breeders, and littermates were used as control mice.

Respiratory enteric orphan serotype 3 Dearing virus infection

Respiratory enteric orphan virus, serotype 3 Dearing (referred to as reovirus-T3D), was used at an inoculation dose of 1.7 × 105 PFU, which we defined as one LD50 in a previous study (79). At predefined endpoints, the pups were sacrificed for collection of brain and lungs to assess viral burden by standard plaque assay. For survival assays, suckling pups were monitored for their degree of encephalitis with predetermined moribund state as an institutionally approved humane endpoint.

Salmonella infection

S. typhimurium SL1344 was used for infection of bone marrow–derived macrophages and in vivo inoculation. All mice were infected with 200 CFU of Salmonella SL1344 via the lateral tail vein. Animals were age- and sex-matched. For assessment of bacterial load, mice were sacrificed at 5 dpi. Their spleen and liver were collected for standardized CFU counts. For the survival study, a predetermined humane endpoint of moribund state was used.

Quantification of reactive oxygen species

For reactive oxygen species production in macrophages, cells were stained with 2′,7′-dichlorodihydrofluorescein diacetate (H2DCFDA; D-399, Molecular Probes–Life Technologies) as per the manufacturer’s instructions, and fluorescence was measured using a microplate reader. For H2O2 quantification in tissues, the Amplex Red Hydrogen Peroxide/Peroxidase Assay Kit (Invitrogen A22188) was used to monitor endogenous production of H2O2. Mock-treated and Salmonella-infected mice (5 dpi) were perfused with 15 ml of 1× phosphate-buffered saline (PBS), and whole brains and spleens were collected and snap-frozen before homogenization. A microplate reader was used to measure either fluorescence with excitation at 560 nm and emission at 590 nm or absorbance at 560 nm. The obtained H2O2 amounts were normalized to the protein concentration in each sample, calculated by the bicinchoninic acid (BCA) assay as instructed by the manufacturer (Thermo Fisher Scientific, 23227).

Human tissue collection and ethics approval

Human specimens, including brain, intestine, spleen, kidney, lung, and mesenteric lymph nodes, were obtained by surgery or by autopsy from donors with known neurological conditions or without neurological conditions (table S1). Tissues were collected with consent provided by the next of kin and per approval of the local ethics boards at participating hospitals. All specimens were fixed in formalin and embedded in paraffin blocks to be used for immunohistochemistry experiments. Unstained sections of human brain infected by rabies virus were a gift from M. Frosch; unstained sections of a subject with heritable PD linked to a mutant SNCA (PARK1) allele (p.A53T) were provided by D. Dickson; unstained PRKN (PARK2) mutant brain sections were provided by J. P. von Sattel; unstained HIV-infected brain sections were obtained from E. Masliah; and unstained sections of human olfactory epithelium were a gift from W. Schulz-Schaeffer.

Immunohistochemistry and microscopy

The immunohistochemistry protocol for human and murine tissue sections and related microscopy were previously described (55, 87). Intact mouse skulls were processed, as recently described (4). Anti-Iba1, anti-LRRK2 (monoclonal antibody, MJFF4; polyclonal antibody, HL-2; each at 1:200 to 1:2000), anti-reovirus, anti-tyrosine hydroxylase, and anti-A60/NeuN (1:1000 to 1:10,000) antibodies as well as staining by cresyl violet and hematoxylin and eosin were carried out using an automated system operated by the Pathology Laboratory of the University of Ottawa (Ottawa, ON). Antibody specificity was also demonstrated by cognate (versus control) peptide absorption experiments (5 to 10 μg/μl), as described (55, 58).

Statistical analyses

Statistical analyses were performed with GraphPad Prism 5 using Mann-Whitney test, log-rank (Mantel-Cox) test for animal survival curves, or one-way analysis of variance (ANOVA) using Bonferroni’s comparison test to analyze multiple groups, where appropriate. Data are presented as means ± SEM, unless otherwise stated in the figure legends. Analyses for immune cells and human tissue expression profiles were performed in R3.2.0 and plotted using GraphPad Prism 5. Mouse survival data were also analyzed for OR calculations, as performed by a statistician, using multivariable Cox regression modeling to determine P values for genotypic differences. CIs were set at 95%: mean ± 2× the SD. Specific information pertaining to statistical analyses can be found in the figure legends. Power analyses were performed with GPower 3.1.9.2.

SUPPLEMENTARY MATERIALS

stm.sciencemag.org/cgi/content/full/11/511/eaas9292/DC1

Materials and Methods

Fig. S1. Mammalian LRRK2 is highly expressed in myeloid cells, and myeloid cell counts in rodent spleens do not change with loss of Lrrk2.

Fig. S2. Bacterial counts in infected spleen and liver for various Lrrk2 genotypes at day 5 after intravenous S. typhimurium inoculation.

Fig. S3. Antibody-mediated immune cell depletion in heterozygous p.G2019S Lrrk2 mice.

Fig. S4. Cytokine and chemokine secretion into primary macrophage culture medium is not altered in an Lrrk2-dependent manner after LPS treatment.

Fig. S5. Analyses of reovirus-3TD–infected brain tissue from wild-type, Lrrk2 knockout, and heterozygous mice as well as from animals expressing the p.G2019S Lrrk2 knockin mutation.

Fig. S6. Additional results for viral titers, viral uptake, and cytokine/chemokine analyses in p.G2019S mutant Lrrk2-expressing mice.

Fig. S7. Additional results for reovirus-T3D infection–associated changes between wild-type and p.G2019S mutant Lrrk2-expressing mice using select readouts.

Fig. S8. ELISA of α-synuclein from two mouse cohorts and representative Western blotting of endogenous α-synuclein and tau proteins in reovirus-T3D–infected mouse brains.

Fig. S9. OR for death from reovirus-T3D–induced encephalitis in mice expressing wild-type Lrrk2 or mutant homozygous Lrrk2 alleles.

Table S1. Details of human specimens used for histological studies and sources of autopsy (surgical) material.

Data file S1. Individual-level data for Figs. 1 to 8.

Reference (101)

REFERENCES AND NOTES

Acknowledgments: We are grateful to patients and their family members for participation in autopsy studies. We thank M. Frosch, E. Masliah, J. Michaud, D. Dickson, W. Schulz-Schaeffer, and J. P. von Sattel for providing unstained sections of human tissues and M. LaVoie for aliquots of anti-LRRK2 (HL-2) antibody and cognate peptide. We thank J. Shen and M. Farrer for sharing Lrrk2-deficient and Snca-deficient mouse lines, respectively, and L. Dong for assistance with skull preparations and automated histochemistry. We thank S. Gupta and A. Agarwal for technical assistance, E. Sabri and J. Harmsen for additional statistical analyses, and S. Gardai, D. Bulman, and A. West for critical discussions. We thank our fellow CLINT investigators D. Gibbings (University of Ottawa), S.H. (Carleton University), D.J.P. (University of Toronto), J.D.R. (University of Montreal), and E. Schurr (McGill University). Funding: This work was supported by the Parkinson Research Consortium of Ottawa (to B.S. and M.H.), Government of Canada Team Grant to CLINT Investigators, Canada Research Chair Program and Project Grant Program (to M.G.S.), NSERC grant (to I.E.H.), Michael J. Fox Foundation for Parkinson’s Research (to D.J.P., J.J.T., E.G.B., S.S., and M.G.S.), the Uttra and Sam Bhargava Family, and the Department of Medicine at The Ottawa Hospital (to M.G.S.). Author contributions: M.G.S., E.G.B., B.S., and J.J.T. designed the study. M.G.S., E.G.B., J.J.T., and S.S. supervised the study. B.S., M.H., I.E.H., and M.L. performed animal studies, viral titer assessments and bacterial counts, genotyping, and statistical analyses. B.S., Y.Y.Z., Q.H.-V., and D.E.-K. carried out cellular studies. B.S. and J.L. performed power analyses and statistical analyses. B.S., J.R., N.L., J.K., A.N., A.A., C.B., and J.M. performed biochemical, genotyping, and histological experiments. B.S., M.H., I.E.H., M.L., J.L., Y.Y.Z., Q.H.-V., D.E.-K., J.L, J.R., N.L., J.M.W., J.D.R., D.J.P., S.S., J.J.T., E.G.B., and M.G.S. performed data analyses. K.C., P.C.M., D.S.P., J.M.W., J.D.R., D.S., D.J.P., and S.H. provided reagents and critical suggestions. M.G.S., B.S., J.J.T., E.G.B., M.H., I.E.H., and M.L. prepared the initial draft of the manuscript and figures. All authors reviewed or edited the manuscript and approved the submitted versions. Competing interests: D.S. is an employee of Novartis Pharmaceuticals. The Ottawa Hospital Research Institute receives annual payments from BioLegend Inc. related to a licensing agreement for an α-synuclein–directed ELISA. M.G.S. has received travel support from the Michael J. Fox Foundation for Parkinson’s Research and has performed paid consultancy work for Genzyme Sanofi. The other authors declare that they have no competing interests. Data and materials availability: All data associated with this study are in the main text and the Supplementary Materials. Lrrk2 G2019S and Lrrk2 D1994S mutant mice have been made available by Novartis through a material transfer agreement. Lrrk2 knockout mice and Lrrk2 R1441C mutant mice were obtained from J. Shen, and Snca knockout mice were provided by M. Farrer. Any other materials and reagents are either commercially available or can be shared using the UBMTA, except for the following reagents: reovirus (T3D) and rabbit anti-reovirus antibody will be provided under MTA with a cost-recovery fee.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article