Background
Niemann-Pick disease type C (NPC) is a neurodegenerative disease inherited in an autosomal recessive pattern [
1], with mutations in the NPC1 gene accounting for approximately 95% of all reported cases and the remaining 5% of the cases resulting from mutations in the NPC2 gene. While NPC affects all cell types in the body, the main disease feature is the progressive neurodegeneration that results in premature death [
1]. Currently, there is no cure or effective therapy available for NPC and the mechanistic etiology of neurodegeneration remains poorly defined [
1‐
4]. Functionally, the biological roles of NPC1and NPC2 are not well defined, a shortcoming that has hampered our understanding of the mechanistic etiology of the disease. At the cellular level, the disease phenotype is broad, affecting multiple functions, such as endosomal lipid accumulation, calcium dysregulation, neuroinflammation, mitochondrial dysfunction, amyloid peptide Aβ accumulation and tau hyperphosphorylation and aggregation. Nevertheless, the pathogenic hierarchy of these cellular dysfunctions remains unresolved. Several laboratories, including ours, have demonstrated that neuroinflammation occurs early in the disease. We have specifically shown an atypical activation pattern of interferon downstream signaling that involves both IFN-γ- and IFN-α-responsive genes in pre-symptomatic
Npc1−/− cerebella. Activation of IFN-γ and IFN-α responsive genes predicts abnormal microglial activation, anti-viral response, antigen-presenting cell and T-lymphocyte activation, and chemotaxis signaling prior to symptomatic onset [
5]. Notably, IP-10/CXCL10 was the only significantly upregulated cytokine detected at this pre-symptomatic stage, suggesting that this effector of both IFN-γ- and IFN-α signaling could be a key early mediator of aberrant neuroinflammation in NPC.
Here, we asked whether the amyloid precursor protein (APP) plays a role in the early interferon-driven aberrant signaling observed in the pre-symptomatic NPC brain. APP is a disease modifier of Niemann-Pick disease type C (NPC). The loss of the
App gene in the NPC mouse model BALB/cNctr-
Npc1miN/J (
Npc1−/−/
App−/−) results in increased neuroinflammation marked by reactive astrocytosis, decreased neuromuscular function, accelerated neuronal death and shorter lifespan [
6]. Thus, APP exerts a protective function in the NPC brain, consistent with the mounting evidence in support of its role as a neuronal stress modulator [
7‐
14]. The specific nature of this neuroprotective role in NPC remains unknown; here, we reasoned that it may exert its protective function by modulating the inflammatory response of the CNS milieu. To explore that possibility, we carried out a comparative genome-wide transcriptome analysis of pre-symptomatic cerebellar tissue samples from
Npc1+/+/App+/+,
Npc1+/+/App−/−,
Npc1−/−/App+/+, and
Npc1−/−/App−/− mice to identify NPC-altered genes and pathways further affected by the loss of APP. Furthermore, we carried out
in vivo protein validation of the key pro-and anti-inflammatory cytokines and chemokines identified in our genome-wide analyses. Our results show that, in pre-symptomatic
Npc1−/−/App−/− cerebella, expression of IFN-γ- and IFN-α-responsive genes is significantly upregulated compared with
Npc1−/−/App+/+ mice, compounding the dysregulation of microglial activation, anti-viral response, activation of antigen-presenting cells, and T-lymphocyte activation and chemotaxis pathways present in the latter. Multiplex analysis further showed elevated expression of IP-10/CXCL10, MIG/CXCL9, RANTES/CCL5, eotaxin/CCL11 and IL-10 prior to symptomatic onset in
Npc1−/−/App−/− cerebella compared with
Npc1−/−/App+/+mice. In the terminal stage, loss of APP caused pleiotropic differential expression of the vast majority of cytokines evaluated. These findings add to the growing evidence in support of a cytoprotective role of APP in the brain and suggest that, in NPC, that role is mediated through the modulation of neuroinflammation.
Discussion
Our comparative and systematic genome-wide transcriptome analyses of
Npc1+/+/App+/+,
Npc1+/+/App−/−,
Npc1−/−/App+/+, and
Npc1−/−/App−/− mice at pre-symptomatic stage revealed that loss of APP function results in severe exacerbation of multiple inflammatory pathways already present in the NPC brain. Specifically, GSEA and IPA
Upstream Analysis showed significantly increased expression of IFN-γ- and IFN-α-responsive genes in the
Npc1−/−/App−/− cerebellar transcriptome (Figs.
1,
2,
3, and
4; 262 IFN-γ-responsive and 84 IFN-α-responsive genes; Figs.
2 and
4), when compared with
Npc1−/−/App+/+ mice (60 IFN-γ-responsive and 23 IFN-γ-responsive genes [
5];), consistent with the significant exacerbation of all four major inflammatory pathways previously identified in this mouse model of NPC [
5], namely activation of microglia, anti-viral response, activation of T-lymphocytes, and chemotaxis of T-lymphocytes (Additional file
4: Figure S4, Additional file
5: Figure S5, Additional file
7: Figure S7 and Additional file
9: Figure S9).
The mechanisms by which APP loss may cause an exacerbation of inflammatory pathways prior to disease onset in NPC is not immediately clear. APP processing is abnormal in the NPC brain, as evidenced by an increase in amyloid peptide Aβ expression, possibly due to the formation of aberrantly enlarged endosomes, a necessary compartment for the generation of Aβ [
6]. Thus, it would appear reasonable to link excess Aβ expression in the NPC with its pathogenesis. However, loss of APP and, by extension, of Aβ, in the NPC brain, leads to decreased life span, increased cholesterol abnormalities and, notably, disruption of tau homeostasis [
6], as well as an early exacerbation of inflammation, as shown here. These findings suggest that Aβ expression is not a primary pathogenic factor in NPC. Rather, given that APP is a multi-potent cytoprotective molecule, whose cleaved products provide beneficial effects against oxidative stress, metabolic stress, and pathogenic infections, it seems more likely that APP plays a homeostatic role in the brain and that loss of that role accelerates NPC onset and progression. For example, both monomeric and oligomeric forms of Aβ have been characterized to possess potent anti-oxidant activity [
7,
8] and the function of APP intracellular domain (AICD) as a transcription factor has recently been shown to directly regulate the cytoprotective mechanisms against oxysterol-mediated stress [
9]. Furthermore, Aβ has potent anti-microbial activity against many strains of pathogens, including bacteria, viruses, and yeast [
10‐
13].
Overall, the available evidence suggests that loss of APP function in the
Npc1−/−/App−/− brain may contribute to the early altered expression of genes directly related to immune response pathways against pathogens, including
Antimicrobial Response and
Antiviral Response identified by IPA analysis (Additional file
5: Figure S5 and Additional file
6: Figure S6). Interestingly, compared with the sole activation of
Antiviral Response identified by IPA in pre-symptomatic NPC, APP loss resulted in an additional enrichment of the larger functional
Antimicrobial Response, which included 31 additional antimicrobial genes (Additional file
6: Figure S6). This increase in anti-microbial function is further highlighted by the activation of genes involved in T-lymphocyte activation and chemotaxis, as well as the activation of antigen presenting cells, all of which are crucial in host-immune response against various strains of pathogens (Additional file
7: Figure S7, Additional file
8: Figure S8, Additional file
9: Figure S9 and Additional file
10: Figure S10).
It is also noteworthy that changes in gene expression in pre-symptomatic NPC as a result of
App deletion (
Npc1−/−/App−/−) translated into increased expression of pro-inflammatory cytokines and chemokines (Fig.
5), even with the loss of one single
App allele. This was the case with the protein expression of IP-10/CXCL10, the central downstream effector of IFN-γ identified in pre-symptomatic
Npc1−/− mice (Fig.
5a [
5];), as well as several other cytokines, including RANTES, eotaxin/CCL11 and IL-10 (Fig.
5). Interestingly, the notion that haploinsufficiency of
APP is a risk factor for neurotoxicity has been proposed in a model of copper-mediated CNS cytotoxicity [
20]. In that study, a single allele loss of
App in mice was sufficient to alter copper homeostasis comparable to that of mice lacking both alleles of
App [
20]. Therefore, it is plausible that dysregulation of APP function may exacerbate the inflammatory response and poor prognosis of NPC in humans.
Functionally, IP-10/CXCL10 is a potent downstream effector of IFN-γ [
21,
22], the master regulator of the adaptive immune activation that is crucial in the transition from the innate immune response to the antigen-specific adaptive immune response [
23]. IP-10/CXCL10 binds to CXCR3, on activated immune cells such as activated T-lymphocytes or natural killer cells to drive the chemotaxis of CXCR3+ cells to the site of inflammation [
21,
24]. Furthermore, IP-10/CXCL10 also plays a major role in the development and antigen-specific activation of T-lymphocytes [
21]. In addition, interferon-inducible T cell alpha chemoattractant (I-TAC/CXCL11) also binds the same CXCR3 receptor to elicit similar physiological functions [
25‐
28]. The fact that T cell infiltration is apparent in the
Npc1−/−/
App−/− cerebellum (Fig.
6) supports the notion that APP loss may exert its deleterious effect through IP-10/CXCL10-driven T-lymphocyte activation and chemotaxis.
In both
Npc1−/−/
App+/− and
Npc1−/−/
App−/− mouse cerebella, another major cytokine significantly increased at 3 weeks of age was eotaxin/CCL11 (Fig.
5c). Eotaxin/CCL11 is a potent eosinophil chemoattractant, implicated in various eosinophil-related pathogenic processes such as asthma and airway inflammation [
29]. While the combined functional roles of eosinophils and eotaxin/CCL11 are widely characterized in the periphery, the precise role of both in the CNS is not well defined [
30]. For example, eotaxin/CCL11 is an anti-inflammatory Th2 cytokine in the CNS in a murine model of multiple sclerosis [
31]. On the other hand, astrocyte-mediated release of eotaxin/CCL11 and subsequent enhancement of neuronal death
via increased production of microglial reactive oxygen species have also been reported [
30]. In the context of the early and widespread activation of IFN-γ-responsive signaling that occurs in pre-symptomatic NPC brains [
5], IFN-γ potentiates the subsequent release of eotaxin/CCL11 in the periphery [
29], thereby suggesting a potential for the co-activation of IFN-γ and eotaxin/CCL11 under certain inflammatory conditions. Interestingly, co-expression of IP-10/CXCL10 receptor CXCR3 and eotaxin/CCL11 receptor CCR5 (whose ligands also include MIP-1α/CCL3, MIP-1β/CCL4, and RANTES/CCL5) have been characterized in autoimmune T-lymphocytes [
32], consistent with the co-activation of CXCR3 and CCR5 as a potential pathologic mechanism involved in autoimmunity.
Loss of APP also showed a significant impact on the expression pattern of cytokines and chemokines in terminal-stage brains, as illustrated in Additional file
13: Figure S13. Interestingly, the overall expression of pro-inflammatory cytokines and chemokines in the terminal stage
Npc1−/−/App+/− or
Npc1−/−/App−/− were relatively lower than that of
Npc1−/−/App+/+ (Additional file
13: Figure S13). While the precise mechanism responsible for this phenomenon remains to be elucidated, one plausible explanation is the significant reduction in brain mass and paralleled neuronal death observed in the
Npc1−/−/App−/− terminal stage cerebella [
6]. Contrary to the classical understanding of neuronal secretion of cytokines, recent evidence consistently highlights neurons as a major source of proinflammatory cytokines and chemokines under various cytotoxic stresses within the CNS [
33‐
35]. The difference in age-at-collection may be another confounding factor for the terminal stage cytokine/chemokine expressions, as the average age for humane-endpoint varied by a week with the successive loss of an
App allele (11.1 weeks for
Npc1−/−/App+/+, 10.4 weeks for
Npc1−/−/App+/−, and 9.4 weeks for
Npc1−/−/App−/−.
Npc1+/+/App+/+).
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.