Aging-induced type I interferon response at the choroid plexus negatively affects brain function

See allHide authors and affiliations

Science  03 Oct 2014:
Vol. 346, Issue 6205, pp. 89-93
DOI: 10.1126/science.1252945

Excess signaling is bad for the aging brain

Preventing antiviral-like responses may protect function in the aging brain. Baruch et al. monitored messenger RNA production in the choroid plexus, the interface between the blood and cerebrospinal fluid, in young and old mice (see the Perspective by Ransohoff). They detected an inflammatory response in older mice not present in the brain of young mice that was also seen in old aged human samples postmortem. Preventing signaling by the cytokine interferon-I, which normally helps in the antiviral response of the immune system, helped prevent the decrease in cognitive function seen in aged mice.

Science, this issue p. 89; see also p. 36


Aging-associated cognitive decline is affected by factors produced inside and outside the brain. By using multiorgan genome-wide analysis of aged mice, we found that the choroid plexus, an interface between the brain and the circulation, shows a type I interferon (IFN-I)–dependent gene expression profile that was also found in aged human brains. In aged mice, this response was induced by brain-derived signals, present in the cerebrospinal fluid. Blocking IFN-I signaling within the aged brain partially restored cognitive function and hippocampal neurogenesis and reestablished IFN-II–dependent choroid plexus activity, which is lost in aging. Our data identify a chronic aging-induced IFN-I signature, often associated with antiviral response, at the brain’s choroid plexus and demonstrate its negative influence on brain function, thereby suggesting a target for ameliorating cognitive decline in aging.

One of the mysteries of brain senescence is to what extent it is influenced by aging of other body tissues (1, 2). Recent studies have suggested that age-related changes in both circulating soluble factors (35) and peripheral immunity (6) are involved in this process. Nevertheless, given the fact that the mammalian central nervous system (CNS) is secluded behind barriers from directly interacting with the blood circulation (7), the underlying mechanism remains unclear.

The choroid plexus (CP), an epithelial monolayer that forms the blood–cerebrospinal fluid (CSF) barrier and produces the CSF (8), serves as a neuroimmunological interface in shaping brain function in health and pathology by integrating signals from the brain with signals coming from the circulation (6, 911). We envisioned that understanding how CP activity is altered in aging might lead to identification of strategies to attenuate aging-associated cognitive decline.

To determine whether aging of the CP reflects aging of the brain and other body tissues, we systematically characterized the mRNA expression profile of young (3-month-old) and aged (22-month-old) murine organs by RNA sequencing. Whereas all examined tissues from aged mice showed increased expression of mRNA transcripts associated with age-related processes, such as immune response (P = 2.66 × 10–05; cluster XIX; fig. S1, A and B), the aged CP exhibited an expression profile corresponding to type I interferon (IFN-I) response (P = 4.25 × 10–07; cluster IX; Fig. 1A and fig. S1, A to C), classically associated with antiviral activity (12). To exclude an aging-unrelated effect such as viral infection, we confirmed the increased expression of IFN-I–dependent genes [e.g., IFN regulatory factor 7 (irf7), IFN-β 1 (ifnβ), and IFN-induced protein with tetratricopeptide repeats 1 (ifit1)] by real-time quantitative polymerase chain reaction (qPCR) in different cohorts of aged mice from various animal centers (Fig. 1B). Closer examination by qPCR and immunohistochemical staining showed that, whereas IFN-I–associated gene expression was increased in the aged CP, expression of type II IFN (IFN-γ)–dependent genes (10) [e.g., intercellular adhesion molecule 1 (icam1), IFN-γ–induced protein 10 (cxcl10), and chemokine (C-C motif) ligand 17 (ccl17)] was decreased (Fig. 1, C and D). Immunohistochemical staining of brain sections of mice, as well as postmortem brain sections from non–CNS-diseased humans, confirmed an aging-associated increase of IFN-I at the CP (Fig. 1E) and suggested that this signature is conserved among species.

Fig. 1 IFN-I–dependent expression program in the aging CP.

(A) Heat map and hierarchical clustering of gene expression in bone marrow (BM), cervical lymph nodes (CLN), colon (CL), hippocampus (HC), inguinal lymph nodes (ILN), liver (LV), lung (LU), mesenteric lymph nodes (MLN), spleen (SP), and thymus (TH) of aged mice, analyzed by RNA sequencing. Results are displayed as log of expression relative to young controls (n = 2 or 3). (B and C) mRNA abundance of IFN-I– and IFN-II–dependent genes in the CP of aged animals (fold change relative to young mice CPs; n = 10 mice per group; bars represent mean ± SEM; *P < 0.05; **P < 0.01; ***P < 0.001; Student’s t test for each pair; data are representative of at least three independent experiments). (D) Representative images of brain sections of young and aged mice, immunostained for claudin-1 (epithelial tight junctions; in green) and either CXCL10 or ICAM1 (in red) (scale bars indicate 25 μm). (E) Representative micrographs of immunohistochemical staining for IRF7 in young and aged mice and for IRF7 and IFN-β in young and aged human postmortem CPs (scale bars, 50 μm; 3V, third ventricle).

The CP epithelium is exposed to brain-derived signals from its apical side, via the CSF, and to peripheral signals from its basal side, via the circulation (8). To differentially examine which of these compartments induces the transcriptional signature of the aged CP, we used a heterochronic parabiosis model (13), in which young and aged mice were surgically connected to share vasculature. Under these conditions, there are bidirectional, although not symmetric, effects on hippocampal neurogenesis and cognitive function between young and aged heterochronic parabionts (3, 4). Transcriptome analyses revealed that the circulation of aged mice failed to induce expression of IFN-I–dependent genes in the CP of young heterochronic partners. It did, however, affect expression of IFN-II–dependent genes, including reduced expression of homing and trafficking determinants required for physiological leukocyte entry via the CP to the CSF [Fig. 2, A and B, and fig. S2, A and B (clusters III and VIII)] (10) and increased expression of ccl11 (Fig. 2C), a chemokine that impairs brain plasticity (3), and its expression by the CP is induced by the cytokine interleukin (IL)–4 when local IFN-γ levels are diminished (6). Under the same experimental setting, the young circulation failed to reverse the expression program of IFN-I–dependent genes in the aged CP yet affected expression of IFN-II–dependent genes (10) (Fig. 2, A and B, and fig. S2, A and B). However, CCL11 expression by the CP of the aged heterochronic parabionts was not affected by the young circulation (Fig. 2C), suggesting that IFN-II regulation of CCL11 expression in the aged CP involves additional factors that are not derived from the circulation. These results indicated that the IFN-II expression program at the aged CP is modulated by both circulation- and brain-derived factors, whereas IFN-I in this compartment might be induced by factors in the CSF emitted from the aged brain. Therefore, we exposed primary cultures of CP epithelial cells of young mice to CSF aspirated from aged or young mice; we found that the CSF of aged but not of young mice induced the IFN-I–dependent response (Fig. 2D). This response was recapitulated when epithelial cells of the CP were treated with a mixture of proinflammatory cytokines that accumulate in the brain during aging and neurodegeneration (14) (fig. S3).

Fig. 2 Effect of signals from the CSF but not from the blood circulation on IFN-I–dependent genes in the CP.

(A) Heat map and hierarchical clustering of gene expression in the CPs of aged (A) and young (Y) iso- (A-A or Y-Y) and hetero- (A-Y) chronic parabionts, analyzed by RNA sequencing and displayed as log of expression relative to young, isochronic CPs (n = 2 or 3). (B and C) mRNA expression of IFN-I–dependent genes, ifit1 and irf7 (B), and the IFN-II–dependent gene, ccl11 (C), in the CPs of parabiotic mice (n = 6 to 8 mice per group). (D) mRNA expression of ifit1 and irf7 in primary cultures of CP cells treated with phosphate-buffered saline (PBS) or CSF of young or aged mice (n = 4 or 5 per group). Throughout the figure, bars represent mean ± SEM; N.S., not significant; *P < 0.05; **P < 0.01; ***P < 0.001; one-way analysis of variance (ANOVA) with Newmann-Kleus post hoc test.

The apparent inverse relations between type I and type II IFN expression programs in the aged CP (Figs. 1 and 2) prompted us to examine their relevance to CP and brain function. Followup of transgenic mice deficient in IFN-II signaling, including IFN-γ receptor knockout (IFN-γR−/−) mice and Tbx21−/− mice [impaired in IFN-γ production by CD4+ T cells (15)], revealed that these animals developed spatial learning and memory deficits (fig. S4, A to H) and displayed reduced hippocampal neurogenesis during adulthood (fig. S4, I to K). Further examination of IFN-γR−/− mice showed that brain functional decline was accompanied by impaired CP activity in supporting physiological leukocyte trafficking to the CSF (fig. S4, L and M). Thus, impaired ability to respond to IFN-γ or reduced IFN-γ in the circulation resulted in cognitive decline and restricted hippocampal neurogenesis during adulthood, before chronological aging (fig. S5). Our observations, together with the essential role of IFN-II in CP function (10) and findings outside the CNS that chronically activated IFN-I signaling interferes with IFN-II–dependent resolution of inflammation (1618), led us to examine whether chronic IFN-I in the aged CP could affect brain function and, if so, whether it involves interfering with local IFN-II–dependent activities.

In vitro exposure of CP epithelial cells to the major IFN-I cytokine, IFN-β, induced a classical “antiviral” response (Fig. 3A) and resulted in reduced expression of insulin-like growth factor (igf1) and brain-derived neurotrophic factor (bdnf) (Fig. 3B), molecules that support neuronal growth, differentiation, and survival and are produced by the CP (11). In vivo administration of neutralizing antibodies to the IFN-I receptor (α-IFNAR) to the CSF of aged mice blocked the IFN-I response program at the CP (Fig. 3C) and resulted in restoration of igf1 and bdnf expression (Fig. 3D and fig. S6), as well as ccl11 expression (Fig. 3E), to levels similar to those found in the CP of young mice. Thus, neutralizing the IFN-I response within the brain affected CP function and suggested that chronically elevated IFN-I interferes with IFN-II activities in the aged CP.

Fig. 3 Restored CP function after neutralization of the aging-induced type I IFN response in the brain.

(A and B) ifit1 and irf7 (A) and bdnf and igf1 (B) mRNA expression in CP epithelial cells cultured for 24 hours with murine IFN-β (n = 4 per group; bars represent mean ± SEM; *P < 0.05; ***P < 0.001, Student’s t test). (C) ifit1 and irf7 mRNA expression in the CP 3 days after α-IFNAR or IgG intracerebroventricular (i.c.v.). administration. (D and E) bdnf and igf1 (D) and ccl11 (E) mRNA expression in the CP 7 days after α-IFNAR or IgG control i.c.v. administration (n = 6 or 7 per group). Bars represent mean ± SEM; *P < 0.05; one-way ANOVA with Newmann-Kleus post hoc test.

To determine whether in vivo neutralization of IFN-I in aging could reverse cognitive dysfunction, we assessed two different measurements of brain plasticity, hippocampal-dependent spatial memory and hippocampal neurogenesis (19) (Fig. 4A). Because age-associated cognitive decline varies across individuals, we first scored memory ability in aged mice by the novel location recognition (NLR) task, in which animals explore two objects, and the exploration time of each object is measured. After 24 hours, one of the objects is relocated, and mice with normal memory spend more time exploring the displaced object, compared with the object that was not moved (20). Cognitively impaired aged mice selected this way (Fig. 4, A and B, and fig. S7) received either α-IFNAR or immunoglobulin G (IgG) injected into their CSF and were tested again in new settings of an NLR test. α-IFNAR–treated mice, unlike the control group, showed improved memory (Fig. 4C), suggesting that neutralizing IFN-I signaling within the brain could restore cognitive function in aged mice. Moreover, the number of newborn neurons, stained for 5-bromo-2'-deoxyuridine (BrdU) (21) and doublecortin (DCX), was increased in the hippocampus of the α-IFNAR–treated group (Fig. 4D), reaching levels comparable to those of aged mice that maintained cognitive function; enhanced hippocampal neurogenesis to a similar extent in aged mice was correlated with improved cognitive function (4). Examination of the hippocampal parenchyma for the effect on age-related neuroinflammation (2), which detrimentally affects adult neurogenesis (22, 23), revealed increased mRNA expression of the anti-inflammatory cytokine, IL-10 (Fig. 4E), and a marked decrease in astrogliosis and microgliosis (Fig. 4, F and G) in the α-IFNAR–treated group.

Fig. 4 Restored brain function after neutralization of the age-induced type I IFN response in the brain.

(A) Scheme illustrating the experimental design. (B) Performance of aged mice in the first NLR task (n = 5 to 18 per group). (C) Performance of α-IFNAR– and IgG-injected cognitively impaired aged mice in the second NLR task (n = 5 or 6 per group). (D and E) Analysis of the hippocampal dentate gyrus (DG) of α-IFNAR or IgG intracerebroventricularly injected cognitively impaired aged mice 14 days after treatment and of young and aged mice that maintained cognitive function (n = 4 to 6 per group). Quantification of hippocampal DCX- and BrdU-labeled newborn neurons (young mice DCX- and BrdU-labeled cell counts were 16,832 ± 873 and 4403 ± 51, respectively) (D); il-10 mRNA expression in the hippocampus (E). (F and G) Quantification (F) and representative images (G) of glial fibrillary acidic protein (GFAP) and ionizing calcium binding adaptor molecule 1 (IBA-1) staining in hippocampal sections (Hoechst in blue; scale bars, 50 μm). Throughout the figure, bars represent mean ± SEM; *P < 0.05; **P < 0.01; one-way ANOVA with Newmann-Kleus post hoc test.

Although originally identified by their antiviral properties, type I IFNs have important roles in attenuating inflammation, both outside and inside the CNS (2426). Nevertheless, we found that blocking IFN-I signaling in the brain of aged mice partially restored cognitive function and hippocampal neurogenesis and attenuated age-related chronic neuroinflammation. It is thus possible that the IFN-I response program at the aged CP represents a physiological mechanism evoked to mitigate age-related neuroinflammation. When this process is not timely terminated, it becomes detrimental to brain plasticity, at least in part by interfering with IFN-II activity in the CP and brain parenchyma, which is needed for physiological immune-dependent brain maintenance and resolution of neuroinflammation in aging and pathology (6, 9, 10, 2729) (fig. S8). Other blood-brain interfaces may undergo similar changes in aging. Our findings may shed light on the association between cerebral levels of IFN-I and memory impairments in several human diseases (25) and the neurological and neuropsychiatric complications that often accompany IFN-I administration to patients (30). Therefore, neutralizing the response to type I IFNs within the CNS might provide an approach to reverse or slow down age-associated cognitive decline or other pathological conditions associated with chronic IFN-I within the CNS.

Supplementary Materials

Materials and Methods

Figs. S1 to S8

References (3142)

References and Notes

  1. Acknowledgments: We thank M. Esiri and A. Troen for their help in obtaining human brain sections, K. I. Mosher for technical assistance in parabiosis experiments, and A. Tsitsou-Kampeli and A. Hutzler for assistance in immunohistochemistry experiments. Supported by a European Research Council (E.R.C.) grant (232835), a European Union Seventh Framework Program (FP7) grant (279017) given to M.S., an E.R.C. grant (309788), and an Israeli Science Foundation grant given to I.A. (1782/11), a National Institute on Aging grant given to T.W.-C. (AG045034), and a Jane Coffin Childs Postdoctoral fellowship given to J.M.C.
View Abstract

Stay Connected to Science

Navigate This Article