2019IL-33 and its decoy sST2 in patients with Alzheimer’s Disease and Mild Cognitive Impairment

Background: Interleukin-33 is a cytokine endowed with pro- and anti-inflammatory properties that plays a still poorly defined role in the pathogenesis of a number of central nervous system (CNS) conditions including Alzheimer’s disease (AD). We analyzed this cytokine and its decoy receptor sST2 in Alzheimer’s disease (AD) and mild cognitive impairment (MCI). Method: IL-33 and sST2 were analyzed in serum and CSF of AD and MCI patients, comparing the results to those obtained in age-matched healthy controls (HC). Because of the ambiguous role of IL-33 in inflammation, the concentration of both inflammatory (IL-1ß and IL-6) and anti-inflammatory (IL-10) cytokines was analyzed as well in serum and cerebrospinal fluid (CSF) of the same individuals. Finally, the effect of IL-33 in an in vitro system of Aß42-stimulated monocytes was examined. Results: As compared to HC: 1) IL-33 was significantly decreased in serum and CSF of AD and MCI; 2) sST2 was increased in serum of AD and MCI but was undetectable in CSF; 3) serum and CSF IL-1ß concentration was significantly increased and that of IL-10 was reduced in AD and MCI whereas no differences were observed in IL-6. In vitro addition of IL-33 to LPS+Aß 42-stimulated monocytes down regulated IL-1b generation in AD, MCI and HC and stimulated IL-10 production in HC alone. IL-33 addition also resulted in a significant reduction of NF-kB nuclear translocation in LPS+Aß42-stimulated monocytes of HC alone. Conclusions: These data support the hypothesis that IL-33 plays a complex anti-inflammatory role that is lost in AD- and MCI-associated neuroinflammation; results herein also suggest a possible use of IL-33 as a novel therapeutic approach in AD and MCI.

background Interleukin-33 (IL-33) is a dual function cytokine produced by endothelial and epithelial cells as well as fibroblast, macrophages, adipocytes, smooth muscle and brain cells [1]. This cytokine is released as a full-length active protein that can be inactivated by caspase-1, -3 and -7-mediated cleavage [2][3] or can processed by different proteases into shorter forms characterized by diverse biological activities [4][5][6]. The regulation of IL-33 biological activity is complex as it is the result of the interplay between the extra-and intra-cellular forms of this cytokine and its ability to bind two different isoforms of ST2, its cognate receptor. Binding of extracellular IL-33 to ST2 on leukocytes, astrocytes and oligodendrocytes [7] mediates the biological effects of this cytokine [8]. On the other hand, when expressed intracellularly, IL-33 binds the p65 subunit of NF-kB [9]; the resulting IL-33/NF-kB p65 complex interferes with NF-kB-dependent transcription by impeding p65mediated transactivation. This causes a negative modulation of NF-kB activity, with a dampening effect on inflammation. An additional actor that modulates the biological activity of IL-33 is the soluble form of ST2 (sST2), a decoy receptor. As is usually the case with decoy receptors, IL-33 binding to sST2 limits its the biological activity [10].
Th2 helper cells and mast cells express ST2 and respond to IL-33, and type 2 innate lymphoid cells (ILC2) are considered to be the signature IL-33-responsive cells [11].
IL-33 plays a yet poorly understood role in the pathogenesis of Alzheimer's disease (AD), a condition where deposition of extracellular amyloid beta (Ab) plaques in the brain and neuronal cell death accompany neuroinflammation. In AD patients the concentration of proinflammatory cytokines, including interleukin IL-1b is increased, possibly as a result of the activation of the NLRP3 inflammasome, and immune regulatory mechanisms, including those mediated by Tregs and PDL-1, are impaired [23][24][25][26][27]. A more subtle form of neuroinflammation is also present in mild cognitive impairment (MCI), a subjective and objective decline in cognitive performance that is greater than expected for an individual's age and education level, but does not meet criteria for the diagnosis of AD [28]. Elderly MCI patients are at high-risk for developing AD, this situation thus represents a borderline condition between normal aging and AD [29].
Correlates of MCI conversion to AD are still poorly defined, even if neuroinflammation is strongly suspected to play a role in this process [29][30][31][32][33].
Recent observations suggest a beneficial role of IL-33 in AD-associated neuroinflammation. To summarize: 1) ex-vivo results in cellular models of AD indicated that over expression of IL-33 decreases Aβ secretion; and 2) autoptic data indicate that IL-33 is significantly reduced in brains of AD patients. Notably, in the APP/PS1 animal model of AD, IL-33 administration was shown to restore the phagolysosomal activity of the microglia, thus enhancing amyloid β (Aβ) clearance, and to polarize monocytes to an anti-inflammatory phenotype. This resulted in a significant amelioration of AD symptoms [34], suggesting the possibility that IL-33 is a neuroprotective cytokine in AD [35]. This hypothesis nevertheless is not supported by other results suggesting that IL-33 is present in high concentrations in the neuropathological lesions of AD brain and can exacerbate AD-associated neuroinflammation [36]. In the attempt to clarify the role of the ST2/IL-33 axis in AD we analyzed these proteins in AD, MCI and HC individuals and explored the effect of IL-33 supplementation in an in vitro system of Aβ-stimulated monocytes.

Patients and controls
Ninety elderly Italian individuals were enrolled in the study by the Neurology Department of the IRCCS Ca' Granda Ospedale Maggiore Policlinico, Italy. Thirthy patients had a diagnosis of AD, 30 patients had a diagnosis of mild cognitive impairment (MCI) and 30 individuals were age and sex matched healthy controls (HC). The clinical diagnosis of AD was performed according to NINCDS-ADRDA work group criteria [28] and further revisions [37], the clinical diagnosis of MCI fulfilled Petersen's operational criteria [29]. Neuropsychological evaluation was performed with a Mini-Mental State Examination (MMSE) [38] and Clinical Dementia Rating Scale (CDR) [39].
All patients underwent a clinical interview, neurological and neuropsychological examination, routine blood tests, brain MRI, and lumbar puncture (LP). The mean age of AD patients (13 males and 17 females) was 74.3 years (age range 54-88 years) and that of MCI patients (8 males and 22 females) was 75.3 years (age range 63-84 years). All the AD patients were enrolled in a Multidimensional Stimulation Rehabilitation program designed for this pathology. Finally, the thirty age-and sex-matched elderly subjects (HC) who were included in the study were selected according to the SENIEUR protocol for immuno-gerontological studies of European Community's Control Action Programme on Aging [40]; their MMSE score was > 28.
The study was approved by the Ethics Committee of Don Gnocchi Foundation and informed written consent was obtained from all the included subjects before study initiation.

Serum and CSF
Serum was collected in vacutainer tubes containing serum separator (Becton Dickinson and Co., Rutherford, NJ, USA) and were centrifuged at 3,000 rpm for 10 min to separate sera; CSF samples were collected by lumbar puncture. Serum and CSF were used immediately or stored at -80°C.

Blood sample collection and cell separation
Whole blood of a subset of individuals (5 AD, 5 MCI and 5 HC) was collected in vacutainer tubes containing ethylenediaminetetraacetic acid (EDTA) (Becton Dickinson). Peripheral blood mononuclear cells (PBMC) were separated on lympholyte separation medium (Cedarlane, Hornby, Ontario, CA) and washed twice in PBS at 1500 rpm for 10 min; viable leukocytes were determined using a TC20 Automated Cell Counter (Biorad, Hercules, CA, USA).

Cell cultures
PBMC (1x10 6 /ml) were cultured in RPMI 1640 supplemented with 10% human serum, 2mM L-glutamine, and 1% penicillin (Invitrogen Ltd, Paisley, UK) and incubated at 37°C in a humidified 5% CO 2 atmosphere for 2 hours in a 12 wells plate for monocyte adhesion. After 2 hours, non-adhering PBMC were harvested and discarded and monocytes grown on plate were either culture in medium alone (unstimulated) or were or primed with 2mg/ml Lipopolysaccharide (LPS) for 2 hours

Western blot
A Qubit Protein Assay Kit was used for protein extraction; total protein amount was

Statistical analysis
Quantitative data were not normally distributed (Shapiro-Wilk test) and were summarized as median and Interquartile Range (IQR) (25° and 75° percentile).
Comparisons between groups were performed used a Kruskal-Wallis ANOVA for each variable. Comparisons among the different groups were made using a 2-tailed Mann-Whitney U test performed for independent samples. Data analysis was performed using the MedCalc statistical package (MedCalc Software bvba, Mariakerke, Belgium).

IL-33 isoforms
Analysis of the different isoforms of IL-33 (full-length; cleaved inactive; cleaved active) was performed by Western blotting in serum and CSF of AD, MCI and HC individuals. Results showed that the full-length IL-33 protein (30kDa band) was present in all the serum and CSF samples analyzed with the exception of two AD sera. In these two particular samples, a 20 KD band, corresponding to the cleaved (c), inactive form of IL-33 [3] was observed. These results are shown in Figure 1A e 1B.

IL-33 concentration was decreased in serum and CSF of both AD and MCI individuals
compared to HC (p value vs. HC: serum, AD p= 0.02; MCI p=0.04; CSF, AD p=0.01; MCI p=0.009). In contrast with these results, serum concentration of the IL-33 decoy receptor sST2 was significantly increased in AD and MCI compared to HC (p value vs. HC: p=0.02 and p= 0.01 respectively) ( Figure 2). CSF sST2 concentration was below the limit of detection of the assay (33pg/ml) in all the analyzed samples (data not shown).

IL-1b, IL-6 and IL-10 serum and CSF concentration
IL-1b serum concentration was significantly increased in AD and MCI sera compared to HC (p=0.01 in both cases) whereas that of IL-10 was significantly reduced in both groups of patients compared to HC (p=0.04 in both cases). No significant differences were detected when IL-6 serum concentration was compared between the three groups. These results are shown in Figure 3.
In CSF, IL-1b concentration was significantly augmented in AD and MCI compared to HC (p=0.02 and p= 0.01 respectively). IL-6 concentration was slightly augmented and IL-10 concentration was marginally reduced in AD and MCI compared to what was seen in HC but these differences did not reach statistical significance ( Figure   3).

Monocytes of AD, MCI and HC individuals (n=8 in all cases) were LPS primed and
Ab 42 -stimulated and IL1b, IL-6 and IL-10 production was measured by ELISA.
Whereas the production of these cytokines was minimal (<20 pg/ml in all cases) and comparable between the three groups of individuals analyzed when monocytes were unstimulated (medium alone) or when IL-33 was added to the culture medium, LPSprimed and Ab 42 -stimulated-monocytes of AD and MCI patients produce significantly augmented quantities of IL-1b (median: AD= 230 ng/ml; MCI= 140ng/ml ) compared to HC (median: 62 ng/ml; AD vs. HC p=0.006; MCI vs. HC p=0.03). Notably, IL-10 production by cells stimulated in the same conditions was significantly reduced in AD (median:129 ng/ml) compared to HC individuals (median: 245 ng/ml)(p=0.01).
Addition of IL-33 to cell cultures resulted in a significant reduction of the production of both IL-1b (median: 36 pg/ml; p=0.04) and IL-6 (median: 616 pg/ml; p=0.01) in HC, and of IL-1b alone in MCI (median: 23 pg/ml; p=0.02) and, although without reaching statistical significance. Finally, LPS+Ab 42 -stimulated IL-10 production was not modified by IL-33 in AD and MCI, but was significantly increased in HC (median: 585 pg/ml; p=0.02) ( Figure 4).
These results support the idea that IL-33 reduces the generation of proinflammatory cytokines and stimulates that of anti-inflammatory cytokines, favoring the establishment of an anti-inflammatory milieu, and suggest that IL-10-mediated anti-inflammatory mechanisms are impaired/exhausted in AD and MCI.

Recombinant IL-33 reduces NF-kB nuclear translocation
Unstimulated as well as LPS primed and Ab 42 -stimulated-monocyte of AD (n=5), MCI (n=5) and HC individuals (n=5) were analyzed to verify the effect of IL-33 on NF-kB nuclear translocation using the FlowSight technology. No differences in NF-kB nuclear translocation were detected in unstimulated cells (data not shown); upon antigenic stimulation, the IL-33 was observed to significantly reduce NF-kB nuclear translocation (p=0.01) in HC alone ( Figure 5). These results suggest that this is a pivotal IL-33 anti-inflammatory mechanism that is possibly lost in MCI and AD.

Discussion
Alzheimer's disease is a highly prevalent form of dementia characterized by the accumulation of extracellular amyloid beta plaques in the brain, neuronal cell death Results herein show that the IL-33/ST2 axis is deeply impaired in MCI and AD. Thus, IL-33 concentration in serum and CSF was significantly reduced whereas the concentration of sST2, the IL-33-specific decoy receptor, was significantly increased in the same patients. In AD, thus, lower amounts of IL-33 are produced and they cannot optimally bind their cognate receptor on the surface of cells, as they are trapped by soluble decoys.
That reduced concentrations of IL-33 are associated with inflammation was indirectly confirmed by the observation that IL1b concentrations were significantly increased whereas that of IL-10 was greatly reduced in the same AD and MCI patients. The anti-inflammatory role of IL-33 was further reinforced by the observation that IL-33 supplementation resulted in the reduction of IL-1b and IL-6 generation by Ab-stimulated cells of AD and MCI. Importantly, IL-33 increased IL-10 production by cells of HC alone, suggesting that the ability of monocytes of AD and MCI individuals to secrete IL-10 in response to IL-33 is deeply defective or exhausted.
IL-1b is augmented in AD. This cytokine promotes amyloid plaque deposition, reduces phagocytic activity by the microglia, stimulates the hyperphosphorylation of t protein, and affects synaptic plasticity. IL-10 production, on the other hand, has repeatedly been shown to be reduced in AD patients [24,41,42], in whom the IL-10 gene SNPs associated with higher production of this cytokine are also less frequently detected [33]. Results herein could thus at least partially justify the genesis of the proinflammatory mileu seen in AD and MCI as the consequence of the reduced amounts of IL-33 seen in these patients.
Notably, IL-33 significantly down-modulated NF-kB nuclear translocation in cells of HC alone, indicating this as a possible mechanisms by which IL-33 favors the maintenance of an anti-inflammatory milieu in individuals in whom dementia is not present. Thus, IL-33 was shown to have a regulatory effect on the NF-kB pathway, which is mediated by its ability to binding the NF-kB p65 subunit [9]. The IL-33/NF-kB p65 complex impedes p65-mediated transactivation; this down regulates NF-kB activity, with a dampening effect on inflammation. The observation that this effect was lost in cells of MCI and AD individuals suggests that the role of IL-33 in the pathogenesis of AD-and MCI-associated inflammation is multifaceted, as this cytokine would exert its effect both at the intracellular and at the extracellular level. To summarize: 1) reduction of IL-33 production per se, 2) increased concentrations of the sST2 decoy receptor, and 3) an impairment of the ability of IL-33 to induce NF-kB nuclear translocation are different mechanism that can explain the proinflammatory role played by IL-33 in AD and MCI.
IL-33 was also reported to induce the polarization of monocytes toward an M2 antiinflammatory phenotype; these monocytes produce IL-10, whose concentration, as indicated above, is greatly reduced in AD and MCI individuals [43][44]. M2 polarization by IL-33 is possibly explained by the observation that the IL-33 binding to ST2 negatively regulates TLR signaling by competing with MyD88 [45][46]. The IL-33/ST2 complex thus suppresses IL-1b generation and the down-stream activation of the TLR signaling pathway by sequestration of MyD88 with the consequence of further inhibiting NF-kB and activating MAP kinases [46]. IL-33 indeed activates ERK1\2 and STAT3, facilitating binding of STAT3 to the -1954 to − 1936 bp sequence upstream of the IL-10 transcription start site, thereby promoting its transcription in macrophages [47]. Although we did not analyze M1 and M2 monocytes in this study, previously published data indicate that M2 cells are significantly reduced in AD. The reduction of M2 monocytes seen in AD could explain why IL-33 supplementation did not increase IL-10 production by monocytes of AD and MCI individuals in our in vitro system and could be a third way to justify why the lower quantities of IL-33 seen in AD results in inflammation in these patients.
IL-33 has repeatedly been suspected to be involved in the pathogenesis of CNS diseases. In particular the administration of recombinant IL-33 was shown to promote recovery in a mouse model of autoimmune encephalomyelitis (EAE) [48][49] and to provide neuroprotection in a mouse model of contusion spinal cord injury (SCI) [43]. In the APP/PS1 animal model of AD, IL-33 attenuated AD pathology and memory deficit and stimulated the polarization of microglia in an anti-inflammatory direction [50][51]. Finally, recent data obtained in a small groups of MCI who did or did not convert to AD over time [31] showed that higher amounts of IL-33-producing CD14+ monocytes were seen in AD-non converters, in whom percentages of