The NLRP3 and NLRP1 inflammasomes are activated in Alzheimer’s disease

Background Interleukin-1 beta (IL-1β) and its key regulator, the inflammasome, are suspected to play a role in the neuroinflammation observed in Alzheimer’s disease (AD); no conclusive data are nevertheless available in AD patients. Results mRNA for inflammasome components (NLRP1, NLRP3, PYCARD, caspase 1, 5 and 8) and downstream effectors (IL-1β, IL-18) was up-regulated in severe and MILD AD. Monocytes co-expressing NLRP3 with caspase 1 or caspase 8 were significantly increased in severe AD alone, whereas those co-expressing NLRP1 and NLRP3 with PYCARD were augmented in both severe and MILD AD. Activation of the NLRP1 and NLRP3 inflammasomes in AD was confirmed by confocal microscopy proteins co-localization and by the significantly higher amounts of the pro-inflammatory cytokines IL-1β and IL-18 being produced by monocytes. In MCI, the expression of NLRP3, but not the one of PYCARD or caspase 1 was increased, indicating that functional inflammasomes are not assembled in these individuals: this was confirmed by lack of co-localization and of proinflammatory cytokines production. Conclusions The activation of at least two different inflammasome complexes explains AD-associated neuroinflammation. Strategies targeting inflammasome activation could be useful in the therapy of AD.


Background
Alzheimer's disease (AD) is a devastating neurodegenerative condition characterized by neuronal cell death and progressive dementia. It is widely accepted that the extracellular accumulation of amyloid-β (Aβ) in senile plaques and the formation of neurofibrillary tangles (NFT) inside neurons as a result of the abnormal phosphorylation of the microtubules-associated tau protein are the main event in the pathogenesis of AD, but the cellular events leading to plaque-induced neuronal dysfunction are less clear [1]. Inflammatory mediators play an essential role in the neuroinflammation observed in AD. In particular interleukin IL-1β is increased in this condition, and the activation of its key regulator, the inflammasome, is suspected to be involved in the pathogenesis of the disease.
The inflammasomes are multiprotein complexes mainly expressed in myeloid cells that are required for the activation of caspase 1 protease and the downstream secretion of two of its substrates, the proinflammatory cytokines IL-1β and IL-18 [2]. Inflammasome-dependent innate immune responses are initiated by Nod-like receptors (NLRs), cytoplasmic pattern recognition receptors that detect invading pathogens. NLRs are activated by bacterial, fungal, or viral molecules that contain pathogen-associated molecular patterns (PAMPs) or by non-microbial danger signals (DAMPs) released by damaged cells [3,4]. NLR activation leads to their oligomerization to form multiprotein inflammasome complexes that serve as platforms for the recruitment, cleavage, and activation of inflammatory caspases. Four inflammasome complexes (NLRP1, NLRP3, IPAF, and AIM2) have been identified. These complexes contain either a specific NLR family protein or AIM2, as well as the PYCARD and/or Cardinal adaptor proteins and procaspases-1, 5 and 8 [5,6]. NLRP3 is the best-characterized inflammasome; its formation requires multiple steps. In a priming step, transcriptionally active signalling receptors induce the NF-kB-dependent induction of NLRP3 itself as well as that of the caspase 1 substrates of the pro-IL-1β family [7,8]. The NLRP3 is, at this stage, in a signalling incompetent conformation; this is modified upon a second signal, that will induce the assembly of a multimolecular complex with PYCARD and caspase 1. Multiple signals, which are potentially provided in combination, thus trigger the formation of an active inflammasome, which, in turn, will stimulate the cleavage and the release of bioactive cytokines including IL-1β and IL-18 [8][9][10][11][12][13][14][15][16][17][18]. Although these cytokines have a beneficial role in promoting inflammation and eliminating infectious pathogens, mutations that result in constitutive inflammasome activation and overproduction of IL-1β and IL-18 have been linked to inflammatory and autoimmune disorders [19][20][21].
The NLRP3 inflammasome complex is suspected to play a role in AD, as its activation in the microglia by Aβ triggers neuroinflammation [14,22,23]. Notably, NLRP3 inflammasome deficiency favours the differentiation of microglia cells to an M2 (anti-inflammatory) phenotype and results in a decreased deposition of amyloid-β in the APP/PSI model of AD; these results reinforce the suggestion that the NLRP3 inflammasome is involved in the pathogenesis of the disease [24]. No conclusive data are nevertheless available in patients. To better investigate the involvement of the inflammasome in AD we performed in-depth analyses of the expression of genes and proteins that are part of the inflammasome complex in individuals with a diagnosis of either AD or mild cognitive impairment (MCI). Results indicated that two different inflammasomes, NLRP3 and NLRP1, are activated in AD but not in individuals with a diangnosis of mild cognitive impairment (MCI).

Results
Up-regulation of inflammasome genes in LPS-primed and Aβ 42 -stimulated-monocytes of AD patients mRNA expression of 84 genes involved in the assembly, the activation, and the down-stream signalling of inflammosomes was quantified by qPCR in all patients and controls. Data are expressed as the fold change (nFold) between the results obtained in unstimulated cells and those obtained upon stimulations of cells with either LPS, Aβ42 or LPS and Aβ42. In this initial set of experiments, pools of cells from donors with the same clinical diagnosis (MCI; MILD AD; severe AD) as well as from HC were created. Data obtained in LPS-primed and Aβ 42− stimulated-monocytes, showed the presence of a significant upregulation involving genes, that codify for the proteins that form the inflammasome in MILD and severe AD, as well as in MCI. In cells of MCI individuals, though, not all the genes that are necessary for the assembly of a fully functional inflammasome complex were up-regulated. To summarize: 1) the expression of NLRP1, NLRP3 and caspase 8 was greatly increased in all groups of patients (nFold >10) compared to HC, with no detectable differences being observed between individuals with a diagnosis of AD or MCI; and 2) PYCARD, caspase 1 and caspase 5 expression was increased in AD (nFold ≥10) but not in MCI individuals. Only minor differences were seen between patients and controls in the expression of these genes in monocytes-stimulated with either LPS or Aβ 42 alone; both stimuli induced IL-1β and IL-6 up-regulation in HC (nFold >10) (Fig. 1).
RT-PCR performed on each individual specimen was used next to verify the expression of NLRP1, NLRP3, PYCARD, caspase 1, caspase 5 and caspase 8 in LPS and Aβ 42− stimulated monocytes of patients and HC. Results of mRNA expression confirmed that: 1) NLRP1, NLRP3 and caspase 8 are significantly increased in AD and MCI individuals compared to HC (p <0.05), with the highest values observed in severe AD, and 2) PYCARD, caspase 1 and caspase 5 are significantly increased in AD compared to MCI and HC (p <0.05) (Fig. 2a).

Up-regulation of inflammasome-related cytokines in LPS-primed and Aβ 42− stimulated-monocytes of AD patients
Stimulation of monocytes with either LPS or Aβ 42 -alone resulted in the moderate increase of IL-1β and IL-6 expression; notably LPS or Aβ 42 -alone were not sufficient to trigger inflammasome assembly, as no differences were detect in these conditions either in NLRP3 or IL-18 mRNA (Fig. 1). Once cells were activated with LPS and Aβ 42 , nevertheless, both IL-1β (nFold >10) and IL-18 (nFold ≥10) mRNA was significantly increased in severe and MILD AD compared to MCI and HC (Fig. 1). Results obtained by single-PCR confirmed that in these experimental conditions IL-1β and IL-18 mRNA (p <0.05 vs. all other groups) is significantly increased in severe and MILD AD, with the highest values in severe AD (Fig. 2b).
IL-33 and IL-37, two relatively novel cytokines that are members of the IL-1 family, were analyzed as well in the study. We found that IL-33 gene expression was greatly increased in MCI (nFold >10) compared to AD and HC; no significant differences, on the other hand, were observed in IL-37 gene expression among groups analyzed (Fig. 1). Single RT-PCR confirmed these results (Fig. 2c).
NLRP3 protein is significantly augmented in LPS-primed and Aβ 42− stimulated-monocytes of AD and MCI patients The expression of NLRP3 protein, the best-characterized protein within the inflammasome complex, was investigated next by Western Blot analyses in LPS-primed and Aβ42-stimulated-monocytes of all patients and controls. Different NLRP3 isoforms are known: the long form (118 KDa) is prevalent in cell lines (THP1 or Jurkat) while the short one (NLRP3s 75 KDa) is seen in primary immune cells [25]. Results of analyses performed on whole-cell lysates confirmed that NLRP3s expression is  (Fig. 3).
LPS-primed and Aβ 42− stimulated monocytes that express inflammasome proteins are significantly augmented in AD Peripheral CD14+ monocytes that express NLRP3 or coexpress NLRP3 and PYCARD, NLRP3 and caspase 1, or NLRP3 and caspase 8 were augmented in LPS-primed and Aβ 42− stimulated-cells of both groups of AD patients compared to MCI individuals and HC. The Fluorochrome Inhibitor of Caspases (FLICA) kit was used to analyze the presence in cells of active caspases; in this method, once inside the cell, the FLICA inhibitor probe binds covalently to active caspases (p20) alone. Staining of active caspase 1 and caspase 8 inhibitors was performed using the green fluorescent probe FAM-YVAD-FMK and FAM-LETD-FMK respectively. Results showed that 1) CD14+/NLRP3+, CD14+/NLRP3+/caspase1+, and CD14+/NLRP3+/Cas-pase8+ immune cells were significantly increased in severe AD patients alone compared to MCI and HC (p <0.05) and 2) CD14+/NLRP3+/PYCARD +, cells were augmented both in severe and MILD AD compared to MCI and HC (MCI and HC vs. moderate AD p <0.05; vs. severe AD p <0.01) (Fig. 4a). LPS-primed and Aβ 42 -stimulated NLRP1-expressing CD14 cells were significantly increased as well in both groups of AD patients compared to MCI and HC (p <0.05), while CD14+/NLRP1+/PYCARD + and CD14+/NLRP1+/caspase1+ immune cells were augmented, although not significantly, in AD patients compared to other two groups. Finally, no differences could be observed in CD14+/NLRP1+/caspase 5+ cells (Fig. 4b). These results confirm those obtained in PCR analyses and indicate that, even if some inflammasome components are up regulated in MCI, fully functional inflammasomes are not assembled in this situation.
Co-localization of inflammasome proteins in LPS-primed and Aβ 42− stimulated monocytes by confocal microscopy analyses Aggregation and activation of inflammasome complexes was further analyzed by evaluating the co-localization of NLRP3 and NLRP1 with PYCARD, caspase 1 or caspase 8 by confocal microscopy. Co-localization efficiency was calculated using the Pearson co-localization coefficient (PCC). Results showed a clear co-localization of NLRP3 with PYCARD, caspase 1 and caspase 8 in LPS-primed and Aβ 42 -stimulated monocytes of AD patients alone; no co-localization could be detected in untreated cells (data not shown). In particular: 1) NLRP3/PYCARD, NLRP1/PYCARD and NLRP3/caspase1 co-localization was increased in both groups of AD patients compared to HC and MCI; 2), NLRP3/caspase8 and NLRP1/cas-pase5 co-localization was increased in severe AD alone; and 3) NLRP1/caspase1 co-localization was seen only in cells of MILD AD individuals. In all cases, with the exception of the NLRP1-caspase 1 complex, the highest PCC values were detected in cells of patients with severe AD disease ( Fig. 5 and Table 1).

Inflammasome-related cytokines production in LPS-primed and Aβ 42− -stimulated -monocytes of AD patients
Aggregation of the inflammasome results in the downstream production of pro-inflammatory cytokines. Because FACS and confocal microscopy data indicated that the different subunits that compose the inflammasome do aggregate in LPS-primed and Aβ 42− -stimulated-monocytes of AD patients, and PCR analyses showed that mRNA for these cytokines is increased in monocytes of AD patients, IL-1β, IL-18, IL-33 and IL-37 were next measured. Results showed that the concentration of IL-1β, IL-18 in supernatants was significantly increased in both groups of AD patients compared to MCI and HC (IL-1β p <0.05; IL-18 p <0.01)( Fig. 6a and b). CD14+/IL-1β+ cells were significantly increased as well in MILD and severe AD compared to MCI and HC (p <0.05) (Fig. 6a), whereas lack of proper reagents prevented IL18-expressing cells to be analyzed by FACS. The fact that IL-1β+ and IL-18 were not increased in MCI is not surprising considering that in this condition NLRPs but not PYCARD expression was augmented, thus preventing the assembly of a functional inflammasome; these data confirm the findings obtained with FACS and confocal analyses.
IL-33 could not be quantified in supernatants (data not shown), probably because the inflammatory mature form of this cytokines is not cleaved and secreted. Results obtained by flow-cytometry, nevertheless showed that CD14+/IL-33+ cells were increased in MCI and in MILD AD, with the lowest percentages of these cells  seen in severe AD (p <0.01) (Fig. 6c). Notably this cytokine was shown to have a multifaceted protective role against AD and to negatively modulate NF-kB activity, dampening inflammation. Finally, CD14+/IL-37+ cells were marginally increased in AD compared to MCI and HC (Fig. 6d).

Discussion
Inflammasomes are intracellular complexes formed by the assembly of multiple subunits that regulate the maturation and the secretion of pro-inflammatory cytokines [2]. At least four different inflammasomes are known; the NLRP3, in particular, is suspected to play a role in the pathogenesis of AD. Thus, in the APP/PSI animal model of AD, NLRP3 up regulation induces the production of IFN1β by microglia [14]; and, on the other hand, its deficiency results in a decreased deposition of amyloid-β [24,[26][27][28][29]. We have previously shown that peripheral monocytes of AD individuals are characterized by an inflammatory profile and a higher surface density of TLR molecules [30]. Further analyses performed in these same individuals allowed us to demonstrate that these cells also express binary complexes formed by Aβ peptides and MHCmolecules, possibly initiating Aβ-specific acquired immune reponses [31]. Herein we investigate the role of the inflammasome in the neuroinflammation that accompanies AD by analyzing peripheral immune cells of patients with MCI or AD. Results obtained using molecular, confocal and cytofluorimetic analyses indicated that the NLRP3 and NLRP1 inflammasomes are indeed activated in AD. The assembly of functional inflammasomes in AD was confirmed by the significantly increased amount of the proinflammatory cytokines IL-1β and IL-18 that were produced by LPS-primed and Aβ 42− -stimulated-monocytes of AD patients. Notably, NLRP1, NLRP3 and caspase 8, but not PYCARD were increased in individuals with a diagnosis of MCI, a condition that is often but not always prodromic to the development of AD. In this situation the assembly of a functional inflammasome is not possible. Consequently, neither a significant co-localization of inflammasome subunits nor increased amount of IL-1β and IL-18 were detected in these individuals. It is interesting to observe that in MCI a partial, initial up regulation of inflammasome proteins limited to NLRP proteins is observed. Only a percentage of MCI progresses to AD; these results thus allow the speculation that the inflammasome is somehow "primed" to be activated in all individuals with a diagnosis of MCI. An increased transcription of the other inflammasome components, the assembly of the functional complex, and the production of proinflammatory cytokines, occurs only in some cases, possibly because of still unknown "triggering" events that act on an "NLRP-primed" background.
NLRP3 and caspase 8 mRNA were up regulated in cells of severe AD patients; CD14+/NLRP3+/caspase8+ cells were increased as well in these individuals in whom the co-localization of caspase 8 with NLRP3 was also detected. Recent data showed that caspase 8 is not only an inducer of cell apoptosis but, together with the Fas-Associated protein with Death Domain (FADD), it interacts with NLRP3. Such interaction is required for caspase 1 activation, as well as for IL-1β and IL-18 secretion [5]. Confocal analyses confirmed that caspase 8 is present in the NLRP3 inflammasome complex, where it is involved in the cleavage of pro caspase 1 and IL-1β. These results support a direct role for caspase 8 in the processing of caspase 1 [6]. Our data confirm these findings, as well as recent results indicating that caspase 8 contributes to both NF-kB-dependent priming and post-translational activation of the NLRP3 inflammasome [5]. Increased NLRP1 and caspase 5 mRNA levels were also detected in cells of individuals with a diagnosis of severe AD. NLRP1 and caspase 5 colocalized in CD14 cells, and higher percentages of CD14+/NLRP1+/caspase 5+ immune cells were also seen in these patients. These data indicate that the NLRP1 inflammasome complex is also activated in severe AD; notably, recent results indicate that SNPs in the NLRP1 gene are associated with AD [32].
Inflammasome assembly leads to the production of proinflammatory cytokines; IL-1β and IL-18 mRNA and secretion were indeed significantly increased in AD. Interestingly, incubation of cells with Aβ 42 was sufficient to induce IL-1β mRNA expression, but a significantly increased production of IL-1β and IL-18 was seen in LPSprimed and Aβ 42 -stimulated cells of AD patients alone. These findings indicate that such experimental conditions result in the assembly of fully functional inflammasomes only in cells of AD patients. Notably, transfection of human PBMC with siRNA specific for NLRP3 and PYCARD greatly reduced IL-1β production, confirming that the production of this cytokine is dependent on inflammasome assembly [33]. That IL-1β plays a role in the pathogenesis of AD has been repeatedly shown. To summarize: this cytokine induces a loss of phagocytic activity by the microglia [34], stimulates the hyperphosphorylation of tau protein [35] and affects synaptic plasticity. As a consequence of these effects, IL-1β can impair learning and memory processes [36,37]. IL-18 has been involved in AD as well, even if fewer data are available. Thus, increased amounts of IL-18 were detected in the brain [38], plasma [39], and peripheral blood lymphocytes of AD [40], and the expression of the IL-18R complex is greatly augmented in peripheral blood cells of MCI and AD individuals [41]. Finally, recent data indicate that IL-18 directly stimulates Aβ production by human neuronlike cells, suggesting a key role for this cytokine in the pathogenesis of AD [42].
Two other novel cytokines, IL-33 and IL-37 were analyzed as well in this study; these cytokines are members of the IL-1β family and are produced upon activation of the inflammasome. Whereas no significant differences were seen in IL-37, mRNA levels of IL-33 were increased in both MILD AD and MCI compared to the values observed in severe AD. Analysis of peripheral CD14+/IL-33+ cells confirmed this observation, as these cells were increased both in MCI and MILD AD compared to severe AD and HC. IL-33 and IL-37 are dual function proteins with both intra and extra-cellular mechanisms of action as, besides being able to bind to their cognate receptors on target cells, they can act intracellularly as nuclear factor. IL-33 is reduced in AD brains [43]. This cytokine is believed to have a neuroprotective role secondary to the reduction of Aβ peptides secretion [44] and the activation of the phagocytosis of amyloid-β peptide by the microglia [45]. IL-33 also interacts intracellularly with the NF-kB p65 subunit [44]; the resulting IL-33/NF-kB p65 complex interferes with NF-kB-dependent transcription by impeding p65-mediated transactivation. This causes a negative modulation of NF-kB activity, with a dampening effect on inflammation [46]. The increase of IL-33 production in MCI and in MILD AD could thus be seen as an attempt of the immune response to reduce neuroinflammation. On the other hand, the observation that the levels of IL-33 seen in severe AD are comparable to those observed in elderly healthy individuals suggest that the IL33-dependent anti-inflammatory effects do not need to be activated in the absence of the pathological alterations present in AD.
Results herein stem from immunological analyses performed in peripheral blood immune cells; a number of data nevertheless provide evidence that the brain is invaded by peripherally derived monocytes/macrophages [47]. In murine models of Alzheimer's disease, mice brains are infiltrated by bone marrow-derived macrophages that associate with amyloid plaques and clear Aβ deposits from the brain [48]. Our data indicate that the CD14+ cells that are likely present in the CNS of AD patients can carry active inflammasome complexes. It also has to be remembered that the role of inflammation in AD is ambiguous. Thus, whether this phenomenon is considered to be responsible for AD-associated neurodegeneration, an alternate hypothesis sees it as a monocyte/macrophages-mediated attempt to slow down the accumulation of Aβ plaques in the brain. Longitudinal studies and analyses performed in animal models will be needed to clarify this issue.

Conclusions
NLRP3 up regulation in mice or human microglia cell line or in brain of AD patients has been previously reported [14,24,49], these are nevertheless the first data showing NLRP3 and NLRP1 inflammasome activation in Aβ stimulated peripheral monocytes of individuals with a diagnosis of AD. It will be interesting to verify whether these data could have a prognostic and/or diagnostic value in the clinical setting. In particular, the observation that some (NLRP1, NLRP3) but not not all (PYCARD, caspase 1) of the genes necessary for the assembly of an inflammosome complex genes are upregulated in MCI suggests that monitoring the transcription rate of, e.g. PYCARD in MCI could offer an early diagnostic tool for AD development. Migration of peripheral monocyte across the blood-brain barrier is likely an important factor in the neuroinflammation that accompanies Alzheimer's disease. Very recent results showed that nucleoside reverse transcriptase inhibitors inhibit the activation of the inflammasome [50]. If neuroinflammation is deleterious in AD, these drugs could be an interesting tool in the treatment of this disease. criteria [51] and the DMS IV-R [52]. All Alzheimer's disease patients underwent complete medical and neurological evaluation, as well as laboratory analysis, and CT scan or MRI. Additional investigations (e.g., EEG, SPECT scan, CSF examination, etc.) were performed in some cases to exclude reversible causes of dementia. Neuropsychological evaluation and psychometric assessment were performed with a neuropsychological battery that included the MMSE [53], Digit Span Forward and Backward, Logical Memory and Paired Associated Words Tests, Token Test, supra Span Corsi Block Tapping Test, Verbal Fluency Tasks, Raven Colored Matrices, the Rey Complex Figure, Clinical Dementia Rating Scale (CDR) [54], and the Hachinski Ischemic Scale. Individuals with a diagnosis of MCI were selected among subjects seen at our Memory Disorders Outpatients Service for the diagnostic evaluation of memory complaints without difficulties in daily activities. MCI diagnosis was based on Petersen's criteria [55] as follows: 1) reported cognitive decline, 2) impaired cognitive function, 3) essentially normal functional activities, and 4) exclusion of dementia. The healthy controls were selected according to the SENIEUR protocol for immunogerontological studies of European Community's Control Action Program on Aging [56,57]. The cognitive status of HC was assessed by administration of MMSE (score for inclusion as normal control subjects >28). Written consent was obtained and ethical approval was granted by the Ethics Committee of the Don C Gnocchi Foundation in Milano, Italy.

Blood sample collection and cell separation
Fifty ml of whole blood were collected in vacutainer tubes containing ethylenediaminetetraacetic acid (EDTA) (Becton Dickinson & Co., Rutherford, NJ, USA). 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 Scepter 2.0 Handheld Automated Cell Counter (Millipore, Billerica, MA).

Confocal microscopy analysis
Unstimulated, LPS-or Aβ 42 -primed, or LPS-primed and Aβ 42 -stimulated-PBMC were cultured on chamber slide (Lab Tek Nalge Nunc Intern. Naperville IL USA) for 24 h at 37°C. Non-adherent cells were then washed away and monocytes were grown on chamber slides, fixed in 4 % paraformaldehyde in PBS for 15 min, and treated for 1 h at room temperature with FLICA staining of active caspase 1 or caspase 8 using green fluorescent FAM-YVAD-FMK probes, following the procedures suggested by the manufacturer (AM-FLICA). Cells were then treated with FIX and PERM Cell kit (eBioscience), and stained with PE or APC or FITC conjugated -mAbs specific for NLRP3 or NLRP1 and PYCARD or caspase 5 for 24 h at 4°C. Finally cells were fixed with paraformaldehyde 1 % for 15 min, washed and mounted on slides using the Vectashield Mounting Medium (Vector Laboratories, Inc, Burlingame, CA, USA). Fluorescent images were acquired on a Leica TCS DMRE spectral laser-scanning confocal microscope (Leica Microsystems, Wetzlar, Germany) with the appropriate filters and laser (488, 633) and a 63× objective lens. Image analysis was performed using the Leica Confocal Software and co-localization index with ImageJ Software. Co-localization indexes were calculated using the plug in JACoP (Justo Another Co-localization Plugin) [62]. The summarized co-localization efficiency data was expressed as Pearson correlation coefficient (PCC) as previously described [63][64][65]. Briefly this coefficient measures the significance of true co-localization. The significance test evaluates the probability that the measured value of r obtained from the two colours is significantly greater than values of r that would be calculated if there were only random overlap. The test is performed by randomly scrambling the blocks of pixels (instead of individual pixels, because each pixel's intensity is correlated with its neighbouring pixels) in one image, and then measuring the correlation of this image with the other (unscrambled) image (Costes et al.) The test produces values in the range −1 + 1, 0 indicating that there is no discernible correlation and −1 and +1 meaning strong negative and positive correlations, respectively.

Statistical analyses
Quantitative data were not normally distributed (Shapiro-Wilk test) and are thus summarized as median and Interquartile Range (IQR) (25°and 75°percentile). Comparisons between groups were analyzed 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. Western Blot data were normally distributed, and were summarized as mean ± standard error. In this case, comparisons were performed using ANOVA and unpaired Student's t test. Data analysis was performed using the MedCalc statistical package (MedCalc Software bvba, Mariakerke, Belgium).