In this imaging genetics analysis, we investigated for the first time the effect of rs1143627 (−31 T > C) loci polymorphism of IL1B gene on brain regional spontaneous activity in aged subjects presenting with aMCI and normally aging participants. We found that, among the CC, CT, and TT genotypes, many brain regions showed significant differences in ALFF between the aMCI patients and the controls. Interestingly, we found that the regions of interaction between the genotypes and the groups showed some overlap with the default mode networks, including the Pcu, the superior frontal gyrus, the anterior cingulum cortex (ACC) and some other frontal regions. Further behavioral significance suggested a role for the rs1143627 polymorphism in regional neuronal activity that has some effects on cognitive aging.
The protein IL-1β, encoded by a 7.5 kb gene with seven exons, is regulated both by distal and proximal promoter elements . The promoter SNP locus rs1143627 (−31 T > C) is located in a TATA-box motif, which markedly affects DNA-protein interactions in vitro, and the change from −31 T to -31C may disrupt the TATA-box, leading to potential reduced or abolished transcriptional activity of the promoter [34, 35].A genome-wide association study (GWAS) confirmed that −31 T > C had the highest LOD score, providing strong, unbiased evidence that this SNP is functional and that IL-1β mRNA expression is a heritable trait .
It has been demonstrated both in vivo and in vitro that basal IL-1β expression is involved in the physiological long-term potentiation (LTP), a process believed to underlie certain forms of learning and memory . At sufficient concentrations, IL-1β has an inhibitory effect on LTP in many regions of the hippocampus . Additionally, the injection of lipopolysaccharide (LPS), a potent inducer of IL-1β expression, into the CA1 regions of the rat hippocampus results in learning and memory deficits . Postmortem brain tissue from patients with early-stage AD showed significantly increased expression of caspase-1 (CASP1), also known as IL-1β converting enzyme (ICE) . A previous meta-analysis of cytokines found a significantly higher concentration of IL-1β in the peripheral blood of AD patients . A case–control study on CSF concluded that intrathecal inflammation precedes the development of AD , and found the levels of IL-1β were significantly correlated with the levels of tau and Aβ in patients with MCI who had progressed to AD at the follow-up nine months later. Furthermore, it was found the IL-1β levels were correlated to MMSE at both baseline and follow-up assessments. There have been many associating IL1B gene SNPs with the risk of occurrence of AD, but the results are controversial (see http://www.alzgene.org/). A previous genetic association study of Chinese participants did not detect any genotypic or allelic frequency differences in IL1B -31C/T between an AD group and a control group, and concluded that the -31C/T polymorphism was not a risk factor for AD , while the result needs to be further confirmed.
Our study showed decreased ALFF in the inferior frontal gyrus, the superior temporal lobe and the middle temporal lobe in aMCI patients, which was consistent with previous hypoperfusion shown by SPECT and hypometabolism shown by PET in aMCI/AD [43, 44]. It is worth noting that the frontal and temporal lobes, together with the parietal and occipital cortices, were previously identified with more accumulation of 11C-Pittsburgh compound B (PIB), consistent with brain regions containing Aβ deposits . A previous resting state fMRI study also found that reductions in regional activation [46, 47], regional homogeneity [10, 48] and functional connectivity [49–52] were associated with both AD and MCI patients. A recent combined structural and resting-state fMRI study  found broad frontotemporal grey matter loss in MCI patients by voxel-based morphometry (VBM) analysis, and found regions of decreased connectivity to the posterior cingulated cortex (PCC), which provides further support for our findings. In the current study, we also observed increased ALFF in the aMCI group, including in the occipital cortex (calcarine), parietal cortex (Pcu) and cerebellar cortex. Jia and colleagues  found increased ALFF/fractional ALFF (fALFF) activity in aMCI patients in several occipital regions, which is consistent with the present study. A previous PET study  showed a significant increase in metabolism in the occipital cortex in aMCI brains. Researchers have also observed that baseline aMCI patients have increased cerebellar ALFF compared to controls . Thus, our results of ALFF in aMCI suggested that aMCI patients might have relatively conserved spontaneous brain activity in these regions. The increased ALFF in the Pcu of aMCI patients in this study was in contrast to previous studies, which showed a decrease in ALFF. A structural MRI study showed that Pcu undergoes significant atrophy in MCI and mild AD . In addition, lower functional connectivity within the default mode network (DMN) in the Pcu and PCC was detected in AD patients by an independent component analysis (ICA) . The PCC and Pcu of aMCI patients displayed decreased ALFF/fALFF in a recent study . We found similar results regarding Pcu in the present study. A recent resting state fMRI study  identified PCC and Pcu hyperfunctional connectivity at baseline in aMCI subjects, and a substantial decrease in these connections was evident at follow-up, compared to matched controls. Additionally, Jones et al.  analyzed task-free fMRI data with both ICA and seed-based analysis, identifying the posterior DMN as having a decline in connectivity, while an increase in connectivity was observed in the Pcu when atrophy correction was applied. A previous study  demonstrated that endogenous neuronal activity regulated the regional concentration of interstitial fluid Aβ levels, which drive local Aβ aggregation. It is uncertain whether some regions with early increased regional neural activity and hyperfunctional connectivity could be a mechanism of compensation for AD pathology or represent a more proximate event leading to cognitive dysfunction. In summary, our results suggested that aMCI patients had abnormal ALFF in their intrinsic brain activity.
In this study, we observed that the rs1143627 C/T genotype had main effects on the frontal cortex, including the superior, middle, inferior and medial frontal gyri, and the ACC. SNPs in the IL1B gene were shown to have effects on the responsiveness of the amygdala and ACC to emotional stimulation in major depression . Additional brain regions that had interactions with groups were in the frontal, parietal and occipital cortices and in the posterior lobe of the cerebellum. A recent genetic neuroimaging study  detected the effect of the IL1B gene on frontal cortex function in schizophrenia. A mammalian genome study identified IL1B gene polymorphisms that modulated scrapie (a type of neurodegenerative disease) susceptibility in sheep and goats, and found IL-1β expression in the cerebellum . These studies supported our findings of the genetic effects of rs1143627 on different brain regions. In addition, the CC genotype showed significantly higher ALFF values than the CT and TT genotypes in both aMCI and control participants, and each genotype showed different or completely opposite results in the association between ALFF and cognitive scores. This study did not find the frequency of the three genotypes between groups, consistent with a previously mentioned study . We did not find any significant effects of the genotypes on cognition in aMCI patients. This was not in contradiction to the behavioral significance, as the pathophysiological process of AD is thought to begin many years, even decades, before the onset of clinical dementia [64, 65]. Thus, further large-scale association replicated and longitudinal studies are needed.
There were biological and technical limitations to this study that must be acknowledged. First, there was biological and clinical heterogeneity in the sample of aMCI participants as recruitment was based only on clinical criteria. Some participants may not display AD pathology, contaminating the sample with non-AD cases. The heterogeneity could be minimized through a combination of CSF and PET biomarkers. Second, as a result of the limitation of the sample size, further adjustment for multiple testing outside the domain of imaging will be necessary and replication in independent samples is also required to further establish the gene-imaging phenotype association. Third, previous resting state fMRI studies [66, 67] have demonstrated very moderate test-retest reliability based on Monte Carlo simulations and, because calculation methods will be improved with time, new analysis methods could be applied to future studies.