Interaction of Genetics, Amyloid-beta and Inflammation as Mediators of Brain Structure Change in Cognitive Decline

doi:10.21203/rs.3.rs-2084869/v1

Download PDF

Research Article

Interaction of Genetics, Amyloid-beta and Inflammation as Mediators of Brain Structure Change in Cognitive Decline

https://doi.org/10.21203/rs.3.rs-2084869/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

In the presence of AD pathologies, the genetic architecture of brain structure changes related to progressive cognitive deterioration is not well examined. Here, we seek to shed light in the interplay of inflammation, amyloid-beta, and genetic background influencing brain structure changes. We studied 12-month changes in neuroimaging, inflammation and Aβ42/Aβ40 in 1322 individuals (ADNI cohort) and ran GWAS as well as SNP effect concordance analysis to test for genetic pleiotropy of identified risk variants. We uncovered genome-wide significant hits for structural change in nine brain regions, as well as change in inflammatory and Aβ42/Aβ40 biomarkers. We also found significant evidence of pleiotropy and concordance for several of the implicated genetic variants. Conditioning on inflammatory and Aβ42/Aβ40 biomarkers, novel variants that underlie brain structure change were identified. Amyloid-beta could interact with inflammatory biomarkers to affect brain structures via SNP-SNP interaction. Our findings point to the Interaction of amyloid-beta and inflammation as mediators of brain structure change and associated cognitive decline in AD and shed light into the underlying genetic background.

Cognitive decline

Inflammatory biomarkers

Amyloid-beta

Magnetic resonance imaging

SNP effect concordance analysis

Alzheimer’s disease (AD), the most common type of dementia, features a gradual and irreversible decline in cognitive and executive function. It is a highly heritable neurodegenerative disorder (heritability estimate of 60–80%), and several large genome-wide association studies (GWASs) have confirmed more than 20 common variants for AD (such as APOE, PSEN1, PIK2B), and revealed rare genetic factors (such as TREM2 and ABCA7)[1]. Nevertheless, a large portion of AD heritability remains unexplained. Extracellular amyloid-beta plaques and intracellular neurofibrillary tangles have been regarded as the two primary pathological hallmarks of AD [2], and recently, a synergistic interaction between amyloid-beta accumulation and inflammation has been recognized as playing a key role in AD pathology [3]. In fact, neuroinflammation has emerged as the third core pathology of AD, providing novel insight into pathogenic mechanisms of AD [4, 5]. Neuroinflammation could be considered a bridge that links amyloid-beta and tau-protein pathologies. The presence of amyloid-beta could activate microglia and astrocytes to promote tau hyperphosphorylation, leading to the propagation of neurofibrillary tangles [6][7]. Inflammatory mediators (including the NLRP3 inflammasome) and components released by activated microglia and astrocytes create a toxic environment, which, in turn, intensifies the inflammatory immune response and causes further damage to neurons [4]. However, to date, the role of neuroinflammation in underlying AD-related cognitive decline is not well understood.

Brain atrophy is an additional characteristic feature of AD. Ventricular enlargement and decreases in brain volume, especially in the hippocampus and entorhinal cortex, correlate with progressive cognitive decline [8]. In addition, white matter hyperintensities (WMH) are considered a predictor of cognitive dysfunction [9]. Mechanisms underlying brain structure changes associated with cognitive dysfunction are thought to overlap with mechanisms involved in the emergence of AD pathologies such as amyloid-beta plaques and neuroinflammation [10]. Cross-sectional GWASs have also been undertaken to identify SNPs and genes that may be implicated in brain structure changes associated with AD [11, 12]. However, the genetic architecture of brain structure changes related to progressive cognitive deterioration over time is still not well examined, especially in the presence of AD pathologies.

Here, we seek to address the gaps in our understanding of the role of inflammation and brain atrophy in AD. We investigated the genetic background and genetic correlation of longitudinal changes in brain structure and levels of AD-related plasma biomarkers (i.e., inflammatory biomarkers and amyloid-beta) as related to cognitive decline over a 12-month period in the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort. Our findings offer insights into the genetic background of pathways that may lead to changes in brain structure associated with cognitive decline through neuroinflammation and Αβ pathology.

Dataset

Data were obtained from the ADNI cohort (http://www.loni.ucla.edu/ADNI/). The ADNI was launched in 2003 as a public-private partnership, led by Principal Investigator Michael Weiner. We analyzed four types of data in this dataset: 1) ADNI1 and ADNIGO/2 genotype data in plink format (BED/BIM/FAM); 2) plasma-based biomarker data at baseline and at the 12-month follow-up assessment; 3) Mini-Mental State Exam scores (MMSE) at baseline and at the 12-month follow-up assessment; and 4) T1-weighted and Fluid-attenuated inversion recovery (FLAIR) neuroimaging data.

Paired t-tests were used to study changes in MMSE for all individuals and for subjects within each group (AD, mild cognitive impairment (MCI), and control (CON)) at 12 months versus the baseline. Inflammation-related proteins and the Aβ40 and Aβ42 peptides were quantified in plasma using a multiplex immunoassay panel. The 12-month change in MMSE, plasma inflammatory biomarkers, and Aβ42/Aβ40 ratio was calculated. For 410 participants (AD = 87, MCI = 277, CON = 46), data for 123 inflammatory biomarkers were available, and for 520 participants (AD = 70, MCI = 338, CON = 112), Aβ42/Aβ40 measurements were available in addition to the inflammatory biomarker scores. Only inflammatory biomarkers that were found to be correlated with MMSE changes over the 12-month period were further investigated in our GWAS.

Neuroimaging analyses

Each participant’s baseline and 12-month follow-up T1-weighted and FLAIR images were analyzed. T1 weighted images were run with FreeSurfer version 6.0 and processed with 2010 Desikan-Killany atlas[13]. FreeSurfer was used to obtain average reconstructed regions of interest (ROIs). WMH segmentations were processed using T1-weighted images as well as reregistered and skull-stripped FLAIR images with the FMRIB Software Library (FSL) toolbox[14]. Quality control (QC) was performed following UCSF guidelines. 12-month changes in brain regions were calculated for 715 participants (AD = 148, MCI = 372, CON = 195) with complete image data. Pairwise correlation analysis between 12-month changes in MMSE and brain regions was used for variable selection (Bonferroni p* =0.0004).

Genetic analyses

Preprocessing: 1,547 participants from ADNI1 and ADNIGO/2 (AD = 547, MCI = 523, CON = 476) with genotype data were included. QC followed the Ricopili pipeline (https://github.com/Ripkelab/ricopili/wiki). Principal component analysis (PCA) and Admixture software were used to assign and cross-validate ethnic descent groups[15]. Following QC, we were left with a dataset of 470,865 SNPs and 1322 European descent participants, which we used for genotype imputation. Imputation was performed using the Michigan Imputation Server with the HRC reference panel[16]. After post-imputation quality control, 5,398,505 SNPs remained available for subsequent analyses.

GWAS: GWAS was performed using the PLINK program (http://zzz.bwh.harvard.edu/plink/) with the additive genetic model. The PCA eigenvectors were included as covariates. Post GWAS, FUMA (http://fuma.ctglab.nl/) was used to visualize genome-wide associations and regional associations[17]. Gene-based analyses were run by MAGMA within FUMA, and the significance level was defined as 2.671*10^− 06.

Tests of pleiotropy and concordance

We used SNP effect concordance analysis (SECA) for pleiotropy and concordance analysis [18]. Independent SNPs were identified by clumping using the PLINK program. Pleiotropy analyses were performed by exact binomial test, and Fisher’s exact tests were employed for concordance analysis. P-values are calculated via permutation. SNPs reaching genome-wide suggestive significance (P1 ≤ 10^− 5) in inflammatory biomarkers or Aβ42/Aβ40 GWAS and surpassing nominally associated significance (P2 ≤ 0.05) in brain structures GWAS were defined as SNPs with shared genetic effects. We also examined whether conditioning on inflammatory biomarkers or Aβ42/Aβ40 could be used to detect novel variants associated with brain structures. Conditional false discovery rate (FDR) results served as outputs for brain structure GWAS p-values (P2) stratified on inflammatory biomarkers or Aβ42/Aβ40 GWAS p-values (P1 ≤ {1.0, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.05, 0.01, 0.001, 0.0001, 0.00001}).

LD score regression analysis

To replicate SECA significant findings, LD score regression (LDSR) was used as an alternative method to estimate the genetic correlation between two phenotypes (Bonferroni p = 0.0042) [19].

Epistasis analysis

To test whether amyloid-beta could interact with inflammatory biomarkers to affect brain structures, SNP–SNP epistatic interactions were investigated using the PLINK program. In this epistasis analysis, we included genome-wide significant SNPs from the GWAS on Aβ42/Aβ40 12-month change and inflammatory biomarker 12-month and significant SNPs through SECA above (P = 0.05).

Correlation between brain structure changes, changes in neuroinflammatory biomarker levels, and cognitive decline over 12 months

Among all individuals with an average age of 73.9 years, MMSE significantly decreased over the 12-month follow-up period (P < 2.2e-16). This significant decrease was also observed when focusing only on the AD and MCI groups (P < 2.2e-16 and P = 5.183e-13). The average MMSE decline in the AD group (2.47) was faster than that observed in the MCI group (0.67) over 12 months. We did not find a significant MMSE decline in the CN group only (P = 0.1732).

We analyzed baseline and 12-month data for participants to test for association between cognitive decline (as measured by MMSE changes) and alterations in brain structures. We found longitudinal alterations for 14 brain structures to be significantly correlated with changes in MMSE scores over the 12-month follow-up period (see Table 1). These brain structures included the ventricles, temporal lobe, cingulate gyrus, insula, and corpus callosum.

Table 1

Correlation between 12-month change in brain structure and cognitive decline
12-months brain structure changes	12-months MMSE changes
12-months brain structure changes	R	P-value
Corpus callosum	0.1914801	1.54E-04**
Left inferior lateral ventricle	-0.3564875	5.21E-13**
Left insula	0.259484	2.34E-07**
Left lateral ventricle	-0.398429	4.44E-16**
Left posterior cingulate	0.2596539	2.29E-07**
Left temporal lobe	0.2816901	1.79E-08**
Right bankssts	0.192511	1.41E-04**
Right inferior lateral ventricle	-0.3421819	4.82E-12**
Right insula	0.2240743	8.80E-06**
Right lateral ventricle	-0.4068034	0.00E + 00**
Right para-hippocampal	0.1897251	1.77E-04**
Right temporal lobe	0.239267	1.99E-06**
Third ventricle	-0.2699022	7.21E-08**
WMH (White Matter Hyperintensities)	-0.2951461	3.38E-09**
Note: Nominally significant (p < 0.05); *Adjusted Significant based on Bonferroni correction (P < 0.0004)

We performed pairwise correlation analysis for cognitive decline (as measured by MMSE changes) and changes in levels of plasma inflammatory biomarkers over 12 months. We found changes in the levels of 18 inflammatory biomarkers in plasma to be significantly correlated with changes in MMSE scores over a 12-month period (see Table 2).

Table 2

Correlation between 12-month change in inflammatory biomarkers and cognitive decline
12-months inflammatory biomarkers changes	12-months MMSE changes
12-months inflammatory biomarkers changes	R	P-value
CD40L	-0.12447622	0.0122820830*
EGF	-0.11267933	0.0235110262*
EGFR	0.14092212	0.0045411686*
GH	-0.09986113	0.0448582464*
ICAM1	-0.10880476	0.0287667676*
IL13	-0.15482585	0.0018014050*
IL16	-0.13136531	0.0082006635*
IL18	-0.11601439	0.0196735462*
IL8	-0.15883339	0.0013601540*
MCP1	-0.11953830	0.0162225071*
MCP2	-0.11959212	0.0161741965*
MDC	-0.11096859	0.0257193907*
MIP1β	-0.15360962	0.0019592224*
NGAL	-0.12497164	0.0119377619*
SCF	-0.11536081	0.0203794662*
SOD1	-0.10161074	0.0412201779*
Sortilin	-0.16857804	0.0006683145*
TNF-a	-0.09885969	0.0470594145*

CD40 Ligand (CD40L), Epidermal Growth Factor (EGF), Epidermal Growth Factor Receptor (EGFR), Growth Hormone (GH), Intercellular Adhesion Molecule 1 (ICAM1), Interleukin-13 (IL13), Interleukin-16 (IL16), Interleukin-18(IL18), Interleukin-8 (IL8), Monocyte Chemotactic Protein 1 (MCP1), Monocyte Chemotactic Protein2 (MCP2), Macrophage-Derived Chemokine (MDC), Macrophage Inflammatory Protein-1 beta (M

Genome-wide association studies of cognitive-decline-correlated changes in neuroimaging endpoints and inflammatory biomarker levels, and Aβ42/Aβ40 ratio

First, we investigated the genetic background of 12-month changes in neuroimaging endpoints as evidence of cognitive deterioration. As outlined above, we found that, over the 12-month follow-up period, changes in the structures of 14 brain regions were correlated with cognitive decline. We proceeded to interrogate whether specific genetic regions might be associated with brain structure changes for these 14 regions via GWAS. We identified nine brain regions with at least one genome-wide significant SNP underlying the brain structure change over the 12-month follow-up period (Table 3 and Supplementary Fig S1). We found that six brain regions, including four ventricles and the left and right temporal lobes, shared genome-wide significant hits in the APOE, APOC1, and TOMM40 (marginal significance) regions on chromosome 19 in the gene-based analyses. The WMH 12-month change GWAS identified SNPs in three loci that surpassed the genome-wide significance threshold. The ZFYVE20 gene was found to be significantly associated with WMH changes in the gene-based GWAS.

Table 3

Genome-wide association study of 12-month changes in neuroimaging that correlate with cognitive decline
SNP ID	Brain regions	CHR (POS)	A1	P	BETA	Nearest Gene Position (dis)
rs10225527	Corpus callosum	7(132320548)	T	3.63E-08	-5.416	PLXNA4/Intronic
rs12332072	WMH	4(65448932)	C	6.12E-10	416	RP11-63H19.1/ Intergenic (23203)
rs12721051	Left lateral ventricle	19(45422160)	C	4.75E-09	496.7	APOC1/Intronic
rs12721051	Right inferior lateral ventricle	19(45422160)	C	2.76E-09	66.9	APOC1/Intronic
rs141577243	Left posterior cingulate	11(14612726)	G	2.77E-09	-66.29	PSMA1/Intronic
rs146128209	Left posterior cingulate	11(14683683)	A	4.00E-08	-63.91	PDE3B/Intronic
rs429358	Left inferior lateral ventricle	19(45411941)	T	5.28E-09	66.66	APOE/Exonic
	Left lateral ventricle	19(45411941)	T	6.24E-09	486.8	APOE/Exonic
	Right inferior lateral ventricle	19(45411941)	T	9.76E-09	64.75	APOE/Exonic
	Right lateral ventricle	19(45411941)	T	1.02E-08	455.4	APOE/Exonic
	Left temporal lobe	19(45411941)	T	3.90E-08	-434.3	APOE/Exonic
rs4420638	Left inferior lateral ventricle	19(45422946)	A	2.71E-09	69.68	APOC1/Downstream (339)
	Left temporal lobe	19(45422946)	A	7.49E-10	-485.4	APOC1/Downstream (339)
	Right temporal lobe	19(45422946)	A	6.33E-10	-472.6	APOC1/Downstream (339)
rs56131196	Right lateral ventricle	19(45422846)	G	5.43E-09	475.2	APOC1/Intronic
rs61965256	WMH	13(75083849)	G	1.75E-08	1000	LINC00347/ Intergenic (34125)
rs6855535	Left inferior lateral ventricle	4(103236295)	G	9.26E-09	81.83	SLC39A8/Intronic
rs6989456	WMH	8(123969957)	C	1.90E-08	377.7	ZHX2/intronic
SNP, single-nucleotide polymorphism; A1, minor allele; CHR, chromosome; Significant genome-wide association P < 5 × 10 − 8.

Next, having identified 18 inflammatory biomarkers that also correlated with cognitive decline over a 12-month period, we proceeded to investigate the genetic background of these changes via GWAS (Table 4 and Supplementary Fig S2). GWAS for changes in the levels of CD40L, EGF, and NGAL identified at least one associated SNP that reached genome-wide significance. In the GWAS for CD40L, chromosome 4q SNPs reached genome-wide significance. The EGF GWAS revealed that SNPs in the chromosome 10q region surpassed the genome-wide significance level of association. Significant SNPs in the EGF-change GWAS can be mapped to the SORCS3 gene. SNPs in chromosome 2q exceeded genome-wide significance in the NGAL GWAS. Gene-based analyses did not reveal any additional genome-wide significant associations.

Table 4

Genome-wide association study of 12-month changes in inflammatory biomarkers and A42/A40 that correlate with cognitive decline
ββ
SNP ID	Phenotype	CHR (POS)	A1	P	BETA	Nearest Gene Position (dis)
rs1913012	CD40L	4 (132688307)	A	2.48E-08	0.2024	SNORA70/Upstream (987)
rs11192350	EGF	10 (106931314)	T	3.81E-09	0.3946	SORCS3/Intronic
rs77419187	NGAL	2 (235573908)	A	2.32E-08	0.1559	ARL4C/intergenic (168210)
rs11651351	Aβ42/Aβ40	17 (73966862)	T	1.57E-09	-0.05582	ACOX1/ Intronic
rs1538785	Aβ42/Aβ40	10 (128804888)	T	2.38E-08	-0.03828	DOCK/ ncRNA_intronic
rs685960	Aβ42/Aβ40	1 (239924486)	G	3.76E-08	-0.03891	CHRM3/ Intronic
rs113874038	Aβ42/Aβ40	2 (42076624)	A	3.82E-08	-0.05998	AC009413.2/ downstream (817)
SNP, single-nucleotide polymorphism; A1, minor allele; CHR, chromosome; Significant genome-wide association P < 5 × 10 − 8.

Our GWAS for changes in the Aβ42/Aβ40 ratio over the 12-month period identified seven genome-wide significant SNPs in four loci (Table 4 and Fig S3). Four SNPs at the DOCK1 locus on chromosome 10q surpassed genome-wide significance. One SNP on chromosome 17q (at the ACOX1 locus), 1q (at the CHRM3 locus), and 2q (inter-genic region) exceeded the genome-wide significance threshold. Gene-based analyses did not reveal any additional genome-wide hits.

Testing for pleiotropy and concordance of genetic variants that underlie brain structure change vs inflammatory biomarkers and Aβ42/Aβ40 ratio

Next, we investigated whether there was evidence for pleiotropy and/or concordance among the genetic loci that we found associated with inflammatory and neuroimaging biomarkers and cognitive decline. Evidence of significant global pleiotropy was found between the genetic loci revealed by the GWAS for CD40L and those of the left inferior lateral ventricle GWAS (P = 0.037), CD40L and right inferior lateral ventricle (P = 0.03), CD40L and WMH (P = 0.002), EGF and Left Inferior lateral ventricle (P = 0.027), EGF and right inferior lateral ventricle (P = 0.04) EGF and WMH (P = 0.000999) and NGAL and left temporal lobe (P = 0. 019). More detailed results can be seen in Table 5, Supplementary Table S1 and Fig S4. We also found all paired inflammatory biomarkers and brain regions with significant global pleiotropy also have significant evidence of positive concordance (P < 0.001). Rs10206871 was identified to confer significant shared genetic risk on both inferior lateral ventricle and EGF and can be mapped to the enhancer of SERTAD2. We also found two SNPs (rs12071618, rs1403044519) have shared genetic effects on WMH and EGF; s12071618 is located in the LRIF1 gene on chromosome 1.

Table 5

Pleiotropy and Concordance results for Inflammatory biomarkers and brain regions.
Inflammatory biomarkers	Brain regions	Pleiotropy P-value	Pleiotropy CI	Concordance P-value	Concordance CI	Direction
CD40L	Left inferior lateral ventricle	0.037*	0.0269–0.0505	0.000999*	5.12e-05-0.00564	+
	Right inferior lateral ventricle	0.03*	0.0211–0.0425	0.000999*	5.12e-05-0.00564	+
	WMH	0.002*	0.0186–0.039	0.000999*	5.12e-05-0.00564	+
EGF	Left inferior lateral ventricle	0.027*	0.0186–0.039	0.000999*	5.12e-05-0.00564	+
	Right inferior lateral ventricle	0.04*	0.0295–0.054	0.000999*	5.12e-05-0.00564	+
	WMH	0.000999*	5.12e-05-0.00564	0.000999*	5.12e-05-0.00564	+
NGAL	Left temporal lobe	0.019*	0.0122–0.0295	0.000999*	5.12e-05-0.00564	+
Note: *significant permutation P value < 0.05

The significant associations between inflammatory biomarkers and brain regions were replicated by LDSR analysis. The associations between CD40L, EGF, and WMH were nominally significant. Other paired inflammatory biomarkers and brain regions reached adjusted significance based on the Bonferroni correction (Supplementary Table S2).

Furthermore, we observed significant global pleiotropy of variants affecting the Aβ42/Aβ40 ratio and the left and right inferior lateral ventricles (P = 0.016, P = 0.000999, respectively), left insula (P = 0.002), corpus callosum (P = 0.000999), and left posterior cingulate (P = 0.000999). Among these brain regions, the two sides of the inferior lateral ventricles also have significant evidence of positive concordance with the Aβ42/Aβ40 ratio (P < 0.001, Supplementary Table S2). No single variant was identified with a significant shared genetic effect on the inferior lateral ventricles and Aβ42/Aβ40 ratio. Only the association between Aβ42/Aβ40 and the right inferior lateral ventricle was confirmed by LDSR (P = 0.049).

Conditional FDR to detect additional genetic loci for brain structure changes

We performed conditional FDR analyses to detect variants associated with brain structures after conditioning on inflammatory biomarkers or Aβ42/Aβ40 (Table 6). Five SNPs, including rs10206871, rs1313924, rs72806398, rs6001426, and rs12466903, were found to be significantly associated with the left inferior lateral ventricle when conditioning on the EGF or CD40L variants. We also revealed novel significant SNPs for WMH when conditioning on EGF or CD40L (Supplementary Table S3); rs12466903 was significant in both tests. When conditioning for the variants of the Aβ ratio, rs72806398 and rs72806398 were identified as associated to changes in the inferior lateral ventricle.

Table 6

Conditional FDR results to detect brain structure changes risk variants
Conditional traits	Brain regions	SNP	CHR (POS)	A1	Nearest gene	P-value in Trait	P-value in Brain	FDR q value
EGF	Left inferior lateral ventricle	rs1313924	4 (103375303)	A		0.025760	2.227e-07	0.01581*
		rs72806398	16 (83487874)	A	CDH13	0.213100	2.811e-07	0.01581*
		rs66626994	22 (39488481)	A	LOC101927202	0.394700	1.994e-07	0.01581*
		rs6001426	22 (39488481)	A	APOC1P1	0.363600	7.494e-07	0.03161*
		rs10206871	2 (65018564)	T	SERTAD2 (Enh5590)	8.508e-06	1.884e-03	0.02449*
	WMH	rs12466903	2 (98330626)	C	ZAP70	0.6563000	1.164e-06	0.03927*
CD40L	Left inferior lateral ventricle	rs1313924	4 (103375303)	A		0.1474000	2.227e-07	0.01274*
		rs72806398	16 (83487874)	A	CDH13	0.3865000	2.811e-07	0.01274*
		rs66626994	22 (39488481)	A	LOC101927202	0.3441000	1.994e-07	0.01274*
		rs6001426	22 (39488481)	A	APOC1P1	0.4835000	7.494e-07	0.02547*
		rs10206871	2 (65018564)	T	SERTAD2 (Enh5590)	1.967e-06	1.884e-03	0.02261*
	WMH	rs12466903	2 (98330626)	C	ZAP70	2.930e-01	1.164e-06	0.03270*
Aβ42/Aβ40	Left inferior lateral ventricle	rs72696804	14 (95106902)	C	SERPINA13P	4.454e-02	5.669e-07	0.03138*
	Right inferior lateral ventricle	rs72806398	16 (83487874)	A	CDH13	7.908e-01	2.811e-07	0.02275*
Note: *significant FDR q value < 0.05

Epistasis Analysis

Based on our findings, we hypothesized that Aβ42/Aβ40 acts on brain structure change via inflammation. Epistatic analyses were implemented to test for interactions between the Aβ42/Aβ40 and inflammatory biomarkers. We found two SNP–SNP epistatic interactions between the Aβ42/Aβ40 ratio and EGF: (1) rs11651351 and rs11192347; and (2) rs685960 and rs10206871. rs11651351 and rs685960 were lead SNPs in Aβ42/Aβ40 GWAS (Supplementary Table S4). rs11192347 was from the set of EGF GWAS-significant SNPs, and rs10206871 was a pleiotropy- and concordance-significant SNP. We also found epistatic interactions between a significant SNP from the Aβ42/Aβ40 GWAS (rs113874038) and a significant SNP from the NGAL GWAS (rs77419187).

Motivated by the lack of understanding of how genetic background influences brain structure changes related to progressive cognitive deterioration over time in the presence of AD pathologies, we investigated the genetic relationships between neuroimaging and AD pathology-related biomarkers. We revealed a genetic correlation between cognitive decline, inflammation, and brain structure changes and identified genetic variants that underlie the related paths. Furthermore, we shed some light on how amyloid-beta could indirectly affect brain structures through SNP interactions with inflammation biomarkers.

To avoid the high variability of biomarkers and leverage the ADNI longitudinal study design, we measured all phenotype changes over a 12-month period. We identified neuroimaging biomarkers that are associated with cognitive decline and revealed genetic regions that underlie the observed changes. APOE and APOC1 genes were identified as genetic regions that underlie the paths that lead to the changes of ventricles and temporal lobes with cognitive decline. APOE and APOC1 genes have been regarded as mediators of amyloid-beta pathology to affect cognitive function and brain atrophy[20, 21]. In our study, APOE and APOC1 highlight ventricle changes, where the hippocampus anatomically lies in and could be indirectly reflected[22]. We also found one significant gene, ZFYVE20, in gene-based analysis as underlying WMH changes (also found associated with cognitive decline). ZFYVE20 has been previously reported to be differentially expressed in copper/iron homeostasis of AD brain[23]. Given the fact that inflammation and brain iron can be involved in the formation of WMH, ZFYVE20 may regulate iron homeostasis as a link with WMH, which further contributes to cognitive deterioration[24].

We identified several genetic regions that underlie inflammatory biomarker changes that are associated with cognitive decline. Intriguingly, amyloid-beta is highly involved in the mechanisms of regulating these inflammatory biomarkers. To our knowledge, this is the first GWAS using dynamic changes of blood inflammatory biomarkers as phenotypes to detect associations in the same set of individuals. CD40L, regarded as a primary regulator of the immune system by activating lymphocyte cells and microglia, plays an important role in amyloidogenesis via effects on microglia of converting amyloid-beta into neurotoxic forms [25, 26]. Plasma CD40L is also involved in the progression of cognitive decline in both cross-sectional and longitudinal evaluation[27]. We found a genome-wide significant hit for changes in CD40L levels, upstream of SNORA70. Similarly, amyloid-beta can induce expression of epidermal growth factor receptor and further increase EGF levels to promote activation of microglia [28, 29]. Inhibition of the EGF pathway in both transgenic fruit fly and mouse AD models results in ameliorating the symptoms of memory loss [28]. Rs1192350, located within SORSC3, is associated with plasma EGF changes in our study. SORSC3 from the VPS10 domain family harboring EGF binding repeats can alter amyloid precursor protein (APP) and thus could lead to cognitive decline [30]. We also found a significant GWAS hit for changes in NGAL levels over the 12-month follow-up period. NGAL is an acute phase molecule released by neutrophil granules. Amyloid-beta can regulate and confine NGAL formation in the brain, together with tumor necrosis factor receptor, leading to neuronal death by promoting microglia and generating a toxic environment in AD [31].

We continued to investigate genetic associations between Aβ42/Aβ40 ratio and cognitive decline over the 12-month follow-up period. Our GWAS highlighted the involvement of the CHRM3, DOCK1, and ACOX1 loci. Aβ42/Aβ40 has long been accepted as a biomarker corresponding to brain amyloidosis[32]. SNPs within the CHRM3, DOCK1, and ACOX1 loci have been previously highly related to amyloid-beta load in AD [33–35]. For example, CHRM3 could express differently to affect brain synaptic plasticity and cognitive function due to the appearance of amyloid-beta[34]. Furthermore, amyloid-beta-mediated neuronal death can disrupt the peroxisomal signaling pathway, in which ACOX1 is heavily involved [35]. These SNPs may underlie the paths that lead to changes of Aβ42/Aβ40 that are related to cognitive decline.

We used our GWAS results to explore the pleiotropic effects of our genome-wide significant SNPs. Indeed, we uncovered global evidence of pleiotropy for genetic variants that underlie both enlarged inferior lateral ventricles, as well as CD40L and EGF levels. These results were replicated using LDSR. Our findings highlight the anatomical structures of cognitive dysfunction that also involve inflammation. The inferior lateral ventricle volume, anatomically adjunct to the hippocampus, is a robust measurement to estimate the hippocampus volume [22]. Inflammation has previously been implicated in hippocampus decrease and cognitive decline [36]. Rs10206871, mapped to the enhancer of SERTAD2, was found to have shared genetic effects on both EGF and inferior lateral ventricles. SERTAD2 could enhance E2F transcription factor activity to control both inflammatory and neuronal cell cycling pathways[37]. We also identified SNPs that influence WMH changes over the 12-month follow-up period, as well as CD40L and EGF. Our findings further support the hypothesis of the possible involvement of inflammatory pathways in WMH formation [9]. The variants rs12071618 at the LRIF1 locus have shared genetic effects on the changes of WMH and EGF. LRIF1, like E2F, is involved in cell cycling, but the role of LRIF1 in cognitive dysfunction is still largely unknown [38].

Furthermore, we uncovered pleiotropy and concordance for genetic variants that underlie changes in NGAL and the left temporal lobe. This finding suggests that NGAL plays a crucial role in temporal lobe atrophy in cognitive decline. NGAL in the brain gives rise to neuronal death by promoting microglia to create an inflammation-related toxic environment [31]. Previous studies revealed that NGAL levels in the temporal lobe, hippocampus, and amygdala were significantly higher for AD patients than healthy controls [39–41]. Thus, we can infer that the high level of NGAL in the temporal lobe could cause neuronal death by enhancing the inflammatory response. It is quite interesting to notice that even though there was evidence of genetic risk of CD40L, EGF, and NGAL affecting brain structures linked to cognitive decline, the mechanism or pathway of NGAL seems quite different from the other two biomarkers.

We found global pleiotropy for genetic variants that underlie changes in Aβ42/Aβ40, which also seem to influence multiple brain regions. This is consistent with previous findings [42–45]. Pleiotropy for variants affecting changes in right inferior lateral ventricle volume and Aβ42/Aβ40 was also confirmed through LDSR. Amyloid-beta may act synergistically with other pathologies to affect brain structure changes [3, 46–48]. We also found that changes in the inferior lateral ventricle volume were genetically associated with levels of CD40L, EGF, and Aβ42/Aβ40. It is possible that both amyloid-beta and inflammation pathologies together contribute to the ventricle enlargement and hippocampus decrease.

We found SNPs that influence brain structures that are associated with cognitive decline as well as inflammatory biomarkers and Aβ42/Aβ40. Thus, by conditioning on inflammatory biomarkers or Aβ42/Aβ40, we were able to identify additional genetic variants affecting brain structure changes, which previously were missed. For example, variants within CDH13 and SERPINA13P have been identified in AD studies, but no connection was established with brain structure before [49, 50]. More studies are needed to understand the role of these variants in brain structure changes associated with cognitive decline.

Given that (1) genetic variants of inflammatory biomarkers and amyloid-beta pathology can also affect brain structures, and (2) inflammation often comes into play upon the emergence of amyloid-beta pathology, we speculated that amyloid-beta and inflammation have a synergistic contribution to brain structures. Therefore, we conducted epistatic analyses to check for interactions between SNPs of Aβ42/Aβ40 and inflammatory biomarkers. We found three paired interactions among Aβ42/Aβ40, EGF, and NGAL. Specifically, for the interaction between rs685960 and rs10206871, rs685960 was a significant SNP with shared genetic effects on EGF and the ventricle, whereas rs10206871 was a leading SNP in Aβ42/Aβ40 GWAS. This supports our speculation that amyloid-beta may interact with inflammation to influence brain structure changes that are associated with cognitive decline.

In conclusion, this is the first genome-wide study exploring the interplay between neuroinflammation, Aβ42/Aβ40, and brain structure changes associated with cognitive decline. Our findings point to the interaction of amyloid-beta and inflammation as mediators of brain structure changes and shed light on the underlying genetic background. Future investigations aimed at characterizing the synergistic contribution of amyloid-beta and inflammation at early stages of cognitive impairment could help in the development of novel treatment approaches for AD.

Ethical Approval

NA. We used public dataset, Alzheimer's Disease Neuroimaging Initiative (ADNI), to perform our analysis.

Competing interests

Authors' contributions

Yin Jin wrote the main manuscript text under Peristera Paschou's supervision. Petros Drineas and Chris Rochet provided valuable feedback and contributed to the final version of the manuscript. All authors reviewed the manuscript.

Funding

This work was supported by a Purdue Institute for Drug Discovery grant to PP and CR.

Availability of data and materials

The dataset supporting the conclusions of this article is available in the Alzheimer's Disease Neuroimaging Initiative repository, [https://adni.loni.usc.edu/].

Reitz C. Genetic loci associated with Alzheimer’s disease. Future Neurol. 2014;
Spires-Jones TL, Hyman BT. The Intersection of Amyloid Beta and Tau at Synapses in Alzheimer’s Disease. Neuron. 2014.
Cullen NC, Mälarstig A nders, Stomrud E, Hansson O, Mattsson-Carlgren N. Accelerated inflammatory aging in Alzheimer’s disease and its relation to amyloid, tau, and cognition. Sci Rep. 2021;
Kinney JW, Bemiller SM, Murtishaw AS, Leisgang AM, Salazar AM, Lamb BT. Inflammation as a central mechanism in Alzheimer’s disease. Alzheimer’s Dement. Transl. Res. Clin. Interv. 2018.
Ransohoff RM. How neuroinflammation contributes to neurodegeneration. Science (80-.). 2016.
Kitazawa M, Yamasaki TR, LaFerla FM. Microglia as a potential bridge between the amyloid β-peptide and tau. Ann N Y Acad Sci. 2004.
Ising C, Venegas C, Zhang S, Scheiblich H, Schmidt S V., Vieira-Saecker A, et al. NLRP3 inflammasome activation drives tau pathology. Nature. 2019;
Du AT, Schuff N, Amend D, Laakso MP, Hsu YY, Jagust WJ, et al. Magnetic resonance imaging of the entorhinal cortex and hippocampus in mild cognitive impairment and Alzheimer’s disease. J Neurol Neurosurg Psychiatry. 2001;
Alber J, Alladi S, Bae HJ, Barton DA, Beckett LA, Bell JM, et al. White matter hyperintensities in vascular contributions to cognitive impairment and dementia (VCID): Knowledge gaps and opportunities. Alzheimer’s Dement. Transl. Res. Clin. Interv. 2019.
Parbo P, Ismail R, Hansen K V., Amidi A, Mårup FH, Gottrup H, et al. Brain inflammation accompanies amyloid in the majority of mild cognitive impairment cases due to Alzheimer’s disease. Brain. 2017;
Furney SJ, Simmons A, Breen G, Pedroso I, Lunnon K, Proitsi P, et al. Genome-wide association with MRI atrophy measures as a quantitative trait locus for Alzheimer’s disease. Mol Psychiatry. 2011;
Shen L, Kim S, Risacher SL, Nho K, Swaminathan S, West JD, et al. Whole genome association study of brain-wide imaging phenotypes for identifying quantitative trait loci in MCI and AD: A study of the ADNI cohort. Neuroimage. 2010;
Fischl B. FreeSurfer. Neuroimage. 2012.
Jenkinson M, Beckmann CF, Behrens TEJ, Woolrich MW, Smith SM. FSL - Review. Neuroimage. 2012;
Alexander DH, Lange K. Enhancements to the ADMIXTURE algorithm for individual ancestry estimation. BMC Bioinformatics. 2011;
Das S, Forer L, Schönherr S, Sidore C, Locke AE, Kwong A, et al. Next-generation genotype imputation service and methods. Nat Genet. 2016;
Watanabe K, Taskesen E, Van Bochoven A, Posthuma D. Functional mapping and annotation of genetic associations with FUMA. Nat Commun. 2017;
Nyholt DR. SECA: SNP effect concordance analysis using genome-wide association summary results. Bioinformatics. 2014;
Bulik-Sullivan BK, Loh P-R, Finucane HK, Ripke S, Yang J, Patterson N, et al. LD Score regression distinguishes confounding from polygenicity in genome-wide association studies. Nat Genet. 2015;47:291–5.
Zhou Q, Zhao F, Lv ZP, Zheng CG, Zheng WD, Sun L, et al. Association between APOC1 polymorphism and alzheimer’s disease: A case-control study and meta-analysis. PLoS One. 2014;
Husain MA, Laurent B, Plourde M. APOE and Alzheimer’s Disease: From Lipid Transport to Physiopathology and Therapeutics. Front. Neurosci. 2021.
Nestor SM, Rupsingh R, Borrie M, Smith M, Accomazzi V, Wells JL, et al. Ventricular enlargement as a possible measure of Alzheimer’s disease progression validated using the Alzheimer’s disease neuroimaging initiative database. Brain. 2008;
Myhre O, Utkilen H, Duale N, Brunborg G, Hofer T. Metal dyshomeostasis and inflammation in Alzheimer’s and Parkinson’s diseases: Possible impact of environmental exposures. Oxid. Med. Cell. Longev. 2013.
Yan S, Sun J, Chen Y, Selim M, Lou M. Brain iron deposition in white matter hyperintensities: A 3-T MRI study. Age (Omaha). 2013;
Giunta B, Rezai-Zadeh K, Tan J. Impact of the CD40-CD40L Dyad in Alzheimers Disease. CNS Neurol Disord - Drug Targets. 2012;
Joshi P, Turola E, Ruiz A, Bergami A, Libera DD, Benussi L, et al. Microglia convert aggregated amyloid-β into neurotoxic forms through the shedding of microvesicles. Cell Death Differ. 2014;
Desideri G, Cipollone F, Necozione S, Marini C, Lechiara MC, Taglieri G, et al. Enhanced soluble CD40 ligand and Alzheimer’s disease: Evidence of a possible pathogenetic role. Neurobiol Aging. 2008;
Wang L, Chiang HC, Wu W, Liang B, Xie Z, Yao X, et al. Epidermal growth factor receptor is a preferred target for treating Amyloid-β-induced memory loss. Proc Natl Acad Sci U S A. 2012;
Christiane-Nolte, Kirchhoff F, Kettenmann H. Epidermal growth factor is a motility factor for microglial cells in vitro: Evidence for EGF receptor expression. Eur J Neurosci. 1997;
Eggert S, Thomas C, Kins S, Hermey G. Trafficking in Alzheimer’s Disease: Modulation of APP Transport and Processing by the Transmembrane Proteins LRP1, SorLA, SorCS1c, Sortilin, and Calsyntenin. Mol. Neurobiol. 2018.
Stock AJ, Kasus-Jacobi A, Pereira HA. The role of neutrophil granule proteins in neuroinflammation and Alzheimer’s disease. J. Neuroinflammation. 2018.
Fandos N, Pérez-Grijalba V, Pesini P, Olmos S, Bossa M, Villemagne VL, et al. Plasma amyloid β 42/40 ratios as biomarkers for amyloid β cerebral deposition in cognitively normal individuals. Alzheimer’s Dement Diagnosis, Assess Dis Monit. 2017;
Shi L. Dock protein family in brain development and neurological disease. Commun. Integr. Biol. 2013.
Cummings DM, Liu W, Portelius E, Bayram S, Yasvoina M, Ho SH, et al. First effects of rising amyloid-β in transgenic mouse brain: Synaptic transmission and gene expression. Brain. 2015;
Berger J, Dorninger F, Forss-Petter S, Kunze M. Peroxisomes in brain development and function. Biochim Biophys Acta - Mol Cell Res. 2016;
Simen AA, Bordner KA, Martin MP, Moy LA, Barry LC. Cognitive dysfunction with aging and the role of inflammation. Ther. Adv. Chronic Dis. 2011.
Tuerxun M, Muhda A, Yin L. The molecular mechanisms of signal pathway activating effect of E2F-1/NF-κB/GSK-3β on cognitive dysfunction of Alzheimer rats. Bioengineered. 2021;
Akram S, Yang F, Li J, Adams G, Liu Y, Zhuang X, et al. LRIF1 interacts with HP1α to coordinate accurate chromosome segregation during mitosis. J Mol Cell Biol. 2018;
Naudé PJW, Nyakas C, E. Eiden L, Ait-Ali D, Heide R, Engelborghs S, et al. Lipocalin 2: Novel component of proinflammatory signaling in Alzheimer’s disease. FASEB J. 2012;
Dekens DW, Naudé PJW, Engelborghs S, Vermeiren Y, Van Dam D, Oude Voshaar RC, et al. Neutrophil Gelatinase-Associated Lipocalin and its Receptors in Alzheimer’s Disease (AD) Brain Regions: Differential Findings in AD with and without Depression. J Alzheimer’s Dis. 2017;
Jang E, Lee S, Kim JH, Kim JH, Seo JW, Lee WH, et al. Secreted protein lipocalin-2 promotes microglial M1 polarization. FASEB J. 2013;
Jonkman LE, Steenwijk MD, Boesen N, Rozemuller AJM, Barkhof F, Geurts JJG, et al. Relationship between β-amyloid and structural network topology in decedents without dementia. Neurology. 2020;
Kara F, Höfling C, Roßner S, Schliebs R, der Linden A, Groot HJM, et al. In Vivo Longitudinal Monitoring of Changes in the Corpus Callosum Integrity During Disease Progression in a Mouse Model of Alzheimer’s Disease. Curr Alzheimer Res. 2015;
Westermark P. Amyloid in the islets of Langerhans: Thoughts and some historical aspects. Ups. J. Med. Sci. 2011.
Schmid S, Jungwirth B, Gehlert V, Blobner M, Schneider G, Kratzer S, et al. Intracerebroventricular injection of beta-amyloid in mice is associated with long-term cognitive impairment in the modified hole-board test. Behav Brain Res. 2017;
Kametani F, Hasegawa M. Reconsideration of amyloid hypothesis and tau hypothesis in Alzheimer’s disease. Front. Neurosci. 2018.
Calvo-Flores Guzmán B, Elizabeth Chaffey T, Hansika Palpagama T, Waters S, Boix J, Tate WP, et al. The Interplay Between Beta-Amyloid 1–42 (Aβ1–42)-Induced Hippocampal Inflammatory Response, p-tau, Vascular Pathology, and Their Synergistic Contributions to Neuronal Death and Behavioral Deficits. Front Mol Neurosci. 2020;
Sakai K, Noguchi-Shinohara M, Ikeda T, Hamaguchi T, Ono K, Yamada M. Cerebrospinal fluid cytokines and metalloproteinases in cerebral amyloid angiopathy-related inflammation. Acta Neurol Scand. 2021;
Liu FF, Zhang Z, Chen W, Gu HY, Yan QJ. Regulatory mechanism of microRNA-377 on CDH13 expression in the cell model of Alzheimer’s disease. Eur Rev Med Pharmacol Sci. 2018;
Grupe A, Abraham R, Li Y, Rowland C, Hollingworth P, Morgan A, et al. Evidence for novel susceptibility genes for late-onset Alzheimer’s disease from a genome-wide association study of putative functional variants. Hum Mol Genet. 2007;

No competing interests reported.

Supplementary.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Interaction of Genetics, Amyloid-beta and Inflammation as Mediators of Brain Structure Change in Cognitive Decline

Status:

Version 1

Abstract

Figures

Introducation

Materials And Methods

Dataset

Neuroimaging analyses

Genetic analyses

Results

Conditional FDR to detect additional genetic loci for brain structure changes

Epistasis Analysis

Discussion

Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1