The TMEM106B rs1990621 protective variant is also associated with increased neuronal proportion

Transcript

1 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. is also rs1990621 TMEM106B The associated with protective variant increased neuronal proportion 1,2 1,2 1,2 ,2 1 , Fabiana G. Farias Kathie A. , Jorge L. Del - , Umber Dube Aguila Zeran Li , 1,2 1,2 1,2 1,2 , Fengxian Maria Victoria Fernandez Mihindukulasuriya , Laura Ibanez , John P. Budde , 4 1,2 1,2 1,2 3 , Joseph , Chengran Yang , Allison M. Lake Wang , Yuetiva Deming , James Perez 1,2 1 ,2 1,2 1,2 , , Kristy Bergmann Joseph D. Bruno A. Benitez , Richard Davenport , Bradley 1 1,2 , ∏ 1,2 , Carlos Cruchaga , Oscar Harari Dougherty 1. Washington University School of Medicine, St Louis, MO, USA 2. al Disorders, St Louis, MO, USA Hope Center for Neurologic 3 TN, USA Vanderbilt University Medical Scientist Training Program, Nashville, . University of Wisconsin - Madison, Madison, WI, USA. 4. ∏ To whom correspondence should be addressed: Carlos Cruchaga, PhD fessor Pro Department of Psychiatry The Hope Center Program on Protein Aggregation and Neurodegeneration ity, School of Medicine Washington Univers 425 S. Euclid Ave. BJC Institute of Heath. Box 8134 St. Louis, MO 63110 - 286 0546 Tel: 314 - 2244 Fax: 314 - 362 - Email: [email protected] 1

2 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. Abstract studies, observed decreased neuronal and increased astrocyte Background: In previous we RNA in parietal brain cortex by using a deconvolution method for in AD cases bulk - proportions . These findings suggested that genetic risk factors associated with AD etiology have a seq specific effect in the cellular composition of AD brains. The goal of this study is to investigate if there are genetic determinants for brain cell compositions. Using cell type composition inferred from transcriptome as a disease status proxy, we Methods: pe association analysis to identify novel loci related to cellular population performed cell ty in total changes in disease cohort. We imputed and merged genotyping data from seven studies a. We of 1,669 samples and derived major CNS cell type proportions from cortical RNAseq dat also inferred RNA transcript integrity number (TIN) to account for RNA quality variances. The neuronal proportion ~ SNP + Age + Gender model we performed in the analysis was: normalized + PC1 + PC2 + median TIN. in the TMEM106B Results: region was significantly A variant rs1990621 located gene 07 - (p= ) and replicated in an independent dataset . The 6.40 × 10 associated with neuronal proportion association became more significant as we combined both discovery and replication datasets in - 09 - 10 This variant is in meta p= ) and joint analysis (p= 7.66 × 10 9.42 × ). analysis ( - multi 10 - tissue ) which was previously identified as a protective variant in 8 high LD with rs1990622 (r2 = 0.9 associated with in d FTD cohorts. Further analyses indicate that this variant is creased neuronal only in AD cohort but also in proportion in participants with neurodegenerative disorders , no t cohort was not observed in a younger cognitive normal elderly . However, this effect The second most significant loci for neuron schizophrenia cohort with a mean age of death < 65. 2

3 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. that ed al proportion as an informative using , which suggest APOE proportion was neuron endophenotype could help identify loci associated with neurodegeneration. shared by aging Conclusion: This result suggested a common pathway i nvolving TMEM106B groups in the present or absence of neurodegenerative pathology may contribute to cognitive preservation and neuronal protection. Introduction Although neuronal loss and synapse dysfunction are the preceding events of cogn itive (AD), neurons do not work or survive by themselves. These Alzheimer’s disease deficits in require supports through intimate collaborations within themselves and with delicate organelles 1 other cell types . The microenvironment of cellular crosstalk, interaction, balance, and circuits maintained by neurons, astrocytes, microglia, oligodendrocytes, and other vascular cells are essential for the brain to carry out functions and fight against insults. ciated risk factors identified across the genome also point to the involvements of AD asso 1,2 cell types apart from neurons multi . APOE4 is related to lipid metabolism and mostly - 3 ABCA7 . Other lipid metabolism related risk genes are expressed in astrocyte and microglia 2,4 2,5,6 , , and identified in all cell types CLU in astrocyte and oligodendrocyte precursor cells 2 . Research interests in the roles of inflammatory response to toxic stimuli or SORL1 in astrocyte iated with immune microbial infection have been escalating recently, and AD risk genes assoc 9 2 7 9 - 2,4 9 2,6 - CR1 CD33 ABI3 , HLA - DRB5 – HLA , DRB1 , , and , , response including TREM2 PLCG2 2 INPP5D expressed in microglia, are mostly expressed in microglia and macrophages. BIN1 2,5 2 ICALM expressed in microglia and endothelial cells are P , and oligodendrocyte, and neurons associated with endocytosis. 3

4 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. In a normal functional brain, astrocytes, microglia, and oligodendrocytes provide trophic supports to neurons and various cell type specific functions. Astrocytes confer multiple functions 10 to fulfill neurons’ metabolic needs including but not limited to providing substrates for 11 oxidative phosphorylation , exerting regulation of excitatory C NS neurotransmitter 12,13 glutamate , and serving as bidirectional communication nodes that talk to both neurons and blood vessels and modulate their activities in an arrangement of functional e ntities named 16 - 14 neurovascular units . Microglia surveil in the extracellular space and look for pathogens or debris to engulf through phag ocytosis. Oligodendrocyte provides insulation to neurons by wrapping around the axons with myelin sheath. However, in an AD diseased brain, these edged swords that play beneficial and/or harmful roles as supporting cells may become double - disease progresse accumulation and clearance are the central events of the amyloid β - s. Amyloid cascade hypothesis. Both astrocyte and microglia have been involved in response to the toxic 17 21 - 19 19 - accumulate stimuli of amyloid plaques. During the early stage, microglia and astrocytes around plaques to phagocytose or degrade those in a protectiv e manner. However, as disease progresses, the chronic and prolonged activation of microglia and astrocytes will be provoked - into a damaging pro inflammatory state and a vicious circle that exacerbate pathology in a harmful manner. Evidence suggested that i ncreased inflammatory cytokine secretion in microglia, and increased production of complement cascade components, and impaired 13 glutamate regulation (unregulated glutamate activity can cause neuronal excitatory cell death) ynaptic loss which ultimately leads to cognitive deficits. Disrupted neuronal may contribute to s plasticity due to myelin loss and dysfunctional neurovascular units further exacerbate the dreadful situation and destroy the harmony of the multi - cell type microenvironment. 4

5 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. Ap art from disturbed homeostatic processes and impaired circuits integrity, cell type composition or proportion is also altered. Brains affected by AD exhibits neuronal loss, that oligodendrocyte loss, astrocytosis, and microgliosis. However, the specific effects pathological mutations and risk variants have on brain cellular composition are often ignored. To investigate the changes of cerebral cortex cell - type population structure and account for the deconvolution associated confounding effects in downstream analysis, we developed an in - silico method to infer cellular composition from RNA - Seq data, which has been documented in our 22 . In summary, we firstly assembled a reference panel to model the previous publication transcriptomic signature of neurons, astrocytes, oligodendrocytes and microglia. The panel was created by analyzing expression data from purified cell lines. We evaluated various digital onvolution methods and selected the best performing ones for our primary analyses. We dec tested the digital deconvolution accuracy on induced pluripotent stem cell (iPSC) derived rification neurons and microglia, and neurons derived from Translating Ribosome Affinity Pu followed by RNA - Seq. Finally, we verified its accuracy with simulated admixture with pre - defined cellular proportions. Once the deconvolution approach was optimized, we calculated the cell proportion in AD n regions of LOAD and ADAD datasets. We found that cases and controls from different brai neuronal and astrocyte relative proportions differ between healthy and diseased brains, and also differ among AD cases that carry different genetic risk variants. Brain carriers of pathogenic presented lower neuronal and higher astrocytes relative PSEN1 or , APP mutations in PSEN2 proportions compared to sporadic AD. Similarly, APOE ε 4 carriers also showed decreased carriers. In contr ast, - neuronal and increased astrocyte relative proportions compared to AD non carriers of variants in TREM2 risk showed a lower degree of neuronal loss than matched AD 5

6 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. cases in multiple independent studies. These findings suggest that different genetic risk factors he cellular composition of AD associated with AD etiology may have gene specific effects in t brains. 23 similar deconvolution In a recently published study named PsychENCODE , a very 22 was taken to infer cell type composition from RNA in our previous study approach as reported - Seq data predominantly drawn from psychiatric disorder cohorts. From the cell fractions inferred Seq data, they found that cell type composition differences can account for more from bulk RNA - than 88% of bulk tissue expression variation observed across the population with a ±4% variance - on a per subject level. Using c ell type compositions as quantitative traits, the authors identified a gene that is associated with both FZD9 - non coding variant closed to the FZD9 gene expression 23 . Interestingly, deletion variants found and the proportion of excitatory layer 3 neurons previously upstream of FZD9 were associated with cell composition changes in Williams 24 , a developmental disorder exhibits mild to moderate intellectual disabilities with syndrome scular problems. This observation re learning deficits and cardiova - emphasized the importance of incorporating cell type composition into RNA - Seq analysis pipeline even in psychiatric disorder cohorts without dramatic changes in cellular composition, not mention the necessity of such actice in neurodegeneration disorders that have significant changes in cell type composition. It pr also demonstrated the great potential of using relative abundance of specific cell types in However , it is unclear if this finding identifying novel variants and genes implicated in disease. more general finding. relate traits or it is - is only applicable to psychiatry a 22 method In this study, we utilized cell - type proportions inferred from our deconvolution search for potential new in dataset enriched for AD cases to perform cell type QTL analysis in a loci that are associated with neurodegeneration disorders. We imputed and merged genotyping or 6

7 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. ( five centered on neurodegeneration - whole genome sequencing data from seven studies N = 1,125 414 ( N = 130 ). From , and GTEx multiple tissue controls ) ) , one schizophrenia cohort ( N = Seq data, we derived cell fractions of four major CNS cell types, including neuron, cortical RNA - astrocyte, microglia, and oligodendrocyte. Using normalized neuronal proportion as quantitative gene region significantly TMEM106B trait, we identified a variant rs1990621 located in the n in all cohorts except schizophrenia subjects. This associated with neuronal proportion variatio 2 variant is in high LD with rs1990622 (r = 0.9 8 ), which was previously identified as a protective 25 variant in FTD cohorts . Variants in this region have also been found to be associated with AD 26 27 TMEM106B is observed in AD brains pathology . In 43 , and downregulation of with TDP - conclusion, we have identified a variant associated with neuronal proportion with potential protective effect in neurodegeneration disorders. Meth ods Study participants The participants were sourced from seven studies with a total sample size of 1,669 Table 1 ( ). Among those, five studies are mainly focused on neurodegenerative disorders N = 681 ) , frontotemporal dementi a ( N = 11 ) , progressive including Alzheimer’s disease ( 82 N = aging ( N = 29 ) , Parkinson Disease ( N = 1 ), as well as ( supranuclear palsy , pathological ) ) N = These samples come from the Mayo, MSSM, Knight cognitive normal individuals ( 540 . . To compare with the Table studies as described in ADRC, DIAN, and ROSMAP 1 ) and bipolar disorders ( N neurodegenerative disorders, we also included schizophrenia ( N = 210 t Table 1 ) . Additionally , ( wo studies, MSSM participants from the CommonMind study ) 34 = 7

8 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. lude some participants contribute more than one tissue data that inc and GTEx, contain multi - Table 1 tissue ) . ( Standard protocol approvals, registrations and patient consents ADRC studies have been approved by the review The protocol of DIAN and Knight - board of Washington University in St. Louis. The protocol of Mayo dataset was approved by the Mayo Clinic Institutional Review Board (IRB). All neuropsychological, diagnostic and autopsy protocols of MSSM dataset were approved by the Mount Sinai and JJ Peters VA Medical Center oards. The religious orders study and the memory and aging project of Institutional Review B ROSMAP was approved by the IRB of Rush University Medical Center. The NIH Common Fund’s GTEx program protocol was reviewed by Chesapeake Research Review Inc., Roswell te’s Office of Research Subject Protection, and the institutional review board Park Cancer Institu of the University of Pennsylvania. Within CommonMind consortium, the MSSM sample protocol was approved by Icahn School of Medicine at Mount Sinai IRB; the Pitt sample as approved by the University of Pittsburgh’s Committee for the Oversight of protocol w Research involving the Dead and IRB for Biomedical Research; the Penn sample protocol was approved by the Committee on Studies Involving Human Beings of the University of Pennsylv ania. All participants were recruited with informed consent for research use. Data collection and generation ortal brains were collected ( m ). - Cortical tissues from various locations of post Table 2 RNA was extracted from lysed tissues and prepared into libr aries of template molecules ready 90% of total for subsequent next - generation sequencing steps. Ribosomal RNAs constitute 80% - RNAs, which are not the targets of this study. To focus on mRNA quantification usually 8

9 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. Seq library - RNAs or enrich for mRNAs during RNA researchers would either remove excessive r 22 29 28 22 took a rRNA preparation. In this study, DIAN , and CommonMind , MSSM , Knight ADRC depletion approach to removed ribosomal RNA from total RNAs to retain a higher mRNA 31,32 33,34 30 y - , ROSMAP content. Whereas, Mayo , and GTEx A enrichment approach to took a pol enrich mRNAs from total RNAs. Genotype information were also collected and sequenced correspondingly. RNAseq paired with genotype data for each participant were either sequenced tudies or downloaded from public - at Washington University for DIAN and Knight ADRC s Table 2 and each study reference(s) for more data database for all the other studies. Please see collection and generation specifications. Data QC and preprocessing Genetic Data plied to each genotyping array or sequence Stringent quality control (QC) steps were ap data. The minimum call rate for single nucleotide polymorphisms (SNPs) and individuals was - 0 6 < 1 × 10 98% and autosomal SNPs not in Hardy - Weinberg equilibrium ( ) were p - value d to verify gender identification. Unanticipated chromosome SNPs were analyze - excluded. X 0.25) among samples were tested by pairwise ≥ duplicates and cryptic relatedness (Pihat 35 - by was used to calculate genome descent. EIGENSTRAT - wide estimates of proportion identity - principal components. The 1000 Genomes Project Phase 3 data (Oc tober 2014), SHAPEIT 36 37 were used for phasing and imputation. Individual genotypes v2.r837 , and IMPUTE2 v2.3.2 imputed with probability < 0.90 were set to missing and imputed genotypes with probability ≥ alyzed as fully observed. Genotyped and imputed variants with MAF < 0.02 or 0.90 were an WGS data quality is censored by filtering IMPUTE2 information score < 0.30 were excluded. 9

10 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. es followed by similar QC approach out reads with sequencing depth DP < 6 and quality GQ < 20 as above for genotyping data. After the QC, all studies including imputed genotype described and WGS data was merged into a binary file using Plink for downstream analysis. PCA and IBD and analyses were performed on the merged binary files using Plink to ke ep European ancestry ). unrelated participants ( Figure 1 and Figure 2 Expression Data 38 FastQC was applied to RNAseq data to examine various aspects of sequencing quality . Outlier samples with high rRNA contents or low sequencing depth were removed from the pool. The remaining samples were aligned to human GRCh37 primary ass - embly using Star with 2 39 Pass Basic mode (ver 2.5.4b) . Alignment metrics were ascertained by applying Picard 40 ncluding reads bias, coverage, ribosomal contents, coding bases, and i CollectRnaSeqMetrics tegrity number (TIN) for each transcript was calculated on etc. Following which, transcript in 41 aligned bam files using RSeQC tin.py (ver 2.6.5). RNAseq coding gene and transcript expression was quantified using Salmon transcript expression quantification (ver 0.7.2) with 42 enome . GENCODE Homo sapiens GRCh37.75 reference g inferred from RNAseq gene Four major central nerve system cell type proportions were 22 . To briefly expression quantification output as documented in our previous deconvolution study explain the deconvolution process, we firstly assembled a reference panel to model the transcriptomic signature of neurons, astrocytes, oligodendrocytes and microglia from purified single cel - l tissue sources respectively. Using the reference panel and the method population 43 44 (PSEA, also named meanProfile in CellMix implementation specific expression analysis ), we calculated four cell type proportions for each subject bulk RNAseq data. For each brain tissue 10

11 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. collection site of each study, outlier values for each cell type proportion were removed. Mean ype of each tissue in each study were subtracted from the deconvolution values for each cell t 3 ). Phenotype information from all results to center all the distributions to zero mean ( Figure studies were merged and unified to the same coding paradigm to enable downstream joint analysis; for example, males are all coded as 1 and females are 2. Data analysis For the discovery phase, ROSMAP dataset was analyzed with linear regression model 45 employed in Plink using normalized neuronal proportion to run quantitative trait analysis. Age, c, sex, PC1, PC2, and median TIN were used as covariates to account fo r potential geneti sequencing results TIN is calculated directly from post phenotypic or technical heterogeneity. - 41 that captures RNA degradation by measuring mRNA integrity directly Results were depicted . 46 as Manhattan plots using R (ver 3.4.3) qqman package (ver 0.1.4). For the replication phase, all the other studies except ROSMAP were combined and - tissue QTL analysis because MSSM and GTEx contain samples with prepressed to run meta Tissue software installation and data preprocessing were - ltiple cortical tissues. Meta mu conducted following a four - step instruction documented in the developer website: 47 http://genetics.cs.ucla.edu/metatissue/install.html . Meta - processing pipeline calls two tissue 47 48 and then followed by Metasoft main functions, firstly MetaTissueMM . MetaTissueMM applies a mixed model to account for the heterogeneity of multiple tissue QTL effects. Metasoft - value for multiple performs the meta - analysis while proving a more accurat e random effect p value based on Bayesian inference to indicate how likely a locus is a - tissue analysis and a m QTL in each tissue. Similarly, results were depicted as Manhattan plots and visually examined. 11

12 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. both discovery and replication studies were combined to For the final merging phase, - tissue analysis by each tissue of each study, a split by maximize sample size. Apart from meta disease status analysis was also performed in the final merging phase. Samples from each tissue of each s tudy were also split into disease categories. Resultant subcategory with less than 20 subjects were removed from the analysis to avoid false results due to too small sample size. ove. Similar data preparation and analysis pipeline were enforced as documented ab 49 to annot ation significant SNPs QTL analysis results were uploaded to Fuma (v1.3.3d) - 06 - 02 06) and ANNOVAR (updated 2017 - 07 - 17). (p - - value < 10 ) with GWAScatalog (e91_r2018 50 implemented in Fuma. - Gene based analysis was also performed by Magma (v1.06) Data availability https://www.synapse.org/#!Synapse:syn5550404 Mayo: https://www.synapse.org/#!Synapse:syn3157743 MSSM: ROSMAP: .org/#!Synapse:syn3219045 https://www.synapse https://www.synapse.org/#!Synapse:syn2759792 CommonMind: https://www.ncbi.nlm.nih.gov/projects/gap/cgi - bin/study.cgi?study_id=phs000424.v7.p2 GTEx: Knight - ADRC: https://www.synapse.org/#!Synapse:syn12181323 http://dian.wustl.edu According to the data request terms, DIAN dat a are available upon request: Results Study Design 12

13 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. The ROSMAP study containing 523 subjects will be the discovery dataset, and the other aset with 1,146 subjects. Altogether, we have six studies are collapsed into replication dat Seq data comprised of 1,669 participants predominantly assembled a set of cortical RNA - , ). Collectively, neurodegenerative Table 1 4 Figure focused on disorders from seven sources ( Mayo, MSSM, Knight ADRC, and ROSMAP studies contributed 664 sporadic AD cases. Apart from sporadic AD, 15 subjects from DIAN study and 2 from Knight - ADRC also harbor PSEN1 , exhibit , and familial AD inheritance patt ern. Other PSEN2 APP mutations that neurodegenerative disorders, including progressive supranuclear palsy (PSP), pathological aging dementia (PA), frontal temporal (FTD), and Parkinson’s Disease (PD), are mainly drawn from Mayo and Knight ADRC datasets. Other psychiatric disorders including schizophrenia and bipolar disorders are contributed by the CommonMind study. Besides, 540 control subjects or neuropsychiatric symptoms were also included. MSSM and dementia cleared of cognitive GTEx also included multiple tissue data, which wer e collected from multiple regions of the same subjects that allow us to perform region specific comparison within the same cohort. was performed in ROSMAP study. In the replication phase, all the analysis Discovery ignals identified from the discovery ROSMAP set. other studies were merged to replicate s Because GTEx and MSSM contain multiple cortical regions collected from the same subjects, 47 specifically designed for multi we also applied meta tissue QTL analysis to - tissue software - perform a mixed model anal ysis with random effects that account for correlated measurements - tissue individuals. To attain the largest available sample size for this study, the from multi discovery and replication sets were merged to perform the merged multi tissue QTL analysis in a - search for additional signals hidden in previously separated discovery or replication analysis due to lack of power. After merged analysis, the cohorts were split into four major disease status 13

14 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. ative disorders) to explore how AD neurodegener - groups (AD, control, schizophrenia, other non different disease strata could impact the results. TMEM106B variants associated with neuronal proportion During discovery phase, ROSMAP dataset (N = 484 after removing outliers from total number of 523 subjects perform cell type proportion QTL analysis. Using ) was used to normalized neuronal proportion as a quantitative trait, the QTL analysis identified more than 10 06 - × Table 3 ). 10 , peaks that passed genome wide suggestive threshold (<1.0 , Figure 5AB 04 - 10 signal rs1990621 (chr7: 12283873) were replicated with p - value = 7.41 × However, only one Figure in the replication dataset (N = 1,052) combining all the other datasets except ROSMAP ( ple size CD 5 ). When the discovery and replication datasets were merged to attained a larger sam (N = 1,536), rs1990621 major allele C is negatively associated with neuronal proportion with p - 09 - ( 6AB , Figure 7AC ), which means the minor allele G is associated 10 × value = 9.42 Figure ocusing on neurodegenerative with increased neuronal proportion in our assembled datasets f disorders. - tissue data were involved that Noticeably, in both replication and merged analyses, multi provided additional power but also posed challenges to the analysis, the same issue faced by the 34,51 GTEx study tissue approach, multiple ti ssues collected from the - . Compared to a tissue - by same subject may help identify QTL by aggregating evidence from multiple tissues, which is analysis of combining each study. However, one violation of such approach is - similar to a meta that the tissues collected from the same subj ect are presumably highly correlated since they shared the same genetic architecture. Thus, it violates the assumption of independency for 47 carrying out a standard meta tissue QTL is the - analysis - . Another challenge of the multi 14

15 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. heterogeneity of the effects, which means a variant may have different effects on different 47 Tissue analytic pipeline tissues. To resolve these issues, we applied the Meta - tissue QTL, the same ( http://genetics.cs.ucla.edu/metatissue/) - specifically designed for multi Tissue - - t issue data. As shown in approach that GTEx took to analyze their multi A , Meta Figure 7 is were displayed as a forest plot with 95% analysis results of rs1990621 for the merged analys value labeled for each tissue of each study. Among them, MSSM and confidence interval and p - - - GTEx are multi - tissue studies while the others are single tissue studies. Meta Tissue used a linear mixed model to captu re the multi - tissue correlation within MSSM and GTEx respectively. 48 - t o predict if a variant has Tissue calculated a m - value Regarding the effect heterogeneity, Meta value is similar to the posterior probability of association based on the - an effect in a tissue. M 48 Bayes factor but with differences specifically designed for detecting whether an effect is 7 Figure is a PM - Plot that integrates evidences analysis. - present in a study included in a meta B eterogeneity of value) sides to interpret the h from both frequentist (p - value) and Bayesian (m - - tissue QTL effects. Variant rs1990621 in ROSMAP and Mayo studies have m - multi values - greater than 0.9, are predicted to have an effect and color coded with red. In CMC study, the m value is less than 0.1, so it is predicted to not to have an effect and color coded with blue. All the - value between 0.1 and 0.9 are predicted with ambiguous effect and color other studies with m - Plot, the variant does have effect coded with green. Based on the forest plot and PM heterogeneity across different ti ssues and studies. In this case, random - effect model will be more suitable to account for effect heterogeneity. Therefore, summary random effect and p - value were reported for the analysis. tissue QTL, a single tissue joint analysis was also performed. In this - Apart from multi - - tissue data to avoid violating the independency case, one tissue region was drawn from the multi 15

16 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. to represent MSSM and assumption. Specifically, BM36 and frontal cortex tissue were selected potential GTEx study respectively. Study coded as dummy variables to account for were sites batch effects . In this joint analysis, the variant rs1990621 is also the top hit with p - value = - 10 × 10 . 7.66 M106B variants observed in neurodegenerative Neuronal protective effect of TME disorders and normal aging participants stratified To explore the effect in different disease categories, the merged dataset was AD neurodegenerative disorders , schizophrenia and control based on disease: AD, other non - . Signification associations between rs1990621 and neuronal proportion were observed in AD (p - - 0 7 - 0 4 - AD neurodegenerative (p - value = 8.19 × 10 10 = 1.95 ), other ), and cognitive non value × 2 0 - normal control (p - ) cohorts , but not in schizo phrenic value = (p - value = 2.94 × 10 cohort 0 - 1 more prominent in , D ) . The effect of the variant wa s 7 9.32 , Figure 10 × Table 4 neurodegenerative cohorts and aging controls with mean age of death greater than 65 years old. However, it absent from younger cohorts such as GTEx controls and CommonMind was schizophrenia participants. Thus, this variant seems to be associated with a neuronal protection mechanism shared by any aging process in the present or absence of neuropathology. nal annotation Functio in the TMEM106B gene region where other variants in The variant rs1990621 is located also and labeled in Figure 8 A . Although the CADD score and are high LD linkage located unctional RegulomeDB score for this variant are not remarkably high to suggest any f 2 consequences ( = 0.9 8 ), a TMEM106B Figure 8 BC ), this variant is in high LD with rs1990622 (r 25 precursor , particularly in granulin variant previous identified to be associated with FTD risk 16

17 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 52,53 is . TMEM106B (GRN) mutation carriers expressed in neurons and microglia, with highest 57 54 - late endosome/lysosome compartments of neurons . A protein expression detected in the 2 ), r with rs1990621 = 0.98 LD located nonsynonymous variant rs3173615 , ( which is also in high two protein isoforms ) produces Figure 8B (the dark blue dot in TMEM106B in the exon 6 of 55,58,59 mechanism (p.T185S) that affect TMEM106B protein level through protein degradation . The impact of other neurodegenerative risk loci on neuronal proportion To investigate what other AD or FTD variants might have an effect in neuronal proportion QTL analysis, we extracted results for 38 SNPs examined in two large scale genome 2 60 ( Ferrari et al . ) ) and FTD focused studies. wide association studies, AD focused ( Lambert et al . gene regions passed genome wide Among those, only variants located in TMEM106B and APOE significant or suggestive threshold. Both rs1990622 Figure 9B ) ( ( Figure 9A and rs2075650 ) , which were associated with were found to be associated with FTD reported in Ferrari et al . ) . The top signals in APOE region are rs283815, neuronal proportion in this study ( Table 5 - 0 5 value < - × 10 . Note that rs429358 is one of the two SNPs with p rs769449, and rs429358 1 .22 Remember that , e 4 allele s coded by rs429358(C) and that determine APOE isoforms. APOE We observed the largest effect for AD risk. t he C allele of rs429358 was rs7412(C), confer s that between rs7412 and association associated with decreased neuronal proportion, but no observed neuronal proportion. (p value = TMEM106B based analysis of our neuronal proportion QTL, - In a gene - 08 - - wide significant threshold followed by APOE 10 × 2.96 ) is the only gene that passed genome 05 - CD 10 the most important gene for sporadic AD risk ( Figure 6 , ). Previous ) × value = 3.2 - (p GWAS for AD risk performed with the International Gen omics of Alzheimer’s Project (IGAP) 17

18 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. genotype showed that AD risk is significantly influenced by the APOE data stratified by 61 TMEM106B APOE . Together with our observation of cellular interaction between and may composition QTL, these results suggest a potential interaction of TMEM106B and APOE play a role in affecting AD risk/vulnerability and cellular composition balance between neurons and astrocytes, and the endosome and lysosome compartments might be the location that the interaction takes place. Discussion The c ommon variant rs1990622 in TMEM106B was first identified to be associated with 25 phosphorylated and ubiquitinated TDP - 43 is the major FTD with TDP - 43 inclusions . Hyper - 62 , which is also present in a broader range of pathological protein for FTD and ALS 63 64 63 , and hippocampal sclerosis . neurodegenerative disorders, including AD , Lewy body disease FTD brains, typical TDP - 43 α - Recent study also suggested distinct TDP - 43 types present in non - 65 type 43 α - type is the typical form conventionally - β type and newly characterized . TDP - type is characterized by its close - observed in temporal, frontal and brainstem regions. TDP - 43 β adjacency to neurofibrillary tangles, which is predominantly observed in limbic system, entorhinal cortex, and subiculum of the hippocampus. These findings including amygdala, variants TMEM106B 43 protein that closely associated with - suggested that pathologic TDP might be the common pathologic substrate linking these neurodegenerative disorders. Multiple are associated lines of evidence have merged and shown that protective variants in TMEM106B 66 with attenuated cognitive deficits or better cognitive performance in ALS , hippocampal 69 67 68 or , presymptomatic FTD sclerosis , and aging groups with various neuropathological burden 70 in the absence of known brain disease ant rs1990621 of . Our study identified a protective vari 18

19 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. is associated with increased neuronal proportion in participants with TMEM106B non neurodegenerative disorders and normal aging - in demented controls. However, this effect is f death less than 65 years old. not observed in a younger schizophrenia cohort with a mean age o shared by aging groups in the TMEM106B This result suggested a common pathway involving present or absence of neurodegenerative pathology that may contribute to cognitive preservation and neuronal protection. TMEM106B demonstrated that a protective variant rs1990621 Our study has identified in protein coding variant may exert neuronal protection function in aging group s. gene region A 2 ( r LD = 0.98 ) produces two protein isoforms (p.T185S) with rs1990621 T h e high in rs3173615 . is degraded faster than the risk S185 allele is protective and the protein carrying this amino acid T185. Thus, the risk allele of this coding variant leads to increased TMEM106B protein variant 55,58,59 overexpression results in enlarged ly sosomes and lysosomal level TMEM106B . 55,71 dysfunction . It has also been shown that TMEM106B may interact with PGRN (the precursor 59 . Although rs3173615 is not included in our genomic data, it is protein for granulin) in lysosome in complete linkage disequilibrium with rs1990621 and rs1990622. It is worth pointing out that the minor phase with - the minor allele of rs1990622, which has a protective effect in FTD, is in allele of rs1990621, which is associated with increased neuronal proportion in our analysis. Despite the fact that our dataset is focused on neurodegeneration, we only have 11 verified FTD role in cases suggesting that TMEM106B might have a general neur onal protection neurodegeneration apart from FTD. a potential involvement of TMEM106B in the suggested observation that This endosome/lysosome pathway may play a role in neurodegenerative disorder risk or vulnerability. continuous lysosomal turnover of cellular contents through Neuronal survival requires 19

20 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 72 . Impaired lysosomal function reduces lysosomal degradative endocytosis and autophagy up of toxic components in the cell. Impaired lysosoma efficiency, which leads to abnormal build l - system has been found to be associated with a broad range of neurodegenerative disorders, 77 77,78 76 - 74 73 79 , Niem , Huntington disease - Pick including AD , FTD ann , ALS , Parkinson disease 8 83 2 80,81 Marie Tooth disease type 2B disease type C , , Charcot - , Creutzfeldt Jakob disease - 84,85 , autosomal dominant hereditary spastic Neuronal ceroid lipofuscinoses (Batten disease) 88 87 86 89 , and , inclusion body myositis paraplegia . Higashi syndrome - , Chediak osteopetrosis Considering the extensive involvements of lysosomal/endosomal compartments in neurodegenerative disorders, it has been proposed that a long and chronic process of abnormal 72 . When lifespan metabolic changes during aging has led to the accumulation of toxic materials forms of neurodegenerative disorders, failures to degrade increases especially in the sporadic these waste products break the proteostasis and the balance maintained by the multicellular interactions, and trigger subsequent chain reactions that lead to neuronal death and outbreaks of neurodegenerative disorders due to different genetic susceptibilities and other disease various etiologies. Although each neurodegenerative disorder has its own characteristic proteopathy, the - cut acros s different disorders. In boundaries of protein pathology distribution are never clear fact, copathology or nonspecific pathology of proteopathy have been observed in most autopsies 90 - , and α - of neurodegenerative disorders, such as TDP 43 discussed above, Lewy body, synuclein TMEM106B etc. Our observation of lysosomal gene associated with neuronal proportion in aging cohorts suggests that the lysosomal pathway might be involved in the common mechanism underlying a broad range of neurodegenerativ e disorders or aging process in general that contribute to neuronal cell death. 20

21 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. Our study has demonstrated the great potential of using cell type composition as ach is quantitative traits to identify QTLs associated with the changes in cell fractions. This appro more powerful for disorders that involve considerably changes in cellular composition, for example, neurodegenerative disorders, and normal conditions during developmental or aging y increase the resolution in processes. The development of recent single cell studies will greatl advancing our knowledge of cellular population changes. More detailed fine mapping of cellular composition from single cell studies together with machine learning algorithms, bulk RNA Seq - apturing cellular fraction changes in the samples, such as deconvolution will be more accurately c different types of neurons or different states of astrocytes or microglia. Regarding scalability, this single cell powered bulk deconvolution approach is preferable for carrying out such cell type composition QTL analysis, because due to the high cost of performing single cell studies, bulk RNA - Seq is more financially feasible to scale up, and with larger sample sizes more hidden signals will be unrevealed with increased statistical power. To conclu TMEM106B associated de, we have identified a protective variant rs1990621 in with increased neuronal proportion through bulk RNA - Seq deconvolution and cell type protective also proportion QTL analysis. This observation replicated previous findings of the 25 . Besides, we also 9990621 h LD with rs1 ariant rs1990622 in FTD risk, which is in hig v ( co determine APOE e 4 isoform with rs7412 C allele ) observed the C allele of rs429358 ed suggest . It it was hypothesized associated with decreased neuronal proportion as potential s APOE and TMEM106B of both in neuronal protection mechanism underlying s vement invol supported previous observation of neurodegenerative and normal aging processes , and 61 related interactions between these two genes in AD cohort TMEM106B . We speculate that lysosomal changes might be involved in the common pathway underlying neuronal death and 21

22 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. e disorders and normal aging cohorts. With larger sample size astrocytosis in neurodegenerativ and higher deconvolution resolution, this approach will reveal more biologically relevant and novel loci associated with changes in cellular composition that are important for understanding both disease etiology and healthy aging. 22

23 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. References - 1 Jones, T. L. Beyond the neuron cellular Henstridge, C. M., Hyman, B. T. & Spires - , 20 Nature reviews. Neuroscience interactions early in Alzheimer disease pathogenesis. 0113 - 1 (2019). - - 018 8/s41583 - 108, doi:10.103 94 - analysis of 74,046 individuals identifies 11 new susceptibility Meta et al. Lambert, J. C. 2 1458, doi:10.1038/ng.2802 (2013). loci for Alzheimer's disease. Nat Genet 45 , 1452 - Gene dose of apolipoprotein E type 4 allele and the risk of 3 Corder, E. H. et al. 923 (1993). Alzheimer's disease in late onset families. - , 921 261 Science Common variants at ABCA7, MS4A6A/MS4A4E, EPHA1, CD33 Hollingworth, P. et al. 4 - 435, and CD2AP are associated with Alzheimer's disease. Nat Genet 43 , 429 doi:10.1038/ng.803 (2011). Genome - wide association study identifies variants at CLU and PICALM 5 Harold, D. et al. , 1088 - Nat Genet th Alzheimer's disease. associated wi 41 1093, doi:10.1038/ng.440 (2009). Lambert, J. C. - wide association study identifies variants at CLU and CR1 6 et al. Genome - 1099, doi:ng.439 [pii] associated with Alzheimer's disease. Nat Genet 41 , 1094 10.1038/ng.4 39 (2009). TREM2 variants in Alzheimer's disease. N Engl J Med 368 , 117 - 127, Guerreiro, R. et al. 7 doi:10.1056/NEJMoa1211851 (2013). et al. Variant of TREM2 Associated with the Risk of Alzheimer's Disease. N Jonsson, T. 8 Engl J Med , doi:10.1056/NEJMoa121 1103 (2012). et al. - Sims, R. 9 Rare coding variants in PLCG2, ABI3, and TREM2 implicate microglial - 49 1384, Nat Genet mediated innate immunity in Alzheimer's disease. , 1373 doi:10.1038/ng.3916 (2017). e of astrocytes as gatekeepers of Stobart, J. L. & Anderson, C. M. Multifunctional rol 10 neuronal energy supply. 7 , 38, Frontiers in cellular neuroscience doi:10.3389/fncel.2013.00038 (2013). Evidence supporting the existence of an activity - dependent astrocyte - 11 Pellerin, L. et al. , 291 elopmental neuroscience 299, doi:10.1159/000017324 neuron lactate shuttle. Dev 20 - (1998). J Neurochem , 1 Fonnum, F. Glutamate: a neurotransmitter in mammalian brain. 11 - 42 12 (1984). Dong, X. X., Wang, Y. & Qin, Z. H. Molecular mechanisms of excitotoxicity and their 13 30 , relevance to pathogenesis of neurodegenerative diseases. Acta pharmacologica Sinica 387, doi:10.1038/aps.2009.24 (2009). - 379 J 14 Tsacopoulos, M. & Magistretti, P. J. Metabolic coupling between glia and neurons. 885 (1996). - Neurosci 16 , 877 15 Simard, M., Arcuino, G., Takano, T., Liu, Q. S. & Nedergaard, M. Signaling at the gliovascular interface. 9262 (2003). J Neurosci 23 , 9254 - Rouach, N., Koulakoff, A., Abudara, V., Willecke, K. & Giaume, C. Astroglial metabolic 16 322 , 1551 - 1555, networks sustain hippocampa l synaptic transmission. Science doi:10.1126/science.1164022 (2008). Neuroinflammation in Alzheimer's disease. Lancet Neurol 14 et al. - 17 Heneka, M. T. , 388 5 (2015). 405, doi:10.1016/s1474 - 4422(15)70016 - 23

24 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. orsett, L. & El Khoury, J. Microglia in Hickman, S., Izzy, S., Sen, P., M 18 1369, doi:10.1038/s41593 - 018 - 0242 - x (2018). Nat Neurosci , 1359 - neurodegeneration. 21 Heppner, F. L., Ransohoff, R. M. & Becher, B. Immune attack: the role of inflammation 19 in Alzheimer disease. Nature reviews. Neur - 372, doi:10.1038/nrn3880 16 oscience , 358 (2015). Pekny, M. Astrocytes: a central element in neurological diseases. 20 et al. Acta Neuropathol - 1513 - 1 (2016). 131 , 323 - 345, doi:10.1007/s00401 - 015 21 Rodriguez - Arellano, J. J., Parpura, V., Zorec, R. & Verkhr atsky, A. Astrocytes in 182, physiological aging and Alzheimer's disease. 323 , 170 Neuroscience - doi:10.1016/j.neuroscience.2015.01.007 (2016). et al. Genetic variants associated with Alzheimer's disease confer different cerebral 22 Li, Z. 10 , 43, doi:10.1186/s13073 - 018 - - type cortex cell Genome Med population structure. 0551 - 4 (2018). Comprehensive functional genomic resource and integrative model for 23 Wang, D. et al. 362 Science the human brain. , doi:10.1126/science.aat8464 (2018). A human neurodevelopmental model for Williams syndrome. l. et a 24 Chailangkarn, T. 536 , 338 - Nature 343, doi:10.1038/nature19067 (2016). Common variants at 7p21 are associated with frontotemporal 25 Van Deerlin, V. M. et al. 42 - 234 - 239, doi:10.1038/ng.536 lobar degeneration with TDP 43 inclusions. Nat Genet , (2010). et al. TMEM106B risk variant is implicated in the pathologic 26 Rutherford, N. J. , 717 - 718, presentation of Alzheimer disease. Neurology 79 doi:10.1212/WNL.0b013e318264e3ac (2012). Satoh, J. TMEM106B expression is reduced in Alzheimer's disease brains. 27 et al. , 17, doi:10.1186/alzrt247 (2014). 6 Alzheimers Res Ther The Mount Sinai cohort of large - scale genomic, transcriptomic and 28 et al. Wang, M. 80185, doi:10.1038/sdata.2018.185 proteomic data in Alzheimer's disease. Sci Data 5 , 1 (2018). Gene expression elucidates functional impact of polygenic risk for et al. Fromer, M. 29 - 1453, doi:10.1038/nn.4399 (2016). schizophrenia. Nat Neurosci 19 , 1442 30 nscriptome data for Alzheimer's Allen, M. et al. Human whole genome genotype and tra , 160089, doi:10.1038/sdata.2016.89 3 Sci Data and other neurodegenerative diseases. (2016). 31 Bennett, D. A., Schneider, J. A., Arvanitakis, Z. & Wilson, R. S. Overview and findings , 628 - 645 (2012). Curr Alz heimer Res from the religious orders study. 9 et al. Overview and findings from the rush Memory and Aging Project. 32 Bennett, D. A. - , 646 9 Curr Alzheimer Res 663 (2012). Nat Genet - , 580 - 585, The Genotype 45 Tissue Expression (GTEx) project. 33 doi:10.1038/ng.2653 (2013). Battle, A., Brown, C. D., Engelhardt, B. E. & Montgomery, S. B. Genetic effects on gene 34 - 213, doi:10.1038/nature24277 (2017). , 204 550 expression across human tissues. Nature et al. Principal components analysis corrects for stratificatio n in genome - 35 Price, A. L. , 904 38 Nat Genet 909 (2006). - wide association studies. 36 Delaneau, O. & Marchini, J. Integrating sequence and array data to create an improved , 3934, 1000 Genomes Project haplotype reference panel. Nature communications 5 34 (2014). doi:10.1038/ncomms49 24

25 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. Howie, B., Fuchsberger, C., Stephens, M., Marchini, J. & Abecasis, G. R. Fast and 37 wide association studies through pre accurate genotype imputation in genome - - phasing. - , 955 44 Nat Genet 959, doi:10.1038/ng.2354 (2012). FastQC: a quality control tool for high throughput sequence data (2010). 38 - seq aligner. Dobin, A. 29 , 15 21, et al. STAR: ultrafast universal RNA 39 - Bioinformatics doi:10.1093/bioinformatics/bts635 (2013). Broad Institute The Picard Pipeline , < http://broadinstitute.github.io/picard/ 40 > ( - seq data. BMC Bioinformatics 41 Wang, L. Measure transcript integrity using RNA et al. - z (2016). 0922 016 - , 58, doi:10.1186/s12859 17 - Patro, R., Duggal, G., Love, M. I., Irizarry, R. A . & Kingsford, C. Salmon provides fast 42 14 and bias - 419, - aware quantification of transcript expression. Nat Methods , 417 doi:10.1038/nmeth.4197 (2017). - 43 specific Carter, R. Population Kuhn, A., Thu, D., Waldvogel, H. J., Faull, R. L. & Luthi - , expression an alysis (PSEA) reveals molecular changes in diseased brain. 8 Nat Methods 945 947, doi:10.1038/nmeth.1710 (2011). - 44 Gaujoux, R. & Seoighe, C. CellMix: a comprehensive toolbox for gene expression - 2212, doi:10.1093/bio informatics/btt351 (2013). deconvolution. Bioinformatics 29 , 2211 et al. PLINK: a tool set for whole - genome association and population - based 45 Purcell, S. - 575 (2007). , 559 Am J Hum Genet linkage analyses. 81 Q and 46 Turner, S. D. qqman: an R package for visualizing GWAS results using Q - , 005165, doi:10.1101/005165 (2014). bioRxiv tan plots. manhat 47 Sul, J. H., Han, B., Ye, C., Choi, T. & Eskin, E. Effectively identifying eQTLs from , - multiple tissues by combining mixed model and meta analytic approaches. PLoS Genet 9 e1003491, doi:10.1371/journ al.pgen.1003491 (2013). - - wide association studies. Han, B. & Eskin, E. Interpreting meta analyses of genome 48 8 , e1002555, doi:10.1371/journal.pgen.1002555 (2012). PLoS Genet and Watanabe, K., Taskesen, E., van Bochoven, A. & Posthuma, D. Functional mapping 49 annotation of genetic associations with FUMA. Nature communications 8 , 1826, - doi:10.1038/s41467 - 017 - 01261 5 (2017). 50 de Leeuw, C. A., Mooij, J. M., Heskes, T. & Posthuma, D. MAGMA: generalized gene - Plos Comput Biol e1004219, 11 , set analysis of GWAS data. doi:10.1371/journal.pcbi.1004219 (2015). - Tissue Expression (GTEx) pilot Consortium, G. T. Human genomics. The Genotype 51 , 648 - 660, analysis: multitissue gene regulation in humans. Science 348 doi:10.1126/science.1262110 (2015). Cruchaga, C. Association of TMEM106B gene polymorphism with age at onset in 52 t al. e - , 581 granulin mutation carriers and plasma granulin protein levels. Arch Neurol 68 586, doi:10.1001/archneurol.2010.350 (2011). TMEM106B regulates progranulin levels a nd the penetrance of FTLD in Finch, N. 53 et al. , 467 - 474, doi:10.1212/WNL.0b013e31820a0e3b GRN mutation carriers. Neurology 76 (2011). Membrane orientation and subcellular localization of transmembrane Lang, C. M. 54 et al. ontotemporal lobar degeneration. protein 106B (TMEM106B), a major risk factor for fr J - 19365, doi:10.1074/jbc.M112.365098 (2012). Biol Chem 287 , 19355 25

26 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. Brady, O. A., Zheng, Y., Murphy, K., Huang, M. & Hu, F. The frontotemporal lobar 55 degeneration risk factor, TMEM106B, regulates lysosomal morphology and fu nction. Hum Mol Genet , 685 - 695, doi:10.1093/hmg/dds475 (2013). 22 The FTLD risk factor TMEM106B and MAP6 control dendritic et al. Schwenk, B. M. 56 trafficking of lysosomes. 33 , 450 - 467, doi:10.1002/embj.201385857 (2014). Embo j 57 Stagi, M., Klein, Z. A., Gould, T. J., Bewersdorf, J. & Strittmatter, S. M. Lysosome size, - temporal dementia modifier motility and stress response regulated by fronto TMEM106B. - , 226 61 Molecular and cellular neurosciences 240, doi:10.1016/j.mcn.2014.07.006 (2014). , A. S. TMEM106B, the risk gene for frontotemporal dementia, is 58 Chen - Plotkin et al. regulated by the microRNA - 132/212 cluster and affects progranulin pathways. J 32 12.2012 (2012). - 11227, doi:10.1523/jneurosci.0521 , 11213 - Neurosci TMEM106B p. T185S regulates TMEM106B protein levels: et al. Nicholson, A. M. 59 - 791, J Neurochem 126 implications for frontotemporal dementia. , 781 doi:10.1111/jnc.12329 (2013). Frontotemporal dementia and its subtypes: a genome - wide association Ferrari, R. et al. 60 686 13 - 4422(14)70065 - 1 (2014). Lancet Neurol study. - , 699, doi:10.1016/s1474 A novel Alzheimer disease locus located near the gene encoding tau protein. 61 Jun, G. et al. Mol Psychiatry 117, doi:10.1038/mp.2015.23 (2016). - , 108 21 62 - 43 in frontotemporal lobar degeneration and Neumann, M. et al. Ubiquitinated TDP , 130 133, doi:10.1126/science.1134108 amyotrophic lateral sclerosis. Science 314 - (2006). - et al. TDP Ortiz, C. 43 immunoreactivity in hippocampal sclerosis and 63 - Amador 61 , 435 - 445, doi:10.1002/ana.21154 (2007). Alzheimer's disease. Ann Neurol Yasuda, H. - Co - morbidity of TDP - 43 proteinopathy in Lewy body Nakashima 64 et al. - 114 007 - 0261 - 2 229, doi:10.1007/s00401 Acta Neuropathol related diseases. - , 221 (2007). et al. Pathological , imaging and genetic characteristics support the 65 Josephs, K. A. Acta Neuropathol 137 , 227 - 238, existence of distinct TDP - 43 types in non - FTLD brains. - - doi:10.1007/s00401 - 018 1951 7 (2019). ment et al. Risk genotypes at TMEM106B are associated with cognitive impair 66 Vass, R. - 380, doi:10.1007/s00401 - in amyotrophic lateral sclerosis. Acta Neuropathol 121 , 373 y (2011). - 0782 - 010 et al. Differential clinicopathologic and genetic features of late - onset 67 Murray, M. E. , 411 128 1007/s00401 - 014 - 1302 - 2 Acta Neuropathol amnestic dementias. - 421, doi:10. (2014). Cognitive reserve and TMEM106B genotype modulate brain damage in Premi, E. 68 et al. , 1784 1791, 140 Brain presymptomatic frontotemporal dementia: a GENFI study. - doi:10.1093/brain/awx103 (2017). et al. Identif ication of genes associated with dissociation of cognitive 69 White, C. C. performance and neuropathological burden: Multistep analysis of genetic, epigenetic, and 14 transcriptional data. PLoS medicine , e1002287, doi:10.1371/journal.pmed.1002287 (2017). eliovich, A. Differential Aging Analysis in Human Cerebral Cortex Rhinn, H. & Ab 70 Cell Syst Identifies Variants in TMEM106B and GRN that Regulate Aging Phenotypes. 4 , 404 - 415.e405, doi:10.1016/j.cels.2017.02.009 (2017). 26

27 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. Hu, F. Elevated TMEM106B levels Zhou, X., Sun, L., Brady, O. A., Murphy, K. A. & 71 exaggerate lipofuscin accumulation and lysosomal dysfunction in aged mice with 5 , 9, Acta Neuropathologica Communications progranulin deficiency. 1 (2017). 0412 - 017 - doi:10.1186/s40478 - 72 Nixon, R. A., Yang, D. S. & Lee, J. H . Neurodegenerative lysosomal disorders: a 4 - 599 (2008). , 590 Autophagy continuum from development to late age. 73 Nixon, R. A. Endosome function and dysfunction in Alzheimer's disease and other 382, neurodegenerative diseases. Neurobiol Aging 26 , 373 - doi:10.1016/j.neurobiolaging.2004.09.018 (2005). et al. Apoptosis and autophagy in nigral neurons of patients with Parkinson's 74 Anglade, P. 31 (1997). - disease. Histology and histopathology , 25 12 Sarkar, S., Davies, J. E., Huang, Z., Tunnacliffe, A. & Rubinsztein, D. C. Trehalose, a 75 independent autophagy enhancer, accelerates the clearance of mutant novel mTOR - huntingtin and alpha , 5641 - 5652, - synuclein. J Biol Chem 282 doi:10.1074/jbc.M609532200 (2007). Webb, J. L., Ravikumar, B., Atkins, J., Skeppe - 76 r, J. N. & Rubinsztein, D. C. Alpha - 278 J Biol Chem Synuclein is degraded by both autophagy and the proteasome. , 25009 25013, doi:10.1074/jbc.M300227200 (2003). et al. Functional multivesicular bodies are required for autophagic 77 Filimonenko, M. J Cell Biol clearanc e of protein aggregates associated with neurodegenerative disease. 500, doi:10.1083/jcb.200702115 (2007). , 485 179 - 78 , 369 - Vonsattel, J. P. & DiFiglia, M. Huntington disease. J Neuropathol Exp Neurol 57 384 (1998). III Lee, J. A., Beigneux, A., - 79 Ahmad, S. T., Young, S. G. & Gao, F. B. ESCRT , 17 Curr Biol dysfunction causes autophagosome accumulation and neurodegeneration. - 1567, doi:10.1016/j.cub.2007.07.029 (2007). 1561 et al. Cell - autonomous death of cerebellar purkinje neurons w ith autophagy in 80 Ko, D. C. PLoS Genet - 95, doi:10.1371/journal.pgen.0010007 Pick type C disease. - Niemann 1 , 81 (2005). - 81 Pick C disease is Pacheco, C. D., Kunkel, R. & Lieberman, A. P. Autophagy in Niemann , 16 1 and responsive to lipid trafficking defects. Hum Mol Genet dependent upon Beclin - 1495 1503, doi:10.1093/hmg/ddm100 (2007). - Lysosomes as key organelles in the pathogenesis of prion Laszlo, L. 82 et al. encephalopathies. 341, doi:10.1002/path.1711660404 (1992). J Pathol 166 , 333 - 83 Shirk, A. J., Anderson, S. K. , Hashemi, S. H., Chance, P. F. & Bennett, C. L. SIMPLE interacts with NEDD4 and TSG101: evidence for a role in lysosomal sorting and J Neurosci Res , 43 82 implications for Charcot - 50, - Tooth disease. - Marie doi:10.1002/jnr.20628 (2005). et al. Participation of autophagy in storage of lysosomes in neurons from 84 Koike, M. Am J Pathol lipofuscinoses (Batten disease). - mouse models of neuronal ceroid 167 , - 9440(10)61253 9 (2005). - 1728, doi:10.1016/s0002 - 1713 85 Involvement of two different cell death pathways in retinal atrophy of Koike, M. et al. 22 , 146 - 161 (2003). cathepsin D - deficient mice. Molecular and cellular neurosciences et al. The hereditary spastic paraplegia protein spastin interacts with the ESCRT - 86 Reid, E. 14 , 19 - 38, III complex - associated endosomal protein CHMP1B. Hum Mol Genet doi:10.1093/hmg/ddi003 (2005). 27

28 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. vsB is Kypri, E., Schmauch, C., Maniak, M. & De Lozanne, A. The BEACH protein L 87 localized on lysosomes and postlysosomes and limits their fusion with early endosomes. - 0854.2007.00567.x 8 Traffic (Copenhagen, Denmark) , 774 - 783, doi:10.1111/j.1600 (2007). 88 body myositis: newest concepts of pathogenesis Askanas, V. & Engel, W. K. Inclusion - , 1 - 14 (2001). 60 J Neuropathol Exp Neurol and relation to aging and Alzheimer disease. lysosomal pathway: emerging roles of CLC 89 Jentsch, T. J. Chloride and the endosomal - 578 640, The Journal of physiology nsporters. chloride tra , 633 - doi:10.1113/jphysiol.2006.124719 (2007). Espay, A. J. et al. Revisiting protein aggregation as pathogenic in sporadic Parkinson and 90 , 329 - 337, doi:10.1212/wnl.00000000000 06926 Alzheimer diseases. Neurology 92 (2019). 28

29 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. Demographic information for cohorts included in the study. AD: Alzheimer’s Disease; FTD: frontal temporal dementia; Table 1 BP: bipolar disease; PSP: progressive supranuclear palsy; PA: pathological aging; PD: Parkinson’s Disease; SCZ: schizophrenia; OTH: other unknown dementia or no diagnosis information. % Male RIN TIN Control AD PSP Region PA PD SCZ BP OTH Age N FTD Discovery 1 86.6 ± 4.59 35.4 7.07 ± 0.99 73.2 ± 5.13 114 338 0 0 0 0 0 0 71 523 ROSMAP Replication 80.4 ± 8.37 48.1 8.16 ± 0.903 77.4 ± 5.94 69 80 260 82 29 0 0 0 0 1 0 Mayo 35.6 6.42 ± 1.77 76.4 ± 2.52 49 170 0 0 84 ± 7.32 0 0 0 0 0 4 MSSM 219 1 83.1 ± 12 42.6 6.44 ± 1.2 79.4 ± 1.91 13 77 11 0 0 1 0 0 6 108 Knight ADRC 50.9 ± 7.08 73.3 5.55 ± 1.09 78.9 ± 0.99 0 15 0 0 0 0 0 0 0 15 1 DIAN 67.7 6.92 ± 0.846 73.8 ± 2.97 125 1 0 0 58.2 ± 9.91 0 0 0 0 4 3 130 GTEx 1 64.6 ± 18 62.3 7.67 ± 0.899 50 ± 7.21 170 0 0 0 0 0 210 34 0 414 CommonMind 426 343 11 82 29 1 210 34 10 1,146 Replication Total 540 681 11 82 29 1 210 34 81 1,669 Merged Total 29

30 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. TCX: temporal cortex; PAR: parietal cortex; CTX: cortex; General information of seven studies evolved in the analysis. Table 2 FCX: frontal cortex; DLPFC: dorsal lateral prefrontal cortex. BM9: dorsal lateral prefrontal cortex; BM10: Anterior prefronta l cortex; BM36: parahippocampal gyrus; BM44: inferior cortex; BM22: superior temporal gyrus; BM24: ventral anterior cingulate frontal gyrus. Mean coverage unit is million. mRNA DNA Library Type Sequencer Mean Coverage Read Length Brain Region Reference Dataset Enrichment type Discovery Bennett 2012; HiSeq - A selection DLPFC 99.2 ± 29.29 WGS 101 Paired end ploy ROSMAP 2000 Bennett 2012 Replication HiSeq Genotype 101 ploy - A selection TCX 158.31 ± 34.04 Allen 2016 Paired end Mayo 2000 BM10 BM22 HiSeq 100 rRNA depletion 35.96 ± 10.04 Single end Wang 2018 WGS MSSM BM36 BM44 2500 - Knight HiSeq PAR Genotype 150 Li 2018 Paired end rRNA depletion 137.87 ± 21.81 4000 ADRC HiSeq rRNA depletion PAR Paired end 149.82 ± 19.68 Genotype Li 2018 150 DIAN 4000 GTEx 2013; BM24 CTX HiSeq ploy - A selection 76 48.28 ± 13.2 WGS Paired end GTEx FCX 2000 Battle 2017 Common HiSeq Fromer 86 ± 21.12 rRNA depletion Paired end BM9 Genotype 100 2016 2500 Mind 30

31 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. rs1990621 (chr7:12283873) major allele C is significantly associated with decreased neuronal proportions. Therefore, G Table 3 allele (MAF = 0.4658) is significantly associated with increased neuronal proportions. Sample Size Beta SE P value Dataset Brain Region Ref Allele - 07 - - 0.3 0.06 6.40 × 10 DLPFC C Discovery 484 04 - 1,052 - 0.13 0.04 7.41 × 10 Replication Multiple C - 09 × 1,536 - 0.16 0.05 9.42 10 Multiple C tissue - Merged meta 31

32 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. associated with decreased neuronal proportions in AD, Control, rs1990621 (chr7:12283873) major allele C is significantly Table 4 - AD neurodegenerative disorders. SCZ: schizophrenia; other: other non AD neurodegenerative disorders, including - and other non ateral prefrontal cortex. TCX: temporal cortex. progressive supranuclear palsy and pathological aging. BM9: dorsal l Sample Size Beta SE P - value Disease Brain Region Ref Allele - 07 10 - 0.26 0.07 1.95 × 639 C AD Multiple 02 - 476 - 0.14 0.06 2.94 × 10 Control Multiple C - 01 10 - 0.01 0.09 9.32 × 189 C BM9 SCZ 04 - 103 - 0.45 0.14 8.19 × 10 Other TCX C 32

33 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. values were reported for variants previously identified - Neuronal proportion cQTL p Table 5 in AD risk (by Lambert et al.) and FTD risk (by Ferrari et al.) studies. cQTL p AD risk p - value FTD risk p - value SNP BP - CHR value 02 - 24 - - × 10 × 5.7 5.59 rs6656401 1 207,692,049 10 08 - - 01 rs35349669 × 10 10 × - 3.2 2 9.80 234,068,476 08 - - 01 rs190982 3.2 × 10 5 - 88,223,420 5.88 × 10 - 01 - 08 7.83 - 1 · 57 × 10 6 10 rs1980493 32,363,215 × - - 01 01 10 - 3 · 43 × 10 3.72 rs3129871 6 32,406,342 × - - 01 02 - 3 · 36 × 10 × 1.04 rs3129882 32,409,530 6 10 09 - 01 - rs9268856 - 5 · 51 6 10 32,429,719 8.12 × 10 × - 01 - 08 32,431,147 1.91 - 1 · 05 × 10 × 10 rs9268877 6 - 01 - 12 2.9 × 10 4.31 10 - 6 rs9271192 32,578,530 × - 01 - 11 - 5.2 × 10 8.62 × rs10948363 47,487,762 6 10 01 - 03 - 10 rs1020004 - 4 · 59 × 7 12,255,778 1.35 × 10 - 08 - 01 × - 1 · 21 × 10 12,265,988 10 rs6966915 7 1.24 - 08 - 02 × - 7 · 88 12,283,787 10 10 × rs1990622 7 1.44 - 01 - 09 × 4.8 × 10 10 37,841,534 - rs2718058 7 9.99 10 - 01 - 10 rs1476679 5.6 × 10 7 - 100,004,446 7.71 × - 01 - 13 143,110,762 × 1.1 × 10 10 - 7 rs11771145 9.67 - 02 - 14 7.4 × 10 27,195,121 10 - 6.38 8 rs28834970 × - 01 - 04 × - 4 · 38 10 8.94 10 27,543,281 9 rs3849942 × 08 - 02 - 47,557,871 10 1.1 × 10 8.60 - 11 rs10838725 × 26 - 01 - rs10792832 9.3 × 10 11 - 85,867,875 4.50 × 10 - 02 - 07 × - 2 · 44 × 10 87,876,911 10 rs302668 11 4.57 - - 01 15 × 9.7 × 10 1.35 10 - 11 rs11218343 121,435,587 09 - 01 - 10 7.9 × 10 53,400,629 - 14 rs17125944 × 1.71 09 - 01 - 10 5.5 × 10 × - rs10498633 14 8.32 92,926,952 - 03 01 - 10 8.89 - 4 · 82 × 10 44,019,712 rs242557 17 × - 01 - 04 44,081,064 - 2 · 80 × 7.48 × 10 rs8070723 17 10 15 - 01 - 10 rs4147929 1.1 × 10 19 - 1,063,443 4.78 × - 04 - 07 10 × - 8 · 81 × 2.11 10 45,395,619 19 rs2075650 06 - 01 - 10 10 rs3865444 3.0 × 19 - 51,727,962 2.02 × - - 01 08 10 55,018,260 3.01 2.5 × 10 × - 20 rs7274581 33

34 bioRxiv preprint first posted online Mar. 20, 2019; . The copyright holder for this preprint doi: http://dx.doi.org/10.1101/583286 (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. Subjects Genotype PCA Plot ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● Pop ● ● ● ● ● 0.075 MSBB HapMap_CEU NIALOAD ● ● ● ● ● ● GTEX HapMap_JPT CMC ● ● ● HapMap_YRI ROSMAP DIAN ● ● 0.050 ● ● MayoADGS MAP ● ● ● PC2 ● ● ● ● ● ● ● ● ● 0.025 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.000 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● − 0.04 0.03 0.01 0.00 0.01 0.02 PC1 Genotype data PCA analysis was performed to select European Genomic PCA analysis. Figure 1 - off lines. HapMap_CEU: - 0.002 and PC2 < 0.008 with red dotted cut ancestry subjects with PC1 < HapMap Utah residents with Northern and Western European ancestry; HapMap_JPT: HapMap kyo, Japan; HapMap_YRI: HapMap Yoruba in Ibadan, Nigeria; MayoADGS: Mayo Japanese in To Clinic study participants; MSBB: MSSM study participants; GTEX: GTEx study participants; ADRC - - DIAN: DIAN study participants; MAP: Knight ADRC participants; NIALOAD: Knight participant s; CMC: CommonMind participants; ROSMAP: ROSMAP participants. 34

35 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. IBD analysis was performed to select unrelated Figure 2 Genomic IBD analysis. - subjects with Z0 > 0.8 and Z1 < 0.2 with red dotted cut off lines. When there are one individual will be dropped from the related pair. related individuals, 35

36 bioRxiv preprint first posted online Mar. 20, 2019; doi: . The copyright holder for this preprint http://dx.doi.org/10.1101/583286 (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. Astrocyte Microglia Oligodendrocyte Neuron B A PAR DIAN − − DIAN PAR − PAR PAR − Knight ADRC Knight ADRC − MSSM − BM10 BM10 MSSM BM22 − MSSM BM22 − MSSM MSSM BM36 − BM36 MSSM − BM44 BM44 MSSM − MSSM − Dataset Dataset − Mayo TCX TCX − Mayo − ROSMAP DLPFC ROSMAP − DLPFC GTEx BM24 BM24 − GTEx − − CTX − GTEx CTX GTEx GTEx FCX GTEx FCX − − BM9 − CommonMind BM9 − CommonMind 60 − 0 75 100 30 0 30 25 50 − 60 25 50 75 100 − 0 20 2 4 100 60 30 0 30 − 0 25 50 75 0 10 100 75 − 50 25 0 10 20 Celltype Proportion (%) Celltype Proportion (%) Major CNS cell type proportions derived from RNAseq datasets with each row representing Figure 3 Cell proportion distribution. each tissue of each study A) raw cell type proportions inferred from the data with vertical bars indicating quantiles within each tissue fter and each cell type. B) cell type proportions were normalized by subtracting the mean from each tissue deconvolution result a removing outliers. 36

37 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. - Seq and paired genotype or WGS data were accessed and preprocessed for Figure 4 Study Design . RNA downstream analysis. Genotype data was censored based on our quality control criteria and imputed as res to select needed. WGS and imputed genotype were merged and followed by PCA and IBD procedu - unrelated European ancestry subjects. RNA Seq data was quality checked with FastQC and aligned to human GRCh37 primary assembly with Star, from which TIN was inferred with RSeQC to account for RNA integrity - d into the analysis. Gene expression were quantified from unaligned RNA variances that we later incorporate - Seq with psedo aligner Salmon for deconvolution procedure. Cell type composition comprised of four major CNS cell type proportions were inferred by performing deconvolution procedure o n gene expression quantification results. Using cell type proportions as quantitative traits, we identified loci in TMEM106B gene region associated with neuronal proportion in our assembled dataset. 37

38 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. phases Manhattan and QQ plots. Figure 5 Discovery and replication Loci located in chromosome 7 were associated with neuronal proportion in ROSMAP discovery dataset and replicated in replication dataset. A) Discovery set Manhattan plot showed seven peaks associated with neuronal proportion a t suggestive threshold. - 07 - . B) QQ plot of value = 6.4 × 10 The peak located in chromosome 7 was labeled, which is for rs1990621 with p the discovery phase analysis. C) Replication set Manhattan plot showed that the peak located in chromosome 7 - 04 . D) QQ plot of the replication 10 × value = 7.41 - signal identified during discovery phase with p replicated the phase analysis. 38

39 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. gene region was based analysis. Merged SNP TMEM106B based and gene - - Figure 6 rs1990621 located in chromosome 7 based - significantly associated with neuronal proportion in cortical RNAseq dataset. A) Manhattan plot showed SNP based genome - wide significant hit located in chromosome 7 with other suggestive SNP hits labeled. B) QQ plot of the SNP - based genome - analys is. C) Manhattan plot showed gene - wide significant hit located in chromosome 7 with other suggestive based analysis. - gene hits labeled. D) QQ plot of the gene 39

40 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. value and confidence - Figure 7 Meta - A) Forest plot showed p Tissue analysis results of rs1990621. interval for rs1990621 for each tissue site of each dataset that included in the Meta - Tissue analysis. - Summary random effect was depicted at the bottom as RE Summary. B) PM Plot of rs1990621 while value (y axis) and m - value (x axis). Red dot indicates that the variant is predicted to have combining both p - an effect in that particular dataset, blue dot means that the variant is predicted to not have an effect, and - value and confidence interval for rs1990621 green dot represents ambiguous prediction. C) Fore st plot p for discovery, replication, and merged datasets. D) Forest plot p value and confidence interval for - rs1990621 when splitting the merged dataset into four main disease categories. 40

41 bioRxiv preprint first posted online Mar. 20, 2019; The copyright holder for this preprint . http://dx.doi.org/10.1101/583286 doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. C A B in view of the hit in chromosome 7 - Variant Figure 8 A) Local plot showed the zoom rs1990621 functional annotation and local plot. with the top lead SNP rs1990621 labeled with dark purple. Nearby SNPs were also mainly located in the TMEM106B gene region and color LD r2 thresholds. B) Bottom panel showed combined CADD score, RegulomeDB score, and Chromatin state of the region shown coded with in the top panel. C) Regulome DB and chromatin state explanation. 41

42 bioRxiv preprint first posted online Mar. 20, 2019; http://dx.doi.org/10.1101/583286 The copyright holder for this preprint . doi: (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. - regions local plot. A) Local plot showed the zoom APOE in view of the hit in chromosome 7 with target SNP TMEM106B Figure 9 and gene rs1990622 labeled with dark purple, and the top leading SNP is rs1990621. Nearby SNPs were also mainly located in the TMEM106B - in view of the hit in chromosome 19 with target SNP region and color coded with LD r2 threshol ds. B) Local plot showed the zoom rs2075650 labeled with dark purple, and the top three leading SNPs are rs283815, rs769449, and rs429358. Nearby SNPs were als o mainly on and color coded with LD r2 thresholds. One gene omitted in this region is SNRPD2. located in the TOMM40/APOE gene regi 42

Related documents

CityNT2019TentRoll 1

CityNT2019TentRoll 1

STATE OF NEW YORK 2 0 1 9 T E N T A T I V E A S S E S S M E N T R O L L PAGE 1 VALUATION DATE-JUL 01, 2018 COUNTY - Niagara T A X A B L E SECTION OF THE ROLL - 1 CITY - North Tonawanda TAX MAP NUMBER ...

More info »
Out of Reach 2018

Out of Reach 2018

2018 of OUT REACH THE HIGH COST OF HOUSING MADE POSSIBLE BY THE GENEROSITY OF:

More info »
Out of Reach 2016

Out of Reach 2016

No Refuge for Low Income Renters MADE POSSIBLE BY THE GENEROSITY OF:

More info »
DER Directory

DER Directory

FAA CONSULTANT DER DIRECTORY May 9, 2019 AIR-6F0, Delegation & Organizational Procedures Branch This directory is generated from information in the FAA Designee Information Network (DIN). If you are a...

More info »
time is the enemy

time is the enemy

is TIME the ENEMY The surprising truth about why today’s college students graduating ... and wha T needs To change aren’t n 1 Time Is the Enemy

More info »
50 Year Trends expanded version

50 Year Trends expanded version

Expanded Edition THE AMERICAN HERI A R CIRP T B I E N L G E C 50 F S I F R T A Y E Y FRESHMAN FIFTY-YEAR TRENDS | 1966-2015 KEVIN EAGAN | ELLEN BARA STOLZENBERG | JOSEPH J. RAMIREZ | MELISSA C. ARAGON...

More info »
Implementation Handbook For The Convention On The Rights Of The Child

Implementation Handbook For The Convention On The Rights Of The Child

IMPLEMENTATION HANDBOOK FOR THE CONVENTION ON THE RIGHTS OF THE CHILD FULLY REVISED THIRD EDITION IMPLEMENTATION HANDBOOK IMPLEMENTATION HANDBOOK FOR THE CONVENTION ON THE FOR THE CONVENTION ON THE RI...

More info »
Abstract and Concrete Categories   The Joy of Cats

Abstract and Concrete Categories The Joy of Cats

Jiˇr ́ı Ad ́amek Horst Herrlich George E. Strecker Abstract and Concrete Categories The Joy of Cats Dedicated to Bernhard Banaschewski

More info »
IWMF Global Report

IWMF Global Report

Global Report on the Status of Women in the News Media INTERNATIONAL WOMEN’S MEDIA FOUNDATION

More info »
What's It Worth? The Economic Value of College Majors

What's It Worth? The Economic Value of College Majors

Anthony P. Carnevale Jeff Strohl Michelle Melton

More info »
S:\FULLCO~1\HEARIN~1\Committee print 2018\Henry\Jan. 9 report

S:\FULLCO~1\HEARIN~1\Committee print 2018\Henry\Jan. 9 report

1 C . RT S. P 115 TH ONGRESS " ! COMMITTEE PRINT 2d Session 115–21 PUTIN’S ASYMMETRIC ASSAULT ON DEMOCRACY IN RUSSIA AND EUROPE: IMPLICATIONS FOR U.S. NATIONAL SECURITY A MINORITY STAFF REPORT PREPARE...

More info »
AndersBehringBreivikManifesto

AndersBehringBreivikManifesto

2011 , London – By Andrew Berwick

More info »
DoD7045.7H

DoD7045.7H

DoD 7045.7-H EPARTMENT OF D EFENSE D F UTURE Y EARS D EFENSE P ROGRAM (FYDP) S TRUCTURE Codes and Definitions for All DoD Components Office of the Director, Program Analysis and Evaluation A pril 2004

More info »
Web Tables—Profile of Undergraduate Students: 2011–12

Web Tables—Profile of Undergraduate Students: 2011–12

Profile of WEB Undergraduate 12 Students: 2011– TABLES EDUCATION OF DEPARTMENT U.S. NCES 2015-1 OCTOBER 2014 67 These Web Tables provide comprehen- distance from home; and participation carried a bala...

More info »
Microsoft Word   cust subm Book 251 Cover.doc

Microsoft Word cust subm Book 251 Cover.doc

Summary of Reciprocity Agreements Between Texas and Other Jurisdictions Vehicle Titles and Registration Division Austin, Texas Issued November 2002

More info »
summer schedule full

summer schedule full

Summer Semester 2019 Class Schedule May 9, 2019 as of Subj Crse CRN Course Type Title Days Time Bldg/Rm Cred Instructor Course Fee Start Date - End Date Campus Part Term BB 321 Financial Accounting I ...

More info »
An Introduction to Computer Networks

An Introduction to Computer Networks

An Introduction to Computer Networks Release 1.9.18 Peter L Dordal Mar 31, 2019

More info »
G:\COMP\PHSA\PHSA.bel

G:\COMP\PHSA\PHSA.bel

G:\COMP\PHSA\PHSA-MERGED.XML PUBLIC HEALTH SERVICE ACT [As Amended Through P.L. 115–408, Enacted December 31, 2018] References in brackets ¿ ø¿ ø are to title 42, United States Code TITLE I—SHORT TITL...

More info »
U.S. Mexico Canada Trade Agreement: Likely Impact on the U.S. Economy and on Specific Industry Sectors

U.S. Mexico Canada Trade Agreement: Likely Impact on the U.S. Economy and on Specific Industry Sectors

United States International Trade Commission U.S. -Mexico -Canada Trade Agreement: Likely Impact on the U.S. Economy and on Specific Industry Sectors April 2019 Publication Number: 4889 Investigation ...

More info »