Study design

The overall aim of our study was to assess the phenotypic and functional effects of a Finnish founder mutation, 5.2 kb TYROBP deletion, that is known to cause the early-onset neurodegenerative disease NHD in biallelic state. We hypothesized that the monoallelic TYROBP deletion may be a risk factor for neurodegeneration later in life and may induce functional changes in the microglial cells. In the first part of the study, we aimed to elucidate the clinical phenotype of the monoallelic TYROBP deletion in a large, well-characterized biobank cohort FinnGen. To achieve this aim, we first identified the founder haplotype and genetic proxy markers for the Finnish TYROBP deletion by utilizing whole genome sequencing (WGS) of three Finnish NHD patients. Using the proxy markers, we conducted phenome-wide association study in FinnGen data freeze 12 (DF12) which contains genome and digital healthcare data on 520,210 Finnish individuals and 2,489 clinical endpoints. To complement the picture of clinical phenotype induced by TYROBP deletion, we present case reports of two monoallelic TYROBP deletion carriers. The first case is a 34-year-old female presenting classic NHD bone cysts in wrists and ankles. The second case is a 75-year-old female diagnosed with idiopathic normal pressure hydrocephalus (iNPH), with amyloid β (Aβ)-positive frontal cortical brain biopsy and cerebrospinal fluid (CSF) biomarker profile indicative of AD-related brain pathology.

In the second part of the study, we aimed to experimentally evaluate the effects of TYROBP deletion using microglia-like models in vitro. Human monocytes for monocyte-derived microglia-like cell (MDMi) differentiation were extracted from peripheral venous blood samples obtained from carriers of the Finnish TYROBP deletion (> 60 years), non-carrier age-matched controls, and NHD patients (30–40 years). MDMi cultures in basal conditions or after treatment with myelin or lipopolysaccharide (LPS) were used for omics and targeted functional analyses as detailed below. To study the effect of TYROBP loss-of-function on M-CSF induced signaling, we created Tyrobp knock-out (KO) mouse microglial BV2 cell lines using CRISPR-Cas9 genome editing and clonal selection or used siRNA to partially silence Tyrobp (Additional file 1, Supplementary Materials and Methods).

All study protocols concerning human samples were approved by Medical Research Ethics Committee of Wellbeing Services County of North Savo. Written informed consent was obtained from all participants. The Ethics Committee of the Hospital District of Helsinki and Uusimaa (HUS) has coordinated the approval for FinnGen Study.

Study subjects

NHD patients, monoallelic TYROBP deletion carriers, and unrelated controls were recruited during 2020–2024 from Neurology clinics at Kuopio University Hospital and Oulu University Hospital, Kuopio University Hospital NPH registry [16], Finnish Geriatric Intervention Study to Prevent Cognitive Decline and Disability (FINGER) [17], Biobank of Eastern Finland, and Auria Biobank. Blood samples for monocyte and/or DNA isolation were collected following written informed consent from each participant.

The FinnGen Study (https://www.finngen.fi/en) is a large biobank-scale research project which combines genome data with digital healthcare data based on national health registers [14]. FinnGen includes samples collected by the Finnish biobanks as well as legacy samples from previous research cohorts that have been transferred to the biobanks. FinnGen Study approved the use of the data in the present work.

The study subjects in FinnGen have provided informed consent for biobank research based on the Finnish Biobank Act. Alternatively, separate research cohorts that were collected prior to the Finnish Biobank Act coming into effect (September 2013) and start of FinnGen (August 2017), were collected based on study-specific consents and later transferred to the Finnish biobanks after approval by the Finnish Medicines Agency Fimea. Participant recruitment followed the biobank protocols approved by Fimea. The Coordinating Ethics Committee of the Hospital District of Helsinki and Uusimaa (HUS) statement number for the FinnGen study is HUS/990/2017. The complete list of ethics committee approval numbers, study permits, and biobank sample and data accession numbers are included in the Declarations at the end of the manuscript.

Genotyping

For non-FinnGen study participants, genomic DNA was extracted from peripheral whole blood using QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany). DNA for a subset of 50 imputed TYROBP deletion proxy marker carriers from the FinnGen cohort was obtained through the Biobank of Eastern Finland.

Library preparation and WGS on Illumina NovaSeq sequencing platform was carried out at Novogene (Novogene (UK) Company Limited, Cambridge, UK). WGS data were initially processed using the nf-core sarek pipeline (release 3.1) with default settings [18], and the GATK GRCh38 as the reference genome. After identifying potential problems in the initial read alignment at and near the deletion breakpoints, WGS data were aligned again to the human reference genome (GRCh38) using a splice-aware aligner STAR (v2.7.9a) [19] with essential non-default settings: –outFilterMultimapNmax 1, –outFilterMismatchNmax 3, –alignIntronMax 10,000 and –alignMatesGapMax 10,000.

TYROBP deletion-specific PCR was carried out as described previously [11]. Genotyping of SNV 19:35,901,079-T-G was carried out by Sanger sequencing. In brief, the target region was amplified by PCR using primers 5’- GCGAACGCAGTCCCTGAATGG- 3’ (forward) and 5’-CCTCCCTCTGGACCCAGTAA- 3’ (reverse) and the PCR product was cleaned using NucleoSpin Gel and PCR Clean-up mini kit (Macherey–Nagel, Düren, Germany). The purified PCR product was combined with the reverse primer and sent to Macrogen Europe (Amsterdam, the Netherlands) for Sanger sequencing. Sanger sequencing was reliable only from reverse direction due to the presence of several poly-T repeats between the forward primer and the variant. APOE genotyping was carried out with pre-designed TaqMan SNP genotyping assays for rs429358 and rs7412 (both from ThermoFisher Scientific). TaqMan SNP genotyping assays were performed according to manufacturer’s instructions, and all samples were assayed in duplicate.

The whole FinnGen cohort has been genotyped with multiple Illumina (Illumina Inc., San Diego, USA) and Affymetrix (Thermo Fisher Scientific, Santa Clara, CA, USA) chip arrays as part of the FinnGen Study. Chip genotype data were imputed using the Finnish population-specific imputation reference panel Sequencing Initiative Suomi project (SISu v4.2), Institute for Molecular Medicine Finland (FIMM), University of Helsinki, Finland (http://sisuproject.fi).

FinnGen analyses

Phenotype information and clinical endpoints in FinnGen are based on different national health registries, including hospital discharge registers, prescription medication purchase registers, and cancer registers. A complete list of FinnGen endpoints and their respective controls are available at https://www.finngen.fi/en/researchers/clinical-endpoints and can be explored at https://risteys.finngen.fi/.

In this study we used summary statistics and data from the FinnGen data release R12. GWAS studies for FinnGen core endpoints were performed with REGENIE 2.2.4. A detailed description of the analytical methods is available at https://github.com/FINNGEN/regenie-pipelines. Phenotype associations for variant rs1244787406-G across 2489 FinnGen core analysis endpoints were visualized using LAVAA [20]. Regional association plots for the endpoint ‘dementia including primary care registry’ were generated with topr 2.0.0 package in R 4.3.2 [21, 22]. Linkage disequilibrium in-sample dosage in the TYROBP-locus (3 Mb window around the lead variant) in FinnGen was computed using LDstore2 [23] and fine-mapping was carried out using SuSiE [24] with the maximum number of causal variants in a locus L = 10.

The map visualizing the regional allele frequency of rs1244787406-G was created in R based on the region of birth for minor and major allele carriers. Kaplan–Meier curves were drawn in R using package survminer v0.4.9 [25].

Radiological imaging

The monoallelic TYROBP deletion carrier and the NHD patient were imaged as part of diagnostic procedure with conventional x-rays (Siemens Ysio Max, Erlangen, Germany) for skeletal features of the hands and feet, and at 3,0 Tesla MRI (Philips Achieva, Best, NL) and 1,5 Tesla MRI (GE Signa Artist, Milwaukee, USA) with standard clinical sequences, including T1, T2, FLAIR, DWI, and susceptibility weighted imaging, for potential brain pathology. The 3D T1 MRI data were also analyzed by brain volumetry software, cNeuro (Combinostics Ltd, Tampere, Finland).

Immunohistochemistry and CSF biomarkers

Diagnostic brain biopsy specimen collected during NPH shunt surgery were used for immunohistochemical analysis. Immunostaining for Iba-1 and Aβ was performed as described previously [26]. Full section brightfield images were obtained with Hamamatsu NanoZoomer-XR Digital slide scanner with 20x (NA 0.75) objective (Hamamatsu Photonics K.K., Shizuoka, Japan) and analyzed as described previously [26]. In short, Aβ plaques were manually outlined in NDP.view2 software (Hamamatsu Photonics K.K.) to obtain the plaque size. The plaque-associated microglia with clearly visible soma were manually counted by an investigator blinded to sample identity.

CSF samples from the same NPH patients were obtained by lumbar puncture. Levels of AD-related biomarkers Aβ42, total Tau (T-Tau), and Tau phosphorylated at Serine 181 (P-Tau 181) were analyzed using a commercial enzyme-linked immunosorbent assay (Innotest, Fujirebio, Ghent, Belgium).

Monocyte isolation and MDMi differentiation

To extract peripheral blood mononuclear cells (PBMCs), 60–100 ml peripheral venous blood was collected from participants and processed within 24 h of sample collection. PBMCs were extracted by density gradient centrifugation over Ficoll-Paque PLUS (#17–1440 - 02, Cytiva, Marlborough, MA, USA) in SepMate- 50 tubes (#85,450 Stemcell Technologies, Vancouver, Canada). PBMC were either cryoprotected in CryoStor CS10 (Stemcell Technologies) or used directly for CD14-positive monocyte isolation using human CD14 MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany) and magnetic-activated cell sorting. When using cryopreserved PBMCs, the cells were thawed at 37 °C for 5 min, washed in RPMI medium containing 20% FBS and incubated for 30 min at 37 °C in RPMI/20% FBS to allow the cells to recover from the thawing prior to monocyte isolation.

To differentiate monocytes into MDMi, monocytes were plated at 0.5 × 106 cells per well in 12-well plates (for RNA-sequencing, LC–MS/MS, and alphaLISA) or in 96-well plates at 1 × 105 cells per well (for conditioned medium). Monocyte differentiation into MDMis was done according to a previously published protocol [27], by culturing the monocytes for 10–12 days in vitro (DIV) under standard humidified environment (+ 37 °C, 5% CO2) in MDMi culture medium consisting of RPMI- 1640 Glutamax (#61,870,036, Gibco, Billings, MT, USA) supplemented with 1% penicillin/streptomycin and a mixture of the following human recombinant cytokines: M-CSF (10 ng/ml, #574,804, Biolegend, San Diego, CA, USA), GM-CSF (10 ng/ml, #215-GM- 010/CF, R&D Systems, Minneapolis, MN, USA), NGF-β (10 ng/ml, #256-GF- 100, R&D Systems), CCL2 (100 ng/ml, #571,404, Biolegend), and IL- 34 (100 ng/ml, #5265-IL- 010/CF, R&D Systems).

RNA extraction and RNA-sequencing

After MDMi differentiation, conditioned medium was replaced with fresh MDMi culture medium or with culture medium containing myelin (25 µg/ml) or LPS (200 ng/ml, O26:B6, L5543, Sigma Aldrich), and the cells were cultured for 24 h prior to sample collection. For total RNA extraction, cells from 2–3 replicate wells/donor were collected into ice-cold PBS and RNA was extracted immediately. Acutely isolated monocytes were placed in Macherey–Nagel™ NucleoProtect RNA reagent (ThermoFisher Scientific) and stored at + 4 °C until RNA extraction. RNA was isolated using High Pure RNA Isolation Kit (11,828,665,001, Roche), and the RNA extracts were stored at − 80 °C until further use. Culture preparation and RNA sequencing of induced pluripotent stem cell (iPSC) and iPSC-derived microglia (iMG) has been described earlier [28].

Library preparation and RNA sequencing was conducted by Novogene (UK) Company Limited. In brief, mRNA enrichment was performed with oligo(dT) bead pulldown, from where the pulldown material was subjected to fragmentation, followed by reverse transcription, second strand synthesis, A-tailing, and sequencing adaptor ligation. The final amplified and size selected library comprised 250–300-bp insert cDNA and paired-end 150 bp sequencing was executed with an Illumina high-throughput sequencing platform. Sequencing yielded 4.4–26.9 million sequenced fragments per sample.

The 150 nucleotide pair-end RNA-seq reads were quality-controlled using FastQC (version 0.11.7) (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Reads were then trimmed with Trimmomatic (version 0.39) [29] to remove Illumina sequencing adapters and poor quality read ends, using as essential settings: ILLUMINACLIP:2:30:10:2:true, SLIDINGWINDOW:4:10, LEADING:3, TRAILING:3, MINLEN:50. Trimmed reads were aligned to the Gencode human transcriptome version 38 (for genome version hg38) using STAR (version 2.7.9a) [19] with essential non-default settings: –seedSearchStartLmax 12, –alignSJoverhangMin 15, –outFilterMultimapNmax 100, –outFilterMismatchNmax 33, –outFilterMatchNminOverLread 0, –outFilterScoreMinOverLread 0.3, and –outFilterType BySJout. The unstranded, uniquely mapping, gene-wise counts for primary alignments produced by STAR were collected in R (version 4.2.2) using Rsubread::featureCounts (version 2.12.3) [30], ranging from 3.3 to 22.1 million per sample. Differentially expressed genes (DEGs) between experimental groups were identified in R (version 4.2.0) using DESeq2 (version 1.36.0) [31] by employing Wald statistic and lfcShrink for FC shrinkage (type = “apeglm”) [32]. Comparisons between TYROBP genotype groups were performed adjusting for the APOE ε genotype, and between monocytes and MDMi cells by adjusting for donor (equivalent of paired test). Pathway enrichment analysis was performed on the gene lists ranked by the pairwise DEG test log2 FCs in R using clusterProfiler::GSEA (version 4.4.4) [33] with Molecular Signatures Database gene sets (MSigDB, version 2022.1) [34].

Proteomics analysis

MDMi were cultured and treated as described above for RNA-sequencing. Cultured MDMi were lysed in RIPA lysis buffer (50 mM Tris–Cl, 150 mM NaCl, 1% NP- 40, 0.05% sodium deoxycholate, 0.01% SDS, pH 7.5) by incubation on ice for 20 min with intermediate vortexing. Cell debris and undissolved material were removed by centrifugation at 16.000 × g for 10 min at 4 °C, and protein concentration was determined using a BCA Assay (Thermo Fisher Scientific). 10 µg of each sample were diluted 1:2 in water and benzonase (Sigma-Aldrich) digestion was performed with 10 units for 30 min at 37 °C to remove remaining DNA/RNA. Protein digestion was performed with 125 ng LysC and 125 µg trypsin (Promega) using the single-pot solid-phase enhanced sample preparation (SP3) [35]. The peptide solution was filtered through 0.22 µm Costar SPIN-X columns (Corning) and dried by vacuum centrifugation. Samples were dissolved in 20 µL 0.1% formic acid using a sonication batch (Hielscher) and the peptide concentration was measured using the Qubit protein assay (Thermo Fisher Scientific).

An amount of 350 ng of peptides were separated on an in-house packed C18 analytical column (15 cm × 75 µm ID, ReproSil-Pur 120 C18-AQ, 1.9 µm, Dr. Maisch GmbH) using a binary gradient of water and acetonitrile (B) containing 0.1% formic acid at flow rate of 300 nL/min (0 min, 2% B; 2 min, 5% B; 70 min, 24% B; 85 min, 35% B; 90 min, 60% B) and a column temperature of 50 °C. A Data Independent Acquisition Parallel Accumulation–Serial Fragmentation (DIA-PASEF) method with a cycle time of 1.4 s was used for spectrum acquisition. Briefly, ion accumulation and separation using Trapped Ion Mobility Spectrometry (TIMS) was set to a ramp time of 100 ms. One scan cycle included one TIMS full MS scan and The DIA-PASEF windows covered the m/z range from 350–1,000 m/z with 26 windows of 27 m/z with an overlap of 1 m/z. The raw data was analyzed using the software DIA-NN version 1.8 [36] for protein label-free quantification (LFQ). A one protein per gene canonical fasta database of Homo Sapiens (download date March 1 st 2023, 20,603 entries) from UniProt and a fasta database with 246 common potential contaminations from Maxquant [37] were used to generate a spectral library in DIA-NN with a library free search which included 10,615 proteins. Trypsin was defined as protease. Two missed cleavages were allowed, and peptide charge states were set to 2–4. Carbamidomethylation of cysteine was defined as static modification. Acetylation of the protein N-term as well as oxidation of methionine were set as variable modifications. The false discovery rate for both peptides and proteins was adjusted to less than 1%. Data normalization was disabled.

Identification of differentially expressed, APOE ε genotype-corrected proteins between experimental groups, non-normalized LFQ intensities as the starting point, and the subsequent pathway enrichment were done as for the RNA-seq data, except using version 4.2.3 for R and version 1.38.3 for DESeq2 [31].

AlphaLISA

CD14 + monocytes were isolated from PBMC as described above and lysed in AlphaLISA lysis buffer. MDMi were prepared from monocytes as described above and lysed in AlphaLISA lysis buffer. The lysates were stored at − 20 °C until measured.

Total DAP12 levels were measured from the cell lysates using AlphaLISA Surefire Ultra Total DAP12 Detection Kit (ALSU-TDAP12-A-HV, Revvity), and the results were normalized to GAPDH levels measured from the same lysate using AlphaLISA Surefire Ultra Human and Mouse Total GAPDH Detection Kit (ALSU-TGAPD-B-HV, Revvity). Samples were measured as duplicates.

Cytokine measurement in conditioned medium

MDMi were treated with 200 ng/ml LPS for 24 h. Conditioned medium was collected, centrifuged at 5,000 × g at + 4 °C for 10 min, supplemented with protease inhibitor, and stored at − 80 °C. IL- 1β, IL- 6, and IL- 10 were measured using a custom U-plex assay (Meso Scale Diagnostics, Rockville, MD, USA). A phase contrast image was captured before the LPS treatment on IncuCyte S3, and the data were normalized to cell confluency. TNFα was measured from the conditioned medium using commercial ELISA kit (88–7346 - 22, Thermo Fisher) according to manufacturer’s instructions. Data is shown as average of three replicate wells/donor measured in duplicates.

Myelin isolation

Myelin was isolated from adult male C57BL/6 J mice using the sucrose gradient protocol described elsewhere [38, 39]. 0.85 M and 0.32 M sucrose solutions were used for the gradient and Beckman Ultracentrifuge with Ti 50.2 rotor for the centrifugation. The concentration of the myelin preparation was determined using Pierce BCA Protein Assay kit (23,225, Thermo Scientific) and stored at − 80 °C until used.

Statistical analysis

Statistical analyses and data visualization were performed in R 4.3.2 [22] and GraphPad Prism v10. Chi-squared test was used to analyze regional allele frequency. Survival analysis comparing disease-free survival of rs1244787406-G carriers and noncarriers was performed in R where Kaplan–Meier curves were drawn using package survminer v0.4.9 [25] and cox proportional hazards model was performed with package survival v3.2–7 [40, 41]. Independent samples T-test was used to analyze normally distributed immunohistochemistry and CSF biomarker data between the APOE ε3ε3 and ε3ε4 groups and cytokine data between monoallelic TYROBP deletion carriers and noncarriers. Additional information on the statistical test, sample size, and technical replicates is given in the figure legends.