# paper.pdf ## Metadata - PDFFormatVersion=1.6 - Language=en - IsLinearized=true - IsAcroFormPresent=false - IsXFAPresent=false - IsCollectionPresent=false - IsSignaturesPresent=false - CreationDate=D:20260616112551+05'30' - Creator=LaTeX with hyperref package - ModDate=D:20260616112623+05'30' - Producer=Acrobat Distiller 26.0 (Windows) - Title=Mitochondrial NADK2‐dependent NADPH controls tau oligomer uptake in human neurons - Custom.WPS-ARTICLEDOI=10.1002/alz.71594 - Custom.WPS-JOURNALDOI=10.1002/(ISSN)1552-5279 - Custom.WPS-PROCLEVEL=3 - xmp:createdate=2015-08-18T08:05:36+05:30 - xmp:modifydate=2016-06-21T14:17:29-04:00 - xmp:metadatadate=2016-06-21T14:17:29-04:00 - pdf:producer=Acrobat Distiller 9.0 (Windows) - pdfx:wps-articledoi=10.1002/alz.71594 - pdfx:wps-journaldoi=10.1002/(ISSN)1552-5279 - pdfx:wps-proclevel=3 - jav:journal_article_version=VoR - prism:doi=10.1002/alz.71594 - dc:format=application/pdf - dc:title=Mitochondrial NADK2‐dependent NADPH controls tau oligomer uptake in human neurons - xmpmm:documentid=uuid:7047ea28-0600-3111-9dc4-41de3357ef2e - xmpmm:instanceid=uuid:4e9e8439-c5c1-3c98-b178-d43f27d5804c ## Contents ### Page 1 Received: 19 October 2025 Revised: 6 May 2026 Accepted: 18 May 2026 DOI: 10.1002/alz.71594R E S E A R C H A R T I C L E Mitochondrial NADK2-dependent NADPH controls tau oligomer uptake in human neurons Evelyn Pardo1 Taylor Kim1 Horst Wallrabe1,2 Kristine E. Zengeler3 Vijay Kumar Sagar1,2 Garnett Mingledorff1 Xuehan Sun1 Ammasi Periasamy1,2 John R. Lukens3 George S. Bloom1,4,5 Andrés Norambuena11Department of Biology, University of Virginia, Charlottesville, Virginia, USA 2W.M. Keck Center for Cellular Imaging, University of Virginia, Charlottesville, Virginia, USA 3Center for Brain Immunology and Glia, Department of Neuroscience, University of Virginia, Charlottesville, Virginia, USA 4Department of Cell Biology, University of Virginia, Charlottesville, Virginia, USA 5Department of Neuroscience, University of Virginia, Charlottesville, Virginia, USACorrespondenceEvelyn Pardo and Andrés Norambuena, Department of Biology, University of Virginia, PO Box 400328, Charlottesville, VA 22904-4328, USA. Email: xgd5ms@virginia.edu and an2r@virginia.eduFunding informationUniversity of Virginia’s School of Medicine; Rick Sharp Alzheimer’s Foundation; Cure Alzheimer’s Fund; Owens Family Foundation; NIH Office of the Director, Grant/Award Number: OD025156; National Institute on Aging, Grant/Award Number: R01AG067048Abstract INTRODUCTION: Reduced brain energy metabolism, mitochondria dysfunction, and extracellular tau oligomer buildup characterize Alzheimer’s disease (AD), but how these phenomena cooperatively promote neurodegeneration is poorly understood. We now report that tau oligomers (TauOs) pathologically coordinate mitochondrial metabolism with increased expression of a plasma membrane (PM) tau receptor.METHODS: Mitochondrial energy metabolism was recorded using two-photon flu- orescence lifetime microscopy of mitochondrial nicotinamide adenine dinucleotide phosphate (NADPH) in live human neurons and PS19 mouse brain.RESULTS: Recombinant or human brain-derived TauOs upregulate expression of the mitochondrial NAD+ kinase, mitochondrial NAD kinase 2 (NADK2), and by extension,de novo NADPH synthesis. This process controls expression of low-density lipoprotein receptor-related protein 1 (LRP1), a major PM receptor for tau, thereby establishing a vicious cycle for further TauO internalization. Upregulation of the NADK2-NADPH pathway was detected in live presymptomatic PS19 mouse brains and in AD patient- derived neurons.DISCUSSION: Upregulation of mitochondrial NADK2-dependent NADPH controls a key step in TauO toxicity and may represent an early stage in human AD.K E Y W O R D SAlzheimer’s disease, brain metabolism, endocytosis, mitochondria, NADPH, PS19, tau spreadingHighlights∙ Tau oligomers (TauOs) upregulate the activity and expression of the mitochon- drial mitochondrial NAD kinase 2–nicotinamide adenine dinucleotide phosphate (NADK2-NADPH) pathway.∙ TauOs upregulate the expression of low-density lipoprotein receptor-related pro- tein 1 (LRP1). This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. © 2026 The Author(s). Alzheimer’s & Dementia published by Wiley Periodicals LLC on behalf of Alzheimer’s Association.Alzheimer’s Dement. 2026;22:e71594. wileyonlinelibrary.com/journal/alz 1 of 20https://doi.org/10.1002/alz.71594 ### Page 2 2 of 20 PARDO ET AL.∙ Mitochondrial NADK2 mechanistically connects TauO uptake to tau toxicity.∙ Dysregulation of the NADK2-NADPH pathway occurs at presymptomatic stages of AD.1 BACKGROUNDNeuronal uptake of pathological forms of tau1,1–5 mitochondrial dysfunction,6 and brain hypometabolism7,8 are key features in Alzheimer’s disease (AD), detectable decades before the onset of clinical symptoms. These phenomena thus represent early biomarkers in AD pathogenesis which coincide with the detection of soluble forms of pathological tau in the cerebrospinal fluid (CSF) of still healthy, presymptomatic humans.9–12 However, the mechanistic connections among these processes have remained elusive. As current available treatments primarily target amyloid-β and tau at prodromal stages of AD, understanding the molecular mechanisms leading to disruption in mitochondrial metabolism at presymptomatic stages could provide new insights for developing future disease-modifying and, hopefully, preventive therapies. Mitochondria are central players in energy metabolism, redox regulation, and biosynthetic pathways.13 These functions directly depend on two important coenzymes contained within the organelle: the pyridine nucleotide, nicotinamide adenine dinucleotide + hydro- gen (NADH), and its phosphorylated form, nicotinamide adenine dinucleotide phosphate (NADPH), which regulate oxidative phos- phorylation and provide reducing power for biosynthetic pathways, respectively.14 Although an optimal NADH concentration is essen- tial for mitochondrial function, it has recently become evident that the mitochondrial NADPH pool also plays a critical role in regu- lating fatty acid, proline, and collagen synthesis to maintain proper cellular functions.15–17 Prolonged imbalance of NADH/NADPH has been linked to diseases like cancer, aging, and neurodegenerative disorders.14 While several metabolic pathways contribute to the cellu- lar levels of NADPH,14 its de novo synthesis in mitochondria requires NAD+ kinase 2 (NADK2).15,17 Although mutations in this enzyme have major neurological consequences associated with early death in humans,18,19 dysregulation of NADK2 and its effect in AD pathogene- sis has not been reported previously. Since its discovery in 1975,20 tau has traditionally been known for its role in regulating microtubule dynamics in neurons.21 Recently, additional functions were reported in multiple other physiological events, such as maintenance of axon initial segment and nuclear mem- brane integrity,22 and regulation of mechanistic target of rapamycin complex 1 (mTORC1) activity on lysosomes.23 Recent studies in induced pluripotent stem cell (iPSC) -derived neurons have revealed the ability of tau to interact with several proteins involved in signaling, protein synthesis, synaptic functioning, and, particularly, mitochondrial activity.24 Interestingly, pathological forms of tau carrying disease- associated mutations disrupt their interactions with mitochondrial proteins, highlighting its potential role in mediating mitochondrial hypometabolism in AD.24 It is well known that pathological tau can form soluble, misfolded tau oligomers (TauOs), which have been pro- posed as the true toxic species25–27 due to their ability to trigger tau misfolding, a precursor of the neurofibrillary tangles that accumulate in human AD brain.28 Pathological tau has been detected in the CSF and plasma of human patients > 10 years before the onset of cognitive decline,11 and TauOs were found to be increased in the CSF of patients at early stages of AD.29 While a previous study showed that subcor- tical injection of TauOs altered the expression of a few mitochondrial proteins in mice,30 the TauOs’ impact on mitochondrial metabolism was not explored. A growing body of evidence suggests that TauOs are transferred along neuro-anatomically connected brain areas. This process involves a series of molecular events, including hyperphosphorylation of tau, formation of seed-competent tau proteoforms, and their release from donor cells. Next, these “abnormal” tau species (or seeds) are taken up by a recipient cell via endocytosis,5,31,32 a key step in this prion-like propagation of tau throughout the brain.33 However, the exact mech- anisms regulating the expression of TauOs receptors at the plasma membrane (PM) have remained unknown. Using two-photon fluorescence lifetime imaging (2P-FLIM), which allows label-free, mitochondrial metabolic imaging in live cells and tissues,23,34–36 we now provide evidence that TauOs amplify their own uptake by upregulating the mitochondrial NADK2-NADPH metabolic pathway in human neurons, obtained by either differentiation of neu- ronal progenitors or by direct conversion of skin fibroblasts from adult AD patients. Importantly, NADK2-NADPH upregulation was also observed in presymptomatic PS19 live mouse brain and pro- tein samples from postmortem human AD brains. Strikingly, we also found that this process controls the expression of low-density lipopro- tein receptor-related protein 1 (LRP1). Since LRP1 is a major reg- ulator of TauO uptake and spread,5 dysregulation of mitochondrial NADPH balance could be a very early contributor to human AD initiation.2 METHODS 2.1 Cell culture and materialsHuman embryonic kidney cells (Lenti-X 293T Cell line from Takara. Cat no. 632180) were grown in Dulbecco’s modified Eagle medium (DMEM)/F12 media (GIBCO) supplemented with HyClone cosmic calf serum (GE Healthcare) and 50 μg/ml gentamycin (GIBCO). ### Page 3 PARDO ET AL. 3 of 20RESEARCH IN CONTEXT1. Systematic review: The search for presymptomatic diag- nosis and disease-modifying therapies for Alzheimer’s disease (AD) requires understanding molecular mech- anisms of AD pathogenesis. Impaired brain energy metabolism and tau toxicity are hallmarks of early AD, but mechanisms explaining how they promote synaptic dysfunction and neuron death, are not fully under- stood. This study aimed at clarifying such pathogenic mechanisms. 2. Interpretation: Soluble extracellular tau oligomers increased mitochondrial nicotinamide adenine din- ucleotide phosphate (NADPH) levels in live human cultured neurons and in vivo pre-symptomatic PS19 mice, a model of human AD. 3. Future directions: Exploring the mechanism(s) involved is a major goal. Examples include refined understanding of (a) how different cell types in the central nervous sys- tem contribute to NADPH levels in the brain; (b) how NADPH contribute to tau spreading and toxicity; and (c) how to leverage new findings to improve AD diagnostics and therapeutics. The ReNcell VM cell line of neuronal progenitors were purchased from Sigma-Aldrich (cat #SCC08). These cells derived from the ventral mesencephalon of a human fetal brain (more information can be found on the CelloSaurus website [ReNcell VM (RRID:CVCL_E921)]. ReNcell VM were differentiated into human-like neurons following a protocol described earlier by us and others.34,35,37 For differentiation, the cells were plated into 12-well plates or 35-mm glass-bottom dishes with DMEM/F12 differentiation media (GIBCO) supplemented with 2 μg/ml heparin (Science Cell), 2% (v/v) B27 neural supplement (GIBCO), and 50 μg/ml gentamycin (GIBCO) without growth factors. One-half of the differentiation medium was replaced every 3 days until the cultures were used for experiments. Human neurons derived from differen- tiated ReNCells VM cells express neuronal markers such as NeuN, synapsin-1, synaptophysin, βIII-tubulin, and microtubule-associated protein 2 (MAP2) as previously reported by Song et al.382.2 Preparation of iNeuronsDirect Conversion of Adult Human Fibroblasts into iNeurons (iN): Primary human dermal fibroblasts were obtained from the Coriell Institute for Medical Research (cat #AG04402-AD, male, 47 years old); cat #AG06265-AD, male, 61 years old; cat #AG06858-control, male, 47 years old; cat #AG02261-control, male, 61 years old). Pri- mary fibroblasts were also obtained from four patients enrolled in study HSR210075 at the University of Virginia (#UVA004AD-PCA, posterior cortical atrophy, male, 79 years old; # UVA007AD, male, 84 years old; UVA006C, male, 79 years old, #UVA008C, male, 79 years old). All the AD cases analyzed are from late-onset Alzheimer’s disease (LOAD) and non–apolipoprotein E (non-APOE) ε4 carriers. Protocols were previously approved by the University of Virginia (UVA) Internal Review Board for Health Sciences Research, and informed consent was obtained from all subjects. Fibroblasts were cultured in minimum essential medium (MEM) (Gibco #11095-080) containing 15% tetracycline-free fetal bovine serum (VWR 76308- 984), 1% NEAA (Gibco #11140050), 1 mM Sodium Pyruvate (Gibco #11360070), and transduced with lentivirus expressing the transcrip- tion factors Ngn2:2A and Ascl1. Transduced fibroblasts were expanded with 1 μg/ml puromycin (Gibco A1113803). Following at least four to five passages post-transduction, cells were considered ‘iN-ready.’ They were trypsinized and pooled into high densities (∼140,000-500,000 cells per cm2 in 2D & 3D culture, respectively) and plated in optically clear 24-well plates (Ibidi USA #82426). 3D cultures were resus- pended in a media/Matrigel mix (5 mg/ml final concentration: Corning #356231) and allowed to polymerize at 37◦C for 40 minutes before adding additional media. The following day, the medium was changed to neuron conversion medium (NC) based on a 1:1 mix of DMEM: F12 (Gibco #11320033) and Neurobasal-A (Gibco #10888022) for three to five weeks. NC contains the following supplements: 1x N2 (Gibco #17502048), 1x B27 (Gibco #17504044), 2 μg/ml doxycycline (Sigma #D9891), 1 μg/ml Laminin (Thermo-Fisher #23017015), 400 μg/ml dibutyryl–cyclic-AMP (Sigma #D0627), 150 ng/ml human recombi- nant Noggin (PeproTech #120-10C), 0.5 μM LDN-193189 (Cayman Chemicals #19396) 0.5 μM A83-1 (Santa Cruz #sc-203791), 3 μM CHIR99021 (Cayman Chemicals #13122), 5 μM Forskolin (BioGems #6652995) and 10 μM SB-431542 (Cayman Chemicals #13031). The medium was changed every two to three days. Once neuronal morphol- ogy was detected (weeks 2-3), 10% Astrocyte Conditioned Medium (ScienCell #1811) was supplemented to the NC medium.2.3 MiceThe PS19 (B6.Cg-Tg(Prnp-MAPT*P301S)PS19Vle/J) tauopathy mouse model was purchased from The Jackson Laboratory (Strain #024841). Animals were housed in a pathogen-free barrier facility with a 12-hour light/12-hour dark cycle and ad libitum access to food and water under a protocol approved by the Institutional Animal Care and Use Commit- tee (IACUC) of the University of Virginia. All pups were genotyped by polymerase chain reaction (PCR), following The Jackson Laboratory’s recommendations. Experiments were performed starting at 8 weeks of age on 3 males per condition cohort of in-house bred mice. Mice were injected intracerebroventricularly (ICV) with an AAVs Scramble (pAAV9- mTagBFP2-U6-scr-shRNA. Catalog #: AAV9S(VB010000-9430hya)- K1) or NADK2 sequence (pAAV9-mTagBFP2-U6-mNADK2-shRNA. Catalog #: AAV9S(VB230209-1263bpw)-K1). Injections were per- formed under standard aseptic surgery conditions. Mice were anaes- thetized with xylazine (10%)/ ketamine (2%). ICV injections were ### Page 4 4 of 20 PARDO ET AL.made at the coordinates (from bregma A/P −0.5mm, M/L +1.0mm, D/V −2.0mm). A Hamilton needle was used to inject the AAV solu- tion at 0.5 μL min−1. Afterward, the skin over the injection site was sutured, and mice were allowed to recover on a warming pad. After 5 weeks, mice were euthanized and processed for analysis. Mice received ibuprofen immediately after surgery and for 2 days afterward if needed. No calculations were performed to predetermine sample size; sample sizes were based on previously published work and cho- sen to support meaningful conclusions, and no specific method of randomization was used.2.4 Lentivirus production and transductionLentiviral particles for shNADK2, shP5CS, shPYCR1, and shLRP1 knockdowns were prepared as follows. The expression plasmids, pLKO.1 (Mission shRNA library from Sigma-Aldrich, see below), and the packaging vectors, pSPAX2 and pMD2.G (Addgene plasmids #12260 and #12259, respectively), were transfected using Lipofectamine 3000 (Thermo-Fisher) into HEK293T cells grown in 15 cm Petri dishes to∼70% confluence in DMEM (GIBCO) supplemented with 10% HyClone cosmic calf serum. Each transfection was performed with 15 μg total DNA at a 50% expression vector, 37.5% pSPAX plasmid, and 12.5% pMD2G plasmid. The lentivirus-conditioned medium was collected 24 and 48 h after the start of transfection. For expression of Ngn2:2A and Ascl1 in dermal fibroblasts, Hek 293 cells were co-transfected with pLVX-UbC-rtTA-Ngn2:2A:Ascl1 (Addgene #127289) and Lenti-X Packaging Single Shots (VSV-G; Takara #631275) following the manu- facturer’s recommendations. Lentiviral particles were concentrated in a Beckman Coulter Optima XE-90 ultracentrifuge for 2 h at 23,000 rpm (90,353 x gmax) at 4◦C in an SW32Ti rotor, resuspended in 400 μL Neurobasal medium, and stored at −80◦C in 40 μL aliquots. Cultured neurons were transduced for 72–96 h before the experiments were performed.2.5 Human brain cell extractsThe Institutional Review Board for Health Sciences Research at UVA approved this study. Human frontal cortex brain samples were pre- pared as we previously described.232.6 Mouse brain cell extractsBrain cortex samples were placed in a Wheaton 1 mL Dounce homogenizer containing 1 mL lysis buffer to cover the sample. Lysis buffer consisted of RIPA buffer (Bioworld, #420200242), 1% HALT™ protease inhibitor cocktail, 1% phosphatase inhibitor cocktail 2 (Sigma Aldrich, #P5726), and 1% phosphatase inhibitor cocktail 3 (Sigma Aldrich, #P0044). Brain homogenates were placed in an Eppendorf 5415C centrifuge at 10,000 rpm (8160 x gmax) for 10 min at 4◦C. Finally, supernatants were trans- ferred to new tubes and used as protein lysates for western blots.2.7 Human shRNA sequencesThe following plasmids were purchased from Mission shRNA library from Sigma-Aldrich: (1) NADK2.1 TRCN00000414087, (2) NADK2.2 TRCN00000424042, (3) PYCR1.1 TRCN0000038979, (4) PYCR1.2 TRCN0000038981, (5) P5CS.1 TRCN0000064849, (6) P5CS.2 TRCN0000064852, (7) NADK1 TRCN00000199808, (8) LRP1 TRCN0000230615, (9) LRP1 TRCN0000230615. These constructs are already cloned into pLKO.1, and lentivirus particles were prepared as described in a previous section. The empty backbone of the pLKO.1 was used as a control.2.8 AntibodiesThe following antibodies were from Cell Signaling Technologies: NADK2 (Proteintech C5orf33 #26352-1-AP; 1/5000), NADK1 (cell signaling #55948; 1/1000), P5CS (Proteintech # 17719-1- AP; 1/1000), PYCR1 (Proteintech #13108-1-AP; 1/5000), PRODH (LSBio #LS-C795520; 1/1000), P5CDH (Proteintech # 11604-1-AP; 1/1000), LRP1 (cell signaling #64099; 1/1000), ColVI (Proteintech #17023-1-AP; 1/5000), T22 (Millipore Sigma #ABN454-1; 1/5000), NeuN (Cell signaling #24307; 1/10000), GAPDH (R) (cell signal- ing #D16H11; 1/5000), GAPDH (M) (Proteintech #60004-1-Ig; 1/5000), GFAP (cell signaling #GA5; 1/5000), Synapsin-1 (cell signaling #D12GA; 1/20000), synaptophysin (cell signaling #D35E4; 1/10000), PSD95 (cell signaling #2507S; 1/1000), MAP2 (cell signaling #4542S; 1/5000), pTau-Thr231(AT180) (Invitrogen #MN1040; 1/3000), pTau- Thr202/205 (AT8) (Invitrogen #MN1020; 1/3000), pTau-Thr231 (Invitrogen #44-746G; 1/500), hTau-HT7 (Invitrogen #MN1000; 1/1000), NMDAR (Novus #NB300-105; 1/5000). βIII-tubulin antibody (cell signaling #5568; 1/10000). βIII-tubulin antibody (Tuj1; 1/10000) was a generous gift by Dr. Anthony Spano. Dr. Lester Binder provided Tau-R1 hybridoma cells (1/1000, respectively).2.9 ImmunoblottingSamples were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using 12% acrylamide/bis-acrylamide gels and transferred to 0.22 μm nitrocellulose (Bio-Rad). Membranes were blocked with Odyssey blocking buffer (LI-COR Biosciences) and were incubated with primary antibodies and secondary IRDye-labeled anti- bodies diluted into antibody buffer (Odyssey blocking buffer diluted 1:1 with phosphate buffered saline (PBS)/0.1% Tween 20). All antibody incubations were overnight at 4◦C, and 3 washes of 5 min each with PBS/0.1% Tween 20 were performed after each antibody step. Mem- branes were quantitatively scanned by an Odyssey imaging station (LI-COR Biosciences). ### Page 5 PARDO ET AL. 5 of 202.10 Other fluorescence microscopy proceduresImmunofluorescence procedure for human neurons. Cells were rinsed in PBS and fixed for 15 minutes in 2% paraformaldehyde. Then, they were washed and permeabilized in washing buffer (0.3% Triton in PBS) once for 5 minutes each, washed with DPBS, and quenched with 300mm glycine in DPBS for 10 minutes. The cells were incubated for 60 minutes in blocking buffer (TBS-LI-COR, Biosciences #927-60001; supplemented with 1% Goat Serum, Southern Biotech #00060-01). Fixed cells were then incubated with primary (ON) and secondary (1.5 hours) antibodies diluted into a blocking buffer, with several PBS washes after each antibody step. DAPI (4′,6-diamidino-2-phenylindole; Thermo-Scientific #62248) was incubated in the last washing step. Finally, coverslips were mounted in each well using Fluoromount-G (Thermo-Fisher #00-4958-02). Immunohistochemistry procedure for mouse brain tissue. After ICV injection, mice were perfused with cold PBS. The brain was then sec- tioned coronally through the middle. The coronal-posterior brain sec- tion was lysed and processed for protein analysis as described above. The coronal-anterior brain section was immersed in 4% paraformalde- hyde (PFA) overnight. This tissue was sent to the UVA Histology facility to be paraffin-embedded and cut into 5-10 μm slices. Paraffin was removed from brain tissue slides using a graded series of xylenes. Sub- sequently, antigen retrieval was performed by submerging brain slides in a beaker with citrate buffer and then microwaving at high power for 2.5 minutes, followed by low power for 8 minutes. The beaker was then gradually cooled to room temperature over 20 minutes, and afterward, it was placed in an ice slurry for another 20 minutes. Brain sections were washed with PBS for 5 minutes and then incubated with blocking solution (Licor Intercept Protein-Free Blocking Buffer PBS (#927-90001) and 0.3% Tween 20) for 1 hour. This was followed by sequential 2-h incubations with primary and secondary antibodies. Multiple PBS washes were performed after each antibody step. Before sealing with a coverslip using Fluoromount G (ThermoFisher # 00- 4958-02), sections were washed for 10 minutes with Autofluorescence Inhibitor Reagent (Millipore # 2160). Both human neurons and mouse tissue were imaged on a Zeiss LSM 880 inverted microscope equipped with an AiryScan 32-channel GaAsP spectral detector array, 2-PMT, a 20x/0.8 Plan Apochromat objective, a 405 nm Diode laser, 488 nm Argon laser, 542 nm DPSS laser, and a 594/633 nm HeNe laser.2.11 Recombinant tauThe human 2N4R tau with a C-terminal His-tag was produced by expression in BL21(DE3) Escherichia coli cells. Protein expression was induced with 0.5mm Isopropyl β-D-1-thiogalactopyranoside (Sigma- Aldrich, I6758) for 2 h at 37◦C. Cells were pelleted by centrifugation, resuspended, and sonicated with Misonix Sonicator 3000 20 times for 30 s each at 60% power. Then, the E. coli lysate was centrifuged in Sor- vall RC6 Plus 6 centrifuges at 18,083 x gmax for 25 minutes with an SLA 1500 rotor. Tau was then purified batch-wise from the supernatant using TALON Metal Affinity Resin (Takara, cat. no. 635502) according to the vendor’s instructions. Finally, purified 2N4R tau was concen- trated, and the buffer was exchanged into 10mm HEPES, pH 7.6, using Amicon Ultra-4 Centrifugal Filters (Millipore, UFC801024).392.12 TauOsOligomeric tau was prepared as described earlier.40 The 2N4R tau iso- form was adjusted to 8 μM in the oligomerization buffer: 10mm HEPES (pH 7.6), 100mm NaCl, 0.1mm ethylenediaminetetraacetic acid, and 5mm dithiothreitol (DTT). The protein was then allowed to oligomer- ize in the presence of 300 μM arachidonic acid (Cayman Chemicals, cat. no. 90010) for 18 h at room temperature in the dark. Oligomerization was verified on Western blots. Oligomerization buffer, prepared in par- allel with oligomeric tau, containing 300 μM arachidonic acid, was used as a vehicle (control) for cell treatments.39 Dr. Rakez Kayed generously donated the human brain-derived tau oligomers (BDTauO).2.13 ATP accumulation assaysWild-type (WT) human neurons (differentiated for 21 days) were grown in 96-well plates (black wall/clear bottom, Costar catalog number #3603). ATP was assayed using the ATPlite 1-step Lumi- nescence Assay System, following the manufacturer’s instructions (Perkin-Elmer).2.14 Subcellular fractioning and mass spectrometry protein identificationSubcellular fractioning was performed according to the vendor’s instructions from the Qproteome® Cell compartment kit (QIAGEN, Catalog #37502). 100 μg of membrane fraction from human neurons WT and NADK2-KO were sent for mass spectrometry analysis at the Biomolecules analysis core facility at UVA. Briefly, the samples were precipitated by acetone-methanol to produce a protein pellet. The pel- let was re-solubilized in a digestion buffer using heat and sonication. The samples were reduced with DTT for 30 minutes at room tem- perature (RT), then alkylated using iodoacetamide for 30 minutes at RT, followed by overnight digestion with 0.5 μg trypsin. The samples were cleaned up using beads and C18 tips. The liquid chromatography– mass spectrometry (LC-MS) system consisted of a Thermo Orbitrap Exploris 480 mass spectrometer system with an Easy Spray ion source connected to a Thermo 3 μm C18 Easy Spray column (through pre- column). A total of 10 μL of the extract was injected, and the peptides eluted from the column by an acetonitrile/0.1 M acetic acid gradient at a flow rate of 0.3 μL/min over 2 hours. The nanospray ion source was operated at 1.9 kV. The digest was analyzed using the rapid switch- ing capability of the instrument, acquiring a full scan mass spectrum to determine peptide molecular weights followed by product ion spec- tra (Top10 HCD) to determine the amino acid sequence in sequential ### Page 6 6 of 20 PARDO ET AL.scans. The data was analyzed by database searching using the Sequest search algorithm against Uniprot Human. The peptides and proteins identified for the sample were analyzed in Scaffold 5.3.3 software.2.15 NADK2, P5CS, and PYCR1 gene knockoutThe LentiCRISPRv2 system (Addgene #98290) was used to assess CRISPR-Cas9-mediated gene knockout of the human NADK2 in pro- genitors ReNCell MV cell line. Two human guide RNA sequences were chosen from the CRISPR pick portal (https://portals.broadinstitute. org/gppx/crispick/public). Primer design, ligation, and transformation were followed from the LentiCRISPRv2 and lentiGuide-Puro: lentivi- ral CRISPR/Cas9 and single guide RNA. Then, plasmids were packed into lentiviral particles (as described before) to transduce cells. Human ReNCell MV cell lines were plated in 96-well plates in a confluence of ∼100 cells per well. After 24 hours, cells were transduced with the NADK2 LentiCRISPR-virus. After 48 hours of transduction, complete antibiotic selection (1 μg/ml puromycin) was applied to the genetically modified cells before proceeding to experiments. After 4 weeks, the resistant cells were detached and expanded. After the expansion, a second round of selection was conducted. Then, the knockout human ReNCell MV cells were differentiated for 21 days for each experiment. The efficiency of gene knockout was established by WB before and after the differentiation process. The sgRNA sequences used in this study were sgNADK2-1; TCCTAGGAATGAGGGAATTG, and sgNADK2-2; TGGCATAAACCCTGTACCTG. sgP5CS-1; GAAGCATGCCAA- GAGAATCG and sgP5CS-2; TGCCAATGGAACCCACCCAA. sgPYCR1-1; GAAGTTGACACCCCACAACA and sgPyCR1-2; CGTGCCTGTGGCATACACGG.2.16 Measurement of intracellular proline levelsA previously described protocol 41–43 was adapted as follows. Briefly, cells were harvested and lysed in cold PBS with 1% Triton X-100. The cells were centrifugated to remove cellular debris (10,000 x g). The supernatants were transferred to a new Eppendorf and boiled in a water bath. Intracellular amino acids were extracted by boiling for 10 minutes. After a centrifugation step (5 minutes, 4◦C, 15,000× g), the supernatant was free of proteins, and intracellular proline was determined as described below. Briefly, 100 μL of the supernatant was incubated with 100 μL of acid-ninhydrin (0.25 g ninhydrin dis- solved in 60/40 glacial acetic acid/6 M phosphoric acid) and 100 μL of glacial acetic acid for 1 hour at 100◦C. The reaction was stopped, and samples were incubated on ice for 5 minutes. The mixture was then extracted with 200 μL toluene. Carefully, the toluene phase was separated, and absorbance at 520 nm was used to determine the proline concentration. All the reactions were performed in trip- licate, and a standard curve ranging in concentrations of 0.05–1mm proline was paralleled to determine the proline concentration of samples.41–432.17 Biotin labeling of tau protein and endocytosis assayA previously described protocol was adapted to label oligomeric tau.44,45 EZ-Link Sulfo-NHS-Biotin (2 mg, Thermo-Scientific #A39258) was reconstituted in water to 2mm. TauOs were incubated with the biotin at 1:1 molar ratio (total tau:biotin) for 30 minutes at room tem- perature or 2 h on ice. To remove unbound biotin and enable buffer exchange into PBS, the solution was centrifuged four times at 10,000× g for 15 minutes each in Amicon Ultra-0.5 Centrifugal Filters (Milli- pore, UFC501096), with the addition of PBS to the solution remaining above the filter at each step. Human neurons were then exposed to 300 nM biotinylated RTauO (bt-RTauO). After 1, 3, and 6 hours of incu- bation, bt-RTauO that remained bound to the outer surface of the PM were reduced and removed with 50mm DTT for 1 h at 37◦C. Cells were washed three times with PBS for 15 minutes and lysed under native conditions with hypotonic buffer (10mm Tris HCl, pH 7.4; 1.5mm MgCl2; 10mm KCl). Protein samples were mixed with 1X NuPage LDS Sample Buffer (Invitrogen #NP0007) without reducing agents and ana- lyzed by western blotting using Streptavidin-IRDye800CW (LI-COR #926-32230) to detect biotinylated tau that had been endocytosed.3 REDUCED NADH AND REDUCED NADPH MEASUREMENTS AND in vivo 2P-FLIM IMAGING 3.1 Human neurons 2P-FLIM imaging and processingHuman neuronal cultures were grown in 35-mm glass-bottom dishes, maintained at 37◦C in 5% CO2/95% air on the stage of the Zeiss LSM- 980 NLO microscope. The laser was tuned to 740 nm with an average power of 7 mW at the specimen plane, and NAD(P)H fluorescence was collected using a 450 to 500 nm emission filter (Objective lens, 40 ×). All 2P-FLIM experiments in human neurons included at least 15 cells per image and 10 fields of view per condition (n ≥ 3 independent exper- iments). This approach yielded over 5,000 ROIs per image; data were expressed as the average ROI per field of view. Each field of view was recorded in the descanned mode, and then each field of view was subjected to a 40-second acquisition in the non- descanned mode. The laser power and acquisition time were selected to ensure enough photons per pixel while avoiding photodamage to cells. FLIM images were processed with SPCImage software (v8.6; Becker & Hickl) at maximum-likelihood fitting at three-component lifetimes for human neurons. Based on Blacker et al,46 enzyme-bound NADH and NADPH lifetimes fall into specific τ2 (1400-4400 picoseconds) and τ3 (> 4400 picoseconds); their fractions of a2% and a3% therefore express the relative abundance of the enzyme-bound coenzymes, charted here. All FLIM image parameter data were exported for further processing in Fiji software (https://hpc.nih.gov/apps/Fiji.html). Here, from thresholded, normalized photon images, pixel-ROIs are isolated based on the more prominent intensity of the mitochondrial ### Page 7 PARDO ET AL. 7 of 20morphology, which were applied to all FLIM parameter data, creating a mitochondria-specific data pool. Those results were analyzed using a custom code in Python software, where all parameters were first exam- ined and filtered for obvious outliers (insignificant numbers). The FLIM metrics of interest – NADH% and NAD(P)H%- were extracted as indi- cators for changes in the mitochondrial metabolic OXPHOS state and plotted for publication-ready charts and statistics in Microsoft Excel. Surgical preparation of animals, in vivo 2P-FLIM imaging of live mouse cortex, and Imaging analysis were performed exactly as described by us before.23,34,353.2 StatisticsAll statistical analyses and graphical representations were performed using GraphPad Prism, version 10 (GraphPad Software, Inc.). For bio- chemical experiments, statistical methods and biological replicates (n) are described in the figure legends. For in vitro experiments with human neurons, each biological replicate consisted of paired samples normalization (vehicle control and treatment condition) processed in parallel. To measure changes in protein levels, the intensity of tar- get protein bands was first normalized to the internal loading control (βIII tubulin) within each lane. Fold changes for the treatment condi- tion were then calculated relative to the corresponding vehicle control (assigned a value of 1) within each matched pair. Statistical signifi- cance was assessed using a one-sample t-test comparing the data to the hypothetical value of 1. Protein expression in mouse brain samples was compared pairwise using a Student’s t-test. Values are presented as mean standard error of the mean (SEM), and statistical significance was determined using p-values generated with a 95% confidence inter- val. Individual significance levels for each figure are described in the specific figure legends. For mass spectrometry-based proteomics, proteins were identified in Scaffold 5 using the ‘OLD’ peptide scoring model, requiring a 99% probability threshold and a minimum of 5 peptides. Differential enrich- ment between conditions was determined via Fisher’s Exact Test, with p-values adjusted for multiple testing using the Benjamini-Hochberg method to control the false discovery rate (FDR). For 2P-FLIM analysis, a paired t-test was used across all assays. Because thousands of ROIs were obtained per image, the average ROI for each field of view was calculated to reduce the sample size and, thus, the number of false pos- itives. All 2P-FLIM data were assumed to be normally distributed based on visual inspection of Gaussian histograms. Descriptive statistics (MS Excel) confirmed that the data distribution was nearly symmetrical, falling within the accepted skewness range of -0.5 to 0.5. A p-value of less than 0.05 was considered statistically significant.‘’4 RESULTS 4.1 TauOs upregulate the expression and activity of the NADK2-NADPH pathway in neuronsTo study the effect of TauOs on mitochondrial metabolism in live neurons and mouse cortex, we use label-free, 2P-FLIM to image NAD(P)H lifetimes.34–36 While spectrally NADPH and NADH are alike, their fluorescence lifetimes and fractions are discriminated by 2P- FLIM (Figure 1A),46 allowing measurement of the relative contribution of each coenzyme in live human neuron cultures and mouse brains (Figure 1B). Tracked changes in the coenzymes’ auto-fluorescent life- time fractions of biochemically active, enzyme-bound NADH (NADH%) or NADPH (NADPH%) act as a proxy for mitochondrial respiration for the former,34 and activation of biosynthetic pathways for the latter (Figure 1A–C).15,17 Human neurons derived from the ReNCell™ MV line of human neuronal progenitors34,35 (hereafter human neurons) were vehicle- treated (control) or exposed for 24 hours to 100 nM of bacteri- ally produced recombinant 2N4R human tau that was assembled into oligomers (RTauOs).39 Under control conditions, NADPH% con- tributed ∼35% of the total fraction, while NADH% accounted for∼65% (Figure 1D). Interestingly, RTauOs treatment led to a ∼31% increase in NADPH% and the corresponding decrease in NADH% in WT neurons (Figure 1D). Additionally, RTauOs also upregulated the expression of NADK2 by ∼30% (Figure 1E), which agrees with the fact that NADPH availability correlates with NAD+ kinase abundance.47 Reciprocally, the RTauOs-mediated increase in NADPH% was reduced by ∼22.5% in NADK2 knocked-down neurons (Figures 1D and S1A), further indicating the specificity of our 2P-FLIM assay. Additionally, RTauOs-mediated imbalance in mitochondrial NADH and NADPH con- tent leads to a 20% decrease in total ATP, likely because of the reduction in the availability of NADH to feed Complex I of the electron transport chain (Figure 1F). NADK2-dependent NADPH regulates mitochondrial proline biosynthesis 15,17 (Figure 1C), an essential process needed to sup- port protein synthesis, cell growth, and proliferation. In addition, dysregulated mitochondrial-derived proline has been associated with cancer cell survival and collagen deposition.48,49 Specific enzymes catalyze mitochondrial proline biosynthesis in a two- step process. First, P5C synthase (P5CS) converts glutamate to glutamic-γ-semialdehyde (GSA), which spontaneously cyclizes to formΔ1-pyrroline-5-carboxylate (P5C). This step is accompanied by the transfer of reducing power from NADPH. Then, P5C is catalytically converted by P5C-reductase (PYCR1) into L-proline, followed by the transfer of reducing potential from NADH or NADPH50 (Figure 1C). These steps affect the mitochondrial content of NADH and NADPH as well as the associated biological processes regulated by them.51 We observed that RTauOs not only increased NADPH% and NADK2 (Figure 1D and E) but also upregulated the expression of P5CS and PYCR1 (Figure 1E). Correspondingly, it also increased the neuronal levels of the two downstream products, proline and the proline-rich protein collagen-VIa (ColVI) (Figure 1G and H). These results were not observed in NADK2, P5CS, or PYCR1-knockdown human neurons (Figures 1G, H, and S1A–C). Importantly, RTauOs did not affect the expression levels of PRODH and P5CDH, which regulate proline catabolism in the mitochondria (Figure 1I).15,17 Importantly, RTauOs did not change the expression of NADK1, the kinase responsible for cytosolic de novo synthesis of NADPH (Figure 2A). Furthermore, RTauO-mediated upregulation of NADK2, proline, and ColVI syn- thesis was unaffected in NADK1 knocked-down human neurons ### Page 8 8 of 20 PARDO ET AL.(A) (C) (D) (E) (H) (I) (F) (G) (B)F I G U R E 1 Tau oligomers (TauOs) upregulate the expression of mitochondrial nicotinamide adenine dinucleotide phosphate (NADPH) and the proline synthesis pathway in a mitochondrial NAD kinase 2 (NADK2) -dependent manner. (A) A simplified illustration showing NAD(P)H photophysical features. The reduced forms of NADPH and NADH [NAD(P)H] are autofluorescent and spectrally identical; their absorption and emission both peak at 740 nm and 460 nm, respectively. However, the fluorescent lifetime can be discriminated by two-photon fluorescence lifetime microscopy (2P-FLIM). This assay allows label-free imaging of mitochondrial activity in live specimens by tracking the differential lifetime of free coenzymes τ1 (∼400 ps) and their fraction NAD(P)H%, enzyme-bound NADH-τ2 (∼1500-4400 ps) and its fraction NADH%, and enzyme-bound NADPH-τ3 (> 4400 ps) and its fraction NADPH%. (B) 2P-FLIM detects the fluorescent lifetimes of the mitochondrial-enriched cofactors, biochemically active enzyme-bound NADH and NADPH in both live neuron cultures and live mouse brains. As described in the Methods section, the images represent the analysis sequence, starting with the collection of the 2P images, continuing with region of interest (ROI) ### Page 9 PARDO ET AL. 9 of 20(Figures 2B–D and S1D). Thus, soluble TauOs specifically activates the mitochondrial NADK2-NADPH-proline metabolic pathway.4.2 NADK2 controls LRP1 expression and TauO uptakeHaving demonstrated that TauOs upregulates NADPH metabolism via NADK2, we then asked how this event contributes to TauOs toxic- ity. To address this question, we performed quantitative protein mass spectrometry (MS) of CRISPR-Cas9-mediated depletion of NADK2 (NADK2-KO; Figure S1F) neurons. These cells were able to differ- entiate into neurons as well as the isogenic WTs (Figure S2A). The analysis identified a total of 371 proteins (Figure 3A) whose expres- sion was significantly affected by NADK2 expression (Supplemental file; p < 0.01299), including 188 proteins whose expression were sig- nificantly reduced in NADK2-KO neurons (Figure 3A). It has been pre- viously reported that the expression levels of two NADPH-dependent enzymes, glutamate dehydrogenase 1 (GLUD1) and isocitrate dehy- drogenase 2 (IDH2), which fuel the tricarboxylic acid (TCA) cycle and support energy production, were significantly affected in sev- eral NADK2-knockout cancer cell lines.15,52 Our proteomic analysis of NADK2-KO human neurons aligns with the prior report, reveal- ing a significant decrease in expression levels of GLUD1 (fold of change = 0.3, p = 0.0054) and IDH2 (fold of change = 0.5, p < 0.00010), as depicted in the volcano plot (Figure 3A). Then, to gain insights into the molecular mechanisms associated with the NADK2 function, we conducted gene ontology (GO) analysis of biological processes, molecular functions, and cellular components using the NADK2-KO human neurons proteomics dataset, which com- prises proteins significantly affected (p < 0.05) by the NADK2-KO neurons. Results showed that proteomic hits were primarily associated with mitochondrial cristae, mitochondrial dynamics, mitochondrial- associated membrane proteins (MAMs), cell surface receptors, elec- tron transport chain (ETC), proline synthesis pathway, and membrane trafficking (Figure S2B and Supplemental file). Within the membrane trafficking group, we found that the expression of the low-density LRP1, a major tau receptor at the PM,5 was reduced by 50% (fold of change = 0.5, p < 0.00010) in NADK2-KO neurons (Figure 3A, Sup- plementary Supplemental file). Importantly, no other receptors known to be involved in TauOs internalization were identified. Consistent with our previous observation that RTauOs upregulate both NADPH% (Figure 1D) and NADK2 expression (Figure 1E) along with proline and ColVI biosynthesis (Figure 1G and H), RTauO treatment also increased LRP1 expression by more than 50% in WT but not in human neurons deficient in either NADK2, P5CS, or PYCR1 enzymes (Figures 3B and S1E–G). In agreement with the role of LRP1 in regulating TauO uptake,5 the typical RTauO-mediated increase in mitochondrial NADPH%, pro- line, and ColVI biosynthesis was not observed in LRP1-KD neurons (Figures 3C–E and S1H). Thus, TauO-mediated upregulation of the NADK2-NADPH pathway requires LRP1 expression. How TauOs impact physiological cellular functions is a matter of intense investigation; the critical step is the TauO uptake by cells. This process occurs by several mechanisms, including receptor-mediated endocytosis and pinocytosis.5,31,32 However, the exact mechanisms regulating the availability of TauOs’ receptors at the PM remain unknown. Since TauOs control the expression of LRP1 in a NADK2- dependent manner (Figure 3B), we hypothesize that NADK2 regu- selection, and ultimately extracting the NADH% and NADPH% fractions (known as “a2%” in FLIM jargon). Scale bar, 20 μm. (C) A simplified cartoon depicting metabolic roles of NADH and NADPH in mitochondria. NADH mainly drives ATP synthesis through mitochondrial oxidative phosphorylation, and NADPH mostly supports reductive pathways, including proline biosynthesis through NADK2-dependent de novo synthesis of NADPH. (D) Recombinant tau oligomers increase NADPH% (enzyme-bound NADPH) in human neurons in a NADK2-dependent manner. Representative 2P-FLIM images of wild-type (WT) or NADK2 knockdown (NADK2-KD, Mission shRNA library from Sigma-Aldrich) neurons treated with a vehicle (control) or oligomers assembled from 100 nM of bacterially produced recombinant 2N4R human tau oligomers (RTauOs) for 24 hours. An RTauOs-mediated increase in NADPH was significantly reduced in NADK2-KD neurons. Scale bar, 20 μm. More than 5,000 ROIs were obtained per image, and the average ROI was calculated for each field of view. Statistical analyses were performed using a two-tailed unpaired Student’s t-test. Error bars represent ± SEM (n = 3). (E) WT human neuron cultures were treated with vehicle (control) or 100 nM of RTauOs for 24 hours. NADK2 (n = 5), P5CS (n = 4), and PYCR1 (n = 5) levels were measured by Western blot (WB) and normalized to the expression level of βIII-tubulin contained in the same samples. Fold changes were calculated for each treatment relative to its matched vehicle control (set to 1). Statistical significance was determined by a one-sample t-test against a baseline of 1. Error bars represent ± SEM. (F) ATP levels were measured in WT human neurons treated with vehicle (control) or 100 nM of RTauOs for 24 h using the ATPlite 1-step Luminescence Assay System (Perkin-Elmer). The data were collected from three independent assays, each containing three replicates per experimental condition. Error bars represent ± SEM. Statistical analyses were performed using a two-tailed unpaired Student’s t-test. (G) Intracellular proline levels were measured using the acid-ninhydrin assay in either WT, NADK2-KD, P5CS-KD, or PYCR1-KD (Mission shRNA library from Sigma-Aldrich) neurons treated with vehicle (control) or 100 nM of RTauOs for 24 hours. Fold changes were calculated for each treatment relative to its matched vehicle control (set to 1). Statistical significance was determined by a one-sample t-test against a baseline of 1. Error bars represent ± SEM (n = 6). (H) Collagen-VIa (ColVI) levels, another read-out for proline synthesis pathway activity, were analyzed in either WT, NADK2-KD, P5CS-KD, or PYCR1-KD (Mission shRNA library from Sigma-Aldrich) neurons treated with vehicle (control) or 100 nM of RTauOs for 24 hours. ColVI expression detected by WB was normalized to the expression of βIII-tubulin contained in the same samples. Fold changes were calculated for each treatment relative to its matched vehicle control (set to 1). Statistical significance was determined by a one-sample t-test against a baseline of 1. Error bars represent ± SEM (n = 4). (I) TauOs do not change the expression of enzymes required for proline catabolism. WT human neuron cultures were treated with a vehicle (control) or 100 nM of RTauOs for 24 hours. PRODH and P5CDH levels were measured by WB and normalized to the expression of βIII-tubulin contained in the same samples. Fold changes were calculated for each treatment relative to its matched vehicle control (set to 1). Statistical significance was determined by a one-sample t-test against a baseline of 1. Error bars represent ± SEM (n = 5). ### Page 10 10 of 20 PARDO ET AL.ColVI III-tubulin NADK1-KD ColVI III-tubulin 0.0 0.5 1.0 1.5 2.0 - + RTauOs p=0.039 p=0.0252 0.0 0.5 1.0 1.5 2.0 p=0.0176 p=0.0042 WT ColVI/ III-tubulin Fold Change(B) (C)0.0 0.2 0.5 0.7 1.0 1.2 1.5 NADK1/ III-tubulin Fold Change NADK1 III-tubulin RTauOs - +(A) NADK1 WT Control RTauO NADK1-KD WT Control RTauOs NADK1-KD WT(D)III-tubulin NADK1-KD - + RTauOs WT - + RTauOs 0.0 0.5 1.0 1.5 NADK2 NADK1 NADK1-KDWT RTauOs Control NADK2/ III-tubulin Fold Change p=0.0072p=0.0192 Control RTauOs ~50 kDa ~40 kDa ~50 kDa Intracellular proline synthesis fold change n.s.F I G U R E 2 Tau oligomer (TauO) -mediated effects on the mitochondrial NAD kinase 2 (NADK2) are independent of NADK1. (A) NADK1 expression was assessed by Western blot (WB) in protein extracts from WT human neurons treated with either vehicle (control) or 100 nM of RTauOs for 24 hours and normalized to the expression level of βIII-tubulin contained in the same samples. Fold changes were calculated for each treatment relative to its matched vehicle control (set to 1). Statistical significance was determined by a one-sample t-test against a baseline of 1. Error bars represent ± SEM (n = 5) (B) Wild-type (WT) and NADK1 knocked down human neurons were treated with vehicle (control) or 100 nM of RTauOs for 24 hours. NADK2 and NADK1 levels were measured by WB and normalized to the expression level of βIII-tubulin contained in the same samples. Fold changes were calculated for each treatment relative to its matched vehicle control (set to 1). Statistical significance was determined by a one-sample t-test against a baseline of 1. Error bars represent ± SEM (n = 3) (C) Intracellular proline levels (n = 4) and (D) ColVI synthesis (n = 4) were monitored in WT or NADK1 knocked-down neurons treated with RTauOs for 24 h, as described before. In both cases, fold changes were calculated for each treatment relative to its matched vehicle control (set to 1). Statistical significance was determined by a one-sample t-test against a baseline of 1. lates the uptake of TauOs by neurons. We performed an endocyto- sis assay using RTauOs labeled with EZ-Link™-Sulfo-NHS-SS-Biotin, which allows reversible labeling of primary amines on proteins. Human neurons were exposed to biotinylated RTauOs (bt-RTauOs) for differ- ent time periods. Before sample harvest, biotin on non-internalized bt-RTauOs was removed by a buffer containing DTT, as internalized bt-RTauOs were protected from the reduction (Figure 3F). While the incorporation of bt-RTauOs showed a steady uptake in WT neurons in a 6-hour time course, its uptake in NADK2-KO cells was reduced by ∼50% (Figure 3G). Knocking down NADK1 expression in human neurons did not affect bt-RTauO uptake (Figure 3H). Interestingly, upregulating the expression of LRP1 (by treating neurons with 100 nM of RTauOs; Figure 3B) for 24 hours prior to the internalization exper- iment further increased the uptake of bt-RTauOs in WT neurons by 25%, but not in NADK2-KO neurons (Figure 3I). Thus, TauOs initi- ate a vicious cycle of uptake by upregulating LRP1 at the PM in a NADK2-dependent NADPH manner.4.3 TauOs upregulate NADK2, NADPH, and LRP1 in human ADTo explore the relevance of these observations to human AD, we treated human neurons with 100 nM of TauOs purified from human AD brains or brain-derived TauOs (BDTauOs) for 24 hours. Like our results with RTauOs (Figure 1E), BDTauO treatment also increased the expression of both NADK2 (Figure 4A) and LRP1, the latter in an NADK2-dependent manner (Figure 4B). Although BDTauOs showed a seemingly lower potency on the expression of NADK2, it delivered a robust, NADK2-dependent increase in NADPH% (Figure 4C). Addi- tionally, we explored NADK2 expression levels in human brain cell extracts taken from deceased patients affected by AD and compared to age-matched, cognitively normal donors. Though a small sample size, we observed a trend toward increasing NADK2 levels in AD brain (Figure S3A–C). Consistent with these observations, large-scale pro- teomic studies analyzing brain and CSF samples from AD patients have ### Page 11 PARDO ET AL. 11 of 20(D) (E)p=0.0192 WT LRP1-KD 0.0 0.5 1.0 1.5 2.0 2.5 p= 0.0013- +ColVI ColVI LRP1-KD RTauOs 0.0 0.5 1.0 1.5 2.0 ColVI/ III-tubulin Fold Change p=0.0137 p< 0.0001 WT III-tubulin III-tubulin Control RTauOs Control RTauOs WT LRP1-KD(F)Streptavidin III-tubulin 1h 3h 6h WTNADK2-KO Streptavidin III-tubulin 1h 3h 6h Streptavidin/ III-tubulin 1h 3h 6h 0.0 0.5 1.0 1.5 2.0 2.5 p=0.0154 bt-RTauOs bt-RTauOs WTNADK2-KO WT NADK2-KO p=0.0003 Streptavidin/ III-tubulin 1h 3h 6h 0.0 0.5 1.0 1.5 2.0 2.5(G)Internalization Cleavage Addition bt-RTauOs LRP1 Reducing agent(H)Streptavidin III-tubulin Streptavidin II-tubulin Pre-treatment WT NADK2-KO(C) (A)III-tubulin LRP1 (~85 kDa) III-tubulin III-tubulin LRP1 III-tubulin LRP1 LRP1/ III-tubulin Fold Change p=0.0422- +RTauOs P5CS-KO PYCR1-KO NADK2-KO WT Control RTauOs P5CS-KO PYCR1-KO NADK2-KO WT LRP1 -7.5 -5.0 -2.5 0.0 2.5 5.0 7.5 Log2(Fold Change) 0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 -Log10 (adj. p-value) LRP1 Upregulated proteins Downregulated proteins NADK2-KO(B)IDH2 GLUD1 NADK1-KD WT NADK1-KD WT 1h 3h 6h 0.0 0.5 1.0 1.5 2.0 2.5 Streptavidin III-tubulin Streptavidin III-tubulin 1h 3h 6h bt-RTauOs Streptavidin/ III-tubulin(I)WT LRP1-KD RTauOs WT 35 46 40 0 5 10 15 20 25 30 35 40 45 50 65 54 60 0 10 20 30 40 50 60 70 WT RTauOs LRP1-KDWT WT RTauOs LRP1-KDWT NADPH% NADH% 20 40 a2% p<0.0001p<0.0001 p<0.0001 p<0.0001 Intracellular proline synthesis fold change p<0.0001 p<0.0001F I G U R E 3 Mitochondrial NAD kinase 2 (NADK2) controls low-density lipoprotein receptor-related protein 1 (LRP1) expression and tau oligomer (TauO) endocytosis. (A) Proteomic analysis from postnuclear membrane fractions obtained from wild-type (WT) or CRISPR-Cas9-mediated depletion of NADK2 human neurons (NADK2-KO) measured by LC/MS/MS. Changes in protein expression were determined using Scaffold software (v5.3) (Fisher’s exact test, p < 0.01, Benjamini–Hochberg). Downregulated (blue) and upregulated (red) proteins in NADK2-KO neurons are represented in the volcano plot (detailed in Supplemental file). (B) LRP1 expression was analyzed by Western ### Page 12 12 of 20 PARDO ET AL.consistently reported that NADK2 is upregulated at various stages of disease progression.53–55 The generation of human-derived adult neurons (iNeurons) through direct conversion of the adult dermal fibroblasts has facilitated the exploration of molecular mechanisms underlying age-related human diseases such as AD.56 We therefore asked whether dysregulation in the NADK2-NADPH pathway could be recapitulated in iNeurons. Our protocol produced iNeurons from healthy or AD age/sex-matched donors that readily expressed neuronal markers, including NeuN,βIII-tubulin, synaptophysin, and Tau (Figure S4A–C). Interestingly, iNeurons obtained from AD patients displayed ∼6-10% higher basal levels of NADPH% compared to iNeurons from cognitively normal con- trols (Figure 4E). Like our observations with RTauOs and BDTauOs (Figures 1D and 4C), higher NADPH% values in AD iNeurons paral- leled an upregulation in the expression of NADK2, P5CS, PYCR1, and LRP1 (Figure 4F and G) and, strikingly, including the detection of T22 antibody28-positive tau oligomers (Figure S4C). Thus, upregulation of the NADK2-NADPH-proline synthesis pathway might represent a novel human AD phenotype.4.4 TauO-mediated upregulation of the NADK2-NADPH pathway occurs at presymptomatic stages of ADPathological tau has been detected in CSF and plasma of human patients more than 10 years before the onset of cognitive decline,11 suggesting that soluble forms of tau could impair mitochondrial metabolism years before disease onset. We took advantage of PS19 mice, which naturally produce and accumulate TauOs in the brain months before synaptic abnormalities and cognitive deficits occur.57 Using 2P-FLIM, we recorded the energy metabolic state in the cortex of live PS19 and WT littermate mice at 1 and 3 months of age (presymp- tomatic), which allows measuring mitochondrial activity in vivo by tracking NADH% and NADPH% through an open-skull window.35 Compared to WT littermates, PS19 mice showed a time-dependent increase of NADPH% of ∼16% at 1 month of age and ∼25% at 3 months of age (Figure 5A). The early rise of NADPH% in presymptomatic PS19 mice paralleled the upregulation of NADK2, P5CS, PYCR1, and LRP1 expression in the cortex of these animals (Figure 5B–E), and with an early detection of T22 antibody-positive tau oligomers detected by western blot in non-reducing conditions (Figure 5F). Next, we asked whether reducing NADK2 expression would ameliorate (at least in part) tau-mediated pathology in PS19 live mouse brains compared to that observed in WT mice. Two month old PS19 mice were injected ICV with AAV9s encoding either a scramble shRNA sequence (PS19-Scr) or NADK2 (PS19-NADK2-KD) shRNA. Five weeks later, transduced neu- rons were visualized by the expression of a blue-fluorescent reporter (BFP2), as demonstrated by its co-distribution with βIII-tubulin (Figure S5A). Reducing NADK2 expression in PS19 brain by ∼25% (Figure S5B) led to a 15% decrease in NADPH% (Figure 6A), a ∼29% decrease in LRP1 expression (Figure 6B) and importantly, a ∼40% reduction in the expression of T22-positive tau oligomers (Figure 6C) without apparent changes in total human tau expression (Figure 6D). In addition, we also observed a corresponding reduction in the expression levels of P5CS and PYCR1 (Figure S5C and D). Early upregulation of the NADK2-NADPH pathway is detected as early as 1-3 months of age in the PS19 mouse cortex, coinciding with blot (WB) of protein samples from human neurons treated for 24 h with vehicle (control) or 100 nM RTauOs. The results were compared to either NADK2-KO, P5CS-KO, or PYCR1-KO neurons. LRP1 expression level was normalized to the expression level of βIII-tubulin contained in the same samples. Fold changes were calculated for each treatment relative to its matched vehicle control (set to 1). Statistical significance was determined by a one-sample t-test against a baseline of 1. Error bars represent ± SEM (n = 4). (C) TauO-mediated increase in mitochondrial nicotinamide adenine dinucleotide phosphate % (NADPH%), intracellular proline, and collagen synthesis requires LRP1 expression. Representative two-photon fluorescence lifetime imaging (2P-FLIM) images of WT or LRP1-KD neurons were treated with vehicle (control) or 100 nM of RTauOs for 24 h, and 2P-FLIM imaging was used to analyze nicotinamide adenine dinucleotide + hydrogen % (NADH%) and NADPH%. Scale bars, 10 μm. As more than 5,000 regions of interest (ROIs) were obtained per image, the average ROI was calculated for each field of view. Error bars represent ± SEM (n = 3). (D) Intracellular proline levels were measured using the acid-ninhydrin assay in WT human neurons treated with a vehicle (control) or 100 nM of RTauOs for 24 h, and the results were compared to LRP1-KD neurons. Fold changes were calculated for each treatment relative to its matched vehicle control (set to 1). Statistical significance was determined by a one-sample t-test against a baseline of 1. Error bars represent ± SEM (n = 5) (E) In a similar experiment, ColVI synthesis was monitored by WB and normalized to the expression level of βIII-tubulin contained in the same samples. Fold changes were calculated for each treatment relative to its matched vehicle control (set to 1). Statistical significance was determined by a one-sample t-test against a baseline of 1. Error bars represent ± SEM (n = 4). (F) Scheme showing the internalization protocol of biotin-tagged RTauOs (bt-RTauOs). (G) NADK2 controls TauO uptake by neurons. 300 nM of biotinylated tau oligomers (bt-RTauOs) were added to either WT or NADK2-KO neuron cultures at 37◦C for the indicated time points. Then, the internalized (DTT-resistant) bt-RTauOs were separated electrophoretically in non-reducing conditions, followed by WB using iRDye800-labeled streptavidin (LICOR). While the internalization of bt-RTauOs in WT neurons increased by ∼50% in a 6-h time course, the corresponding incorporation in NADK2-KO neurons remained steady. Line graphs show the quantification at the 6-h time-point obtained in three independent experiments and analyzed using a two-tailed unpaired Student’s t-test. Internalized bt-TauOs was normalized to the expression level of βIII-tubulin contained in the same samples. Error bars represent ± SEM. (H) An endocytosis assay of bt-RTauOs in WT and NADK1 knocked-down neurons was performed as described in F and G. (I) TauOs stimulated their own uptake by a NADK2-dependent mechanism. WT and NADK2-KO neurons were treated with 100 nM RTauOs for 24 hours before performing the bt-RTauO uptake assay, as described in (F). The internalized bt-TauOs were separated electrophoretically under native conditions, followed by WB, and were normalized to the expression level of βIII-tubulin contained in the same samples. Lines in each graph show the quantification obtained from three independent experiments and analyzed using a two-tailed unpaired Student’s t-test. Error bars represent ± SEM (n = 3). ### Page 13 PARDO ET AL. 13 of 20(D) (A)Control LRP1 LRP1 BDTauOs LRP1/ III-tubulin Fold Change p=0.05 0.0 0.5 1.0 1.5 2.0- +NADK2-KO WT WT NADK2-KO III-tubulin III-tubulin p=0.0472 BDTauOs(B)III-tubulin NADK2- +BDTauOs NADK2/ III-tubulin Fold ChangeControl BDTauO p=0.0024 0 0.5 1.0 1.5(F) p=0.0219 LRP1 III-tubulin 0.0 0.5 1.0 1.5 2.0 2.5 LRP1/ III-tubulin Fold Change ADN ADN(E)6N 4AD 45 48 0 10 20 30 40 50 60 55 52 0 10 20 30 40 50 60 NADPH% N AD N AD p<0.0001 p<0.0001 NADH% 20 40 a2% P5CS III-tubulin NADK2 ADN III-tubulin III-tubulin PYCR1 Enzyme/ III-tubulin Fold Change p=0.0009 p=0.0388 p=0.0493 N AD P5CS PYCR1 NADK2 0 0.5 1.0 1.5 2.0 2.5 iNeurons(C)NADK2-KO BDTauOs WT WT 39 51 44 0 10 20 30 40 50 60 61 49 56 0 10 20 30 40 50 60 70 WT BDTauOs NADK2-KOWT WT BDTauOs NADK2-KOWT 20 40 a2% NADPH% NADH% p<0.0001 p<0.0001 p<0.0001 p<0.0001F I G U R E 4 The mitochondrial mitochondrial NAD kinase 2–nicotinamide adenine dinucleotide phosphate (NADK2-NADPH) pathway is upregulated in human Alzheimer’s disease (AD) models. (A) Wild-type (WT) human neuron cultures were treated with vehicle (control) or 100 nM of tau oligomers purified from human AD brain (brain-derived tau oligomers [BDTauOs]) for 24 h. NADK2 levels were measured by Western blot (WB) and normalized to the expression level of βIII-tubulin contained in the same samples. Fold changes were calculated for each treatment relative to its matched vehicle control (set to 1). Statistical significance was determined by a one-sample t-test against a baseline of 1. Error bars represent ± SEM (n = 3). (B) Low-density lipoprotein receptor-related protein 1 (LRP1) expression levels were analyzed in WT neurons as described above and compared to NADK2-KO neurons (n = 3). (C) Representative two-photon fluorescence lifetime imaging (2P-FLIM) images of WT or NADK2-KO human neurons treated with vehicle (control) or 100 nM of BDTauOs for 24 h. 2P-FLIM was used to monitor changes in NADH% and NADPH%. BDTauOs-mediated increase in NADPH was significantly reduced in NADK2-KO neurons. Scale bars, 20 μm. As more than 5,000 regoins of interest (ROIs) were obtained per image, the average ROI was calculated for each field of view. Statistical analyses were performed using a two-tailed unpaired Student’s t-test. Error bars represent ± SEM (n = 3). (D) The experiments were performed with adult human neurons obtained by direct conversion of dermal fibroblasts (iNeurons) from three AD and three age- and sex-matched cognitively normal donors. Representative 2P-FLIM images of human iNeurons from a cognitively normal control (N, patient #6; male, 76 years old) or AD (AD, patient #4; male, 76 years old) donor are shown. Scale bars, 10 μm. Statistical analyses were performed using a two-tailed unpaired Student’s t-test. Each treatment was compared to its corresponding control condition. Error bars represent ± SEM. (E) NADK2, P5CS, and PYCR1 or (F) LRP1 expression levels in healthy and AD iNeurons were analyzed by WB against the indicated proteins. The expression levels of the indicated markers were normalized against βIII-tubulin contained in the same samples. In E and F, fold changes were calculated for each treatment relative to its matched vehicle control (set to 1). Statistical significance was determined by a one-sample t-test against a baseline of 1. Error bars represent ± SEM (n = 3). ### Page 14 14 of 20 PARDO ET AL.1 2 3 1 2 3 WT PS19 1 2 3 1 2 3 WT PS19 III-tubulin 1 Mo 3 Mo 250 150 100 75 50 (kDa) T22(F)3 Mo WTPS19 1 Mo Cranial Window(A)10 40a2%(B)1 2 3 1 2 3 WT PS19 NADK2 III-tubulin NADK2 III-tubulin p=0.0261 p=0.0288 NADK2/ III-tubulin (relative units) 0.5 1.0 1.5 2.0 1 Mo 3 Mo 1 Mo 3 Mo WT PS19(E)0.0 0.5 1.0 1.5 2.0 p=0.0147 n.s. LRP1/ III-tubulin (relative units) 1 2 3 1 2 3 WT PS19 LRP1 III-tubulin LRP1 III-tubulin 1 Mo 3 Mo P5CS III-tubulin 0.0 0.2 0.4 0.6 0.8 1.0 P5CS/ III-tubulin (relative units) p=0.03 p=0.0382 P5CS III-tubulin(C)1 2 3 1 2 3 WT PS19 PYCR1 III-tubulin PYCR1/ III-tubulin (relative units) n.s. 0.0 0.5 1.0 1.5 2.0 p=0.0336 PYCR1 III-tubulin(D)1 2 3 1 2 3 WT PS19 1 Mo 3 Mo 1 Mo 3 Mo WT PS19 WT PS19 WT PS19 1 Mo 3 Mo 1 Mo 3 Mo 1 Mo 3 Mo56 49 44 51 0 20 40 60 80 100 120 WT PS19 64 55 36 45 WT PS191 Mo 3 Mo NAD(P)H % NADH% NADPH% p<0.0001p<0.0001F I G U R E 5 Human tau oligomers (TauOs) increase the expression of the mitochondrial NAD kinase 2–nicotinamide adenine dinucleotide phosphate (NADK2-NADPH) pathway in live mouse brains. (A) Representative two-photon fluorescence lifetime imaging (2P-FLIM) imaging in WT and presymptomatic (1 and 3 months old) PS19 mouse cortex through an open-skull window. An increase in NADPH% was observed in the PS19 mouse cortex, likely indicating upregulation in mitochondrial NADK2 activity. As more than 5,000 regions of interest (ROIs) were obtained per image, the average ROI was calculated for each field of view. Statistical analyses were performed using a two-tailed unpaired Student’s t-test. Three animals/age/groups/genetic backgrounds were used for recording NADH% and NADPH%. Error bars represent ± SEM. (B–E) NADK2, P5CS, PYCR1, and LRP1 protein levels are upregulated in presymptomatic PS19 mice. Cortical protein samples from WT or PS19 brains were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE, followed by Western blot (WB) of the indicated proteins. The expression levels of NADK2, P5CS, PYCR1, and LRP1 were normalized against βIII-tubulin detected in each sample. Bar graphs show the quantification obtained using three mice/group/genetic background and analyzed using a two-tailed unpaired Student’s t-test. Error bars represent ± SEM. (F) Upregulation of the NADK2-NADPH pathway in the cortex of presymptomatic PS19 mice correlated with the early detection of tau oligomers. Cortical protein samples from WT or PS19 brains were electrophoretically separated under non-denaturing conditions and followed by WB using T22 antibody-positive TauOs to identify the presence of high molecular weight tau species. Three mice/group/genetic background were used. ### Page 15 PARDO ET AL. 15 of 20PS19Cranial Window Scr NADK2-KD 60 66 40 34 0 20 40 60 80 100 120 PS19 p=0.0102Scr NADK2-KD(C) (D) (A) (B) (F)(E)1 2 3 1 2 3 Scr NADK2-KD III-tubulin 250 150 100 75 50 (kDa) T22 0 2 4 6 8 10 p=0.0458 T22/ III-tubulin (relative units) Scr NADK2-KD LRP1 III-tubulin 1 2 3 1 2 3 Scr NADK2-KD LRP1/ III-tubulin (relative units) p=0.0449 Scr NADK2-KD 0.0 0.1 0.2 0.3 0.4 PS19 PS19 NADH% NADPH%ICV injectionAT8 III-tubulin 1 2 3 1 2 3 Scr NADK2-KD AT8/ III-tubulin (relative units) Scr NADK2-KD PS19 250 150 100 75 50 (kDa) AT180 III-tubulin 1 2 3 1 2 3 Scr NADK2-KD AT180/ III-tubulin (relative units) Scr NADK2-KD PS19 250 150 100 75 50 (kDa) 0.0 0.2 0.4 0.6 p=0.0032 0.0 0.2 0.4 0.6 0.8 1.0 p=0.0253 HT7 III tubulin 1 2 3 1 2 3 Scr NADK2-KD HT7/ III-tubulin (relative units) PS19 250 150 100 75 50 (kDa) 10 40 a2% 0 5 10 15 20 25 p=0.2035 Scr NADK2-KD NAD(P)H %F I G U R E 6 Restoring neuronal mitochondrial NAD kinase 2 (NADK2) expression levels lowers tau oligomer (TauO) -mediated toxicity in PS19 mice. (A) Polycistronic AVVs expressing blue fluorescent protein (BFP) and a shRNA against NADK2 or AAVs carrying a scramble shRNA (Scr) were injected into the left ventricle of two months old PS19 mice. Five weeks later, two-photon fluorescence lifetime imaging (2P-FILM) and protein analysis were performed on the contralateral side. Representative 2P-FLIM imaging in the PS19 mouse cortex through an open-skull window shows that shNADK2-PS19 has a ∼20% decrease in nicotinamide adenine dinucleotide phosphate % (NADPH%) compared to Scr control PS19. We analyzed ∼100K region of interest (ROI)/condition/image, which are averaged to reduce the sample size (see refs35 for full details). Statistical analyses were performed using a two-tailed unpaired Student’s t-test. Three or four animals/groups were used. Error bars represent ± SEM. (B) Similar to the experiment described in A, cortical protein samples from Scr or NADK2-KD PS19 brains were electrophoretically separated and followed by Western blot (WB) using the LRP1 antibodies. Each treatment was compared to its corresponding control conditions. Three mice/group/genetic background were used and analyzed using a two-tailed unpaired Student’s t-test. Error bars represent ± EM. (C) Cortical protein samples from Scr or NADK2-KD PS19 brains were electrophoretically separated under non-denaturing conditions and followed by WB using T22 antibody-positive tau oligomers to detect the expression of high molecular weight tau species. Reducing NADK2 expression led to a∼20% decrease in the expression of TauOs. Each treatment was compared to its corresponding control conditions. Three mice/group/genetic background were used and analyzed using a two-tailed unpaired Student’s t-test. Error bars represent ± SEM. (D) Cortical protein samples from ### Page 16 16 of 20 PARDO ET AL.the first detection of TauOs (Figure 5A and F), and the emergence of synaptic or neuronal loss in this transgenic mouse.57,58 We inves- tigated whether restoring NADK2 expression levels at the onset of synaptic dysfunction in the 3-month-old PS19 brain could improve tau pathology in these mice. Our results indicate that, consistent with the decrease in tau oligomers observed in the PS19-NADK2-KD mice (Figure 6C), well-known AD-associated tau phosphorylation sites, serine-202/threonine-205 (AT8), and threonine-231 (AT180), were also reduced in these mice (Figure 6E and F). Moreover, postsynaptic markers such as PSD95, NMDA receptor, and MAP2, were significantly increased in NADK2-knockdown PS19 mice (Figure S5E–G), along with a notable reduction in GFAP (Figure S5H), suggesting decreased astrogliosis. However, no differences were detected in the presynap- tic proteins synapsin-1 and synaptophysin, suggesting the existence of mitochondria-independent regulatory mechanisms controlling their expression (Figure S5I and J). Overall, these results indicate that TauOs regulate their own recep- tor’s expression at the PM through a mechanism involving upregulation of the mitochondrial NADK2-NADPH pathway. This suggests that TauOs exert toxicity, at least in part, via mitochondrial NADK2-NADPH long before symptom onset.5 DISCUSSIONUsing 2P-FLIM imaging of the mitochondrial-enriched coenzymes, NADH and NADPH,36 we now report that TauOs upregulates the expression and activity of the highly conserved mitochondrial NADK2- NADPH-Proline pathway.59 This process was observed in several models, including live neuronal progenitor-derived human neurons (Figure 1), patient-derived AD adult iNeurons (Figure 4D–F), and presymptomatic PS19 mouse brain (Figure 5). Notably, we found this process to control the expression of LRP1, a main TauO receptor at the PM, thereby establishing a vicious cycle for further TauO internal- ization. As the latter process represents a key step in the propagation of toxic tau through the brain,5 our results suggest that TauOs control their own uptake by upregulating mitochondrial NADK2 (Figure 7). NAD+ and NADP+ coenzymes and their reduced forms (NADH and NADPH) act as cofactors and/or substrates for numerous enzymes to maintain cellular redox balance and energy metabolism.60 NADK2 is the mitochondrial kinase responsible for de novo NADP+ synthesis,61 enabling the synthesis of proline and collagen, which are major play- ers in cell growth, proliferation,15,17 and cell survival.48,49 Interestingly, loss-of-function mutations in NADK2 have been shown to have pro- found neurological consequences in humans,62 but NADK2’s role in AD had not been explored. Here we report that human TauOs upregulates the expression levels of NADK2 and biochemically active enzyme-bound NADPH in presymptomatic PS19 mice at 1 and 3 months old (Figure 5A and B), an event that coincides with alterations in the expression of synaptic proteins in this transgenic model.57,63 These abnormalities were reverted by normalizing NADK2 levels in PS19 mice (Figures 6 and S5), likely because of reduced LRP1- dependent internalization of tau oligomers in the somatodendritic compartment, where LRP1 is mainly localized.58 It is important to high- light that reducing NADK2 levels also decreased the levels of TauOs and phospho-tau species in PS19 mice without affecting overall hTau expression levels (Figure 6D–F). Thus, an important area of future stud- ies will be to explore the role of NADK2-NADPH in tau seeding and spreading. Aligned with our findings, several large-scale proteomic studies examining brain samples from AD, as well as CSF, have consistently identified NADK2 and LRP1 as significantly upregulated across dif- ferent periods of disease progression.53–55 Thus, our observations suggest that dysregulation of this mitochondrial metabolic pathway is an early event in AD pathogenesis and highlights NADK2 as a potential novel AD biomarker. How does the upregulation of NADK2 contribute to AD? Besides the evidence provided here, showing the effect in regulating TauO uptake by neurons via LRP1, TauOs could also impact brain energy metabolism, reduce ATP levels (Figure 1F), and decrease NAD+ avail- ability in the mitochondria. This metabolite and its reduced form, NADH, are important coenzymes and substrates regulating cellular metabolism. Decreased NAD+ levels are known to be linked to vari- ous age-related diseases, including cancer and cognitive decline.64 In fact, NAD+ supplementation was shown to reduce tau hyperphospho- rylation, neuroinflammation, and memory impairment in mouse models of AD.65,66 Although we did not quantitatively measure NAD+ in our experimental conditions, it is tempting to speculate that TauOs con- tribute to the decrease in NAD+ by increasing its consumption by NADK2, thereby boosting NADPH levels. Along these lines, it is impor- tant to highlight that the NADPH not only functions as a cofactor and/or substrate for numerous enzymes to maintain cellular redox balance, biosynthetic reactions, and overall metabolism.60 NADPH also regulates gene expression by inhibiting histone deacetylase- 3 (HDAC3) by a mechanism independent of its cognate metabolic functions.67 In fact, we recently reported that TauOs control gene expression in primary neuron cultures through a mechanism involv- ing methylation of histone H3.39 Additional consideration of HDAC3 is important as this enzyme controls the expression of genes regulating circadian rhythms.68 Thus, TauO-mediated alteration of the NADK2- Scr and NADK2-KD mice were electrophoretically separated under non-denaturing conditions and followed by WB using HT7 antibody to detect human tau expression levels. Bands at ∼50, 100, and 150 kDa were quantified (shown in black arrows). No differences were observed between the Scr and NADK2-KD mice. (E,F) Phospho-tau protein levels are downregulated in PS19-NADK2-KD mice, correlated with the reduction in TauOs. Cortical protein samples from Scr or NADK2-KD PS19 brains were separated electrophoretically under native conditions, followed by WB using two well-known phospho-tau antibodies (AT8 and AT180). The expression levels of AT8 and AT180 were normalized against βIII-tubulin detected in each sample. Each treatment was compared to its corresponding control conditions. Bar graphs show the quantification obtained using three mice/group/genetic background and analyzed using a two-tailed unpaired Student’s t-test. Error bars represent ± SEM. ### Page 17 PARDO ET AL. 17 of 20NADPH pathway could also impinge on circadian rhythm, which has been proposed as another risk factor linked to AD.69 It is possible to speculate that TauOs have broad physiological consequences by affect- ing NADPH’s cell-regulated functions that are both dependent and independent of its role as an enzyme cofactor (Figure 7). Perturba- tions in metabolic pathways have also been identified as some of the earliest measurable changes before the appearance of clinical symp- toms in AD. Metabolomic studies have described proline metabolism as one of the pathways significantly impacted by the disease’s progres- sion, both in AD animal models and human AD patients.70–73 Here, we found that TauOs increased collagen VI levels (Figure 1H), which agrees with a previous report showing increased collagen expression in AD models.74 As activation of proline metabolism and collagen synthesis have been associated with increased cell survival in cancer models,48,49 it is possible that TauOs trigger a similar mechanism to promote neu- ronal survival needed to support its own spreading through the brain. Thus, early changes in the NADK2-NADPH pathway may be an early contributor to AD progression. While our study provides strong evidence of dysregulation of the NADK2-proline pathway and brain energy metabolism by TauO in vivo, these findings were restricted to male mice. Sex-related difference in AD-related cognitive impairment has been known for ∼30 years 75 are driven by non-genetic (cardiovascular, inflammation and sexLRP1OthersCollagen Glutamate NADP++H+ P5C NAD(P)H NAD(P)++H+ GSAP5CS PYCR1P5CDH PRODHProline NADPHNAD+ Plasma MembraneTauO? LRP-1LRP-1 NADK2 ProlineProtein Synthesis Mitochondria Others non-canonical functionsF I G U R E 7 Working model. Mitochondrial mitochondrial NAD kinase 2 (NADK2) -dependent nicotinamide adenine dinucleotide phosphate (NADPH) facilitates Tau oligomers’ (TauOs’) entry into human neurons. TauOs upregulate the expression of NADK2 and stimulate the de novo synthesis of mitochondrial NADPH. Notably, this upregulation of NADPH was observed in the PS19 live mouse brain (which naturally produces human TauOs) months before cognitive decline onset, and in “inducible AD neurons” obtained from direct conversion of human dermal fibroblasts. Activation of this NADK2-dependent biochemical pathway somehow controls the expression of the LRP1, a main TauO receptor in the neuronal plasma membrane. This vicious cycle, whereby TauOs orchestrate mitochondrial NADPH metabolism and membrane trafficking to stimulate their own uptake, implies that TauO-mediated dysregulation of NADK2 is one of the earliest molecular missteps in Alzheimer’s disease (AD) pathogenesis. hormones) and genetic factors (e.g., APOE genotype).76 Between the former, estrogen has been recognized as a key regulator of brain energy metabolism impacting glucose transport,77 mitochondrial biogenesis and oxidative phosphorylation capacity.78,79 Estrogen also controls the expression of genes found downregulated in LOAD, including those involved in mitochondrial function.80 Since female brains show signif- icant hypometabolism during perimenopause,81 it will be important to investigate whether the NADK2-NADPH-proline pathway plays a role in these sex-specific metabolic responses. We provide evidence that TauOs upregulate LRP1 expression (Figures 3B and 4B). While the precise mechanism was not fully explored in this study, this upregulation may involve TauO-mediated regulation of LRP1 gene expression and/or protein turnover. Interest- ingly, LRP1 has been shown to interact with PSD95 at the neuronal cell surface.58 Here, we demonstrate that reducing NADK2 expres- sion in the PS19 mouse brain restores both LRP1 and PSD95 to levels observed in WT animals (Figure S5E). Whether TauO-induced changes in PSD95 modulates its interaction with LRP1 requires further exploration. LRP1 is an endocytic receptor that delivers multiple extracellular lig- ands into cells, including APOE lipoproteins, which are crucial for lipid uptake needed for energy production.82 These observations suggest that the upregulation of LRP1 expression by TauOs could be a compen- satory response to restore energy levels. While this is a plausible idea, the finding that neuron-specific LRP1 knockout mice do not exhibit reduced viability or immediate neuronal loss 58 suggests that LRP1 may not be a primary regulator of neuronal nutritional status. Given its clin- ical implications of our observations, future studies are warranted to elucidate the mechanisms by which TauOs regulate LRP1 expression and its functional consequences‘’. Therefore, unveiling molecular mechanisms disrupted by TauOs and how they favor their own cellular uptake are key for understand- ing molecular missteps occurring at presymptomatic stages of AD and Tauopathies like frontotemporal dementia, progressive supranu- clear palsy, and others,83 and for developing therapeutic interventions. Future work will explore NADK2’s contribution to TauO uptake in microglia and astrocytes and whether decreasing NADK2 levels alone and/or in combination with dietary supplementation with NAD+ pre- cursors such as nicotinamide riboside or nicotinamide mononucleotide ameliorate or delay AD onset.AUTHOR CONTRIBUTIONSEvelyn Pardo and Andrés Norambuena conceived, designed, initiated, and led the entire study. Evelyn Pardo conducted all the neuron cul- ture studies, managed the animal colony, performed genotyping, and, along with Andrés Norambuena and Vijay Kumar Sagar, coordinated live mouse experiments. Evelyn Pardo generated CRISPR/CAS9 cells and wrote the first draft of the paper, and, along with Andrés Noram- buena, served as the main editor of the original text. Vijay Kumar Sagar performed 2P-FLIM experiments in the live mouse brain. Horst Wallrabe and Vijay Kumar Sagar analyzed 2P-FLIM experiments in the live cells and mouse brains. Taylor Kim provided technical support throughout the study. Ammasi Periasamy provided extensive technical ### Page 18 18 of 20 PARDO ET AL.support for 2P-FLIM. Xuehan Sun generated the initial TauO prepara- tion. Garnett Mingledorff performed the culture of iNeurons. Kristine E. Zengeler and John R. Lukens provided extensive technical and intel- lectual support for in vivo experiments. George S. Bloom analyzed data and provided intellectual feedback throughout the study. All authors read, edited, and approved the submission of the manuscript.ACKNOWLEDGMENTSWe thank Dr. Anthony Spano for providing the βIII-tubulin antibody (Tuj1) and Dr. Rakez Kayed for his generous gift of human brain- derived Tau oligomers. We also thank the W.M. Keck Biomedical Mass Spectrometry Laboratory (funded by a grant from the Uni- versity of Virginia’s School of Medicine). We thank Dr. Dilza Silva for her assistance with MS analysis and Hyeseon Seo for her guid- ance on the statistical analysis. Funding for this work was provided by NIH/NIA Grant R01AG067048 (to AN); The Commonwealth of Virginia’s Alzheimer’s and Related Diseases Research Award Fund, administered by the Virginia Center on Aging, College of Health Pro- fessions, Virginia Commonwealth University Award 22-7 (to AN); the Owens Family Foundation (GSB); Cure Alzheimer’s Fund (to AN and GSB); Rick Sharp Alzheimer’s Foundation (GSB); and the NIH/Office of the Director for funds to purchase a Zeiss 780 and 980 microscopes used in this study (OD016446 and NIH-OD 025156 to AP). Purification of BDTauO was supported by NIH grant AG072458 (to Rakez Kayed, PhD).CONFLICT OF INTEREST STATEMENTThe authors have no conflict of interest to report. Author disclosures are available in the Supporting Information.CONSENT STATEMENTInformed consent was obtained from all subjects before sample col- lection. Receipt of human tissues was granted through University of Virginia’s Institutional Review Board for Health Sciences Research.ORCIDAndrés Norambuena https://orcid.org/0000-0003-2977-1709REFERENCES1. Clavaguera F, Bolmont T, Crowther RA, et al. Transmission and spreading of tauopathy in transgenic mouse brain. Nat Cell Biol. 2009;11(7):909-913. doi:10.1038/ncb1901 2. Ruan Z, Pathak D, Venkatesan Kalavai S, et al. Alzheimer’s disease brain-derived extracellular vesicles spread tau pathology in interneu- rons. Brain. 2021;144(1):288-309. doi:10.1093/brain/awaa376 3. Colom-Cadena M, Davies C, Sirisi S, et al. Synaptic oligomeric tau in Alzheimer’s disease — a potential culprit in the spread of tau pathology through the brain. Neuron. 2023;111(14):2170-2183. doi:10.1016/j. neuron.2023.04.020 4. Mate De Gerando A, Welikovitch LA, Khasnavis A, et al. Tau seeding and spreading in vivo is supported by both AD-derived fibrillar and oligomeric tau. Acta Neuropathol. 2023;146(2):191-210. doi:10.1007/ s00401-023-02600-1 5. Rauch JN, Luna G, Guzman E, et al. LRP1 is a master regulator of tau uptake and spread. Nature. 2020;580(7803):381-385. doi:10.1038/ s41586-020-2156-5 6. Wang W, Zhao F, Ma X, Perry G, Zhu X. Mitochondria dysfunc- tion in the pathogenesis of Alzheimer’s disease: recent advances. Mol Neurodegener. 2020;15(1):30. doi:10.1186/s13024-020-00376-6 7. Womack KB, Diaz-Arrastia R, Aizenstein HJ, et al. Temporoparietal hypometabolism in frontotemporal lobar degeneration and associated imaging diagnostic errors. Arch Neurol. 2011;68(3):329-337. doi:10. 1001/archneurol.2010.295 8. Landau SM, Mintun MA, Joshi AD, et al. Amyloid deposition, hypometabolism, and longitudinal cognitive decline. Ann Neurol. 2012;72(4):578-586. doi:10.1002/ana.23650 9. Palmqvist S, Insel PS, Stomrud E, et al. Cerebrospinal fluid and plasma biomarker trajectories with increasing amyloid deposition in Alzheimer’s disease. EMBO Mol Med. 2019;11(12):e11170. doi:10. 15252/emmm.201911170 10. Janelidze S, Berron D, Smith R, et al. Associations of plasma phospho-tau217 levels with tau positron emission tomography in early Alzheimer disease. JAMA Neurol. 2021;78(2):149. doi:10.1001/ jamaneurol.2020.4201 11. Barthélemy NR, Li Y, Joseph-Mathurin N, et al. A soluble phosphory- lated tau signature links tau, amyloid and the evolution of stages of dominantly inherited Alzheimer’s disease. Nat Med. 2020;26(3):398- 407. doi:10.1038/s41591-020-0781-z 12. Mattsson-Carlgren N, Andersson E, Janelidze S, et al. Aβ deposi- tion is associated with increases in soluble and phosphorylated tau that precede a positive tau PET in Alzheimer’s disease. Sci Adv. 2020;6(16):eaaz2387. doi:10.1126/sciadv.aaz2387 13. Spinelli JB, Haigis MC. The multifaceted contributions of mitochon- dria to cellular metabolism. Nat Cell Biol. 2018;20(7):745-754. doi:10. 1038/s41556-018-0124-1 14. Xie N, Zhang L, Gao W, et al. NAD+ metabolism: pathophysiologic mechanisms and therapeutic potential. Signal Transduct Target Ther. 2020;5(1):227. doi:10.1038/s41392-020-00311-7 15. Tran DH, Kesavan R, Rion H, et al. Mitochondrial NADP+ is essential for proline biosynthesis during cell growth. Nat Metab. 2021;3(4):571- 585. doi:10.1038/s42255-021-00374-y 16. Kay EJ, Zanivan S, Rufini A. Proline metabolism shapes the tumor microenvironment: from collagen deposition to immune evasion. Curr Opin Biotechnol. 2023;84:103011. doi:10.1016/j.copbio.2023.103011 17. Zhu J, Schwörer S, Berisa M, et al. Mitochondrial NADP(H) generation is essential for proline biosynthesis. Science. 2021;372(6545):968- 972. doi:10.1126/science.abd5491 18. Pomerantz DJ, Ferdinandusse S, Cogan J, et al. Clinical heterogeneity of mitochondrial NAD kinase deficiency caused by a NADK2 start loss variant. Am J Med Genet A. 2018;176(3):692-698. doi:10.1002/ajmg.a. 38602 19. Houten SM, Denis S, Te Brinke H, et al. Mitochondrial NADP(H) defi- ciency due to a mutation in NADK2 causes dienoyl-CoA reductase deficiency with hyperlysinemia. Hum Mol Genet. 2014;23(18):5009- 5016. doi:10.1093/hmg/ddu218 20. Cleveland DW, Hwo SY, Kirschner MW. Purification of tau, a microtubule-associated protein that induces assembly of micro- tubules from purified tubulin. J Mol Biol. 1977;116(2):207-225. doi:10. 1016/0022-2836(77)90213-3 21. Santarella RA, Skiniotis G, Goldie KN, et al. Surface-decoration of microtubules by human tau. J Mol Biol. 2004;339(3):539-553. doi:10. 1016/j.jmb.2004.04.008 22. Parra Bravo C, Naguib SA, Gan L. Cellular and pathological functions of tau. Nat Rev Mol Cell Biol. 2024;25(11):845-864. doi:10.1038/s41580- 024-00753-9 23. Norambuena A, Sun X, Wallrabe H, et al. SOD1 mediates lysosome- to-mitochondria communication and its dysregulation by amyloid-βoligomers. Neurobiol Dis. 2022;169:105737. doi:10.1016/j.nbd.2022. 105737 24. Tracy TE, Madero-Pérez J, Swaney DL, et al. Tau interactome maps synaptic and mitochondrial processes associated with neurodegen- ### Page 19 PARDO ET AL. 19 of 20eration. Cell. 2022;185(4):712-728.e14. doi:10.1016/j.cell.2021.12. 041 25. Castillo-Carranza DL, Gerson JE, Sengupta U, Guerrero-Muñoz MJ, Lasagna-Reeves CA, Kayed R. Specific targeting of tau oligomers in Htau mice prevents cognitive impairment and tau toxicity following injection with brain-derived tau oligomeric seeds. J Alzheimers Dis. 2014;40(s1):S97-S111. doi:10.3233/JAD-132477 26. Lasagna-Reeves CA, Castillo-Carranza DL, Sengupta U, et al. Alzheimer brain-derived tau oligomers propagate pathology from endogenous tau. Sci Rep. 2012;2:700. doi:10.1038/srep00700 27. Ghag G, Bhatt N, Cantu DV, et al. Soluble tau aggregates, not large fibrils, are the toxic species that display seeding and cross-seeding behavior. Protein Sci. 2018;27(11):1901-1909. doi:10.1002/pro.3499 28. Lasagna-Reeves CA, Castillo-Carranza DL, Sengupta U, et al. Identifi- cation of oligomers at early stages of tau aggregation in Alzheimer’s disease. FASEB J. 2012;26(5):1946-1959. doi:10.1096/fj.11-199851 29. Sengupta U, Portelius E, Hansson O, et al. Tau oligomers in cere- brospinal fluid in Alzheimer’s disease. Ann Clin Transl Neurol. 2017;4(4):226-235. doi:10.1002/acn3.382 30. Lasagna-Reeves CA, Castillo-Carranza DL, Sengupta U, Clos AL, Jackson GR, Kayed R. Tau oligomers impair memory and induce synap- tic and mitochondrial dysfunction in wild-type mice. Mol Neurodegener. 2011;6(1):39. doi:10.1186/1750-1326-6-39 31. Wu JW, Herman M, Liu L, et al. Small misfolded tau species are internalized via bulk endocytosis and anterogradely and retrogradely transported in neurons. J Biol Chem. 2013;288(3):1856-1870. doi:10. 1074/jbc.M112.394528 32. Holmes BB, DeVos SL, Kfoury N, et al. Heparan sulfate proteo- glycans mediate internalization and propagation of specific proteo- pathic seeds. PNAS. 2013;110(33):E3138-E3147. doi:10.1073/pnas. 1301440110 33. Mudher A, Colin M, Dujardin S, et al. What is the evidence that tau pathology spreads through prion-like propagation?. Acta Neuropathol Commun. 2017;5(1):99. doi:10.1186/s40478-017-0488-7 34. Norambuena A, Wallrabe H, Cao R, et al. A novel lysosome-to- mitochondria signaling pathway disrupted by amyloid-β oligomers.EMBO J. 2018;37(22):e100241. doi:10.15252/embj.2018100241 35. Norambuena A, Sagar VK, Wang Z, et al. Disrupted mitochondrial response to nutrients is a presymptomatic event in the cortex of the APP SAA knock-in mouse model of Alzheimer’s disease. Alzheimers Dement. 2024;20(10):6844-6859. doi:10.1002/alz.14144 36. Lakowicz JR. Principles of fluorescence spectroscopy. 3rd ed. Springer; 2006. doi:10.1007/978-0-387-46312-4 37. Choi SH, Kim YH, Hebisch M, et al. A three-dimensional human neural cell culture model of Alzheimer’s disease. Nature. 2014;515(7526):274-278. 38. Song Y, Subramanian K, Berberich MJ, et al. Data descriptor: a dynamic view of the proteomic landscape during differentiation of rencell VM cells, an immortalized human neural progenitor line. Sci Data. 2019;6: 190016. doi:10.1038/sdata.2019.16 39. Sun X, Eastman G, Shi Y, et al. Structural and functional damage to neu- ronal nuclei caused by extracellular tau oligomers. Alzheimers Dement. 2024;20(3):1656-1670. doi:10.1002/alz.13535 40. Combs B, Tiernan CT, Hamel C, Kanaan NM. Production of recombi- nant tau oligomers in vitro. In: Methods in cell biology. Academic Press Inc.; 2017:45-64. doi:10.1016/bs.mcb.2017.06.005 41. Jones B, Balasubramaniam M, Lebowitz JJ, et al. Activation of proline biosynthesis is critical to maintain glutamate homeostasis during acute methamphetamine exposure. Sci Rep. 2021;11(1):1422. doi:10.1038/ s41598-020-80917-7 42. Chen C, Wanduragala S, Becker DF, Dickman MB. Tomato QM- like protein protects Saccharomyces cerevisiae cells against oxidative stress by regulating intracellular proline levels. Appl Environ Microbiol. 2006;72(6):4001-4006. doi:10.1128/AEM.024 28-05 43. Pandhare J, Donald SP, Cooper SK, Phang JM. Regulation and function of proline oxidase under nutrient stress. J Cell Biochem. 2009;107(4):759-768. doi:10.1002/jcb.22174 44. Puangmalai N, Bhatt N, Montalbano M, et al. Internalization mecha- nisms of brain-derived tau oligomers from patients with Alzheimer’s disease, progressive supranuclear palsy and dementia with Lewy bod- ies. Cell Death Dis. 2020;11(5):1-16. doi:10.1038/s41419-020-2503- 3 45. Rauch JN, Chen JJ, Sorum AW, et al. Tau internalization is regulated by 6-O sulfation on heparan sulfate proteoglycans (HSPGs). Sci Rep. 2018;8(1):1-10. doi:10.1038/s41598-018-24904-z 46. Blacker TS, Mann ZF, Gale JE, et al. Separating NADH and NADPH fluorescence in live cells and tissues using FLIM. Nat Commun. 2014;5:1-9. doi:10.1038/ncomms4936 47. Pollak N, Niere M, Ziegler M. NAD kinase levels control the NADPH concentration in human cells*. J Biol Chem. 2007;282:33562-33571. https://api.semanticscholar.org/CorpusID.595722 48. Kay EJ, Paterson K, Riera-Domingo C, et al. Cancer-associated fibrob- lasts require proline synthesis by PYCR1 for the deposition of pro-tumorigenic extracellular matrix. Nat Metab. 2022;4(6):693-710. doi:10.1038/s42255-022-00582-0 49. Westbrook RL, Bridges E, Roberts J, et al. Proline synthesis through PYCR1 is required to support cancer cell proliferation and survival in oxygen-limiting conditions. Cell Rep. 2022;38(5):110320. doi:10.1016/ j.celrep.2022.110320 50. Geng P, Qin W, Xu G. Proline metabolism in cancer. Amino Acids. 2021;53(12):1769-1777. doi:10.1007/s00726-021-03060-1 51. Chalecka M, Kazberuk A, Palka J, Surazynski A. P5C as an interface of proline interconvertible amino acids and its role in regulation of cell survival and apoptosis. Int J Mol Sci. 2021;22(21):11763. doi:10.3390/ ijms222111763 52. Kim D, Kesavan R, Ryu K, et al. Mitochondrial NADPH fuels mito- chondrial fatty acid synthesis and lipoylation to power oxidative metabolism. Nat Cell Biol. 2025;27(5):790-800. doi:10.1038/s41556- 025-01655-4 53. Drummond E, Kavanagh T, Pires G, et al. The amyloid plaque proteome in early onset Alzheimer’s disease and Down syndrome. Acta Neu- ropathol Commun. 2022;10(1):53. doi:10.1186/s40478-022-01356- 1 54. Johnson ECB, Dammer EB, Duong DM, et al. Large-scale proteomic analysis of Alzheimer’s disease brain and cerebrospinal fluid reveals early changes in energy metabolism associated with microglia and astrocyte activation. Nat Med. 2020;26(5):769-780. doi:10.1038/ s41591-020-0815-6 55. Hondius DC, Van Nierop P, Li KW, et al. Profiling the human hippocam- pal proteome at all pathologic stages of Alzheimer’s disease. Alzheimers Dement. 2016;12(6):654-668. doi:10.1016/j.jalz.2015.11.002 56. Mertens J, Paquola ACM, Ku M, et al. Directly reprogrammed human neurons retain aging-associated transcriptomic signatures and reveal age-related nucleocytoplasmic defects. Cell Stem Cell. 2015;17(6):705- 718. doi:10.1016/j.stem.2015.09.001 57. Yoshiyama Y, Higuchi M, Zhang B, et al. Synapse loss and microglial activation precede tangles in a P301S tauopathy mouse model. Neuron. 2007;53(3):337-351. doi:10.1016/j.neuron.2007.01.010 58. May P, Rohlmann A, Bock HH, et al. Neuronal LRP1 functionally asso- ciates with postsynaptic proteins and is required for normal motor function in mice. Mol Cell Biol. 2004;24(20):8872-8883. doi:10.1128/ MCB.24.20.8872-8883.2004 59. Fichman Y, Gerdes SY, Kovács H, Szabados L, Zilberstein A, Csonka LN. Evolution of proline biosynthesis: enzymology, bioinformatics, genet- ics, and transcriptional regulation. Biol Rev. 2015;90(4):1065-1099. doi:10.1111/brv.12146 60. Xiao W, Wang RS, Handy DE, Loscalzo J. NAD(H) and NADP(H) redox couples and cellular energy metabolism. Antioxid Redox Signal. 2018;28(3):251-272. doi:10.1089/ars.2017.7216 ### Page 20 20 of 20 PARDO ET AL.61. Ohashi K, Kawai S, Murata K. Identification and characterization of a human mitochondrial NAD kinase. Nat Commun. 2012;3(1):1248. doi:10.1038/ncomms2262 62. Zhang R, Zhang K. Mitochondrial NAD kinase in health and dis- ease. Redox Biol. 2023;60:102613. doi:10.1016/j.redox.2023. 102613 63. Martinez P, Patel H, You Y, et al. Bassoon contributes to tau-seed propagation and neurotoxicity. Nat Neurosci. 2022;25(12):1597-1607. doi:10.1038/s41593-022-01191-6 64. Covarrubias AJ, Perrone R, Grozio A, Verdin E. NAD+ metabolism and its roles in cellular processes during ageing. Nat Rev Mol Cell Biol. 2021;22(2):119-141. doi:10.1038/s41580-020-00313-x 65. Fang EF, Hou Y, Palikaras K, et al. Mitophagy inhibits amyloid-β and tau pathology and reverses cognitive deficits in models of Alzheimer’s disease. Nat Neurosci. 2019;22(3):401-412. doi:10.1038/s41593-018- 0332-9 66. Hou Y, Lautrup S, Cordonnier S, et al. NAD + supplementation normalizes key Alzheimer’s features and DNA damage responses in a new AD mouse model with introduced DNA repair defi- ciency. PNAS. 2018;115(8):E1876-E1885. doi:10.1073/pnas.171881- 9115 67. Li W, Kou J, Qin J, et al. NADPH levels affect cellular epigenetic state by inhibiting HDAC3–Ncor complex. Nat Metab. 2021;3(1):75-89. doi:10.1038/s42255-020-00330-2 68. Alenghat T, Meyers K, Mullican SE, et al. Nuclear receptor corepres- sor and histone deacetylase 3 govern circadian metabolic physiology.Nature. 2008;456(7224):997-1000. doi:10.1038/nature07541 69. Musiek ES, Xiong DD, Holtzman DM. Sleep, circadian rhythms, and the pathogenesis of Alzheimer disease. Exp Mol Med. 2015;47(3):e148- e148. doi:10.1038/emm.2014.121 70. Jasbi P, Shi X, Chu P, et al. Metabolic profiling of neocortical tis- sue discriminates Alzheimer’s disease from mild cognitive impair- ment, high pathology controls, and normal controls. J Proteome Res. 2021;20(9):4303-4317. doi:10.1021/acs.jproteome.1c00290 71. Butterfield DA, Favia M, Spera I, Campanella A, Lanza M, Castegna A. Metabolic features of brain function with relevance to clinical fea- tures of Alzheimer and Parkinson diseases. Molecules. 2022;27(3):951. doi:10.3390/molecules27030951 72. Tondo M, Wasek B, Escola-Gil JC, et al. Altered brain metabolome is associated with memory impairment in the rTG4510 mouse model of tauopathy. Metabolites. 2020;10(2):69. doi:10.3390/ metabo10020069 73. Kim YH, Shim HS, Kim KH, et al. Metabolomic analysis identifies alter- ations of amino acid metabolome signatures in the postmortem brain of Alzheimer’s disease. Exp Neurobiol. 2019;28(3):376-389. doi:10. 5607/en.2019.28.3.376 74. Cheng JS, Dubal DB, Kim DH, et al. Collagen VI protects neurons against Aβ toxicity. Nat Neurosci. 2009;12(2):119-121. doi:10.1038/nn. 2240 75. Laws KR, Irvine K, Gale TM. Sex differences in cognitive impairment in Alzheimer’s disease. World J Psychiatry. 2016;6(1):54-65. doi:10.5498/ wjp.v6.i1.54 76. Gamache J, Yun Y, Chiba-Falek O. Sex-dependent effect of APOE on Alzheimer’s disease and other age-related neurodegenerative disor- ders. Dis Model Mech. 2020;13(8):1-21. doi:10.1242/dmm.045211 77. Rettberg JR, Yao J, Brinton RD. Estrogen: a master regulator of bioenergetic systems in the brain and body. Front Neuroendocrinol. 2014;35(1):8-30. doi:10.1016/j.yfrne.2013.08.001 78. Kemper MF, Stirone C, Krause DN, Duckles SP, Procaccio V. Genomic and non-genomic regulation of PGC1 isoforms by estrogen to increase cerebral vascular mitochondrial biogenesis and reactive oxygen species protection. Eur J Pharmacol. 2014;723:322-329. doi:10.1016/j. ejphar.2013.11.009 79. Nilsen J, Irwin RW, Gallaher TK, Brinton RD. Estradiol in vivo regula- tion of brain mitochondrial proteome. J Neurosci. 2007;27(51):14069- 14077. doi:10.1523/JNEUROSCI.4391-07.2007 80. Ratnakumar A, Zimmerman SE, Jordan BA, Mar JC. Estrogen activates Alzheimer’s disease genes. Alzheimers Dement. 2019;5(1):906-917. doi:10.1016/j.trci.2019.09.004 81. Brinton RD, Yao J, Yin F, Mack WJ, Cadenas E. Perimenopause as a neurological transition state. Nat Rev Endocrinol. 2015;11(7):393-405. doi:10.1038/nrendo.2015.82 82. Herz J, Strickland DK. LRP: a multifunctional scavenger and signaling receptor. J Clin Invest. 2001;108(6):779-784. doi:10.1172/JCI13992 83. Lane-Donovan C, Boxer AL. Disentangling tau: one protein, many ther- apeutic approaches. Neurotherapeutics. 2024;21(2):e00321. doi:10. 1016/j.neurot.2024.e00321SUPPORTING INFORMATIONAdditional supporting information can be found online in the Support- ing Information section at the end of this article.How to cite this article: Pardo E, Kim T, Wallrabe H, et al. Mitochondrial NADK2-dependent NADPH controls tau oligomer uptake in human neurons. Alzheimer’s Dement. 2026;22:e71594. https://doi.org/10.1002/alz.71594