Mitochondrial dysfunction is observed in AD across numerous tissues. Reductions in cytochrome oxidase (COX) Vmax have been observed in brain, blood, and skin from AD subjects (Parker and Parks, 1995; Bosetti et al., 2002; Cardoso et al., 2004; Swerdlow et al., 2017). Across models of AD, including mouse and cell models, alterations to mitochondrial oxygen consumption, reactive oxygen species (ROS), and mitophagy are observed (Arduino et al., 2015; Guo et al., 2017; Swerdlow et al., 2017). Mitochondrial phenotypes in specific cell types are difficult to study in postmortem human and animal models of AD. However, induced pluripotent stem cells (iPSC) have emerged as a tool to study AD and understand cell type specific changes in mitochondrial function.

iPSC derived neurons have mitochondrial dysfunction independent of AD pathological hallmarks. Prior studies indicate iPSC derived neurons from sporadic AD patients show increased oxidative stress and mtDNA damage/mutations. These observations were independent of AD pathological hallmarks including tau neurofibrillary tangles and Aβ plaques (Birnbaum et al., 2018). iPSC derived neurons from sporadic and familial AD subjects show impaired mitophagy linked to lysosomal defects (Martin-Maestro et al., 2017). Further studies have examined overexpression of APP and tau in iPSC derived neurons, which impaired mitophagy. Due to changes observed in iPSC derived neurons, they have been used to test potential therapeutics to remedy mitophagy failure in AD (Martin-Maestro et al., 2019a,b).

iPSC derived neurons often show elevated Aβ in addition to increased mtDNA mutation burden and altered mitochondrial function (Lee et al., 2022). Familial AD derived neurons also recapitulate increased Aβ and tau pathology/accumulation with reduced mitochondrial trafficking (Li et al., 2019). Studies comparing iPSC derived neurons with postmortem brain tissue from sporadic AD subjects have identified overlapping proteomic profiles between sample types. Mitophagy and autophagy were major implicated pathways between iPSC derived neurons and postmortem brain samples (Verheijen et al., 2022). Overall iPSC derived models likely recapitulate disease mechanisms observed in postmortem human tissue and could be a relevant model for testing therapeutics.

Studies examining iPSC derived astrocytes in the context of mitochondrial function are limited. iPSC derived astrocytes treated with Aβ resulted in mitochondrial dysfunction and a shift in bioenergetic pathways to glycolysis and fatty acid oxidation (Zysk et al., 2023). A separate study also reported changes in mitochondrial energy metabolism and mitochondrial transcriptome in iPSC derived astrocytes (Ryu et al., 2021). Beyond iPSC derived astrocytes, cerebral organoids are potential models to recapitulate cell-to-cell interactions, however limited data are available examining mitochondrial dysfunction in these models.

Here, we compared iPSC neurons and astrocytes from the same individuals to examine cell type specific changes in mitochondrial function from non-demented (ND) and sporadic AD (AD) donors. We leveraged age and sex matched iPSC models from the University of Kansas Alzheimer’s Disease Research Center (KU ADRC) clinical cohort and WiCell.

We examined COX Vmax across iPSC models. iPSC derived neurons (iNeurons) derived from AD subjects had reduced COX Vmax (Figure 1A) and this was driven by cells derived from female subjects (Figure 1B). iPSC derived astrocytes (iAstrocytes) showed no change in COX Vmax (Figures 1C,D). However, iPSC derived cerebral organoids showed reduced COX Vmax (Figure 1E) regardless of sex (Figure 1F). To determine if these observations were due to a change in expression of COX components, we examined the protein expression of COX4I1. There was no change in COX4I1 protein expression for any model type (data not shown).

To further examine mitochondrial function, we used seahorse assays to examine mitochondrial respiration, glycolytic flux, and ETC function. Seahorse analysis of ETC function showed that iNeurons had reduced flux at complexes I, II, and IV (Figure 2A). iNeurons also showed reduced baseline, uncoupled, and ATP-linked mitochondrial respiration (Figure 2B) with increased glycolytic flux (Figure 2C). Comparatively, iAstrocytes had increased ETC flux at complexes I-III, with reduced mitochondrial respiration at baseline, uncoupled, and ATP-linked respiration (Figures 2D,E). Glycolytic flux was also increased in iAstrocytes (Figure 2F). Seahorse tracings are shown in Supplementary Figure S2.

Altered mitochondrial respiration can be a consequence or can cause alterations to mitochondrial membrane potential and reactive oxygen species (ROS). As such, we measured mitochondrial membrane potential, mitochondrial superoxide, and hydrogen peroxide levels. Mitochondrial membrane potential was increased in female iNeurons and iAstrocytes from AD subjects (Figures 3A,B) but reduced in male cells derived from AD subjects. Mitochondrial superoxide was decreased in female iNeurons from AD subjects but increased in male iNeurons from AD subjects (Figure 3C). iAstrocytes showed an increase in mitochondrial superoxide in female cells derived from AD subjects, with no change in male derived cells (Figure 3D). iNeurons and iAstrocytes had an increase in hydrogen peroxide in female cells derived from AD subjects, but not in males (Figures 3E,F).

Calcium dynamics can profoundly change mitochondrial function and are critical to bioenergetic homeostasis. We measured total cellular calcium and mitochondrial calcium content. Total cellular calcium was not changed between ND and AD derived iNeurons and iAstrocytes (Figures 4A,C). However, iNeurons had an increase in mitochondrial calcium in female cells derived from AD subjects but not in males (Figure 4B). iAstrocytes showed increased mitochondrial calcium regardless of sex, in AD derived cells (Figure 4D).

Deficits in mitochondrial turnover through mitophagy and mitochondrial biogenesis have been noted in AD models. Here we measured basal mitophagy levels and mitochondrial biogenesis using adenoviral vectors (EFGP-mCherry COX8 and MitoTimer). The number of mitochondria localized to lysosomes was reduced in female iNeurons derived from AD subjects, but not in males (although a trend was present, Figures 5A–C). iAstrocytes derived from female AD subjects had increased mitochondria within lysosomes but no difference was observed between male AD and ND donors (Figures 5G,H). Overall, iNeurons from female AD subjects had an increase in the number of mature and immature mitochondria (Figures 5D–F). iNeurons derived from male AD subjects had a decrease in the number of mature and immature mitochondria (Figures 5D–F). iAstrocytes showed no differences in mitochondrial number, either mature or immature (data not shown).

To further understand changes to mitochondria in iPSC derived AD models, we examined mitochondrial dynamics. Mitochondrial fission and fusion are critical for bioenergetic function and mitophagy. Both iPSC derived neurons and astrocytes from AD subjects had increased expression of phosphorylated DRP1 at Serine 637 (Figures 6A,C). There was no change in total expression of DRP1 (data not shown). iPSC derived neurons and astrocytes from AD subjects also had decreased MFN1 expression (Figures 6B,D). We did observe sex differences for phosphorylated DRP1 in the iPSC derived neurons, but not in astrocytes. MitoTracker intensity was increased in female derived iPSC neurons from AD subjects but decreased in those derived from male AD subjects (Figure 6E). iPSC derived neurons from AD subjects had reduced mitochondrial branch number, junction number, and maximum branch length (Figures 6F–I). We did not observe sex differences for these findings. MitoTracker intensity was increased in female iPSC derived astrocytes from AD subjects but no in those derived from male AD subjects. iPSC astrocytes derived from female AD subjects had reduced mitochondrial branch number, junction number, but increased maximum branch length (Figures 6K–N). These observations were not found in iPSC derived astrocytes from male AD subjects. Supplementary Figure S3 shows western blots for Figure 6.

To determine if these models recapitulate phenotypes observed in AD, we measured Aβ, APP, and GFAP expression. We also examined relationships between mitochondrial membrane potential and Aβ, as we have previously described strong correlations between these outcomes (Wilkins et al., 2022). iNeurons derived from AD subjects had increased Aβ₄₂ in media samples for both male and female cohorts (Figure 7A). Aβ₄₀ was increased in media samples for female AD subjects only in iNeurons (Figure 7B). We had previously shown that mitochondrial membrane potential correlates with Aβ secretion, and this was recapitulated here in female derived iNeurons (Figure 7C) (Wilkins et al., 2022). Total full length amyloid precursor protein expression (APP) was decreased in female but increased in male AD derived iNeurons (Figures 7D,E) Finally, iAstrocytes derived from female AD subjects had an increase in glial fibrillary acid protein (GFAP) expression (Figures 7F,G).

iPSC derived models are gaining popularity in the AD research field. However, it is important to understand their phenotypes and nuances. Here we describe mitochondrial and AD pathological hallmarks in iPSC derived models. We considered cell type and sex specific differences, as these variables can and will impact study design. Most prior research has focused on neurons and have failed to consider sex differences. The importance of non-neuronal cell types in AD pathology have been noted and require further study. Furthermore, sex specific differences are imperative to understand given the increased risk of AD for females.

It is important to note that iPSCs from the KU ADRC clinical cohort are derived from fibroblasts upon autopsy which allows for a definitive diagnosis of AD with neuropathological exam. This is an important consideration in studies and allows for the possibility of linking clinical data with in vitro data. iPSCs purchased from WiCell were derived from fibroblasts but not upon autopsy, so a true definitive diagnosis of AD is not available (Israel et al., 2012). The gold standard for AD diagnosis is neuropathological exam at autopsy.

We observed mitochondrial phenotypes in iPSC derived neurons, astrocytes, and cerebral organoids in age and sex matched models derived from ND and AD subjects. Neuronal and cerebral organoid AD models showed reduced COX Vmax, while astrocytes did not. This finding is interesting and suggests a difference in physiology between single cell cultures and co-culture models. COX Vmax deficits have been noted across cell, animal, and human AD models (Kish et al., 1992; Curti et al., 1997; Cardoso et al., 2004; Crugnola et al., 2010). However, in iPSC derived models the finding of COX Vmax reduction in AD could be cell type and sex specific.

Mitochondrial and glycolytic flux were changed in both neuron and astrocyte derived models. Overall, we did not observe sex differences in these measures. The most significant finding is that both neuron and astrocyte derived models increase glycolytic flux and reduce mitochondrial flux. This suggests bioenergetic uncoupling and the possibility that these cell types could be competing for the same fuel sources (Fernandez-Fernandez et al., 2012; Qi et al., 2021). It is currently accepted that astrocytes provide bioenergetic support to neurons in the form of lactate, and that astrocytes typically produce more ATP from glycolysis while neurons leverage mitochondrial respiration for ATP production (Machler et al., 2016; Steinman et al., 2016; Muraleedharan et al., 2020; Yamagata, 2022). If both cell types are shifting to a more glycolytic state in AD, this could impair energy homeostasis and requires further investigation.

Both iPSC derived neurons and astrocytes had reduced ATP-linked respiration, consistent with an overall decrease in mitochondrial respiration. The iPSC derived neurons also had reduced leak rate, but as discussed below had increased indicators of ROS. These data suggest an overall uncoupling of mitochondrial respiration and potential reductions in redox homeostasis. Given that substrates are saturating in the Seahorse assays, this suggests that mitochondrial deficits are either intrinsic, i.e., enzyme deficits, or extrinsic, i.e., an inability of cells to transport substrates like glucose. Reductions in COX Vmax suggest intrinsic deficits in ETC enzyme structure and function, however future studies should examine glucose transport and uptake.

We also observed cell-type and sex specific changes to mitochondrial membrane potential and ROS levels. Mitochondrial membrane potential showed sex differences in both iPSC derived neurons and astrocytes, where female AD models were hyperpolarized, and male models were depolarized. This was consistent between cell types and could lead to altered ATP production and ROS. There were significant changes in mitochondrial superoxide and hydrogen peroxide levels in both neuron and astrocyte AD models, and these could reflect bioenergetic stress and failure of antioxidant enzyme pathways. We also observed an increase in mitochondrial calcium in female AD neuron and astrocyte models. Increased mitochondrial calcium is associated with bioenergetic stress, protein misfolding, and apoptosis (Jadiya et al., 2019; Calvo-Rodriguez et al., 2020; Ryan et al., 2020; Calvo-Rodriguez and Bacskai, 2021). Increased mitochondrial calcium has been noted across several AD models.

Mitophagy failure is a hallmark of AD and is recapitulated in iPSC models, however, there appear to be sex and cell-type specific differences. Both male and female AD subject iPSC derived neurons had reduced mitochondrial localization to lysosomes. The decreased mitochondrial lysosome content in the iPSC derived neurons was also supported by an increase in the total number of mitochondria, both mature and immature. In iPSC derived astrocyte models female derived AD cells had increased mitochondria localized to lysosomes while males showed no change. These results highlight potential cell type specific deficits in mitophagy pathways. Our data are consistent with findings of prior studies. Defects in lysosome degradation in fibroblast and iPSC models of familial AD have been described (Martin-Maestro et al., 2017). In postmortem human brain phosphorylated ubiquitin (pS65-Ub) levels are increased and localize with mitochondria (Hou et al., 2020). Other studies have indicated an increase in mitochondrial content in postmortem brain with possible increases in mitochondria within autophagic vesicles. These studies have been reviewed elsewhere recently (Mary et al., 2023). Overall, it’s not clear where the lesion is in mitophagy pathways in AD.

Mitochondrial dynamics are also altered in the iPSC derived models from AD subjects. Mitochondrial dynamics appear to be shifted away from fusion, with reductions in MFN1 expression and reduced mitochondrial branch lengths. Interactions between mitochondria also are reduced in both iPSC derived neurons and astrocytes from AD subjects. DRP1 phosphorylation is increased at Ser637, which is consistent with inhibition of fission (Knott et al., 2008). Given that the mitochondrial branch lengths are reduced the inhibition of fission through DRP1 could be a compensatory mechanism. Future studies should focus on understanding the context and consequences of these findings.

AD pathological hallmarks are observed in iPSC derived neurons. Secreted Aβ₄₀ and Aβ₄₂ were increased in AD neuron models with changes in total APP protein expression. iPSC derived neurons are not in culture for extended time periods and therefore do not develop Aβ plaques. Furthermore, GFAP expression in astrocyte AD models were higher in female cells. Sex differences in neuroinflammation phenotypes in AD have been noted previously in human subjects (Buckley, 2022; Coales et al., 2022). Prior studies have also shown that iPSC derived neurons from sporadic AD subjects have increase Aβ levels and phosphorylated tau (Ochalek et al., 2017).

iPSC studies leveraging female cells can be difficult due to unknown expression of maternal or paternal X chromosomes. This genetic variation could confound results, as different iPSC clones from the same donor could have an active maternal X chromosome or an active paternal X chromosome and this could yield varying phenotypes. However, our results indicate that sex as a biological variable is a critical component to consider in study design. While our study revealed interesting data, there are limitations. These include a limited number of iPSC donor cells and lack of power to examine effects of apolipoprotein E (APOE) genotype. However, our data highlight that future studies need to consider sex as a biological variable and be designed with enough power to detect sex differences. Consideration of sex differences is imperative for future therapeutic development and fundamental understanding of AD risk and cause.

Mitochondrial dysfunction is a hallmark of AD. Mitochondria are critical to proteostasis, and loss of proteostasis is an additional hallmark of AD. We have previously reported that mitochondrial membrane potential correlates with Aβ production and secretion, and these results were recapitulated here (Wilkins et al., 2022). The mitochondrial cascade hypothesis places mitochondria upstream of Aβ plaques and tau tangles in AD (Swerdlow et al., 1842). Supporting this hypothesis are studies which show direct relationships between mitochondrial function and Aβ and tau pathologies (Fukui et al., 2007; Pinto et al., 2013; Kukreja et al., 2014; Wilkins et al., 2015; Weidling et al., 2020; Wilkins et al., 2022). Overall, future studies should consider cell-type specific effects and work to understand the relationship between mitochondria and proteostasis in AD.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the University of Kansas Medical Center Institutional Review Board. The patients/participants provided their written informed consent to participate in this study.

HW conceptualized, obtained funding for the work, developed and performed assays, and wrote the manuscript. RS, JV, and MH edited the manuscript. KF, JS, BH, BT, CL, TS, VC, CAG, NS, JD, and CMG assisted with data collection. All authors contributed to the article and approved the submitted version.

This study was supported by the Margaret “Peg” McLaughlin and Lydia A. Walker Opportunity Fund, Alzheimer’s Association Grant 23AARG-1023294, the Transgenic and Gene Targeting Institutional Facility at the University of Kansas Medical Center, the University of Kansas Cancer Center (NIH P30 CA168524) the University of Kansas Alzheimer’s Disease Center P30AG035982, R21TR003589, R00AG056600, and R01AG078186.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.