The Mitochondrial Enzyme 17βHSD10 Modulates Ischemic and Amyloid-β-Induced Stress in Primary Mouse Astrocytes

Visual Abstract


Introduction
Alzheimer's disease (AD) and postischemic neurodegeneration are two of the main conditions causing dementia worldwide (Pluta et al., 2021). Clinical, epidemiological and experimental studies have shown that the two conditions may share the same or closely related pathologic mechanisms, including brain metabolic dysfunction (Kunz and Iadecola, 2008;Hammond et al., 2020), disruption of the blood-brain barrier (Zlokovic, 2008;Giraud et al., 2015), neuroinflammation and amyloid accumulation in the extracellular space (Jobst et al., 1994;Pluta et al., 1998;Grothe et al., 2018). Furthermore, AD increases the risk of stroke (Tolppanen et al., 2013) and vice versa (Ballard et al., 2000;Pluta and Ułamek, 2008). This association may be key in identifying the causative mechanisms driving neurodegeneration in these pathologies.
Although astrocytes are essential players in all of the aforementioned mechanisms (Guo et al., 2012;Chaitanya et al., 2014;Grubman et al., 2019;Habib et al., 2020), the role of their mitochondria has been mostly neglected because of the well-established observation that astrocytes are normally glycolysis-dependent (Pellerin and Magistretti, 1994;Magistretti and Pellerin, 1999;Magistretti, 2006). However, disease-associated metabolic stress marked by low levels of brain glucose (Mosconi, 2005;Oh et al., 2016) imposes metabolic load on astrocytic mitochondria to use alternative energy fuels to sustain the brain (Hoyer et al., 1991;Guzmán and Blázquez, 2004;Nilsen et al., 2007;Xu et al., 2016;Polyzos et al., 2019). Furthermore, immortalized hippocampal astrocytes from 3xTg-AD mice (3Tg-iAstro cells) have an impaired bioenergetics profile characterized by reduced glycolysis and mitochondrial respiration, accompanied by an increased generation of reactive oxygen species (ROS; Dematteis et al., 2020). Impairing mitochondrial respiration in astrocytes also increases astrogliosis, and neurodegeneration in a mouse stroke model (Fiebig et al., 2019). In this regard, mitochondrial bioenergetics and metabolic flexibility would be a key determinant in the survival and onset of neurodegeneration.
17b -hydroxysteroid dehydrogenase type 10 (17b HSD10) is a mitochondrial short-chain dehydrogenase/reductase (SDR; He et al., 1998) with a wide-ranging substrate specificity (Powell et al., 2000;He and Yang, 2006). It includes the catalysis of the third step of b -oxidation , catabolism of isoleucine (Luo et al., 1995), and neurosteroid metabolism of 17b -estradiol, allopregnanolone, and 3a,5aÀ3,21-dihydroxypregnan-20-one (Belelli and Lambert, 2005). Independent of its catalytic function, 17b HSD10 has a structural role in mitochondria as one of the three components of ribonuclease P complex which is essential for mitochondrial tRNA processing (Holzmann et al., 2008;Deutschmann et al., 2014). Missense mutations in the gene encoding 17b HSD10 cause progressive neurodegeneration which affects motor skills, speech, and vision in childhood (Zschocke, 2012). 17b HSD10 has also been known under the name of amyloid binding alcohol dehydrogenase (ABAD) because of its capacity to bind b -amyloid (Ab ; Lustbader et al., 2004;Yan et al., 2007), which is a key pathologic marker in AD (Querfurth and LaFerla, 2010). The complex was later proposed to obstruct the binding of 17b HSD10 to cyclophilin D, leading to increased formation of mitochondrial permeability pore . However, it is still unclear whether this is the sole mechanism through which 17b HSD10 affects AD pathology and whether it is only neurons that mediate these mechanisms. He et al. (2005) proposed that 17b HSD10 is overexpressed in astrocytes surrounding amyloid plaques and proposed that astrocytic 17b HSD10 might be important for allopregnanolone metabolism in these cells. While there is no further confirmation of their immunohistological data, others have suggested that 17b HSD10 might not be detected in astrocytes (Fukuzaki et al., 2008). Nevertheless, a recent transcriptomic analysis of the adult mouse brain also shows that 17b HSD10 levels are significantly higher in astrocytes as compared with neurons (Saunders et al., 2018). Although these initial studies proposed that astrocytic 17b HSD10 may be important for normal and disease physiology, the non-neuronal role of the protein remains elusive.
The current study addresses this lack in literature and offers direct evidence that the protein is involved in the astrocytic response to metabolic and amyloidogenic stress. To isolate astrocyte-specific events in the absence of other cell types, we examined the structural and enzymatic role of 17b HSD10 in an in vitro culture model of murine astrocytes. We first established that the enzyme is expressed and active at comparable levels in cortical, hippocampal, and cerebellar astrocytes. Furthermore, we found that the expression and enzymatic activity of 17b HSD10 are affected by amyloidogenic and ischemic stress. In turn, the overexpression of catalytically active (as opposed to inhibited or mutated and inactive) 17b HSD10 inhibits respiration and has multiple implications for mitochondrial function and astrocytic response to ischemia and Ab treatment.

Materials and Methods
Primary culture and tissue slice preparation The current methodology was designed and performed in accordance with the United Kingdom Animals (Scientific Procedures) Act 1986 and was approved by the University of St Andrews Animal Welfare and Ethics Committee.
The culture medium was replaced every 48 h, which was always preceded by a gentle wash with PBS to remove loosely attached populations of microglia and oligodendrocytes (Lange et al., 2012). After a week, in vitro cultures were briefly washed with PBS1TrypLE (5%) to further facilitate the removal of non-astrocytic populations. The cultures were then dissociated with 2 mL TrypLE (ThermoFisher, #12604-039) for 8 min at 37°C and seeded into appropriate cell culture vessels precoated with PDL at a density of 15,000 cells per cm 2 . The resulting cell populations had comparable purity levels as indicated by the traditional astrocytic marker GFAP (glial fibrillary acidic protein) and the pan-astrocytic marker aldehyde dehydrogenase 1 family, member L1 (ALDH1L1; Cahoy et al., 2008) with preparations comprising over 98% ALDH1L1 1 and over 85% GFAP 1 cells.
Tissue slices were prepared by humanely killing neonatal and adult male mice, isolating and slicing the brain using McIlwain tissue chopper to generate 350-mm-thick brain slices. The slices subsequently placed in 4% paraformaldehyde (PFA) and fixed overnight at 4°C.
The desired sequences were PCR-amplified out of the pcDNA3 vector while 59 XbaI and 39 SalI restriction sites were added. The newly generated donor sequences as well as the recipient lentiviral construct pLenti-CMV-GFP-Puro (658-5; Addgene, #17448, a gift from Eric Campeau and Paul Kaufman) were digested with restriction enzymes XbaI and SalI (NEB). The GFP sequence was removed and the desired MTS-17b HSD10/ Y168G were inserted instead. One Shot TM Stbl3 TM competent Escherichia coli (ThermoFisher, #C737303) were used to produce and isolate the newly generated pLenti-CMV-MTS-17b HSD10/Y168G-Puro. QIAGEN Miniprep (QIAGEN, #27104) and Maxiprep (QIAGEN, #12163) kits were used to purify DNA from overnight liquid bacterial cultures following manufacturer's instructions.

Alamar blue
Viability of primary astrocytes was assessed via an alamar blue assay (ThermoFisher, #DAL1025). Cells were seeded into PDL-coated black wall clear bottom 96-well plates (Greiner, #655090). Following experimental treatment, alamar blue was added (10% v/v) and incubated for 4 h at 37°C. Resazurin, the active compound of the assay, is metabolized to resorufin in living cells. Resazurin fluorescence was detected using SpectraMaxM2e spectrophotometer (Molecular Devices) using 570-nm excitation and 590-nm detection wavelengths. The signal was corrected for background fluorescence and the viability was calculated as a % control (untreated) cells.

Lactate dehydrogenase (LDH) assay
Cytotoxicity was determined using LDH assay (ThermoFisher, #88953), a colorimetric assay detecting LDH release into culture medium following cell membrane rupture. Following manufacturer's instructions, 50% (v/v) LDH reaction mix was added to cell medium removed from the treated astrocytes. Following a 30 min incubation at room temperature, LDH concentration was measured by detecting absorbance at 490 and 680 nm using a SpectraMaxM2e spectrophotometer (Molecular Devices). The % cytotoxicity was calculated via the following formula: %cytotoxicity ¼ LDH treatment À LDH spont LDH max À LDH spont Â 100: LDH treatment indicates LDH activity measured in astrocytes exposed to different treatments, LDH spont is the level of activity in water-treated control; LDH max is the maximum LDH activity in medium where cells were lysed with 10Â lysis buffer (supplied with the kit).

Immunocytochemistry and immunohistochemistry
Mitochondria were stained with MitoTracker Red CMXRos (ThermoFisher, #M7512). The reagent was prepared as 1 mM stock solution in DMSO and further diluted to 100 nM working concentration in serum-free, phenol red-free astrocyte culture medium. The staining solution was incubated with astrocytes grown on PDL-coated glass coverslips (VWR, #631-0150P) for 20 min at 37°C.

Mitochondrial superoxide detection
Mitochondrial superoxide generation in astrocytes was determined using MitoSOX Red Mitochondrial Superoxide Indicator (ThermoFisher, #M36008). The reagent allows for a highly selective detection of mitochondrial superoxide in live cells, whereby the oxidation of a cationic derivative of dihydroethidum by superoxide generates a highly fluorescent product. The reagent was stored as a 5 mM stock in DMSO, which was diluted further to a 5 mM working solution in serum-free, phenol red-free astrocyte culture medium. The cells were incubated with the working solution for 5 min, washed with PBS and fluorescence was measured using FLUOstar Optima microplate reader (BMG; excitation at 510 nm and emission at 620 nm). Background fluorescence was subtracted, and superoxide generation was calculated as a % control by normalizing to the fluorescence signal recorded in control cells. H 2 O 2 -treated cells were used as a positive control confirming MitoSOX reliably detected oxidative stress generation in primary astrocytes.
Following each experiment, cells in each well were lysed on ice in RIPA buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM sodium orthovanadate, and 1 mM NaF) and the protein concentration was determined using BCA assay following the manufacturer's instructions (ThermoFisher, #23225). Data from each experiment was exported using Wave Software (Agilent) and normalized to blank control wells and to the protein concentration measured for each well. and exported for further statistical analysis.

17bHSD10 enzymatic activity assay
The enzymatic activity of 17b HSD10 measured using the fluorogenic probe (À)-cyclohexenyl amino naphthalene alcohol [(À)-CHANA; Muirhead et al., 2010]. The oxidation of this molecule results in the formation of a highly fluorescent ketone cyclohexenyl amino naphthalene ketone (CHANK). Furthermore, the selectivity of the probe has been previously confirmed in HEK293T cells, where showed minimal fluorescence in control and high metabolism in 17b HSD10-overexpressing cells (Muirhead et al., 2010). CHANA was kindly provided by Dr. Laura Aitken (University of St Andrews), reconstituted in DMSO and further diluted into standard astrocyte cell culture medium buffered with 10 mM HEPES before the experimental procedure. Cells were seeded into in 96-well black wall plates, experimentally treated and exposed to CHANA with final assay concentration of 20 mm. The conversion of CHANA to CHANK was measured over 90 min at 37°C using FLUOstar Optima microplate reader (BMG; excitation wavelength 380 nm, emission wavelength 520 nm). The assay quality was ensured for each experiment by, calculating Z' values: where the SDs (s ) and the means (m) of triplicate positive (p) and negative (n) controls were measured. Z' values above 0.5 are accepted to indicate reliable activity detection and excellent assay quality (Zhang et al., 1999) and only Z' values between 0.6 and 0.8 were used for the currently reported results. The slope of fluorescence intensity change generated by CHANA to CHANK metabolism was calculated and used as 17b HSD10 activity indicator. All values were normalized and shown as % change from control cells and presented as "CHANA turnover" throughout the Results.

Mitochondrial morphology measurement and colocalization analysis
The mitochondrial network morphology was assessed using the ImageJ-supported toolset MiNA (Mitochondrial Network Analysis) developed by Valente et al. (2017). The source code for the analysis is kindly provided by the authors in GitHub repository: https://github.com/StuartLab/ MiNA/tree/master.
Confocal microscopy images of MitoTracker Red CMXRos-stained mitochondrial networks were preprocessed to obtain high-contrast images permitting for the reliable quantification of mitochondrial network morphology in ImageJ. The output of MiNA was used to evaluate branching morphologies exhibited by mitochondrial networks of primary astrocytes. Optimizing the analysis protocol for this model allowed to distinguish between unbranched structures (puncta and rods) versus branched network formations (Valente et al., 2017). Briefly, the workflow started with the conversion of the images to binary, skeletonizing and analyzing each skeleton. The currently described analysis uses four parameters discussed in the original paper by Valente et al. (2017) to characterize mitochondrial structure: (1) mitochondrial footprint: showing the area of detected mitochondrial signal (mm 2 ); (2) mean branch length: indicating the average length of all the lines used to represent the mitochondrial structures (mm); (3) total branch length: showing the mean of the sum of the lengths of branches for each independent structure representing part of the mitochondrial skeleton in a single cell (mm); (4) network branching: showing the mean number of attached lines used to represent each structure.
Colocalization of 17b HSD10 and the astrocytic marker GFAP was quantified using confocal images of acute cortical, hippocampal, and cerebellar brain slices. The images shown in Extended Data Figure 1-1 were analyzed through the JACoP plugin in ImageJ (Bolte and Cordelières, 2006). The Manders' coefficient was used to quantify the overlap of the GFAP signal with the 17b HSD10 signal to show what portion of the GFAP-positive cells also express 17b HSD10. Manders' coefficient is based on Pearson's correlation coefficient with average intensity values being removed from the equation (Manders et al., 1992). The coefficient varies from 0 (no colocalization) to 1 (100% colocalization; Bolte and Cordelières, 2006).

Stress treatments
Oxygen deprivation of primary astrocytes was conducted in a Don Whitley H35 hypoxystation. The oxygen sensors of the incubated station were calibrated before each use and set to 0.5% O 2 , 5% CO 2 at 37°C according to manufacturer's instructions. The PBS and culture medium used to wash and treat the cells were preconditioned in the hypoxystation for 1 h before the start of the experiment. The astrocytic cultures were washed with deoxygenated PBS and exposed to 0.5% O 2 levels, 5% CO 2 at 37°C for 6 h. The cultures were re-oxygenated by being transferred to normal culture conditions.
Oligomeric Ab (1-42) was prepared from lyophilized synthetic Ab (1-42) (1 mg) dissolved in 221.53 ml 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP; Sigma #105228) and dispensed into 10 ml aliquots which were desiccated in a ventilated safety cabinet for 60 min. The resulting peptide films were sealed and stored at À20°C. A final concentration of 5 mM stock was obtained by adding 2 ml DMSO, followed by a 15-min sonication and addition of 48-ml PBS. The sample was vortexed for 30 s and used for treatment of primary astrocytic cultures. The treatment was administered for 48 h before experiments.

Statistical analysis
Statistical analysis was performed using SPSS Statistics 26 (IBM) while figures were created with GraphPad Prism (GraphPad Software). BioRender was used to create the facilitating figures, including the graphical abstract. The currently presented data constitutes at least three independent biological replicates (i.e., independent primary culture preparations). All data have been tested for normal distribution using Shapiro-Wilk test. Comparisons between more than two groups were conducted using one-way ANOVA, while multiple groups and conditions were compared using twoway or mixed ANOVA. Preceding ANOVA, data were also tested for homoscedasticity using Levene's test. When conducting repeated measures or mixed ANOVA, Mauchly's sphericity test was conducted as well. If homoscedasticity or sphericity assumptions were violation, Greenhouse-Geisser corrected values were reported. Since mitochondrial morphology analysis required treating individual astrocytes as datapoints, we utilized a mixed model analysis with culture replicate as a random factor to control for changes that could be attributed to differences between cultures. Tukey's post hoc comparisons were used following ANOVA. The selected confidence interval for statistical significance was 95%; therefore, only p , 0.05 was considered significant. Further statistical details are provided in Table 1.

Results
Astrocytes from the mouse cortex hippocampus and cerebellum show similar levels of 17bHSD10 expression and activity Because of the limited amount of literature addressing the role of 17b HSD10 in astrocytes, there had been no evidence of the 17b HSD10 protein being expressed in these cells. While one study suggested that this protein may be expressed in hypertrophic astrocytes surrounding amyloid plaques in AD patients , others claimed that 17b HSD10 may not be at all found in murine astrocytes (Fukuzaki et al., 2008). Therefore, our first aim was to confirm whether the protein is present and enzymatically active in astrocytes from different regions of the mouse brain.
We found that 17b HSD10 immunostaining co-localized with mitochondria in cultured mouse astrocytes from the cortex, hippocampus, and cerebellum (Fig. 1A). Since previous reports show that total 17b HSD10 protein is higher in the hippocampal as compared with cortical and cerebellar tissue , we next asked whether these disparities could be driven by differences in the astrocytic population. The protein expression levels were similar between astrocytes from the three brain regions (Fig. 1B,C). To confirm that potential differences were not obscured by a variation in mitochondrial levels, COXIV was used as a mitochondrial control. As expected, COXIV was similarly expressed in the three astrocytic samples and normalizing 17b HSD10 levels to COXIV did not reveal further differences ( Fig. 1C). We provide further immunohistochemical evidence that 17b HSD10 is endogenously expressed in noncultured cortical, hippocampal, and cerebellar astrocytes of neonatal (2 and 7 d of age) and adult mice (two and four months of age; Extended Data Figs. 1-1, 1-2, 1-3, 1-4).
Next, we employed an enzymatic assay, which can report 17b HSD10-mediated break-down of CHANA to its fluorescent metabolite CHANK (Muirhead et al., 2010) to compare 17b HSD10 activity between astrocytes ( Fig. 1D; Materials and Methods, Mitochondrial respiration). No differences were found in the levels of 17b HSD10 activity in astrocytes from the cortex, hippocampus, and cerebellum ( Fig. 1D,E). Therefore, 17b HSD10 is expressed and enzymatically active in astrocytes from different regions of the murine brain. Any reported 17b HSD10 expression and activity differences between brain regions in normal conditions could not be attributed to astrocytes alone when controlling for total cell numbers and total protein content.
The overexpression of the enzyme was comparable between both wt-17b HSD10 and mut-17b HSD10 ( Fig. 2A). Overexpression of either variant caused no changes in the viability and mitochondrial superoxide levels, while only wt-17b HSD10 overexpression increased the reported levels of 17b HSD10 activity in cortical astrocytes (Fig. 2B).
While confirming the successful mitochondrial targeting of the overexpressed protein (Fig. 2C), we also found that overexpression of 17b HSD10, both wt and mut, induced apparent changes in mitochondrial network architecture. In order to test and quantify these observations, we used the ImageJ-supported toolset MiNA (Valente et al., 2017). 17b HSD10 overexpression did not alter the total area occupied by mitochondria (mean area across conditions: 691.21 6 10.31 mm; Fig. 2D), which is consistent with the equal mitochondrial number suggested by the mitochondrial loading control (COXIV) reported in Figure 1. On the other hand, branch parameters indicated that mitochondrial networks with overexpressed wt-17b HSD10 and mut-17b HSD10 have increased length of individual structures by 16% (Fig. 2E) and elongation of branched mitochondrial structures by .40% (Fig. 2F). However, the number of mitochondrial network branches remained unchanged (Fig. 2G). These findings were consistent with elongated mitochondrial phenotype, pointing to altered mitochondrial dynamics.
Overexpression of catalytically active, but not mutated/inactive, 17bHSD10 inhibits mitochondrial respiration in cortical astrocytes After characterizing our model systems, we proceeded to addressing the effect of wt-17b HSD10 and mut-17b HSD10 overexpression on mitochondrial respiration in cortical astrocytes (Fig. 3). Utilizing mitochondrial stress testing (Fig. 3A), we aimed to assess the effects of 17b HSD10 overexpression on basal mitochondrial respiration and the capability of astrocytes to respond to an energetic demand by upregulating mitochondrial respiration.
The first finding was that the overexpression of overexpressed wt-17b HSD10 did not affect mitochondrial respiration (Fig. 3C,D). However, it decreased maximal respiratory capacity by 45% from control (decrease by 5539 6 931 pmol/min/mg protein) and spare respiratory capacity by 65% (decrease by 4306 6 610 pmol/min/ mg protein), without affecting basal respiratory capacity, ATP production and proton leak (Fig. 3D). Control and mut-17b HSD10-overexpressing cells were able to upregulate their respiration by almost 70% from baseline. While wt-17b HSD10-overexpressing cells were also able to increase their respiration when challenged to reach maximum oxygen consumption rate (OCR) through FFCP administration which induced maximal oxygen consumption at Complex IV (t (20) = 2.145, p = 0.044), this increase was significantly reduced as compared with the other two conditions, reaching only 30% above baseline (Fig. 3E).
In order to confirm whether all the respiratory changes associated with wt-17b HSD10 overexpression were caused by the excess enzymatic activity, we employed a known 17b HSD10 inhibitor. AG18051 which is currently the most potent available 17b HSD10 inhibitor (IC 50 92 nM; Kissinger et al., 2004;Vinklarova et al., 2020). When primary astrocytes overexpressing wt-17b HSD10 were pretreated with 20 mm AG18051, all respiratory parameters were restored to control levels ( Fig. 3C-E). This further confirmed that the excess enzymatic activity of the protein was the main driver of respiratory inhibition in these cells. Furthermore, the inhibitor increased maximal respiration and spare respiratory capacity in cortical astrocytes with endogenous levels of 17b HSD10 (Extended Data Fig. 3-1).
We next sought out to find whether mitochondrial decoupling or respiratory inhibition through different ETC enzymes would affect 17b HSD10 activity. Mitochondrial decoupling though addition of FCCP did not affect 17b HSD10 activity (Fig. 3F). However, inhibition of respiration through Complex I (via Rot), Complex III (via AA), and Complex V (using OM), caused a significant decrease in 17b HSD10 activity in all astrocytes ( Fig. 3F; Extended Data Fig. 3-1).
Therefore, our findings suggest that excessive 17b HSD10 activity decreases maximal astrocytic respiration and spare respiratory capacity, without affecting baseline respiratory levels. However, inhibition of ETC Complex I, III, or V causes activity down-regulation of both endogenous and overexpressed 17b HSD10 (both mut and wt). Altogether, these findings suggest that while 17b HSD10 activity is dependent on ETC activity, overexpressed activity of the enzyme could inhibit the ability of these cells to upregulate their mitochondrial respiration to meet a sudden increase in energy demand.
Ischemia-reoxygenation increases 17bHSD10 activity and expression in astrocytes, affecting ROS generation and mitochondrial network morphology Ischemic conditions during stroke are characterized by the disruption of oxygen and nutrient supply to the affected part of the brain. The complex nature of this insult implicates varying degrees of neurodegeneration, and the recovery is heavily dependent on astrocyte mediatednutrient support to neurons (Rossi et al., 2007;Chaitanya et al., 2014). One early study by Yan et al. (2000) found that 17b HSD10 expression is elevated in neurons proximal to the infracted area in a mouse stroke model. Overexpressing 17b HSD10 in this model was found to enhance TCA flux and ATP levels, while decreasing neurologic deficit and infarction volume . The relevance of astrocytes, which are a major factor in the recovery from this type of injury was never addressed (Rossi et al., 2007;Guo et al., 2012;Chaitanya et al., 2014).
Therefore, we employed a simplified in vitro stroke model to address this lack in literature. Primary cortical astrocytes were incubated in a hypoxystation with 0.5% O 2 , 5% CO 2 at 37°C for 6 h and returned to normal conditions for 2 h before assessment. When glucose (5.5 mM) was present in their environment, 17b HSD10 levels remained similar to control (Fig. 4A,B). However, 17b HSD10 protein levels were elevated when no glucose was present during IR. Although COXIV expression levels remained stable in these conditions, VDAC1 was increased in the absence of glucose. These findings are consistent with the knowledge that the mitochondrial permeability transition pore (mPTP) opens following IR injury (Reichert et al., 2001) D G F E B C A Figure 2. Overexpressing 17b HSD10 in cortical astrocytes induces elongated mitochondrial networks, regardless of 17b HSD10 enzymatic activity. A, 17b HSD10 expression levels were assessed through Western blot analysis 10 d after lentiviral transduction. A representative immunoblot of three independent primary culture preparations and the quantification of 17b HSD10 in these cells showed that the lentiviral protocol induced 20-fold increase in protein expression with both the mutant and wt versions of 17b HSD10 as compared with control astrocytes (F (2,18) = 16.67, p , 0.001). Bar graphs represent the mean 6 SEM. Analysis through one-way ANOVA with Tukey's post hoc comparisons on n = 6 independent culture preparations. B, 17b HSD10 enzymatic activity as measured by CHANA turnover rate was significantly increased only when the wt variant of 17b HSD10 was overexpressed, while the mutated version did not cause significant increase (F (2,24) = 13.24, p , 0.001). Bar graphs represent the mean 6 SEM. Analysis through one-way ANOVA with Tukey's post hoc comparisons on n = 9 independent culture preparations. C, Representative confocal microscopy images of astrocytes immunostained for 17b HSD10 and mitochondrial dye MitoTracker Red CMXRos confirmed mitochondrial expression of the enzyme in both control and overexpressed conditions. Scale bars: 10 mm. The arrows with dotted lines in the control image indicate predominant mitochondrial morphology in normal conditions; arrows with solid lines in mut-17b HSD10 and wt-17b HSD10 panels indicate abnormal elongated and highly branched network morphology. D, Mitochondrial footprint was uniform across conditions (F (2,751) = 0.308, p = 0.75, n.s.). E, Mean branch length was greater in mut-17b HSD10 and wt-17b HSD10 populations (F (2,712) = 30.16, p , 0.004).  . Increased 17b HSD10 activity decreased mitochondrial respiration, while ETC inhibition decreased the activity of the enzyme and AG18051 countered these effects. A, A schematic representation of the ETC targets and utilized inhibitors used for the mitochondrial bioenergetic test. B, Timescale of the experiment, inhibitor administration, and measurement parameters utilized in the respiratory test. C, Mitochondrial respiration was assessed by measuring OCR which was normalized to the protein content in each sample. This profile was further used to calculate the respiratory parameters in the next panel. D, Overexpression of mut-17b HSD10 did not affect the respiratory parameters (ps . 0.05); however, wt-17b HSD10 overexpression reduced maximal respiration (F (3,40) = 14.22, p , 0.001), as well as spare respiratory capacity (F (3,40) = 26.737, p , 0.001) in cortical astrocytes, while basal respiration (F (3,40) = 2.04, p = 0.12, n.s.), ATP production (F (3,40) = 1.93, p = 0.14, n.s.), and proton leak (F (3,40) = 1.43, p = 0.25, n.s.) remained unaffected. These effects were recovered to baseline when cells were pretreated with 17b HSD10 inhibitor AG18051 (20 mm; ps . 0.05). E, Similarly, while all cells with normal activity levels of 17b HSD10 were able to upregulate respiration when challenged, high activity of the protein reduced this metabolic compensation (F (3,40) = 13.89, p , 0.001). F, Both endogenous and overexpressed 17b HSD10 activity was inhibited by OM, AA, and Rot (control: F (4,30) = 29.99, p , 0.001; mut-17b HSD10: F (4,25) = 13.73, p , 0.001 and wt-17b HSD10 (F (4,25) = 12.09, p , 0.001), while mitochondrial decoupling with FCCP did not affect 17b HSD10 activity (ps . 0.05). Graphs represent the mean 6 SEM. Compounds: FCCP (carbonyl cyanide-4 (trifluoromethoxy)phenylhydrazone: 4 mM), OM (1.5 mM), AA (5 mM), Rot (5 mM). Treatment time: acute injection. AG18051 (20 mm) was administered for 24 h before the experiment. One-way between-subjects ANOVA with Tukey's post hoc comparisons (n = 4-6 independent primary culture preparations); n.s, non-significant, *p , 0.05, **p , 0.01, ***p , 0.001. Abbreviations: cyanide-4-(trifluoromethoxy) phenylhydrazone (FCCP), antimycin A (AA), oligomycin (OM), rotenone (ROT). Extended Data Figure 3-1 shows the effects of AG18051 on respiratory function of astrocytes with endogenous 17b HSD10 levels.
This metabolic supplementation prevented the increase of both 17b HSD10 and VDAC1 expression levels (Fig. 4A,B).
In addition to altering mitochondrial protein expression, IR caused a decrease in astrocytic viability (Fig. 4C) with a concomitant increase in cytotoxicity as measured by LDH release (Fig. 4D), while overexpressing 17b HSD10 did not affect these changes. Although the timescale of the current experiment did not assess long-lasting effects on astrocytes, the current observations suggest that 17b HSD10 might be a crucial factor for metabolic recovery following such an injury.
As expected, based on observed protein expression shifts, the detected 17b HSD10 activity increased following an IR insult, and this change was evident in normal, as well as astrocytes with overexpressed mut-17b HSD10 and wt-17b HSD10 (Fig. 5A). These changes were mirrored by mitochondrial superoxide generation, whereby IR induced an increase in all cells, but the effect was particularly pronounced in those with overexpressed wt-17b HSD10 (Fig. 5B). Importantly, both enzymatic activity elevation and superoxide generation were countered by the addition of AG18051. However, the inhibitor reduced IR-induced cytotoxicity only in wt-17b HSD10-overexpressing astrocytes (Fig. 5C).
The current results were also consistent with the literature showing that cerebral ischemia causes increased mitochondrial fission and increased mitophagy in astrocytes following IR (Lan et al., 2018; Quintana et al., 2019). The phenotype we observed in normal astrocytes exposed to IR is consistent with mild mitochondrial fragmentation ( Fig. 5F-H) and pronounced reduction of mitochondrial footprint (Fig. 5E). Interestingly, although overexpression of either wt-17b HSD10 or mut-17b HSD10 caused mitochondrial network elongation with a hyperfused phenotype, mut-17b HSD10 overexpression did not prevent mitochondrial fragmentation in astrocytes exposed to IR, while the opposite was seen with wt-17b HSD10 overexpression. Additionally, mut-17b HSD10 overexpression inhibited the decrease of mitochondrial territory observed in normal astrocytes exposed to IR. On the other hand, wt-17b HSD10 overexpression not only halted the decrease, but further increased mitochondrial footprint following the insult (Fig. 5E). What is more, wt-17b HSD10overexpressing cells did not exhibit the characteristic mitochondrial fragmentation (Fig. 5F-H).
Collectively, our data suggested that mut-17b HSD10 overexpression caused a mitochondrial hyperfused phenotype at baseline, and was not able to counter fragmentation associated with IR. The wt-17b HSD10 overproduction not only countered mitochondrial fragmentation but potentiated mitochondrial elongation following IR. Excessive 17b HSD10 activity also increased mitochondrial superoxide generation in IR conditions.
injury (Shao et al., 2020). Although the current results do not address the underlying mechanisms, we report that abnormally high 17b HSD10 enzymatic activity (following wt-17b HSD10 overexpression) potentiates ROS, alters network morphology, and halts mitochondrial reduction in cortical astrocytes exposed to IR. We speculate that the phenotype is likely to affect the mitochondrial network balance and hamper the recovery following this type of injury; however, further mechanistic studies should address these phenomena in more detail. Figure 5. 17b HSD10 activity increased following IR and this affected superoxide generation and mitochondrial network morphology. A, 17b HSD10 activity increased following IR in all astrocytes and AG18051 inhibited this elevation (F (4,45) = 20.88, p , 0.001); n = 3 independent primary culture preparations. Analysis via two-way between subjects ANOVA with Tukey's post hoc test. B, Superoxide generation was increased following the insult and the effects were ameliorated by AG18051 in all astrocytes but mut-17b HSD10-overexpressing cells. The increase was of higher magnitude in wt-17b HSD10-overexpressing cells, and the AG18051-mediated decrease was particularly pronounced in this condition (interaction: F (4,27) = 3.18, p = 0.029). Analysis on n = 3 independent biological replicates using two-way between subjects ANOVA with Tukey's post hoc test. C, AG18051 did not rescued the cytotoxic effect induced by IR only in astrocytes overexpressing wt-17b HSD10 (interaction: F (4,111) = 6.03, p , 0.001). D, Representative confocal images of mitochondrial networks in astrocytes in control and IR conditions. Arrows with dotted line indicate fragmented mitochondrial morphology with reduced branching, while arrows with solid line show elongated and highly branched network morphology. Scale bars: 10 mm. E, Mitochondrial footprint changes following IR depended on 17b HSD10 expression and catalytic activity (interaction between treatment and 17b HSD10 expression phenotype: F (2,243) = 96.80, p , 0.001) with substantial reduction in normal astrocytes, nonsignificant effect in mut-17b HSD10-overexpressing cells and an increase in the wt-17b HSD10-overexpressing group. F, Average branch length decreased following IR and this was significant in 17b HSD10-overexpressing astrocytes (main effect of 17b HSD10 expression: F (2,243) = 14.29, p = 0.015). G, Total branch length changes following IR depended on 17b HSD10 expression and catalytic activity (interaction between treatment and 17b HSD10 expression phenotype: F (2,261) = 7.77, p = 0.041) with only wt-17b HSD10-overexpressing astrocytes resisting the decrease in this parameter following IR insult. H, Network branching also showed significant interaction between treatment and 17b HSD10 expression phenotype (F (2,243) = 22.14, p = 0.007), whereby wt-17b HSD10 showed an increase of branch number following treatment. Analysis via two-way between subjects ANOVA with Tukey's post hoc test, n = 3 independent biological replicates, 45 cells per condition; n.s, non-significant, *p , 0.05, **p , 0.01, ***p , 0.001. Abbreviations: ischemiareoxygenation (IR). Treatment: hypoxia, 6 h (0 mM glucose 0.5% O 2 ; 5% CO 2 at 37°C) followed by 2 h of reoxygenation.
Oligomeric Ab (1-42) alters 17bHSD10 expression and activity in astrocytes, inducing ROS generation and alterations in mitochondrial network architecture Until now, the literature studying the role of 17b HSD10 in AD has mainly focused on neuronal cells, where Ab binds 17b HSD10 inducing a number of neurotoxic effects (Yan et al., 1999;Lustbader et al., 2004;Takuma et al., 2005;Yan and Stern, 2005;Borger et al., 2011). However, non-neuronal populations such as astrocytes provide essential metabolic support to the brain and clear amyloid plaques from the brain. While soluble Ab (1-42) oligomers rather than fibrils have been proposed to drive the main toxic events in AD pathology, 17b HSD10 also has a higher affinity for binding Ab (1-42) as compared with Ab (1-40) (Chen et al., 2017;De et al., 2019;Ciudad et al., 2020) and the affinity is determined by the oligomerization of the peptide (Hemmerová et al., 2019). In line with these studies, we opted for utilizing oligomeric Ab (1-42) treatment in astrocytes.
First, we confirmed that similar to observations in neurons and total expression in cortical matter Lustbader et al., 2004;Yan and Stern, 2005), cortical astrocytes exposed to oligomeric Ab (1-42) upregulated their 17b HSD10 protein levels, without affecting COXIV and VDAC1 (Fig. 6A,B). Importantly, the overexpression of wt-17b HSD10 also caused a significant decrease of viability in an amyloid-rich environment (Fig. 6C). Cytotoxicity levels were significant in all cells treated with amyloid; however, the overall magnitude did not exceed 5% (Fig. 6D). Yan et al. (1999) reported that the currently studied mutated variant of 17b HSD10 (Y168G), can bind Ab  with similar affinity to the wt variant. This allowed us to study the impact of 17b HSD10 overexpression in astrocytes in an amyloid-rich environment, while only eliminating the activity-mediated aspect of the interaction. Interestingly, we found that cortical astrocytes overexpressing the catalytically inactive form of the protein (mut-17b HSD10) showed similar response to amyloid in the context of viability, 17b HSD10 activity, ROS generation, mitochondrial territory and morphology.
We further studied mitochondrial architecture in these cells and found that Ab  reduced mitochondrial territory in all astrocytes (Fig. 7E). While mitochondrial fragmentation was also observed in all amyloid-treated samples, the phenotype was particularly exacerbated when wt-17b HSD10 was overexpressed in cortical astrocytes (Fig. 7F-H).

Discussion
Astrocytes play an essential role in neuroprotection and recovery from ischemic insult (Guo et al., 2012;Zamanian et al., 2012;Cai et al., 2020) and AD-associated metabolic and amyloidogenic stress (De Strooper and Karran, 2016;Liddelow et al., 2017;Grubman et al., 2019;Mathys et al., 2019). Although their energy metabolism has multiple implications in these pathologies, the role of their mitochondria has been generally overlooked since the majority of their ATP is derived through glycolysis (Pellerin and Magistretti, 1994;Magistretti and Pellerin, 1999;Magistretti, 2006). It is now becoming clear that mitochondria are crucial for astrocytic metabolic flexibility during metabolic challenges, malignant transitions, and region-dependent vulnerabilities to Figure 7. The catalytic activity of overexpressed 17b HSD10 decreased following treatment with oligomeric Ab  , and this was associated with exacerbated superoxide generation and mitochondrial fragmentation. A, Oligomeric Ab (1-42) reduced detected 17b HSD10 activity only in wt-17b HSD10-overexpressing astrocytes, and AG18051 co-administration abolished this difference (interaction between treatment and 17b HSD10 expression: F (4,63) = 178.45, p , 0.001). B, Oligomeric Ab (1-42) increased superoxide generation in all astrocytes and the effect was of greatest magnitude in the wt-17b HSD10-overexpressing group, while AG18051 countered the increase in all conditions (F (2,63) = 10.34, p , 0.001). Bar graphs show mean 6 SEM; n = 3 independent primary cultures; two-way between subjects ANOVA with Tukey's post hoc comparisons. C, AG18051 did not rescue the cytotoxic effect induced by Ab (1-42) (ps . 0.05). D, Representative confocal images of astrocytic mitochondrial networks in control and Ab (1-42) treatment. Arrows with dotted line indicate fragmented mitochondrial morphology which was particularly pronounced in wt-17b HSD10-overexpressing astrocytes. Scale bars: 10 mm. E, Mitochondrial footprint was reduced in all astrocytes following amyloid treatment (treatment main effect: F (1,702) = 30.79, p = 0.031) in all three groups of astrocytes (F (2,702) = 0.50, p = 0.640). F, Changes in mean branch length depended on both 17b HSD10 expression and amyloid treatment (interaction F (2,702) = 21.17, p = 0.007). G, Total branch length also displayed a significant interaction (F (2,702) = 12.58, p = 0.019) whereby amyloid reduced summed branch length only in cells overexpressing the catalytically active form of the protein. H, Network branching also showed significant interaction between treatment and protein expression (F (2,702) = 13.59, p = 0.016) with only wt-17b HSD10-overexpressing cells showing significant decrease in the number of mitochondrial branches following amyloid treatment. Analysis via two-way between subjects ANOVA with Tukey's post hoc test, n = 3 biological replicates with 120 cells analyzed per treatment; n.s, non-significant, *p , 0.05, **p , 0.01, ***p , 0.001. Treatments: Ab (1-42) 1 mM for 48 h; AG18051 20 mm.
neurodegeneration (Polyzos et al., 2019;Ryu et al., 2019;Dematteis et al., 2020;Habib et al., 2020). The current study provides novel insights into the role of astrocytic mitochondria in pathology by showing that 17b HSD10, a hypothesized metabolic switch between energy and neurosteroid metabolism (Borger et al., 2011), has increased expression and catalytic activity following metabolic and amyloidogenic stress, which subsequently affects their mitochondrial function.
Existing reports on the role of 17b HSD10 in astrocytes are predominantly theoretical, however they propose that the protein is upregulated in astrocytes surrounding amyloidogenic plaques, which might disrupt steroid metabolism in the brains of AD patients He and Yang, 2006). Our findings support the idea that 17b HSD10 has an important role in astrocytic metabolism and neurodegenerative pathology. We further found that astrocytes from the murine cortex, hippocampus, and cerebellum express enzymatically active 17b HSD10 at comparable levels. Furthermore, overexpressing either the normal or the mutated (catalytically inactive) version of the enzyme did not alter the viability ( Fig. 2A), baseline mitochondrial respiration (Fig. 3D) and ROS generation (Figs. 5B, 7B) in cortical astrocytes under normal conditions. However, mitochondrially targeted overexpression of the protein, regardless of its catalytic activity, induced elongated mitochondrial architecture without affecting overall mitochondrial content these cells. Indeed, such elongation has been documented by Bertolin et al. (2015), who showed that 17b HSD10 promotes this phenotype by interfering with Drp1 recruitment to mitochondria. We further extended these findings by showing that the mechanism is not dependent on the catalytic activity of 17b HSD10, and that the phenomenon also impacts mitochondrial territory and network architecture during stress. However, the exact mechanistic underpinnings of these observations must be addressed in future studies.
Inhibition of mitochondrial ETC Complexes I, III, and V decreased 17b HSD10 enzymatic activity. However, overexpression of wt-17b HSD10 caused a decline in the ability of astrocytes to upregulate their respiration when metabolically challenged. Unlike the overexpression of the wt variant, the catalytically inactive form did not affect maximal respiration and spare respiratory capacity. The hypothesis that respiratory inhibition is catalytically-dependent was confirmed by the fact that AG18051, a potent 17b HSD10 inhibitor (Kissinger et al., 2004), countered respiratory inhibition in cortical astrocytes overexpressing wt-17b HSD10 and caused a small increase in spare respiratory capacity in astrocytes with endogenous 17b HSD10 levels. These findings are not consistent with existing literature on cancer biology showing that 17b HSD10 overexpression in PC12 cells increases their mitochondrial respiration through Complex IV activity upregulation (Carlson et al., 2015).
However, other studies found that mitochondrial respiration in neurons of 17b HSD10-overexpressing mice remains similar to control (Tieu et al., 2004;Takuma et al., 2005), while amyloid-rich environment induces a Complex IVmediated inhibition of the process (Takuma et al., 2005). In contrast, 17b HSD10 overexpression increases respiration and ATP production in mouse models of Parkinson's disease (Tieu et al., 2004) and ischemia . Therefore, the general discrepancy in the literature is related to the cell type and the metabolic demands under which 17b HSD10 is being studied. Furthermore, most of the literature characterizing 17b HSD10 impact on respiration utilized methodology where respiratory complex activity was measured in lysed samples through spectrophotometric methodology (Tieu et al., 2003;Takuma et al., 2005;Carlson et al., 2015), rather than intact cells as was performed in the current study. Importantly, ETC configuration in astrocytes is reportedly characterized by higher levels of free Complex I, while in neurons Complex I and Complex III are assembled into supercomplex formations. Previously published evidence suggests that these differences render the astrocytic ETC less "efficient" and produces higher levels of ROS in respiration (Lopez-Fabuel et al., 2016). Such differences could explain why the current observations are not consistent with data in cancer or neuronal cell models, further emphasizing that the role of 17b HSD10 could be dependent on cell type and specific metabolic conditions and stresses.
While 17b HSD10 enzymatic activity was dependent on ETC function in cortical astrocytes, excessive activity of the enzyme impaired the ability of these cells to upregulate their respiration during metabolic challenge. These changes could have profound impact on astrocytic function and survival in stress conditions associated with disease. As an example of such a malignant state, ischemic insult causes severe bioenergetic stress which forces astrocytes to use their glycogen reserves (Cai et al., 2020;Köhler et al., 2021), while reoxygenation imposes a severe load on mitochondrial respiratory chain, driving reverse electron transport through Complex II, causing elevated ROS generation in the form of superoxide generation at Complex I (Chouchani et al., 2014). Interestingly, our results showed that ischemia elevated 17b HSD10 expression and activity, however hypoxia in the presence of glucose did not. These observations were expected, since astrocytes are capable of anaerobic glycolysis to ensure lactate production for energy requirements and neuronal supply (Dienel and Hertz, 2005;Hertz, 2008). Therefore, 17b HSD10 might be particularly relevant in conditions where astrocytes cannot rely on glycolytic metabolism for their survival, suggesting that the enzyme is particularly important for metabolic flexibility in conditions of reduced glucose supply regardless of aerobic or anaerobic conditions. Consistent with literature, IR-exposed astrocytes had increased superoxide generation (Kowaltowski and Vercesi, 1999), increased voltage-dependent anion channel 1 (VDAC1) expression (Yang et al., 2021) and mitochondrial network fragmentation while decreasing mitochondrial territory (Lan et al., 2018;Quintana et al., 2019). The elimination of damaged mitochondria following ischemic stroke is crucial for recovery (Lan et al., 2018) and it is activated through several mechanisms, which include ischemia-induced mPTP opening, mitochondrial membrane potential disruption (Reichert et al., 2001;Wang et al., 2012), ROS elevation (Wang et al., 2012;Fan et al., 2019), and Drp1-mediated activation (Zuo et al., 2014).
However, the overexpression of catalytically active 17b HSD10 (wt-17b HSD10) increased both mitochondrial territory and fragmentation, while exacerbating ROS production. While we speculate that mitophagy and fusion-fission mechanisms underpin these effects, further studies should address these mechanisms in detail. While ROS activate mitophagy (Wang et al., 2012;Fan et al., 2019), this serves to reduce oxidative damage following IR injury (Livingston et al., 2019). Importantly, 17b HSD10 has been reported to induce mitochondrial elongation by abrogating Drp1 recruitment to mitochondria (Bertolin et al., 2015), while Drp1dependent mitophagy ensures the clearance of damaged mitochondria (Zuo et al., 2014). Pharmacological or siRNAmediated inhibition of the pathway causes decreased mitochondrial fragmentation, accumulation of damaged mitochondria through selective blocking of mitophagy and increased ROS generation following IR (Zuo et al., 2014), which could explain the currently observed phenotype. Although 17b HSD10 expression did not affect astrocytic survival immediately after the insult, the overexpressed 17b HSD10 activity is likely to potentiate mitochondrial damage in the long-term by simultaneously affecting ROS production, mitochondrial network morphology, mitophagy, and respiration. Yan et al. (2000) showed that high levels of neuronal 17b HSD10 in a mouse model of ischemic stroke are associated with higher ATP, which correlated with increased acetyl-CoA flux, increased ketone body (b -hydroxybutyrate; BHB) utilization and better recovery of the animals. Although such adaptation facilitates short-term recovery, astrocytic up-regulation of fatty acid oxidation in low glucose conditions imposes higher load on mitochondria for metabolic adaptation and escalating ROS and associated damage (Polyzos et al., 2019). This further suggests that the enzymatic role of 17b HSD10 in b -oxidation of ketone bodies such as BHB  and short branch-chained fatty acids (Luo et al., 1995) is of key importance to astrocytic metabolism during glucose deprivation.
Cerebral ischemia also increases the risk of developing AD (Ballard et al., 2000;Pluta and Ułamek, 2008). In addition, there are multiple mechanisms proposed to drive AD pathology and among the key hallmarks are amyloid pathology (Karran et al., 2011;Hammond et al., 2020), complex brain metabolic dysfunction (Costantini et al., 2008) and mitochondrial aberrations (Swerdlow, 2018). Furthermore, Yan et al. (1999) found that both Ab  and Ab  bind to 17b HSD10 which inhibits the enzymatic activity of the protein. This interaction was identified as a central neurotoxic event in AD pathology and preventing this interaction was found to be neuroprotective in a mAPP mouse model of AD (Lustbader et al., 2004). Consistent with early findings by He et al. (2005), we provided evidence that 17b HSD10 expression increases in cortical astrocytes exposed to an amyloid-rich environment. Furthermore, overexpressing the catalytically active (and not the mutant) version of the protein reduced astrocytic viability following amyloid treatment. This was accompanied by inhibition of overexpressed wt-17b HD10 activity, which was in line with findings by Yan et al. (1999) showing that Ab (1-42) inhibits the enzymatic activity of the protein and 17b HSD10 overexpression exacerbates amyloid pathology (Yan et al., 1999;Lustbader et al., 2004;Takuma et al., 2005). We also observed exacerbated ROS generation in amyloid-treated astrocytes overexpressing wt-17b HSD10. These effects are similar to observations in neurons from AD mice co-overexpressing mutant APP and 17b HSD10 (Tg mAPP/17b HSD10), which have increased ROS production, reduced ATP levels and compromised mitochondrial integrity (Takuma et al., 2005). Importantly, ROS production was inhibited by AG18051-mediated inhibition of 17b HSD10. Therefore, the novel generation of inhibitors which are currently being developed (Aitken et al., 2017(Aitken et al., , 2019Boutin, et al., 2021;Fišar et al., 2021) hold promise for effectively modifying 17b HSD10-mediated effects in pathology.
In contrast to mitochondrial elongation following severe metabolic stress, mitochondrial fragmentation induced by Ab  was exacerbated by elevated wt-17b HSD10. It is important to note that while ischemia-reoxygenation potentiated the activity of the enzyme (which was associated with increased elongation), Ab  inhibited the activity of overexpressed wt-17b HSD10. Although mitochondrial fragmentation is a known consequence of amyloid-induced toxicity, there are multiple mechanisms that have been identified to drive this change, including elevated ROS, alterations in the levels of mitofusin 1 and 2 (Mfn1/2) levels, Opa1 (optic atrophy 1), and Tomm40, as well as dynamin-related protein 1 (Drp1) and CypD (Manczak et al., 2013;Owens et al., 2015;Park et al., 2015). It is yet to be confirmed which mechanism could be down-stream of the currently reported fragmentation. However, it is clear that the effects of amyloidogenic and metabolic stress on mitochondrial architecture and function in astrocytes are severely affected by alterations in 17b HSD10 enzymatic activity.
Overall, our data revealed that 17b HSD10 regulates mitochondrial function in astrocytes and alters their response to amyloidogenic and metabolic stress. We confirmed that the enzyme is expressed and active in astrocytes from different brain regions. Importantly, 17b HSD10 affects mitochondrial metabolism in astrocytes differently to what has been reported in cancer and neuronal models. Our results further showed that it was particularly the catalytic function of the protein that affects spare respiratory capacity in astrocytes, while 17b HSD10 chemical inhibition counters this effect. We further found that the protein bears multiple implications for astrocytic recovery, oxidative stress, and mitochondrial network morphology in stress conditions. In light of recent findings showing that astrocytic mitochondrial metabolism might be key to their metabolic flexibility and region-dependent vulnerability in neurodegeneration, our findings suggest that 17b HSD10 might be an important regulator. Another important question that remained beyond the scope of the current study is to find out whether 17b HSD10 might have different effects in astrocytes isolated from different brain regions, since the protein is upregulated only in cortical and hippocampal areas of AD brains.
Despite these pending questions, the current study highlights the role of astrocytic 17b HSD10 as a potential mediator of pathologic mechanisms and an exciting potential target for developing meaningful therapeutics.