Visualizing Intracellular Localization of Natural-Product-Based Chemical Probes Using Click-Correlative Light and Electron Microscopy

Abstract

Flavonoids such as sterubin and fisetin—and derivatives thereof—show strong neuroprotective effects in vitro as well as in vivo, combined with negligible toxicity and can therefore be considered novel treatment options for neurodegenerative diseases such as Alzheimer’s disease. However, their subcellular locations responsible for neuroprotection and exact modes of action still remain unclear. Here, we present chemical probes based on both flavonoids sterubin and fisetin that were utilized in fluorescence microscopy and click-correlative light and electron microscopy to detect and visualize the localization of specific intracellular targets. We successfully adapted the workflow of correlative light and electron microscopy to a click-chemistry-based approach in a murine hippocampal cell line (HT22) on ultrathin resin sections making visualization of a small molecule for the first time possible in this setup. Utilizing this newly adapted technique, we could demonstrate that sterubin and fisetin show specific enrichment in the endoplasmic reticulum.

Graphical Abstract

INTRODUCTION

More than 55 million people worldwide suffer from dementia making the disease one of the top ten leading causes of death. Alzheimer’s disease (AD), a subtype of dementia, may contribute to about 60–70% of all cases.¹ This disease is characterized by protein abnormalities like the accumulation of amyloidβ as senile plaques and tau proteins as neurofibrillary tangles. Increased oxidative stress and neuroinflammation driving protein accumulation are characteristics of AD and might be causative for the disease.^2-4 Current pharmacotherapy treatments are highly limited, including cholinesterase inhibitors, NMDA inhibitors, and amyloidβ antibodies.^5,6 Those medications also lead to severe side effects while not being able to cure the disease. The search for novel treatments is hampered by the lack of knowledge about the actual cause of the disease. Furthermore, the one-target approach in drug discovery was not successful up to now. Therefore, an approach to counteract AD with compounds displaying a broad neuroprotective behavior seems more promising in addressing the multifactorial nature of the disease.⁷

Natural products as potent antioxidants have gained interest in the search for novel treatments. Due to their anti-inflammatory and neuroprotective properties by acting on several intracellular pathways, they provide a good basis for developing new treatment strategies. Flavonoids such as sterubin and fisetin have shown excellent neuroprotective properties, both in vitro and even in vivo at a comparatively low dosage.^8-11

Several studies have been conducted to evaluate their exact mode of action, showing profound effects on various intracellular pathways.^8,12-15 However, the exact localization of the compounds and their respective targets enabling their action remain elusive.

To localize compounds in cells by fluorescence microscopy, derived probes containing alkyne-handles for further modification with fluorophores via a bio-orthogonal Cu(I)-catalyzed [3 + 2] azido-alkyne cycloaddition (CuAAC) can be utilized.¹⁶ Since fluorescence microscopy alone cannot clearly tell the subcellular localization of a compound, correlative light and electron microscopy (CLEM) can be pursued. Therein, fluorescence images are correlated with electron microscopy (EM) images, in which cellular ultra-structures like organelles can be seen.¹⁷ This approach is superior to simple fluorescence microscopy in imaging the subcellular locations of small molecules in cells.

In this work, we adapted the super resolution array tomography (srAT) CLEM workflow using embedding in LR white, structured illumination microscopy (SIM) and SEM¹⁷ from tissue sections to a click-chemistry based approach on neuronal HT22 cells to track the enrichment areas of chemical probes based on sterubin and fisetin on ultrathin slices. Other groups have applied bioorthogonal chemistry in CLEM before, e.g., by metabolically labeling biomolecules with bioorthogonal functional groups which can later be reacted to dyes capable of photooxidation. Illuminating this fluorescent dye in the presence of diaminobenzidine led to an osmiophilic product which could later be imaged with EM.¹⁸ Challenges of this method lie in a potentially highly pronounced deposition of oxidized DAB which might impact cellular ultrastructure for EM imaging. Our herein developed method does not affect the cellular ultrastructure. In another interesting previous paper, the van Kasteren group applied a click-CLEM method they developed to image the degradation of bacteria in phagocytes using a two-step activity-based protein profiling method, thereby imaging a covalent enzyme inhibitor.¹⁹ They used confocal microscopy which displays a lower resolution compared to high-resolution SIM. However, they overcame this issue elegantly by using direct stochastic optical reconstruction microscopy (dSTORM) in a later work on cryosectioned bacteria.²⁰ Since dSTORM is quite a time-consuming method, especially for imaging multiple fluorescent channels, we chose SIM in our work since we deemed the resolution good enough and multichannel imaging is more feasible with SIM. Furthermore, we used SEM instead of TEM like the van Kasteren group because while it has somewhat lower resolution, with SEM it is possible to image the slices on the glass slide using both fluorescent and electron microscopy thereby making imaging, especially of long slice series, much easier. A similar approach to the herein presented one was used to label sphingolipids in bacteria,²¹ but our new adaption of the CLEM method now enables for the first time to our knowledge the visualization of small molecules in cells using the srAT approach.

RESULTS AND DISCUSSION

Aiming to visualize intracellular targets of the flavonoids sterubin and fisetin, probes 1 and 2 have been developed by attachment of an alkyne-linker for functionalization by CuAAC (Chart 1). The mode of action of the probes was compared to their parent molecules using different phenotypic screening assays as well as looking at changes in expression of intracellular proteins by Western blotting. Fluorescence microscopy studies showed clear intracellular localization of the compounds, which was further investigated using Click-CLEM. We adapted the workflow used for tissue sections to our application with CuAAC in a HT22 cell pellet. Using this newly adapted approach, we saw localization of the probes mainly in the endoplasmic reticulum (ER), partly in mitochondria, and in nuclear structures. These results were further supported by fluorescence microscopy studies using costaining of ER and mitochondria.

Chart 1.

Design strategy for chemical probes based on sterubin and fisetin

Synthesis.

Aiming to keep the mode of action of the flavonoids the same while introducing an alkyne-linker, the 7-OH-group was chosen as the site of modification (Chart 1). The catechol group as a key pharmacophore cannot be altered without losing neuroprotective potency.²² Furthermore, modification of the 7-OH-group in previous studies has even shown to improve biological effects.^23,24 A pentylene linker was chosen for both compounds since it proved effective in our previous study²⁵ and we assumed that it is not so much the linker length that might influence biological activity but rather the introduction of the linker itself.

Sterubin-alkyne 1 was synthesized similar to Hofmann et al.²⁴ via acetyl-protection of all hydroxyl groups, followed by selective deprotection of the 7-OH-group analogous to Mattarei et al. (Scheme 1).²⁶ Intermediate 5 was further reacted with 5-iodo-1-pentyne 6 in a Williamson ether synthesis. The intermediate was deprotected without further purification under acidic conditions to give the sterubin-alkyne probe 1 (Scheme 1).

Scheme 1. Synthesis of Sterubin-Alkyne 1 and Fisetin-Alkyne 2, Chemical Probes Derived from Sterubin and Fisetin^a.

^aReagents and conditions: i. Ac₂O, I₂, RT, 3 h; ii. imidazole, DCM, −15 °C → RT, 2 h; iii. 1. K₂CO₃, 5-iodo-1-pentyne 6, DMF RT, overnight; 2. 10% HCl in MeOH, 50 °C, 1 h; iv. K₂CO₃, 5-iodo-1-pentyne 6, acetone, 65 °C, overnight; v. 1. KOH, EtOH, 30 min; 2. 3,4-Bis(methoxymethoxy)benzaldehyde 10; RT, overnight; vi. 1. NaOH, H₂O₂, dioxane, EtOH, RT, overnight; 2. HCl, MeOH, 50 °C, 3 h.

For synthesis of fisetin-alkyne 2, 2,4-dihydroxy acetophenone 7 was first reacted with 5-iodo-1-pentyne 6, followed by an aldol-condensation with 3,4-bis(methoxymethoxy)-benzaldehyde 10. Oxidative cyclization with NaOH and H₂O₂²⁷ gave the MOM-protected fisetin-probe which was then deprotected under acidic conditions to give the fisetin-alkyne probe 2 (Scheme 1).

Cell-Based Assays Confirming Similarity to the Parent Compounds.

To determine if the chemical probes have the same activities as the parent compounds, several phenotypic screening assays mimicking distinct age-associated stresses were performed.⁷ Compounds were first investigated for neurotoxicity on the murine hippocampal cell line HT22 showing that compound 1 is nontoxic and compound 2 is only slightly toxic at the highest concentration tested (Figure 1A). The compounds were further tested for their neuroprotective properties against oxytosis, ferroptosis and ATP-depletion. Oxytosis is a form of oxidative stress-induced cell death. It can be induced by treatment with glutamate which causes a reduction in glutathione (GSH) levels in the cell, leading to the accumulation of reactive oxygen species (ROS) and subsequent cell death.²⁸ GSH-depletion is also a hallmark of the aging brain and more severe in the AD brain.²⁹ Therefore, protection against GSH-depletion could be important in AD treatment and prevention. Ferroptosis is closely related to oxytosis and is induced via inhibition of glutathione-peroxidase 4 (GPX4) which is downstream in the cascade induced by glutamate treatment.³⁰ Another hallmark of the aging brain is a reduction in neuronal energy production which can be mimicked by inducing ATP depletion.³¹ Treating cells with iodoacetic acid (IAA), an irreversible inhibitor of glyceraldehyde 3-phosphate dehydrogenase, leads to decreased ATP production via the glycolytic pathway.

Figure 1.

Phenotypic screening assays in HT22 hippocampal nerve cells and neuroinflammation in BV-2 microglial cells. (A) Neurotoxicity of sterubin, sterubin-alkyne, fisetin, and fisetin-alkyne. (B,C,D) Neuroprotective effects of sterubin, sterubin-alkyne, fisetin, and fisetin-alkyne against (B) oxytosis induced by 5 mM glutamate, (C) ferroptosis induced by 300 nM RSL3, and (D) in vitro-ischemia induced by 15 μM iodoacetic acid. Data are normalized to the survival of control cells not treated with the different insults and are presented as means ± SEM of three independent experiments. Statistical analysis was performed using One-way ANOVA followed by Dunnett’s multiple comparison post-test using GraphPad Prism 10 referring to untreated control cells (A) or cells treated with the respective insult only (B,C,D) (red). Levels of significance: *p < 0.05; **p < 0.01; ***p < 0.001, ****p < 0.0001. (E) Modulation of GSH in HT22 cells exposed to oxytosis. Total GSH levels in HT22 cells treated with compounds (10 μM) in the absence or presence of glutamate (5 mM). Graphs show the percentage of total GSH normalized to control DMSO treatment. Data are given as means ± SEM of three independent experiments §§p < 0.01 relative to control DMSO. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 relative to control DMSO. #p < 0.05, ##p < 0.01, ###p < 0.001 relative to glutamate control. (F) Effects of sterubin, sterubin-alkyne, fisetin, and fisetin-alkyne on NO production in LPS-induced neuroinflammation in BV-2 cells. Data are given as means ± SEM of three independent experiments and relative to BV-2 cells treated with LPS only, which was set at 100%.

As shown in Figure 1, the probes show comparable effects against oxytosis (Figure 1B) and ATP depletion (Figure 1D) and enhanced protection of the probes 1 and 2 in the ferroptosis assay (Figure 1C) compared to their parent compounds due to yet unknown reasons. We also looked at the ability of the compounds to modulate GSH levels in healthy and glutamate-stressed cells.³² Results (Figure 1E) show that the probes as well as their parent flavonoids can increase basal GSH levels as well as maintain GSH levels during glutamate-induced toxicity.

Compounds 1 and 2 also were tested for their ability to protect BV2 microglial cells against neuroinflammation induced by bacterial LPS. NO production was quantified to assess the degree of neuroinflammation in the cells. Results (Figure 1F) show a comparable potency of the probes to protect against neuroinflammation with fisetin-alkyne being slightly less active than fisetin itself. No compound was toxic to BV2 cells (Supporting Information (SI), Figure S1).

Summarizing the results of the phenotypic screening assays, probes 1 and 2 show similar biological activity as the flavonoids sterubin and fisetin themselves. This is of utmost importance in testing for intracellular localizations of the parent compounds since a different activity could result in different intracellular targets. The probes show considerable neuroprotection, even at low concentrations, with negligible toxicity, and are further able to reduce neuroinflammation in microglial cells.

Intracellular Signaling Pathways.

Sterubin and fisetin are known to have a strong influence on the levels of intracellular proteins such as the nuclear proteins nuclear factor erythroid 2-related factor 2 (Nrf2) and activating transcription factor 4 (ATF4).⁸ Nrf2 is a transcription factor that is recruited into the nucleus under stress conditions or by Nrf2 inducers. There it binds to the antioxidant-response element (ARE) in the promoter region of specific genes and induces the expression of antioxidant proteins including GSH synthetic enzymes. This makes Nrf2 crucial for maintenance of cellular redox balance.³³ ATF4 is another transcription factor playing an important role in the maintenance of GSH levels in nerve cells.³⁴ Since sterubin and fisetin showed induction of Nrf2 and ATF4 in HT22 cells,^8,35 probes 1 and 2 were tested for similar effects. After treatment with the compounds for 4 h, nuclear extracts of HT22 cells were submitted to Western Blot analysis (Figure 2). The results show concentration-dependent induction of both Nrf2 and ATF4 for the flavonoids and their probe-derivatives. The probes show slightly diminished effects which is to be expected due to the linker introduction, but overall a similar trend could be observed. This upregulation of transcription factors involved in the cellular antioxidant response might at least partially explain the overall neuroprotective effects of the compounds.^8,35

Figure 2.

Western blot analysis of Nrf2- and ATF4-induction by sterubin, sterubin-alkyne, fisetin, and fisetin-alkyne in nuclear fractions of HT22 cells. (A) Representative blots. Quantification of the results given as mean ± SEM (n = 3) as shown in A: (B) Nrf2 induction by sterubin and sterubin-alkyne; (C) Nrf2 induction of fisetin and fisetin-alkyne; (D) ATF4 induction by sterubin and sterubin-alkyne; (E) ATF4 induction by fisetin and fisetin-alkyne. Statistical significance refers to DMSO-treated controls with *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Nrf2 can be induced by oxidative stress and inducing compounds, but it can also be activated by phosphorylation through extracellular regulated kinases (ERKs) leading to increased transcriptional activity of Nrf2.³³ ERK is furthermore involved in various cellular processes, such as transmission of cellular signals and promotion of neuronal plasticity and thus is related to learning and memory processes.³⁶ It also plays a role in neurodegenerative diseases such as AD.³⁶ We investigated time-dependent influence of all compounds on phosphorylation and thereby activation of ERK in HT22 cells. Western blot analysis shows increased phosphorylation after 5 min for sterubin and sterubin-alkyne and after 5 and 10 min for fisetin and fisetin-alkyne (Figure 3). With increased incubation time, phosphorylation decreases again. This is important, since chronic ERK activation is correlated with neurodegeneration.³⁷

Figure 3.

Time-dependent phosphorylation of ERK1 and ERK2 by sterubin, sterubin-alkyne, fisetin, and fisetin-alkyne in HT22 cells. (A) Representative blots. Quantification of the results given as mean ± SEM (n = 3) as shown in (B) ERK1 phosphorylation by sterubin and sterubin-alkyne; (C) ERK1 phosphorylation by fisetin and fisetin-alkyne; (D) ERK2 phosphorylation by sterubin and sterubin-alkyne; (E) ERK2 phosphorylation by fisetin and fisetin-alkyne. Statistical significance refers to DMSO-treated controls with *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001..

Probes 1 and 2 still show an effect on intracellular proteins, albeit less so than the effects mediated by the parent compounds. Nevertheless, we deemed the probes suitable for identification of their intracellular localization since they follow a very similar trend as the parent compounds.

Fluorescence Microscopy with mito-GFP HT22 Cells.

For first visualization experiments, HT22 cells stably expressing green-fluorescent protein in the mitochondria (mito-GFP) were used. Mitochondria play an important role in sustaining energy in the brain and their dysfunction is closely related to AD.³⁶ Furthermore, sterubin and fisetin are both able to restore mitochondrial health under oxytotic/ferroptotic conditions.¹⁵ Therefore, mitochondria might be a potential target of sterubin and fisetin. To evaluate this hypothesis, mito-GFP HT22 cells were treated with 2 μM of each probe and after fixation subjected to CuAAC with Cy3-azide. Fluorescence imaging shows a broad distribution of the compounds in the cells with specific enrichment areas (SI, Figure S2). In some areas, green and red fluorescence signals show an overlap, indicating enrichment in the mitochondria. Other enrichment areas of the probes do not clearly correlate with the mitochondrial signal, indicating other intracellular targets. This is in agreement with previous findings that mitochondria are not required for neuroprotection mediated by sterubin and fisetin.¹⁵

Click-Correlative Light and Electron Microscopy.

To get a more detailed view of the intracellular localization of the probes, CLEM based on the srAT technique is the method of choice. In the principal approach, tissue or, in this case, a cellular pellet is embedded with the resin LR White and cut into ultrathin slices. These slices are first imaged via fluorescence imaging using methyl green costaining to visualize the DNA and therefore nuclear structures. This allows unbiased correlation of the fluorescence images with the EM images.³⁷ Channels were aligned using TetraSpeck beads. Signals in the channel of interest are further analyzed regarding their overlap with specific intracellular substructures.¹⁷ Since there are no antibodies for specific compounds as sterubin and fisetin, CuAAC can be performed to couple a fluorophore to the compound before embedding the sample into LR White. This variant of CLEM has previously been applied in bacteria, showing the general feasibility of pre-embedding staining.²¹ We adapted this workflow to the alkyne-modified probes of sterubin and fisetin in HT22 cells.

Cells were treated with the indicated concentrations of the probes and after fixation and permeabilization subjected to CuAAC with AF488-azide in suspension. AF488 was chosen for its ability to retain fluorescence after being embedded into LR White. Afterward, cells were fixed as a pellet, embedded in LR White, and cut into ultrathin sections (100 nm). After imaging and correlation of fluorescence and SEM images, the intracellular localization of the probes was investigated. The previously described probe Taxifolin-CA was used as a positive control since it binds covalently to its target proteins (SERCA, ANT-1) and is known to be enriched in mitochondria associated membranes (MAMs).²⁵ An overview of all samples is given in SI, Figure S3.

For sterubin-alkyne 1 localization in the ER can be seen (Figure 4; enlarged section can be found in SI, Figure S4). This is consistent with the findings in the first fluorescence experiment with mito-GFP cells indicating enrichment in regions near the nuclear envelope which is by definition part of the ER continuum and close to mitochondria (SI2). The part of the ER interacting with the mitochondria is called MAMs. Current evidence suggests that MAM-dysfunction plays an important role in AD pathology.^38,39 In particular, alterations in the early stages of AD such as mitochondrial dysfunction, impaired Ca²⁺ homeostasis, oxidative stress, and abnormal autophagy are strongly correlated with abnormal MAM function.⁴⁰ Enrichment of probe 1 in the ER close to the mitochondria indicates that possible targets of sterubin are proteins relevant to MAM formation and/or function. Thus, a possible mode of action underlying the neuroprotective effects exerted by sterubin might be counteracting abnormalities in MAM function in the AD brain. There is no evidence suggesting that enrichment in the ER is solely due to the lipophilicity of the compound since then also staining of other membranes such as the cytoplasmic outer membrane would be expected.

Figure 4.

Click-correlative light and electron microscopy images of HT22 cells treated with 2 μM of sterubin-alkyne. Green signal: AF488 attached to the probe; red signal: methyl green. Upper row shows the image of the merged fluorescence channels after channel alignment taken with SIM, the SEM image, and the merged image showing both fluorescence channels correlated with the SEM image after unbiased correlation; AF488+SEM only displays the probe signal and the SEM image. Images on the right are higher magnified sections of images on the left. Scale bar equals 1 μm (left) or 100 nm (right).

Fisetin-alkyne 2 interestingly shows a correlation with the heterochromatin of the nucleus and substructures of the nucleolus, as well as the ER (Figure 5; enlarged section can be found in SI, Figure S5). A crosstalk between the different fluorescence channels could be ruled out by imaging samples without nuclear costaining, so the signal clearly derives from the fluorophore-tagged probe. This might be due to fisetin targeting nuclear transcription factors which would explain the aforementioned effects on Nrf2 and ATF4. Enrichment of fisetin in the nucleolus is also consistent with previous findings.⁴¹

Figure 5.

Click-correlative light and electron microscopy images of HT22 cells treated with 2 μM of fisetin-alkyne. Green signal: AF488 attached to the probe; red signal: methyl green. Upper row shows the image of the merged fluorescence channels after channel alignment taken with SIM, the SEM image, and the merged image showing both fluorescence channels correlated with the SEM image after unbiased correlation; AF488+SEM only displays the probe signal and the SEM image. Images on the right are higher magnified sections of images on the left. Scale bar equals 1 μm (left) or 100 nm (right).

Imaging was performed on consecutive slices of the pellet to ensure that specific fluorescence signals are not just an artifact. Images of the slice series are depicted in SI, Figure S6-S11. To test for specific effects of the compounds in neuronal cells, the same experiment was performed with HEK293T cells as a negative control. We found that there was no clear localization of the compounds in these cells suggesting that their targets are specific to neuronal cells (SI, Figure S12). This further supports the idea that ER enrichment is not based on the lipophilicity of the compounds.

Co-staining Experiments.

To confirm the intracellular localizations of the probes found by click-CLEM, costaining experiments were performed. We first tried to perform antibody-staining on the resin sections of the click-CLEM experiment, but unfortunately both tested antibodies did not work properly on the resin even under several different conditions. Therefore, we stained the ER of HT22 with a livecell ER tracker (ER-Tracker Red) after probe incubation in a single-cell fluorescent experiment. The probe was then subjected to a CuAAC reaction with AF488 and imaged. Methanol fixation was used unlike in the other experiments because the permeabilization normally necessary for CuAAC strongly diminished the ER tracker signal although there seems to be some cross-linking of AF488 to nuclear structures with methanol fixation (SI, Figure S13). Fluorescence imaging clearly shows colocalization of the fluorescence signals derived from the ER tracker and AF488 further supporting the enrichment of the fisetin and sterubin probes in the ER (Figure 6). Quantification of colocalization was performed with ImageJ yielding Pearson scores of 0.684 and 0.764 for the sterubin alkyne and fisetin alkyne, respectively. Pearson score for the negative control is 0.471 due to some background fluorescence in the AF488 channel.

Figure 6.

Microscopic analysis of HT22 cells treated with Sterubin- and Fisetin-Alkyne or DMSO as a negative control. Representative microscopic images of HT22 cells treated with 2 μM Sterubin- (A,B,C) or Fisetin-Alkyne (D,E,F) or a corresponding DMSO concentration (G,H,I) for 30 min. Orange signal: ER tracker BOPIDY TR gilbenclamide; green signal: AF488 attached to the probes; blue signal: DAPI. To visualize the colocalization, a line was plotted in the merged images (magnified images). Charts on the right show quantification of the fluorescence profile along this line for each channel. The X-axis indicates distance, and the Y-axis represents fluorescence intensity in arbitrary units. Scale bar equals 5 μm.

CONCLUSIONS

We report the development of two chemical probes based on the neuroprotective flavonoids sterubin and fisetin by introduction of an alkyne-linker. These probes were evaluated regarding their neuroprotective behavior in various phenotypic screening assays as well as their effects on intracellular proteins, i.e. Nrf2, ATF4, and ERK phosphorylation. All assays suggested a comparable mode of action of the probes to their parent compounds, thereby validating their use as molecular probes to characterize the intracellular localization of the flavonoids. The probes were then used in fluorescence microscopy and a novel, specifically adapted version of the click-CLEM method to investigate sites of subcellular enrichment. The adaptation of the CLEM method to a CuAAC-based cellular setup proved to be successful, and this technique provides a novel tool to study the distribution of small molecules in cells with enhanced correlation compared to fluorescence microscopy alone, since it spares the trial-and-error process of testing various organelle-specific antibodies for colocalization. This method also could further enable a 3D-array of the sections analogous to Markert et al.³⁷

Application of the probes in this assay showed enrichment in the ER and partly in mitochondria for the sterubin-alkyne probe 1 and enrichment in the ER and heterochromatin for the fisetin-alkyne probe 2. These areas of enrichment provide a strong suggestion as to the mode of action for neuroprotection, since dysfunction of mitochondria and mitochondria-associated membranes in the ER is highly correlated with the pathology of AD. Furthermore, these findings are indications that different flavonoids–despite their strong chemical similarities–can have very different localizations in specific subcellular structures and potentially modes of action.

EXPERIMENTAL SECTION

Chemistry.

Synthetic procedures can be found in the Supporting Information.

Cell Culture General Procedures.

General procedures and methods for phenotypic screening assays and Western blots can be found in the Supporting Information.

Click-CLEM Experiment.

For the click-CLEM experiment, HT22 cells were grown in 100 mm dishes to confluency. Cells were treated with 2 μM sterubin-alkyne, 2 μM fisetin-alkyne, 5 μM taxifolincinnamic acid-probe, or 1 μM DMSO for 30 min at 37 °C. Afterward, cells were rinsed twice with PBS, then scraped into 500 μL PBS. After centrifugation (1100 rpm, 4 min), the supernatant was removed, and the cell pellet was resuspended in 4% PFA + 0.05% glutaraldehyde in PBS and fixed for 15 min at 37 °C with shaking. Cells were then permeabilized with 0.5% Triton X-100 in PBS at RT for 15 min with shaking. CuAAC was performed on a cell suspension with the Click-iT Plus Alexa Fluor 488 Picolyl Azide Toolkit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s procedure with 5 μM AF488. Afterward, the cellular pellet was washed 4 x with PBS, then 4% PFA + 0.05% glutaraldehyde in PBS were added. After centrifugation (13500 rpm, 5 min), the cellular pellet was fixed overnight at 4 °C in the dark. The next day, the pellet was washed 3 x with cold PBS and incubated in ammonium chloride (50 mM ammonium chloride in PBS) for 15 min. Following several washes in water, the samples were embedded in LR White resin according to the following Progressive Lowering of Temperature (PLT) protocol. The initial dehydration step of 30% ethanol in water happens on ice (2 × 15 min), then the following steps (50%, 70% (with 0.2% uranyl acetate), 90% (with 0.2% uranyl acetate) and 100% ethanol; 2 times 30 min each) were performed at −20 °C. A mixture of 100% ethanol and LR White (1:1) was incubated overnight and resin infiltration was performed within 24 h, with washes after 1 h, 4 h, 20 and 24 h at 4 °C. The resin was polymerized at 42 °C for 3 days. 100 nm sections were produced using a special diamond knife with a boat large enough to accommodate glass slides (histo Jumbo diamond knife, DiATOME). Slides were submerged in the boat before sectioning.¹⁷ Then the desired number of sections was cut without interruption. Ultrathin sections were stained with Methyl Green (1:10000) for 30 min⁴² rinsed off and mounted in Mowiol and high precision coverslips. Fluorescence images were acquired using an ELYRA S.1 superresolution structured illumination microscope (Zeiss) and channel alignment was performed using TetraSpeck fluorescent beads. The samples were then processed for imaging in the SEM by removing the coverslip, washing off the mounting medium and contrasting with heavy metals as follows: the sections are first incubated in 2.5% uranyl acetate in ethanol for 15 min and in 50% Reynolds’ lead citrate in H₂O for 10 min.⁴³ Carbon coating was performed with the Compact Coating Unit CCU-010 (Safematic, Bad Ragaz, Switzerland) resulting in a ca. 3 nm coating. Images were taken with the JSM-7500F scanning electron microscope (JEOL Germany, GmbH, Freising, Germany). Registration of the fluorescence and electron microscopy images was accomplished with Inkscape 0.91 (Inkscape Community) using the heterochromatin pattern as an intrinsic landmark in both channels. For an unbiased correlation the channel of interest (AF488 signal) was hidden and only the DNA signal (Methyl Green) was used. After the correlation the channel of interest was revealed, and all channels could be merged. More details on the procedure can be found in Markert et al.¹⁷

Fluorescence Microscopy.

Four ×10⁴ HT22 cells were grown on 24 well plates with coverslips. The next day, compounds were added in indicated concentrations and incubated for 30 min at 37 °C. After washing with HBSS, cells were incubated with ER tracker BODIPY TR gilbenclamide at 4 μM in HBSS for 30 min at 37 °C. Then, cells were washed 2 times with warm PBS, and fixed with ice-cold methanol for 10 min at −20 °C. The CuAAC reaction with AF488 was performed according to the manufacturer’s instructions. After washing with PBS, nuclei were costained with DAPI. Three additional PBS washing steps were performed prior to placing the coverslips on slides with mounting media (Prolong glass). Images were acquired using a Zeiss Elyra S.1 SIM - Super Resolution Microscope. DAPI, AF488 and BODIPY TR were excited sequentially using 405, 488, and 561 laser lines, respectively and collected using the corresponding filters. Final analysis of the images was performed using ImageJ software.

Supplementary Material

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acschembio.4c00849.

• NMR spectra and LC-MS chromatograms of target compounds, experimental procedures, materials, analytical methods, additional fluorescence and click-CLEM images (PDF)

ABBREVIATIONS

AD

Alzheimer’s disease

ARE

Antioxidant response element

ATF4

Activating transcription factor 4

CLEM

Correlative light and electron microscopy

CuAAC

Cu(I)-catalyzed [3 + 2] azido-alkyne cycloaddition

EM

Electron microscopy

ER

Endoplasmic reticulum

ERK

Extracellular regulated kinase

GSH

Glutathione

GPX4

Glutathione peroxidase 4

IAA

Iodoacetic acid

LPS

Lipopolysaccharide

MAM

Mitochon-dria-associated membrane

Nrf2

Nuclear factor erythroid 2-related factor 2

ROS

Reactive oxygen species

SEM

Scanning electron microscopy