Reduced secretion of LCN2 (lipocalin 2) from reactive astrocytes through autophagic and proteasomal regulation alleviates inflammatory stress and neuronal damage

ABSTRACT

LCN2/neutrophil gelatinase-associated lipocalin/24p3 (lipocalin 2) is a secretory protein that acts as a mammalian bacteriostatic molecule. Under neuroinflammatory stress conditions, LCN2 is produced and secreted by activated microglia and reactive astrocytes, resulting in neuronal apoptosis. However, it remains largely unknown whether inflammatory stress and neuronal loss can be minimized by modulating LCN2 production and secretion. Here, we first demonstrated that LCN2 was secreted from reactive astrocytes, which were stimulated by treatment with lipopolysaccharide (LPS) as an inflammatory stressor. Notably, we found two effective conditions that led to the reduction of induced LCN2 levels in reactive astrocytes: proteasome inhibition and macroautophagic/autophagic flux activation. Mechanistically, proteasome inhibition suppresses NFKB/NF-κB activation through NFKBIA/IκBα stabilization in primary astrocytes, even under inflammatory stress conditions, resulting in the downregulation of Lcn2 expression. In contrast, autophagic flux activation via MTOR inhibition reduced the intracellular levels of LCN2 through its pre-secretory degradation. In addition, we demonstrated that the N-terminal signal peptide of LCN2 is critical for its secretion and degradation, suggesting that these two pathways may be mechanistically coupled. Finally, we observed that LPS-induced and secreted LCN2 levels were reduced in the astrocyte-cultured medium under the above-mentioned conditions, resulting in increased neuronal viability, even under inflammatory stress.

Abbreviations: ACM, astrocyte-conditioned medium; ALP, autophagy-lysosome pathway; BAF, bafilomycin A₁; BTZ, bortezomib; CHX, cycloheximide; CNS, central nervous system; ER, endoplasmic reticulum; GFAP, glial fibrillary acidic protein; GFP, green fluorescent protein; JAK, Janus kinase; KD, knockdown; LCN2, lipocalin 2; LPS, lipopolysaccharide; MACS, magnetic-activated cell sorting; MAP1LC3/LC3, microtubule-associated protein 1 light chain 3; MTOR, mechanistic target of rapamycin kinase; NFKB/NF-κB, nuclear factor of kappa light polypeptide gene enhancer in B cells 1, p105; NFKBIA/IκBα, nuclear factor of kappa light polypeptide gene enhancer in B cells inhibitor, alpha; OVEX, overexpression; SLC22A17, solute carrier family 22 member 17; SP, signal peptide; SQSTM1, sequestosome 1; STAT3, signal transducer and activator of transcription 3; TNF/TNF-α, tumor necrosis factor; TUBA, tubulin, alpha; TUBB3/β3-TUB, tubulin, beta 3 class III; UB, ubiquitin; UPS, ubiquitin-proteasome system

Introduction

Infection-causing bacteria utilize iron from the host for their growth and survival. These bacteria produce strong iron-utilizing molecules known as siderophores [1]. LCN2/neutrophil gelatinase-associated lipocalin/siderocalin/24p3 (lipocalin 2) is a small (180 amino acids long) secretory protein that binds to the iron-siderophore complex, inhibiting iron utilization and growth of infectants [2,3]. Similar to other secretory proteins, previous studies have reported that LCN2 can be modified by removing its N-terminal signal peptide (SP), which is essential for targeting it into secretory pathways during de novo translation, followed by N-glycosylation [4,5]. Moreover, Lcn2 expression is transcriptionally increased upon activation of NFKB/NF-κB (nuclear factor of kappa light polypeptide gene enhancer in B cells 1, p105) signaling in the presence of inflammatory stressors such as lipopolysaccharide (LPS) or pro-inflammatory cytokines such as TNF/TNF-α (tumor necrosis factor) [6,7]. The expression of Lcn2 is very low in most organs under normal conditions; however, under disease conditions, Lcn2 levels are dramatically increased in the uterus, kidney, and even in the brain [8–11]. In the central nervous system (CNS), microglia and astrocytes are the two major cell types that produce and secrete LCN2 during neuroinflammatory stress conditions [12,13].

CNS disease progression is often accompanied by neuroinflammation and disruption of iron homeostasis, resulting in oxidative and metabolic stress and eventually irreversible neuronal loss [14]. Microglia, which constitute approximately 5–10% of cells in the brain, are parenchymal macrophages that play multiple roles in CNS physiology under normal and inflammatory stress conditions [15,16]. Under normal conditions, microglia regulate synaptic homeostasis [17,18]. In addition, microglia produce complements, which are pivotal components of phagocytosis, and engulf some myelin or adult neural progenitor cells to tune the brain parenchyma [19,20]. The classification and nomenclature of microglia has been broadly applied based on the dichotomous concept that microglia are activated and polarized to either M1 (neurotoxic) or M2 (neuroprotective) depending on the pathological conditions [21,22]. Recently, however, a number of experienced researchers have departed from this traditional view and have suggested that activated microglia cannot simply be categorized into M1 or M2, and are more heterogeneous at the single cell level [23]. The heterogeneous characteristics of disease-associated or even homeostatic (under basal conditions) microglia are observed more often in vivo than in vitro. Interestingly, these heterogeneous properties of microglia reflect the results of their communications with different cell types in their surroundings as an adaptive response to pathological conditions or fine-tuning of brain damages [24]. Communications between microglia and astrocytes in response to LPS are essential for reactive astrogliosis [25].

Astrocytes, the most abundant glial cells in the CNS, perform several functions related to brain homeostasis [26,27]. Similar to microglia, astrocytes undergo molecular and cellular remodeling in response to altered CNS microenvironment during the progression of neurodegenerative diseases, such as Alzheimer or Parkinson disease [28]. Conventionally, reactive astrocytes were categorized into A1 or A2 subtypes based on their altered transcriptomes, which have been regarded as neurotoxic and neuroprotective astrocytes, respectively [29–31]. Recently, heterogeneous characteristics of reactive astrocytes have been continuously reported in in vivo disease models, and it has emerged that the traditional A1 or A2 classification may not match the neurotoxic or neuroprotective function of astrocytes [32]. Hence, to properly understand astrocyte reactivation, it should be defined by the integration of various factors beyond the traditional concept. One of the critical factors of astrocyte reactivation is the cell signaling with transcription factors, including NFKB and JAK-STAT3 signaling pathways, both of which have been well-understood at the cellular level. The other is various secretory proteins that are produced and secreted by reactive astrocytes, of which LCN2 is extensively produced by neurotoxic reactive astrocytes, inducing apoptosis in damaged neurons [33]. A recent study revealed that in a mouse model of Alzheimer disease, severely reactive astrocytes generate hydrogen peroxide, which causes neuronal loss and eventually cognitive defects [34]. Therefore, several pharmacological strategies have been developed to target activated microglia and reactive astrocytes based on their integrating properties under specific neurological disorder conditions to alleviate the pathogenesis [35,36].

Other pharmacological approaches employed for CNS diseases target two major pathways of protein catabolism: the ubiquitin-proteasome system (UPS) and autophagy-lysosome pathway (ALP) [37–40]. The proteasome is a multiprotein complex composed of 19S regulatory particles, which recognize ubiquitylated target proteins, and 20S core particles, which catalyze the degradation of target proteins [41]. Reduced proteasome activity is closely related to the formation of insoluble protein deposits and the onset and progression of several neurodegenerative diseases [42,43]. Autophagy is initiated by the formation of double-membrane autophagosomes, which deliver intracellular cargo, including proteins, lipids, DNA, RNA, and even mitochondria, to lysosomes, resulting in their subsequent degradation by acidic enzymes [44]. Several lines of evidence suggest that the impairment of the autophagic pathway is associated with the accumulation of protein aggregates and promotes extensive neuronal loss in the CNS [45,46]. However, attempts to reduce insoluble protein aggregates in treating neurodegenerative diseases, including Alzheimer disease, have shown limited success [47,48]. Therefore, further studies regarding cellular protein catabolic pathways and their regulatory effects in CNS models under normal and disease conditions are warranted.

In this study, using primary astrocytes, we investigated whether the neurotoxic potential of LCN2 secreted from reactive astrocytes can be reduced by modulating the intracellular protein degradation pathways. We found that the neurotoxicity of LPS-induced reactive astrocytes was significantly reduced under two different conditions: direct proteasome inhibition and indirect autophagy activation via MTOR (mechanistic target of rapamycin kinase) inhibition. Treatment with the proteasome inhibitor bortezomib reduced Lcn2 expression and several reactive astrocyte markers by suppressing NFKB signaling. We also demonstrated that treatment with the MTOR complex inhibitor torin 1 induced autophagy activation not only in non-reactive astrocytes but also in LPS-induced reactive astrocytes. Autophagy activation in the LPS-induced reactive astrocytes resulted in rapid degradation of de novo synthesized LCN2 before its secretion. Our data suggest mechanistic insights on how the secretory LCN2 is regulated at the RNA and protein levels, and how the neurotoxicity of LCN2 can be alleviated. In this study, we also investigated how the regulation of protein degradation pathway affects the expression of traditional pan, A1, and A2 markers. We did not intend to support the traditional concept of binary polarization of reactive astrocytes, but to validate whether altered expression of these markers is accompanied when LPS-induced neurotoxicity is alleviated.

Results

LPS-induced reactive astrocytes reduce neuronal viability by producing LCN2

Our previous study that used a mixed culture of neurons and astrocytes showed that increase in the number of astrocytes and astrocytic expression of Lcn2 is associated with neuronal apoptosis [49]. To better understand the relationship between Lcn2 expression in astrocytes and neuronal apoptosis, we first cultured primary astrocytes isolated from the mouse brain and demonstrated the characteristics of these cells using immunofluorescence staining (Figure 1A). LPS treatment significantly enhanced LCN2 production in a dose-dependent manner (Figure 1B). In addition, LCN2 levels significantly increased after one day of LPS treatment, where the levels plateaued for the next two days (Figure 1C). Next, we investigated the effect of LPS treatment on astrocyte reactivation whether it could increase the expression of pan, A1, or A2 markers. Our results showed that LPS treatment upregulated pan and A1 marker expression; however, the expression of A2 markers was downregulated following LPS treatment (Figure 1D). In agreement with previous studies, our data suggest that LPS induces acute immune responses in astrocytes and transforms quiescent astrocytes to reactive astrocytes with high expression of A1 markers, which are known to robustly secrete LCN2 [25,50].

Figure 1.

Reactive astrocytes secrete LCN2 and induce neuronal loss under LPS-treated conditions. (A) Primary astrocytes isolated from mouse pup brains were stained with antibodies against a neuronal marker, TUBB3/β3-TUB (tubulin, beta 3 class III), and an astrocyte marker, glial fibrillary acidic protein (GFAP). DNA was visualized by staining with DAPI. The percentage of TUBB3-positive neurons and GFAP-positive astrocytes was calculated based on the DAPI-positive total cell number (n = 3, >50 cells). (B) Immunoblot detection of LCN2 in astrocytes treated with low to high concentrations of LPS (0–1000 ng/mL). (C) LCN2 levels in astrocytes exposed to 100 ng/mL of LPS for 1 to 3 d were detected via immunoblotting. For immunoblot analysis, TUBA/α-TUB (tubulin, alpha) was used as a loading control. (D) Primary astrocytes were treated with 100 ng/mL of LPS or PBS for 1 d. The mRNA levels of pan markers (Serpina3n, Cxcl10, Osmr, and Cp), A1 markers (H2-T23, Serping1, H2-D1, Ugt1a, Fbln5, Srgn, Ggta1, and Gbp2), and A2 markers (Clcf1, S100a10, and Emp1) in the negative control group (−LPS) and LPS-treated group (+LPS) were measured via qRT-PCR, normalized against Gapdh levels, and expressed as a fold change relative to the levels in the negative control group (n = 3; two-tailed unpaired t-test). (E) Immunoblot detection of LCN2 secreted from astrocytes cultured under various conditions, including transduction with lentivirus (LV) and treatment with 100 ng/mL of LPS. Lcn2 knockdown (KD) and LCN2 overexpression (OVEX) were performed via LV infection. LV MOCK was used as a negative control. For the astrocyte-conditioned medium (ACM), Coomassie blue staining of polyacrylamide gels was used as a loading control. Depending on the conditions described above, the ACM was numbered between 1 and 8. The protein concentration in ACM was about 3 mg/mL based on the BCA protein assay. (F) Primary neurons were isolated from mouse embryos and treated with the numbered ACM (total 3 mg of protein) for 3 d. The viability of cultured neurons was measured using the MTT assay (n = 3; Kruskal-Wallis one-way ANOVA, followed by Tukey’s multiple comparison test). Neuronal viability was expressed as a percentage relative to the control (ACM 1). Representative images of cells or immunoblots are shown. qRT-PCR and MTT assay data are expressed as the means ± SEM from the indicated number of samples. *p < 0.05; **p < 0.01; ***p < 0.001 vs. the control or between two groups as indicated by the horizontal bars. n.s, not significant. Scale bar: 100 μm.

Previous studies have shown that LCN2 secreted by reactive astrocytes can induce apoptosis in degenerating and normal neurons [33]. In contrast, LCN2 secreted from damaged neurons can propagate specific signals to surrounding cells, resulting in the repair of damaged nerve tissues [51]. To evaluate how the increased LCN2 in our astrocyte reactivation model affects the neuronal viability, we prepared primary neurons and astrocyte-conditioned medium (ACM), in which astrocytes were transduced with negative control (MOCK), Lcn2 knockdown (KD), or LCN2 overexpression (OVEX) lentiviruses, followed by treatment with LPS or not (Figure S1A). First, we determined the levels of secreted LCN2 in the ACM (Figure 1E). When we monitored the viability of neurons treated with various ACM, reduced viability was observed in neurons treated with ACM from LPS-treated reactive astrocytes and LPS-treated LCN2-overexpressing astrocytes (#5 and #8 in Figure 1F, respectively). However, the ACM from control astrocytes, astrocytes transduced with MOCK lentiviruses, and LPS-treated Lcn2 KD reactive astrocytes had no or marginal effect on neuronal viability (#1, #2, and #7 in Figure 1F, respectively). These results indicate that robust secretion of LCN2 by reactive astrocytes can induce neuronal death by itself. The secreted LCN2 may also increase neurotoxicity through a complex reaction with other secreted molecules during LPS stimulation. Taken together, our data recapitulate previous results, raising an interesting question on how to reduce LCN2 secretion in LPS-induced reactive astrocytes that may be beneficial for neuroprotection.

Proteasome inhibition reduces Lcn2 expression by suppressing NFKB signaling

Previous studies suggest that Lcn2 expression is upregulated by activating the NFKB signaling pathway [6,7]. To activate NFKB signaling, inflammatory stressors (e.g., LPS and TNF) bind to their receptors (e.g., TLR [toll-like receptor] and TNFRSF [tumor necrosis factor receptor superfamily]) on the plasma membrane, initiating a signaling cascade that facilitates the UPS-mediated degradation of NFKBIA/IκBα (nuclear factor of kappa light polypeptide gene enhancer in B cells inhibitor, alpha) [52]. The released NFKB RELA/p65 subsequently translocates to the nucleus and binds to the consensus sequences of its target genes as a transcription factor. We hypothesized that proteasome inhibition in LPS-induced reactive astrocytes could downregulate the expression of Lcn2 by suppressing NFKB signaling. When reactive astrocytes were treated with a proteasome inhibitor (MG132), proteasome activity was significantly reduced, and the levels of ubiquitylated proteins were increased compared to the controls, suggesting that MG132 can inhibit the proteasome even at a low concentration (Figure 2A, B). Consistent with our hypothesis, LCN2 levels significantly decreased in LPS-treated reactive astrocytes following MG132 treatment (Figure 2B). Bortezomib (Velcade®) has been widely used to treat multiple myeloma patients, sharing its mode of action as a proteasome inhibitor with MG132. As bortezomib has more experimental references directly related to NFKB signaling, its usage was more suitable to validate our hypothesis. Similarly, proteasome activity and LCN2 levels in astrocytes were also significantly decreased by bortezomib treatment (Figure 2C, D).

Figure 2.

Proteasome inhibition alters the expression of NFKB target genes and astrocyte reactivation markers. (A) LPS-treated (or untreated) and MG132-treated (0, 1, and 5 μM) astrocytes were subjected to the proteasome activity assay (n = 3). (B) Immunoblot detection of ubiquitin conjugates (UB_n), LCN2, and GFAP in astrocytes treated with PBS (−LPS) or 100 ng/mL of LPS (+LPS), and low to high concentrations of MG132 (0, 1, and 5 μM) for 1 d. (C) The proteasome activity assay was performed as described in (A) following treatment with bortezomib (BTZ) instead of MG132 (n = 3). (D) Immunoblot detection was performed as described in (B) following treatment with BTZ instead of MG132. TUBA was used as a loading control in the immunoblot analysis. (E, F) Astrocytes were treated with 100 ng/mL of LPS (or untreated) and low to high concentrations of MG132 (0, 1, and 5 μM) for 1 d. Subsequently, the mRNA levels of Lcn2 and Tnf were measured via qRT-PCR, normalized against Gapdh levels, and expressed as a fold change relative to the control (n = 3; one-way ANOVA, followed by Tukey’s multiple comparison test). (G to J) Astrocytes were treated with 100 ng/mL of LPS (or untreated) and low to high concentrations of BTZ (0, 1, and 5 μM) for 1 d. The mRNA levels of Lcn2, Tnf, Il1a, and C3 were measured via qRT-PCR, normalized against Gapdh levels, and expressed as a fold change relative to the control (n = 3; one-way ANOVA, followed by Tukey’s multiple comparison test). (K to N) Primary astrocytes were treated with 100 ng/mL of LPS (or untreated) and 1 μM of BTZ (or untreated) for 1 d. Thereafter, mRNA levels of pan, A1, and A2 markers in the control, BTZ-, LPS-, and both BTZ and LPS-treated groups were measured via qRT-PCR. The mRNA levels were normalized against Gapdh levels and expressed as a fold change relative to the levels in the control group (n = 3; one-way ANOVA, followed by Tukey’s multiple comparison test). Representative images of immunoblots are shown. qRT-PCR data are expressed as the means ± SEM from the indicated number of samples. *p < 0.05, **p < 0.01; ***p < 0.001 vs. the control or between two groups as indicated by the horizontal bars. n.s, not significant.

Next, we evaluated whether the expression of Lcn2 and other NFKB target genes (such as Tnf, Il1a, and C3) was altered by proteasome inhibition. The levels of Lcn2 and Tnf were increased in astrocytes following LPS treatment; however, additional treatment with MG132 or bortezomib ameliorated this increase (Figure 2E-H). Moreover, the expression of other NFKB target genes, including Il1a and C3, was upregulated in the presence of LPS alone but not in the presence of both LPS and bortezomib (Figure 2I, J). Interestingly, in most cases, the elevated expression of pan and A1 markers following LPS treatment was reversed by co-treatment of LPS and bortezomib (Figure 2K-M). We also noticed that some of the A2 markers of reactive astrocytes, whose expression levels were reduced by LPS treatment, were upregulated under these conditions (Figure 2N).

To further investigate whether proteasome inhibition indeed prevented NFKB signaling activation at the molecular level, alterations in NFKBIA/IκBα and NFKB RELA/p65 levels were also evaluated in astrocytes. Upon LPS treatment, NFKBIA levels decreased in astrocytes within 1 h; however, the total NFKB RELA levels remained unchanged (Figure 3A, Figure S2A). As expected, co-treatment of LPS and bortezomib prevented this decrease in NFKBIA levels. Moreover, increase in the levels of phosphorylated NFKBIA, which is known to be a prerequisite for NFKBIA degradation, was observed in astrocytes only when LPS and bortezomib were co-treated (Figure S2A). Based on these results, we speculated that NFKBIA was phosphorylated and subsequently degraded upon exposure to LPS. In contrast, blocking NFKBIA degradation through proteasome inhibition increased the levels of phosphorylated NFKBIA. Next, using nucleus/cytoplasm fractionation, we observed that the level of NFKB RELA in the nucleus increased following LPS treatment, which was reversed by co-treatment of LPS and bortezomib (Figure 3B). In addition, LPS-induced nuclear translocation of NFKB RELA in astrocytes was dramatically reduced by bortezomib treatment, consistent with immunoblot results (Figure 3C). We also validated JAK-STAT3 signaling and C3 levels, which are also known as markers of reactive astrocytes, under LPS treatment conditions (Figure S2B, C). We observed that the levels of phosphorylated STAT3 were increased by LPS treatment regardless of the presence of bortezomib, and C3 levels were also increased slightly by LPS treatment.

Figure 3.

Proteasome inhibition suppresses the activation of NFKB signaling. (A) Immunoblot detection of UB_n, NFKB RELA/p65, NFKBIA/IκBα, LCN2, and GFAP in astrocytes treated with PBS (−LPS) or 100 ng/mL of LPS (+LPS) and 1 μM of BTZ (or untreated) for 1 or 24 h. (B) Cytosolic and nuclear fractions isolated from astrocytes treated with PBS (−LPS) or 100 ng/mL of LPS (+LPS) and 1 μM of BTZ (or untreated) for 1 h were subjected to immunoblot detection of NFKBIA and NFKB RELA. TUBA and LMNB (lamin B1) were used as loading controls for the cytosolic and nuclear proteins, respectively. (C) Primary astrocytes treated with PBS (−LPS) or 100 ng/mL of LPS (+LPS) and DMSO (−BTZ) or 1 μM of BTZ (+BTZ) for 1 h were immunostained for GFAP and NFKB RELA. DNA was visualized with DAPI. (D, E) During the course of treatment with 1 μM BTZ, Lcn2 and Tnf expression levels were determined via qRT-PCR. The mRNA levels were normalized against Gapdh levels and expressed as a fold change relative to the levels in the negative control group (n = 3). (F) UB_n, LCN2, and GFAP levels in LPS-pretreated reactive astrocytes were detected via immunoblot analysis during the course of treatment with 1 μM BTZ. (G) Immunoblot detection of UB_n, NFKB RELA, NFKBIA, LCN2, and GFAP in astrocytes pretreated with PBS (−LPS) or 100 ng/mL of LPS (+LPS) for 24 h, and then treated with DMSO (−BTZ) or 1 μM of BTZ (+BTZ) for 1 and 3 h. For immunoblot analysis, TUBA was used as a loading control. Representative images of cells or immunoblots are shown. qRT-PCR data are expressed as the means ± SEM from the indicated number of samples. Scale bar: 20 μm.

Next, we monitored whether bortezomib post-treatment could attenuate LCN2 production in reactive astrocytes that had already responded to LPS. To address this, we evaluated the effects of bortezomib treatment on NFKB signaling and Lcn2 expression in LPS-pretreated astrocytes. Post-treatment with bortezomib reduced the upregulated expression of Lcn2 and Tnf in a time-dependent manner (Figure 3D, E). Additionally, the LPS-induced increase in LCN2 levels upon LPS pretreatment was significantly reduced after bortezomib treatment in a time-dependent manner (Figure 3F). The suppressive effect of bortezomib on NFKB signaling through the stabilization of NFKBIA was effective, even in LPS-pretreated cells (Figure 3G). Taken together, our data suggest that proteasome inhibition not only reduces the expression of pro-inflammatory genes, including Lcn2, at the RNA level by suppressing the NFKB signaling pathway but also induces altered expression of traditional pan, A1, and A2 markers even under LPS-treated conditions. Although the alteration of these traditional markers could not be directly linked to the characteristics of reactive astrocytes (neurotoxic or neuroprotective), bortezomib treatment inhibits the LPS-induced NFKB signaling pathway in astrocytes, which is likely to form a novel unique type of reactive astrocytes.

LCN2 is degraded by the autophagy-lysosome pathway inside cells

After our finding that neurotoxic LCN2 production could be reduced at the RNA level by suppressing NFKB signaling, we further investigated the regulatory potential of LCN2 in different pathways, including post-translational and secretory regulation. Since our results indicated that LCN2 is not a proteasome target protein, we speculated that the synthesized LCN2 is degraded through the autophagy-lysosome pathway. Treatment with bafilomycin A₁, an inhibitor of macroautophagic flux, induced the accumulation of LCN2 in LPS-treated cells (Figure 4A). Since LCN2 is a secretory protein, it is possible that bafilomycin A₁ may have prevented the secretion of this protein, leading to its accumulation in cells. We observed that the levels of secreted LCN2 in the medium was not altered by bafilomycin A₁ treatment, suggesting that the secretion of LCN2 was not affected by bafilomycin A₁ (Figure 4B). However, bortezomib treatment reduced the level of secreted LCN2, similar to the reduced intracellular LCN2 level (see Figure 2D). Moreover, the expression of Lcn2 mRNA did not increase following bafilomycin A₁ treatment, in contrast to LCN2 protein levels (Figure 4C). Interestingly, bafilomycin A₁ treatment slightly decreased the expression levels of Lcn2 and Tnf even under inflammatory stress conditions (Figure 4C, D). Taken together, these data suggest that de novo synthesized LCN2 can be degraded within cells via the autophagy-lysosome pathway prior to its secretion.

Figure 4.

The autophagy-lysosome pathway degrades intracellular LCN2. (A) Astrocytes treated with PBS (−LPS) or 100 ng/mL of LPS (+LPS) and DMSO (0 nM) or 50 to 100 nM of bafilomycin A₁ (BAF) were subjected to immunoblot detection of UB_n, LCN2, and GFAP. (B) Immunoblot detection of secreted LCN2 from astrocytes treated with 100 ng/mL LPS (or untreated), 1 μM BTZ (or untreated), and 100 nM BAF (or untreated). For ACM, equal loading was evaluated via Coomassie blue staining of polyacrylamide gels. (C, D) Astrocytes were treated with 100 ng/mL of LPS and the indicated concentrations of BAF. Lcn2 and Tnf expression levels were determined via qRT-PCR, normalized against Gapdh levels, and expressed as a fold change relative to the control (n = 3; one-way ANOVA, followed by Tukey’s multiple comparison test). (E, F) Astrocytes were pretreated with 100 ng/mL of LPS for 1 d to induce the de novo synthesis of LCN2 before cycloheximide (CHX) chasing. In (E), LPS-induced intracellular LCN2 in astrocytes was chased with a medium containing 10 μg/mL of CHX, 100 ng/mL of LPS, and DMSO (−BAF) or 100 nM of BAF for up to 80 min. In (F), after adding 1 μM of BTZ, LPS-induced intracellular LCN2 was chased for up to 80 min according to the conditions described above. Relative LCN2 levels in cells over time were normalized to TUBA and expressed as a percentage relative to the zero-chasing time (0 min) (n = 3). (G) Before CHX chase, LCN2 was induced in astrocytes after treatment with 100 ng/mL of LPS for 1 d. During CHX chase for up to 60 min, LCN2 and GFAP levels were monitored at the cellular level with anti-LCN2 and anti-GFAP antibodies, and DNA was visualized with DAPI. Representative images of cells or immunoblots are shown. qRT-PCR and CHX chase data are expressed as the means ± SEM from the indicated number of samples. **p < 0.01; ***p < 0.001 vs. the control or between two groups as indicated by the horizontal bars. n.s, not significant. Scale bar: 20 μm.

Next, to investigate the rapid degradation of LCN2 in cells, the half-life of the LCN2 protein in astrocytes was determined using the conventional cycloheximide chase assay. After treatment with cycloheximide, which inhibits a wide range of ribosomal translation processes, half of the LCN2 in LPS-pretreated cells was degraded in approximately 30 min (Figure 4E). In contrast, bafilomycin A₁ treatment attenuated LCN2 degradation. Since proteasome inhibition downregulated Lcn2 expression at the mRNA level, the half-life of LCN2 was not altered under bortezomib-treated conditions (Figure 4F). Consistent with these results, immunofluorescence signals for LCN2 were reduced in astrocytes following cycloheximide treatment in a time-dependent manner (Figure 4G). These results indicate that intracellular LCN2 secreted during the immune response is likely degraded via the autophagy-lysosome pathway, which is a plausible mode of LCN2 protein regulation.

Activation of the autophagic flux reduces LCN2 levels in reactive astrocytes

The autophagy-lysosome pathway is categorized based on the characteristics of its cargo or its selectivity [53]. Selective macroautophagy, which is highly conserved in eukaryotes and has been well-studied at the genetic and molecular levels, is often described in terms of flux. This flux can be divided into the following steps: formation of autophagosomes for cargo sequestration, the fusion of mature autophagosomes and lysosomes (autolysosome formation), and cargo degradation by acidic hydrolases in the lysosomal lumen [44,53]. Autophagic flux is delicately regulated at the molecular level, and the MTOR signaling pathway plays an important role in autophagosome and lysosome formation [44]. Next, we investigated whether LCN2 is regulated post-translationally via the MTOR signaling-ALP axis in reactive astrocytes. Torin 1, a direct inhibitor of the MTOR complex and activator of autophagic flux, was used in this study. We found that intracellular LCN2 levels were reduced by torin 1 treatment (Figure 5A). In addition, autophagic flux activation was observed by torin 1 treatment through decreased levels of SQSTM1/p62 (sequestosome 1) and increased ratio of LC3-II:I (Figure 5B). Moreover, phosphorylated RPS6KB/p70S6K, a conventional MTOR target protein, was increased following LPS treatment, but not in the presence of torin 1 (Figure S3A). As the intracellular LCN2 levels were reduced, we observed that LCN2 secretion was also reduced by torin 1 treatment (Figure 5C). Interestingly, we also found that LPS-induced LCN2 levels were reduced when MTOR was inhibited by nutrient starvation (Figure S3B). However, we could not observe further activation of MTOR and increase in LCN2 levels by NV5138, a brain-selective MTOR activator (Figure S3C) [54]. Taken together, our results suggest that torin 1 can activate autophagic flux in both non-reactive and LPS-induced reactive astrocytes via MTOR inhibition, resulting in more rapid degradation of LCN2.

Figure 5.

Autophagic flux activation reduces the levels of LPS-induced LCN2. (A) Immunoblot analysis was performed to detect intracellular LCN2 and GFAP levels in astrocytes treated with PBS (−LPS) or 100 ng/mL of LPS (+LPS) and DMSO (0 μM) or two different concentrations of torin 1 (0.5 and 2 μM). (B) In the absence or presence of LPS (100 ng/mL) and torin 1 (0.5 μM), alterations of autophagy-related proteins, including MAP1LC3/LC3 (microtubule-associated protein 1 light chain 3) and SQSTM1/p62 (sequestosome 1), were measured by immunoblotting. The LC3-II:I ratio was calculated based on the intensity of each band (n = 3; Kruskal-Wallis one-way ANOVA, followed by Tukey’s multiple comparison test). (C) Immunoblot detection of secreted LCN2 from astrocytes treated with LPS and torin 1 as described in (A). (D to G) The mRNA levels of pan, A1, and A2 markers in the control, torin 1-, LPS-, and both torin 1 and LPS-treated groups were measured via qRT-PCR, normalized against Gapdh levels, and expressed as a fold change relative to the levels in the control group (n = 3; one-way ANOVA, followed by Tukey’s multiple comparison test). (H) After treatment with LPS (100 ng/mL), torin 1 (2 μM), and bafilomycin A₁ (BAF, 100 nM), primary astrocytes were stained with anti-GFAP and anti-LCN2 antibodies, and DNA was visualized with DAPI (top). Under the same conditions, astrocytes were stained with anti-LCN2 antibody, and the DNA and lysosomal contents were visualized with DAPI and LysoTracker, respectively (bottom). Representative images of cells or immunoblots are shown. qRT-PCR data are expressed as the means ± SEM from the indicated number of samples. *p < 0.05, **p < 0.01; ***p < 0.001 vs. the control or between two groups as indicated by the horizontal bars. n.s, not significant. Scale bar: 20 μm.

Next, we investigated whether there is an interaction between MTOR signaling (and the related autophagic flux) and LPS-induced astrocyte reactivation. We first found that LPS-induced Lcn2 expression was decreased slightly by torin 1 treatment, while that of Tnf was significantly decreased, regardless of LPS treatment (Figure S3D, E). Moreover, increased levels of pan and A1 markers upon LPS exposure were reversed by torin 1 treatment (Figure 5D-F). We also noticed that some A2 markers, whose expression levels were reduced by LPS treatment, were upregulated by torin 1 treatment, although there were less alterations in A2 markers (Figure 5G). These results suggest that MTOR signaling may be involved in the LPS responses of astrocytes via regulation of the NFKB and other signaling pathways.

To further confirm our observations, LCN2, SQSTM1, and lysosomal contents were visualized in cells under LPS exposure conditions with or without torin 1 or bafilomycin A₁ treatment. Immunofluorescence staining showed that LCN2 expression in LPS-treated astrocytes decreased following torin 1 treatment; however, it was increased by bafilomycin A₁ treatment (Figure 5H, top). Interestingly, lysosomal contents in LPS-induced reactive astrocytes were increased by torin 1 treatment (Figure 5H, bottom). This phenomenon was reversed when bafilomycin A₁ was used instead of torin 1. As bafilomycin A₁ interferes with the maintenance of lysosomal acidity, it affects LysoTracker staining intensity. Our results also suggest an inverse correlation between intracellular LCN2 levels and intact lysosomal content. Furthermore, immunofluorescence staining revealed that the levels of LCN2 and SQSTM1 were significantly higher in cells treated with bafilomycin A₁, and that colocalization of LCN2 and SQSTM1 was observed within cells (Figure S4A). In the presence of torin 1, the levels of LCN2 and SQSTM1 decreased, and colocalization of LCN2 and SQSTM1 was likewise decreased. Thus, it can be concluded that intracellular LCN2 levels are positively correlated with the number of autophagosomes.

In this study, endogenous LCN2 in astrocytes, whose levels were increased by LPS exposure, was found to be the cargo protein of the autophagy-lysosomal pathway. However, according to previous reports, LCN2 itself may be able to reduce autophagic flux [55,56]. To evaluate this possibility in the primary astrocytes, we performed Lcn2 knockdown or LCN2 overexpression in astrocytes before LPS treatment for the detection of autophagy-related proteins. Although further validation is required, our data suggest that neither LPS treatment nor the level of LCN2 per se has a significant effect on autophagic flux in astrocytes (Figure S4B). Thus, our data suggest that intracellular LCN2 can be degraded by the autophagy-lysosome pathway and that the activation of autophagic flux accelerates its degradation prior to secretion.

Purified astrocytes secrete less LCN2 through proteasome inhibition and autophagy activation

In this study, we used primary astrocytes isolated from mouse cortex without immunopanning or sorting. This conventional astrocyte culture (often called MD astrocyte culture) does not contain pure population of astrocytes and may include some microglia [57]. Based on our experience, the percentage of microglia is less than 10%. Since our results using MD astrocyte culture may have in vitro limits due to multi-population issues, we performed pure astrocyte culture following magnetic-activated cell sorting (MACS), and evaluated LCN2 levels and astrocyte reactivation [58]. Interestingly, although Lcn2 expression increased slightly by LPS treatment, LCN2 was not observed in ACM, collected from astrocytes purified by MACS (MS astrocytes) even under LPS treatment conditions (Figure 6A, C). As the purity of MS astrocytes is higher than that of MD astrocytes, we speculated that the LPS-stimulated microglia may be a prerequisite for Lcn2 upregulation in the MS astrocytes, as observed in other study [25]. If microglia present in our MD astrocyte cultures mediate the LPS-induced astrocyte reactivation, ACM collected from LPS-treated MD astrocytes should contain enough secreted molecules to induce MS astrocyte reactivation. In fact, when MS astrocytes were supplied with ACM from LPS-treated MD astrocytes, Lcn2 expression levels were dramatically increased compared to those supplied with the control MD ACM (Figure 6A, B). Furthermore, the increased Lcn2 expression levels under this condition was significantly reduced by bortezomib treatment in MS astrocytes but not by torin 1 treatment (Figure 6B). We also found that altered expression of several reactive markers in MS astrocytes by LPS-stimulated MD ACM were significantly reversed by bortezomib treatment, although the effect was less significant by torin 1 treatment (Figure 6D). In addition, LCN2 secretion increased significantly when MS astrocytes were treated with the LPS-stimulated MD ACM, but not in the presence of bortezomib or torin 1 (Figure 6E). Therefore, regulation of LCN2 levels by bortezomib or torin 1 is a bona fide event that occurs in pure astrocytes. Taken together, these results suggest that the characteristics of LPS-induced MD astrocyte reactivation and LCN2 secretion can be reproduced in purified MS astrocytes. Thus, MD astrocytes can be used to study the effect of bortezomib and torin 1 on LPS-induced LCN2 production and secretion in reactive astrocytes.

Figure 6.

Effects of bortezomib and torin 1 in LPS-treated MS astrocytes. (A) MS astrocytes were treated with LPS (or untreated), with ACM collected from control MD astrocytes (MD ACM, −LPS), or with ACM collected from LPS-treated MD astrocytes (MD ACM, +LPS). The Lcn2 expression levels were determined via qRT-PCR, normalized against Gapdh levels, and expressed as a fold change relative to the control (n = 3; one-way ANOVA, followed by Tukey’s multiple comparison test). (B) During MD ACM (+LPS) treatment, MS astrocytes were co-treated with BTZ (1 μM) or torin 1 (1 μM) for 24 h. The mRNA levels of Lcn2 in MS astrocytes treated with MD ACM (−LPS or +LPS) or MD ACM (+LPS) with BTZ or torin 1 were measured via qRT-PCR, normalized against Gapdh levels, and expressed as a fold change relative to the control (n = 3; one-way ANOVA, followed by Tukey’s multiple comparison test). (C) ACM collected from MS astrocytes (MS ACM), which were treated with LPS (or untreated), was subjected to immunoblot detection of LCN2 and Coomassie blue staining as a loading control. (D) Under the same conditions as in (B), the mRNA levels of pan, A1, and A2 markers were measured via qRT-PCR, normalized against Gapdh levels, and expressed as a fold change relative to the levels in the control group (MD ACM (−LPS) (n = 3; one-way ANOVA, followed by Tukey’s multiple comparison test). (E) MS ACM, which was collected from MS astrocytes treated with MD ACM (−LPS or +LPS) in the presence of DMSO (−BTZ or −Torin 1), BTZ, or torin 1, was subjected to immunoblot detection of LCN2. To use as negative control (NC) and positive control (PC), MD ACM (−LPS or +LPS, respectively) was diluted to 1:3 in MS medium. Representative images of immunoblots are shown. qRT-PCR data are expressed as the means ± SEM from the indicated number of samples. *p < 0.05, **p < 0.01; ***p < 0.001 vs. the control or between two groups as indicated by the horizontal bars. n.s, not significant.

N-terminal signal peptide is required for the degradation and secretion of LCN2

Next, we investigated the mechanism regulating LCN2 degradation via the autophagy-lysosome pathway. Although it has been reported that LCN2 is resistant to extracellular peptidases [59], our results suggest that LCN2 can be rapidly degraded within cells. Therefore, we speculated that the difference between secreted and nascent (currently synthesized) LCN2 was important for its degradation. To address the mechanism regulating LCN2 degradation based on our speculation, we explored different moieties of secreted and nascent LCN2. The canonical secretory pathway is conserved in mammals and requires a specialized and consensus amino acid sequence known as signal peptide (SP) [60]. The nascent polypeptide chain with an N-terminal SP was translocated into the lumen of the endoplasmic reticulum (ER), where SP was cleaved by a signal peptidase to produce the secreted protein without the SP. Because LCN2 has an intrinsic SP moiety, we generated a construct harboring an SP-deletion LCN2 mutant (LCN2 Δ1-20) (Figure 7A). Additionally, an LCN2-green fluorescent protein (GFP) construct was generated to evaluate whether LCN2 can alter the destiny of other proteins that are stable and not secreted. We observed that LCN2-GFP acquired fluorescence properties while maintaining the intrinsic degradation properties of LCN2 in astrocytes (Figure S5).

Figure 7.

Signal peptide is essential for the secretion and degradation of LCN2. (A) Graphical representation of LCN2, a mutant LCN2 with 20 N-terminal amino acids deleted (LCN2 Δ1-20), and LCN2 tagged with a C-terminal GFP (LCN2-GFP). (B, C) Immunoblot detection of LCN2, LCN2-GFP, LCN2 Δ1-20, SQSTM1, and MAP1LC3 in HEK293T cells based on the types of LCN2 constructs transfected. After transfection, cells were treated with control vehicle (DMSO), torin 1, or BAF for 1 d. TUBA was used as a loading control. Asterisks indicate nonspecific bands. (D) Conditioned medium collected from HEK293T cells under various conditions as described above was subjected to immunoblot detection of secreted LCN2, LCN2-GFP, and LCN2 Δ1-20. For HEK293T-conditioned medium, Coomassie blue staining of polyacrylamide gels was used as a loading control. (E, F) Cycloheximide (CHX) chase of exogenous LCN2 or LCN2 Δ1-20. HEK293T cells were transfected with LCN2 or LCN2 Δ1-20 before BAF and CHX treatment. One day after transfection, intracellular LCN2 was chased in a medium containing 10 μg/mL CHX with or without BAF for up to 80 min. Changes in the LC3-II:I ratio over time indicate BAF activity. TUBA was used as a loading control. CHX-chased LCN2 or LCN2 Δ1-20 levels were normalized to TUBA levels and expressed as a percentage relative to the zero-chasing time (0 min) (n = 3). Representative immunoblots are shown. CHX chase data are expressed as the means ± SEM from the indicated number of samples.

Ectopic Lcn2 was strongly expressed since it was under the control of a viral promoter, unlike endogenous Lcn2, whose expression is tightly regulated by an endogenous promoter. The overexpression of LCN2, LCN2 Δ1-20, and LCN2-GFP in HEK293T cells were confirmed via immunoblotting (Figure 7B, C). Torin 1 treatment decreased the levels of LCN2 and LCN2-GFP, whereas bafilomycin A₁ increased the levels of these proteins, which is consistent with previous results (see Figure 4A and 5A). Immunoblot analysis of cells overexpressing LCN2 and LCN2-GFP following treatment with an autophagy activator or inhibitor suggested that LCN2 may be a prominent target of the autophagy-lysosome pathway. Notably, the basal levels of LCN2 Δ1-20 were higher than those of LCN2, and the level of LCN2 Δ1-20 was decreased by torin 1; however, it was unaffected by bafilomycin A1 (Figure 7B). Moreover, except for LCN2 Δ1-20, LCN2 and LCN2-GFP were secreted from HEK293T cells, and the effects of autophagy activators and inhibitors on LCN2 secretion were similar to those observed in astrocytes (Figure 7D, see Figure 4B, 5C, 6E).

The half-lives of ectopically expressed LCN2 and LCN2 Δ1-20 were also evaluated using the cycloheximide chase assay. Surprisingly, the half-life of ectopically overexpressed LCN2 was similar to that of endogenous LCN2 in astrocytes, whereas the half-life of LCN2 Δ1-20 was significantly increased, regardless of bafilomycin A₁ treatment (Figure 7E, F). The decrease in intracellular LCN2 levels, even under lysosome inhibition conditions (bafilomycin A₁ treatment), was probably due to differences in the secretory characteristics of astrocytes and HEK293T cells. Although further investigation is required to explore the effect of torin 1 on LCN2 Δ1-20, our results suggest that LCN2-SP is essential for the degradation and secretion of LCN2.

Regulation of proteasome and autophagy reduces the neurotoxicity of LCN2 secreted from reactive astrocytes

We next investigated whether proteasome inhibitors or autophagy activators could rescue the neurotoxic effects of secreted LCN2 from reactive astrocytes. To address this, ACM was collected from reactive astrocytes cultured under different conditions (Figure S1A). Incubation with ACM from LPS-induced cells reduced the viability of primary neurons, whereas incubation with ACM from LPS- and bortezomib-treated astrocytes maintained the viability of neurons (Figure 8A, Figure S6A). Furthermore, ACM prepared from LPS- and torin 1-treated astrocytes also promoted the survival of cultured primary neurons (Figure 8B, Figure S6B). To further validate that LPS-induced neurotoxicity depends on LCN2 levels, we performed Lcn2 knockdown in astrocytes and collected ACM after LPS exposure, which was then treated to the primary neurons. Although LPS-induced neurotoxicity was not completely overcome by Lcn2 knockdown in reactive astrocytes, it was greatly improved compared to the presence of LCN2 in reactive astrocytes. These results suggest that the major neurotoxin in ACM from LPS-treated astrocytes is LCN2, and bortezomib and torin 1-induced LCN2 reduction favors neuronal survival even in the presence of LPS (Figure 8A, B). Before ACM treatment, the levels of secreted LCN2 in the conditioned medium from astrocytes treated with LPS and/or torin 1 were also evaluated. We found that secreted LCN2 was only observed in the ACM prepared from astrocytes treated with LPS only (Figure 8C). Next, we prepared hippocampal neurons from mouse pups to reproduce these experiments. Consistent with the results from embryonic cortical neurons, hippocampal neurons incubated with ACM from LPS- and torin 1-treated astrocytes had higher survival rates than those incubated with ACM from LPS only-treated astrocytes (Figure 8D). Moreover, immunofluorescence staining revealed that some hippocampal neurons underwent apoptosis when cultured with ACM from LPS-treated cells, which was rescued by the presence of torin 1 (Figure 8E). We also investigated the effect of ACM collected from hippocampal astrocytes on hippocampal neurons and observed the similar results (Figure S6C). These results suggest that the effect of ACM on neurons can be applied in a more physiologically relevant manner by using ACM from the same region of the brain where neurons were originated. In addition, the astrocyte cultures in our experimental set-up contain microglia, which were needed for astrocytes to produce and secrete LCN2 in response to the LPS stimulation. Thus, our in vitro experimental observations may be transferable to the in vivo situation harboring more cells such as microglia. Overall, our experiments demonstrated the alleviation of LCN2 neurotoxicity and the neuroprotective effects of proteasome inhibition and autophagy activation under neuroinflammatory stress conditions.

Figure 8.

Neuronal viability is restored by proteasome inhibition and autophagy activation. (A, B) Astrocytes were infected with scrambled (−Lcn2 KD) or Lcn2 KD lentivirus (+Lcn2 KD) for 1 d, and then treated with PBS (−LPS) or LPS (+LPS), and control vehicle (DMSO [−BTZ, −Torin 1]), BTZ (+BTZ), or torin 1 (+Torin 1) for 2 d. Thereafter, the ACM was collected and treated to primary cortical neurons for 3 d, and the viability of neurons was measured using the MTT assay. Cell viability was expressed as a percentage relative to the control (n = 3; one-way ANOVA, followed by Tukey’s multiple comparison test). (C) Immunoblot detection of LCN2 secreted from astrocytes into ACM after LPS and torin 1 treatment. For ACM, Coomassie blue staining was used as a loading control. (D) The MTT assay was performed to measure the viability of hippocampal neurons treated with ACM collected from astrocytes and with LPS and torin 1 as described above. Cell viability was expressed as a percentage relative to the control (n = 3; Kruskal-Wallis one-way ANOVA, followed by Tukey’s multiple comparison test). (E) After ACM treatment, primary hippocampal neurons were immunostained with TUBB3 and Cleaved CASP3/CC3 (cleaved-caspase 3). DNA was visualized with DAPI. Representative images of cells or immunoblots are shown. MTT assay data are expressed as the means ± SEM from the indicated number of samples. *p < 0.05; **p < 0.01; ***p < 0.001 vs. the control or between two groups as indicated by the horizontal bars. n.s, not significant. Scale bar: 20 μm.

Discussion

Significant advances have been made in our current understanding of reactive astrocytes and their functions under various neuropathological conditions [28]. Notably, in the research field of neuroinflammation, these context-dependent in vivo studies have emerged beyond the traditional concept of reactive astrocytes (A1 or A2) or activated microglia (M1 or M2) [23,32]. Reactive astrocytes communicate with CNS-resident and CNS-infiltrating peripheral immune cells, including microglia, and this cooperation alters the CNS microenvironment, promoting or limiting pathogenesis. During this process, reactive astrocytes secrete various molecules, including chemoattractants, cytokines, and complement factors. LCN2, which is secreted by astrocytes under various neuropathological conditions, is considered an acute-phase protein and a pan-reactive astrocyte marker. In this study, we showed that cultured primary astrocytes treated with LPS led to the increased expression of A1 markers of cells and the production and secretion of LCN2. Moreover, LCN2 containing ACM is neurotoxic and can induce apoptosis in primary cultured neurons, consistent with previous reports [25,33]. Furthermore, our research showed that neurotoxic LCN2 levels could be significantly reduced by regulating two intracellular protein degradation pathways: the ubiquitin-proteasome system (UPS) and autophagy-lysosome pathway (ALP). Moreover, the reduced production and secretion of LCN2 resulted in neuroprotection, even under LPS-induced stress conditions.

To date, there have been controversial opinions whether LCN2 secreted from reactive astrocytes is neurotoxic or neuroprotective [61,62]. Under inflammatory stress conditions induced by LPS, Lcn2 expression is increased in reactive astrocytes with high expression of A1 markers, which have been considered to induce neurotoxicity, supporting that LCN2 can cause neuronal death. On the other hand, other study showed that, when astrocytes were treated with LCN2 secreted from damaged neurons during ischemic stroke, the expression of neurotrophic factor was increased and became reactive with increased expression of A2 markers [51]. Thus, the controversial effects of LCN2 in vivo may need to be further evaluated in a context-dependent manner depending on the stress conditions. Interestingly, it has been reported that LPS injection into mouse cerebellum induced microglia activation and cerebellar neuronal loss, resulting in an acute ataxia phenotype [63]. Although it is unclear whether LCN2 served as a critical neurotoxic factor in this case, our data suggest that robust increase in LCN2 levels can act as a major neurotoxic factor in vivo under LPS-induced stress conditions.

Based on our observations, the question arises as to how LCN2 secreted from reactive astrocytes promotes the death of nearby neurons, but not the astrocytes themselves. One possibility is the presence and function of the LCN2 receptor, also known as SLC22A17 (solute carrier family 22 member 17) [64]. SLC22A17 binds to ligands, including LCN2, and facilitates endocytosis. During this process, owing to the molecular characteristics of LCN2, intracellular iron concentrations could be increased by LCN2 uptake, which may lead to cell death. Interestingly, SLC22A17 is highly expressed in the brain, although it is still unclear whether its expression is altered in neurons under LPS-induced stress conditions [65]. In addition, previous studies have shown that LCN2-mediated cell death is closely related to SLC22A17 expression [33,64]. Since the molecular function of SLC22A17 is involved in the regulation of elements essential for cell survival, including intracellular iron homeostasis [66], a neuroprotection strategy based on the regulation of neurotoxic LCN2, but not SLC22A17, is more practical for therapeutic purposes.

LCN2 is an inducible secretory protein and considered protease-resistant because of its easy detection in the serum after certain types of injury [3,59]. However, intracellular LCN2, which is directly produced or entered cells via endocytosis, is inevitably exposed to intracellular proteolytic pathways, such as autophagy. Interestingly, another group showed that LCN2 treatment inhibited autophagic flux in cardiomyocytes [55]. Furthermore, a recent study revealed that increased LCN2 levels in a mouse model of dry AMD (age-related macular degeneration) reduced autophagy, resulting in the activation of inflammasome-ferroptosis processes [56]. Based on these studies, intracellular accumulated LCN2 is associated with the autophagic machinery and acts as a suppressor of autophagic flux. However, our research suggests that newly synthesized LCN2 can be degraded by autophagy before secretion. One possible reason for this is that LCN2 has a signal peptide that is synthesized first at its N-terminus, which may provide a chance to be regulated by the endoplasmic reticulum (ER)-associated protein quality control [60]. This quality control system is well-conserved in eukaryotes and removes misfolded or surplus proteins in the ER lumen through the UPS or ALP. Notably, we showed that ALP, but not the UPS, is a candidate for LCN2 degradation. Although it remains unknown whether endocytosed LCN2 from the extracellular space can also be degraded by ALP, our data suggest that LCN2 without a signal peptide may not be a target of autophagy. Thus, ALP activation facilitated the degradation of newly synthesized LCN2, resulting in reduced secretion.

In this study, we evaluated bortezomib as a potent drug that suppresses NFKB signaling in LPS-treated reactive astrocytes to reduce neurotoxicity. Consistent with previous reports, our results suggest that bortezomib treatment suppresses canonical NFKB signaling and the upregulation of Lcn2 expression in LPS-induced reactive astrocytes. Moreover, we demonstrated that bortezomib treatment could alter the expression of reactive astrocyte markers, suggesting that the LPS-induced reactive astrocytes may acquire novel unique functions via proteasome inhibition. However, it should not be overlooked that the accumulation of misfolded proteins via proteasome inhibition can also induce ER stress in cells. In moderate cases, it alters intracellular metabolites and de novo translation; however, it can induce apoptosis in severe cases [67]. We also used torin 1 as an activator of ALP to reduce LCN2 secretion and neurotoxicity. Torin 1 is widely used in biochemical research to inhibit MTORC1 and MTORC2, which phosphorylates target proteins that regulate fundamental cellular responses, including ALP [68]. The torin 1-induced inhibition of MTORC1 causes the degradation of ALP target proteins; however, it also facilitates nonspecific protein degradation by activating both the UPS and ALP [69]. Further in-depth investigations are required to warrant the intracellular action of bortezomib and torin 1 prior to the use of these reagents for therapeutic purposes.

In summary, our study revealed two potentially interesting neuroprotective mechanisms that target LPS-induced reactive astrocytes and reduce the neurotoxicity from robust LCN2 secretion (Figure 9). First, proteasome inhibition could suppress canonical NFKB signaling in reactive astrocytes, resulting in the downregulation of Lcn2 expression and partial suppression of astrocyte reactivation. Second, the inhibition of MTOR signaling could activate ALP in LPS-induced reactive astrocytes, promoting intracellular degradation of the secretory protein LCN2, thereby reducing its secretion.

Figure 9.

Graphical representation of LPS-induced neurotoxicity and alleviation mechanisms of proteasome inhibition and autophagy activation in astrocytes. Inflammatory stress induced by LPS stimulation promotes degradation of NFKBIA/IκBα and nuclear translocation of NFKB RELA/p65, which increases the expression of Lcn2, making astrocytes reactive. LCN2 is rapidly degraded by autophagy-lysosome pathway, but undegraded LCN2 can be secreted from the cells and act as neurotoxin (top). Upon proteasome inhibition, degradation of NFKBIA and nuclear translocation of NFKB RELA are inhibited, thus Lcn2 expression levels are reduced along with secreted amount of LCN2 (bottom left). Upon activation of autophagy, degradation of LCN2 is promoted and the amount of secreted LCN2 is subsequently decreased (bottom right).

Materials and methods

Mouse studies

Wild-type CD-1 (ICR) mice were housed in plastic cages with ad libitum access to food and water under a 12-h light/dark cycle. All experiments with mice (breeding, euthanasia, isolation of primary cells from embryonic or postnatal mouse brains) were approved by the University of Seoul Institutional Animal Care and Use Committee (approval no. UOS-IACUC-2020-03-A, UOS IACUC-2021-01-TA). All animal procedures were performed in accordance with relevant guidelines and regulations approved by the UOS IACUC.

Primary cell isolation and cell culture

Primary glial cells (MD astrocytes, or simply described as astrocytes in this study) were isolated from the mouse cortex or hippocampus on postnatal day 1, as previously described with slight modifications [70]. Briefly, mouse pups were sacrificed, and their whole brains were removed and kept in ice-cold Hanks’ balanced salt solution (HBSS, calcium/magnesium-free; Gibco™, 14175095). After resection of the olfactory bulbs and cerebellums from the whole brain, the meninges and any non-cortical forebrain or non-hippocampal tissues were removed using surgical instruments and a dissecting microscope. The cortex that was divided into 4–5 pieces or the entire hippocampus was kept in a conical tube containing HBSS, and 5 mL of 0.5% trypsin-EDTA solution (Gibco™, 25300062) was added to the collected pieces (eight cortical hemispheres from four mouse pups or ten pairs of hippocampal regions from ten mouse pups in each conical tube) after the removal of HBSS. The mixture was then incubated for 20 min in a 37°C water bath, with shaking every 5 min. After trypsinization, 5 mL of complete medium (Dulbecco’s modified Eagle medium [DMEM; Gibco™, TFS-12800017] supplemented with 10% fetal bovine serum [FBS; Gibco™, 26140079], 1× L-glutamine [Corning®, 25–005-CI], and 1× antibiotics-antimycotics [Corning®, 30–004-CI]) was added to the cell mixture, inverted several times, and centrifuged at 300 × g for 5 min at room temperature. After gently removing the supernatant, 10 mL of fresh complete medium was added to the mixture and the cortical or hippocampal pieces were triturated via resuspension. The dissociated cell suspension was spread on a poly D-lysine (Sigma-Aldrich, P0899)-coated 100-mm culture dish, and the cells were maintained at 37°C in a 5% CO₂ incubator, replacing the spent medium with fresh medium every two days. On day 7 in vitro (DIV7), the prepared cells were counted via Trypan blue staining using a cell counter (Logos Biosystems) and seeded on culture plates or dishes at a density of approximately 2.0 to 4.0 × 10² cells/mm², depending on the experiments.

Primary neuronal cell isolation from the embryonic cortex or postnatal hippocampus was performed as previously described with slight modifications [49,71]. As mentioned above, the brains were isolated from mouse embryos (14 days post coitum) or pups. The embryonic or postnatal whole brains were processed by removing the meninges and other regions except the cortex using surgical instruments before collection into HBSS-containing conical tubes. HBSS was carefully removed from the conical tube, and 5 mL of 0.5% trypsin-EDTA solution was added to the tissues. After trypsinization for 20 min with gentle rocking, 5 mL of complete medium was added to the tissues, inverted several times to quench the trypsinization, and centrifuged at 300 × g for 5 min at room temperature. The supernatant was carefully removed, 10 mL of fresh complete medium was added, and the regions of interest were triturated via resuspension. The dissociated cell suspension was then strained through a 70-μm nylon mesh, and the collected cells were subsequently counted via Trypan blue staining using a cell counter (Logos Biosystems). Depending on the experiments, the counted cells were seeded on culture plates or dishes at densities ranging from 2.0 to 4.0 × 10² cells/mm². The seeding medium was carefully removed after 4 h of culture, and fresh complete neuronal cell culture medium (Neurobasal™ medium [Gibco™, 21103049] supplemented with 1× B-27™ supplement [Gibco™, 17504044], 1× GlutaMax™ [Gibco™, 35050061], and 1× antibiotics-antimycotics) was added. Half of the spent medium was changed every two days.

HEK293T cells were purchased from the American Type Culture Collection (CRL-3216) and maintained in complete medium at 37°C in a 5% CO₂ incubator. Fresh complete medium was added to the cells every two days, and the cells were split in ratios ranging from 1:20 to 1:10 to prevent over-confluence every three to four days. Detailed information on the cell culture reagents used in this study are summarized in Table 1.

Table 1.

Summary of cell culture reagents used in this study.

Complete medium for HEK293T cell and astrocyte culture
Reagents	Manufacturer	Catalog No.	Description
DMEM	Gibco™	TFS-12800017	Made and used according to the manufacturer’s protocol.
Fetal bovine serum	Gibco™	26140079	Added to 10% of the complete medium.
L-Glutamine (200 mM)	Corning®	25–005-CI	Added to 1% (2 mM) of the complete medium.
Antibiotics-antimycotics	Corning®	30–004-CI	Added to 1% of the complete medium.
Complete Neurobasal medium for neuron culture
Neurobasal™ medium	Gibco™	21103049
B-27™ supplement	Gibco™	17504044	Added to 2% of the complete medium.
GlutaMAX™ supplement	Gibco™	35050061	Added to 1% of the complete medium.
Antibiotics-antimycotics	Corning®	30–004-CI	Added to 1% of the complete medium.
General cell culture reagents
1× Phosphate-buffered saline			Made and used according to the laboratory protocol.
1× HBSS	Gibco™	14175095	No calcium, no magnesium, and no phenol red
Trypsin-EDTA (0.05%)	Gibco™	25300062
Poly-D-lysine hydrobromide	Sigma-Aldrich	P0899	Used for the plate coating at a final concentration of 50 μg/mL.
LUNA™ cell counting slides	Logos Biosystems	L12001
Trypan blue stain, 0.4%	Logos Biosystems	T13001
Other reagents
LPS (from E. coli)	Sigma-Aldrich	L4391
MG132	Sigma-Aldrich	474790
Bortezomib	Selleckchem	S1013
Bafilomycin A1	Enzo Life Sciences	BML-CM110
Torin 1	Selleckchem	S2827
NV5138	MedChemExpress	HY-114384
EBSS	Sigma-Aldrich	E3024

Magnetic-activated cell sorting for pure astrocyte culture

Magnetic-activated cell sorting (MACS) was performed using an Anti-ACSA-2 MicroBead Kit (ACSA-2 [astrocyte cell surface antigen-2]; Miltenyi Biotec, 130–097-678), as previously described with slight modifications [58]. Cortical regions were isolated from the mouse brain on postnatal day 5 and divided into small pieces. After completion of cortical piece preparation, collected pieces were enzymatically digested using Neural Tissue Dissociation Kit (papain-based; Miltenyi Biotec, 130–092-628), following the manufacturer’s protocol. Enzymatic tissue digestion was carried out for 15 min in a 37°C water bath, and digested tissues were treated subsequently with DNase and incubated for additional 10 min. Thereafter, cortical pieces were gently triturated with fire-polished glass pipette and dissociated cell suspension was filtered through a cell strainer to remove undigested tissues or large tissue debris. After several washing with 0.5% bovine serum albumin (BSA; Sigma-Aldrich, A1933) in PBS (pH 7.4, 137 mM NaCl, 2.7 mM KCl, and 10 mM Na₂HPO₄), cells were incubated with anti-FcR blocker and anti-ACSA-2 MicroBead sequentially on ice. After 30 min of incubation, ACSA-2-positive cells were captured in LS Columns (Miltenyi Biotec, 130–042-401), which were located in QuadroMACS™ Separator (Miltenyi Biotec), while the cell suspension was flowing down. The column was first washed to remove the negative cell suspension and the ACSA-2-positive cell suspension was subsequently eluted and collected. Purified cells were counted and seeded on culture plates or dishes at densities ranging from 2.0 to 4.0 × 10² cells/mm² depending on the experiments. The pure astrocytes (MS astrocytes) were cultured with MS medium (50% Neurobasal™ medium + 50% DMEM supplemented with 1× B-27™ supplement, 1× GlutaMax™, 1× antibiotics-antimycotics, and 5 ng/mL of recombinant human HBEGF (R&D Systems, 259-HE-050) for seven days to grow.

Cell starvation and general cell culture

For cell starvation, primary astrocytes were seeded on 60-mm culture dish two days before LPS (E.coli origin; Sigma-Aldrich, L4391) treatment. One day after LPS treatment, cells were briefly washed with PBS and subsequently incubated with fresh complete medium (with or without LPS) or Earle’s balanced salt solution (EBSS [Sigma-Aldrich, E3024], with or without LPS) for 4 h. After induction of cell starvation, samples were collected and subjected to immunoblot analysis. For general cell culture, cells were seeded on appropriate culture plates or dishes one day before transfection or virus transduction. One day after transfection or transduction, cells were briefly washed with PBS and incubated with LPS or treated with other reagents/chemicals for one day. When the cells were not transfected (or not infected), they were seeded two days prior to treatment with LPS or other reagents/chemicals to synchronize cell harvest times regardless of transfection. In case of LPS pretreatment conditions, cells were treated with LPS for one day, followed by other drugs for 4 to 18 h. Schematic illustrations of the cell culture methods according to the various experimental conditions are shown in Figure S1.

Transfection and lentivirus production

Conventional transfection of HEK293T cells to deliver plasmids was performed using the standard calcium phosphate method. Six hours prior to transfection, the cells were plated on culture plates or dishes at densities ranging from 2.0 to 4.0 × 10² cells/mm², depending on the experiments. The mixture used for transfection was prepared (1 μg of plasmid per 1.0 × 10⁵ cells, 125 mM CaCl₂, 1× HEPES-buffered saline, and 5 mM HEPES, pH 7.3-buffered water to adjust the final volume), mixed well via vortexing, and incubated for 20 min at room temperature. The mixture containing plasmid precipitates was added to the cells, which were incubated at 37°C overnight in a 5% CO₂ incubator. After incubation, the cells were washed twice with PBS, and fresh medium was added.

Lipofectamine (Lipofectamine™ 3000 Transfection Reagent; Invitrogen, L3000015)-mediated transfection of primary astrocytes was carried out following the manufacturer’s protocol. Briefly, primary astrocytes were seeded on glass coverslips (12-mm diameter) placed in a 24-well plate at a density of approximately 6.0 × 10² cells/mm² and transfected with plasmids using Lipofectamine. After transfection, cells were washed with PBS and the medium was replaced with complete medium.

Lentivirus production was performed using the standard calcium phosphate method, as mentioned above. Twenty-four hours after changing the medium, the supernatant was collected using a 15 mL syringe and strained with a 0.45-μm membrane filter (GVS, FJ25ASCCA004FL01). The strained supernatant was mixed with a Lenti-X™ Concentrator (Takara Bio, 631231) and incubated at 4°C overnight. After incubation, the mixture was centrifuged at 1,500 × g for 45 min at 4°C, the viral pellet was resuspended in PBS, titrated using a qPCR Lentivirus Titration Kit (Applied Biological Materials, LV900), and stored in aliquots at −80°C. Detailed information on the plasmids used in this study is summarized in Table S1.

Immunoblot analysis

Cells were lysed in radioimmunoprecipitation assay buffer (50 mM Tris-HCl, pH 7.5, 200 mM NaCl, 1% NP-40 [Sigma-Aldrich, 56741], 1% sodium deoxycholate [Sigma-Aldrich, 30,970], 0.1% sodium dodecyl sulfate [SDS; Sigma-Aldrich, L3771]) with a protease inhibitor cocktail (1 mM PMSF [Roche Diagnostics GmbH, 10837091001], 1 μg/μL aprotinin [Roche Diagnostics GmbH, 10236624001], 1 μg/μL leupeptin [Roche Diagnostics GmbH, 11017101001]) via resuspension. If needed, a phosphatase inhibitor cocktail (Quartett, PPI 1041) was also added to the lysis buffer. After 30 min of incubation on ice, the lysed samples were centrifuged at 15,000 × g for 5 min at 4°C, and the supernatant was collected. The protein concentration of each sample was determined using the Pierce^TM BCA Protein Assay Kit (Thermo Fisher Scientific, 23225) and a SpectraMax® M2^e microplate reader (Molecular Devices). Proteins (15–30 μg) were separated via SDS polyacrylamide gel electrophoresis at 25 mA (per gel) for 60–75 min using a PowerPac^TM Basic power supply (Bio-Rad). After electrophoresis, the separated proteins were transferred onto a polyvinylidene fluoride (PVDF) membrane (Millipore, ISEQ00010) at 100 V for 1 h 30 min using a PowerPac^TM HC power supply (Bio-Rad). The transferred membranes were blocked with 5% skim milk (Genomic Base, SKI400) in TBST (Tris-buffered saline [TBS; 50 mM Tris-HCl, pH 7.5, 150 mM NaCl] containing 0.05% Tween® 20 [Sigma-Aldrich, P1379]) buffer for 1 h at room temperature and incubated with the primary antibodies suspended in 5% skim milk in TBST buffer at 4°C overnight. After primary antibody incubation, the membranes were washed with TBST buffer and subsequently incubated with the corresponding secondary antibodies suspended in 5% skim milk in TBST buffer for 1 h at room temperature. Chemiluminescent signals were detected, and images were captured using an enhanced chemiluminescence solution (ATTO, WSE-7120L) with a ChemiDoc system (LI-COR Bioscience). Detection of immunoblot bands was performed using the image analysis software (Image Studio) equipped on the ChemiDoc system. The band with the highest intensity was set to 100% and the intensities of other bands were normalized to this band by adjusting the exposure time. Thus, the immunoblot bands were in the linear range for comparison with no saturation of band intensities. Band intensities were quantified using ImageJ software when necessary (version 1.8.0). The antibodies used in this study are summarized in Table 2.

Table 2.

Summary of antibodies used in this study.

Primary antibodies	Isotype	Supplier	Cat. No.	WB	IF
TUBA/α-Tubulin	Mouse IgG	Santa Cruz Biotechnology	SC-32293	1:1,000
LMNB (lamin B1)	Mouse IgG	Santa Cruz Biotechnology	SC-6217	1:1,000
ACTB/β-Actin	Mouse IgG	Santa Cruz Biotechnology	SC-47778	1:1,000
Ubiquitin	Mouse IgG	Santa Cruz Biotechnology	SC-8017	1:1,000
LCN2 (lipocalin 2)	Rabbit IgG	Thermo Fisher Scientific	PA579590	1:500 − 1,000	1:200
LC3B	Rabbit IgG	Cell Signaling Technology	2775S	1:1,000
TUBB3/β3-Tubulin	Mouse IgG	Santa Cruz Biotechnology	SC-51670	1:1,000	1:200
GFAP	Mouse IgG	Millipore	MAB360	1:1,000
GFAP	Chicken IgG	Millipore	AB5541		1:500
SQSTM1/p62	Mouse IgG	Thermo Fisher Scientific	MA527800	1:500 − 1,000
SQSTM1/p62	Rabbit IgG	Cell Signaling Technology	5114	1:1,000	1:1,000
RPS6KB/p70S6K	Rabbit IgG	Cell Signaling Technology	9202	1:1,000
p-RPS6KB/p70S6K (Thr389)	Rabbit IgG	Cell Signaling Technology	9205	1:1,000
NFKBIA/IκBα	Mouse IgG	Thermo Fisher Scientific	MA515132	1:1,000
p-NFKBIA/IκBα (Ser32, Ser36)	Mouse IgG	Thermo Fisher Scientific	MA515224	1:1,000
NFKB RELA/p65	Rabbit IgG	Thermo Fisher Scientific	510,500	1:500 − 1,000	1:100
C3 (complement component 3)	Rabbit IgG	Bioss	BS-4877 R	1:1,000
STAT3	Rabbit IgG	Cell Signaling Technology	9132	1:500
p-STAT3 (Tyr705)	Mouse IgG	Santa Cruz Biotechnology	sc-8059	1:500
Cleaved CASP3 (caspase 3)	Rabbit IgG	Cell Signaling Technology	9661S		1:1,000
Secondary antibodies	Label	Supplier	Cat. No.	WB	IF
Goat anti-mouse IgG	HRP	Enzo Life Sciences	ADI-SAB-100-J	1:10,000
Goat anti-rabbit IgG	HRP	Enzo Life Sciences	ADI-SAB-300-J	1:10,000
Goat anti-mouse IgG	Alexa Fluor 488	Thermo Fisher Scientific	A-11001		1:1,000
Goat anti-mouse IgG	Alexa Fluor 555	Thermo Fisher Scientific	A-21422		1:1,000
Goat anti-rabbit IgG	Alexa Fluor 488	Thermo Fisher Scientific	A-11006		1:1,000
Goat anti-rabbit IgG	Alexa Fluor 555	Thermo Fisher Scientific	A-21428		1:1,000
Goat anti-chicken IgG	Alexa Fluor 488	Thermo Fisher Scientific	A-11039		1:1,000

Cycloheximide chase assay

To inhibit de novo protein synthesis, LPS-pretreated cells (100 ng/mL) were treated with 10 μg/mL of cycloheximide (Sigma-Aldrich, 01810) with or without other chemicals (e.g., bortezomib [Selleckchem, S1013] and bafilomycin A₁ [Enzo Life Sciences, BML-CM110]). After incubation, the treated cells were harvested and lysed with a radioimmunoprecipitation assay buffer, and immunoblot analysis was performed as described above. LCN2 and TUBA/α-tubulin band intensities were quantified using ImageJ software (version 1.8.0). LCN2 levels (normalized to TUBA levels) were expressed as LCN2 remaining (%) relative to the control.

Quantitative RT-PCR (qRT-PCR) analysis

Total RNA was isolated from cultured cells or prepared from tissues using TRI reagent (Molecular Research Center, TR118), following the manufacturer’s protocol. Before reverse transcription, the RNA samples were quantified using NanoDrop™ One (Thermo Fisher Scientific, 13400518) and treated with DNase I (Thermo Fisher Scientific, 18068015) for 15 min at room temperature. To generate 100 μL of cDNA from each RNA sample, reverse transcription was performed with 2 μg of total RNA per sample with a SuperiorScript II reverse transcriptase (Enzynomics, RT005M), according to the manufacturer’s protocol, using a Veriti™ 96-Well Fast Thermal Cycler (Thermo Fisher Scientific, 4375305). A total of 5 μL of cDNA was used for qRT-PCR, which was performed using SYBR Master Mix (Enzynomics, RT500M) and an iCycler system with the iCycler iQ software (version 2.0; Bio-Rad). The mRNA expression levels were normalized to that of Gapdh. Information on the primers used for qRT-PCR is summarized in Table S2.

Immunocytochemistry and immunofluorescence analysis

Before cell seeding, cover glasses were placed on a 100-mm culture dish coated with poly-D-lysine (50 μg/mL in autoclaved ddH₂O). After incubation, the coated cover glasses were washed twice with a sufficient volume of autoclaved ddH₂O and transferred onto 24-well cell culture plates or 60-mm cell culture dishes containing complete culture medium. The cells were trypsinized, counted, and then seeded onto plates or dishes at densities ranging from 2.0 to 4.0 × 10² cells/mm², depending on the experiments. After treatment and incubation with the appropriate chemicals, the cells on the cover glasses were briefly washed with PBS, transferred to a humidified chamber, and fixed with 4% paraformaldehyde at 4°C overnight. After fixation, the cells were permeabilized with 0.3% Triton™ X-100 (Sigma-Aldrich, X100) in PBS for 15 min at room temperature and blocked with 3% BSA in PBS for 1 h at room temperature. Subsequently, the cells were incubated with the appropriate antibodies diluted in 3% BSA in PBS at 4°C overnight. The primary antibodies were washed three times with 0.03% Tween-20 in PBS, followed by incubation with Alexa Fluor 488- or 555-conjugated anti-mouse or rabbit IgG (Invitrogen) with 0.1 μg/mL of 4′,6-diamidino-2-phenylindole (DAPI) for 1 h at room temperature. The secondary antibodies were washed three times with 0.03% Tween-20 in PBS and the cells were finally mounted on slides using ProLong™ Gold Antifade Reagent (Thermo Fisher Scientific, P36930). Finally, immunofluorescence images were obtained using an LSM 800 confocal microscope and ZEN Blue software (Carl Zeiss).

LysoTracker staining

Cells were stained with 250 nM LysoTracker^TM Red DND-99 (Thermo Fisher Scientific, L7528) for 2 h in a dark chamber. After staining, the cells were fixed in 4% paraformaldehyde for 1 h at room temperature. Then, the cells were washed three times with PBS for 10 min each time and subjected to immunofluorescence staining with anti-LCN2 antibody, as mentioned above. The cells were then mounted using ProLong™ Gold Antifade Reagent. After mounting, the cells were observed under a confocal fluorescence microscope (LSM 800, Carl Zeiss).

Proteasome activity assay

Cells were lysed in a lysis buffer (10% glycerol, 5 mM MgCl₂, 100 mM NaCl, 5 mM NaH₂PO₄, 0.5% NP-40, 5 mM ATP (Sigma-Aldrich, A6419), and 1 mM DTT, with 1 μg/mL aprotinin, 1 μg/mL leupeptin, and 1 mg/mL PMSF as protease inhibitors) before the assay. After determining the protein concentration of the cell lysates using the Pierce^TM Bradford Protein Assay Kit (Thermo Fisher Scientific, 23,238), total cell lysates (10 μg) were subjected to a proteasome activity assay. To measure chymotrypsin-like activity, the fluorogenic substrate 0.2 mM suc-LLVY-AMC (Millipore, APT280) was added to an assay buffer (50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mg/mL BSA, 1 mM ATP, 1 mM DTT). The intensity of free AMC was measured using a SpectraMax® M2^e microplate reader (Molecular Devices) at excitation and emission wavelengths of 380 nm and 460 nm, respectively.

Statistical analysis

For multiple comparisons and analyses with more than two groups, parametric one-way analysis of variance (ANOVA) was performed for data that passed Shapiro-Wilk normality test and equal variance test. For data that were not under normal distribution, non-parametric one-way ANOVA on ranks (Kruskal-Wallis test) was used instead. These analyses were followed by pairwise multiple comparisons using a post-hoc Tukey’s test. For the comparison between two groups, two-tailed unpaired Student’s t-test was performed with 95% confidence. All statistical analyses were carried out using a SigmaPlot 14.5 software.

Funding Statement

This work was supported by the National Research Foundation of Korea [2021R1A6A3A13044664] to BKJ; National Research Foundation of Korea [2021R1F1A1049089, 2022R1F1A1063323] to KYR; National Research Foundation of Korea [RS-2022-00165439] to DEK.

Disclosure statement

The authors declare no competing interests.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15548627.2023.2180202