Valosin-containing protein is a key mediator between autophagic cell death and apoptosis in adult hippocampal neural stem cells following insulin withdrawal

Since VCP has not been studied in NSCs before, first, we examined the expression level of VCP in HCN cells. VCP was highly expressed in HCN stem cells in vitro, compared to the total hippocampal tissue derived from 8-week-old rat and embryonic hippocampal primary neurons after 9 days in vitro (Fig. 1a). To further examine whether VCP is abundantly expressed in HCN cells than hippocampal neurons in vivo, we performed immunohistochemical analysis in 8-week-old rat brain tissue, since HCN cells were derived from the adult rat hippocampus of the same age [13]. VCP was highly expressed in the NSCs in the subgranular zone (SGZ) of the dentate gyrus, as revealed by co-localization with glia fibrillary acidic protein (GFAP), a NSC maker in SGZ [32]. VCP was also expressed in the granule neurons of the DG in adult hippocampus, when microtubule-associated protein 2 (MAP2) was used as a neuronal marker [33] (Fig. 1b). The time course analysis of VCP protein and mRNA expression levels was performed following insulin withdrawal. From here, we denote the insulin-containing and insulin-depleted culture conditions as I(+) and I(-), respectively. Consistent with our prior studies [13‐15], high level of autophagy was induced in I(-) (Fig. 1c). Autophagy induction was verified by an increase in the level of type II of microtubule-associated protein 1 light chain 3 (LC3-II), since this is the most reliable biochemical marker of autophagy [34‐36]. VCP protein level was decreased 48 h following insulin withdrawal (Fig. 1c). To further confirm the decrease of VCP protein level following insulin withdrawal in ex vivo HCN culture, we measured VCP levels in organotypic hippocampal slice culture prepared from 8-week-old rat with vimentin as a NSC marker [37, 38]. VCP expression level was highly enriched in vimentin-immunoreactive stem cells in the SGZ in the presence of insulin, but its level significantly decreased in insulin-deprived slices (Fig. 1d). These results with organotypic slice cultures were well accordant with the data obtained from HCN cell culture, demonstrating the decrease of VCP protein level following insulin withdrawal in ex vivo and in vitro conditions. Despite the decrease of VCP protein levels in HCN cells and hippocampal slices, the levels of VCP mRNA expression were not changed by insulin withdrawal in HCN cells (Fig. 1e). These data suggest that VCP may be subject to post-translational regulation. To test whether VCP itself is a substrate for autophagic degradation or proteasome in insulin-deprived HCN cells, autophagic flux inhibitor, Bafilomycin A1 (BafA1) was treated following insulin withdrawal [39]. Interestingly, BafA1 (30 nM for 3 h) significantly prevented VCP degradation suggesting that autophagy may be responsible for degradation of VCP in insulin-deprived HCN cells (Fig. 1f).

VCP is known as a positive regulator in autophagosome maturation. To test whether VCP ATPase activity is required for ACD in HCN cells, HCN cells were treated with VCP ATPase activity inhibitor, N ² , N ⁴ -dibenzylquinazoline-2,4-diamine (DBeQ) [40], in insulin-depleted media and cell death was measured. We tested a wide range of DBeQ concentrations both in I (+) and I (-) HCN cells and chose 0.5 μM concentration of DBeQ, based on marginal effect on cell death in I (+) but significant increase of cell death in I (-) HCN cells (Fig. 2a). Hereafter, this concentration of DBeQ has been used throughout this study. Since cell death in I(-) HCN cells was boosted by DBeQ, we wondered if this increase was due to the continuance of the default ACD or switch to different mode of PCD. To determine whether ACD was affected by DBeQ in I(-) HCN cells, the different kinds of PCD inhibitors were treated (Fig. 2b). Consistent with the absence of apoptotic indices, Z-VAD-FMK failed to reduce cell death in I(-) HCN cells (data not shown), which was well documented in our prior studies [13‐15]. However, interestingly, Z-VAD-FMK significantly decreased cell death in DBeQ-treated I(-) HCN cells (Fig. 2b), suggesting that apoptosis was induced by VCP inhibition in I (-) HCN cells, which otherwise underwent genuine ACD. Staurosporine (STS), a common apoptosis inducer, was used as a positive control of apoptosis in I(+) HCN cells. STS-induced cell death was efficiently prevented by a broad-spectrum caspase inhibitor, Z-VAD-FMK (Fig. 2b). Also, necrostatin-1, a widely used necroptosis inhibitor [41], did not change cell death rate in DBeQ-treated I(-) HCN cells. This pharmacological inhibitor study suggests that VCP inhibition triggered a switch of cell death mode from ACD to apoptosis in I(-) HCN cells. Further analyses were performed to demonstrate the induction of apoptosis following VCP inhibition in I (-) HCN cells. Cleaved, active form of caspase 3 and chromatin condensation was observed in DBeQ-treated I(-) HCN cells (Fig. 2c, d). A marked increase in Annexin V positive staining indicating the derangement of phosphatidylserine distribution during apoptosis was also observed in DBeQ-treated I(-) HCN cells (Fig. 2e). Taken together, these data show that inhibition of VCP switches ACD to apoptosis in insulin-deprived HCN cells.

To further test whether genetic suppression of VCP expression also switches ACD to apoptosis, VCP-targeting small interfering RNA (siRNA) was transfected in HCN cells following the presented experimental scheme (Fig. 3a–b). In support of the results obtained with DBeQ treatment, knockdown of VCP did not alter cell death rate in I(+) condition, but significantly enhanced cell death in I(-) HCN cells (Fig. 3c). Augmented cell death in I(-) HCN cells by VCP knockdown was substantially diminished by Z-VAD-FMK (Fig. 3c) and accompanied by caspase-3 activation (Fig. 3d). Collectively, these data indicate that genetic knockdown of VCP expression induced a switch of PCD mode from ACD to apoptosis in I(-) HCN cells in a similar fashion as pharmacologic inhibition of VCP. However, it should be noted that VCP depletion or inactivation did not induce apoptosis at basal state in I(+) HCN cells, which is in sharp contrast with the previous studies reported in yeast, Caenorhabditis. elegans, HeLa, and several other mammalian cells [31, 42‐49].

To exclude the possibility that DBeQ induces apoptosis independent of inhibiting VCP, we treated VCP-suppressed I(-) HCN cells with DBeQ. Unlike the elevation of cell death in NT siRNA-treated HCN cells, DBeQ failed to increase cell death in I(-) HCN cells depleted of VCP (Fig. 3e). These data suggest that DBeQ-induced apoptosis in I(-) condition was VCP-dependent. Also, to test whether inhibition of VCP by DBeQ was due to perturbation of non-autophagy related functions of VCP independently of autophagy, we stably knocked down Atg7 in HCN cells before DBeQ treatment. HCN cells were transduced with the lentivirus expressing Atg7-targeting shRNA and subject to puromycin selection. Atg7 knockdown significantly decreased ACD in I(-) HCN cells, consistent with our previous studies [14, 15]. Of particular interest, DBeQ failed to increase cell death in I(-) HCN cells depleted of Atg7, nor did Z-VAD-FMK affect cell death (Fig. 3f). These data strongly demonstrate that ACD is a prerequisite to DBeQ-induced switch to apoptosis in HCN cells and DBeQ-induced apoptosis proceeds following yet-to-be-discovered mechanism for interaction between ACD and apoptosis, but is not non-specific cytotoxic event.

Our findings that loss of VCP activity switched ACD to apoptosis suggest that VCP is required for the progress of ACD in I(-) HCN cells. However, it is not known whether VCP functions at the autophagosome maturation step, as in other studies [27‐29], or plays new roles in the autophagy of HCN cells upon insulin withdrawal. To elucidate what role VCP may play in the progression of autophagy in HCN cells, we depleted VCP function by siRNA or using the same chemical inhibition approach of VCP activity and examined the autophagy flux rate. VCP inactivation by DBeQ or siRNA knockdown in I(+) HCN cells led to a modest increase in LC3-II level (Fig. 4a–b). On the other hand, DBeQ treatment or VCP siRNA knockdown in I(-) HCN cells substantially decreased LC3-II levels (Fig. 4c–d). When autophagy flux was blocked by addition of BafA1, there was a marked increase in the amount of LC3-II and they reached the same level between BafA1 control and BafA1/DBeQ or BafA1/VCP siRNA in I(+) HCN cells (Fig. 4a–b). In the case of I(-) HCN cells, BafA1 also increased LC3-II levels. However, interestingly, BafA1-induced accumulation of LC3-II was significantly less in BafA1/DBeQ or BafA1/VCP siRNA than BafA1 control in I(-) HCN cells (Fig. 4c–d).

An increase in LC3-II level may indicate increased formation of autophagosomes with high autophagy flux. Yet, impairment of autophagy flux due to blocking of autophagosome-lysosome fusion and failure of autophagosome maturation can also lead to an increase in LC3-II. Therefore, mild increase following DBeQ-treatment or VCP depletion in I(+) HCN cells can be due to either increase or impairment in autophagy flux. If an increase of LC3-II observed in VCP-depleted HCN cells in I(+) was due to high rate of autophagy flux, BafA1 treatment would lead to higher amount of LC3-II than BafA1 I(+) control cells. However, the amount of LC3-II was the same, suggesting that the overall autophagy flux rate remained the same regardless of VCP inactivation during the basal autophagy in I(+). Therefore, this modest increase in LC3-II by VCP depletion can be interpreted due to partial inhibition of autophagosome maturation with the yield of the same amount of LC3-II accumulation following full inhibition of autophagy flux by BafA1. On the other hand, complete blocking of autophagy flux by BafA1 disclosed considerably smaller amount of LC3-II in VCP depleted I(-) HCN cells than I(-) control, indicating reduced rate of autophagy flux. Therefore, a decrease in LC3-II by DBeQ or VCP siRNA before BafA1 treatment imply the similar decrease of autophagy flux rate in I(-) HCN cells. Quantitative analysis of LC3 mRNA level did not show significant change by VCP inactivation (Fig. 4e). Taken together, these data demonstrate that VCP positively regulates autophagosome maturation at basal autophagy in I(+), as reported by others. In contrast, VCP regulates overall autophagy flux rate in insulin-deprived HCN cells when autophagy activity is high.

Modest changes in the level of LC3-II may not be detected due to saturation of Western blotting signals. To further validate different roles of VCP for autophagy between I(+) and I(-) conditions, next, we adopted monomeric RFP-GFP tandem fluorescent LC3 (mRFP-GFP-LC3) and measured autophagy flux. Co-existence of GFP and RFP signals in autophagosomes before maturation and RFP signals in autolysosomes after maturation are based on the acid-sensitive quenching of EGFP protein. DBeQ treatment in I(+) increased the percentage of yellow puncta, but without significant increase in the total puncta number, suggesting attenuation of autophagosome maturation by DBeQ in I(+) HCN cells. On the other hand, LC3 puncta formation was strikingly increased in I(-) HCN cells and this increase was greatly reduced by DBeQ. However, the percentages of yellow puncta remained similar between I(-) and I(-)/DBeQ conditions, indicating that DBeQ inhibited autophagy induction rather than maturation step in I(-) HCN cells (Fig. 5).

Based on our results indicating the reduced autophagy level by VCP inactivation in I(-) HCN cells, we postulated that VCP may regulate autophagy initiation signaling in HCN cells following insulin withdrawal. Our previous study by Ha et al. demonstrated that glycogen synthase kinase 3β (GSK-3β) induced ACD in HCN cells following insulin withdrawal [15]. Both pharmacological and genetic inactivation of GSK-3β significantly decreased ACD, while activation of GSK-3β increased autophagic flux and caused more cell death without inducing apoptosis following insulin withdrawal. This report uncovered GSK-3β as one of the critical upstream signaling kinases for ACD in I(-) HCN cells. GSK-3β activation, as revealed by a decrease in the inhibitory phosphorylation of serine nine following insulin withdrawal, was significantly prevented by both pharmacological and genetic inactivation of VCP (Fig. 6a–b). Mammalian target of rapamycin (mTOR) is known for a nutrient sensor and negative regulator of autophagy [50]. mTOR activation in relation to autophagy suppression is reflected by various phosphorylation sites including serine 2448 in response to insulin or nutrient stimulation [51, 52]. S2448 of mTOR was dephosphorylated in I(-) HCN cells. However, its phosphorylation level was rescued by VCP inaction surprisingly even in the absence of insulin (Fig. 6a–b). In our previous report, we demonstrated that calpain 2 was degraded by UPS in I(-) HCN cells, but conditions leading to calpain accumulation and high calpain activity utterly switched the default ACD to apoptosis in I(-) HCN cells [14]. In accordance with the induction of apoptosis in VCP-inactivated I(-) HCN cells, calpain 2 level was also restored to much higher level than I(-) (Fig. 6a). Knockdown of VCP in I(+) induced mild changes in the phosphorylation level of mTOR (data not shown). Compared with dramatic recovery of mTOR phosphorylation by VCP knockdown in I(-), these data indicate that deficit in autophagy induction in I(-) HCN cells depleted of VCP is like a direct result of VCP inactivation specifically in I(-) condition.

Double FYVE-domain-containing protein 1 (DFCP1) interacts with phosphatidylinositol 3-phosphate (PtdIns(3)p), which is required for the early step of autophagosome formation. Therefore, DFCP1 is regarded as an autophagy initiation marker [53]. A transient transfection of DFCP1-GFP construct gave the high number of DFCP1 puncta in I(-) HCN cells, compared with the negligible number in I(+). However, DBeQ treatment strongly suppressed the formation of DFCP1 puncta, again suggesting that VCP activity is required for autophagy initiation in I (-) HCN cells (Fig. 6c). This novel role of VCP in autophagy initiation has not been reported before. Taken together, these data suggest that VCP contributes to the maturation of autophagosomes to autolysosomes at basal autophagy process under I(+) condition. However, when autophagy flux is intensified upon insulin withdrawal, VCP is critically involved in the initiation stage of autophagy, as well exemplified in insulin-deprived HCN cells, which undergoes high autophagy level and cell death.

The following antibodies and reagents were used: VCP antibody was purchased from Cell signaling Technology (Danvers, MA, USA) or Abcam (Cambridge, United Kingdom). β-actin, calpain2, GSK-3β (S9), mTOR (S2448) and cleaved caspase-3 antibodies were purchased from Cell Signaling Technology; LC3B antibody from Sigma-Aldrich (St. Louis, MO, USA); MAP2 antibody from Abcam; GFAP and Vimentin antibodies from Millipore (Billerica, MA, USA). DBeQ, staurosporine, lactacystin and BafA1 were purchased from Sigma-Aldrich, Z-VAD-FMK from R&D Systems (Minneapolis, MN, USA), necrostatin-1 from Enzo Life Sciences (Farmingdale, NY, USA). They were diluted in dimethyl sulfoxide at appropriate concentrations.

HCN cells were grown as described previously by Chung et al (2015) [14].

HCN cells were seeded a cell density of 1.5 × 10⁵ cells/mL. HCN cells were suspended in 100 μl of Nucleofector Kit solution and were transfected with siRNA specific for rat VCP (Dharmacon, Lafayette, CO, USA) by using a Nucleofector Kit (Lonza, Basel, Switzerland) according to the manufacturer’s instructions with minor modification. After the nucleofection, the cells were incubated for 1 day and re-seeded in plates according to the experimental designs. pMXs-puro GFP-DFCP1 for mouse was purchased from Addgene (Cambridge, MA, USA). pMXs-puro GFP-DFCP1 and mRFP-GFP-LC3 constructs were transfected by Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. Transfection was performed in culture medium without penicillin/streptomycin for 4 h. After, HCN cells were incubated with culture medium.

Cell death was measured in a 96-well plate at a cell density of 1 × 10⁵ cells/mL. HCN cells were stained with diluted Hoechst 33342 (Invitrogen, Carlsbad, CA, USA) and propidium iodide (PI; Sigma-Aldrich) (1 % volume of media in the well, final 1/2000 and 1/1000 dilution) for 20 min at 37 °C. Cell counting was performed under a fluorescence microscope and was analyzed using Pixcavator IA.

HCN cells were harvested and lysed in radioimmunoprecipitation assay buffer (RIPA buffer, Sigma-Aldrich) with 1× protease cocktail inhibitors (Thermo Scientific, Waltham, MA, USA) and 1× phosphatase cocktail inhibitors (Thermo Scientific), 1 mM dithiothreitol and 1 mM phenylmethylsulfonyl fluoride for 30 min on ice. After centrifugation (16,100 × g, 10 min), protein concentrations were measured using the BCA protein assay reagent (Thermo Scientific). Proteins were loaded into the gel and electrotransfered to polyvinylidene difluoride (PVDF) membrane with a semi-dry electrophoretic transfer cell (Bio-Rad, Richmond, CA). Membranes were blocked for 1 h at room temperature in a blocking solution consisting of 5 % nonfat dry milk, 0.1 % Tween 20, and Tris-buffered saline (TBST). The membranes were then incubated overnight with the primary antibodies diluted in the blocking solution and washed three times for 10 min. The membranes were incubated for 1 h at room temperature with peroxidase-conjugated secondary antibodies diluted in blocking solution. After washing, protein expression was detected by using a chemiluminescence detection kit (Thermo Scientific).

All procedures for the care and use of laboratory animals were approved by the Institutional Animal Care and Use Committee (IACUC) at Daegu Gyeongbuk Institute of Science and Technology (DGIST). The 8-week-old rats were perfused with 1× PBS and 4 % PFA. After dissection, the post-fixation was performed with 4 % PFA for 12 h. The brain was immersed in 30 % sucrose in PBS at 4 °C until it sank. The brain was embedded in OCT compound and frozen on dry ice. The brain was cut into 40 μm serial fee-floating coronal sections. The brain sections were washed with 1× PBS three times for 10 min. Blocking was performed for 1 h into 5 % normal donkey serum (NDS) (Jackson ImmunoResearch Laboratories Inc., West Grove, PA, USA) with 1 % BSA and 0.3 % Triton X-100. The brain sections were incubated with primary antibody diluted in 5 % NDS and 0.3 % Triton X-100 for 18 h at 4 °C. After washing with 1× PBS, the brain sections were incubated with secondary antibody diluted in 5 % NDS and 0.3 % Triton X-100 for 2 h at room temperature in dark. Hoechst 33342 (Invitrogen) (1/10000 dilution) was stained for 10 min at room temperature. After washing, the brain sections were mounted on slide glass. The imaging was processed under laser scanning microscopes 700 (Zeiss, Oberkochen, Germany).

Total RNA was isolated using the ImProm-II Reverse Transcriptase kit (Promega, Madison, WI, USA) and cDNA was synthesized. qRT-PCR was performed with the CFX96 Real-Time System (Bio-Rad) and TOPreal™ qPCR 2× PreMIX (SYBR Green with low ROX) (Enzynomics, Daejeon, Republic of Korea). β-actin was used as the reference gene for normalization. Primers used for VCP, LC3 and β-actin were as follows: VCP forward, 5′- AGAGCAACCTTCGTAAAGCC-3′, and reverse, 5′-AACAACTGAGACACGATGCG-3′; LC3 forward 5′- AGTGGAAGATGTCCGGCTCA-3′, and reverse, 5′- GCTTCTCACCCTTGTATCGCT-3′; β-actin forward, 5′- GTCCACCCGCGAGTACAACCTT-3′, and reverse, 5′- TTGCACATGCCGGAGCCGTT-3′.

Cells were trypsinized and harvested in phosphate-buffered saline (PBS). Centrifuged cells were washed in PBS and re-suspended with 100 μl of 1× Annexin V binding buffer (10 mM HEPES, pH 7.4), 140 mM NaCl, 2.5 mM CaCl₂). For labeling, Annexin V-FITC (BD Biosciences, San Jose, CA, USA) was added and incubated for 15 min at room temperature. After the incubation, 400 μl of Annexin V binding buffer was added to dilute the sample. Samples were kept in ice, protected from light, and used for the analysis within an hour. Analysis was taken by flow cytometry using the Accuri C6 Flow Cytometer System (BD Biosciences) according to the manufacturer’s instructions.

Pregnant rats were purchased from Koatech (Namyangju-si, Gyeonggi-Do, Republic of Korea) and primary rat hippocampal neurons were prepared and cultured from rat embryos at day 17 (E-17). In brief, E-17 hippocampi were dissected and dissociated in a serum-free environment, and maintained in a serum-containing medium. Medium consists of Neurobasal medium (Gibco) supplemented with 100 U/ml penicillin (HyClone, Logan, UT, USA), 100 μg/ml streptomycin (HyClone), 2 mM L-glutamine (Sigma-Aldrich), and 2 % B27 supplement (Gibco). Three to four days after plating, medium was changed by one-third of the volume per well with fresh medium, which contained cytosine β-D-arabinofuranoside (Sigma-Aldrich) to a final concentration of 2 μM. The culture continued for up to 2 to 3 weeks until used for the experiment with change of medium by one-third of the volume every 3–4 days.

Organotypic hippocampal slices were prepared according to the method described by Simoni and Yu [59] slight modifications due to experimental designs. Adult female Sprague Dawley rats (7–8 weeks of age from Koatech) were decapitated following all institutional and ethical regulations regarding animal handling. Brains were rapidly removed and placed into a ice-cold slicing medium composed of 0.02 mM HEPES in Earle’s Balanced Salt Solution (EBSS; Welgene, Republic of Korea). Hippocampi were dissected out in fresh chilled slicing medium. Extracted hippocampi were then placed on the Teflon stage of a manual tissue slice chopper (Leica, Wetzlar, Germany) for coronal sectioning at 350 μm. Four hippocampal slices were seeded onto each Millicell culture plate insert (Millipore) in 6-well plates. After a 2-day stabilization period, hippocampal slices were deprived of insulin for 48 h prior to further experimental use.

All data from at least three independent experiments were analyzed as mean ± standard deviation (SD). Statistical significance was determined by the unpaired Student’s t-test or one-way analysis of variance (ANOVA) by Tukey’s multiple comparison tests using Graphpad Prism (GraphPad Software, San Diego, CA, USA).