NM-induced keratinocyte cell death
Histopathology of skin lesions reveals that NM targets basal epidermal keratinocytes in skin tissue and causes cell death.21 Therefore, we used human HaCaT keratinocytes as an in vitro NM exposure model. We found that treatment with different concentrations (0.1, 1, 5, 10, 20, 50, 100, and 200 μM) of NM for 24 h significantly inhibited cell viability in a dose-dependent manner, and that the LC50 of NM was 33.94 μM in keratinocytes (Fig. 1a). As shown in Fig. 1b, NM robustly changed the morphology of keratinocytes with irregular cell shape, weakened cell membrane refractive index; cell debris and dead cells were visible. Moreover, with PI staining, we also found that NM markedly increased keratinocyte cell death (Fig. 1c, d and Supplementary Fig. S1). These results suggested that NM dose-dependently caused keratinocyte cell death. Accordingly, 20 μM NM (resulting in ~40% decrease in cell viability) was used in subsequent experiments to clarify the underlying mechanism of NM-induced dermal toxicity.
NM-induced autophagy in keratinocytes
The ratio of light chain 3 beta 2 (LC3B2) to β-actin (ACTB) is an important indicator of autophagy.22 NM (≥10 μM) was found to increase the ratio of LC3B2 to ACTB compared to the control and decrease the level of sequestosome 1 (SQSTM1/p62) (Fig. 2a, b). 3-methyladenine (3-MA, an inhibitor of early autophagy stages22), markedly inhibited NM-induced LC3 conversion and p62 degradation (Fig. 2c, d). To monitor autophagic flux, LC3B2 levels were measured in the presence of chloroquine (CQ), which inhibits autophagosome-lysosome fusion.22 CQ resulted in a greater accumulation of LC3B2 in keratinocytes after incubation with NM (20 µM) for 24 h compared to that in cells treated with CQ alone (Fig. 2c, d). Moreover, tandem fluorescent red fluorescent protein (RFP)- green fluorescent protein (GFP)-LC3 staining is another useful tool for distinguishing autophagy pathway intermediates, which allows one to evaluate the extent of autophagosome and autolysosome formation simultaneously, because LC3 puncta labeled with both GFP and RFP represent autophagosomes, whereas those labeled with RFP alone represent autolysosomes.23 As shown in Fig. 2e, f, NM (20 µM) notably increased the numbers of GFP and RFP dots per cell. In the merged images, more free red dots than yellow dots were seen, indicating markedly induced autolysosome formation compared with autophagosomes, and suggesting that NM increased autophagic flux. Meanwhile, bafilomycin A1 (BafA1), which inhibits the acidification of organelles and, subsequently, autophagosome-lysosome fusion,24,25 challenge resulted in further accumulation of autophagosomes in NM-treated keratinocytes compared to cells treated with BafA1 alone (Fig. 2e, f). These results suggested that NM promoted cellular autophagic flux in keratinocytes.
To further characterize NM-induced autophagy in keratinocytes, cells were transfected with the GFP-LC3B plasmid, a specific marker of phagophores and autophagosomes, and visualized by fluorescence microscopy.22 NM (20 µM) notably increased the number of GFP-LC3B dots compared with the control group, which then decreased in the presence of 3-MA (Fig. 2g). Moreover, transmission electron microscopy, the most valid method for both qualitative and quantitative analysis of autophagy,22 showed that more vacuoles were present in NM-treated keratinocytes, but the number decreased when the cells were pretreated with 3-MA (Fig. 2h). These results suggested that NM treatment upregulated the entire autophagy process in keratinocytes.
The overactivation of autophagy contributed to NM-induced keratinocyte cell death
3-MA, CQ, and small interfering RNA (siRNA) targeting ATG5 were used to investigate the role of autophagy in NM-caused keratinocyte cell death. As shown in Fig. 3a–f and Supplementary Fig. S1, 3-MA and CQ as well as ATG5 siRNA transfection ameliorated NM-induced keratinocyte cell death. In addition, 3-MA, CQ and ATG5 siRNA had no obvious effects on keratinocyte cell viability (Fig. 3a–f and Supplementary Fig. S1). These results indicated that autophagy contributed to the cytotoxicity of NM in keratinocytes. In 2013, Levine et al.26 described “autosis” as a subtype of autophagy-dependent cell death, which is a Na+, K+-ATPase–regulated form of cell death. It has been demonstrated that autosis can be inhibited by digoxin (a Food and Drug Administration [FDA]-approved antagonist of Na+, K+-ATPase); meanwhile, BafA1 does not reduce autosis. Therefore, BafA1 and digoxin were further used to investigate the role of autosis in NM-induced keratinocyte cell death. As shown in Supplementary Fig. S2, BafA1 pretreatment significantly attenuated NM-induced cell death of keratinocytes; in contrast, digoxin had no effect on NM-caused keratinocyte cell death. These results indicated that NM-induced autophagic cell death of keratinocytes, which was not autosis.
NM-induced autophagy via the AMPK-ULK1 pathway in keratinocytes
Previous studies have demonstrated that AMPK is a key regulator of autophagy.16 Our results showed that NM resulted in a dose-dependent AMPK activation, as monitored by AMPK phosphorylation (Fig. 4a, b). Compound C (CC, a potent AMPK inhibitor) abolished NM-induced AMPK activation and caused a decrease in autophagy (Fig. 4c, d). Meanwhile, NM-stimulated autophagy was also diminished by AMPK siRNA in keratinocytes (Fig. 4e, f). In addition, CC and AMPK siRNA treatment notably inhibited NM-caused cell death of keratinocytes (Fig. 4g, h). Together, these results indicated that AMPK inhibition suppressed autophagy thereby attenuating NM-caused cell death of keratinocytes.
Since AMPK can activate autophagy through the activation of ULK1 or by inhibiting mTOR signaling,16 we next examined alterations in the AMPK/ULK1 and mTOR pathways induced by NM. As shown in Fig. 5a, b, the phosphorylation of ULK1 and mTOR was increased by NM in a dose-dependent manner. AMPK inhibition or knockdown significantly abolished NM-induced increase of pULK1 level, however, had no obvious effect on NM-induced pmTOR expression in keratinocytes (Fig. 5c–f). Moreover, inhibition of ULK1 by SBI-0206965 (SBI) significantly prevented NM-induced autophagy and restored cell viability following NM treatment (Supplementary Fig. S3a, d). Although the level of pmTOR was increased by NM, the mTOR inhibitor rapamycin (RAPA) had no obvious effect on NM-induced LC3 conversion or p62 degradation, indicating that NM-mediated autophagy activation occurred independently of mTOR regulation in keratinocytes (Fig. 5g, h). In addition, RAPA treatment had no effect on NM-caused cell death or ULK1 activation in keratinocytes (Fig. 5i, j and Supplementary Fig. S4a, b). Thus, we concluded that the AMPK-ULK1 pathway was required for NM-induced autophagy and keratinocyte cell death.
NM-activated AMPK via the Ca2+-CaMKKβ pathway in keratinocytes
In view of the finding that the Ca2+-CaMKKβ pathway plays an important role in the regulation of AMPK-mediated autophagy.27,28 Therefore, the effects of NM on Ca2+ influx and CaMKKβ activity were investigated. As shown in Fig. 6a, b, NM significantly increased the intracellular Ca2+ content. NM-induced a significant increase of pCaMKKβ expression in a dose-dependent manner (Fig. 6c, d), indicating the activation of CaMKKβ by NM. Moreover, STO-609 (a CaMKK-α/β inhibitor) treatment markedly inhibited NM-induced AMPK and ULK1 activation and autophagy induction, ultimately attenuating NM-induced cell death (Fig. 6e, f and Supplementary Fig. S3b, d). In addition, STO-609 had no obvious effect on NM-induced pmTOR expression in keratinocytes (Supplementary Fig. S3b). Meanwhile, CaMKK-β siRNA also abolished the activation of AMPK and the induction of autophagy in NM-stimulated keratinocytes (Fig. 6g, h). Additionally, it has been demonstrated that AMPK monitors the cellular energy status which is usually stimulated by inhibition of mitochondrial ATP synthesis.29 Accordingly, ATP levels were detected in NM-treated keratinocytes, we found that NM slightly but not significantly increased ATP levels (Supplementary Fig. S5a), indicating that NM-activated AMPK in an ATP independent manner in keratinocytes. These results suggested that the Ca2+-CaMKKβ pathway was required for AMPK activation and subsequent NM-induced autophagy in keratinocytes.
NM-induced Ca2+ influx in a TRPV1-dependent manner in keratinocytes
TRPV1 is a non-selective ion channel that regulates the influx of Ca2+ and is considered as a primary cellular sensor of thermal and chemical stimulation in the skin.13 Thus, the involvement of TRPV1 in the NM-induced increase of intracellular Ca2+ content was also explored. As shown in Fig. 7a, b, NM significantly induced TRPV1 expression. The TRPV1 antagonist capsazepine (CPZ) notably inhibited NM-stimulated Ca2+ influx and CaMKKβ activation, thereby, inhibiting AMPK and ULK1 activation; meanwhile, CPZ also inhibited NM-induced autophagy and cell death of keratinocytes (Fig. 7c–f and Supplementary Fig. S3c, d). When TRPV1 was knocked down by TRPV1 siRNA transfection, similar results were found (Fig. 7g–j). These data clearly indicated that TRPV1 was required for Ca2+ accumulation and the subsequent activation of the CaMKKβ-AMPK-autophagy signaling pathway induced by NM in keratinocytes.
As NM-stimulated Ca2+ influx, the effect of NM on lysosomal function and reactive oxygen species (ROS) generation was investigated. LysoSensor Green DND-189 dye is an acidotropic probe that accumulates in acidic organelles, such as lysosomes, and exhibits a pH-dependent increase in fluorescence intensity upon acidification.30 NM increased the LysoSensor Green DND-189 fluorescence intensity (Supplementary Fig. S5b), suggesting lysosomal pH alteration in keratinocytes. As shown in Supplementary Fig. S5c–e, NM induced the expression of the lysosomal-associated membrane protein 1 (LAMP1) and cathepsin D (CTSD) and increased the fluorescence intensity of cells stained by LysoTracker Green (LTG) fluorescent dye, which stains acidic compartments, particularly lysosomes.31 In addition, ROS levels were detected by dichloro-dihydro-fluorescein diacetate (DCFH-DA) staining, and we found that NM markedly increased ROS levels in keratinocytes (Supplementary Fig. S5f). These results suggested that NM both induced lysosomal activity and increased ROS levels in keratinocytes. However, CPZ pretreatment abolished NM-induced lysosomal activity and increase of ROS levels (Supplementary Fig. S5b–f). Together, these results, combined with the above data, collaboratively demonstrate that TRPV1 is a potential target for the treatment of NM-caused dermal toxicity.
NM-induced autophagy via activating the TRPV1 signaling pathway in the skin of SKH-1 hairless mice
We additionally investigated NM-induced dermal toxicity in skin in vivo. Epidermal thickness, microvesication (epidermal-dermal separation), and epidermal denudation have been considered as the primary injury end points in NM-exposed dorsal mouse skin.3 In accordance with previous observations, NM significantly increased skin epidermal thickness in post 24-h exposure and microvesication and epidermal denudation in skin post 72-h exposure, which were significantly attenuated by CQ treatment (Fig. 8a). Accumulating evidence has revealed that NM causes a strong inflammatory response in skin;32 therefore, the effect of CQ on NM-induced inflammation was determined in vivo. As expected, CQ significantly decreased NM-caused cyclooxgenase 2 (COX2) and matrix metalloproteinase 9 (MMP9) expression in skin post 72-h exposure (Supplementary Fig. S6a, b). CQ also inhibited NM-induced autophagy in skin (Fig. 8b–d). As shown in Fig. 8e–h, NM increased the levels of TRPV1, pCaMKKβ, pAMPK, pULK1, and pmTOR, indicating that the TRPV1 signaling pathway was activated in NM-exposed skin. Moreover, CPZ treatment abolished the effect of NM on the TRPV1 signaling pathway thereby attenuating NM-induced autophagy, which, subsequently, ameliorated NM-caused skin injury (Fig. 8a–j and Supplementary Fig. S6a, b). The results suggested that NM-induced autophagy via activating the TRPV1 signaling pathway, ultimately, causing cutaneous injury in vivo.