Niclosamide Triggers Non-Canonical LC3 Lipidation

Autophagy is a highly- evolutionarily-conserved catabolic pathway activated by various cellular stresses. Recently, non-canonical autophagy (NCA), which does not require all of the ATG proteins to form autophagosome or autophagosome-like structures, has been found in various conditions. Moreover, mounting evidence has indicated that non-canonical LC3 lipidation (NCLL) may reflect NCA. We and others have reported that niclosamide (Nic), an anti-helminthic drug approved by the Food and Drug Administration, could induce canonical autophagy via a feedback downregulation of mTOR complex 1. In this study, we found that Nic could also induce NCLL, which is independent of the ULK1 complex and Beclin 1 complex, but dependent on ubiquitin-like conjugation systems. Although bafilomycin A1 and concanamycin A, two known V-ATPase inhibitors, significantly inhibited Nic-induced NCLL, Nic-induced NCLL was demonstrated to be independent of V-ATPase. In addition, the Golgi complex and vimentin were involved in Nic-induced NCLL, which might be a platform or membrane source for Nic-induced LC3-positive structures. These results would be helpful to broaden our understanding of the working mechanisms of Nic and evaluate its pharmacological activities in diseases.


Introduction
Drug discovery and development are costly and lengthy processes. Therefore, drug repurposing has attracted increasing attention in recent years [1]. Drug repurposing means finding novel functions for existing drugs, which is more economical and faster than the canonical drug development strategy, as new uses of old drugs that have been well characterized and approved for human use could be evaluated rapidly in clinical trials [2].
Niclosamide (Nic), a salicylanilide derivative, is on the World Health Organization's list of essential medicines and has been approved by the Food and Drug Administration (FDA) as an anti-helminthic drug. Considerable studies have suggested that Nic is a multifunctional drug and has the potential to be a candidate for the treatment of various diseases [3]. One of its biological activities is to induce autophagy. We and others have demonstrated that Nic could induce canonical autophagy on ubiquitin-like conjugation systems. LC3 could also be an appropriate marker for the majority of NCA, and non-canonical LC3 lipidation (NCLL) may reflect NCA.
In this study, we are the first to discover that Nic could initiate NCLL, indicating that Nic may also induce NCA. This non-canonical process required integration of the Golgi complex and involved vimentin. We also found that bafilomycin A1 (Baf) could inhibit Nic-induced NCLL at an early stage in a V-ATPase-and Ca 2+ -independent manner. Determination of the precise mechanisms of NCLL induced by Nic from different angles could be helpful to evaluate its multiple pharmacological activities.

Immunoblot Assay
For the immunoblot assay, 20-30 µg of whole cell lysates were prepared for Western blot analysis. Samples were run on 10% or 12% SDS-PAGE gels and transferred to PVDF membranes (Millipore). After blocking and incubation with primary and secondary antibodies, specific proteins were detected using enhanced chemiluminescence Western blot agent (Millipore, Billerica, MA, USA). Images were taken using Kodak Image Station 4000 (Carestream Health, Rochester, NY, USA) or ImageQuant LAS 4000mini (GE, Uppsala, Sweden) with the companion software.

Immunostaining Assay
Cells were grown on 96-well plates and fixed in 4% paraformaldehyde for 15 min. Cells were washed twice in phosphate-buffered saline (PBS) followed by permeabilization with 0.1% Triton X-100 for 15 min and blocked in PBS containing 5% bovine serum albumin for 1 h. Primary antibodies were used at 4 • C overnight, followed by secondary antibodies for 1 h at room temperature. Fluorescence images were taken using EVOS FL Auto (Life Technologies, Bothell, WA, USA). For manual quantification of the puncta formation, at least 3 optical fields with at least 50 cells per experimental condition were analyzed. Data from repeated experiments were subjected to statistical analysis.

Molecule Capture
To measure the interaction between immobilized Nic and flowing cell lysates, the PlexArray HT System (Plexera LLC, Woodinville, WA, USA) was used to monitor the whole procedure in real time. The optical architecture and operation details of the PlexArray HT System were as described previously [24]. Cell lysates were prepared in phosphate-buffered saline with Tween-20 (PBST, pH 7.4), and glycine-HCl (pH 2.0) was used as the regeneration buffer. A typical binding curve was obtained by flowing the cell lysate sample at 2 µL/s for 300 s of association and then using PBST to wash the biochip 3 times at 2 µL/s for 300 s. Then, we used trypsin (Promega, Madison, WI, USA) to digest the proteins binding on the chip at 37 • C for 12 h. The peptide was separated by the Nano Acquity UPLC System (Waters, Milford, MA, USA) and identified by Q Exactive (Thermo Scientific, Waltham, MA, USA).

Statistical Analysis
The results shown are representative of at least three independent experiments. The data are presented as the mean ± standard deviation. Statistical significance was assessed by Student's t-test. p < 0.05 was considered significant.

Nic-Induced NCLL Is Independent of ULK1 Complex and Beclin 1 Complex, But Dependent on Ubiquitin-Like Conjugation Systems
The ULK1 complex and Beclin 1 complex are essential for the early stage of autophagy. Unexpectedly, we found that Nic could still trigger LC3 lipidation in FIP200KO-MEFs, ULK1KO-MEFs, and Beclin 1KD-U251, cells as well as in wild-type MEFs ( Figure 1A,B and Supplementary Figure S1G), whereas EBSS-induced canonical LC3 lipidation was abolished in these knockout/knockdown cells ( Figure 1B and Supplementary Figure S1E,F). Consistent with the LC3-II formation, the number of GFP-LC3-positive structures was significantly increased with Nic treatment ( Figure 1C,D). However, the non-canonical LC3 lipidation and GFP-LC3 puncta formation induced by Nic were abolished in ATG5KO-MEFs ( Figure 1C-E). Moreover, we checked the role of ATG4B in Nic-initiated NCLL. In canonical autophagy, ATG4B is responsible for the LC3 conversion. In line with ATG5KO-MEFs, there was no LC3-II formation following Nic treatment in ATG4BKO-HeLa cells ( Figure 1F). Taken together, these data suggested that Nic could induce NCLL. In Nic-induced NCLL, the ULK1 complex and Beclin 1 complex were dispensable, whereas the ubiquitin-like conjugation systems were imperative.

Nic Recruits the ATG12-ATG5-ATG16L1 Complex via a WIPI2-Independent Pathway
In canonical autophagy, PI3P, a product of PI3KC3, is required for autophagosome biogenesis and maturation [25]. The generation of PI3P on the phagophore facilitates the recruitment of two ubiquitin-like conjugation systems. It has been reported that this process is meditated by WIPI2, a PI3P effector protein. WIPI2 could directly bind to ATG16L1 and act immediately upstream of ATG16L1 [26]. Meanwhile, ATG16L1 binds to ATG15-ATG12 to form the ATG12-ATG5-ATG16L1 complex, which serves as an E3-like enzyme to generate LC3-II. Consequently, we evaluated the formation of WIPI2, ATG16L1, and ATG12 puncta after Nic stimuli. In wild-type MEFs, WIPI2, ATG16L1, and ATG12 puncta were all induced by Nic, but only ATG16L1 and ATG12 dots were detected in FIP200KO-MEFs (Figure 2A-C and Supplementary Figure S2A-C). Subsequently, we analyzed the colocalization of ATG12 and ATG16L1 puncta and the conjugation of ATG5 and ATG12. ATG12-and ATG16L1-positive structures were almost completely colocalized ( Figure 2D and Supplementary Figure S2D), and ATG5-ATG12 conjugation was unaffected ( Figure 2E) with Nic treatment. Moreover, we found that Nic-induced ATG16L1-and ATG12-positive puncta were also colocalized with Nic-induced LC3-positive structures ( Figure 2F,G and Supplementary Figure S2E,F). These data indicated that the recruitment of ATG12-ATG5-ATG16L1 complex induced by Nic was independent of WIPI2.

Nic Recruits the ATG12-ATG5-ATG16L1 Complex via a WIPI2-Independent Pathway
In canonical autophagy, PI3P, a product of PI3KC3, is required for autophagosome biogenesis and maturation [25]. The generation of PI3P on the phagophore facilitates the recruitment of two ubiquitin-like conjugation systems. It has been reported that this process is meditated by WIPI2, a PI3P effector protein. WIPI2 could directly bind to ATG16L1 and act immediately upstream of ATG16L1 [26]. Meanwhile, ATG16L1 binds to ATG15-ATG12 to form the ATG12-ATG5-ATG16L1 complex, which serves as an E3-like enzyme to generate LC3-II. Consequently, we evaluated the formation of WIPI2, ATG16L1, and ATG12 puncta after Nic stimuli. In wild-type MEFs, WIPI2, ATG16L1, and ATG12 puncta were all induced by Nic, but only ATG16L1 and ATG12 dots were detected in FIP200KO-MEFs ( Figure  S2E,F). These data indicated that the recruitment of ATG12-ATG5-ATG16L1 complex induced by Nic was independent of WIPI2.

Baf Is Able to Inhibit Nic-Induced NCLL
Bafilomycin A1 (Baf), chloroquine (CQ), and ammonia chloride (AC) have been commonly used as inhibitors of the late stage of canonical autophagy [27]. However, we found that only Baf could inhibit the Nic-induced LC3-II formation and GFP-LC3 punctation in FIP200KO-MEFs, whereas CQ and AC had no extra effect on Nic-activated NCLL ( Figure 3A,B). Similar to Baf, concanamycin (ConA), another well-known vacuolar-type H + -ATPase (V-ATPase) inhibitor, could also attenuate Nic-induced LC3 lipidation ( Figure 3A). In wild-type MEFs, Baf alone could already decrease Nic-induced LC3-II formation; while both Baf and 3MA were present, LC3 lipidation was almost completely abolished ( Figure 3C,D and Supplementary Figure S3A,B). This suggested that NCLL, which was inhibited by Baf, and canonical autophagy, which was blocked by 3MA, coexisted in WT-MEFs with Nic treatment. In addition, in FIP200KO-MEFs, Baf could eliminate the ATG16L1 and ATG12 puncta triggered by Nic without altering the expression level of ATG16L1 and ATG5-ATG12 conjugation ( Figure 3E,F and Supplementary Figure S3C), indicating that Baf might directly interfere with ATG16L1 or its upstream signal. Taken together, Baf could be an early phase inhibitor of Nic-induced NCLL.

Baf Is Able to Inhibit Nic-Induced NCLL
Bafilomycin A1 (Baf), chloroquine (CQ), and ammonia chloride (AC) have been commonly used as inhibitors of the late stage of canonical autophagy [27]. However, we found that only Baf could inhibit the Nic-induced LC3-II formation and GFP-LC3 punctation in FIP200KO-MEFs, whereas CQ and AC had no extra effect on Nic-activated NCLL ( Figure 3A,B). Similar to Baf, concanamycin (ConA), another well-known vacuolar-type H + -ATPase (V-ATPase) inhibitor, could also attenuate Nic-induced LC3 lipidation ( Figure 3A). In wild-type MEFs, Baf alone could already decrease Nic-induced LC3-II formation; while both Baf and 3MA were present, LC3 lipidation was almost completely abolished ( Figure 3C,D and Supplementary Figure S3A,B). This suggested that NCLL, which was inhibited by Baf, and canonical autophagy, which was blocked by 3MA, coexisted in WT-MEFs with Nic treatment. In addition, in FIP200KO-MEFs, Baf could eliminate the ATG16L1 and ATG12 puncta triggered by Nic without altering the expression level of ATG16L1 and ATG5-ATG12 conjugation ( Figure 3E,F and Supplementary Figure S3C), indicating that Baf might directly interfere with ATG16L1 or its upstream signal. Taken together, Baf could be an early phase inhibitor of Nic-induced NCLL.

Nic-Induced NCLL Is Independent of V-ATPase and Ca 2+
We have shown that Baf and ConA, two V-ATPase inhibitors, could block Nic-induced NCLL. V-ATPase consists of two domains, V0 and V1, and contains at least 13 subunits. To investigate the role of V-ATPase in NCLL, we knocked down three subunits of V-ATPase, ATP6V1A, ATP6V1D, and ATP6V0c, respectively. However, in HeLa cells, Nic-induced LC3-II formation was unaffected by the knockdown of ATP6V1A or ATP6V1D ( Figure 4A,B), even in the presence of 3MA ( Figure 4D,E). ATP6V0c was the reported target of Baf and ConA, but neither knocking it down, nor overexpressing it could change the LC3-II level upregulated by Nic ( Figure 4C,F). These results indicated that V-ATPase might not be required for Nic-induced NCLL.  Interestingly, Baf has also been observed to be a moderately good inhibitor of P-ATPase [28], and SERCA, a type of P-ATPase, shows an intermediate sensitivity to Baf [29]. Thapsigargin (TG), a plant-derived sesquiterpene lactone, has been identified as a specific inhibitor of SERCA [30]. It was used widely as an ER stressor to induce autophagy, but recently shown in fact to block autophagy [31]. However, we found that TG could not inhibit Nic-triggered NCLL ( Figure 4G). Subsequently, since both V-ATPase and SERCA have effects on the uptake and/or efflux of Ca 2+ , we applied two chemicals to change the intracellular calcium concentration to investigate the role of Ca 2+ in Nic-induced NCLL. The Western blot results showed that both ionomycin, an ionophore that could (siATP6V1A) or ATP6V1D (siATP6V1D) was transfected to WT-HeLa cells, followed by treatment with Nic (10 µM) for 6 h, and then, the LC3-II formation was analyzed. © The siRNA of ATP6V0c (siATP6V0c) was transfected to FIP200KO-MEFs, followed by treatment with Nic (10 µM) for 6 h, and then, the LC3-II formation was evaluated. (D,E) The siRNA of ATP6V1A (siATP6V1A) or ATP6V1D (siATP6V1D) was transfected to WT-HeLa cells, followed by treatment with Nic (10 µM) in the presence or absence of 3MA (10 mM) for 6 h, and then, the LC3-II formation was analyzed. (C,F) The Flag-ATP6V0c plasmid was transfected to 293A cells, followed by treatment with Nic (10 µM) in the presence or absence of 3MA (10 mM) for 6 h, and then, the LC3-II formation was analyzed. (G-I) FIP200KO-MEF were treated by Nic with or without TG (3 µM), ionomycin (Iono, 2 µM), or BAPTA (10 µM) for 6 h, and then, LC3-II formation was analyzed by immunoblotting.
Interestingly, Baf has also been observed to be a moderately good inhibitor of P-ATPase [28], and SERCA, a type of P-ATPase, shows an intermediate sensitivity to Baf [29]. Thapsigargin (TG), a plant-derived sesquiterpene lactone, has been identified as a specific inhibitor of SERCA [30]. It was used widely as an ER stressor to induce autophagy, but recently shown in fact to block autophagy [31]. However, we found that TG could not inhibit Nic-triggered NCLL ( Figure 4G). Subsequently, since both V-ATPase and SERCA have effects on the uptake and/or efflux of Ca 2+ , we applied two chemicals to change the intracellular calcium concentration to investigate the role of Ca 2+ in Nic-induced NCLL. The Western blot results showed that both ionomycin, an ionophore that could raise the intracellular level of calcium, and BAPTA, a chelator that could reduce the intracellular level of calcium, were not able to block Nic-induced non-canonical LC3-II formation ( Figure 4H,I). These data demonstrated that Ca 2+ was dispensable for Nic to trigger NCLL.

Nic-Induced NCLL Requires Intact Golgi Complex
It has been reported that the Golgi complex is vital for several types of NCA [23,32]. To confirm whether the Golgi apparatus is required for Nic-induced NCLL, brefeldin A (BFA) and golgicide A (GCA), two well-studied Golgi toxins that cause Golgi complex collapse, were used. Although both BFA and GCA were demonstrated to trigger canonical autophagy, they were all able to reduce the formation of LC3-II and GFP-LC3 puncta induced by Nic ( Figure 5A-D). Furthermore, Nic-induced LC3 puncta overlapped with the cis-Golgi complex marker, GM130 ( Figure 5E). Unexpectedly, the expression level of GM130 was significantly decreased by Nic, while TGN38 was not affected, and Baf could not reverse it ( Figure 5F). These results suggested that integrated Golgi was required for Nic-induced NCLL, and the Golgi apparatus, especially the cis-Golgi complex, may contribute to autophagosome formation following Nic treatment.

Nic-Induced NCLL Requires Intact Golgi Complex
It has been reported that the Golgi complex is vital for several types of NCA [23,32]. To confirm whether the Golgi apparatus is required for Nic-induced NCLL, brefeldin A (BFA) and golgicide A (GCA), two well-studied Golgi toxins that cause Golgi complex collapse, were used. Although both BFA and GCA were demonstrated to trigger canonical autophagy, they were all able to reduce the formation of LC3-II and GFP-LC3 puncta induced by Nic ( Figure 5A-D). Furthermore, Nic-induced LC3 puncta overlapped with the cis-Golgi complex marker, GM130 ( Figure 5E). Unexpectedly, the expression level of GM130 was significantly decreased by Nic, while TGN38 was not affected, and Baf could not reverse it ( Figure 5F). These results suggested that integrated Golgi was required for Nic-induced NCLL, and the Golgi apparatus, especially the cis-Golgi complex, may contribute to autophagosome formation following Nic treatment.

Vimentin Is Involved in Nic-Induced NCLL
Recently, CQ, a conventional inhibitor of autophagy flux, was found to induce non-canonical LC3 lipidation [33], and this process is regulated by osmotic imbalances. Since both Nic and CQ could induce canonical autophagy through mTORC1 suppression and were all able to induce NCLL,

Vimentin Is Involved in Nic-Induced NCLL
Recently, CQ, a conventional inhibitor of autophagy flux, was found to induce non-canonical LC3 lipidation [33], and this process is regulated by osmotic imbalances. Since both Nic and CQ could induce canonical autophagy through mTORC1 suppression and were all able to induce NCLL, we also explored the role of osmotic imbalances and water flux in the non-canonical LC3 lipidation triggered by Nic. Two mechanistically-distinct aquaporin inhibitors, phloretin and mercury chloride, were used to inhibit water influx. It was observed that both phloretin and mercury chloride could partially inhibit NCLL induced by Nic (Supplementary Figure S4), but the inhibitory effect was not as significant as Baf. In order to find a more precise target protein, we utilized "molecule capture" and got more than 90 protein candidates ( Figure 6A and Supplementary Table S1). Among these candidates, vimentin got the highest score. Currently it is the only one we have found that could influence Nic-induced NCLL. We found that knocking vimentin down could significantly inhibit Nic-induced NCLL ( Figure 6B-D), as well as AMDE1-and CCCP-induced NCLL, two known NCLL [23,34] (Supplementary Figure S5B,C), while the mRNA level of LC3B was not affected by knocking vimentin down (Supplementary Figure S5A). In addition, neither Nic nor Baf could influence the expression level of vimentin ( Figure 6E), indicating that the interaction between vimentin and other proteins may play a critical role in Nic-induced NCLL. Otherwise, since vimentin is a structural protein, it is also possible that vimentin acts as a platform for recruiting NCLL-required proteins. these candidates, vimentin got the highest score. Currently it is the only one we have found that could influence Nic-induced NCLL. We found that knocking vimentin down could significantly inhibit Nic-induced NCLL ( Figure 6B-D), as well as AMDE1-and CCCP-induced NCLL, two known NCLL [23,34] (Supplementary Figure S5B,C), while the mRNA level of LC3B was not affected by knocking vimentin down (Supplementary Figure S5A). In addition, neither Nic nor Baf could influence the expression level of vimentin ( Figure 6E), indicating that the interaction between vimentin and other proteins may play a critical role in Nic-induced NCLL. Otherwise, since repurposiis a structural protein, it is also possible that vimentin acts as a platform for recruiting NCLL-required proteins.

Discussion
Nic is an FDA-approved anti-helminthic drug. It has been used in millions of people to treat tapeworm infection for nearly 40 years, whereas its mechanism of action has not been well elucidated. Mounting evidence has demonstrated that Nic modulates diverse signaling pathways and biological processes, including uncoupling of oxidative phosphorylation, autophagy, and the Wnt/β-catenin signaling pathway [3], indicating that beyond application in parasitic infection

Discussion
Nic is an FDA-approved anti-helminthic drug. It has been used in millions of people to treat tapeworm infection for nearly 40 years, whereas its mechanism of action has not been well elucidated. Mounting evidence has demonstrated that Nic modulates diverse signaling pathways and biological processes, including uncoupling of oxidative phosphorylation, autophagy, and the Wnt/β-catenin signaling pathway [3], indicating that beyond application in parasitic infection treatment, it can also be exploited in other diseases. Here, we show a new function of Nic to help enlarge its application fields.
Autophagy dysfunction has been linked to a wide range of human diseases, and considerable enthusiasm has emerged to discover and develop autophagy inducers for the prevention and treatment of diseases [13]. However, this treatment strategy has certain limitations. The mutations or polymorphisms of ATG genes would possibly suppress the autophagy induced by candidate compounds. Therefore, NCA may possess a much wider prospect of application as a treatment target than canonical autophagy. An interesting point is how to find novel modulators of NCA. Although LC3 lipidation is not observed in ATG5/ATG7-independent alternative autophagy, in most types of NCA, functional autophagosomes or autophagosome-like structures are still decorated by LC3. Thus, NCA might represent LC3-positive structures and could be detected by the LC3 lipidation assay. In this study, we found that Nic might initiate NCA via evaluating the expression level of LC3-II and the number of GFP-LC3 puncta.
The ULK1 complex and Beclin 1 complex are involved in the initiation of canonical autophagy. However, in the absence of ULK1 or Beclin 1, Nic could still induce LC3 lipidation ( Figure 1A-C), indicating that Nic-induced NCLL could be initiated in a ULK1-and Beclin 1-independent manner. These observations suggested that Nic could be used to treat diseases when the canonical initiation systems are compromised. Moreover, considerable enthusiasm has emerged for using Nic to treat cancer [3]. It has been demonstrated that Nic could treat more than 11 types of cancers in cell and animal models. Currently, on ClinicalTrial.gov, four clinical trials of Nic are recruiting participants for prostate cancer and colorectal cancer treatment. It is known that Beclin 1-independent NCA is induced by many proapoptotic compounds [14]. Nic is also able to trigger apoptosis, and a recent study reported that Nic-induced Wnt signaling inhibition in colorectal cancer is mediated by autophagy [6]; thus, there is a possibility that Nic-induced Beclin 1-independent NCA is involved in the therapy process and results in an autophagic cell death.
In canonical autophagy, PI3P, a product of PI3KC3, is required for autophagosome biogenesis [25]. However, recently, VPS34-independent NCA has been reported. In VPS34 −/− T lymphocytes and sensory neurons, LC3 lipidation and autophagosomes have been observed [35,36]. However, WIPI2 puncta, an effector of PI3P, was only found in WT-MEFs, but not observed in FIP200KO-MEFs with Nic treatment (Figure 2A and Supplementary Figure S2A). Thus, PI3P may not be required for Nic-induced NCLL. The inhibitory effect of 3MA ( Figure 3C,D and Supplementary Figure S3A,B) was due to its function in Nic-induced canonical autophagy, as canonical autophagy and NCLL coexisted in WT-MEF with Nic treatment.
ATG16L1 has several functional domains that play an essential role in autophagy. The N-terminal of ATG16L1 is responsible for interacting with ATG5-ATG12 conjugation to form the ATG12-ATG5-ATG16L1 complex, which performs as an E3-like enzyme to generate LC3-II [37]. The central region of ATG16L1 could bind to FIP200 and WIPI2, and this region is indispensable for starvation-activated canonical autophagy, but is not required for glucose deprivation-induced ULK1-independent NCA [26,38]. The C-terminal WD40 domain of ATG16L1 was recently reported to be used to distinguish canonical and non-canonical autophagy [39]. In this study, although we did not investigate which domain of ATG16L1 is required for Nic-induced NCLL, we found that in the absence of FIP200, Nic could also recruit ATG16L1 via a WIPI2-independent way (Figure 2A,B and Supplementary Figure S2A,B). In line with the previous study, we also found that Baf, a V-ATPase inhibitor, could abolish the ATG16L1 dots' formation and LC3 lipidation triggered by Nic ( Figure 3E), suggesting Baf could inhibit Nic-induced NCLL at the early stage. Thus, we predicted that V-ATPase activity should be required for Nic-initiated non-canonical LC3 lipidation. However, neither knockdown of ATP6V1A, ATP6V1D, or ATP6V0c, nor overexpression of ATP6V0c could inhibit Nic-induced NCLL ( Figure 4A-F), indicating that V-ATPase may not be required, and the exact targets of Baf need further studies.
Various membrane structures, the Golgi complex, ER, mitochondria, and plasma membrane, could contribute to autophagosomes' biogenesis, and among these, the Golgi complex may serve as a common platform for canonical autophagy and NCA [40]. It has been reported that Beclin 1-independent NCA induced by αSNAP depletion is associated with Golgi complex fragmentation [41]. In contrast, disrupting the Golgi complex by BFA or GCA could suppress cis-unsaturated fatty acid-induced Beclin 1-independent NCA [32]. Here, we demonstrated that BFA and GCA also inhibited Nic-triggered NCLL ( Figure 5A-D). Furthermore, we observed that Nic-induced LC3 puncta overlapped with the cis-Golgi marker, GM130 ( Figure 5E), indicating that the Golgi complex could potentially be the source of the membrane for Nic-induced LC3-positive structures. Meanwhile, we found that Nic could decrease the expression level of GM130, and there were some Golgi fragments after Nic treatment ( Figure 5E,F). However, disruption of Golgi architecture and function has been widely observed in neurodegenerative diseases and the deletion of GM130 would result in cell death [42], which suggest that Nic might be toxic for neurons and may not be a good choice for neurodegenerative disease treatment.

Conclusions
In summary, we have provided evidence of Nic-induced ULK1-and Beclin 1-independent, but conjugation system-dependent NCLL, which could be significantly inhibited by Baf at the early stage. In addition, an intact Golgi complex and vimentin were involved in Nic-induced NCLL, which might be a membrane source or a platform for Nic-induced LC3-positive structures. Moreover, we have also provided several mechanistic views on how Nic could modulate NCLL, which would be helpful for an in-depth exploration of the pharmacological actions of Nic.