Autophagy is considered to be an important switch for facilitating normal to malignant cell transformation during colorectal cancer development. Consistent with other reports, we found that the membrane receptor Neuropilin1 (NRP1) is greatly upregulated in colon cancer cells that underwent autophagy upon glucose deprivation. However, the mechanism underlying NRP1 regulation of autophagy is unknown. We found that knockdown of NRP1 inhibits autophagy and largely upregulates the expression of aldo-keto reductase family 1 B10 (AKR1B10). Moreover, we demonstrated that AKR1B10 interacts with and inhibits the nuclear importation of glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and then subsequently represses autophagy. Interestingly, we also found that an NADPH-dependent reduction reaction could be induced when AKR1B10 interacts with GAPDH, and the reductase activity of AKR1B10 is important for its repression of autophagy. Together, our findings unravel a novel mechanism of NRP1 in regulating autophagy through AKR1B10.
Colorectal cancer is one of the leading causes of cancer death worldwide (Puccini and Lenz, 2018; Hubbard and Grothey, 2015; Siegel et al., 2014; Schmoll and Stein, 2014). The incidence of the disease is still increasing. More than 1.9 million new cases and ∼940,000 deaths of colorectal cancer were expected to occur worldwide during 2020 (Sung et al., 2021). In recent decades, although the treatment of colorectal cancer has been improved significantly (Schmoll and Stein, 2014; Palta et al., 2014), the overall survival rate of patients with advanced, metastatic or recurrent colorectal cancer is still far from satisfactory (Hubbard and Grothey, 2015; Siegel et al., 2014; Schmoll and Stein, 2014). Therefore, it is an arduous and urgent task to elucidate the molecular mechanism of the development and occurrence of colorectal cancer.
Among various pathogenic factors and mechanisms of colorectal cancer, autophagy is considered to be an important switch for facilitating normal to malignant cell transformation (Burada et al., 2015). Autophagy is a mechanism by which normal cells exercise surveillance. It protects them from transforming into malignant tumors by removing damaged organelles and aggregated proteins, and reducing active oxides, mitochondrial abnormalities and DNA damage. However, autophagy also supports the formation of tumors by promoting the acquisition of essential nutrients for cell metabolism and growth, and by inhibiting cell death and increasing drug resistance. Therefore, it is believed that autophagy can prevent the development of cancer in the early stage, but once cancer is established, the enhancement of autophagy can often lead to the survival and growth of cancer cells (Amaravadi et al., 2016; White, 2012). It is generally accepted that autophagy plays a dual and contradictory role in the development of cancer, but the exact mechanism of autophagy leading to cancer has not yet been fully determined and seems to vary according to the situation. At present, the study of autophagy related to colorectal cancer mainly focuses on several molecules, such as microtubule associated protein 1 light chain 3 (LC3, herein referring to MAPLC3A and MAPLC3B), beclin1 and ATG5, but the results are contradictory (Burada et al., 2015). Therefore, the mechanisms need to be further explored. In our study, we unravelled a NRP1-mediated autophagy pathway that NRP1 affects the expression of AKR1B10 and then the interaction between AKR1B10 and GAPDH.
As a transmembrane protein, NRP1 acts as a co-receptor for a number of extracellular ligands, including class III/IV semaphorins, certain isoforms of vascular endothelial growth factor (VEGF) and transforming growth factor beta (TGFβ) (Kolodkin et al., 1997; Soker et al., 1998; Glinka and Prud'homme, 2008). NRP1 is involved in a wide range of physiological and pathological processes. Consistent with other reports (He et al., 2018; Yang et al., 2019a,b; Pang et al., 2014), we found that the expression of NRP1 was induced by autophagy induction. However, it is unclear whether or how it could regulate cell autophagy. We then knocked down NRP1 in human colon cancer cells and revealed its role in regulating cell autophagy. RNA-seq analysis showed that the expression of AKR1B10 increases significantly after NRP1 knockdown.
AKR1B10, also known as aldose reductase-like-1 (ARL-1), was first identified from human hepatocellular carcinoma. AKR1B10 is an NADPH-dependent reductase that can reduce aldehydes, methylglyoxal and diacetyl (Cao et al., 1998). AKR1B10 is secreted through a lysosome-mediated non-classical pathway regulated by heat shock protein, and can be detected in the serum of cancers (Luo et al., 2011, 2013). AKR1B10 expression is regulated by the KEAP1-NRF2 pathway and p53 in colorectal cancer (Nishinaka et al., 2011; Ohashi et al., 2013). In mice, aldose reductase deletion enhances autophagy in transverse aortic constriction hearts (Baba et al., 2018). The inhibition of aldose reductase activates autophagic response during fasting, and increases clearance of aldehyde-protein adducts, which could maintain cellular homeostasis during starvation (Zhang et al., 2019a,b,c). This indicates that AKR1B10 may play a role in autophagy through reducing harmful aldehyde protein adducts to maintain glycolysis homeostasis. In normal tissues, AKR1B10 is primarily expressed in the colon and small intestine; however, the expression of AKR1B10 is decreased in colorectal cancer, but is increased in pancreatic cancers, lung small cell cancers, breast cancers and cervical cancers (Chung et al., 2012; Fukumoto et al., 2005; Ma et al., 2012; Yoshitake et al., 2007). When primary human tumors progress to advanced metastatic disease, cancer cells use high level glucose metabolism and gain more energy by promoting autophagy, which can fuel cellular metabolism. Autophagy therefore plays a key role in adhesion dynamics during epithelial-to-mesenchymal transition (Poillet and White, 2019; Mowers et al., 2018; Takamura et al., 2011a,b; Kimmelman and White, 2017). Our research shows AKR1B10 inhibits glucose-deprivation autophagy; this may imply that low AKR1B10 expression in colon cancer promotes tumor development by upregulated autophagy. Recently, studies have shown that AKR1B10 as a aldehyde-ketone reductase reduces carbonyl group and DL-glyceraldehyde in some anti-cancer drugs, and some inhibitors had developed in cancer chemotherapy (Huang et al., 2016; van Weverwijk et al., 2019; Shi et al., 2020; Matsunaga et al., 2012a,b). Exploring the mechanism between AKR1B10 and autophagy or cancer metabolism is important for understanding tumorigenesis, development and aggressive malignancy. Thus, AKR1B10 protein is a new potential therapeutic target in human cancers.
GAPDH is a housekeeping gene well known for its function in glycolysis. It catalyses the conversion of glyceraldehyde-3-phosphate to D-glycerate 1,3-bisphosphate, which is then rapidly converted to 3-phosphoglycerate and generates ATP (Seidler, 2013). In addition to its importance in metabolism, GAPDH is also involved in intracellular membrane trafficking, DNA repair, gene expression, apoptosis and autophagy (Jung et al., 2014; Sirover, 2012, 2011). In glucose metabolism, GAPDH activity can be activated by acetylation to promote cell proliferation and tumor growth (Li et al., 2014), methylated by CARM1 to regulate glucose metabolism (Zhong et al., 2018), and phosphorylated by AMPK to translocate to the nucleus during starvation-induced autophagy and thus maintain cellular energy homeostasis (Chang et al., 2015; Yang et al., 2018). In this study, we reveal that AKR1B10 interacts with GAPDH and repressed nuclear importation of GAPDH, through which AKR1B10 blocked the autophagy.
NRP1 promotes autophagy induced by glucose starvation in human colon cells
In accordance with other reports (He et al., 2018; Yang et al., 2019a,b; Pang et al., 2014), we also found that the expression of NRP1 was increased with the induction of autophagy by glucose starvation in HT-29 cells (Fig. 1A,B). To further explore the role of NRP1 in autophagy, we developed two NRP1 knockdown HT-29 cell lines named, shNRP1#1 and shNRP1#2 (Fig. 1C), using a lentivirus-based shRNA technique with two distinct targeting shRNAs. We also complemented shNRP1#1 with NRP1 (Fig. 1D) to investigate the rescue effect of NRP1. Cells were treated with complete medium or glucose-deprived medium for 18 h, with or without 10 μM chloroquine for inhibiting or not inhibiting lysosome degradation, and autophagic flux was measured. Compared with the control cells, autophagic flux decreased in NRP1 knockdown cell lines (Fig. 1E-G). When we performed the rescue experiment with the replenishment of NRP1 into the shNRP1#1 cell, autophagic flux was restored, precluding any off-target effects of the NRP1 shRNA (Fig. 1E-G). We then used confocal microscopy to examine both autophagosome and autolysosome LC3+ puncta in HT-29 cells of the control, shNRP1#1 and rescue-NRP1, which were all infected with mRFP-GFP-LC3 lentivirus for stable expression. According to the principle of mRFP-GFP-LC3 fluorescence, mRFP-GFP tandem fluorescently labeled LC3 (tfLC3) displays both GFP and mRFP signals before fusing with lysosomes. When autophagosomes fuse with lysosomes, the green fluorescence is quenched with only red fluorescence left, because green fluorescence from the GFP protein is sensitive to acid. Therefore, spots with both red and green fluorescence indicate autophagosomes; spots with only red fluorescence but not green fluorescence indicate autolysosomes. We found that autophagosome and autolysosome LC3+ puncta both decreased in NRP1 knockdown cells treated with glucose-deprived medium for 18 h, and the phenomena was restored in NRP1-rescued cells (Fig. 1H,I). Collectively, our results demonstrate that NRP1 could promote autophagy driven by glucose starvation.
AKR1B10 is downregulated by NRP1 and plays a negative role in regulating autophagy
To explore how NRP1 promoted autophagy driven by glucose starvation, we analyzed the different expression of genes between the control and NRP1 knockdown cells using RNA-seq analysis. As a result, compared with control cells, AKR1B10 was strikingly increased in shNRP1 cells (Fig. 2A,B). To further confirm the downregulation of AKR1B10 by NRP1, we checked the expression of AKR1B10 in NRP1 knockdown cells (Fig. 2C) and NRP1 rescued cells (Fig. 2D,E) through western blotting and qRT-PCR. These data implied that NRP1 might promote autophagy by downregulating the expression of AKR1B10; therefore, AKR1B10 could have a negative role in regulating autophagy.
To verify the role of AKR1B10 in glucose deprivation-induced autophagy, we designed two distinct AKR1B10 shRNA primers, developed lentivirus-infected HT-29 cells and screened for stable expression cells of shctrl, shAKR1B10#1 and shAKR1B10#2 (Fig. 3A). We cultured cells in complete medium or glucose-deficient medium with or without 10 μM chloroquine for 18 h. We found that, after treatment with glucose-deprived medium, autophagic flux greatly increased in AKR1B10 knockdown cells compared with HT-29 shctrl cells (Fig. 3B-D). However, when AKR1B10 was replenished in shAKR1B10 cells, the increased autophagic flux reversed (Fig. 3E-H). According to confocal microscopy observations, autophagosome and autolysosome LC3+ puncta were obviously increased in AKR1B10 knockdown cells (Fig. 3I), and thus correlated with the statistical analysis (Fig. 3J). In addition, we overexpressed AKR1B10 in another human colon cancer cell line, HCT116, which has low expression of AKR1B10, and found that the autophagic flux was depressed by the overexpression of AKR1B10 (Fig. S1). These results clearly showed AKR1B10 inhibited autophagy induced by glucose starvation.
The reductase activity of AKR1B10 is essential for the repressing autophagy of AKR1B10
AKR1B10 is an NADPH-dependent aldose-keto reductase, which can catalyze the NADPH-dependent reduction of a wide variety of aldehydes and ketones, such as glyceraldehyde, methylglyoxal, diacetyl and aromatic aldehydes (Endo et al., 2009). It has been reported that site mutation of amino acid K125L or V301L damages the catalytic activity of AKR1B10 towards all-trans-retinaldehyde. For example, either mutation loses reductase activity of AKR1B10 towards DL-glyceraldehyde (Gallego et al., 2007). To explore whether AKR1B10 reductase activity regulates glucose starvation-induced autophagy, we used AKR1B10 inhibitor oleanolic acid (OA) (Takemura et al., 2011a,b) to inhibit the reductase activity of AKR1B10 and check the autophagic flux with corresponding stimulation in HT-29 cells. As shown, when cells were treated with 10 μM OA, the autophagic flux increased (Fig. 4A), which was the same as in shAKR1B10 cells. As AKR1B10 expression increased in shNRP1 cells, the reductase activity of AKR1B10 should be correspondingly increased. Thus, we treated shNRP1 cells with 10 μM OA to inhibit AKR1B10 activity, and we found that shNRP1 cells treated with OA showed high autophagic flux (Fig. 4B), which indicated the promotion of cell autophagy. When the amino acid site K125 or V301 of AKR1B10 was mutated into K125L or V301L, AKR1B10 lost its reductase activity. So, we respectively constructed and replenished AKR1B10 (K125L), AKR1B10 (V301L) and AKR1B10 (K125L and V301L) mutants in AKR1B10 knockdown cells (Fig. 4C). After stimulating the cells with glucose starvation with or without chloroquine, we found that all three rescued AKR1B10 mutants lost their ability to inhibit autophagic flux (Fig. 4D-K). Furthermore, the results were also confirmed by the counting of autophagosome and autolysosome LC3+ puncta in shctrl, AKR1B10 rescue, AKR1B10 (K125L) rescue, AKR1B10 (V301L) rescue and AKR1B10 (K125L and V301L) rescue cells (Fig. S2). Therefore, the data revealed that the reductase activity was very important for AKR1B10 to inhibit autophagy.
AKR1B10 represses autophagy through interacting with GAPDH and preventing it from translocating into the nucleus
To further explore the molecular mechanisms of AKR1B10 in inhibiting autophagy, we performed a co-immunoprecipitation (Co-IP) assay in AKR1B10 rescue cells that were treated with glucose-deprived medium, and then protein samples were collected and analyzed by mass spectrometry. We found several AKR1B10-interacting protein candidates, and GAPDH was one of the candidates attracting our attention (Fig. 5A). Confirmed by western blotting, we found that GAPDH interacted with AKR1B10 and that the interaction slightly decreased over time (Fig. 5B). Furthermore, through mass spectrometry analysis, we found that site S118 of AKR1B10 was phosphorylated (Fig. 5C). We also checked the GAPDH-interacting effects of mutant AKR1B10s that lost the reductase activity. Compared with the wild-type AKR1B10, the mutated AKR1B10 (K125L), AKR1B10 (V301L) and AKR1B10 (K125L and V301L) had the weaker interacting effects with GAPDH, although they could still interact with GAPDH (Fig. 5D). In view of the fact that the S118 site of AKR1B10 changed its phosphorylation status when stimulated by autophagy, we also constructed two related point mutants, AKR1B10 (S118A) and AKR1B10 (S118D), to test their interaction with GAPDH. AKR1B10 (S118A) lost the ability to be phosphorylated, whereas AKR1B10 (S118D) mimicked phosphorylated status (Fig. 5E). As shown, the phosphorylation status at the S118 site of AKR1B10 did not obviously affect the interaction between AKR1B10 and GAPDH (Fig. 5E). The amino acid site E250 of GAPDH is essential for its nucleus import (Bae et al., 2008). When E250 was mutated to M250, losing its nuclear-entry ability, we found that the site-mutated GAPDH had a much stronger interaction with AKR1B10 (Fig. S3).
Collectively, AKR1B10 can directly interact with GAPDH, although the reductase activity-related mutations of AKR1B10 weaken the interaction with GAPDH, and the phosphorylation status of AKR1B10 has little effect on their interaction. As AKR1B10 has the reductase activity and the ability to reduce its substrates, we speculated that GAPDH could be reduced by AKR1B10. To test this hypothesis, we first expressed and purified recombinant proteins of AKR1B10, AKR1B10 (K125L), AKR1B10 (V301L), AKR1B10 (K125L and V301L) and GAPDH from Escherichia coli. Second, we performed two reduction reactions: one was with a mixture of substrate DL-glyceraldehyde and AKR1B10 or mutants, and the other was with a mixture of GAPDH and AKR1B10 or mutated AKR1B10s. These two reactions were separately incubated with 0.1 μM NADPH at 35°C in vitro. The NADP+/NAPDH ratio of the products was determined by liquid chromatography with mass spectrometry (LC/MS). The result showed that OA could inhibit AKR1B10 reductase activity and three mutants lost most of their reductase activities (Fig. 6A). As a result, we found that the NADP+/NADPH ratio increased when the mixture of AKR1B10 and GAPDH was incubated, which indicated a successful reduction reaction. However, the NADP+/NADPH ratio did not change when GAPDH was incubated with various AKR1B10 reductase mutations (Fig. 6B), which indicated failed reduction reactions.
Previously, it has been reported that GAPDH was phosphorylated by AMPK and then translocated from the cytoplasm into the nucleus and thus enhanced autophagy upon glucose starvation in HEK293 (Chang et al., 2015; Yang et al., 2018). To test whether AKR1B10 could regulate GAPDH redistribution from cytosol to nucleus, we performed a nucleus-cytosol separation experiment from control cells and AKR1B10 knockdown cells that were treated by glucose-deprived medium for 12 h. We found GAPDH translocated to the nucleus in HT-29 cells without glucose induction, and AKR1B10 knockdown promoted nuclear translocation of GAPDH (Fig. 6C). In contrast, AKR1B10 rescue blocked the translocation of GAPDH to the nucleus (Fig. 6D,F). Unlike AKR1B10 rescue, we found that the rescue of K125L or V301L of AKR1B10 reductase activity mutations could not effectively block the translocation of GAPDH from the cytosol to the nucleus (Fig. 6E,F). Correspondingly, the immunofluorescence assay supported the notion that the reductase activity of AKR1B10 hindered the nuclear import of GAPDH (Fig. 6G,H). Taken together, the data implied GAPDH was downstream of AKR1B10, and AKR1B10 reduced GAPDH to inhibit its nuclear importation upon glucose starvation, thus inhibiting the autophagy.
Consistent with some previous reports (He et al., 2018; Yang et al., 2019a,b; Pang et al., 2014), we found that NRP1 accumulated in HT-29 cells after the induction of autophagy by glucose deprivation. However, this was somewhat contradictory to other reports that suggest hypoxia and nutrient deprivation stimulate the rapid loss of NRP1 expression in both endothelial and carcinoma cells of breast cancer and prostate cancer (Tao et al., 2003). The difference could be caused by the different cell lines used or the different stimulating conditions used. In our western blot assay, we found that a slightly downshifted band of NRP1 mainly contributed to the accumulation of NRP1. The qPCR results confirmed the accumulation of NRP1. As several human NRP1 variants were generated by alternative splice (Hendricks et al., 2016; Gagnon et al., 2000; Cackowski et al., 2004; Huang et al., 2019a,b), we speculated that the downshifted band could be a certain kind of NRP1 splice variant. Recently, two human NRP1 splice variants from colorectal cancer cells, generated by skipping exon 4 and exon 5, have been identified. The loss of these two exons causes defects in N-linked glycosylation and plays a critical role in the regulation of the endocytic trafficking of NRP1 (Lamb et al., 2013). In this study, we could not determine the splice variant of NRP1 accumulated in HT-29 cells by autophagy induction. We believe that the identification of the NRP1 variant would be greatly helpful to explore the role of NRP1 in autophagy.
NRP1 has been reported to be involved in a previously undiscovered endocytic pathway, which initiated intracellular trafficking of CendR peptides. CendR endocytosis could be enhanced by nutrient depletion, such as removal of glucose or amino acids, and there is a concomitant increase of NRP1 expression at the cell surface (Pang et al., 2014). Endocytosis has a close link with autophagy in terms of the formation of early endosome derived from the plasma membrane, which fuses to autophagosome and then to lysosome to be degraded (Lamb et al., 2013; Tooze et al., 2014; Yu et al., 2018). In our study, we found that autophagy could increase the expression of NRP1, and NRP1 in turn could promote the autophagy. The expression changes of NRP1 mediated by autophagy induction needs to be further detailed. Here, we demonstrated that knockdown of NRP1 increased the expression of AKR1B10, an aldose reductase, which impeded the progression of autophagic flux induced by glucose depleting.
AKR1B10 is a new member of aldose reductase family 1. It uses NADPH as a co-enzyme to reduce aldehydes to alcohols. The expression of AKR1B10 can be regulated by the transcription factor NRF2 (also known as NRF2) and NRF2 binding protein, KEAP1 (Nishinaka et al., 2011; Agyeman et al., 2012). AKR1B10 can also be regulated by p53 protein at the transcription level and can participate in p53-regulated cell apoptosis (Ohashi et al., 2013). AKR1B10 secretion is regulated by the G protein-coupled receptor signaling pathway (Luo et al., 2011). Some studies have found that phorbol ester can inhibit the expression of AKR1B10 through the ERK/c-Jun signaling pathway (Nishinaka et al., 2015). In our study, we found that NRP1 could repress AKR1B10 through the p62-NRF2 pathway. In NRP1 knockdown cells, P62 (also known as SQSTM1) expression was increased (Fig. S5A). As P62 accumulation leads to the stabilization of NRF2 and the transcriptional activation of NRF2-targeting cellular defence enzyme genes (Lau et al., 2010; Komatsu et al., 2010), we constructed two shP62 cell lines. The results showed that knockdown of P62 caused a dramatic decrease in the nuclear importation of NRF2 and expression of AKR1B10 (Fig. S5B,C). NRF2 is one of the major transcriptional factors involved in AKR1B10 gene regulation (Nishinaka et al., 2011), which was verified when we used NRF2 activator bardoxolone methyl to detect AKR1B10 expression (Fig. S5D) (Yang et al., 2020; Tian et al., 2019). Thus, we speculated that NRP1 downregulates AKR1B10 through the P62-NRF2 axis. We supposed that NRP1 inhibits P62 and then the nuclear translocation of NRF2, which leads to the decrease of AKR1B10 expression. However, the additional mechanisms need to be explored.
Although AKR1B10 is an aldose reductase, there has been no report that AKR1B10 can reduce GAPDH. For the first time, we found that AKR1B10 could directly interact with GAPDH and cause an NADPH-dependent reduction reaction between them. However, we still do not know how AKR1B10 interacts with GAPDH and causes the reduction reaction. We speculate that certain carbonyl groups on GAPDH could be reduced by AKR1B10. This needs to be further explored in the future. It has been reported that overexpression of AKR1B10 can inhibit cellular autophagy induced by 6-tert-butyl-2,3-epoxy-4-benzoquinone (Matsunaga et al., 2012a,b). Our research work shows that AKR1B10 inhibits autophagy induced by glucose starvation through its reductase activity, and its reductase inhibitor OA can reverse this inhibition. In addition, we also found that reductase activity-related point mutations of AKR1B10, either K125L or V301L, result in a loss of ability to repress autophagy.
GAPDH has recently been found to be involved in the regulation of autophagy. It was found that in the autophagy induced by glucose starvation rather than amino acid starvation, the 122nd serine (ser122) of GAPDH in the cytoplasm was phosphorylated by activated AMPK. The phosphorylated GAPDH was then transferred from the cytoplasm to the nucleus and directly interacted with, and activated, the histone deacetylase SIRT1 in the nucleus, thus activating autophagy (Chang et al., 2015). GAPDH is thus a pivotal and central regulator of autophagy under glucose deficiency. SIRT1 activated by GAPDH deacetylated the K49 and K51 sites of LC3 in the nucleus, and the deacetylated LC3 interacted with ATG7 from the nucleus to the cytoplasm to initiate autophagy (Huang et al., 2015). Another report found that GAPDH was not only redistributed from the cytoplasm to the nucleus but was also redistributed to cytoplasmic membranes after being phosphorylated by AMPK, thus inhibiting the transport of cop I vesicles and multiple transport pathways in cells, so that cells can reduce energy consumption to maintain cell homeostasis when they are hungry (Yang et al., 2018).
As we all know, AMPK is a key molecule in regulating glucose metabolism and autophagy during glucose deprivation (Ha et al., 2015; Kim et al., 2011). We also tried to find out whether NRP1 or AKR1B10 affect AMPK in no glucose-induced autophagy. We found that AMPK phosphorylation was decreased in NRP1 knockdown cells treated by no glucose medium for 12 h, and increased in NRP1 rescue cells (Fig. S4A,B); meanwhile, AKR1B10 could inhibit AMPK phosphorylation (Fig. S4C,D). Furthermore, rescue of reductase activity-related mutants of AKRB10 could not effectively repress AMPK phosphorylation (Fig. S4E-G), which indicates that reductase activity of AKR1B10 is important for AMPK phosphorylation. These results implied that NRP1 and AKR1B10 may be upstream of AMPK to regulate autophagy and the detailed mechanism needs to be further explored, and indicates that AKR1B10 downregulates autophagy not only through interacting with GAPDH to inhibit its nuclear transport but also through inhibiting AMPK phosphorylation.
In this study, we unravelled the novel mechanism that NRP1 can promote autophagy through repressing AKR1B10 expression. AKR1B10 interacts with GAPDH and reduces GAPDH to prevent nuclear entry of GAPDH, thus impeding the progress of autophagy upon glucose starvation. In addition, our studies indicated that AKR1B10 downregulates autophagy not only through interacting with GAPDH to inhibit its nuclear transport but also through inhibiting AMPK phosphorylation. These findings may further mechanistic insights into the regulation of autophagy in human colon cancer.
MATERIALS AND METHODS
Reagents and antibodies
Reagents used in this study were as follows: no glucose medium (GIBCO, 11966025); chloroquine (Sigma-Aldrich, C6628; Ju et al., 2010); puromycin (Solarbio, P8230; Liang et al., 2019); Oleanolic Acid (Selleck, S2334; Zhang et al., 2019a,b,c); and anti-FLAG M2 beads (Sigma-Aldrich, A2220; Roach et al., 2007). The primary antibodies used in this study were as follows: rabbit-anti-human AKR1B10 antibody (1:1000; Abcam, ab96417; Wang et al., 2017); rabbit-anti-LC3 antibody (1:1000; Proteintech, 14600-1-AP; Zhang et al., 2019a,b,c); rabbit-anti-NRP1 antibody (1:1000; Abcam, ab81321; Mei et al., 2020); mouse-anti-GAPDH antibody (1:1000; Proteintech, 6004-1-AP; Ding et al., 2018); mouse-anti-GAPDH antibody for immunofluorescence (1:100; Santa Cruz Biotechnology, sc47724; Morozova et al., 2010); mouse-anti-HA antibody (1:1000; Sigma-Aldrich, H9658; Zhang et al., 2016); mouse-anti-flag antibody (1:1000; Sigma-Aldrich, F3165; Heo et al., 2010); rabbit-anti-β-tubulin antibody (1:1000; Proteintech, 10068-1-AP; Liu et al., 2019); mouse-anti-β-actin antibody (1:1000; Proteintech, 66009-1-Ig; Huang et al., 2019a,b); mouse IgG (H+L) highly cross adsorbed secondary antibody (1:1000; Invitrogen, A32728; Ogueta et al., 2018); goat-anti-mouse secondary antibody (1:10000; Sigma-Aldrich, A0412; Liang et al., 2017); goat-anti-rabbit secondary (1:10000; Sigma-Aldrich, A6154; Liang et al., 2017); bafilomycin A1 (Abcam, ab120497; Liao et al., 2018); bardoxolone methyl (MCE, |HY-13324; Wyler et al., 2019); pi-AMPK(1:1000; Cell Signaling Technology, 2535; Chang et al., 2015); AMPK (1:1000; Cell Signaling Technology, 2532; Chang et al., 2015); P62 (1:1000; Proteintech, 18420-1-AP; Arhzaouy et al., 2019); NRF2 (1:1000; Proteintech, 16396-1-AP; Yamada et al., 2019); and LaminA/C (1:1000 Proteintech, 10298-1-AP; Yang et al., 2019a,b). We used 5% bovine serum albumin as an antibody dilution.
Cell culture, transfection and infection
HT29, HCT116 and V293T cell lines were obtained from the Cancer Research Center of Xiamen University (Xiamen, Fujian, China) (Zhong et al., 2020). These cell lines were cultured in Dulbecco's modified Eagle medium (Gibco, C11995500BT) supplemented with 10% fetal bovine serum, 100 IU penicillin and 100 mg/ml streptomycin (Gibco, C15140122) at 37°C in a humidified incubator containing 5% CO2. No glucose medium was used as the source of glucose starvation to induce autophagy. For lentivirus production, V293T cells were transfected with lentiviral vectors and virus-packing plasmids pMDL, pREV and pVSV-G by calcium phosphate transfection. The virus-containing medium was collected 48-60 h later and added to cells with 10 μg/ml polybrene, then centrifuged at 800 g for 30 min. Infectious medium was changed with fresh medium 6-12 h later.
The sequences encoding human NRP1, AKR1B10 and GAPDH were amplified from HT29 cell line cDNA and cloned into the pLV-EF1a-puro vector and pCDNA3.3 vector. The autophagy marker mRFP-GFP-LC3 (obtained from Plasmid, 21074) was replaced into pLV-EF1a-puro. Details of the sequences are available upon request. Primer sequences used to amplify these genes were as follows: hNRP1-F, 5′-GAGAGGGGGCTGCCGCTCCTCTGCGCCGT-3′; hNRP1-R, 5′-TCATGCCTCCGAATAAGTACTCTGTGTA-3′; hAKR1B10-F, 5′-GCCACGTTTGTGGAGCTCAGT-3′; hAKR1B10-R, 5′-TCAATATTCTGCATTGAAGGGATAGT-3′; hAKR1B10(K125L)-F, 5′-CCTTTTCCCCTTAGATGATAAAGG-3′; hAKR1B10(K125L)-R, 5′-CCTTTATCATCTAAGGGGAAAAGG-3′; hAKR1B10(V301L)-F, 5′-GGCCTGTAACTTGTTGCAATCCT-3′; hAKR1B10(V301L)-R, 5′-AGGATTGCAACAAGTTACAGGCC-3′; AKR1B10(S118A)-F, 5′-ACAGGGATTCAAGGCTGGGGATGACCTTT-3′; AKR1B10(S118A)-R, 5′-AAAGGTCATCCCCAGCCTTGAATCCCTGT-3′; AKR1B10(S118D)-F, 5′-ACAGGGATTCAAGGATGGGGATGACCTTT-3′; AKR1B10(S118D)-R, 5′-AAAGGTCATCCCCATCCTTGAATCCCTGT-3′; GAPDH(E250M)-F, 5′-CTGCCGTCTAATGAAACCTGCCA-3′; GAPDH(E250M)-R, 5′-TGGCAGGTTTCATTAGACGGCAG-3′; GAPDH-F, 5′-GGGAAGGTGAAGGTCGGAGTCAAC-3′; GAPDH-R, 5′-TTACTCCTTGGAGGCCATGT-3′; ptfLC3-F, 5′-GCCTTCTCCGAGGACGTCATCAAG-3′; ptfLC3-R, 5′-TCACAAGCATGGCTCTCTTCCTGT-3′.
RNAi and Real-time PCR
For the shRNA constructs, shRNA oligo was cloned into lenti-based shRNA vector pLV-H1-EF1α-puro (from Biosettia, San Diego, CA, USA) according to the manufacturer's instructions. The sequences of shRNA oligos for human NRP1 and AKR1B10 were as follows: shNRP1#1, 5′-GCTACGACCGGCTAGAAAT-3′; shNRP1#2, 5′-GGGCAACAACAACTATGATAC-3′; shAKR1B10#1, 5′-GGTTCTGATCCGTTTCCATAT-3′; shAKR1B10#2, 5′-GAACAAACCTGGACTGAAATA-3; shP62#1, 5′-GCTCACCGTGAAGGCCTACCT-3′; shP62#2, 5′-GGCGCACTACCGCGATGAGGA-3′.
A non-target shRNA oligonucleotide (5′-CAACAAGATGAAGAGCACCAA-3′) was used as a negative control. All constructs were verified by DNA sequencing. For real-time PCR, primers for NRP1, AKR1B10 and GAPDH were as follows: NRP1-RT-F, 5′-CCCCAAACCACTGATAACTCG-3′; NRP1-RT-R, 5′-AGACACCATACCCAACATTCC-3′; AKR1B10-RT-F, 5′-ACCTGTTCATCGTCAGCAAG-3′; AKR1B10-RT-R, 5′-CATCCCCAGACTTGAATCCC-3′; GAPDH-RT-F, 5′-ACATCGCTCAGACACCATG-3′; GAPDH-RT-R, 5′-TGTAGTTGAGGTCAATGAAGGG-3′.
Measurement of autophagic flux
Cells were cultured with no glucose medium and treated with or without chloroquine (10 μM) for hours. Cells were then harvested, lysed with RIPA buffer (Beyotime, P0013B; Shen et al., 2007), measured for protein concentration by BCA method, normalized for total protein content and subjected to immunoblot analysis using anti-LC3 antibody. Autophagic flux was measured by comparison of the levels of LC3-II after no glucose medium with or without chloroquine treatment. Statistical analysis of ratios of LC3-II/β-actin under autophagic flux was undertaken using ImageJ software. At least three independent experiments were performed.
Western blotting and immunoprecipitation
Cells were washed twice with PBS and lysed in RIPA buffer for 20 min. After centrifugation, the samples were subjected to SDS-PAGE gels. Following electrophoresis, proteins in gels were transferred to PVDF membranes (Fang et al., 2016) and the blocked membranes were blotted with antibodies. The chemiluminescent signals were detected using an ECL substrate kit (NCM Biotech; Zhou et al., 2019). At least three different time points were selected for exposure and the results were analyzed using ImageJ software to ensure that band detection was within the linear range. Western blots shown are representative of the appropriate one with band detection within the linear range. For the immunoprecipitation assay, cell extract was mixed with anti-Flag M2 beads in a ratio of 1 ml of extract per 10 μl of beads at 4°C overnight. The beads were then pelleted at 800 g for 3 min and washed with lysis buffer three times. The beads were subjected to elution with lysis buffer containing 100 μg/ml 3×Flag peptide. Western blotting of the cell lysates and immunoprecipitates was performed using anti-Flag, anti-HA and other antibodies, as indicated.
Immunostaining and confocal microscopy
Cells were fixed in 4% formaldehyde for 30 min at room temperature and then incubated with 0.5% Triton X-100 for 20 min at 4°C. The cells were then incubated with appropriate primary antibodies overnight at 4°C. Next, they were incubated with a fluorescent-labeled secondary antibody for 2 h at room temperature. After nuclear counterstaining of the cells with Hoechst 33342, confocal images were obtained using a Zeiss LSM 880+Airyscan with a 63× oil objective. Image processing was carried out using Zen 2.3 blue edition confocal acquisition software. Cells were randomly selected, and the amount of GAPDH redistributed from the cytosol to the nucleus was counted manually in a two-blinded manner.
Tandem mRFP-GFP-LC3 assay
Experimental cells infected with mRFP-GFP-LC3 lentivirus were treated with glucose-deprived medium for 18 h. Next, cells were fixed in 4% formaldehyde for 30 min at room temperature and then incubated with 0.5% Triton X-100 for 20 min at 4°C. Confocal images were obtained using a Zeiss LSM 880+Airyscan with a 63× oil objective. Image processing was carried out using Zen 2.3 blue edition confocal acquisition software. Cells were randomly selected, and the numbers of GFP puncta and mRFP puncta were counted manually in a two-blinded manner. The numbers of autophagosome puncta and numbers of autolysosome puncta were calculated based on the numbers of GFP puncta and mRFP puncta. The statistical significance (P-value) was analyzed using Student's t-test in GraphPad Prism6.
Recombinant protein purification and in vitro enzyme reductase activity assay
Full-length human AKR1B10 (wild-type, K125L, V301L or K125L and V301L mutant) and GAPDH were cloned into pGEX-4T1 and expressed as the glutathione S-transferase-tagged form in E. coli BL21 by induction with 0.1 mM isopropyl β-D-thiogalactopyranoside for 12 h at 30°C. The recombinant proteins were purified using glutathione-sepharose4B beads, eluted with reduced glutathione and analyzed by western blotting. In vitro enzyme reductase activity reaction, GAPDH and AKR1B10 (wild-type, K125L, V301L or K125L and V301L mutant) were mixed in the buffer consisting of 135 mM Na3PO4 (pH 7.0), 50 mM KCL, 25 mM MgCl2, 0.2 mM DTT and 0.1 mM NADPH in a total volume of 0.2 ml. The mixture was incubated at 35°C for 40 min. AKR1B10 reductase activity to GAPDH was determined by high performance (HP)LC/MS-based NADP+/NADPH analysis. To make the NADP+ and NADPH standard curve, the NADP+ concentration gradient was as follows: 0.1 mM, 0.05 mM, 0.025 mM, 0.0125 mM, 0.00625 mM, 0.003125 mM and 0 mM, in a total volume of 0.2 ml. The NADPH concentration gradient was as follows: 0.1 mM, 0.05 mM, 0.025 mM, 0.0125 mM, 0.00625 mM, 0.003125 mM and 0 mM, in a total volume of 0.2 ml.
HPLC/MS-based analysis of NADP+/NADPH
For NADP+/NADPH analysis, liquid chromatography using an AB SCIEX ExionLC was prepared and all chromatographic separations were performed using an SeQuant ZIC-pHILIC column (5 μm, 2.1×100 mm internal dimensions, PN: 1.50462.0001). The column was maintained at 25°C, and the injection volume of all samples was 2 μl. The mobile phase consisted of 15 mM ammonium acetate and 3 ml/l ammonium hydroxide (>28%) in LC/MS grade water (mobile phase A) and LC/MS grade 90% (v/v) acetonitrile: HPLC water (mobile phase B) ran at a flow rate of 0.2 ml/min. The samples were separated using the following gradient program: 95% B held for 1 min, increased to 50% B in 6 min, held for 1 min, and post time was set for 2 min. The HPLC system was coupled to an AB SCIEX QTRAP5500 tandem mass spectrometer. Quantification was achieved by multiple reaction monitoring in negative ion mode.
Flag-AKR1B10-rescued cells were immediately washed twice with PBS and harvested by scraping and centrifugation at 100 g for 10 min. The harvested cells were washed with PBS and lysed for 30 min on ice in lysis buffer [(20 mM Tris-HCl, 150 mM NaCl, 1% Triton X-100 (pH 7.5)] with protease inhibitor cocktail. Cell lysates were then spun down at 20,000 g for 30 min. The soluble fraction was collected and immunoprecipitated overnight with anti-Flag M2 antibody-conjugated agarose at 4°C. Resins containing protein complexes were washed three times with lysis buffer. Proteins were then eluted twice with 0.1 mg/ml of 3×Flag peptide in lysis buffer for 15 min each time, and elutions were pooled for a final volume of 100 μl. Proteins in the elution were precipitated with 20% trichloroacetic acid, and the pellet was washed twice with 1 ml cold acetone and dried in speedvac. The protein pellet was dissolved in 1% sodium deoxycholate (SDC)/10 mM TCEP/40 mM CAA/Tris-HCl (pH 8.5). Subsequently, 1% SDC was diluted to 0.5% with water. The protein concentration was measured with Pierce 660 nm protein assay reagent (Thermo Fisher Scientific). Trypsin (Sigma-Aldrich) was added with a ratio of 1:100 (trypsin: protein). The tubes were kept at 37°C for 12-16 h. Peptides were desalted with SDB-RPS StageTips.
Peptides were dissolved in 0.1% formic acid and analyzed using sequential window acquisition of all theoretical mass spectra (SWATH-MS). MS analysis was performed on a TripleTOF 5600 (Sciex) mass spectrometer coupled to a NanoLC Ultra 2D Plus (Eksigent) HPLC system. Peptides was first bound to a 5 mm×500 μm trap column packed with Zorbax C18 5-μm 200-Å resin using 0.1% (v/v) formic acid/2% acetonitrile in water at 10 μl/min for 5 min, and then separated using a gradient from 2 to 35% of buffer B [buffer A 0.1% (v/v) formic acid, 5% DMSO in water, buffer B 0.1% (v/v) formic acid and 5% DMSO in acetonitrile] on a 35 cm×75 μm in-house pulled emitter-integrated column packed with Magic C18 AQ 3-μm 200- Å resin. The gradient time was 120 min for the immunoprecipitation dataset. For SWATH-MS, the mass spectrometer was operated such that a 250-ms survey scan (time-of-flight-MS) which was collected in 350-1500 m/z was performed followed by 32 100-ms MS/MS (MS2) experiments or 100 33-ms MS2 experiments. MS2 scans were collected at 100-1800 m/z. The fixed 25 Da MS2 experiments used an isolation width of 26 m/z (containing 1 m/z for the window overlap) to cover the precursor mass range of 400-1200 m/z.
The 100 variable isolation windows were ‘399.5-409.9, 408.9-418.9, 417.9-427.4, 426.4-436, 435-443.6, 442.6-450.8, 449.8-458, 457-464.8, 463.8-471.1, 470.1-476.9, 475.9-482.8, 481.8-488.6, 487.6-494, 493-499, 498-504.4, 503.4-509.3, 508.3-514.3, 513.3-519.2, 518.2-524.2, 523.2-529.1, 528.1-534.1, 533.1-539, 538-543.5, 542.5-548.5, 547.5-553, 552-558, 557-562.5, 561.5-567, 566-571.5, 570.5-576, 575-580.5, 579.5-585, 584-589.5, 588.5-594, 593-598, 597-602.5, 601.5-607, 606-611.1, 610.1-615.6, 614.6-620.1, 619.1-624.6, 623.6-628.6, 627.6-633.1, 632.1-637.6, 636.6-642.1, 641.1-646.6, 645.6-651.1, 650.1-655.6, 654.6-660.1, 659.1-665.1, 664.1-669.6, 668.6-674.5, 673.5-679, 678-684, 683-688.5, 687.5-693.4, 692.4-698.4, 697.4-703.3, 702.3-708.7, 707.7-713.7, 712.7-719.1, 718.1-724.5, 723.5-729.9, 728.9-735.3, 734.3-740.7, 739.7-746.5, 745.5-751.9, 750.9-757.8, 756.8-763.6, 762.6-769.5, 768.5-775.3, 774.3-781.2, 780.2-787, 786-793.3, 792.3-800.1, 799.1-806.4, 805.4-813.1, 812.1-820.3, 819.3-827.5, 826.5-835.2, 834.2-843.3, 842.3-851.4, 850.4-859.9, 858.9-868.9, 867.9-878.4, 877.4-888.3, 887.3-899.1, 898.1-910.3, 909.3-922.9, 921.9-936, 935-949.5, 948.5-963.4, 962.4-978.7, 977.7-994.9, 993.9-1015.6, 1014.6-1042.2, 1041.2-1070.1, 1069.1-1100.7, 1099.7-1140.7 and 1139.7-1196.5’.
SWATH-MS wiff files were converted to centroid mzXML using the qtofpeakpicker tool and profile mzXML files using proteoWizard MSConvert V.3.0.447. Centroid mzXML files were analyzed using DIA-Umpire (Tsou et al., 2015). DIA-Umpire was run with default settings except for BoostComplementaryIon=false. Profile mzXML files were split into a number of MS2 mzXML files and 1 MS1 mzXML file according to the SWATH window using the in-house script. Group-DIA software was composed of four modules: alignment; analysis; identification; and validation. For the generation of the internal library, only ‘alignment’ and ‘analysis’ modules were performed. Retention time in multiple runs was first aligned using MS1 intensity. MS1 and MS2 features were first extracted in a single run and were then concatenated across all runs. Precursors and the XICs similarity of product ions were compared, and the pair of precursor and product ions were then extracted. The generated presudospectra were stored in mgf and mzML formats.
The mgf files from DIA-umpire and Group-DIA were converted to mzML files, which were analyzed using Trans-Proteomic Pipeline (Version 5.0) software. mzML files were subjected to a database search using Comet (Version 2017.01) and X!tandem (Version 2013.06.15.1, native and k-score) against Swissprot human (downloaded in September, 2018) appendant with common contaminants and reversed sequence decoys (41,298 entries includes decoys for human). The search parameters were set as follows: parent monoisotopic tolerance 50 ppm; product ion tolerance 0.1 Da; modification 57.021464@C; potential modification mass 15.994915@M; and maximum missed cleavage sites 2. The pep.xml search results were validated and scored using PeptideProphet with parameters –OARPd -dDECOY and combined by iProphet with parameters DECOY=DECOY. Mayu (version 1.07) was used to determine iProphet probability corresponding to a 1% peptide false discovery rate (FDR). The peptide ions passing the 1% FDR were inputted into SpectraST for library building with the CID-QTOF setting. The retention time of peptides in the sptxt file was replaced with iRT time using spectrast2spectrast_irt.py script (downloaded from www.openswath.org), and iRT peptides used for retention time normalization were endogenous peptides. The sptxt file was made into a consensus non-abundant spectral library with the iRT retention time using spectraST.
The consensus sptxt files were converted to tsv using spectrast2tsv.py script, which was then converted to a TraML file with the TargetedFileConverter tool, which is integrated into OpenMS software (Version 2.2.0) (Röst et al., 2016a,b). In OpenSWATH analysis, ciRT peptide (Parker et al., 2015) and iRT peptides (Escher et al., 2012) were used for retention time normalization. The XIC extraction window is 20 min. An extended version of Pyprophet (Reiter et al., 2011; Rosenberger et al., 2017) (Pyprophet-cli v0.19; www.github.com/PyProphet) was employed for FDR estimation. A protein FDR of 1% at global level was applied in the non-phosphoproteomic dataset, and a 1% global peptide FDR was set for the phosphoproteomic dataset. The filtered results were inputted into TRIC software for cross-run alignment. The parameters in TRIC (Röst et al., 2016a,b) were set as follows: --method LocalMST--realign_method lowess_cython --max_rt_diff 60 --mst:useRTCorrection True --mst:Stdev_multiplier 3.0 --target_fdr 0.01 --max_fdr_quality 0.05.
The statistical significance of the results obtained from qPCR, immunofluorescence and NADP+/NADPH analysis was determined using Student's t-test in GraphPad Prism6.The results are shown as the mean±s.d., and P<0.05 was considered statistically significant. (*P<0.05, **P<0.01, ***P<0.001, ****P<0.0001). Immunofluorescence analyses were performed manually in a two-blinded manner.
We thank Jiahuai Han from Xiamen University for generous help and guidance; Chuanqi Zhong for mass spectrometry analysis at the State Key Laboratory of Cellular Stress Biology, Innovation Centre for Cellular Signaling Network, School of Life Sciences, Xiamen University; Yunwu Zhang from the Institute of Neuroscience at the School of Medicine, Xiamen University for kindly gifting us the mRFP-GFP-LC3 plasmid; and Jingru Huang for confocal imaging and Cixiong Zhang for HPLC/MS analysis at the Core Facility of Biomedicine, Xiamen University.
Conceptualization: W.L., T.W.; Methodology: W.L., C.L., Z.H., Z.L., W.Z.; Software: W.L., C.L., L.S., C.Z.; Validation: L.S., Z.L.; Formal analysis: W.L., Z.H., Z.L., P.M.; Investigation: L.S., S.W.; Resources: C.Z.; Data curation: C.L., Z.H., W.Z., P.M.; Writing - original draft: W.L.; Writing - review & editing: T.W.; Supervision: T.W.; Project administration: S.W., F.L., J.Y., T.W.; Funding acquisition: F.L., J.Y., T.W.
This work was supported by the National Basic Research Program of China (973 Program 2015CB553800), the National Natural Science Foundation of China (31401180 and 81773770); the Special Fund of Public Welfare Research Institutes in Fujian Province (2018||R1036-1, 2018||R1036-3 and 2019R1001-2); and the research fund of Shanghai Jiangxia Blood Technology Co., Ltd (0070-K81B0019).
The authors declare no competing or financial interests.