The Aryl Hydrocarbon Receptor Undergoes Chaperone-Mediated Autophagy in Triple-Negative Breast Cancer Cells

The aryl hydrocarbon receptor (AHR) is a ligand-activated signaling molecule expressed in many cell types, including triple-negative and non-triple-negative breast cancer cells. It affects breast cancer growth and crosstalk with estrogen receptor signaling. Normally, this receptor is degraded shortly after ligand activation via the 26S proteasome. Here, we report that AHR undergoes chaperone-mediated autophagy in MDA-MB-468 triple-negative breast cancer cells. This lysosomal degradation of AHR exhibits the following characteristics: (1) it is triggered by 6 amino-nicotinamide, starvation, and piperazinylpyrimidine compound Q18; (2) it is not observed in non-triple-negative breast cancer cells (MCF-7, T47D, and MDA-MB-361); (3) it can be inhibited by progesterone receptor B but not estrogen receptor alpha; (4) it can be reversed by chloroquine but not MG132; (5) it requires LAMP2A; and (6) it involves AHR-HSC70 and AHR-LAMP2A interactions. The NEKFF sequence localized at amino acid 558 of human AHR appears to be a KFERQ-like motif of chaperone-mediated autophagy, responsible for the LAMP2A-mediated AHR protein degradation.


Introduction
Autophagy plays an important role in various normal physiological functions, such as immune responses, liver and cardiac function, aging, and responses to infection [1]. In addition, it is implicated in disease development such as cancer, metabolic disorders, and neurodegeneration. Autophagy can be categorized into macroautophagy, microautophagy, and chaperone-mediated autophagy [2]. Macroautophagy generally involves the sequestration of organelles into the formation of the double-membrane structure of autophagosome, followed by lysosomal fusion for degradation. Microautophagy involves the lysosomal internalization of the whole organelle, such as mitochondria and peroxisomes, and a region of cytosol for degradation. Chaperone-mediated autophagy (CMA) allows the selective degradation of client proteins by guiding them directly into lysosomes for degradation (for general review, see [3]). This CMA-mediated lysosomal entry requires the HSC70 chaperone to bring client proteins to the CMA receptor LAMP2A at the lysosomal membrane [4]. This client protein-LAMP2A interaction starts the process of unfolding the client protein before internalization into lysosomes for degradation by hydrolytic enzymes [5]. CMA client proteins contain the HSC70 interaction motif, which has been mapped as a pentapeptide sequence initially shown as KFERQ [6]. Since then, many KFERQ-like motifs have been reported which follow the general rules of this CMA signature motif.
The aryl hydrocarbon receptor (AHR) is a ligand-activated transcription factor essential for a variety of cellular functions; for example, this receptor is involved in xenobiotic sensing [7], normal physiological functions (such as liver development [8,9], immune The aryl hydrocarbon receptor (AHR) is a ligand-activated transcription factor essential for a variety of cellular functions; for example, this receptor is involved in xenobiotic sensing [7], normal physiological functions (such as liver development [8,9], immune response [10,11], and hematopoietic stem cell differentiation [12]), and many diseases (such as breast and other cancers [13][14][15], cardiac disorders [16,17], respiratory disorders [18,19], metabolic disorders [20,21], and macular degeneration [22,23]). AHR resides in the cytoplasm as a complex containing AHR, a dimer of HSP90, p23, and XAP2 [24][25][26]. The canonical signaling mechanism of this receptor is initiated by ligand binding to AHR, causing the conformational change necessary for revealing the nuclear localization sequence of AHR. The nuclear internalization of the AHR complex can then occur via a pendulindependent mechanism [27]. The binding of aryl hydrocarbon receptor nuclear translocator (ARNT) to AHR dissociates the AHR complex [28], resulting in an active transcription factor in the form of AHR/ARNT heterodimer. This transcription factor then binds to the dioxin response element (DRE) enhancer, activating the transcription of a battery of target gene-most notably is the cytochrome P450 genes cyp1a1, cyp1a2, and cyp1b1 [29]. Interestingly, another heterodimeric complex of AHR-AHR/RelB-has been reported to regulate different AHR-dependent gene transcription [30].
After ligand activation, AHR undergoes ubiquitination and is degraded via the 26S proteasome as early as 1-2 h after ligand treatment [31]. The displacement of HSP90 from AHR in the cytoplasm by geldanamycin (GA) also causes the degradation of AHR via the same ubiquitin-proteasome system [32]. We observed that the down-regulation of p23 promotes AHR protein degradation without ligand treatment [33], unveiling the potential dynamics of AHR protein synthesis and degradation in maintaining the AHR cellular levels. However, the mechanisms responsible for the degradation of AHR without the addition of an exogenous ligand are largely uncharacterized.
Q18 (with the structure shown in Figure 1a) is a synthetic piperazinylpyrimidine molecule which was originally designed as a tyrosine kinase inhibitor [34]. Although there is no tyrosine kinase target identified thus far, Q18 appears to alter the AHR function. With the use of Q18, we serendipitously unveiled a new mechanism that degrades the AHR protein. Here, we provide evidence supporting that AHR is a CMA client protein which undergoes lysosomal degradation in triple-negative breast cancer (TNBC) cells. . Cells were treated with 10 μM of Q18 or DMSO (vehicle) for the indicated times. All lanes contain 60 μg of whole cell lysates. The amount of AHR was normalized by β-actin. The normalized value of AHR from Q18 treatment was divided by the normalized value from DMSO treatment to calculate the relative abundance of AHR at each timepoint. Y-axis represents the fold difference in Q18 over DMSO at each timepoint. Results are means ± SD of three independent experiments. One-way ANOVA with Dunnett's multiple comparisons test was used to determine statistical significance. ** p < 0.01, *** p < 0.001, **** p < 0.0001 when compared with zero timepoint in MDA-MB-468 and MDA-MB-231 cells. ns, not significant. MCF-7 and MDA-MB-361 data were all statistically insignificant when compared with the zero timepoint.

Suppression of the AHR Protein Levels in Triple-Negative, but Not in Non-Triple-Negative, Breast Cancer Cells by Q18
We treated two TNBC and two non-TNBC cells with 10 μM of Q18 for various time periods to determine its effect on the AHR protein levels. We observed that the AHR protein levels were significantly suppressed after Q18 treatment by 50% as early as 2 and 24 . Cells were treated with 10 µM of Q18 or DMSO (vehicle) for the indicated times. All lanes contain 60 µg of whole cell lysates. The amount of AHR was normalized by β-actin. The normalized value of AHR from Q18 treatment was divided by the normalized value from DMSO treatment to calculate the relative abundance of AHR at each timepoint. Y-axis represents the fold difference in Q18 over DMSO at each timepoint. Results are means ± SD of three independent experiments. One-way ANOVA with Dunnett's multiple comparisons test was used to determine statistical significance. ** p < 0.01, *** p < 0.001, **** p < 0.0001 when compared with zero timepoint in MDA-MB-468 and MDA-MB-231 cells. ns, not significant. MCF-7 and MDA-MB-361 data were all statistically insignificant when compared with the zero timepoint.

Results
2.1. Suppression of the AHR Protein Levels in Triple-Negative, but Not in Non-Triple-Negative, Breast Cancer Cells by Q18 We treated two TNBC and two non-TNBC cells with 10 µM of Q18 for various time periods to determine its effect on the AHR protein levels. We observed that the AHR protein levels were significantly suppressed after Q18 treatment by 50% as early as 2 and 24 h, respectively, in triple-negative MDA-MB-468 and MDA-MB-231 cells ( Figure 1B). This suppression was not observed in non-triple-negative MCF-7 cells within a 24 h treatment period ( Figure 1C, left). Although we observed a 50% suppression of the AHR protein levels by Q18 in non-triple-negative MDA-MB-361 cells, the effect was transient-the AHR levels were reverted back to normal in 4 h ( Figure 1C, right). Collectively, the AHR protein levels were uniquely suppressed by Q18 in TNBC cell lines.

Degradation of AHR in Triple-Negative Breast Cancer Cells Is Dependent on Progesterone Receptor but Not Estrogen Receptor
Since TNBC cells do not express estrogen receptor α (ERα) and progesterone receptor B (PR-B), we performed transient transfection experiments to alter the cellular ERα or PR-B levels in an effort to address whether the Q18 suppression of the AHR protein levels is ERα or PR-B dependent. We observed that the down-regulation of the PR-B levels in PR-B positive T47D cells using the PR-B-specific short hairpin RNA (shRNA) sensitized the Q18 suppression of the AHR protein levels (Figure 2A), supporting the notion that the absence of PR-B in MDA-MB-468 cells allows the downregulation of AHR by Q18 to occur. Conversely, the exogenous expression of ERα in MDA-MB-468 cells had no effect on the Q18-mediated AHR protein degradation ( Figure 2B), ruling out the involvement of ERα in this Q18 effect. Although it was very evident that Q18 failed to alter the AHR protein levels in MDA-MB-468 cells after the transient transfection of a PR-B-expressing plasmid ( Figure 2C, left), we were unable to detect any exogenous PR-B expression by Western blot analysis after transfection. In this experiment, we compared the AHR levels of cells transfected with a PR-B-expressing plasmid with cells transfected with the same plasmid, except that 75% of the PR-B cDNA was removed (which we referred to as the control). The transfected pr-b message levels were at a very low level, as reflected by a Cq number beyond 30 ( Figure 2C, right). Nevertheless, the endogenous pr-b message levels were consistently two orders of magnitude lower than the transfected levels at 24 h after transfection. It is conceivable that very small amounts of PR-B expression in MDA-MB-468 cells would be sufficient to suppress the Q-18-mediated AHR degradation.

Q18 is an AHR Antagonist
Next, we addressed whether Q18 could act as an AHR ligand and, in turn, cause AHR protein degradation. We treated the rat H4G1.1c3 cells stably transfected with a plasmid carrying a DRE-driven GFP cDNA with 1 µM of Q18-a 10-fold lower concentration than what was used for the MDA-MB-468 cell study, since 10 µM of Q18 caused significant H4G1.1c3 cell death. We detected minimal GFP expression when comparing with the treatment of an AHR ligand 3MC at a 1 µM concentration for 12 h, suggesting that Q18 is not an AHR ligand and cannot activate the DRE-driven gene transcription ( Figure 3A, top panel). When we pre-treated the cells with 1 µM of Q18 two hours before the 1 µM 3MC treatment, we observed that there was noticeable reduction in GFP fluorescence in the presence of Q18, suggesting that Q18 acts as an AHR antagonist, analogous to the AHR antagonist CH223191 ( Figure 3A, middle and bottom panels). Q18 did not alter the AHR protein levels in H4G1.1c3 cells ( Figure 3B), suggesting that the lack of GFP expression in the presence of Q18 in H4G1.1c3 cells is not due to a lesser amount of AHR to activate the GFP gene transcription. Next, we examined whether Q18 could cause the ligand-dependent AHR gel shift formation. We used the Pichia-expressed human AHR and ARNT [35] to form the AHR gel shift complex, which was βNF-, AHR-, and ARNT-dependent ( Figure 3C, lanes 1-4). Co-chaperone p23 was used to promote the gel shift formation [36]. We observed that, unlike an AHR ligand βNF, Q18 was not able to generate the AHR gel shift complex at a 10 µM concentration ( Figure 3C, lane 12), consistent with our rat H4G1.1c3 cell study that showed that Q18 is not an AHR ligand. Q18 could, however, inhibit the formation of the βNF-dependent AHR gel shift complex at a 50 µM concentration, although higher concentrations did not show more inhibition ( Figure 3C, lanes 8-10). The AHR antagonist CH223191 was used as the control to show the concentration-dependent inhibition of the βNF-dependent AHR gel shift complex formation ( Figure 3C, lanes 5-7). Collectively, data from the DRE-driven GFP expression and gel shift studies supported that Q18 is an AHR antagonist which competes with an AHR ligand in binding to the ligand-binding domain of AHR.
fected with a PR-B-expressing plasmid with cells transfected with the same plasmid, except that 75% of the PR-B cDNA was removed (which we referred to as the control). The transfected pr-b message levels were at a very low level, as reflected by a Cq number beyond 30 (Figure 2c, right). Nevertheless, the endogenous pr-b message levels were consistently two orders of magnitude lower than the transfected levels at 24 h after transfection. It is conceivable that very small amounts of PR-B expression in MDA-MB-468 cells would be sufficient to suppress the Q-18-mediated AHR degradation. showing that the amount of all pr-b messages were all very low so that no statistics was performed.

Q18 is an AHR Antagonist
Next, we addressed whether Q18 could act as an AHR ligand and, in turn, cause AHR protein degradation. We treated the rat H4G1.1c3 cells stably transfected with a plasmid carrying a DRE-driven GFP cDNA with 1 μM of Q18-a 10-fold lower concentration than what was used for the MDA-MB-468 cell study, since 10 μM of Q18 caused significant after Srf I and Eco RV digest to remove 75% of PR-B cDNA (control). All lanes for Western blot analysis contain 60 µg of whole cell lysates. All Western blot (A-C) data of AHR were normalized by total protein stain. All the results are means ± SD of three independent experiments. Two-way ANOVA with Sidak's multiple comparisons test was used in 2A for statistical analysis. * p < 0.05, *** p < 0.001 when compared between PR-B shRNA and scramble shRNA. Unpaired t test was used in 2B and 2C for statistical analysis. ns, not significant. * p < 0.05 when compared to control. formation of the βNF-dependent AHR gel shift complex at a 50 μM concentration, although higher concentrations did not show more inhibition (Figure 3c, lanes 8-10). The AHR antagonist CH223191 was used as the control to show the concentration-dependent inhibition of the βNF-dependent AHR gel shift complex formation (Figure 3c, lanes 5-7). Collectively, data from the DRE-driven GFP expression and gel shift studies supported that Q18 is an AHR antagonist which competes with an AHR ligand in binding to the ligand-binding domain of AHR. Figure 3. Effect of Q18 on AHR ligand binding. (A) Treatment of Q18, CH223191, and 3MC in rat H4G1.1c3 cells stably transfected with a plasmid carrying a DRE-driven GFP cDNA (left top diagram). Cells were treated for 12-14 h (as indicated in the left bottom timeline diagram) with DMSO, 1 μM 3MC, 1 μM Q18, 10 μM CH223191, 1 μM 3MC plus 1 μM Q18, or 1 μM 3MC plus 10 μM CH223191. Q18 could not cause GFP formation but could block the GFP formation caused by 3MC. CH223191 was used as the AHR antagonist control. Top panel represents one experiment; middle and bottom panels represent another experiment. These two experiments were repeated two more times with similar results. (B) Western blot analysis to show the effect of Q18 on the AHR protein levels when the stable H4G1.1c3 cells were treated with 1 μM of Q18 for 24 h. The AHR protein levels were normalized by total protein stain. The error bar represents the mean ± SD of three independent experiments and the representative Western images are shown. (C) Effect of Q18 on the formation of the AHR/ARNT/DRE gel shift complex. Gel shift complex was formed in the presence of 10 μM of βNF as the AHR ligand (lanes 1-4). A total of 10 μM of Q18 could not form the AHR gel shift complex but could suppress the βNF-dependent AHR gel shift formation at a 50 μM concentration (lanes 8-10 and 12). CH223191 (CH) was used as the AHR antagonist control (lanes 5-7 and 11). Arrows indicate the AHR/ARNT/DRE gel shift complex and free DRE conjugated with IRD700 (free probe). This gel shift experiment was repeated once with similar results. ns, not significant. . Cells were treated for 12-14 h (as indicated in the left bottom timeline diagram) with DMSO, 1 µM 3MC, 1 µM Q18, 10 µM CH223191, 1 µM 3MC plus 1 µM Q18, or 1 µM 3MC plus 10 µM CH223191. Q18 could not cause GFP formation but could block the GFP formation caused by 3MC. CH223191 was used as the AHR antagonist control. Top panel represents one experiment; middle and bottom panels represent another experiment. These two experiments were repeated two more times with similar results. (B) Western blot analysis to show the effect of Q18 on the AHR protein levels when the stable H4G1.1c3 cells were treated with 1 µM of Q18 for 24 h. The AHR protein levels were normalized by total protein stain. The error bar represents the mean ± SD of three independent experiments and the representative Western images are shown. (C) Effect of Q18 on the formation of the AHR/ARNT/DRE gel shift complex. Gel shift complex was formed in the presence of 10 µM of βNF as the AHR ligand (lanes 1-4). A total of 10 µM of Q18 could not form the AHR gel shift complex but could suppress the βNF-dependent AHR gel shift formation at a 50 µM concentration (lanes 8-10 and 12). CH223191 (CH) was used as the AHR antagonist control (lanes 5-7 and 11). Arrows indicate the AHR/ARNT/DRE gel shift complex and free DRE conjugated with IRD700 (free probe). This gel shift experiment was repeated once with similar results. ns, not significant.

Q18 Promotes the Autophagy-Mediated AHR Protein Degradation and Does Not Affect the AHR Transcript Levels
Next, we investigated whether the suppression of the AHR protein levels by Q18 is caused by a reduction in the AHR transcript and/or the promotion of the AHR protein degradation. We observed that there was no statistically significant difference in the AHR transcript levels in the presence of 10 µM of Q18 in MDA-MB-468 cells when comparing each timepoint (2-24 h) with the zero timepoint ( Figure 4A, left). We performed an actinomycin D experiment to determine whether Q18 could increase the degradation of the AHR transcript in MDA-MB-468 cells. We observed that the AHR transcript was degraded similarly in the presence or absence of Q18 when transcription was inhibited by actinomycin D (Figure 4A, right). Collectively, our data suggested that the suppression of the AHR protein levels by Q18 is not mediated through transcriptional and RNA stability mechanisms. Next, we measured the AHR protein half-life in MDA-MB-468 cells when protein synthesis was inhibited by 30 µg/mL of cycloheximide. We observed that AHR protein was degraded significantly faster in the presence of Q18 ( Figure 4B), supporting the notion that Q18 triggers AHR protein degradation. This Q18-dependent AHR protein degradation was not affected by the proteasome inhibitor MG132, since the presence of MG132 did not significantly change the AHR protein half-life ( Figure 4C), suggesting that proteasomal degradation is not involved in the Q18-mediated AHR degradation. On the contrary, the degradation of the AHR protein by Q18 was retarded significantly in the presence of the autophagy inhibitor chloroquine ( Figure 4D), showing that Q18 causes the lysosomal degradation of AHR.
Next, we investigated whether the suppression of the AHR protein levels by Q18 is caused by a reduction in the AHR transcript and/or the promotion of the AHR protein degradation. We observed that there was no statistically significant difference in the AHR transcript levels in the presence of 10 μM of Q18 in MDA-MB-468 cells when comparing each timepoint (2-24 h) with the zero timepoint (Figure 4a, left). We performed an actinomycin D experiment to determine whether Q18 could increase the degradation of the AHR transcript in MDA-MB-468 cells. We observed that the AHR transcript was degraded similarly in the presence or absence of Q18 when transcription was inhibited by actinomycin D (Figure 4a, right). Collectively, our data suggested that the suppression of the AHR protein levels by Q18 is not mediated through transcriptional and RNA stability mechanisms. Next, we measured the AHR protein half-life in MDA-MB-468 cells when protein synthesis was inhibited by 30 μg/mL of cycloheximide. We observed that AHR protein was degraded significantly faster in the presence of Q18 (Figure 4b), supporting the notion that Q18 triggers AHR protein degradation. This Q18-dependent AHR protein degradation was not affected by the proteasome inhibitor MG132, since the presence of MG132 did not significantly change the AHR protein half-life (Figure 4c), suggesting that proteasomal degradation is not involved in the Q18-mediated AHR degradation. On the contrary, the degradation of the AHR protein by Q18 was retarded significantly in the presence of the autophagy inhibitor chloroquine (Figure 4d), showing that Q18 causes the lysosomal degradation of AHR.   Results are means ± SD of: n = 3 for DMSO group; n = 4 for CHX plus Q18 and CHX plus Q18 plus CQ group. n represents the number of independent experiments. Two-way ANOVA with Tukey's multiple comparisons test was used in statistical analysis. *** p < 0.001 compared between CHX plus Q18 and CHX plus Q18 plus CQ-treated cells. Results are means ± SD of: n = 3 for DMSO group; n = 4 for CHX plus Q18 and CHX plus Q18 plus CQ group. n represents the number of independent experiments. Two-way ANOVA with Tukey's multiple comparisons test was used in statistical analysis. *** p < 0.001 compared between CHX plus Q18 and CHX plus Q18 plus CQ-treated cells.

Degradation of AHR in Triple-Negative Breast Cancer Cells Is LAMP2A-Mediated
Since chloroquine blocked the Q18-mediated AHR protein degradation, we investigated whether LAMP2A, an essential protein responsible for importing client protein into lysosomes for CMA [37], is involved in the degradation of AHR. We used a LAMP2 antibody from Santa Cruz Biotechnology (H4B4, sc-18822) which detects LAMP2A in a broad region. This antibody has been used by other researchers to detect LAMP2A expression [38]. We observed that LAMP2A could be down-regulated in MDA-MB-468 cells via the transient transfection of a plasmid carrying the LAMP2A-specific antisense cDNA. When the LAMP2A expression was suppressed in MDA-MB-468 cells, the degradation of the AHR protein was significantly reversed from 11-13% to 46-55% content up to 8 h of Q18 treatment ( Figure 5A). During this 8 h Q18 treatment of the empty plasmid transfected cells, the amount of the LAMP2A protein increased while AHR was substantially reduced, suggesting that Q18 may upregulate LAMP2A, leading to AHR degradation. Next, we addressed whether the accumulated LAMP2A could interact with AHR. We performed a co-immunoprecipitation experiment and observed that AHR was co-precipitated with LAMP2A and vice versa in a time-dependent manner when cells were treated with Q18 ( Figure 5B). In addition, the activation of CMA by 6-AN or starvation (via HBSS treatment) also upregulated the LAMP2A expression and degraded AHR in MDA-MB-468 cells, although the timing of the AHR suppression varied between these two treatments ( Figure 5C). However, a reduction in the AHR levels by GA, a reported CMA activator, did not show any apparent difference in the LAMP2A protein levels. Results from the Western blot analysis showed that CQ could significantly block the 6-AN-dependent AHR degradation in the cycloheximide experiment ( Figure 5D). AHR might be partially degraded by 26S proteasome after 6-AN treatment, since MG132 was able to reverse some of the AHR degradation, but this effect was not statistically significant ( Figure 5D). On the contrary, the treatment of MDA-MB-468 cells with GA showed a more pronounced degradation of AHR which was not reversed by either CQ or MG132 ( Figure 5E).

Treatment of Q18, 6-AN or Starvation in MDA-MB-468 Cells Causes Interaction between AHR and LAMP2A in Proximity Ligation Assay
Next, we examined whether the activation of CMA in MDA-MB-468 cells could induce the AHR-LAMP2A interaction using a proximity ligation assay to detect interaction in situ. GAPDH, which is a CMA client protein [39], was used as a positive control to show the interaction between LAMP2A and client protein during CMA. The treatment of Q18 increased the interaction between GAPDH and LAMP2A, showing that Q18 triggers CMA in MDA-MB-468 cells after 8 h of treatment ( Figure 5F). Likewise, the interaction between AHR and LAMP2A was noticeably increased when the cells were treated with Q18 for 8 h. Other means of CMA activation by either 6-AN or starvation also showed a clear increase in the AHR-LAMP2A interaction at 24 and 48 h, respectively ( Figure 5G). CMA activation by 6-AN and starvation was confirmed using the GADPH control. However, we failed to detect any interaction between AHR and LAMP2A when MDA-MB-468 cells were treated with 1 µg/mL of GA for 24 h ( Figure 5H).

Treatment of Q18 Increases the LAMP2A Expression in TNBC but Not in Non-TNBC Cells
Next, we examined whether the LAMP2A expression could be upregulated by Q18 to cause AHR degradation. We observed that Q18 treatment for up to 48 h increased the LAMP2A expression in TNBC (MDA-MB-468 and MDA-MB231) but not in non-TNBC (MCF-7 and T47D) cells ( Figure 5I). PR-B expression was detected in non-TNBC, which was not affected by Q18 treatment. The expression of PR-B was not detected in TNBC cells.

6-AN Promotes AHR-HSC70 and AHR-LAMP2A Interaction in MDA-MB-468 Cells
It has been reported that HSC70 escorts cargo to LAMP2A during CMA [3]. Therefore, we examined whether AHR might interact with HSC70 in a manner that is dependent upon CMA activation. We performed co-immunoprecipitation experiments using anti-AHR SA-210 in MDA-MB-468 cells and observed that AHR interacted with HSC70 with or without CMA activation by Q18 or 6-AN. However, either 10 µM of Q18 or 6-AN significantly enhanced the AHR-HSC70 interaction by 5-fold and 1.25-fold, respectively ( Figure 6A,B), showing that the activation of CMA can increase the interaction between AHR and HSC70. Similar to the earlier data showing that Q18 triggered the AHR-LAMP2 interaction ( Figure 5B), the activation of CMA by 6-AN showed a small (but significant) quantifiable increase in the interaction between AHR and LAMP2 ( Figure 6B). . The Western bands were detected using proteinspecific antibodies (SA-210 for AHR, B-6 for HSC70, and H4B4 for LAMP2A). Input represents 1% of sample to start the experiment. One control (0 h for A and DMSO for B) of each data set was arbitrarily set as one to normalize the three independent experimental data sets, thus no error bar was presented for these conditions. Results are means ± SD of three independent experiments. Two-tailed unpaired t test with Welch's correction were used for statistical analysis. * p < 0.05 when compared to the control. ns, not significant.

AHR Contains a CMA Signature Motif
Human AHR contains three putative sequences-QKTVK, QDVIN, and NEKFFwhich follow the general requirements of a CMV signature motif (Figure 7a). When we transiently expressed the GFP fusion of the full-length human AHR in MDA-MB-468 cells, we observed that the GFP-AHR fusion was degraded significantly within six hours of the Q18 treatment (Figure 7b), validating that the GFP fusion of AHR has the same fate as the endogenous AHR when treated with Q18. However, when we monitored the levels of the GFP fusion of an AHR construct (amino acid 1-295 of human AHR, C∆553) which contains two out of the three putative signature motifs; we observed that it remained stable up to 8 h of Q18 treatment in MDA-MB-468 cells (Figure 7c), suggesting that the two putative motifs close to the bHLH domain are not used to trigger AHR degradation. Next, we generated GFP-AHR mutants E559A/F561A and E559A/F561L, which changed the NEKFF motif into NAKAF and NAKLF, respectively. Our transient transfection results revealed that both NAKAF and NAKLF mutants were more resistant to degradation up to 24 h when compared to the NEKFF-containing GFP-AHR in a statistically significant manner, supporting that NEKFF is a KFERQ-like motif that is responsible for AHR undergoing CMA (Figure 7d). . The Western bands were detected using protein-specific antibodies (SA-210 for AHR, B-6 for HSC70, and H4B4 for LAMP2A). Input represents 1% of sample to start the experiment. One control (0 h for A and DMSO for B) of each data set was arbitrarily set as one to normalize the three independent experimental data sets, thus no error bar was presented for these conditions. Results are means ± SD of three independent experiments. Two-tailed unpaired t test with Welch's correction were used for statistical analysis. * p < 0.05 when compared to the control. ns, not significant.

AHR Contains a CMA Signature Motif
Human AHR contains three putative sequences-QKTVK, QDVIN, and NEKFFwhich follow the general requirements of a CMV signature motif ( Figure 7A). When we transiently expressed the GFP fusion of the full-length human AHR in MDA-MB-468 cells, we observed that the GFP-AHR fusion was degraded significantly within six hours of the Q18 treatment ( Figure 7B), validating that the GFP fusion of AHR has the same fate as the endogenous AHR when treated with Q18. However, when we monitored the levels of the GFP fusion of an AHR construct (amino acid 1-295 of human AHR, C∆553) which contains two out of the three putative signature motifs; we observed that it remained stable up to 8 h of Q18 treatment in MDA-MB-468 cells ( Figure 7C), suggesting that the two putative motifs close to the bHLH domain are not used to trigger AHR degradation. Next, we generated GFP-AHR mutants E559A/F561A and E559A/F561L, which changed the NEKFF motif into NAKAF and NAKLF, respectively. Our transient transfection results revealed that both NAKAF and NAKLF mutants were more resistant to degradation up to 24 h when compared to the NEKFF-containing GFP-AHR in a statistically significant manner, supporting that NEKFF is a KFERQ-like motif that is responsible for AHR undergoing CMA ( Figure 7D). Results are means ± SD of three independent experiments. One-way ANOVA with Dunnett's multiple comparisons test (bar graph) and two-way ANOVA with Sidak's multiple comparisons test (line graph) were used for statistical analysis. * p < 0.05, ** p < 0.01, and *** p < 0.001 when compared with zero timepoint. ns, not significant. There is no difference between GFP-AHR and endogenous AHR degradation. Bottom diagram and top left images show the degradation of the transfected GFP-AHR upon Q18 treatment in MDA-MB-468 cells. The amount of AHR and GFP-AHR was normalized by total protein stain, whereas each lane contained 60 μg of protein loaded. (C) MDA-MB-468 cells were transiently transfected with pEGFP-C2-C∆553 or pEGFP-C2 plasmid (empty vector, EV). Cells were treated with 10 μM of Q18 or DMSO (vehicle) for 0-8 h. The graph represents the normalized amount of GFP-CΔ553 at 4 and 8 h after Q18 or DMSO treatment with the amount at zero hour set as one. Results are means ± SD of three independent experiments. Unpaired t test was used for statistical analysis. There is no difference between Q18 and vehicle treatment. EGFP-CΔ553 was stable in the presence of Q18 for up to 8 h. (D) MDA-MB-468 cells were transiently transfected with pGFP 2 -N2-AHR (NEKFF, black), pGFP 2 -N2-AHR E559A/F561A (NAKAF, blue), or pGFP 2 -N2-AHR E559A/F561L (NAKLF, red) and then treated with 10 μM of Q18 for 0, 8, or 24 h. Both NAKAF and NAKLF mutants were statistically more stable than GFP-AHR with NEKFF in the presence of Q18. Results are means ± SD (with error bars of above half only) of five independent experiments, except that the 24 h transfection of pGFP 2 -N2-AHR (black) and pGFP 2 -N2-AHR E559A/F561L (red) was from six independent experiments. The table represents the mean values ± SD with zero timepoints set as one for comparison. Two-way ANOVA with Turkey's multiple comparisons test was used for statistical analysis. * p < 0.05, ** p < 0.01, and *** p < 0.005 when compared with pGFP 2 -N2-AHRtransfected cells.

Discussion
Results from the high-throughput gene expression screening revealed that the ahr and cyp1a1 transcripts were upregulated by 2.5-and 8-fold, respectively, in MDA-MB-468 cells after treatment with 10 μM of Q18 for 24 h (unpublished data). We were able to confirm the CYP1A1 upregulation ( Figure 8a); however, no upregulation of AHR at the transcript level has been observed within the 24 h treatment period of Q18 (Figure 4a). To our surprise, the AHR protein is rapidly reduced upon Q18 treatment. Even though it is un- Results are means ± SD of three independent experiments. One-way ANOVA with Dunnett's multiple comparisons test (bar graph) and two-way ANOVA with Sidak's multiple comparisons test (line graph) were used for statistical analysis. * p < 0.05, ** p < 0.01, and *** p < 0.001 when compared with zero timepoint. ns, not significant. There is no difference between GFP-AHR and endogenous AHR degradation. Bottom diagram and top left images show the degradation of the transfected GFP-AHR upon Q18 treatment in MDA-MB-468 cells. The amount of AHR and GFP-AHR was normalized by total protein stain, whereas each lane contained 60 µg of protein loaded. (C) MDA-MB-468 cells were transiently transfected with pEGFP-C2-C∆553 or pEGFP-C2 plasmid (empty vector, EV). Cells were treated with 10 µM of Q18 or DMSO (vehicle) for 0-8 h. The graph represents the normalized amount of GFP-C∆553 at 4 and 8 h after Q18 or DMSO treatment with the amount at zero hour set as one. Results are means ± SD of three independent experiments. Unpaired t test was used for statistical analysis. There is no difference between Q18 and vehicle treatment. EGFP-C∆553 was stable in the presence of Q18 for up to 8 h. (D) MDA-MB-468 cells were transiently transfected with pGFP 2 -N2-AHR (NEKFF, black), pGFP 2 -N2-AHR E559A/F561A (NAKAF, blue), or pGFP 2 -N2-AHR E559A/F561L (NAKLF, red) and then treated with 10 µM of Q18 for 0, 8, or 24 h. Both NAKAF and NAKLF mutants were statistically more stable than GFP-AHR with NEKFF in the presence of Q18. Results are means ± SD (with error bars of above half only) of five independent experiments, except that the 24 h transfection of pGFP 2 -N2-AHR (black) and pGFP 2 -N2-AHR E559A/F561L (red) was from six independent experiments. The table represents the mean values ± SD with zero timepoints set as one for comparison. Two-way ANOVA with Turkey's multiple comparisons test was used for statistical analysis. * p < 0.05, ** p < 0.01, and *** p < 0.005 when compared with pGFP 2 -N2-AHR-transfected cells.

Discussion
Results from the high-throughput gene expression screening revealed that the ahr and cyp1a1 transcripts were upregulated by 2.5-and 8-fold, respectively, in MDA-MB-468 cells after treatment with 10 µM of Q18 for 24 h (unpublished data). We were able to confirm the CYP1A1 upregulation ( Figure 8A); however, no upregulation of AHR at the transcript level has been observed within the 24 h treatment period of Q18 ( Figure 4A). To our surprise, the AHR protein is rapidly reduced upon Q18 treatment. Even though it is unclear how CYP1A1 can be upregulated by Q18 while the AHR expression is actually suppressed, Q18 is nevertheless an effective tool to eventually help us to unveil how TNBC cells regulate the AHR protein levels without exogenous ligand treatment. AHR and its partner ARNT have been linked to autophagy. AHR levels are inversely proportional to the autophagy potential in non-small cell lung cancer cells; the activation of AHR causes less metastatic potential by downregulating autophagy, resulting in lesser EMT progression [40]. However, the activation of AHR in another cell type may trigger macroautophagy to repair cellular damage caused by external agents. For example, researchers have shown that particulate matter induced LC3/p62-mediated macroautophagy by activating AHR in keratinocytes [41]. It has been reported that autophagy may protect the liver against alcohol-induced damage, which, interestingly, is ARNT-dependent, since ARNT knockout mice were resistant to such damage [42]. Recently, we reported that AHR undergoes LC3-mediated autophagy in HeLa cells [43]. However, in contrast to what we observed here in MDA-MB-468 cells, the activation of CMA does not degrade AHR in HeLa cells. Interestingly our data support the selective degradation of AHR via CMA in TNBC cells, which adds another layer of complexity to these vital mechanisms involving autophagy.
clear how CYP1A1 can be upregulated by Q18 while the AHR expression is actually suppressed, Q18 is nevertheless an effective tool to eventually help us to unveil how TNBC cells regulate the AHR protein levels without exogenous ligand treatment. AHR and its partner ARNT have been linked to autophagy. AHR levels are inversely proportional to the autophagy potential in non-small cell lung cancer cells; the activation of AHR causes less metastatic potential by downregulating autophagy, resulting in lesser EMT progression [40]. However, the activation of AHR in another cell type may trigger macroautophagy to repair cellular damage caused by external agents. For example, researchers have shown that particulate matter induced LC3/p62-mediated macroautophagy by activating AHR in keratinocytes [41]. It has been reported that autophagy may protect the liver against alcohol-induced damage, which, interestingly, is ARNT-dependent, since ARNT knockout mice were resistant to such damage [42]. Recently, we reported that AHR undergoes LC3-mediated autophagy in HeLa cells [43]. However, in contrast to what we observed here in MDA-MB-468 cells, the activation of CMA does not degrade AHR in HeLa cells. Interestingly our data support the selective degradation of AHR via CMA in TNBC cells, which adds another layer of complexity to these vital mechanisms involving autophagy. . RT-qPCR was performed as described in the Materials and Methods section. The fold changes of gene transcript levels between Q18 treated and DMSO vehicle control treated cells were normalized by 18S. One-way ANOVA with Dunnett's multiple comparisons test was used in analyzing the Q18 treatment data. **** p < 0.0001 showed that there is a significant difference in the cyp1a1 mRNA levels at the 24 h timepoint versus zero timepoint. The error bars represent means ± SD, n = 3. (B) Western blot analysis showing the effect of CH223191 on the AHR protein levels in MDA-MB-468 cells. Cells were treated with 10 μM of CH223191, 10 μM of Q18, or DMSO (vehicle) for 24 h. All lanes contain 60 μg of whole cell lysates. The amount of AHR was normalized by total protein stain. The normalized value of AHR from the treatment group was divided by the normalized value from DMSO treatment to calculate the relative abundance of AHR. The error bar represents mean ± SD of three independent experiments.
One of the first questions we asked was whether Q18 degrades AHR via the wellestablished mechanism involving the proteasomal degradation of AHR after ligand binding [31]. Our data revealed that Q18 is not an AHR ligand; rather, it is an AHR antagonist. However, this Q18 effect on the AHR degradation is not mediated through its antagonistic activity since CH223191, unlike Q18, did not alter the AHR protein levels in MDA-MB-468 cells up to 24 h of treatment (Figure 8b). One-way ANOVA with Dunnett's multiple comparisons test was used in analyzing the Q18 treatment data. **** p < 0.0001 showed that there is a significant difference in the cyp1a1 mRNA levels at the 24 h timepoint versus zero timepoint. The error bars represent means ± SD, n = 3. (B) Western blot analysis showing the effect of CH223191 on the AHR protein levels in MDA-MB-468 cells. Cells were treated with 10 µM of CH223191, 10 µM of Q18, or DMSO (vehicle) for 24 h. All lanes contain 60 µg of whole cell lysates. The amount of AHR was normalized by total protein stain. The normalized value of AHR from the treatment group was divided by the normalized value from DMSO treatment to calculate the relative abundance of AHR. The error bar represents mean ± SD of three independent experiments.
One of the first questions we asked was whether Q18 degrades AHR via the wellestablished mechanism involving the proteasomal degradation of AHR after ligand binding [31]. Our data revealed that Q18 is not an AHR ligand; rather, it is an AHR antagonist. However, this Q18 effect on the AHR degradation is not mediated through its antagonistic activity since CH223191, unlike Q18, did not alter the AHR protein levels in MDA-MB-468 cells up to 24 h of treatment ( Figure 8B).
In addition to the early discovery that client proteins such as RNase A and GAPDH are preferentially uptaken into lysosomes for degradation via the CMA mechanism [42], hypoxia inducible factor 1α (HIF-1α), a PAS protein-like AHR, has been reported as a CMA client protein [44,45]. Carboxyl terminus of HSC70-interacting protein (CHIP, aka STUB1), a co-chaperone E3 ligase [46,47], forms the K63-linked ubiquitinated HIF1α, which in turn triggers the degradation of HIF-1α via the LAMP2A-mediated CMA mechanism [38,48]. Interestingly, CHIP has been shown (more than a decade ago) to degrade the unliganded AHR, and this degradation can be protected in the presence of XAP2 [49]. The down-regulation of CHIP in human liver Hepa-1c1c7 cells, surprisingly, does not alter the AHR protein levels or function [50], possibly due to the cell-specific involvement of CHIP in the regulation of the AHR protein levels. This is consistent with our finding that the degradation of AHR is mediated through CMA in TNBC cells but not in non-TNBC and HeLa cells.
It is commonly known that MDA-MB-468 and MDA-MB-231 cells are ER-, PR-, and HER2-negative, whereas MCF-7 and T47D cells are ER-and PR-positive but HER2-negative. MDA-MB-361 cells, on the other hand, are ER-and HER2-positive but might express PR-B [51,52], which might explain our observation that Q18 reduces the AHR levels in MDA-MB-361 cells initially but that the levels can be quickly reverted back, since the Q18 effect is PR-B-sensitive. Q18 triggers AHR protein degradation in triple-negative but not in non-triple-negative breast cancer cells, partly because PR-B expression blocks autophagy-dependent AHR degradation. It appears that PR-B activity prevents AHR from CMA-mediated degradation-this hypothesis is consistent with the literature that progesterone increases the AHR protein levels in the uterus of rats [53] and rabbits [54]. It was reported that the activation of the PR-B signaling suppresses the AHR function in T47D cells [55]. This crosstalk should be more complicated, since PR-B can stabilize the AHR protein levels, which might contribute to any sustained action of AHR in T47D cells. Since the expression levels of LAMP2A are inversely proportional to PR-B ( Figure 5I), it would be intriguing to examine whether PR-B can act as a transcriptional suppressor of the LAMP2A gene transcription or as a modulator of the LAMP2A protein half-life.
To validate whether the activation of CMA is the cause of AHR protein degradation, we subjected MDA-MB-468 cells to a condition where CMA is triggered-the treatment of 6-AN [56] or starvation [37,39,57]. The results from our proximity ligation assay prove that AHR interacts with LAMP2A in situ when CMA is activated. In all cases, with the treatment of Q18, 6-AN, or HBSS (to mimic starvation), the LAMP2A expression is enhanced, followed by an increase in the AHR-LAMP2A interaction, leading to AHR protein degradation, which can be reversed by an autophagy inhibitor chloroquine. Our data are consistent with other researchers' observation that the treatment of Hep3B cells with digoxin, which appears to increase lysosomal degradation, increases the LAMP2A expression [45]. Increased LAMP2A expression at the lysosomal membrane is likely due to a decrease in the LAMP2A protein degradation and/or the shuttling of LAMP2A from lysosomal cytosol to membrane [58]. In addition, 6-AN and Q18 enhance the interaction between HSC70 and AHR, supporting the notion that HSC70 escorts AHR to LAMP2A for CMA. Although GA has been shown as a CMA activator after 24 h of treatment at 2 µM concentration using isolated lysosomes [56], we were not able to see any activation of autophagy at that concentration-there was no increase in the LAMP2A expression and the AHR protein levels could not be restored by chloroquine after GA treatment in MDA-MB-468 cells. We cannot rule out the possibility that the timing and dose should be adjusted if cells rather than isolated lysosomes are used for the study. Nevertheless, it is interesting that the AHR protein levels were pronouncedly suppressed after 24 h of GA treatment and neither chloroquine nor MG132 could restore the levels. It is well known that GA causes the ligand-independent degradation of AHR via the 26S proteasome and that this degradation can occur within 2 h of GA treatment [32]. It is conceivable that AHR in MDA-MB-468 cells might be degraded to a great extent shortly after GA treatment and remain at low levels for up to 24 h. Thus, exposing cells with chloroquine or MG132 at the last 8 h of the 24 h GA treatment window did not elicit any inhibitory effect on degradation. Collectively, our data strongly support the notion that TNBC cells use CMA to degrade AHR selectively and that Q18 upregulates CMA to degrade AHR. Interestingly, this Q18dependent CMA activation is PR-B dependent and thus is not observed in non-TNBC cells, revealing potentially important crosstalk between the PR-B signaling and autophagy mechanisms in breast cancer cells.
A pentapeptide sequence KFERQ was discovered to be the signature motif of CMA client proteins [6]. The general requirement of the pentapeptide signature motif is as follows: (1) one acid residue (D, E) with two basic (K, R) and one hydrophobic (F, V, L, I) residues or (2) one acid residue (D, E) with one basic (K, R) and two hydrophobic (F, V, L, I) residues. Both scenarios should have a Q or N as the first or last residue of the pentapeptide. This signature motif is thought to involve in the HSC70-mediated interactions that brings client proteins into lysosomes for degradation. Interestingly, human AHR contains putative CMA motifs-one putative signature sequence at amino acid 558 (NEKFF) and two quasi sequences at amino acid 20 (QKTVK) and amino acid 59 (QDVIN). The deletion construct of AHR (amino acid 1-295, EGFP-C∆553) lacking the NEKFF sequence is rather stable in MDA-MB-468 cells after transient transfection-this observation rules out the involvement of the two quasi sequences for CMA. On the other hand, when two of the five amino acids of NEKFF are mutated in the GFP fusion of AHR, GFP-AHR becomes more stable upon Q18 treatment, supporting the notion that NEKFF is a bona fide motif that is responsible for the CMA-mediated AHR degradation. This motif is remarkably similar to the human HIF-1α CMA motif (NEFKL) and, interestingly, both the human AHR and HIF-1α motif are not conserved in mouse [48]. In summary, we have provided supportive evidence to conclude that AHR is a CMA client protein which can be degraded selectively via CMA in TNBC cells. Mechanistic studies to better understand how cells utilize lysosomal degradation to regulate their AHR protein levels are warranted.

Co-Immunoprecipitation
Protein G magnetic beads (Thermo Fisher Scientific, Waltham, MA, USA), which had been preincubated with co-immunoprecipitation antibody for 30 min at room temperature, were incubated with whole cell lysates (2 mg) overnight at 4 • C with rotation. Protein G beads were then washed with the buffer (HEDG containing 150 mM NaCl and 0.05-0.1% Tween-20) three times at 4 • C. Western blot analysis was performed to visualize the results.

Proximity Ligation Assay
A proximity ligation assay was performed according to the protocol of Duolink Reagents Red (Sigma-Aldrich, St. Louis, MO, USA). Briefly, cells were fixed on glass slides and permeabilized using cold methanol and then incubated with corresponding rabbit and mouse antibodies for 1 h at 37 • C. Anti-rabbit PLUS and anti-mouse MINUS probes were added and incubated for another hour at 37 • C. Ligation (1:40 dilution, 30 min at 37 • C) and polymerization (1:80 dilution, 100 min at 37 • C) were then performed successively. Slides were then mounted before imaging with DAPI for 2 min at room temperature. Fluorescence images were acquired using a Keyence BZ-X700 fluorescence microscope.

Gel Shift Assay
Pichia expressed human AHR and ARNT were used to form the ligand-dependent AHR gel shift complex, as described previously [35], with modifications. In brief, 6His-AHR and 6His-ARNT expressed in 9 mL and 2 mL of Pichia culture, respectively, were enriched by metal affinity purification into 0.75 mL of HEDG except with 50% glycerol. 6His-human p23 from 600 mL of bacterial culture was likewise enriched into 1.5 mL of PBS and was used to promote the formation of the AHR gel shift complex [36]. AHR gel shift complex was formed by adding the AHR ligand βNF (10 µM) into HEDG containing AHR (2 µL), ARNT (0.5 µL), and p23 (1 µL), followed by addition of dIdC (3 µg) and DRE-IRD700 (0.5 pmol). A LI-COR CLx imager was used to visualize the results when electrophoresis was complete.

Statistical Analysis
All the data are reported as means ± SD and analyzed using Prism GraphPad version 8. Statistical analysis was determined by one-way or two-way ANOVA with multiple comparisons, or Student's t test. The number of asterisk indicates the p value of statistical significance as follows: p > 0.05 (ns, not significant), p < 0.05 (*), p < 0.01 (**), p < 0.001 (***), and p < 0.0001 (****).

Conclusions
Similar to 6-AN and starvation, Q18 promotes CMA in MDA-MB-468 cells. The activation of CMA in these cells triggers the degradation of the AHR protein selectively, since human AHR contains a CMA-like motif at amino acid 558-562 (NEKFF), which is essential for this degradation. This degradation is mediated through LAMP2A and appears to occur in breast cancer cells when PR-B is lacking.