Divergent effects of acute and chronic PPT1 inhibition in melanoma

ABSTRACT

Macroautophagy/autophagy-lysosome function promotes growth and survival of cancer cells, making them attractive targets for cancer therapy. One intriguing lysosomal target is PPT1 (palmitoyl-protein thioesterase 1). PPT1 inhibitors derived from chloroquine block autophagy, have significant antitumor activity in preclinical models and are being developed for clinical trials. However, the role of PPT1 in tumorigenesis remains poorly understood. Here we report that in melanoma cells, acute siRNA or pharmacological PPT1 inhibition led to increased ferroptosis sensitivity and significant loss of viability, whereas chronic PPT1 knockout using CRISPR-Cas9 produced blunted ferroptosis that led to sustained viability and growth. Each mode of PPT1 inhibition produced lysosome-autophagy inhibition but distinct proteomic changes, demonstrating the complexity of cellular adaptation mechanisms. To determine whether total genetic loss of Ppt1 would affect tumorigenesis in vivo, we developed a Ppt1 conditional knockout mouse model. We then crossed it into the BrafCA, PtenloxP, Tyr:CreERT2 melanoma mouse model to investigate the impact of Ppt1 loss on tumorigenesis. Loss of Ppt1 had no impact on melanoma histology, time to tumor initiation, or survival of tumor-bearing mice. These results suggest that chemical PPT1 inhibitors produce different adaptations than genetic PPT1 inhibition, and additional studies are warranted to fully understand the mechanism of chloroquine derivatives that target PPT1 in cancer.Abbreviations: 4-HT: 4-hydroxytamoxifen; BRAF: B-Raf proto-oncogene, serine/threonine kinase; cKO: conditional knockout; CRISPR-Cas9: clustered regularly interspaced short palindromic repeats-CRISPR-associated protein 9; DC661: A specific PPT1 inhibitor; DMSO: dimethyl sulfoxide; Dox; doxycycline hyclate; Easi-CRISPR: efficient additions with ssDNA inserts-CRISPR; GNS561/ezurpimtrostat: A PPT1 inhibitor; Hug: human guide; iCas: inducible CRISPR-Cas9; KO: knockout; LC-MS/MS: Liquid chromatography-tandem mass spectrometry; LDLR: low density lipoprotein receptor; NFE2L2/NRF2: NFE2 like bZIP transcription factor 2; NT: non-target; PTEN: phosphatase and tensin homolog; PPT1: palmitoyl-protein thioesterase 1; RSL3: RAS-selective lethal small molecule 3; SCRIB/SCRB1: scribble planar cell polarity protein; Tyr:CreERT2: tyrosinase-driven Cre recombinase fused with the tamoxifen-inducible mutant ligand binding domain of the human estrogen receptor; UGCG: UDP-glucose ceramide glucosyltransferase; WT: wild-type.

Introduction

Autophagy-lysosome function is often upregulated in cancer cells, promoting growth and survival [1–3]. Autophagy is the process by which cellular constituents are sequestered in autophagic vesicles and trafficked to the acidic environment of the lysosome for degradation [4]. Evidence suggests that lysosomal catabolism plays a crucial metabolic function in autophagy’s role in recycling macromolecules to fuel further growth [5]. The lysosomal enzyme PPT1 (palmitoyl-protein thioesterase 1) regulates the acidification of the lysosome by directing subcellular localization of palmitoylated vacuolar-type H⁺-ATPase subunits [6–8]. Mutations in PPT1 lead to the childhood lysosomal storage disorder neuronal ceroid lipofuscinosis/NCL, a rare neurodegenerative disorder that unfolds over many years [9]. In this disease, before neurodegeneration is clinically detectable, many years of normal development can be observed despite chronic loss of PPT1 enzymatic activity in every cell in the body. This observation supports the view that PPT1 could be a potentially druggable target in advanced cancer, especially with agents that do not cross the blood-brain barrier. Chloroquine derivatives such as hydroxychloroquine, DC661 [8], DQ661 [7], and GNS561/ezurpimtrostat [10] bind to and inhibit PPT1, and potently inhibit autophagy. A clinical trial of ezurpimtrostat (NCT05874414) is underway in advanced cholangiocarcinoma. Chemical inhibition of PPT1 produces significant loss of viability in cancer cells, yet in the DepMAP screen [11] PPT1 is not a dependency gene. Our previous work has demonstrated that chemical PPT1 inhibition with DC661, PPT1 knockdown with siRNA, and chronic knockout of PPT1 all produce significant lysosomal dysfunction and autophagy inhibition [7,8,12,13]. This raises the question of whether genetic PPT1 inhibition produces changes similar to chemical inhibition or whether adaptation allows cells to survive the stress of genetic PPT1 inhibition.

Melanoma has high levels of autophagy and lysosome function at baseline [14]. The melanosome, a key organelle for melanin production, is a lysosome-derived organelle [15]. Potent PPT1 inhibitors like DC661 produce significant viability loss in melanoma models [8,12]. Lysosomal inhibition augments BRAF inhibition in BRAF mutant melanoma [16]. Here, we report a comparison of the proteomes of chronic CRISPR knockout of PPT1, acute knockdown of PPT1, and chemical PPT1 inhibition in melanoma cell lines. We also report a novel conditional Ppt1 knockout (KO) mouse model. This model was generated using two different CRISPR-Cas9 targeting strategies. The conditional Ppt1 KO mouse was crossed into a genetically engineered mouse model/GEMM of melanoma. We chose the BrafCA, PtenloxP, Ppt1loxP, Tyr:CreERT2 melanoma [17] as a model because tumors can be induced in a spatially and temporally regulated manner, and tumors generated in this model phenocopy the histology and growth kinetics of human melanoma tumors. Acute chemical or genetic inhibition with siRNA of PPT1 produced significant loss of viability in human melanoma cell lines, but chronic PPT1 KO with CRISPR-Cas9 produced no loss of viability. A proteomic comparison of acute versus chronic genetic PPT1 inhibition found that these interventions produced strikingly different protein expression patterns and pathway alterations that suggested that iron metabolism and ferroptosis pathways play an important but complex role in adaptation to PPT1 inhibition. Finally, our conditional knockout mouse model demonstrated that complete genetic loss of Ppt1 did not affect tumor initiation, tumor growth rate or survival of mice bearing induced melanoma tumors.

Results

Acute but not chronic PPT1 inhibition leads to loss of viability in cancer cells

To compare the effects of constitutive PPT1 knockout and acute genetic or pharmacological PPT1 inhibition, we first generated A375P sgNon-Target (NT) and A375P sgPPT1 human guide (Hug) KO cell lines using CRISPR-Cas9 technology, resulting in PPT1-deficient cells as confirmed by western blot (Figure 1A) and enzymatic activity analysis [18] (Figure 1B). Constitutive chronic knockout of PPT1 did not significantly affect the growth of melanoma cells (Figure 1C). Knockout of PPT1 with doxycycline-inducible CRISPR-Cas9 system (Figure 1D), in which reduction of PPT1 protein levels was observed by day 3 of doxycycline treatment, also did not affect the growth of cells (Figure 1E), despite the loss of PPT1 enzymatic activity (Figure 1F). Next, we used siRNA to knock down PPT1 in A375P sgNT cells (Figure 1G). Acute PPT1 knockdown with siRNA produced a 90% reduction in PPT1 biochemical activity in melanoma cells (Figure 1H). Acute knockdown of PPT1 produced significant viability loss in melanoma cells (Figure 1I). Pharmacological PPT1 inhibition with DC661 [8] produced a significant loss of PPT1 enzymatic activity (Figure 1J) and loss of viability in melanoma cells (Figure 1K).

Figure 1.

Cytotoxicity observed by acute but not chronic PPT1 inhibition. (A) Immunoblot demonstrating PPT1 knockout in A375P sgPPT1 Hug1 and Hug3 clones compared to A375P sgNT cells. sg: small guide RNA; Hug: human guide; NT: non-target. (B) Van Diggelen PPT1 biochemical assay in the indicated cells. n = 3, repeated twice. ****P ≤ 0.0001; one way ANOVA test. (C) Cell growth at 4 days after plating the indicated cell lines at 1X10⁵. n = 3, repeated 3 times, ns: nonsignificant, one way ANOVA. (D) Immunoblot demonstrating PPT1 knockdown in A375P cells after doxycycline (Dox) induction of Cas9 expression. Cells with (+) or without (-) Dox treatment were collected on the indicated days after the start of treatment. (E) A375P iCas9 cell growth at 6 days after Dox treatment. (F) A375P iCas9 Van Diggelen PPT1 biochemical assay at 6 days after Dox treatment. (G) Immunoblot of A375P sgNT cells transfected with siNT or siPPT1 for 48 h. (H) Van Diggelen PPT1 biochemical assay in indicated cells 48 h after siRNA transfection. n = 3. (I) Cell growth 48 h after siRNA transfection, n = 3. (J) Van Diggelen PPT1 biochemical assay in A375P cells treated with 3 µm DC661 for 24 h, n = 3. (K) Cell number after treatment with 3 µm DC661 for 24 h. All cell growth assays and enzyme assays were performed in triplicate. *P ≤ 0.05; ****P ≤ 0.0001; one way ANOVA was used when more than 2 groups were compared (B, C, and F), two-tailed unpaired t test was used when 2 groups were compared (E, H-K).

Proteomic changes during acute versus chronic PPT1 inhibition

To elucidate the distinctions between chronic PPT1 knockout and acute genetic or chemical inhibition, we conducted an unbiased proteomic analysis using LC-MS/MS. The cells analyzed included A375P sgNT and A375P sgPPT1 Hug1, A375P sgPPT1 Hug3, A375P sgNT siNT, and A375P sgNT siPPT1. We confirmed the effective genetic suppression of PPT1 in these cells by mass spectrometry (Figure 2A). Unsupervised hierarchical clustering revealed different patterns in the proteomic profiles resulting from constitutive compared to acute PPT1 inhibition (Figure 2B). Of note, siNT cells had a different expression pattern than sgNT cells reflecting the stress experienced by cells during transfection. siNT and siPPT1 cells were more similar than sgNT or sgPPT1 cells. Using a defined threshold (absolute fold change ≥ 1.5 and adjusted p-value of ≤ 0.05), we found that constitutive PPT1 knockout (Hug1) produced 338 significantly increased proteins and 118 significantly decreased proteins compared to non-target control. Acute PPT1 knockdown produced 138 and 107 significantly increased or decreased proteins, respectively (Figure 2C). The largest significant protein changes observed in A375P sgNT siPPT1 and A375P sgPPT1 samples are shown in Tables 1 and 2. We next compared the proteome changes induced by genetic inhibition of PPT1 (A375P sgNT, A375P sgPPT1 Hug1, A375P sgNT siNT, and A375P sgNT siPPT1) to a previously published proteomic dataset of A375P cells treated with DMSO or DC661 3 µM [12,13]. In this comparison, chemical PPT1 inhibition induced 182 and 46 significantly increased and decreased proteins, respectively (Figure 2D). Strikingly, comparing significantly upregulated proteins across the three conditions (constitutive PPT1 KO, acute PPT1 knockdown, and chemical PPT1 inhibition), only 6 proteins (ADGRG1, CD74, IL13RA2, ITM2C, NEU1, and TPRG1L) were commonly increased (Figure 2E). Only 42 proteins were significantly elevated both in acute and chronic genetic PPT1 inhibition conditions. Only 24 proteins were upregulated in common between acute chemical and acute genetic PPT1 inhibition. Of note, UGCG (UDP-glucose ceramide glucosyltransferase) is among the 24, and was previously shown to be a primary driver of lipidome remodeling that limits cell death from lysosomal inhibition in response to acute PPT1 inhibition with either DC661 or siPPT1 [13].

Figure 2.

Proteomic changes during acute versus chronic PPT1 inhibition. (A) MS abundance of PPT1 protein in A375P cells in which PPT1 is depleted by either CRISPR-Cas9, or siRNA. NT: non-target, sg: small guide RNA. (B) Unsupervised hierarchical cluster comparing proteomes of constitutive Cas9 (sg) and acute siRNA (si) PPT1 inhibition (C) Volcano plots of sgPPT1 vs. sgNT and siPPT1 vs. siNT, (D) Volcano plot of DC661 vs. control (DMSO), reanalyzed from reference 12. Red = q-value ≤0.05 and ≥ 1.5 fold increase; Blue = q-value ≤0.05 and ≥ 1.5 fold decrease. (E) Venn diagram demonstrating overlapping proteins that are significantly increased with sgPPT1 vs. sgNT, siPPT1 vs. siNT, and DC661 vs. control (DMSO).

Table 1.

Top 20 proteins significantly increased and decreased in chronic and acute genetic PPT1 inhibition (q ≤ 0.05).

sgPPT1 vs sgNT				siPPT1 vs siNT
Protein Name	Fold Increase	Protein Name	Fold Decrease	Protein Name	Fold Increase	Protein Name	Fold Decrease
GSTM3	40.48	PPT1	−15.41	ENC1	30.25	ALAS1	−13.88
GSTM5	38.06	LMO7	−6.63	HMOX1	15.50	JMJD1C	−8.85
EEF1A2	20.77	ENC1	−6.05	UBL5	11.92	PPT1	−8.14
SPICE1	14.65	CCN1	−5.00	SLC4A2	10.90	PTGES	−6.40
MAP1A	10.64	CTH	−4.29	HPX	8.00	RWDD4	−6.13
COLEC12	9.51	ALDH1L1	−3.93	SDC4	5.82	ISCU	−5.81
STRA6	8.50	ALCAM	−3.86	KHDRBS3	4.71	CTH	−5.64
HLA-DQB1	8.25	TAX1BP1	−3.64	ARHGEF12	4.56	GDF15	−5.36
GBP2	6.33	RBMS2	−3.60	PHLDA2	4.56	PNISR	−4.85
CKB	5.94	AXL	−3.43	IGF1R	4.15	FUNDC2	−4.12
B3GNT7	5.54	ALDH1L2	−3.25	RAD1	3.97	CCN1	−3.66
ARHGEF12	5.54	AFDN	−3.15	CCND1	3.89	NECTIN2	−3.46
UBE2C	5.19	ATG13	−3.02	ATP5IF1	3.77	ISYNA1	−3.06
S100A2	5.18	DDR2	−2.96	KLHL21	3.50	TSEN15	−2.86
ICAM1	5.09	IGHMBP2	−2.87	TMEM59	3.40	NMT2	−2.75
MRFAP1	4.79	MBNL2	−2.78	PFKFB3	3.25	FBXW9	−2.71
AGRN	4.28	METAP2	−2.67	FAT1	3.07	PRMT9	−2.67
TNC	4.26	ASNS	−2.59	PLIN2	2.89	WASHC3	−2.66
SPARC	4.13	GPT2	−2.58	PLAT	2.85	SPRYD4	−2.65
ISG20	3.63	CBS	−2.56	NRP1	2.78	IVNS1ABP	−2.58

Table 2.

Top 20 proteins significantly increased and decreased in siPPT1/siNT vs sgPPT1/sgNT (q ≤ 0.05).

siPPT1/siNT vs sgPPT1/sgNT
Protein Name	Fold Increase	Protein Name	Fold Decrease
ENC1	94.67	GSTM3	−606.81
HMOX1	32.75	GSTM5	−42.82
HPX	20.20	EEF1A2	−12.99
SLC4A2	16.76	MAP1A	−11.99
BACE2	8.18	CKB	−11.53
TMEM59	7.30	PTBP2	−8.63
PLAT	6.27	STRA6	−8.27
KLHL21	6.27	B3GNT7	−8.18
CCND1	5.80	GBP2	−7.25
LPXN	5.62	S100A2	−7.08
LRRC58	5.43	MMP2	−6.85
TMBIM1	5.35	SNX21	−6.02
PDCD2	5.04	RIC8B	−5.67
IGF1R	4.65	TAF4	−5.67
NDFIP2	4.51	COL6A1	−5.66
CLTB	4.28	FUNDC2	−5.49
ZNF280C	4.27	SLC6A6	−5.45
AEN	4.14	UPP1	−5.36
RAD51	4.06	UBE2C	−5.30
CALCOCO1	4.06	RASA2	−5.24

To identify pathway alterations that may explain the differences between acute PPT1 inhibition (siPPT1) and constitutive PPT1 inhibition (sgPPT1), we compared the differentially expressed proteins (q-value ≤0.05 and absolute fold change ≥ 1.5) in A375P sgNT siPPT1 and A375P sgPPT1 samples. DAVID pathway analysis of sgPPT1 compared to sgNT revealed that sgPPT1 was deficient in proteins involved in ferroptosis and NFE2L2/NRF2-regulated signaling pathways among the significantly enriched canonical pathways (Figure 3A). Both pathways were not significantly enriched when comparing acute siPPT1 vs. siNT (Figure 3B). However, proteins involved in ferroptosis signaling and NFE2L2/NRF2-regulated signaling pathways were upregulated in siPPT1 compared to sgPPT1 (Figure 3C). In the siPPT1 versus sgPPT1 comparison, protein levels were normalized to their respective controls to account for the differences in the proteome profile of siNT and sgNT (Figure 2B). Consistent with the pathway analysis, we found that ferroptosis induction by RSL3 was blunted in A375P sgPPT1 cells compared to A375P sgNT cells (Figure 3D). In contrast, RSL3 produced significantly greater ferroptosis in A375P siPPT1 compared to A375P siNT cells (Figure 3E). This is corroborated by previously published work that shows that chemical PPT1 inhibition induces ferroptosis [12]. These findings suggest that chronic PPT1 loss leads to an adaptation that avoids ferroptosis. To determine if tumor cells can adapt similarly to chronic genetic Ppt1 inhibition in vivo, we developed a Ppt1 conditional knockout (cKO) mouse and crossed it into a mouse model of melanoma.

Figure 3.

Pathway alterations in acute and chronic PPT1 inhibition. (A) Top enriched canonical pathways identified by DAVID pathway analysis when comparing significantly increasedor decreased proteins in sgPPT1 vs. sgNT. (B) Top enriched pathways identified when comparing significantly increased or decreased proteins in siPPT1 vs. siNT. (C) Top enriched pathways identified when comparing normalized siPPT1 vs. sgPPT1. Enriched pathways with Fisher’s exact test p-value ≤0.01 were considered significant. (D) Bodipy-C11 analysis indicated blunted ferroptosis induction in A375P sgNT and sgPPT1 after treatment with 0.5 μM RSL3. One way ANOVA was used to compare the groups. *P ≤ 0.05; ***P ≤ 0.001 (E) Bodipy-C11 analysis after 0.5 μM RSL3 treatment indicated enhanced ferroptosis induction in siPPT1 treated A375P compared to cells that received siNT. One way ANOVA was used to compare the groups. ***P ≤ 0.0001; ****P ≤ 0.0001. RFU: relative fluorescence units.

A Ppt1 conditional knockout mouse

We employed two methods to create a Ppt1 cKO mouse. Table S1 lists the sequences used. The first involved the sequential insertion of LoxP sites flanking the first exon of the Ppt1 gene (Figure 4A). After several failed attempts at simultaneous LoxP insertions, we were able to derive 3’ LoxP mice from one of the injections, and subsequently injected the 5’ Ppt1 sgRNA, 5’ LoxP repair template, and Cas9 mRNA into oocytes in vitro fertilized with sperm from Ppt1 3’ LoxP fl/fl mice. We also used Easi-CRISPR [19], a one-step approach to create the cKO animals. Fertilized WT C57Bl/6J eggs were injected with Cas9 mRNA, two guide RNAs, and a long single-stranded repair oligonucleotide that contained both the 5’ and 3’LoxP sites closer to and flanking the first exon of Ppt1 (Figure 4B). Founder #6 was the only positive founder with both LoxP sites using the sequential approach (Figure 4C) so the descendants are referred to as line 6. We obtained 5 potential Easi-CRISPR founders by genotyping (Figure 4D). A screening strategy outlined in Miura [20] using primers overlapping the LoxP sites for cloning and sequencing eliminated founder 32 as it lacked one of the LoxP sites, but confirmed the other 4 Easi-CRISPR founders had both LoxP sites (Figure S1A). However, founder 37 had a mutation near the LoxP site and founder 34 had a mutation in the Ppt1 exon 1 (data not shown). After confirming perfect LoxP sequences, founders were bred with wild-type C57Bl/6J mice and the first generation N1 pups were genotyped. Mosaicism with some pups missing a LoxP site was noted in this first generation from all founders (Figure S1B-C) except Easi-CRISPR founder 25 (Figure S1D) whose descendants are referred to as line 25. We confirmed the LoxP sites were intact in N1 pups by sequencing and that the LoxP sites were functional by an in vitro Cre recombinase assay. After treatment of genomic DNA with Cre recombinase, PCR bands of the expected sizes for deletion of Ppt1 exon 1 were observed in line 6 fl/fl, line 25 fl/fl & line 39 fl/WT pups. (Figure S1E). All three genotypes in the expected Mendelian ratios were obtained from N1-crossed mice (representative genotyping in Figure S1F). Homozygous Ppt1^fl/fl mice were normal, with none of the defects and early death seen in whole-body ppt1 KO mice [21]. To study the role of Ppt1 in melanoma tumorigenesis, we next generated the BrafCA, PtenloxP, Ppt1loxP, Tyr:CreERT2 mice by crossing BrafCA, Pten^fl/fl, Tyr:CreERT2 mice [17] with Ppt1^fl/fl mice. The genotypes of mice used for the study were 1) BrafCA^+/-, Pten^fl/fl, Ppt1^fl/fl Tyr:CreERT2^+/- 2) BrafCA^+/-, Pten^fl/fl, Ppt1^fl/WT Tyr:CreERT2^+/- 3) BrafCA^+/-, Pten^fl/fl, Ppt1^WT/WT Tyr:CreERT2^+/-. In this inducible model of melanoma, application of 4-hydroxytamoxifen (4-HT) to the skin of C57Bl/6 mice led to hyperpigmentation by roughly 12 days and eventually tumor growth beginning about 3 weeks after treatment with similar gross morphology in the presence or absence of Ppt1 (Figure 4E). PCR on DNA isolated from tumors showed the expected shifts in band size for deletion of Ppt1 exon 1 (Figure 4F), activation of constitutively active Braf, and deletion of Pten exon 5 in the tumors (Figure S2A-C).

Figure 4.

A Ppt1 conditional knockout mouse. (A and B)Schema for generating a Ppt1 conditional knockout allele using 2 CRISPR-Cas9 strategies that were pursued concurrently. (A) Sequential LoxP site insertions Cas9 mRNA, 2 guide RNAs (sgRNA) and 2 repair templates (LoxP sites represented by arrowheads – red 3’ LoxP, yellow 5’ LoxP) were injected into the pronucleus of fertilized C57Bl/6J oocytes. Mice with the 3’ LoxP site were obtained and then bred together. Sperm from a mouse homozygous for the 3’LoxP site was used to in vitro fertilize oocytes which were then injected with Cas9 mRNA, the 5’ guide RNA, and the 5’ LoxP repair template. Black arrowheads indicate primer locations on either side of the LoxP insertion sites used for screening in (C). (B) Easi-CRISPR – Cas9 mRNA, 2 guide RNAs, and a long single stranded repair template with both LoxP sites and exon 1 were injected into fertilized oocytes. Black arrowheads indicate primer locations on either side of the LoxP insertion sites used for screening in (D). (C) PCR screening of the sequential founders with primer pairs Ppt1 F3/R3 and 3’ Ppt1 F3/R3. Only #6 (red number) was positive for both 5’ and 3’ LoxP sites. (D) PCR screening of Easi-CRISPR founders with primer pairs Ppt1 F7/R7 and Ppt1 205 F/611 R. Founders #25, 32, 34, 37 & 39 (red numbers) appeared to have both 5’ LoxP and 3’ LoxP sites. (E) Representative images of 4-HT induced Braf mutant pten null melanoma tumors with the indicated Ppt1 genotypes. (F) PCR of the entire 1381 bp Ppt1 F3/611 R amplicon (diagram indicates location of the primers) of DNA isolated from tails and from 4-HT induced melanoma tumors demonstrates deletion of Ppt1 exon 1 by Cre Recombinase in ppt1^fl/fl and Ppt1^fl/WT tumors in both line 6 and line 25, but not in wild-type Ppt1 mice. Expected bands for deletion of Ppt1 exon 1 were 384 bp for line 6 and 847 bp for line 25. G: Genomic DNA, T: tumor DNA.

Ppt1 is dispensable for melanoma tumor growth

Heterozygous or homozygous loss of Ppt1 had no discernible impact on the rate of tumor initiation (Figure 5A), or the rate of tumor growth (Figure 5B). There was no difference in median survival (time to 2000 mm³ tumor volume) in mice with any of the three genotypes (Figure 5C). Subgroup analysis separating mice generated from line 6 or line 25 found similar results (Figure S3A-B), where loss of Ppt1 had no impact on growth of tumors. Similarly, Ppt1 loss produced no significant difference in growth of tumors or survival of mice in females or males (Figure S3C-F). To ensure that PPT1 enzymatic activity was truly lost, a PPT1 biochemical activity assay demonstrated a significant decrease in PPT1 activity in Ppt1^fl/WT and ppt1^fl/fl tumors compared to Ppt1^WT/WT tumors (Figure 5D). Although other cells within the tumor are expected to still produce PPT1 protein, a western blot analysis of tumor tissue indicated reduced PPT1 in Ppt1^fl/fl mice (Figure S4 A-B). There was no difference in the histological analysis of Ppt1^WT/WT or ppt1^fl/fl tumors (Figure 5E). Inspection of the draining ipsilateral inguinal lymph nodes showed enlargement and intense black pigmentation, suggestive of metastases in tumor-bearing mice regardless of Ppt1 genotype. (Figure S4C and Table S2). No visible metastasis to other organs was noted. Finally, knocking out Ppt1 in YUMM 1.7 cells, generated originally from a tumor growing in the BrafCA, Pten^fl/fl, Cdkn2a^fl/fl, Tyr:CreERT2 mouse model [22] produced no significant difference in viability (Figure 5F–H). Similar to human melanoma cells, CRISPR knockout of Ppt1 in mouse YUMM 1.7 cells produced reduced ferroptosis sensitivity compared to a non-target control (Figure 5I).

Figure 5.

Ppt1 is dispensable for melanoma tumor growth. (A) Time to tumor formation by genotype in line 6 and line 25 (WT n = 17; Ppt1^fl/WT n = 38; ppt1^fl/fl n = 49) graphed is time to first appearance of black dots after 4-HT treatment, and time to coalescence (merging of the tumor dots). Representative images are shown. (B) Tumor growth measured weekly after 4-HT treatment in both lines by genotype (WT n = 43; Ppt1^fl/WT n = 65; ppt1^fl/fl n = 65); ANOVA comparing tumor volumes on the last day. (C) Kaplan-Meier survival curve (time to 2000 mm³ tumor volume) by genotype in both lines. (D) Van Diggelen PPT1 enzyme assay in tumor samples from 3 mice of each Ppt1 genotype with duplicate technical replicates. (E) Representative H&E slides of melanoma tumors of each genotype at 100X magnification. (F) Immunoblot demonstrating Ppt1 knockout in YUMM1.7 sgPpt1 Msg1 and Msg3 clones compared to YUMM1.7 sgNT cells. sg: small guide RNA; Msg: mouse guide; NT: non-target. (G) Van Diggelen PPT1 biochemical assay in the indicated cells. n = 3, repeated twice. (H) Cell growth at 3 days after plating the indicated cell lines at 1X10⁵. n = 3. (I) Bodipy C-11 fluorescence of YUMM1.7 cells treated as indicated for 24 h demonstrated a blunted response to RSL3 ferroptosis induction in sgPpt1 cells. One way ANOVA test was used to compare multiple groups. ns: nonsignificant,*p ≤ 0.05, **p ≤ 0.005, ***P ≤ 0.001; ****P ≤ 0.0001; RFU: relative fluorescence units.

Discussion

This study provides valuable insights into the role of the lysosomal thioesterase PPT1 in tumorigenesis, particularly in melanoma, and sheds light on the differential effects of acute and chronic PPT1 inhibition, achieved through both genetic knockout and acute chemical or genetic inhibition, on melanoma cell survival. One of the central findings of this study is the observation that loss of Ppt1 does not significantly impact the rate of tumor initiation, tumor growth, or overall survival in a genetically engineered mouse model of melanoma. Likewise, genetic knockout of human PPT1 did not affect cell growth rates. This finding challenges the hypothesis that PPT1 is essential for melanomagenesis. It suggests that targeting PPT1 alone may not be an effective strategy for inhibiting melanoma tumor growth.

An intriguing aspect of the study is the distinction between acute and chronic PPT1 inhibition. Acute genetic or pharmacological PPT1 inhibition led to significant loss of viability in melanoma cells, while chronic PPT1 knockout had no such effect. Even in the case of the inducible CRISPR-Cas9 knockout of PPT1, complete PPT1 knockout took a few days to achieve, this was enough time for the cells to adapt and avoid cell death, unlike acute PPT1 knockdown or chemical PPT1 inhibition. This differential response suggests that melanoma cells can adapt to prolonged PPT1 deficiency, allowing them to survive and proliferate. Understanding the mechanisms underlying this adaptation warrants further study. The proteomic analysis revealed distinct protein expression profiles in response to acute and chronic PPT1 inhibition, as well as chemical inhibition. Only a small subset of proteins was commonly regulated across all three conditions, highlighting the complexity of cellular responses to PPT1 inhibition.

Previous work has demonstrated that acute PPT1 inhibition, either with DC661 treatment or siPPT1 produces lysosomal lipid peroxidation, lysosomal membrane permeabilization, autophagy inhibition, and immunogenic cell death [12]. Constitutive knockout of PPT1 was also shown to produce autophagy inhibition [12]. While phenotypically, acute, chronic and chemical PPT1 inhibitions produce similar findings, proteomic changes were strikingly different amongst these three conditions. In response to chemical PPT1 inhibition with DC661, numerous proteins associated with cholesterol and sphingolipid metabolism were significantly upregulated, resulting in a striking remodeling of the lipidome and assembly of UGCG-dependent lipid rafts that limit cell death from lysosomal inhibition [13]. UGCG was among the 24 upregulated proteins in common between acute chemical and acute genetic PPT1 inhibition. However, examining the 20 most increased and decreased proteins in constitutive PPT1 knockout and acute PPT1 knockdown, we did not find significant changes in other lipid metabolism proteins (e.g. LDLR, SCRIB/SCRB1) that are modulated by DC661. This suggests that increased levels of lipid metabolism proteins observed with DC661 are either independent of PPT1 inhibition or attributable to the specific kinetics of PPT1 inhibition that can be achieved with chemical inhibition compared to genetic inhibition.

There are ongoing clinical trials involving PPT1 inhibitors in advanced cholangiocarcinoma, indicating the translational potential of targeting PPT1 in cancer therapy. However, the findings from this study emphasize the need for a nuanced approach to clinical trials, considering factors such as the duration of PPT1 inhibition and patient-specific molecular profiles.

This study provides valuable insights into the role of PPT1 in melanoma tumorigenesis and highlights the complex interplay between acute and chronic PPT1 inhibition. It suggests that targeting PPT1 alone may not be sufficient in melanoma therapy, but modulation of ferroptosis and antioxidant pathways could offer a potential strategy for enhancing treatment efficacy. Further research is needed to unravel the intricate mechanisms underlying melanoma adaptation to PPT1 inhibition and to develop more precise and effective therapeutic approaches for melanoma patients.

Materials and methods

Cell culture

Human A375P (CRL-3224) was purchased from ATCC. Cell lines were tested for Mycoplasma biannually by University of Pennsylvania Core facilities and authenticated using short-tandem repeat fingerprinting by Wistar Institute Genomics core. Preparation of constitutive Cas9 A375P PPT1-non-target and KO cells were described previously [12]. A375P cells were cultured in RPMI (Invitrogen, 11875093) culture media supplemented with 10% fetal bovine serum (Sigma-Aldrich, F6178) and 1× penicillin-streptomycin solution (Gibco, 15140–122). Cells were grown at 37°C in the presence of 5% CO₂. Some cells were treated with 3 µM DC661 (Selleckchem, S8808) for 24 h. Transfection with siRNA is described below. For cell growth assays, cells were plated at 1X10⁵ in 60 mm dishes in triplicate. On designated days, cells were collected by trypsin and resuspended in media. Cell samples were mixed 1:1 with 0.2% trypan blue in PBS (Gibco, 14190–136). The mixture was incubated for 1 min and cells were counted on a Cellometer (Nexcelom, Lawrence, MA).

A375P iCas9 cells

Following Zhang lab protocols available through Addgene, guide RNA oligos (IDT) with the sequences in the table below, were annealed, phosphorylated, and then ligated into Esp3I (BsmBI) cut TLCV2 (a gift from Adam Karpf; Addgene, 87360) Ligated plasmids were transformed into Stbl3 cells (Invitrogen, C737303). Sequencing of plasmid preps confirmed the presence of the desired guide RNA sequences. 10 µg TLCV2, 7.5 µg psPax2 (a gift from Didier Trono; Addgene, 12260), and 4 µg pVSV-G (a gift from Akitsu Hotta; Addgene, 138479) were transfected with Lipofectamine 3000 (Invitrogen, L3000015) into 293FT cells (Invitrogen, R70007) to generate lentivirus that was subsequently used to transduce A375P cells followed by puromycin selection. Clones were grown from single cells in 96 well plates. Treatment with 1 µg/ml doxycycline hyclate (Sigma, D9891) induced co-expression of Cas9 and EGFP. No GFP expression was noted in cells without Dox treatment (data not shown). Cells were collected at different days after Cas9 induction for western, blot, cell counts, and PPT1 enzyme activity analysis.

Guide	Forward sequence	Reverse Sequence
Non-Target	CACCGGTAGCGAACGTGTCCGGCGT	AAACACGCCGGACACGTTCGCTACC
Human PPT1 sg1	CACCGTTTGGACTCCCTCGATGCCC	AAACGGGCATCGAGGGAGTCCAAAC
Human PPT1 sg3	CACCGGCCTCGTGCAAGCCGAATAC	AAACGTATTCGGCTTGCACGAGGCC

YUMM1.7 CRISPR-Cas9 cells

Mouse YUMM1.7 [22] cells were a gift from Dr. Xiaowei Xu (University of Pennsylvania). YUMM1.7 cells were grown in DMEM/F-12 (Corning, 10–092-CV), 1% MEM NEAA (Gibco, 11140–050), 10% FBS and 1% penn-strep. For YUMM1.7 PPT1 non-target and KO cells, guide RNA oligos (IDT) were annealed, phosphorylated, and then ligated into BbsI digested and dephosphorylated pSpCas9(BB)-2A-Puro (PX459) V2.0 plasmid, a gift from Feng Zhang (Addgene, 62988), following Zhang lab protocols available through Addgene. Ligated plasmids were transformed into Stbl3 cells. Sequencing of plasmid preps confirmed the presence of the desired guide RNA sequences. YUMM1.7 cells were transfected using Lipofectamine 3000, followed by puromycin selection. Clones were grown from single cells in 96-well plates. Ppt1 knockdown was confirmed by western blot. Sequences for guide RNA oligos are as follows:

Guide	Forward sequence	Reverse sequence
Non-Target	CACCGTAGCGAACGTGTCCGGCGT	AAACACGCCGGACACGTTCGCTAC
Mouse Ppt1 sg1	CACCGTTTGGACTCCCCCGATGCCC	AAACGGGCATCGGGGGAGTCCAAA
Mouse Ppt1 sg 3	CAACCGTACGTCCTGTCTCTAGAGAT	AAACATCTCTAGAGACAGGACGTA

Van Diggelen PPT1 enzyme assay

PPT1 enzyme activity was measured following published protocols [18]. Briefly, cells were sonicated in water plus protease inhibitors. 4-HT induced tumor lysates were prepared with RIPA buffer (Cell Signaling Technology, 9806S) with protease inhibitor cocktail tablets cOmplete (Roche, 04 693 159 001) and PhosSTOP (Roche, 04 906 837 001). The protein concentration of the lysates was measured with Pierce™ BCA Protein Assay Kit (Thermo Scientific, 23225). One to 5 µg of protein in 10 µl total volume with H₂O were incubated with 20 µl of McIlvains phosphate/citric-acid pH 4.0 buffer (Boston BioProducts, BB-2440) containing 0.64 mM 4-methylumbelliferyl 6-thio-palmitate-β-D-glucopyranoside substrate (Cayman Chemical, 19524), 15 mM DTT (Roche, 93911520), 0.375% Triton X-100 (Roche, 10789704001), and 0.1 U β-glucosidase from almonds (Sigma-Aldrich, G4511) in opaque sided clear bottom 96 well plates (Falcon, 35337). After 1 h of incubation at 37°C, the reaction was stopped with 200 µl 0.5 M NaHCO₃, 0.5 M Na₂CO₃, pH 10.7 buffer containing 0.025% Triton X-100. Fluorescence was measured at 358 nm excitation and 448 nm emission on a SpectaMax i3X (Molecular Dynamics, Chatsworth, CA) and graphed in GraphPad Prism after background subtraction of wells with no added protein.

siRNA transfection

Constitutive Cas9 A375P PPT1-non-target cells were transfected with Non-target siRNA (Santa Cruz Biotechnology, sc-44236) or human PPT1 siRNA (Santa Cruz Biotechnology, sc-105216) using Lipofectamine 2000 (Invitrogen, 11688027). Cells were then collected 48 h later for western blot, cell counts, PPT1 enzyme activity assay, and proteomic analysis.

Ferroptosis measurement using BODIPY-C11

Ferroptosis induction was measured using C11-BODIPY (BODIPY 581/591 C11; Thermo Fisher Scientific, D3861). A375P or YUMM1.7 cells were seeded in a 6-well plate and treated with either DMSO or 0.5 μM RSL3 (Cayman Chemical, 19288) for 24 h. Cells were harvested by centrifugation at 300 × g for 3 min. Cells were resuspended in 500 μL 1X HBSS (Gibco, 14175079) containing 0.5 μM C11-BODIPY and incubated at 37°C for 15 min. Cells were pelleted as above and resuspended in 500 μL 1X HBSS, and fluorescence intensity was measured on an Accuri C6 Plus cytometer (BD Biosciences, Franklin Lakes, NJ) using FL1 FITC.

Immunoblotting

Whole cell lysates were prepared using RIPA buffer or SDS lysis buffer and protease inhibitors. Protein concentration was measured using Pierce™ BCA Protein Assay Kit (Thermo Scientific, 23225). Thirty µg of protein was used for SDS-PAGE then transferred onto PVDF membranes (Bio-Rad, 1620177). Membranes were blocked with 5% milk in 1X TBST (Cell Signaling Technology, CS-9997) and incubated overnight at 4°C with primary antibody in 2.5% BSA (Sigma-Aldrich, A9418-100G). Primary antibodies included PPT1 for human samples (OriGene, CF800490; 1:1000), PPT1 for mouse samples (GeneTex, GTX110677; 1:5000), Cas9 (Cell Signaling Technology, 65832S; 1:1000) and ACTB/β-actin (Cell Signaling Technology, 3700S; 1:5000). Membranes were washed with 1X TBST and incubated for 1 h at room temperature with species-specific horseradish peroxidase-conjugated secondary antibodies (Cell Signaling Technology, 7076S and 7074S; 1:5000). Membranes were subsequently washed and developed using Pierce ECL Western Blotting substrate (Thermo Scientific, 32106) and autoradiography films (Lab Scientific, XAR ALF 1318).

Ppt1 conditional knockout mouse (Ppt1 cKO)

Mice with LoxP sites inserted on either side of exon 1 in the mouse Ppt1 gene were generated in collaboration with the University of Pennsylvania CRISPR-Cas9 Mouse Targeting Core, University of Pennsylvania Transgenic and Chimeric Mouse Facility, and the Children’s Hospital of Philadelphia Transgenic Core. For line 6, originally Cas9 mRNA, 2 guide RNAs, and 2 separate repair templates with the LoxP sites were injected into fertilized C57BL/6J oocytes. Founder mice were screened by PCR for insertion of LoxP sites into Ppt1 with primers outside of the repair template. All sequences were ordered from IDT, and are listed in Table S1. Detection of insertion of the 5’ LoxP upstream of the mouse Ppt1 exon 1 in the sequential line used Ppt1 F3/R3 primers with TopTaq DNA Polymerase (Qiagen, 200203) followed by Topo TA cloning (Invitrogen, 45–0641). Insertion of the 3’ LoxP site was detected with Ppt1 3’ F3/R3 primers and Q5 High Fidelity Taq with GC enhancer (New England Biolabs, M0493L) and touchdown PCR, followed by Zero Blunt Topo cloning (Invitrogen, 45–0245). Cloned amplicons were sequenced using PENN Core facilities. After repeated failure to obtain mice with both LoxP sites on the same chromosome, sperm from a mouse homozygous for just the 3’ LoxP site was used to in vitro fertilize C57BL/6J oocytes that were then subjected to pronuclear microinjection with Cas9 mRNA, the 5’ guide RNA and 5’ LoxP repair template. For line 25 we utilized the Easi-CRISPR technique [19,20]. Cas9 mRNA, 2 guide RNAs and a long single-stranded DNA repair template with both LoxP sites in place closer to exon 1 of the Ppt1 gene were injected into fertilized C57Bl/6J oocytes. Founders were screened by PCR with Ppt1 F7/R7 for the 5’ LoxP region and Ppt1 205 F/611 R for the 3’ LoxP region. Easi-CRISPR founders were further screened and cloned as recommended in [20] using primers overlapping the LoxP sites paired with primers outside the repair template. TopTaq PCR reactions used the 5’ LoxP overlap primer paired with Ppt1 611 R, and the 3’ LoxP overlapping primer was paired with Ppt1 F7. Amplicons were Topo TA cloned and sequenced. Founders with perfect sequence were backcrossed to wild-type C57BL/6J mice and the resulting N1 pups were screened and sequenced as well. Mosaicism was noted in the first generation of pups from founder 6 and all the Easi-CRISPR founders except for line 25 in that some N1 pups lacked one of LoxP sites. Heterozygous N1 pups with both LoxP sites and perfect sequence were bred together to generate homozygous Ppt1 cKO mice. PCR with DreamTaq (Thermo Scientific, K1081) was used for routine genotyping. Due to inconsistent results with the Line 6 3’ Ppt1 F3/R3 primers, new primers Ppt1 423F/868 R were selected and used for routine genotyping. PCR band sizes were compared to 1 KB Plus Ladder (Invitrogen, 10787018), and gels were imaged on a Gel Doc XR+ (Bio-Rad, Hercules, CA).

DNA isolation

Founder mice tails for screening or pup toe tips for genotyping were digested with proteinase K (Denville, CB3210–5) in lysis buffer (10 mM Tris, pH 8.0, 100 mM NaCl, 10 mM EDTA, 0.5% SDS) overnight at 56 degrees. DNA was precipitated in isopropanol and resuspended in TE. Cultured cells or tumor tissue DNA was isolated with the Qiamp DNA Mini Kit (Qiagen, 51304)

Melanoma model

Homozygous Ppt1^fl/fl mice were crossed with Tyr:CreERT2; BrafCA; PtenloxP (exon 5) mice [17]. Genotyping with primers listed in Table S1 was as previously described [17,23–25]. 4-hydroxytamoxifen (4-HT, 70% Z-isomer; Sigma, H6278) was dissolved in DMSO and diluted to 1.9 mg/ml. To induce tumor formation, on three consecutive days 0.5 µl of the 1.9 mg/ml 4-HT was pipetted onto a small area of shaved and Nair treated skin on the right back flank of mice heterozygous for both BrafCA and Tyr;CreERT2 transgenes and carrying either WT Ppt1, Ppt1^fl/WT, or Ppt1^fl/fl alleles. All animal experiments were approved by the University of Pennsylvania Institutional Animal Care and Use Committee (IACUC; approval number 805835) and were performed in an Association for the Assessment and Accreditation of Laboratory Animal Care (AAALAC) accredited facility, maintained at 20 ± 2°C, 42% humidity, with a 12-h light – dark cycle and free access to food and water. Mice cohorts ranged in age from 6.5 to 14 weeks at the start of 4-HT treatment. Tumor volumes were measured weekly in millimeters with digital calibers (Thermo Fisher, 14-648-17). Length, width and depth were multiplied to determine cubic tumor volume. The criteria for euthanasia under this protocol were lethargy (none observed), tumor ulceration or tumor volume of 2,000 mm³.

Cre recombinase assay

A 20 μl reaction containing 1 μg of genomic DNA with and without 10 units of recombinant Cre recombinase (New England Biolabs, M0298S) in 1× buffer was incubated at 37°C for 1 h. One μl of the Cre reaction mix was used as template for PCR reactions with Ppt1 F3/868 R primers to amplify the entire 1381 bp amplicon (1313 bp in WT). Cre deletion of Ppt1 exon 1 was predicted to yield a 384 bp band in line 6 and an 847 bp band in lines 25 & 39.

Protein extraction and digestion for liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis

A375P melanoma cells (0.7 × 10⁶) were cultured in 60-mm culture dishes. Frozen cell pellets were lysed with 50 mM Tris, pH 7.4, 1% SDS, 150 mM NaCl, 1 mM EDTA, 0.15 mM PMSF (Sigma-Aldrich, 10837091001), 1 µg/ml pepstatin (Sigma-Aldrich, P4265), and 1 µg/ml leupeptin (Sigma-Aldrich, L2884). Clarified lysates (25 µg each) were electrophoresed 0.5 cm into pre-cast NuPAGE (Thermo Fisher Scientific, NP0301) Bis-Tris gel followed by fixing and staining with colloidal Coomassie Brilliant Blue. The entire stained gel lanes were excised, de-stained, reduced with tris (2-carboxyethyl) phosphine (Thermo Fisher Scientific, 20490), alkylated with iodoacetamide (Sigma-Aldrich, I1149), and digested with 10 ng/ml modified trypsin as described previously [26]. Digests (2 µg) were analyzed by LC-MS/MS on a Q-Exactive Plus mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) in-line with a Vanquish Neo UHPLC system (Thermo Fisher Scientific). Tryptic peptides were injected onto an Acclaim PepMap^TM 100 trap column (75 μm i.d. x 2 cm packed with 3 μm C18 resin; Thermo Fisher Scientific, 16435) and separated by reversed phase HPLC on a BEH C18 nanocapillary analytical column (75 μm i.d. x 25 cm, 1.7 μm particle size; Waters, 186008795) using a 245-min gradient formed by 0.1% formic acid in water (mobile phase A) and acetonitrile (mobile phase B). Full MS spectra were acquired at 70,000 resolution with a scan range of 350–1800 m/z, automatic gain control (AGC) target of 3e6, and maximum injection time (max IT) of 50 ms. Data-dependent MS2 spectra were acquired for the top 20 most abundant ions at 17,500 resolution with an isolation width of 1.5 m/z, AGC target of 5e4, and max IT of 50 ms. Peptide match was set to preferred and unassigned and singly charged ions were rejected.

MS data were analyzed using MaxQuant 2.4.2.0 (http://www.maxquant.org) with a UniProt human sequence database (21 July 2022) and a common contaminants database [27]. Tryptic peptide specificity with a maximum of 2 missed cleavages, fixed modification on cysteine (carbamidomethylation), and variable methionine oxidation and N-terminal acetylation were used in the search. A cutoff of 1% false discovery rate (FDR) was used for peptides and protein identifications. Match between runs was enabled; proteins were quantified using label-free quantification [28].

Statistical analysis was performed using Perseus 2.0.10.0 [29,30]. Proteins were required to be identified by at least 2 unique peptides and have 3 valid values (non-zero quantitation) within a sample group. Contaminants, reverse proteins, and proteins identified only by site modified peptides were removed from the dataset. Missing values were imputed from a normal distribution. Pairwise comparisons between conditions were performed at the protein level using a Student’s t-test with permutation-based FDR with s0 = 0.1 and 250 randomizations. For siPPT1 versus sgPPT1 comparison, protein intensities were normalized to the average of their respective controls. Significant changes were defined as q-value ≤0.05 and an absolute fold change ≥ 1.5. The Database for Annotation, Visualization, and Integrated Discovery (DAVID, Ver. 2021; http://david.ncifcrf.gov [31,32] was used for a combined KEGG (Kyoto Encyclopedia of Genes and Genomes; www.genome.jp/kegg/pathway.html; [33]) and Wikipathways (https://www.wikipathways.org; [34]) database pathway enrichment analysis. For each comparison (sgPPT1 vs. sgNT, siPPT1 vs. siNT and siPPT1/siNT vs sgPPT1/sgNT), gene sets of significant proteins having q-value ≤0.05 and ≥ 1.5 increases or decreases were input separately, and all proteins identified in the dataset were used as the background proteome. Fisher’s Exact Test p-value ≤0.01 was considered significant.

Statistical analysis

Statistical significance was determined using Student’s unpaired, two-tailed t-test when comparing two samples with relatively equal variances. The one-way ANOVA was used when more than two groups were compared or in the case of unequal variance. Kaplan-Meier survival curves were computed with survival event defined as death, humane endpoint for sacrifice as defined by the animal protocol, or tumor volume ≥2000 mm³. A null hypothesis was significantly rejected if the p-value was less than 0.05. Statistical analysis was conducted using GraphPad Prism 10.

Funding Statement

This work was supported by HHS/NIH/NCI grants [P30 CA016520-45, PO1 CA114046, RO1 CA266404, P50CA261608, R50CA221838, and P30 CA010815]. This work was also supported by Commonwealth Universal Research Enhancement Program (CURE) funds to the University of Pennsylvania (RKA). Equipment use was enabled by the Center for Molecular Studies in Liver and Digestive Diseases [NIH-P30-DK050306].

Disclosure statement

RKA is an inventor on patents related to dimeric chloroquines. The patents are licensed to Pinpoint Therapeutics, and RKA is a scientific founder. RKA is a consultant for Deciphera, Tasca Therapeutics, and gets research funding from Novartis, Bristol-Myers Squibb, Deciphera, Springworks, Merck. The other authors declare no conflicts of interest.

Data availability statement

The proteomics data for Figure 2 have been deposited into the MassIVE public repository (http://massive.ucsd.edu/ProteoSAFe/static/massive.jsp) with accession MSV000089020 and the ProteomeXchange repository (http://proteomecentral.proteomexchange.org) with accession PXD032120.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15548627.2024.2403152