Lysyl oxidase-like 2 (LOXL2) is a member of the scavenger receptor cysteine-rich (SRCR) repeat carrying LOX family. Although LOXL2 is suspected to be involved in histone association and chromatin modification, the role of LOXL2 in epigenetic regulation during tumorigenesis and cancer progression remains unclear. Here, we report that nuclear LOXL2 associates with histone H3 and catalyzes H3K36ac deacetylation and deacetylimination. Both the N-terminal SRCR repeats and the C-terminal catalytic domain of LOXL2 carry redundant deacetylase catalytic activity. Overexpression of LOXL2 markedly reduced H3K36 acetylation and blocked H3K36ac-dependent transcription of genes, including c-MYC, CCND1, HIF1A, and CD44. Consequently, LOXL2 overexpression reduced cancer cell proliferation in vitro and inhibited xenograft tumor growth in vivo. In contrast, LOXL2 deficiency resulted in increased H3K36 acetylation and aberrant expression of H3K36ac-dependent genes involved in multiple oncogenic signaling pathways. Female LOXL2-deficient mice spontaneously developed uterine hypertrophy and uterine carcinoma. Moreover, silencing LOXL2 in cancer cells enhanced tumor progression and reduced the efficacy of cisplatin and anti-programmed cell death 1 (PD-1) combination therapy. Clinically, low nuclear LOXL2 expression and high H3K36ac levels corresponded to poor prognosis in uterine endometrial carcinoma patients. These results suggest that nuclear LOXL2 restricts cancer development in the female reproductive system via the regulation of H3K36ac deacetylation.

Significance:

LOXL2 loss reprograms the epigenetic landscape to promote uterine cancer initiation and progression and repress the efficacy of anti–PD-1 immunotherapy, indicating that LOXL2 is a tumor suppressor.

The mammalian lysyl oxidase (LOX) family has five members (LOX, LOXL1–4). All five members carry an N-terminal signal peptide end and a conserved C-terminal copper(II)-dependent ε-amine oxidase catalytic domain. LOX was originally discovered as a secreted extracellular enzyme that oxidizes lysine residues of collagen into allysine residues (1, 2). LOX has also been frequently reported to be involved in cell proliferation and cancer metastasis regulation (3, 4). In addition to carrying the LOX family conserved catalytic domain, LOXL2–4 also carry the four scavenger receptor cysteine-rich (4xSRCR) repeats within their N-terminal region (2). We recently disclosed that LOXL3 is a dual specificity enzyme responsible for STAT3 deacetylation and deacetylimination in CD4+ Th17 cell differentiation (5). LOXL2 plays pivotal roles in creating the pathologic microenvironment for fibrotic diseases as well as malignant transformation and metastasis (4, 6–20). LOXL2 has been widely reported to directly interact with the elastin precursor tropoelastin to catalyze its deamination (1, 2). Notably, LOXL2 undergoes nuclear translocation. Nuclear LOXL2 either interacts with the transcription factor SNAI1 for epithelial–mesenchymal transition (EMT) or binds to histones to presumably catalyze epigenetic regulation. However, the effect of LOXL2 on the oxidative deamination of H3K4me3 remains controversial (21–23). Therefore, the role of LOXL2, especially in epigenetic gene regulation for tumorigenesis and cancer-related pathologic progression, remains largely unclear.

H3K36me3, a well-defined epigenetic modification, is associated with transcriptional regulation, premRNA splicing, and DNA-repair pathways (24, 25). H3K36 can be acetylated by the histone acetyltransferase GCN5 (26). Although H3K36me3 is restricted to the coding regions (24, 25), H3K36ac is localized predominantly in the promoters of RNA polymerase II–transcribed genes (26). H3K36 modification is cell-cycle related, i.e., whereas H3K36me3 peaks in the G1 phase when nonhomologous end joining (NHEJ) occurs, H3K36ac peaks in the S–G2 phase when homologous recombination prevails (27). Although H3K36me3 with a tumor-suppressive role is identified in numerous cancers, little is known about the function of H3K36ac in human cancers.

In this work, we provide evidence that nuclear LOXL2 is a deacetylase responsible for H3K36ac deacetylation and deacetylimination in vitro. LOXL2 overexpression reduced cancer cell proliferation in vitro and inhibited xenograft tumor growth in vivo. In contrast, LOXL2 deficiency resulted in H3K36 hyperacetylation and enhanced the expression of multiple genes responsible for tumor progression. LOXL2 deficiency in female mice caused aberrant cell proliferation and signal transduction, which eventually promoted organ enlargement or carcinogenesis in the uterus. Moreover, silencing LOXL2 in murine U14 cells enhances tumor progression by reducing the efficacy of cisplatin and anti-programmed cell death 1 (PD-1) combination immunotherapy. Additionally, low nuclear LOXL2 expression was further associated with a poor prognosis in uterine endometrial cancer patients. Thus, nuclear LOXL2 physiologically regulates organ expansion and retards cancer progression, presumably via H3K36 deacetylation/deacetylimination in the female reproductive system.

Animals

Loxl2loxP/loxP mice were crossed to C57/BL6 CMV-Cre transgenic mice to generate CMV-Cre; Loxl2loxP/+ mice and control littermates. CMV-Cre; Loxl2loxP/loxP mice were generated by inbreeding CMV-Cre; Loxl2loxP/+ mice. The genotypes of CMV-Cre; Loxl2loxP/loxP mice were determined by using PCR amplification of tail DNA. Mice were housed in specific pathogen-free rodent facilities of Shanghai Institutes for Biological Sciences, CAS, Shanghai. The animal study was approved by the Institutional Animal Care and Use Ethics Committee of Shanghai Institutes for Biological Sciences, CAS, Shanghai. Mice were randomly allocated to each experimental group. All mice were euthanized for sacrifice.

BrdUrd incorporating assay

BrdUrd (50 mg/kg body weight) was administered by intraperitoneal injection every 12 hours for 3 days until the mice were sacrificed. Mice tissues were fixed with 4% formaldehyde for 10 minutes and underwent subsequently optimal cutting temperature compound embedding. Sections (6-μm-thick) were used for BrdUrd immunofluorescence staining.

IHC and immunofluorescence

Paraffin-embedded sections were performed with a series of dewaxing, rehydration, antigen retrieval, and quenching of endogenous peroxidase activity. After blocking with 5% goat serum, the sections were labeled with the corresponding primary antibody at 4°C overnight and anti-mouse/rabbit secondary antibody (Dako REAL EnVision Detection System) for 1 hour at room temperature. For IHC, detection was developed by 3,3′-diaminobenzidine (DAB) substrate. Images were captured with an Olympus DP70 camera system. For immunofluorescence, Alexa fluorescently labeled secondary antibodies were used at a 1:300 dilution. Nuclear DNA was counterstained with 100 nmol/L DAPI (28). Images were performed using an Olympus FV3000 confocal microscope. The antibodies and reagents used are listed in Supplementary Table S1.

Plasmids

Gene constructs used for mammalian cell expression were cloned into the pcDNA3.1 with tag (3×FLAG, HA, Myc). Vectors (3–6 μg) were transfected into HEK293T cells (1 × 106) using lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. Lentiviral particles were produced in HEK293T cells for lentiviral infections to inhibit targeted gene expression in HeLa or U14 cells according to Addgene protocols (https://www.addgene.org/protocols).

Cell culture and cell growth assays

HeLa, KLE, HEK293T, and U14 cells were purchased from the Cell Bank of Type Culture Collection of the Chinese Academy of Sciences. All cells were verified to be free of Mycoplasma by PCR. Cells were cultured in DMEM or MEM medium supplied with 10% fetal bovine serum (FBS) and 100 U/mL penicillin–streptomycin in a 5% CO2 incubator (Thermo Scientific) at 37°C. Cell viability was determined on the indicated days using the Cell Counting Kit-8 according to the manufacturer's instructions. For the colony formation assay, 1×103 HeLa cells were cultured in 6-well plates for 6 days. Colonies were fixed with methanol and then stained with 0.1% crystal violet for 10 minutes. For G2–M phase synchronization, cells were treated with 1 μmol/L nocodazole for 24 hours. The cell cycle was detected by Agilent NovoCyte flow cytometry. All cell culture media and reagents were listed in Supplementary Table S1.

Mice tumor model

Four-week-old female BALB/c mice and nude mice were purchased from Vital River Laboratory Animal Technology Co., Ltd. and all the animal experiments were approved by the Institutional Animal Use and Care protocols of Wenzhou Medical University. Mice were randomly allocated to each experimental group. HeLa cells (1 × 106) and KLE cells (2 × 106) were injected subcutaneously into 4-week-old female nude mice. Mice were sacrificed 4 weeks after injection and tumors were collected. The tumor volume was measured using the following formula: 0.5 × length × width2. The U14 mouse tumor (No.14 of cervical carcinoma), a cervical squamous cell carcinoma, was an ectopically induced carcinoma by treating the uterine cervix with 20-methylcholanthrene in 1958 (29). U14 cells (1 × 106) were injected subcutaneously into the right flank of female BALB/c mice (30). For the combinatory immunity therapy, the mice were intraperitoneally administered 200 μg of anti–PD-1 antibody and cisplatin (2 mg/kg body weight) on days 7, 10, 13, 16, and 19 after inoculation of U14 cells.

Isolation of tumor-infiltrating lymphocytes

The fresh tumor tissue was cut into small pieces and incubated with 1 mg/mL collagenase D, 0.1 mg/mL hyaluronidase, 0.1 mg/mL DNase I, and 2% FBS in RPMI-1640 and treated with gentleMACS Dissociator (Miltenyi Biotech) to dissociate of tumor tissue into single cells for flow cytometry. Isolated cells were filtered through 40-μm strainer, immune cells were purified by Percoll-gradient centrifugation. To detect CD8+ tumor-infiltrating lymphocytes, isolated immune cells were stained with CD45-PE-CY7, CD3-PE, and CD8-PerCP. Live cells were stained with LIVE/DEAD Fixable Violet Kit (Invitrogen) and subsequently analyzed by using NovoCyte flow cytometry.

Western blot

Cells or tissue pieces were lysed in RIPA buffer and followed by sonication. Protein concentration was quantified by using a bicinchoninic acid protein assay kit. Proteins are separated in an acrylamide/bis-acrylamide gel by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and then electro-blotted onto a polyvinylidene (PVDF) membrane. The membrane was blocked with 5% nonfat milk and incubated with primary antibody overnight at 4°C, and subsequently incubated with corresponding horseradish peroxidase (HRP)-conjugated secondary antibody. Western Lightning Plus enhanced chemiluminescence reagent (PerkinElmer) and Bio-Rad Gel Doc XR+ Gel Documentation System were used to visualize protein levels. Western blots were quantified with ImageJ software (NIH, Bethesda, MD). The antibodies and reagents used are listed in Supplementary Table S1.

Immunoprecipitation

Cell lysates in RIPA buffer were supplemented with protease inhibitor cocktail. The indicated antibody was used to incubate with lysate samples at 4°C overnight according to the manufacturer's instructions of a reversible immunoprecipitation system kit (Merck Millipore). After rinsing the protein G or A sepharose beads with RIPA buffer, supernatants were subsequently used for Western blotting.

Nuclear and cytoplasmic extraction

Nuclear and cytoplasmic extracts were separated according to the manufacturer's instructions of NE-PER Nuclear and Cytoplasmic Extraction Reagents, and extracts were subsequently used for Western blotting.

Mass spectrometric analysis

Transfected HeLa lysate was separated by SDS-PAGE and stained with Coomassie brilliant blue. The H3 band was excised from the gel for MS-grade trypsin protease digestion, followed by MALDI-TOF spectra analysis. After the LOXL2 deacetylation/deacetylimination activity assay, the mixture was loaded into a Thermo LTQXL MS system. Using Thermo Xcalibur software (version 2.2), liquid chromatography/mass spectrometry (LC/MS) data were converted into total ion current chromatograms representing all ions according to their corresponding m/z values, charge state, intensity, and retention time. The mixture was then enriched and desalted using an m-C18 ZipTip (Millipore) and eluted directly onto a MALDI plate containing 2 mL a-cyano-4-hydroxy cinnamic acid (CHCA) saturated solution in 50% acetonitrile (ACN) and 0.1% trifluoroacetic acid (TFA). MALDI-TOF spectra were acquired on a Voyager DE Pro mass spectrometer from Applied Biosystems (5). The following gradient elution programs were used: linear gradient (A: 0.1% formic acid; B: 80% acetonitrile and 0.1% formic acid) from 3% to 95% B (0–30 minutes), isocratic 100% B (30–40 minutes), 100%–3% B (40–41 minutes). The injected volume was 1 μL, and the flow rate 0.3 mL/minute.

Real-time quantitative PCR

Total RNA was extracted with a TRIzol reagent. RNA reverse transcription was performed with an iScript cDNA synthesis kit. RT-qPCR was performed with SsoFast EvaGreen Supermix on an ABI-7900HT machine (Applied Biosystems). The expression of target genes was normalized by internal control (β-Actin or GAPDH), relative gene expression levels were analyzed using the ΔΔCt method. The sequences of primers are presented in Supplementary Table S3.

Human clinical samples

Cancer patient samples were collected from Xiangya Hospital, Central South University, including uterine endometrial cancer (n = 101), stromal sarcoma (n = 35), normal endometrium (n = 10) and cervical cancer (n = 48). All human cancer tissues used in this study were approved by the Institutional Review Board of Xiangya Hospital, Central South University. Written informed consent was obtained from each patient before sample collection. IHC staining results were scored in a semiquantitative scoring manner considering both staining intensity and percentage of positive staining cells, as previously described (31). Briefly, staining intensity (I) was recorded as 0 (negative), 1 (weak staining), 2 (moderate staining), and 3 (strong staining), respectively. In addition, the percentage (P) of positively staining cells was recorded as 0 (positive staining in less than 10% cells), 1 (positive staining in 10%–40% cells), 2 (positive staining in 40%–70% cells), and 3 (positive staining in more than 70% cells). The final staining index (S) was calculated as the sum of the two scores: S = I + P. We classified staining results as negative (S ≤ 2), middle positive (2 < S ≤ 4), and highly positive (S > 4). The distribution of the clinicopathologic data in this study cohort is given in Supplementary Table S7.

The Cancer Genome Atlas

Transcriptome sequencing data and patient survival data of human cancer patients were obtained from The Cancer Genome Atlas (TCGA) via the Broad GDAC Firehose (Broad Institute; https://www.cancer.gov/tcga). LOXL2 genetic alteration and frequencies (including deletion, mutation, and amplification) in cancer patients were defined.

Statistical analysis

All experiments were performed at least three times, as noted in figure legends. We chose a sample size based on pilot experiments and literature protocols in the field to achieve at least 80% power and a two-sided type I error of 5%. The chosen sample size was sufficient to determine statistical significance in our established tumor model. Blinding was not performed due to personnel availability. Random allocation and quantitative measurement using instruments and kits in our experiments minimized biased assessments (32). All data were expressed as the mean ± SD processed by GraphPad Prism 8.0. Data were tested for the homogeneity of variances and normality. Statistical comparisons between the two groups were assessed with an unpaired, two-tailed Student t test. Survival analysis was determined by Kaplan–Meier followed by a log-rank test. Statistical significance is stated in the figure legends. A P < 0.05 was considered statistically significant (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Availability of data and materials

The sequencing data have been deposited in the Gene Expression Omnibus database (accession number: GSE213868). All data created and analyzed in this work are involved in this published article (and its Supplementary Data files).

LOXL2 in nuclei is responsible for H3K36ac deacetylation

To investigate the role of LOXL2 in histone deacetylation, we started by analyzing LOXL2 protein expression levels and cellular localization in various mice tissues. LOXL2 was highly expressed in female reproductive organs, including both the uterus and ovary (Fig. 1A). This was essentially in line with the previous finding that LOXL2 expression was more intensive in the reproductive system (8). Although LOX family members have the potential for extracellular matrix cross-linking upon secretion to the extracellular space, IHC analysis revealed a nuclear distribution pattern of LOXL2 in the uterus but not in the kidney (Fig. 1B). Consistently, Western blotting and immunostaining confirmed LOXL2 as a nuclear protein in the human cervical carcinoma cell line HeLa and human endometrial cancer cell line KLE (Fig. 1C,E).

Figure 1.

LOXL2 in nuclei is responsible for H3K36ac deacetylation. A, The expression profiling of LOXL2 in mice tissues. B, IHC staining of LOXL2 in mice uterus and kidney. Scale bar, 100 μm. C, LOXL2 expression pattern in cytoplasmic and nuclear fractions of HeLa cells. D, Cytoplamic fraction, nuclear fraction, and chromosome-associated fraction were prepared from HeLa cells and analyzed by Western blot. E, The nucleus location of LOXL2 in HeLa and KLE cells. Scale bar, 20 μm. F, Histone H3 proteins were prepared from HEK293T cells (1×106) with transfection of Flag-tagged LOXL2 plasmid (3 and 6 μg) or 3 μg pcDNA3.1-3×Flag empty vector (EV). H3 acetylation and methylation on indicated residues were analyzed with respective antibodies. G, HA-tagged H3 was cotransfected with Myc-GCN5 and LOXL2-Flag plasmid, and cell lysates were subjected to Western blot analysis. H, Schematic diagram of LOXL2 structure (UniProt: Q9Y4K0) with 4×SRCR repeats in N-terminal domain (ND) and the amine oxidase in C-terminal domain (CD). The four cysteines were mutated into four alanines (C97A, C231A, C364A, C477A; 4CA), whereas two histones were mutated into two glutamines (H626Q, H628Q; 2HQ). SP-NES, signal peptide-nuclear exporting signal. I, LOXL2 full-length (FL), LOXL2-ND, and LOXL2-CD were transiently expressed in HEK293T cells, followed by Western blot analysis. J, H3K36ac peptide was incubated with purified Flag-LOXL2-FL, LOXL2-ND, and LOXL2-CD protein for in vitro deacetylation assay and then subjected to liquid chromatography–mass spectrometry (LC-MS) analysis. One percent Bull serum albumin in deacetylation buffer was used as blank control. K, LOXL2 wild-type (WT), LOXL2–2HQ-mutant, and LOXL2-4CA-mutant were transiently overexpressed in HEK293T cells, followed by Western blot analysis. L, Top, peptides from in vitro deacetylation assay were analyzed with LC-MS according to molecular weight and mass-to-charge ratio. Bottom, mass/mass spectrum analysis for H3K36 and H3K36ac peptides extracted from the above in vitro deacetylation assay. All data represent the mean ± SD; n = 3. **, P < 0.01.

Figure 1.

LOXL2 in nuclei is responsible for H3K36ac deacetylation. A, The expression profiling of LOXL2 in mice tissues. B, IHC staining of LOXL2 in mice uterus and kidney. Scale bar, 100 μm. C, LOXL2 expression pattern in cytoplasmic and nuclear fractions of HeLa cells. D, Cytoplamic fraction, nuclear fraction, and chromosome-associated fraction were prepared from HeLa cells and analyzed by Western blot. E, The nucleus location of LOXL2 in HeLa and KLE cells. Scale bar, 20 μm. F, Histone H3 proteins were prepared from HEK293T cells (1×106) with transfection of Flag-tagged LOXL2 plasmid (3 and 6 μg) or 3 μg pcDNA3.1-3×Flag empty vector (EV). H3 acetylation and methylation on indicated residues were analyzed with respective antibodies. G, HA-tagged H3 was cotransfected with Myc-GCN5 and LOXL2-Flag plasmid, and cell lysates were subjected to Western blot analysis. H, Schematic diagram of LOXL2 structure (UniProt: Q9Y4K0) with 4×SRCR repeats in N-terminal domain (ND) and the amine oxidase in C-terminal domain (CD). The four cysteines were mutated into four alanines (C97A, C231A, C364A, C477A; 4CA), whereas two histones were mutated into two glutamines (H626Q, H628Q; 2HQ). SP-NES, signal peptide-nuclear exporting signal. I, LOXL2 full-length (FL), LOXL2-ND, and LOXL2-CD were transiently expressed in HEK293T cells, followed by Western blot analysis. J, H3K36ac peptide was incubated with purified Flag-LOXL2-FL, LOXL2-ND, and LOXL2-CD protein for in vitro deacetylation assay and then subjected to liquid chromatography–mass spectrometry (LC-MS) analysis. One percent Bull serum albumin in deacetylation buffer was used as blank control. K, LOXL2 wild-type (WT), LOXL2–2HQ-mutant, and LOXL2-4CA-mutant were transiently overexpressed in HEK293T cells, followed by Western blot analysis. L, Top, peptides from in vitro deacetylation assay were analyzed with LC-MS according to molecular weight and mass-to-charge ratio. Bottom, mass/mass spectrum analysis for H3K36 and H3K36ac peptides extracted from the above in vitro deacetylation assay. All data represent the mean ± SD; n = 3. **, P < 0.01.

Close modal

Given that nuclear LOXL2 was substantially incorporated into the chromatin fraction (Fig. 1D; Supplementary Fig. S1A), LOXL2 might indeed be involved in histone modifications. In HEK293T cells transiently transfected with LOXL2, we analyzed histone acetylation. Among all analyzed histone acetyl-lysine sites, H3K36ac was markedly deacetylated by LOXL2 in a dose-dependent manner (Fig. 1F; Supplementary Fig. S1B). Although LOXL2 was reported by others for H3K4me3 demethylation in vitro (23), we failed to detect apparent demethylation activity of LOXL2 on H3K4me3 or H3K36me3 (Fig. 1F). GCN5 is responsible for H3K36 acetylation induction (26), and we transiently overexpressed GCN5 in HEK293T cells alone or along with LOXL2. GCN5-induced H3K36 acetylation was abolished by LOXL2 coexpression (Fig. 1G), further supporting the conclusion that LOXL2 is able to reverse H3K36 acetylation induction by GCN5 in vitro.

LOXL2 N-terminal and C-terminal domains carry independent deacetylase activity

In the case of LOXL3, both the N-terminal SRCR repeats and the C-terminal domain carry deacetylase activity (5). To determine whether LOXL2 resembles LOXL3 in this regard, we generated N-terminal domain carrying 4xSRCR repeat (ND, aa 1–544) and C-terminal domain (CD, aa 548–774) constructs (Fig. 1H; Supplementary Fig. S1C and S1D) following bioinformatic (UniProtKB: Q9Y4K0) and ▵1–2SRCR-LOXL2 (aa 318–774) structural analyses (33). H3K36ac was efficiently deacetylated by full-length LOXL2 (FL), LOXL2-CD, and LOXL2-ND in HEK293T cells (Fig. 1I). Furthermore, mass spectrometric analysis showed that all three purified forms of LOXL2 catalyzed H3K36ac deacetylation (Fig. 1J; Supplementary Fig. S1E), suggesting that both the C-terminal and N-terminal domains of LOXL2, which is independent of each other, carry deacetylase activity. In addition, LOXL2 with either two C-terminal copper(II)-binding site mutations (H626Q, H628Q; 2HQ) or four disulfide bond site mutations (C97A, C231A, C364A, C477A; 4CA) partially abrogated H3K36ac deacetylation activity in HEK293T cells (Fig. 1H and K; Supplementary Fig. S1F and SG). Although purified wild-type LOXL2 protein was fairly efficient in removing the acetyl-group from the H3K36ac peptide, purified LOXL2 with 2HQ or 4CA mutation partially abolished the deacetylation activity (Supplementary Fig. S1H–S1L). Mass/mass analysis of the spectra further confirmed that H3K36ac was significantly deacetylated by the wild-type LOXL2 protein in vitro (Fig. 1L), indicating that the N-terminal disulfide binding sites (C97, C231, C364, and C477) and the C-terminal copper(II)–binding sites (H626 and H628) are responsible for the deacetylase activity of the LOXL2 N-terminal domain and CD, respectively.

LOXL2 has the capacity to catalyze H3K36ac deacetylimination

Because the LOX family bears amine oxidase activity (5), LOXL2 might also catalyze acetyl-lysine deacetylimination, similar to LOXL3 (Supplementary Fig. S2). To gain insights into the deacetylimination property of LOXL2, we synthesized an H3 peptide carrying multiple acetyl-lysine residues (K23, K27, and K36) and performed in vitro deacetylimination analysis. Mass spectra showed that purified LOXL2 protein preferentially deacetyliminated H3K36ac into H3K36al (allysine) over other lysine residues of the peptide, suggesting that LOXL2 has the capacity to catalyze H3K36ac deacetylimination (Fig. 2A). During deacetylimination or deamination, i.e., oxidation of acetyl-lysine or lysine, H2O2 is released as a product (Supplementary Fig. S3A). We synthesized a series of peptides and incubated them with purified LOXL2 protein for H2O2 production analysis. Of note, in addition to H3K36ac allysine conversion, only a small fraction of acetyl-H3K37 was converted into allysine (Fig. 2AC; Supplementary Fig. S3B and S3C), suggesting that LOXL2 had the capacity to catalyze deamination and deacetylimination sites specifically.

Figure 2.

LOXL2 catalyzes H3K36ac deacetylimination in addition to deacetylation. A, Acetyl-H3 peptide was incubated with purified Flag-LOXL2 protein for in vitro deacetylimination assay and followed by MALDI-TOF spectra analysis. One percent bull serum albumin (BSA) in deacetylimination buffer was used as blank control. B, Acetyl-H3 peptides that were used for in vitro deacetylimination assay are shown in the table. C, LOXL2 catalyzed deacetylimination of acetyl-H3 peptides listed in B, and the production of the H2O2 level was detected. BAPN (50 μmol/L) was used to inhibit LOXL2 deacetylimination activity. D, HEK293T cells transiently transfected with LOXL2-WT and LOXL2-2HQ. Biotin hydrazide was followed by avidin affinity precipitation of histone H3 with allysine and was analyzed with Western blot. E, HEK293T cells transiently transfected with GCN5 along with LOXL2-WT or LOXL2-2HQ were subjected to affinity precipitation and Western blot analysis. All data represent the mean ± SD; n = 3. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 2.

LOXL2 catalyzes H3K36ac deacetylimination in addition to deacetylation. A, Acetyl-H3 peptide was incubated with purified Flag-LOXL2 protein for in vitro deacetylimination assay and followed by MALDI-TOF spectra analysis. One percent bull serum albumin (BSA) in deacetylimination buffer was used as blank control. B, Acetyl-H3 peptides that were used for in vitro deacetylimination assay are shown in the table. C, LOXL2 catalyzed deacetylimination of acetyl-H3 peptides listed in B, and the production of the H2O2 level was detected. BAPN (50 μmol/L) was used to inhibit LOXL2 deacetylimination activity. D, HEK293T cells transiently transfected with LOXL2-WT and LOXL2-2HQ. Biotin hydrazide was followed by avidin affinity precipitation of histone H3 with allysine and was analyzed with Western blot. E, HEK293T cells transiently transfected with GCN5 along with LOXL2-WT or LOXL2-2HQ were subjected to affinity precipitation and Western blot analysis. All data represent the mean ± SD; n = 3. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

Using avidin beads, we pulled down biotin hydrazide, which reacted with allysine (aldehyde) via Schiff's base reaction (Supplementary Fig. S3D). In comparison with the 2HQ-mutant, wild-type LOXL2 in HEK293T cells led to more histone H3 pulldown by the biotin hydrazide–avidin system (Fig. 2D; Supplementary Fig. S3E), suggesting that LOXL2 indeed catalyzed acetyl-H3 lysine residues into H3-allysine residues. Consistently, in vitro measurement of H2O2 revealed that wild-type LOXL2 but not the 2HQ-mutant deacetylated lysine residues of histone H3 protein purified from calf thymus into allysine (Supplementary Fig. S3F). When GCN5 was ectopically expressed in HEK293T cells, wild-type LOXL2 but not the 2HQ-mutant showed much stronger deacetylimination induction (Fig. 2E), further validating that the amine oxidase domain in the C-terminus of LOXL2 has the capacity to catalyze H3K36ac deacetylimination.

LOXL2 reduces the H3K36ac level and inhibits cell growth

Given that LOXL2 could specifically deacetylate or deacetyliminate H3K36ac in vitro, we next determine the cellular effect of LOXL2 in LOXL2-overexpressed cancer cells. The constitutive H3K36 acetylation of parental HeLa cells was dramatically reduced with the ectopic introduction of LOXL2 (Fig. 3A). Histone H3 peptide with allysine36 was recovered from HeLa cells with ectopic LOXL2 overexpression (Fig. 3B; Supplementary Fig. S3G), suggesting that LOXL2-dependent deacetylimination or a LOXL2-mediated two-step reaction (i.e., deacetylation followed by oxidation) occurred in these cells. Nocodazole treatment induced G2–M phase arrest (34), and flow cytometry analysis showed that a higher fraction of nocodazole-treated HeLa cells significantly delayed the reentry of the G0–G1 phase after LOXL2 overexpression (Fig. 3C). Coincidently, LOXL2 overexpression in HeLa cells significantly attenuated colony formation capability in vitro and xenograft tumor growth in vivo (Fig. 3DG). In contrast, LOXL2 knockdown efficiently reduced endogenous LOXL2 expression and increased H3K36ac levels in HeLa cells (Supplementary Fig. S4A–S4C). The percentage of cells entering the S-phase and the expression of proliferating cell nuclear antigen were increased in HeLa cells upon LOXL2 knockdown (Supplementary Fig. S4D–S4G). In addition, LOXL2 overexpression in KLE cells remarkably inhibited cell growth and xenograft tumor formation (Supplementary Fig. S4H–S4L).

Figure 3.

LOXL2 inhibits cell growth and abolishes H3K36ac peaks in the gene promoter. A, Western blot analysis of H3K36ac and H3K36me3 expression level in pcDNA3.1-3×Flag empty vector (EV) and LOXL2 overexpressed HeLa cells. B, The allysine36 residue of histone H3 was extracted from HeLa cells by SDS-PAGE and followed by MALDI-TOF spectra analysis subsequently. C, Flow cytometry analysis of the cell cycle in LOXL2-overexpressed HeLa cells after treatment with 1 μmol/L nocodazole for 24 hours. D, LOXL2-WT, LOXL2-2HQ, LOXL2-4CA, and empty vector–overexpressed HeLa cells were analyzed by colony formation assay in vitro. E–G, LOXL2-WT, LOXL2-2HQ, LOXL2-4CA, and empty vector–overexpressed HeLa cells were used for xenograft tumor growth in vivo. Tumor size (E), tumor growth (F), and tumor weight (G) were recorded after HeLa cell challenge in female nude mice (n = 5). H, H3K36ac peaks in the whole gene or TSS region of control or LOXL2-overexpressed HeLa cells. I, Differential analysis of H3K36ac peaks in control or LOXL2-overexpressed HeLa cells. Approximately 800 H3K36ac peaks were reduced by LOXL2 overexpression, and gene ontology analysis is shown. J, The distribution of H3K36ac peaks surrounding c-MYC, CCND1, HIF1A, and CD44 gene promoters in control and LOXL2-overexpressed HeLa cells. K, The mRNA expression level of c-MYC, CCND1, HIF1A, and CD44 in empty vector and LOXL2 (WT, 2HQ, and 4CA)-overexpressed HeLa cells. All data represent the mean ± SD; n = 3 or 5. *, P < 0.05; **, P < 0.01.

Figure 3.

LOXL2 inhibits cell growth and abolishes H3K36ac peaks in the gene promoter. A, Western blot analysis of H3K36ac and H3K36me3 expression level in pcDNA3.1-3×Flag empty vector (EV) and LOXL2 overexpressed HeLa cells. B, The allysine36 residue of histone H3 was extracted from HeLa cells by SDS-PAGE and followed by MALDI-TOF spectra analysis subsequently. C, Flow cytometry analysis of the cell cycle in LOXL2-overexpressed HeLa cells after treatment with 1 μmol/L nocodazole for 24 hours. D, LOXL2-WT, LOXL2-2HQ, LOXL2-4CA, and empty vector–overexpressed HeLa cells were analyzed by colony formation assay in vitro. E–G, LOXL2-WT, LOXL2-2HQ, LOXL2-4CA, and empty vector–overexpressed HeLa cells were used for xenograft tumor growth in vivo. Tumor size (E), tumor growth (F), and tumor weight (G) were recorded after HeLa cell challenge in female nude mice (n = 5). H, H3K36ac peaks in the whole gene or TSS region of control or LOXL2-overexpressed HeLa cells. I, Differential analysis of H3K36ac peaks in control or LOXL2-overexpressed HeLa cells. Approximately 800 H3K36ac peaks were reduced by LOXL2 overexpression, and gene ontology analysis is shown. J, The distribution of H3K36ac peaks surrounding c-MYC, CCND1, HIF1A, and CD44 gene promoters in control and LOXL2-overexpressed HeLa cells. K, The mRNA expression level of c-MYC, CCND1, HIF1A, and CD44 in empty vector and LOXL2 (WT, 2HQ, and 4CA)-overexpressed HeLa cells. All data represent the mean ± SD; n = 3 or 5. *, P < 0.05; **, P < 0.01.

Close modal

LOXL2 mediates H3K36ac-dependent gene expression

Histone acetylation induces an open chromatin confirmation by reducing positive charges and generally increases the transcription of nucleosome-bound DNA (35). Given that H3K36ac is a conserved epigenetic marker actively associated with transcriptional activation during cell proliferation (26, 27), we performed differential ChIP-seq analysis between H3K36ac in native and LOXL2-overexpressing HeLa cells. The abundance and intensity of H3K36ac peaks as well as the width of H3K36ac peaks were strikingly reduced around the promoter region or along the gene in the whole genomic region after LOXL2 overexpression (Fig. 3H). By mapping the promoter sequencing results to the control chromatin sequence, we found more than 3,000 H3K36ac peaks enriched in the promoter regions in native HeLa cells, but the number dropped to approximately 2,000 upon ectopic expression of LOXL2. Among the total H3K36ac peaks, approximately 800 H3K36ac peaks in the promoter region were significantly abolished upon LOXL2 overexpression (Fig. 3I), suggesting that LOXL2 inhibits the transcription of H3K36ac-dependent genes. Gene ontology analysis revealed that LOXL2-dependent H3K36ac deacetylation remarkably affected the transcription of genes involved in cell adhesion and cell division (Fig. 3I). Individual promoter analysis confirmed that H3K36ac was enriched in the promoter region of genes such as c-MYC, CCND1, HIF1α, and CD44, whereas LOXL2 overexpression markedly reduced H3K36ac peak intensity in these gene promoter regions (Fig. 3J). RT-qPCR analysis showed that LOXL2 overexpression significantly reduced the expression of these cancer progression-related genes in HeLa cells and KLE cells (Fig. 3K; Supplementary Fig. S4L). As the Cell Counting Kit-8 (CCK-8) and xenograft tumor formation assays showed, genetic ablation of c-MYC, CCND1, HIF1α, or CD44 notably inhibited HeLa cell and KLE cell proliferation and the xenograft tumor growth (Supplementary Fig. S5). Thus, LOXL2 inhibited the promoter activities of those genes involved in cell-cycle progression and signal transduction by specifically targeting H3K36ac.

LOXL2 deficiency induces spontaneous carcinogenesis in the uterus

We next analyzed the role of LOXL2-mediated H3K36ac deacetylation in vivo and generated CMV-Cre; Loxl2loxP/loxP mice (hereafter referred to as Loxl2−/− mice). CMV-Cre mice express a transgene containing Cre under the transcriptional control of a human cytomegalovirus (CMV) minimal promoter. In our transgenic strain, the loxP-flanked murine Loxl2 gene was deleted in all tissues and germ cells (Fig. 4A and B; Supplementary Fig. S6A–S6C). The offspring percentage of Loxl2−/− mice was approximately half of what was expected. The reproductive capacity of Loxl2−/− mice, evaluated by the number of offspring per nest and overall offspring numbers, was significantly lower than that of their control littermates (Supplementary Fig. S6D). These findings are consistent with a previous study reporting that the deletion of Loxl2 provoked perinatal lethality (36). Interestingly, female Loxl2−/− mice developed uterine enlargement (Fig. 4C; Supplementary Fig. S6E), which resembled the uterine enlargement observed in Loxl1−/− mice (37, 38). Of note, 12.1% (7/58) of the female Loxl2−/− mice spontaneously developed endometrial carcinoma (4/7) or cervical squamous cell carcinoma (3/7), which occurred in endometrial or cervical sections in Loxl2−/− mice (Fig. 4D). Together, these results demonstrate that LOXL2 might be a novel tumor suppressor in the mice uterus.

Figure 4.

LOXL2 deficiency results in uterine hypertrophy and carcinogenesis in mice. A, Schematic diagram of Loxl2 knockout strategy in mice. B, LOXL2 expression level in Loxl2+/+, Loxl2+/−, and Loxl2−/− mice uterus. C, The uterus diameters were compared between 16-week-old Loxl2+/+ and Loxl2−/− mice (n = 10). D, The carcinomas occurred in cervical or endometrial sections in Loxl2−/− mice. Percentage of mice with carcinoma in 10-month-old Loxl2−/− mice was calculated (n = 58). E, Whole-cell extracts prepared from Loxl2+/+ and Loxl2−/− mice uterus were subjected to Western blot analysis with indicated antibodies. F, Immunofluorescence for H3K36ac in Loxl2+/+ and Loxl2−/− mice uterus (n = 5). H3K36ac intensity from the above samples was calculated by ImageJ software. Scale bars, 100 μm. G, BrdUrd incorporation assay in Loxl2+/+ and Loxl2−/− mice uterus (n = 5). Scale bars, 100 μm. H, Immunofluorescence for Ki67 in Loxl2+/+ and Loxl2−/− mice uterus (n = 5). Scale bars, 50 μm. I, ChIP-qPCR assay showed that the H3K36ac level was enriched in Myc, Ccnd1, Hif1a, and Cd44 gene promoters of Loxl2−/− mice uterus (n = 3). All data represent the mean ± SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 4.

LOXL2 deficiency results in uterine hypertrophy and carcinogenesis in mice. A, Schematic diagram of Loxl2 knockout strategy in mice. B, LOXL2 expression level in Loxl2+/+, Loxl2+/−, and Loxl2−/− mice uterus. C, The uterus diameters were compared between 16-week-old Loxl2+/+ and Loxl2−/− mice (n = 10). D, The carcinomas occurred in cervical or endometrial sections in Loxl2−/− mice. Percentage of mice with carcinoma in 10-month-old Loxl2−/− mice was calculated (n = 58). E, Whole-cell extracts prepared from Loxl2+/+ and Loxl2−/− mice uterus were subjected to Western blot analysis with indicated antibodies. F, Immunofluorescence for H3K36ac in Loxl2+/+ and Loxl2−/− mice uterus (n = 5). H3K36ac intensity from the above samples was calculated by ImageJ software. Scale bars, 100 μm. G, BrdUrd incorporation assay in Loxl2+/+ and Loxl2−/− mice uterus (n = 5). Scale bars, 100 μm. H, Immunofluorescence for Ki67 in Loxl2+/+ and Loxl2−/− mice uterus (n = 5). Scale bars, 50 μm. I, ChIP-qPCR assay showed that the H3K36ac level was enriched in Myc, Ccnd1, Hif1a, and Cd44 gene promoters of Loxl2−/− mice uterus (n = 3). All data represent the mean ± SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

LOXL2 deletion accelerates H3K36ac level and cell proliferation in the uterus

To confirm that LOXL2-mediated H3K36ac deacetylation is involved in spontaneous carcinogenesis in the uterine organ, we first analyzed the acetyl-lysine level of histones in the Loxl2−/− mice uterus. LOXL2 deficiency remarkably increased the H3K36ac level without affecting other types of acetylation. In the uterus of Loxl2−/− mice, no noticeable changes were observed in the expression level of other LOXL family members or HDACs, such as LOXL3–4, HDAC1–3, and SIRT1–3, further supporting that LOXL2 is responsible for H3K36ac deacetylation in vivo (Fig. 4E and F; Supplementary Fig. S6F and S6G). Both the Ki67 intensity and BrdUrd incorporation rate showed that cell proliferation was dramatically increased in Loxl2−/− mice uterus (Fig. 4G and H). ChIP-qPCR assays also confirmed that the H3K36ac intensity in Loxl2−/− mice uterus was enriched in the promoter of genes related to cancer progression, such as c-Myc, Ccnd1, Hif1α, and Cd44 (Fig. 4I). Consistently, LOXL2 deletion significantly increased the expression level of these genes in Loxl2−/− mice uterus (Supplementary Fig. S6H and S6I).

LOXL2 deletion activates oncogenic signaling pathways during tumor progression

To further understand how LOXL2 deficiency induces uterine hypertrophy and carcinogenesis, we performed microarray analysis. Although the expression of cell cycle– and DNA replication–related genes was markedly elevated in enlarged uterus, the PI3K–AKT pathway was also upregulated in LOXL2-deficient uterus (Fig. 5A and B). The Hippo pathway, critical for organ enlargement and cancer development (39), remained largely unaffected in Loxl2−/− mice (Fig. 5B and C), suggesting that LOXL2 deficiency regulates uterine enlargement independent of the Hippo pathway. Consistently, the expression of genes related to cytokine signaling, including those encoding interleukin and chemokine family members, multiple protumorigenic signal transduction pathway members, and cell proliferation–related proteins, was markedly elevated in human uterine endometrial cancer samples with low LOXL2 expression (Supplementary Fig. S7A). A volcano plot further showed that the expression of cell-cycle genes, such as c-Myc, Ccnd1, and Cdk2, was markedly upregulated in an enlarged uterus (Fig. 5C), further supporting our ChIP results (Fig. 3J and Fig. 4I). RT-qPCR assays also confirmed the increased expression of genes related to cell-cycle progression in the enlarged uterus and uterine carcinoma (Fig. 5D and E). In addition, LOXL2 knockout caused uterine cell apoptosis in Loxl2−/− mice (Fig. 5F and G; Supplementary Fig. S6J). Altogether, our data demonstrate that the pathogenesis of spontaneous carcinogenesis in the uterus is closely related to LOXL2-mediated H3K36ac deacetylation, but the various effects of LOXL2 deficiency that promote cancer also enhance tumor progression.

Figure 5.

LOXL2 deletion activates multiple signaling pathways for tumor promotion. A, Heatmap of gene expression in Loxl2+/+ mice uterus (Loxl2+/+-UT) and enlarged uterus of Loxl2−/− mice (Loxl2−/−-UE; n = 3). B, Gene ontology analysis for Loxl2−/−-UE versus Loxl2+/+-UT (n = 3). C, Volcano plot showed gene expression profiles of Loxl2−/−-UE versus Loxl2+/+-UT. Upregulated cell-cycle genes are shown in the red area. Black spots, no significant change in gene expression of the Hippo pathway (n = 3). D, RT-qPCR analysis confirmed upregulated genes in Loxl2−/−-UE (n = 3). E, RT-qPCR analysis confirmed upregulated genes in the uterine carcinoma of Loxl2−/− mice (Loxl2+/+-UC; n = 3). F and G, Immunofluorescence staining of cleaved-caspase-3 in Loxl2−/−-UE. (n = 5). Scale bars, 200 μm. All data represent the mean ± SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 5.

LOXL2 deletion activates multiple signaling pathways for tumor promotion. A, Heatmap of gene expression in Loxl2+/+ mice uterus (Loxl2+/+-UT) and enlarged uterus of Loxl2−/− mice (Loxl2−/−-UE; n = 3). B, Gene ontology analysis for Loxl2−/−-UE versus Loxl2+/+-UT (n = 3). C, Volcano plot showed gene expression profiles of Loxl2−/−-UE versus Loxl2+/+-UT. Upregulated cell-cycle genes are shown in the red area. Black spots, no significant change in gene expression of the Hippo pathway (n = 3). D, RT-qPCR analysis confirmed upregulated genes in Loxl2−/−-UE (n = 3). E, RT-qPCR analysis confirmed upregulated genes in the uterine carcinoma of Loxl2−/− mice (Loxl2+/+-UC; n = 3). F and G, Immunofluorescence staining of cleaved-caspase-3 in Loxl2−/−-UE. (n = 5). Scale bars, 200 μm. All data represent the mean ± SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

Low nuclear LOXL2 expression together with a high H3K36ac level is indicative of a poor prognosis in uterine endometrial cancer

To assess the pathologic significance of LOXL2 in clinical oncology, we first analyzed LOXL2 expression in TCGA data. LOXL2, which resides on human chromosome 8p21.2–21.3 (40, 41), was also deleted or mutated in human uterine cancer patient samples (Supplementary Fig. S7B–S7D). Next, the nuclear LOXL2 and H3K36ac levels in uterine endometrial cancer samples and cervical cancer samples were assessed by IHC. Low nuclear LOXL2 expression was associated with high H3K36ac intensity in uterine endometrial cancer and cervical cancer (Fig. 6A and B; Supplementary Fig. S7E and S7F). Kaplan–Meier analysis showed that low nuclear LOXL2 expression and a high H3K36ac level were remarkably correlated with reduced recurrence-free survival in patients (Fig. 6C; Supplementary Table S7). The histogram showing the frequency distribution of the H3K36ac level and LOXL2 expression scores showed that nuclear LOXL2 expression tended to be negatively correlated with the H3K36ac level in uterine endometrial cancer and cervical cancer (Fig. 6D; Supplementary Fig. S7G). Notably, uterine endometrial cancer patients with a high H3K36ac level or insufficient nuclear LOXL2 expression had significantly worse overall survival than the other patients (Fig. 6C and E). These data indicate that low nuclear LOXL2 expression together with a high H3K36ac level can potentially serve as a prognostic biomarker for uterine endometrial cancer patients.

Figure 6.

Low LOXL2 expression with a high H3K36ac level predicts a poor prognosis in uterine endometrial cancer patients. A, IHC staining for H3K36ac and LOXL2 in human uterine endometrial cancer samples. Scale bars, 100 μm. B, H3K36ac and LOXL2 expression patterns in human normal endometrium (n = 10), stromal sarcoma (n = 35), and uterine endometrial cancer (n = 101) samples. C, Kaplan–Meier analysis showed that high H3K36ac or low nuclear LOXL2 expression significantly reduced the overall survival rate of uterine endometrial cancer patients. D, Histogram frequency distribution diagram of H3K36ac and LOXL2 expression scores in normal endometrium, stromal sarcoma, and uterine endometrial cancer, respectively. E, The death rate of uterine endometrial cancer patients was analyzed.

Figure 6.

Low LOXL2 expression with a high H3K36ac level predicts a poor prognosis in uterine endometrial cancer patients. A, IHC staining for H3K36ac and LOXL2 in human uterine endometrial cancer samples. Scale bars, 100 μm. B, H3K36ac and LOXL2 expression patterns in human normal endometrium (n = 10), stromal sarcoma (n = 35), and uterine endometrial cancer (n = 101) samples. C, Kaplan–Meier analysis showed that high H3K36ac or low nuclear LOXL2 expression significantly reduced the overall survival rate of uterine endometrial cancer patients. D, Histogram frequency distribution diagram of H3K36ac and LOXL2 expression scores in normal endometrium, stromal sarcoma, and uterine endometrial cancer, respectively. E, The death rate of uterine endometrial cancer patients was analyzed.

Close modal

LOXL2 inhibition reduces the response to cisplatin + anti–PD-1 combination immunotherapy

The combination of platinum-based chemotherapy and PD-1 blockade has emerged as a first-line treatment for multiple malignancies (42). Given the crucial role of the LOXL2–H3K36ac axis in tumorigenesis and patient prognosis, we speculated that LOXL2 may also regulate the response to anti–PD-1 immunotherapy. We used only the mouse uterine cervix tumor cell line U14 (30) and knocked down LOXL2 expression. The expression of H3K36ac was dramatically upregulated in LOXL2-knockdown U14 cells, suggesting that the biochemical and functional properties of LOXL2 were also conserved in murine U14 cells (Fig. 7A; Supplementary Fig. S2; Supplementary Fig. S8A). PD-1 and PD-L1 expression is critical for anti–PD-1 therapeutic efficacy (42). We found that U14 cells had significantly high PD-L1 expression and that CD8+ T cells had upregulated PD-1 levels, indicating a potential immunotherapeutic effect in the U14 tumor model (Supplementary Fig. S8B–S8D). Next, we constructed a syngeneic tumor model with immunocompetent BALB/c mice (30). Seven days after U14 tumor challenge, the tumor-bearing mice received anti–PD-1 blockade treatment together with cisplatin every three days (Fig. 7B). LOXL2 inhibited U14 tumor growth and H3K36ac-regulated gene expression in vivo. Importantly, cisplatin and anti–PD-1 antibody in combination profoundly retarded U14 tumor growth. However, the combination of LOXL2 inhibition and cisplatin + anti–PD-1 therapy exerted a significant resistance effect compared with that in the control group, indicating that LOXL2 reduced the resistance phenotype of U14 cells to the combined immunotherapy (Fig. 7CE; Supplementary Fig. S8E and S8F; Supplementary Table S8). Furthermore, the number of tumor-infiltrating cytotoxic CD8+ T cells in the LOXL2-knockdown tumors was barely changed upon the combined treatment, whereas the control group showed significantly increased CD8+ T-cell infiltration within tumors upon combined treatment (Fig. 7F and G). In contrast, LOXL2 overexpression significantly attenuated U14 tumor growth and enhanced responses to combined immunotherapy compared with the control group (Fig. 7B; Supplementary Fig. S8G–S8K; Supplementary Table S8). Taken together, our results demonstrate that LOXL2 inhibition reduces the response of U14 tumor cells to cisplatin and anti–PD-1 combination immunotherapy.

Figure 7.

LOXL2 inhibition promotes the resistance to cisplatin + anti–PD-1 combined immunotherapy. A, shLOXL2 efficiently inhibited LOXL2 expression in U14 cells (n = 3). B, Schematic of the cisplatin and anti–PD-1 combined immunotherapy in the U14 tumor model. C, Photographs of representative U14 tumors following the indicated treatments on day 21. D and E, Tumor growth and tumor weight at 21 days after U14 challenge and treated with indicated therapy (n = 6). F and G, Quantification of tumor-infiltrating CD8+ T lymphocytes in U14 tumor following the indicated treatments (n = 4). H, Graphic illustration of H3K36ac deacetylation by LOXL2. LOXL2 specifically catalyzes H3K36ac deacetylation or deacetylimination to restrain cancer progression-related gene expression. All data represent the mean ± SD. *, P < 0.05; **, P < 0.01; ns, nonsignificant.

Figure 7.

LOXL2 inhibition promotes the resistance to cisplatin + anti–PD-1 combined immunotherapy. A, shLOXL2 efficiently inhibited LOXL2 expression in U14 cells (n = 3). B, Schematic of the cisplatin and anti–PD-1 combined immunotherapy in the U14 tumor model. C, Photographs of representative U14 tumors following the indicated treatments on day 21. D and E, Tumor growth and tumor weight at 21 days after U14 challenge and treated with indicated therapy (n = 6). F and G, Quantification of tumor-infiltrating CD8+ T lymphocytes in U14 tumor following the indicated treatments (n = 4). H, Graphic illustration of H3K36ac deacetylation by LOXL2. LOXL2 specifically catalyzes H3K36ac deacetylation or deacetylimination to restrain cancer progression-related gene expression. All data represent the mean ± SD. *, P < 0.05; **, P < 0.01; ns, nonsignificant.

Close modal

A secreted form of LOXL2 has long been considered to play a critical role in the oxidative deamination of collagen and tropoelastin in the extracellular space. LOXL2 in cells has been reported for histone oxidation, heterochromatin reorganization, and transcriptional regulation induction (21–23, 36, 43). We previously reported that LOXL3 is an deacetylase responsible for STAT3 deacetylation during CD4+ Th17 differentiation (5). Recently, we reported that the synaptic AMPA receptors, GluR1 and GluR2 (GluR1/2), were acetylated upon glutamate stimulation in cells. Acetylated-GluR1/2 were deacetylated by LOXL2 in the cytoplasm (44). Like LOXL3, both the N-terminal 4xSRCR repeats and the C-terminal catalytic domain carry redundant deacetylase activity (5), which bestows LOXL2 as the most efficient histone deacetylase responsible for H3K36ac deacetylation induction in the nuclei. LOXL2 and LOXL3 also carry the capacity to catalyze deacetylimination by removing acetyl-amine from acetyl-lysine residues, leading to allysine formation. In addition, the lysine flanking sequences of the substrate protein such as H3K36ac here or H3K4me3 reported previously by others could be critical for LOXL2 catalytic activity (23).

Histone acetylation, an epigenetic mechanism closely related to active gene transcription, occurs at lysine residues and is reversibly regulated by HATs and HDACs in mammals (45, 46). The level of H3K36ac is enriched in euchromatin and peaks in multiple gene promoter regions in plants (47). Depletion of the H3K36 methyltransferase NSD2 increases the H3K36ac intensity and promotes oncogenesis- and adipogenesis-associated gene expression (48). Under the conditions used, LOXL2 deficiency significantly increased the H3K36ac intensity and enhanced the H3K36ac-regulated gene expression for cell proliferation. In contrast, LOXL2 overexpression reduced the H3K36ac intensity and inhibited cell growth, suggesting that LOXL2-mediated H3K36ac deacetylation is critical for the regulation of cell proliferation–related genes. Emerging data show that epigenetic regulation in the promoter region of CD44 and HIF1a is also modulated by multiple epigenetic factors (49, 50). LOXL2 deficiency increased the H3K36ac intensity at the promoters of Hif1a and Cd44 and therefore upregulated the transcriptional activation of Hif1a and Cd44 in mice uterus. As a transcription factor, HIF1a in turn turns on LOX genes (LOX, LOXL2, and LOXL4) for expression (49).

As an amine oxidase, LOXL2 aberrant expression has been suspected for either its oncogenic activity (4, 6–16, 18–20) or tumor suppressor activity (41, 51, 52). For instance, an elevated LOXL2 protein level in the cytoplasm promotes EMT in cervical cancer (8), whereas LOXL2 deficiency does not affect dermal development, homeostasis, or tumor stroma formation induced by DMBA/TPA (53). Therefore, LOXL2 seems to behave quite differently even oppositely in various types cells or organs. Given that H3K36M mutation impairs the differentiation of mesenchymal progenitor, results in undifferentiated sarcoma, and is associated with chondroblastoma development, dysregulation of H3K36 modification can be tightly associated with developmental defect and cancer development (54). In Loxl2−/− mice, an elevated or constitutive H3K36 acetylation was strongly correlated with uterine enlargement and carcinoma. Uterine enlargement can be observed in patients with fibroid, adenomyosis, or endometrial carcinoma, pronouncedly associating with cell-cycle progression (55). In addition to numerous cell-cycle–promoting and cell proliferation genes, genes involved in various oncogenic pathways were all upregulated in Loxl2−/− mice, suggesting a critical role of LOXL2-in H3K36ac-dependent genes for expression. The inverse correlation between LOXL2 expression and the H3K36ac level predicts a poor prognosis in patients with uterine endometrial cancer. Thus, LOXL2 can serve as a new epigenetic regulator for uterine cancer progression and a putative target for uterine cancer therapy.

Pembrolizumab (a PD-1 inhibitor) is a second-line treatment for malignant gynecologic tumors, and a clinical trial showed that pembrolizumab improves the prognosis of patients with PD-1–positive cancer (42). However, the efficacy of anti–PD-1 treatment for uterine cancer is low, and the underlying antagonistic mechanism is unclear. H3K4 demethylase KDM5B enhances responses to anti–PD-1 therapy in the mouse melanoma model, suggesting that epigenetic regulators could induce robust antitumor immune responses and overcome resistance to immunotherapy (56). Consistently, LOXL2 deficiency promotes the expression of LOXL2–H3K36ac axis–dependent cell cycle and proliferation genes and attenuates the antitumor effects of cisplatin and anti–PD-1 combined immunotherapy in mouse U14 tumors. LOXL2 increased cytotoxic CD8+ T-cell infiltration into the tumors, suggesting a conversion of the tumor immune environment from a noninflamed “cold tumor” to an inflamed “hot tumor.” Given that the infiltration of CD8+ T cells is strongly associated with the progression of cervical cancer, PD-1 expression on CD8+ T cells is an independent prognostic factor for cervical cancer patients (57, 58). Our results provide a novel insight into LOXL2 epigenetic regulation to thereby improve the efficacy of chemotherapy and anti-immune checkpoint therapy via their combination.

In summary, LOXL2 in nuclei controls cellular proliferation in the uterus to restrict organ overgrowth and tumor development via H3K36 epigenetic regulation (Fig. 7H).

Z. Chang reports grants from the National Nature Science Foundation of China during the conduct of the study and grants from the National Nature Science Foundation of China outside the submitted work. Y. Chin reports grants from the National Key Basic Research Program of China, the National Key Research and Development Project, and Wenzhou Leading Talent Innovation and Entrepreneurship Project during the conduct of the study. No disclosures were reported by the other authors.

X. Lu: Data curation, validation, investigation, writing–original draft, writing–review and editing. D.E. Xin: Resources, investigation, visualization, writing–original draft. J.K. Du: Investigation. Q.C. Zou: Data curation, methodology. Q. Wu: Software, formal analysis. Y.S. Zhang: Resources. W. Deng: Methodology, project administration. J. Yue: Resources, visualization. X.S. Fan: Software, project administration. Y. Zeng: Data curation, validation. X. Cheng: Formal analysis, investigation. X. Li: Project administration. Z. Hou: Software, visualization, methodology. M. Mohan: Software, methodology. T.C. Zhao: Resources, project administration. X. Lu: Resources. Z. Chang: Project administration. L. Xu: Project administration. Y. Sun: Methodology, project administration. X. Zu: Resources, project administration. Y. Zhang: Software, visualization. Y.E. Chin: Conceptualization, supervision, funding acquisition, writing–original draft, writing–review and editing.

The authors thank Gang Wang for data analysis assistance, Yumei Wang for animal facility, and Xiaoyan Ding for transgenic animal consultancy. This research was supported by the National Key Basic Research Program of China (Grant Nos. 81820108023, 82030077, and 81530083), the National Key Research and Development Project (2016YFC1302402 and 2018YFC1705505), and the Wenzhou Leading Talent Innovation and Entrepreneurship Project (RX2016001).

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.

Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).

1.
Wen
B
,
Xu
LY
,
Li
EM
.
LOXL2 in cancer: regulation, downstream effectors and novel roles
.
Biochim Biophys Acta Rev Cancer
2020
;
1874
:
188435
.
2.
Meier
AA
,
Moon
HJ
,
Toth
Rt
,
Folta-Stogniew
E
,
Kuczera
K
,
Middaugh
CR
, et al
.
Oligomeric states and hydrodynamic properties of lysyl oxidase-like 2
.
Biomolecules
2021
;
11
:
1846
.
3.
Wang
TH
,
Hsia
SM
,
Shieh
TM
.
Lysyl oxidase and the tumor microenvironment
.
Int J Mol Sci
2016
;
18
:
62
.
4.
Barker
HE
,
Chang
J
,
Cox
TR
,
Lang
G
,
Bird
D
,
Nicolau
M
, et al
.
LOXL2-mediated matrix remodeling in metastasis and mammary gland involution
.
Cancer Res
2011
;
71
:
1561
72
.
5.
Ma
L
,
Huang
C
,
Wang
XJ
,
Xin
DE
,
Wang
LS
,
Zou
QC
, et al
.
Lysyl oxidase 3 is a dual-specificity enzyme involved in STAT3 deacetylation and deacetylimination modulation
.
Mol Cell
2017
;
65
:
296
309
.
6.
Peinado
H
,
Moreno-Bueno
G
,
Hardisson
D
,
Pérez-Gómez
E
,
Santos
V
,
Mendiola
M
, et al
.
Lysyl oxidase-like 2 as a new poor prognosis marker of squamous cell carcinomas
.
Cancer Res
2008
;
68
:
4541
50
.
7.
Torres
S
,
Garcia-Palmero
I
,
Herrera
M
,
Bartolomé
RA
,
Peña
C
,
Fernandez-Aceñero
MJ
, et al
.
LOXL2 is highly expressed in cancer-associated fibroblasts and associates to poor colon cancer survival
.
Clin Cancer Res
2015
;
21
:
4892
902
.
8.
Tian
J
,
Sun
HX
,
Li
YC
,
Jiang
L
,
Zhang
SL
,
Hao
Q
.
LOXL 2 promotes the epithelial-mesenchymal transition and malignant progression of cervical cancer
.
Onco Targets Ther
2019
;
12
:
8947
54
.
9.
Salvador
F
,
Martin
A
,
López-Menéndez
C
,
Moreno-Bueno
G
,
Santos
V
,
Vázquez-Naharro
A
, et al
.
Lysyl oxidase-like protein LOXL2 promotes lung metastasis of breast cancer
.
Cancer Res
2017
;
77
:
5846
59
.
10.
Kasashima
H
,
Yashiro
M
,
Kinoshita
H
,
Fukuoka
T
,
Morisaki
T
,
Masuda
G
, et al
.
Lysyl oxidase-like 2 (LOXL2) from stromal fibroblasts stimulates the progression of gastric cancer
.
Cancer Lett
2014
;
354
:
438
46
.
11.
Wong
CC
,
Tse
AP
,
Huang
YP
,
Zhu
YT
,
Chiu
DK
,
Lai
RK
, et al
.
Lysyl oxidase-like 2 is critical to tumor microenvironment and metastatic niche formation in hepatocellular carcinoma
.
Hepatology
2014
;
60
:
1645
58
.
12.
Barry-Hamilton
V
,
Spangler
R
,
Marshall
D
,
McCauley
S
,
Rodriguez
HM
,
Oyasu
M
, et al
.
Allosteric inhibition of lysyl oxidase-like-2 impedes the development of a pathologic microenvironment
.
Nat Med
2010
;
16
:
1009
17
.
13.
Peng
DH
,
Ungewiss
C
,
Tong
P
,
Byers
LA
,
Wang
J
,
Canales
JR
, et al
.
ZEB1 induces LOXL2-mediated collagen stabilization and deposition in the extracellular matrix to drive lung cancer invasion and metastasis
.
Oncogene
2017
;
36
:
1925
38
.
14.
Mahjour
F
,
Dambal
V
,
Shrestha
N
,
Singh
V
,
Noonan
V
,
Kantarci
A
, et al
.
Mechanism for oral tumor cell lysyl oxidase like-2 in cancer development: synergy with PDGF-AB
.
Oncogenesis
2019
;
8
:
34
.
15.
Hong
X
,
Yu
JJ
.
Silencing of lysyl oxidase-like 2 inhibits the migration, invasion and epithelial-to-mesenchymal transition of renal cell carcinoma cells through the Src/FAK signaling pathway
.
Int J Oncol
2019
;
54
:
1676
90
.
16.
Barker
HE
,
Bird
D
,
Lang
G
,
Erler
JT
.
Tumor-secreted LOXL2 activates fibroblasts through FAK signaling
.
Mol Cancer Res
2013
;
11
:
1425
36
.
17.
Kim
BR
,
Dong
SM
,
Seo
SH
,
Lee
JH
,
Lee
JM
,
Lee
SH
, et al
.
Lysyl oxidase-like 2 (LOXL2) controls tumor-associated cell proliferation through the interaction with MARCKSL1
.
Cell Signal
2014
;
26
:
1765
73
.
18.
Cuevas
EP
,
Eraso
P
,
Mazón
MJ
,
Santos
V
,
Moreno-Bueno
G
,
Cano
A
, et al
.
LOXL2 drives epithelial-mesenchymal transition via activation of IRE1-XBP1 signalling pathway
.
Sci Rep
2017
;
7
:
44988
.
19.
Canesin
G
,
Cuevas
EP
,
Santos
V
,
López-Menéndez
C
,
Moreno-Bueno
G
,
Huang
Y
, et al
.
Lysyl oxidase-like 2 (LOXL2) and E47 EMT factor: novel partners in E-cadherin repression and early metastasis colonization
.
Oncogene
2015
;
34
:
951
64
.
20.
Peinado
H
,
Del Carmen Iglesias-de la Cruz
M
,
Olmeda
D
,
Csiszar
K
,
Fong
KS
,
Vega
S
, et al
.
A molecular role for lysyl oxidase-like 2 enzyme in snail regulation and tumor progression
.
EMBO J
2005
;
24
:
3446
58
.
21.
Herranz
N
,
Dave
N
,
Millanes-Romero
A
,
Pascual-Reguant
L
,
Morey
L
,
Díaz
VM
, et al
.
Lysyl oxidase-like 2 (LOXL2) oxidizes trimethylated lysine 4 in histone H3
.
FEBS J
2016
;
283
:
4263
73
.
22.
Millanes-Romero
A
,
Herranz
N
,
Perrera
V
,
Iturbide
A
,
Loubat-Casanovas
J
,
Gil
J
, et al
.
Regulation of heterochromatin transcription by Snail1/LOXL2 during epithelial-to-mesenchymal transition
.
Mol Cell
2013
;
52
:
746
57
.
23.
Cebrià-Costa
JP
,
Pascual-Reguant
L
,
Gonzalez-Perez
A
,
Serra-Bardenys
G
,
Querol
J
,
Cosín
M
, et al
.
LOXL2-mediated H3K4 oxidation reduces chromatin accessibility in triple-negative breast cancer cells
.
Oncogene
2020
;
39
:
79
121
.
24.
Gopalakrishnan
R
,
Marr
SK
,
Kingston
RE
,
Winston
F
.
A conserved genetic interaction between Spt6 and Set2 regulates H3K36 methylation
.
Nucleic Acids Res
2019
;
47
:
3888
903
.
25.
Wagner
EJ
,
Carpenter
PB
.
Understanding the language of Lys36 methylation at histone H3
.
Nat Rev Mol Cell Biol
2012
;
13
:
115
26
.
26.
Morris
SA
,
Rao
B
,
Garcia
BA
,
Hake
SB
,
Diaz
RL
,
Shabanowitz
J
, et al
.
Identification of histone H3 lysine 36 acetylation as a highly conserved histone modification
.
J Biol Chem
2007
;
282
:
7632
40
.
27.
Pai
CC
,
Deegan
RS
,
Subramanian
L
,
Gal
C
,
Sarkar
S
,
Blaikley
EJ
, et al
.
A histone H3K36 chromatin switch coordinates DNA double-strand break repair pathway choice
.
Nat Commun
2014
;
5
:
4091
.
28.
Lu
XF
,
Cao
XY
,
Zhu
YJ
,
Wu
ZR
,
Zhuang
X
,
Shao
MY
, et al
.
Histone deacetylase 3 promotes liver regeneration and liver cancer cells proliferation through signal transducer and activator of transcription 3 signaling pathway
.
Cell Death Dis
2018
;
9
:
398
.
29.
Gu
B
,
Feng
H
,
Dong
J
,
Zhang
H
,
Bian
X
,
Liu
Y
.
The establishment and characterization of a continuous cell line of mouse cervical carcinoma
.
Chin J Clin Oncol
2008
;
5
:
44
8
.
30.
Chang
M
,
Wang
M
,
Wang
M
,
Shu
M
,
Ding
B
,
Li
C
, et al
.
A multifunctional cascade bioreactor based on hollow-structured Cu(2) MoS(4) for synergetic cancer chemo-dynamic therapy/starvation therapy/phototherapy/immunotherapy with remarkably enhanced efficacy
.
Adv Mater
2019
;
31
:
e1905271
.
31.
Salvesen
HB
,
Stefansson
I
,
Kalvenes
MB
,
Das
S
,
Akslen
LA
.
Loss of PTEN expression is associated with metastatic disease in patients with endometrial carcinoma
.
Cancer
2002
;
94
:
2185
91
.
32.
Zhang
Z
,
Zhang
Y
,
Xia
S
,
Kong
Q
,
Li
S
,
Liu
X
, et al
.
Gasdermin E suppresses tumour growth by activating anti-tumour immunity
.
Nature
2020
;
579
:
415
20
.
33.
Zhang
X
,
Wang
Q
,
Wu
J
,
Wang
J
,
Shi
Y
,
Liu
M
.
Crystal structure of human lysyl oxidase-like 2 (hLOXL2) in a precursor state
.
Proc Natl Acad Sci U S A
2018
;
115
:
3828
33
.
34.
Harper
JV
.
Synchronization of cell populations in G1–S and G2–M phases of the cell cycle
.
Methods Mol Biol
2005
;
296
:
157
66
.
35.
Coursey
T
,
Milutinovic
M
,
Regedanz
E
,
Brkljacic
J
.
Arabidopsis histone reader EMSY-LIKE 1 binds H3K36 and suppresses geminivirus infection
.
J Virol
2018
;
92
:
e00219
18
.
36.
Martin
A
,
Salvador
F
,
Moreno-Bueno
G
,
Floristán
A
,
Ruiz-Herguido
C
,
Cuevas
EP
, et al
.
Lysyl oxidase-like 2 represses Notch1 expression in the skin to promote squamous cell carcinoma progression
.
EMBO J
2015
;
34
:
1090
109
.
37.
Liu
X
,
Zhao
Y
,
Gao
J
,
Pawlyk
B
,
Starcher
B
,
Spencer
JA
, et al
.
Elastic fiber homeostasis requires lysyl oxidase-like 1 protein
.
Nat Genet
2004
;
36
:
178
82
.
38.
Liu
G
,
Daneshgari
F
,
Li
M
,
Lin
D
,
Lee
U
,
Li
T
, et al
.
Bladder and urethral function in pelvic organ prolapsed lysyl oxidase like-1 knockout mice
.
BJU Int
2007
;
100
:
414
8
.
39.
Ma
S
,
Meng
Z
,
Chen
R
,
Guan
KL
.
The Hippo pathway: biology and pathophysiology
.
Annu Rev Biochem
2019
;
88
:
577
604
.
40.
Zhan
XH
,
Jiao
JW
,
Zhang
HF
,
Xu
XE
,
He
JZ
,
Li
RL
, et al
.
LOXL2 upregulates phosphorylation of ezrin to promote cytoskeletal reorganization and tumor cell invasion
.
Cancer Res
2019
;
79
:
4951
64
.
41.
Zhan
P
,
Shen
XK
,
Qian
Q
,
Zhu
JP
,
Zhang
Y
,
Xie
HY
, et al
.
Down-regulation of lysyl oxidase-like 2 (LOXL2) is associated with disease progression in lung adenocarcinomas
.
Med Oncol
2012
;
29
:
648
55
.
42.
Pembrolizumab OK'd for cervical cancer
.
Cancer Discov
2018
;
8
:
904
.
43.
Iturbide
A
,
Pascual-Reguant
L
,
Fargas
L
,
Cebrià
JP
,
Alsina
B
,
Cebrià
JP
,
García de
Herreros A
, et al
.
LOXL2 oxidizes methylated TAF10 and controls TFIID-dependent genes during neural progenitor differentiation
.
Mol Cell
2015
;
58
:
755
66
.
44.
Li
XR
,
Cheng
X
,
Sun
J
,
Xu
YS
,
Chen
N
,
Hao
Y
, et al
.
Acetylation-dependent glutamate receptor GluR signalosome formation for STAT3 activation in both transcriptional and metabolism regulation
.
Cell Death Discov
2021
;
7
:
11
.
45.
Kim
H
,
Jo
H
,
Seo
HD
,
Park
HS
,
Lee
D
.
Chd1p recognizes H3K36Ac to maintain nucleosome positioning near the transcription start site
.
Biochem Biophys Res Commun
2018
;
503
:
1200
6
.
46.
Lin
C
,
Li
H
,
Liu
J
,
Hu
Q
,
Zhang
S
,
Zhang
N
, et al
.
Arginine hypomethylation-mediated proteasomal degradation of histone H4-an early biomarker of cellular senescence
.
Cell Death Differ
2020
;
27
:
2697
709
.
47.
Wang
WW
,
Angulo-Ibanez
M
,
Lyu
J
,
Kurra
Y
,
Tong
Z
,
Wu
B
, et al
.
A click chemistry approach reveals the chromatin-dependent histone H3K36 deacylase nature of SIRT7
.
J Am Chem Soc
2019
;
141
:
2462
73
.
48.
Zhuang
L
,
Jang
Y
.
Depletion of Nsd2-mediated histone H3K36 methylation impairs adipose tissue development and function
.
Nat Commun
2018
;
9
:
1796
.
49.
Schito
L
,
Semenza
GL
.
Hypoxia-inducible factors: master regulators of cancer progression
.
Trends Cancer
2016
;
2
:
758
70
.
50.
Chen
C
,
Zhao
S
,
Karnad
A
,
Freeman
JW
.
The biology and role of CD44 in cancer progression: therapeutic implications
.
J Hematol Oncol
2018
;
11
:
64
.
51.
Matsuo
A
,
Tanida
R
,
Yanagi
S
,
Tsubouchi
H
,
Miura
A
,
Shigekusa
T
, et al
.
Significance of nuclear LOXL2 inhibition in fibroblasts and myofibroblasts in the fibrotic process of acute respiratory distress syndrome
.
Eur J Pharmacol
2021
;
892
:
173754
.
52.
Schmidt
H
,
Semjonow
A
,
Csiszar
K
,
Korsching
E
,
Brandt
B
,
Eltze
E
.
Mapping of a deletion interval on 8p21–22 in prostate cancer by gene dosage PCR
.
Verh Dtsch Ges Pathol
2007
;
91
:
302
7
.
53.
Kober
KI
,
Cano
A
,
Géraud
C
,
Sipilä
K
,
Mobasseri
SA
,
Philippeos
C
, et al
.
Loxl2 is dispensable for dermal development, homeostasis and tumour stroma formation
.
PLoS One
2018
;
13
:
e0199679
.
54.
Lu
C
,
Jain
SU
,
Hoelper
D
,
Bechet
D
,
Molden
RC
,
Ran
L
, et al
.
Histone H3K36 mutations promote sarcomagenesis through altered histone methylation landscape
.
Science
2016
;
352
:
844
9
.
55.
Huvila
J
,
Pors
J
,
Thompson
EF
,
Gilks
CB
.
Endometrial carcinoma: molecular subtypes, precursors and the role of pathology in early diagnosis
.
J Pathol
2021
;
253
:
355
65
.
56.
Zhang
SM
,
Cai
WL
,
Liu
X
,
Thakral
D
,
Luo
J
,
Chan
LH
, et al
.
KDM5B promotes immune evasion by recruiting SETDB1 to silence retroelements
.
Nature
2021
;
598
:
682
7
.
57.
Fan
P
,
Li
X
,
Feng
Y
,
Cai
H
,
Dong
D
,
Peng
Y
, et al
.
PD-1 expression status on CD8+ tumour infiltrating lymphocytes associates with survival in cervical cancer
.
Front Oncol
2021
;
11
:
678758
.
58.
Karim
R
,
Jordanova
ES
,
Piersma
SJ
,
Kenter
GG
,
Chen
L
,
Boer
JM
, et al
.
Tumor-expressed B7-H1 and B7-DC in relation to PD-1+ T-cell infiltration and survival of patients with cervical carcinoma
.
Clin Cancer Res
2009
;
15
:
6341
7
.