Association of Histone H3 Trimethylation in Circulating Monocytes with Lack of Early Recurrence in Patients with Bladder Cancer following BCG Induction Therapy

Study participants

A subset of seven patients from our previous study [4], recruited between September 2018 and March 2019, was selected for ChIP-seq analysis. Patients were recruited following informed consent and protocol approval by the Queen’s University Health Sciences and Affiliated Hospitals Research Ethics Board (approval number UROL-282-13). All resected tumors were at the Ta stage but varied in grade and American Urological Association risk (see Table 1). Four of these participants suffered neoplastic recurrence within a year while three remained in remission at the time of this study. For ChIP-qPCR validations, seven additional patients with Ta or T1 stage tumors (recruited between March 2019 and October 2019; Table 2) were selected. Three of these patients suffered recurrences within a year of therapy initiation. All patients were BCG naïve prior to enrollment in this study.

Blood collection and cell isolations

Peripheral blood (20 mL) was collected during patient clinical visits at two time-points (Fig. 1A): before the first BCG instillation (week one) and before the sixth BCG instillation (week six). Blood was collected in Vacutainer K2-EDTA-K tubes (Becton Dickinson, Mississauga, ON) and transferred to the laboratory within two hours for isolation of peripheral blood mononuclear cells (PBMCs) by density gradient centrifugation using Lymphoprep (STEMCELL Technologies, Vancouver, Canada) according to the manufacturer’s instructions. Cells were frozen in 8% DMSO/fetal bovine serum at a concentration of 1×10⁷ cells per vial (1 mL) and stored in liquid nitrogen. Monocytes were subsequently isolated from thawed PBMCs using an EasySep^TM Human Monocyte Isolation Kit (STEMCELL Technologies).

OPEN IN VIEWER

Fig. 1

H3K4me3 methylation at promoters of innate immune memory-related genes. A) peripheral blood from patients with NMIBC was collected before the first (week 1) and last (week 6) intravesical BCG instillation of the first induction phase. TURBT: transurethral resection of bladder tumor; BCG: bacillus Calmette-Guérin. B) H3K4me3 in monocytes at promoters of TNFα and TLR4 genes before the first BCG instillation (week 1) and the last (week 6). Blue solid shapes: no-early recurrence (NR) patients used for ChIP-seq; red solid shapes: early recurrence (R) patients used for ChIP-seq; circles: week 1 (pre-BCG); squares: week 6 (before 6th BCG instillation).

Chromatin immunoprecipitation

For each chromatin immunoprecipitation (ChIP) reaction, one million freshly purified monocytes were pelleted, and their chromatin was cross-linked in 1% paraformaldehyde for 10 minutes at room temperature. Cross-linking of chromatin was stopped by adding glycine to a final concentration of 125 mM and incubating under rotation for five minutes at room temperature. Cell pellets were washed twice with ice-cold PBS and stored at –80°C until sonication. Before sonication, nuclei were prepared according to the truChIP Chromatin Shearing Kit (Covaris, Woburn, MA) low cell protocol. Sonication was performed using a Covaris M220 focused ultrasonicator under the following conditions: PIP = 75, duty factor = 5%, CPB = 200, time = 10 minutes, temperature = 7°C. Immunoprecipitation was performed on fragmented chromatin using a rabbit monoclonal anti-H3K4me3 antibody (Cell Signaling Technology, Danvers, MA) and SimpleChIP Plus Sonication IP Kit reagents (Cell Signaling Technology), from which antibody and protein G magnetic beads amounts were adjusted for the number of cells used.

DNA amplification

Immunoprecipitated DNA was analyzed by qPCR using SYBR Green PCR Master Mix (Applied Biosystems – Fisher Scientific, Ottawa, ON, Canada) on a Roche Lightcycler 480. Amplification conditions included one cycle of 10 minutes at 95°C followed by 40 cycles of amplification (20 seconds at 95°C and one minute at 60°C per cycle). Percent input values were calculated to quantify the relative levels of immunoprecipitated histone. Primers used are listed in Table 3.

Sequencing of immunoprecipitated DNA and analysis

Chromatin immunoprecipitation sequencing (ChIP-seq) libraries were constructed using SimpleChIP ChIP-seq DNA Library Prep Kit for Illumina and SimpleChIP ChIP-seq Multiplex Oligos for Illumina (Cell Signaling Technology) according to the manufacturer’s instructions. ChIP-seq libraries were subjected to quality control and the size of each library was quantified so that library concentrations (nM) could be calculated. Libraries were pooled and purified prior to dilution to 4 nM and confirmed in triplicate using a Qubit 4 Fluorometer (ThermoFisher Scientific, Mississauga, ON, Canada). The library pool was denatured and diluted to 1.8 pM in 1.3 mL of HT1 solution and paired-end sequenced with the Illumina NextSeq550 system using the HIGH output v2.5 flowcell resulting in >25 million PE reads per sample. Data were stored in the secure server of the Queen’s University Centre for Advanced Computing. Demultiplexing and Fastq generation were processed using bcl2fastq conversion software (Illumina). Fastq files were aligned using Bowtie against HG38 and duplicates were filtered using Samtools. Peak calling was performed using HOMER [10] with length between peaks and length of a peak set to 370 and 250, respectively.

To identify peaks that best discriminate between patients who suffered from early disease recurrence and patients with no early recurrence at different time points, we used a machine learning-based feature selection algorithm (MFeaST) [11]. The algorithm relies on a set of machine learning techniques to rank the features based on their ability to discriminate between two groups and is highly effective at identifying molecular patterns among samples. The following pair-wise comparisons were performed: early recurrence week one (pre-BCG) versus no early recurrence week one (post-BCG), and early recurrence week six versus no early recurrence week six. To prepare for the MFeaST, the peaks data were segregated by chromosome. For each chromosome, a data matrix was generated where the intensity of a peak was recorded for each sample and for each genomic location block of 400 base pairs. Blocks with no peaks in more than one sample were filtered out. For each comparison category, the processed data were run twice through MFeaST with leave-one-out validation and all available algorithms. First, for each chromosome, the top 5% of the ranked blocks in each chromosome were selected and combined into a single data matrix. The combined data were then passed through MFeaST to generate the final combined ranking. From the top 5% of combined top ranked peaks, only the promoter and exon peaks were used in the subsequent analysis. For each peak, a median-based fold-change value was calculated in each comparison category (e.g., week one versus week six, etc.). Peaks with log2 fold change of ≥2 were used in pathway analysis to identify enriched peaks (2625 genes for week one and 2735 genes for week six). The aggregate output files from these data are presented in the following location (DOI: 10.17632/vgt4jzvfg4.1). Raw data are available upon request to the corresponding authors.

Gene Ontology (GO) enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were performed by applying the “clusterprofiler” package [12]. GO terms were prioritized using a corrected false discovery rate (FDR;<0.05). The significantly enriched KEGG signaling pathways were displayed in the form of a dot plot [13]. Genes involved in Wnt, AMPK and MAPK signaling pathways were selected for validation. The Integrative Genomics Viewer software (IGV, v2.9.4) [14] was used to visualize the selected candidates to confirm the genomic locations of the peaks to be used in the validation.

Validation of differentially active pathways

To validate the newly identified pathways enriched in patients with no early recurrences, we performed an independent H3K4me3 immunoprecipitation experiment on monocytes from the same patients as well as an additional independent cohort using the monoclonal antibody (Cell Signaling Technology) used earlier for ChIP-seq. Primers within the identified differential peaks were used to quantify H3K4me3 in immunoprecipitated chromatin by qPCR on the co-precipitated DNA (see Table 2 for primers).

Quantification of transcript levels

RNA was extracted from flash frozen monocytes of NMIBC patients using a PureLink^TM RNA Mini Kit (Invitrogen). One hundred nanograms of RNA was used for reverse transcription using an AMV 1st Strand cDNA Synthesis Kit (Roche). qPCR was performed using SYBR Green PCR Master Mix (Applied Biosystems) on a Roche Lightcycler 480 system with the following amplification conditions: one cycle of 10 minutes at 95°C followed by 40 cycles of amplification (20 seconds at 95°C and one minute at 60°C per cycle). GAPDH was used as a reference for relative quantitation by ΔCt. The sequences for primers used were as follows: TNFα sense 5’-CCTCTCTCTAATCAGCCCTCTG-3’; TNFα antisense 5’-GAGGACCTGGGAGTAGATGAG-3’; TLR4 sense 5’-AGACCTGTCCCTGAACCCTAT-3’; TLR4 antisense 5’-CGATGGACTTCTAAACCAGCCA-3’; GAPDH sense 5’-GGAGCGAGATCCCTCCAAAAT-3’; GAPDH antisense 5’-GGCTGTTGTCATACTTCTCATGG-3’.

Statistical analysis

Data processing and visualization were performed in MATLAB (Mathworks, Inc., MA, USA, v. R2020a) and GraphPad Prism (GraphPad Software Inc., MA, USA, v. 9.5.1). Clustering results were generated using the t-distributed stochastic neighbor embedding (tSNE) [15]. A two-sided t-test was used to assess differences between no-early recurrence and early recurrence groups at weeks one and six. A one-sided t-test was used for the analysis of the validation results, since the direction of the effect was known. Assumptions of all statistical tests were checked and satisfied. A p-value <0.05 was considered statistically significant.

H3K4me3 at inflammatory gene promoters

Since BCG-induced trained immunity leads to the accumulation of histone 3 lysine 4 trimethylation at pro-inflammatory gene promoters [16], we performed ChIP on chromatin extracted from monocytes using an H3K4me3-specific antibody followed by gene promoter-specific amplification of immunoprecipitated DNA. Specifically, we amplified DNA associated with the promoters of the TNFα and TLR4 genes (known to play a role in trained immunity [17]) in monocytes isolated before treatment (week one) and prior to the last weekly instillation of BCG (week six). Patients who remained recurrence-free for one year (no-early recurrence) had significantly higher H3K4me3 modifications at gene promoters for both TNFα and TLR4 locations in monocytes isolated prior to BCG induction, compared to patients who suffered early recurrences (Fig. 1B, Supplementary Table 1). However, the significance of these differences weakened after five rounds of BCG instillations (Supplementary Table 1). This indicates that, for some NMIBC patients, inflammatory genes may be more transcriptionally accessible even before BCG induction therapy, and could indicate a favorable response to treatment.

Epigenomic profiles of monocytes from patients with and without early recurrence

To determine whether monocytes from patients without early recurrences had a specific epigenomic profile, we performed ChIP-seq on DNA associated with chromatin regions immunoprecipitated with the anti-H3K4me3 antibody. We selected monocytes from the three patients without early recurrences previously identified as showing higher H3K4me3 histone modifications at inflammatory gene promoters as well as the four patients who suffered high-grade early recurrences (Table 1). ChIP-seq profiles were generated using monocytes from these patients at week one and week six. Initial analysis of the peak-calling data showed that the patient group without recurrence at one year had more peaks with average higher scores at the promoter and exon locations compared to the patient group with early recurrence at week one and week six (differences not statistically significant, Supplementary Table 2). To identify genomic locations that are most discriminating between the early recurrence and no early recurrence groups, we performed feature selection using MFeast [11]. The number of peaks at the selected locations was significantly higher in the no early recurrence group compared with the early recurrence group at both week-one and week-six time points (p = 0.02, <0.001, respectively, Supplementary Table 2). Consequently, on average, the peak scores at the selected locations were also higher in the group with no early recurrence at week one and week six (p = 0.04, Supplementary Table 2, Supplementary Figure 1B). Clustering analysis of the selected peak locations revealed that the H3K4me3 epigenomic profiles of patients with and without early recurrence localized to distinct clusters before BCG treatment was initiated (Fig. 2A), suggesting the presence of a distinct chromatin accessibility profile in patients responding successfully to BCG immunotherapy.

GO and KEGG pathway analyses of H3K4me3-associated genes enriched in the groups with and without early recurrence revealed that, at week one, the most significantly enriched pathways in the group without early recurrence included genes involved in Wnt signaling (overlapping multiple GO pathways), as well as in AMPK signaling. At week six, elements of the MAPK signaling pathway were the most enriched in the group without early recurrence (Fig. 2B).

OPEN IN VIEWER

Fig. 2

High-dimensional arrangement and pathway enrichment from H3K4me3 epigenomic profiles of monocytes from patients with NMIBC. A) Two-dimensional t-SNE embeddings of data from no early recurrence (NR, blue) and early recurrence (R, red) samples at week 1 (left panel) and week 6 (right panel). B) Gene Ontology (GO) terms mostly enriched among genes upregulated in recurrence-free patients at week1 one (left panel) and week 6 (right panel). C) Example of the LEPR promoter element that shows noticeable differences in monocytes from recurrence-free (blue) versus early recurrence (red) patients.

Validation of H3K4me3 target genes enriched in monocytes from patients with no early recurrence

To confirm that these identified pathways were associated with monocytes from patients without early recurrent disease, we first performed an independent ChIP on a sample taken from the same population of monocytes used for ChIP-seq. Due to limitations in the number of monocytes isolated from each patient, we were only able to perform this confirmation experiment using monocytes from five of the seven patients used for the ChIP-seq study (set 1, two patients with early recurrence and three without early recurrence, indicated by * in Table 1). We first evaluated the H3K4me3 enrichment at promoter sequences from a gene involved in Wnt signaling (FZD5) that was repeatedly identified in the pathways upregulated in monocytes from week-one patients without early recurrence, as well as promoter sequences of genes from the AMPK signaling pathway (LEPR, IRS2; LEPR peaks visualized in Fig. 2C). Results showed that the levels of H3K4me3 associated with these three gene promoter regions were higher in recurrence-free patients than in patients with early recurrence at week one (Fig. 3A, Supplementary Table 3). We also found that the levels of H3K4me3 enriched for two genes involved in MAPK signaling (AKT3, MET) initially identified following ChIP-seq pathway analysis were higher at week six in monocytes from recurrence-free patients when compared with monocytes from patients with early recurrence, although the difference was not significant (Fig. 3B, Supplementary Table 3). These results show that the candidate genes identified in the ChIP-seq study were representative of H3K4me3 enrichment present in the patients’ monocytes. The levels of H3K4me3 enrichment obtained for the genes involved in the identified pathways were used to cluster the patients before and at the end of treatment by t-SNE. We found that the patients evaluated by qPCR for enriched pathways genes were clustering distinctly between early recurrence and no early recurrence (Fig. 3C). This showed that the clustering observed from the epigenomic profiles was maintained when using a subset of enriched genes quantified independently.

OPEN IN VIEWER

Fig. 3

Validation of gene promoter elements identified as enriched in patients without early recurrence. A) H3K4me3 in monocytes at the promoter sequences of FZD5, LEPR and IRS2. B) H3K4me3 in monocytes at promoter sequences of AKT3 and MET. Circles: week 1 (pre-BCG); squares: week 6 (before 6th BCG instillation) C) Two-dimensional t-SNE embeddings of ChIP-qPCR data at week 1 (left panel) and week 6 (right panel) obtained from monocyte samples of patients without early recurrence (NR, blue) and patients with early recurrence (R, red). Blue solid shapes: monocytes from patients with no early recurrence (NR) used for ChIP-seq; red solid shapes: monocytes from patients with early recurrence (R) used for ChIP-seq.

Further ChIP validation experiments using monocytes from seven independent patients with NMIBC (set 2, three with early recurrences and four without early recurrences) revealed that the higher enrichment of genes (AKT3 and MET) from the MAPK pathway at week six was maintained in this new subset of recurrence-free patients when compared to the new subset of patients who suffered early recurrence (Fig. 4B, Supplentary Table 3). However, the enrichment of pathways associated with recurrence-free status at week one (FZD5 for Wnt pathway, LEPR and IRS2 for AMPK pathway) was not maintained (Fig. 4A). These week-one pathway genes were more enriched at week six in patients without early recurrence when compared with week-six patients with early recurrence (Fig. 4B). Clustering of the candidate genes’ enrichment showed separation according to recurrence status at week six, but not at week one (Fig. 4C). Together, these results indicate that when expanded to patients not profiled in our H3K4me3 study, patients who remain recurrence-free for at least one year may not be identified through a single type of epigenomic H3K4me3 profile at week one of BCG induction therapy, but their profiles at week six reveal a stronger association with patients that did suffer early disease recurrence.

OPEN IN VIEWER

Fig. 4

Methylation levels on monocytes from an independent validation cohort of patients at gene promoters previously identified as enriched in the original group of patients without early disease recurrence. A) H3K4me3 in monocytes at gene promoters for FZD5, LEPR and IRS2 genes. B) H3K4me3 in monocytes at gene promoters for AKT3 and MET genes. C) Two-dimensional t-SNE embeddings of ChIP-qPCR data from monocytes of a validation cohort at week one (left panel) and week six (right panel). Blue solid shapes: monocytes from patients with no early recurrence (NR) used for ChIP-seq; red solid shapes: monocytes from patients with early recurrence (R) used for ChIP-seq.