A panel of epigenetically dysregulated Wnt signaling pathway genes for non-invasive diagnosis of pediatric acute lymphoblastic leukemia

Abstract

BACKGROUND:

Genetic and epigenetic dysregulation of Wnt signaling pathway is widely linked up with abnormal proliferation and/or epithelial-to-mesenchymal transition, in different cancer cell types.

OBJECTIVE:

In the present research, we have tested whether promoter DNA methylation of a set of Wnt/non-Wnt genes such as [cadherin-2 (CDH2)], “present in circulation”, could serve as “bone-marrow biopsy surrogate” and help in diagnosing the status, sub-type or treatment outcome in pediatric acute lymphoblastic leukemia (ALL) patients.

METHODS:

Promoter DNA methylation was quantified in the bisulfite modified blood from the pediatric ALL patients ( $n=$ 86) in comparison with age-matched cancer-free subjects ( $n=$ 28), using real-time methylation specific PCR followed by rigorous statistical validations.

RESULTS:

The observed methylation index, sensitivity and specificity of selected molecular markers (viz., SALL1, WNT5 $\alpha$ , LRP1b, CDH2) in patients’ liquid-biopsies was clinically significant showing high positive correlation in the pre-B ALL cases ( $p$ -value $<$ 0.001). A substantial drop in promoter methylation signal of the follow-up/post-treatment patients was also noted ( $p$ -value $<$ 0.001), which suggested an impending role of minimally invasive liquid-biopsy approach in the diagnosis and/or therapeutic monitoring of pediatric leukemia.

CONCLUSIONS:

Whilst the reported metadata provides useful insight into the plausible involvement of epigenetic glitches in leukemogensis, our findings strengthen the remarkable functional consequences of dysregulated Wnt signaling genes in the hematological malignancies besides offering a novel panel of epigenetic marks.

Keywords

Acute lymphoblastic leukemia biomarker bisulfite conversion CpG island diagnosis promoter hypermethylation Wnt signaling pathway

1. Introduction

With 450,000 new cases and over 320,000 deaths, leukemia (a hematological malignancy) has been placed alongside the top-ranked cancers, worldwide [1, 2]. Amongst the subtypes of leukemia, acute lymphoblastic leukemia (ALL) is most frequently recorded in infants (onset at an age $\leqslant$ 12 months) and the children (aged $\leqslant$ 15 years). Although, nationally representative reports pertaining to the prevalence of ALL in Pakistan are lacking; nonetheless, some prospective studies have documented that pediatric ALL accounts for around 35% of the total registered leukemia cases in contrast to the western world where it represents only 25% [3, 4]. Contrary to the other forms of leukemia, pediatric-/infant- ALL, is multifaceted; hence the scientific classification standards and perfect biomarkers for its diagnosis and follow-up monitoring, are lacking [5, 6]. To this end, only bone marrow biopsy (BMB) can clearly diagnose the disease and judge the treatment effect and therefore patients with ALL undergo BMB examination at least three times, from onset- to the end- of first treatment cycle. Owing to the invasive nature of BMB, potential side-effects and expense-related trepidations of the affected families (coming from poor socioeconomic background), there exists a substantial need for alternative procedures and molecular markers for effective diagnosis and/or prognosis of pediatric ALL.

Despite remarkable advancements made in the fields of human biology and pathophysiology, there are still several ‘grey areas’ in our understanding of the underlying, concealed molecular perturbations that trigger and essentially drive the genesis- and/or progression of hematological malignancies, including ALL [5, 6, 7, 8, 9]. Conventionally, the genetic abnormalities, copy number variations and the chromosomal rearrangements are regarded as the key drivers contributing towards malignant transformation of lymphoid progenitors into the ‘leukemic’ cells, in the B- and/or the T-cell lineages. However, accumulating evidences suggest that epigenetic alterations such as hypo-/hyper-methylation of promoter DNA, histone modifications, acetylations, etc., also trigger the leukemogensis by activating the oncogenes and/or silencing the tumor suppressor genes [7, 8, 9, 10, 11, 12]. Studies have shown that in diverse cancers, promoter DNA and/or CpG islands display raised methylation levels contrasting an overall reduction in the methylation frequencies at the genome-wide scale. Whereas, promoter DNA hypermethylation of certain tumor suppressor genes has been linked to cancer chemotherapeutic resistance; the treatment with methylation inhibitors (e.g., 5 ${}^{\prime}$ -azacytidine), has also shown therapeutic benefits advocating the importance of epigenetic priming mechanisms in leukemogenesis [9, 10, 11, 12, 13, 14, 15, 16, 17, 18].

Owing to the potential involvement of epigenetic glitches in leukemic gene regulations, we here tested the hypothesis whether promoter DNA methylation of a “set of genes” present in circulation, could help in diagnosing the status, sub-type or treatment outcome in pediatric ALL patients, either alone or in combination with available stream of tests. To assess and explore the potential gene signatures with functional CpG site methylation, extensive data mining was performed; interestingly, the search came up with a set of genes that mostly linked up at the interface of Wnt canonical/non-canonical signaling pathways along with a distantly linked non-Wnt gene CDH2 [a modulator of cell migration, proliferation, and stem cell differentiation]. The findings of our wet-lab study are interesting and provide useful insight into the plausible involvement of aberrant promoter DNA methylation in derailing an otherwise finely-tuned normal Wnt signaling cascade, besides offering a panel of gene signatures for potential non-invasive, liquid biopsy-based diagnosis of ALL.

2. Materials and methods

2.1 Study population and sample collection

20 subjects (15 pre-treatment ALL $+$ 5 controls) were enrolled in the test phase of the study. The validation phase, however, was comprised of additional 94 subjects that included pre- ( $n=$ 71) and post-treatment ( $n=$ 25) pediatric ALL patients (age $<$ 15 year) along with age- and gender-matched cancer-free controls ( $n=$ 23). The study was undertaken in accordance with the guidelines duly approved by the Institutional Ethical Review Board [284/15-SBS]. Blood samples were collected from the patients admitted at the “Oncology Ward, The Children’s Hospital, Lahore, Pakistan” with the informed consent of their guardians. Follow-up blood samples were collected at the time of complete remission i.e., blast cells $<$ 5% of mononuclear cells.

2.2 Selection of marker set – systematic data mining

In order to select the candidate genes, whose promoter DNA methylation is associated with ALL, we conducted a systematic review of the literature published during 1997–2019. A total of 337 research articles were retrieved through searching different databases, using keywords such as “Pediatric Acute Lymphoblastic Leukemia”, “Biomarker”, “Diagnostic”, “Hypermethylation”, and “Promoter”. After excluding 56 duplicate records, 281 studies were screened to identify the eligible studies for analysis. From these, 241 studies were further excluded including the 46 review articles, 50 book chapters, 17 encyclopedia records, 56

Figure 1.

(A) PRISMA flow diagram detailing the literature search and inclusion and exclusion criteria for studies. (B) Venn-diagram showing different sample types used in the reported literature. (C) Selection of candidate gene signatures; the frequency of promoter methylation (%) has been plotted against the number of articles reporting their hypermethylation. The genes displaying methylation frequency $\geqslant$ 70% and reported in at least 2 articles, have been selected and marked with arrow. (D) Proteins encoded by the selected candidate genes (encircled with orange lines); and their functional protein-protein interaction analysis using STRING online tool (www.string-db.org).

conference abstracts, 1 mini-review, 1 short communication, 7 studies investigating the role of miRNAs, 3 addressing adult ALL, 4 analyzing the promoter hypermethylation in other leukemia types and 46 monitoring the genetic/genomic/proteomic biomarkers (Fig. 1A). Forty studies were thereafter screened for eligibility; however, individual patient data (IPD) could be sought only for 36 studies. Amongst these, 3 were further excluded as lacking % methylation frequency for individual genes thus only 33 articles could be considered for final analysis.

Table 1

Sequence of the primer pairs used for qMSP

Gene		Sense (5’ $\to$ 3’)	Antisense (5’ $\to$ 3’)
SALL1	M	TCGTAAAGAGGGTTAATTTAGTGTC	TTACTACAAACCCCTATCTCTCCG
	U	TTGTAAAGAGGGTTAATTTAGTGTTG	TTTACTACAAACCCCTATCTCTCCA
RSPO1	M	GTTTTCGGTCGCGTTTTGAC	TACGCTAATACTCTCCGTCGCG
	U	TGAGGGTTTTTGGTTGTGTTTTGAT	CCCTACACTAATACTCTCCATCACA
LRP1b	M	GCGTTTTAGGTTTTAAATTAGTTAGC	GCGACAATCCAATAAATACGC
	U	GTGTTTTAGGTTTTAAATTAGTTAGTGA	AAACTACACAACAATCCAATAAATACAC
WNT5 $\alpha$	M	GTATTTTTCGGAGAAAAAGTTATGC	TACAACCGCGAATTAATATAAACG
	U	GGTATTTTTTGGAGAAAAAGTTATGTG	ACAACCACAAATTAATATAAACATC
CDH2	M	TAGTACGTACGATTCGGAGC	CGACGAAAACCTTAACCG
	U	GGTGTAGTATGTATGATTTGGAGT	CAACAACAAAAACCT TAACCA
BCL2	M	ACGTTACGTATAGGAAATCGGTC	CCTTCGCTAACAACGACG
	U	GTGATGTTATGTATAGGAAATTGGTT	ACCTTCACTAACAACAACAAC
Actin $\beta$	IC	TGGTGATGGAGGAGGTTTAGTAAGT	AACCAATAAAACCTACTCCTCCCTTAA
DAPK		Acquired from ZymoResearch – Cat. # D5014

M: methylated primer; U: unmethylated primer; IC: internal control primers.

Figure 2.

The genomic location of CpG islands and binding position of methylated and unmethylated primer sets in the promoter regions of the study genes.

From the 33 studies, we sorted out 46 candidate genes based on the fact that these were reported in at least two studies. Most of the reported data was involving bone marrow aspirates (21 articles) and blood samples (15 articles), along with eight studies employing cell lines, two involving bone marrow biopsies and only one with mononuclear cell samples (Fig. 1B). To narrow down the selection of gene signatures, we applied another filter i.e., genes with reported methylation frequency of $\geqslant$ 70%, present only in bone marrow aspirate or blood samples (Fig. 1C). The two-tier criteria came up with a panel of four candidate genes i.e., SALL1, WNT5 $\alpha$ , LRP1b, and RSPO1. Using STRING (www.string-db.org), a protein-protein interaction and functional enrichment analysis software, all four genes were found associated with each other, closely or distantly (Fig. 1D); cadherin 2 (CDH2) was additionally included in the list, for analysis.

2.3 Sodium bisulfite modification of DNA

Genomic DNA extraction from the whole blood and its bisulfite modification was performed using EZ DNA Methylation-Direct ${}^{\text{TM}}$ kit (Zymo Research, CA, USA). For bisulfite conversion of human methylated- and non-methylated control DNAs (Zymo Research, CA, USA), EZ DNA Methylation-Lightning ${}^{\text{TM}}$ kit, was used.

2.4 Primer designing and real-time methylation specific PCR (qMSP)

The promoter sequences of all candidate genes were retrieved from the eukaryotic promoter database (EPD). Using MethPrimer v.2.0. software, pairs of methylated- and unmethylated primers were designed for each of the selected set of genes (Table 1). This tool initially performs in-silico bisulphite conversion of the sequence and then designs two sets of primers, one for the converted sequence and other for the original sequence to distinguish methylated and the unmethylated DNA sequences. The genomic location of CpG islands and binding position of methylated and unmethylated primer sets in the promoter regions is presented in Fig. 2.

All qMSP reactions were performed in BioRad CFX-96 Real-Time PCR System. The reaction mixture (25 $\mu$ l) contained 100 ng of bisulfite-treated DNA, 0.3 $\mu$ M forward and reverse primer pair (methylated- and unmethylated in separate tubes) and 2X Maxima SYBR Green/ROX qPCR Master Mix (Thermo Scientific Inc.). Thermal cycling profile included an enzyme activation step (95 ${}^{\circ}$ C, 10 min.) followed by 50 cycles of denaturation (95 ${}^{\circ}$ C, 15 sec.), annealing (30 sec.) and extension (72 ${}^{\circ}$ C, 30 sec.) [19]. The melt curve was analyzed at 65–95 ${}^{\circ}$ C with 0.05 ${}^{\circ}$ C increment in each step. Positive- (human methylated DNA), negative- (human non-methylated DNA) DNA standards along with unique sets of primers (for DAPK/BCL2) were included to assess the efficiency of bisulfite-mediated conversion of DNA (Zymoreserach D5014). Actin $\beta$ was used as internal control to remove the conversion bias. All reactions were performed at least in duplicate and the quantification cycles (Cq value) determined. 2 ${}^{-\Delta\Delta\text{Cq}}$ method was applied to calculate ratio of unmethylated- to amplifiable bisulfite-treated methylated DNA using DAPK and BCL2 as reference genes. The percentage methylation index (% MI) for each gene was calculated using following formula.

MI (%) $=$ (2 ${}^{-\Delta\Delta\text{Cq}}$ Methylated/2 ${}^{-\Delta\Delta\text{Cq}}$ Methylated $+$ 2 ${}^{-\Delta\Delta\text{Cq}}$ Unmethylated)*100

2.5 Bisulfite amplicon sequencing

Bisulfite amplicon sequencing (BAS) of all five genes from two randomly selected paired ALL samples (pre- and post-treatment) was performed to validate the results obtained from the qMSP analysis. The bisulfite converted DNA was subjected to PCR amplification using primers listed in Table 1. The PCR products were purified using PureLink Quick Gel Extraction kit and sequence analyzed using the standard automated Sanger sequencing (Macrogen, Korea).

2.6 Statistical analysis

The Origin and SPSS statistics ver.21 software, were used for statistical analyses. Methylation frequencies in the patient- and the control groups were compared using Fisher’s exact test; while Spearman non-parametric test was applied to assess correlation. The diagnostic and prognostic significance of the selected set of gene(s) was predicted using Receiver Operator Characteristics (ROC) and Kaplan-Meier survival analyses, respectively. $p$ -value $\leqslant$ 0.05 was considered as statistically significant.

3. Results

3.1 Promoter DNA methylation analysis

The clinicopathological features of pediatric ALL patients (both discovery- and validation-phase) are listed in Suppl. Table A. The patients’ group was dominated by pre-B ALL cases, median age ranged between 2.5 months to 15 years with male to female ratio being over 2:1. Only four patients had a family history of ALL; 7 cases were Philadelphia positive, 4 had CNS involvement, 21 showed hyperdiploidy while 48 presented normal karyotype. Hypodiploidy (an unfavorable risk feature associated with poor patient outcome), which according to WHO is manifested by 1–2% pediatric ALLs only [7, 8 and references therein], was found in around 20% cases (17/86); out of these 4% were baby girls while 16% were boys. Diagnosis of all the cases was based on the standard morphological, cytochemical and immunophenotypic criteria, as per the WHO classification guidelines (Suppl. Fig. S1).

Figure 3.

The calculated promoter methylation index (% MI) of SALL1, LRP1b, WNT5 $\alpha$ , RSPO1 and CDH2 genes in Pre-B ALL (PB-All), Pre-T ALL (PT-All) patients and healthy controls (HC).

The promoter methylation/demethylation pattern of five selected candidate genes was analyzed in the pre- and post-treatment ALL patients and the results were compared with the control group (Fig. 3). The individual methylation frequencies of SALL1, LRP1b, WNT5 $\alpha$ and RSPO1 in the pre-treatment ALL patients ranged between 80–89% whereas the same were below 15% in the control group; thus associating ‘hyper-methylation’ of these promoter with leukemia (Table 2). The highest relative fold change ( $>$ 5-fold) was observed in WNT5 $\alpha$ and SALL1 while the lowest (2-fold) in RSPO1. Following subtype categorization, a remarkable difference in the methylation frequencies/indices of the candidate genes was noticed, in the Pre-B and the Pre-T ALL patients. Although the p-value for each gene in each case was statistically significant ( $\leqslant$ 0.001); the former presented markedly higher promoter DNA methylation signals than the later. Contrary to the other four genes, the promoter of CDH2 was hypermethylated in the controls while demethylated in the ALLs.

Table 2

Methylation profile of ALL patients and cancer-free control

Candidate genes		Methylation frequencies (%)			Mean methylation index (%) $\pm$ SD			$P$ -value
		Pre-treatment patients	Post-treatment patients	Control group	Pre-treatment patients	Post-treatment patients	Control group
SALL1	ALL	89.53	04.0	3.57	80.26 $\pm$ 23.02	47.32 $\pm$ 9.82	19.33 $\pm$ 14.68	$<$ 0.001
	Pre-B ALL	92.72	06.3		83.35 $\pm$ 20.26	47.32 $\pm$ 9.41	19.33 $\pm$ 14.68
	Pre-T ALL	83.87	33.3		75.13 $\pm$ 26.15	41.96 $\pm$ 8.67	19.33 $\pm$ 14.68
LRP1b	ALL	86.04	00.0	10.7	79.90 $\pm$ 27.62	26.58 $\pm$ 6.77	24.48 $\pm$ 22.72	$<$ 0.001
	Pre-B ALL	92.72	06.3		85.25 $\pm$ 21.29	26.58 $\pm$ 6.77	24.48 $\pm$ 22.72
	Pre-T ALL	74.19	22.2		69.86 $\pm$ 34.25	21.32 $\pm$ 9.32	24.48 $\pm$ 22.72
WNT5 $\alpha$	ALL	84.88	08.0	3.57	79.91 $\pm$ 28.61	19.64 $\pm$ 7.23	18.83 $\pm$ 13.18	$<$ 0.001
	Pre-B ALL	92.72	00.0		87.55 $\pm$ 22.35	19.64 $\pm$ 7.23	18.83 $\pm$ 13.18
	Pre-T ALL	70.96	11.1		65.97 $\pm$ 33.01	14.32 $\pm$ 5.98	18.83 $\pm$ 13.18
RSPO1	ALL	80.23	16.0	14.3	71.02 $\pm$ 29.70	28.77 $\pm$ 24.43	32.61 $\pm$ 21.84	$<$ 0.001
	Pre-B ALL	94.54	25.0		83.30 $\pm$ 21.63	28.77 $\pm$ 24.43	32.61 $\pm$ 21.84
	Pre-T ALL	54.83	00.0		49.20 $\pm$ 29.32	22.02 $\pm$ 6.49	32.61 $\pm$ 21.84
CDH2	ALL	17.44	84.0	89.3	31.22 $\pm$ 27.88	79.13 $\pm$ 24.63	84.76 $\pm$ 25.57	$<$ 0.001
	Pre-B ALL	07.27	100		25.41 $\pm$ 20.49	78.04 $\pm$ 16.64	84.76 $\pm$ 25.57
	Pre-T ALL	35.48	67.0		41.07 $\pm$ 35.56	79.8 $\pm$ 15.79	84.76 $\pm$ 25.57

SD: Standard deviation.

Table 3

Correlation of clinical features with promoter methylation index of each gene

Clinical features	SALL1 (M)		LRP1b (M)		WNT5 $\alpha$ (M)		RSPO1 (M)		CDH2 (DM)
	Rho ${}^{*}$	$p$ -value	Rho	$p$ -value	Rho	$p$ -value	Rho	$p$ -value	Rho	$p$ -value
RBC10 ${}^{6}$ / $\mu$ l	0.67	0.009	0.71	0.001	0.45	0.001	0.63	0.008	0.57	0.002
Hb (g/dl)	0.58	0.008	0.25	0.005	0.36	0.007	0.54	0.006	0.94	0.009
WBC10 ${}^{3}$ / $\mu$ l	0.97	0.005	0.94	0.002	0.75	0.004	0.89	0.002	0.89	0.007
PLT10 ${}^{3}$ / $\mu$ l	0.81	0.003	0.84	0.009	0.59	0.003	0.89	0.001	0.79	0.004
L (%)	0.88	0.008	0.21	0.020	0.66	0.001	0.88	0.027	0.58	0.001
N (%)	0.79	0.001	0.21	0.090	0.57	0.002	0.97	0.006	0.47	0.007
M (%)	0.93	0.002	0.32	0.040	0.71	0.009	0.86	0.004	0.84	0.004
Blast cells (%)	0.86	0.003	0.23	0.020	0.64	0.007	0.78	0.002	0.84	0.003
CNS manifestation	$-$ 0.20	0.070	0.60	0.040	0.56	0.004	$-$ 0.6	0.041	0.13	0.059
Philadelphia chromosome	0.87	0.090	0.92	0.060	0.99	0.001	0.78	0.002	0.45	0.007
Family history	0.32	0.080	0.12	0.018	0.44	0.007	0.26	0.081	0.21	0.092
Splenomegaly	0.24	0.090	0.26	0.095	0.36	0.004	0.31	0.072	0.38	0.100

${}^{*}$ Rho: Spearman correlation coefficient; M: Methylation; DM: Demethylation.

Table 4

Representative receiver operating characteristics (ROC) of the informative sets of genes for the detection of ALL

Gene set	Patients’ group	TP/FN	Sensitivity (%) ${}^{**}$	PPV ${}^{\#}$ (%)	FP/TN	Specificity (%) ${}^{***}$	NPV ${}^{\#\#}$
(%)
SALL1	Pre-B ALL	51/04	93	98	01/27	96	87
	Pre-T ALL	26/05	84	96			84
	ALL	77/09	90	99			75
SALL1, LRP1b	Pre-B ALL	51/04	93	94	03/25	89	86
	Pre-T ALL	23/08	74	88			76
	ALL	74/12	86	96			68
SALL1, LRP1b, WNT5 $\alpha$	Pre-B ALL	51/04	93	94	03/25	89	86
	Pre-T ALL	22/09	71	88			74
	ALL	73/13	85	96			66
SALL1, LRP1b, WNT5 $\alpha$ , CDH2 ${}^{}$*	Pre-B ALL	51/04	93	94	03/25	89	86
	Pre-T ALL	20/11	65	87			70
	ALL	71/15	83	96			63
SALL1, LRP1b, WNT5 $\alpha$ , RSPO1	Pre-B ALL	51/04	93	93	04/24	86	86
	Pre-T ALL	18/13	55	82			65
	ALL	69/17	80	95			59
SALL1, LRP1b, WNT5 $\alpha$ , RSPO1, CDH2 ${}^{}$*	Pre-B ALL	51/04	93	93	04/24	86	86
	Pre-T ALL	18/13	55	82			65
	ALL	69/17	80	95			59

TP: True Positive; TN: True Negative; FP: False Positive; FN: False Negative. ${}^{*}$ Demethylation. ${}^{**}$ Sensitivity $=$ TP/(TP $+$ FN)*100. ${}^{***}$ Specificity $=$ TN/(TN $+$ FP)*100. ${}^{\#}$ Positive Predictive Value (PPV) $=$ TP/(TP $+$ FP)*100. ${}^{{\#}{\#}}$ Negative Predictive Value (NPV) $=$ TN/(TN $+$ FN)*100.

Figure 4.

Receiver operating characteristic (ROC) curves depicting prognostic significance of SALL1, LRP1b, WNT5 $\alpha$ , RSPO1 and CDH2 promoter methylation in combined ALL (A), Pre-B ALL (B) and Pre-T ALL (C) cases, along with Kaplan-Meier survival analysis (D) as a function of % methylation index (%MI).

Further, the methylation levels were assessed in the post-treatment patients. Notwithstanding the fact that analysis was performed with a limited number of paired pre- and post-treatment blood samples ( $n=$ 86 vs 25), the observed decrease in methylation frequencies was remarkable; this trend towards demethylation of candidate marker genes was comparable to the cancer-free controls (Table 2). BAS, performed for each gene with representative pre- and post-treatment ALL samples, validated the qMSP results (Suppl. Fig. S3). The methylated CpGs reverted to their native-like pattern in the post-treatment subjects, reinforced the plausible role of dysregulated methylation in leukemia patients.

3.2 Association study of the candidate marker genes

Spearman correlation coefficient was calculated to evaluate the relationship between the % MI value of each candidate marker gene and the clinicopathological features (Table 3). With a few exceptions, almost all features showed a very strong, statistically significant correlation with % MI (highlighted).

The sensitivity and specificity of the candidate genes, when worked out using ROC analysis, yielded an area under the curve (AUC) of 0.96, 0.92, 0.87 and 0.83 for SALL1, WNT5 $\alpha$ , LRP1b and RSPO1, respectively (Fig. 4A). The potential of individual and/or paired genes to detect Pre-B ALL cases was significantly higher than the Pre-T ALL, as revealed by the respective AUCs and the spearman correlation coefficient (Fig. 4B and C; Table 4). A significant monotonic relationship was found between all gene pairs; SALL1/WNT5 $\alpha$ and SALL1/LRP1b exhibiting the stronger, followed by the LRP1b/WNT5 $\alpha$ . RSPO1 due to its least TP/FN ratio resulted in an overall decrease in the detection sensitivity of the panel; however, by excluding RSPO1, the sensitivity raised up to 83% with a specificity of 86% envisaging the usefulness of panel for potential liquid-biopsy based diagnosis of ALL (Fig. 4C). Finally, patients with hypermethylated promoters of the candidate genes showed poor overall survival as depicted by the Kaplan-Meier survival analysis (Fig. 4D); patients afflicted with hypodiploidy also showed poor outcome (data not shown).

Figure 5.

Curated pathway depicting the mechanism of Wnt signaling in the normal (A, left half) and the leukemic lymphoblasts (B, right half). Detailed mechanism in normal cells is summarized under the ‘Discussion’ section. In case of leukemia, the promoters regulating the transcription of SALL1, LRP1b, WNT5 $\alpha$ and RSPO1 are hypermethylated and hence transcriptionally inactive (1B). CDH2 promoter, however, is unmethylated and overexpressing an otherwise poorly expressed CDH2 (2B). Due to the limited- or complete loss of WNT5 $\alpha$ and RSPO1 expression, tankyrase remains bound to the intracellular domain of frizzled receptor (4B), resultantly, the multi-protein scaffold is not formed (5B). Also, in the absence of SALL1 (red asterisk), HDAC fails to bind TCF/LEF whereas $\beta$ -catenin, which initially is localized to the cell membrane (3B), translocates to the nucleus and binds to TCF/LEF promoter (6B). A series of like-events end up with the transcriptional activation of oncogenes [Myc, Cyclin D1, TCF-1, MMP-7, CTLA4 etc., (7B)], involved in leukemogenesis.

4. Discussion

Wingless type (Wnt) signaling cascade is amongst the key signaling pathways that play important role in the normal cell growth, proliferation and development of many organ systems; and hence its dysregulation is believed to be linked up with abnormal cellular proliferation and/or epithelial-to-mesenchymal transition (EMT), in different solid tumors [20, 21, 22, 23]. Recent studies have documented that impaired regulation of Wnt proteins and their antagonists also have adverse impact on the thymocytes proliferation, lymphocytes differentiation, and/or the self-renewal of hematopoietic stem cells [[24, 25, 26] and references therein].

In the present study, the genes selected to assess the promoter DNA methylation status in pediatric ALL vs non-cancer controls (i.e., SALL1, WNT5 $\alpha$ , RSPO1, LRP1B and CDH2), were interestingly encoding the proteins that were either closely or distantly related to the Wnt signaling pathways (Fig. 1D). Amongst these, spalt-like protein 1 (SALL1) is a member of highly conserved spalt family zinc-finger transcription factors that positively regulates the canonical Wnt signaling cascade [27]; its silencing through promoter hypermethylation or other like transcription repression events has been linked with increased risk of hematological- (such as CML) and non-hematological malignancies (such as breast, head and neck carcinomas) [27, 28, 29].

While respectively acting as a ligand for the members of leucine-rich repeat-containing G-protein coupled receptors (LGR) 4/5/6 and the frizzled family of seven transmembrane receptors, the R-spondin 1 (RSPO1) and the Wnt family protein type 5A (WNT5A) modulate/activate the canonical and non-canonical Wnt pathways [24, 30, 31]. Likewise, the large-sized transmembrane protein low-density lipoprotein receptor-related protein 1B (LRP1B; genomic sequence spanning 1.9 Mb, coding sequence extending 16.5 Kb, total number of exons 91), is a member of LDL-receptors family that performs a myriad of functions at the cellular-, extracellular-, and the cell membrane level. Unlike LRP5/LRP6, the LRP1B is not directly linked with Wnt pathway (Fig. 1D); its unique role in embryonic development and as a signal modulator, however, is evident from several recent reports [32]. Furthermore, mutations (deletion/point), copy number variations, and aberrant expression of LRP1B, reported in different malignancies such as gastric-, lung-, breast-, oral-, esophageal-, neuronal-, renal-, and ovarian carcinomas [32, 33], also make it a strong candidate for analysis in hematological malignancies.

Our last selected gene, CDH2, encodes cadherin 2 protein (CDH2, also known as N-cadherin), a calcium-dependent transmembrane glycoprotein involved in the intercellular- and the cell-extracellular matrix adhesions. Unlike the co-family member CDH1 (E-cadherin), the CDH2 is either absent or meekly expressed in the normal epithelium. However, loss of CDH1 with concomitant aberrant overexpression of CDH2 has been described in several epithelial- and non-epithelial/hematological malignancies [24, 33, 34]. This “cadherin-switch” is believed to have crucial role in promoting the EMT associated events i.e., cancer cells’ migration (distant metastasis), localized invasion and tumor aggressiveness through the modulation, augmentation and/or polarization of important signaling pathways such as Wnt/ $\beta$ -catenin, fibroblast growth factor/receptors (FGF/FGFR) and Rho-family GTPase signaling [35].

Studies have shown that the CpG islands in promoters regulating the transcription of SALL1, WNT5 $\alpha$ , LRP1B, and RSPO1 are normally unmethylated, hence fully expressed in the non-leukemic healthy subjects [14, 24]. Consistent with earlier observations are the results of present study wherein the liquid biopsy samples of control subjects (85–96%) showed promoter DNA unmethylation of the aforesaid genes while hypermethylation (or transcriptional silencing) of CDH2 (Fig. 2, Table 2). Classically, the binding of expressed WNT5 $\alpha$ to the extracellular domain of frizzled receptor and the RSPO1 to LRP1B, brings about a conformational change in the cytosolic domain of frizzled that liberates the tankyrase, which in turn phosphorylates the GSK3/APC and mediates the assembly of a multi-protein scaffold consisting of GSK-3, APC, Axin, WTX, PP2A and $\beta$ -catenin (Fig. 5). Formation of this multi-protein scaffold triggers the ubiquitination and proteasomal degradation of $\beta$ -catenin by cytosolic ubiquitin ligases (Skp and B-Trcp). In the absence of $\beta$ -catenin, histone deacetylase complex (HDAC) binds to the TCF/LEF thus leading to the transcriptional repression of many oncogenes including Myc, CyclinD1, TCF1, MMP-7, Axin2, CD44, CTLA4 etc. [24, 25, 26, 36, 37]. Whereas mutations in the N-terminal region of $\beta$ -catenin make it resistant to the proteasomal degradation, functional loss of Wnt antagonists (like WNT5 $\alpha$ ) also leads to the activation of Wnt pathway, which along with CDH2-mediated $\beta$ -catenin sequestration and nuclear translocation, synergize with known driver mutations hence ends up with the transcriptional activation of several oncogenes [24, 31, 33, 34, 38].

In the present study, more than 85% B-ALL patients showed promoter DNA “hypermethylation” of otherwise “unmethylated” genes with parallel demethylation of CDH2 (Tables 2 and 3; Fig. 2). Reversal to the ‘normal promoter methylation’ patterns ( $p<$ 0.001; Table 3), however, was observed in patients following treatment with the prescribed therapies. These results, validated through bisulfite sequencing, are in line with the earlier studies, augmenting the role of reversible epigenetic glitches in the leukemic transformation of normal cells, in the B-/T-cell lineages. In the survival analysis, the promoter DNA hypermethylation was found associated with poor patient outcome (Fig. 3D).

While early-stage, cost-effective diagnosis is a critical component of success in the cancer management [39, 40] and a way to lessen the burden of complex, life-threatening disease(s); being an early event in leukemogenesis, the epigenetically dysregulated Wnt signaling pathway genes can provide important clues and thus facilitate early-on diagnosis of ALL in children [2, 24, 41, 42, 43, 44]. Whereas the findings of this research are encouraging in terms of panel sensitivity and specificity in liquid biopsies, and hence expected to have great impact on the ALL diagnosis and routine monitoring of treatment- or disease recurrence, the results need to be assessed in light of their limitations. For instance, in validation phase, we could enroll only a moderate (yet statistically significant) number of pediatric ALL patients and age-matched controls due to the reluctance of many families to give blood samples of their infants/children; difficulty in keeping track of the patients (coming from remote and distant areas of the countries) for the follow-up analyses was another great challenge. Furthermore, while designing the methylated/un-methylated primer pairs of the selected genes, we ignored scanning of the entire CpG island/shores. Since different sites in the CpG islands/promoter region differently impact the transcriptional activation and repression, undertaking ‘global methylome’ analysis instead of a site-specific analysis, may help in filling up the research gaps, still left behind. Lastly, to make this non-invasive liquid biopsy approach a bit more cost-effective in clinical oncology, we have used high-throughput qMSP technique instead of the next generation bisulfite targeted sequencing (NGS). qMSP have shown consistent results with up-to-the-mark accuracy level. The results were validated using BAS (Suppl. Fig. S3), nonetheless, NGS for an even better and complete coverage of the epigenomic alterations, may be considered, in future investigations.

In summary, our results demonstrate that the promoter regions of SALL1, WNT5 $\alpha$ , LRP1B, and CDH2 are epigenetically dysregulated in ALL patients and their assessment in liquid biopsies may be an alternative/complementary approach thereby enabling non-invasive diagnosis and treatment monitoring of pediatric leukemia, in the clinical settings.

Disclosure of Interest

Authors of this manuscript have no competing interests to declare.

Funding

This study was partially supported by grants from University of the Punjab and the Higher Education Commission, Pakistan.

Supplementary data

The supplementary files are available to download from http://dx.doi.org/10.3233/CBM-200814.

sj-docx-1-cbm-10.3233_CBM-200814.docx - Supplemental material

Supplemental material, sj-docx-1-cbm-10.3233_CBM-200814.docx

Footnotes

Acknowledgments

Authors extend special thanks to all blood sample donors, both pediatric ALL patients and healthy volunteers as well as their families, who made this study possible. We are also thankful to the doctors (especially Dr. Samina) and paramedical staff of Children Hospital Lahore, who not only helped us in sample collection but also convincing the families to participate in this study.

References

https://gco.iarc.fr/ (Global Cancer Observatory).

Bispo

J.A.B.

Pinheiro

P.S.

and Kobetz

E.K.

, Epidemiology and etiology of leukemia and lymphoma, Cold Spring Harbor Perspectives in Medicine 10 (2020), a034819.

Ahmad

Shah

K.A.

Hussain

Haq

A.U.

Ullah

Khan

and Rahman

N.U.

, Prevalence of acute and chronic forms of leukemia in various regions of khyber pakhtunkhwa, pakistan: needs much more to be done!, Bangladesh Journal of Medical Science 18 (2019), 222–227.

Maaz

Badar

Mahmood

and Al Nassir

, High infection related mortality in pakistani children with acute lymphoblastic leukemia during remission induction chemotherapy: review of data from a single institution, Journal of Cancer and Applied Specialities2 (2016), 1–7.

Pui

C.H.

Relling

M.V.

and Downing

J.R.

, Acute lymphoblastic leukemia, New England Journal of Medicine 350 (2004), 1535–1548.

Pui

C.H.

Carroll

W.L.

Meshinchi

and Arceci

R.J.

, Biology, risk stratification, and therapy of pediatric acute leukemias: an update, Journal of Clinical Oncology 29 (2011), 551.

Brown

Inaba

Annesley

Beck

Colace

Dallas

DeSantes

Kelly

Kitko

and Lacayo

, Pediatric acute lymphoblastic leukemia, version 2.2020, NCCN clinical practice guidelines in oncology, Journal of the National Comprehensive Cancer Network 18 (2020), 81–112.

Vasuki

Holden

E.J.

Kovesi

and Micklethwaite

, Semi-automatic mapping of geological structures using UAV-based photogrammetric data: an image analysis approach, Computers & Geosciences 69 (2014), 22–32.

Camino

A.G.

Guerrero

I.M.

and Orad

A.G.

, Genetic susceptibility in childhood acute lymphoblastic leukemia, Medical Oncology 34 (2017), 1–9.

10.

Jaenisch

and Bird

, Epigenetic regulation of gene expression: how the genome integrates intrinsic and environmental signals, Nature Genetics 33 (2003), 245–254.

11.

Jones

P.A.

and Baylin

S.B.

, The epigenomics of cancer, Cell 128 (2007), 683–692.

12.

McCabe

M.T.

Brandes

J.C.

and Vertino

P.M.

, Cancer DNA methylation: molecular mechanisms and clinical implications, Clinical Cancer Research 15 (2009), 3927–3937.

13.

Berdasco

and Esteller

, Aberrant epigenetic landscape in cancer: how cellular identity goes awry, Developmental Cell 19 (2010), 698–711.

14.

Rani

Mathur

Gupta

Gogia

Kaur

Dhanjal

J.K.

Sundar

Kumar

and Sharma

, Genome-wide DNA methylation profiling integrated with gene expression profiling identifies PAX9 as a novel prognostic marker in chronic lymphocytic leukemia, Clinical Epigenetics 9 (2017), 1–14.

15.

Milani

Lundmark

Kiialainen

Nordlund

Flaegstad

Forestier

Heyman

Jonmundsson

Kanerva

and Schmiegelow

, DNA methylation for subtype classification and prediction of treatment outcome in patients with childhood acute lymphoblastic leukemia, Blood, The Journal of the American Society of Hematology 115 (2010), 1214–1225.

16.

Issa

J.P.J.

, DNA methylation as a therapeutic target in cancer, Clinical Cancer Research 13 (2007), 1634–1637.

17.

Narayan

Freddy

A.J.

Xie

Liyanage

Clark

Kisselev

Un Kang

Nandula

S.V.

McGuinn

and Subramaniyam

, Promoter methylation-mediated inactivation of PCDH10 in acute lymphoblastic leukemia contributes to chemotherapy resistance, Genes, Chromosomes and Cancer 50 (2011), 1043–1053.

18.

Perrard

Morel

Meznad

Paget-Bailly

Dalstein

Guenat

Mourareau

Clavel

Fauconnet

and Baguet

, DNA demethylation agent 5azadC downregulates HPV16 E6 expression in cervical cancer cell lines independently of TBX2 expression, Oncology Letters 19 (2020), 1074–1081.

19.

Shahid

Shaheen

Shahid

Akhtar

M.W.

and Sadaf

, mir-16-5p as a suitable reference gene for normalization of quantitative real time PCR in acute lymphoblastic leukemia, Pak Journal of Zoology 51 (2019), 747–754.

20.

Suzuki

Watkins

D.N.

Jair

K.W.

Schuebel

K.E.

Markowitz

S.D.

Chen

W.D.

Pretlow

T.P.

Yang

Akiyama

and Van Engeland

, Epigenetic inactivation of SFRP genes allows constitutive WNT signaling in colorectal cancer, Nature Genetics 36 (2004), 417–422.

21.

Lee

H.C.

Kim

and Wands

J.R.

, Wnt/Frizzled signaling in hepatocellular carcinoma, Frontiers in Bioscience11 (2006), 1901–1915.

22.

Yongx

, Helicobacter pylori upregulates nanog and oct4 via Wnt/ß-catenin signaling pathway to promote cancer stem cell-like properties in human gastric cancer, Cancer Letters 374 (2016), 292–303.

23.

Dong

J.J.

Ying

and Shi

K.Q.

, Expression of the Wnt ligands gene family and its relationship to prognosis in hepatocellular carcinoma, Cancer Cell International 19 (2019), 1–11.

24.

Chiarini

Paganelli

Martelli

A.M.

and Evangelisti

, The role played by Wnt/β-catenin signaling pathway in acute lymphoblastic leukemia, International Journal of Molecular Sciences 21 (2020), 1098.

25.

Haseeb

Pirzada

R.H.

Ain

Q.U.

and Choi

, Wnt signaling in the regulation of immune cell and cancer therapeutics, Cell 8 (2019), 1380.

26.

Luis

T.C.

Ichii

Brugman

M.H.

Kincade

and Staal

F.J.

, Wnt signaling strength regulates normal hematopoiesis and its deregulation is involved in leukemia development, Leukemia 26 (2012), 414–421.

27.

Wang

Han

Zhong

Hsueh

E.C.

Robbins

Kiefer

S.M.

and Zhang

, SALL1 functions as a tumor suppressor in breast cancer by regulating cancer cell senescence and metastasis through the NuRD complex, Molecular Cancer 17 (2018), 1–21.

28.

Chi

Zhang

Jia

Cong

and Hu

, Spalt-like transcription factor 1 (SALL1) gene expression inhibits cell proliferation and cell migration of human glioma cells through the Wnt/β-catenin signaling pathway, Medical Science Monitor Basic Research 25 (2019), 128.

29.

Misawa

Imai

Mochizuki

Endo

Mima

Ishikawa

Kawasaki

Yamatodani

and Kanazawa

, Epigenetic modification of SALL1 as a novel biomarker for the prognosis of early-stage head and neck cancer, Journal of Cancer 9 (2018), 941.

30.

Shao

Zheng

Wang

Xin

Song

and Zhao

, Biological functions of macrophage-derived Wnt5a, and its roles in human diseases, Oncotarget 7 (2016), 67674.

31.

Lento

Congdon

Voermans

Kritzik

and Reya

, Wnt signaling in normal and malignant hematopoiesis, Cold Spring Harbor Perspectives in Biology 5 (2013), a008011.

32.

Dietrich

M.F.

Van Der Weyden

Prosser

H.M.

Bradley

Herz

and Adams

D.J.

, Ectodomains of the LDL receptor-related proteins LRP1b and LRP4 have anchorage independent functions in vivo, PLOS One 5 (2010), e9960.

33.

Nusse

and Clevers

, Wnt/β-catenin signaling, disease, and emerging therapeutic modalities, Cell 169 (2017), 985–999.

34.

Yang

and Zhang

, Cadherin signaling in cancer: its functions and role as a therapeutic target, Frontiers in Oncology 9 (2019), 989.

35.

Mrozik

K.M.

Blaschuk

O.W.

Cheong

C.M.

Zannettino

A.C.W.

and Vandyke

, N-cadherin in cancer metastasis, its emerging role in haematological malignancies and potential as a therapeutic target in cancer, BMC Cancer 18 (2018), 1–16.

36.

Arqués

Chicote

Puig

Tenbaum

S.P.

Argilés

Dienstmann

Fernández

Caratù

Matito

and Silberschmidt

, Tankyrase inhibition blocks Wnt/β-catenin pathway and reverts resistance to PI3K and AKT inhibitors in the treatment of colorectal cancer, Clinical Cancer Research 22 (2016), 644–656.

37.

Feder

Schröger

K.E.

Rawat

V.P.

Kirsten

Metzeler

Kraus

J.M.

Döhner

Kestler

H.A.

and Feuring-Buske

, Differences in expression and function of LEF1 isoforms in normal versus leukemic hematopoiesis, Leukemia 34 (2020), 1027–1037.

38.

Kong

Zhao

Tian

and Hai

, GAS2 promotes cell proliferation and invasion and suppresses apoptosis in pediatric T-cell acute lymphoblastic leukemia and activates wnt/β-catenin pathway, OncoTargets and Therapy 13 (2020), 1099.

39.

Khoo

B.L.

Shang

C.H.

Lim

C.T.

Chng

W.J.

and Han

, Liquid biopsy for minimal residual disease detection in leukemia using a portable blast cell biochip, NPJ Precision Oncology 3 (2019), 1–12.

40.

Heitzer

Haque

I.S.

Roberts

C.E.

and Speicher

M.R.

, Current and future perspectives of liquid biopsies in genomics-driven oncology, Nature Reviews Genetics 20 (2019), 71–88.

41.

Gomez

J.R.

Velasco

A.J.

Agirre

Castillejo

Navarro

Barrios

Andreu

E.J.

Prosper

Heiniger

and Torres

, Transcriptional silencing of the Dickkopfs-3 (Dkk-3) gene by CpG hypermethylation in acute lymphoblastic leukaemia, British Journal of Cancer 91 (2004), 707–713.

42.

Zhao

Tawatao

Cottam

H.B.

Sen

Leoni

L.M.

Kipps

T.J.

Corr

and Carson

D.A.

, Activation of the Wnt signaling pathway in chronic lymphocytic leukemia, Proceedings of the National Academy of Sciences 101 (2004), 3118–3123.

43.

Gomez

J.R.

Velasco

A.J.

Castillejo

J.A.

Agirre

Barrios

Navarro

Molina

F.J.

Calasanz

M.J.

Prosper

and Heiniger

, Promoter hypermethylation of cancer-related genes: a strong independent prognostic factor in acute lymphoblastic leukemia, Blood 104 (2004), 2492–2498.

44.

Mazieres

You

Lee

A.Y.

Mikami

McCormick

and Jablons

D.M.

, Inhibition of Wnt16 in human acute lymphoblastoid leukemia cells containing the t (1; 19) translocation induces apoptosis, Oncogene 24 (2005), 5396–5400.