Differential gene expression profiling linked to tumor progression of splenic marginal zone lymphoma

The genetic events that lead to aggressive transformation of cases of splenic marginal zone lymphoma (SMZL) after the chronic clinical stage have not been well understood. We aimed to find candidate genes associated with aggressive features of SMZL. We have successfully established two SMZL cell lines, designated SL-15 and SL-22, derived from the same patient’s tumor clone in chronic and aggressive phases, respectively. Microarray analysis identified cell cycle-associated genes—specifically PLK1—as the most significantly upregulated in primary aggressive SMZL cells compared with cells from chronic phase. EPHA4 and MS4A1 (CD20) were found to be downregulated dramatically. These gene expression patterns were reproduced in both cell lines. Genetic knockdown of PLK1 resulted in inhibition of cell proliferation and induction of apoptosis in SL-22 cells, which expressed higher levels of PLK1 than SL-15 cells. SL-22 cells needed higher concentrations of chemical PLK1 inhibitors to achieve greater effects. In addition, we found homozygous deletion of the MS4A1 gene as a newly identified molecular mechanism of CD20-negative conversion. Our findings are expected to stimulate further studies on whether PLK1 could be a potential therapeutic target for this tumor. Furthermore, cases with CD20-negatively converted lymphomas should be screened for the genomic loss of MS4A1.

Splenic marginal zone lymphoma (SMZL), also called splenic lymphoma with villous lymphocytes, is a rare B-cell neoplasm involving the spleen, bone marrow, and usually peripheral blood 1 . Most patients with SMZL show a chronic course with a median survival of around 10 years, whereas in a subset of patients the disease transforms to a more aggressive course with rapidly progressive and treatment-resistant form with increased mortality [2][3][4] . In the last few years, molecular genetic studies have identified a plethora of somatic mutations in cases of SMZL 5,6 . The most frequently mutated genes are KLF2 and NOTCH2, with a prevalence of 20-40% 7-9 and ~10-25% 6,8,10 , respectively. Inactivation of KLF2 and upregulation of NOTCH2 are involved in the physiological differentiation and proliferation of splenic marginal zone B cells, which might contribute to lymphomagenesis 2 . However, the genetic changes underlying the transformation of SMZL into a high-grade aggressive malignancy remain unknown. Although recognition of the sequential gene expression profiles during progression from chronic to aggressive phases of SMZL is helpful in revealing markers for tumor progression, the rarity of the disease, coupled with a lack of suitable in vitro study systems, might have hindered the biologic and genetic investigation of the aggressive transformation of SMZL. This study aimed to identify candidate genes associated with aggressive features of SMZL.
One approach to understand malignant transformation is by comparing gene expression of tumor cells derived from a chronic phase to their evolved malignant counterparts. Cell lines represent invaluable tools for research on rare diseases such as SMZL. Our previous study described an SMZL cell line, SL-15, established form a tumor in a chronic phase 11 . The case had a prolonged chronic clinical course with a good therapeutic response to monotherapy using the anti-CD20 monoclonal antibody rituximab, but later transformed into an aggressive disease. We have again successfully established another cell line, designated SL-22, from the transformed and aggressive tumor in the same patient. Comparison of the primary lymphoma cells as well as their evolved cell lines derived from a single patient with SMZL in two different phases of the disease has provided an opportunity to study sequential gene expression profiles during such transformation. In this study, microarray analysis showed a differential gene expression profile between SMZL cells derived from the chronic and aggressive clinical phases. We raised several therapeutic potential targets especially linked to cell cycle regulation, most notably PKL1, for further investigation of the genetic basis of SMZL transformation.
Rituximab-based treatment is a valid therapy for SMZL and is associated with a high overall response rate (~90%), with complete remission in more than half of these responding cases 2,12 . Although consecutive treatment with rituximab further improves the complete remission rate in patients with SMZL, acquired resistance to this drug has become a considerable problem. Studies have suggested that loss of CD20 expression is a major mechanism in such resistance [13][14][15][16] . Epigenetic mechanisms, in part, might contribute to the downregulation of CD20 expression, but the molecular mechanisms are still unclear. One of the major limitations in defining the mechanism of CD20-negative conversion from CD20-positive lymphomas after rituximab use is the lack of a laboratory model by which unlimited supplies of CD20-negative clones derived from CD20-positive cells can be studied repeatedly and extensively. So far, a few CD20-negative cell lines have been established from patients with CD20-positive lymphomas treated with rituximab 17,18 . However, paired CD20-positive and -negative cell lines derived from the same clones before and after rituximab use, respectively, have been lacking. In this context, our two lymphoma cell lines, SL-15 and SL-22, are valuable for studying the negative conversion of CD20. By utilizing these cell lines, we show here that genomic deletion of the MS4A1 (CD20) gene is another molecular mechanism in the loss of CD20 expression.

Results
Comparative characterization of the cell lines. The Epstein-Barr virus (EBV)-immortalized SL-15 and SL-22 cell lines were established from a single patient with SMZL. The cell surface marker profile of SL-22 cells was similar to that of SL-15 cells, except for being negative for CD20 expression. The karyotype of SL-22 cells was identical to that of the primary SMZL cells at an aggressive phase (Fig. 1A), confirming that these cells were derived from the clone of the patient's tumor cells. The two cell lines possessed common chromosome aberrations, including a unique t(9;14) chromosomal translocation involving 9q13 and 14q32, where PAX-5 and the immunoglobulin (Ig) heavy-chain gene are located, respectively 11 , indicating that the SL-15 and SL-22 lines had evolved from the same clone. Southern blot analysis of DNA showed that SL-22 cells exhibited a rearrangement of the Ig heavy-chain gene bands identical to those of SL-15 cells (Fig. 1B), also signifying that the two cell lines were clonally identical. Clearly SL-15 and SL-22 cells are paired SMZL cell lines derived from the same clone.

Differential gene expression profiles between different clinical periods of SMZL.
We compared gene expression profiles of the paired primary SMZL cells derived from the chronic (designated PB-15 cells) and aggressive (PB-22 cells) clinical phases using microarray analysis. A list of the differentially expressed genes was formed under criteria of 2.54-fold upregulation (Z-score > 2) and downregulation (Z-score < -2) in PB-22 cells compared with PB-15 cells (Table 1). A total of 1161 upregulated genes and 1112 downregulated genes were identified and further subjected to gene ontology (GO) analysis using the DAVID analysis. In this, the Functional Annotation Clustering tool identified several significantly upregulated clusters of genes. Annotation cluster 1 showed the highest enrichment score of 10.79 and included genes linked to the cell cycle, cell division, and mitosis (Table 2). Furthermore, pathway analysis (KEGG_PATHWAY) also identified the cell cycle pathway (P = 1.16 × 10 −7 ) as the most significantly overexpressed one (Table 2). These results indicated that dysregulated expression of genes associated with cell cycle regulation was involved in the aggressive transformation of the disease in our patient.
GO analysis was also performed on the downregulated genes. Annotation cluster 1 with the highest enrichment score of 28.63 included genes assigned to GO terms involved in cadherin and cell adhesion ( Table 2). The EPHA4 (erythroprotein-producing hepatocellular receptor A4) gene, which was the most strongly downregulated gene (Table 1), belonged to the category of cell adhesion in this cluster. Annotation cluster 2 with an enrichment score of 13.91 contained genes linked to cell membrane functions, including membrane-spanning 4A (MS4A) family genes such as MS4A1 (CD20) and MS4A7. KEGG pathway analysis also identified the MS4A family of genes in the inclusive hematopoietic cell lineage pathway as being significantly downregulated (Table 2).
Changes in the expression levels of KLF2 and NOTCH2 were also investigated. Our microarray analysis showed that KLF2 expression was significantly downregulated in PB-22 cells (18.5-fold lower; Table 1). A higher expression of NOTCH2 (1.5-fold higher) was observed, although this was not significant. , and UBE2C. The differential expression patterns were confirmed for all genes in both the primary tumors and cell lines ( Fig. 2A). Among the cell cycle related-genes upregulated, PLK1 showed the greatest difference in expression, as demonstrated by both microarray analysis and RT-qPCR. Immunoblot analysis also validated differential expression patterns of the protein polo-like kinase 1 (PLK1; Fig. 2B).
We also validated the expression levels of EPHA4 by RT-qPCR, because microarray analysis identified this as the most downregulated gene ( Table 1). The EPHA4 expression levels were dramatically suppressed in PB-22 (P < 0.001) and SL-22 cells (P = 0.020) compared with those in PB-15 and SL-15 cells, respectively ( Fig. 2A). The reduced expression of EphA4 in PB-22 and SL-22 cells was also confirmed at the protein level (Fig. 2B).
Antiproliferative effect of PLK1 inhibition evaluated using PLK1-specific small hairpin RNA (shRNA) and small interfering RNA (siRNA). PLK1, belonging to the family of serine/threonine protein kinases, plays a key role in centrosome maturation, bipolar spindle formation, and cytokinesis during mitosis 19 . It is likely to be one of the key molecular candidates associated with malignant transformation of SMZL, and thus could be a potential therapeutic target. In this context, we assessed the effects of PLK1 inhibition through applying PLK1-specific shRNA on the proliferation and apoptosis of SL-22 cells. Transduction of this shRNA into the cells resulted in significant decreases in PLK1 gene expression at the RNA and protein levels compared with transduction of control shRNA (Fig. 3A,B). Genetic knockdown of PLK1 caused a significant decrease in cell proliferation and an increase in apoptosis in SL-22 cells (Fig. 3C,D). Next, we conducted experiments to determine whether the antiproliferative effects on PLK1 inhibition would be mediated through cell cycle inhibition. Inhibition of PLK1 expression caused a significant increase in the proportion of the cell population at the G2/M phase of the cell cycle (Fig. 3E). Likewise, transduction of siRNA targeting different sequence of PLK1 caused a significant decrease in cell proliferation through cell cycle inhibition and an increase in apoptosis in SL-22 cells compared with the control siRNA transfected cells ( Supplementary Fig. S1).
Differential sensitivity to chemical inhibition of PLK1. Next, we performed experiments to determine whether there would be a difference in the antiproliferative effect of chemical inhibition of PLK1 between SL-15 and SL-22 cell lines. Both cell lines showed no difference in cell growth in RPMI 1640 medium supplemented with 10% fetal calf serum without PLK1 inhibitors ( Supplementary Fig. S2). The cells were treated with various concentration of volasertib, a selective PLK1 inhibitor, for 48 h. This resulted in efficient growth inhibition of both cell types at higher concentrations, but the two cell lines had differing sensitivities to this drug (Fig. 4A). The 50% growth inhibition (EC 50 ) value of SL-22 (25 nM) was 1.9-fold higher than SL-15 (13 nM). The EC 75 values (42 nM for SL-22 and 17 nM for SL-15) distinguished the drug sensitivity more clearly, with a 2.5-fold difference. These differences in drug sensitivity were apparent when volasertib was used at concentrations ranging from 20 to 40 nM. Similar results were obtained when another PLK1 inhibitor, BI 2536, was used (data not shown).
Exposure of SL-15 and SL-22 cells to volasertib for 24 h induced apoptosis in both lines. However, consistent with the results of the cell proliferation assay, SL-15 cells were more susceptible to apoptosis than SL-22 cells at 50 nM (P < 0.01; Fig. 4B). We next compared the effects of volasertib on the cell cycle between SL-15 and SL-22   Table 1). The MS4A1 gene, located on chromosome 11q12, belongs to the MS4A gene family with at least 18 subgroups (MS4A1-MS4A18) 20 . Karyotypes of PB-22 and SL-22 cells had no chromosomal translocations and deletions involving the 11q12 region, so we suspected a partial genomic deletion around MS4A1 in these cells. Confirming this, we identify a homozygous MS4A1 deletion (Fig. 5). We explored the deleted span of the genomic DNA on chromosome 11q12. A 487-kilobase (kb) region from MS4A3 to MS4A13 within the MS4A cluster region was deleted. An upstream114-kb segment containing OOSP1 and OOSP2 showed further loss, whereas the region containing TCN1 was preserved. Overall, at least a 600-kb region around MS4A1 was missing on chromosome 11q12. This suggests that MS4A1 along with its neighboring genes was lost after rituximab treatment.

Discussion
In recent years, there have been advances in understanding the somatic gene alterations involved in SMZL.
However, current knowledge about the genetic basis of SMZL is still incomplete and cannot explain the transitions in gene expression that enable progression to the aggressive form of the disease. We have established two lymphoma cell lines, SL-15 and SL-22, from a single patient with SMZL in the chronic and transformed aggressive phases, respectively. Cytogenetic studies and Southern blot analysis of Ig gene rearrangements confirmed the identical origin of the two cell lines. Although a few SMZL cell lines have been reported-one (VL51) from clinically stable SMZL 21 and two (Karpas 1718 and UCH1) from the refractory form of the disease 22, 23 -the paired availability of our cell lines from a case of SMZL that manifested a long chronic phase before progression to an aggressive clinical course is invaluable in investigating the transformation mechanism of SMZL. Somatic mutations affecting KLF2 and NOTCH2 appear to be the commonest genomic aberrations in SMZL, and studies suggest that cases with these mutations have an inferior outcome 2, 7-9 . However, the changing expression of KLF2 and NOTCH2 during tumor progression has not been delineated. Our findings showed that the trends toward lower expression of KLF2 and higher expression of NOTCH2 were linked tumor progression. We found a low frequency of somatic mutations in KLF2 and NOTCH2 in our samples, indicating that DNA mutations were unlikely to be responsible for the differential gene expression. Further studies are needed to confirm these findings in a large cohort.
Our microarray analysis identified specifically upregulated genes in transformed-aggressive SMZL cells compared with cells from the chronic stage. Both Functional Annotation Clustering and KEGG pathway analysis using DAVID showed that 'cell cycle' was the most significantly upregulated GO term. The cell cycle pathway is dominated with genes regulating cell proliferation and mitosis such as PLK1, E2F2, MAD2L1, AURKB, CDCA5, CCNA2, CCNB1, CCNB2, CDK1, CDK2, PTTG1, and UBE2C. Differences in the expression levels of all these selected genes were verified by RT-qPCR in both primary cells and their corresponding cell lines. Among these genes, PLK1 was shown to have the largest differences in expression by microarray analysis as well as a RT-qPCR,  Table 2. Enriched gene ontology (GO) functions and KEGG pathways for the upregulated and downregulated genes. 1 The underlined terms include PLK1. 2 The underlined term includes EPHA4. 3 The underlined terms include MS4A1 (CD20) and MS4A7. suggesting that higher PLK1 expression appears to be associated with a more severe grade of malignancy. Therefore, the role of PLK1 on cell proliferation and apoptosis was investigated further using the SMZL cell lines. Genetic knockdown of PLK1 through shRNA-and siRNA-mediated RNA interference caused a reduction in cell proliferation through cell cycle inhibition and an increase in apoptosis. Although the selective PLK1 inhibitor volasertib also showed antiproliferative effects in both SL-15 and SL-22 cells, there were clear associations between the levels of PLK1 expression and the sensitivity of cells to volasertib. SL-22 cells, which expressed higher levels of PLK1 than SL-15 cells, needed higher concentrations of volasertib to achieve more efficient inhibition of cell proliferation and induction of apoptosis. The EC 50 value of SL-22 shown in this study was similar to that of multiple cell lines derived from various cancer tissues, including carcinomas of the colon (HCT 116, EC 50 = 23 nM) and lung (NCI-H460, EC 50 = 21 nM), as shown in previous studies 24 . Thus, these data suggest that volasertib could serve as a potential therapeutic agent against PLK1-expressing SMZL tumors, as shown in many forms of solid cancer. Indeed, PLK1 overexpression has been found in a variety of cancers in advanced stages, and several PLK1 inhibitors are currently in various stages of clinical trials 25 . In certain cancer types, such as invasive breast cancers and renal cell carcinomas, PLK1 has significantly higher expression levels in late than in early stages 26 , which is in line with the results of the present study. Among hematological malignancies, PKL1 is often overexpressed in acute myeloid leukemia 27,28 . A few studies have suggested that high-grade non-Hodgkin's lymphomas show a trend toward higher expression levels of PLK1 than low-grade forms 29,30 . Our findings suggest that upregulation of PLK1 might be involved in the biological aggressiveness of SMZL and promote its progression. In this context, although high levels of PLK1 expression should be confirmed in more patients with advanced SMZL, clinical management with combination chemotherapy including PLK1 inhibitors is worth studying in the future. It is notable that CD20 expression was lost in the late-stage tumor cells in our patient. Several studies have suggest that epigenetic mechanisms are linked to the loss of CD20 expression after rituximab treatment, and that the its expression can be restored by DNA methyltransferase inhibitors and histone deacetylase inhibitors [14][15][16][17] . Our study revealed a genomic deletion of the entire MS4A1 gene along with its neighboring genes on chromosome 11q12. Based on the literature to date, Nakamaki et al. 31 reported a case in which the MS4A1 gene was deleted after rituximab-containing chemotherapy for the treatment of diffuse large B-cell lymphoma (DLBCL). Their report showed that genomic loss was observed intensively around a region including MS4A1 and MS4A5, and spanned a 700-kb region involving some genes of the MS4A family. Consistent with that report, here we showed the genomic loss of a 600-kb segment around MS4A1. Loss of CD20 expression leads to the potential loss of a therapeutic target during relapse and/or disease progression of CD20-positive B-cell malignancies, and is often associated with poor prognosis for the patients at that point 14,32 . Accordingly, clinicians and researchers should note that genomic losses around MS4A1 are missed by conventional karyotyping, so it is necessary to recognize this genomic MS4A1 deletion as a new molecular mechanism for CD20-negative conversion.
Our microarray identified EPHA4 as the most downregulated gene. GO analysis revealed that EPHA4 belonged to the category of 'cell adhesion' in annotation cluster 1 with the highest enrichment score. For these reasons, we payed attention to EPHA4 as a representative downregulated gene. EPHA4 downexpression was verified at both the RNA and protein levels by RT-qPCR and immunoblotting, respectively. Eph comprises the largest family of receptor tyrosine kinases, being composed of nine EphAs and five EphBs 33 . Recent evidence indicates that the Eph receptors have both tumor-promoting and suppressing activities, depending on their expression pattern in different tumor types; thus, some of the EPH genes are oncogenic and are upregulated in various cancers 33 . On the other hand, the EPH genes can act as tumor suppressors, and loss of their expression is evident in some tumors, for example, EphB4 in colorectal and breast cancers 34,35 . Likewise, EphA4 has also been found to have a multifaceted function as a tumor suppressor and promoter in some solid cancers 33,[36][37][38] . However, its role in the pathogenesis of hematological malignancies has not been fully determined. DNA methylation of EPHA4 has been observed in cases of acute lymphoblastic leukemia 39 , and EphA4 expression has been shown to inhibit lymphocyte proliferation 40 . Thus, a potential role of EphA4 as a tumor suppressor in lymphoid malignancies is currently receiving increasing attention. We hypothesize that signaling pathways involved in EphA4 might be associated with the aggressive transformation of SMZL, and this should be clarified by further studies. Of note, Koivula et al. 41 showed that a low level of EPHA4 expression was associated with poor overall survival in patients with DLBCL. Interestingly, EPHA4 was found to be one of the most important genes associated with the responsiveness to rituximab in cases of B-cell lymphoma 41 . In this context, it is plausible that the dramatic downregulation of EPHA4 might be caused, in part, in association with CD20 downregulation, thereby directly or indirectly contributing to the poor prognosis in our patient.
In summary, we have presented a differential gene expression profile associated with tumor progression of SMZL, and have identified specific genes for further studies to identify the molecules involved in the transformation process of this disease. Some of the gene expression changes reported here, specifically PLK1, might be involved in the biological aggressiveness of SMZL and could serve as potential therapeutic targets. Although a limitation of the current study was that only one case of transformed SMZL and a pair of cell lines was analyzed, future studies promise to elucidate the important issues. Another limitation of our study is the use of EBV immortalization to create the SMZL cell lines. Although their differential gene expression profile was confirmed in the primary SMZL cells, the experimental results should be interpreted with caution. If our findings are confirmed, we hope that PKL1 inhibitors will prove efficacious in improving the outcome of patients with advanced SMZL, who have limited therapeutic options. Furthermore, we have demonstrated a homozygous MS4A1 deletion as a unique molecular mechanism of CD20-negative relapse in a patient with B-cell lymphomas. This finding suggests that cases of B-cell lymphomas with loss of CD20 should be screened for the genomic loss of MS4A1. Such screening will help identify patients who need early intensive treatments including stem cell transplants to overcome a CD20-negative relapse of B-cell lymphomas, because genomic deletion of MS4A1 appears to be an irreversible event that leads to the permanent loss of the immunotherapeutic target, and ultimately to reduced survival of the patient.

Materials and Methods
Cell lines. The EBV-immortalized SL-15 cell line was established from a 53-year-old man with SMZL in a chronic phase. The detailed characteristics have been reported 11 . The SL-15 line was demonstrated to be derived from the clone of the patient's primary lymphoma.
Complete remission was achieved in this patient after rituximab-based treatment, but the disease relapsed three times. After 3 years and 4 months of repeated rituximab monotherapy for each relapsed disease, the patient developed bilateral pleural effusions and ascites infiltrated with lymphoma cells. The patient died of the progressive disease with resistance to rituximab-inclusive combination chemotherapy when his white blood cell count was 60.0 × 10 9 /l with 70% lymphoma cells. The immunophenotype of these cells was similar to that of the lymphoma cells at diagnosis, with the exception that CD20 expression became negative. The karyotype of the lymphoma cells was 47, XY, add(3)(p13), add(3)(p13), t(9;14)(p13;q32), add(10)(q24), add(11)(q21), + add (11). der(11:13)(q10;q10), + 12, and add(16)(p11.2), showing a close resemblance to that of the lymphoma cells at diagnosis 11 . These findings indicated that the lymphoma cells from the pre-and post-rituximab therapy were of the same clonal origin. Following informed consent from the patient, peripheral blood was obtained 2 weeks before his death, and mononuclear cells were separated by Ficoll-Hypaque density gradient centrifugation. The cells were cultured under the same conditions used for the establishment of SL-15 11 . The cells began to proliferate after a week from initiation of the culture and could then be regularly passaged. The cell line was designated SL-22 and characterized as described 11 . This study was approved by the Ethics Committee of Kochi Medical School, Kochi University, Japan. All experiments were performed in accordance with the relevant guidelines and regulations.
Oligonucleotide microarray. The CodeLink Human Whole Genome Bioarray (Applied Microarrays, Tempe, AZ, USA) was used to define and compare gene-expression profiles between primary SMZL cells derived from the chronic and aggressive clinical phases. The array targets most of the known and predicted genes of the human genome, and is composed of approximately 55,000 probes designated to blind to conserved exons. Labeling of complementary DNA targets, hybridization, and scanning of the arrays were carried out following the manufacturer's instructions. Raw intensity measurements of all probe sets were background-corrected, normalized, and converted into expression measurements using the MicroArray Data Analysis Tool Version 3.2 (Filgen, Nagoya, Japan). All microarray data were submitted to Gene Expression Omnibus (https://www.ncbi. nlm.nih.gov/geo/) under accession number GSE94318. Differentially expressed genes were identified using a cutoff fold change of >2.54. GO analysis and pathway analysis (KEGG_PATHWAY) were performed using the DAVID Bioinformatics Resource 6.7 online software (https://david.ncifcrf.gov/).

Real-time quantitative reverse-transcription polymerase chain reaction (RT-qPCR). Real-time
RT-qPCR was used to validate selected data from microarray experiments in both primary lymphoma cells and their corresponding cell lines. Total RNA was extracted using High Pure RNA Tissue kits (Roche Diagnostics, Tokyo, Japan). The total RNA was treated with DNase to avoid any amplification of genomic DNA and reverse-transcribed using the SuperScript III First-Strand Synthesis System (Life Technologies, Tokyo, Japan). An aliquot of cDNA was subjected to qPCR analysis. The reaction was conducted in triplicate on a StepOnePlus thermocycler (Life Technologies) with SYBER green PCR master mix containing 0.4 μM of each primer. The primer sequences used to determine the gene expression are listed in Supplementary Table S1. The β-globin (HBB) gene was amplified to confirm the presence of PCR-amplifiable cDNA. The PCR conditions were 10 min at 95 °C, followed by 50 cycles of 15 s at 95 °C and 1 min at 60 °C. Relative gene expression levels in PB-22 and SL-22 cells were calculated using the 2 −ΔΔCt method 42 , with the β-actin (ACTB) gene used as a housekeeping control, and the value was expressed as an n-fold change relative to that in PB-15 and SL-15 cells, respectively. Statistical analysis was performed at the ΔCt stage using unpaired two-tailed Student's t-tests. A statistically significant difference was defined as P < 0.05. Immunoblot analysis. Immunoblotting was performed as described 43 . The following antibodies were used: rabbit monoclonal anti-PLK1 (clone 208G4; Cell Signaling Technology, Danvers, MA, USA); mouse monoclonal anti-EPHA4 (clone 4C8H5; Thermo Fisher Scientific, Waltham, MA, USA); mouse monoclonal anti-β-actin (clone AC-74; Merck KGaA, Darmstadt, Germany); IRDye 680RD goat anti-rabbit IgG (LI-COR Biosciences, Lincoln, NE, USA); and IRDye 800CW goat anti-mouse IgG (LI-COR Biosciences). Bands were visualized using ODYSSEY CLx (LI-COR Biosciences). Signal intensities were quantified using ImageJ software (NIH, Bethesda, MD, USA; https://imagej.nih.gov/ij/). Levels of proteins were normalized to that of β-actin. DNA sequencing analysis. Nested PCR was performed using Platinum SuperFi DNA polymerase (Thermo Fisher Scientific). To amplify exon 1-2 and exon 3 of KLF2 and exon 34 of NOTCH2, the PCR conditions were 10 s at 98 °C (30 s for the first cycle), followed by 20 cycles of 30 s at 58 °C and 1.5 min at 72 °C (5 min for the last cycle) for the first round of PCR and 30 cycles for the second round of PCR. Amplification products were electrophoresed on 2% agarose gel and stained with ethidium bromide. Presence of somatic mutations in the KLF2 (exon 1-3) and in the NOTCH2 (exon 34) were investigated by Sanger sequencing as described 7,9,10 . The primer sequences used for nested PCR and sequencing analysis are listed in Supplementary Table S2.
Plasmid construction and production of PLK1-specific shRNA and siRNA. The doxycycline-regulated and RNA polymerase II-inducible shRNA-expression plasmid vector, designated as pTRE3G1, was generated for this study. The pTRE3G1 plasmid vector contains the following constructs: reverse tetracycline activator coding region derived from pCMV-TET3G (Takara Bio Inc., Shiga, Japan); an improved variant of the copepod Pontellina plumata green fluorescent protein (max GFP) coding region derived from pmaxGFP (Lonza, Basel, Switzerland); the GAPDH promoter from nucleotides -376 to + 183 relative to the transcription start site; a tetracycline-response element promoter derived from pTRE3G (Takara Bio); 5′ and 3′ miR-155 flunking region derived from pcDNA 6.2-GW/EmGFP-miR (Thermo Fisher Scientific); and a Luciferase shRNA coding region derived from pSingle-tTS-Anti-Luc (Takara Bio Inc.) for control shRNA. The DNA fragments were synthesized using the GeneArt Strings system (Thermo Fisher Scientific) and cloned into SalI-PciI restriction enzyme sites of pcDNA3.1 (Thermo Fisher Scientific) using the In-Fusion HD cloning kit (Takara Bio Inc.), according to the manufacturer's instructions.

PLK1 inhibition.
For genetic inhibition of PLK1, cells were transfected with PLK1 shRNA plasmid vector, PLK1 siRNA, or their controls on Nucleofector (Lonza) using C solution and the D-23 program. The cells transfected with shRNA plasmid vectors were isolated 2 days after transfection by sorting maxGFP-expressing cells on a FACSAria II flow cytometer (Becton Dickinson, Mountain View, CA, USA). The isolated cells were treated with 1 μg/ml of doxycycline for 48-96 h to induce the expression of shRNA before further experiments. The transfection efficiency of siRNAs was more than 85%, as determined using fluorescent siRNA. For chemical inhibition of PLK1, cells were treated with the PLK1 inhibitor volasertib (BI 6727) (ChemScene, Monmouth Junction, NJ, USA) or BI 2536 (ChemScene) at concentrations of 5-50 ng/ml. Cell proliferation, apoptosis and cell cycle analyses. For cell proliferation assays, cells were seeded in 96-well plates (8 × 10 4 cells/well) and viable cells were counted after 48 h or every 24 h on a FACSCalibur flow cytometer (Becton Dickinson) by gating out cells stained with propidium iodide as described 43 . For apoptosis assays, cells were stained with annexin V-phycoerythrin and 7-amino-actinomycin D according to the manufacturer's instructions. For cell cycle analysis, cells were fixed in cold 70% ethanol, treated with RNase, and stained with propidium iodide. Cells were analyzed using a FACSCalibur as above, and all flow cytometry data were analyzed using CellQuest Pro software (Becton Dickinson). All experiments were performed in triplicate.
Real-time qPCR for detecting genomic DNA of the MS4A gene cluster. Real-time qPCR was used to detect and quantify genomic DNA of the MSA4 gene cluster and their neighboring genes. Genomic DNA was extracted using the phenol-chloroform method. The reaction was conducted in duplicate with 200 ng of extracted DNA and SYBR green PCR master mix containing 0.4 μM of each primer. The primer sequences used to determine the gene levels are listed in Supplementary Table S3. The PCR conditions were 10 min at 95 °C, followed by 30 cycles of 15 s at 95 °C and 1 min at 60 °C. Relative gene loads in SL-22 cells were calculated using the 2 −ΔΔCt method, with the β-globin (HBB) gene used as housekeeping control, and the value was expressed as an n-fold change relative to that in SL-15 cells. The PCR products were separated electrophoretically on 2.0% agarose gels, visualized with ethidium bromide staining, and photographed. Data Availability. All data generated or analyzed during this study are included in this published article and its Supplementary Information files.