Common variants in MMP20 at 11q22.2 predispose to 11q deletion and neuroblastoma risk

MYCN amplification and 11q deletion are two inversely correlated prognostic factors of poor outcome in neuroblastoma. Here we identify common variants at 11q22.2 within MMP20 that associate with neuroblastoma cases harboring 11q deletion (rs10895322), using GWAS in 113 European-American cases and 5109 ancestry-matched controls. The association is replicated in 44 independent cases and 1902 controls. Our study yields novel insights into the genetic underpinnings of neuroblastoma, demonstrating that the inherited common variants reported contribute to the origin of intra-tumor genetic heterogeneity in neuroblastoma.


Supplementary information Supplementary Note 1
In the supplementary note, we highlighted the novel genomic regions of recurrent focal events and candidate genes discovered in this study. Beside the commonly observed MNA, 11q deletion and 1q deletion variations in neuroblastoma (Supplementary Fig. 1), focal amplifications of ALK, CCND1, LIN28B, MDM2 and 19q13.42 observed in our study have been implicated in neuroblastoma previously [1][2][3][4][5] . However, to our knowledge, recurrent focal amplifications of MYC, ZFHX3, KRAS, RRAS2 and CYTH1 have not been reported in neuroblastoma primary tumors before.
We further confirmed that the amplified genes were significantly overexpressed in neuroblastoma tumors and cell lines by examining gene expression data from 100 primary tumors and 29 cell lines (Supplementary Table 1).

Focal amplifications on other genomic regions
MYC amplification: two cases of MYC high-level amplification were detected including one case showing both MYC and ZFHX3 high-level amplification (Supplementary Fig. 3). Two cases of lowlevel MYC amplification were detected including one case showing both MYC and ZFHX3 amplification found another case showing both MYC and ZFHX3 low-level amplification. None of the four cases was observed with MNA. Amplification of MYC and ZFHX3 (ATBF1) has been reported in the SJNB12 cell line, but MYC amplification has not been observed in neuroblastoma primary tumors before 7,17 . A previous study also reported that chromothripsis resulted in the amplification (low-level) of MYC in one case 18 . In addition, MYC was highly expressed rather than MYCN in Neuroblastoma-derived cell lines lacking amplified MYCN 19,20 . Here, our results showed MYC amplification existed not only in cell lines but also in neuroblastoma primary tumors, and suggest that MYC amplification may be a rare alternative mechanism instead of MNA in neuroblastoma carcinogenesis. ZFHX3 (ATBF1) amplification: two cases of ZFHX3 high-level amplification were detected including one case showing both MYC and ZFHX3 high-level amplification. Seven cases were found with lowlevel ZFHX3 amplifications and only one of them exhibited low-level MNA. ZFHX3 has been reported to function as a tumor suppressor in several cancers 21,22 . ZFHX3 also plays a role in multiple other biological processes that are regulated by progesterone-PR, including cell proliferation, cell differentiation and tumorigenesis in the mammary gland 23 .
KRAS amplification: two cases were found (Supplementary Fig. 3). One case exhibited MNA and another case displayed MDM2 amplification. As a member of RAS oncogene family, KRAS amplification has been reported in a variety of different cancers [24][25][26] . Moreover, a recent study implicated recurrent new mutations of KRAS in relapse neuroblastomas 27 .
RRAS2 amplification: one case was found (Supplementary Fig. 3). RRAS2 amplification was also detected in the cell line, CHLA-136.  CCND1 amplification: two cases of high-level amplification were detected. Both cases showed MNA (Supplementary Fig. 3). CCND1 amplification has been reported in diverse tumor types 32,33 , and has been reported as a rare event in neuroblastoma 2,7,34 . However, besides the two high-level amplification of CCND1, we found 56 cases with low-level CCND1 amplification (56/628, 8.9%) in our study. Moreover, we found low-level CCND1 amplification co-occurred with 11q deletion in LIN28B amplification: one case was found that displayed MNA. Three cases were identified with low-level HACE1-LIN28B amplification. Among them, one exhibited MNA. Common variants within HACE1-LIN28B locus have been shown to contribute to neuroblastoma susceptibility. Significant growth inhibition was observed upon depletion of LIN28B in neuroblastoma cell lines, and low HACE1 and high LIN28B expression were associated with worse overall survival in neuroblastomas 41 . The LIN28B amplification has been reported previously as a rare event 3 . and ANTXR1, which were not detected in tumors. MEIS1 is a homeobox gene and has been associated with many cancers. MEIS1 amplification was also detected in IMR32 previously [42][43][44] .

Focal amplifications detected in cell lines
ANTXR1 also involves in cell attachment and migration, and has been associated with many cancers [45][46][47] .

Focal amplifications confirmed by gene expression data
Amplifications of chromosome 2p: three tumor samples and two cell lines with ALK amplification have mRNA expression data. The expression levels of ALK in the three tumor samples with ALK amplification were much higher than the tumor samples without ALK amplification. Though much lower than in tumor samples with ALK amplification, the expression levels of ALK in two cell lines with ALK amplification were also higher (Supplementary Fig. 4).  (Supplementary Fig. 4).
MYC amplification: one tumor sample with MYC amplification has mRNA expression data. The expression of MYC in this tumor sample was much higher than other tumor samples and the cell line samples (Supplementary Fig. 4).
LIN28B amplification: one tumor sample with LIN28B amplification has mRNA expression data.
The expression of LIN28B in this tumor sample was significantly higher compare to other tumor samples and the cell line samples (Supplementary Fig. 4).
CYTH1 amplification: one tumor sample with CYTH1 amplification has mRNA expression data.
The amplified region is chr17, 76,509,989-76,758,251 including DNAH17 and CYTH1, however, only the expression of CYTH1 in this tumor sample was significantly higher than other tumor samples and the cell line samples (Supplementary Fig. 4).

Chromothripsis
Chromothripsis is a shredding of chromosomal regions and subsequent random reassembly of the fragments, which leads to multiple segmental Gains and/or Losses including loss of tumor suppressors and oncogene amplifications 63,64 . A previous study suggested that the neuroblastoma with chromothripsis were associated with poor prognosis 18 . While chromothripsis was typically identified by whole genome sequencing 18 , several hallmarks of chromothripsis can be observed in our study such as alternating copy number states, loss of heterozygosity and high level of breakpoints within confined chromosomal regions 65,66 .
In this study, we found 46 (