Abstract

Uniparental disomy (UPD) is defined as the inheritance of both homologs of a given genomic region from only one parent. The majority of UPD includes an entire chromosome. However, the extent of UPD is sometimes limited to a subchromosomal region (segmental UPD). Mosaic paternal UPD (pUPD) of chromosome 11 is found in approximately 20% of patients with Beckwith–Wiedemann syndrome (BWS) and almost all pUPDs are segmental isodisomic pUPDs resulting from mitotic recombination at an early embryonic stage. A mechanism initiating a DNA double strand break (DSB) within 11p has been predicted to lead to segmental pUPD. However, no consensus motif has yet been found. Here, we analyzed 32 BWS patients with pUPD by SNP array and searched for consensus motifs. We identified four consensus motifs frequently appearing within breakpoint regions of segmental pUPD. These motifs were found in another nine BWS patients with pUPD. In addition, the seven motifs found in meiotic recombination hot spots could not be found within pUPD breakpoint regions. Histone H3 lysine 4 trimethylation, a marker of DSB initiation, could not be found either. These findings suggest that the mechanism(s) of mitotic recombination leading to segmental pUPD are different from that of meiotic recombination. Furthermore, we found seven patients with paternal uniparental diploidy (PUD) mosaicism. Comparison of clinical features between segmental pUPDs and PUDs showed that developmental disability and cardiac abnormalities were additional characteristic features of PUD mosaicism, along with high risk of tumor development. We also found that macroglossia was characteristic of segmental pUPD mosaicism.

Introduction

Uniparental disomy (UPD) is defined as the inheritance of both homologs of a given genomic region from only one parent. The lack of a maternal allele is defined as paternal UPD (pUPD), and the lack of a paternal allele is maternal UPD (mUPD). Isodisomy shows two identical copies of one of the parental alleles, and heterodisomy shows two distinct alleles derived from the same parent (1). The majority of UPD includes an entire chromosome; however, the extent of UPD is sometimes limited to the subchromosomal region (segmental UPD). UPD of an entire chromosome occurs due to gamete complementation, trisomy rescue, monosomy rescue, compensatory UPD and mitotic non-disjunction at post-fertilization stage, whereas segmental UPD is caused by mitotic recombination during the early embryonic stage and therefore shows mosaicism (2,3). UPD can cause imprinting disorders when it occurs in the imprinted region, and recessive diseases when two copies of a recessive gene mutation are inherited from a carrier parent. UPD is also involved in cancer through the inactivation of tumor-suppressor genes, the activation of oncogenes or alteration of imprinting (2–4).

Beckwith–Wiedemann syndrome (BWS) is an imprinting disorder showing prenatal and postnatal macrosomia, macroglossia, abdominal wall defects and a predisposition to tumorigenesis, among other features. In the etiologies of BWS, mosaic pUPD of chromosome 11 is found in approximately 20% of BWS patients (5). Almost all pUPD in BWS is segmental isodisomic pUPD resulting from mitotic recombination at an early embryonic stage (6). Although mechanisms of meiotic recombination, including consensus motifs and molecular machinery, have been well studied, those of mitotic recombination, especially the consensus motifs leading to segmental UPD, are less known. DNA double strand breaks (DSBs) are programed and catalyzed by meiosis-specific topoisomerase-II-like SPO11 endonuclease in meiotic recombination (7), but in mitotic recombination DSBs are not programed and can be caused by other factors, including endogenous metabolic reactions, replication stress, exogenous chemotherapeutic drugs or ionizing radiation (8). Although the factors and mechanisms involved in the occurrence of segmental pUPD of chromosome 11 are unknown, a mechanism initiating DSBs within 11p and leading to segmental pUPD is predicted for the following reasons. First, imprinted chromosomal regions are historical hot spots of meiotic recombination (9,10). Second, segmental UPD limited to 11p is found in almost all BWS patients with pUPD, but is rare in other imprinted disorders (9), indicating that 11p is a region prone to mitotic recombination.

BWS patients with pUPD mosaic frequently show hemihyperplasia and have a high risk of embryonal tumors, including Wilms' tumor and hepatoblastoma (5,11). Furthermore, a small number of paternal uniparental diploidy (PUD) mosaic cases have been reported in the past two decades, showing clinical features of BWS and various phenotypic attributes, including high risks of tumor development (4,5,11). Although the relationship between the level of pUPD and clinical features has been investigated (1,12,13), the mosaic ratio in peripheral blood does not generally correlate with phenotypic presentation.

In this study, we performed detailed analyses on 36 patients with pUPD of chromosome 11. We first identified consensus motifs appearing within the breakpoint regions of pUPD. We also found characteristic features of PUD in addition to tumor development.

Results

The mosaic ratio and the length of pUPD

We analyzed 36 patients with pUPD of chromosome 11 in this study (Table 1). First, we screened pUPD for mosaic ratios using genomic DNA from peripheral blood with microsatellite markers at 11p15. We confirmed gain of methylation at H19DMR and loss of methylation at KvDMR1 by methylation-sensitive Southern blots (data not shown). The median of the mosaic ratio was 62% (15–100%) and 25 patients (69%) showed more than 50% (Table 1 and Supplementary Material, Fig. S1).

Table 1.

Patients analyzed in this study

Patient ID Sex Elliot et al. (33DeBaun et al. (34Weksberg et al. (35Mosaic ratio SNP array Extent of pUPD Birth length (cm) Birth length (SD) Birth weight (g) Birth weight (SD) Gestational age (day) 
BWS003 69% Seg-pUPD      
BWS044 45%  PUD*      
BWS045 98% +(SNP5.0) PUD   4256 6.82 244 
BWS050  58% Seg-pUPD 58.0 6.25 4290 4.25 265 
BWS053 54% Seg-pUPD 53.0 3.43 3637 3.39 255 
BWS060   70% Seg-pUPD 51.0 1.98 3370 2.11 259 
BWS107 51% Seg-pUPD 52.8 2.16 3830 2.21 276 
BWS109 62% Seg-pUPD 44.5 0.74 2839 3.57 230 
BWS113 35% Seg-pUPD 51.4 1.69 3834 2.64 269 
bwsh21-002   91% PUD 41.5 −0.09 1940 −0.08 235 
bwsh21-006 90% Whole chromosome 11 49.0 0.58 3666 2.76 264 
bwsh21-013 37% Seg-pUPD 51.5 1.43 4020 2.71 275 
bwsh21-022 74% Seg-pUPD 51.0 2.73 4335 6.05 248 
bwsh21-024 70% Seg-pUPD 52.0 2.38 4350 4.35 262 
bwsh21-025 58% Seg-pUPD 52.8 2.21 3856 2.26 275 
bwsh21-027a  82%  PUD* 34.2 0.29 1211 2.46 184 
bwsh21-028 28% Seg-pUPD 51.0 1.56 4126 3.85 266 
bwsh21-030  62% Seg-pUPD 49.0 −0.13 3090 0.23 278 
bwsh21-036 60% Seg-pUPD 51.0 1.98 2980 1.35 259 
bwsh21-042  96% Seg-pUPD   4236 3.29 280 
bwsh21-051  34% PUD   3450 3.11 253 
bwsh21-056  62% Seg-pUPD 54.0 2.53 4440 3.67 283 
bwsh21-066  24% Seg-pUPD 46.6 −0.26 2830 0.35 258 
bwsh21-070 86% Seg-pUPD 53.0 2.08 4126 2.69 280 
bwsh21-072  88% Seg-pUPD 53.0 1.83 4272 2.82 285 
bwsh21-078   72% PUD 52.0 2.15 3738 2.99 267 
bwsh21-086  25% Seg-pUPD 51.5 1.61 3804 2.42 271 
bwsh21-095 17% Seg-pUPD 53.4 2.06 4402 3.14 286 
bwsh21-099 71% Seg-pUPD 54.0 5.16 2980 2.68 240 
bwsh21-104  43% Seg-pUPD 51.0 0.90 3996 2.66 280 
bwsh21-109 79% Seg-pUPD 45.5 0.80 3305 4.76 236 
bwsh21-110  44% Seg-pUPD 49.8 0.87 3212 1.11 267 
bwsh21-118 65% Seg-pUPD 51.1 1.85 3545 2.54 262 
bwsh21-119  40%  Seg-pUPD   3200 3.37 245 
bwsh21-123    100% PUD 47.0 1.16 2050 −0.08 241 
bwsh21-125 15%  Seg-pUPD 52.0 1.92 4315 3.69 271 
Patient ID Sex Elliot et al. (33DeBaun et al. (34Weksberg et al. (35Mosaic ratio SNP array Extent of pUPD Birth length (cm) Birth length (SD) Birth weight (g) Birth weight (SD) Gestational age (day) 
BWS003 69% Seg-pUPD      
BWS044 45%  PUD*      
BWS045 98% +(SNP5.0) PUD   4256 6.82 244 
BWS050  58% Seg-pUPD 58.0 6.25 4290 4.25 265 
BWS053 54% Seg-pUPD 53.0 3.43 3637 3.39 255 
BWS060   70% Seg-pUPD 51.0 1.98 3370 2.11 259 
BWS107 51% Seg-pUPD 52.8 2.16 3830 2.21 276 
BWS109 62% Seg-pUPD 44.5 0.74 2839 3.57 230 
BWS113 35% Seg-pUPD 51.4 1.69 3834 2.64 269 
bwsh21-002   91% PUD 41.5 −0.09 1940 −0.08 235 
bwsh21-006 90% Whole chromosome 11 49.0 0.58 3666 2.76 264 
bwsh21-013 37% Seg-pUPD 51.5 1.43 4020 2.71 275 
bwsh21-022 74% Seg-pUPD 51.0 2.73 4335 6.05 248 
bwsh21-024 70% Seg-pUPD 52.0 2.38 4350 4.35 262 
bwsh21-025 58% Seg-pUPD 52.8 2.21 3856 2.26 275 
bwsh21-027a  82%  PUD* 34.2 0.29 1211 2.46 184 
bwsh21-028 28% Seg-pUPD 51.0 1.56 4126 3.85 266 
bwsh21-030  62% Seg-pUPD 49.0 −0.13 3090 0.23 278 
bwsh21-036 60% Seg-pUPD 51.0 1.98 2980 1.35 259 
bwsh21-042  96% Seg-pUPD   4236 3.29 280 
bwsh21-051  34% PUD   3450 3.11 253 
bwsh21-056  62% Seg-pUPD 54.0 2.53 4440 3.67 283 
bwsh21-066  24% Seg-pUPD 46.6 −0.26 2830 0.35 258 
bwsh21-070 86% Seg-pUPD 53.0 2.08 4126 2.69 280 
bwsh21-072  88% Seg-pUPD 53.0 1.83 4272 2.82 285 
bwsh21-078   72% PUD 52.0 2.15 3738 2.99 267 
bwsh21-086  25% Seg-pUPD 51.5 1.61 3804 2.42 271 
bwsh21-095 17% Seg-pUPD 53.4 2.06 4402 3.14 286 
bwsh21-099 71% Seg-pUPD 54.0 5.16 2980 2.68 240 
bwsh21-104  43% Seg-pUPD 51.0 0.90 3996 2.66 280 
bwsh21-109 79% Seg-pUPD 45.5 0.80 3305 4.76 236 
bwsh21-110  44% Seg-pUPD 49.8 0.87 3212 1.11 267 
bwsh21-118 65% Seg-pUPD 51.1 1.85 3545 2.54 262 
bwsh21-119  40%  Seg-pUPD   3200 3.37 245 
bwsh21-123    100% PUD 47.0 1.16 2050 −0.08 241 
bwsh21-125 15%  Seg-pUPD 52.0 1.92 4315 3.69 271 

Seg-pUPD, segmental paternal UPD of chromosome 11; PUD, paternal uniparental diploidy; PUD*, PUD identified by microsatellite analysis with microsatellite markers mapped to chromosomes 1, 2, 5, 6, 8, 9, 11, 14, 15 and 16 (see Supplementary Material, Fig. S2); +, patient met the clinical criteria for BWS or analyzed by SNP array 6.0.

aKaryotype was 46,XY/46,XX mosaic; genomic DNA of BWS044 was extracted from frozen lung tissue.

Next, we performed SNP array analysis to investigate the length of pUPDs. We analyzed 31 patients with the Genome-Wide Human SNP Array 6.0 (Affymetrix, Santa Clara, CA) and one with the Genome-Wide Human SNP Array 5.0 (Affymetrix) (Table 1). Because patients with pUPD show mosaicism, pUPD is considered to be a consequence of mitotic recombination at an early embryonic stage (6). A breakpoint of mitotic recombination must be present between a proximal probe with homozygous calls and a distal probe showing heterozygous calls. We used Nexus Copy Number software (BioDiscovery Inc., El Segundo, CA, USA) to determine proximal and distal probes and considered the length of pUPD to be the position of the proximal probe. Twenty-five patients (78%) showed segmental pUPD limited to the short-arm of chromosome 11, one (3%) showed pUPD extending to the long-arm of chromosome 11, one (3%) showed pUPD of the entire chromosome 11 and, surprisingly, five (16%) showed pUPD of the whole genome, denoted as PUD (Fig. 1 and Supplementary Material, Table S1). The instance of pUPD encompassing the entire chromosome 11 was probably caused by mitotic non-disjunction at the post-fertilization stage, as markers on chromosome 11 showed isodisomy (data not shown). We also found two additional patients with PUD by microsatellite marker analysis using markers mapped to chromosomes 1, 2, 5, 6, 8, 9, 11, 14, 15 and 16 (Table 1 and Supplementary Material, Fig. S2). The total number of patients with PUD was 7 out of 36 (19%). Among them, six were female, and one was phenotypically male with 46,XY/46,XX mosaicism (Table 1). All pUPDs started from the telomere of 11p, and the minimal length of pUPD (found in bwsh21-042) was 2.71 Mb and included both H19DMR and KvDMR1. The proximal probe of the minimal pUPD was within KCNQ1 (Supplementary Material, Fig. S3).

Figure 1.

The lengths of pUPDs obtained from SNP array analyses. The lengths of pUPDs are indicated by the gray bars. Twenty-five patients showed segmental pUPD of 11p, one showed pUPD extending to 11q, one showed pUPD of the entire chromosome 11 and five showed PUD.

Figure 1.

The lengths of pUPDs obtained from SNP array analyses. The lengths of pUPDs are indicated by the gray bars. Twenty-five patients showed segmental pUPD of 11p, one showed pUPD extending to 11q, one showed pUPD of the entire chromosome 11 and five showed PUD.

Search for consensus motifs for mitotic recombination associated with segmental pUPD

A mechanism initiating DNA DSBs within 11p leading to segmental pUPD has been predicted (see ‘Introduction’ section), and accordingly, we searched for consensus DNA motifs associated with mitotic recombination. We defined a pUPD breakpoint region, which consisted of a 5 kb region upstream of the proximal probe, a 5 kb region downstream of the distal probe and the interval region between the two probes (Fig. 2A). From data sets of 26 segmental pUPD patients analyzed by SNP arrays, the average distance between the proximal and distal probes was 8462 ± 10 309 bp (Supplementary Material, Table S1). Therefore, the average size of pUPD breakpoint regions was 18,462 bp. We searched for consensus motifs using two methods (Fig. 2B) (detailed in ‘Materials and Methods’ section). One method was based on multiple alignments; the other was based on BLAST searches. In the multiple alignments method, we simply aligned the 26 sequences of all the breakpoint regions using CLC Genomics Workbench software (CLC bio, Aarhus, Denmark). After the alignment, we searched for candidate motifs using our own criteria (detailed in ‘Materials and Methods’ section) and selected 24 candidate motifs (Table 2). We compared the appearance frequency of each motif within all the breakpoint regions of the 26 patients with those within chromosome 11, using a one-sided Fisher's exact test. Seven motifs were significantly enriched in the breakpoint regions (P < 0.0001) with significant odds ratios (significant P values are indicated with bold numbers in Table 2). Then, for each patient, we also compared the appearance frequency of each motif in each breakpoint region with the overall motif frequency in chromosome 11. One motif, CCNNNCT, showed P < 0.01 and was detected in 11 patients (motifs and the number of the patients are indicated in boldface in Table 2). Furthermore, we calculated the q-value of the motif based on a Poisson distribution and found a q-value lower than 0.01 in the breakpoint regions of four patients (Figs. 3, 4 and Table 3, and Supplementary Material, Fig. S4A and Table S2).

Table 2.

Candidate consensus motifs found in this study

 Motifs All breakpoint regions
 
Chromosome 11
 
P value (Fisher's exact test) Odds ratio 95% confidence interval Number of patients with P < 0.01 (Fisher's exact test) Frequency (%) (n = 26) 
Observed appearance number Estimated maximum appearance numbera Observed appearance number Estimated maximum appearance numberb 
Multiple alignments CCNNNCT 2724 65 861 16 475 732 482 918 826 2.20E16 1.212 1.173583 to inf 11 42.3 
TGCCTG 332 79 684 1 750 684 580 876 296 1.27E08 1.382 1.259788 to inf 19.2 
GGNNTNNTGG 257 47 752 1 382 108 348 194 080 1.98E06 1.356 1.219454 to inf 7.7 
CCCNGNTANTTT 55 39 953 204 828 291 108 662 4.75E06 1.956 1.543381 to inf 15.4 
CTNAGCCTCC 102 47 907 472 030 349 104 158 1.19E05 1.575 1.327024 to inf 11.5 
CANNNCT 2597 65 988 17 539 548 481 855 010 5.94E05 1.081 1.045867 to inf 15.4 
GCTNAG 637 79 379 3 967 470 578 659 510 6.09E05 1.170 1.094914 to inf 7.7 
GGTNTC 487 79 529 2 989 012 579 637 968 1.26E−04 1.187 1.100137 to inf 7.7 
CTNGNNCA 481 59 531 2 979 652 433 990 596 2.82E−04 1.177 1.089651 to inf 15.4 
CCNCANCNTC 149 47 860 812 032 348 764 156 4.04E−04 1.337 1.161934 to inf 11.5 
CCTNGNNCA 159 53 185 903 630 387 514 348 1.45E−03 1.282 1.119301 to inf 11.5 
CCANCANNCCC 71 43 574 354 484 317 442 060 1.58E−03 1.459 1.186244 to inf 11.5 
GANTAC 408 79 608 2 557 464 580 069 516 1.63E−03 1.162 1.069201 to inf 19.2 
ATGNNNANACCC 54 39 954 257 738 291 055 752 2.16E−03 1.526 1.201219 to inf 7.7 
TCNCAA 696 79 320 4 641 000 577 985 980 1.12E−02 1.093 1.025283 to inf 11.5 
CCAAAG 252 79 764 1 586 988 581 039 992 1.28E−02 1.157 1.039344 to inf 7.7 
AGAATCANTT 23 47 986 106 054 349 470 134 2.47E−02 1.079 1.579389 to inf 3.8 
CCANCA 789 79 227 5 372 224 577 254 756 3.14E−02 1.070 1.007894 to inf 7.7 
TANTTT 1447 78 569 10 222 394 572 404 586 1.26E−01 1.031 0.9866789 to inf 15.4 
TANTNNGG 384 59 628 2 692 092 434 278 156 2.35E−01 1.039 0.9529627 to inf 3.8 
AGAATCA 65 68 520 459 134 498 935 424 4.20E−01 1.031 0.8298448 to inf 3.8 
ATNCTC 520 79 496 3 783 390 578 843 590 4.99E−01 1.001 0.9295047 to inf 0.0 
ATNCNTC 387 68 198 3 005 054 496 389 504 9.03E−01 0.937 0.8601813 to inf 3.8 
TANNNAAA 1106 58 906 9 323 262 427 646 986 1.00E+00 0.861 0.8186622 to inf 3.8 
BLAST search CTNGG 2508 93 511 14 428 102 684 724 274 2.20E16 1.273 1.230768 to inf 11 42.3 
GNCAGG 1004 79 012 5 446 714 577 180 266 2.20E16 1.347 1.277023 to inf 11 42.3 
CTNGGCC 231 68 354 1 033 552 498 361 006 4.20E12 1.630 1.629504 to inf 19.2 
GATTACAGG 116 53 228 485 420 387 932 558 2.76E08 1.742 1.48424 to inf 10 38.5 
CTCCCAAA 112 59 900 480 090 436 490 158 1.50E07 1.700 1.44447 to inf 23.1 
TCACTGCA 103 59 909 438 152 436 532 096 3.26E07 1.713 1.444874 to inf 19.2 
TGAGCCAC 95 59 917 398 658 436 571 590 5.12E07 1.736 1.453828 to inf 11.5 
GCTGGAGT 94 59 918 393 848 436 576 400 5.49E07 1.739 1.454647 to inf 19.2 
GCTAATTTT 79 53 265 353 574 388 064 404 3.68E05 1.628 1.338528 to inf 19.2 
ANCTCNTG 243 59 769 1 362 686 435 607 562 5.10E05 1.300 1.165374 to inf 15.4 
TAGAGA 248 79 768 1 515 722 581 111 258 3.89E−03 1.192 1.070076 to inf 7.7 
CACCATGC 46 59 966 220 272 436 749 976 4.57E−03 1.521 1.171744 to inf 7.7 
TGGTCT 210 79 806 1 276 028 581 350 952 5.80E−03 1.199 1.06592 to inf 11.5 
CAGTG 770 95 249 5 202 704 693 949 672 2.04E−02 1.078 1.014928 to inf 3.8 
CACCA 671 95 348 4 570 618 694 581 758 4.45E−02 1.069 1.002227 to inf 7.7 
Meiotic recombination motif CCCCACCCC 20 53 324 95 394 388 322 584 4.55E−02 1.527 1.011738 to inf 7.7 
GGGGGT 97 79 919 599 586 582 027 394 6.22E−02 1.178 0.9884445 to inf 7.7 
CCACGTGG 60 006 29 302 436 940 946 2.19E−01 1.491 0.6492949 to inf 3.8 
CCTCCCTG 25 59 987 153 790 436 816 458 2.26E−01 1.184 0.8229634 to inf 0.0 
CCTCCCT 72 68 513 522 626 498 871 932 5.05E−01 1.003 0.8168118 to inf 3.8 
TACTGTTC 60 006 46 488 436 923 760 6.14E−01 0.409 0.9397834 to inf 0.0 
CCNCCNTNNCCNC 23 36 907 176 826 268 727 940 6.30E−01 0.947 0.6472155 to inf 0.0 
 Motifs All breakpoint regions
 
Chromosome 11
 
P value (Fisher's exact test) Odds ratio 95% confidence interval Number of patients with P < 0.01 (Fisher's exact test) Frequency (%) (n = 26) 
Observed appearance number Estimated maximum appearance numbera Observed appearance number Estimated maximum appearance numberb 
Multiple alignments CCNNNCT 2724 65 861 16 475 732 482 918 826 2.20E16 1.212 1.173583 to inf 11 42.3 
TGCCTG 332 79 684 1 750 684 580 876 296 1.27E08 1.382 1.259788 to inf 19.2 
GGNNTNNTGG 257 47 752 1 382 108 348 194 080 1.98E06 1.356 1.219454 to inf 7.7 
CCCNGNTANTTT 55 39 953 204 828 291 108 662 4.75E06 1.956 1.543381 to inf 15.4 
CTNAGCCTCC 102 47 907 472 030 349 104 158 1.19E05 1.575 1.327024 to inf 11.5 
CANNNCT 2597 65 988 17 539 548 481 855 010 5.94E05 1.081 1.045867 to inf 15.4 
GCTNAG 637 79 379 3 967 470 578 659 510 6.09E05 1.170 1.094914 to inf 7.7 
GGTNTC 487 79 529 2 989 012 579 637 968 1.26E−04 1.187 1.100137 to inf 7.7 
CTNGNNCA 481 59 531 2 979 652 433 990 596 2.82E−04 1.177 1.089651 to inf 15.4 
CCNCANCNTC 149 47 860 812 032 348 764 156 4.04E−04 1.337 1.161934 to inf 11.5 
CCTNGNNCA 159 53 185 903 630 387 514 348 1.45E−03 1.282 1.119301 to inf 11.5 
CCANCANNCCC 71 43 574 354 484 317 442 060 1.58E−03 1.459 1.186244 to inf 11.5 
GANTAC 408 79 608 2 557 464 580 069 516 1.63E−03 1.162 1.069201 to inf 19.2 
ATGNNNANACCC 54 39 954 257 738 291 055 752 2.16E−03 1.526 1.201219 to inf 7.7 
TCNCAA 696 79 320 4 641 000 577 985 980 1.12E−02 1.093 1.025283 to inf 11.5 
CCAAAG 252 79 764 1 586 988 581 039 992 1.28E−02 1.157 1.039344 to inf 7.7 
AGAATCANTT 23 47 986 106 054 349 470 134 2.47E−02 1.079 1.579389 to inf 3.8 
CCANCA 789 79 227 5 372 224 577 254 756 3.14E−02 1.070 1.007894 to inf 7.7 
TANTTT 1447 78 569 10 222 394 572 404 586 1.26E−01 1.031 0.9866789 to inf 15.4 
TANTNNGG 384 59 628 2 692 092 434 278 156 2.35E−01 1.039 0.9529627 to inf 3.8 
AGAATCA 65 68 520 459 134 498 935 424 4.20E−01 1.031 0.8298448 to inf 3.8 
ATNCTC 520 79 496 3 783 390 578 843 590 4.99E−01 1.001 0.9295047 to inf 0.0 
ATNCNTC 387 68 198 3 005 054 496 389 504 9.03E−01 0.937 0.8601813 to inf 3.8 
TANNNAAA 1106 58 906 9 323 262 427 646 986 1.00E+00 0.861 0.8186622 to inf 3.8 
BLAST search CTNGG 2508 93 511 14 428 102 684 724 274 2.20E16 1.273 1.230768 to inf 11 42.3 
GNCAGG 1004 79 012 5 446 714 577 180 266 2.20E16 1.347 1.277023 to inf 11 42.3 
CTNGGCC 231 68 354 1 033 552 498 361 006 4.20E12 1.630 1.629504 to inf 19.2 
GATTACAGG 116 53 228 485 420 387 932 558 2.76E08 1.742 1.48424 to inf 10 38.5 
CTCCCAAA 112 59 900 480 090 436 490 158 1.50E07 1.700 1.44447 to inf 23.1 
TCACTGCA 103 59 909 438 152 436 532 096 3.26E07 1.713 1.444874 to inf 19.2 
TGAGCCAC 95 59 917 398 658 436 571 590 5.12E07 1.736 1.453828 to inf 11.5 
GCTGGAGT 94 59 918 393 848 436 576 400 5.49E07 1.739 1.454647 to inf 19.2 
GCTAATTTT 79 53 265 353 574 388 064 404 3.68E05 1.628 1.338528 to inf 19.2 
ANCTCNTG 243 59 769 1 362 686 435 607 562 5.10E05 1.300 1.165374 to inf 15.4 
TAGAGA 248 79 768 1 515 722 581 111 258 3.89E−03 1.192 1.070076 to inf 7.7 
CACCATGC 46 59 966 220 272 436 749 976 4.57E−03 1.521 1.171744 to inf 7.7 
TGGTCT 210 79 806 1 276 028 581 350 952 5.80E−03 1.199 1.06592 to inf 11.5 
CAGTG 770 95 249 5 202 704 693 949 672 2.04E−02 1.078 1.014928 to inf 3.8 
CACCA 671 95 348 4 570 618 694 581 758 4.45E−02 1.069 1.002227 to inf 7.7 
Meiotic recombination motif CCCCACCCC 20 53 324 95 394 388 322 584 4.55E−02 1.527 1.011738 to inf 7.7 
GGGGGT 97 79 919 599 586 582 027 394 6.22E−02 1.178 0.9884445 to inf 7.7 
CCACGTGG 60 006 29 302 436 940 946 2.19E−01 1.491 0.6492949 to inf 3.8 
CCTCCCTG 25 59 987 153 790 436 816 458 2.26E−01 1.184 0.8229634 to inf 0.0 
CCTCCCT 72 68 513 522 626 498 871 932 5.05E−01 1.003 0.8168118 to inf 3.8 
TACTGTTC 60 006 46 488 436 923 760 6.14E−01 0.409 0.9397834 to inf 0.0 
CCNCCNTNNCCNC 23 36 907 176 826 268 727 940 6.30E−01 0.947 0.6472155 to inf 0.0 

Total length of all breakpoint regions is 480 099 bp. The length of chromosome 11 is 134 452 384 bp (NCBI36/hg18).

aCalculation formula is (length of all breakpoint regions)/(length of each motif) − observed appearance number.

bCalculation formula is ((length of chromosome 11)/(length of each motif) − observed appearance number in chromosome 11) × 26; inf, infinity.

Table 3.

Appearance of individual motifs in each patient's breakpoint region

Patient ID CCNNNCT
 
CTNGG
 
GNCAGG
 
GATTACAGG
 
Total count of appearances Number of windows q-value Total count of appearances Number of windows q-value Total count of appearances Number of windows q-value Total count of appearances Number of windows q-value 
BWS003          53 26 0.00431 
BWS050             
BWS053          0.00443 
BWS060             
BWS107             
BWS109       0.00567 0.00443 
BWS113          0.00443 
bwsh21-013             
bwsh21-022             
bwsh21-024 11 0.00215          
bwsh21-025    10 0.00599       
bwsh21-028          0.00443 
bwsh21-030 21 0.00417    26 0.00365 34 17 0.00443 
bwsh21-036             
bwsh21-042    20 0.00599       
bwsh21-056    20 0.00599       
bwsh21-066          0.00381 
bwsh21-070 35 0.00233    0.00162    
bwsh21-072             
bwsh21-086       0.00567 0.00443 
bwsh21-095 21 0.00417 10 0.00599 0.00162 0.00443 
bwsh21-099             
bwsh21-104          0.00443 
bwsh21-109             
bwsh21-110       0.00567    
bwsh21-118             
Patient ID CCNNNCT
 
CTNGG
 
GNCAGG
 
GATTACAGG
 
Total count of appearances Number of windows q-value Total count of appearances Number of windows q-value Total count of appearances Number of windows q-value Total count of appearances Number of windows q-value 
BWS003          53 26 0.00431 
BWS050             
BWS053          0.00443 
BWS060             
BWS107             
BWS109       0.00567 0.00443 
BWS113          0.00443 
bwsh21-013             
bwsh21-022             
bwsh21-024 11 0.00215          
bwsh21-025    10 0.00599       
bwsh21-028          0.00443 
bwsh21-030 21 0.00417    26 0.00365 34 17 0.00443 
bwsh21-036             
bwsh21-042    20 0.00599       
bwsh21-056    20 0.00599       
bwsh21-066          0.00381 
bwsh21-070 35 0.00233    0.00162    
bwsh21-072             
bwsh21-086       0.00567 0.00443 
bwsh21-095 21 0.00417 10 0.00599 0.00162 0.00443 
bwsh21-099             
bwsh21-104          0.00443 
bwsh21-109             
bwsh21-110       0.00567    
bwsh21-118             
Figure 2.

Strategy for finding consensus motifs within pUPD breakpoint regions. (A) Definition of a pUPD breakpoint region. Top shows a representative image of allelic deference generated by SNP array analysis. Bottom shows the definition of pUPD breakpoint regions. Proximal probe (prox probe) with homocalls and distal probe (dis probe) with heterocalls were determined by Nexus Copy Number software. The genomic positions corresponded to NCBI36/hg18. (B) Strategy for searching for consensus motifs. Two methods were used: (1) simply multiplying alignments of all breakpoint regions; and (2) comparing pUPD breakpoint regions one by one using BLAST search (see the details in ‘Materials and Methods’ section). Candidate motifs were then selected using our own criteria. Statistical analysis with Fisher's exact test was performed and q-values were calculated based on Poisson distributions.

Figure 2.

Strategy for finding consensus motifs within pUPD breakpoint regions. (A) Definition of a pUPD breakpoint region. Top shows a representative image of allelic deference generated by SNP array analysis. Bottom shows the definition of pUPD breakpoint regions. Proximal probe (prox probe) with homocalls and distal probe (dis probe) with heterocalls were determined by Nexus Copy Number software. The genomic positions corresponded to NCBI36/hg18. (B) Strategy for searching for consensus motifs. Two methods were used: (1) simply multiplying alignments of all breakpoint regions; and (2) comparing pUPD breakpoint regions one by one using BLAST search (see the details in ‘Materials and Methods’ section). Candidate motifs were then selected using our own criteria. Statistical analysis with Fisher's exact test was performed and q-values were calculated based on Poisson distributions.

Figure 3.

Appearance and q-value of each consensus motif. Number of motif appearances (count) within 500 bp are shown by blue bars. The reciprocals of the q-values within 500 bp are shown by green bars. Breakpoint regions of the patients who show significant q-values within their breakpoint regions are indicated at the bottom.

Figure 3.

Appearance and q-value of each consensus motif. Number of motif appearances (count) within 500 bp are shown by blue bars. The reciprocals of the q-values within 500 bp are shown by green bars. Breakpoint regions of the patients who show significant q-values within their breakpoint regions are indicated at the bottom.

Figure 4.

Representative breakpoint regions with significant q-values. Significant peaks of reciprocal q-values (more than 100) are seen in all breakpoint regions. Numbers of motif appearances (count) within 500 bp are shown by blue bars. The reciprocals of q-values within 500 bp are shown by green bars.

Figure 4.

Representative breakpoint regions with significant q-values. Significant peaks of reciprocal q-values (more than 100) are seen in all breakpoint regions. Numbers of motif appearances (count) within 500 bp are shown by blue bars. The reciprocals of q-values within 500 bp are shown by green bars.

Aside from the multiple alignments method, we also used a method based on BLAST to search consensus motifs (Fig. 2B). BLAST searches regions of local similarity between two nucleotide sequences and can find additional candidate motifs. We compared sequences of all breakpoint regions from 26 patients one by one using BLAST. A total of 325 pairwise sequences generated by selecting 2 out of the 26 breakpoint regions were analyzed. The top five sequences with the lowest E-values were chosen from each BLAST comparison for a total number of chosen sequences of 123. Then, we aligned the 123 sequences using CLC Genomics Workbench software (CLC bio) and searched for candidate motifs using the same criteria as above (detailed in ‘Materials and Methods’ section). We obtained 15 candidate motifs (Table 2). Next, we compared the frequency of each motif within all breakpoint regions of the 26 patients with those within chromosome 11, using a one-sided Fisher's exact test. Ten motifs were significantly enriched in the breakpoint regions (P < 0.0001) with significant odds ratios (significant P values are indicated with bold numbers in Table 2). We also compared the appearance frequency of each motif in each breakpoint region with that of the appearances on the entire chromosome 11 for each patient. Three motifs, CTNGG, GNCAGG and GATTACAGG, showed P < 0.01 in at least 10 patients (motifs and the number of the patients are indicated in boldface in Table 2). Furthermore, we calculated the q-value of the three motifs based on a Poisson distribution and found q-values lower than 0.01 in the breakpoint regions of at least four patients (Figs 3, 4 and Table 3, and Supplementary Material, Fig. S4B–D and Table S2). We identified four consensus motifs from the breakpoint regions of segmental pUPD patients using the two methods above. The four motifs were presented in 16 out of 26 breakpoint regions (62%). These results suggested that the motifs are involved in mitotic recombination.

Replication analysis for the existence of the four motifs in other segmental pUPD breakpoint regions

To confirm whether the four motifs existed in breakpoint regions of other patients with segmental pUPD, we performed the same analyses mentioned above on nine segmental pUPD patients who had been reported in Romanelli et al. (1). These patients had been analyzed to find the length of UPD through the use of Illumina Human610-Quad BeadChips (1). The four motifs were enriched with P < 0.0001 with a significant odds ratio in all breakpoint regions (Supplementary Material, Table S3). At a patient level, all motifs were also found in a number of patients with P < 0.01 (Supplementary Material, Table S3). Furthermore, the four motifs showed significantly low q-values in the patients (Supplementary Material, Table S4 and Fig. S5). The results of this replication analysis supported our results.

Analysis of the consensus motifs associated with meiotic recombination within segmental pUPD breakpoint regions

In addition, we analyzed our patients for the appearance of seven motifs associated with meiotic homologous recombination (10,14). However, there was no significance in the appearance frequency within all breakpoint regions (Table 2), suggesting that the motifs associated with meiotic recombination were not associated with mitotic recombination in somatic cells. Taken together, the four motifs we found in this study were specifically associated with mitotic recombination to generate segmental pUPD of chromosome 11 in a certain number of patients.

Search for protein binding sites in the four consensus motifs

In yeast, the recombination machinery is recruited to transcription binding sites (15). To determine whether the four motifs were transcription factor binding sites, we performed a search using the JASPAR database (http://jaspardev.genereg.net) (Supplementary Material, Table S5). Three of the motifs, CCNNNCT, CTNGG and GNCAGG, contained candidate transcription factor binding sites. However, factors that might be associated with mitotic and meiotic recombination were not found. There were no associated factors for the final motif, GATTACAGG. The results suggest that specific transcription factors do not bind to the four motifs and that the binding of transcription factors is not involved in mitotic recombination leading to segmental pUPD.

Genomic and epigenomic characteristics of pUPD breakpoint regions

We investigated genomic and epigenomic characteristics of pUPD breakpoint regions using the International Human Epigenome Consortium (IHEC) database (http://epigenomesportal.ca/ihec/index.html) and the NCBI epigenomics database (http://www.ncbi.nlm.nih.gov/epigenomics/). Among the 26 breakpoint regions, 19 overlapped genic regions (four in promoters, twelve in gene bodies, two at the gene's end and one over the whole gene) and seven existed in intergenic regions (Supplementary Material, Table S6). Four breakpoint regions contained a total of five CpG islands (four in promoters, and one in a gene body). In addition, we examined whether the positions of breakpoint regions were associated with common fragile sites (CFSs). CFSs are specific chromosomal regions that preferentially form gaps or breaks on metaphase chromosomes under replication stress. Using data reported by Fungtammasan et al. (16), in which aphidicolin-induced CFSs and nonfragile control regions (NFRs) were defined, eight and seven breakpoint regions were located within two CFSs and three NFRs, respectively (Supplementary Material, Table S6). The results suggested that these genomic characteristics are not involved in mitotic recombination.

Since H3K4 trimethylation is enriched at active hotspots in mouse spermatocytes (17) and mitotic recombination in BWS should occur at the pre-implantation stage, we searched for histone modifications in the human ES cell line H9, though the developmental period of the ES cells was later than that of the mitotic recombination in BWS (Supplementary Material, Table S6). H3K4 trimethylation was found only in three regions, but was associated with gene promoters, suggesting active transcription. We found other modifications, such as H3K27 acetylation, H3K27 trimethylation, H3K36 trimethylation, H3K4 monomethylation and H3K9 trimethylation; however, there was only one positive breakpoint region for each modification. We also searched chromatin accessibility obtained by DNase-seq in the NCBI epigenomics database and found DNase hypersensitivity sites in only seven regions (four in promoters, two in gene bodies and two in intergenic regions). Three sites were found in gene promoters associated with H3K4 trimethylation, again suggesting active transcription. The paucity of positive data suggests that epigenetic modifications are not involved in mitotic recombination, in contrast to meiotic recombination.

The relationship among variations of segmental pUPD and clinical features

As mentioned above, the mosaic ratio and the length of segmental pUPD varied among patients. We analyzed the relationship between the mosaic ratio and the length of segmental pUPD. For this study, a patient with pUPD of whole chromosome 11 was included and the length of pUPD was given by the common logarithm of the length. We found no correlation between the mosaic ratio and the length of pUPD as previously described (Supplementary Material, Fig. S6A) (1). We also analyzed the relationship between the variations of segmental pUPD and clinical features, such as birth length, birth weight, gestational age, macroglossia, macrosomia, abdominal wall defects, ear creases/pits, neonatal hypoglycemia, hemihyperplasia, nevus flammeus, renal anomalies, genital anomalies, visceromegaly, developmental disabilities, cardiac anomalies and tumor development. However, no correlation was found between them (Supplementary Material, Fig. S6B and Table S7).

Characteristic clinical features of PUDs

Because we found seven PUD patients among the 36 patients with pUPD analyzed in this study (Table 1), we compared the clinical features of PUDs with those of segmental pUPDs, which included a pUPD of the entire chromosome 11. For this study, we included 14 PUD patients who had been previously reported (18–28). Birth length, birth weight and gestational age did not differ between patients (Supplementary Material, Fig. S7). As for other clinical features, macroglossia was less frequent in PUDs than in segmental pUPDs, indicating that macroglossia was specific to segmental pUPDs (Table 4). In addition, developmental disabilities, cardiac anomalies and tumor development occurred significantly more often in PUDs after the Bonferroni correction was applied (P = 0.0017, 0.0014 and 0.0000, respectively) (Table 4). Examples of developmental disabilities in PUDs were speech delays, unsteady gaits and seizures. Cardiac anomalies included atrial septal defects, patent foramen ovale, left ventricular hypertrophy and small descending aorta. The tumors that developed included nonmalignant tumors of the liver, adrenal glands, pancreas and breast, in addition to malignant tumors, such as Wilms' tumor, hepatoblastoma and neuroblastoma. This result suggests that in addition to high risk of tumor development, developmental disabilities and cardiac anomalies are also characteristic features of PUD.

Table 4.

Comparison of clinical features between segmental pUPDs and PUDs

 Seg-pUPD PUD P value (χ2-test) 
Macroglossia 27/29 8/20 0.0001* 
Macrosomia 24/28 9/18 0.0087 
Abdominal wall defect 20/29 11/19 0.4329 
Ear creases/pits 16/28 4/16 0.0394 
Neonatal hypoglycemia 22/29 17/19 0.2137a 
Hemihyperplasia 20/29 13/20 0.7711 
Nevus flammeus 9/27 7/20 0.9105 
Renal anomaly 2/28 2/19 0.5362a 
Genital anomaly 3/27 3/17 0.4255a 
Visceromegaly 16/28 13/21 0.7372 
Developmental disability 1/24 8/17 0.0017*a 
Cardiac anomaly 1/27 8/18 0.0014*a 
Tumor development 4/25 17/21 0.0000* 
 Seg-pUPD PUD P value (χ2-test) 
Macroglossia 27/29 8/20 0.0001* 
Macrosomia 24/28 9/18 0.0087 
Abdominal wall defect 20/29 11/19 0.4329 
Ear creases/pits 16/28 4/16 0.0394 
Neonatal hypoglycemia 22/29 17/19 0.2137a 
Hemihyperplasia 20/29 13/20 0.7711 
Nevus flammeus 9/27 7/20 0.9105 
Renal anomaly 2/28 2/19 0.5362a 
Genital anomaly 3/27 3/17 0.4255a 
Visceromegaly 16/28 13/21 0.7372 
Developmental disability 1/24 8/17 0.0017*a 
Cardiac anomaly 1/27 8/18 0.0014*a 
Tumor development 4/25 17/21 0.0000* 

*Significant with Bonferroni correction (P < 0.00384).

aFisher's exact test.

Discussion

DNA DSBs are the critical events that initiate both meiotic and mitotic recombination (17,29). For meiotic recombination, DSBs are programed and catalyzed by meiosis-specific topoisomerase-II-like SPO11 endonuclease. More than 25 000 crossover hot spots and several consensus motifs have been identified in human to date (7,10,14). Among the consensus motifs, 13-mer CCNCCNTNNCCNC exists in at least 40% of all hot spots and is recognized by PRDM9, which catalyzes the trimethylation of histone H3 lysine 4 on the nearest nucleosome and is estimated to recruit and activate the SPO11-containing recombination initiation complex to initiate programed DSBs (7,17,30). On the other hand, DSBs for mitotic recombination are not programed and can be caused by variable factors, including endogenous metabolic reactions, replication stress or exogenous chemotherapeutic drugs and ionizing radiation (8). Although a certain mechanism leading to the segmental pUPD of chromosome 11 has been predicted (9,10), no consensus motif has been found, even though the length of segmental pUPD in BWS patients has been analyzed using microsatellite markers and SNP arrays (1,31). The failure to find consensus motifs was probably due to the long intervals between two microsatellite markers or the small number of patients analyzed by SNP arrays. In this study, we analyzed a number of patients using a high density SNP array. In addition, we used two different methods to search consensus motifs. Under these conditions, we could successfully identify four consensus motifs from mitotic recombination breakpoint regions. The method based on BLAST search might be effective and found three out of the four consensus motifs. The motifs existed in 62% of the breakpoint regions and may be involved in the etiology of segmental pUPD in BWS patients. However, we could not find any other conserved genomic characteristics in the breakpoint regions. On the other hand, factors involved in meiotic recombination, such as the seven consensus motifs and H3K4 trimethylation, were not found. Our results suggest that the consensus motifs found in this study are associated with the initiation of DSBs, and that the mechanism involved in mitotic recombination leading to segmental pUPD of 11p differs from that of meiotic recombination.

The relationship between the mosaic ratio of pUPD and clinical features has been previously investigated (1,12,13). We could not find any correlation among the mosaic ratio, the length of pUPD or clinical features of the patients as previously described (1). Although Smith et al. reported that two patients with severe presentations of BWS showed an extremely high mosaic ratio of pUPD in lymphocytes (13), the mosaic ratio in peripheral blood does not generally correlate with phenotypic presentation because the phenotypic presentation should correlate with the mosaic ratio in the organs and tissues (12).

PUD was found in 19% (7 out of 36) of pUPD patients in this study. In a previous study, two PUD patients were found out of 11 pUPD patients (32). The total frequency of PUD in pUPD patients thus became 19% (9 out of 47). PUDs showed more frequent developmental disabilities, cardiac anomalies and tumor development than segmental pUPDs. Although frequent tumor development has been previously described (4), we found that developmental disabilities and cardiac anomalies were additional common features. These features might depend on the mosaic ratio, the imprinting status of other chromosomes and paternally inherited recessive mutations, as previously described (4,18). In addition, the frequencies of tumor development in segmental pUPD and PUD were 16% (4/25) and 81% (17/21) in this study. The frequency of tumor development in pUPDs has been thought to be approximately 25% (5,11). However, this value would be lower than expected because PUD patients, most of whom developed tumors, are present among pUPD patients with relatively high frequency (19%).

In conclusion, we successfully identified four consensus motifs within breakpoint regions of segmental pUPD of chromosome 11. To understand the precise mechanism of mitotic recombination, future work should analyze more patients and focus on detailed epigenomic analysis in earlier developmental stages, such as the pre-implantation stage. We found that developmental disabilities and cardiac anomalies are additional characteristic features of PUD, along with tumor development. Since PUD mosaicism can occur more frequently than expected, patients with pUPD of chromosome 11 should be analyzed for UPD of other chromosomes.

Materials and Methods

BWS patients with pUPD

Thirty-six patients with pUPD of chromosome 11, who met one of the clinical criteria for BWS as described by Elliott et al., DeBaun et al. or Weksberg et al. except for bwsh21-123, were included in this study (Table 1) (33–35). Genomic DNA was extracted from peripheral blood using a FlexiGene DNA Kit (QIAGEN, Hilden, Germany). Genomic DNA of BWS044 was extracted from frozen lung tissue. The presence of pUPD of 11p15 was screened using tetranucleotide repeat markers (D11S1984, HUMTH01, D11S1997) at 11p15 as previously described (4,36). Other markers, such as D11S2001, D11S1983 and D11S2000 at 11p13-q22, were also used to analyze pUPD. The mosaic ratio of pUPD was calculated using the peak heights of the paternal and maternal alleles as follows: mosaic ratio = (k − 1)/(k + 1) × 100, where k is the peak height ratio of paternal to maternal allele in the patient (37). The average mosaic ratio of informative markers was adopted as the mosaic ratio of each patient (Table 1).

This study was approved by the Ethics Committee for Human Genome and Gene Analyses of the Faculty of Medicine, Saga University. Written informed consent was obtained from the patients or guardians.

SNP array analysis

To analyze the length of pUPDs, 31 patients were analyzed with the Genome-Wide Human SNP Array 6.0 (Affymetrix, Santa Clara, CA) and 1, BWS045, was analyzed with the Genome-Wide Human SNP Array 5.0 (Affymetrix) (Table 1). The Genome-Wide Human SNP Array 6.0 features 1.8 million genetic markers, including more than 906 600 probes for SNPs and more than 946 000 probes for the copy number variation. The average and median inter-marker spacings were 3230 and 1270 bp, respectively. The genotypes were analyzed using Genotyping Console™ (GTC) 4.0 analysis (Affymetrix). The genomic positions of the SNPs corresponded to NCBI36/hg18. Copy numbers and allele ratios were analyzed using Nexus Copy Number software 6.0 (BioDiscovery, Hawthorne, CA). To detect the pUPD region, the homozygous value threshold was set to 0.6.

Extraction of consensus motifs around mitotic recombination breakpoints associated with segmental pUPD

We analyzed DNA sequences of pUPD breakpoint regions obtained from SNP array data of 26 BWS patients, which consisted of a 5 kb region upstream of a proximal probe, a 5 kb region downstream of a distal probe and the interval region between the two probes (NCBI36/hg18). We used Nexus Copy Number software (BioDiscovery) with SNP-FASST2 segmentation to determine the proximal and distal probes. The settings of SNP-FASST2 segmentation in the Nexus Copy Number software were as follows: significance threshold = 5.0E−7, maximum contiguous probe spacing = 1000 kb, minimum number of probes per segment = 3, high gain = 0.7, gain = 0.1, loss = −0.15, big loss = −1.1, homozygous frequency threshold = 0.95, homozygous value threshold = 0.8, heterozygous imbalance threshold = 0.4 and minimum LOH length = 500 kb. The HapMap control set provided by the manufacturer was used as a control. To extract candidate motifs, we used two methods, multiple alignments and BLAST search (Fig. 2B). In multiple alignments, we simply aligned all the breakpoint regions obtained from the 26 pUPD patients using CLC Genomics Workbench 5.1 software (CLC bio, Aarhus, Denmark) with progressive alignment algorithm (gap open cost = 10, gap extension cost = 1). After the alignment, we selected candidate motifs using criteria as follows: from 6 mer to 13 mer in length, at least the first two consecutive nucleotides showing more than 75% identity, at least the last two consecutive nucleotides showing more than 60% identity, nucleotides showing less than 60% identity defined as ‘N’ and elimination of the sequences including four or more consecutive same nucleotides. In the BLAST search, we compared pUPD breakpoint regions one by one using BLAST (http://www.ncbi.nlm.nih.gov/BLAST/). A total of 325 pairwise sequences were generated by taking and analyzing 2 of the 26 breakpoint regions at a time. The top five sequences with the lowest E-values were obtained from each BLAST comparison. The E-value is a parameter that describes the number of hits one can ‘expect’ to see by chance. The lower the E-value, the more significant the match is (http://blast.ncbi.nlm.nih.gov/Blast.cgi?CMD=Web&PAGE_TYPE=BlastDocs&DOC_TYPE=FAQ#expect). The obtained sequences with the lowest E-values from all BLAST comparisons were aligned using CLC Genomics Workbench 5.1 software (CLC bio) with the same alignment algorithm mentioned above. After the alignment, we selected candidate motifs using the same criteria mentioned above.

Statistical analyses of the candidate motifs

We statistically analyzed the candidate motifs selected by multiple alignment and BLAST searches. First, we compared the appearance frequency of each motif in all breakpoint regions of all 26 patients with that from the appearance frequency in the total length of chromosome 11 of all 26 patients using a one-sided Fisher's exact test. The appearance frequency of each motif in all breakpoint regions was obtained by dividing the observed appearance number by the estimated maximum appearance number. Because Fisher's exact test requires the elimination of overlapping events, the estimated maximum appearance number in all breakpoint regions was calculated using the following formula: (the total length of all breakpoint regions)/(the length of each motif) – (observed appearance number in all breakpoint regions). The appearance frequency of each motif in total length of chromosome 11 was obtained by dividing the observed appearance number by the estimated maximum appearance number. The estimated maximum appearance number in chromosome 11 was calculated using the following formula: ((the total length of chromosome 11)/(the length of each motif) − (observed appearance number in chromosome 11)) × 26. The motifs which showed P < 0.0001 and an odds ratio >1 were selected.

Next, we compared the appearance frequency of each motif at the level of each patient. The appearance frequency in the breakpoint region of each patient was compared with that over chromosome 11 using a one-sided Fisher's exact test. The appearance frequency of each motif in each patient was obtained by dividing the observed appearance number by the estimated maximum appearance number. The estimated maximum appearance number in each breakpoint region was calculated using the following formula: (the length of each breakpoint region)/(the length of each motif) − (observed appearance number in each breakpoint region). The appearance frequency of each motif over chromosome 11 was obtained by dividing the observed appearance number by the estimated maximum appearance number. The estimated maximum appearance number in chromosome 11 was calculated using the following formula: (the total length of chromosome 11)/(the length of each motif) − (observed appearance number in chromosome 11). We selected motifs that showed P < 0.01 and were detected in at least 10 patients.

Finally, we counted the number of appearances of each motif within chromosome 11 with a window of 500 bp (250 bp overlapped) and calculated q-values for each motif based on a Poisson distribution corrected with the Benjamini-Hochberg method. The motifs with a q-value less than 0.01 within the breakpoint region of any patient were further selected. The motif count and reciprocal q-values were plotted by means of GenomeJack Browser (Mitsubishi Space Software, Tokyo, Japan).

Search for transcription factor binding motifs

The four candidate motifs were searched for in the JASPAR CORE vertebrata (http://jaspardev.genereg.net), a database for transcription factor binding sites. Candidate motifs that produced a score greater than 90% in JASPAR were selected.

Statistical analysis of the relationship between clinical information and variations of pUPD

We performed bivariate analysis to evaluate the relationship between the mosaic ratio and length of segmental pUPDs, and between clinical features (birth length, birth weight and gestational age) and variations of pUPD. For this analysis, the length of pUPD was indicated by a common logarithm of the distance from a telomere of 11p to the proximal probe of the breakpoint. A P-value <0.05 was considered statistically significant. To evaluate the relationship between other clinical features (macroglossia, macrosomia, abdominal wall defects, ear creases/pits, neonatal hypoglycemia, hemihyperplasia, nevus flammeus, renal anomalies, genital anomalies, visceromegaly, developmental disabilities, cardiac anomalies and tumor development) and variations of pUPD, multiple logistic regression analysis was used. The estimated odds ratio increased significantly when the 95% confidence intervals did not include 1.0.

For the comparison of segmental pUPDs with PUDs, the differences in birth length, birth weight and gestational age were evaluated by a Mann–Whitney U test, where P < 0.05 was also considered statistically significant. For the comparison of other clinical features between segmental pUPDs and PUDs, we employed the χ2-test or Fisher's exact test. In this case, we used a corrected P-value with a Bonferroni correction. P < 0.00384 was considered statistically significant.

Supplementary Material

Supplementary Material is available at HMG online.

Funding

This study was supported, in part, by a Grant for Research on Intractable Diseases from the Ministry of Health, Labor and Welfare (H26-nanchitou(nan)-ippan-035 to H.S.); a grant for Child Health and Development from the National Center for Child Health and Development (26-13 to H.S.); a grant from Japan Agency for Medical Research and Development (AMED) (15Aek0109034h0002 to H.S.); a Grant-in-Aid for Challenging Exploratory Research (26670169 to H.S.); a Grant-in-Aid for Scientific Research (C) from the Japan Society for the Promotion of Science (25461554 to K.H.); the Cooperative Research Project Program of the Medical Institute of Bioregulation, Kyushu University (H.S.) and a grant from the Mother and Child Health Foundation (K.H.).

Acknowledgements

We thank all patients and their families for participating in this study and all doctors who provided patient information and samples.

Conflict of Interest statement. None declared.

References

1
Romanelli
V.
,
Meneses
H.N.
,
Fernandez
L.
,
Martinez-Glez
V.
,
Gracia-Bouthelier
R.
,
Fraga
M.
,
Guillen
E.
,
Nevado
J.
,
Gean
E.
,
Martorell
L.
et al
. (
2011
)
Beckwith–Wiedemann syndrome and uniparental disomy 11p: fine mapping of the recombination breakpoints and evaluation of several techniques
.
Eur. J. Hum. Genet.
 ,
19
,
416
421
.
2
Yamazawa
K.
,
Ogata
T.
,
Ferguson-Smith
A.C.
(
2010
)
Uniparental disomy and human disease: an overview
.
Am. J. Med. Genet. C Semin. Med. Genet.
 ,
154C
,
329
334
.
3
Lapunzina
P.
,
Monk
D.
(
2011
)
The consequences of uniparental disomy and copy number neutral loss-of-heterozygosity during human development and cancer
.
Biol. Cell
 ,
103
,
303
317
.
4
Ohtsuka
Y.
,
Higashimoto
K.
,
Sasaki
K.
,
Jozaki
K.
,
Yoshinaga
H.
,
Okamoto
N.
,
Takama
Y.
,
Kubota
A.
,
Nakayama
M.
,
Yatsuki
H.
et al
. (
2015
)
Autosomal recessive cystinuria caused by genome-wide paternal uniparental isodisomy in a patient with Beckwith–Wiedemann syndrome
.
Clin. Genet.
 ,
88
,
261
266
.
5
Soejima
H.
,
Higashimoto
K.
(
2013
)
Epigenetic and genetic alterations of the imprinting disorder Beckwith–Wiedemann syndrome and related disorders
.
J. Hum. Genet.
 ,
58
,
402
409
.
6
Henry
I.
,
Puech
A.
,
Riesewijk
A.
,
Ahnine
L.
,
Mannens
M.
,
Beldjord
C.
,
Bitoun
P.
,
Tournade
M.F.
,
Landrieu
P.
,
Junien
C.
(
1993
)
Somatic mosaicism for partial paternal isodisomy in Wiedemann–Beckwith syndrome: a post-fertilization event
.
Eur. J. Hum. Genet.
 ,
1
,
19
29
.
7
Myers
S.
,
Freeman
C.
,
Auton
A.
,
Donnelly
P.
,
McVean
G.
(
2008
)
A common sequence motif associated with recombination hot spots and genome instability in humans
.
Nat. Genet.
 ,
40
,
1124
1129
.
8
Mehta
A.
,
Haber
J.E.
(
2014
)
Sources of DNA double-strand breaks and models of recombinational DNA repair
.
Cold Spring Harb. Perspect. Biol.
 ,
6
,
a016428
.
9
Engel
E.
(
2006
)
A fascination with chromosome rescue in uniparental disomy: Mendelian recessive outlaws and imprinting copyrights infringements
.
Eur. J. Hum. Genet.
 ,
14
,
1158
1169
.
10
Sandovici
I.
,
Kassovska-Bratinova
S.
,
Vaughan
J.E.
,
Stewart
R.
,
Leppert
M.
,
Sapienza
C.
(
2006
)
Human imprinted chromosomal regions are historical hot-spots of recombination
.
PLoS Genet.
 ,
2
,
e101
.
11
Choufani
S.
,
Shuman
C.
,
Weksberg
R.
(
2013
)
Molecular findings in Beckwith–Wiedemann syndrome
.
Am. J. Med. Genet. C Semin. Med. Genet.
 ,
163C
,
131
140
.
12
Itoh
N.
,
Becroft
D.M.
,
Reeve
A.E.
,
Morison
I.M.
(
2000
)
Proportion of cells with paternal 11p15 uniparental disomy correlates with organ enlargement in Wiedemann–Beckwith syndrome
.
Am. J. Med. Genet.
 ,
92
,
111
116
.
13
Smith
A.C.
,
Shuman
C.
,
Chitayat
D.
,
Steele
L.
,
Ray
P.N.
,
Bourgeois
J.
,
Weksberg
R.
(
2007
)
Severe presentation of Beckwith–Wiedemann syndrome associated with high levels of constitutional paternal uniparental disomy for chromosome 11p15
.
Am. J. Med. Genet. A
 ,
143A
,
3010
3015
.
14
Myers
S.
,
Bottolo
L.
,
Freeman
C.
,
McVean
G.
,
Donnelly
P.
(
2005
)
A fine-scale map of recombination rates and hotspots across the human genome
.
Science
 ,
310
,
321
324
.
15
Petes
T.D.
(
2001
)
Meiotic recombination hot spots and cold spots
.
Nat. Rev. Genet.
 ,
2
,
360
369
.
16
Fungtammasan
A.
,
Walsh
E.
,
Chiaromonte
F.
,
Eckert
K.A.
,
Makova
K.D.
(
2012
)
A genome-wide analysis of common fragile sites: what features determine chromosomal instability in the human genome
?
Genome Res.
 ,
22
,
993
1005
.
17
Baudat
F.
,
Imai
Y.
,
de Massy
B.
(
2013
)
Meiotic recombination in mammals: localization and regulation
.
Nat. Rev. Genet.
 ,
14
,
794
806
.
18
Yamazawa
K.
,
Nakabayashi
K.
,
Matsuoka
K.
,
Masubara
K.
,
Hata
K.
,
Horikawa
R.
,
Ogata
T.
(
2011
)
Androgenetic/biparental mosaicism in a girl with Beckwith–Wiedemann syndrome-like and upd(14)pat-like phenotypes
.
J. Hum. Genet.
 ,
56
,
91
93
.
19
Kalish
J.M.
,
Conlin
L.K.
,
Bhatti
T.R.
,
Dubbs
H.A.
,
Harris
M.C.
,
Izumi
K.
,
Mostoufi-Moab
S.
,
Mulchandani
S.
,
Saitta
S.
,
States
L.J.
et al
. (
2013
)
Clinical features of three girls with mosaic genome-wide paternal uniparental isodisomy
.
Am. J. Med. Genet. A
 ,
161A
,
1929
1939
.
20
Inbar-Feigenberg
M.
,
Choufani
S.
,
Cytrynbaum
C.
,
Chen
Y.A.
,
Steele
L.
,
Shuman
C.
,
Ray
P.N.
,
Weksberg
R.
(
2013
)
Mosaicism for genome-wide paternal uniparental disomy with features of multiple imprinting disorders: diagnostic and management issues
.
Am. J. Med. Genet. A
 ,
161A
,
13
20
.
21
Gogiel
M.
,
Begemann
M.
,
Spengler
S.
,
Soellner
L.
,
Goretzlehner
U.
,
Eggermann
T.
,
Strobl-Wildemann
G.
(
2013
)
Genome-wide paternal uniparental disomy mosaicism in a woman with Beckwith–Wiedemann syndrome and ovarian steroid cell tumour
.
Eur. J. Hum. Genet.
 ,
21
,
788
791
.
22
Romanelli
V.
,
Nevado
J.
,
Fraga
M.
,
Trujillo
A.M.
,
Mori
M.
,
Fernández
L.
,
Pérez de Nanclares
G.
,
Martínez-Glez
V.
,
Pita
G.
,
Meneses
H.
et al
. (
2011
)
Constitutional mosaic genome-wide uniparental disomy due to diploidisation: an unusual cancer-predisposing mechanism
.
J. Med. Genet.
 ,
48
,
212
216
.
23
Wilson
M.
,
Peters
G.
,
Bennetts
B.
,
McGillivray
G.
,
Wu
Z.H.
,
Poon
C.
,
Algar
E.
(
2008
)
The clinical phenotype of mosaicism for genome-wide paternal uniparental disomy: two new reports
.
Am. J. Med. Genet. A
 ,
146A
,
137
148
.
24
Giurgea
I.
,
Sanlaville
D.
,
Fournet
J.C.
,
Sempoux
C.
,
Bellanné-Chantelot
C.
,
Touati
G.
,
Hubert
L.
,
Groos
M.S.
,
Brunelle
F.
,
Rahier
J.
et al
. (
2006
)
Congenital hyperinsulinism and mosaic abnormalities of the ploidy
.
J. Med. Genet.
 ,
43
,
248
254
.
25
Bryke
C.
,
Garber
A.
,
Israel
J.
(
2004
)
Evolution of a complex phenotype in a unique patient with a paternal uniparental disomy for every chromosome cell line and a normal biparental inheritance cell line. The Annual Meeting of The American Society of Human Genetics, Toronto, Canada, Program Nr. 823
.
26
Hoban
P.R.
,
Heighway
J.
,
White
G.R.
,
Baker
B.
,
Gardner
J.
,
Birch
J.M.
,
Morris-Jones
P.
,
Kelsey
A.M.
(
1995
)
Genome-wide loss of maternal alleles in a nephrogenic rest and Wilms’ tumour from a BWS patient
.
Hum. Genet.
 ,
95
,
651
656
.
27
Reed
R.C.
,
Beischel
L.
,
Schoof
J.
,
Johnson
J.
,
Raff
M.L.
,
Kapur
R.P.
(
2008
)
Androgenetic/biparental mosaicism in an infant with hepatic mesenchymal hamartoma and placental mesenchymal dysplasia
.
Pediatr. Dev. Pathol.
 ,
11
,
377
383
.
28
Johnson
J.P.
,
Waterson
J.
,
Schwanke
C.
,
Schoof
J.
(
2014
)
Genome-wide androgenetic mosaicism
.
Clin. Genet.
 ,
85
,
282
285
.
29
Lisby
M.
,
Rothstein
R.
(
2015
)
Cell biology of mitotic recombination
.
Cold Spring Harb. Perspect. Biol.
 ,
7
,
a016535
.
30
Baudat
F.
,
Buard
J.
,
Grey
C.
,
Fledel-Alon
A.
,
Ober
C.
,
Przeworski
M.
,
Coop
G.
,
de Massy
B.
(
2010
)
PRDM9 is a major determinant of meiotic recombination hotspots in humans and mice
.
Science
 ,
327
,
836
840
.
31
Cooper
W.N.
,
Curley
R.
,
Macdonald
F.
,
Maher
E.R.
(
2007
)
Mitotic recombination and uniparental disomy in Beckwith–Wiedemann syndrome
.
Genomics
 ,
89
,
613
617
.
32
Eggermann
T.
,
Heilsberg
A.K.
,
Bens
S.
,
Siebert
R.
,
Beygo
J.
,
Buiting
K.
,
Begemann
M.
,
Soellner
L.
(
2014
)
Additional molecular findings in 11p15-associated imprinting disorders: an urgent need for multi-locus testing
.
J. Mol. Med. (Berl)
 ,
92
,
769
777
.
33
Elliott
M.
,
Bayly
R.
,
Cole
T.
,
Temple
I.K.
,
Maher
E.R.
(
1994
)
Clinical features and natural history of Beckwith–Wiedemann syndrome: presentation of 74 new cases
.
Clin. Genet.
 ,
46
,
168
174
.
34
DeBaun
M.R.
,
Tucker
M.A.
(
1998
)
Risk of cancer during the first four years of life in children from The Beckwith–Wiedemann Syndrome Registry
.
J. Pediatr.
 ,
132
,
398
400
.
35
Weksberg
R.
,
Shuman
C.
,
Beckwith
J.B.
(
2010
)
Beckwith–Wiedemann syndrome
.
Eur. J. Hum. Genet.
 ,
18
,
8
14
.
36
Higashimoto
K.
,
Jozaki
K.
,
Kosho
T.
,
Matsubara
K.
,
Fuke
T.
,
Yamada
D.
,
Yatsuki
H.
,
Maeda
T.
,
Ohtsuka
Y.
,
Nishioka
K.
et al
. (
2014
)
A novel de novo point mutation of the OCT-binding site in the IGF2/H19-imprinting control region in a Beckwith–Wiedemann syndrome patient
.
Clin. Genet.
 ,
86
,
539
544
.
37
Sasaki
K.
,
Soejima
H.
,
Higashimoto
K.
,
Yatsuki
H.
,
Ohashi
H.
,
Yakabe
S.
,
Joh
K.
,
Niikawa
N.
,
Mukai
T.
(
2007
)
Japanese and North American/European patients with Beckwith–Wiedemann syndrome have different frequencies of some epigenetic and genetic alterations
.
Eur. J. Hum. Genet.
 ,
15
,
1205
1210
.