1. Timing, rates and spectra of human germline mutation
Becca Ahrens-Nicklas
Genetics Journal Club
2/25/2016

2. Why do de novo mutations matter?
There has been an increasing realization that de novo mutations cause disease
This includes numerous syndromes: Noonan, CHARGE, etc.
Also de novo changes are enriched in populations with more common disorders including congenital heart disease, intellectual disability, autism, etc…

3. Previous estimates of de novo mutation rates
Nature Reviews Genetics 13, (August 2012)

4. De novo mutations and mosaicism
Somatic
Gonosomal
Germline
Nature Reviews Genetics 14, (May 2013)

5. What influences de novo mutation frequencies? Paternal age
Ann. Eugenics (1947)

6. De novo mutations when do they happen?
Males 160 genome replications at 20, 610 by age 40

7. What influences de novo mutation frequencies?
Nature Reviews Genetics 13, (August 2012)

9. Major questions addressed
How often and when do de novo mutations arise?
How frequently is germline mosaicism in parents the cause of apparent de novo mutations?
What does the germline mutational spectra look like? Does it vary between paternally or maternally inherited mutations?

10. Major questions addressed
How often and when do de novo mutations arise?
How frequently is germline mosaicism in parents the cause of apparent de novo mutations?
What does the germline mutational spectra look like? Does it vary between paternally or maternally inherited mutations?

11. Methods to quantify de novo mutations (DNM)
Genome sequencing of three multisibling families, 24.7X coverage
4,881 candidate mutations
Resequence 4,141 sites at 139X coverage
768 germline DNM *
Inherited Variant
False positive
False Positives:
1) Located in segmental duplicaitons and simple repeats
2) >5% reads show alternative allele in one parent
3) Located in regions with high numbers of misaligned reads
49,893 candidate DNMs

12. Methods to quantify de novo mutations (DNM)
Genome sequencing of three multisibling families, 24.7X coverage
4,881 candidate mutations
Resequence 4,141 sites at 139X coverage
768 germline DNM *
Inherited Variant
False positive
False Positives:
1) Located in segmental duplicaitons and simple repeats
2) >5% reads show alternative allele in one parent
3) Located in regions with high numbers of misaligned reads
49,893 candidate DNMs
*Final numbers corrected for inaccessible regions,
83.1% of genome was accessible

13. How often do de novo mutations arise?
On average there are 64 (range 43-84) DNMs per child, when this number is adjusted for inaccessible genomic regions, it increases to 76.9 per child
Average mutation rate of 1.28 x mutations per nucleotide per generation at a mean paternal age of years

14. How often do de novo mutations arise?

15. How often do de novo mutations arise?
* 78% of DNM are of paternal origin

16. How often do de novo mutations arise?
* The # of DNM increases with paternal age by 2.87 mutations/ year

17. The paternal age effect on rate of de novo mutation varies between families

18. Major questions addressed
How often and when do de novo mutations arise?
How frequently is germline mosaicism in parents the cause of apparent de novo mutations?
What does the germline mutational spectra look like? Does it vary between paternally or maternally inherited mutations?

19. Mosaicism as an etiology of de novo mutations
Somatic
Gonosomal
Germline
Nature Reviews Genetics 14, (May 2013)

20. Mosaicism as an etiology of de novo mutations: Suspicious pedigree
* Multiple affected children from an apparently unaffected parent

21. Mosaicism as an etiology of de novo mutations: Suspicious pedigree
* Multiple affected children from an apparently unaffected parent

22. Methods to quantify mosaicism as the cause of de novo mutations (DNM)
Deep sequencing of all (768) validated DNM at 567x coverage
Identification of recurrent DNMs shared by at least 2 siblings, that are not constitutively heterozygous in either parent
Identification DNMs with significant excess of reads in a single parent. This could suggest low level mosaicism in blood.

23. Methods to quantify mosaicism as the cause of de novo mutations (DNM)
Deep sequencing of all (768) validated DNM at 567x coverage
Identification of recurrent DNMs shared by at least 2 siblings, that are not constitutively heterozygous in either parent
Identification DNMs with significant excess of reads in a single parent. This could suggest low level mosaicism in blood.

24. Estimation of recurrence risk of any apparent de novo mutation: 1.2%

25. Methods to quantify mosaicism as the cause of de novo mutations (DNM)
Deep sequencing of all (768) validated DNM at 567x coverage
Identification of recurrent DNMs shared by at least 2 siblings, that are not constitutively heterozygous in either parent
Identification DNMs with significant excess of reads in a single parent. This could suggest low level mosaicism in blood.
4 DNM
6 DNM
19 DNM

26. Estimation of power to detect gonosomal mosaicism
80% power to detect 1% mosaicism in paternal blood
90% power to detect 2% mosaicism in paternal blood

27. How many de novo mutations are actually due to mosaicism
How many de novo mutations are actually due to mosaicism? At what level of mosaicism are they present?

28. How many de novo mutations are actually due to mosaicism
How many de novo mutations are actually due to mosaicism? At what level of mosaicism are they present?
4 variants only seen in siblings, not detectable in low level mosaicism in a parent’s blood

29. How many de novo mutations are actually due to mosaicism
How many de novo mutations are actually due to mosaicism? At what level of mosaicism are they present?
6 variants shared by siblings and were detectable in low level mosaicism in a parent’s blood

30. How many de novo mutations are actually due to mosaicism
How many de novo mutations are actually due to mosaicism? At what level of mosaicism are they present?
19 variants not shared by siblings but were detectable in low level mosaicism in a parent’s blood

31. Take home messages using the data from this table:
How many de novo mutations are actually due to mosaicism? At what level of mosaicism are they present?
Take home messages using the data from this table:
1) Recurrence risk of an apparent de novo mutation is 1.3%
2) 4.2% of germline mutations may be mosaic in >1% of parental blood cells

32. We know that de novo mutations more often come from dad… is this true for germline mosaic mutations?
Remember among the total 768 de novo mutations the ratio of paternal to maternal inheritance was 3.5:1
BUT among the 25 de novo mutations with detectable parental mosaicism the ratio of paternal to maternal inheritance was 1:1

33. Mosaic apparent de novo mutations occur early in development
Mosaic apparent de novo mutations occur early in development. These have equal likelihood in both sexes because there are similar numbers of replications at these early stages.

34. Limitations of the study
Mutations that arise in very early post-zygotic cell divisions may have such high levels of somatic mosaicism that they would be missed in this study.
Conversely, some mutations may be present in blood (or other tissue) but at levels of mosaicism below the level of detection of this study.

35. Major questions addressed
How often and when do de novo mutations arise?
How frequently is germline mosaicism in parents the cause of apparent de novo mutations?
What does the germline mutational spectra look like? Does it vary between paternally or maternally inherited mutations?

36. Methods to characterize the mutational spectra of de novo mutations (DNM)
Compiled data from 6 sources to obtain 6,570 high-confidence DMNs,
10% had known parental inheritance data
Low resolution mutational spectra analysis-
Calculated the frequency of the 6 basic point mutations types
High resolution mutational spectra analysis-
Categorized each DNM based on the mutation and the 3’ and 5’ base.
Calaculated the frequency of each possibility
Compared the pattern of mutation types to known “mutational signatures” from the cancer field

37. Definitions of types of point mutations
DNA substitution mutations are of two types. Transitions are interchanges of two-ring purines (A G) or of one-ring pyrimidines (C T): they therefore involve bases of similar shape. Transversions are interchanges of purine for pyrimidine bases, which therefore involve exchange of one-ring and two-ring structures.     Although there are twice as many possible transversions, because of the molecular mechanisms by which they are generated, transition mutations are generated at higher frequency  than transversions. As well, transitions are less likely to result in amino acid substitutions (due to "wobble"), and are therefore more likely to persist as "silent substitutions" in populations as single nucleotide polymorphisms (SNPs).

38. Transitions are more common than transversions
What type of de novo mutations are the most common? Low resolution spectra analysis
Transitions are more common than transversions
Remember CpG sites spontaneously deaminate leading to TpG sites, which increases the C>T mutations

39. What type of de novo mutations are the most common
What type of de novo mutations are the most common? Low resolution spectra analysis
Neither parent of origin nor age of the father change what type of de novo mutations occur.
Remember CpG sites spontaneously deaminate leading to TpG sites, which increases the C>T mutations

40. More evidence that parent of origin does not change the types of de novo mutations that occur
In a separate cohort of 2,453 individuals, all the SNV seen on the X and Y chromosomes were classified. There was no difference in the mutation spectra.
Remember CpG sites spontaneously deaminate leading to TpG sites, which increases the C>T mutations

41. The frequency of all 96 possible mutations triplets were calculated.
What type of de novo mutations are the most common? high resolution spectra analysis
The frequency of all 96 possible mutations triplets were calculated.
(96 is all of the combinations of the 6 basic point mutations + the flanking 3’ and 5’ bases)
Remember CpG sites spontaneously deaminate leading to TpG sites, which increases the C>T mutations

42. DNM published in this study
What type of de novo mutations are the most common? high resolution spectra analysis
This mutational signature of de novo mutations was then compared to a catalog of published mutational signatures of somatic mutations in different cancers.
Catalog of mutational signatures
Signature of
DNM published in this study
Remember CpG sites spontaneously deaminate leading to TpG sites, which increases the C>T mutations
Nature. 500, (Aug 2013)

43. DNMs published in this study
What type of de novo mutations are the most common? high resolution spectra analysis
This mutational signature of de novo mutations was then compared to a catalog of published mutational signatures of somatic mutations in different cancers.
Catalog of mutational signatures
Signature of
DNMs published in this study
Remember CpG sites spontaneously deaminate leading to TpG sites, which increases the C>T mutations
* DNMs match signature 1 and 5 which are the signatures of the majority of sponatneous preneoplastic somatic mutations

44. What type of de novo mutations are the most common
What type of de novo mutations are the most common? high resolution spectra analysis
The same mutational processes underlying preneoplastic somatic mutations operate in the germ cell to yield DNM.
Remember CpG sites spontaneously deaminate leading to TpG sites, which increases the C>T mutations

45. What type of de novo mutations are the most common
What type of de novo mutations are the most common? Effect of germline methylation status
Methylated CpG sites spontaneously deaminate, leading to TpG sites (C:G>T:A mutations)
To detect if methylated sites were common sites of DNM, the authors obtained cell-line methylation data for a testis cell line (from the ENCODE database)
23.5% of all CpG sites in the testis database were methylated
Among CpG sites both present in the testes database and found as DNM in this study 12/13 (92%) were methylated
Remember CpG sites spontaneously deaminate leading to TpG sites, which increases the C>T mutations

46. What type of de novo mutations are the most common
What type of de novo mutations are the most common? Effect of germline methylation status
Methylated CpG sites spontaneously deaminate, leading to TpG sites (C:G>T:A mutations)
To detect if methylated sites were common sites of DNM, the authors obtained cell-line methylation data for a testis cell line (from the ENCODE database)
23.5% of all CpG sites in the testis database were methylated
Among CpG sites both present in the testes database and found as DNM in this study 12/13 (92%) were methylated
Remember CpG sites spontaneously deaminate leading to TpG sites, which increases the C>T mutations
Therefore methylated CpG sites in germ cells are more likely than
unmethylated sites to give rise to de novo mutations.

47. Conclusions 1 How often and when do de novo mutations arise?
The rate of DNM is 1.28 x 10-8 mutations per nucleotide per generation.
The ratio of paternal : maternal origin is 3.5 : 1. Presumably this is due to larger numbers of post-pubertal cell divisions in male germ cells.
On average the number of DNM in a child increases by 2.9 mutations / year of paternal age. (But this can vary between families).

48. Conclusions 2
How frequently is germline mosaicism in parents the cause of apparent de novo mutations?
Recurrence risk of an apparent de novo mutation is 1.3%
But this increases to 24% for DNMs that are mosaic at > 1% in parental blood
And increases to 50% for DNMs that are mosaic at > 6% in parental blood
Should deep sequencing be used to stratify real recurrence risk?
Among all de novo mutations, paternal origin is 3x more common, but in the case of parental mosaicism there is no sex difference.
So, in a patient with a maternally-derived mutation there is a higher risk it is due to maternal mosaicism, and therefore a higher recurrence risk.

49. Conclusions 3
What does the germline de novo mutational spectra look like?
Transitions are much more common than transversions.
Mutational spectra among germline DNM are very similar to those seen in somatic mutations that lead to neoplams. This suggests similar mutational mechanisms.
Methylated CpG sites are hotspots for mutations in germ cells.

50. Conclusions 3
What does the germline de novo mutational spectra look like?
Transitions are much more common than transversions.
Mutational spectra among germline DNM are very similar to those seen in somatic mutations that lead to neoplams. This suggests similar mutational mechanisms.
Methylated CpG sites are hotspots for mutations in germ cells.

51. A model of germline mutation rates

52. Questions, thoughts?

53. Supplementary Table 1